T. Lund PhD, Department of Immunology and Molecular Pathology, UCL, 46 Cleveland Street, London W1T 4JF, UK. E-mail: email@example.com
A number of immunomodulatory molecules are present in the placenta, including cytokines, prostaglandins, progesterone and indoleamine 2,3-dioxygenase. An undefined factor capable of down-regulating T-cell activity has recently been reported  as being produced by short-term cultures of placental fragments. By careful repetition of these studies we have confirmed that chorionic villi isolated from term placenta produce a low molecular weight, heat stable factor capable of inhibiting the IL-2-dependent proliferation of mouse CTLL-2 cells. This activity was not due, however, to a previously unknown immunosuppressive molecule, but rather to prostaglandin E2 (PGE2). Expression of cyclooxygenase (COX)-2 was detected in the syncytiotrophoblast of chorionic villi explants using immunohistochemistry. Culture of the explants in the presence of the COX-1/COX–2 inhibitors indomethacin and diclofenac, or with the COX-2-selective inhibitor DFP, blocked the production of the immunosuppressive factor. The immunosuppressive activity was restored by adding PGE2 to the supernatants obtained from diclofenac-inhibited explants. A number of different receptors are involved in mediating the biological effects of prostaglandins. By utilizing selective antagonists of individual receptors, we have established that the immunosuppressive effect of PGE2 on CTLL-2 cells is exerted via the EP4 receptor. Thus, addition of an EP4-selective antagonist, but not of EP1 or EP3 antagonists, abolished the immunosuppressive effect of PGE2 on CTLL-2 cells. This may have implications for attempts to selectively manipulate T-cell responses.
During pregnancy the lack of an overt immunological attack on the semiallogeneic fetus defies the normal rules for transplant rejection . The mother encounters fetal cells and therefore paternal antigens during pregnancy, but her immune system is rendered transiently tolerant to them . There is evidence that during pregnancy the immune response of the mother is shifted from a Th1 cell-mediated immune response towards a more humoral Th2-directed response [4–8]. Although T-cell numbers in the uteroplacental tissues decrease early on during pregnancy, large numbers of CD3–CD16–CD56+ large granular lymphocytes are present and these, together with non-leukocytic placental cells, secrete a number of cytokines including IL-4, IL-5 and IL-10, thus suppressing differentiation of Th0 cells into Th1 cells and favouring formation of Th2 cells [9,10].
In addition to cytokines, a number of other mediators directly affect the immune system during pregnancy, including indoleamine 2,3-dioxygenase [11–13] and prostaglandin E2 (PGE2) [14,15]. PGE2 can interact with a number of different specific receptors, some of which elevate the intracellular level of cAMP which, in lymphocytes, has immunomodulating properties which can lead to anergy [16,17]. Thus it has been shown that PGE2 is capable of inducing T cell anergy to specific antigens [18,19]. Dendritic cells undergoing maturation in the presence of PGE2 are polarized toward IL-12-deficient cells that direct the differentiation of naive T cells into antigen-specific Th2 cells [20–22]. However, the effect of PGE2 may vary depending on the mode of T cell activation. Thus, in contrast to the above findings, Reiser and colleagues  have demonstrated that PGE2 can be a potent enhancer of IL-12 production by human dendritic cells. PGE2 also affects T cell differentiation directly by inhibiting the production of Th1 cytokines including IL-2 and IFN-γ without affecting the production of the Th2 cytokines IL-4, IL-5 and IL-10 [14,15,17,24–26].
The placenta may produce other, uncharacterized, factors capable of modulating the immune response. Chaouat and his colleagues demonstrated that chorionic villi from human term placenta, trophoblasts, and choriocarcinoma cell lines secrete molecules capable of suppressing the proliferation of CTLL-2 and other T cells, inhibiting mixed lymphocyte reactions, and preventing graft versus host reactions in mice [1,27–29]. A major active component of the immunosuppressive activity was an undefined heat-stable molecule of less than 3 kDa. Chaouat proposed that this factor could, under appropriate conditions, bind to proteins produced by the placenta or elsewhere, thereby explaining the fact that molecules such as human chorionic gonadotrophin and α-fetoprotein have been reported as having immunosuppressive properties which disappear when the protein is highly purified or used in recombinant form . We have investigated the immunosuppressive material produced by the chorionic villi of term placenta and report here that the factor suppressing the IL-2-dependent proliferation of CTLL-2 cells is PGE2. This function is exerted through the EP4 family of receptors.
