Critical Paracrine Interactions Between TNF-α and IL-10 Regulate Lipopolysaccharide-Stimulated Human Choriodecidual Cytokine and Prostaglandin E₂ Production¹

Preterm labor resulting in delivery occurs in ∼5–10% of all pregnancies (1). Despite advances in the understanding of the mechanisms that result in preterm birth, this rate has remained relatively constant for decades. Between 30 and 70% of very preterm births are associated with an ascending intrauterine infection in which microorganisms, originating from the vagina, rise through the choriodecidua and subsequently colonize the chorion, amnion, amniotic fluid, and ultimately the fetus (2). In response to the infective pathogen, the maternal immune system initiates an inflammatory response that frequently results in the onset of preterm labor.

The choriodecidua is a tissue composed of interdigitating fetal and maternal cells. In an ascending intrauterine infection it is the first tissue colonized by the microbial pathogen and is the main barrier to progression of the infection into the amniotic cavity. The response of the choriodecidua to the presence of the microorganism is likely an integral part in determining the severity, extent, and consequences of the infection.

Human gestational membranes produce a number of proinflammatory cytokines (e.g., IL-1β, TNF-α, IL-6, and IL-8) and PGs, both constitutively and in response to inflammatory stimuli and bacterial cell wall products such as LPS (3, 4, 5, 6). At parturition, both at term as well as preterm, there is an increase in the production of these proinflammatory mediators within the uterus, but in the case of intrauterine infection, this response is significantly increased (4, 7, 8). Previous studies have shown that LPS treatment of pregnant mice leads to an increase in IL-1β and PGE₂ production by decidual caps in vitro (9, 10) and preterm delivery (3). Mice treated with LPS also have elevated circulating levels of both TNF-α and IL-10 (11). These findings indicate that cytokines and PGs produced by gestational tissues as a result of LPS treatment might play an important role in the initiation of preterm labor leading to delivery.

In addition to proinflammatory cytokines, gestational tissues can also produce anti-inflammatory cytokines such as IL-10 (12, 13). IL-10 is a potent inhibitor of the production of IL-1α, IL-6, IL-8, and TNF-α by human monocytes and macrophages (11). It is produced by chorion, decidual, and placental tissues (14, 15, 16, 17, 18, 19, 20) and has been detected in the amniotic fluid of women during late gestation (21, 22, 23). IL-10 production by decidual cells in vitro has been reported to increase upon treatment with IL-1β and bacterial cell wall products (14, 15, 16, 17). IL-10 inhibits cytokine and PG production by human chorion, decidual, and placental cells in vitro (24, 25, 26, 27, 28, 29), although the effects of endogenous IL-10 on the local inflammatory response have not been determined. However, it has been reported that treatment with IL-10 prevented LPS-induced preterm delivery in mice (30) as well as rats (31). IL-10 thus has therapeutic potential for the treatment of intrauterine infection-associated preterm labor.

The present studies were conducted to characterize auto regulatory interactions between pro- and anti-inflammatory cytokines in choriodecidual explants upon stimulation with LPS. Although it has been established that LPS elicits an inflammatory response within gestational tissues, the significance and role of local mediators in the elaboration of the response is not clear. Potential targets for intervention could potentially be identified if they werefound to be key mediators in the response. Explants were chosen as the model of study to ensure that the structural and cellular architecture of the choriodecidual membrane was maintained.

DMEM-199 culture was obtained from Irvine Scientific (Santa Ana, CA). FCS and streptavidin-alkaline phosphatase were purchased from Invitrogen (Auckland, New Zealand). Bovine γ-globulin and LPS (serotype 055:B5) were purchased from Sigma-Aldrich (St. Louis, MO). Human IL-1β was a generous gift of Immunex (Seattle, WA). IL-1R antagonist (IL-1Ra)³ was a gift from Synergen (Boulder, CO). For experimental studies, the human rTNF-α was provided by Dr. J. Fraser (Department of Molecular Medicine, University of Auckland, Auckland, New Zealand). Recombinant IL-1β and anti-IL-1β antiserum were purchased from R&D Systems (Minneapolis, MN). Abs to IL-10 and TNF-α were purchased from BD PharMingen (San Diego, CA). Tritiated PGE₂ was purchased from Amersham Pharmacia Biotech (Aylesbury, U.K.). Recombinant IL-10 and TNF-α used to calibrate the ELISAs were purchased from R&D Systems.

