Prostaglandin E2 Regulates Both Interleukin-2 and Granulocyte ...

BIOLOGY OF REPRODUCTION 58, 143-151 (1998)

Prostaglandin E2 Regulates Both Interleukin-2 and Granulocyte-Macrophage Colony-Stimulating Factor Gene Expression in Bovine Lymphocytes' V. Emond, 3 ,4 M.A. Fortier,3 4 ,s B.D. Murphy,6 and R.D. Lambert2,3,4,5

Unite de Recherche en Ontoge'nie et Reproduction, Centre de Recherche du Centre Hospitalier de I'Universite Laval,3 Centre de Recherche en Biologie de la Reproduction,4 and Departement d'Obstetrique et Gynecologie,5 University Laval, Ste-Foy, Quebec, Canada G1 V 4G2 Centre de Recherche en Reproduction Animale,6 Universite de Montreal, St-Hyacinthe, Quebec, Canada J2S 7C6 ABSTRACT Prostaglandin E2 (PGE 2) is known to inhibit interleukin-2 (IL2) production by human peripheral blood lymphocytes (PBL) and to increase granulocyte-macrophage colony-stimulating factor (GM-CSF). In many species with hemochorial placentation, down-regulation of IL-2 appears necessary to impede early embryonic demise, whereas up-regulation of GM-CSF increases embryonic growth and survival. It is not known whether the same mechanisms are involved in a species with a less invasive placenta. PGE2 is synthesized during early bovine gestation by the endometrium and by the embryo, and it may therefore be involved in regulating IL-2 and GM-CSF in this species. Our goal was to evaluate the impact of PGE 2 on cellular proliferation and on IL-2 and GM-CSF gene expression in bovine PBL. Incorporation of [3H]thymidine was used to study DNA synthesis. Gene expression was estimated by semiquantitative polymerase chain reaction using bovine-specific primers and by Northern analysis using amplified bovine cDNAs as probes. The DNA synthesis and IL-2 mRNA levels of bovine PBL stimulated by concanavalin A (ConA) were greatly reduced by PGE 2 in direct-treatment studies. Under the same conditions, GM-CSF gene expression was also inhibited. However, pretreatment of PBL for 72 h with ConA and PGE2, followed, after washing, by an incubation with ConA alone for 12 h resulted in reduced DNA synthesis, stable expression of IL-2, and a dramatic increase of GM-CSF mRNA levels. This is the first evidence in the bovine model that direct treatment with PGE2 down-regulates IL-2 and GM-CSF mRNA levels and that preconditioning with PGE2 stimulates GM-CSF gene expression. We propose that PGE2, either from embryonic or from endometrial compartments, induces bovine PBL to undergo functional changes, affecting cellular proliferation and cytokine production inorder to accommodate the developing conceptus. INTRODUCTION Early after conception, the embryo expresses antigens of paternal origin and can be considered a semiallogeneic graft. Given that the bovine embryo expresses paternal major histocompatibility complex (MHC) molecules of class I at Day 7 of pregnancy [1], and since MHC antigens can influence the outcome of bovine gestation [2], the conceptus appears to be recognized immunologically but not rejected. Two general mechanisms have been proposed to exAccepted August 29, 1997. Received May 28, 1997. 'This work was supported by grant 98-ER-2421 from the Fonds pour la Formation de Chercheurs et l'Aide la Recherche (FCAR) and a grant to R.D.L. from the Natural Sciences and Engineering Research Council of Canada (NSERC). 2Correspondence: Raymond D. Lambert, Ontog6nie et Reproduction, Centre de Recherche du Centre Hospitalier de I'Universit6 Laval, 2705 Blvd Laurier, Ste-Foy, PQ, Canada G1V 4G2. FAX: (418) 654-2765; email: [email protected] 143

plain how the conceptus is recognized by and escapes maternal immune defenses: 1) local production of regulatory signals with immunomodulatory properties [3, 4] and 2) concealed expression of fetal antigens and expression of MHC antigens with low polymorphism, such as HLA-G [5, 6]. In fact, it is critical that a complex communication network be established among the conceptus, the endometrium, and the local immune cells to allow gestation to go to term. Messengers involved include cytokines, growth factors, and prostaglandins (PGs). Such events in early pregnancy have been well studied in species with hemochorial placentation such as the mouse [7-11] and the human [1216]. Species with less invasive types of placenta lack characterization and, as conceptus exposure to maternal tissues differs widely among species, mechanisms involved in conceptus tolerance are expected to vary accordingly. The human trophoblast is bathed in maternal blood. On the other hand, the bovine conceptus and the maternal bloodstream are separated by multiple layers of epithelial, stromal, and endothelial cells in intercaruncular areas; and the deepest invasion is localized in cotyledons where only trophoblast interdigitates and fuses with uterine epithelial cells [17-19]. Due to these physical constraints, it is expected that interactions between the bovine conceptus and local maternal leukocytes are likely to involve mechanisms different from those in the human or the mouse. The immunomodulatory role of PGE2 has been studied extensively in assays of cell proliferation and of cytokine production such as interleukin-2 (IL-2) and granulocytemacrophage colony-stimulating factor (GM-CSF). In species with hemochorial placentation, PGE 2 has been shown to play a key role in promoting embryo survival by countering the disruptive potential of natural killer (NK) cells and lymphokine-activated killer cells [8, 14, 20]. In particular, it was demonstrated that PGE 2 can induce IL-2-activated NK cells to differentiate to less embryotoxic forms [11]. To date, the only confirmed effect of PGE2 on bovine leukocytes is the inhibition of T-cell proliferation in vitro [21]. Sources of PGE 2 in the pregnant uterus include the conceptus [22, 23], endometrial cells, and macrophages [24]. Recent findings have revealed that the conceptus protein responsible for maternal recognition, interferon tau (IFN-T) [25], dramatically stimulates PGE 2 synthesis in cultured bovine endometrium [26]. Thus, the bovine conceptus can regulate PGE 2 concentrations in the vicinity of the conceptus by modulating both its own PGE2 production and that of endometrial cells. It is expected that failure to modulate PGE 2 production could jeopardize fetal survival. IL-2 and GM-CSF are known to have opposite actions during pregnancy. IL-2 is associated with spontaneous abortions in the mouse [9, 20, 27] and the human [28], and GM-CSF stimulates conceptus growth [9, 29-31]. Previous

