Luteolysis: A Neuroendocrine-Mediated Event

John A. McCracken, Edward E. Custer, Justin C. Lamsa

Abstract

In many nonprimate mammalian species, cyclical regression of the corpus luteum (luteolysis) is caused by the episodic pulsatile secretion of uterine PGF, which acts either locally on the corpus luteum by a countercurrent mechanism or, in some species, via the systemic circulation. Hysterectomy in these nonprimate species causes maintenance of the corpora lutea, whereas in primates, removal of the uterus does not influence the cyclical regression of the corpus luteum. In several nonprimate species, the episodic pattern of uterine PGF secretion appears to be controlled indirectly by the ovarian steroid hormones estradiol-17β and progesterone. It is proposed that, toward the end of the luteal phase, loss of progesterone action occurs both centrally in the hypothalamus and in the uterus due to the catalytic reduction (downregulation) of progesterone receptors by progesterone. Loss of progesterone action may permit the return of estrogen action, both centrally in the hypothalamus and peripherally in the uterus. Return of central estrogen action appears to cause the hypothalamic oxytocin pulse generator to alter its frequency and produce a series of intermittent episodes of oxytocin secretion. In the uterus, returning estrogen action concomitantly upregulates endometrial oxytocin receptors. The interaction of neurohypophysial oxytocin with oxytocin receptors in the endometrium evokes the secretion of luteolytic pulses of uterine PGF. Thus the uterus can be regarded as a transducer that converts intermittent neural signals from the hypothalamus, in the form of episodic oxytocin secretion, into luteolytic pulses of uterine PGF. In ruminants, portions of a finite store of luteal oxytocin are released synchronously by uterine PGF pulses. Luteal oxytocin in ruminants may thus serve to amplify neural oxytocin signals that are transduced by the uterus into pulses of PGF. Whether such amplification of episodic PGF pulses by luteal oxytocin is a necessary requirement for luteolysis in ruminants remains to be determined. Recently, oxytocin has been reported to be produced by the endometrium and myometrium of the sow, mare, and rat. It is possible that uterine production of oxytocin may act as a supplemental source of oxytocin during luteolysis in these species. In primates, oxytocin and its receptor and PGF and its receptor have been identified in the corpus luteum and/or ovary. Therefore, it is possible that oxytocin signals of ovarian and/or neural origin may be transduced locally at the ovarian level, thus explaining why luteolysis and ovarian cyclicity can proceed in the absence of the uterus in primates. However, it remains to be established whether the intraovarian process of luteolysis is mediated by arachidonic acid and/or its metabolite PGF and whether the central oxytocin pulse generator identified in nonprimate species plays a mediatory role during luteolysis in primates. Regardless of the mechanism, intraovarian luteolysis in primates (progesterone withdrawal) appears to be the primary stimulus for the subsequent production of endometrial prostaglandins associated with menstruation. In contrast, luteolysis in nonprimate species appears to depend on the prior production of endometrial prostaglandins. In primates, uterine prostaglandin production may reflect a vestigial mechanism that has been retained during evolution from an earlier dependence on uterine prostaglandin production for luteolysis.

I. INTRODUCTION

This review concerns the phenomenon of luteolysis or regression of the corpus luteum, which terminates the female reproductive cycle of many mammals. Luteolysis is characterized by an initial decline of progesterone secretion that is commonly designated as functional luteolysis as distinct from structural or morphological luteolysis which, as the name suggests, signifies the subsequent change in the cellular structure of the gland and its gradual involution in the ovary to form a small scar composed of connective tissue. This latter structure, known as the corpus albicans, persists in the ovary, often for several weeks. However, the distinction between functional and structural regression of the corpus luteum is not well defined and, indeed, they may constitute part of an ongoing process. Luteolysis appears to have evolved in some species as a mechanism to increase reproductive efficiency. In most regularly cycling mammals when the female does not conceive following ovulation, the corpus luteum which forms subsequently in the ovary is “removed” to permit a new ovarian cycle to begin. In this way, after a relatively short interval of time, a further opportunity is provided for the female to conceive. In a few mammals, typified by the dog, no mechanism appears to exist to curtail the life span of the corpus luteum and, in unmated females or those failing to conceive, the corpora lutea last for a period approaching the length of pregnancy. The main secretory product of the corpus luteum in most mammals is the steroid hormone progesterone, which is the hormone of pregnancy in mammals. Progesterone is considered to be essential for pregnancy maintenance by inducing a state of quiescence in the myometrium (128) and by suppressing the maternal immune response to fetal antigens (626,627, 666). In addition to providing a uterine environment suitable for the development of the embryo, progesterone is also responsible for the reduction of cyclic ovarian activity during pregnancy in most mammals and is responsible, in part, for mammary development. The corpus luteum also plays a key role in regulating the length of the ovarian cycle in most cyclic mammals, and the extension of the life span of the corpus luteum and progesterone secretion is necessary in most species to maintain gestation in its early stages.

Although progesterone is the hormone which, from the evolutionary point of view, is responsible for viviparity, and in some cases ovoviviparity (98), this review discusses the regulation of the life span of the corpus luteum in cyclic mammals. Progesterone is also produced by the placenta in some species, and in these, the placenta usually becomes the dominant source of progesterone during the later stages of pregnancy, e.g., sheep, horse, and human. Such a change in the source of progesterone for pregnancy maintenance is designated as the luteoplacental shift. In other species such as the goat, pig, rabbit, and mouse, the corpus luteum (or in polyovulatory species, corpora lutea) remains the principal source of progesterone throughout pregnancy. In some of these species, such as the goat and mouse, the initiation of labor appears to involve a luteolytic mechanism. In species where the corpus luteum is the sole source of progesterone, ovariectomy (removal of the corpus luteum) will terminate pregnancy at any stage, whereas in those species where the placenta produces progesterone, the corpus luteum can be removed when placental production is high enough to maintain pregnancy. In women with ovarian dysgenesis, a donor fertilized ovum will develop in the uterus when exogenous progesterone and estrogen are given to replace the secretion by the corpus luteum. When placental steroid production is adequate to maintain pregnancy, pregnancy will continue to term without further administration of exogenous hormones (431, 512).

A.  Early History of the Corpus Luteum

Coiter (117) described the presence of cavities filled with a yellow solid in the ovary, but it was de Graaf (146) who gave the first definitive description of these structures noting that their number appeared to be related to the number of fetuses in utero. Malpighi (440) provided an accurate microscopic description of these structures and was the first to apply the name corpus luteum, literally the yellow body. Beard (51) postulated that corpora lutea were responsible for the suppression of ovulation and estrus during pregnancy, and about that time, Prenant (566) suggested that the corpus luteum might be a gland of internal secretion directly benefiting the egg with which it appeared to be associated. It was, however, Fraenkel (218), a pupil of Gustav Born, who demonstrated that corpora lutea were necessary for implantation and the subsequent maintenance of pregnancy in the rabbit. Corner and Allen (124) and Allen and Corner (14) prepared a relatively pure alcoholic extract of corpora lutea from sows and showed that this extract maintained pregnancy in ovariectomized rabbits. A few years later, the isolation of the pure crystalline hormone was reported simultaneously by four groups (93, 290, 634, 731). Slotta et al. (634) named the compound progesterone and suggested a structural formula, and in the same year, the compound was synthesized by Butenandt and Westphal (92).

B.  Corpus Luteum of the Ovarian Cycle/Classification of Cycles

There is probably no area of mammalian physiology where interspecies variation is so manifest as in the endocrine regulation of the ovarian cycle. To appreciate the comparative aspects of luteolysis in controlling the reproductive cycle in mammals, a brief classification and description of the various types of cycles are outlined below. A more detailed description of reproductive cycles in mammals is described elsewhere (24, 622). Ovarian cycles of the more common eutherian mammals can be divided into five groups as follows: 1) primate menstrual cycle, e.g., monkey and humans; 2) domestic animal cycles, e.g., sheep, cow, goat, pig, horse, and guinea pig; 3) laboratory rodent cycles, e.g., rat, mouse, and hamster; 4) reflex ovulators, e.g., rabbit, cat, mink, ferret, and camel; and 5) canine cycle, e.g., dog and wolf.

As shown in Figure 1, the phenomenon of luteolysis occurs in all species in groups 1 and2, whereas in groups 3 and 4, luteolysis occurs in some species primarily as a means of terminating pseudopregnancy. In group 5, there is no evidence for the occurrence of luteolysis so that the corpus luteum persists for a period equivalent to that of pregnancy.

Fig. 1.

Occurrence of luteolysis for controlling life span of corpus luteum during cycle or during pseudopregnancy in different species. E, estradiol-17β; P, progesterone; LH, luteinizing hormone.

1.  Primate menstrual cycle

In primates, ovulation and corpus luteum formation occur under gonadotropic control. Repetitive ovarian cycles occur with distinct follicular and luteal phases. Toward the end of the luteal phase of an infertile cycle, the corpus luteum, which produces both progesterone and estrogen, undergoes luteolysis, thus terminating the cycle. The withdrawal of progesterone causes the endometrial pseudodecidua to degenerate resulting in menstruation. Menstruation occurs only in humans and old world nonhuman primates because, unlike the new world primates, they have a spiral arteriolar complex in the uterus, the contraction of which precipitates endometrial shedding. In nonhuman primates, the midcycle preovulatory surge of estrogen causes an increase in sexual receptivity (estrus), whereas in the human female, there is no overt increase in sexual receptivity at this time. Thus nonhuman primates exhibit a transitional form of both estrous and menstrual cycles. After a fertile mating in primates, the corpus luteum is rescued by chorionic gonadotropin (CG) secreted by the implanting blastocyst. Progesterone continues to be secreted by the corpus luteum to maintain pregnancy until the placenta assumes this function, i.e., the luteoplacental shift.

2.  Domestic animal cycles

Ovulation and corpus luteum formation are under gonadotropic control, and repetitive cycles occur with a relatively short follicular phase and a relatively long luteal phase. During the brief follicular phase, the rise of estrogen from the preovulatory follicle(s) induces a short period of sexual receptivity (estrus) as well as inducing the preovulatory surge of luteinizing hormone (LH). In this way, mating is synchronized with ovulation. In the infertile cycle, the corpus luteum undergoes luteolysis which terminates the luteal phase and a new ovarian cycle is initiated. In the event of a fertile mating, the uterine luteolytic signal is subverted and the corpus luteum continues to secrete progesterone for the maintenance of pregnancy. In some species (e.g., pig, goat, and mouse), the corpus luteum is the sole source of progesterone throughout pregnancy, whereas in others (e.g., sheep and horse), the placenta later becomes the major source of progesterone for the maintenance of pregnancy. The guinea pig is included in this category because its reproductive cycle is very similar to domestic animals.

3.  Laboratory rodent cycles

Laboratory rodents, e.g., mouse, rat, and hamster, have developed a reproductive strategy that allows them to ovulate and thus potentially conceive every 4–5 days. The high frequency of ovulation is possible because these mammals, while they ovulate spontaneously, do not develop a fully functional secretory corpus luteum. Thus there is no inhibition of gonadotropin activity, which allows follicular development and ovulation to occur within a few days. Laboratory rodents require coital stimulation to produce a fully functional secretory corpus luteum either during pregnancy, in the event of a fertile mating, or during pseudopregnancy in the case of a sterile mating. However, most rodents develop a luteolytic mechanism that curtails the length of pseudopregnancy. Thus reproduction in these rodents is highly efficient since, by not spontaneously developing a luteal phase (i.e. unless they are mated), they have an opportunity to conceive every 4–5 days.

4.  Induced or reflex ovulatory cycles

Waves of follicular development occur during which there is a relatively long period of sexual receptivity. A mating stimulus is required for the neurogenic release of LH and thus induction of ovulation and the development of corpora lutea. The corpora lutea are maintained in both fertile matings (pregnancy) and infertile matings (pseudopregnancy). The latter may last as long as a normal pregnancy (e.g., ferret) or may be terminated prematurely by a luteolytic mechanism (e.g., rabbit). Generally, these species are reproductively efficient, since ovulation is precipitated only by a mating stimulus.

5.  Canine cycle

Canines exhibit only about two ovulatory cycles per year. Ovulation and corpus luteum formation occur under gonadotropic control. A functional corpus luteum is formed that secretes progesterone throughout pregnancy after a fertile mating. In unmated animals or after an infertile mating, the corpus luteum persists for a period approximating the normal duration of pregnancy. During pseudopregnancy, or so-called phantom pregnancy, females exhibit nest building and mammary development and often begin to lactate. Hysterectomy has no effect on the life span of corpora lutea in the dog, although it may shorten the duration of anestrus (316). It appears that a luteolytic mechanism has not evolved in these animals to curtail the life span of the corpus luteum when pregnancy fails to occur, and no signal from the embryo is required to extend the life span of the corpus luteum. Reproduction in these animals is rather inefficient, since they do not have repetitive cycles and thus they have an opportunity to conceive only about twice per year.

In marsupials such as the red kangaroo, “uterine” pregnancy is shorter than the length of the ovarian cycle, and the presence of an embryo in the marsupial pouch does not extend the cycle or inhibit subsequent ovulation (681). This is in contrast to mammals that show spontaneous cycles, ovulation, corpus luteum formation, and regression where the presence of an embryo in the uterus is required to extend the life span of the corpus luteum for varying periods of time during pregnancy. [For reviews on the effects of the embryo on corpus luteum life span, see Moor (499), Heap et al. (296), Webley and Hearn (708), Stouffer et al. (658), and Hamberger et al. (282); see also sect. vi]. It is likely that these differences in reproductive patterns are meaningful in terms of the adaption of species to reproduce more efficiently under different environments and life patterns. Comparative aspects of corpus luteum development and function have been reviewed previously (357, 524, 591).

C.  Development of the Corpus Luteum

Under the influence of the preovulatory surge of LH from the anterior pituitary, the mature follicle ruptures and expels the ovum. The wall of the follicle collapses in folds, and capillaries invade the developing corpus luteum probably under the influence of angiogenic and mitogenic factors that may include basis fibroblast growth factor (261, 517), platelet-derived growth factor (358), insulin-like factor I (661, 349), heparin binding growth factor (264), and vascular endothelial growth factor (572). In domestic animals and primates, the major luteotropin appears to be LH, a luteotropin being defined as a substance that promotes the growth of the corpus luteum and stimulates the production of progesterone. For example, hypophysectomy in primates (220, 746) and in sheep (351, 353) causes regression of the corpus luteum, an effect which can be reversed by exogenous LH. Also, immunoneutralization of circulating LH with antibodies against bovine LH causes regression of the corpus luteum in sheep (231,455; see Fig. 2). In primates, progesterone secretion is dependent on the pulsatile secretion of LH throughout the luteal phase (220, 746), whereas in the sheep, secretion of progesterone appears to be independent of LH pulses, since secretion can be maintained with minimal basal levels of LH (478). It may be that pulsatile LH secretion is not an absolute requirement per se for corpus luteum function in primates, since GC is released in a continuous nonpulsatile fashion but clearly maintains the corpus luteum during the establishment of pregnancy (see sect. vi). Luteinizing hormone pulses appear to be necessary in the cow for development of a fully functional corpus luteum but, as in the sheep, only basal levels are required to maintain progesterone secretion later in the luteal phase (548). The reduction in progesterone during luteolysis removes the progesterone blockade of tonic gonadotropin releasing hormone (GnRH)/LH pulses, and the consequent increase in pulse frequency promotes the growth of the preovulatory follicle and increases estrogen secretion for the next cycle. Although pituitary prolactin released by a mating stimulus in rodents is critical for maintaining the corpus luteum in a progesterone-secreting mode (225), there is little evidence to support a role for prolactin in luteal function in ruminants or primates. For example, in sheep, the infusion of ovine prolactin into the ovarian artery did not stimulate progesterone secretion (473). Moreover, the administration of 2-bromo-α-ergocryptine, a potent inhibitor of prolactin secretion, did not affect progesterone levels or the cycle length (521). In primates, prolactin does not alter progesterone production by dispersed luteal cells from monkeys (656) or women (668) in the presence or absence of human CG (hCG). However, in the canine, both prolactin and LH are required to maintain the secretory function of the corpus luteum (121).

Fig. 2.

Progesterone secretion rate (solid bar) and ovarian blood flow (●) after systemic infusion of antiserum against bovine LH (3.84 ml/h for 12 h). Progesterone secretion rate showed a marked, but continuous, decline within 1 h of beginning infusion. [From McCracken et al. (455).]

In primates, the corpus luteum is a significant source of inhibin, a heterodimeric glycoprotein, which was initially found in the developing follicle and having an ascribed function of inhibiting follicle-stimulating hormone (FSH) secretion. Plasma levels of inhibin are elevated during the luteal phase in primates (476) and are reduced by GnRH antagonists and stimulated by hCG (640). It is possible that luteal inhibin in primates acts to block the release of pituitary FSH, thus accounting for the lack of follicular development during the luteal phase, whereas in species that do not produce large amounts of luteal inhibin, waves of follicular development occur during the luteal phase, e.g., sheep and cow. Different forms of inhibin appear to be secreted by the corpus luteum (inhibin A) versus the follicle (inhibin B) (269). Moreover, it has been shown that inhibin A will reduce bioactive FSH blood levels and prevent follicular development in monkeys (498). In primates, the “rescue” of the human corpus luteum with exogenous hCG results in increased inhibin production that may extend the inhibition of follicular development in early pregnancy (331). However, there is no evidence to suggest that inhibins are involved in the process of luteolysis in primates, at least in the marmoset monkey (223). Relaxin is also produced by the cyclic corpus luteum in several species including human (714), pig (620), and rat (258). However, the function of luteal relaxin is presently unclear, and no specific action of relaxin has so far been associated with luteolysis. Oxytocin is synthesized by the corpus luteum of some species, notably ruminants, and may play a contributory role in luteolysis (see sects.iii and iv).

The overall rate of growth of the corpus luteum in most species is extremely rapid. For example, in the cow, the weight of the corpus luteum 3 days after ovulation averages 640 mg, whereas on day 14, the average weight is 5.1 g (201). Most of this rapid increase in mass is due to hypertrophy of granulosa and theca cells as well as some mitotic division of the latter. Also, there is a rapid mitotic division and growth of endothelial cells and fibroblasts. As measured by 85 rypton clearance, the mature ovine corpus luteum has a high capillary blood flow, on the order of 1 ml · g−1 · min−1, in keeping with its high metabolic activity (177). In addition to connective tissue fibrocytes and endothelial cells lining capillaries, most species have two types of steroidogenic cells, both of which secrete progesterone. The smaller steroidogenic cells (<20 μm) are thought to be derived from the theca interna, and the large steroidogenic cells (20–30 μm) are thought to be derived from the granulosa cells lining the follicle wall (12). There is also evidence that at least some small steroidogenic cells may be transformed into large steroidogenic cells as the corpus luteum matures (12, 195). The small steroidogenic cells secrete low basal levels of progesterone but respond to LH with an increase in progesterone production, whereas large steroidogenic cells secrete high basal levels of progesterone but are unresponsive to LH stimulation (204, 382). Two types of steroidogenic luteal cells have been identified in the corpus luteum of sheep (204), cow (687), sow (413), rat (635), rabbit (324), rhesus monkey (307), and human (528). However, in the mare, in which ovulation occurs into an ovulation fossa and which develops secondary corpora lutea during pregnancy, it appears that only the granulosa lutein cells contribute to the formation of the mature corpus luteum (83). The large steroidogenic cells of ruminants possess secretory granules containing oxytocin that may contribute to the luteolytic process (see sects. iii and iv). The cellular composition of the ovine corpus luteum is shown diagrammatically in Figure 3. It should be noted that, although the large steroidogenic cells constitute only 4% of the total number of cells, they constitute 25% of the cellular volume of the corpus luteum (195, 586).

Fig. 3.

Major cell types in ovine corpus luteum. Relative number of cells (%) and relative cell volume (%) of each cell type in corpus luteum are shown diagrammatically. Remaining number of cells (7%) are composed of plasma cells, lymphocytes, and granular leukocytes. Residual tissue volume (33%) is occupied by vascular lumina and intercellular spaces. Granulosa-derived large luteal cell constitutes 4% by number but 25% by volume of gland. [Based on data from Rodgers et al. (586) and Farin et al. (195).]

In contrast to nonprimate species, large and small steroidogenic cell types in the monkey are responsive to LH and CG stimulation in vitro, but as the luteal phase progresses, only the large cells are responsive (307). Human large and small steroidogenic cells also show different characteristics in that basal progesterone produced by large cells is only twice that of small cells (528), whereas in monkeys and sheep, the large cells produce more than 10 times the amount of progesterone as the small cells. Thus it is apparent that considerable differences exist among species as to the specific characteristics of large and small steroidogenic cells. More detailed information on the characteristics of large and small steroidogenic cell types in different species has been reviewed previously (227, 655).

