Activated Neutrophils Propagate Fetal Membrane Inflammation and Weakening through ERK and Neutrophil Extracellular Trap–Induced TLR-9 Signaling Free

Preterm birth, delivery of the baby prior to 37 wk of gestation, affects over 10% of live births worldwide (1) and is the major cause of neonatal mortality and morbidity within the infant’s first year of life. Preterm birth is also associated with learning disabilities and increased risk of long-term complications for the offspring (2). Despite these grave consequences, the causes and pathogenesis of spontaneous preterm birth remain incompletely understood. From previous work, we know that infection is an important risk factor for preterm birth (1, 3), and chorioamnionitis, inflammation of the placenta and fetal membranes (FMs), is observed in over 40% of preterm birth cases (4, 5). Chorioamnionitis is characterized by an infiltration of neutrophils to the FMs (6), although the function of these cells at the maternal–fetal interface remains unclear.

Neutrophils are short-lived phagocytic immune cells that are increasingly being recognized for having multiple functions during both physiological and pathological processes. Classically, neutrophils are rapid responders of the innate immune system that defend the body against invading pathogens; however, they have also been reported to interact with the adaptive immune system and be involved in chronic inflammation, cancer, and autoimmunity (7). In addition to phagocytosis, neutrophils can neutralize pathogens by degranulation, the release of cytoplasmic granules containing antimicrobial proteins and enzymes, and the release of neutrophil extracellular traps (NETs), webs of DNA that are decorated with histones and antimicrobial enzymes that can entrap and neutralize microbes (8–10).

Recently, we reported that neutrophils can be recruited by resting human FMs, and this recruitment was significantly increased if FMs were first exposed to low-dose bacterial LPS (11). Furthermore, conditioned media (CM) from LPS-stimulated FMs activated neutrophils to release high levels of many proinflammatory cytokines and chemokines as well as increased neutrophil degranulation, reactive oxygen species production, and NET release (11). Intriguingly, resting FM-CM also induced neutrophil NET release, albeit at lower levels, as well as low levels of cytokine production. Neutrophil recruitment and NET release under resting conditions was confirmed in vivo in a mouse model and has also been reported in pregnant women (12). We posit that these recruited neutrophils may not affect FM function under resting conditions and may instead be important for protective surveillance at the FMs because NETs may be a quick defensive mechanism against potential invading pathogens (13). Conversely, the products of neutrophil activation after exposure to LPS-stimulated FM-CM (FM+LPS) may deleteriously affect FM function, leading to premature preterm rupture of the membranes (PPROM) and preterm birth.

Extending upon our experimental model previously described (11), this study examined the effects of FM-activated neutrophil CM on subsequent FM viability and their secretion of matrix metalloproteinases (MMPs), tissue inhibitors of MMPs (TIMPs), and inflammatory cytokines. The signaling pathways activated in FMs by neutrophil CM were also investigated. In this study, we present the first evidence, to our knowledge, that neutrophils that were preactivated with FM+LPS induced FMs to increase their secretion of proinflammatory IL-6, IL-8, and GRO-α and the markers of membrane weakening, MMP-9 and PGE₂, without affecting FM viability. We demonstrate a role for ERK MAP kinase signaling in mediating some of these FM responses. Furthermore, we found that NETs in the activated neutrophil CM contributed to FM MMP-9 release, and this was mediated by the DNA sensor, TLR-9.

Human FM tissue and blood collection was approved by Yale University’s Human Research Protection Program (no. 0607001625).

Human FMs (37–41 wk of gestation) were collected from planned uncomplicated Caesarean deliveries without labor or known infection/inflammation. FM explants with the amniotic and chorionic layers intact were prepared as previously described (14, 15). FM explants were treated with or without 1 ng/ml LPS isolated from Escherichia coli O111:B4 (Sigma-Aldrich, St Louis, MO) in serum-free OptiMEM (Life Technologies, Grand Island, NY). We selected this dose of LPS as we previously reported that 1 ng/ml was sufficient to induce a robust cytokine response by human FMs (14) without directly inducing neutrophil activation (11). After 24 h, cell-free FM-CM was collected under sterile conditions and stored at −80°C. CM from three separate FMs were pooled together to reduce the influence of patient variability. Prior to treating neutrophils, 2% FBS (Gemini Bio, West Sacramento, CA) was added into the pooled FM-CM.

