|
|
||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Female Reproductive Tract |
1
Unité de Recherche en Ontogénie et Reproduction,3
Centre de Recherche du Centre Hospitalier de l'Université Laval, Centre de Recherche en Biologie de la Reproduction,4
Département d'Obstétrique et Gynécologie,5 Université Laval, Ste-Foy, Quebec, Canada G1V 4G2
Nova Scotia Agricultural College (Plant and Animal Sciences),6 Truro, Nova Scotia, Canada B2N 5B1
| ABSTRACT |
|---|
|
|
|---|
) and prostaglandin E2, increase the expression of cyclooxygenase (COX)-2 or granulocyte-macrophage colony-stimulating factor (GM-CSF) in immune and nonimmune cells of the bovine endometrium. Two experiments were conducted to verify the validity of this hypothesis in vivo. In experiment 1, the in vivo expression of COX-2 and GM-CSF during early pregnancy was monitored. Uteri from heifers were collected at different days (d) of the estrous cycle and pregnancy (P). In experiment 2, the effects of intrauterine infusions of IFN-
on the expression of COX-2 and GM-CSF were analyzed. Immunohistochemistry was performed on uterine sections, and image analysis was used to evaluate the staining intensity in the conceptus, the luminal epithelium (LE), and the subepithelial stroma. In experiment 1, staining for COX-2 was maximal between d18P and d24P, both in the LE and in the conceptus, whereas staining for GM-CSF reached a plateau between d18P and d30P in the LE. In experiment 2, in response to IFN-
, COX-2 was up-regulated in the LE of the ipsilateral horn, whereas GM-CSF was enhanced in both uterine horns. The current report supports the view that the conceptus, through its secretion of IFN-
, stimulates maternal epithelial expression of COX-2 and GM-CSF during the peri-attachment period in the cow.
conceptus, female reproductive tract, growth factors, pregnancy, uterus
| INTRODUCTION |
|---|
|
|
|---|
), the pregnancy recognition signal, produced by the bovine conceptus between Days (d) 1228 of pregnancy (P), peaking at d15Pd19P [810]. Indeed, IFN-
inhibits the expression of estrogen receptors in the ovine uterine epithelium, preventing up-regulation of the oxytocin receptor (OTR) by estradiol and subsequent generation of luteolytic pulses of prostaglandin (PG) F2
by oxytocin (OT) [6]. In cattle, the relationship between IFN-
, estrogen receptor, and OTR is not clear, and IFN-
may directly inhibit the expression of OTRs [11, 12].
Pregnancy or intrauterine treatment with IFN-
also induces the expression of several proteins that can be classified as chemokines, antiviral proteins, cytokines, cell surface markers, signal transducers, and transcription factors. In the endometrium, IFN-
has been shown to stimulate granulocyte chemotactic protein-2 [13] and monocyte chemotactic protein-1 and -2 [14], IFN-
-inducible protein 10 kDa [15], IFN-stimulated gene product 17 [16], 2'5'-oligoadenylate-synthetase [17], 18U and Leu-13 [18], major histocompatibility complex class I and beta-2-microglobulin [19], Mx protein [20, 21], signal transducer and activator of transcription 1 and 2, interferon regulatory factor (IRF) 1 and IRF9 [22]. In the ovine uterus, the receptor for type I interferons is predominantly expressed in the luminal epithelium (LE) and shallow glands (SG) but also in the caruncular stroma [23]. However, binding of IFN-
is limited to the LE and SG [23, 24]. In immune cells, IFN-
stimulates the expression of interleukin (IL)-4, IFN-
[25], IL-2, and IL-10 [26]. It is also active in vivo across species, preventing fetal resorptions [27] and experimental allergic encephalomyelitis [28] in the mouse.
Additionally, IFN-
has been shown to regulate PG synthesis in endometrial cells in vitro, even though reports are contradictory because the responses appear to depend upon cell system and dose. At low doses, IFN-
inhibits the expression of cyclooxygenase (COX)-2 and the synthesis of PGs in endometrial epithelial and stromal cells, whereas at high doses it is stimulatory [2932]. Thus, IFN-
would either suppress the expression of COX-2 and the production of luteolytic PGF2
or induce COX-2 and the synthesis of luteoprotective/luteotropic PGE2 [3335]. However, COX-2 itself does not discriminate between PGF2
and PGE2 because it catalyzes the transformation of arachidonic acid into the same precursor for both PGF2
and PGE2, PGH2, which is rapidly converted by PGF-synthase [36] or PGE-synthase [37], among others [38]. It is notable that in many species, endometrial expression of COX-2 is stimulated during the attachment period [3942]. Granulocyte-macrophage colony-stimulating factor (GM-CSF) is another important factor expressed in the uterine epithelium of the mouse [43], human [44], and ruminants [45, 46] because it is known to promote the survival and growth of embryos in these species [4750].
