Main

The uterus undergoes many changes during pregnancy and must achieve enormous expansion to accomodate the growing fetus, and to support the fetus through sustained muscle tone, without generating propagated contractions. This state of quiescence-the absence of coordinated contractions-is critical to the successful outcome of pregnancy, but remains poorly understood. The delicate balance between maintenance of tone and resistance to propagated contractions continues to the end of gestation and the onset of labor, when the uterus becomes active and empties its contents through rhythmic, forceful, organized contractile waves. In this review, we discuss the mechanisms governing these responses and new insights into their regulation during pregnancy. We focus on emerging areas of research and cite recent review articles concerning well studied areas.

UTERINE SMOOTH MUSCLE: FUNCTIONAL STRUCTURE

Cellular

The four main types of smooth muscle tissue are distinguished by the type of contraction (tonic or phasic) and whether or not the muscle is excitable by electrical stimulation (i.e. generates an action potential (1). Uterine smooth muscle is highly excitable, and normally undergoes spontaneous rhythmic contractions that vary in frequency and amplitude. The nature of the regional variation in spontaneous contractile activity in the uterus is not known. Specialized pacemaker cells have been hypothesized as the initiators of activity, but have not been characterized; spontaneous activity may be a property of many myometrial muscle cells, with sites of initiation of electrical activity throughout the myometrium (2–4).

For a multicellular muscle to contract as an organ, the contraction of individual cells must be coordinated by sequential depolarization of cells. The uterus contracts as a three-dimensional system of cells, the individual responses of which are communicated to neighbors in a distinct pattern. Depolarization of one cell leads to activation of only neighboring nondepolarized cells, so that a wave of activation proceeds in a specific direction, a process required for organized contractions and thus transportation of the uterine contents. During gestation, this coordination is far less effective than during labor, allowing gestation to be maintained. However, abortion can be induced by strong pharmacologic stimulation at any point in gestation, indicating that ineffective electrical coupling can be surmounted. Therefore, the inhibition is relative rather than absolute, and the ability to respond is maintained.

Contractile Apparatus

Although largely considered an obligatory step in excitation-contraction coupling, activation of the contractile apparatus-the actin-activated myosin ATPase-recently has been recognized as a regulated process. Moreover, changes in components of the contractile apparatus exert significant effects on the ability of Ca2+ to activate contraction (5,6). Ca2+ - and calmodulin-activated MLC kinase is the principal control mechanism for contractility, and the intracellular level of ionized Ca2+ is the most important controlling feature [reviewed in Bárány and Bárány (7)]. However, other actin-related proteins, troponins, caldesmon, and perhaps calponin also are phosphorylated under physiologic conditions and regulate Ca2+-sensitivity of the contractile apparatus, allowing a finer degree of control of the response to a given Ca2+ level (6). This so-called "thin-filament-based" regulation enables smooth muscle cells to adjust the amount of force developed in response to the same Ca2+ level. Moreover, thin filament-based regulation may explain how force is maintained when levels of MLC (20 kD) phosphorylation or of Ca2+ are low. This mechanism could be important to both maintenance of gestational quiescence and subsequent augmentation of coordinated contractility during labor. For example, stretch has been found to increase the expression of caldesmon, an inhibitory molecule, in the uterus (5). Much work is now being done to elucidate the role of these regulatory mechanisms in uterine contractile function (6).

Another emerging area of interest is the role of small GTP-binding proteins, such as RhoA and Rac-1, relatives of Ras, which regulate the assembly of filamentous, nonmuscle actin, leading to alterations in cell shape as well as capacity for force development. Force development (i.e. contractility) at a given level of activation (by Ca2+) is determined in part by the extent of attachment (focal adhesive contacts) between actin and the intermediate filaments of the cytoskeleton within cells, as well as to other muscle cells. Increased numbers of focal contacts provide a stiffer scaffold and result both in enhanced force transfer from cell to cell as well, and in increased shortening length and force development. Further, increased MLC phosphorylation should enhance smooth muscle contractility, as predicted by studies demonstrating that intracellular Ca2+ concentration is not always a good predictor of either myosin phosphorylation or contractility. The relevance of this mechanism to smooth muscle contraction became apparent with the discovery that RhoA can regulate myosin phosphorylation (8)-the critical step controlling cross-bridge cycling rate in contraction. The interaction of RhoA and the MLC regulatory subunit of MLC-P leads to inhibition of MLC-P activity, thereby increasing myosin phosphorylation and hence contraction and/or stress fiber formation (Fig. 1). Guanosine 5′-O-(thiotriphosphate) can increase MLC phosphorylation through the inhibition of MLC-P, and RhoA may be involved. RhoA interacts with protein kinase N (related to protein kinase C) and with Rho kinase. Both kinases become activated upon interaction or binding with GTP-bound (active) RhoA. Kimura et al. (8) discovered that Rho-kinase phosphorylates the catalytic subunit of MLC-P, which prevents activation of MLC-P by the regulatory subunit, possibly by facilitating release of MLC-P from the catalytic subunit. Thus the general concept (9) is that Rho activation (possibly by growth factor-mediated signaling events) leads to its redistribution to the plasma membrane and subsequently the interaction with or recruitment of protein kinase N, Rho kinase, and their substrates (which already may be at the membrane). Still unclear is the turn-off mechanism for this, which likely would involve activation of RhoA's guanosine triphosphate hydrolase via guanosine triphosphate hydrolase-activating proteins. Interestingly, the attachment sites for stress fibers, the focal adhesions, are the points at which the cytoskeleton communicates with the extracellular matrix via the integrin receptors; thus, one might expect that force changes could signal through the integrin receptors, possibly to regulate the RhoA signaling mechanism in a feed-back of feed-forward manner. Levels of RhoA and Rac-1 are increased more than 3-fold in myocytes from pregnant human uterus (versus nonpregnant) (10), and in the rat uterus, OT stimulation or KCl depolarization increases translocation of RhoA from the sarcolemma to the cytoskeleton (11). These observations suggest that the RhoA-based signaling pathway may promote enhanced contractility at term.

