This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Challis, J. R.G. Articles by Lye, S. J. Search for Related Content PubMed PubMed Citation Articles by Challis, J. R.G. Articles by Lye, S. J.

Endocrine and Paracrine Regulation of Birth at Term and Preterm¹

Departments of Physiology (J.R.G.C., S.G.M., W.G., S.J.L.) and of Obstetrics and Gynaecology (J.R.G.C., S.G.M., S.J.L.), University of Toronto, Toronto, Ontario, Canada M55 1A8; Program in Development and Fetal Health (J.R.G.C., S.J.L.), Samuel Lunenfeld Research Institute, Mount Sinai Hospital, Toronto, Ontario, Canada M5G 1X5; MRC Group in Fetal and Neonatal Health and Development (J.R.G.C., S.J.L.); Department of Obstetrics and Gynaecology, and Cellular and Molecular Medicine (W.G.), University of Ottawa, Ottawa, Ontario, Canada K1H 8L6

	Abstract

Top Abstract I. Introduction II. Regulation of Myometrial... III. Pregnancy: Phase 0... IV. Myometrial Activation: Phase... V. Myometrial Stimulation: Phase... VI. Application to Clinical... References

 
We have examined factors concerned with the maintenance of uterine quiescence during pregnancy and the onset of uterine activity at term in an animal model, the sheep, and in primate species. We suggest that in both species the fetus exerts a critical role in the processes leading to birth, and that activation of the fetal hypothalamic-pituitary-adrenal axis is a central mechanism by which the fetal influence on gestation length is exerted. Increased cortisol output from the fetal adrenal gland is a common characteristic across animal species. In primates, there is, in addition, increased output of estrogen precursor from the adrenal in late gestation. The end result, however, in primates and in sheep is similar: an increase in estrogen production from the placenta and intrauterine tissues. We have revised the pathway by which endocrine events associated with parturition in the sheep come about and suggest that fetal cortisol directly affects placental PGHS expression. In human pregnancy we suggest that cortisol increases PGHS expression, activity, and PG output in human fetal membranes in a similar manner. Simultaneously, cortisol contributes to decreases in PG metabolism and to a feed-forward loop involving elevation of CRH production from intrauterine tissues. In human pregnancy, there is no systemic withdrawal of progesterone in late gestation. We have argued that high circulating progesterone concentrations are required to effect regionalization of uterine activity, with predominantly relaxation in the lower uterine segment, allowing contractions in the fundal region to precipitate delivery. This new information, arising from basic and clinical studies, should further the development of new methods of diagnosing the patient at risk of preterm labor, and the use of scientifically based strategies specifically for the management of this condition, which will improve the health of the newborn.

I. Introduction

II. Regulation of Myometrial Contractions

III. Pregnancy: Phase 0 of Parturition

IV. Myometrial Activation: Phase 1 of Parturition

A. Activation: role of fetal hypothalamic-pituitary-adrenal (HPA) maturation

B. Activation mechanism by which cortisol changes placental steroid and PG synthesis

C. HPA function in the primate fetus and activation of parturition

D. HPA maturation in the primate fetus

E. Placental progesterone and human pregnancy: the enigma of the progesterone block

V. Myometrial Stimulation: Phase 2 of Parturition

A. Stimulation: role of oxytocin

B. Stimulation: role of PGs

C. Stimulation: role of CRH

VI. Application to Clinical Preterm Labor

	I. Introduction

Top Abstract I. Introduction II. Regulation of Myometrial... III. Pregnancy: Phase 0... IV. Myometrial Activation: Phase... V. Myometrial Stimulation: Phase... VI. Application to Clinical... References

 
PARTURITION is the process by which the fetus is expelled from the uterus to the extrauterine environment. Parturition results from a complex interplay of maternal and fetal factors. It requires that the uterus, which has been maintained in a relative state of quiescence during pregnancy, develops coordinated contractility and that the cervix dilates in a manner that allows passage of the fetus through the birth canal. To be successful, parturition requires also that maturation of those fetal organ systems necessary for extrauterine survival has occurred, and that the maternal organism has undergone the changes necessary for lactation in the postpartum period. It is not surprising, therefore, that synchronous maturation of the fetus and stimulus to increased uterine activity should be desirable, and much evidence suggests that it is the fetus itself that triggers both these series of events.

Preterm birth, where there is asynchrony between the labor process and fetal maturation, occurs in 8–10% of all pregnancies, and its incidence has changed little in the past 40 yr (1). Indeed, factors such as low socioeconomic status of some inner-city populations, the tendency for women to choose to start a family at an older age, and the impact of fertility treatment are contributing to an increase in the incidence of preterm delivery (2, 3). Improved neonatal care, however, continues to reduce the mortality rate due to prematurity, although preterm birth remains the primary cause of neonatal death. In North America, the cost of caring for infants in the neonatal intensive care nursery during the first months of life has been estimated at $5–6 billion annually (3). That figure does not take into account the extraordinary emotional stress to the family of the prematurely delivered infant. Nor does it take into account the long-term costs required for chronic care of these infants, some of whom have major motor and/or mental handicaps and/or long-term neuro-developmental complications. To prevent preterm birth effectively, we need to understand the fundamental processes that switch the myometrium from its relative quiescence during pregnancy to the activated and contractile state at the time of labor. We will develop the thesis that regulation of myometrial function requires both endocrine and mechanical controls. Furthermore, it is now evident that the cause of preterm labor may vary at different times during pregnancy and will not necessarily reflect acceleration of the processes at term gestation. The ability to recognize these various causes of premature delivery, in a clinical setting, and then provide appropriate treatment remains a major clinical challenge. Furthermore, it is evident that prevention of preterm delivery may not always be desirable, particularly if the fetus is allowed to develop in a hostile intrauterine environment.

Causes of preterm birth in general fall into three categories: iatrogenic, where there is demonstrable complication of pregnancy such as preeclampsia or fetal distress that requires obstetrical intervention; premature rupture of the fetal membranes with or without infection; and, idiopathic preterm labor. The relative importance of these causes varies. However, most sources consider that approximately 30–40% of preterm birth is associated with an underlying infective process, and 40–50% of preterm births are idiopathic.

In this review, we will focus attention on experimental studies in the sheep, the species of choice for many investigators concerned with understanding the processes of birth (4). We shall then extrapolate from the sheep to an understanding of parturition in primates, particularly in the human. Our central thesis is that the processes of birth are remarkably similar, at a fundamental level, across species, and in both sheep and human the fetus, through activation of its hypothalamic-pituitary-adrenal (HPA) axis, plays a central and crucial role. We shall examine how the fetal HPA axis may be activated in response to a stress circumstance during pregnancy, e.g., hypoxemia, such as that perhaps associated with reduced uteroplacental perfusion in preeclampsia. It will be apparent that the fetal signal provokes increased outputs of stimulatory PGs and other uterotonins from intrauterine tissues. It is evident now that there is a progression from fetal to maternal control of intrauterine PG production. Furthermore, the regulation of PG synthesis and metabolism in fetal trophoblasts and maternal uterus is effected by different mechanisms.

	II. Regulation of Myometrial Contractions

Top Abstract I. Introduction II. Regulation of Myometrial... III. Pregnancy: Phase 0... IV. Myometrial Activation: Phase... V. Myometrial Stimulation: Phase... VI. Application to Clinical... References

 
During pregnancy, myometrial activity is characterized by poorly coordinated contractures, or the Braxton-Hicks contractions of human gestation (5). In late pregnancy, the uterus undergoes preparedness for the stimuli that lead to contractility and labor (6, 7). Those stimuli may be local, maternal, mechanical, or fetal (8). The contracture pattern of uterine activity has been observed in several species, including the sheep, baboon, and rhesus monkey (9). The development of coordinated uterine contractions at term results in a myometrium that is excitable, generating high-frequency, high-amplitude contractions. It is spontaneously active and responds to exogenous uterotonins. The transition of the myometrium from a quiescent to an active state has been termed "activation." When this has occurred the myometrium can then undergo "stimulation" in response to endogenous and/or exogenous agonists (8).