144

EMOND ET AL.

studies with rabbit allantoic fluid have demonstrated that PGE 2, along with transforming growth factor 32, is responsible for the regulation of lymphocyte functions, including IL-2 and GM-CSF production [32]. In the present study, semiquantitative reverse transcription-polymerase chain reaction (RT-PCR) and Northern blot analysis were used to demonstrate the response of bovine peripheral blood lymphocytes (PBL) to PGE 2 in terms of IL-2 and GM-CSF gene expression. Their reciprocal response suggests that, during early bovine gestation, PGE 2 can modulate the immune function through the regulation of IL-2 and GM-CSF

The incorporated radioactivity was then evaluated by liquid scintillation counting. For RT-PCR and Northern blot studies, the cells (2 x 106 cells per well) were distributed in 24-well plates (Becton Dickinson) with ConA (5 ,ag/ml), FBS (6.7% v:v), and RPMI 1640, in a final volume of 2.25 ml/well. Various concentrations of PGs were added as described above for concentration-response studies. The plates were incubated at 37°C under a humidified atmosphere and 5% CO 2 for 6 h (IL-2 studies) and 24 h (GM-CSF studies). Before harvesting, the cells were observed under a microscope to detect aggregation and blastogenesis, two qualitative parameters related to lymphocyte activation.

MATERIALS AND METHODS Materials

Cell Culture: Preconditioning

Percoll was obtained from Pharmacia Biotech (Baie d'Urf, PQ, Canada). Streptomycin and penicillin were supplied by ICN Biochemicals (Aurora, OH) and fetal bovine serum (FBS) by Hyclone Laboratories (Logan, UT). Hanks' Balanced Salt Solution (HBSS), RPMI 1640, TRIzol, and RT-PCR enzymes and reagents were purchased from Gibco BRL (Burlington, ON, Canada). The [methyl- 3H]thymidine was supplied by Dupont (Mississauga, ON, Canada), the deoxynucleotide triphosphates by Boehringer Mannheim (Laval, PQ, Canada), and Qiaex and Qiaquick kits by Qiagen (Chatsworth, CA).

For RT-PCR and Northern blot studies, lymphocytes were cultured as described above for 72 h. The cells were collected, pooled, and washed twice with HBSS and then resuspended in RPMI containing 10% FBS. They were then distributed in 24-well plates (2 x 106 cells per well) with ConA (5 pxg/ml), FBS (6.7% v:v), and RPMI 1640, supplemented with antibiotics, to a final volume of 2.25 ml/well. The plates were incubated at 370C under a humidified atmosphere and 5% CO 2 for 12 h. Before harvesting, the cells were observed under a microscope to detect aggregation and blastogenesis. PBL collected at the end of the 72-h pretreatment were distributed in 96-well flat-bottom plates (2 x 105 cells per well) with ConA (5 g/ml), FBS (6.7% v:v), and RPMI 1640, to a final volume of 225 ill/well. The 3H-TdR incorporation was measured as described above for direct treatment.

Preparation of Bovine Lymphocytes Blood from cows was collected in sodium heparin vacutainers. PBL were prepared in 50-ml conical tubes (Becton Dickinson, Lincoln Park, NJ) using a 44% Percoll/ HBSS (v:v) density gradient. The diluted Percoll (20 ml/20 ml of blood) was added to the bottom of the tube with a sterile pasteur pipette. After centrifugation at 800 X g for 20 min at room temperature, cells were separated into three phases. PBL in the middle phase (Percoll), between the erythrocyte and plasma phases, were collected and washed three times with HBSS. The PBL were then resuspended in RPMI 1640 containing streptomycin (100 ng/ml), penicillin (100 IU/ml), and decomplemented FBS (10%). Cell Culture: Direct Treatment Incorporation of [3H]thymidine (3H-TdR) was used to evaluate DNA synthesis, as there is general acceptance that it reflects cell proliferation. For incorporation studies, PBL (2 x 105 cells per well) were distributed in 96-well flatbottom plates (Becton Dickinson) with 5 pxg/ml concanavalin A (ConA; Laboratoire Mat Inc., Beauport, PQ, Canada), FBS (6.7% v:v), and RPMI 1640, in a final volume of 225 l/well. Various concentrations (0.1-100 ng/ml) of PGs (Sigma, Oakville, ON, Canada) were added for concentration-response studies. These concentrations are comparable to those used in similar experiments [11, 21, 32]. Nonstimulated control cultures had no ConA, while ethanol (0.02% v:v final), the vehicle for both PGE 2 and PGF 2 , was added to the control medium containing no PGs (PGO). The plates were incubated at 37C under a humidified atmosphere and 5% CO 2 for 24 h. After addition of 0.2 p.Ci 3 H-TdR (2.0 Ci/mmol), incubation was resumed for a further 24 h. The cells were harvested on glass fiber filters (Wallac, Turku, Finland) with a multiple cell harvester (Skatron, Lier, Norway), and the filter disks corresponding to each well were placed into vials with 3 ml of biodegradable scintillant (Amersham, Oakville, ON, Canada).