We have previously demonstrated, with cultured bovine endometrial cells and lymphocytes, that IFN-
or PGE2 enhance the secretion of PGE2 via COX-2 stimulation [29] and the expression of GM-CSF [51, 52]. In the present study, immunohistochemistry (IHC) was used to localize both COX-2 and GM-CSF in the bovine endometrium. Regulation of these two proteins during the estrous cycle and early pregnancy, and after intrauterine infusions of IFN-
, was assessed by image analysis. The results suggest that the conceptus, via its production of IFN-
, up-regulates COX-2 and GM-CSF during the peri-implantation period, thus delaying luteolysis and redirecting the maternal response to its advantage, favoring its own growth and escaping immune rejection [29, 51, 52].
| MATERIALS AND METHODS |
|---|
|
|
|---|
All procedures performed were in accordance with the guidelines of the Canadian Council on Animal Care and were reviewed and approved by the Nova Scotia Agricultural College Animal Care and Use Committee. In experiment 1, 12 sexually mature, healthy mixed-breed beef heifers were slaughtered at government-inspected abattoirs on Days 0 (estrus), 7, 16, and 18 of the estrous cycle (n = 3 per day). To confirm the stage of the estrous cycle, animals were observed for estrous behavior, and blood samples taken at estrus and slaughter were assayed for progesterone (P4) by RIA with a Coat-A-Count kit (Diagnostic Products Corporation, Los Angeles, CA). Ovarian dating was also performed at collection according to the criteria of Ireland et al [53]. Another group of 18 heifers was observed for at least one natural estrous cycle before breeding by artificial insemination. Reproductive tracts were collected on Days 7, 16, 18, 21, 24, and 30 of gestation (n = 3 per day). These days were chosen based on the IFN-
timeframe (d16P, d18P, d21P, and d24P), with d7P and d30P respectively serving as controls before and after the major output of IFN-
[810]. Pregnancy was confirmed by examination of the ovaries and the presence of embryonic tissue. Uteri from all animals were dissected into 1.5-cm3 cross-sectioned blocks, frozen in liquid nitrogen (-179°C) at the collection site, and transferred to an ultra-low-temperature freezer (-80°C) for storage.
In experiment 2, six healthy, sexually mature mixed-breed beef heifers (1.52 yr of age; 520 ± 31-kg bodyweight) showing a cycle length of 1920 days were randomly assigned to control (n = 3) or IFN-
treatments (n = 3) and treated midcycle with Estrumate (500 mg cloprostenol, Schering Canada Inc., Pointe Claire, QC, Canada) to synchronize estrus. Animals were observed at least twice daily for estrus throughout the experiment. In the morning of Day 14 after estrus (Day 0), each heifer received either vehicle (5 ml, 0.1% BSA in PBS) or recombinant ovine IFN-
(5 ml, 0.25 mg, equivalent to 2.5 x 107 antiviral units) transcervically into the uterine body. Infusions were repeated three times at 12-h intervals. The animals were slaughtered in the afternoon of Day 16, 7 h after the last infusion. Both uterine horns, ipsilateral [ipsi] and contralateral (referring to the corpus luteum), were collected and dissected into 1.5-cm3 cross-sectioned blocks. Fixation in 4% paraformaldehyde was allowed to proceed for 4 h. Samples were then dehydrated and embedded in paraffin and stored at room temperature until processed.
Antibodies and Recombinant Ovine IFN-
A monoclonal mouse antibody against recombinant bovine GM-CSF, GM-CSF17.2, was purchased from VMRD (Pullman, WA). The negative control, mouse immunoglobulins G1 (IgG1), was acquired from Dako Diagnostics (Mississauga, ON, Canada), and the biotinylated goat-anti-mouse secondary antibody was obtained from Jackson Immunoresearch Laboratories (West Grove, PA). Polyclonal rabbit antisera against ovine COX-1 (COX-1 241) and COX-2 (COX-2 243) were donated by Merck Frosst (Kirkland, QC, Canada). Biotinylated goat-anti-rabbit secondary antibody was obtained from Dako Diagnostics. Recombinant ovine IFN-
[54] was kindly donated by Dr. Fuller W. Bazer (College Station, TX). Its amino acid sequence is 80% homologous to bovine IFN-
[55, 56] and it cross-reacts in cattle [29, 32, 52, 57].