MATERIALS AND METHODS
Preparation of human placental supernatants
Human placental supernatants (HPS) were obtained as described by Menu . Briefly, human placentas were obtained at term from caesarean deliveries. Chorionic villi were isolated from surrounding tissue and further cut with scissors to obtain 1–3-mm3 pieces. The fragments were washed five times in serum free RPMI-1640 culture medium (GIBCO BRL, Paisley, UK). Ex-plants of chorionic villi were cultured at 37°C in 5% CO2, using 25 cm3 tissue culture flasks with 20–30 fragments per 10–15 ml serum-free RPMI-1640 supplemented with 100 Units/ml penicillin and 100 μg/ml streptomycin (GIBCO BRL) and 25 μM 2-mercaptoethanol. Supernatants were collected after 48 h, centrifuged for 20 min at 30 000 g at 4°C to remove particulate material and stored in aliquots at –80°C until use. To obtain low molecular weight material supernatants were ultrafiltered using a Centriprep-3 centrifugal filter unit (Millipore, Watford, UK) with a 3-kDa cut-off. HPS obtained from different placentas are designated using numbers (HPS1-HPS17) and the numbering of the low molecular weight fitrates correspond to the HPS from which they were obtained. In some experiments the filtrate was heated at 100°C for 2h prior to use.
In experiments investigating the effect of cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2) inhibition, the chorionic villi explants were prepared as above and after the washing procedure prior to culture were incubated for 1 h in the presence of indomethacin (2 μg/ml), diclofenac (3 μg/ml) (both from Sigma, Poole, UK) or the COX-2-selective inhibitor DFP 3-(2-propyloxy)-(4-methyl-sulphonylphenyl)-(5,5-dimethyl)-furanone] (1 μg/ml) (a gift from Merck-Frosst, Canada). The explants were then washed again and cultured in fresh medium supplemented with the same amounts of indomethacin, diclofenac or DFP. Following 48h of culture the supernatants were harvested and processed as above.
Immunoregulatory activity of placental supernatants
Placental supernatants were tested for their immunosuppressive activity using an IL-2-dependent CTLL-2 cell proliferation assay. CTLL-2 cells were grown in RPMI-1640 containing 2 mML-glutamine (GIBCO BRL), 100 Units/ml penicillin and 100 μg/ml streptomycin, 25 μM 2-mercaptoethanol, 10% heat-inactivated fetal calf serum (HIFCS) and 10 ng/ml recombinant human interleukin-2 (rhIL-2) (Peprotech Inc, Rocky Hill, NJ, USA). Prior to use, the CTLL-2 cells were washed in modified Eagle’s medium (MEM, GIBCO BRL) containing 2% HIFCS and incubated for 2–3 h without IL-2. The cells were then cultured (5 × 104 cells/ml) in 96-well flat-bottomed microtitre plates in a final volume of 100 μl with a final concentration of 0·6 ng/ml of rhIL-2 and in the presence of placental supernatant or filtrate. Following incubation for 24 h, their proliferation was determined using a 4-h pulse of 20 μl per well of 5 mg/ml MTT (3-4,5-Dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide) (Sigma) in PBS, or 1 μCi per well 3H]-thymidine (Amersham Pharmacia Biotech). The MTT reactions were stopped by adding 100 μl per well of 10% sodium dodecyl sulphate in 0·01M HCl, and after an overnight incubation the amount of formazan produced was measured using a dual wavelength (570 and 630 nm) microtitre plate reader.