RNA Isolation At the end of cultures, cells were recovered, pooled, and pelleted at 800 g for 10 min. They were then lysed in TRIzol reagent (1 ml/107 cells), stored at -80 0 C, and processed within a month. Total RNA was extracted using TRIzol according to the manufacturer's instructions. RNA samples were resuspended in water containing diethylpyrocarbonate (0.05% v:v) and stored at -80°C. Before analysis, RNA was quantified by measuring absorbance at 260 nm. RT-PCR Analysis Total RNA samples (400 ng) from each treatment were heated at 65°C for 10 min with oligo-dT primers (0.5 ,ug) in a final volume of 10 ,ul. Reverse transcription was then performed for 1 h at 37C in a mixture containing RT buffer (single-strength), dithiothreitol (10 mM DTT), deoxynucleotide triphosphates (1.25 mM dNTPs), and Muloney murine leukemia virus reverse transcriptase (200 U RTase) in a final volume of 20 [l1. The RT reaction was terminated by heating at 94°C for 10 min, and reaction volumes were then brought to 65 I1. A control without RTase was performed at the same time to ensure absence of contaminating DNA in the RNA templates. Gene expression was determined using PCR amplification of the cDNAs obtained. Each reaction was run with 5 l of RT template or negative control, MgC1 2 (1.5 mM), PCR buffer (single-strength), dNTPs (0.2 mM), sense and antisense primers (10 RIM of each), and Taq DNA polymerase (1.5 U) in a final volume of 50 1. The samples were overlaid with autoclaved light mineral oil (35 p), and

PGE 2 REGULATES IL-2 AND GM-CSF IN BOVINE PBL

145

PCR amplifications were performed for 25-29 cycles for IL-2, 25-30 for GM-CSF, and 18-26 for -actin. Denaturation was performed at 94°C (30 sec), annealing at 55°C (30 sec), and elongation at 72°C (30 sec); lastly, the PCR was followed by a 10-min extension (72°C). Reactions were analyzed on 1.8% agarose gels containing 0.1 ±Lg/ml ethidium bromide. Bands were visualized by UV illumination and quantified by densitometry using ANA Mac NIH V1.0 software. Intensity of each band was normalized to the intensity of the corresponding -actin band as an internal control. The controls without RTase were negative for each amplification. Identity of RT-PCR products was evaluated by size and confirmed by restriction digestion. Primers All primers were chosen with the aid of the OLIGO 4.01 primer analysis software (National Biosciences, Plymouth, MN) using published bovine cDNA sequences. Expected PCR product lengths were 413 base pairs (bp) for IL-2, 421 bp for GM-CSF, and 349 bp for -actin, the internal control. Sense (S) and antisense (AS) primers were (5'->3'): IL-2 (S) CTT-GTC-TTG-CAT-TGC-ACT-AAC-TCT-TG; IL-2 (AS) TCC-ATT-TGT-TCA-GAA-ATT-CTA-CAG-CG; GM-CSF (S) ATG-TGG-CTG-CAG-AAC-CTG-CTT-C; GM-CSF (AS) CTG-GTT-CCC-AGC-AGT-CAA-AGG-G; 3-actin (S) CAA-CTG-GGA-CGA-CAT-GGA-GAA-GATCTG-GCA; 3-actin (AS) GAG-GAT-CTT-CAT-GAGGTA-GTC-TGT-CAG-GTC. Northern Blot Analysis Total RNA (10 or 20 pxg) was subjected to electrophoresis on a 1.2% agarose-2.2 M formaldehyde gel and transferred by capillary action to a nylon membrane (Qiagen). After 4 h of prehybridization in a 50% formamide solution at 42°C, the membrane was sequentially hybridized overnight at 42°C in the same solution, to which were added [ao- 32 P]dCTP-labeled DNA probes. Membranes were washed (0.5% SDS, single-strength SSC [0.15 M sodium chloride and 0.015 sodium citrate], 65°C) and exposed to x-ray film with an intensifying screen at -80°C for 24 h (GM-CSF), 48 h (IL-2), and 2-4 h (-actin). Membranes were stripped between hybridizations by soaking in a boiling 0.1% SDS solution. Autoradiograms were quantified by densitometry using ANA Mac NIH V1.0 software. Intensity of each band was normalized to the intensity of the corresponding 3-actin band as an internal control. Probes DNA templates for probes were obtained by gel excision of the various PCR products and extracted by Qiaex using the manufacturer's protocol. The probes were thus 413 bp for IL-2, 421 bp for GM-CSF, and 349 bp for -actin. The [ca- 32 P]dCTP-labeled DNA probes were produced by random-priming protocol using 50 ng DNA and the T7 Quickprime kit (Pharmacia Biotech) and were purified by Qiaquick following the manufacturer's procedure. Resultant specific activity averaged 1 X 109 dpm/Lg. Statistical Analysis Each experiment was run in three replicates. Replicates for RT-PCR correspond to the three different numbers of cycles used to amplify each sample in an experiment. Data were expressed as percentages of or fold increases in relation to PGO controls and were examined by analysis of

FIG. 1. Time course of ConA-induced levels of 3H-TdR incorporation, and IL-2 and GM-CSF mRNA levels detected by RT-PCR analysis. A)The peak of stimulation was reached after 48 h of stimulation with 5 g/ml ConA. Values are means ( SEM) of three replicate wells within a single experiment. B) RT-PCR products of IL-2 (26 cycles) and GM-CSF (28 cycles) are shown after electrophoresis. A representative experiment is illustrated. C) Quantification of PCR products for IL-2 and GM-CSF by image analysis. IL-2 and GM-CSF gene expression reached their respective maximums after 6 and 24 h. Values, presented in percentages of the maximal value obtained throughout time, are means ( SEM) of a representative experiment in which three replicates were amplified at 24, 26, and 28 cycles, respectively, for IL-2 and at 26, 28, and 30 cycles for GM-CSF. Two additional experiments were performed. One gave identical results, while the other showed a decline in IL-2 mRNA levels after 24 h.

variance using super ANOVA software (Abacus Concepts, Berkeley, CA). The effect of each concentration-response treatment was determined using orthogonal contrasts. RESULTS IL-2 and GM-CSF Gene Expression Is Induced by ConA As an initial step, we determined the DNA synthesis of bovine PBL in response to increasing doses of ConA. We found that maximal DNA synthesis was reached after 48 h of stimulation with 5 Lg/ml ConA (Fig. 1A). RT-PCR amplifications of IL-2 and GM-CSF mRNAs following ConA stimulation are illustrated in Figure 1, B and C. Expression of IL-2 mRNA was elevated at 6 h, and a plateau was maintained through 48 h in two of three experiments. In

146

EMOND ET AL.