Immunohistochemistry
In experiment 1, cryostat cross-sections (6 µm) were prepared from frozen uteri with a Shandon SME cryostat (Fisher Scientific, Nepean, ON, Canada). IHC was performed twice on sections from a total of 30 blocks, each from 18 pregnant or 12 cyclic cows, with three animals per day postestrus. Two or three sections, each from different animals, were mounted on each SuperfrostPlus slide (Fisher Scientific) coated with 3-aminopropyltriethoxysilane (Sigma-Aldrich, Oakville, ON, Canada). Sections were air dried and fixed with 4% paraformaldehyde in PBS (pH 10) for 30 min. Immunostaining procedures were immediately performed at room temperature unless otherwise noted. First, endogenous peroxidase activity was quenched with 3% H2O2 in PBS for 15 min. Nonspecific binding sites were then blocked with 10% goat serum in PBS for 1 h. Antibodies were diluted in 0.5% goat serum in PBS (1/4000 for anti-COX sera or nonimmune serum; 1 µg/ml for GM-CSF-specific antibody or IgG1 control) and applied overnight at 4°C. Sections were subsequently incubated with biotinylated secondary antibody (goat-anti-rabbit for COXs, goat-anti-mouse for GM-CSF) for 40 min and with ABC (Vector Laboratorie, Burlington, ON, Canada) reagent (Elite for GM-CSF) for 30 min. Immunostaining was revealed by using 3-amino-9-ethylcarbazole (AEC) for 12 min. Harris hematoxylin was used for counterstaining, and aqueous mounting medium (Fisher Scientific) was applied on the sections.
In experiment 2, cross-sections (4 µm) were prepared from paraffin-embedded uteri with a microtome. IHC was performed twice on a total of 12 sample blocks, one from each uterine horn of six animals (three controls and three treated). Two sections, one from a control animal the other from an IFN-treated animal, were mounted on each precoated slide. Sections were air dried, incubated at 60°C for 30 min, and kept at 4°C until used. Immunostaining procedures were performed at room temperature unless otherwise noted. First, slides were deparaffinized in xylene and washed with ethanol. Endogenous peroxidase activity was then quenched with 3% H2O2 in methanol for 20 min. After rehydratation, sections were incubated in 0.1 M citrate buffer for 15 min at 95°C but only for GM-CSF immunostaining (this step was not necessary to detect both COXs). The blocking, antibody, and ABC steps were the same as for experiment 1 except for the antibody concentrations (anti-GM-CSF 4 µg/ml; anti-COX 1/2000). AEC was left to reveal for 15 (COXs) or 30 (GM-CSF) min. Mayer hematoxylin was used for counterstaining, and aqueous mounting medium was applied on the sections.
Image Analysis
Slides were observed under a Zeiss Axioskop2 Plus microscope (Toronto, ON, Canada) linked to a digital camera from Diagnostics Instruments (Sterling Heights, MI). Images were captured by the Spot software (Diagnostics Instruments) and analyzed with Image-Pro Plus from Media Cybernetics (Silver Springs, MD). For COX-1 studies, an average of 10 pictures per section were taken, and the number of positive isolated cells in the subepithelial stroma were counted. For COX-2 and GM-CSF studies, an average of 20 pictures per section were taken. Areas of interest, either the conceptus, LE, or subepithelial stroma (S), were separately cropped and then submitted to densitometry analysis. The amount of binucleate cell migration is very extensive between d21P and d30P, rendering it difficult to distinguish between LE, hybrid syncytium, and trophoblastic cells. When the exact nature of the cells was uncertain, they were excluded from both groups (conceptus and LE). Color-cube-based segmentation was used to select only shades of red (AEC) in the area of interest. Integrated optical density (IOD) of the red staining was measured after standard OD calibration.
Statistical Analysis
Assignment of animals to treatments was made at random. Data (dependent variables: intensity of COX-2/GM-CSF staining in the LE/S/conceptus, or number of COX-1-positive cells in the S) were subjected to least-squares factorial ANOVA by SuperANOVA software (Abacus Concepts, Berkeley, CA) followed by regression analysis or paired t-tests when main effects were significant (P < 0.05). In experiment 1, the effects of day within status (cyclic or pregnant) were examined by regression analysis. Data from days 7, 16, and 18 were also examined statistically across status. No significant difference was found between animals within the same group of days (animal; animal x horn, ANOVA, P > 0.05). After analysis, IOD/area data (±SEM) were converted to percentages (d7P or d18P = 100%) for ease of reporting. In experiment 2, main sources of variations included IFN-
treatment (0 or 1 mg total), side of the uterine horn (ipsi vs. contralateral), and associated interaction (treatment x horn). No significant difference was found between animals within the same treatment groups (animal; animal x horn, ANOVA, P > 0.1). IOD/area data (±SEM) were also converted to percentages (untreated control = 100%) after analysis.
