The uterus is a dynamic organ that undergoes distinct molecular and functional changes during the menstrual cycle, implantation, pregnancy and parturition. Steroid hormones are the predominant driving force for uterine transformation, but there is also a paracrine and autocrine contribution via the local release of cytokines from both immune (leucocytes, maternal and foetal macrophages, natural killer and T cells), and non-immune tissues (uterine epithelia, cervix and uterine smooth muscle). This review provides a brief overview of the role played by cytokines during pregnancy and parturition and begins to explore how disturbances in cytokine networks may lead to impaired reproductive success. We hope that by identifying key areas of interest and controversy in field, the review will serve as a useful starting point for readers. It is important to note that much of the literature originates from animal studies and, given the complexity of cytokine interactions, we would caution extrapolation between species. We have, where possible, referred to pertinent human data.
Complex cytokine networks play an important role in a wide range of reproductive and pregnancy related processes. Here, we review the current knowledge concerning the impact of cytokines on uterine physiology and pathophysiology. Cytokines influence a range of uterine functions during the menstrual cycle, implantation, pregnancy and labour. The synergistic interactions between individual cytokines are intricate and dynamic, and modulated by pregnancy hormones. It is not surprising therefore, that perturbations to cytokine signalling are associated with adverse pregnancy outcomes, such as miscarriage, pre-eclampsia, preterm labour and foetal brain injury. Further insight into the complexity of cytokine networks will be required to develop novel therapeutic strategies for the treatment of cytokine imbalances in pregnancy.
Cytokines and their inflammatory networks
Cytokines are an extensive array of pleiotropic glycoproteins involved in the regulation of all biological processes. Although their traditional role is commonly perceived in relation to their immunoregulatory properties, cytokines also have a range of mitogenic and proapoptotic functions on non-immune cells (1). They operate as parts of highly complex integrated networks that exhibit marked multiple stimulatory/antagonistic interactions, synergism and a degree of functional redundancy. The complexity of their network regulation is due to the unique properties of cytokines, which include pleiotropism, where each cytokine has multiple target cells in an array of different organs, and where responses may differ according to cell type. Cytokines also act cooperatively to potentiate and modulate each other’s actions in order to induce specific effects (2). They are also capable of mutual antagonism, wherein different cytokines have opposing actions (3). Finally, functional redundancy (whereby several different cytokines act individually on a cell type to induce the same response) is often mediated via a common receptor complex (4). Redundancy is of particular relevance in reproduction as a salvage pathways, highlighted by the fact that various knockout mouse models expected to be infertile actually maintain pregnancies to term and deliver normally (5). Individual cytokines are further able to induce different effects based on their exposure time and absolute concentration: for example, tumour necrosis factor (TNF)-α can trigger apoptotic pathways, while paradoxically also being able to induce opposing bio-regulatory effects, such as mitogenesis and differentiation (6). Fine tuning of cytokine-driven responses is further mediated by an array of membrane bound and soluble receptors, their specific localisation, and the temporal regulation of intracellular signalling cascades (7, 8). The correct operation of cytokine networks is essential for much of normal physiological homeostasis, and this central role is underlined by the fact that inflammatory/immune cytokine-mediated deregulations are associated with a plethora of pathological reproductive conditions, which typically exhibit changes in both local and systemic cytokine profiles (9–11).