Neutrophils were isolated from the peripheral blood of healthy women and used immediately as previously described (11). Briefly, blood was collected into heparin-coated vacutainers (BD Biosciences, San Jose, CA), and neutrophils were isolated by density centrifugation over Histopaque 1077/1119 layers (Sigma-Aldrich). Hypotonic RBC lysis was performed, and the isolated neutrophil fraction was >95% pure as determined by flow cytometric analysis of CD16b staining (no. 302005, 1:100 dilution; BioLegend, San Diego, CA). Neutrophils were treated with no treatment (NT; OptiMEM supplemented with 2% FBS alone), untreated FM-CM (FM), FM+LPS, or as a control LPS alone (1 ng/ml) for 1 h before removing this media and any associated cytokines, changing the medium to fresh RPMI 1640 media with 2% FBS for 4 h at 37°C. We have previously demonstrated that this culture protocol does not negatively affect neutrophil viability (11). Thus, cell-free neutrophil culture supernatants were collected under sterile conditions and stored at −80°C.

Term human FM explants were collected as described above, rested in DMEM/F12 media supplemented with 10% FBS overnight, and then treated with neutrophil CM for 24 h. In some experiments, FMs were pretreated with the ERK inhibitor, SCH772984 (10 nM; Selleckchem, Houston, TX), or the TLR-9 antagonist, ODN2088 (TTAGGG, 1 μM, A151; InvivoGen, San Diego, CA), for 1 h at 37°C prior to the addition of neutrophil CM. In other cases, DNase I derived from bovine pancreas (1 U/ml, no. 10104159001; Roche, Indianapolis, IN) was added to neutrophil CM for 30 min at room temperature and then the CM incubated at 70°C for 20 min and cooled prior to adding to the FMs. Neutrophil CM was incubated at 70°C for 20 min without DNase I as the baseline control. Incubation of activated neutrophil CM at 70°C for 20 min without DNase I did not affect its ability to alter FM function. After 24 h of coculture, the neutrophil CM and any associated factors was removed, and FMs were transferred to a serum-free OptiMEM culture system for an additional 24 h to collect freshly secreted FM-derived factors for downstream analysis. FM supernatants and tissues are collected and stored at −80°C until analysis.

The effects of neutrophil CM on FM caspase 3 activity was measured using the Caspase-Glo assay (Promega, Madison, WI). Briefly, 10 μg of whole cell lysates were incubated at room temperature in the dark for 1 h with the caspase 3 substrate. Following incubation, luminescence was measured using a TD-20/20 luminometer (Turner Designs, Sunnyvale, CA). The amount of luminescence detected as relative light units was proportional to caspase activity. All samples were assayed in triplicate.

The effects of neutrophil CM on FM viability were examined by measuring lactate dehydrogenase (LDH) release in the FM supernatants using the CyQUANT LDH Cytotoxicity Assay, according to the manufacturer’s instructions (Thermo Fisher Scientific, Waltham, MA). All samples were assayed in duplicate.

FM release of the following cytokines G-CSF (DY214), GRO-α (DY275), IL-1β (DY201), IL-6 (DY206), IL-8 (DY208), MCP-1 (DY279), and TNF-α (DY210) and soluble markers of membrane weakening MMP-1 (DY901B), MMP-2 (DY902), MMP-9 (DY911), PGE₂ (KGE004B), TIMP-1 (DTM100), and TIMP-2 (DTM200) were quantified by ELISA in duplicate (R&D Systems, Minneapolis, MN).