| RESULTS |
|---|
|
|
|---|
The expression of COX-1 in the conceptus (Fig. 1, FH) and major compartments (epithelium, stroma, blood vessels, and myometrium) of the uterus was either very weak or nonexistent, with the exception of intensely stained, leukocyte-like cells that were concentrated in the S but absent from the LE or glandular epithelium (Fig. 1). A few similar COX-1-positive cells could also be found deeper in the myometrium, particularly around blood vessels (not shown). COX-1 staining was both nuclear and cytoplasmic (Fig. 1I). Nonimmune serum, used as a negative control for COX-1 antiserum, showed no reactivity (Fig. 6G). The number of COX-1-positive cells in the S did not vary throughout the estrous cycle (regression analysis, P > 0.5), whereas it was doubled from d7P to d16P, decreasing (regression analysis, R2 = 0.25, slope = -4, P < 0.005) to cycle-level afterward (Fig. 3A). When d7, d16, and d18 were compared across status (pregnant vs. cyclic) by regression analysis, an effect of status was found (increased in pregnant cows, P < 0.05).
|
|
|
Immunohistochemical Localization of COX-2 in the Bovine Uterus
As shown in Figure 2, immunostaining for COX-2 was most intense in the trophectoderm layer of the conceptus, particularly in uninucleate trophoblastic cells (Fig. 2, F and G). On the other hand, underlying mesoderm was negative (Fig. 2, FI) and binucleate cells were weakly stained (Fig. 2, GI). In maternal tissues at all stages of the cycle or pregnancy, the staining intensity followed this order: LE > SG > deep glands (DG) > S. The exceptions are d16P (Fig. 2E) in which all maternal compartments expressed COX-2 at a similar level and d30P (Fig. 2I) where COX-2 was darker in SG than in LE (not shown). COX-2 staining was also reduced at attachment sites in fused epithelial cells (Fig. 2, H and I), and a faint signal was sometimes apparent in the myometrium (Fig. 6B). The compactness of the stroma did not permit the separate analysis of the three contributors to stromal COX-2, which are fibroblasts, capillaries, and leukocytes. No difference was observed between caruncular and intercaruncular regions (not shown). In general, intracellular localization of COX-2 was cytoplasmic and also perinuclear, particularly in trophoblastic cells (Fig. 6A). Again, nonimmune serum showed no staining (Fig. 6G).
|
COX-2 Immunostaining Is Increased in LE During Early Pregnancy
The intensity of the red staining (Fig. 2) in the conceptus, LE, and S was evaluated by image analysis (Fig. 3, BD). COX-2 staining linearly decreased with time (Fig. 3B) in the conceptus between d18P and d30P (R2 = 0.3, slope = -4, P < 0.005). In the LE, on the other hand, no significant effect of day was observed (Fig. 3C) during the cycle (regression analysis, P > 0.5), even though COX-2 intensity appeared stronger at d16C, the day of luteolysis. However, COX-2 was significantly stimulated during pregnancy (Fig. 3C) in the LE from d7P to d30P (R2 = 0.20, slope = 10, P < 0.005). COX-2 was expected to follow the pattern of IFN-
production, thus being weakly expressed at d7P, increasing through d21P, and decreasing afterward. When examined separately by regression analysis, the up-regulation through time was confirmed between d7P and d21P (R2 = 0.4, slope = 19, P < 0.0005) but not between d21P and d30P, which rather showed a decreasing tendency, a regression plot with a negative slope (R2 = 0.1, slope = -13, P > 0.1). When d7, d16, and d18 were compared across status (pregnant vs. cyclic) by regression analysis, no effect of status was found (P > 0.1). In the S (Fig. 3D), no significant effect of day on COX-2 staining was observed either during the cycle or the pregnancy (regression analysis, P > 0.1). When d7, d16, and d18 were compared across status (pregnant vs. cyclic), again no effect of status was found (regression analysis, P > 0.1).
Immunohistochemical Localization of GM-CSF in the Bovine Uterus
Immunostaining for GM-CSF was observed in the conceptus at the surface of the trophectoderm at d18P (Fig. 4F) and sometimes in d21Pd24P in regions where the attachment process was not advanced (Fig. 6F). In maternal tissues, GM-CSF staining was concentrated in the apical portions of the LE and glandular epithelium, as well as in the basement membrane (Fig. 4). The staining grades were similar to those observed with COX-2: LE > SG > DG > S. However, expression levels were similar among these tissue types at d16P (Fig. 4E and 6C) and darker in SG compared with LE at d30P (not shown). On the other hand, the myometrium was positive only at d30P (Fig. 6D). GM-CSF expression in the stroma could mainly be credited to capillaries, leukocyte-like cells, and fibroblasts in the stratum compactum. In all cell types, GM-CSF was clearly confined to the cytoplasm. No difference was observed between caruncular and intercaruncular regions (not shown). The strong signal typical of the LE also decreased at attachment sites in syncytial regions (Fig. 4, H and I). An example of "advanced" fusion at d21P between trophoblast and LE is illustrated in Figure 6E (vs. Fig. 4G), along with an unfused region that probably comes from the extremities of the conceptus at d24P (Fig. 6F vs. Fig. 4H). Control IgG1, targeting a nonmammalian protein, was negative (Fig. 6H).