We have found it useful to divide the uterine phenotype into different stages of the parturition process (10). The uterus is relatively quiescent during 95% of pregnancy, corresponding to phase 0 of parturition. Activation corresponds to phase 1 and is effected predominantly by mechanical input, and through regulation by uterotrophins such as estrogen. Stimulation corresponds to phase 2, when endogenous uterotonins, including PGs and oxytocin (OT), act on the activated myometrium. Postpartum involution corresponds to phase 3. In this sequence of events, the "initiation" of parturition corresponds to the transition from phase 0 to phase 1, although clearly one could argue that initiation started much earlier in gestation (11).

Contraction of the myometrium at term or preterm depends upon conformational changes in the actin and myosin molecules, which allow actin and myosin filaments to slide over each other, ultimately leading to a shortening of the myocyte (Fig. 1 and Refs. 12, 13). The confirmational changes (involving cross-bridge cycling of the myosin head) require ATP, which is generated by myosin after phosphorylation of the 20-kDa light chains of myosin by the enzyme myosin light chain kinase (MLCK). This enzyme is central to signaling pathways that both stimulate and inhibit myometrial contractions (14, 15). MLCK is activated through interaction with the calcium binding protein calmodulin (CAM), which in turn requires 4 Ca²⁺ ions for its own activation. Binding of calcium-activated CAM to MLCK induces a conformational change in the enzyme, allowing MLCK to phosphorylate the 20-kDa light chains of myosin. MLCK can also undergo phosphorylation by protein kinase A (PKA, cAMP-activated protein kinase), which reduces the affinity of the enzyme for calcium calmodulin (Ca-CAM) and leads to its inactivation (14, 16). Regulation of MLCK has been reviewed extensively (17, 18). It is evident that activity of this enzyme is altered by intracellular pathways that regulate levels of calcium and of cAMP and is critical for the development of uterine contractility. Uterotonins generally increase intracellular calcium levels ([Ca²⁺]_i), by increased influx of Ca²⁺ through receptor-operated channels, or release of calcium from intracellular stores including sarcoplasmic reticulum (see Ref. 19). Agents that inhibit myometrial activity do so by increasing intracellular levels of cyclic nucleotides cAMP or cGMP, which in turn inhibit release of calcium from intracellular stores or reduce MLCK activity. Binding of agents such as ß-adrenergic agonists, relaxin and prostacyclin, to myometrial receptors activates adenylate cyclase activity, leading to an increase in cAMP generation, while uterine inhibitors such as nitric oxide (NO) activate guanyl cyclase, increasing cGMP. In collaborative studies, Pato et al. (20) characterized MLCK purified from pregnant sheep myometrium. The enzyme had an apparent molecular mass of 160 kDa and high substrate specificity for myosin light chains. Sheep myometrial MLCK has an absolute requirement for Ca²⁺ and CAM for activation; in the absence of Ca-CAM, MLCK is inactive. On binding Ca-CAM, MLCK undergoes a conformational change that exposes the catalytic site, which can then phosphorylate the 20-kDa myosin light chains to initiate contraction. Relaxation is achieved either by dephosphorylation of MLC-20 by the catalytic subunits of type 2A phosphatase (21) or by reduction in MLCK activity. The latter is achieved, as discussed, by reduction in [Ca²⁺]_i, resulting in dissociation of Ca-CAM from MLCK. Sheep myometrial MLCK is also a substrate for PKA, which phosphorylates serine residues on the sheep myometrial enzyme in the presence or absence of bound Ca-CAM. The ability of PKA to inhibit myometrial MLCK activity, even in the presence of agonists that increase [Ca²⁺]_i, provides a biochemical rationale for the finding that agents that increase intracellular cAMP inhibit uterine contractions even in the presence of calcium-activating agents such as OT and stimulatory PG. Ca-CAM can also activate phosphodiesterase to increase the breakdown of cAMP.

View larger version (20K): [in this window] [in a new window]   Figure 1. Cartoon of a myometrial cell indicating the intracellular biochemical pathways involved in regulating contractions. MLCK is central to uterine contractility. It is activated by Ca-CAM after an increase in intracellular calcium levels. This increase is generated by the action of various uterotonins: PGF acting through PGF receptor (FP), OT acting through OTR. Agents that increase cAMP (ß-agonists) or cyclic GMP, or NO donors decrease uterine contractility. AA, Arachidonic acid; Atosiban OTR antagonist.

Regulation of myometrial calcium levels has been reviewed extensively (see Refs. 12, 22, 23, 24). Free resting Ca²⁺ increases from 150 nM to about 500 nM during contraction through influx of extracellular Ca²⁺ or by the release of Ca²⁺ from intracellular binding sites or intracellular organelles (25, 26). Extracellular Ca²⁺ enters cells through receptor-operated or voltage-gated channels. Release of intracellular Ca²⁺ from sarcoplasmic reticulum is activated through the phosphoinositol (PI) pathway. Binding of a uterotonin to its plasma membrane receptor activates a G protein transducer, coupled to phospholipase C, which frees inositol trisphosphate (IP3) and diacylglycerol (27, 28). Free IPs, especially IP3, increase cellular calcium from intracellular storage sites. Interestingly, IP3 binding in myometrium was inhibited by calcium, suggesting that this might provide a mechanism for regulating the IP3 response by oscillating [Ca²⁺]_i. Diacylglycerol formed during IP3 turnover may stimulate PKC to phosphorylate cellular proteins such as MLCK or be rapidly phosphorylated by diacylglycerol kinase to phosphatidic acid, a naturally occurring Ca²⁺ ionophore, or lead to release of arachidonic acid by cellular lipases, resulting in production of eicosanoids (see below).

Function of the myometrium during labor at term or preterm requires highly developed cell-to-cell coupling, effected through formation of intercellular GAP junctions within adjacent cell membranes (14, 29). The proteins forming GAP junctions are termed connexins and are classified according to their apparent molecular weights (30). Connexins are arranged into hexameric hemichannels, which become aligned across adjacent cells to form an interconnecting pore that allows low-resistance electrical or ionic coupling between the cells and provides a pathway for metabolite transfer (31). Hundreds of individual channels arrange themselves into an organized plaque to form a GAP junction. Regulation of connexins occurs at the level of transcription and translation (31, 32); mechanisms also operate to control transport of connexin protein to the cell membrane and to direct assembly into connexons, through apposition, clustering, and formation of functional channels (33, 34). This complex process is poorly understood, although it is influenced by steroids and by mechanical stretch (35). GAP junction formation requires the presence of cell adhesion molecules, and in early studies, Meyer et al. (36) showed that appearance of GAP junctions in transfected S180 cells was blocked by coincubation with antisera to liver cell adhesion molecule.

Garfield (see Refs. 14, 16) established clearly that an absence of GAP junctions in the pregnant myometrium was responsible for high-input resistance of these smooth muscle cells and poor coordination of uterine contractions. There is a massive increase in numbers of GAP junctions with the onset of labor, which significantly enhances electrical coupling and allows the myometrium to develop synchronous high amplitude contractions (37). An increase in GAP junctions with labor onset has been found in all species studied. In the rat, levels of connexin-43 (CX-43) mRNA and protein were low during pregnancy but increased some 48 h before labor (38, 39). Highest levels of mRNA and protein were found during delivery itself. This is critical because the half-life of GAP junctions may be as short as 1–2 h, and hence continued synthesis would be required to maintain labor. Increases in CX-43 mRNA have been reported in sheep and human myometrium with the onset of labor and correlated with increases in CX-43 protein (37, 38). Permeability of GAP junctions may be facilitated through phosphorylation at consensus serine and tyrosine sites within the cytoplasmic domain of CX-43. Garfield (14) demonstrated that cell-to-cell communication in the myometrium is reduced by elevated [Ca²⁺]_i and increased levels of cAMP. Importantly, more recent studies have shown that the pattern of CX-43 in myometrium during pregnancy differs from that of CX-26. Connexin-26 expression is elevated in midgestation in the rat and appears to be associated more with uterine quiescence (7, 8).