Total protein lysates from FM explants were collected by tissue homogenization, and Western blot analysis was performed as previously described (16). Membranes were probed with the following primary Abs from Cell Signaling Technology (Danvers, MA): p p38 MAPK (no. 9211S; 1:10,000 dilution), total p38 MAPK (no. 9212; 1:30,000 dilution), p-ERK (no. 9101; 1:1000 dilution), and total ERK (no. 4695S; 1:2000 dilution). Chemiluminescence was detected using an Amersham Imager 680 (General Electric, Boston, MA), and semiquantitative densitometry was performed using the Gel Logic 100 (Eastman Kodak, Rochester, NY) and Carestream software (Carestream Molecular Imaging, New Haven, CT). Levels of p-protein were normalized against the total amount of that specific protein.

Each experiment was performed at least five times using different FMs and neutrophil CM. The number of experiments performed using independent neutrophil preparations that data were pooled from are indicated in the figure legends as n. All data are reported as mean ± SEM of pooled experiments. Statistical significance was set at p < 0.05 and determined using Prism software (GraphPad, La Jolla, CA). For normally distributed data, significance was determined using either a repeated measures one-way ANOVA for multiple comparisons or a paired t test. For data not normally distributed, significance was determined using a Friedman test followed by a Dunn multiple comparisons test or the Wilcoxon matched-pairs signed-rank test. This is indicated in the figure legends.

To determine whether neutrophil CM affected FM apoptosis, FM explants were exposed to media (NT) or CM from resting peripheral blood neutrophils, neutrophils exposed to LPS, neutrophils exposed to resting FM-CM (FM), or neutrophils exposed to FM+LPS for 24 h. Then, the treated FMs were collected, and a caspase 3 activity assay was carried out to quantify apoptosis. Throughout all experiments, 1 ng/ml of LPS was used as we previously reported that this dose was sufficient to induce a robust cytokine response by human FMs (14) without directly inducing neutrophil activation (11). FM caspase 3 activity was not significantly different in any treatment groups compared with the media control (Fig. 1A). To more globally assess FM viability, LDH release by the FMs treated with neutrophil CM was also quantified. No significant differences in LDH release were observed between the different treatment groups (Fig. 1B).

To determine the effects of neutrophil activation on FM inflammation, FMs were pretreated with media or CM from neutrophils exposed to media and LPS, resting FM-CM, or FM+LPS for 24 h, and then placed into fresh media for 24 h to avoid cytokine carry-over from the neutrophil CM. FM secretion of inflammatory cytokines and chemokines was then investigated by ELISA. We were unable to detect FM secretion of IL-1β or TNF-α under any experimental conditions (data not shown). However, we observed that treatment of FMs with CM from neutrophils activated by FM+LPS significantly increased FM secretion of IL-6 by 14.69 ± 8.93–fold, IL-8 by 6.75 ± 2.33–fold, and GRO-α by 5.09 ± 1.65–fold compared with treatment of FMs with CM from neutrophils exposed to resting FM-CM (Fig. 2). FM secretion of G-CSF and MCP-1 were not significantly affected by treatment with CM from neutrophils exposed to FM-CM compared with control conditions (data not shown).

To investigate whether neutrophil activation can affect FM weakening, FMs were treated with media (NT) or CM from resting neutrophils, neutrophils exposed to LPS, neutrophils exposed to resting FM-CM (FM), or neutrophils exposed to FM+LPS. Exposure of FMs to CM from neutrophils treated with resting FM-CM significantly increased MMP-9 secretion compared with untreated FMs (Fig. 3C) but did not significantly affect FM release of MMP-1, MMP-2, TIMP-1, TIMP-2, or PGE₂ compared with untreated FMs (Fig. 3). In contrast, exposure of FMs to CM from neutrophils activated with FM+LPS significantly increased FM release of MMP-9 by 2.51 ± 0.49–fold (Fig. 3C) and PGE₂ by 1.97 ± 0.40–fold (Fig. 3F) compared with FMs exposed to CM from neutrophils treated with resting FM-CM. Interestingly, exposure of FMs to CM from neutrophils activated with FM+LPS also significantly increased FM secretion of TIMP-1 compared with FMs exposed to CM from neutrophils exposed to resting FM-CM, but levels were not significantly different compared with untreated FMs (Fig. 3D). CM from neutrophils activated with FM+LPS did not significantly affect FM release of MMP-1, MMP-2, or TIMP-2 compared with treatment with CM from resting neutrophils or untreated FMs (Fig. 3).