|
GM-CSF Immunostaining Is Increased in LE During Early Pregnancy
The intensity of the red staining (Fig. 4) in the LE and S was evaluated by image analysis, first by comparing between days of the estrous cycle or pregnancy. In the LE (Fig. 5A), no significant effect of day was observed during the cycle (regression analysis, P > 0.5). On the other hand, GM-CSF was significantly stimulated during pregnancy in the LE from d7P to d30P (R2 = 0.25, slope = 11, P < 0.01). As with COX-2, GM-CSF was expected to follow the pattern of IFN-
secretion. When examined separately by regression analysis, the up-regulation of GM-CSF in LE through time was confirmed between d7P and d21P (R2 = 0.38, slope = 14, P < 0.005), but a plateau was rather obtained between d21P and d30P (R2 = 0, slope = -0.5, P > 0.5). When d7, d16, and d18 were compared across status (pregnant vs. cyclic) by regression analysis, no effect of status was found (P > 0.5). Stromal expression of GM-CSF (Fig. 5B) was not statistically altered through time either during the cycle or the pregnancy (regression analysis, P > 0.5). When d7, d16, and d18 were compared across status (pregnant vs. cyclic) by regression analysis, again no effect of status was found (P > 0.1).
|
COX-1-Positive Cells Are Not Significantly Decreased after IFN-
Treatment
In experiment 2, the weak expression of COX-1 was not regulated in the LE or S in response to IFN-
treatment (Fig. 7, A and B). Isolated COX-1-positive cells in the S were counted (Fig. 8A) after treatment, and their number was comparable with the untreated control (treatment; treatment x horn, ANOVA, P > 0.1). Again, nonimmune serum showed no reactivity (Fig. 7I).
|
|
COX-2 Immunostaining Is Increased in LE of the Ipsi Horn after IFN-
Treatment
After IFN-
treatment, the staining for COX-2 (Fig. 7, CF) in the LE was not significantly enhanced (Fig. 8B) when only the effect of treatment was examined (ANOVA, P > 0.05). However, the effect of the interaction between treatment and side of the uterine horn (Fig. 8C) was significant (treatment x horn, P < 0.05). The striking stimulation of COX-2 observed in the ipsi horn to the corpus luteum (Fig. 7, E vs. F) reached a 3.6 ± 0.5-fold increase over untreated control (control x ipsi vs. IFN x ipsi, paired t-test, P < 0.01). On the other hand, the expression of COX-2 in the LE of the contralateral (contra) horn (Fig. 7, C vs. D) was not significantly different (Fig. 8C) between treatment and control (control x contra vs. IFN x contra, paired t-test, P > 0.5). Abundance of COX-2 in both horns of control cows was similar (control x ipsi vs. control x contra, paired t-test, P > 0.1), whereas levels of COX-2 were higher after treatment in ipsi horns compared with contralateral (IFN x ipsi vs. IFN x contra, paired t-test, P < 0.05). The signal for COX-2 was not significantly altered (treatment; treatment x horn, ANOVA, P > 0.1), even when ipsi and contralateral horns were compared, in the stroma (Fig. 8D), the glandular epithelium, the blood vessels, and the myometrium (not shown). Nonimmune serum showed no staining (Fig. 7I).
GM-CSF Immunostaining Is Increased in LE after IFN-
Treatment
Expression levels of GM-CSF (Fig. 7, G and H) in the LE (Fig. 8E) were significantly increased after treatment (ANOVA, P < 0.05). The intensity of the signal reached 1.5 ± 0.2-fold over control values (control vs. IFN, paired t-test P < 0.05). Immunostaining was not different between both uterine horns (treatment x horn, ANOVA, P > 0.5). In the stroma (Fig. 8F), immunostaining for GM-CSF was not significantly altered (treatment; treatment x horn, ANOVA, P > 0.1), and no variation could be found between treatment and control in glandular epithelium, blood vessels, and myometrium (not shown). Control IgG1 showed no reactivity (Fig. 7J).