	III. Pregnancy: Phase 0 of Parturition

Top Abstract I. Introduction II. Regulation of Myometrial... III. Pregnancy: Phase 0... IV. Myometrial Activation: Phase... V. Myometrial Stimulation: Phase... VI. Application to Clinical... References

 
Studies in different species have indicated that a variety of different inhibitors may play upon the myometrium during pregnancy. Withdrawal of one or more of these may predict the onset of delivery; precocious withdrawal may predict the onset of premature parturition. Such an inhibitor, PTH-related peptide (PTHRP), is produced in myometrium, and its rate of transcription is increased by progesterone and transforming growth factor ß (TGFß) (40). PTHRP receptor mRNA has also been localized to rat myometrial tissue, suggesting that the protein may act in an autocrine/paracrine fashion through specific receptors to activate the G_s subunits of G proteins and increase intracellular levels of cAMP (40, 41, 42).

Relaxin also elevates myometrial cAMP and inhibits OT-induced turnover of phosphoinositide (PI) by the action of cAMP-dependent protein kinase. Relaxin exerts a dual role in the inhibition of myometrial contractility and in the regulation of connective tissue changes in the cervix (43, 44). Porter and colleagues (45, 46) were among the first to show that relaxin suppressed spontaneous uterine contractility in the rat and guinea pig, but sensitivity to OT was preserved. Thus, the major action of relaxin is one of frequency modulation (47). Hansell et al. (48) and others have demonstrated that relaxin is expressed in the human fetal membranes, decidua, and placenta, consistent with its exerting paracrine/autocrine effects on intrauterine tissues (49, 50, 51). Relaxin gene expression is dramatically up-regulated in patients with preterm, premature rupture of membranes (PPROM) (49). Relaxin receptors have been localized to decidua and chorionic trophoblast cells, and the protein acts through these to up-regulate expression of matrix metallo-proteinases (MMP), especially MMP1, MMP3, and MMP9. Similarly, relaxin increases MMP expression in cervical tissue at term. Administration of exogenous relaxin stimulates separation of the pubic symphysis in those species in which it is a prerequisite for delivery (52). In addition, in pigs and rats, relaxin appears necessary for maintaining evolution of spontaneous uterine contractility in late pregnancy and for maintaining a high frequency of live births (43). In vitro studies have shown that relaxin blocks the action of stimulants such as OT, carbachol, and norepinephrine on the myometrium, through mechanisms involving PKA-mediated phosphorylation of PLC-linked G proteins. This in turn inhibits IP3 turnover and the increase in [Ca²⁺]_i (22). Although the precise role that relaxin plays during pregnancy remains to be determined, it may be particularly useful in maintaining uterine quiescence during the period when progesterone concentrations are falling and estrogen levels are beginning to increase before the onset of labor (see Ref. 12). In addition, there are reports that relaxin may act centrally to increase circulating plasma OT and vasopressin concentrations by an opioid-independent mechanism (53). It is now known that OT is produced within human intrauterine, choriodecidual tissues. It remains to be established whether a similar relationship exists between relaxin and OT synthesized within the intrauterine compartment in women.

Lye and Challis (54, 55) first showed, some 20 yr ago, that prostacyclin infused into nonpregnant sheep inhibited uterine contractility in vivo. In parallel studies a similar inhibitory effect of prostacyclin was observed on human myometrium (56), and it is clear now that prostacyclin represents the major eicosanoid present within the pregnant myometrium of many species (57), including human. In human term pregnant myometrial strips maintained in vitro, the initial response to PGI2 was contraction, but this was followed by relaxation (58, 59). It is now recognized that PGI2 acts through specific IP receptor species to increase adenylate cyclase activity and elevate intracellular cAMP (60). Other agents such as CRH also stimulate output of cAMP from myometrial cells and act synergistically with PGI2 in a paracrine/autocrine fashion (61). The role of CRH in pregnancy maintenance and parturition will be discussed later in this review.

More recently, interest has arisen over the potential role of NO as an endogenous inhibitor of myometrial contractility (62). Increases in endogenous synthesis of NO by administration of the NO precursor L-arginine, or the NO donor sodium nitroprusside, inhibit myometrial contractions in the rat and human (62). Nitroprusside has been shown to decrease force and 20-kDa myosin light chain phosphorylation in human myometrial strips, although the tissue is not as sensitive as vascular smooth muscle. Nitric oxide synthase (NOS) isoforms have been detected using RT-PCR in human fetal membranes and choriodecidua (62). Levels of mRNA-encoding inducible NOS (iNOS) are highest in human myometrium at preterm, not in labor patients, and decrease with a corresponding fall in iNOS protein in myometrium collected at term (see Ref. 10). Several authors have suggested that NO acts in a paracrine manner, potentially in conjunction with progesterone to effect myometrial quiescence during pregnancy, although this position has been disputed. There is a decrease in NOS activity of decidua and myometrium in species such as rat before parturition in a manner that would presumably diminish its inhibitory influence on the uterus. Furthermore, studies by Chwalisz and Garfield (62) have shown that, at term in the rat, there is a corresponding increase in NO production by inflammatory cells of the cervix, indicating a role for NO in cervical effacement and relaxation as its influence on the myometrium is diminished.

Other inhibitors of uterine activity include calcitonin gene-related peptide (CGRP), vasoactive intestinal polypeptide (VIP), and endogenous ß-adrenergic agonists (63). These compounds act through increasing intracellular cAMP and/or decreasing intracellular calcium availability (64).

	IV. Myometrial Activation: Phase 1 of Parturition

Top Abstract I. Introduction II. Regulation of Myometrial... III. Pregnancy: Phase 0... IV. Myometrial Activation: Phase... V. Myometrial Stimulation: Phase... VI. Application to Clinical... References

 
The switch from myometrial quiescence to myometrial activation is essential to enable the muscle to respond to the stimulation provided by the high levels of uterotonic agonists and to generate the synchronous, high-amplitude, high-frequency contractions of labor. We have proposed that myometrial activation results from coordinated expression of a cassette of proteins, termed contraction-association proteins, or CAPs (12). CAPs include ion channels [which determine the resting membrane potential and hence excitability of myocytes (65)], agonist receptors [e.g., to OT and PG (60)] and GAP junctions [permitting cell-to-cell coupling (16)].

Overall regulation of myometrial activity is genetically regulated (Fig. 2). Different species have gestations of varying lengths, and studies involving embryo transfer suggest that it is the genotype of the fetus that controls the length of pregnancy. For example, when sheep embryos from short gestation or long gestation breeds were implanted into random gestation-age recipients, parturition occurred at the appropriate time for the fetal rather than maternal genotype (66, 67). There is a variety of mechanisms by which the fetal genotype can influence pregnancy length, and we have proposed that it includes both endocrine and mechanical signals. In initial studies, Ou and Lye (68) found, using unilaterally pregnant rats, that while expression of CAP genes, CX-43 and OT receptor (OTR), increased in the gravid uterine horn in labor, there was no parallel increase in the nongravid horn, even though both horns were exposed to the same systemic hormonal changes. Next, these workers showed that when a small 3-mm diameter tube was placed into one uterine horn of bilaterally ovariectomized nonpregnant animals, there was a significant increase in mRNA levels encoding CX-43 in that horn, compared with the contralateral horn. Control experiments showed that this result was not due to the presence of a foreign body within the uterus. Administration of progesterone to these animals blocked stretch-induced increases in CX-43 expression.

View larger version (15K): [in this window] [in a new window]   Figure 2. The onset of labor is dictated by the fetal genome proceeding through either a fetal growth pathway with increases in uterine stretch or fetal endocrine pathway involving activation of the fetal HPA axis. These two arms are not independent because changes in progesterone and estrogen modulate the ability of uterine stretch to increase expressions of genes associated with myometrial activation.