Physiological functions of inflammation in pregnancy
Seminal cytokines in mediating uterine receptivity
Animal models have indicated that exposure of the maternal tract to seminal plasma triggers a sperm-independent endometrial inflammatory reaction, which features extensive infiltration and activation of macrophages, dendritic cells, and granulocytes (12). This marked, transient inflammatory response conditions the maternal immune system to tolerate the conceptus and to organise endometrial molecular/cellular changes aimed at favouring embryo development and implantation. Furthermore, it dissipates by the time of blastocyst hatching/implantation, by which stage resident endometrial leukocytes exhibit an immunopermissive phenotype (13–15). This inflammatory response is triggered by prevalent seminal plasma factors such as transforming growth factor (TGF)-β, regulated upon activation, normal T-cell expressed and secreted (RANTES), macrophage inflammatory proteins (MIP)-1α/β, monocyte chemoattractant protein (MCP)-1 and prostaglandins. In turn, these stimulate uterine epithelial cell release of other pro-inflammatory cytokines, including granulocyte-macrophage colony-stimulating factor (GM-CSF), a phenomenon partly regulated by the steroid milieu (16–18). GM-CSF produced in this way is thought to target macrophages, granulocytes and dendritic cells in the underlying stroma. In the mouse, other seminal cytokines such as eotaxin, interleukin (IL)-1β, IL-4, IL-9 and granulocyte-colony stimulating factor (G-CSF) may orchestrate the chemotaxis, relocation and function of endometrial immune effector cells, and modulate the production of abortifacient factors such as interferon (IFN)-γ (19). In vitro studies of murine/human cervical explants and endometrial epithelial cells further implicate IL-6, IL-8, IL-10, IL-12, and leukaemia inhibitory factor (LIF) in this process. As such, both seminal exposure and subsequent local alterations in cytokine ratios (e.g. IL-10 : IL-12) depress immune cell-mediated defences in the lower genital tract in humans (17, 20), and modulate uterine killer cell activity and prime the uterus for pregnancy in swine (21). Murine models suggest that endometrial lymphocytes relocate to implantation sites and other mucosal tissues/lymph nodes as part of the development of maternal tolerance of the foetal allograft, a phenomenon mediated via maternal hyporesponsiveness to ejaculate/conceptus paternal major histocompatibility complex class I antigens (14, 22). This mechanism is thought to underlie the observation that partner-specific exposure to seminal plasma in women reduces the risk of developing pregnancy complications allied with poor trophoblast invasion/placentation patterns, such as pre-eclampsia (23).
Cytokines in implantation, placentation and foetal immunotolerance
Endometrial luminal cytokines may also have preimplantation embryotrophic properties, as has been proposed for G-CSF, GM-CSF and LIF in humans and mice (24–26). A tightly regulated temporal–spatial expression of endometrial cytokines is required to orchestrate both the early stages of implantation and trophoblast invasion of the decidua and maternal vasculature (27). In the first instance, cytokines act as mediators of the embryo-maternal paracrine dialogue associated with apposition, attachment and invasion (28). Although LIF and the IL-1 system are central to embryo implantation per se, IL-11 signalling is required for both blastocyst hatching/attachment and the associated uterine decidualisation process in women (25, 29). The difficulty in interpreting the absolute requirement of these cytokines in regulating implantation is unclear. In this respect, although LIF expression is essential to murine implantation, this requirement is more nebulous in women (11). The IL-12/IL-15/IL-18 system interacts with endometrial leukocytes, natural killer cells in particular, and participates in local angiogenesis and tissue remodelling (25). The IL-1 system cooperates in this process through the regulation of matrix metalloproteinase activity, whereas vascular endothelial growth factor (VEGF) acts as an angiogenic factor to promote neovascularisation at the implantation site (29). The implantation process remains, however, relatively poorly understood, and involves many more cytokines, such as epidermal growth factor (EGF) and heparin-binding EGF (mouse), which play additional roles in peri-implantation endometrial changes and trophoblast differentiation (30). The signalling process can be localised and temporally regulated by downstream signalling, as highlighted by LIF responsiveness in murine implantation. Murine blastocysts produce LIF, which the uterine luminal epithelium responds to by phosphorylation and nuclear translocation of signal transducer and activator of transcription (STAT)3. Localisation of the response is determined by STAT3 activation at implantation sites only, whereas temporal regulation is determined by STAT3 activation which occurs only on day 4 of pregnancy, despite extensive LIF receptor expression throughout the uterine epithelium during the entire preimplantation period (31). Cytokines also participate in regulating trophoblast invasiveness. On one hand, uterine natural killer cell-derived IFN-γ has been implicated in inhibiting invasion through increased extravillous trophoblast cell apoptosis and reduced levels of active proteases (32). However, paradoxically, human decidual natural killer cells can also promote trophoblast invasion and neo-angiogenesis through the production of IL-8, IFN-inducible protein-10, VEGF and placental growth factor (33). In addition, human trophoblast itself uses cytokines such as TNF-α-related apoptosis-inducing ligand (TRAIL) to induce endothelial and smooth muscle cell apoptosis during the physiological remodelling of the uterine spiral arteries; this removes their vasoconstrictive properties to meet the nutritional demands of the foetus (34).