II.  LUTEOLYSIS

A.  Role of the Uterus in Corpus Luteum Regression

A physiological explanation for the cyclical regression of the corpus luteum, which is observed most clearly in the group of mammals showing spontaneous ovulation and corpus luteum formation, proved to be elusive. Initially, because of the emphasis placed on the importance of pituitary support for the maintenance of the corpus luteum, it was considered that withdrawal of pituitary luteotrophins such as LH and/or prolactin might cause the cyclical regression of the corpus luteum. However, with the advent of more sophisticated methods for measuring circulating pituitary gonadotropins, it became apparent that, at the time of corpus luteum regression, measurable levels of these hormones were present in ruminants (455, 729) and primates (40, 329, 482). Thus it appeared that complete withdrawal of pituitary gonadotropins from the circulating blood was unlikely to be a general mechanism for the induction of corpus luteum regression. Moreover, in primates, the administration of GnRH pulses or LH itself does not prevent the timely regression of the corpus luteum (746). Thus not only is there no evidence for a complete withdrawal of gonadotropins, but even methods which sustain basal gonadotropin levels do not prevent luteolysis.

The importance of the uterus in the control of corpus luteum regression was first reported by Loeb (422, 423), who demonstrated that hysterectomy in the cyclic guinea pig abolished cycles and caused abnormal persistence of the corpora lutea. Similar effects were subsequently observed in the cyclic sheep, cow, pig, and mare and in the pseudopregnant hamster, rabbit, and rat [for a review on this topic, see Anderson et al. (21)]. In primates, however, hysterectomy has no effect on the length of the ovarian cycle or the life span of the corpus luteum either in women (52, 56, 166, 422,480) or in monkeys (89, 103, 298, 516). The persistence of corpora lutea in nonprimate species after hysterectomy suggested that the uterus produced a substance that caused the cyclical regression of corpora lutea, although this view was not universal (508, 522).

Unlike primates, all species that show corpus luteum maintenance after hysterectomy possess a bicornuate uterus. In most of these species, a unilateral effect of hysterectomy was demonstrated, i.e., removal of one uterine horn causes maintenance of corpora lutea in the ovary on the side on which hemihysterectomy was performed (21, 245, 452,455). That surgery per se caused the observed effects of hemihysterectomy was ruled out by observations made in sheep, cows, and guinea pigs with congenital absence of one uterine horn (21, 68,457). These subjects showed a persistence of the corpora lutea in the ovary on the side where the uterine horn was absent, provided that vascular connections with the remaining uterine horn were also absent (531). In contrast, in women with congenital absence of the uterus, no marked alteration in ovarian cyclicity was observed (84, 127, 137, 224, 474), thus supporting the finding that hysterectomy does not influence luteal function in primates.

B.  Identification of PGF as the Mammalian Luteolysin

1.  Sheep

The sheep has long been used as a model animal for the study of luteolysis, partly because the effects of hysterectomy on corpus luteum function are very pronounced and partly because the size of the sheep permits the collection of amounts of ovarian and uterine venous blood adequate for repetitive hormonal measurements (144, 158, 501,623). An important advance in understanding the role of the uterus in the regulation of luteolysis came with the development of the ovarian autotransplant model at the Worcester Foundation in 1966 (252, 253). Such a development was a logical extension of the use of ex vivo organ perfusion systems (302, 588, 715) and tissue superfusion (667) for the elucidation of steroid biosynthetic pathways as pioneered at the Worcester Foundation. After the success of the adrenal autotransplant model at the Howard Florey Institute in Melbourne, Australia (475), Gregory Pincus had the foresight to establish the ovarian autotransplant model at the Worcester Foundation, thanks to the collaboration and surgical skills of Dr. James Goding from the Melbourne adrenal group. The ovarian autotransplantation technique involves the excision of the ovary with its vascular pedicle and transplanting it with vascular anastomoses of the ovarian artery and the utero-ovarian vein into a jugulocarotid skin loop (Fig. 4). The remaining ovary is excised and discarded with the uterus remaining in situ (252, 253, 455). Such a model permits long-term access to the arterial and venous supply of the ovary in the conscious, unstressed animal. Substances of interest can be infused directly into the arterial supply of the ovary, and frequent samples of ovarian venous blood can be obtained readily (455, 473). Moreover, ovarian blood flow can be determined by timing the collection of a given volume of ovarian venous blood (Fig. 4). After measuring the concentration of a specific hormone in ovarian venous plasma, the direct ovarian secretion rate of the hormone can then be calculated as mass per unit time (253, 473). However, animals with ovarian autotransplants showed a prolonged luteal phase (>100 days) apparently because the ovary in the jugulocarotid loop was separated from the uterus in the abdomen. Such an observation indicated that the putative control of luteal regression by the uterus in the sheep could not be mediated via the systemic circulation. Although the ovarian autotransplant model proved valuable for investigating the direct intraovarian effect of pituitary gonadotropins on ovarian steroid secretion in vivo (473), the lack of regular ovarian cyclicity was a disadvantage. Therefore, a method for autotransplanting the uterus and ovary together as one block of tissue to jugulocarotid loops was developed (455, 458, 465). This procedure was based on a method for transplanting the uterus and ovary subcutaneously with vascular anastomoses to the carotid artery and jugular vein (288). When one uterine horn and its adjacent ovary were transplanted with vascular anastomoses to a jugulocarotid loop, regression of the corpus luteum occurred at the expected time and steroid secretion, and LH levels were similar to the normal cycling intact sheep (455, 458, 465). These transplantation studies not only supported previous findings in utero-ovarian relationships (21), but proved unequivocally in the sheep that the uterus and ovary had to be contiguous for the occurrence of cyclic regression of the corpus luteum and for a normal cyclic pattern of hormone secretion. Cross-circulation experiments between ovarian and utero-ovarian transplants indicated that uterine venous blood possessed luteolytic activity at the time of corpus luteum regression (451, 452, 455, 456). This finding was supported by evidence that ovine uterine venous plasma, collected at the time of corpus luteum regression, had a progesterone-suppressing effect when infused into an ovarian artery (42, 96).

Fig. 4.

Diagram of technique for intra-arterial infusion of autotransplanted ovary in sheep and periodic collection of ovarian venous blood. With inflation of pneumatic cuff above carotid arterial pressure, carotid arterial blood containing an infusate supplies ovary. [From McCracken et al. (473), reproduced by permission of the Society for Endocrinology.]

During the course of these cross-circulation experiments, Pharriss and Wyngarden (552) proposed that PGFmight fulfill the role of a uterine luteolysin because it was relatively abundant in the uterus (174, 553) and was a potent venoconstrictor (167). Pharriss and Wyngarden (552) showed that large amounts of PGFinjected into rats caused a shortening of pseudopregnancy and a reduction of the progesterone content of corpora lutea. When PGF was infused into the arterial supply of the transplanted ovary in the sheep (Fig. 5), the results mimicked the luteolytic factor in uterine vein blood in that corpus luteum regression occurred and a new ovarian cycle was initiated (451,459, 466). In addition, when PGF was infused systemically, no effect on corpus luteum function was observed in these animals (451). The negative results of systemic infusions of PGF were explained partly by a dilution effect and partly by the rapid clearance of PG from the blood after one passage through the lungs in some species (199, 555). This latter observation strengthened the view that PGF was the luteolytic hormone released periodically from the uterus and capable of acting locally on the adjacent ovary to cause corpus luteum regression. The manner in which PGF from the uterus might reach the ovary without passing through the systemic circulation clearly had to be investigated.

Fig. 5.

Progesterone secretion rate and peripheral plasma levels of LH following intra-arterial infusion of PGF (2.5 μg · h−1 · 6 h−1) into autotransplanted ovary of sheep during luteal phase. This infusion rate was minimal effective dose required to induce luteolysis when given as a constant infusion.

Previous studies involving ligation of uterine and ovarian blood vessels had indicated the importance of a vascular pathway for the local effect of the uterus on ovarian function in several species (39, 245, 456). In sheep, the ovarian artery is closely attached to the surface of the utero-ovarian vein and transverses the vein in an extremely tortuous manner before entering the hilus of the ovary (Fig. 6). This curious vascular anatomy had been suspected previously of having functional importance at least in the cow (694). The possibility was considered that substances might diffuse from the utero-ovarian vein into the ovarian artery and reach the ovary directly without passing through the systemic circulation. Early evidence that direct local connections might exist between the ovary and the uterus was recorded in the giant fruit bat where unilateral hypertrophy of the uterine horn next to the ovary containing a newly formed corpus luteum was observed (445). In this case, it appeared that greater quantities of ovarian hormones reached the uterus from the ovary than would reach the uterus via the general circulation.

Fig. 6.

Anatomic relationship of ovarian artery and utero-ovarian vein in sheep and experimental design used to demonstrate countercurrent transfer mechanism. [From McCracken et al. (459). Reprinted with permission from Nature; Copyright 1972 Macmillan Magazines Limited.]

The initial demonstration of a countercurrent transfer mechanism in the ovarian vascular pedicle involved the transfer of [3H]PGF from the uterine vein to the ovarian artery in the sheep (455). Later studies indicated that ∼1% of [3H]PGF infused into a uterine vein appeared in the adjacent ovarian artery after a lag of 20–30 min (see Fig. 7) (459). Thus the presence of a luteolytic factor in uterine vein blood at the time of luteolysis had been identified that was mimicked by infusions of PGF into the arterial supply of the ovary. Moreover, a pathway had been found whereby a small portion of PGF from the uterus (∼1%) could bypass the systemic circulation and reach the ovary directly. It remained to be determined whether PGF was secreted by the uterus into the uterine vein during the course of luteolysis in amounts large enough to have a luteolytic effect after local transfer to the ovary. This was achieved when PGF was measured by gas chromatography/mass spectrometry (GC/MS) (267, 452, 455,459) and by rat fundal strip bioassay (69) in samples of uterine vein plasma from a series of individual sheep around the time of corpus luteum regression. From these results, it was calculated that during luteolysis in the sheep, each uterine horn secreted 25 μg PGF/h into the uterine vein. When this quantity of PGF was infused into a uterine vein in situ, premature corpus luteum regression was induced consistently in the adjacent ovary (459, 675), but no effect on the corpus luteum was observed when the same amount of PGF was administered into the peripheral circulation (451). These infusion experiments not only confirmed the countercurrent transfer process in the utero-ovarian vascular pedicle, but also established the role of PGF as a uterine luteolytic hormone in the sheep (459).

Fig. 7.

Time course of countercurrent transfer of [3H]PGF from utero-ovarian vein to adjacent ovarian artery. [From McCracken et al. (459). Reprinted with permission from Nature; Copyright 1972 Macmillan Magazines Limited.]

As shown in Figure 8, GC/MS analysis of ovine uterine vein blood collected frequently over the time of luteolysis indicated that PGF was secreted by the uterus as a series of pulses each lasting ∼1 h and occurring at intervals of ∼6–9 h (45, 456). Similar pulses of PGF were also observed during luteolysis when measured by radioimmunoassay (RIA) (41, 673). Additional evidence for the pulsatile release of PGF from the ovine uterus during luteolysis was obtained by the measurement of the primary metabolite of PGF, 15-keto-13,14-dihydro-PGF (PGFM), in peripheral blood of sheep (373, 549). Pulses of PGFM occurred in the peripheral blood with a similar frequency to PGFmeasured in uterine vein blood. The role of PGF as a uterine luteolytic hormone was supported by the finding that systemic administration of indomethacin, an inhibitor of PG synthesis (527, 561), or the intrauterine administration of indomethacin (416) delayed or prevented luteal regression in several species. Also, immunization against PGF, either passively (193, 194) or actively (589,600), delayed regression of the corpus luteum in the sheep.

Fig. 8.

Concentration of PGF, progesterone, estradiol-17β, and LH in utero-ovarian vein plasma sampled every 2 h during luteolysis in a sheep with a utero-ovarian autotransplant. Preovulatory rise of estradiol concentration is muted because of concomitant 5- to 10-fold increase in uterine blood flow. [From Barcikowski et al. (45); © The Endocrine Society.]

The proposed countercurrent mechanism for the transfer of PGF in the utero-ovarian pedicle of the sheep initially met with some skepticism. One report by Coudert et al. (126) failed to confirm the transfer of [3H]PGF in the uterovascular pedicle of the sheep, although in a companion paper, a transfer of xenon gas was demonstrated (125). However, at this time, it was also shown that unlabeled PGF infused into the uterus of the cow caused a persistent elevation of PGF in ovarian arterial plasma, but not in carotid arterial plasma (313), which supported the existence of a countercurrent mechanism. The countercurrent transfer of [3H]PGF was confirmed subsequently by another group using the sheep (398). It was shown that the percentage of PGF transferred from the uterine vein to the ovarian artery was mass dependent. This explained why Coudert et al. (125) failed to demonstrate a transfer of PGF, since the use of unlabeled PGF as a carrier decreased the transfer of [3H]PGF below the limit of detection of their method (0.7%). In sheep and cattle, experiments involving the anastomosis of the uterine vein or the ovarian artery in the hemihysterectomized animal have provided good support for the local transport of uterine luteolysin to the ovary (245). However, a large quantity of PGF given by the systemic route causes luteolysis because enough PGF escapes metabolism by the lungs to act on the ovary (456). Moreover, in many species including the sheep, cow, and pig, the corpus luteum is resistant to PGF during the early part of the luteal phase (see below under different species).

An anatomic study of the utero-ovarian blood vessels and lymphatics in eight sheep demonstrated that, although no direct lymphatic connections existed between the uterus and the ovary, some uterine lymphatics were observed to lie adjacent to the ovarian artery in a fashion similar to the utero-ovarian vein. It was suggested that the uterine lymphatics might also serve as a possible route of transfer for uterine products to the ovary (113). Later studies in the sheep indicated that both the vena tubarius (15) and uterine lymph (2) contained a relatively high concentration of PGF, indicating that these two structures may also participate in the transfer process. Indeed, it was later shown that [3H]PGF, given as a single injection into the uterine lumen of sheep, reached a peak in a uterine lymphatic in 50 min while peak radioactivity in the adjacent uterine vein was achieved within a few minutes (295). Also, in this study, it was shown that the injection of [3H]PGF into the uterine lumen resulted in a transfer of radioactivity to both the adjacent and, to a lesser extent, the opposite ovary (295). However, because [3H]PGF was not measured in ovarian arterial plasma but rather in ovarian venous plasma after passing through the ovary, the dynamics of PGF transfer into the ovarian artery were not established in this study. Receptors for PGF are abundant in the ovary, and these may have bound a portion of the infused [3H]PGF, thus causing an underestimate of the percent of transfer of [3H]PGF in this study (0.3%). Moreover, the lymphatics adjacent to the cannulated lymph vessel were ligated, which most likely concentrated the flow of lymph from the uterus into the utero-ovarian vascular pedicle (295). Nonetheless, the observed slower transport of PGF from the uterine lumen into the uterine lymphatics compared with its rapid transfer into the uterine vein suggests that lymphatic transport may serve to extend the duration over which a pulse of PGF secreted by the uterus acts at the ovarian level in this species.

Fig. 9.

Concentration of 15-keto-13,14-dihydroprostaglandin Fmetabolite (PGFM) in peripheral plasma of cyclic cow during luteolysis. Pulses of PGFM representing uterine secretion of PGFcoincide with decline of progesterone during luteolysis. [From Kindahl et al. (370), Copyright 1984, with permission from Elsevier Science.]

Other studies in the sheep revealed that progesterone and several other ovarian steroids are transferred from the ovarian vein to the ovarian artery, indicating the presence of an ovarian-ovarian countercurrent system in this species (178, 462, 469). Thus it would appear that the sheep has two types of countercurrent systems in the vasculature of the reproductive system: 1) a utero-ovarian countercurrent system for the transfer of PGF from the uterus to the ovary and 2) an ovarian-ovarian countercurrent system for the transfer of ovarian steroids back to the ovary and the Fallopian tube and probably also to the uterus. A countercurrent mechanism has also been identified in the human ovarian vascular pedicle (57, 58), although the evidence indicates an ovarian-ovarian countercurrent transfer with no evidence of a utero-ovarian transfer. Such an anatomic difference in the ovarian vascular pedicle in the human may explain, at least in part, why the effects of hysterectomy on corpus luteum function differ in the primate from those in nonprimate species.

2.  Cow

Initially, a low-molecular-weight protein was proposed as the luteolytic substance in the cow (429), a view which was later modified to include arachidonic acid (618, 619,285). However, other evidence indicated that, in the cow, PGF is the luteolytic hormone that is released cyclically from the uterus at the time of corpus luteum regression. Prostaglandin F is elevated as a series of pulses in uterine venous blood during luteolysis in this species (509). Also, it has been demonstrated that levels of PGFM in the peripheral blood of cows show pulsatile increases during luteolysis (371, 372, 374), thus supporting the role of PGF as a luteolytic hormone (Fig. 9). Wild ruminants such as the reindeer also show a pulsatile release of PGF during luteolysis (590) (see Fig.10).

Fig. 10.

Concentration of progesterone (dotted line) and PGFM (solid line) in peripheral plasma of reindeer during luteolysis. Horizontal axis shows time in hours. [From Ropstad et al. (590).]

A countercurrent mechanism for the transfer of PGFappears to exist in the utero-ovarian pedicle of the cow (313). Prostaglandin F injected into the uterus caused an elevation of PGF in ovarian arterial plasma for several hours after PGF levels had subsided in the jugular vein and carotid artery. However, a transfer of endogenous PGF in the uterine horn to the ovarian artery in the cow following systemic challenges of oxytocin was not demonstrated (485). The authors point out that they took no control samples of peripheral arterial blood and in addition were unable to show a difference in concentration of PGFbetween venous and arterial plasma, both of which gave readings of ∼200 pg/ml. Further studies in the cow confirmed that a countercurrent transfer of PGF occurs in this species (670). This was achieved by demonstrating a concentration gradient of PGF between an ovarian artery and a peripheral artery. There is also evidence that PGF may act, in part, through the systemic circulation in the cow, since normal cycles were observed in beef cattle after hemihysterectomy regardless of which ovary contained a corpus luteum (698). Moreover, when the ovary was transplanted with vascular anastomoses to the carotid artery and the jugular vein in a cow, and the other ovary was removed, estrous cycles were of normal duration (46). When one ovary was transplanted in the cow, the intact uterus with both uterine horns was left in situ in the abdomen and the other ovary was removed. Thus PGF secreted by the whole intact uterus during luteolysis would produce a peripheral level of PGF approximately twice that of the hemihysterectomized cow. The conclusion that PGF may act partly via the systemic circulation in the cow is supported by the finding that only 65% of [3H]PGF is metabolized in one passage through the lungs in the cow (138), whereas in the sheep, >99% is metabolized (139). Some of the apparent discrepancies concerning the local countercurrent transfer versus the systemic transport of PGF in the cow may be related to the use of different breeds of cattle or to the amount of uterine tissue removed during hemihysterectomy. In the alpaca, a member of the camel family indigenous to South America, there is evidence that the luteolytic effect of the uterus is mediated locally in the right uterine horn but is mediated both systemically and locally in the left uterine horn (198).

3.  Sow

Early work on the effect of hysterectomy on luteal function in the sow was carried out by du Mesnil du Buisson and Dauzier (169), who showed that bilateral regression of the corpora lutea in both ovaries of this polyovulatory species occurred after hemihysterectomy. However, if the amount of tissue in the remaining horn was reduced to ∼25%, the corpora lutea in the opposite ovary no longer regressed (170). This indicated that there may be both a local and a systemic effect of the uterus on the corpora lutea in the sow. Indeed, in the reproductive tract of the sow, there is now anatomic evidence that a crossover of the venous drainage between both uterine horns may occur (148). Subsequently, PGF was identified and measured by RIA in uterine vein blood of the sow at luteolysis as shown in Figure11 (251). A local effect of PGF in the luteolytic process in the sow was shown when luteolysis was induced by infusions of PGF into an adjacent uterine vein (251), although there was some evidence that the opposite ovary was also affected. This may be caused by some of the infused PGF reaching the opposite ovary either systemically (388) or by lymphatic connections with the opposite uterine horn (385). It is likely that PGF can act in part via the systemic circulation, since ∼40% of [3H]PGF infused into the pulmonary artery traverses the lungs unchanged (139). This finding explains the evidence for both a local and systemic luteolytic action of PGF in the sow. Bazer and colleagues (50, 494, 495) have presented additional evidence that PGF is the luteolysin in the sow and have suggested also that the reduction in uterine PGF secretion in early pregnancy is caused by a change in the uterus from an endocrine function to an exocrine function so that PGF is secreted into the uterine lumen (50). It has been known for some time that exogenous estrogen, rather than shortening the cycle as in other species, appears to prolong the life span of the corpora lutea in the pig (235).