2

The molecular mechanisms by which stretch increases CX-43 and OTR expression remain to be determined (69). In other systems, such as cardiac myocytes, stretch activates multiple intracellular signaling pathways through shear stress response elements in the promoter of some stretch-responsive genes (70). The CX-43 gene contains such an element, suggesting that if wall tension contributes to the regulation of CAP genes in the myometrium, regulation of uterine growth through pregnancy will be important in determining the level of shear stress. Lye and colleagues (8) have argued that, during pregnancy, uterine growth follows three distinct phases: an initial phase during the first trimester where uterine growth is due to hyperplasia and controlled by endocrine factors, a second phase during the second and third trimester in which growth is closely matched to increased fetal size, and a final phase in which there is a decline in uterine growth in comparison to fetal growth, and hence an increase in uterine wall stretch and tension. They speculate that progesterone is necessary to support stretch-induced hypertrophy of the uterus during midgestation in concert with increasing fetal size. Near term, the fall in progesterone, observed in most animal species (see below), leads to a decline in uterine growth relative to fetal growth and hence increased tension development, which in turn results in increased CAP gene expression and contributes to myometrial activation. Since the decrease in circulating progesterone appears critical for the altered influence of stretch on myometrial CAP gene expression, we shall consider the endocrine pathways that result in progesterone withdrawal.

A. Activation: role of fetal HPA maturation
More than 35 yr ago, Professors Sir Graham (Mont) Liggins and Geoffrey Thorburn, working in the sheep and goat, showed conclusively in those species that the fetus, in utero, appeared to provide the trigger mechanism for the onset of parturition and that it did so through activation of the fetal HPA axis. An endpoint of activation of this axis is progesterone withdrawal. We shall suggest that the primate fetus similarly affects gestation lengths through activation of the HPA axis. However, in human gestation there is no systemic progesterone withdrawal, and we shall argue that, in women, sustained circulating concentrations of progesterone are indeed required at term to effect regionalization of myometrial contractility and promote relaxation of the lower uterine segment.

Early studies in the fetal sheep showed that ablation of the fetal pituitary gland, the fetal adrenal gland, pituitary stalk section, or lesioning of the fetal paraventricular nucleus (PVN) resulted in prolongation of gestation (71, 72, 73), whereas the infusion to the fetal lamb in utero of ACTH or of a glucocorticoid resulted in premature parturition within 3–5 days of beginning the infusion. These studies provided experimental verification of the concept developed from observations of naturally occurring prolonged gestation in sheep attributable to ingestion of the teratogen Veratrum californicum at a specific time of gestation. In those animals, gestation length was prolonged by up to 60 or 70 days, although fetal growth continued. Fetuses exhibited gross malformations, including cyclopean characteristics. At autopsy, the pituitary and adrenal glands were remarkably hypoplastic as a result of impaired pituitary development at an early gestational age (see Ref. 81).

Several groups of workers provided clear evidence for maturation of fetal HPA function in the sheep fetus during late gestation (74, 75, 76). There are progressive increases in fetal plasma ACTH_1–39 and cortisol in the plasma of the late-gestation fetal sheep (77, 78, 79, 80); the initial increases in ACTH precede the rise in cortisol (79), but fetal cortisol increases in an exponential fashion over the last 10 days of gestation, with highest concentrations being established immediately before term (80). This is consistent with the fact that ACTH is important in the development of the adrenal cortex in late gestation. Similar maturation of pituitary adrenocortical function has been demonstrated in several other species, including the guinea pig, which represents a species that gives birth to mature young. The prepartum surge of cortisol is important for the maturation of several organ systems, particularly the lungs and kidneys (see Ref. 81). It is also critical for normal development of programming of the brain. However, the simultaneous increase in fetal plasma ACTH and cortisol has remained somewhat of a paradox because, under normal circumstances, one would expect elevations in fetal plasma cortisol concentration to inhibit further ACTH secretion. Mechanisms have developed to override the influence of negative feedback in the fetus in late gestation, a relationship now described in the guinea pig as well as in the sheep (see below).

Recent studies have explored the molecular mechanisms underlying changes in fetal pituitary adrenocortical activation in late gestation sheep. Developmental changes in CRH mRNA in the fetal hypothalamic PVN were examined by in situ hybridization (82). By day 60 of gestation, CRH mRNA was detectable in the fetal PVN. There was an increase in CRH mRNA expression by day 120 of gestation and a further substantial up-regulation of CRH gene expression in the last 20 days of pregnancy. This was followed by a decrease in CRH expression in the PVN of the newborn lamb. Throughout development, expression of CRH mRNA appears to be confined to parvocellular fields of the PVN, with no expression detected in magnocellular neurons (82). Recent studies have confirmed that the changes in CRH mRNA are translated to CRH peptide in the fetal hypothalamus, indicating a close association between transcription and translation of the CRH gene during development.

In the fetal pituitary, expression of the ACTH precursor, POMC, is detectable in the inferior region of the pars distalis by day 60 of gestation. Levels of POMC mRNA in the superior and inferior regions of the pars distalis increased with progression of gestation until around day 120, when there was a further increase in expression, peaking at term (83, 84). The increase in POMC expression is combined with a remarkable reorganization of the corticotrophs toward the inferior aspect of the fetal pituitary gland. This pattern of expression was sustained in the newborn lamb. In the fetal pars intermedia, the developmental profile of POMC mRNA was quite different. Relatively high levels were present by day 60 of gestation; these increased between days 60 and 100 and then remained relatively constant for the remainder of gestation. Early controversy concerning changes in expression of POMC mRNA in fetal pituitary tissue appears to result from differences in methodologies. The use of in situ hybridization clearly allows separation of different zones of the fetal pituitary gland, whereas erroneous results may have been obtained through use of Northern blot analysis (85, 86). In a recent carefully conducted study obtaining pituitary tissue from fetuses at specific times in late gestation and during the labor process itself, the lack of negative feedback on POMC mRNA, and the sustained increase in POMC mRNA levels, was clearly demonstrated (87). The change in regional distribution of POMC mRNA in the pars distalis may indicate the transition of fetal-like to adult-like corticotrophs that has been described at this time (see below). Changes in POMC mRNA in the pars distalis are reflected by increased levels of ir-ACTH and by increased immunostaining for ACTH in pituitary corticotrophs (83, 84); at term ir-ACTH-positive cells represent about 15% of the total cell number in the pars distalis.

Arginine vasopressin (AVP) is also an important regulator of fetal pituitary ACTH secretion and is expressed in the fetal PVN relatively early in gestation (88). AVP mRNA is present in the supraoptic nucleus, PVN, and the accessory magnocellular nuclei by day 60 of gestation (82). Differential expression of magnocellular and parvocellular AVP is evident in the PVN by day 80. In magnocellular neurons, AVP mRNA increases with gestational age, whereas parvocellular expression of AVP remains relatively unchanged. Levels of AVP mRNA increase dramatically in both regions of the PVN in the newborn lamb. It is suggested that magnocellular AVP is involved primarily in fetal fluid homeostasis, while parvocellular AVP is important in stimulation of the pituitary corticotroph (84). There is a close correlation between AVP mRNA levels and ir-AVP in the anterior hypothalamus, as there is for CRH. The increase in parvocellular AVP mRNA in the newborn may be associated with the stress of the novel extrauterine environment. Axons containing AVP and OT have been identified in a zone of the pars distalis adjacent to the pars intermedia in fetal sheep. These axons are probably those of magnocellular neurons and may represent a mechanism by which magnocellular AVP and OT directly affect ACTH release in vivo.

CRH and AVP induced dose-dependent increases in ACTH output from ovine fetal pituitary cells in vitro (89); at equimolar concentrations AVP was more potent than CRH. Simultaneous administration of CRH and AVP showed an additive interaction between the neuropeptides (90). Treatment with CRH significantly increases POMC mRNA levels in sheep pituitary cells harvested at day 120 and day 138 of gestation. However, CRH treatment of cells collected from fetuses at term failed to affect POMC synthesis. AVP increased POMC mRNA levels in cells obtained at day 138 of gestation; in pituitary cells from late-gestation fetuses, AVP and CRH are equally potent in the induction of POMC synthesis. Cortisol has little negative feedback effect on basal output of ACTH in these cells but inhibits CRH-stimulated ACTH output and POMC gene expression.