All procedures involving human placentas were approved by the Auckland Ethics Committee. Placentas were obtained with informed consent from women undergoing elective Cesarean section at term before the onset of labor. After manually removing the amnion, choriodecidual membranes were washed carefully in medium to remove residual RBCs without causing damage to the integrity of the membrane. Tissue explants (6-mm disks) were excised with a cork borer as described previously (32). Explants were pooled and randomly distributed into six-well plates (six explants per well, three wells per treatment) containing medium supplemented with 10% FCS and antibiotics (33). The explants were allowed to equilibrate overnight at 37°C in a humidified atmosphere of 5% CO₂/95% air. The following day, media were replaced with serum-free media containing 0.1% bovine γ-globulin and antibiotics. Explants were then treated with the various test substances or the appropriate vehicle control. At the indicated time points, the media were collected. Production rates were normalized to the wet weight of the explants in the individual wells.

Commercially available neutralizing Abs were used to neutralize the effects of the LPS-stimulated TNF-α and IL-10. The Ab at various concentrations was added concurrently with the LPS and supernatants were harvested 24 h later. The explants were pretreated with the IL-1Ra for 1 h before the addition of LPS and supernatants were harvested 24 h later. The concentrations selected were chosen according to manufacturer’s specifications for neutralizing the cytokines (34, 35).

IL-1β, IL-10, and TNF-α were measured by ELISA as previously described. (32, 33). PGE₂ production was determined by RIA as described previously using an antiserum generated in our laboratory (36).

Cytokine and PG production rates, calculated as picograms per milligram of wet weight/24 h, are represented as a percent of control for each experiment (mean ± SEM). Results are presented as pooled data from multiple experiments performed in triplicate. Statistical significance was determined by ANOVA followed by Dunnett’s test. A value of p < 0.05 was considered to be significant.

Results presented in Fig. 1 show the time course of choriodecidual IL-1β (Fig. 1,a), IL-10 (Fig. 1,b), TNF-α (Fig. 1,c), and PGE₂ (Fig. 1,d) production at 2, 8, 12, and 24 h after stimulation with LPS (5 μg/ml). The production of IL-1β was significantly elevated compared with control at the 8 h time point, reaching maximum levels at 12 h post-LPS. PGE₂ production followed a similar time course to that of IL-1β. IL-10 production was significantly elevated 4 h after LPS stimulation and maximal production rates were observed 8 h after stimulation. TNF-α production was maximal 4 h after LPS stimulation and subsequently decreased by approximately one-third over 4–24 h; production of TNF-α at 24 h post-LPS stimulation was similar to that observed at the 2-h time point. The basal production of TNF-α leveled off after 8 h and remained constant for the subsequent time points studied. The basal production of the other cytokines as well as PGE₂ also showed similar patterns. This pattern of TNF-α production remained evident when the data were expressed as picograms per milligram of tissue rather than normalized to control (Fig. 2).

LPS induced a significant increase in IL-1β, IL-10, and PGE₂ production. Immunoneutralization of TNF-α bioactivity resulted in a significant reduction in LPS-stimulated production of IL-1β (Fig. 3,a), IL-10 (Fig. 3,b), and PGE₂ (Fig. 3 c) over a 24-h incubation period. A significant reduction of IL-1β production was observed with the TNF-α neutralizing Ab at a concentration of 3 μg/ml, while LPS-stimulated production rates of IL-10 and PGE₂ were significantly reduced at 1–3 μg/ml anti-TNF-α Ab.

The effects of IL-10 immunoneutralization on cytokine production by LPS-stimulated choriodecidual explants are presented in Fig. 4. Although there was no significant effect on LPS-stimulated IL-1β production over the 24-h culture period (Fig. 4,a), IL-10 neutralization resulted in a significant increase in the production of TNF-α (Fig. 4,b), and PGE₂ (Fig. 4,c) at all concentrations of anti-IL-10 neutralizing Ab tested (0.3–3 μg/ml). In the case of TNF-α, neutralization of IL-10 resulted in a 4-fold increase in the production of TNF-α after 24 h stimulation with LPS. The addition of exogenous IL-10 (25–100 ng/ml) resulted in a decrease in IL-1β production (Fig. 5,a). The inhibitory effects of IL-10 on TNF-α production were more pronounced (Fig. 5,b), with all concentrations of IL-10 tested (1–100 ng/ml) significantly inhibiting TNF-α production. The effect was concentration-dependent, with the higher concentrations of IL-10 exerting the greatest degree of inhibition. Addition of IL-10 had no effect on PGE₂ production by LPS-stimulated choriodecidual explants (Fig. 5 c).