| DISCUSSION |
|---|
|
|
|---|
treatment. Indeed, the secretion of IFN-
(between d12Pd28P, peaking at d15Pd19P [8, 9]) by uninucleate trophoblastic cells [10] is concomitant with dramatic cellular modifications that happen to both the uterine epithelium and the conceptus. Early attachment in ruminants is divided in three main stages [3, 4, 58]. The first is characterized by the exponential elongation of the conceptus; the second involves apposition between the trophoblast and the flattened uterine epithelium; and the third is the adhesion stage during which trophoblastic binucleate cells fuse with LE cells to form hybrid syncitia, accounting for 50% of the maternal layer by d24P. The attachment begins in the vicinity of the embryo then spreads throughout, explaining why fusion of trophoblastic and epithelial cells is less advanced at the extremities of the conceptus. Uterine LE is quickly reconstituted after d30P, however, and giant multinucleate cells are absent from d40P. On the basis of results obtained in vitro, it has been proposed that IFN-
stimulates COX-2, PGE2, PGF2
, and GM-CSF in the endometrium [29, 51, 52]. Expression patterns of COX-2 and GM-CSF reported in the present study fit well with this hypothesis. COX-2 and GM-CSF are strongly expressed during early pregnancy in regions where fusion has not yet occurred, whereas the staining for the two proteins is reduced at sites of attachment (in the hybrid syncytium) as was reported for IFN-
[10]. This supports the view that COX-2 and GM-CSF are important mediators during the elongation and apposition periods, preparing the materno-fetal interface for the adhesion stage. It is possible that COX-2 or GM-CSF levels would be re-established after d30P when the LE is reconstituted, but this question could not be addressed in the current study.
Because the embryo attachment process and the embryonic signal (IFN-
) are very similar between the cow and the ewe, it has been hypothesized that molecular events occurring during pregnancy recognition should also be closely related. However, the expression of COXs in the endometrium differs significantly between the two ruminant species. COX-1 is expressed at steady-state levels in the sheep during the cycle and early pregnancy in both epithelial and stromal cells [39]. In contrast, COX-1, known as a constitutive isoform in most tissues, is very weakly expressed in the bovine endometrium at all stages, confirming previous results obtained during the cycle [37] and third-trimester pregnancy [59]. In the ewe, endometrial expression of COX-2 is limited to the LE and SG [39, 60] and is detectable only at peak expression between Day 12 and Day 17 of the cycle or pregnancy [39]. In the cow, staining for COX-2 is observed at all stages, even in the stroma and in DG. Nonetheless, four major resemblances remain between the two species regarding the expression of COXs: first, COX-1 is undetectable in the conceptus [61]; second, COX-2 expression in the conceptus decreases with time [61]; third, endometrial COX-2 is high in the late phase of the estrous cycle [37, 39, 60]; and fourth, endometrial COX-2 is stimulated during early pregnancy [39, 60, 62]. Thus, in both species, COX-2 is high in the endometrium at the time of onset of luteolysis and maximal at the materno-fetal interface during establishment of pregnancy. COX-2 null female mice have reproduction failures at multiple levels [63]. In accordance, the current study supports the assertion that COX-2 is responsible for the regulation of PG synthesis in the endometrium and conceptus. However, the physiological role of COX-1-positive cells in the bovine stroma remains to be determined. Their exact identity is unknown at this time, although they are presumed to be leukocytes [38] based on their pattern of distribution [64, 65].
In the present study, the increase in COX-2 expression after treatment with IFN-
was observed only in the ipsi horn of the uterus. The same side-specific effect was reported when studying the regulation of the EP2 receptor [66]. On the other hand, there is no similar difference in the regulation of GM-CSF, and IFN receptors are expressed uniformly in the ipsi and the contralateral horns [67]. Because the antiluteolytic effects of IFN-
are P4 dependent [68], and because P4 stimulates endometrial synthesis of PGs [30, 39, 69] and is more concentrated in the ipsi horn [70], it is possible that the effect of IFN-
on the PG axis is favored in the ipsi horn. During pregnancy, PGE2 concentrations are more elevated in the ipsi horn, and this may reflect both the contribution of the conceptus and the up-regulation of COX-2 in the maternal epithelium [71]. It is well known that COX-2 is an important regulator of the implantation process [63, 72] and that attachment begins in the ipsi horn before spreading to the contralateral horn [58]. Moreover, embryos transferred in the ipsi horn have greater chances of survival, whereas embryos deposited in the contralateral horn have a tendency for transuterine migration [73].
