Is human labor at term an inflammatory condition?

Abstract

Parturition at term in normal pregnancy follows a predictable sequence of events. There is some evidence that a state of inflammation prevails in the reproductive tissues during labor at term, but it is uncertain whether this phenomenon is the initiating signal for parturition. The absence of a clear temporal sequence of inflammatory events prior to labor casts doubt on the concept that normal human labor at term is primarily the result of an inflammatory cascade. This review examines evidence linking parturition and inflammation in order to address whether inflammation is a cause of labor, a consequence of labor, or a separate but related phenomenon. Finally, we identify and suggest ways to reconcile inconsistencies regarding definitions of labor onset in published research, which may contribute to the variability in conclusions regarding the genesis and maintenance of parturition. A more thorough understanding of the processes underlying normal parturition at term may lead to novel insights regarding abnormal labor, including spontaneous preterm labor, preterm premature rupture of the fetal membranes, and dysfunctional labor, and the role of inflammation in each.

Human labor has features suggestive of inflammatory processes. However, data indicate that inflammation may not be necessary for labor initiation at term. Further research is needed to understand the events leading to normal human parturition.

Introduction

The onset of labor is regarded as an event initiated by the withdrawal of pregnancy maintenance mechanisms. Despite decades of intensive research, the molecular signals responsible for timely spontaneous labor are poorly described [1]. Major difficulties in understanding the biology of parturition stem from variability in the processes of parturition observed among different mammalian species, limitations of our ability to make direct observations in humans, inconsistencies in definitions of labor onset in the research literature [2–4], and methods focused on gene transcription (i.e., mRNA) rather than proteins to understand the birthing process [5, 6].

Inflammation is a physiological protective response mechanism that defends and re-establishes cellular homeostasis in conditions of infection, tissue stress, and injury. In the female reproductive tract, inflammation is a consistently noted component of many physiological processes, including ovulation, menstruation, and implantation [7, 8]. The cellular components responsible for both innate immunity (monocytes, neutrophils, eosinophils, and natural killer (NK) cells) and the adaptive immune system (T and B lymphocytes) play crucial roles in the molecular control of inflammation and its resolution in the reproductive tract [9]. Nonetheless, acute and chronic inflammation are increasingly recognized as contributing to reproductive dysfunction, including pregnancy- and labor-related complications, such as recurrent miscarriage [10], preeclampsia [11], and intrauterine growth restriction [12, 13].

It is clear that infection and/or inflammation contribute to a significant proportion of abnormal (preterm) labors (i.e., that these processes can lead directly to labor and/or preterm premature rupture of membranes (PPROM)) [14]. The evidence supporting this includes (i) administration of bacteria or bacterial-derived products to experimental animals results in premature parturition [15–17], (ii) both intrauterine [18] and extrauterine maternal infections, such as pyelonephritis [19], pneumonia [20], or periodontal disease [21], predispose pregnant patients to preterm delivery, and (iii) endogenous, noninfectious molecules classified as “damage-associated molecular patterns” (DAMPs) released during cell stress, injury, and death can trigger sterile intra-amniotic inflammation leading to preterm labor (PTL) and preterm birth (PTB) [22].

Inflammation has also been implicated in the mechanisms of spontaneous parturition at term. It has been noted that the cellular changes accompanying successful completion of pregnancy share features with chronic inflammation, such as vasodilation, immune cell infiltration, and localized production of molecular mediators that can abrogate an inflammatory stimulus [23]. Supporting evidence includes infiltration of leukocytes and upregulation of cytokines/chemokines in the cervix, fetal membranes, and myometrium during term labor (TL) [24, 25]. Cytokines, such as interleukin (IL)-1β and tumor necrosis factor (TNF)-α induce production of prostaglandins (PGs) and metalloproteinases that are capable of promoting uterine contractions [26], cervical ripening [27], and rupture of membranes [28]. However, whether this phenomenon is the primary driver of labor remains unknown, in part because much of the evidence derives from studies comparing patients at “term in labor” (TIL) with those at “term not in labor” (TNL). This comparison is flawed as a method of identifying mechanisms involved in initiation of labor, as most TIL patients are already in a well-established phase of labor or even advanced labor or postpartum. Conversely, “not in labor” spans a variable period of time that similarly may or may not reflect the activation of signaling cascades that would have led to spontaneous labor had not a cesarean (C/S) or induction of labor first been performed. Therefore, comparisons of patients undergoing “elective C/S prior to onset of labor” with persons in “established labor” may not address factors involved in labor initiation, as it is usually unknown for patients undergoing scheduled C/S how far into the future spontaneous labor onset might have occurred.

In this review, we summarize existing literature addressing the hypothesis that spontaneous human labor at term is at its root an inflammatory process. We do not address whether inflammation can lead to labor, as this is already firmly established (see above). Rather, our goal is to determine whether inflammation is a driver of labor, a consequence of labor, or an epiphenomenon (e.g., whether it occurs during labor in preparation for resolution of pregnancy). This is particularly important, as understanding these processes can provide insights into the pathological activation of pathways associated with preterm delivery.

Methods and approach

We searched PubMed for articles from inception to January 2022, focusing on molecular and cellular events occurring in association with normal parturition in humans and animal models, as well as cell lines from gestational tissues. For the identification of relevant studies, the following search terms were used: “labor,” “inflammation,” “gene expression,” “microarray analysis,” “transcriptome,” “genome-wide analysis,” and “protein expression profiling.” We also screened the reference citations of the above papers for relevant studies and reviews to identify possible missing publications. Studies that focused on abnormal labors, such as preeclampsia, preterm labor and birth, or chorioamnionitis, were deemphasized.

What is known about the mechanisms of normal parturition and the role of inflammation in these mechanisms?

Our understanding of the mechanisms regulating human parturition is largely derived from studies in rodents and other model organisms (see [4, 29, 30] for limitations and benefits of using animal models to study human parturition). When integrated with evidence from humans [25, 31] and nonhuman primate models [32], it seems that parturition is a synchrony of multiple mechanisms wherein redundancy assures delivery in the absence of other (pathological) triggers. Mechanisms proposed to have the capacity to trigger parturition include (i) functional progesterone withdrawal [33, 34], (ii) surfactant protein (SP)-A secreted by the maturing fetal lungs [35], (iii) circulating cell-free fetal DNA (cffDNA) in the maternal blood [36], (iv) cellular senescence of the fetal tissues [37, 38], and (v) uterine stretch [39–41]. Below we summarize the evidence supporting each proposed mechanism as an inflammatory activator of spontaneous human parturition, along with evidence suggesting that these mechanisms do not depend upon inflammatory intermediates (Table 1). Other potential mechanisms, including a rise in corticotropin-releasing hormone levels [42–56], reactive oxygen species-induced oxidative stress [57–76], nitric oxide [77–90], hydrogen sulfide [91–93], and endoplasmic reticulum stress [94], are beyond the scope of this paper.

Table 1.

Major proposed mechanisms for the onset of term human labor and their potential inflammatory underpinnings.