Although the maternal immune system has been primed by seminal exposure to tolerate the foetal allograft, the maternal hormonal environment contributes to maintaining the status quo. Progesterone also acts at the foeto–maternal interface, in part contributing to support IL-3, IL-4, IL-5 and IL-10 production, which inhibit Th1 responses and favour allograft tolerance in women (35, 36). Although local T cells can produce many such cytokines, major Th2 production appears to derive from nonlymphoid placental/decidual tissues, particularly trophoblast (35). The complexity of the cytokine, chemokine and growth factor networks is further demonstrated in recent protein profiling data that suggest that the early pregnancy human decidua provides both a pro-inflammatory environment to support placentation, while paradoxically also harbouring anti-inflammatory cytokine-producing leukocytes that promote immunotolerant uterine natural killer and CD14+ monocyte phenotypes (37).
Contribution of cytokines to uterine expansion in early pregnancy
Following implantation, both the uterine compartment and uterine smooth muscle need to expand (without initiating contraction) to accommodate the developing foetus. There is a paucity of detailed information concerning the mechanisms that initiate uterine smooth muscle and foetal membrane growth in early pregnancy, but it is likely that mechanical stretch induces the release of cytotrophic substances. In vitro, mechanical stretch of human myometrial cells is associated with the synergistic production of cytokines (38). Both IL-1β and stretch activate increased uterine smooth muscle cell prostaglandin H synthase 2 and IL-8 mRNA expression via a mitogen-activated protein kinase dependent mechanism, but there appears to be a differential pattern of transcription factor activation and subsequent gene expression (39). These in vitro data have been interpreted to be important in relation to priming the uterus for term labour (or earlier in the case of multiple pregnancy), but it is possible that similar mechanisms also promote cell hypertrophy and hyperplasia during early and mid pregnancy.
Immune responses at the cervical barrier
The cervix and associated mucus plug (which contains antimicrobial proteins) form an important defence system against uterine invasion by micro-organisms during human pregnancy (40–42). From early pregnancy to the postpartum period, the cervix undergoes extensive remodelling (42); a dynamic process temporally influenced by hormonal (43–46) and cytokine related events (47–50). Research in this area has focused mainly on delineating the cascades linking progesterone withdrawal, cytokine induction of matrix metalloproteinases (MMPs) and collagen breakdown during cervical ripening at term (see below). Related mechanisms may be involved in cervical softening in the first trimester when the collagen bundles become less tightly packed (42). For example, cultured cervical epithelial cells from nonpregnant women clearly have the ability to basally secrete macrophage-CSF, TGF-β1, IL-1α, IL-β and IL-8. Moreover, endocervical epithelial cell lines produce IL-6, IL-7 and RANTES, and further respond to the exogenous application of other cytokines, including IFN-γ and TNF-α (51, 52). Cytokines may also influence natural antimicrobial responses in the cervical mucus and plug, as production of secretory leucocyte protease inhibitor and elafin have to be shown to be elevated by cytokines in vitro (53).
Involvement of cytokines in preparation for labour
Immunoregulators play three interrelated roles at the time of parturition: they are involved in cervical ripening, promote foetal membrane weakening/rupture, and may enhance myometrial muscle excitability and contractility (Fig. 1). As such, advancing gestation is characterised by an increased inflammatory responsiveness in women (54), which parallels the increased circulatory inflammatory burden observed in murine models (55). Strikingly, data from pregnant mice highlight that changes in cytokines are strictly compartmentalised and independently regulated. There are sharply contrasting serum and amniotic fluid profiles between mid- and late pregnancy; this adaptation may allow maternal inflammatory priming in preparation for labour, while maintaining low intra-amniotic cytokine levels until the time of delivery, likely to avoid detrimental inflammatory effects to the foetus (56). A plethora of cytokines is involved in this process (e.g. IL-1β, IL-6, GM-CSF), but many of the changes in profiles from mid- to late murine pregnancy appear to revolve around tightly regulated alterations in IL-12 (p40) and its putative antagonism of IL-12-mediated responses through its antagonism of the IL-12 (p70) heterodimer (55–57).