Studies by Lu et al. (91) showed that ovine fetal pituitary membranes expressed CRH receptor activity as early as day 100 of gestation. CRH binding increased to its highest levels at around day 135 (term, 145–150 days) and then decreased progressively through late gestation (92). Recent studies have extended these measurements to show that levels of mRNA encoding fetal pituitary CRH-receptor type I may follow a similar profile (J. C. Rose, personal communication), and this may account for the altered outputs of ACTH in response to CRH stimulation in vivo (see below). Factors regulating CRH receptor expression have been examined in vivo and in vitro. In vitro studies indicated that CRH down-regulated expression of its own receptor and cortisol produced a similar attenuation of binding activity (92).

In vivo studies demonstrated that CRH was more potent than AVP in stimulating ACTH output by pituitary tissue from chronically catheterized fetal sheep in late gestation (93, 94). The response profiles, however, are quite different. AVP induced a transient rise in plasma ACTH while CRH stimulated a more sustained increase (95). Subsequently, it was demonstrated that AVP concentrations are about 5 times those of CRH in the hypophyseal portal circulation of adult sheep (96), and it remains possible that the relative importance of AVP in fetal corticotroph activation in utero may be greater than that of CRH (97). Fetal pituitary responsiveness to CRH increases between day 110 and 125 and then decreases toward term (79). This relative insensitivity of the pars distalis to CRH may reflect the increase in negative feedback influence of rising endogenous cortisol concentrations, or the decrease in CRH binding sites indicated above (79). Simultaneous administration of CRH and AVP results in an ACTH response that is greater than when either neuropeptide is administered independently, and the interaction is synergistic in nature, at least at around day 115 of gestation (95). CRH and AVP affect the corticotrophs through different second messenger systems. CRH exerts this action through up-regulating a G_s-adenylate cyclase-linked membrane receptor and increasing intracellular levels of cAMP (89). AVP acts through V_1b receptors to stimulate PI turnover, stimulating phospholipase C and activating protein kinase C.

POMC is processed through different endoproteases, prohormone convertase 1 (PC-1) and prohormone convertase 2 (PC-2), to yield a spectrum of products. Recent studies have demonstrated that both PC-1 and PC-2 are present in fetal sheep pituitary tissue in late gestation. However, expression of these enzymes does not change with labor, and it seems unlikely that the increase in ACTH output is attributable to altered prohormone convertase activity (87, 98). However, the pattern of POMC-derived peptides from the fetal pituitary does change in the plasma of the fetal lamb in late pregnancy (99). Several groups of investigators have reported that large molecular weight POMC-derived ACTH precursor peptides are present in the circulation (100). The concentrations of these larger molecular weight forms decrease prepartum, whereas those of ACTH_1–39 increase. Because the larger molecular weight peptides may act to antagonize the action of ACTH_1–39 on adrenocortical cells (101, 102, 103), a decrease in their concentration prepartum would presumably facilitate ACTH action and an increase in adrenal glucocorticoid secretion (104). The sources of these peptides may be different (105, 106, 107). Studies in hypothalmo-pituitary-disconnected fetuses have led to the suggestion that the pars intermedia may be a potential source of large molecular weight peptides, whereas the pars distalis is the primary source of ACTH_1–39. In addition, the ovine fetal lung and placenta express POMC mRNA and contain ir-ACTH. It is not clear whether these potential sources of ACTH contribute to circulating ACTH_1–39 in a meaningful manner or whether the peptides have paracrine/autocrine actions within the tissues of origin.

Thus, the temporal relationship between hypothalamic-CRH and pituitary POMC expression is consistent with the simultaneous increase in plasma ACTH and cortisol observed in late gestation (84, 108, 109, 110). Nevertheless, the mechanism by which CRH mRNA and POMC mRNA increase in the presence of high plasma glucocorticoid concentrations is not clear. One possible mechanism is that, in the fetus, glucocorticoid feedback thresholds within the brain and pituitary become modified. This may occur at several levels (Fig. 3). We have reported that glucocorticoids up-regulate expression of corticosteroid binding globulin (CBG) mRNA in the fetal liver, and of circulating CBG, which is the opposite of the response in adult sheep (111, 112). In the fetus, the pattern of CBG glycosylation varies from that in adult animals, but the glycoprotein increases in concentration in the fetal circulation and maintains a relatively constant free cortisol concentration for most of pregnancy (112, 113). Near term, however, the increase in adrenal cortisol output exceeds the CBG binding capacity, resulting in a sudden increase in free cortisol concentration over the final hours before birth (114). It appears that this increase in free cortisol before parturition is a consistent observation across different animal species (115). More recently, we have demonstrated expression of CBG mRNA and the presence of CBG immunoreactive protein in other fetal tissues including the kidney, pancreas, and pituitary (115). CBG mRNA has been localized to fetal pituitary cells by in situ hybridization, and its pattern of distribution appears to differ from that of POMC, with greater abundance in superior regions of the gland. As yet, there are no studies demonstrating colocalization of CBG with ACTH-producing cells in fetal pituitary tissue.

View larger version (32K): [in this window] [in a new window]   Figure 3. Summary of events associated with maturation and development of the HPA axis in the fetal sheep. Increased expression of CRH from the PVN of the hypothalamus drives increased expression of POMC in the anterior pituitary. POMC is processed to ACTH, which drives the adrenal gland. In the fetus normal negative effects of cortisol on the hypothalamus and pituitary are diminished through increases in systemic corticosteroid binding globulin (CBG), pituitary 11ß HSD, and diminished expression of GR in the pituitary and hypothalamus.

A further mechanism by which glucocorticoid feedback could be altered locally is through modification of corticosteroid receptor expression (117). The ovine fetal pituitary expresses type II glucocorticoid receptor (GR) from relatively early in gestation, and the levels of GR mRNA increase toward term (118), consistent with glucocorticoid effects in modulating the switch from fetal to adult corticotroph cell types in the pituitary (106). During the course of labor, there is a dramatic decrease in levels of GR mRNA in the fetal pars distalis, suggesting that the potential for glucocorticoid negative feedback decreases in the pituitary during the course of labor. More important, perhaps, is the demonstration that there are decreases in immunoreactive GR in the hypothalamic PVN near term. These changes were specific to CRH- and AVP-positive parvocellular neurons. More recently, we showed that GR mRNA levels in the PVN of fetal sheep and guinea pigs decrease in late gestation, and in fetal sheep levels of GR mRNA in the hippocampus also fall prepartum. The hippocampus represents a major site of glucocorticoid feedback for HPA function, and there are a number of direct and indirect connections between the limbic system and the PVN. Together these data suggest that a reduction in the potential for glucocorticoid feedback occurs in late gestation in brain structures that are central to glucocorticoid negative feedback action (119).

In addition to classic feedback processes, there are several other mechanisms by which fetal HPA axis activation may occur. Expression of pro-enkephalin mRNA rises to a maximum in the parvocellular PVN of fetal sheep at day 135 of gestation and then decreases in older animals (120). A fall in hypothalamic pro-enkephalin mRNA occurs with intrafetal infusion of cortisol at day 135, suggesting that the prepartum rise in endogenous cortisol may inhibit parvocellular pro-enkephalin synthesis. CRH and met-enkephalin are present in the same secretory granules in rodents, and met-enkephalin inhibits CRH-stimulated ACTH secretion from fetal pituitary cells in vitro. Thus, a decrease in met-enkephalin production may facilitate corticotroph function near term (120). OT has been implicated in the control of ACTH secretion in adult sheep, and OT stimulates ACTH output from the fetal pituitary cells in vitro. OT mRNA is present in both magnocellular and parvocellular fields of the PVN and SON and follows a similar developmental profile to AVP mRNA, raising the possibility that it too may influence fetal pituitary function.