Mechanism	Summary	Evidence supporting mediation by inflammation	Contradicting evidence
Functional progesterone withdrawal	During most of pregnancy, progesterone via PR-B promotes myometrial quiescence by suppressing responsiveness to proinflammatory stimuli. With advancing gestation, increased bioavailability of estrogen increases the stability and transrepressive activity of PR-A. This in turn decreases progesterone responsiveness by inhibiting the transcriptional activity of PR-B, leading to a proinflammatory state and local production of PGs that increase myometrial contractions and promote labor.	• hTERT-HM cells expressing more PR-A than PR-B have higher expression of proinflammatory genes compared to those with more PR-B than PR-A [99]. • Progesterone inhibits NF-κB activation, COX-2 expression in human myometrium [107], and secretion of proinflammatory cytokines in human placenta [108].	• PRs may regulate GJA1 (connexin 43) expression in the absence of NF-κB activation in hTERT-HM cells [34]. • In humans, myometrial inflammation is observed only after labor is established [162] . • Myometrial inflammation is not observed in RU486- induced labor in mice [115–117].
SP-A	Fetal lung SP-A accumulates in amniotic fluid of mice as term approaches. SP-A induces NF-κB to increase inflammation, and subsequently labor.	• SP-A can induce CAP expression in human and mouse myometrium and proinflammatory gene expression in mouse myometrium [119, 121].	• Deficiencies of SP-A (or of proteins regulating SP-A’s production), and overexpression of SP-A do not consistently affect gestation length and/or expression of proinflammatory mediators in mice [119, 120, 122, 123] • SP-A concentrations decrease or remain unchanged (but do not increase) during labor in human amniotic fluid, decidua, and chorioamniotic membranes [126–128]. • SP-A suppresses inflammation in human amnion, decidua, and decidual macrophages, and suppresses both inflammation and delivery in preterm mice administered an inflammatory stimulus [124–127, 129].
cffDNA	Hypomethylated cffDNA released from senescent fetal tissues into maternal circulation triggers inflammation via TLR9 in immune cells, leading to labor.	• Extracellular cffDNA stimulates IL-6 production in macrophages [135].	• Although overall cell-free DNA concentration in maternal circulation increases with labor onset, the methylation ratio is higher in labor than in term patients not in labor [137].
SASP	As fetal tissues senesce, various proinflammatory molecules are released into maternal intrauterine compartments leading to labor via NF-κB activation.	• Senescence markers are capable of activating NF-κB in immune cells [139–142].	• Senescence markers have been consistently found only in preterm laboring decidua, and not in term laboring or nonlaboring decidua [144].
Uterine stretch	Prolonged stretch at the end of pregnancy stimulates the expression of gastrin-releasing peptide and various cytokines and chemokines, and recruits immune cells to the myometrium. These in turn promote parturition.	• Mechanical stretch elicits COX-2 expression in primary human uterine myocytes [149], and NF-κB-mediated proinflammatory cytokine and chemokine production in hTERT-HM cells [39, 154, 155]. • Uterine overdistension in nonhuman primates is associated with elevated proinflammatory cytokines [40].	• Most evidence comes from twin studies or animal studies—human singleton labor is not well characterized. • In both human primary myocytes and cell lines, expression of CAPs may be regulated via non-inflammatory mechanisms [34, 121].

For other proposed mechanisms, see text.

Abbreviations: CAP, contraction-associated proteins; cffDNA, cell-free fetal DNA; GJA1, gap junction alpha 1 or connexin 43; hTERT-HM, an immortalized human myometrium cell line; NF-κB, nuclear factor κB; PG, prostaglandin; PR, progesterone receptor; RU486, mifepristone; SASP, senescence-associated secretory phenotype; SP, surfactant protein; TLR, Toll-like receptor.

Functional progesterone withdrawal

Progesterone, a hormone produced in humans initially by the corpus luteum and later by the placenta, plays a crucial role in the maintenance of uterine quiescence during pregnancy. In most mammals, circulating progesterone levels decline toward the end of term pregnancy, eventually triggering the onset of labor. In rodents, this decline is mediated through regression of the corpus luteum (luteolysis). In humans, in contrast, circulating progesterone remains elevated until birth, leading to the concept that a “functional” progesterone withdrawal occurs before parturition [95]. This is supported by the fact that administration of mifepristone (RU486, a progesterone receptor antagonist) induces labor in both humans [96] and mice [97]. Progesterone action is mediated through its two nuclear receptor (PR) isoforms, PR-A and PR-B. PR-A is a truncated form of PR-B that lacks the first 164 N-terminal amino acids. In human myometrial cells, PR-B is the principal ligand-dependent transcriptional activator of progesterone-responsive genes, whereas PR-A is a ligand-activated repressor of the transcriptional activity mediated by PR-B [98]. There is evidence that the relative ratios of PR-B and PR-A in the uterine myometrium play a decisive role in promoting either relaxed or contractile phenotypes, respectively [99].

Mechanisms by which a functional progesterone withdrawal might occur include (i) changes in the ratio of PR isoforms [100]; (ii) decline in PR coactivators and/or repressors [33]; (iii) local progesterone withdrawal in myometrial cells mediated by the progesterone metabolizing enzyme 20-α hydroxysteroid dehydrogenase (20αHSD) [34]; (iv) regulation of microRNA-200 and its targets ZEB1 and ZEB2 proteins, which regulate local progesterone metabolism [101]; (v) increased circulating estrogen levels [102, 103] and enhanced estrogen receptor (ER)-α activity [100, 102] near term promoting a cascade of proinflammatory events leading to the decline in PR function [104]; and (vi) reduction of PR transcriptional activity by increasing nuclear factor (NF)- κB activity [105].

The prevailing hypothesis is that the pregnancy-conserving actions of progesterone are mediated through PR-B by preventing several proinflammatory events (including inflammatory cell infiltration [106], activation and nuclear translocation of NF-κB [107], and expression of proinflammatory cytokines [108]); by suppressing production of contraction-associated proteins (CAPs) [109]); and by promoting the expression of genes that tend to stabilize pregnancy, such as prostaglandin dehydrogenase (Hpgd) [110]. As pregnancy advances, prolabor stimuli such as uterine distension and fetal membrane-derived production of PGs promote PR-A expression [40, 99, 111, 112]. The predominance of PR-A tends to repress the anti-inflammatory actions of progesterone, leading to a proinflammatory state within the myometrial compartment [99]. During labor, activation of NF-κB in amnion cells upregulates prostaglandin synthase 2 (PTGS2/COX2) and represses PR activity [105]. A concurrent increase in inflammatory cytokines activates the NF-κB pathway [113] and increases the steady-state levels of PR-A in myometrial cells, further elevating inflammation [114]. Taken together, these lines of evidence connect inflammation to functional progesterone withdrawal and initiation of labor at term.

Evidence that labor initiated by functional progesterone withdrawal is not mediated through inflammation

Parturition involves a complex interplay of various hormonal and other signals that allows for transformation of the uterus from a quiescent to a contractile state. A study conducted in mice investigated the relative contributions of hormonal and inflammatory signaling in the uterus modulating the onset of contractions [115]. RNA isolated from myometrium collected at multiple time points during gestation and in labor from mice given RU486 (a progesterone receptor antagonist) or vehicle control was sequenced for pathway analysis. Transcriptome changes unique to spontaneous term labor were predominantly related to muscle contraction, chemotaxis, and inflammation. Notwithstanding this observation, genes uniquely associated with RU486-induced PTL were primarily associated with tissue remodeling (cell adhesion and proteolysis). Inflammatory gene networks were not overexpressed in RU486-induced labor. These findings, together with other evidence demonstrating that PR knockdown does not affect inflammatory gene expression in uterine myocytes [116] and that RU486-induced labor does not involve infiltration of macrophages or neutrophils [117], support the conclusion that labor induced by progesterone withdrawal can proceed in the absence of inflammatory gene activation.