Figure 1
figure 1

Pathways regulating MLC-P activity, and thereby myosin phosphorylation, and hence contraction and/or stress fiber formation. This figure depicts the roles of small GTP-binding proteins in the activation of smooth muscle myosin ATPase-to cause shape changes or contraction-through effects exerted on MLC-P. The key concept is that dephosphorylation of MLC (i.e. phosphatase activity) rather than MLC kinase activity is the target of regulation, but the end result is increased MLC phosphorylation and subsequent crossbridge cycling via the myosin ATPase. Activation of Receptor A (perhaps by contractile agonists or growth factors) promotes GTP binding to RhoA, which then can bind to and activate Rho-kinase. Rho-kinase-mediated phosphorylation of the MLC phosphatase's regulatory subunit causes an inhibition of the phosphatase, allowing increased levels of phosphorylated MLC and increased contractility. ECM, extracellular matrix; FAK, focal adhesion kinase; GAPs, guanosine triphosphate hydrolase activating proteins; GEFs, GDP-GTP exchange factors; MBS MLC-P regulatory subunit; MLCK, MLC kinase; the circled P indicates phosphorylated MLC.

Full size image

MECHANISMS OF CONTRACTION

Membrane Potential Controls Myocyte Responsiveness

Depolarization of the uterine smooth muscle cell (i.e. membrane potential is made more positive relative to the resting state) occurs when an action potential is initiated through the regenerative influx of cations. In nerve cells Na+ ions carry the bulk of the action potential current, but in smooth muscle cells, the depolarizing action potential current is carried by Ca2+ ions, which enter via voltage-gated Ca2+ channels [reviewed in Wray (12)]. The repolarization phase of the action potential is the result of both K+ ion efflux and the inactivation of Ca2+ channels as potential drops. Although still poorly understood, pacemaker-generated action potentials likely are initiated through instability of membrane permeability to Na+ (increased) or, possibly, K+ (decreased). Depolarization is a function of the resting membrane potential, and is blocked if the potential is sufficiently negative relative to the resting potential, i.e. hyperpolarized. Pharmacologic agents can also activate (or inhibit) ion channel activities, and stimulate (or inhibit) contractile activity. The role of Ca2+ is discussed below.

Calcium-Mediated Contractile Activation

A variety of hormones, neurotransmitters, and pharmacologic agents influence the contractile state of uterine smooth muscle cells [reviewed in Fuchs (13)]. Ultimately, the ability of any of these agents to affect response is governed by the final common pathway of Ca2+ ion mobilization, which couples excitation to contraction. Activation causes the intracellular concentration of ionized free Ca2+ to rise from a resting level of ∼130 nM (14) toward the 200-400 nM level, enabling Ca2+-activated MLC kinase to increase myosin ATPase activity and cross-bridge cycling and muscle cell shortening [reviewed in Word (5) and Bárány and Bárány (7)]. Immediately after elevation of intracellular Ca2+ levels, homeostatic mechanisms act to restore the resting condition by active removal of the excess Ca2+ from the interior through energy-driven extrusion mechanisms (pumps) and channels present in the membranes of intracellular Ca2+ storage sites (SR) and in the plasmalemma.

Calcium entry. Ca2+ mobilization is regulated through channels that gate ion flux through the plasmalemmal and SR membranes. The plasmalemmal channels can be activated by electrical changes (voltage-operated channels), receptors/second messengers (e.g. putative G protein-coupled channels), and forces such as stretch and pressure (see below).

Calcium channels. The properties of myometrial ion channels and their regulation by voltage and agonist-occupied receptors have been reviewed recently (15). The actions of channel-specific inhibitors indicate a major role for both Ca2+ and K+ channels in controlling uterine contractility (Fig. 2). Specifically, the L-type (dihydropyridine-sensitive or dihydropyridine receptor) VOC, and the high conductance Ca2+-activated K+ channels ("maxiK" or "KCa"), related to the Drosophila "slo" channel (16), exert the greatest influence on intracellular Ca2+ ion concentration and contractility in myometrium. VOC and KCa channels operate in a reciprocal manner. Depolarization-induced VOC activation causes Ca2+ influx, and the increased intracellular Ca2+ levels activate KCa channels, permitting K+ efflux and restoration of membrane potential, thereby inactivating the VOC. In the interval between the completion of these two cycles, the muscle cell contracts, and if chemical activators remain available to their receptors, the activation phase is repeated in a cyclical manner.

Figure 2
figure 2

Roles of Ca2+ and K+ channels in controlling smooth muscle activity/contractility and relaxation. The two types of channels are thought to act in a reciprocal manner to control contractility. Calcium entry via the dihydropyridine receptor/VOC activates the contractile apparatus and also activates Ca2+ release from SR stores, providing a locally elevated Ca2+ concentration in the region of sarcolemmal KCa channels. The locally elevated Ca2+ (sparks) are sufficient to activate KCa channels, which repolarize the sarcolemma through potassium efflux, and thereby eliminate the voltage potential driving the VOCs. DHPR, dihydropyridine receptor; TEA, tetraethylammonium chloride.