Addition of IL-1Ra had no statistically significant effect on production of IL-10 (Fig. 6,a) or TNF-α (Fig. 6 b) by LPS-stimulated choriodecidual explants over 24 h. However, there was a significant reduction (40%) in PGE₂ production at 1 μg/ml IL-1Ra.

The resident cells in the choriodecidua are of pivotal importance in the inflammatory response elicited by microbial colonization of the gestational membranes. In tissues with histological chorioamnionitis, the chorion and decidua become infiltrated with leukocytes, which in some cases accumulate at the margin between the amnion and chorion (37). There is little doubt that the cytokines released by choriodecidual membranes upon LPS stimulation play a role in the elaboration of the inflammatory response; however, whether they are principal drivers of the inflammatory reaction or secondary modulators has been a matter for debate (38). Because both positive and negative regulators of the inflammatory reaction are produced in response to infection, the degree of stimulation observed would presumably depend upon the balance of pro/anti-inflammatory cytokines released in response to contact with microbial products, which in this study was modeled using a bacterial LPS preparation from Escherichia coli. E. coli is one of the most common bacteria associated with neonatal infections (39) and infection of the amniotic cavity (1, 40) The results presented in this study strongly argue that in the choriodecidua, cytokines are involved in both positive and negative feedback loops to modulate the inflammatory response to bacteria, which in turn regulates the production of uterotonic PGs.

We have previously reported that the exogenous treatment of choriodecidual explants with TNF-α and IL-1β results in increased production of both PGE₂ and IL-10 after 24 h (32, 41). These results confirmed the immunomodulatory role of TNF-α and IL-1β in the regulation of anti-inflammatory cytokines as well as PGs. Results reported in this study demonstrated that the TNF-α produced by choriodecidual explants in response to LPS exposure is a crucial modulator of further cytokine and PG production. TNF-α is a potent proinflammatory mediator and has been implicated as the primary cause of many immunopathologies ranging from arthritis (42, 43, 44) to septic shock (45, 46), in addition to preterm labor (47). Mice treated with Abs or antagonists to TNF-α are subsequently protected from LPS-induced death (45, 48, 49). Our data would suggest that, in the choriodecidua, the inflammatory response elicited by bacteria might be mediated and modulated to a large extent by the actions of endogenous cytokines. TNF-α is the critical stimulating factor in regulating choriodecidual IL-1β, IL-10, and PGE₂ production, although it should be noted that neutralization of TNF-α did not result in a complete inhibition of LPS-stimulated cytokine and PG production. The concentrations of the Abs used were chosen based on the manufacturer’s recommendations (34, 35) and confirmed by removal of immunodetectable TNF-α in preliminary experiments. However, the lack of complete inhibition could reflect inadequate penetration of the neutralizing Abs as they were added concurrently with the LPS treatment. Additionally, cytokine measurements reflect the concentrations present in the media, whereas the localized concentrations of the cytokines within the tissues may be significantly higher. Alternatively, the LPS treatment could have resulted in the production of other as yet unidentified factors that may also exert proinflammatory actions. IL-1β may be a candidate factor because it is a potent proinflammatory cytokine that has been shown to be released by decidual cells and to increase PG production by gestational tissues through increased expression of PGH synthase (PGHS)-II (50). Results presented in this study demonstrated that IL-1β production was significantly increased 8 h after LPS stimulation, while PGE₂ production did not reach a maximum until 12–24 h after LPS stimulation. Therefore, the IL-1β produced may contribute to PGE₂ production as it is present in significantly increased levels 16 h before the medium was harvested. Antagonism of IL-1 activity by IL-1Ra resulted in a decrease in PGE₂ production, supporting this conclusion. Immunoneutralization of IL-10 resulted in an increase in PGE₂ production. However, addition of exogenous IL-10 failed to exert any effects on PGE₂ production despite a reduction in IL-1β and TNF-α production. These apparently conflicting data may be explained by the possibility that IL-10 may not be having a direct effect itself on PGE₂ production, but that an additional factor(s) may be produced upon LPS treatment of the choriodecidual explants. These data are not consistent with those of others showing inhibition of PG production by IL-10 in gestational membranes (51, 52).