A study using two types of human myometrial cells (the immortalized hTERT-HM cell line and term myometrium from nonlaboring and laboring patients) demonstrated that PRs differentially interact with activator protein (AP)-1, a transcription factor, to regulate the expression of the gap junction protein alpha 1 (GJA1/Cx43), which is responsible for transmission of signals between adjacent cells in the myometrium [34]. During gestation, the progesterone-bound PR-B complex interacts with cJun/JunB homodimers of AP-1 to suppress the expression of GJA1. At the end of gestation, increased PR-A, especially in the unliganded state, transactivates the expression of GJA1 by interacting with FRA2/JunD heterodimers. A concurrent increase in 20αHSD causes local metabolism of progesterone in laboring myometrial tissue, leading to functional withdrawal of progesterone/PR-B activity. Collectively, these findings raise the possibility that the myometrial contractile phenotype can be achieved without the involvement of inflammatory pathways.

Fetal lung SP-A

Pulmonary SP-A is synthesized by fetal alveolar type II cells and nonciliated bronchial epithelial cells starting in the 28th week of human pregnancy, reaching functional levels in the 34th week [118]. There is evidence from the mouse that SP-A from the fetal lung can initiate parturition through mechanisms involving the activation of NF-κB and inflammation in the pregnant uterus [35]. In a follow-up report, single- or double-deficient SP-A and SP-D female mice deliver normally at term in their first pregnancies, but mice doubly deficient in SP-A and SP-D exhibit a protracted time to labor in subsequent pregnancies [119]. In a related line of investigation, steroid receptor coactivators (SRC)-1 and -2 (genes that, among other activities, regulate SP-A production) were studied in pregnant mice. Dams heterozygous for SRC-1 and -2 and bearing SRC-1 and -2 deficient embryos experience a delay in parturition of approximately 38 h. Uterine tissues from mothers carrying SRC1- and 2-deficient embryos demonstrated diminished NF-κB activity and lower expression of contraction-associated genes. Importantly, the ovaries of these mice had impaired luteolysis [120]. These findings suggest that SP-A may serve a modulatory role in the timing of labor. In human myometrial cells, SP-A and SP-D increase the expression of oxytocin receptor (OXTR) and GJA1 [121], and in murine myometrial tissues, SP-A promotes proinflammatory cytokines and CAP gene expression [119]. These findings support the concept of fetal control of timing of parturition through SP-A, and that SP-A’s role might be exerted through inflammatory pathways.

Evidence that SP-A is not a signal for human parturition

Despite the above evidence, doubt persists regarding these proposed processes, including (a) the absence of delayed delivery in SP-A-deficient mice [122]; (b) the unusual emergence of such a delay in SP-A/SP-D double knockouts only in the second pregnancy; (c) the apparent indirect mediation of SRC-1 and -2 effects through delayed luteolysis [119, 120]; (d) mice overexpressing rat SP-A have normal length of pregnancy, similar to that of C57BL/6 wild-type (WT) mice [123]; (e) we previously reported a remarkable anti-inflammatory and labor-suppressing (not labor-enhancing) effect of SP-A on PTL induced by bacteria and bacteria-derived proinflammatory products such as lipopolysaccharide and peptidoglycan [124, 125]; (f) in humans, concentrations of SP-A have been found to decrease during labor in amniotic fluid [126] and decidua [127], rather than increasing, with no signs of fetal macrophage infiltration in the myometrium after labor. Similarly, SP-A mRNA levels are decreased (not increased) in pregnant patients undergoing preterm delivery in the absence of chorioamnionitis [128]; and (g) several in vitro studies provide support for the inflammation-suppressive action of SP-A in human amnion [126], decidua [127], and decidual macrophages [129]. In sum, we consider the questions of whether SP-A signals for the onset of labor, and if it does, whether this occurs via an inflammatory mechanism, to be unsettled.

Cell-free fetal DNA

Throughout pregnancy, a range of molecules, including proteins, peptides, lipids/lipid metabolites, and nucleic acids (normally predominantly intracellular), are released into the maternal circulation during turnover of uteroplacental cells. During apoptosis and following cellular damage, stress, or necrosis, the concentrations of these “sterile particles” in the extracellular compartment increase, where they may trigger inflammatory responses through innate immune receptors. These sterile particles are implicated in the pathogenesis of pregnancy-related complications [130]. Patients with sterile intra-amniotic inflammation have rates of PTB and neonatal morbidities similar to patients with intra-amniotic infection [22].

Cell-free fetal DNA is a DAMP released from trophoblasts and fetal membranes into the maternal circulation as a part of normal turnover during pregnancy [131]. The proinflammatory properties of cffDNA are attributed to the presence of large proportions of unmethylated CpG sequences (more common in fetal than adult DNA and in microbial compared to mammalian DNA) that can induce inflammation through toll-like receptor (TLR) -9 signaling in immune cells [132]. Studies have shown that cffDNA concentration in maternal plasma is correlated with gestational age (a feature thought to be related to senescence of placentas and fetal membranes toward the end of pregnancy) [133] and is found at higher levels during labor at term, followed by a rapid decline after birth [134]. This evidence, along with other studies that demonstrate the propensity of extracellular cffDNA to stimulate production of IL-6 by naïve macrophages [135], underpins the idea that cffDNA may act as an inflammatory signal initiating parturition [136].

Cell-free fetal DNA as a trigger to parturition has not been established definitively

A prospective cohort study of 55 pregnant patients examined the relationship between total cell-free DNA concentrations, methylation ratios, and inflammatory markers in maternal blood samples collected at the following time points: 28 and 36 weeks of gestation, and at admission for delivery [137]. This study found that spontaneous labor at term is associated with increased (not decreased) methylation of cell-free DNA compared to nonlaboring controls. Given that maternal cell-free DNA is methylated compared to placental (fetal) DNA, which is hypomethylated, the possibility that cffDNA serves as a trigger for parturition requires further investigation.

Fetal tissue senescence

Throughout pregnancy, fetal tissues undergo extensive growth and remodeling to accommodate the developing fetus. As pregnancy advances, the fetal membranes and placenta attain an irreversible senescent phenotype, thereby losing their functional and mechanical properties. Senescence of the fetal membranes generates signals of aging, collectively referred to as “senescence-associated secretory phenotype” (SASP). The SASP consists of a myriad of cytokines, chemokines, matrix-degrading enzymes, apoptotic inducers, and their ligands that are propagated via extracellular vesicles from the fetal side to the maternal side (uterine and cervix) to cause parturition by promoting inflammation [38].

A study conducted in mice examined expression of molecular hallmarks of senescence together with pro- and anti-inflammatory cytokines in the fetal membranes [138]. The expression of senescence markers such as phosphorylated-p38-mitogen activated protein kinase and phosphorylated-glycogen synthase kinase 3β was highest at gestation day (GD) 18 (1 day prior to delivery). The concentrations of the proinflammatory cytokines IL-6 and IL-8 were significantly elevated on the day of delivery.

At the molecular level, SASP markers, together with other DAMPs, such as high-mobility group box 1 (HMGB1) [139–141] and uric acid [142] (released during senescence-associated cellular injury), attract immune cells and engage receptors, such as TLRs, leading to a NF-κB signaling cascade that initiates parturition. Together, the findings suggest the presence of a progressive process of senescence and oxidative stress in the fetal membranes toward the end of gestation, which may contribute to an enhanced inflammatory state leading to parturition [138, 143].