Full size image

Quantal Ca2+ release. Recently, a much clearer picture has emerged [reviewed in Bootman and Berridge (17) and Berridge (18)] of the mechanisms of fine control of Ca2+ mobilization in smooth muscle and other cell types. Ca2+ is released from intracellular storage sites in quantal units, variously called "sparks" or "STOCs" (spontaneous transient outward currents) in muscle cells and "puffs" or "bumps" in nonmuscle cells. The receptors/channels responsible for these types of Ca2+ release are the SR RyR (Figs. 2 and 3) and the IP3R (Fig. 3), respectively. The RyRs are activated by Ca2+ flowing through VOCs, whereas IP3Rs are activated by receptor-linked PLC activation. These release receptors have two important properties. They have a bell-shaped sensitivity to Ca2+; low concentrations prompt more Ca2+ release (positive feed-back), whereas at high concentrations (> micromolar), release is inhibited (negative feedback). The positive feedback mechanism also has been called "Ca2+-induced Ca2+ release." The brief (70-500 ms) punctate Ca2+ release events occur just beneath the plasma membrane and produce only small current fluctuations because they activate only local neighboring ion channels (possible only one or a few clustered channels). Although small, the "sparks" are thought to play a major role in the generation of global Ca2+ signals within the cell through the recruitment of VOCs. In the myocardium, SR Ca2+ sparks now are thought to be initiated through the intimate association of the VOCs in the T-tubule invaginations of the sarcolemma with the subjacent RyRs of the SR, the VOC-derived Ca2+ serving to activate sufficient RyRs to activate global Ca2+ release from the SR (19). In smooth muscle, low intensity global signals stimulate contraction, whereas conversely, focused high intensity signals (e.g., "sparks") in the submembranous region bring about relaxation. In other tissues, receptor activation leads to a proportionate increase in IP3 concentration, which increases both the frequency of sparks and the Ca2+ sensitivity of the neighboring IP3Rs. This acts to coordinate the release of the sparks and results in their summation as waves of Ca2+. The progressive recruitment process allows graded responses despite the regenerative or positive feedback nature of the spark. Most important is the spatial segregation of Ca2+ signals. For example, in arterial smooth muscle, the global elevation of Ca2+ levels (waves) triggers contraction, whereas sparks produced by RyR located near the sarcolemma cause relaxation. This elaborate interaction between IP3R and RyR to control Ca2+ levels, and thus smooth muscle tone, has been described recently (Fig. 2) (17,20). The concept is that increased intracellular Ca2+ concentrations, as a result of VOC and IP3R activation, lead to increased SR Ca2+ content, which increases the frequency and amplitude of spark emissions from the RyR. Because these sparks are located near the sarcolemma, they can activate KCa channels, which act to restore local membrane potential and thereby feedback inhibit VOC activation. The proposed feedback mechanism to restore resting tension is conceived of as acting in parallel with the homeostatic Ca2+ restorative systems (pumps and uptake mechanisms). These models provide an important framework for understanding the dynamics of myometrial tone throughout gestation, and suggest several potential sites of action for the mediators of pregnancy-induced changes. The functional aspects of myometrial smooth muscle SR are still poorly understood (12). Further studies will be needed to determine the relevance of the calcium spark mechanisms to myometrial function, although it serves at least as a useful conceptual model. At least two subtypes of the RyR are reported to be expressed at the mRNA level in near-term human myometrium (21).

Figure 3
figure 3

Mechanisms of fine control of intracellular Ca2+ mobilization in smooth muscle; role of receptors/channels. When SR Ca2+ stores are depleted, IP3R activation is coupled to activation of store-operated Ca2+ channel, leading to replenishment of SR Ca2+ stores. cADPR, cyclic ADP ribose; CICR, Ca2+-induced Ca2+ release; ROC, receptor-operated Ca2+ channel; OC, store-operated Ca2+ channel.

Full size image

Potassium Channels

High conductance KCa channels regulate contractility in vascular (22) as well as uterine smooth muscle (23–25). Of note, cAMP-generating agents (e.g. β-adrenergic agonists) (15), cGMP-generating agents (e.g. NO) (26), and NO itself, through an apparently direct effect (27), all activate KCa channels to promote relaxation. Current research is directed at understanding the mechanisms through which these agents regulate KCa channels in terms of the roles of receptor-channel coupling, channel phosphorylation, and modification of critical sulfhydryl groups by oxygen radicals such as NO in the regulation of this channel. The molecular identify of the KCa channel has been determined (28), but little is known about K+ channel expression in the myometium, although functional differences in KCa channels present in laboring versus nonlaboring human myometrium have been reported (29).

Another type of K+ channel, the ATP-sensitive K+ channel, also affects the contractility of uterine and other smooth muscles, probably by NO release as well as by other mechanisms. A voltage-sensitive K+ channel with distribution in heart and uterus has been identified through molecular homology studies (30), and may have a function in the labor process (see below).

Stretch and Force-Mediated Signaling

Stretch can induce contraction of the myometrium (31) and other muscles. The mechanisms of stretch-induced activation have been studied most extensively in myocardium [reviewed in Crozatier (32)]. Stretch-activated ion channels, the open time of which is increased upon stretch or deformation (e.g. due to pressure or shear stress), are still poorly understood, but are thought to gate Cl- (33) or Na+ (34) ions. The stretch-induced Ca2+-mobilization process may involve activation of a Gi-coupled PLC (35); however, the effector molecules responding to stretch (i.e. the mechanotransducers) are poorly understood. A force/pressure-sensing system in contractile (36,37) and also noncontractile (38,39) cells is mediated through stretch-induced changes in the interactions between transmembrane integrins, their extracellular matrix ligands, and their intercellular cytoskeletal ligands (40).

Stretch not only initiates contractile activity, but also modifies contractile responses in the heart through "mechanoelectrical feedback" (41). However, the role of mechanoelectrical feedback in the regulation of myometrial contractility has not been studied.

Increased shear force increases human vascular endothelial cell expression of cyclooxygenase-2, superoxide dismutase, and endothelial constitutive NOS genes (42) as well as endothelial cell NO production, NOS activity, and endothelial constitutive NOS mRNA and protein expression in ovine pulmonary artery (43). Whether or not the increasing myometrial stretch that occurs with advancing gestation has an equivalent effect on myometrial NO synthesis, thereby increasing the quiescent potential, is unknown.

Stretch and Hypertrophy

The hypertrophy of uterine myocytes is one of the earliest responses to ovarian hormones and likely modifies the cellular responses to stretch and pressure through effects on electrical activity (via ion channels) and membrane-cytoskeletal interactions. Current studies of the influence of cellular swelling on signaling are focused on the effects of ischemia and reperfusion on myocardial cells (44), but may lead to concepts from which we can extrapolate increased understanding of myometrial smooth muscle cells. For example, a defect in the coupling of intracellular Ca2+ sparks emitted by the SR RyRs and sarcolemmal VOCs is implicated in the pathophysiology of myocardial dysfunction in hypertension-induced myocardial hypertrophy and congestive heart failure (45). The data suggest that hypertrophy may alter the spatial orientation between the VOCs and the SR RyRs such that sparks (see above) couple inefficiently to VOC activation, thereby reducing the Ca2+ available for excitation-contraction coupling-in effect reducing the VOC recruitment by the sparks. The response of the pregnant myometrium to stretch-induced activation is likely to be markedly attenuated before term. The myometrium could provide a model for understanding how hypertrophy affects cellular signaling, and vice versa.