FIG. 3. Effects of PGE, and PGF2, on levels of IL-2 mRNA detected by Northern blot analysis ("a" indicates without ethanol). A) Northern blots of IL-2 and -actin were exposed at -80°C for 48 h and 4 h, respectively. A representative experiment is illustrated. B) Using image analysis, the ratios of IL-2/3-actin signals were determined. PGE2 reduced IL-2 gene 11%. Values, presented in percentages of the value expression by 88 obtained in the absence of PGs, are means of data (+ SEM) from two experiments. **p < 0.0001 compared to the PG0 control. FIG. 2. Effects of PGE2 and PGF 2,, on H-TdR incorporation, and IL-2 and GM-CSF mRNAs levels detected by RT-PCR analysis ("-a" indicates without ethanol). A) The H-TdR uptake is reduced by up to 80 ± 2% in the presence of PGE 2. Values, presented in percentages of the value obtained in the absence of PGs, are means ( SEM) of n experiments with three replicate wells. For 100 ng/ml PGE2, n = 7; for 1-10 ng/ml PGE, and 100 ng/ml PGF,, n = 5; for 1-10 ng/ml PGF2 , n = 3; and for 0.1 ng/ml PGE2, n = 1. **p < 0.0001 compared to the PGO control. B) RTPCR products of IL-2 (26 cycles) and 6-h -actin (26 cycles) are shown after electrophoresis. A representative experiment is illustrated (ND: not determined in the illustrated experiment). C) RT-PCR products of GM-CSF (29 cycles) and 24-h -actin (24 cycles) are shown after electrophoresis. A representative experiment is illustrated (ND: not determined in the illustrated experiment). D) Using image analysis, the ratios of IL-2/[-actin signals and GM-CSF/[3-actin signals were determined. PGE, reduced IL-2 gene expression by up to 67 _+ 5% and GM-CSF by up to 54 + 8%. Values, presented in percentages of the value obtained in the absence of PGs, are means (+ SEM) of n experiments (1-10 ng/ml PGE 2, n = 3; 100 ng/ml PGE2, n = 2; 0.1 ng/ml PGE2 and 1, 10, 100 ng/ml PGF 2 ,, n = 1). For each experiment, three replicates were amplified at 25-29 cycles for IL-2, 20-27 cycles for 13-actin (6 h), 28-30 for GM-CSF, and 20-25 for 13-actin (24 h). ** p < 0.0001; * p < 0.005 compared to the respective PGO control.

the third, the decline occurred at 24 h rather than 48 h. GM-CSF was elevated at 24 h, and the elevation remained through 48 h. The times at each plateau (6 h for IL-2; 24 h for GM-CSF) were chosen to study the effect of PGs on IL-2 and GM-CSF gene expression. Direct Treatment with PGE2 Down-Regulates Lymphocyte Functions As shown in Figure 2A, bovine PBL DNA synthesis was inhibited in a concentration-dependent manner by PGE2. At

100 ng/ml, PGE 2 induced 3 H-TdR incorporation in PBL, but at a level of only 20 ± 2% (p < 0.0001) of the control value. PGF 2, had no significant effect (p > 0.3). In agreement with this finding, observation of bovine PBL under a microscope revealed that increasing amounts of PGE 2 decreased cell aggregation and lowered the percentage of blastic cells, while cells treated with PGF 2, looked like the stimulated control (PGO) cells. Figure 2, B and D, also demonstrates that increasing doses of PGE 2 inhibited IL-2 5% (p < 0.0001) of the gene expression, reaching 33 control value at 100 ng/ml PGE 2. PGF 2, had no significant effect (p > 0.3). Northern analysis (Fig. 3, A and B) gave similar results, with an inhibition of IL-2 mRNA levels of 11% (p < 0.0001) by PGE 2 . Figure 2, C-D, also 88 illustrates that increasing doses of PGE 2 inhibited GM-CSF gene expression while low concentrations had a stimulatory effect. PGE 2 at 0.1 ng/ml raised GM-CSF mRNA levels by 68 ± 11% (p < 0.005), but an inhibitory effect was observed at 100 ng/ml, to 46 + 8% (p < 0.0001) of the control value. PGF 2,, on the other hand, had no significant effect (p > 0.3). Under direct-treatment conditions, DNA synthesis and IL-2 and GM-CSF gene expression exhibited a typical concentration-dependent response to PGE 2. IL-2 and GM-CSF Gene Expression Is Induced by Restimulation of PGE2 -Preconditioned Lymphocytes with ConA As shown in Figure 4, A and B, both IL-2 and GM-CSF mRNA levels reached a peak 12 h after the end of the

PGE2 REGULATES IL-2 AND GM-CSF IN BOVINE PBL

FIG. 4. Time course of IL-2 and GM-CSF mRNA levels detected by RTPCR analysis after preconditioning with PGE2 (100 ng/ml) and restimulation. A) RT-PCR products of IL-2 (28 cycles) and GM-CSF (28 cycles) are shown after electrophoresis. A representative experiment is illustrated. B) Quantification of PCR products of IL-2 and GM-CSF by image analysis. IL-2 and GM-CSF gene expression both were maximal at 12 h. Values, presented in percentages of the maximal value obtained throughout time, are means ( SEM) of a single experiment in which three replicates were amplified at 24, 26, and 28 cycles, respectively, for IL-2 and at 26, 28, and 30 cycles for GM-CSF.

pretreatment with PGE 2 (100 ng/ml). The effect of preconditioning with PGs on both IL-2 and GM-CSF gene expression was therefore assayed at this time. Preconditioning with PGE2 Modulates Lymphocyte Functions As illustrated in Figure 5, A and B, PBL preconditioning with PGs for 72 h followed by a stimulation with ConA alone for 12 h had no significant effect (PGE 2 p > 0.5; PGF2, p > 0.3) on IL-2 gene expression as compared to the PGO control. GM-CSF mRNA levels were increased in a concentration-dependent manner when PGE 2 was present in the pretreatment medium (Fig. 5, A and B). A maximum stimulation of 18 + 2-fold (p < 0.0001) of the control value was obtained with 100 ng/ml PGE 2. Again, PGF 2. had no significant effect (p > 0.7). Northern analysis showed the same tendency; GM-CSF mRNA levels were 7 2-fold higher (p < 0.005) than levels in the controls without PGE 2 or with PGF2 (Fig. 6, B and C). Figure 6A indicates that PBL preconditioning with PGE 2 for 72 h followed by a 48h stimulation with ConA alone led to an inhibition of DNA