To investigate the intracellular signaling pathway activated in FMs by the neutrophil CM, Western blotting for key inflammatory pathways was performed using FM lysates. We did not observe any changes in the levels of FM p-p38 MAP kinase compared with total p38 protein or p-p65 compared with total p65 (data not shown). However treatment of FMs with CM from neutrophils exposed to FM+LPS significantly increased FM p-ERK/total ERK levels by 4.52 ± 1.39–fold compared with FMs treated with CM from neutrophils exposed to resting FM-CM (Fig. 4A). When SCH772984, which specifically blocks ERK activity, was added to the FM culture system, the increased release of IL-8 and MMP-9 by FMs in response to CM from neutrophils activated by FM+LPS was significantly reduced. SCH772984 inhibited FM secretion of IL-8 by 26.9 ± 9.1% and MMP9 by 27.1 ± 6.4% in response to activated neutrophil CM (Fig. 4). SCH772984 alone or in the presence of CM from neutrophils exposed to resting FM-CM did not significantly affect FM secretion of IL-8 or MMP-9 (Fig. 4B, 4C).

To examine whether NETs released from activated neutrophils played a role in modulating FM function, DNase I was employed to degrade NETs from the neutrophil CM prior to culturing with FM explants. The presence of DNase I did not significantly affect the elevated FM secretion of IL-8 induced by activated neutrophil CM (Fig. 5A). However, the presence of DNase I significantly reduced the increase in FM secretion of MMP-9 by 49.0 ± 6.3% induced by this activated neutrophil CM (Fig. 5B). Exposure to DNase I alone or in treatments with CM from neutrophils exposed to resting FM-CM did not significantly affect FM secretion of IL-8 or MMP-9 (Fig. 5). Because cell-free DNA can activate TLR-9 and we have previously shown that FMs express this receptor (15), we investigated whether NETs in the neutrophil CM could activate FM TLR-9 to induce the downstream inflammatory and tissue weakening phenotypes observed. The presence of a TLR-9 antagonist, ODN2088, significantly reduced the ability of activated neutrophil CM to induce increased FM secretion of MMP-9 by 56.2 ± 10.5% (Fig. 5B); however, TLR-9 antagonism did not affect the increased FM IL-8 secretion (Fig. 5B).

Preterm birth is the leading cause of neonatal mortality and continues to be a major healthcare challenge worldwide. Despite intense research, the etiologies and pathogenesis of preterm birth remain unclear, although intrauterine infection is an important cause and chorioamnionitis is commonly reported in 40–70% of preterm birth cases. Even in the absence of preterm birth, chorioamnionitis can significantly affect fetal development in utero (17–20); thus, it is key to understand how chorioamnionitis occurs and the downstream effects of neutrophil recruitment on FM function. This study employed a reductionist approach using bacterial LPS to broadly model a Gram-negative bacterial infection at the maternal–fetal interface to elucidate the mechanisms of interactions between FMs and neutrophils that may lead to PPROM and preterm birth.

Using this model, we had previously characterized the direct responses of human FMs to bacterial LPS (14, 15) as well as how this leads to neutrophil recruitment and activation (11). We reported that FM+LPS release a wide range of chemotactic signals that leads to the recruitment and activation of neutrophils that can then release a large range of chemokines and proinflammatory cytokines at high levels as well as degranulate and release NETs through a vital pathway (11). In this current work, we further build upon those observations to show that this activated neutrophil phenotype deleteriously affects FM function, specifically by augmenting FM inflammation and inducing markers of FM weakening, which may contribute to PPROM and preterm birth. To specifically study how the activated neutrophil phenotype can affect FM function, neutrophils were first preactivated with FM+LPS, which was subsequently removed before generating the activated neutrophil CM. Similarly, to assay the FM response to this activated neutrophil CM, FM were exposed to neutrophil CM for 24 h before changing into fresh media to quantify the secretion of FM-derived factors rather than neutrophil-derived factors. Although the authors recognize that a coculture model may be more physiological, this approach allows us to specifically determine the effects of FM-induced neutrophil activation on subsequent FM function, which would otherwise be missed/inconclusive using other methods. This method also enables us to model what may be happening at the maternal–fetal interface during low-grade infections/inflammation as we have demonstrated that human FMs are able to respond to lower concentrations of LPS than neutrophils (11, 14).