Evidence that SASP may not be associated with term labor

A study examined senescence-associated markers in placental-decidual biopsies from patients with singleton vaginal term delivery (37–41 weeks’ gestation), preterm delivery (25–36 weeks), and nonlaboring patients undergoing C/S (31–36 weeks) [144]. Senescence-associated beta-galactosidase (SA-β-gal) and histone-γH2AX were used as indexes of cellular senescence. A clear signature of senescence, i.e., positive SA-β-gal and γH2AX, was observed only in preterm laboring decidua, with little or no staining in term decidua-placental sections (whether laboring or not). The finding that decidual senescence is present in preterm but not term labor speaks against gestational tissue aging as a trigger for normal parturition and leaves open the possibility that fetal tissue senescence is a consequence rather than a cause of labor.

Uterine stretch

During pregnancy, the uterine muscle (myometrium) undergoes dramatic transformation by increasing in both cell number (hyperplasia) and size (hypertrophy) to accommodate the growing fetus [145]. At the end of pregnancy, in response to hormonal and inflammatory signals, uterine myocytes assume a contractile phenotype to promote parturition [146]. Studies in primary myometrial cultures [147] and myometrial strips [148] have shown that prolonged mechanical stretch is associated with upregulation of CAP genes such as OXTR and PTGS2 [149]. These findings are further supported by animal studies showing that uterine stretch is associated with increased expression of OXTR, PTGS2, and GJA1 [150–152]. Although the pathways by which myometrial contractility is enhanced are not fully elucidated, evidence suggests that stretch stimulates the expression of gastrin-releasing peptide (a smooth muscle stimulatory agonist) [153], enhances immune cell recruitment into the myometrium, and induces secretion of a range of cytokines and chemokines [39, 154, 155]. Excessive stretch (as in multiple pregnancy or polyhydramnios) is associated with an increased risk of PTL and PTB [155]. Studies in a nonhuman primate model show that uterine overdistention by balloon inflation is associated with significant elevation of proinflammatory cytokines that precedes labor and correlates with the inflammatory response observed with human twin PTL [40]. It may be inferred that pathological increased uterine stretch can activate inflammatory signals that might trigger labor. However, as has been noted above, two objections to this conclusion persist: (1) while stretch-induced inflammation may be sufficient to induce labor, whether it is necessary has not been established; and (2) whether stretch is a trigger for normal singleton labor at term (as opposed to multiple gestation or pathological stretch) is not known.

Evidence of inflammation in gestational tissues at term

In preparation for labor and delivery, a complex interplay between maternal and fetal factors allows for at least three types of transformation: the quiescent uterus becomes rhythmically contractile; the cervix softens, effaces, and dilates; and the tensile strength of the fetal membranes weakens. In the following section, we have gathered evidence that the above events are associated with a core inflammatory response. In the subsequent section, we present evidence that calls into question whether these inflammatory responses are necessary for labor onset or propagation.

Uterine myometrium

Multiple data sources have demonstrated spatio-temporal activation of NF-κB and inflammation in the uterine myometrium toward the end of gestation. In one study, fundal and lower uterine segment tissues were sampled at C/S from pregnant patients before the onset of labor at 35 weeks’ gestation and in active labor at 39 weeks’ gestation. This study showed activation of NF-κB specifically in the upper region in laboring patients [156]. Another study reported activation of NF-κB in both the fundus and lower segments, both before and after the onset of labor at term [157]. Histochemical analysis of myometrial biopsies from pregnant patients who delivered by C/S before the onset of labor at term (37 weeks’ gestation), pregnant patients during active term labor (cervical dilatation >4 cm and <9 cm), and nonpregnant individuals undergoing hysterectomy for benign disease, revealed that inflammatory cells, predominantly neutrophils and macrophages, infiltrate human myometrium during spontaneous term labor [158]. Several gene expression profiling studies have identified enrichment of genes involved in the inflammatory response and chemotaxis in term laboring myometrium compared to nonlaboring samples [24, 159–161].

In normal pregnancies, infiltration of myometrium by polymorphonuclear leukocytes (PMNs) is rarely detected before labor. A study examined the presence of inflammation in gestational tissues in term pregnancies (39 weeks) without clinical infection and with intact fetal membranes [162]. The study population consisted of (1) 159 subjects who underwent elective C/S without labor; (2) 22 subjects who underwent medically indicated C/S without labor; and (3) 35 subjects who underwent unplanned C/S after the onset of labor. Histological analysis revealed that cellular infiltration and myometrial inflammation occurs after the onset of labor at term. In contrast, the myometrium in the nonlaboring subjects did not exhibit an inflammatory infiltrate. Thus, in this study, though term labor was associated with myometrial inflammation, infiltration by inflammatory cells was not observed before the onset of labor.

Uterine cervix

Cervical remodeling occurs slowly during most of pregnancy, but accelerates following the infiltration of leukocytes near term [163, 164]. A study investigated morphological and inflammatory changes in cervical biopsies obtained at term from patients delivered by C/S prior to the onset of labor (37–39 weeks, unripe cervix) and in term patients (38–41 weeks) immediately following vaginal delivery. These were compared to cervix samples of nonpregnant people undergoing hysterectomy (not matched by age) [163]. Cervix biopsies from term laboring and nonlaboring groups showed reduction in collagen content and increased CD68 staining compared to nonpregnant controls. The authors conclude that macrophages play a role in cervical ripening at term, though this conclusion is limited by the fact that CD68 is expressed in both myeloid and nonmyeloid cell types [165, 166].

In rodents, similar cervical remodeling characteristics are found with a peak in macrophage numbers at GD18 (1 day prior to birth) that decline to nonpregnant levels by the day after birth, [167]. This observation suggests that migration of macrophages into the cervix occurs during the peripartum period and may play a role in the process of parturition.

A recent study examined the role of macrophages in remodeling the cervix in preparation for birth [168]. CD11b-dtr mice (in which temporary macrophage depletion can be induced by injection of diphtheria toxin (DT)) exhibited depletion of macrophages in the cervix within 24 h of DT treatment on GDs 14 and 16. Although macrophages repopulated the cervix to levels similar to controls by GD18, severe impairment in cervical remodeling was observed. This result suggests that cervical macrophages play a role in remodeling in preparation for birth; however, it is confounded by the fact that DT exposure led to widespread fetal mortality.

Taken together, we believe the above findings imply that macrophages are important for processes that remodel the cervix in preparation for parturition; however, this has not been demonstrated conclusively.

Choriodecidua

The decidua comprises the innermost layer of the uterus, and among other activities is thought to coordinate responses at the maternal–fetal interface during labor. Leukocyte infiltration of the decidua occurs during labor [169, 170], which could activate parturition processes in adjacent maternal and fetal tissues through the release of matrix degrading enzymes, cytokines, and PGs [171]. A study examined gene expression and protein concentration to understand whether a distinct chemokine profile in the decidua is associated with leukocyte infiltration during labor at term [172]. The choriodecidua was sampled from TNL (37–42 weeks, elective C/S) and normal TIL (37–42 weeks, cervical dilation >3 cm and regular contractions) in an area distant from the membrane rupture site and placenta. Among the genes differentially expressed, mRNAs and proteins for several interleukins and chemokines were consistently upregulated in choriodecidua in TIL compared to TNL. The increase in chemokine concentrations directly correlated with macrophage numbers. Another study examined the frequency of histologic inflammation in decidual biopsies obtained from pregnant subjects undergoing term elective or nonelective C/S prior to the onset of labor with intact membranes, and from subjects who underwent nonelective C/S after the onset of labor for various obstetrical indications [162]. Infiltration of the decidua by PMNs was rare in this cohort before the onset of labor. However, after the onset of labor, decidual inflammation was observed, and it correlated with cervical dilation. The study also demonstrated that decidual inflammation precedes and potentially facilitates myometrial inflammatory events—while decidual inflammation was observed without myometrial inflammation, the reverse was not true (for further details on temporal activation of myometrial inflammation in this study, see “Uterine myometrium” above). A separate study examined changes in choriodecidual gene expression prior to labor (elective C/S) and after normal term labor (37–42 weeks, after vaginal delivery) to identify key pathways associated with labor [173]. Extensive inflammatory activation was identified in choriodecidua following labor, predominantly upregulation of genes regulating leukocyte trafficking and cytokine signaling. Together, the above findings support that choriodecidual inflammatory processes are activated during labor.