147

FIG. 5. Effects of PGE2- and PGF,,-preconditioning on levels of IL-2 and 2 GM-CSF mRNAs detected by RT-PCR analysis ("-a" indicates without ethanol). A) RT-PCR products of IL-2 (27 cycles), GM-CSF (27 cycles), and 3-actin (22 cycles) are shown after electrophoresis. A representative experiment is illustrated. B) Using image analysis, the ratios of IL-2/-actin signals and GM-CSF/P-actin signals were determined. IL-2 gene expression is not affected by PGE 2 (p > 0.3), while GM-CSF is increased by up to 18 + 2-fold. Values, presented as fold increase over PG0 control, are means ( SEM) of n experiments (1, 10, 100 ng/ml PGE2, n = 2; 1, 10, 100 ng/ml PGF2,,, n = 1). For each experiment, three replicates were amplified at 25-29 cycles for IL-2, 25-29 for GM-CSF, and 20-22 for 3actin. **p < 0.0001; * p < 0.005 compared to the respective PGO control.

synthesis. Only 30 + 10% (p < 0.0001) of the control value was obtained when PGE 2 was present in the preconditioning medium, while PGF 2 had no significant effect (p > 0.3). Accordingly, under a microscope, cell aggregation was decreased by PGE2 but not PGF 2,. DISCUSSION During the bovine estrous cycle, uterine T lymphocytes increase in number from Day 11 onward [33]. CD4+ cells, typically T-helper lymphocytes, are located in the endometrial stroma and outnumber CD8+ (suppressor/cytotoxic) cells, which constitute the prevalent lymphocytes in the uterine epithelium [33]. Conversely, a decrease of intraepithelial lymphocytes is observed during bovine pregnancy, from around Day 21 [34]. This is preceded by the remarkable conceptus elongation occurring between Day 11 and Day 22 of pregnancy [5]. This a critical time of bovine embryonic development, since it involves production of the conceptus secretory protein (peak at Days 15-19) [35], now known as IFN-r, that maintains the corpus luteum [25, 36]. It also has strong immunosuppressive potential [37, 38]. The endometrial epithelial and stromal cells respond to this signal by substantially increasing their secretion of PGE 2 [26], a known immunomodulator [11, 21, 24]. Since IFN-'r strongly stimulates uterine PGE2 , it is likely that as IFN-T production falls, uterine PGE 2 is diminished. From Day 18 to 20, shortly after the peak of IFN-'r, the

148

EMOND ET AL.

FIG. 6. Effects of PGE2- and PGF,,-preconditioning on levels of 3H-TdR 2 incorporation and GM-CSF mRNA detected by Northern blot analysis (" a" indicates without ethanol). A) The 3H-TdR uptake is reduced by 68 ± 10% by PGE2 . Values, presented in percentages of the value obtained in the absence of PGs, are means ( SEM) of two experiments with three replicate wells. ** p < 0.0001 compared to the PG0 control. B) Northern blots of GM-CSF and -actin were, respectively, exposed at -80C for 24 h and at room temperature for 2 h. A representative experiment is illustrated. C) Using image analysis, the ratios of GM-CSF/13-actin signals were determined. PGE 2 increased GM-CSF gene expression by 7 2fold. Values, presented as fold increase over PG0 control, are means (± SEM) of two experiments. * p < 0.005 compared to the PG0 control.

first intimate connections between embryonic and maternal tissues occur when multiple trophoblastic papillae reach uterine glands [39] and extensive adhesion between trophoblast and epithelium is observed [17]. Embryonic attachment, early in the fourth week of gestation, is characterized by interdigitation of trophoblastic and maternal villi. The binucleate trophoblastic cells fuse with epithelial cells to form multinucleate giant cells, comprising around 50% of the maternal epithelial surface by Day 24 [17-19]. Trophoblastic cells are most certainly in contact with endometrial leukocytes, and the present observations suggest that the events accompanying early trophoblastic infiltration around Days 18-24 trigger immunorecognition and immunomodulation mechanisms. The present study confirms an immunomodulatory role

for PGE2 and extends this observation to the cow. Indeed, direct treatment with concentrations of PGE2 ranging from 1 to 100 ng/ml causes a general down-regulation of bovine PBL functions. Cell proliferation was inhibited as expected [21], and IL-2 and GM-CSF mRNA levels were decreased. It is generally accepted that T-cell activation requires IL-2 production [12, 40]. The major inhibition of IL-2 gene expression obtained early during treatment with PGE2 could thus explain subsequent declines in GM-CSF mRNA levels and DNA synthesis. This suggests that PGE 2 in sufficient amounts can maintain lymphocytes in a near-dormant state similar to that of control without ConA. PGE 2 is known to inhibit IL-2 production [32, 40] and GM-CSF production [32, 41] and gene expression [42] in the human. It has also been shown to decrease IL-2 and IL-2 receptor production in the mouse [12]. On the other hand, 0.1 ng/ml PGE 2 stimulated GM-CSF gene expression, with no effect on IL-2 mRNA or DNA synthesis. Thus, the present study has shown a biphasic effect of PGE 2 acute exposure on GMCSF gene expression: a stimulation at low concentrations and an inhibition at high concentrations. Biphasic effects of PGE 2 have been reported to be mediated by PGE receptors of different affinities [43]. Other studies have confirmed that GM-CSF responses to PGE 2 are indeed complex and depend on the types of lymphocytes involved [44, 45]. It should be noted that the design of these experiments did not permit determination of whether the change in mRNA levels was due to transcription rate or to change in mRNA degradation. Direct-treatment studies indicate that PGE 2 can have an impact on proliferation and cytokine production on bovine PBL. These regulatory effects occur as early as at 6 h of stimulation and at concentrations of PGE 2 comparable to physiological levels, since bovine allantoic fluid contains a maximum of 6 ng/ml of PGE 2 (unpublished results) while the uterine vein has 2.24 ng/ml of PGE 2 and 3.96 ng/ml of PGE 2 metabolite [46]. Results obtained in direct-treatment protocols are believed to mimic the sudden in vivo exposure of endometrial T cells to high amounts of PGE2, a situation likely to occur during the extensive output of IFNT by the conceptus. The quiescent state of the immune cells, reflected in down-regulated lymphocyte functions, may be related to the success of bovine gestation, since around Days 15-22 the conceptus expands and begins its first intraepithelial infiltrations, being therefore directly exposed to maternal immune cells. Local reduction of IL-2 production would prevent the activation of NK cells into lymphokine-activated killer cells that could be harmful to the conceptus [7, 15, 20]. The in vitro preconditioning of bovine PBL is a model that mimics the effect of endometrial [26, 47, 48] or conceptus [22, 23] fluctuations of PGs. High levels of PGE 2 during the peak of IFN-T production are expected to desensitize the immune cells, decreasing their responsiveness to various stimuli, as shown in other cell types [49]. Reduced gene expression of IL-2 following pretreatment is in agreement with this interpretation. The semiquiescent cells thus obtained may preserve the conceptus by keeping cell proliferation and IL-2 low to prevent the appearance of cytotoxic cells. However, elevated GM-CSF mRNA levels indicate that responses to PGE 2 are, again, complex. The fact that there are dual effects of PGE2 on GM-CSF gene expression, depending on the type of stimulation (acute vs. chronic exposure), suggests 1) that some immune cellular functions in the immediate vicinity of the conceptus are preserved while others are suppressed, 2) that subtypes of