Many proinflammatory cytokines have been implicated in the pathogenesis of preterm birth, including IL-1β, IL-6, IL-8, and TNF-α (21–25). In this study, we reported that the exposure of FMs to CM from FM+LPS-stimulated neutrophils induced the upregulation of a very specific inflammatory cytokine signature, including IL-6, IL-8, and GRO-α but not the classic inflammatory cytokines IL-1β or TNF-α (24, 26). IL-8 and GRO-α are important chemotactic factors that can augment neutrophil recruitment to the maternal–fetal interface whereas IL-6 is a proinflammatory cytokine that can also facilitate neutrophil chemotaxis and prime neutrophils for activation (27, 28). Moreover, IL-8 has also been reported to be able to induce NET release by neutrophils in a paracrine manner (29). Thus, neutrophils may play a key role in propagating inflammation at the FMs during an infection by amplifying their own recruitment and activation. Indeed, although it has previously been reported that neutrophil depletion cannot prevent preterm birth in mice induced by high-dose bacterial LPS or E. coli (30, 31), neutrophils do contribute to utero–placental inflammation in vivo (30).

In this study, we also demonstrated that although neutrophil CM did not affect FM viability, FM-activated neutrophils were able to specifically upregulate MMP-9 and PGE₂ release by FMs that may lead to FM weakening. In contrast, the secretion of MMP-1, MMP-2, and TIMP-2 by FMs was not affected by exposure to activated neutrophil CM. Along with MMP-2, MMP-9 is a gelatinase that degrades collagen IV, the major structural component of the FMs, prior to parturition (32, 33). Similar to our observations, the expression levels of MMP-9, but not MMP-2, have been reported to increase in the FMs (34) and sera (35) of women with preterm labor, whereas another study reported that both MMP‐2 and ‐9 activities are dramatically increased in human FMs during labor and are higher at preterm birth (36). MMP-9 levels in the FMs and amniotic fluid can also be increased in response to TNF-α (37), LPS (37), or microbial invasion of the amniotic cavity (38), strongly implicating this MMP in the pathogenesis of preterm birth in women with infection or chorioamnionitis.

Although we observed that there was a significant increase in FM TIMP-1 secretion in response to CM from FM-activated neutrophils compared with FMs treated with CM from neutrophils exposed to resting FMs, this was due to a reduction in TIMP-1 secretion with the latter treatment rather than the activated neutrophil CM stimulating increased de novo secretion. This lack of change in TIMP levels in response to neutrophil CM came as a surprise to us as these are the main TIMP activity, and it has been reported that during FM rupture in both normal or pathological parturition, TIMP-1 and TIMP-2 concentrations are reduced, whereas the level of active MMP-9 is increased (32). Conversely, it has also been reported that placental TIMP-1 and TIMP-2 expression levels were not altered in women delivering preterm compared with delivery at term (39), so it appears that TIMP levels may be variable depending on the timing of sampling and the tissue studied. That TIMP levels are unchanged whereas MMP-9 levels are increased in FMs exposed to CM from activated neutrophils correlates with clinical observations in which women with preterm birth were reported to have increased circulating MMP-9:TIMP-1 and MMP-9:TIMP-2 ratios compared with gestational age-matched controls (35) and supports the notion that activated neutrophil CM can induce FM weakening.