In some experiments CTLL-2 cells were preincubated with the PGE2 receptor antagonists L-818 638, L-826 266 and/or L-161 982  (each at 1 μM/ml, a gift from Merck-Frosst, Canada) for 1·5 h prior to the addition of either PGE2 or the placental supernatant or filtrate.
Measurement of PGE2 in placental supernatants
The concentration of PGE2 was measured using a standard radioimmunoassay. Briefly, PGE2 standards (100 μl, prepared in assay buffer comprising 50 mM tris pH 7·4, 0·1% gelatin, 0·05% sodium azide) or test samples (1–100 μl) were mixed with 100 μl aliquots of rabbit anti-PGE2 Sigma, UK] in assay buffer; 100 μl aliquots of 3H]PGE2 (Amersham International, Amersham, Bucks, UK), also prepared in assay buffer, were then added to the samples or standards and additional buffer provided to give a final reaction volume of 400 μl. Following vortex mixing, the tubes were left overnight at 4°C. After this incubation period, unbound radioactivity was removed by the addition of 200 μl 50 mM tris buffer pH 7·4 containing 0·02% charcoal and 0·004% dextran. The charcoal mixture was incubated with the standards or samples at 4°C for 10 min Incubates were then centrifuged at 800 g for 10 min at 4°C, and the supernatant containing the antigen–antibody complex decanted and mixed with 4 ml of Picofluor scintillation fluid. Bound radioactivity within the samples was determined using a liquid scintillation counter (Beckman; Type 3801).
Four micron thick cryostat sections of the chorionic villi were fixed using acetone. Following inactivation of endogenous peroxidase activity with methanol/H2O2 and blocking of non-specific binding with 2% normal rabbit serum, the sections were incubated with a 1 : 25 dilution of an IgG1 mouse monoclonal anti-COX-2 (Cayman Chemical Company, Ann Arbor, MI, USA) or control IgG1 antibody for 1 h. Following washing, the binding of the primary antibody was detected by incubation with a 1 : 50 dilution of a peroxidase-labelled rabbit antimouse immunoglobulin (DAKO, Ely, UK). COX-2 expression was localized using the peroxidase substrate DAB and the slides counterstained using Harris’s haematoxylin.
Presentation of the experiments and statistical analysis
All the experiments presented are representative examples. In each case all the data shown are consistent with the results found in all other experiments, with the number of occasions a particular HPS or filtrate was used varying from two to 12 times. Statistical analysis was carried out using unpaired Student’s t-test.
Human placental supernatants inhibit CTLL-2 cell proliferation
The chorionic villi explants from term placenta secreted a factor which was able to inhibit the IL-2-dependent proliferation of CTLL-2 cells in a concentration-dependent manner, as determined using both an MTT reduction assay (Fig. 1a) and 3H]-thymidine incorporation (data not shown). All the human placental supernatants tested, obtained from 17 individuals, suppressed CTLL-2 proliferation, albeit with differences in the degree of inhibition. The immunosuppressive factor has a molecular weight of less than 3 kDa, being present in the filtrate following ultrafiltration using a Centriprep-3 centrifugal filter unit (Fig. 1b), and is resistant to heating to 100°C for 2 h (Fig. 1b). The suppression was not caused by a general cytotoxic effect of the HPS or the filtrate because adding these to the mouse T cell hybridoma DO.11, the mouse B cell line Bu.12 and to the human EBV-transformed B cell line WH3·72 caused no reduction in viability even after 24h (data not shown). Indeed, a slight stimulation of the Bu.12 B-cell line was observed.