COORDINATION OF CONTRACTILE ACTIVITY

Coordination of contractions among cells is achieved by coupling responses of individual muscle cells to their neighbors. Coupling between cells is achieved chemically and/or electrically. Chemical coupling is the process whereby a chemical (excitatory or inhibitory) has access to multiple cells in a region, and may affect all cells having receptors. For example, norepinephrine release by sympathetic nerves may initiate contraction of a sufficient number of cells to allow propagation of action potentials, via electrical conductance. Electrical coupling is the process whereby low resistance bridges physically connect individual cells and enable the transfer of electrical signals (carried by ions or second messengers) from one cell to its neighbors. The so-called "gap junctional" protein subunits, "connexins," arrayed in multimeric structures called connexons [reviewed in Beyer (46) and Finbow and Pitts (47)] are considered to provide these low resistance bridges between cells in the heart and in smooth muscles, and also in nonmuscular cells in a variety of tissues.

Gap Junctions

Gap junction distribution is not necessarily homogenous in tissues (48), and the relation between junctional patterns and intercellular communication remains poorly understood. Also, the junctions are not necessarily composed of a single type of connexin (49), suggesting that the functional aspects of the junctions may be related to the heterogeneity of subunit composition.

One of the best studied gap-junctional subunits is C43, a 43-kD protein expressed in myocardium and myometrium as well as other cells. The biophysics of electrical conduction along specialized pathways characterized by cells that express C43 in distinctive patterns is of special interest in the myocardium. Directional flow of electrical excitation from the sinoatrial node is achieved through a discontinuous rather than a continuous expression of C43 in cells along the conduction path, with sinoatrial nodal, C43-negative cells alternating with C43-expressing atrial myocytes (50). Expression patterns of myocardial junctional proteins illustrate important aspects of their function, and that these patterns are not necessarily intuitive. Clearly, much remains to be learned about how arrays of connexons dictate their varied functions, particularly as applied to the myometrium.

Gap-junction dynamics. The dynamics of myometrial gap junctions show patterns of regulation that suggest a role in facilitating the process of parturition. However, the spatial pattern of C43 expression among uterine myocytes has not been explored extensively. Ultrastructural studies in rodents reveal that gap-junctional surface area increases immediately before term, reaching a maximum at the onset of labor, then declining rapidly (51,52). In the human uterus, the increase is more gradual throughout gestation. The gap-junctional area is increased by estrogen and decreased by progesterone. The expression of myometrial C43 is estrogen-dependent, although only partial sequences of the canonical estrogen response element have been identified in the C43 promoter region (53), suggesting a role for other elements, in particular the activator protein 1 element (54). Estrogen increases C43 expression in rat myometrium, an action opposed by progesterone treatment (55). Estrogen increases expression more in circular than in longitudinal muscle, and C43 expression is unevenly distributed in the circular muscle, forming patchy areas of high expression. At parturition, however, C43 expression is evenly distributed and present in both muscle layers of the rat uterus (55,56), suggesting that widespread myometrial expression of C43 is important to parturition.

PGs also increase C43 expression, and may underlie the preterm increase in C43 expression in human myometrium (57). Despite a major increase in C43 protein and assembled gap junctions, the level of C43 mRNA has been found either not to change or to increase up to 5-fold in rodent uteri at term (52,58), suggesting that junctional assembly is the more critically timed process in gap junction dynamics.

Recently, an additional 45-kD subunit of connexin was demonstrated in rat myometrium. The 45-kD connexin expression was found to be at a similar level in the nonpregnant and pregnant uterus, to decrease just before term, and then to increase in the postpartum period (58).

Gap-junctional transmission. In addition to the presence of formed gap junctions, the function (i.e. permeability) of these connections also can be regulated, although the mechanism responsible for this is not known. Recently, physiologic regulation of conductance in vivo was demonstrated by the finding that electrical conductance between rat myometrial cells was increased in animals laboring at term compared with preterm or postpartum uteri (59). However, C43 assembly into multimeric complexes (connexons) is currently the best documented mechanism for controlling gap-junctional coupling between myometrial smooth muscle cells.

Myometrial gap junctions are lower in number before term, increase by late in gestation, and are fully formed before the onset of labor. However, whether gap junctions are critical to the parturition process, facilitative, or merely correlative is not known. Perhaps chemical coupling-achieved through an enormous increase in the local concentration of excitatory autacoids (e.g., OT and prostaglandins)-is sufficient to ensure adequate coordination of contractility for delivery. Of note, elimination of C43 expression in the heart (see above) disrupts key conduction pathways, producing impaired function, but the myocardium still contracts. This suggests that intercellular coupling does not depend solely on C43, and hence may be primarily enhanced by, rather than fundamentally dependent on, gap junctional communication.

Coordination of Excitation

The three-dimensional spread of excitation has been explored through mathematical modeling in an attempt to understand how the wave of depolarization spreads, how the wave of Ca2+ mobilization is elicited by the electrical excitation, and how the wave spreads in the orderly manner required to achieve a coordinated, synchronized contraction. In one model, contractile waveforms arise primarily through action potential-based communication between cells via gap junctions (60). Also, waveforms were affected profoundly by the location of the initiating pacemaker cell. Although this model predicted waveforms resembling intrauterine pressure waves, transmission, based solely on action potentials, would be faster than actually observed. Another group (61) examined monolayers of human uterine myocytes to examine Ca2+ waves (see above) by confocal microscopy. The expected Ca2+ waves were found within the myocytes, could be elicited by OT treatment, and were independent of external Ca2+ concentration, all of which supports their emanation from intracellular Ca2+ stores. Waves of Ca2+ traveled from cell to cell in a coordinated fashion, propagating in a radial pattern from the initiating cell, demonstrating that coordination of response (i.e. coordinated oscillations in Ca2+ levels among cells) is an inherent property of myometrium. However, few cells contracted repetitively despite the continued presence of OT. Another model (62) may better explain the discrepancy between the speed of intercellular action potential spread (about 10 cm/s) and that of intercellular Ca2+ waves (about 4 µm/s); action potentials spread throughout muscle bundles to initiate contraction over a wide area, and initiate the slower Ca2+ waves that spread throughout the bundle. Each myocyte contracts for approximately 20 s before relaxing, in reasonable agreement with the value of 15 s predicted by the model. The predicted pressure wave closely approximated a genuine contraction event, and the largest myocyte bundle size allowed by the model was similar to that observed. The mechanism for the spread of intercellular Ca2+ waves, which was predicted to be nonelectrical because of the much slower speed of propagation when compared with the action potential, was examined in monolayers of human myocytes (62) and was a diffusible substance, the action of which was prevented by an inhibitor of cyclooxygenase, likely acting in a paracrine manner.