PGs are key mediators of parturition that can stimulate myometrial contractility, cervical remodeling, and extracellular matrix degradation, leading to rupture of the FMs and preterm birth (40–42). PGE₂ is one of the main phosphatidylglycerols found in amniotic fluid, and levels have been reported to be higher in women with preterm labor and intraamniotic infection than in women without infection (42). In this study, we found that whereas FMs did not directly respond to low-dose LPS by increasing their secretion of PGE₂, treatment of FMs with CM from neutrophils that were previously activated with LPS-stimulated FMs significantly upregulated PGE₂ expression, suggesting that the FM PGE₂ response to infection may be secondary through the recruitment and activation of neutrophils. Interestingly, it has been reported that PGE₂ can upregulate MMP9 expression in FMs in vitro (33), potentially forming a feed-forward loop for FM weakening.

To further elucidate the signaling mechanism activated in FMs in response to neutrophil CM, we investigated the canonical inflammatory pathways, NF-κB and p38 MAP kinase. Neither pathways were activated in FMs after treatment with neutrophil CM for 48 h; however, the ERK MAP kinase signaling pathway was activated. Indeed, ERK and downstream RAS signaling has been associated with preterm birth (43), and the ERK pathway can also be activated by the stretch in the myometrium that is key during labor and parturition (44). This pathway appears to be responsible for both the FM inflammatory and weakening effects of activated neutrophil CM. In these experiments, we focused on IL-8 and MMP-9 secretion by FMs, if these responses were robust in response to activated neutrophil CM, and if these factors were crucial for neutrophil recruitment/activation (causing FM inflammation) and FM weakening, respectively. Because we have previously reported that NETs are present in this neutrophil CM, and if insufficiently cleared, NETs may act as danger signals (45), we investigated whether the NETs in the activated neutrophil CM were responsible for the FM inflammatory and membrane weakening responses. Cell-free DNA that constitute NETs have been reported to be proinflammatory by activating the DNA sensor TLR-9 (46, 47). Indeed, degradation of NETs using DNase I or inhibition of TLR-9 activation significantly reduced FM MMP-9 release in response to activated neutrophil CM but, interestingly, did not significantly affect the FM IL-8 response. In a model of liver ischemia/reperfusion injury, the use of PAD4 inhibitor to inhibit NET formation or DNase I to degrade NETs can protect hepatocytes from injury and death (48), raising the interesting possibility that perhaps a similar treatment may be possible to delay PPROM and preterm birth in women with intrauterine injection or chorioamnionitis.

Throughout these studies, it was surprising to observe that CM from neutrophils exposed to resting FMs did not affect FM function at all, despite having shown that neutrophils respond to resting FMs by secreting low levels of cytokines as well as NETs (11). In fact, at times, neutrophils preconditioned with resting FM-CM induced less of a response than neutrophils isolated directly from the peripheral blood and not exposed to FM-CM, suggesting that whereas resting FMs do recruit neutrophils, resting FMs may be actively keeping them quiescent, despite inducing vital NET release. This observation suggests that the composition of the NETs released in response to resting FM-CM may be different to those released in response to FM+LPS. That resting neutrophil CM did not affect FM viability or cytokine secretion supports our hypothesis that during normal pregnancy, neutrophils play a surveillance role at the FMs to protect the tissue against pathological infections. Indeed, we observed that neutrophils and NETs are found in low numbers in normal murine placentae at gestational day 15.5 (11), and these structures have also been reported in women in vivo (12).

In summary, this study provides the first evidence, to our knowledge, of a direct effect of neutrophil activation on human FM function by activation of FM ERK signaling and demonstrated that NETs may be perceived as a danger signal by FM TLR-9. Exposure of FMs to activated neutrophil CM augmented FM inflammation and markers of weakening in the presence of a bacterial infection. Thus, increased neutrophil recruitment and activation in chorioamnionitis may act in a feed-forward mechanism to contribute to FM inflammation and weakening, leading to increased risk of PPROM in women with intrauterine infection.

We thank all the donors and staff of Yale-New Haven Hospital and the Yale University Reproductive Sciences Biobank for the donation of blood and FM for this study.

The authors have no financial conflicts of interest.