Fig. 11.

Concentration of PGF and progesterone in utero-ovarian vein plasma and concentration of progesterone in peripheral plasma during luteolysis in sow. [From Gleeson et al. (251), Copyright 1974, with permission from Elsevier Science.]

Bazer and co-workers (219) have proposed that endogenous estrogen secreted during early pregnancy may be responsible for inhibition of luteolysis by switching the secretion of PGF from the venous side of the uterine circulation to the uterine lumen (endocrine vs. exocrine). In the peripheral blood of sows, PGFM is elevated as a series of pulses at the time of luteolysis, a finding which serves to confirm the role of PGF as a luteolytic hormone in this species (374). An unusual feature of luteolysis in the sow is that corpora lutea are refractory to exogenous PGF for about the first 10–12 days of the 20- to 21-day cycle (156), although PGFmay be active earlier in the cycle if infused into the adjacent utero-ovarian vein (388), if a synthetic analog is used (274), or if repeated injections are given (187). The refractoriness to PGF early in the cycle may be explained by the finding that the number of receptors for PGF in the corpus luteum of the sow increases up to sevenfold as the luteal phase progresses (232). This increase did not occur in pregnant or estrogen-treated pigs (234). Moreover, these investigators suggested that uterine PGF in the cyclic pig might cause the induction of its own receptor in the corpus luteum (187).

4.  Mare

In the mare, there is evidence for a more dominant systemic effect of the uterus on the corpus luteum. Although total hysterectomy results in luteal maintenance (246, 649), hemihysterectomy failed to establish a unilateral relationship (246). Furthermore, in contrast to other species such as the sheep (161) and cow (333), where a much lower dose of PGFis luteolytic when given by the intrauterine route versus the systemic circulation, both these routes of administration were equally effective in the mare (162). This is in keeping with the finding that the ovarian artery in the mare has very little direct contact with the uterine vein compared with the sheep and cow (244). Confirmation of the role of PGF as a luteolytic hormone came from the measurement of PGF in the uterine vein (163) and the elevation of PGFM in the peripheral blood (Fig. 12) at the time of luteolysis (514). The role of PGF in the control of luteolysis in the mare has been reviewed previously (335, 613,650, 745). In view of the lack of a local control of the ovary by the uterus in the mare, it is probable that, like the sow, a proportion of PGF secreted by the uterus may escape metabolism by the lungs and thus may act systemically as a mediator of luteolysis.

Fig. 12.

Concentration of PGFM and progesterone in peripheral plasma in mare during luteolysis. [From Stabenfeldt et al. (650).]

5.  Goat

The reproductive pattern of the goat is rather similar to the sheep in many respects, but there are two important differences between these two species. First, the cycle length in the goat is longer at 19–21 days compared with 16–17 days in the sheep. Second, the goat relies exclusively on progesterone secreted by the corpus luteum for the maintenance of pregnancy, since the goat placenta, unlike the sheep, does not produce progesterone. As in the sheep, hysterectomy in the cyclic goat results in the maintenance of the corpus luteum (130). Prostaglandin F is produced by the uterus in a pulsatile fashion and induces spontaneous luteolysis in this species (317, 370). Pulses of two metabolites of PGF are observed on days 16–18 of the cycle, which coincides with the decline in peripheral plasma progesterone levels (Fig. 13). Parturition in the goat appears to involve a luteolytic mechanism, whereby PGF secreted by the fetoplacental unit at term causes a decrease in progesterone production by the corpus luteum and, hence, contributes to the initiation of labor (131). Evidence for a role of PGF in labor was demonstrated by the infusion of PGF into a uterine vein adjacent to the corpus luteum of the pregnant goat, which caused regression of the corpus luteum and the induction of abortion or labor (129). Similarly, surgical removal of the corpus luteum before term in the goat results in abortion, an effect which could be prevented by the administration of progesterone (131). In a recent study, no increase in PGF was detected before the preparturient decline in progesterone in goats (214). However, the authors point out that blood samples were collected only once daily before labor so that potential episodes of pulsatile PGF secretion during preparturient luteolysis would not have been detected.

Fig. 13.

Concentration of 2 metabolites of PGF and progesterone in peripheral plasma during estrous cycle and after mating in goat. Horizontal solid bar represents time of estrus, and 2 arrows indicate time of mating. [From Kindahl et al. (370).]

6.  Guinea pig

After the sheep, which, because of its size, offers substantial advantages as an experimental animal model for luteolysis, the strongest evidence for PGF as a luteolytic hormone is found in the guinea pig (321, 561). The major pieces of evidence are 1) a luteolytic effect of administered PGF, 2) an elevation in PGFlevels in uterine venous blood at the time of spontaneous luteolysis,3) prolonged cycles in animals treated with indomethacin or immunized against PGF, and 4) increased uterine PGF secretion after treatment with exogenous estrogen. The correct vascular anatomy for a countercurrent system appears to be present in the guinea pig (147), which is in keeping with the known unilateral effect of the uterus on the life span of corpora lutea in this species. In addition, xenon-133 gas dissolved in saline is transferred locally from one uterine horn to the adjacent ovary in the guinea pig (175). Of several species examined, the clearance rate of labeled PGF from the systemic circulation was most rapid in the guinea pig (262). Such a finding indicates that, as in the sheep, there is an absolute requirement for a local transfer of PGF between the uterus and the ovary for luteolysis in the guinea pig. As in several other species, PGF(measured as PGFM in peripheral blood) is secreted in a pulsatile fashion from the uterus of the guinea pig during luteolysis (Fig.14) (183). Uterine PGF production is suppressed during the establishment of pregnancy so that corpora lutea continue to secrete progesterone for the maintenance of pregnancy (562). The corpora lutea are necessary for the maintenance of pregnancy up until about day 28 when the placenta assumes this function (297).

Fig. 14.

Concentration of PGFM and progesterone in peripheral plasma of guinea pig during luteolysis. Hourly blood samples were collected daily between hours of 9:00 a.m. and 4:00 p.m.[Adapted from Elger et al. (183).]

7.  Other laboratory animals

Hysterectomy is known to prolong the life span of the corpus luteum in the pseudopregnant rat (80), rabbit (25), and hamster (95), but not in the mouse (154). In addition, PGF now has been shown to be luteolytic when administered in vivo to the pseudopregnant rat (552), rabbit (275), hamster (390), and Mongolian gerbil (160). Prostaglandin F will also terminate pregnancy in the rat, rabbit (275), hamster (277, 391), and mouse (47). This last effect seems to be primarily a luteolytic one since exogenous progesterone prevents abortion in most of these species, although higher doses of PGF were required to terminate pregnancy than to shorten pseudopregnancy in the mouse (47) and rat (111). The requirement for a luteolytic mechanism for the induction of labor in mice was demonstrated by the finding that mice rendered deficient in the gene for the PGF receptor do not show the normal prepartum drop in progesterone and do not exhibit parturition [see Sugimoto et al. (660) and sect. v A]. Because of the small size of these experimental animals, there have been few direct measurements of levels of PGF in blood. However, PGF levels were found to be elevated in the peripheral blood of pseudopregnant hamsters at the time of luteolysis (610).

In the rabbit, in which luteal function is estrogen dependent, plasma progesterone levels begin to decline on day 14 of pseudopregnancy in both intact and hysterectomized animals, indicating that the initial decline of plasma progesterone does not depend on the uterus. On day 17, a marked drop in progesterone levels occurs only in intact pseudopregnant animals and is accompanied by an increase in uterine venous PGF levels (432). In the rabbit, like the mare, there is evidence that the luteolytic effect of the uterus may be mediated systemically, since no unilateral effect of the uterus on the corpus luteum can be demonstrated (328). The importance of the systemic route is supported by the lack of an anatomic basis for a countercurrent transfer system between the uterus and the ovary (147) and by the absence of a local transfer of xenon-133 from the uterus to the ovary in the rabbit (175). The possibility that a metabolite of PGF is the systemic component causing luteolysis is suggested by the finding that the metabolite 13,14-dihydro-PGF was fourfold more luteolytic than PGF when infused systemically in the pseudopregnant rabbit (354). Like the pig, there is evidence in the rabbit that estrogen, rather than hastening the onset of luteolysis, actually has a luteotropic action in the corpus luteum, possibly by protecting it against PGF (276), or by changing the secretion of PGF by the endometrium from an endocrine to an exocrine mode as proposed in the sow (494,495).

8.  Primate

As in most other species, basal levels of LH in the human appear to be essential to maintain the secretory function of the corpus luteum (690). In the rhesus monkey, bilateral lesions in the arcuate nucleus of the hypothalamus caused a cessation of ovarian ovulatory activity that could be restored by chronic circhoral infusions of GnRH (381). If plasma levels of LH were reduced to undetectable levels during the midluteal phase by halting GnRH infusions in these lesioned monkeys, plasma progesterone fell to undetectable levels. However, when LH levels were restored 3 days later by resuming circhoral GnRH infusions, the corpus luteum resumed a normal pattern of progesterone secretion but regressed at the expected time (330). These studies suggest that LH acts to promote progesterone synthesis by the corpus luteum but that other factors are responsible for the loss of function and structural integrity of the primate corpus luteum during luteolysis.

Because the primate corpus luteum undergoes luteolysis in the absence of the uterus, it has long been considered that luteolysis in primates might be an intraovarian event. Early studies suggested that estrogen produced by the primate corpus luteum mediated luteolysis (380). Subsequent work indicated that estrogen may act by increasing PGF levels in the ovary. This view was based on the finding that exogenous estrogen increased the concentration of PGF in ovarian venous blood (29) and that indomethacin blocked estrogen-induced luteolysis in the rhesus monkey (30). However, it was found that high levels of estrogen (10 μg/ml) inhibited progesterone synthesis by human luteal cells, both in the presence and absence of indomethacin (672). Later studies suggested that the luteolytic effect of exogenous estrogen in the primate may be due to its suppression of pituitary gonadotropin secretion rather than a direct effect on the ovary (605, 716). Moreover, estrogen receptors are absent in all cell types of the primate corpus luteum (306, 657). The finding that administration of either an aromatase inhibitor (186, 748) or an estrogen antagonist (10) does not prolong the life span of the corpus luteum in monkeys indicates that estrogen may not be a direct mediator of luteolysis in primates. However, it is possible that estrogen may have indirect actions in the ovary or the corpus luteum other than via estrogen receptors. For example, co-oxidation of steroidal estrogens by purified PG synthase results in stimulation of PG formation (145). It has been suggested that progesterone promotes its own synthesis and maintains the structural integrity of the corpus luteum (591). Although until recently there has been limited data to support this hypothesis, the finding that both steroidogenic and nonsteroidogenic cell types in the corpus luteum of the primates possess receptors for progesterone suggests that progesterone may have direct autoregulatory actions in the corpus luteum itself (306, 657).

Although there is evidence that PGF or its analogs are luteolytic in monkeys (376, 464, 593, 611, 662,719), the evidence for a luteolytic effect of these compounds in the human is less marked. Several attempts have been made to demonstrate a luteolytic effect of PGF infused intravenously during the luteal phase of the cycle in women. However, the results have been equivocal. Some studies reported a fall in plasma progesterone concentration or a shortening of the cycle (308,410), whereas others reported no effect (346, 412). These results may be due in part to the rapid metabolism of the administered PGF and thus the low amounts of PGF reaching the ovary. Also, a refractoriness to PGF has been reported in the early part of the luteal phase in the pig (494) and in the cow (569), and later studies in the pig indicated that this may be due to a low density of PGF receptors in the early to mid luteal phase (232). It is possible that such refractoriness to PGF may also exist for the first half of the luteal phase in primates. The intravenous infusion of two analogs of PGF (17-phenyl-PGF and 15-methyl-PGF), which are more stable in the circulation, were found to depress plasma progesterone levels in women during the early luteal phase. However, progesterone levels recovered rapidly after stopping two 8-h infusions and the cycle length was generally normal (403). The dose of the analogs employed was relatively low (10 μg · kg−1 · h−1 for 8 h) compared with doses of the other analogs given to monkeys (up to 200 μg · kg−1 · h−1 for 6 h). Moreover, the human studies were conducted early in the luteal phase, a time when the corpus luteum is refractory to PGF in several species. However, inhibition of progesterone synthesis by PGF has been demonstrated in vitro using cultured human granulosa cells (304).

In the cynomolgus monkey, the infusion of PGF into the hilus of the ovary containing the corpus luteum caused a temporary fall in progesterone levels (464). Prostaglandin F infused over several days in very low doses directly into the corpus luteum of the monkey caused premature luteolysis (33, 36, 300). In women, 0.5–1.0 mg PGFinduced luteal regression when given as a single injection directly into the corpus luteum but not when injected into the adjacent stroma (383). More recently, higher doses of PGFinjected directly into the human corpus luteum in vivo caused premature regression of the corpus luteum and shortening of the cycle length (60). There is, however, some disagreement as to whether or not endogenous levels of PGF rise in the primate corpus luteum at the time of luteolysis (107, 322, 546, 625,664). It is possible that stromal tissue of the ovary might be a significant source of PG in the primate especially since there are reports on the synthesis of PG by ovarian tissues of several species including the rat (110), rabbit (108), pig (544), and monkey (720). More direct evidence for a role of PGF in human corpus luteum regression was reported by Aksel et al. (9) who found that PGF levels in ovarian venous blood were elevated in the late luteal phase.

Although treatment with indomethacin, an inhibitor of PG synthesis, delays the onset of luteolysis in a number of species (320, 402,527), such an effect was not observed in the rhesus monkey (442). In the monkey study, the drug was administered either once per day orally (2 monkeys) or injected twice daily in sesame oil subcutaneously (1 monkey) at a dose of 30 mg · kg−1 · day−1. The length of the luteal phase was normal as determined by plasma progesterone levels. Approximately the same dose of indomethacin, given to monkeys late in pregnancy, caused a delay in parturition but also caused extreme flaccidity of the uterus, a 50% fetal mortality rate, and severe oligo hydramnios. The authors concluded that, although PG may play a role in normal parturition, they do not appear to be involved in the process of luteolysis in primates. However, the twice daily administration of indomethacin to monkeys may not have been sufficient for total suppression of ovarian PG synthesis, especially since PG levels were not actually measured in this tissue. In other studies, it was necessary to administer indomethacin every 4 h to inhibit PG-mediated effects (632). Although the regimen of indomethacin administered to monkeys was sufficient to interfere with labor, indomethacin could have inhibited phosphodiesterase and elevated myometrial cAMP levels, thus explaining the flaccidity of the uterus and the delay in labor. The results of other studies also argue against a luteolytic role for PGF synthesized locally in the ovary of primates. The direct infusion of an inhibitor of PG synthesis, sodium meclofenamate, into the midcycle corpus luteum of the rhesus monkey, rather than preventing luteolysis, caused premature luteal regression and shortening of the cycle (598). The authors concluded that this was due to the inhibition of luteotropic PG such as PGE2 and PGI2 which have been reported to increase cAMP and/or progesterone production in vitro by luteal tissue from monkeys (497) and women (284). However, subsequent studies showed that sodium meclofenamate also inhibits gonadotropic support of luteal cells in vitro (749). The intraluteal infusion of PGE2 inhibited PGF-induced luteal regression in monkeys, but a similar infusion of PGE2 during the time of corpus luteum regression did not extend its life span (750). However, it is possible that the level of PGE2 was not high enough to counteract putative endogenous luteolysins including PGF.

It has also been suggested that PGF levels in the primate ovary and/or corpus luteum may not necessarily change during the luteal phase but rather that the second messenger systems mediating PGF action undergo a maturational process toward the time of luteolysis (481, 708). It is proposed that, when the appropriate second messenger systems mature, basal levels of PGF may cause an elevation in intracellular calcium that activates protein kinase C and cyclic nucleotide phosphodiesterase, thus inhibiting the cAMP-mediated luteotropic action of LH. The addition of the PGF analog cloprostenol to incubations of luteal tissue from the marmoset monkey inhibited both the stimulation of cAMP by hCG and the stimulation of progesterone production by dibutyryl cAMP, suggesting that PGF may act at sites before and after the generation of cAMP (708). There is also evidence that arachidonic acid could play a role in luteolysis in primates. Ottobre et al. (534) found that arachidonic acid inhibited progesterone production by luteal cells obtained from rhesus monkeys in the presence of either indomethacin or nordihydroguaiaretic acid. The authors proposed that arachidonic acid may inhibit gonadotropin-stimulated progesterone production without conversion to either a cyclooxygenase or a lipoxygenase product, most likely by direct activation of protein kinase C. In conclusion, although there is considerable evidence that PG of ovarian and/or luteal origin may participate in the induction of corpus luteum regression in primates, further studies will be required to resolve this question (see sect. iv B).

III.  HORMONAL REGULATION OF UTERINE PROSTAGLANDIN F SYNTHESIS

The release of arachidonic acid from phospholipids by phospholipase A2(PLA2) activation is considered to be the rate-limiting step in PG synthesis (389, 399, 454). Released arachidonic acid is then rapidly converted to PGF by PG synthetase, also called cyclooxygenase (COX), which exists in both a constitutive (COX-1) and an inducible form (COX-2) (692). It was established early on that the site of PGF synthesis in the uterus is the endometrium, since high concentrations of PGF were found in sheep (721, 722) and guinea pig (560) endometrium. Estradiol was found to stimulate the synthesis of PGF by the uterus in the guinea pig (71), rat (102), and sheep (97). It was suggested that estrogen increased uterine PGF production by stimulating the activity of enzymes controlling PG synthesis in the endometrium (281, 356). However, it was found that a prior period of exposure to progesterone enhanced estrogen-stimulated production of endometrial PGF in several species (45, 97, 102, 217, 416,597). Estradiol has been shown to increase the activity of PLA2, an enzyme which liberates arachidonic acid from phospholipid stores (74, 155) and increases the activity of PG synthase (356, 733). The priming effect of progesterone on endometrial PGF synthesis is considered to be due to the accumulation of lipids in the endometrium, since progesterone increases lipid accumulation in rats (76,77). Progesterone may also enhance PGF synthesis by increasing the concentration and activity of endometrial PG synthase (173, 571).

Fig. 15.

Peak concentration of PGF in utero-ovarian venous plasma after a 10-min period of mechanical stimulation of ovine uterus during cycle and in early pregnancy. [From McCracken (453).]

Subsequently, it became apparent in the sheep that estrogen and progesterone primarily controlled endometrial PGFsynthesis by regulating the concentration of endometrial oxytocin receptors. Such a finding was based on the observation that mechanical stimulation of the uterus evoked the secretion of PGFonly early and late in the cycle (Fig. 15), whereas no effect was observed during the midluteal phase (453, 723). Because mechanical stimulation of the reproductive tract causes an elevation of neurohypophysial oxytocin in peripheral blood of sheep (583) and goats (70) via the centrally acting Ferguson reflex (197), it was considered that oxytocin might be responsible for the observed effect of mechanical stimulation on PGF secretion. This seemed plausible, since exogenous oxytocin had previously been shown to cause premature luteolysis in the cow (23), an effect which had been postulated to be due to oxytocin-stimulated PGFsecretion from the uterus (452). As shown in Figure16, the infusion of oxytocin in a physiological range (200 pg/min) into the arterial supply of the ovine uterus mimicked the effects of mechanical stimulation of the uterus in that oxytocin evoked a modest increase in PGF secretion early in the luteal phase (day 3), had no effect onday 8, but caused a marked increase late in the luteal phase on day 14, around the time of luteolysis (580, 581,723). Therefore, it seemed likely that the cyclical variation in the ability of oxytocin to evoke uterine PGF secretion was due to a cyclical variation in the concentration of endometrial oxytocin receptors. This view was supported by the finding that previously established target tissues for oxytocin, such as the mammary gland and the oviduct, contained binding sites for oxytocin (643) and that estrogen enhanced oxytocin binding in the rat uterus (642). Moreover, it was reported that oxytocin-induced uterine PGF secretion in anestrous sheep was enhanced by exogenous estrogen (612). Subsequently, it was demonstrated that oxytocin-stimulated PGF production by ovine endometrium in vitro was positively correlated with the relative abundance of oxytocin receptors in this tissue (582). The concentration of endometrial and myometrial oxytocin receptors during the ovine estrous cycle is shown in Figure 17.