PGE 2 REGULATES IL-2 AND GM-CSF IN BOVINE PBL

149

FIG. 7. Cross-talk between the conceptus, endometrial cells, and maternal immune cells: effects on conceptus accommodation in the cow. PGE2 is produced by the embryo [22, 23], and its synthesis in epithelial and probably stromal uterine cells is stimulated by IFN-, [26]. It locally suppresses the immune system by inhibiting lymphocyte proliferation [211 and by down-regulating IL-2 gene expression (this study). PGE2 could also promote conceptus growth [9, 29-31] or reorient local cytokine production by stimulating GM-CSF gene expression (this study).

lymphocytes can respond differently one from another, or 3) that receptors with different affinities and signaling pathways are involved. Among probable roles of lymphocyte GM-CSF, sustenance of conceptus growth [9, 29-31], regulation of the functions of fellow immune cells like macrophages or NK cells [50, 51], and reorientation of the local cytokine network [52] have been proposed. Indeed, GM-CSF has been shown to stimulate IL-10 production in a choriocarcinoma cell line [52] and in human lymphocytes (unpublished results), giving support to the hypothesis of a T-helper-2 bias during early pregnancy [53]. Given that uterine epithelial cells also produce GM-CSF during pregnancy [54, 55], the relative contribution of immune versus nonimmune cells to these biological effects of GM-CSF is yet a matter of speculation. From these data we suggest that the bovine conceptus orchestrates early uterine accommodation mechanisms by modulating PGE 2 production at the feto-maternal interface (Fig. 7). It can exert its influence by directly targeting resident leukocytes of the uterus with its own PGE2 or by inducing, via IFN-'r, uterine epithelial and probably stromal cells to synthesize PGE 2 [26]. This causes a decrease in lymphocyte proliferation [21] and IL-2 gene expression, thus reducing immune rejection. In increasing GM-CSF gene expression, PGE 2 could also promote conceptus growth [9, 29-31]. Taken together, these findings suggest that the bovine conceptus plays an active role in its protection against immune rejection and turns the local maternal immune system to its advantage by modulating harmful and beneficial cytokine gene expression. Overall, data presented here suggest that PGE 2 can have diverse effects that depend on exposure time, conditions of stimulation, and PGE 2 concentrations. This regulation is specific to PGE 2 since PGF 2 , another PG synthesized by the bovine embryo [22, 23] and endometrium, could not induce the same effects even though it has the potential to generate similar intracellular messengers. The tremendous increase in GM-CSF mRNAs following PGE 2 removal is in agreement with previous findings obtained using human PBL [32]. This is the first evidence in the bovine model that direct treatment with PGE 2 down-regulates IL-2 and

GM-CSF mRNA levels, and that preconditioning with PGE 2 stimulates GM-CSF gene expression. This work demonstrates that PGE 2 can induce bovine PBL to undergo functional changes, and it suggests an important immunoregulatory role for PGE 2 during early bovine gestation. Although the results presented here have been discussed in terms of reproductive biology, the same mechanisms can likely apply to any physiological or pathophysiological situations involving local variations of PGE 2 production. Such variations are known to occur during embryo implantation in several animal species [56], during inflammation [57], and even during tumor development [58]. ACKNOWLEDGMENTS The authors would like to thank EJ. Dugr6, D. Lacroix, S. Gagnon, E. Asselin, and C. Lgard for their helpful discussions on many aspects of the work.

REFERENCES 1. Templeton JW, Tipton RC, Garber T, Bondioli K, Kraemer DC. Expression and genetic segregation of parental BoLA serotypes in bovine embryos. Anim Genet 1987; 18:317-322. 2. Aguilar B, Vos PLAM, Beckers JF, Hensen EJ, Dieleman SJ. The role of the major histocompatibility complex in bovine embryo transfer. Theriogenology 1997; 47:111-120. 3. Bergeron D, Audette M, Lambert RD. Regulation of leukocyte interleukin 2 and interleukin 2 receptor gene expression by rabbit blastocoelic fluid. J Reprod Fertil 1996; 106:143-151. 4. Ouellette MJ, Dubois CM, Bergeron D, Roy R, Lambert RD. TGFbeta-2 in rabbit blastocoelic fluid regulates CD4 membrane expression-possible role in the success of gestation. Am J Reprod Immunol 1997; 37:125-136. 5. Ellis S. HLA G: at the interface. Am J Reprod Immunol 1990; 23: 84-86. 6. Kovats S, Main E, Librach C, Stubblebine M, Fisher S, DeMars R. A class I antigen, HLA-G, expressed in human trophoblasts. Science 1990; 248:220-223. 7. Drake BL, Head JR. Murine trophoblast can be killed by lymphokineactivated killer cells. J Immunol 1989; 143:9-14. 8. Lala PK. Interruption of murine pregnancy by activation of antigennon-specific killer cells in the endometrium with indomethacin, high dose IL-2 or a combination. Res Immunol 1990; 141:159-164. 9. Chaouat G, Menu E, Clark DA, Dy M, Minkowski M, Wegmann TG. Control of fetal survival in CBA x DBA/2 mice by lymphokine therapy. J Reprod Fertil 1990; 89:447-458.