Amniotic fluid

In normal pregnancy, cytokine levels increase in the amniotic fluid from the second to the third trimester and may participate in parturition by stimulating local production of PGs [174]. A study analyzed cytokine levels in amniotic fluid collected from elective amniocentesis for karyotype analysis in the early second trimester (14–16 weeks’ gestation) and from elective term C/S deliveries (37.6–40.3 weeks’ gestation) [175]. A comparison of cytokine levels revealed higher concentrations of proinflammatory cytokines during the third trimester. In a cross-sectional study of 375 subjects, proinflammatory cytokine concentrations were measured in amniotic fluid from patients with either spontaneous onset of labor at term resulting in vaginal delivery or elective C/S without signs of labor [176]. In laboring patients, amniotic fluid was collected from the vaginal pool at the time of artificial rupture of membranes or after spontaneous rupture of membranes, and in some cases immediately after delivery. In patients undergoing elective C/S, amniotic fluid was collected with a syringe directly after incision of the membranes. Analysis revealed higher concentrations of proinflammatory cytokines in laboring patients than in the elective C/S group. Although the different methods of sample collection is a limitation, this evidence supports the concept that an elevated inflammatory load in the amniotic fluid is a feature of spontaneous labor at term.

Fetal membranes

As with other gestational tissues, inflammatory changes have been seen in human fetal membranes before and during labor at term [177]. A study determined mRNA and protein content of IL-1β and IL-8 in fetal membranes collected at pregnancy termination during the first trimester (up to 12 completed weeks’ gestation) or second trimester (up to 23 weeks’ gestation), and between 24 and 37 weeks following indicated C/S. After 37 weeks, membranes were collected at C/S prior to labor or after spontaneous vaginal delivery [178]. IL-1β and IL-8 mRNA concentrations were increased in third trimester amnion and choriodecidua as compared with the first and second trimester. A further increase in expression of IL-1β mRNA was found after term labor. IL-1β protein content was low during pregnancy but significantly increased after labor as compared with each of the other groups. On the other hand, IL-8 mRNA concentration was not significantly different between samples obtained before and after labor at term. IL-8 protein was relatively low during pregnancy. There was a significant increase in IL-8 protein content of amnion after labor compared with first and second trimester amnion. No significant difference was observed in IL-8 protein content at term before and after labor. A prospective cohort study examined transcriptional profiles of chorioamniotic membranes between patients at term with no labor (38.7–39.2 weeks, delivered by C/S) and those delivered vaginally after spontaneous labor at term (39.2–40.7 weeks) to identify the biological processes involved in normal labor [179]. Microarray and Gene Ontology analysis indicated increased expression of transcripts associated with neutrophil recruitment and activation, chemotaxis, monocyte recruitment and differentiation in TIL samples relative to nonlabor controls.

The amnion and choriodecidua progressively weaken before they rupture at term. This weakening occurs in a restricted area overlying the cervix, described as the zone of altered morphology (ZAM), as opposed to regions far from the cervix, termed as the zone of intact morphology (ZIM) [180]. A study examined transcriptional gene expression changes and lymphocyte activation in the ZAM and ZIM areas collected before labor after C/S at 28–29 weeks’ gestation (early preterm), 33–34 weeks (moderate preterm), and at term before labor (TNL). These were compared with gene expression in the choriodecidua at the site of rupture after spontaneous vaginal delivery at term (TIL) in samples collected as part of a prior cohort study [181, 182]. At full-term gestation and before labor (TNL), the ZAM region demonstrated graft rejection molecular signatures and higher proportions of activated decidual NK cells and M1-like macrophages compared with TNL ZIM. At the same time, the membrane of chorionic trophoblasts displayed highly polymorphic HLA haplotypes in TNL ZAM but not in TNL ZIM. In TIL samples, genes associated with inflammatory, wounding, and acute phase responses were significantly enriched. Collectively, these results suggest that within the ZAM region before membrane rupture there is a shift in immune activation, initiation of a graft rejection-like program, and selective exposure of HLA determinants of fetal origin. After membrane rupture, inflammation can occur in the ZAM region.

Placenta

Inflammation-associated placental lesions are well described in pregnancies complicated by infection-associated preterm delivery and PPROM. A retrospective cohort study examined maternal and fetal vascular and inflammatory changes after spontaneous vaginal term deliveries (39.3 ± 1.3 weeks’ gestation) and elective C/S deliveries without labor (38.7 ± 0.93 weeks’ gestation) in uncomplicated pregnancies [183]. Based on the presence of an inflammatory neutrophil infiltrate at two or more sites on the chorionic plate and extraplacental membrane, the maternal inflammatory lesions were divided into early, acute subchorionitis; intermediate acute chorioamnionitis; and late, severe chorioamnionitis. The fetal inflammatory response in the umbilical cord was also classified as early, umbilical phlebitis; intermediate, umbilical arteritis; and concentric umbilical perivasculitis. Histologic examinations revealed a higher rate of maternal inflammatory lesions in placentas from pregnant people after vaginal delivery as compared with placentas after C/S delivery without labor. However, no difference was observed in fetal vascular and inflammatory lesions between the groups. These findings suggest that maternal inflammatory changes exist in placentas from spontaneous term delivery (even in the absence of clinical infection).

Maternal peripheral blood

A successful pregnancy requires that the maternal immune system adapt to prevent fetal rejection. A large body of evidence supports the conclusion that maternal immunologic function is altered locally at the maternal–fetal interface, which may be necessary to facilitate crucial pregnancy events such as implantation, maintenance of gestation, and initiation of parturition [184]. Immunologic changes that occur at the systemic level have been less intensely studied. In a longitudinal cohort study, peripheral blood collected during all trimesters as well as 6 weeks and 6 months postpartum was analyzed for proinflammatory cytokines [185]. In this study, each individual’s 6-month postpartum blood sample served as their own comparison. Serum concentrations of cytokines and chemokines (including TNF-α) were elevated during the second and third trimester of pregnancy when compared with the postpartum period, while some cytokines (interferon (IFN)-γ and MCP-1) were repressed, suggesting a complex balancing act in immune regulation in peripheral blood during pregnancy.

A cohort study characterized peripheral blood leukocyte priming and activation in TNL and in TIL samples (37–42 weeks’ gestation) to identify the molecular and biochemical events accompanying labor [186]. Term labor was characterized by an overall increase in monocyte and neutrophil counts in maternal circulation. In addition, the migratory response of neutrophils to the chemotactic signal peptide formyl-methionyl-leucyl-phenylalanine was significantly greater in laboring compared with nonlaboring people. These changes were accompanied by a general increase in the levels of proinflammatory cytokines in the blood of subjects in active labor compared to nonlaboring controls [187]. A separate study evaluated the association between peripheral markers of maternal inflammation and the clinical onset of labor at term in serum samples collected at routine visits from term nulliparous participants [188]. Maternal blood serum IL-1β, -4, -6, -8, -10, TNF-α, and IFN-γ were analyzed in samples of term participants who labored spontaneously within 48 h of sample collection and matched control subjects who labored spontaneously at least 14 days after enrollment. The findings revealed significantly higher serum levels of IL-1, IL-6, and TNF-α in subjects who labored within 48 h compared with the delayed labor group, suggesting elevated maternal inflammation in the initiation of term labor.