Fig. 16.

Peak concentration of PGF in utero-ovarian venous plasma after a 10-min infusion of oxytocin (200 pg/min) into a uterine artery on different days of cycle in sheep. [From McCracken (453).]

Fig. 17.

Concentration of high-affinity binding sites for oxytocin in endometrium and myometrium during estrous cycle of sheep. [From McCracken (453).]

Later studies showed that oxytocin infused into the uterine artery of sheep evoked a small increase in uterine PGF secretion following 6 h of systemic administration of a physiological level of estrogen (0.5 μg/h), whereas systemic administration of progesterone alone (500 μg/h) did not cause any increase in oxytocin-stimulated uterine PGF secretion. During a 10-day continuous infusion of progesterone, the enhancing effect of estrogen on oxytocin-stimulated uterine PGFsecretion was blocked on days 2 and 6 (Fig.18). However, after 10 days of progesterone treatment, the enhancing effect of estrogen was restored, but the response to intra-arterial oxytocin was 100-fold greater than the response to oxytocin before progesterone treatment (453). A model for the hormonal regulation of PGF synthesis in the endometrial cell is shown in Figure 19. It is proposed that, in the intact cycling sheep, estrogen enhances the formation of endometrial receptors for oxytocin, but, during the luteal phase, progesterone reduces the concentration of endometrial oxytocin receptors by blocking the action of estrogen. Progesterone also has a direct nongenomic inhibiting effect on uterine oxytocin receptor binding in the rat which is thought to be mediated by guanine nucleotide-induced changes in conformation of the oxytocin receptor (266). In view of this finding, progesterone may block oxytocin action, not only by inhibiting estrogen-induced upregulation of the oxytocin receptor, but also by a direct nongenomic inhibitory action on the oxytocin receptor itself. Thus the uterus becomes refractory to oxytocin during the luteal phase in terms of PGF secretion. However, progesterone eventually downregulates its own receptor (483,695) and decreases mRNA for its receptor (116) so that toward the end of the luteal phase, estrogen action is no longer suppressed because of the loss of progesterone action, and estrogen now induces the formation of endometrial oxytocin receptors. The decline in progesterone as luteolysis progresses may also stablize the oxytocin receptors by removing the nongenomic inhibitory action of progesterone on the oxytocin receptors (266). The large amplification of endometrial PGF secretion induced by oxytocin at the end of the luteal phase most likely results from the priming effect of progesterone on lipid precursors in the endometrium during the luteal phase (77, 453). Other studies support the view that progesterone eventually downregulates its own receptor. For example, the administration of exogenous progesterone, soon after ovulation, shortens the length of the cycle in sheep (535, 734) and cattle (236). When progesterone is administered to ovariectomized sheep for 12 days, oxytocin receptors were initially inhibited but were upregulated after 10 days, consistent with a loss of progesterone action due to downregulation of its receptor (689). When progesterone action was blocked using the progesterone receptor antagonist RU-486 during the early to midluteal phase in sheep, luteolysis failed to occur and the cycle was extended to at least 24 days (504). The authors concluded that progesterone action is required to prime the endometrium for subsequent PGF synthesis and that the timely downregulation of progesterone receptors by progesterone is an important component in timing the onset of luteolysis.

Fig. 18.

Effect of a 12-h infusion of estradiol-17β (E2-17β) ondays 2, 6, and 10, superimposed on a continuous 10-day infusion of progesterone, on oxytocin-stimulated PGF secretion rate from autotransplanted uterus in ovariectomized sheep. [From McCracken (453).]

Fig. 19.

A model for endocrine control of PGF synthesis in endometrial cell of sheep during luteolysis. [From McCracken et al. (461).]

Direct evidence was found later for an increase in oxytocin receptors in the ovine endometrium and myometrium following the loss of progesterone action in the presence of estradiol (405). Loss of progesterone action was simulated by stopping a 5-day infusion of progesterone (500 μg/h) in three ovariectomized sheep continuously infused with estradiol (0.5 μg/h). Uterine oxytocin and estradiol receptors were markedly upregulated in parallel between 6 and 12 h after progesterone withdrawal (Fig.20). Recent studies in sheep confirm the critical role of estrogen and oxytocin receptors during luteolysis. In this species, only luminal epithelial cells of the endometrium express receptors for estrogen and oxytocin at the time of luteolysis, whereas progesterone receptors are not detectable in luminal or glandular epithelia during this period (645, 701, 702). In addition to the upregulation of oxytocin receptors toward the end of the luteal phase, there is also evidence that the inducible form of the cyclooxygenase enzyme (COX-2) increases in the luminal epithelial cells of the ovine uterus (109). Estradiol slightly increased COX-2 expression, but only after progesterone priming. Moreover, oxytocin administration upregulates COX-2 expression in ovine endometrium (27, 90), which may amplify the production of oxytocin-stimulated PGF production during luteolysis in the sheep.

Fig. 20.

Time course of response of uterine receptors for oxytocin, estradiol, and progesterone in 3 ovariectomized sheep after withdrawal of a 5-day infusion of progesterone (500 μg/h) in presence of a continuous infusion of estradiol (0.5 μg/h). Endometrial and myometrial tissues were biopsied at −1 h and at +3, +6, and +12 h after stopping progesterone infusion. A parallel time-dependent increase in estradiol and oxytocin receptors occurred in both endometrial and myometrial tissue. nRe, nuclear estradiol receptor; cRe, cytosolic estradiol receptor; ROT, oxytocin receptor; cRp, cytosolic progesterone receptor. [Adapted from Leavitt et al. (405).]

The action of oxytocin on endometrial PGF synthesis in the sheep was suggested to occur via activation of the phospholipase C pathway and that the hydrolysis of phosphoinositides to diacyglycerol and inositol phosphate promoted the formation of arachidonic acid for PGF production (206). However, on a quantitative basis, the amount of arachidonic acid released by this mechanism likely would be inadequate to account for the very large production of PGF during luteolysis. Later studies indicated that the activation of PLA2 by oxytocin plays a dominant role in oxytocin-stimulated PGF synthesis in the ovine endometrium (406). In the rat, the increased expression of oxytocin binding sites observed at proestrus and at the onset of labor appears to be mediated, at least in part, by an estrogen-induced upregulation of oxytocin receptor gene expression (400). However, in this study, the downregulation of oxytocin receptor binding by progesterone could not be explained wholly by an effect on oxytocin mRNA, and it was suggested that progesterone could act at translational or posttranslational levels. However, recent findings in the rat indicate that, during pregnancy, progesterone has a direct nongenomic inhibitory action on uterine oxytocin receptors which blocks oxytocin action (266).

Although endometrial PG in primates are not required for luteolysis, their role in endometrial physiology has been of considerable interest since they were first identified in human menstrual fluid (174). Comparative information on PG synthesis in the primate endometrium is important because it may represent a remnant of evolution from an earlier dependence on uterine PG for luteal regression. Moreover, there may be similarities in the regulation of endometrial PG synthesis between the primate and the guinea pig. The participation of PG in the process of menstruation was underlined by the observation that PGF, administered to women during the luteal phase of the cycle, caused menstrual-like bleeding (732). Further studies confirmed that human endometrium (4, 164, 430, 631) and rhesus monkey endometrium (150, 181, 182) are an abundant source of PG, primarily PGF and, to a lesser extent, PGE2 and PGI2. In addition to their role in promoting vascular changes associated with menstruation, their overproduction appears to contribute to the pathogenesis of dysmenorrhea (430, 554). The control of PG synthesis in the primate endometrium appears to be regulated primarily by estradiol and progesterone. Progesterone has a priming effect on the endometrium, probably by increasing the incorporation of arachidonic acid into cellular phospholipids (165, 336). At the same time, progesterone appears to inhibit the synthesis of PG as evidenced by the marked inhibition of PG production by the addition of progesterone to cultures of human endometrium (3, 99, 604) and monkey endometrium (181, 182). Estradiol, in the presence of serum, stimulates PG production by cultured human endometrium (3,604), most likely by stimulating PG synthase and phospholipase activity. Because progesterone added to cultures of luteal phase human and rhesus monkey endometrium markedly inhibited PG production, both in the absence and presence of estradiol, it appears that progesterone withdrawal alone acts as a sufficient stimulus to cause a large increase in endometrial PG production in primates (181,182). Thus both progesterone priming and progesterone withdrawal appear to play important roles in controlling PG production in the primate endometrium. In nonprimates, luteolysis depends on the prior production of PG (PGF) by the endometrium to initiate luteolysis, whereas in primates, it appears that intraovarian luteolysis (progesterone withdrawal) is the primary stimulus for the subsequent endometrial PG production associated with menstruation.

The progesterone receptor antagonist RU-486 blocks progesterone action at its target sites in the uterus and brain and is, thus, a useful pharmacological agent to induce the loss of progesterone action experimentally. The administration of progesterone receptor antagonists to women during the luteal phase induces premature luteolysis, stimulates endometrial PG synthesis, and causes the early onset of menstruation (48, 693). It is unclear whether the luteolytic action of RU-486 is mediated directly on the corpus luteum or indirectly via the pituitary gland (693). It should be noted that RU-486 exhibits both agonistic and antagonistic activity that is observed in the central nervous system (120, 530) as well as in the primate endometrium depending on whether or not this tissue has been exposed previously to progesterone (513).

IV.  REGULATION OF PULSATILE UTERINE PROSTAGLANDIN F SECRETION

A.  Role of the Central Oxytocin Pulse Generator

Although it was established that estrogen and progesterone indirectly control endometrial PGF secretion by regulating the formation of oxytocin receptors in several nonprimate species, the regulation of systemic oxytocin during the estrous cycle was largely unknown. Oxytocin is a nonapeptide hormone synthesized as part of a high-molecular-weight precursor in the hypothalamic magnocellular neurons where it is packaged into secretory granules. The latter reach the neurohypophysis via axonal transport during which the precursor is cleaved into oxytocin and its oxytocin neurophysin before their release into the bloodstream (417). Early evidence indicated that, in addition to regulating uterine PGFsynthesis, estrogen and progesterone might also control the secretion of oxytocin from the neurohypophysis during luteolysis (453). Such a notion was supported by the finding that peaks of oxytocin neurophysin occurred synchronously with peaks of PGFM in the peripheral plasma of sheep during luteolysis (190). Subsequently, it was reported that peaks of oxytocin also occurred synchronously with peaks of PGFM in the peripheral blood of sheep during luteolysis (209). Because RIA methods for measuring plasma oxytocin levels were not readily available initially, circulating levels of oxytocic activity in the blood of conscious ovariectomized and intact sheep were measured by a biometric method (471,607). These biometric measurements were later supported by RIA measurements of plasma oxytocin (470). Small pulses of intramyometrial pressure (IMP) were observed to occur synchronously in both uterine horns of ovariectomized sheep with a mean duration of 5.9 min and a pulse interval of 14.3 min (607). The basal frequency (20 min) of these pulses was maintained in ovariectomized sheep by the administration of low levels of estrogen (0.05 μg/h). Confirmation that pulses of IMP in the ovariectomized sheep were due to small intermittent pulses of oxytocin from the neurohypophysis was obtained by demonstrating that the infusion of 0.01 mU of oxytocin given over 1 min into one uterine artery produced an ectopic burst of IMP only in the infused horn while an injection of 10.0 mU into a jugular vein caused an ectopic burst of IMP in both uterine horns. Also, the infusion of a potent antagonist of oxytocin action (44) into one uterine artery for 30 min blocked the 20-min pulses of IMP only in the infused horn. Lastly, as shown in Figure21, peaks of oxytocin were observed to occur in jugular plasma (10 pg/ml) which were synchronous with the 20-min pulses of IMP (470). Others have also reported a pulsatile pattern of oxytocin secretion in peripheral plasma of sheep during the estrous cycle (241, 491). During labor in women, the frequency of small pulses of oxytocin in peripheral blood increases from a resting level of 2.6 to 10.4 pulses/h (230). The amplitude of these pulses of oxytocin increased only slightly from a resting level of 2.0 pg/ml plamsa to 3.0 pg/ml plasma during second- and third-stage labor (230).

Fig. 21.

Concentration of oxytocin in jugular venous plasma sampled at 1-min intervals during endogenous episodes of intramyometrial pressure that occur every 20 min in intact sheep and also in ovariectomized sheep provided that the latter are maintained on low estradiol (0.05 μg/h). IMP, intramyometrial pressure. [From McCracken et al. (461).]

In cyclic sheep, the amplitude of small pulses of IMP was high around the time of estrus, but their amplitude declined during the luteal phase and increased again around day 13 of the cycle. However, at this time, large hour-long episodes of IMP began to occur at intervals of ∼8 h (471). Measurement of oxytocin in jugular plasma indicated that plasma levels of oxytocin increased markedly during these hour-long bursts of IMP, whereas plasma levels of vasopressin remained unchanged (Fig.22). The concentration of oxytocin during the first large burst of IMP reached ∼200 pg/ml of plasma, but peak concentration of oxytocin declined by ∼50% during each subsequent burst of IMP (460, 461). Other studies have also reported large pulses of oxytocin or its neurophysin during luteolysis in sheep (190, 209), goats (122), and cows (697). These results suggested that the large hour-long episodic releases of oxytocin in ruminants, interacting with rising levels of endometrial oxytocin receptors, evoked the large episodic pulses of uterine PGF that cause luteolysis in these species. However, because of the unexpected discovery that the ovine corpus luteum contained large amounts of oxytocin (703) and that luteal oxytocin was discharged by the systemic administration of an analog of PGF(208), it was unclear what proportion of the elevated levels of oxytocin, observed during luteolysis in sheep, emanated from the posterior pituitary versus the corpus luteum.

Fig. 22.

Concentration of oxytocin (●) and vasopressin (○) in jugular venous plasma during successive large bursts of intramyometrial pressure observed during luteolysis in sheep. Peak levels of oxytocin declined as luteolysis progressed. [From McCracken et al. (461).]

B.  Oxytocin in the Corpus Luteum

The reports of relatively large quantities of oxytocin in the corpus luteum of the sheep (703) and cow (202) confirmed and extended earlier findings which had indicated the presence of oxytocic activity in extracts of the corpus luteum as determined by the milk-secreting activity of these extracts in the goat and cat (434, 533). Exogenous PGF and PGE2 were observed to cause a gradual but modest increase in peripheral plasma levels of oxytocin in women receiving these compounds for the induction of labor. Because a similar increase was seen in men, it was suggested that PG caused a release of oxytocin from the posterior pituitary (243). In pigs, peripheral levels of oxytocin were also reported to be elevated following an intramuscular injection of PGF(184), but not in anesthetized lactating rats, where inhibition was observed (567). In the cow, the administration of several PGF analogs caused a large elevation of oxytocin in jugular venous blood that reached a peak within 15–20 min (602). However, in the sheep, the intramuscular injection of cloprostenol, an analog of PGF, caused a rapid increase in the secretion of oxytocin from the ovary containing the corpus luteum, but not from the opposite ovary or the brain (208). This latter finding suggested that the elevation of oxytocin observed in the cow following PG treatment also came from the corpus luteum. At that time, it was unclear whether oxytocin was simply taken up and stored in the corpus luteum or whether it was actively synthesized in this tissue. Later studies in both the cow (337, 341) and sheep (339,340, 345) showed that the gene for oxytocin is fully expressed in the developing corpus luteum shortly after ovulation, thus confirming de novo synthesis of oxytocin in the corpus luteum of ruminants. Messenger RNA for oxytocin in the corpus luteum of the sheep is abundant for the first 6 days after ovulation during which time it is translated into a finite pool of oxytocin (339, 345).

It was proposed that, during luteolysis in sheep, all of the oxytocin in the corpus luteum was discharged during each pulse of uterine PGF secretion and that the interval between pulses of PGF was determined by the time required for synthesis and replenishment of luteal oxytocin (212). This seems unlikely, however, since levels of mRNA for oxytocin in the corpus luteum (Fig. 23) are very low at the time of luteolysis in the sheep (339). The site of oxytocin synthesis in the corpus luteum was shown to be the large steroidogenic cells in the cow (196, 200, 202) and sheep (587, 671). Moreover, oxytocin and its neurophysin have been identified in the small electron-dense granules found in large steroidogenic cells in ruminants (200, 576, 671). These oxytocin-containing granules are released by exocytosis, indicating that this process is involved in the regulation of oxytocin secretion from the corpus luteum (Fig. 24). In addition to domestic ruminants such as the sheep and cow, there is evidence that oxytocin is also produced by the corpus luteum in some species of deer (211). Studies in nonruminant species, such as the pig, rat, and rabbit, indicate that the concentration of oxytocin in luteal tissue is in nanograms per gram rather than micrograms per gram as found in ruminants (699, 700).

Fig. 23.

Relative concentration of oxytocin (OT) mRNA in luteal tissue obtained from sheep throughout estrous cycle. Levels of oxytocin mRNA declined precipitously around day 6 of cycle. [From Ivell et al. (339).]

Fig. 24.

Electron micrograph showing 2 oxytocin-containing granules undergoing exocytosis after an infusion of a subluteolytic level of PGF (100 pg/min) into ovarian artery of anesthetized sheep on cycle day 12. (J. Aghajanian, E. E. Custer, and J. A. McCracken, unpublished data).

With the use of RIA, small quantities of oxytocin have been detected in human luteal tissue (143, 361, 601, 704) and in nonhuman primates including the cynomolgus monkey (360), the baboon (363), the rhesus monkey (467), and the marmoset monkey (179). In addition, oxytocin extracted from the ovaries of the cynomolgus monkey (19) and from baboon and human corpora lutea and stromal tissue (365) was shown to be biologically active using a rat milk ejection assay or a rat uterine strip bioassay, respectively. The detection of oxytocin mRNA using cDNA probes, a sensitive PCR or RT-PCR suggests that oxytocin is indeed synthesized in small quantities in the corpus luteum of the human (573), the baboon (338), and the marmoset monkey (179). However, some studies were unable to detect oxytocin by RIA in the corpus luteum of the marmoset (705) or the human (32). It is unclear whether luteal oxytocin is actually secreted by the ovary in primates as has been shown to occur in ruminants (206, 208, 294, 387, 396,397, 706). Some studies in primates report increased levels of oxytocin in ovarian venous plasma during the luteal phase in the human (362) and in the marmoset monkey (179), whereas others failed to find elevated levels of oxytocin in the ovarian vein of the human (31) or rhesus monkey (467). Moreover, peripheral levels of oxytocin in women which reach a peak at midcycle have been shown to be derived solely from the neurohypophysis and not from the ovaries (624). Nonetheless, low levels of oxytocin produced within the primate ovary could act in a paracrine or autocrine manner to control luteal function in the primate, especially since receptors for oxytocin have been found in primate luteal tissues (179, 228, 364). Although there are conflicting reports of the effect of exogenous oxytocin on progesterone synthesis by primate luteal tissue in vitro (59,575, 669), at least two reports indicate that oxytocin infused in vivo into the corpus luteum of the rhesus monkey (33,36) or human (61) causes a shortening of the luteal phase. In the human study (61), oxytocin, infused into the nonluteal ovary, had no effect on cycle length. Moreover, the infusion of a PG synthesis inhibitor simultaneously with oxytocin into the corpus luteum prevented regression of the corpus luteum. Thus the general components present in the utero-ovarian system of a number of nonprimate species for regulating the life span of the corpus luteum (namely, oxytocin and its endometrial receptor and endometrial PGF and its luteal receptor) are also present within the ovary of the primate. The presence of these components may be relevant to the current view of an intraovarian mechanism for luteolysis in the primate which is independent of the uterus.