150

EMOND ET AL.

10. Kanzaki H, Crainie M, Lin H, Yui J, Guilbert LJ, Mori T, Wegmann TG. The in situ expression of granulocyte-macrophage colony-stimulating factor (GM-CSF) mRNA at the maternal-fetal interface. Growth Factors 1991; 5:69-74. 11. Linnemeyer PA, Pollack SB. Prostaglandin E2-induced changes in the phenotype, morphology, and lytic activity of IL-2-activated natural killer cells. J Immunol 1993; 150:3747-3754. 12. Lala PK, Kennedy TG, Parhar RS. Suppression of lymphocyte alloreactivity by early gestational human decidua. II. Characterization of the suppressor mechanisms. Cell Immunol 1988; 116:411-422. 13. Daya S, Johnson PM, Clark DA. Trophoblast induction of suppressortype cell activity in human endometrial tissue. Am J Reprod Immunol 1989; 19:65-72. 14. Parhar RS, Yagel S, Lala PK. PGE2-mediated immunosuppression by first trimester human decidual cells blocks activation of maternal leukocytes in the decidua with potential anti-trophoblast activity. Cell Immunol 1989; 120:61-74. 15. King A, Loke YW. Human trophoblast and JEG choriocarcinoma cells are sensitive to lysis by IL-2-stimulated decidual NK cells. Cell Immunol 1990; 129:435-448. 16. Hill JA, Polgar K, Anderson DJ. T-helper -type immunity to trophoblast in women with recurrent spontaneous abortion. JAMA 1995; 273:1933-1936. 17. King GJ, Atkinson BA, Robertson HA. Development of the bovine placentome from days 20 to 29 of the gestation. J Reprod Fertil 1980; 59:95-100. 18. Wooding FB. The role of the binucleate cell in ruminant placental structure. J Reprod Fertil Suppl 1982; 31:31-39. 19. Wooding FB. Frequency and localization of binucleate cells in the placentomes of ruminants. Placenta 1983; 4:527-539. 20. Lala PK, Scodras JM, Graham CH, Lysiak JJ, Parhar RS. Activation of maternal killer cells in the pregnant uterus with chronic indomethacin therapy, IL-2 therapy, or a combination therapy is associated with embryonic demise. Cell Immunol 1990; 127:368-381. 21. Low BG, Hansen PJ. Actions of steroids and prostaglandins secreted by the placenta and uterus of the cow and ewe on lymphocyte proliferation in vitro. Am J Reprod Immunol Microbiol 1988; 18:71-75. 22. Shemesh M, Milaguir F, Ayalon N, Hansel W. Steroidogenesis and prostaglandin synthesis by cultured bovine blastocysts. J Reprod Fertil 1979; 56:181-185. 23. Lewis GS, Thatcher WW, Bazer FW, Curl JS. Metabolism of arachidonic acid in vitro by bovine blastocysts and endometrium. Biol Reprod 1982; 27:431-439. 24. Scodras JM, Parhar RS, Kennedy TG, Lala PK. Prostaglandin-mediated inactivation of natural killer cells in the murine decidua. Cell Immunol 1990; 127:352-367. 25. Bazer FW, Thatcher WW, Hansen PJ, Mirando MA, Ott TL, Plante C. Physiological mechanisms of pregnancy recognition in ruminants. J Reprod Fertil Suppl 1991; 43:39-47. 26. Asselin E, Bazer FW, Fortier MA. Recombinant ovine and bovine interferons tau regulate prostaglandin production and oxytocin response in cultured bovine endometrial cells. Biol Reprod 1997; 56: 402-408. 27. Tangri S, Raghupathy R. Expression of cytokines in placentas of mice undergoing immunologically mediated spontaneous fetal resorptions. Biol Reprod 1993; 49:850-856. 28. Marzi M, Vigano A, Trabattoni D, Villa ML, Salvaggio A, Clerici E, Clerici M. Characterization of type I and type 2 cytokine production profile in physiologic and pathologic human pregnancy. Clin Exp Immunol 1996; 106:127-133. 29. Athanassakis I, Bleackley RC, Paetkau V, Guilbert L, Barr PJ, Wegmann TG. The immunostimulatory effect of T cells and T cell lymphokines on murine fetally derived placental cells. J Immunol 1987; 138:37-44. 30. Loke YW, King A, Gardner L, Carter NP Evidence for the expression of granulocyte-macrophage colony-stimulating factor receptors by human first trimester extravillous trophoblast and its response to this cytokine. J Reprod Immunol 1992; 22:33-45. 31. Garcia-Lloret MI, Morrish DW, Wegmann TG, Honore L, Turner AR, Guilbert LJ. Demonstration of functional cytokine-placental interactions: CSF-I and GM-CSF stimulate human cytotrophoblast differentiation and peptide hormone secretion. Exp Cell Res 1994; 214:4654. 32. Fortin M, Ouellette MJ, Lambert RD. TGFI and PGE in rabbit 2 2 blastocoelic fluid can modulate GM-CSF production by human lymphocytes. Am J Reprod Immunol 1997; (in press). 33. Cobb SP, Watson ED. Immunohistochemical study of immune cells in

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

44.

45.

46.

47.

48.

49.

50.

51.

52.