GESTATIONAL REGULATION OF UTERINE CONTRACTILITY AND OF RESPONSES TO SIGNALS

The Switch to Local and Humoral Control in Pregnancy

An intriguing characteristics of the pregnant uterus is the virtual disappearance of sympathetic, cholinergic, and peptidergic nerves (63,64) and also apparently NO-containing nerves (65). Such a change suggests that control of uterine contractility switches from autonomic to humoral-based. Indeed, the major endogenous contractile agonists of labor are generated locally: prostaglandins in the fetal chorion and maternal decidua, and OT in the decidua.

Gonadal Steroids Alter Membrane Potential

The hormonal milieu of pregnancy causes both hypertrophy of the uterine smooth muscle cells and, in some species, a hyperpolarization of the sarcolemma. Although estrogen alone can induce hypertrophy, both estrogen and progesterone cause hyperpolarization of the myometrial smooth muscle cell membrane (66). The mechanism of estrogen- and progesterone-mediated hyperpolarization is thought to be an increase in membrane permeability to K+, suggesting an effect on membrane ion channels (67). The hyperpolarization of the smooth muscle cell membrane reduces the excitability of the cells by electrical stimuli, by increasing the degree of depolarization required to elicit action potentials. Although there is evidence for a reduced membrane potential during gestation, this is not apparent in all species, suggesting that such a change is not critical to quiescence (67).

Pregnancy Alters Contractile Signaling

During gestation, the ability of the uterus to respond to activators of contraction is diminished (68) and reflects reduced concentrations of receptors. Thus contractile activators still can act, but the uterus normally remains quiescent. The synthesis of PGs by fetal chorion laeve normally is minimal before term (except perhaps in preterm labor associated with infection; see "Infection and labor," below), but increases markedly with labor (24). Interestingly, the umbilical cord is a major site of PGE2 production at term and contributes the bulk of the PGE2 present in amniotic fluid (69). However, there is no evidence that PGs can cross the fetal membrane in the absence of infection or rupture (70). Synthesis of OT within the uterus also is reduced before term and increases markedly at term to provide an autocrine mechanism for promoting contractions during labor.

The up-regulation of enzymes that degrade contractile agonists such as enkephalinase (which in human chorion laeve inactivates the potent contractile agonist endothelin-1), oxytocinase (degrades OT), and PGDH (degrades eicosanoids) will inhibit contraction; these are expressed at high levels during gestation (71,72). There is a significant decline in PGDH mRNA in chorion from spontaneously laboring women at term compared with those undergoing elective cesarean section (73). Women experiencing idiopathic (i.e. in the absence of infection) preterm labor also had a lower level of PGDH mRNA compared with term, elective section patients. Other investigators observed an increase in PGDH around the time of labor onset (70). Thus the fetal membranes have a high capacity to destroy endogenous contractile agonists during gestation, but it is not clear whether a decline in PGDH facilitates labor at term.

Expression of the key PG-synthesizing enzyme, cyclooxygenase-2, is low early in pregnancy, then becomes highly induced in the amnion near term in rats, sheep, and humans. In human amnion, one study estimated a ratio of 100:1 for cyclooxygenase-2 versus cyclooxygenase-1 mRNA (74). Thus, the availability of contractile agonists within the uterus may be maintained at a subthreshold level through reduced synthesis before term as well as by active catabolism.

Regulation of Ion Channels by Pregnancy

mRNA for VOC-type Ca2+ channels (measured by a polymerase chain reaction) increases before term, and also with antiprogesterone-induced preterm labor in the rat, but mRNA levels decline during labor (75).

Na+ channels (fast type) are present in myometrium, and their apparent concentration (inferred from current density) increases near term in rats (76). Also, a putative voltage-regulated Na+ channel, highly expressed in human heart and uterus (77), identified through molecular cloning analysis (78), has been localized to uterine nerves and myometrium in mice. In pregnancy, neuronal expression of this Na+ channel disappeared, whereas myometrial expression appeared, reaching a maximum at term; the Na+ channel co-localized with C43 (79). This regulatory pattern suggests a role in parturition, and the co-localization with C43 may indicate a functional role in junctional transmission (electrical coupling). An unusual voltage-gated K+ channel, noted above (30), also is regulated by pregnancy; it is expressed at higher levels at term in the mouse uterus, and this expression declines markedly after birth (80), suggesting a role in parturition.

Activation/Stimulation

Receptors for uterine stimulatory agonists. Many studies have documented low levels of OT receptors in myometrium and decidua before term and a marked increase just before (or during) labor. The receptors for OT, acetylcholine, norepinephrine, and ETs are members of the seven-transmembrane domain receptor structural family, which act via heterotrimeric Gqαβγ to activate PLCβ and to initiate the IP3-diacylglycerol signaling system (Fig. 4). PG receptors are members of the same G protein-coupled receptor family [reviewed in Coleman et al. (81)]. Molecular cloning studies have identified four distinct receptors for E-type PGs (82–85), and one each for F type PG receptor, I-type (prostacyclin) (86), and D-type PGs. The E-type PG receptor 3 receptor family comprises a group of several subtypes resulting from alternative mRNA splicing (87), each of which couples to different effector pathways (88). Such diversity in signaling may explain conflicting observations of the biochemical response to prostanoids in the uterus. The presence of I-type PG (prostacyclin) receptors in the uterus has not yet been reported, although it should be noted that prostacyclin has efficacy at E-type PG receptor as well as I-type PG (prostacyclin) receptors.