C.  Relative Contribution of Neurohypophysial and Luteal Oxytocin

Although earlier studies had established that the secretion of PGF from the uterus in several species is controlled by oxytocin acting on its endometrial receptor (see sect.iii), the source of oxytocin initially was considered to be the neurohypothesis. However, because of the discovery of substantial amounts of oxytocin in the corpus luteum of the sheep and other ruminants (see sect. vi B), the possibility arose that some or all of the increases in oxytocin observed in the peripheral circulation of ruminants during luteolysis might originate from the corpus luteum. Therefore, to determine the relative contributions of the neurohypophysis and the corpus luteum to the large pulses of oxytocin observed during luteolysis in sheep, two model systems were developed to exclude the corpus luteum as a source of oxytocin. In the first model, ovariectomized sheep, given low levels of estrogen (0.05 μg/h) to maintain the basic frequency of the oxytocin pulse generator, were infused either with exogenous estradiol equivalent to the ovarian secretion rate around the time of estrus (1 μg/h) or with exogenous progesterone equivalent to luteal phase progesterone secretion rate (500 μg/h). The infusion of estrogen (1 μg/h for 12–36 h) evoked a series of four to six large bursts of IMP in both uterine horns, each lasting 1–2 h at intervals of 3 h (472). The measurement of oxytocin in jugular plasma during these bursts of IMP (Fig. 25) revealed a series of low-amplitude high-frequency pulses of oxytocin, none of which exceeded 20 pg/ml (461). Similarly, in a separate group of ovariectomized sheep, the withdrawal of a 10-day infusion of exogenous progesterone (500 μg/h) superimposed on basal estradiol infusion (0.05 μg/h) also resulted in a series of large bursts of IMP in both uterine horns (Fig.26), similar in magnitude and duration to those observed after an infusion of high estrogen (1 μg/h). The onset of these large bursts of IMP occurred ∼24 h after beginning the infusion of high estrogen or after the withdrawal of exogenous progesterone. In the second model, using intact cycling sheep, the corpus luteum was surgically removed (luteectomy) during the luteal phase to exclude the corpus luteum as a source of oxytocin and to subject the animals to the withdrawal of endogenous progesterone (461). In all luteectomized animals, a series of large bursts of IMP occurred in both uterine horns simultaneously similar to those observed following high estrogen infusion or the withdrawal of exogenous progesterone in ovariectomized sheep. It was concluded that increasing the circulating levels of estradiol, or withdrawing progesterone in the ovariectomized animal or by withdrawing progesterone by luteectomy in the intact animal, caused the neurohypophysial oxytocin pulse generator to alter its frequency, thus producing a series of high-frequency pulses of oxytocin (460,461). The interval between the large bursts of IMP observed in animals subjected to hormonal manipulation was about 3 h, whereas during natural luteolysis, the intervals between episodes of IMP are ∼6–9 h. This difference may be explained by the fact that animals under experimental hormone infusions were subjected to acute changes in hormone levels, whereas during natural luteolysis, endogenous hormone levels change more gradually.

Fig. 25.

Concentration of oxytocin in 2 ovariectomized sheep maintained on low estradiol (0.05 μg/h) after infusion of high-level estradiol (1.0 μg/h) for 12 h (A) or 36 h (B). Horizontal lines indicate occurrence of high-intensity bursts of intramyometrial pressure that began 24 h after beginning infusion of high-level estradiol. [From McCracken et al. (461).]

Fig. 26.

Large bursts of intramyometrial pressure in an ovariectomized sheep maintained on low estradiol (0.05 μg/h) after cessation of a 10-day infusion of progesterone (500 μg/h). Large bursts occurred 24 h after progesterone withdrawal, and interpulse interval averaged 3 h. [From McCracken et al. (461).]

The concentration of oxytocin in peripheral plasma during the first luteolytic pulse of PGF in the intact cycling animal is ∼200 pg/ml (Fig. 22), whereas the level of oxytocin during the bursts of IMP in the hormone-treated ovariectomized animal is ∼20 pg/ml. Therefore, at the onset of luteolysis in the intact cycling sheep, the contribution of the neurohypophysis to circulating levels of oxytocin is ∼10%, whereas the corpus luteum contributes ∼90% of the oxytocin at this time. However, as luteolysis progresses, the magnitude of each increase in the plasma level of oxytocin declines by ∼50% with each successive episode of oxytocin release (Fig. 22). Thus the relative contribution of oxytocin from the neurohypophysis will be proportionately larger as luteolysis progresses (460,461).

D.  Regulation of Supplemental Secretion of Luteal Oxytocin

Several studies suggested that the systemic administration of PGF evoked the secretion of oxytocin from the ovine corpus luteum. This proposal was based on the prior observation that releases of oxytocin or its neurophysin usually occurred synchronously with pulses of PGF from the uterus (190, 209,319, 502). In addition, a PGF analog given systemically caused the secretion of ovarian oxytocin (208) and the systemic infusion of high levels of PGF in intact ewes increased peripheral levels of oxytocin-associated neurophysin, but not in ovariectomized ewes (706). Thus there was evidence that PGFcould stimulate luteal oxytocin secretion, although it was unclear from these reports whether the effect was direct or indirect. The direct effect of PGF on luteal oxytocin secretion was demonstrated by giving 10-min infusions of very low levels of PGF (5–100 pg/min) intra-arterially into the ovary of conscious sheep. These low levels of PGF evoked the secretion of luteal oxytocin without any effect on progesterone secretion (397). Prostaglandin F infused unilaterally into one uterine lymphatic in the anesthetized sheep released oxytocin from both ovaries (294), and the authors concluded that PGF acted indirectly, perhaps via the systemic circulation. In a subsequent study, the infusion of PGF into an ovarian lymphatic released oxytocin from the adjacent but not the opposite ovary (206). The authors suggested that the previously observed stimulation of oxytocin secretion by both ovaries after unilateral infusion of PGF into a uterine lymphatic was due to lymphatic connections between the two uterine horns. It is also possible that luteal oxytocin, released into the systemic circulation by exogenous PGF infused into one uterine lymphatic, caused an endogenous release of PGF in the opposite uterine horn and thus a local release of oxytocin from the opposite ovary. Such an explanation would require the presence of endometrial oxytocin receptors.

When subluteolytic levels of PGF were given continuously into an ovarian artery of the sheep for 10 h, luteal oxytocin was secreted during the first hour followed by desensitization. Recovery of the response to additional subluteolytic infusions of occurred after 6 to 9 h (396). In subsequent studies, it was found that immediately after desensitization to a subluteolytic infusion of PGF (100 pg/min for 2 h), a luteolytic infusion of PGF (2,500 pg/min for 2 h) not only evoked an additional secretion of luteal oxytocin, but also caused progesterone secretion to decline (134,135; see Fig. 27). The basis for this differential sensitivity of the corpus luteum to PGF is presently unknown. However, as discussed in section v A, possible explanations include the existence of high- and low-affinity states of the PGF receptor or differential increases in second messengers in response to different concentrations of ligand. Regardless of the explanation, small increases in circulating levels of oxytocin, caused by the periodic increase in the frequency of the central oxytocin pulse generator, would be expected to stimulate the low levels of uterine PGF secretion which, in turn, have been shown to initiate the large supplemental release of luteal oxytocin.

Fig. 27.

Ovarian secretion rate of progesterone and oxytocin in conscious sheep on day 12 of cycle after intra-arterial infusion of saline, a subluteolytic level of PGF (100 pg/min), and a luteolytic level of PGF (2,500 pg/min), each for a 2-h period. Secretion rate of progesterone begins to decline only during luteolytic infusion of PGF, indicating a differential sensitivity of corpus luteum to PGF. [From Custer et al. (134).]

E.  Proposed Model for Neuroendocrine Control of Luteolysis

To integrate the hormonal factors controlling the pulsatile secretion of uterine PGF and hence luteolysis in the sheep, a model has been developed to explain how the ovarian steroid hormones control this process (Fig.28).

1
Toward the end of the luteal phase, loss of progesterone action occurs due to downregulation of its own receptor, both in the hypothalamus and in the endometrium, which results in the return of estrogen action in these tissues.
2
Returning estrogen action will stimulate the hypothalamic oxytocin pulse generator to secrete high-frequency bursts of low levels of oxytocin intermittently and simultaneously upregulate endometrial oxytocin receptors in the uterus.
3
Low levels of PGF (subluteolytic) will be released from the uterus due to the interaction of posterior pituitary oxytocin and endometrial oxytocin receptors.
4
Low levels of PGF are sufficient to initiate the release of luteal oxytocin, acting via a high sensitivity state of the PGF receptor in the corpus luteum (HFPR) to initiate a supplemental secretion of luteal oxytocin.
5
The supplemental release of luteal oxytocin will now amplify the release of endometrial PGF.
6
Prostaglandin F production by the endometrium will now be high enough to activate the low sensitivity state of the PGF receptor (LFPR) and will inhibit progesterone secretion (luteolysis) and release additional luteal oxytocin, hence reinforcing endometrial PGF synthesis. Such a closed loop system will continue until the PGF receptor response system becomes desensitized, thus curtailing the supplemental release of luteal oxytocin.

Fig. 28.

Proposed model for neuroendocrine regulation of pulsatile secretion of PGF during luteolysis in sheep. For explanation of numbers, see text. [From McCracken et al. (461).]

The interval to the subsequent luteolytic pulse of uterine PGF will most likely depend on two factors:1) the next high-frequency burst of posterior pituitary oxytocin via the hypothalamic pulse generator, and 2) recovery of the endometrial oxytocin receptor that may be downregulated by oxytocin, at least during the early stages of luteolysis when the concentration of oxytocin receptors is low. Thus the uterus appears to act as a transducer that converts neural signals (oxytocin pulse generator) into uterine PGF pulses that are required for luteolysis. In the sheep and other ruminants, luteal oxytocin appears to act as a supplemental source of oxytocin that amplifies these neural signals (oxytocin pulse generator) and hence increases the magnitude of luteolytic pulses of uterine PGF. The observed desensitization of the corpus luteum to PGFafter about 1 h cannot be involved in regulating the pulse frequency of uterine PGF secretion because the frequency of PGF secretion is similar in hormone-treated ovariectomized or luteectomized sheep (443,630; see below). However, it is likely that desensitization of the corpus luteum to PGF controls the supplemental episodes of oxytocin secretion from this tissue. In nonruminant species, such as the sow and the mare, which do not synthesize large quantities of luteal oxytocin, uterine PGF is also released in a pulsatile manner and may be regulated solely by the central oxytocin pulse generator. However, it has recently been shown that oxytocin is synthesized in the endometrium of the cyclic sow (678), the cyclic mare (53, 707), and the pregnant rat (407). Therefore, it is possible that uterine oxytocin could serve to amplify the uterine production of PGF during luteolysis in the sow and mare and during labor in the rat.

The regulation of the central oxytocin pulse generator by estrogen and progesterone described above has remarkable similarities to the regulation of endometrial oxytocin receptors by estrogen and progesterone (453, 471). At the end of the cycle, progesterone influence on the endometrium wanes due to the catalytic loss of progesterone receptors by progesterone (116, 483,695). The loss of progesterone action and the consequent return of estrogen action then upregulates endometrial oxytocin receptors which, when interacted with oxytocin, will evoke PGFsecretion (461; see also Fig. 19). Both estrogen and progesterone receptors are present in the hypothalamus of most species including rodents (73), sheep (409), and cats (49), and progesterone has been shown to downregulate its own receptor in the hypothalamus (72, 496). Thus the general components of the estrogen and progesterone receptor system in the uterus are also present in the hypothalamus. It is likely that, as in the uterus, a similar downregulation of progesterone receptors by progesterone secreted during the luteal phase will occur in the hypothalamus. Such loss of progesterone action will upregulate estrogen receptors in critical neurons and, hence, allow circulating levels of estrogen to increase the frequency of the hypothalamic oxytocin pulse generator. The likelihood that estrogen acts directly on oxytocin-containing neurons in the hypothalamus is significantly enhanced by the finding that the β-form of the estrogen receptor colocalizes with oxytocin-containing neurons of the supraoptic and paraventricular nuclei in the rat (326).

In addition to excitation of hypothalamic neurons (7), gonadal steroids have been shown to upregulate oxytocin and vasopressin gene expression in the hypothalamus (16, 94), thus potentially amplifying steroid regulatory effects on the oxytocin pulse generator. The mechanism by which estrogen intermittently alters the frequency of the central oxytocin pulse generator is presently unknown. It has been suggested that the magnocellular neurons form an “electrical syncitium” which permits the magnocellular neurons to fire synchronously and precipitate an episode of oxytocin release (417). More recently, nitric oxide (NO) has been suggested as a mediator of central neurotransmitter action in oxytocin-containing neurons (22). Moreover, exogenous estrogen increases calcium-dependent NO synthase (NOS) activity in guinea pig brain (711). In rats, however, exogenous estradiol increases neuronal NOS expression in the ventromedial nucleus, but not in the paraventricular or supraoptic nuclei, the primary location of the magnocellular neurons (105). Interestingly, during dehydration in rats, the central inhibition of NOS activity was accompanied by augmentation of oxytocin release (663). Further studies will be required to determine whether or not changes in central NOS activity or expression may be involved in mediating estrogen-induced changes in the activity of the central oxytocin pulse generator. Adrenomedulin, a recently discovered vasodilator peptide (377), has been localized in oxytocin-containing neurons in the hypothalamus of the rat (682). However, a potential mediating role for adrenomedulin in the hypothalamic control of oxytocin secretion remains to be investigated. Because of the proposed role for the central oxytocin pulse generator as a pacemaker for luteolysis, it might be expected that section of the pituitary stalk would prevent or delay luteolysis. Rather surprisingly, section of the pituitary stalk in the sheep does not prevent normal cyclic regression of the corpus luteum (151, 439). However, the above observation is explained by the discovery that within 1 or 2 days of stalk section, posterior pituitary hormone secretion returns to normal from the axons proximal to the site of section (278, 438).

Several other studies support the view that estrogen and progesterone can act simultaneously at the level of the hypothalamus and the uterus. For example, the injection of a luteolytic level of estrogen during the luteal phase of the cycle in sheep causes a premature upregulation of endometrial oxytocin receptors, the appearance of premature PGF pulses (presumably initiated by the effect of estrogen on the central oxytocin pulse generator) and the premature regression of the corpus luteum (312). Further evidence is seen from a study in which ovariectomized sheep were treated with a regimen of estrogen and progesterone to simulate the levels attained during the cycle. Pulses of PGFM (PGF metabolite) were observed in peripheral blood at about the same time and frequency as those observed in intact cycling sheep, but the magnitude of PGFM pulses was reduced by ∼75% compared with the intact sheep (630). The appearance of pulses of PGFM in the ovariectomized animals most likely reflects stimulation of uterine PGF secretion by oxytocin derived solely from the central oxytocin pulse generator. The observed reduction in the magnitude of PGFM pulses can be explained by the absence of a supplemental secretion of luteal oxytocin. It is not known whether the reduced pulses of PGF (PGFM) would have been adequate to initiate luteolysis had the corpus luteum been present. Similarly, pulses of PGFM of reduced magnitude have been observed in luteectomized sheep treated with exogenous progesterone (443). It has been shown in the cow that a 60–75% reduction of oxytocin content of the corpus luteum does not prevent luteolysis in this species (387). However, the remaining oxytocin in the corpus luteum could have contributed to the magnitude of PGF pulses that presumably occurred in these animals. Removal of ∼70% of the granulosa cells from the bovine preovulatory follicle reduced plasma levels of progesterone from ∼7.0 to 1.5 ng/ml during the ensuing luteal phase, but cycle lengths were normal (484). Although not measured in the latter study, the oxytocin content of the modified corpora lutea was most likely low because of the reduction of the oxytocin-producing granulosa cells. Because luteolysis occurred at the normal time, these findings provide additional support for the view that luteal oxytocin may not be an absolute requirement for luteolysis in the cow. The key role of oxytocin and its uterine receptor in regulating lutolysis in sheep and cattle is underlined by the finding that a continuous systemic infusion of oxytocin delays or prevents luteolysis (210, 242, 323,386), which is thought to be due to chronic downregulation of the endometrial oxytocin receptor. Moreover, luteolysis is prevented or delayed in sheep (603) and goats (123) immunized against oxytocin or by treatment of goats with an oxytocin receptor antagonist (318).

In species that depend on the presence of the uterus for luteolysis, termination of the ovarian cycle via the luteolytic action of uterine PGF secretion appears to be mediated by the action of estrogen and progesterone, not only by regulating endometrial oxytocin receptors, but also by regulating the central oxytocin pulse generator. Thus, in addition to the well-established interdependence of ovarian steroid hormones and the anterior pituitary gonadotropins (FSH, LH, and prolactin) required to initiate follicular growth, ovulation, and corpus luteum function, a second interdependence appears to exist between ovarian steroid hormones and the posterior pituitary hormone oxytocin that is required to terminate the cycle in nonprimate species. Thus, for both the initiation and termination of the reproductive cycle, there is now good evidence for a close interaction between the ovary and the brain.

V.  MECHANISM OF LUTEOLYTIC ACTION OF PROSTAGLANDIN F

A.  Receptors for Prostaglandins

Although the involvement of receptors in the action of PG had been postulated (553), the first direct evidence for the existence of PG receptors was provided by Fried et al. (226), who showed that biologically inactive 7-oxa-acetylenic analogs of PGF antagonized the uterotonic action of natural PG. Subsequently, using radioactively labeled PG, binding sites were identified in a variety of tissues. Specific receptor binding sites for PGF in the corpus luteum have been identified in the sheep (43, 368, 369, 556, 557,569, 717, 730), rat (91, 428), sow (232), and human (559, 570). Evidence has accumulated that the luteolytic action of PGF is mediated by means of these specific plasma membrane receptors. For example, the luteolytic potency of PG analogs parallels their affinity for the PGF receptor in luteal membrane preparations (557), although some of the increased activity of these analogs in vivo is due also to their resistance to metabolism by the 15-OH-PG-dehydrogenase enzyme (464, 696). Furthermore, in the bovine (569) and porcine (232) corpus luteum, an increase in the number of PGF receptors occurs during the late luteal phase, which may explain the reduced luteolytic action of PGF when given early in the luteal phase. However, resistence to PGF early in the cycle may not be explained solely by the lack of luteal PGFreceptors, since high-affinity PGF receptors are present by day 2 in the cow (730), but rather to the lack of expression of other mediators (680).

In addition to binding studies, PG receptors have been characterized pharmacologically by comparing the evoked response of different PG and their synthetic analogs in bioassay systems (119). With the use of these systems, it was found that some prostanoid receptors exist as subtypes. Thromboxane A2 appeared to have two subtypes and PGE2 had three subtypes, designated as EP1, EP2, and EP3, whereas PGF, PGI2, and PGD2 appeared to exist as single types. More recently, receptors for prostanoids have been cloned, the first of which was thromboxane A2 (311). This study indicated that thromboxane A2 is a G protein-coupled rhodopsin type of receptor with seven transmembrane domains. Because there was only a 10–20% homology with other rhodopsin type receptors, it was proposed that the prostanoids constitute a subfamily of the rhodopsin type receptor family. On the basis of this model, screening of mouse cDNA libraries revealed the presence of seven different prostanoid receptors including the cloning of three receptor subtypes for PGE2 (511). These cloning studies are in good agreement with the previous pharmacological classification of the prostanoid receptors. In addition to the prostanoid receptors encoded by different genes, alternative splicing of the EP3 receptor transcript has yielded several isoforms of this subtype that couple to various G proteins to induce specific second messenger systems (450,515).

The PGF receptor has been cloned in several tissues including the corpus luteum in the mouse (659), rat (378, 394), cow (595), human (6,394), and sheep (263). The amino acid sequence and the proposed seven transmembrane segments for the bovine PGF receptor are shown in Figure29 (596). The existence of a single type of receptor for PGF is supported by the finding of a single mRNA species for the PGF receptor in ovine luteal tissue (180, 263, 523). However, equivocal evidence has accumulated with respect to high- and low-affinity sites for the PGF receptor based on binding affinities to cell membranes or separated whole cells from corpora lutea or liver. Some studies indicate only high-affinity binding sites for PGF (91, 595, 717), whereas others report the occurrence of both high- and low-affinity binding sites for PGF (43, 232, 518, 558, 568, 735). In the sheep, small steroidogenic cells were reported to possess a low-affinity PGF receptor while large steroidogenic cells possessed both high- and low-affinity PGFreceptors (43). However, Scatchard analysis of saturation binding studies did not reveal the presence of the low-affinity site identified by displacement analysis (43). It is also possible the low-affinity site represents a receptor for a different PG that may cross-react weakly with PGF. Moreover, mRNA for the PGF receptor has been localized in large steroidogenic cells in the sheep (348), although it was found in both large and small steroidogenic luteal cells as well as in endothelial cells in the cow (441).

Fig. 29.

Deduced amino acid sequence for bovine PGF receptor. Positions of 7 transmembrane segments (I–VII) were tentatively assigned by hydropathy analysis. Arrows indicate potentialN-glycosylation sites in amino-terminal region. Asterisks denote potential phophorylation sites by protein kinase C. [From Sakamoto et al. (596).]