The mechanisms underlying the onset of labour remain unclear. Uterine quiescence is maintained by increased progesterone receptor transcriptional activity, which is depressed prior to parturition. Mouse models suggest that the process is determined by foetal pulmonary maturation: the foetus triggers the onset of labour via surfactant protein-A secretion into amniotic fluid. In turn, this promotes foetal amniotic fluid macrophage expression of nuclear factor-κB and IL-1β, and relocation to the uterine wall, where they elicit an inflammatory response, which triggers uterine contractility and negatively impacts on progesterone receptor maintenance of uterine quiescence (58). These changes precede the marked increase in cytokine production typically associated with labour, which may be attributable to paracrine phenomena allied to prostaglandin and cytokine synthesis in the labouring uterine wall (59).
In human parturition, the contribution of surfactant has been less well studied, but the mechanism of functional progesterone withdrawal is generally accepted. In theory, progesterone withdrawal at term will remove constraints on the immune system and permit the initiation of a mild inflammatory process and labour. Indeed, there is increasing evidence pointing to the inflammatory nature of human parturition (48, 55). Recent microarray studies indicate an acute inflammatory gene expression signature in human foetal membranes at the time of labour, which is consistent with the initiation of neutrophil and monocyte recruitment. (61). Leucocytic influx also features an increase in IL-1β, IL-6 and IL-8 production in the cervix and myometrium during labour (62, 63). In the cervix, this promotes cervical ripening via the activation of MMPs and the inhibition of their tissue inhibitors. Cytokine-producing invading leukocytes in human myometrium also appear to be trafficked and suitably localised through coordinated production of intracellular adhesion molecule-1 as part of the propagation of the inflammatory trigger of labour (64).
The impact of inflammatory cascades on myometrial activity has been investigated both in vivo and in vitro. Experimental models of preterm labour have demonstrated that introduction of bacterial products or IL-1β into the amniotic cavity of pregnant animals leads to cytokine synthesis, up-regulation of Toll-like receptors, and premature uterine contraction (65–71). Rodent and simian models also suggest that this may be associated with the activation of prostaglandin signalling, since cyclo-oxygenase (COX)-2 inhibitors dampen cytokine induced contractility (72–74).
There is also an increasing body of in vitro evidence that cytokines play a role in enhanced intracellular Ca2+ signalling, cADP ribose production, increased phosphodiesterase activity, increased prostaglandin/prostaglandin receptor biosynthesis and modulation of oxytocin receptor expression in cultured human myometrium (75–85), all of which increase cell excitability. We have shown clearly in primary cultures of human uterine smooth muscle that IL-1β enhances uterine excitability and increases intracellular calcium availability (a trigger for uterine contraction). IL-1β modulates protein expression of TRPC3 (a putative calcium channel) and sarcoplasmic reticulum Ca2+-ATPases (75, 76), initiates spontaneous [Ca2+]i oscillations and augments store operated and receptor operated calcium influx (75). The IL-1β induced increase in TRPC3 expression does not appear to be influenced by COX-2 mediated prostaglandin synthesis, as nimesulide (a COX-2 specific inhibitor) does not inhibit this response (76). Interestingly, IL-1β stimulates arachidonic acid (the prostaglandin synthase substrate) (86), which is also a known activator of TrpC3 channels (87). These events are further potentiated by inflammatory events in the foetal membranes (61).
Inflammatory pathophysiology in pregnancy
Cytokines as markers of pregnancy-related disorders
Given the importance of the timely regulation of cytokine networks during pregnancy, deregulations of this system at both local and systemic levels are invariably characteristic of adverse pregnancy outcome. In this respect, cytokine signalling (i.e. involving both appropriate levels of cytokines and/or their receptors) is pivotal to processes such as implantation. Thus, in women, Th1-type response perturbations resulting from altered endometrial RANTES, IL-1α and IL-6 signalling at (normally immune privileged) implantation sites have been implicated in recurrent miscarriage due to an altered maternal immune response, effects on decidual tissue remodelling and angiogenesis, and deregulated trophoblast differentiation/invasion (88, 89). Although cytokines may mediate these effects directly, downstream effectors such as MMPs may also be implicated (90). Local alterations in cytokine profiles tend to remain relatively compartmentalised: indeed, although cervical mucus IL-6 and IL-8 levels are significantly higher in patients prior to miscarriage, these alterations are not always reflected in the circulation (91). Nevertheless, local changes can have profound effects: as such, although endometriosis causes obvious local cell-mediated immunological alterations in peritoneal fluid, these alterations may translate to ovarian follicles, thereby potentially compromising oocytes therein, and adversely affecting subsequent embryo viability (92).