In fetal sheep, the kinetically determined production of cortisol from the adrenal gland increases during the last 20–25 days of gestation (77, 121). In part, this results from the increase in drive to the adrenal from rising levels of ACTH, but, in part, it is attributable to maturation of fetal adrenal function (122). Indeed, in hypophysectomized fetuses treated with a continuous low-level infusion of ACTH, plasma cortisol concentrations increased and parturition occurred at around the normal time, consistent with fetal adrenal maturation as the overriding influence (123).

Ovine fetal adrenal responsiveness changes dramatically during the course of pregnancy (124, 124, 125, 126). Adrenal cells collected from animals at days 50–70 of gestation secrete cortisol in response to ACTH stimulation in amounts similar to or greater than adrenal tissue from term fetuses (127). However, between approximately days 90–110 of pregnancy the adrenal is relatively insensitive to ACTH stimulation (124). It is now clear that this pattern of response is due, in large part, to decreased gene expression of P450_C17 and P450_SCC steroidogenic enzymes in fetal adrenal cortical cells at midgestation (128, 129). The abundance of mRNAs for these enzymes is increased by ACTH administration to the fetus (130, 131). Although 3ß-HSD may be rate limiting to cortisol production in the first half of pregnancy (132), immunoreactive (ir)-3ß-HSD-positive cells are present throughout the zona fasiculata of the fetal adrenal cortex from day 50 until term (133). The midgestational decrease in P450_C17 may result by TGFß inhibiting ACTH-induced stimulation to P450_C17, as demonstrated in vitro in ovine fetal and adult adrenal cells (134). Recent studies have demonstrated that ACTH receptor mRNA is detectable from around day 60 of gestation (135). There is a modest increase through pregnancy and then a substantial increase between days 126–128 and days 140–141 (135). Thus, the low level of basal adrenal responsiveness to ACTH around day 100 of gestation is not due to lack of ACTH receptor expression, but may be attributable, in part, to very low concentrations of ACTH in the fetal circulation at that time (136). The increase in ACTH receptor expression in late gestation would appear to contribute to increased adrenal responsiveness near term. The factors responsible for up-regulating ACTH receptor mRNA abundance are unclear (137). These may include ACTH itself, cortisol, or local intraadrenal interaction with IGF-II and/or decreased influence of TGFß (138, 139, 140).

Both in vivo and in vitro studies have shown that fetal adrenal maturation can be advanced by ACTH_1–24 administration (110, 141, 142, 143). Exogenous ACTH in vivo enhances coupling between ACTH receptor and adenylate cyclase and enhanced capacity for cAMP generation (144, 145, 146). ACTH treatment in vivo also increased expression and activity of P450_C17, P450_C11, P450_C21, and 3ß-HSD (130, 147). The adrenal responds to ACTH early in gestation, although continued trophic input is required to maintain increased levels of gene expression. Interestingly, when ACTH was administered to fetuses in vivo as pulses, rather than as a continuous infusion, it led to a pattern of fetal adrenal steroidogenesis that favored cortisol over corticosterone output (i.e., directed P450_C17 activity). Thus, the pulse pattern of endogenous ACTH secretion in vivo may affect the pattern of adrenal activation (148, 149).

These studies suggest that ACTH-induced increases in adrenal steroidogenic enzymes, particularly P450_C17, is essential to allow C21 steroids to proceed through the 17-hydroxy pathway leading to cortisol biosynthesis (130, 150). An obligatory role for an increase in ACTH drive to the fetal adrenal as a prerequisite for increased responsiveness, however, has been challenged recently. When hypophysectomized fetal sheep were infused at a constant, but low level of ACTH, there was a normal rise in fetal cortisol concentration; later, maternal progesterone levels decreased and birth occurred at about the expected time (123). The molecular mechanisms underlying this fascinating result clearly require elucidation.

We have hypothesized that fetal stress, perhaps reflected in diminished fetal arterial P0₂, constitutes a stimulus for preterm birth. Experimental hypoxemia has been used extensively to investigate fetal HPA activation (151, 152). Many studies have shown that even modest reductions in fetal arterial P0₂ induce robust increases in fetal plasma ACTH and cortisol concentrations (153, 154). Release of CRH and AVP into the hypophysial portal system is abolished in the hypothalamo-pituitary-disconnected (HPD) fetus (152), and these animals are incapable of mounting an ACTH response to stress, implying that increased ACTH output requires hypothalamic input. Studies by Akagi and colleagues (155) demonstrated that changes in fetal P0₂ of only 4–5 mm Hg were adequate to elicit increased ACTH concentrations in the circulation of the fetal lamb. This level of oxygen change is similar to that seen during spontaneous contractures in late gestation sheep, raising the possibility that uterine activity itself may contribute part of the stimulus to increased fetal HPA maturation. Whether chronic stress is a stimulus to birth at term (156) or contributes only to some cases of preterm labor is unclear at the present time.

At 135 days’ gestation, hypoxia (P0₂ reduction by 8 mm Hg) significantly increased CRH mRNA in parvocellular PVN and POMC mRNA in the pars distalis within 6 h. This response, however, was attenuated by concurrent infusion of cortisol, indicating effective glucocorticoid feedback mechanisms in vivo at this time (157). After 48 h of sustained hypoxemia, levels of POMC in the pars distalis were elevated, but expression in the pars intermedia was decreased (158). This suggests differential regulation of these two zones of the fetal pituitary, consistent with observations that dopamine, likely from the fetal arcuate nucleus, tonically inhibits pituitary POMC synthesis, and this inhibition is exacerbated in the presence of hypoxemia. Infusion of bromocriptine, a dopamine D2 receptor agonist at day 130 of gestation, produced a 50% decrease in pars intermedia POMC mRNA levels, without affecting POMC mRNA in the pars distalis (159). Thus, the fetal D2 receptor system is functional in late pregnancy, but the fetal pars intermedia does not appear to secrete ACTH_1–39 in amounts that alter fetal adrenal function.

Activation of fetal HPA function in response to hypoxemia, however, is a critical aspect of the story leading to preterm birth (160, 161). A sustained pulsatile hypoxemic stimulus is adequate to up-regulate HPA gene expression, plasma ACTH, and cortisol concentrations. It is reasonable to predict that sustained hypoxemia in conditions of fetal compromise predisposes to fetal HPA activation and would result in premature birth (162, 163).

B. Activation mechanism by which cortisol changes placental steroid and PG synthesis (Fig. 4)
Fetal cortisol acts on the sheep placenta to alter the pattern of steroidogenesis; as a result, progesterone output falls and estrogen concentrations increase (164, 165, 166, 167). These changes in placental steroid output are associated with increased expression and activity of placental P450_C17 (168, 169). This is a critical difference between the sheep and the human, where this enzyme is not induced in the placenta at term. Ovine placental tissue contains P450_arom activity, and up-regulation of this gene also occurs in late gestation. For many years the general thesis has been that placental estrogen production is limited in ovine pregnancy and occurs in abundance only at term with the induction of placental P450_C17 as a result of glucocorticoid action (170, 171, 172). The fall in progesterone and later increase in maternal and fetal estrogen concentrations have been considered as providing the stimulus to increased PG output by intrauterine tissues, with consequent increase in myometrial contractility (173, 174, 175, 176, 177, 178, 179, 180).

View larger version (20K): [in this window] [in a new window]   Figure 4. Endocrine pathways leading to the onset of parturition in sheep. A, current model; B, proposed sequence of hormone events. In the current model, activation of the fetal HPA axis leads to increased cortisol thought to up-regulate expression of P450C17 in the placenta. In the new proposed hypothesis, increased fetal adrenal output of cortisol results in up-regulation of prostaglandin synthase (PGHS)-2 gene expression in the placenta with increased production of PGE2. PGE2 feeds back to further up-regulate fetal HPA function but is itself responsible for up-regulation of P450C17 gene expression in the placenta. Increased placental estrogen is required for up-regulation of PGHS-2 in maternal tissues but not in fetal tissues. Thus, with the onset of parturition there is progression from fetal trophoblast within the placenta to the maternal uterine tissues.