Also, as discussed earlier, the action of PGF on the ovine corpus luteum in vivo indicates a differential sensitivity to PGF, since subluteolytic concentrations of PGF (25–250 pg/min), infused into the arterial supply of the ovary, selectively evoke the secretion of luteal oxytocin without any effect on progesterone secretion rate (396). Moreover, desensitization to a subluteolytic infusion of PGF was apparent after 1 h after which the gland was refractory to subluteolytic infusions of PGF for 6–9 h (396). On the basis of the reported dissociation constant of 4–6 nM for the luteal PGF receptor in the sheep (717), the infusion of a subluteolytic level of PGF (100 pg/min) would occupy <1% of the available PGF receptors. Therefore, the termination of the oxytocin response after 1 h is unlikely to be caused by receptor saturation, suggesting the involvement of other factors such as desensitization of second messenger systems. However, a luteolytic infusion of PGF (2,500 pg/min), following desensitization to a subluteolytic infusion (see Fig. 27), immediately evoked an additional release of oxytocin and now caused a decline in progesterone secretion rate (132, 134, 135). It has also been reported that cultured ovine granulosa cells or early luteal phase cells show a differential sensitivity to PGF in terms of oxytocin and progesterone release (152). This observed differential sensitivity of the corpus luteum to PGFcould be due to the existence of high- and low-affinity states of the PGF receptor as has been demonstrated for the leukotriene B4 receptor in human myeloid cells (633). Alternatively, there could be a single affinity state of the PGF receptor that shows a dose-dependent increase in second messengers with different concentrations of ligand. In NIH 3T3 cells, there is evidence for two forms of the PGF receptor, one that couples to adenylate cyclase and does not exhibit selectivity between PGF and PGE2 and another which couples to phospholipase C and shows a 40-fold selectivity toward PGF over PGE2 (273). In osteoblastic MC3T3-E1 cells, evidence points to a single PGF receptor that uses diverse signal transduction systems linked to phospholipase C activation (279). Thus it is possible that different concentrations of PGF may activate different signal transduction/second messenger systems in luteal cells, thus explaining the differential sensitivity of the ovine corpus luteum to PGF.

Prostaglandin F receptors in the ovine corpus luteum are localized on large luteal cells (608), which are the sole source of luteal oxytocin (587) and which secrete most of the basal output of progesterone (289). Moreover, in situ hybridization studies also localized the PGFreceptor message to large luteal cells in the ovine corpus luteum (523). Treatment of ovine luteal cells with PGF induces a transient increase in intracellular calcium followed by a sustained elevation that is observed in large, but not small, luteal cells (325, 448, 710, 727). Therefore, it is possible that the differential effects of low and high concentrations of PGF on the ovine corpus luteum in vivo may be controlled by concentration-dependent changes in intracellular Ca2+. It has been suggested that phospholipase C-activated inositol 1,4,5-trisphosphosphate (IP3) production releases Ca2+ in an all or nothing fashion from a series of intracellular Ca2+ pools (538). Thus increasing concentrations of agonist, in this case PGF, may recruit a series of Ca2+pools as IP3 concentration rises within the cell.

In mice lacking the gene encoding the receptor for PGF(FP), the estrous cycle, ovulation, fertilization, and implantation are normal (660). However, in FP-deficient mice, the prepartum decline in progesterone is absent, and the animal does not undergo parturition even when given exogenous oxytocin. A reduction of progesterone levels by performing ovariectomy 24 h before term caused an upregulation of uterine receptors for oxytocin and normal parturition in the FP-deficient mice. These experiments indicate that a luteolytic action of PGF is required in mice to diminish progesterone levels and thus permit the initiation of labor. Although uterine oxytocin receptors were upregulated in conjunction with labor in these mice, it is not clear that oxytocin is essential for labor in mice, since parturition is normal in oxytocin-deficient mice, although they fail to lactate (520,741). The luteolytic role for PGF in the induction of labor in mice is supported by the finding that mice lacking the gene encoding cytosolic PLA2 also had abnormal parturition (75, 686). The observed reduction in PG production in these mice is in keeping with the key role of PLA2 in cleaving the PG precursor arachidonic acid from phospholipids. Moreover, the administration of a progesterone receptor antagonist (RU-486) at term to substitute for the luteolytic decline in progesterone corrected the defect in labor seen in the PLA2-deficient mice (686).

B.  Functional Luteolysis

A recurring problem in studies on the effect of PGF on progesterone production by luteal tissue both in vitro and in vivo is the use of pharmacological concentrations of PGF. Although such studies provide support for the luteolytic action of PGF, the high levels used in many studies, particularly those designed to investigate the mechanism of action of PGF, may be confounding. The use of pharmacological levels of PGF is also evident when PGF is administered in vivo and the corpus luteum is removed at various time points to determine the levels of putative mediators of luteolysis in the harvested tissue. High concentrations of PGF are likely to produce exaggerated responses and may also produce artifactual biochemical effects in the tissue. Thus the results of many studies have to be carefully reevaluated depending on the concentration of PGF employed to achieve the responses. To some extent, this is also true for the method employed to separate various cell types from luteal tissue in an attempt to evaluate the cellular site(s) of PGF action. Most studies use enzymatic digestion of luteal tissue to separate the various cell components. The effect of such enzymatic digestion on the biochemistry of the cells and their plasma membrane receptors is difficult to evaluate. For example, mechanically dissociated luteal cells from the corpus luteum of the mare bound significantly more LH than did enzymatically dissociated cells from the same tissue (83). Moreover, enzymic separation of ovine luteal cell types causes the depletion of 99% of the oxytocin content of large luteal cells (587). Even conditions of tissue incubation have to be carefully evaluated, since changing the incubation medium from ovine luteal slices was alone sufficient to cause a maximum release of oxytocin into the medium (136).

Functional luteolysis, i.e., the decline in progesterone secretion, takes place over a period of ∼24–36 h in the sheep (Fig.30), and in primates, functional luteolysis occurs over a period of ∼48 h (746). However, in both sheep (464) and marmoset monkeys (221,223), treatment with high doses of exogenous PGF, or one of its synthetic analogs, can produce an accelerated onset and shortened duration of functional luteolysis. Because the release of PGF from the uterus occurs in a series of pulses in several nonprimate species, experiments were designed to mimic the pulsatile pattern of endogenous PGF during luteolysis observed in sheep. When a physiological pattern of PGF secretion was mimicked by giving five 1-h-long incremental infusions of PGF into the arterial supply of the transplanted ovary at 6-h intervals, permanent luteal regression was observed (471, 606). Moreover, a single hour-long infusion of PGF, given once daily for 4 days, caused a temporary fall in progesterone secretion after each infusion, followed by recovery to control values (606). Corpus luteum regression did not occur with this protracted infusion regimen, further supporting the view that a finite number of endogenous PGF pulses, occurring at relatively short intervals over a period of ∼24 h, is a requirement for the physiological regression of the corpus luteum in this species. The total amount of PGF required to cause luteolysis when given in a pulsatile fashion was 1/40 of the minimal amount of PGF required to cause luteolysis when given as a continuous infusion (see Fig. 5). Thus a pulsatile secretion of uterine PGF appears to be advantageous for luteolysis.

Fig. 30.

Alteration of pulsatile pattern of progesterone secretion in utero-ovarian venous plasma after onset of spontaneous luteolysis in conscious sheep. Pulse amplitude of progesterone secretion diminished after first luteolytic pulse of PGF. Blood samples were collected hourly on days 13–16.

To obtain luteal tissue representative of the biochemical events occurring during natural luteolysis, a model was developed in which PGF was administered systemically to intact cycling sheep. Prostaglandin F, infused at 20 μg/min for 1 h, caused a decline in peripheral plasma progesterone that reached a nadir between 6 and 8 h and recovered by 12 h postinfusion, whereas the same infusion rate of PGFM was ineffective (133). The results indicated that, at the dose used (20 μg/min), enough of the infused PGF (probably ∼1%) had passed across the lungs to act on the ovary. A regimen of hour-long systemic infusions of PGF (20 μg/min), based on the frequency of endogenous releases of PGF in the sheep (743), caused a fall in progesterone that closely mimicked the rate of decline in progesterone observed during natural luteolysis in sheep (133). Such a model permits the removal of the corpus luteum for biochemical analysis at selected times during the process of what is essentially physiological regression of the corpus luteum. Thus a specific pattern and duration of pulses is required to mimic the rate of decline in plasma progesterone during natural luteal regression in this species. This finding is supported by the observation that hysterectomy performed during the luteolytic process in sheep restored the secretory function of the corpus luteum and corpus luteum regression does not occur (500). Recently, it has been suggested that the luteolytic action of PGF may be enhanced by the delayed induction of PGG/H synthetase-2 within the ovine corpus luteum (680). Such an induction of the above enzyme may amplify the initial luteolytic signal of PGF by producing additional PGF within the corpus luteum. The observed delay in the induction of the enzyme may sensitize the corpus luteum to subsequent episodes of PGF and may account for the increased luteolytic efficiency of a pulsatile regimen of PGF infusions in sheep (606).

A variety of agents have been proposed to explain the mechanism of action of PGF in causing functional luteolysis. The range of proposed mediators of PGF action is very large, and a selection of these is shown in Table1. This selection is not intended to be comprehensive or all inclusive, but rather to illustrate the diversity of mechanisms proposed. It is beyond the scope of this review to give details of all of these. However, some of the more recently proposed mediators of luteolysis are discussed in some detail. Recent reviews on luteolysis, including mechanism of action, have been published with emphasis on the rat (54), the primate (481, 563,708), and the guinea pig and primate (563). For example, PGF-induced luteolysis has been correlated with luteal membrane composition and fluidity (100, 260), which may prevent LH receptor aggregation (427). Such a blockade of LH action could prevent coupling to adenylate cyclase with a decline in cAMP and progesterone synthesis (160, 304,393). There is also considerable evidence that PGF acts at the postreceptor level, since PGF activates the phospholipase C pathway and stimulates a rapid increase in intracellular Ca2+ in bovine luteal cells (140, 142). Mobilization of Ca2+from intracellular stores may result from phospholipase C-catalyzed generation of IP3 and binding of the IP3 to IP3 receptor (64). In addition to the phospholipase C pathway, there is evidence that PGFactivates the phospholipase D pathway, producing two second messengers, phosphatidic acid and diacylglycerol (189). Recent studies in Chinese hamster ovary cells, transfected with the bovine PGF receptor, indicate that activation of the phospholipase D pathway by PGF is mediated by both protein kinase C-dependent and protein kinase C-independent pathways (418). Regardless of the mechanism, the consequent increase in intracellular Ca2+ will further activate protein kinase C, which has been implicated in the luteolytic action of PGF (see below). In addition to the activation of protein kinase C via diacylglycerol from the phospholipase C or phospholipase D pathways, PGF could also activate the PLA2 pathway to generate arachidonic acid, which may directly activate the γ-isoform of protein kinase C (425). An important role for protein kinase C in luteolysis is based on the antigonadotropic action of phorbol esters (activators of protein kinase C) in luteal cells from human (1,141) and sheep (726).

View this table:
Table 1.

Proposed mediators of functional luteolysis

C.  Antisteroidogenic Action

A key step in the process of steroidogenesis is the cleavage of cholesterol to pregnenolone by the cholesterol side chain cleavage enzyme (P-450scc) located on the inner membrane of mitochondria (280). Thus potential mechanisms of action for the acute antisteroidogenic effect of PGF could involve a disruption of cholesterol transport into the mitochondrion or by an effect on P-450scc. Early studies of an effect of PGF on cholesterol utilization by bovine luteal cells suggested that PGF inhibits lipoprotein-stimulated progesterone production in vitro (542, 543). Further work revealed that PGF did not inhibit lipoprotein-stimulated increases in cellular cholesterol but actually increased mitochondrial cholesterol content (272). In this study, 25-hydroxycholesterol, which diffuses freely into mitochondria, stimulated progesterone production in the presence of PGF, leading to the conclusion that PGF may exert its luteolytic effect at a site after cholesterol transport to mitochondria, but before cholesterol side chain cleavage (272). In this connection, it is relevant that luteal tissue from regressing porcine luteal tissue is unable to utilize low-density lipoprotein as a source of cholesterol for progesterone synthesis (82). In the rat, there is evidence that PGF may interfere with cholesterol transport. Treatment with PGF decreased both mRNA levels and the activity of sterol carrier protein-2, a protein considered to play a major role in transporting cholesterol to mitochondria (477). However, in ovine luteal cells, activation of protein kinase C appears to inhibit cholesterol side chain cleavage, since treatment with phorbol 12-myristate 13-acetate (an activator of protein kinase C) partially blocks 25-hydroxycholesterol-stimulated progesterone production, whereas this activator had no effect on cells made protein kinase C deficient (728). Moreover, PGF was found to block lipoprotein-stimulated progesterone production in normal cells but had no effect on protein kinase C-deficient cells (725). The above findings suggest that steroidogenesis is more dependent on cholesterol delivery to the steroidogenic enzymes than on the absolute levels of these enzymes. This point has been clearly demonstrated in experiments in which transport-independent hydroxylated cholesterol substrates were shown to enhance steroid production in late luteal phase tissue or in PGF-treated luteal tissue despite low basal levels of steroid production (81, 724). These studies clearly demonstrate that P450scc and 3β-hydroxysteroid dehydrogenase were functional when cholesterol substrate was supplied. Taken together, the above studies in the sheep suggest that PGF may act before and after cholesterol transport to mitochondria. Such a proposal is supported by the finding that, in long-term culture of ovine luteal cells, PGF causes a partial inhibition of 22-hydroxycholesterol-stimulated progesterone production (205). Therefore, PGF may have multiple sites of action that mediate the antisteroidogenic effect of PGF.

A mechanism to explain cholesterol transport across the mitochondrial membrane has recently been provided by Clark et al. (115), who isolated and sequenced a mitochondrial protein which stimulated steroidogenesis in transfected MA-10 Leydig tumor cells. They named the 37-kDa protein StAR (for steroidogenic acute regulatory protein) and proposed that it facilitates the transport of cholesterol across the mitochondrial membrane to provide substrate for the synthesis of pregnenolone by the P-450scc enzyme (653). In view of the reported blockade of cholesterol transport by PGF in bovine (272) and ovine luteal cells (205, 725), StAR becomes an attractive candidate for the antisteroidogenic action of PGF. Indeed, several recent studies have reported that PGF treatment, both in vivo and in vitro, reduces the levels of StAR mRNA in ovine and bovine luteal tissue (348, 547). However, PGFcould also inhibit translation of mRNA for StAR. Because StAR itself has a very high utilization rate (653), a reduction in translation by PGF would be expected to have a relatively rapid effect on progesterone synthesis in luteal cells. The precise mechanism by which PGF acts on StAR remains to be determined. However, it is significant that removal of LH by hypophysectomy causes a decline in StAR mRNA in luteal tissue that can be restored by treatment with LH (348). An abrogation of LH action by PGF at the cellular level would be expected to diminish StAR, thus providing a plausible explanation for the antisteroidogenic action of PGF. It has recently been proposed that the 70-kDa heat shock protein-70 (HSP70) in MA-10 cells may play a role in regulating StAR and thus steroid synthesis (421). This is an interesting connection because HSP70 has been proposed as a mediator of PGF-induced luteolysis in the rat (366), human (367), and sheep (479). However, steroidogenesis can occur in the ovary, the testes, and the placenta in the apparent absence of a functional StAR protein (114).

Several recent studies have reported that endothelin-1 (ET-1), a 21-amino acid peptide synthesized by endothelial cells, inhibits gonadotropin-stimulated progesterone production by porcine granulosa cells (213, 342) and bovine luteal tissue (247-249). In the bovine study, coculture of luteal cells with endothelial cells increased the sensitivity of the culture to the antisteroidogenic effects of PGF (247). Further studies indicated that ET-1 inhibited progesterone production from steroidogenic luteal cells, an effect which was blocked by an ET-1 receptor antagonist (248). In the latter study, PGF elevated ET-1 content in luteal tissue, and the addition of an ET-1 receptor antagonist blocked PGFinhibition of progesterone production by luteal slices. A subsequent study indicated that the level of ET-1 and the expression of prepro-ET-1 was elevated in bovine corpora lutea around the time of natural regression and was detectable as soon as 2 h after exogenous treatment with PGF (248). Prostaglandin F, oxytocin, and vasopressin augmented ET-1 production in endothelial cells, and PGF also induced ET-1 expression in these cells after 15–90 min of incubation (249). These authors proposed that the antisteroidogenic effect of PGF may be enhanced by ET-1 production from luteal endothelial cells. The ET-1, so released, may then act locally on ET-1 receptors on steroidogenic luteal cells to diminish progesterone production, an effect which may be amplified by PGF-induced release of oxytocin from steroidogenic luteal cells. Further evidence that ET-1 may directly affect steroidogenesis is provided by the report that ET-1 inhibits progesterone production by dispersed ovine luteal cells, an effect which was blocked by an ET-1 receptor antagonist (310). Moreover, the administration of an ET-1 receptor antagonist directly into the ovine corpus luteum diminished the luteolytic effect of 10 mg Lutalyse given 30 min after the administration of the ET-1 receptor antagonist (309). With the use of a microdialysis system in bovine luteal tissue in vitro, ET-1 is reported to potentiate the decrease in progesterone caused by PGF(492). Because ET-1 depresses progesterone in vitro, its antisteroidogenic action is unlikely to be initiated by a decrease in blood flow, although it may contribute to the subsequent decrease as luteolysis progresses.

In a recent study in the cow (529), an intramuscular injection of cloprostenol (500 μg) caused a sharp peak in oxytocin concentration in ovarian venous blood as well as a gradual decline in progesterone levels. An increase in the concentration of ET-1 in ovarian venous blood was also observed that lasted ∼8 h. Moreover, after a similar intramuscular injection of cloprostenol, ET-1 levels in microdialysate from the corpus luteum in vivo increased as progesterone in the microdialysate declined. Thus there is considerable evidence that ET-1 may play a regulatory role in PGF-induced luteolysis in ruminants. The proposal that PGFstimulates ET-1 release from endothelial cells would require the presence of receptors for PGF in these cells. Indeed, a recent report has shown that mRNA for the PGF receptor is expressed in both steroidogenenic and endothelial cells in the bovine corpus luteum (441). The reports that ET-1 inhibits both gonadotropin-stimulated progesterone production by porcine granulosa cells (213, 342) and basal production by human granulosa cells (352), together with the report that the ET-1 gene and protein are expressed in human granulosa cells (435), indicates that further studies on the role of ET-1 in luteolysis are warranted.

An additional regulatory mechanism for the control of luteal function may be the manner in which gonadotropins in capillary blood reach their steroidogenic target cells in the corpus luteum. Binding of LH to ovarian and testicular blood vessels has been reported in the rat (88). Moreover, endothelial cells in the ovarian and testicular microvasculature are reported to bind and transport hCG to adjacent endocrine cells by a process termed “transcytosis” (238). Endothelial cells in other organs did not exhibit a capacity for transcytosis. At present, it is unknown if transcytosis of gonadotropins across endothelial cells occurs within the corpus luteum. However, a rapid block of gonadotropin uptake in vivo by rat corpora lutea has been reported following exogenous treatment with PGF (55). Therefore, a blockade of transcytosis may explain this rapid action of PGF. Taken together with the proposed intermediary role for endothelins in luteolysis (248, 249), the role of endothelial cells in luteolysis requires further investigation.

D.  Blood Flow

Initially, it was suggested that PGF caused luteolysis by virtue of a constricting effect on blood flow through the utero-ovarian vein (550-552). In addition, the early paradoxical finding that PGF stimulated progesterone production by luteal tissue in vitro (648), but depressed progesterone secretion in vivo (466), underlined the possible importance of an intact vascular bed as a condition for luteolysis. However, emphasis on blood flow diminished when subsequent studies showed that PGF could depress progesterone production in vitro during long-term incubations at lower doses (149, 304, 527). Early studies employing radioactively labeled microspheres (375, 526, 674) and later studies employing microspheres and/or Doppler flow measurements (216,525) implied that PGF caused luteolysis by a reduction in ovarian or corpus luteum blood flow. However, later investigations, also using microspheres, but at lower doses of PGF, indicated that no change in total or regional ovarian blood flow could be detected at the onset of PGF-induced luteolysis (85, 343).