However, when pregnancy-related disorders can be attributed to host-specific inflammatory/immune deregulations, dysfunctions in cytokine responsiveness can be measured peripherally in an array of immune effector cells in vitro. In this respect, in vitro stimulation assays have been used to investigate aberrant lymphocyte inflammatory responsiveness, proliferation and CD62L expression in women with a history of recurrent miscarriage, whereas peripheral blood natural killer cell Th1/Th2 ratio increases have been determined in women with pre-eclampsia (93, 94). Cytokine imbalance is also a key feature in women at high risk of preterm labour. Numerous correlations between cytokines (e.g. IL-6, IL-8 and MCP-1) in cervical fluid, amniotic fluid and serum with chorioamnionitis, preterm labour and premature rupture of membranes have been reported (60, 95). This has led to the hypothesis that high cervicovaginal cytokine concentrations (indicative of vaginal infection) are associated with premature cytokine mediated cervical shortening (via MMP activation) and subsequent intrauterine infection (60). However, other studies (96, 97) have suggested that a suppressed immune response to vaginal microorganisms may result in a permissive environment for intrauterine infection. The two hypotheses are not necessarily mutually exclusive and may depend on a host of other factors such as gene–environment interactions and level of infection.
Indeed, there is a clear familial pattern in many pregnancy complications, such as unexplained recurrent miscarriage, pre-eclampsia, preterm rupture of foetal membranes and spontaneous preterm labour (98–101), indicating a genetic susceptibility to a deregulation in inflammatory response, where an otherwise innocuous inflammatory stimulus triggers pregnancy complications in susceptible individuals as part of gene–environment interactions. Cytokine and their receptor polymorphisms may explain some of the variation of risk between women. High secretory genotypes of IFN-γ and IL-10 have been associated with an increased risk of recurrent early pregnancy loss (102). Similarly, in preterm labour, polymorphisms in the IL-1RN have been reported (103) and difference frequencies of single nucleotide polymorphisms in genes encoding for TNF and TNF receptors 1 and 2 have been reported in ethnic groups (104). The additional interactive increase in susceptibility to preterm labour is exemplified by the associated with carriage of the high secretory variant A allele of the TNF-α 308 (G/A) single nucleotide polymorphism and bacterial vaginosis (105, 106). Although variant allele carriage and bacterial vaginosis each increase the risk of premature labour by 1.8- and 1.6-fold, respectively, this escalates markedly to 10.1-fold when both these risk factors are present together (106). Genome-wide screens for susceptibility markers may soon reveal specific genotypes with predictive value for identifying women at high risk of pregnancy complications.
The involvement of cytokines in reproductive biology is extensive, and our understanding of the specific contributions of these immunomodulators in relation to pregnancy remains relatively limited. We are, however, beginning to comprehend the impact that the complexity and compartmentalisation of cytokine networks, in concert with hormones, has on pathophysiological situations in pregnancy. It is increasingly accepted that research focused on the actions of individual cytokines is no longer sufficient for developing our knowledge of pregnancy related pathophysiology. Equally, it is clear that, to fully appreciate the intricacy and timing of inflammatory interactions in pregnancy, further research will need to take a ‘systems biology’ approach. It will only be through integration of bioinformatics and experimental information (e.g. in vivo and in vitro data from human/animal cells and tissues) coupled with the utilisation of powerful genomic, transcriptomics and proteomic technologies and genetic screening that will we be able to gain the necessary insight to develop novel, effective and targeted immunomodulatory therapies to improve pregnancy outcome.
R.T. is HEFCE funded and is further supported by the Biotechnology and Biological Sciences Research Council, Action Medical Research (UK Charity No: 208701), Tommy’s the Baby Charity (UK Charity No: 1060508) and the Dr Hadwen Trust (UK Charity No: 261096). NMO is supported by Cerebra (UK Charity No: 1089812) and the Sir Richard Stapley Trust (UK Charity No. 313812).