Inflammatory crosstalk exists between gestational tissues

Molecular crosstalk between maternal–fetal tissues is important for maintaining pregnancy and the onset of labor. Studies using intrauterine co-culture models have shown that crosstalk occurs either by direct cell contact or through cell factors released to the extracellular space. Such interactions have been shown to promote proinflammatory amplification and uterine transition for parturition [52, 154, 162, 189–192].

Inconsistencies in the evidence supporting inflammation as the primary driver of labor

Although each of the models and mechanisms described above provides insights into fundamental questions regarding human parturition, a key question remains: Do myometrial and fetal membrane inflammation occur before the onset of labor, or do they develop after the establishment of labor? A corollary question is whether the presence of inflammation within laboring tissues serves the purpose of advancing labor or whether it is an outcome of labor, or perhaps preparatory for postpartum resolution and/or tissue remodeling. In this section, we present contrary considerations to the characterization of inflammatory mechanisms as required for the onset of normal parturition at term.

A cohort study examined decidual gene expression patterns in normal pregnant subjects who delivered at term with spontaneous labor (TIL, 38.7–39.3 weeks), and those who had elective C/S at term without labor (TNL, 38.9–39.1 weeks) to understand the role of the decidua in the orchestration of proinflammatory pathways during spontaneous term parturition [6]. A total of 46 genes associated with decidual development, sex steroid and PG signaling, and pro- and anti-inflammatory pathways were interrogated by high-throughput quantitative real-time PCR. Genes involved in development and steroid signaling pathways differed significantly between TIL and TNL. Among 23 proinflammatory genes interrogated, only 3 genes were differentially expressed between the two groups, and of these only 1 was increased in TIL (IL-1β) while 2 were decreased (CCL5 and CCL2). Expression of anti-inflammatory IL4 and IL10 did not change. The study concluded that with the initiation of parturition, the decidual expression of anti-inflammatory mediators decreases, while the expression of proinflammatory cytokines increases—a surprising conclusion given the above results, and possibly reflective of pre-conceived notions regarding the role of inflammation in labor. The study is further limited by lack of information regarding mode of delivery (vaginal or C/S) and an imprecise description of term labor (“cervical change resulting in delivery”). An additional limitation lies in the fact that a decline in the decidual expression of galectins and glycodelins, which are at best indirectly involved in regulating anti-inflammatory cytokines [193–195], was considered sufficient to establish the notion that parturition is associated with a decline in anti-inflammatory mediators.

The activation of NF-κB is a key step in the regulation of immune and inflammatory responses, including via increased production of the induced form of prostaglandin synthase (COX2). In human fetal membranes, particularly the amnion, NF-κB activity is increased in association with labor. It is thought that NF-κB in turn induces parturition-related genes. One study categorized amnion cells collected from patients following pre-labor elective C/S (at 39 weeks’ gestation) as “nonactivated” (low NF-κB/low COX2 activity) or “activated” (high NF-κB/high COX-2 activity) and used microarray to identify genes associated with amnion activation [196]. Among 19 000 genes interrogated, 919 genes were found to be increased in “activated” amnion. Gene Ontology and Ingenuity Pathway Analysis identified only two major networks in “activated” amnions: (i) cell death, cancer, and morphology, and (ii) cell cycle, embryonic development, and tissue development. A directed search for inflammatory genes in “activated” amnion identified upregulation of IL-8, IL-1RAP, thrombospondin, monoglyceride lipase, nuclear transcription factor X box binding, and macrophage migration inhibitory factor. On this basis, the study concluded that amnion activation is largely an inflammatory event that occurs as a prelude to the onset of labor. We note that this study relies on a circular argument—that the major transcription factor acting as an inflammatory “switch” (NF-κB) is relied upon to differentiate activation status. Nonetheless, “activated” amnion showed upregulation of only 7 of the 19 000 genes interrogated with a direct or indirect role in inflammation. Conventional markers of inflammation, such as TNF-α, IL-6, and CCL5, were not upregulated in “activated” amnion.

A study examined when myometrial inflammation occurs in relation to the onset of labor using samples from TNL (average 39.3 weeks), term early labor (TeL, 38.4 weeks, cervical dilation <3 cm), and term established labor (TestL, 39.6 weeks, cervical dilation >3 cm) [197]. Labor was defined as the presence of regular uterine contractions every 3–4 min. Activation of the AP-1 and NF-κB systems, cytokine profiles, and inflammatory cell infiltration were assessed. With the onset of labor, NF-κB activity progressively increased, being greater in both TeL and TestL compared to TNL samples. In contrast, AP-1(c-Jun) levels remained unchanged. TeL was characterized by relatively high mRNA concentrations of CXCL11, a chemoattractant, while all other inflammatory mediators were comparable to TNL. On the other hand, TestL samples featured increased protein and mRNA levels of proinflammatory cytokines compared to TNL and/or TeL samples. In addition, the absolute number of neutrophils was higher in TeL myometrium, but these cells were confined to intravascular or perivascular regions and had not infiltrated the myometrium. Together, these findings suggest that with the exception of NF-κB activity, markers of inflammation are present only with established labor, and not preceding or concurrent with the onset of labor or in early labor.

In a related line of investigation, the levels of inflammatory cytokines in the amnion, choriodecidua parietalis, and decidua basalis samples from TNL, TeL, and TestL were measured to understand the source of inflammatory stimulus to labor [198]. In the amnion, decidua basalis, and choriodecidua parietalis higher concentrations of cytokines and chemokines were observed in TestL samples. However, choriodecidua parietalis showed an increase in the protein levels of IL-1β and IL-6 in the TeL compared to TNL samples. The mRNA levels of prolabor genes (PTGS2 and PGDH) did not show any change with advancing gestation or the onset of labor in the fetal membranes. These findings contrast to many studies that have shown term labor is associated with an upregulation of inflammation and prolabor genes in fetal tissues [173, 179, 183]. The findings also suggest that the inflammatory stimulus to labor may begin in the choriodecidua parietalis, and that signals originating in the decidua may act on the myometrium to drive prolabor gene expression. Collectively, these findings support the hypothesis that myometrial inflammation is a consequence rather than a cause of labor.

A molecular shift from immune activation to suppression of inflammatory responses occurs weeks before term delivery

A study combined untargeted mass spectrometry, proteomic technology, and single-cell mass cytometry immunoassay approaches to evaluate the biological processes that precede spontaneous labor at term [199]. Blood samples (plasma and whole blood) were collected three times during the last 100 days of pregnancy (including the day of onset of spontaneous labor) from 53 patients. All 53 patients were in the first stage of labor at sample collection, among which 39 were in the latent phase and delivered within 11 h, and 14 were in the active phase and delivered within 4 h. Among the most informative metabolic features, the plasma concentration of steroid hormone metabolites (i.e., products related to progesterone and cortisol) increased 2 to 4 weeks before labor. The levels of plasma pregnenolone sulfate, an upstream substrate for the production of 17-hydroxyprogesterone, decreased over time, stagnating around 30 days before the day of labor. In contrast, the percentage of singlet live (DNA⁺cPARP⁻CD235⁻CD61⁻) granulocytes, Janus-kinase-STAT, and myeloid differentiation primary response 88 (MyD88) signaling pathways in CD56^dimCD16⁺ NK cells decreased over time. The innate and adaptive immune cell response to IFN-α and granulocyte-macrophage colony-stimulating factor in vitro diminished with approaching labor. These changes coincided with an approximate 2.5-fold increase in the plasma protein concentration of IL-1 receptor type 4 (IL-1R4), an inhibitor of IL-33 proinflammatory cytokine activity. In summary, these findings are in opposition to prior studies [185–188] and show decreased systemic inflammation during labor and parturition.