Pregnancy recognition in ruminants occurs when IFN-
from the trophoblast prevents the increase of OTRs in the LE, thus disrupting luteolytic pulses of PGF2
[6]. Because a d16 embryo can be successfully transferred to a cycling cow at d16 (the day that luteolysis is initiated [74]), it has been suggested that IFN-
also has a quick, direct effect on endometrial synthesis of PGs [29]. In the present study, the expression of COX-2 is strongly enhanced in the LE during the pregnancy recognition period and after IFN-
treatment. COX-2 might be involved in opposed physiological events, forcing the female reproductive tract either to return to the cycle or to accommodate the embryo, because both luteolysis/OT treatment [75, 76] and pregnancy recognition/IFN-
treatment are accompanied by an elevated expression of COX-2 in the LE [39, 60, 62]. Indeed, COX-2 generates PGH-2, the common precursor for PGF2
, PGE2, and PGI2to name only those PGs well known to participate at the materno-fetal interfacewhich are respectively luteolytic [77], antiluteolytic/luteotropic [3335, 78], and required during decidualization and implantation in mice [72]. It can thus be expected that the fate of the corpus luteum depends in part upon the balance between luteotropic PGE2 and luteolytic PGF2
. Even though an elevated COX-2 expression during the periods of luteolysis and pregnancy recognition would result in the simultaneous synthesis of both PGs, PGE2 is not sufficient to prevent luteolysis during a natural cycle and endogenous PGF2
is not sufficient to trigger luteolysis during early pregnancy [79, 80]. Results therefore suggest that the corpus luteum is differentially resistant to both types of PGs.
Indeed, levels of PGF2
remain elevated during early pregnancy [62, 81, 82] or in cultured conceptus [61, 83], just like the expression of COX-2, and luteolysis is nonetheless prevented [79, 80]. It is well known that the pulsatility in PGF2
production, rather than the absolute quantity, is the major factor that triggers luteolysis [77]. Because this pulsatility is itself regulated by pulses of OT and that OTRs are undetectable during early pregnancy [12, 84], high quantities of endometrial COX-2 and PGF2
do not seem to be a critical issue at this time. On the other hand, PGE2 production is also stimulated during early pregnancy [71, 81] and PGE2 secretion in bovine conceptus from d10P to d19P is increased in vitro [83, 85]. The high COX-2 staining in the endometrium and the conceptus between d18P and d24P supports an increase in PGs during early pregnancy. Enzymes downstream of COX-2, like PGE synthase [37, 86] and PGF synthase [36], could also be regulated to favor the synthesis of one type of PG, because there is a correlation between the up-regulated COX-2 and the increase in PGE2, not PGF2
, in the endometrium of the ewe [87]. Indeed, COX-2 and PGE synthase exhibit similar patterns of expression during the cycle in the bovine endometrium [37], and there is a correlation between the expression of COX-2 and PGE synthase and the production of PGE2 in vitro [86]. Additionally, PGE2 exhibits functions on immune cells that are not regulated by PGF2
(or PGI2) and that could be critical to maternal tolerance of the conceptus [88]. Such functions include the differentiation of leukocytes [89, 90] into cells that are less lytic [91], more granulated [92], or that produce growth factors [51] or Th2-type cytokines [90, 93]. It would be interesting to look at the spatio-temporal expression of both PGE synthase and PGF synthase in the endometrium to monitor any regulation in the expression of enzymes downstream of COX-2. In summary, we propose that the up-regulation of COX-2 is beneficial in establishing pregnancy even if it is accompanied by higher PGF2
, because endogenous PGF2
is unable to trigger luteolysis at this time, probably because it lacks pulsatility and is counterbalanced by the concomitant increase in PGE2, which also has the potential to prevent immune rejection of the conceptus.
We report here that GM-CSF was localized on the bovine trophectoderm for only a short period, between d18P and d24P, in regions of the trophectoderm where attachment was not advanced. It was also found to be expressed in human and murine trophoblast [94, 95], although it was undetectable in d17 ovine conceptus [45]. The present report confirms the expression of GM-CSF in the LE, SG, DG, endometrial vasculature, and isolated cells in the stroma during the estrous cycle in the cow [46]. GM-CSF is secreted in the uterine lumen because it is detectable in flushings from nonpregnant but particularly from pregnant cows [46]. Accordingly, endometrial GM-CSF is up-regulated during pregnancy in both the ewe [96] and the cow, and the moderate stimulation of GM-CSF in the LE after exposure to IFN-
also fits with this expression pattern. GM-CSF is known to stimulate human and murine placento-fetal growth, differentiation and survival, in vitro and in vivo [47, 48]. Moreover, GM-CSF promotes development of in vitro-produced bovine embryos [97] and may up-regulate IFN-
production [45, 98, 99]. Culture of murine embryos with GM-CSF enhances blastocyst formation and attachment, as well as glucose metabolism and viability of the inner cell mass [49]. A similar regimen prevents inner cell mass apoptosis in human embryos [50]. In GM-CSF knock-outs, as expected, resorptions are increased and fetal weights are reduced because of placental dysfunctions, and even postnatal deaths are more frequent [100]. Thus, these present results suggest that the bovine endometrium accommodates the conceptus by providing a powerful growth/antiapoptotic factor to assist it during the demanding period of elongation and attachment.