53.

the bovine endometrium at different stages of the oestrous cycle. Res Vet Sci 1995; 59:238-241. Vander Wielen AL, King GJ. Intraepithelial lymphocytes in the bovine uterus during the oestrous cycle and early gestation. J Reprod Fertil 1984; 70:457-462. Bartol E Roberts RM, Bazer FW, Lewis G, Godkin J, Thatcher WW. Characterization of proteins produced in vitro by periattachment bovine conceptuses. Biol Reprod 1985; 32:681-693. Helmer SD, Hansen PJ, Thatcher WW, Johnson JW, Bazer FW. Intrauterine infusion of highly enriched bovine trophoblast protein-I complex exerts an antiluteolytic effect to extend corpus luteum lifespan in cyclic cattle. J Reprod Fertil 1989; 87:89-101. Skopets B, Li J, Thatcher WW, Roberts RM, Hansen PJ. Inhibition of lymphocyte proliferation by bovine trophoblast protein-1 (type I trophoblast interferon) and bovine interferon-alpha II. Vet Immunol Immunopathol 1992; 34:81-96. Assal-Meliani A, Kinsky R, Martal J, Chaouat G. In vivo immunosuppressive effects of recombinant ovine interferon-tau (trophoblastin): r.oTP (roIFN-tau) inhibits local GVH reaction in mice (PLN assay), prevents fetal resorptions, and favors embryo survival and implantation in the CBA/J x DBA/2 mice combination. Am J Reprod Immunol 1995; 33:267-275. Guillomot M, Guay P Ultrastructural features of the cell surfaces of uterine and trophoblastic epithelia during embryo attachment in the cow. Anat Rec 1982; 204:315-322. Chouaib S, Welte K, Mertelsmann R, Dupont B. Prostaglandin E2 acts at two distinct pathways of T lymphocyte activation: inhibition of interleukin 2 production and down-regulation of transferrin receptor expression. J Immunol 1985; 135:1172-1179. Sottile A, Venza I, Venza M, Valenti A, Teti D. PGE -induced im2 munoregulation mediated by cytokine production from cultures of human peripheral T lymphocytes. Immunopharmacol Immunotoxicol 1996; 18:27-36. Borger P, Kauffman HF, Vijgen JL, Postma DS, Vellenga E. Activation of the cAMP-dependent signaling pathway downregulates the expression of interleukin-3 and granulocyte-macrophage colony-stimulating factor in activated human T lymphocytes. Exp Hematol 1996; 24:108115. Senior J, Marshall K, Sangha R, Baxter GS, Clayton JK. In vitro characterization of prostanoid EP-receptors in the non-pregnant human myometrium. Br J Pharmacol 1991; 102:747-753. Quill H, Gaur A, Phipps RP. Prostaglandin E2 -dependent induction of granulocyte-macrophage colony-stimulating factor secretion by cloned murine helper T cells. J Immunol 1989; 142:813-818. Li TK, Fox BS. Effect of prostaglandin E2 (PGE 2) on IL-3/granulocyte-macrophage colony-stimulating factor production by T helper cells. Mode of stimulation and presence of costimulation can determine response to PGE .2 J Immunol 1993; 150:1680-1690. Heuwieser W, Grunert E, Hoppen HO. Maternal and fetal prostaglandin concentrations during late gestation in dairy cattle. Prostaglandins 1993; 45:35-46. Silvia WJ, Ottobre JS, Inskeep EK. Concentrations of prostaglandins E2 , F2 , and 6-keto-prostaglandin F1, in the utero-ovarian venous plasma of nonpregnant and early pregnant ewes. Biol Reprod 1984; 30: 936-944. Asselin E, Goff AK, Bergeron H, Fortier MA. Influence of sex steroids on the production of prostaglandins F2, and E2 and response to oxytocin in cultured epithelial and stromal cells of the bovine endometrium. Biol Reprod 1996; 54:371-379. Fortier MA, Boulet AP, Dugre FJ, Lambert RD. Local alteration in adenylate cyclase activity and stimulation response at implantation site in rabbit endometrium during early pregnancy. Biol Reprod 1990; 42: 106-113. Mayer P, Lam C, Obenaus H, Liehl E, Besemer J. Recombinant human GM-CSF induces leukocytosis and activates peripheral blood polymorphonuclear neutrophils in nonhuman primates. Blood 1987; 70:206-213. Singh SM, Sone S, Inamura N, Ogura T. Up-regulation by granulocyte-macrophage colony-stimulating factor (GM-CSF) of induction of lymphokine (IL-2)-activated killer (LAK) cells by human blood monocytes. Int J Cancer 1989; 44:170-176. Bennett WA, Brackin MN, Whitworth NS, Cowan BD. Granulocyte macrophage-colony-stimulating factor increases interleukin-10 secretion by the bewo cloned choriocarcinoma cell-line. Biol Reprod 1996; 54(suppl 1):83 (abstract). Wegmann TG, Lin H, Guilbert L, Mosmann TR. Bidirectional cyto-

PGE 2 REGULATES IL-2 AND GM-CSF IN BOVINE PBL kine interactions in the maternal-fetal relationship: is successful pregnancy a TH2 phenomenon? Immunol Today 1993; 14:353-356. 54. Robertson SA, Mayrhofer G, Seamark RE Uterine epithelial cells synthesize granulocyte-macrophage colony-stimulating factor and interleukin-6 in pregnant and nonpregnant mice. Biol Reprod 1992; 46: 1069-1079. 55. Giacomini G, Tabibzadeh SS, Satyaswaroop PG, Bonsi L, Vitale L, Bagnara GP, Strippoli P, Jasonni VM. Epithelial cells are the major source of biologically active granulocyte macrophage colony-stimu-

151

lating factor in human endometrium. Hum Reprod 1995; 10:32593263. 56. Kennedy TG. Eicosanoids and blastocyst implantation. In: Mitchell MD (eds.), Eicosanoids in Reproduction. Boston: CRC Press, Inc.; 1990: 123-138. 57. Gallin JI. Inflammation. In: Paul WE (ed.), Fundamental Immunology, 3rd ed. New York: Raven Press; 1993: 1015-1032. 58. Lupulescu A. Prostaglandins, their inhibitors and cancer. Prostaglandins Leukotrienes Essent Fatty Acids 1996; 54:83-94.