The above examples show that upregulation of inflammatory pathways in labor is an inconsistent finding.

Genome-wide association studies do not show robust evidence of association of inflammatory genes in the initiation of term labor

Several lines of evidence have suggested that birth timing and pregnancy phenotypes are defined by both the maternal and the fetal genomes [200, 201]. Nonetheless, little is known about the specific maternal and fetal genetic contribution to gestational duration and parturition. A two-stage genome-wide association study (GWAS) of more than 40 000 research participants identified three genomic loci associated with gestational length [202]. Among the three maternal loci, wingless-type MMTV integration site family member (WNT4) was found robustly associated with gestational duration. Functional experiments and analysis showed that a noncoding variant in the WNT4 gene region, rs3820282, mechanistically contributes to birth timing by regulating estrogen receptor signaling. The association at the WNT4 loci suggests the critical roles of the decidua, the maternal–fetal interface, and estrogen signaling in decidual cells for the duration of pregnancy.

A large-scale GWAS meta-analysis of more than 80 000 infants [203] reported that a locus on chromosome 2q13 (2q13) in the fetal genome was significantly associated with gestational duration. The locus harbors a variant rs759482 encompassing three genes that encode proteins in the IL-1 family of proinflammatory cytokines. The rs759482-associated regulation of gestational duration was characterized by diminished IL1A and IL36G levels, and upregulation of IL36RN, which encodes an antagonist to the IL36 receptor. It is evident across this study that the fetal proinflammatory locus encompassing the 2q13 locus is associated with gestational length. What is less clear is whether inflammatory genes under the 2q13 locus play a critical role in term labor, as the GWAS meta-analysis was conducted in post-term newborns.

Evidence that enhanced immune cell activity within the myometrium and the cervix at term occurs primarily for the purposes of postpartum tissue repair and remodeling

A study conducted primarily in mice investigated the mechanism of uterine activation during labor by evaluating myometrial infiltration by immune cells and cytokine protein levels throughout late gestation (GD15 and GD18), at spontaneous term labor (TL after delivery of at least one pup on GD19) and postpartum (2–6 h after delivery) [117]. Monocytes increased at GD18 and were dramatically upregulated through the early postpartum period. On the other hand, neutrophils were present in myometrium throughout late gestation, but significantly increased at TL and postpartum. The expression of multiple proinflammatory cytokines and chemokines was significantly upregulated during TL compared to GD15. Protein expression analysis, however, revealed that the majority of the cytokine proteins were elevated shortly after birth, temporally following changes in cytokine mRNA transcripts. The mRNA expression of genes encoding contraction-associated proteins (GJA1, OXTR, and PTGS2) was significantly higher in myometrial tissues from TL mice. In contrast to the continued and increased postpartum production of inflammatory mediators, the expression of contraction-associated genes (i.e., genes required specifically for labor) decreased immediately after TL to the levels detected on GD15. This observation suggests that, at least in part, the role of inflammation is to participate in postpartum resolution. Such postpartum processes include repair and regeneration of smooth muscle cells [204], cellular proliferation, differentiation, apoptosis [145], and angiogenesis [204, 205].

A cohort study investigated transcriptional changes associated with labor by comparing myometrial biopsies taken from patients prior to (NIL) and after the onset of spontaneous labor (LAB) by high-throughput RNA sequencing [206]. Labor was defined as regular contractions <3 min apart, membrane rupture, and cervical dilatation >2 cm. Microarray and principal component analysis revealed enrichment of genes in the PGE2 synthesis pathway and those associated with chemotaxis and leukocyte infiltration in LAB samples compared to NIL controls; however, whether these have a role in labor onset was not established. The transcripts upregulated in LAB biopsies encoded proteins, including tissue inhibitor of matrix metalloproteinase-1 (TIMP1), metallothionine (MT2A), IL-10, dual specificity phosphatase1 (DUSP1), and DNA damage inducible transcript 4 gene (DDIT4), may be more important for uterine repair and involution than for controlling the onset of labor.

A separate study primarily conducted in C3/HeN mice across pregnancy (GD15 and GD18), at delivery (GD19), 1 day postpartum, and in nonpregnant controls investigated whether macrophage migration into the uterus and cervix plays a role in parturition [167]. During pregnancy, uterine macrophages increased on GD15 relative to nonpregnant controls, declined before birth, and increased again postpartum. However, macrophage numbers in the cervix peaked on GD18, and then declined to nonpregnant levels by the day after birth. These findings imply that the return of macrophages into the uterus during the postpartum period contributes to the postpartum restoration of uterine tissue to the nonpregnant state, whereas macrophage infiltration before the day of delivery into the cervix facilitates cervical remodeling during the processes of parturition.

A study examined the timing of inflammatory cell migration, expression of chemokines, and the presence of proinflammatory molecules in the cervix through comparisons between WT and steroid 5α reductase type1 null mice (Srd5a1^−/−), before (GD15) and during cervical ripening (GD18.75), and postpartum (GD19, 2–4 h postpartum) [207]. Srd5a1 knockout mice have a mutation in the Srd5a1 gene that impairs local (uterus and cervix) progesterone metabolism, leading to accumulation of progesterone, compromised cervical ripening, and either delayed parturition or failure to deliver [208, 209]. Thus, studies with Srd5a1^−/− mice provide insights into how local progesterone affects the distribution and activation of inflammatory cells in the cervix and further their role in cervical ripening. Despite differences in timing and duration of labor, the distribution and molecular profile of macrophages did not differ between WT and mutant mice on GD15 or 18.75. However, the distribution of macrophages increased in the cervical stroma of WT mice between GD15 and postpartum. In contrast, neutrophils and/or monocytes (but not macrophages or eosinophils) redistributed throughout the cervical tissue during ripening (GD18.75), but not in Srd5a1^−/− mice (attributed to elevated tissue progesterone levels). The activity of myeloperoxidase (an enzyme made primarily by neutrophils and to a lesser extent in monocytes) was not detectable in pregnant cervices from mice at GD18.75 in either genotype, but was detectable postpartum. Targeted depletion of neutrophils in WT mice had no effect on the timing or success of parturition. A separate study demonstrated that cervical macrophages before ripening remain undifferentiated and inactive (neither M1 nor M2 markers are expressed). Within a few hours postpartum, gene and protein expression characteristics of M2 macrophages (including Ym1, Arg1, and Il13ra1) and, to a lesser extent, M1 macrophages (Il1a, Tnfα and Mcp1), were increased [164]. These findings, together with other evidence demonstrating that immune-modulating genes (Il6, Tnfα, and Il1α) are not induced in the cervix until the onset of labor [210], suggest that (i) normal cervical ripening does not require a typical inflammatory response involving neutrophil activation, and (ii) macrophages are recruited to the cervix before parturition and may remain inactive until birth, at which time they differentiate into both M1 and M2 phenotypes. M1-like macrophages participate in processes such as removal of the large amounts of disorganized extracellular matrix accumulated during cervical ripening; whereas M2-like macrophages suppress excessive tissue damage, thus promoting rapid tissue repair of the cervix.