Although no change in total ovarian blood flow at the time of functional luteolysis was observed in the sheep (451) or rat (537), it was considered possible that PGF might cause a regional change in flow through the corpus luteum without affecting total ovarian blood flow. Therefore, capillary blood flow was measured in the corpus luteum in both the autotransplanted and in situ ovary of the sheep before and after an intra-arterial infusion of PGF (176,177). A miniaturized Geiger-Muller probe was inserted into the corpus luteum of each animal, and krypton-85 dissolved in saline was infused periodically into the arterial supply of the ovary. The clearance rate of krypton-85 from 1 mm of luteal tissue nearest the probe was determined by measuring β-emissions. Capillary blood flow, calculated from t/2 (401) indicated that functional luteolysis, defined as the time when progesterone levels fell to 50% of mean controls (176), occurred after 1–2 h, several hours before any change in capillary flow could be detected (177). In support of this view, there was no evidence of anoxic damage in ovine luteal tissue before the reduction of progesterone secretion in the early stage of luteolysis in sheep (683). This is consistent with the finding that total ovarian blood flow in sheep does not decrease until after the onset of luteolysis in the intact cycling sheep (449).

In the rabbit in which estrogen is the major luteotrophin controlling progesterone secretion, the withdrawal of estrogen caused a precipitous drop in peripheral progesterone concentration within 24 h, whereas luteal blood flow was unchanged (357). Although it is possible to demonstrate a reduction in total blood flow through the ovary or corpus luteum with very high doses of PGF(38, 375, 451), no acute changes in total or regional blood flow were demonstrated with more physiological doses of PGF (85, 177, 343, 537). However, the effect of pulsatile infusions of PGF on luteal blood flow over the duration of luteolysis has not been reported. As discussed earlier, a single 1-h-long systemic infusion of PGF in sheep will depress plasma progesterone levels, which reach a nadir after 6–8 h and which recover after 12 h. An additional 1-h infusion of PGF given 8 h later prevents recovery, and progesterone continues to further decline (133). Whether this synergism of pulses of PGF is due to the enhancement of an antisteroidogenic effect or the result of reduced blood flow remains to be determined. In the sheep, exogenous PGF induces both functional and structural regression of the corpus luteum in a manner similar to the changes seen during natural regression, but functional luteolysis was found to precede structural changes (683). The decrease in blood flow, associated with the longer term reduction in secretory function and size of the corpus luteum, may play a role in structural luteolysis. The later structural changes in the ovine corpus luteum resemble changes associated with anoxic damage (144, 532). Also, in this species, phagocytosis of endothelial cells is observed during structural luteolysis and may contribute to the continued reduction in blood flow by reducing the capillary bed (536). Moreover, endothelins, which are proposed to play a role in functional luteolysis (248, 249), may contribute to the reduction in luteal blood flow during structural luteolysis by virtue of their vasoconstricting properties. In the human corpus luteum, the number of capillary blood vessels was reduced after the onset of luteolysis, then their number declined slowly over several weeks (493). In the latter study, functional regression of the steroidogenic cells in the corpus luteum preceded regression of the vasculature, which is in keeping with the concept that blood flow reduction is not the initiating factor in functional luteolysis. If it were, then necrosis of the steroidogenic cells due to anoxia would predominate, which is not the case. Because apoptosis (see sect.v E) is the major feature observed in steroidogenic cells during luteolysis, it seems likely that the subsequent regression of the vasculature in the corpus luteum is related to the earlier decline in the steroidogenic capacity/metabolic requirements of the gland. Thus the key to understanding how PGF causes regression of the corpus luteum may lie in how it abrogates steroidogenesis.

The corpus luteum of the rat and guinea pig lacks innervation (685), and denervation of the rat ovary did not affect luteal blood flow (240), suggesting that sympathetic control of blood flow in the corpus luteum is unlikely. Because PGF receptors are localized on steroidogenic luteal cells (608), the primary antisteroidogenic action of PGF most likely occurs in these cells. Degenerative changes in the cellular components of the corpus luteum, including the endothelial cells, may explain the reduction of blood flow as luteolysis progresses. This view is supported by the observed loss of endothelial cells during luteal regression (37, 195). As luteolysis proceeds, a decline in luteal blood flow occurs in parallel with the decline in progesterone levels (216, 525).

E.  Structural Luteolysis

Because of its transient nature, the corpus luteum is an unusual endocrine gland. It grows extremely rapidly, functions in the cyclic animal for 10 to 14 days, and then involutes to form the corpus albicans, composed of connective tissue and collagen. The process of involution is termed “structural” luteal regression, as distinct from “functional” regression, which signifies the decline of the progesterone secretory capacity of the corpus luteum. In most species, the distinction between functional and structural regression is not clearly documented, and the two events may not be entirely separate. However, in unmated rodents such as the rat, the corpora lutea of the short 4-day cycle do not become fully functional, although they persist for several cycles (424), indicating that in this species, structural and functional capacities are separate entities. Although PGF causes functional luteolysis in the rat, it does not cause structural luteolysis, the latter being caused by prolactin only after functional luteolysis has occurred (54).

A wide variety of agents have been proposed as mediators of structural luteolysis, and a selection of these is listed in Table2. It has been reported that macrophages and other immune cells play a role in luteal regression via the release of tumor necrosis factor-α (TNF-α) and other cytokines (63). Tumor necrosis factor-α may also be involved in an antisteroidogenic effect during luteolysis, since TNF-α inhibits LH-stimulated progesterone production in bovine luteal cells (63). Mutual histocompatability complex antigens have been suggested as a possible trigger for releases of immune cell cytokines (62). It is likely that immune cells and macrophages participate in structural luteolysis. For example, macrophages have been observed to phagocytose luteal cells in the postpartum corpus luteum of the guinea pig (536). Monocyte chemoattractant protein-1 (MCP-1), a 15-kDa protein, is a potent chemotactic agent for monocytes/macrophages that is expressed in regressing corpora lutea of the rat (677). Moreover, in hypophysectomized rats, MCP-1 expression increased in response to exogenous prolactin (79), a known luteolytic stimulus in this species. Immunoreactive MCP-1 was localized in the intracellular spaces of regressing corpora lutea, and it increased as luteolysis progressed (79). In the sheep, a 1-min infusion of PGF (1 μmol) into the ovarian artery at mid cycle caused a rapid expression of MCP-1 at 1 and 4 h postinfusion, an effect which was replicated by an infusion of polymyristic acid (a protein kinase C activator). In the same study, a similar rapid increase in MCP-1 mRNA was observed in the corpora lutea of cows treated with intramuscular 25 mg PGF at mid cycle, but not on day 4 when the corpus luteum is not fully responsive to PGF (679). The authors concluded that MCP-1 may play an important role in structural luteolysis by increasing the migration of monocytes/macrophages into the corpus luteum. In the latter study, MCP-1 was not found in cultured large luteal cells either before or after PGF treatment. In a similar study, a luteolytic regimen of PGF in sheep (2.5 mg im followed by 10 mg 2 h later) was shown to increase the expression of MCP-1 in the corpus luteum at 2 h, increasing at 4 and 8 h and declining at 16 h after PGF treatment (293). In this study, MCP-1 also was not localized in large luteal cells. However, other cellular components of the corpus luteum may be responsible for MCP-1 expression, since endothelial cells, fibroblasts, and macrophages themselves express MCP-1 (414). Thus, during luteal regression, MCP-1 is a potential mediator of the invasion of macrophages which may serve to phagocytose cellular components as well as produce cytokines and other factors. However, because hysterectomy extends the life span of the corpus luteum in many species, the stimulus for the involvement of macrophages and other immune cells most likely follows the initial onset of functional luteolysis and apoptotic cell death due to PGF (see below).

View this table:
Table 2.

Proposed mediators of structural luteolysis

Changes in lysosomal function have also been observed in association with luteolysis. Lysosomes, which are abundant in regressing sheep corpora lutea, exhibit increased fragility (157), an effect which was reproduced by exogenous PGF in rats (712). Later studies indicated that PGFincreased the release of marker enzymes (β-glucouronidase and acyl hydrolases) from isolated lysosomal preparations (350,713). It is possible the lysosomal enzymes may play a role in the degenerative changes associated with the structural involution of the corpus luteum. Physical changes in the structure of lipids in the plasma membrane of luteal cells undergoing luteolysis have been reported (86). This would be compatible with a role of lysosomes in structural regression, particularly since it has been proposed that the plasma membrane influences lysosomal function (665).

During structural regression, considerable tissue remodeling takes place as the corpus luteum becomes the corpus albicans. Tissue metalloproteinases have been suggested as mediators of such tissue remodeling, particularly of the extracellular matrix (65). The corpus luteum produces specific tissue inhibitors of metalloproteinases (TIMP) that may help to maintain the structural integrity of this gland. The most abundant metalloproteinase inhibitor in the corpus luteum of the sheep (639), cow (638), pig (637), and human (172) is TIMP-1. In the marmoset corpus luteum, TIMP-1 levels decline significantly 24 h after inducing luteolysis with a GnRH antagonist or PGF-induced luteolysis (171). A decline of TIMP-1 would be consistent with increased activity of metalloproteinases during structural regression. In the human corpus luteum, levels of TIMP-1 were relatively constant during the luteal phase (172), whereas in the cow, expression of luteal TIMP-1 increases up to 24 h after PGF-induced luteolysis (638). It is possible that TIMP may have some effects on steroidogenesis, since within the sequences expressed in the bovine corpus luteum, there is a 124 base homology between StAR and TIMP (291). Moreover, there is evidence that TIMP has a direct stimulatory effect on steroidogenesis, at least in the testis (78). In addition, the regulation of the extracellular matrix in the corpus luteum by the proteinase inhibitor plasminogen activator inhibitor type 1 has been suggested to play a role in tissue remodeling of the corpus luteum in the rat (419), sheep (636), and rhesus monkey (420).

Apoptosis, derived from the Greek “apo” meaning leaf and “ptosis” meaning to drop, is a term that was coined to designate programmed cell death. Generally, it is used to describe a series of morphological changes observed during cell death (355). Cell death has been categorized into two types, namely, necrosis and apoptosis (738). According to these authors, cell necrosis is characterized by swelling of the cell and rupture of the plasma and organelle membranes resulting in lysis of the cell. Classically, cell necrosis is caused by injurious agents such as toxins or prolonged ischemia. In contrast, apoptosis denotes a selective or programmed cell death that is characterized by chromatin condensation and nuclear shrinkage in the presence of membrane integrity. Later, protruberances in the plasma membrane form into globules (apoptotic bodies) containing cytoplasm or nuclear fragments, which are phagocytosed by macrophages and even by adjacent tissue cells. Exudative inflammation, such as that which accompanies necrosis, is not present. Because lysis of the cellular components of the corpus luteum does not occur during regression of this tissue, the term luteolysis may be something of a misnomer. However, the term luteolysis is so firmly entrenched in the literature that its use will, no doubt, continue. Recently, it has been proposed that the removal of apoptotic cells by macrophages is mediated by a receptor whose interaction with “nonself” components such as bacterial lipopolysaccharide triggers an inflammatory response, whereas interaction of the receptor with “self” components (apoptotic cells) initiates phagocytosis (153). Apoptosis can be caused by a loss of trophic support or by a wide variety of different stimuli. Examples of physiological apoptosis include regression of the thymus gland by glucocorticoids and the loss of tadpole tails by the secretion of thyroid hormone (738).

Considerable evidence has accumulated that structural luteolysis is yet another example of apoptosis. Apoptosis has been reported to occur during corpus luteum regression in ruminants (347, 599,751). In regressing luteal cells from the rat, endonuclease activity is increased (747), consistent with the view that apoptosis is involved in granulosa and luteal cell death in the rat (676). During structural luteolysis in pseudopregnant mice (which occurs between days 11 and 13 postmating), there is a good correlation between the onset of luteal cell apopotsis and the maximum expression of mRNA for the PGF receptor (292). Moreover, the Fas/Fas ligand system has been implicated in the apoptotic process that occurs during luteolysis in mice (594). The protooncogene c-myc is one of several proteins associated with cell proliferation (737) and cell death (188). In the marmoset corpus luteum, c-myc is present throughout the luteal phase but undergoes a transient rise and transfer to the nucleus of luteal cells during PGF or GnRH-induced luteolysis (221). Recent work suggests that different cell death pathways may exist in the marmoset monkey during natural luteolysis and during luteolysis induced by PGF or by a GnRH antagonist. However, the predominant forms of cell death under these conditions corresponded neither to apoptosis nor to necrosis (222). In the rat corpus luteum, the protooncogene c-jun is implicated in the luteolytic action of PGF (359). It is possible that these, or other proteins, may play a permissive role in structural regression following the blockade of tropic support of steroidogenesis during luteolysis. Detailed biochemical aspects of apoptosis in a variety of tissues, including follicular granulosa cells and luteal cells, have been reviewed recently (20, 327,609).

VI.  ABROGATION OF LUTEOLYSIS IN EARLY PREGNANCY

Because progesterone is required for the maintenance of pregnancy in cyclic mammals that normally undergo luteolysis at the end of the cycle, mechanisms have evolved to inhibit luteolysis in the event of pregnancy. Several strategies exist in different species to achieve this objective. Early studies in the sheep (501) demonstrated that the transfer of an embryo to the uterus of a nonpregnant sheep whose cycle was synchronized to the donor prevented regression of the corpus luteum in the recipient even if transferred as late as day 13 of the cycle (the time of the normal onset of luteolysis). The same effect was achieved when homogenates of embryos or their secreted products were infused into the uterus of nonpregnant recipient sheep (185, 499). The active component produced by the embryo was proteinaceous and was tentatively termed “trophoblastin” (447). When PGF was identified as the mediator of luteolysis in the sheep (459) and several other species (321, 468), it seemed likely that the presence of an embryo prevented luteolysis either by suppressing uterine production of PGF(anti-prostaglandin-secreting effect) or by the production of a substance that protected the corpus luteum against the luteolytic action of PGF (luteoprotective effect). Indeed, there is evidence that one or both effects may occur in a number of species. During the establishment of pregnancy in women, endometrial PG production is low in both intrauterine and extopic pregnancies (5). However, the mechanism for the suppression of uterine PG production in primates most likley results from continued secretion of progesterone via the rescue of the corpus luteum by hCG (see sect.vi B).

A.  Antiprostaglandin-Secreting Effect of Pregnancy

Early studies indicated that uterine pulses of PGF, normally secreted during luteolysis, were diminished or absent at this time in the pregnant sheep (45,673) (see Fig. 31). Such a reduction in PGF pulses during the establishment of pregnancy was subsequently confirmed by the measurement of the primary metabolite of PGF (PGFM) in the peripheral blood of sheep (see Fig. 32) (549,742). A similar reduction in uterine PGFproduction during the establishment of pregnancy was also reported in the cow (341), pig (495), goat (370), and guinea pig (564). However, in some of these species, the basal level of PGF is elevated in early pregnancy. For example, during early pregnancy in the goat, the pulsatile pattern of PGF is suppressed, but basal levels of PGF, measured as PGFmetabolites, are higher than in the nonpregnant animal (Fig. 13). A similar elevation of basal levels of PGF during early pregnancy in the sheep has also been reported (742). The significance of these higher basal levels of PGF during early pregnancy remains to be determined. Because the interaction of oxytocin with its receptors was implicated in the secretion of PGF in sheep (581), it seemed likely that the reduction of PGF during early pregnancy might be related to the suppression of endometrial oxytocin receptors. Indeed, it was found that the infusion of oxytocin into the uterine artery of sheep on day 16 of pregnancy did not evoke the secretion of PGF, as was observed in nonpregnant sheep at this time (453). Moreover, the concentration of oxytocin receptors in the endometrium was found to be 33 fmol/mg protein in day 16 pregnant animals, whereas in the cyclic animal on day 16, the concentration was >250 fmol/mg protein (453,471). Such a difference in oxytocin receptor levels between pregnant and nonpregnant sheep was also observed by Sheldrick and Flint (617). These studies suggested that trophoblastin, the protein secreted by the embryo (447), suppressed the formation of endometrial oxytocin receptors and thus reduced the magnitude of PGF pulses during early pregnancy in sheep.

Fig. 31.

Episodic pulses of PGF in utero-ovarian venous plasma in cyclic nonpregnant and early pregnant sheep over normal time of luteolysis. [From Barcikowski et al. (45).]

Fig. 32.

Concentration of PGFM in jugular venous plasma of cyclic nonpregnant (A) and early pregnant sheep (B) during normal time of luteolysis. Large episodic pulses of PGFM seen during luteolysis were diminished or absent during early pregnancy. [From Peterson et al. (549), Copyright 1976, with permission from Elsevier Science.]

A protein similar to trophoblastin that is secreted by the sheep preimplantation blastocyst between days 13 and 21of pregnancy was identified and designated as ovine trophoblast protein-1 (OTP-1) (254). Moreover, OTP-1 receptors are present in the epithelium of the luminal and superficial glands of the endometrium on day 16 of pregnancy in sheep (255). Ovine trophoblast protein-1 was sequenced and found to be an interferon-like protein (332), later designated interferon-τ (584). Infusion of conceptus secretory proteins, recombinant ovine interferon-τ, or bovine interferon-α1 in cycling sheep prevents the formation of oxytocin receptors and extends the length of the cycle (203, 446,487, 488, 539, 647, 652, 688). Thus it is well established that, in sheep, the luteolytic pattern of PGF pulses is abrogated by interferon-τ secreted by the preimplantation embryo. A similar protein is also secreted by the bovine trophoblast during early pregnancy (584) and prevents luteolysis when infused into the uterus of cyclic cows (286). The mechanism by which interferons block the formation of endometrial oxytocin receptors in early pregnancy in the sheep is considered to be due to suppression of transcription of the estrogen receptor gene and, hence, lack of upregulation of the oxytocin receptor (395, 646). Thus interferon-τ, secreted by the blastocyst during days 13–21 of pregnancy, appears to rescue the corpus luteum from luteolysis by temporarily extending the blockade of estrogen action, which is mediated by progesterone up until about day 13. It has also been reported that recombinant bovine interferon-τ blocks oxytocin-induced expression of cyclooxygenase-2 and prostaglandin F synthase in cultured bovine endometrial cells (739). Thus interferon-τ may act at several loci to inhibit endometrial prostaglandin synthesis during early pregnancy in ruminants. When the trophoblast is surgically removed from the ovine uterus on day 21 after mating, the corpus luteum persists for a further 20–40 days (447, 471). This finding indicates that, once the luteolytic process is abrogated between days 13 and21 by OTP-1, a luteolytic event cannot be mounted for a considerable period of time. The mechanism for the eventual curtailment of the life span of the corpus luteum, extended by removal of the embryo, is not known.

During early pregnancy in sheep, ovarian oxytocin is discharged from the ovine corpus luteum over days 13–20, although pulses of oxytocin or oxytocin neurophysin in plasma are much lower and occur less frequently than in the cyclic sheep (192). Ovarian oxytocin is also discharged during the same time period in the hysterectomized sheep (616). The stimulus for the secretion of luteal oxytocin in early pregnancy or in the hysterectomized sheep is unclear, but it could originate in the ovary itself, since systemic treatment with indomethacin suppressed surges of neurophysin in peripheral plasma, whereas intrauterine treatment was ineffective (191). In support of an intraovarian mechanism, small peaks of oxytocin occur at ∼12-h intervals on cycledays 14–17 from the autotransplanted ovary in the sheep (Custer and McCracken, unpublished data). This model is equivalent to hysterectomy, since the uterus is left in situ and does not influence the transplanted ovary. It remains to be determined whether the peaks of oxytocin secreted from the transplanted ovary are spontaneous or controlled by humoral mechanisms.

In the pig, estrogen production by the implanting blastocyst may act to divert the secretion of endometrial PGF from an endocrine mode to an exocrine mode, thus accounting for the absence of PGF in uterine venous blood (50, 270,271). Such a proposal is supported by the presence of high levels of PGF in uterine fluids during early pregnancy (744). However, it is unclear how much of the accumulated PGF comes from the blastocysts, which also produce PGF (415). In perifused porcine endometrium, PGF secretion was reoriented by calcium ionophore estradiol and prolactin (271), and receptors for prolactin have been found in porcine endometrium (740), suggesting that both estrogen and prolactin may be involved in the endocrine/exocrine shift in pigs. Trophoblasts in pigs also secrete interferons (392), but these proteins may not have an antiluteolytic effect, since the infusion of several trophoblast interferons in cyclic animals does not extend the cycle (408), although total trophoblast proteins cause an increase in PGE2 secretion (287). During the establishment of pregnancy in pigs on days 11–12postconception, secretion of PGF in uterine venous blood increased transiently and was accompanied by a marked increase in the secretion of PGE (112). The increase in PGE coincided with an increase in progesterone secretion, suggesting that, in the pig, PGE may have a luteotropic effect as well as a luteoprotective effect (see sect. vi B).