Figure 4
figure 4

Receptor-coupled, G-protein-mediated regulation of the IP3-diacylglycerol signaling mechanism of Ca2+ related smooth muscle activation. This scheme illustrates both the general pathway for distinct signaling via cAMP (on the left side) or PLC (right side) and also the possible mechanisms for signal modification ("cross-talk") via Gαi inhibition of adenylyl cyclase and cAMP-dependent protein kinase-mediated phosphorylation and inhibition of PLC. DAG, diacyl glycerol; PKA, protein kinase A; PKC, protein kinase C; PLC β2, phospholipase-Cβ2; PTX, pertussis toxin.

Full size image

Receptor-coupled effectors and myometrial G proteins also are targets for regulation by estrogen and progesterone, and by pregnancy. Estrogen increases PLC activity in response to adrenergic, but not cholinergic, stimulation (89), and decreases the expression of Gs (90) in the rabbit uterus. With advancing gestation, Gs concentrations increase (91) in the human pregnant uterus, whereas in rat myometrium, the concentrations of Gi decrease and of Gq increase (92). These changes are consistent with the effects of pregnancy in promoting uterine quiescence through alterations in receptor-effector coupling.

Stimulatory agonists. ET, a potent stimulator of myometrial contractions, is synthesized by endometrial stromal cells. ET receptors have been localized to myocytes of the rat uterus (93), and estrogen treatment increases receptor concentration. ET-1 also is synthesized in the myometrium during the early postpartum period (94), suggesting a paracrine role for ET-1 in parturition, as for OT (see below). ET-1 binding increases in the myometrial membrane fraction of laboring rats (95). In the human uterus, responses to ET are mediated by the ETA receptor subtype (96). ET-1 has been localized in both the decidua and myometrium of the late pregnant human uterus, but not in the nonpregnant uterus (97). However, another study failed to detect ET-1 immunoreactivity in either term-pregnant or nonpregnant human myometrium (98). Both ETA and ETB receptors were found in human myometrium, but receptor levels were not affected by pregnancy (97). Potent efficacious contractile agonists such as ET and OT probably also contribute to the control of bleeding in the immediate postpartum period as well as facilitating labor.

Inhibitors of Contractility

Use of agents that inhibit uterine contractions has been the main strategy for the control of premature labor. Unfortunately, identification of highly effective, uterine-selective inhibitory agents remains elusive. The principal reasons are 3-fold. First, difficulties in accurately diagnosing preterm labor have hampered the identification of truly effective agents, making it hard to design clinical trials. Second, available agents do not have a high degree of uterine selectivity and are of limited use because of their effects on the myocardium and the vasculature, including those of the fetus. Third, the nature of gestation as a process characterized by highly redundant control mechanisms may require simultaneous inhibition of several components of systems for safe, prolonged inhibition of uterine contractions.

Endogenous inhibitors. Progesterone block. Although production of progesterone is important to the maintenance of gestation in sheep and rodents, the decline in progesterone production, which marks the end of gestation, cannot be detected in primates. Csapo's (99) progesterone block hypothesis for the mechanism of gestational quiescence, based on observations in rabbits, seems to be of paramount importance in all species except humans and nonhuman primates. Failure to observe these progesterone changes in humans has led to the concept that antagonism of progesterone response, rather than reduced production, overrides the progesterone block in humans at term. Transforming growth factor-β blocks the inductive effect of progesterone on in vitro expression of enkephalinase activity in human endometrial stromal cells (100). The relevance of this mechanism to parturition, if any, remains to be established. Considering all the available data, labor apparently ensues equally well in the presence and absence of progesterone [reviewed in Olson et al. (101)].

Karalis et al. (102) suggested that cortisol may be a physiologic antagonist of progesterone action. CRH concentrations (derived from placental trophoblasts) in maternal plasma increase exponentially during gestation, reaching a maximum at term. CRH itself (103) may determine the timing of parturition, because CRH concentrations rise more rapidly in women whose pregnancies resulted in spontaneous preterm labor than in those delivering postterm, and cortisol production is increased only slightly in pregnancy. Inducible NOS expression is down-regulated by glucocorticoids; thus the putative quiescing role of endogenous NO could be associated with the CRH regulatory mechanism. The importance of CRH in human parturition needs clarification.

Receptors and signaling processes. Fetal membranes composed of amnion-chorion and attached decidua produce a substance that inhibits VOC-type Ca2+ channel activity (104). Fetal membranes also inhibit contractions induced by prostaglandins but not by OT, suggesting an unusual type of agonist-selectivity for this effect (105). OT receptor coupling to PLC-β activation is inhibited by cAMP-dependent protein kinase-mediated phosphorylation (Fig. 4), although the target of the kinase (e.g. PLC, Gq, Gi, or receptor) was not identified (106). This "cross-talk" between opposing receptor systems has been confirmed in a reconstituted cellular system, as well as with HL60 cells, where agonist (chemotactic peptide) activation of the β2 isoform of PLC was inhibited by cAMP-dependent protein kinase-catalyzed phosphorylation of its serine residues (107). The inhibition was found to be mediated via Giβγ subunits, which are released from Giα upon agonist activation of the receptor [for a general review, see Neer (108)]; inhibition does not occur through the Gq-coupled PLC-activating receptors. These findings suggest that the observed pertussis toxin sensitivity of the rat myometrial PLC response to OT (106,109) may be explained by Gi protein acting via this cAMP-dependent mechanism to inhibit OT response. Such cross-talk between receptor systems (Fig. 4) clearly has the potential to account for the mechanism of such enigmatic phenomena as the cAMP-mediated "relaxation" of smooth muscle. Rigorous testing of this concept is warranted.