Inhibitors of COX-1 and COX-2 abrogate the suppressor activity of human placental supernatants
It has been shown previously that the placenta produces PGE2, a molecule with known immunosuppressive properties [14,15]. PGE2 is synthesized from PGH2, which is itself synthesized from arachidonic acid by either the COX-1 or COX-2 enzymes. In cells expressing large amounts of COX-2 it appears that the PGH2 formed is saturating for the prostaglandin synthetase enzymes and that the formation of PGE2 is favoured . Using immunohistochemistry we detected abundant expression of COX-2 in the syncytiotrophoblast cells of the placental explants used in the current experiments (data not shown). We therefore wished to establish if the production of the immunosuppressive factor by the chorionic villi could be abrogated using the COX-1/COX-2 inhibitors indomethacin and diclofenac, or the COX-2-selective inhibitor DFP. HPS obtained in the presence of the inhibitors failed to block the IL-2-dependent proliferation of the CTLL-2 cells (Fig. 2), indicating that the immunosuppression is caused by prostanoids derived from the activity of COX-2.
Prostaglandin E2 is the major low molecular weight CTL-suppressive material in human placental supernatants
In light of the above observations, we measured the levels of PGE2 in the HPS obtained from explants of chorionic villi cultured in the presence or absence of the COX-1/COX-2 inhibitor indomethacin. The supernatants contained varying amounts of PGE2 (0·71–4·95 ng/ml) which in each case were substantially reduced when the chorionic villi were cultured in the presence of indomethacin (Table 1). Commercially available PGE2 (Sigma) mimicked the effect of HPS inhibition of CTLL-2 cell proliferation in a concentration-dependent manner while commercially available PGF2α (Sigma), another prostaglandin that is produced within the placenta, had no effect (Fig. 3a). Furthermore, PGE2 retained its activity after heating for 2 h at 100°C (Fig. 3a).
Table 1. PGE2 concentration in indomethacin-treated supernatants; The concentration of PGE2 measured by radioimmunoassay in HPS obtained from explants from five different placentas cultured for 48 h with (+) or without (−) 2μg/ml indomethacin
Further confirmation for the role of PGE2 was obtained from experiments in which supernatants from placental chorionic villi cultured in the presence of the COX inhibitor diclofenac were supplemented with amounts of PGE2 similar to those found in active (non-COX inhibitor-treated) supernatants. This procedure restored, in a concentration-dependent manner, the inhibition of CTLL-2 proliferation (Fig. 3b). Taken together, the data obtained strongly indicate that the major CTLL-2 immmunosuppressive low molecular weight heat stable material produced by 48h culture of chorionic villi explants is PGE2.
PGE2-mediated immunosuppression occurs via the EP4 receptor
Prostaglandins exert their function via specific receptors . At least four groups (EP1-EP4) of receptors have been characterized with distinct binding properties and tissue distribution. They belong to the seven transmembrane family of receptors and exert their function via G-proteins. The EP2 and EP4 receptors function by increasing the level of cAMP, an event associated with the inhibition of IL-2 and IL-2 receptor expression in Th1 cells. The receptor responsible for the suppression of the IL-2-dependent proliferation of CTLL-2 has not been identified previously. In order to establish which receptor is involved, we preincubated CTLL-2 cells with a set of novel antagonists having selectivity for the different EP receptor subgroups. When CTLL-2 were incubated with PGE2 in the presence of the EP-4-specific antagonist L-161 982, immunosuppressive activity of PGE2 was abrogated whereas the EP-1 specific antagonist L-818 638 and the EP-3 specific antagonist L-826 266 both failed to influence the effect of PGE2 on CTLL-2 (Fig. 4a). Thus, PGE2 exerts its function on CTLL-2 cells via the EP4 receptor. Similarly, the immunosuppressive effect of both unfractionated HPS and of the <3 kD filtrate was specifically reduced only by the EP-4 antagonist (Fig. 4b).
Chaouat and his colleagues have described a placental low molecular weight heat stable factor capable of down-regulating the expression of the IL-2 receptor on mouse T cells, and of inhibiting the mixed lymphocyte reaction, graft-versus-host rejection and NK cytotoxicity in vitro[1,27–29]. They did not identify their factor, but have recently referred to their molecular definitions of the material as being due to biochemical artefacts caused by the sticky nature of the active moiety . Using their methodology, we have been able to reproducibly obtain a <3 kDa heat stable immunosuppressive factor from placental supernatants and identify this factor as PGE2 synthesized by placental COX-2. Indeed, the biological properties attributed to the factor(s) described in the Chaouat studies mimic the PGE2 regulation of T cell function.