Another study assayed mRNA and protein concentrations of select proinflammatory cytokines in cervical biopsies from (a) 16 nonpregnant fertile patients undergoing hysterectomy for nonmalignant disease; (b) 19 term-pregnant patients undergoing elective C/S with no signs of labor (unripe cervices); and (c) 20 pregnant patients with spontaneous cervical ripening and vaginal delivery (postpartum) [211]. Both mRNA and protein levels of IL-8, IL-6, and G-CSF were increased in pregnancy compared to the nonpregnant state, and to a greater extent in postpartum cervix.

Taken together, the patterns of leukocyte infiltration and expression of inflammatory cytokines locally in the uterus and in the cervix peripartum indicate that inflammatory mediators are important factors in postpartum repair and remodeling.

Conclusion

It has long been noted that human labor shares certain molecular features with inflammatory processes. A link between infection/inflammation and preterm labor is firmly established. In this paper, we aimed to explore whether normal spontaneous human labor at term can properly be considered an inflammatory state, and, more specifically, whether inflammation is a critical precursor in that process. We organized the existing data to understand whether inflammatory activation occurs before, during, or after labor onset (Table 2). There was no well-defined infiltration sequence of inflammatory immune cells within uterine tissues prior to labor, with the exception of the uterine cervix [162, 173, 174, 183, 197], nor systemic increases in inflammatory cytokines/chemokines in healthy individuals before the onset of labor [185]. Markers of inflammatory activation, including NF-κB, chemokines, cytokines, and influx of immune cells, were apparent during active labor in the uterine myometrium [197], fetal membranes [181], and choriodecidua cells [196]; however, whether these are necessary for labor onset or develop only after labor processes are underway is not established.

Table 2.

Evidence for and against inflammation as a stimulus of labor onset in reproductive compartments, with an emphasis on timing (before, during, or after labor initiation).

Compartment	Timing of inflammation in relation to labor onset
Uterine myometrium	• NF-κB activation has been detected before labor onset in humans in some studies [157], but not in others [156, 197]. • Studies determining leukocyte infiltration and/or enrichment of proinflammatory cytokine and chemokine genes have only detected them after labor is already established [24, 158–161].
Uterine cervix	• Proinflammatory cytokine/chemokine genes and/or proteins are enriched at term without labor (compared to nonpregnant), and are further enriched after delivery [207, 211]. • Cervical macrophage numbers peak the day before labor and decline after birth in mice [167]. • In mice, cervical macrophages remain inactive and undifferentiated until postpartum, with subsequent induction of proinflammatory cytokine genes [164, 207].
Choriodecidua	• Leukocytes are occasionally but not consistently detected before labor onset [162]. • Uterine macrophages decrease before birth and increase again postpartum in rats [172]. • Leukocytes are more commonly detected after labor is underway or after delivery than before labor [169, 170, 172]. • Proinflammatory cytokine/chemokine genes and/or proteins are only enriched during labor or after delivery [6, 169, 171–173].
Amniochorion	• In the area overlying the cervix, immune cells display proinflammatory activation before labor [180]. • Proinflammatory cytokines are higher before labor the day of delivery compared to the day before delivery in mice [177–179, 196].
Amniotic fluid	• Concentrations of proinflammatory cytokines are higher in the third trimester compared to the second, and higher intra- or postpartum compared to before labor [174–176].
Placenta	• Neutrophil infiltration in maternal placental vasculature is common even in normal (uninfected) pregnancy and delivery [183].
Maternal blood	• The data are inconsistent: some studies find increased proinflammatory cytokine concentrations before labor [185, 188], while others find a prominent increase in proinflammatory cytokine concentration only after labor onset [186, 187, 199]. • Leukocyte priming and activation gradually decrease before labor (compared to earlier in the third trimester), and increase again after labor onset [186].

Abbreviation: NF-κB, nuclear factor κB.

We also found consistent evidence that proinflammatory cytokines are elevated in the postpartum period compared to antepartum [212], leading us to conclude that, whether or not inflammation is necessary for the onset or maintenance of spontaneous labor in the absence of infection, it is likely that inflammation plays a role in postpartum resolution and tissue remodeling.

Many studies described in this review adopt a “transcript-centric” approach (i.e., use mRNA as a proxy for protein levels) to identify key molecular mechanisms of the birthing process [5, 6, 160]. This method is flawed, because protein activity depends not only on mRNA copy number but also on many other factors (e.g., regulatory elements in 5′- and 3′ untranslated regions, micro-RNAs, and protein–protein interactions, to name but a few) [5].

Immune cell modifications seem necessary for parturition in the choriodecidua overlying the cervix, which may be utilized for membrane weakening and rupture [181]. Although inflammatory responses are observed in the placenta, amniotic fluid, and maternal circulation, they appear to be mild, and whether they are required for the normal initiation of labor is not clear [183]. The evidence that inflammation is rarely detected before the onset of labor in the myometrium, and that its role in local rupture of fetal membranes (which most commonly occurs after labor is underway) without inducing widespread changes, raises the possibility that inflammation is not the main driver but rather a consequence of parturition. This assumption is supported by GWAS that have suggested that inflammation contributes to the birth process but comes after initiating and facilitating signals [202, 203].

To date, human studies have provided incomplete information regarding mechanisms underlying parturition. Both ethical and practical issues limit tissue sampling during and after pregnancy. With the possible exception of nonhuman primates, animal models do not fully represent the complexity of physiological systems active in pregnant humans. A significant drawback of existing studies is the imprecise definition of “term in labor,” and the extent to which this state can be used to learn anything about the initiation of labor processes. Typically, such subjects are sampled while they are in well-established or even advanced labor or postpartum. There is little uniformity regarding the definitions of these terms. Also inconsistently defined is whether spontaneous rupture of fetal membranes prior to the onset of uterine contractions is a labor event. Given that normal labor in terms of its progression and duration is not universal and varies between individuals, the transition between different phases of labor is only subjectively determined [213].

We recommend that future research aimed at understanding molecular mechanisms include standard definitions of labor onset that incorporate cervical dilation and/or effacement, and frequency of uterine contractions, together with clinically relevant markers to enhance comparability of research findings. Use of the clinical concept of “term” should be avoided in mechanistic research on timing of parturition, because it refers to a wide gestational age range and a rigid cutoff of 37 weeks of pregnancy. Pregnancies a few days before and after this cutoff are designated as “preterm” and “term,” respectively, even though they are likely to be very similar. For the purposes of mechanistic research, it would be more useful to draw wider boundaries, for example by restricting “preterm” to <36 weeks and “term” to >38 weeks.

Furthermore, labor should be specified as either “spontaneous” or “induced,” and augmented with uterotonics or not. If such methods of induction or augmentation of labor are employed (as occurs in most labors in the USA), their use should be specified, and results from such patients analyzed separately as appropriate (depending on the questions being asked and whether it can be demonstrated that samples obtained from such patients do not differ from those in spontaneous, unaugmented labor).

The answer to the question of whether inflammation is a necessary precursor to the onset of spontaneous term labor is unclear. It seems that, at least in some cases, authors have concluded that term labor is associated with inflammation without solid evidence [6, 196], or consider withdrawal of suppression of chemokines at the maternal–fetal interface as the mechanism underlying labor at term without demonstrating robust activation of proinflammatory genes [5].

Further research is needed to understand the events leading to normal human parturition.

Authors’ contributions

KNC wrote the first draft. MS and EH provided critical input. EH conceived the idea and provided guidance in writing the manuscript. All authors have reviewed the final draft of this manuscript.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Conference Presentation: Presented in part at the American Society for Reproductive Immunology Annual Meeting and Chinese Society of Reproductive Immunology Annual Meeting, Shanghai, China, June 30, 2018.