Nitric oxide. Uterine NO production is increased during pregnancy but declines markedly at term, and this NO production is difficult to detect in the nonpregnant uterus (reviewed in Sladek et al. (110)). Expression of inducible NOS is localized to myometrial smooth muscle cells and decidual epithelium of the pregnant rat (65). Although all three isoforms of NOS are found in the uterus under different conditions, only inducible NOS expression follows a pattern compatible with a role in uterine quiescence. Expression of inducible NOS in human myometrium falls markedly with term or preterm labor (111), suggesting that NO may be an important component of the endogenous system that maintains uterine quiescence before term. Augmentation of these signaling mechanisms may offer innovative approaches to tocolysis, although this needs to be rigorously confirmed, and the safety to both mother and fetus, must be demonstrated clearly.

Although cellular sources are unknown, all three NOS isoforms are expressed in the cervix. Cervical inducible NOS is up-regulated in laboring rats (112), suggesting a role in the increased cervical distensibility required for parturition. Comparing d 19 with term, the greatest change was in inducible NOS expression, which, however, increased by less than 10%. The physiologic relevance of this modest change remains unclear.

Exogenous inhibitors (tocolytic agents). One clinical approach to inhibiting uterine contractions in preterm labor has been to use tocolytic agents such as nifedipine that reduce intracellular Ca2+ ion concentrations by inhibiting VOC-type Ca2+ channels. However, these agents lack uterine selectivity and produce cardiovascular side effects, such as reduction in uterine and umbilical blood flow (113). β-Adrenoreceptor agonists (e.g. ritodrine and terbutaline) reduce intracellular Ca2+ ion concentration via the cAMP signaling cascade. These agents are important, but are limited by the development of maternal pulmonary edema and marked tachyphylaxis through β-adrenoreceptor down-regulation (113). Although cAMP also may effect relaxation by reducing the sensitivity of MLC kinase to Ca2+, this mechanism has not been proven. Magnesium sulfate acts by inhibiting Ca2+ influx, probably by interaction with the cytoplasmic face of VOC-type Ca2+ channels (114).

Other tocolytic strategies have been to use receptor antagonists to block OT or PGF2α. Although theoretically sound, these approaches have had limited success. High endogenous agonist concentrations may limit the efficacy of competitive OT antagonists. In a random, double-blind clinical trial in 112 women experiencing preterm labor, one such agonist, atosiban, reduced contraction frequency about 2-fold greater than did placebo (115). More recently, a nonrandomized study of 62 patients demonstrated a 70% success rate for delay of delivery by more that 48 h, an effect equivalent to that of ritodrine, but with fewer side effects (116). Although a 48-h delay in parturition may not seem to be of obvious importance to a fetus that is born months prematurely, it is, in fact, critical to survival. Even a 24-h delay in delivery permits the administration of lung-maturing glucocorticoids such as betamethasone, which switch on surfactant production and secretion by alveolar type II cells. Although outside the scope of this review, surfactant is critically important in providing sufficient alveolar surface tension to facilitate alveolar air sac expansion so that gas exchange can occur and respiratory distress syndrome be prevented. Thus, any delay in parturition provides a window of opportunity to prepare the lungs for their role as a gas-exchange organ after delivery, and is fundamental to the survival of premature infants (117).

Suppression of prostaglandin production using nonsteroidal anti-inflammatory drugs such as indomethacin must be controlled carefully, especially the duration of use (113), to prevent side effects such as prenatal constriction of the ductus arteriosus, the patency of which is maintained by PGE2 (118). The inducible isoform of PG synthetase H, or cyclooxygenase-2, is largely responsible for the increase in PG production at term (119,120). Cyclooxygenase-2-specific inhibitors could be effective tocolytic agents and be free of adverse effects on the ductus arteriosus and other fetal organs.

A possible approach to tocolysis is by manipulation of the NO signaling pathway in the myometrium. NO is the active chemical agent mediating the effects of smooth muscle relaxants such as the nitroso vasodilators (e.g. nitroglycerine). Thus nitroglycerine or a similar NO donor eventually may prove to be another possible tocolytic agent. Likewise, inhibition of cGMP breakdown, by a specific cGMP phosphodiesterase inhibitor, may reduce uterine contractility.

Specific inhibitors of ATP-sensitive K+ channels and activators are available; however, their use as tocolytic agents is not especially promising because their inhibitory action on the pregnant myometrium is partial at best (15). Agents that activate KCa channels, such as a recently characterized African herbal glycoside, dehydrosoyasaponin I (121), offer some tocolytic potential but lack uterine specificity, which may limit their clinical use.

Prepartal Gestational Regulation

Activation/normal labor. Signaling and parturition: Conversion from quiescence to activation. The approach and onset of labor at term involves changes in many components of the contractile signaling system. Local production of contractile agonists increases, and structural alterations of the cytoarchitecture increase the effectiveness of endogenous agonists. Despite decades of study, the mechanisms that signal the onset of active labor in humans remain incompletely understood.

Alterations in myometrial structure and functional regulation at term. A major change in the ability of myometrial cells to communicate with neighboring cells is the increased synthesis of the gap-junctional proteins, C43 in particular, late in gestation. Closer to term, the gap-junctional proteins are assembled into connexons, which form the gap-junctions. The increased electrical coupling between adjacent cells facilitates the coordination of myometrial excitation, allowing the contraction to develop propulsive force. However, it has not been established that connexons are necessary for coordination of contractions.

OT receptors in both myometrial and decidual cells are markedly up-regulated just before the onset of labor. The dynamics of the recently cloned human OT receptor during gestation have been described. OT receptor protein is present in the nonpregnant myometrium at reduced levels, which increase nearly 300-fold (mRNA levels) at parturition (122). The receptors are distributed unevenly throughout the myometrium, rather than being homogeneously expressed from cell to cell. Decidual OT receptors are thought to couple OT to PG production in a paracrine amplification mechanism to enhance the contractile action of OT on the myometrium. In contrast to posterior pituitary OT, which is unchanged, local production of OT in the amnion, chorion, and decidua (123,124) increases markedly at term; this may serve as a paracrine mechanism for OT stimulation of contractions [reviewed in Zingg et al. (125)]. Estrogen increases uterine OT production at term in both rat and human uterus (123,126,127).