In the horse, PGF secretion by the uterus is diminished in early pregnancy, thus preventing regression of the corpus luteum (163). The equine blastocyst does not elongate as in ruminants and pigs but maintains contact with the entire endometrium by traversing the uterus every 2 h during pregnancy recognition (411). Such intrauterine migration by the blastocyst is thought to inhibit PGF secretion, but the mechanism for inhibition is not established. However, endometrial oxytocin receptors are not suppressed during early pregnancy in the mare (614) as they are in ruminants (229, 471). Moreover, neither endogenous oxytocin (614) nor exogenous oxytocin (256) causes a release of uterine PGF in mares during early pregnancy. These findings suggest that substances produced by the equine trophoblast may inhibit uterine oxytocin-stimulated PGF secretion without inhibiting the density of endometrial oxytocin receptors. Betweendays 40 and 150 of pregnancy in the mare, endometrial cups secrete large amounts of gonadotropin, which causes the formation of multiple accessory corpora lutea. The latter secrete progesterone until the placenta assumes this role later in pregnancy (13, 613, 745). In the rat, cervical stimulation during mating reflexly stimulates pituitary secretion of prolactin, which causes the corpus luteum to secrete progesterone (641). However, in pseudopregnant animals, the uterus secretes PGF on days 10–12, resulting in luteolysis (104). In the pregnant guinea pig, the luteolytic secretion of uterine PGF is abrogated, and progesterone production is extended until the placenta secretes a luteotropic substance (562). In the rabbit, which is a reflex ovulator, pregnancy is maintained by progesterone secreted by the corpora lutea, the latter being maintained by follicular estrogen (233). However, during pseudopregnancy, the corpora lutea undergo premature luteolysis due, in part, to the secretion of PGF by the uterus (432). The endometrial concentration of prostaglandin is suppressed in both early intrauterine and ectopic pregnancies in women (5), indicating that secretory products of the conceptus do not directly inhibit endometrial prostaglandin synthesis. This suppression of endometrial prostaglandins in primates most likely is mediated via the continued secretion of progesterone by the corpus luteum, which is rescued through the production of hCG by the implanting blastocyst in both intrauterine and ectopic pregnancies (see sect. vi B).

B.  Luteoprotective Effect of Pregnancy

In addition to the suppression of endometrial PGFsynthesis during early pregnancy, there is evidence that the corpus luteum in some species is more resistant to the luteolytic action of PGF. In the sheep and cow, experiments involving the anastomosis of the uterine vein or the ovarian artery from the pregnant horn to the nonpregnant horn indicated that blastocyst luteotrophins may overcome the effect of luteolysins (245). Also, the injection of 200 μg PGF into an ovarian artery of seven early pregnant sheep caused luteolysis in only two animals (444). A similar amount of PGF injected into follicles of early pregnant sheep caused luteolysis in only 17% of the animals as compared with 83% in nonpregnant sheep undergoing the same treatment (334, 565). The injection of exogenous estradiol at doses causing premature luteolysis in cyclic sheep is less effective in pregnant sheep (379). In later studies, the corpus luteum of pregnant sheep proved more resistant to a luteolytic injection of 100 μg cloprostenol when multiple embryos were present (510). Similarly, the amount of exogenous PGF required to induce luteolysis in pregnant sheep was more than that required to cause luteolysis in nonpregnant sheep, the resistance to PGF being greatest on days 13–15 (628). It is possible that PGE2, produced by the blastocyst or the endometrium, may counteract the luteolytic effect of PGF. As a stimulator of cAMP and a vasodilator, PGE2 has properties that are opposite to PGF (304). It has also been shown that PGE2 can counteract the luteolytic effects of PGF for short periods of time (118, 305, 436,565). Moreover, an increase in PGE2 in uterine venous blood has been reported during early pregnancy in sheep (629). Because of its structural similarity to PGF, a small amount of PGE may be transported locally from the uterus to the ovary by countercurrent transfer from the uterine vein to the ovarian artery. With the use of cultured bovine endometrial cells, it has been proposed that interferon-τ may transform the response of the endometrium to oxytocin from stimulation of PGF to stimulation of PGE2 (26). Thus PGE2 may be one of the factors that protects the corpus luteum from luteolysis during early pregnancy in ruminants. In support of this view, the infusion of PGE2 into the uterus of nonpregnant ewes delays luteolysis (437). Similar effects of uterine infusion of PGE2 have been observed in cows (574). However, PGE2 infused into the ovary in a physiological range did not prevent PGF-induced luteolysis in sheep (471).

During early pregnancy in the pig, PGE2 is increased in the uterine lumen (237). The intraluteal or intrauterine infusion of PGE2 has a luteoprotective effect against exogenous or endogenous PGF in pigs provided that sufficiently high doses of PGE2 are employed (8,215). Collectively, these studies provide good evidence that PGE2 has a luteoprotective/luteotropic role during the establishment of pregnancy in pigs. It is suggested that the changing pattern of PGF and PGE secretion during early pregnancy in the pig is due to the effect of estrogen produced by the blastocyst (112) and/or by an estrogen-mediated increase in endometrial prolactin receptors (271, 740). Both prolactin and indomethacin will extend the life span of the corpus luteum in the hysterectomized pseudopregnant rats (66).

In primates, progesterone from the corpus luteum is required to maintain early pregnancy until placental production of progesterone begins. The timing of the luteoplacental shift varies among primates. In humans, luteectomy performed before the fifth week of gestation, but not after the seventh week, resulted in abortion (128). In rhesus monkeys, neither the corpus luteum nor the ovaries are necessary to maintain pregnancy from the third week after fertilization (259, 315). The extension of the life span of the corpus luteum during pregnancy in primates is caused by the action of CG, which is secreted by the blastocyst around the time of implantation (299, 314). In addition to extending the life of the corpus luteum, CG promotes steroidogenic and peptinergic functions of the primate corpus luteum (654). Hearn et al. (301) suggested that, in addition to CG, the blastocyst in primates may secrete other factors that contribute to the rescue of the corpus luteum in early pregnancy. Because the corpus luteum is rescued in both eutopic and ectopic pregnancies in women (5), such additional factors would have to act systemically. However, it has been demonstrated that a regimen of exogenous hCG given to simulate the levels seen in early pregnant monkeys will effectively rescue the corpus luteum in cyclic animals (658), indicating that hCG, in the absence of other trophoblastic factors, is capable of rescuing the corpus luteum.

The presence of progesterone receptors in the primate corpus luteum (306, 657) provides support for the view that progesterone may have an autoregulatory role by promoting its own secretion from the corpus luteum (591). Because of the apparent lack of estrogen receptors in the primate corpus luteum, it has been suggested that, instead of the classic upregulation of progesterone receptors by estrogen, the ovulatory surge of LH may serve this function (657). On this basis, it is possible that CG may perform a similar function by upregulating progesterone receptors during the rescue of the corpus luteum. In rhesus monkeys in which the medial basal hypothalamus is lesioned, the secretion of progesterone by the corpus luteum can be maintained by exogenous pulses of GnRH given every hour (330). Temporary withdrawal of GnRH supports reduced plasma progesterone to basal levels. When GnRH pulses were restored after 3 days, there was a large surge in plasma LH, progesterone levels were partially restored (50%), and corpus luteum regression occurred at the normal time (330). Thus the loss of progesterone secretion alone did not promote regression because the corpus luteum recovered at least partially. When trilosane, an inhibitor of steroid biosynthesis, was given to monkeys for 2 days during mid cycle, progesterone fell to basal levels and did not subsequently recover (168). The different results of these two studies (partial recovery versus nonrecovery) makes it difficult to assess the role of progesterone in maintaining luteal function in primates. It has been suggested that there may be an age-dependent decrease in the responsiveness of the primate corpus luteum to LH so that the corpus luteum of the late luteal phase is unable to maintain its secretory function, with rescue of the corpus luteum being achieved by CG that may overcome the diminished responsiveness of the aging gland to LH (746). Such a loss in sensitivity to LH could still be due to the intraluteal generation of endogenous luteolysins that may be overcome by CG during the establishment of pregnancy.

Because the corpus luteum is resistant to the luteolytic effect of PG analogs in early pregnant monkeys, i.e., when CG is present, it was suggested that one function of CG in primates is to override or block the action of potential luteolysins such as PGF(464). Moreover, it has been demonstrated that a regimen of hCG administered systemically to rhesus monkeys during the midluteal phase prevents the induction of corpus luteum regression by an intraluteal infusion of PGF (34). However, rhesus monkey luteal tissue previously exposed to PGFin vivo exhibited a decrease in progesterone production when treated later with hCG in vitro (35). It has been suggested that the human corpus luteum may be rescued during early pregnancy through the inhibition of apoptic cell death (621). It has also been proposed that an increase in luteotropic PG, such as PGE2, PGI2, and PGD2 within the corpus luteum, may play a role in the rescue of the corpus luteum during early pregnancy in women (282). The inhibitory effect of PGF in human luteal function in vitro can be overcome by supramaximal concentrations of hCG (283). Whether the antiluteolytic effect of hCG is mediated by the local production of luteotropic prostaglandin remains to be established. There is evidence that PGF activates the phosphatidylinositol pathway in the corpus luteum of primates, whereas hCG neither activates this pathway nor alters its activation by PGF (534). Collectively, these findings support the view that CG may rescue the corpus luteum in early pregnancy, not by blocking the ovarian synthesis of a luteolysin, but rather via a luteotropic stimulus that may counteract the luteolytic effect of potential luteolysins such as PGF. A more detailed description of luteolysis and the rescue of the corpus luteum in women and nonhuman primates is provided in several recent reviews (282, 301, 545, 746).

VII.  UNIFYING HYPOTHESIS FOR LUTEOLYSIS

Although it is well established that PGF is the common mediator of luteolysis in most nonprimate species, current evidence indicates that the brain (hypothalamic-pituitary system) is not only responsible for initiating the ovarian cycle via the interaction of ovarian steroids and pituitary gonadotropins (FSH and LH) but also is responsible for terminating the ovarian cycle via the posterior pituitary neuropeptide oxytocin. Thus, in nonprimate species, the uterus can be regarded as a transducer that converts neural signals in the form of episodes of oxytocin secretion from the hypothalamic-neurohypophysial system into luteolytic episodes of PGF secretion from the endometrium. Luteolytic levels of uterine PGF reach the ovary, either via local countercurrent transfer from uterus to ovary, or via the systemic circulation. The removal of the uterus (“the transducer”) effectively interrupts the process of luteolysis and extends the life span of the corpus luteum. The models shown in Figure33 represent comparative hypothetical models for the regulation of luteolysis in the major groups of cyclic mammals that are considered to exhibit regular cyclical luteolysis.

Fig. 33.

Comparative hypothetical models for regulation of luteolysis in major groups of mammals that exhibit regular cyclical regression of corpus luteum (see text for explanation). OTR, oxytocin receptor; OPG, oxytocin pulse generator.

A.  Ruminants

The uterus can be regarded as a transducer that converts neurohypophysial oxytocin signals, generated by the central oxytocin pulse generator, into pulses of PGF which then causes luteolysis. However, in ruminants, a finite store of oxytocin in the corpus luteum may act to supplement oxytocin signals from the central oxytocin pulse generator and hence amplify PGF pulses from the uterus. It remains to be determined whether such an amplification of PGF by luteal oxytocin is a necessary requirement for luteolysis in ruminants.

B.  Nonruminants

Because the corpora lutea of most nonruminants contain very little oxytocin, it is likely that these species rely mainly on the neurohypophysis as a source of oxytocin, with the central pulse generator signals of oxytocin being transduced by the uterus into luteolytic PGF pulses without amplification by supplemental releases of luteal oxytocin. However, it is possible that oxytocin produced locally by the uterus in the sow and mare may serve as a supplemental source of oxytocin during luteolysis in these species.

C.  Primates

Because hysterectomy is without effect on ovarian cyclicity, and because oxytocin and oxytocin receptors have been identified in the primate corpus luteum and/or ovary, oxytocin from the ovary and or a central oxytocin pulse generator could act on ovarian oxytocin receptors to generate low levels of intraovarian PGFthat may then cause luteolysis. The transducing function of the uterus would thus be bypassed in this paradigm. However, when the uterus is present, as in the intact subject, the withdrawal of progesterone action in the uterus due to intraovarian luteolysis acts as a major stimulus for endometrial PGF synthesis and hence the initiation of the vascular and contractile events associated with menstruation.

In the foregoing review, considerable information has been cited to support the model illustrated for the control of luteolysis in ruminants. A question that presently remains unanswered is whether the supplemental secretion of luteal oxytocin is an absolute requirement for luteolysis in ruminants. The evolutionary advantage of the amplification of neural signals (oxytocin pulse generator) by supplemental releases of luteal oxytocin is presently unclear. However, it is possible that such an amplification has evolved to ensure that the process of luteolysis occurs efficiently and in a time frame that optimizes the neuroendocrine events later in the cycle, e.g., the synchronization of estrous behavior and ovulation to achieve optimum conception. Indeed, manipulation of the rate of decline in plasma progesterone during luteolysis in sheep is reported to influence the cellular composition of the ovulatory follicle and also the length of the subsequent cycle (718). In nonruminants, such as the pig and horse, in which PGF is also secreted in a pulsatile manner, it appears that the central oxytocin pulse generator may play a dominant role in luteolysis, since these species have little ovarian oxytocin but show an increase in PGF in response to exogenous oxytocin (101, 256). Moreover, pulses of oxytocin in peripheral blood have been associated with pulses of uterine PGF in the horse (691) and the pig (384, 489). In the pig, mRNA levels for oxytocin precursor are elevated in the endometrium late in the luteal phase (678), and immunoreactive oxytocin has been detected in the uterine lumen (490). Oxytocin mRNA and immunoreactive peptide have also been detected in the endometrium of the mare during the estrous cycle (53, 707). The relative contributions of the neurohypophysis and the endometrium to oxytocin blood levels in the pig and the mare are presently unknown, but the endometrium could act as a supplemental source of oxytocin in these species during luteolysis. In the rat, mRNA for oxytocin and the peptide itself increase in the endometrium during the course of pregnancy, and it is proposed that the endometrium is the primary source of oxytocin during labor in this species (407). However, oxytocin was not found in the endometrium of nonpregnant cyclic rats (407).

Although PGF is released in a pulsatile manner in the guinea pig during luteolysis (183), there is little information on the role of oxytocin in luteolysis in this species, although the ovary does not appear to be a major source (699,700). However, there is some disagreement as to whether oxytocin stimulates endometrial PGF synthesis in this species (404, 579), and an earlier study in the guinea pig indicated that exogenous oxytocin did not have any marked effect on the length of the cycle (159). Toward the time of parturition in the guinea pig, there is an increase in oxytocin receptor binding (11), and mRNA levels for the oxytocin receptor in the uterus are also elevated at this time (R. Ivell, personal communication). Information on the concentration of oxytocin receptors in the guinea pig uterus during the estrous cycle has not been reported. It is possible that the synthesis of endometrial PGF in the guinea pig is controlled directly by estrogen and progesterone (507) and in this respect may resemble regulation of endometrial PGF synthesis in the primate (181, 182).

The major components controlling luteolysis in the nonprimate species, namely, oxytocin, oxytocin receptors, PGF, and PGF receptors, have now been identified in the primate ovary, thus providing a plausible model for the current view that luteolysis in primates is most likely an intraovarian process operating independently of the uterus. Thus it is possible that a paracrine interaction of estradiol, oxytocin, and PGF may occur within the primate ovary to promote luteolysis (433). It has recently been reported that a metabolite of progesterone, namely, 5β-dihydroprogesterone, can inactivate the oxytocin receptor in the human myometrium via a nongenomic conformational change in the receptor (266). Thus nongenomic regulation of the oxytocin receptor in the primate ovary by a progesterone metabolite might regulate the action of oxytocin in the ovary and/or corpus luteum of the primate and hence play a role in mediating lutolysis. A central oxytocin pulse generator has not been identified in primates. However, estrogen-stimulated neurophysin and oxytocin have been observed to increase in both monkeys and women around the time of ovulation and, to a lesser extent, toward the end of the luteal phase (17, 18, 585,624). Further studies will be required to determine what role, if any, central oxytocin plays in mediating luteolysis in primates. Although the primate endometrium is an abundant source of prostaglandins that may play a role in the vascular events associated with menstruation, the production of PG in the endometrium is not a requirement for luteolysis that occurs in the absence of the uterus. The withdrawal of progesterone as a result of intraovarian luteolysis appears to be the principal stimulus for the increase in endometrial PG synthesis at the time of menstruation in old world primates including women (see Refs. 181, 182). Thus, in nonprimates, uterine production of PGF initiates luteolysis, whereas in primates, intraovarian luteolysis and the withdrawal of progesterone precedes the synthesis of endometrial progesterone associated with menstruation.

VIII.  CONCLUDING REMARKS

Since the original observation on the influence of the uterus on the cyclical regression of the corpus luteum by Leo Loeb in 1923 (422), it took 50 years to establish the identity of PGF as a uterine luteolytic hormone (459). One might ask the question, What has been achieved since and what remains to be accomplished? As described in this review, the role of PGF as a luteolytic hormone has been amply confirmed in mammalian species that depend on the presence of the uterus for corpus luteum regression. Such a discovery proved to be a major breakthrough in the livestock industry where PGF and its synthetic analogs are now routinely used to regulate the breeding of domesticated animals. For example, PGF is used to synchronize estrus for more effective artificial insemination procedures as well as for the now routine practice of embryo transfer [for a review on practical uses of PGF, see McCracken and Schramm (468)]. Basic mechanisms regulating uterine PGF secretion in nonprimates have been established, although it is apparent that considerable species variation exists. It is now clear that estradiol-17β and progesterone have multiple roles in controlling the luteolytic process, not only by regulating the enzymes necessary for the endometrial biosynthesis of PGF, but also by controling endometrial receptors for oxytocin in a number of species. Moreover, evidence has accumulated that ovarian steroids also modulate the synthesis and secretion of oxytocin from the hypothalamic/posterior pituitary system, which explains, at least in part, the pulsatile nature of uterine PGF secretion. However, further investigations will be required to establish the relative importance to the process of luteolysis of oxytocin produced in the corpus luteum of ruminants and in the endometrium of nonruminant species.

The recent elegant knockout studies of the PGF receptor (660) and of the PLA2 enzyme (75,686) in mice have demonstrated convincingly the requirement for PGF-mediated luteolysis before labor in this species. However, it has to be borne in mind that there are major species differences in the physiology of reproduction so that one has to be cautious in direct extrapolation to other species. Nonetheless, it is likely that genetic models will continue to shed light on the complex mechanisms involved in the regulation of luteolysis in different mammals. An explanation for the proposed intraovarian mechanism for luteolysis in primates has remained elusive. However, new information of various potential regulatory factors in the primate ovary continue to emerge. For example, the recent observation that uterine oxytocin receptors in humans are subject to nongenomic conformational control by 5β-dihydroprogesterone (266) suggests a potential regulatory mechanism for the control of oxytocin receptors within the primate ovary. In this regard, studies on ovarian 5β-reductase activity in the primate ovary may prove to be a fruitful avenue of investigation. Moreover, in primates, androgen receptors have been detected in the corpus luteum, primarily in the vascular elements, but the role of androgens in corpus luteum function in primates is still largely unexplored (657).

The precise biochemical mechanism for the luteolytic action of PGF in the corpus luteum remains unsolved. However, present evidence indicates that PGF acts on plasma membrane receptors in the luteal cells to abrogate the steroidogenic action of LH as well as via the disruption of multiple intracellular mediators downstream of the LH receptor. Loss of luteal cell function may lead to apoptotic cell death which, in turn, signals phagocytosis of the cellular components leading to the structural involution of the gland. Further basic studies on the growth and regression of the corpus luteum are clearly indicated, not only for understanding basic reproductive processes, but also because the mechanisms controlling the function of this transient structure may have important implications for understanding the basis of controlled and uncontrolled growth and regression of other tissues. It is even more apt today that molecular biology is revealing an extraordinary array of new peptides. When the molecular approach runs out, it will take the concerted effort of many bright minds in the discipline of physiology to determine their biological function, if any (519).

Acknowledgments

We are grateful to Dr. Walter Elger and Dr. Wendell Leavitt for providing Figures 14 and 20, respectively. We thank Merrilyn G. Nay for typing the manuscript and Judy M. Nordberg for reference tracing.

The work in the authors’ laboratory was supported by National Institutes of Health Grants HD-08129 and HD-20290, by United States Department of Agriculture Grant 98-35203-6635, and by general support funds from the Worcester Foundation.

REFERENCES