There are other troubling features of the currently accepted model (185). Several groups of investigators have used either immunohistochemical techniques for localization of PGHS-1/-2, or PGHS-2, or in situ hybridization for PGHS-2 mRNA, or measurements of PGHS and/or PGHS-2 activity in ovine placental cells and microsomal preparations (186, 187, 188, 189), to show that PG production by the sheep placenta increases progressively through the last 20–25 days of gestation (190, 191, 192, 193, 194, 195). Placental output of PGs is not confined to the immediate 24–48 h before spontaneous parturition (196, 197). The increase in PGHS expression and activity in the placenta correlates closely with the progressive increase in plasma PGE₂ concentrations in the circulation of the chronically catheterized fetal lamb (191, 198). The increase in circulating PGE₂ in the fetus bears a striking temporal relationship to the increase in plasma cortisol concentration (198, 199). Louis et al. (200) first reported, more than 25 yr ago, that infusion of PGE₂ into the ovine fetus in late gestation stimulated an increase in the plasma cortisol concentration at a time when the fetal adrenal gland was relatively unresponsive to ACTH stimulation. Later studies have shown that the effect of PGE₂ infused into the fetus on fetal HPA function could be exerted at any one or all of the hypothalamic, pituitary, or adrenal levels (201, 202). Thus, the progressive increase in output of PGE₂ appears to contribute to the drive to fetal HPA function and augments the stimulus supplied by ACTH to the fetal adrenal (201, 203). Indeed, fetal PGE₂ infusion will provoke premature delivery of the ovine fetus (204). Placental PGE₂ output would not be subjected to negative feedback regulation by cortisol and may contribute to the apparent lack of negative feedback relationship between ACTH and cortisol in the late gestation ovine fetus.

Recent studies have suggested that in addition to PGE₂ stimulating output of cortisol by the fetal adrenal gland (205, 206), fetal cortisol, and not estrogen, may affect placental PGHS-2 activity and contribute to the rise in fetal plasma PGE₂ concentrations. Evidence in support of this suggestion included the observation that infusion of estrogen into fetal lambs in late pregnancy was without stimulatory effect on levels of placental PGHS-2 mRNA (207), although estrogen infusion into nonpregnant adult sheep did increase PGHS-2 expression in the endometrium (see also below). Studies with human amnion cell cultures and chorion trophoblast cells have suggested that glucocorticoids may up-regulate PGHS-2 gene expression in these tissues. Infusion of cortisol to fetal sheep in late gestation also increased levels of PGHS-2 mRNA and immunoreactive PGHS-2 protein (by Western blotting) in placental trophoblast cells. This effect was independent of changes in estrogen, since a similar stimulation of placental PGHS-2 mRNA levels was observed when cortisol was infused in the absence or presence of the aromatase inhibitor, 4-hydroxyandrostenedione.

Using immunohistochemistry we showed that the P450_C17 enzyme and PGHS-2 both localized to trophoblast epithelial cells, but not binucleate cells in ovine placentomes (208). Moreover, the appearance of ir-PGHS-2 clearly preceded that of P450_C17. Collectively, therefore, these data offer strong reasons to refute the current model of endocrine events occurring in the placenta of the sheep in late gestation and suggest that a different sequence likely pertains. This is summarized in Fig. 4. We have argued elsewhere that during late gestation in the fetal sheep, increased output of cortisol from the fetal adrenal gland progressively up-regulates PGHS-2 gene expression in placental trophoblast cells (208). The mechanism of this action remains unresolved. It may depend on trophoblast-specific transcription factors generated in response to elevations of cortisol, or it could be a direct action of cortisol since early studies reported a full GRE consensus sequence at approximately 760 bp upstream from the PGHS-2 transcription start site. We suggest that increased PGHS-2 expression in the sheep placenta contributes to increased PGE₂ output into the fetal circulation. Fetal PGE₂ drives the fetal HPA axis in a positive feed-forward fashion (Fig. 4). PGE₂, and not cortisol, is responsible for up-regulation of P450_C17 in placental trophoblast cells. This occurs in a manner analogous to the effect of PGE₂ on P450_C17 induction in ovine and bovine adrenal tissue. Ovine placental tissue expresses PGE receptor subtypes (EP1-EP4), but any changes in their expression during the course of late gestation remain to be determined (see Ref. 208). We have suggested further that increased P450_C17 in the placenta allows the conversion of C21 5 steroids directly through to 5 C19 steroids, precursors for estrogen biosynthesis, as demonstrated by Flint et al. (209) and Mason and colleagues (210) some years ago. A crucial difference of the current hypothesis is that this change is superimposed on an already substantial basal output of estrogen by the sheep placenta (measured as conjugated estrogens in maternal plasma and urine), and contributes principally to the terminal increase in maternal estradiol concentrations. This increase in estrogen is required for expression of CAP genes in the ovine myometrium and for expression of PGHS-2 in maternal endometrial tissue, predominantly endometrial epithelium. We have found that whereas the increase in placental (fetal trophoblast) expression of PGHS-2 after intrafetal cortisol administration was unaffected by concurrent infusion of 4-hydroxyandrostenedione, maternal endometrial up-regulation of PGHS-2 and output of 13–14 dihydro-15-keto PGF₂ (PGFM) into the maternal circulation occurred with cortisol infusion but was blocked by concurrent administration of the aromatase inhibitor (211). Thus, in sheep it appears that the fetal placenta and maternal endometrium exist as two separate sites of PG synthesis in late gestation and that these are differentially regulated. In fetal placenta, PGHS-2 is increased by cortisol, independent of changes in estrogen output, whereas in maternal uterine tissue, up-regulation of PGHS-2 and maternal plasma PGFM is dependent upon increased estrogen production (Fig. 4).

Current studies are directed at examining this hypothesis further. Using immunohistochemistry and Western blot analysis, it is evident that GR is expressed in ovine placental tissue, predominantly in uninucleate trophoblast cells. Estrogen receptor (ER) mRNA and activity have been demonstrated in maternal endometrium but is apparently lacking in placental trophoblast (212). Hence, it is difficult to envisage how estrogen could provide a stimulus to placental PG production as previously hypothesized. It remains to be shown whether glucocorticoids affect placental PGHS activity directly or indirectly. However, in early studies we have demonstrated that glucocorticoids increase output of PGE₂ by ovine placental trophoblast cells maintained in culture, and this effect is abolished by addition of meloxicam, a specific inhibitor of PGHS-2 activity.

C. HPA function in the primate fetus and activation of parturition
The role of the human and subhuman primate fetus in controlling gestation length has been, until recently, less clearly defined than that of the sheep fetus. However, over the past few years it has become apparent that mechanisms leading to activation of fetal HPA function in primates bear considerable similarity to processes in sheep, and that fetal cortisol and fetal adrenal C19 steroids appear to play an important role. In 1933, Malpas (213) in a study of gestation length in human pregnancies complicated with anencephaly concluded that "... . the fetal pituitary and adrenal glands was responsible for the trigger to the neuromuscular expulsive mechanism that led to the onset of labor. " Early observations indicated that the mean length of gestation in anencephaly, after exclusion of cases with polyhydramnios, was similar to controls, but the proportions of preterm and postmature births were both higher (see Ref. 12). Similar results have been obtained after experimental anencephaly in rhesus monkeys (214). In monkeys, fetal hypophysectomy predisposed to prolongation of gestation (215), but fetal adrenalectomy was without effect on gestational length, although five of eight fetuses died in that study (see Ref. 12). Initial studies indicated that removal of the fetus, but leaving the placenta in utero (fetectomy) had little effect on gestation length. However, more recent studies have indicated clearly that placental retention after fetectomy was significantly longer (195 days) compared with 164 days in controls (216). Fetectomy in baboon pregnancy did not affect gestation length, although maternal estradiol concentrations fell to basal values and progesterone concentrations were reduced by 20–45% (217, 218, 219). Overall, these experiments are difficult to interpret. The numbers and observations are invariably small, no attempt is generally made to sustain uterine volume and the stretch stimulus to the myometrium, and it is technically very difficult to operate on the primate fetus without stimulating uterine contractility.