PGE2 affects both the activation and proliferation of T cells by interacting with specific cell surface receptors to stimulate cAMP accumulation, thereby activating the cAMP-dependent protein kinase A (PKA). Unlike other cAMP stimulators, such as the β-adrenergic receptor agonist isoproterenol, PGE2 activates both the cytoplasmic (PKAI) and membrane-located (PKAII) isoenzymes to the same extent . The increased cAMP levels affect the PKA pathway by inhibiting Ca2+ mobilization and reducing the turnover of inositol 3-phosphates . This modulates the activity of the AP-1 and NF-AT transcription factors, reducing their binding to the promoter regions of IL-2 and the IL-2 receptor . Not only is very little IL-2 and IL-2 receptor produced [14–16,26,35], but the cAMP-mediated effect on PKA also inhibits the production of other Th1 cytokines such as IFN-γ, without affecting the production of the Th2 cytokines IL-4 and IL-10 [12,25,26]. Thus, PGE2 switches the production of antibodies towards IgG1 and IgE classes . In addition, the elevated levels of cAMP negatively regulate the expression of c-myc thereby having an effect on T cell proliferation . PGE2 can induce anergy in activated T cells by increasing the turnover of IL-2 mRNA  and by affecting the IL-2 receptor signalling pathway directly through selectively suppressing the expression of the Janus kinase JAK3 in both naive and activated human T cells . In murine cell lines such as CTLL-2, PGE2 suppresses IL-2 and IL-4 dependent proliferation, as we have observed here for IL-2. However, the effect of PGE2 can be reversed by addition of excess IL-2 (or IL-4) , suggesting that the IL-2 signalling pathway in this species may not dramatically down-regulate JAK3 levels. In addition to T cells, natural killer (NK) cells are also present in the uterine decidua . Recent observations have indicated that IL-15-mediated NK cell cytotoxicity and interferon-γ production is suppressed by PGE2.
Production of PGE2 by the placenta is thought to contribute to ensuring that the maternal immune system ignores the paternal alloantigens, thus preventing rejection of the semiallogeneic fetus. PGE2 can interact with a number of receptors, EP1-EP4, with different tissue distribution and function. EP3 and EP4 are expressed in most tissues; EP1 expression is restricted to kidney, lung and stomach, whereas EP2 is not constitutively expressed but can be induced by various stimuli . Each of these receptors mediates their function through G-protein linked signalling pathways, albeit by activating distinct G-proteins. The EP1 receptor induces an increase in free Ca2+ possibly by regulating Ca2+ channel gating, EP3–ligand interaction results in an inhibition of adenylate cyclase, whereas both EP2 and EP4 mediate an increase in cAMP concentration . Naive and primed mouse T cells express predominantly EP3 and EP4, with EP4 being expressed more abundantly on Th1 than on Th2 cells . NK cells also express the EP4 receptor . Using a new set of receptor-specific antagonists we demonstrate here that the effect of PGE2 on IL-2-dependent CTLL-2 proliferation is selectively inhibited by an EP4 antagonist and unaffected by the antagonists selective for EP1 and EP3 receptors. This suggests that the immunosuppressive effect of PGE2 on T cells is mediated by the EP4 receptor, and opens up the possibility of using selective antagonists of the different PGE2 receptors for the deliberate skewing of immune responses towards a desired goal, for example favouring Th1-mediated cytotoxic responses where immunity to tumours is required. The antagonists may also give further insight into the mechanisms by which possible aggressive actions of NK cells on the fetal allograft may be subdued.
N. Kvirkvelia was a recipient of a Fellowship from The Royal Society of Great Britain. We thank Dr Robert Young, Merck-Frosst Center for Therapeutic Research (Canada) for kindly providing DFP and the PGE2 antagonists. This work was supported in part by The Cleveland General and Immunological Trust.