The anti-inflammatory effects of IL-10 have been well-documented (12). Previous studies have demonstrated that IL-10 can decrease TNF-α production in human macrophages and monocytes (53). Cassatella et al. (13) demonstrated that IL-10 diminished the levels of TNF, IL-1-β, and IL-8 mRNA late after the onset of stimulation of polymorphonuclear leukocytes with LPS. Previous studies in gestational tissues indicated that IL-10 inhibits basal- as well as LPS-stimulated PGE₂ production by intact fetal membranes (52). In the present study, we observed that IL-10 was produced at the time when TNF-α levels were beginning to decrease, and one interpretation of these data is that the effects of IL-10 on PGE₂ production are mediated by its inhibition of TNF-α production. It has been reported that the inhibitory effects of IL-10 on TNF-α production are independent of its ability to suppress the effects of NF-κB activation (54, 55). Studies conducted on human monocytes demonstrated that IL-10 can down-regulate surface expression of the TNF receptor while increasing production of the soluble TNF-α receptor, in addition to inhibiting the release of TNF-α itself (56, 57). The effects of IL-10 on TNF-α receptor expression in human gestational tissues are unknown at this time. It is possible that IL-10 could suppress TNF-α receptor expression in addition to decreasing TNF-α production, thus exerting a more complete antagonism of TNF-α effects. IL-10 has also been reported to increase IL-1Ra production from LPS-stimulated human polymorphonuclear leukocytes (58) as well as the decidua (59), offering the possibility of an additional route of action of IL-10 in the gestational membranes.

IL-10 has been shown to inhibit PGHS-II mRNA expression in human monocytes (60) and neutrophils (61), and also in IL-1β- and TNF-α-stimulated trophoblast cells (51). However, PG activity can also be regulated by catabolism. 15-Hydroxyprostaglandin dehydrogenase (PGDH) is an NAD-dependent cytoplasmic enzyme that reduces the biological activity of PGs by catalyzing the first reaction in the catabolic pathway for PG degradation and inactivation. PGDH is expressed and functions in human placenta and fetal membranes (62, 63). A reduction of chorion PGDH activity and expression has been demonstrated in association with preterm birth, particularly in pregnancies with intrauterine infection (64). In villous and chorionic trophoblasts, IL-1β increases levels of PGHS-II mRNA and decreases PGDH mRNA in villous trophoblasts; both of these effects can be reversed by IL-10 (51). Hence, in addition to its effects mediated by TNF-α inhibition, IL-10 could also both decrease the synthesis and increase catabolism of PGs in these tissues.

The results of the present studies are summarized in the schematic pathway presented in Fig. 7. Upon LPS exposure, the human choriodecidua membranes initially produce TNF-α, which in turn induces the production of IL-1β, PGE₂, and IL-10. Endogenous IL-1β potentiates the production of PGE₂ while IL-10 inhibits the production of TNF-α and PGE₂. The choriodecidua is composed of cells of both fetal and maternal origin. Previous studies have characterized cytokine production by chorion and decidual cells and also identified the localization of cytokine receptors present (7, 47, 50, 59). Apart from resident cells of these tissues, other potential sources of cytokine production are the lymphocytes that infiltrate as the onset of labor approaches (65, 66, 67). All the placentas used in this study were from patients undergoing elective Cesarean sections at term. However, it is feasible that decidual lymphocytes may have contributed to cytokine and PG production and interacted with the resident cells of the choriodecidua to modulate the responses.

From our data we conclude that the paracrine/autocrine role of choriodecidual TNF-α appears to be pivotal in the elaboration of the inflammatory response to LPS. These studies suggest that pharmacological agents that inhibit TNF-α synthesis or actions might be an effective therapeutic intervention aimed at preventing or delaying infection-associated preterm delivery. IL-10 may be such an agent as results presented in this study demonstrated that itcan inhibit TNF-α and IL-1β production by LPS-stimulated choriodecidual membranes and is also involved in the suppression of PG production.

We thank the theater staff at National Women’s Hospital for the collection of placentas, with a very special thanks to Oliva Tupusi for the organization of patient consent. We also thank and acknowledge Elizabeth Robinson (Department of Community Health, University of Auckland, New Zealand) for her advice on the statistical analysis of the data.