In intact rhesus monkeys, as in the baboon and human, there is an increase in maternal estrogen concentrations in late gestation that parallels an increase in the concentrations of fetal adrenal C19 steroids, particularly DHEA and DHEA-sulfate (DHAS) (220, 221). Maternal estrogen concentrations increase progressively and then more rapidly in the later phases of human gestation; estriol, derived in substantial part from precursors of fetal adrenal origin, rises rapidly in maternal plasma and urine in late pregnancy at term, and in preterm labor (221). When androstenedione was infused into pregnant rhesus monkeys at about three-quarters of the way through gestation, there was an increase in maternal plasma estrogen concentrations and premature birth (222). This effect was blocked by the coinfusion of the aromatase inhibitor 4-hydroxyandrostenedione, which prevented maternal endocrine changes and changes in fibronectin in the fetal membranes and inhibited the nocturnal increases in uterine myometrial contractility (223). Elevations of maternal systemic estrogen concentrations by infusion increased myometrial activity, but did not produce premature delivery or fetal membrane changes. It was suggested that in the primate, as in the sheep, estrogen is important for the normal processes of parturition. The failure of exogenous estrogen to stimulate sustained uterine contractility, even though locally produced estrogen formed after C19 steroid infusion was effective, led the authors to suggest that the estrogen had to be generated near to its site of paracrine/autocrine action (223).

D. HPA maturation in the primate fetus
There is emerging strong evidence that maturation of HPA function occurs in the primate fetus in a manner generally analogous to that discussed above in the sheep fetus. Excellent reviews by Pepe and Albrecht (221, 224) and by Mesiano and Jaffe (225) have provided detailed analyses of pituitary-adrenal function in the primate fetus. In the human, baboon, and monkey fetus the pituitary is necessary for adrenal maturation and steroidogenesis, at least during the second half of gestation. Adrenal development is impaired in anencephalic human fetuses. In the baboon fetus treated in late gestation with betamethasone, there was suppression of fetal pituitary POMC mRNA and reductions in fetal adrenal weight, and 3ß-HSD fetal adrenal ACTH receptor mRNA levels (221). The authors concluded that increased expression of fetal adrenal ACTH receptor and mRNA species encoding steroidogenic enzymes depended upon fetal pituitary ACTH stimulation.

In the human fetus, ACTH activity is present in the pituitary by 5 weeks’ gestational age, and CRH- and AVP-like activity is present in the fetal hypothalamus by approximately 12 weeks gestation (226). CRH_1–41, in addition to a large molecular weight form of CRH, are contained within the human fetal hypothalamic tissue. CRH and AVP synergize in promoting ACTH release from the human fetal pituitary tissue in early gestation, and the stimulatory effect of CRH and ACTH output was reproduced by 8-bromo-cAMP (see Ref. 12).

Levels of POMC mRNA in anterior pituitary tissue from fetal baboons increased significantly from mid (day 100) and late (day 165) gestation (term = day 184) in nontreated animals, and there was a corresponding increase in pituitary cells expressing ACTH peptide (227, 228). In the baboon it has been suggested that this increase in fetal pituitary POMC mRNA levels might be associated with increased pituitary CRH receptor activity, rather than increased expression of CRH peptide in hypothalamic nuclei. However, administration of estrogen to midgestation baboons resulted in an increase in levels of POMC mRNA- and ACTH-positive corticotrophs in pituitary tissue to values that approached, but remained significantly different from, those at term (228). Pepe et al. (229) have argued that this increase in POMC is secondary to an effect of estrogen on placental 11ß-HSD activity, particularly 11ß-HSD-2. In previous studies, these investigators have shown increased expression of placenta 11ß-HSD-2 in the baboon during pregnancy and have shown that activity of this enzyme is increased by treatments that increase estrogen and decreased with inhibition of estrogen production or action (221, 229). In midgestation, the relatively lower levels of placenta 11ß-HSD-2 allow passage of maternal cortisol into the fetal compartment and relative suppression of fetal HPA activity (221). With increased 11ß-HSD-2 activity at day 160, there would be diminished maternal cortisol reaching the fetus (230), allowing the fetal HPA axis to escape from the presumed negative feedback of maternal cortisol. This would allow increases in POMC gene expression, ACTH output, and fetal adrenal maturation. These results are compatible with observations that production of cortisol by the primate fetal adrenal gland is relatively low for much of gestation (231, 232). The bulk of the gland is occupied by the fetal zone with relative deficiency of 3ß-HSD, and predominant formation of C19 5 steroids, particularly DHAS (233, 234, 235). In late gestation, there is an increase in ACTH receptor mRNA and 3ß-HSD activity in the definitive zone of the fetal adrenal, and a decrease in ACTH receptor mRNA and formation of DHAS in the fetal zone (236, 237, 238). The expression of fetal adrenal enzymes P450_C17 and P450_SCC remained relatively unchanged during gestation. Thus, there are subtle differences between fetal adrenal development in the primate and sheep. In the former, expression of 3ß-HSD appears rate limiting toward adrenal cortisol output whereas in the ovine species, expression of P450_C17 appears to regulate fetal adrenal steroidogenesis.

In primate pregnancy, estrogen production in the placenta depends extensively on the provision of C19 precursor steroids, predominantly from the fetal adrenal gland (239, 240). Fetal adrenal DHAS can be converted to estrone and estradiol in the placenta, and approximately 50% of circulating maternal estrone and estradiol are derived from placental aromatization of fetal DHAS; the remainder is formed from maternal adrenal C19 steroids (239, 241). Activation of the pituitary-adrenal axis of the fetus occurs in late gestation. There is a progressive increase in the concentration of DHAS in the fetal circulation, which mirrors an increase in maternal plasma estriol concentration (maternal estriol is formed in the placenta from the precursor 16-hydroxy-DHAS that is 90% of fetal origin and formed in the fetal liver from adrenal DHAS). This pattern of fetal adrenal activation, reflected in plasma DHAS concentrations, resembles the time course of increase for plasma cortisol in the fetal sheep. Recent studies have shown that the fetal adrenal in primates is divided into the outer adult zone that produces predominantly aldosterone, the fetal zone that produces DHAS, and the transitional zone, interposed between the adult and fetal cortex, which produces predominantly cortisol (225). Thus, the elegant studies of Mesiano and Jaffe (225) and Coulter and colleagues (242), have shown that P450_scc is expressed throughout the primate fetal adrenal gland. P450_C17 is not expressed in the definitive zone but is expressed in the transitional and fetal zones. P450_C21 is expressed throughout the gland. 3ß-HSD is not expressed in the fetal adrenal at midgestation but is expressed in the definitive and transitional zone in late gestation fetuses. P450_C11 is expressed in the transitional zone in midgestation and throughout the fetal adrenal cortex in late gestation. ACTH stimulates steroidogenesis in the transitional and fetal zone; the major products in late pregnancy are cortisol from the former and DHAS from the latter. Both in vitro and in vivo studies show dependence on ACTH for fetal adrenal steroidogenesis. More recent studies, however, have indicated that CRH, potentially of placental origin (see below), can also stimulate the fetal zone to produce DHAS (243). In addition, this zone of the fetal adrenal appears to respond to trophic inputs from the fetal pituitary other than ACTH. ER-/ß mRNA is also expressed in fetal and definitive-transitional zones of the baboon fetal adrenal cortex at mid- and at late gestation (244). The presence of ER in the adrenal cortical cells provides an additional mechanism by which estrogen mediates ACTH-dependent functional maturation of the primate fetal adrenal gland. In addition, previous studies had shown that estrogens increase availability of LDL-cholesterol as precursor for adrenal steroidogenesis (245, 246).

The difference in fetal adrenal architecture between the sheep and primate fetus has been regarded by many as a clear obstacle to extrapolating from the sheep model of parturition to the primate. However, it is now apparent that similarities between these species are greater than the perceived differences (247). In both the sheep and primate fetus the fetal adrenal produces increased amounts of cortisol in late gestation (247). It is relatively unprofitable to make detailed comparison of the minutiae of temporal changes