It long has been argued that the estrogen:progesterone ratio governs uterine excitability throughout gestation, and that an increase in this ratio underlies activation of the uterus at term. The ratio could be increased by increasing estrogen production without changing progesterone. The fetal adrenal steroid, DHEAS, which can be converted to estrogen by placental enzymes, may be involved in such a regulatory mechanism whereby placental estrogen feedback inhibits fetal adrenal DHEAS production (101). Late in gestation, this inhibition may be lost, leading to increased DHEAS and maternal estrogen production, as seen at term. The increased estrogen could overcome the suppressive effect of progesterone on OT receptor expression as well as increase OT and prostaglandin production, all of which promote uterine activation and labor. However, although DHEAS administration to the primate fetus does increase maternal estrogen production and uterine contractility, maternal administration of exogenous estrogen does not alter contractility. Thus, the role of DHEAS may involve a complex fetal-placental interaction involving other, as yet un-identified mediators.

ET concentrations in human amniotic fluid are higher at mid-trimester than at term (128) leaving open the question of a role for fetal membrane-derived ET in the labor process (and also the significance of enkephalinase). PG production by both the uterine decidua in rat (119) and fetal chorion laeve is increased at term, providing a source of contractile PGs. However, PGs or ET released into the amniotic compartment may not reach the myometrium until membrane rupture.

Lessons from Knockouts

Gene deletion ("knockout") is one approach to analyzing the putative role of biologic signals. Genes for several prominent signaling molecules have been deleted, but the only reported putative uterine inhibitory-related deletion is the gene for inducible NOS. Deleting this gene had no apparent effect on gestation or parturition (129–131), despite differences in the phenotype produced [reviewed in Cobb and Danner (132)]. Genes for other putative stimulatory signals, the OT receptor (133), cyclooxygenase-1 (134), and cyclooxygenase-2 (135), also have been deleted, again with no apparent effect on gestation or parturition. The cyclooxygenase-2 knockout in mice did lead to inhibition of ovulation.

Deletion of the C43 gene produced an atrioventricular conduction block, which prevented the increased cardiac output needed to sustain oxygenation and perfusion and survival (136). Disruption of the progesterone receptor gene causes infertility and inability to support implantation; therefore, gestational changes cannot be studied in these mice (137). Prolactin receptor gene deletion also prevented implantation TBASE, ID: 4223; no published report yet). Rat and human myometrium and endometrium contain binding sites for epidermal growth factor; estrogen increases epidermal growth factor receptor levels (13). Deletion of the estrogen receptor gene also produced infertility, but the mice were otherwise grossly normal (138). Of note, the estrogen receptor gene knockout mice had no uterotrophic response (progesterone receptor induction) to epidermal growth factor (139).

Estrogen receptor gene overexpression, on the other hand, produced delayed parturition and prolonged labor (140). A second molecular form of the estrogen receptor (designated estrogen receptor β) is expressed in the uterus, complicating interpretation of the estrogen receptor α knockout phenotype (141).

During gestation and parturition, uterine activity clearly is controlled by a complex network of redundant controls so that eliminating a single component may not be sufficient to derail the mechanism. The implication of this interpretation is that strategic targeting of multiple genes may be required to perturb the gestational mechanisms sufficiently to gain insight into the regulatory interactions.

Infection and Labor

Considerable attention has been paid to the role of infection and consequent inflammation of the uterine decidua and fetal membranes in the etiology of preterm labor. Indeed, intrauterine infection and the ensuing production of inflammatory mediators (cytokines and growth factors) have been documented in preterm labor associated with premature rupture of the membranes. The inflammatory mediators IL-1β and tumor necrosis factor-α cause up-regulation of cyclooxygenase-2 in rat endometrial cell lines (119), consistent with the enhanced prostaglandin production by the infected uterus. The intensity with which intrauterine infection stimulates delivery has led to speculation that all labor is the result of an infection-like process. However, the initiator of this process has not been determined because most labor is associated with the absence of detectable infection.

An understanding of timing mechanisms could shed light on the changes inherent to premature labor, and thereby suggest better tocolytic therapies. Of concern is the ability to sort out the contribution of inflammatory mediators to normal and premature labor to understand the most enigmatic problem of noninfection, "idiopathic" preterm labor, which may account for the bulk of premature births for which tocolytic intervention could be of medical value (i.e., not complicated by prematurely ruptured membranes) (113). If normal labor is not due to bacterial infection, then perhaps the hormonal changes that lead to alterations in cytokine and growth factor production at term need only be sought, which narrows the targets considerably. Alternatively, it can be argued that, in normal labor, there is a loss of resistance to infection-related signaling molecules (e.g., cytokines), which is maintained throughout most of gestation.

The predictive value of intrauterine infection on premature delivery has been the subject of several recent multicenter clinical studies (142–145). Although the studies confirm that elevated amniotic fluid IL-6 levels (144) are an indicator of an early state of intrauterine infection and the presence of bacterial vaginosis is associated with an increased risk of spontaneous preterm birth, elevation of cervicovaginal fetal fibronectin levels, a short cervix, and previous preterm birth were found (145) to be stronger predictors of preterm birth. Earlier studies demonstrated that bacterial vaginosis is associated with elevated fetal fibronectin levels (146). Thus, at present, elevated cervical fetal fibronectin levels appear to be one of the strongest predictors of premature birth and are closely associated with the presence of intrauterine infection.

SUMMARY

This review has focused largely on subcellular events that govern smooth muscle contraction and relaxation, using the recognized physiologic and biochemical changes in the myometrium during pregnancy and gestation as a unifying theme to illustrate the relevance of the regulatory phenomena. Thus, we have discussed roles of small GTP binding proteins, ion channels, and "sparks," and integrin-cytoskeletal interactions in cells other than myometrium and even nonmuscle cells to illustrate their potential for affecting myometrial function. It is often the case that advances in other fields provide insight into mechanisms that subsequently are found to be relevant to the myometrium (a good example is MLC kinase in the regulation of tracheal smooth muscle contractility). We believe that the advances in what appear to be distantly related disciplines will serve as guiding principles in the quest for better understanding uterine function; only further experimentation will determine whether we are correct. The role of stretch as a signaling mechanism for both uterine quiescence and labor is a particularly exciting prospect for research, which is likely to provide important insight into uterine function. If any of the topics in this review spark a new research interest in the reader, then we will have fulfilled our goal.