The present study lends support to our previous hypothesis built on data obtained with in vitro experiments, that IFN-
stimulates both COX-2 [29] and GM-CSF [52] in the endometrium. However, a few differences were observed. In vitro, IFN-
induced the expression of COX-2 in both epithelial and stromal cells and up-regulated GM-CSF in stromal cells and leukocytes. In vivo, the effect on immune cells could not be verified because their GM-CSF expression was not intense enough to allow exact discrimination between intraepithelial leukocytes and epithelial cells and between stromal leukocytes and capillaries or fibroblasts. On the other hand, the expression pattern of the two proteins in the LE in vivo followed the window of IFN-
synthesis and contrasted with an absence of correlation in the stroma. Moreover, in cows treated with IFN-
, COX-2 and GM-CSF were increased only in the LE, not in the stroma. These differences could be explained in part because, usually, epithelial and stromal cells are cultured separately in vitro, whereas they are free to interact in vivo where the LE is also exposed to higher concentrations of mediators from the conceptus, compared with the stroma. Discrepancies were also reported among in vitro studies conducted by different teams [2931], and distinct responses to IFN-
might depend upon the concentration or isoform used [32]. Overall, this reminds us that care must be taken in evaluating observations made in vitro, and that performing in vivo protocols in parallel often offers physiological clarification. Although this may seem obvious, our own experience and examples from other labs prove this notion can often be forgotten. The sole expression of IFN receptors or any type of response to the molecule in vitro does not guarantee that stromal cells are a direct target of IFN-
or that the response obtained really has a physiological counterpart. Three observations suggest that IFN-
acts primarily on cells of the LE and shallow glandular epithelium (sGE): 1) IFN receptors are mainly expressed in these cells [23], 2) binding of IFN-
is limited to these cells after intrauterine infusions [23], and 3) IFN-
is detectable only in these maternal cells during pregnancy [10, 24]. Thus, it is likely that the increase in several proteins (or encoding mRNAs) in the stroma or DG after in vivo treatments with IFN-
[13, 1622] could be induced by intermediary molecules originating from the LE/sGE. Intraepithelial leukocytes are also a likely candidate because immune cells are very sensitive to IFN-
[52] and respond to it with increased production of IFN-
[25], another IFN capable of duplicating some of the effects triggered by IFN-
[101].
This paper describes the first evidence that COX-2 and GM-CSF are up-regulated in vivo at the materno-fetal interface during the first month of gestation in the cow, particularly during the window of production of IFN-
by the conceptus. In accordance, COX-2 and GM-CSF were increased in the LE but not in the stroma of cows after intrauterine infusions of IFN-
. The results of the present study thus support the hypothesis that the embryo orchestrates the regulation of maternal mediators in favor of its survival and growth in two major ways during pregnancy recognition: 1) by modulating PG synthesis to protect both the corpus luteum from destruction and the conceptus from immune maternal rejection and 2) through GM-CSF, an antiapoptotic factor that promotes feto-placental development.
| ACKNOWLEDGMENTS |
|---|
and Dr. Peter J. Hansen for providing rbGM-CSF and bovine tissues that were used in preliminary experiments. The authors are also grateful to Dr. Ted Semple and the staff at NSAC, Westmorland Institute, and HUB meat packers for assistance with the animals and are thankful to Ms. Christine Légaré, M. Nicholas Robert, and M. Sylvain Picard for their technical advice. | FOOTNOTES |
|---|
2 Correspondence: Raymond D. Lambert, Ontogénie et Reproduction T1-49, Centre de Recherche du Centre Hospitalier de l'Université Laval, 2705 Blvd. Laurier, Ste-Foy, QC, Canada G1V 4G2. FAX: 418 654 2765; ray.lambert{at}crchul.ulaval.ca ![]()
Received: 25 April 2003.
First decision: 22 May 2003.
Accepted: 28 August 2003.
| REFERENCES |
|---|
|
|
|---|
This article has been cited by other articles:
![]() |
N. Krishnaswamy, G. Danyod, P. Chapdelaine, and M. A. Fortier Oxytocin Receptor Down-Regulation Is Not Necessary for Reducing Oxytocin-Induced Prostaglandin F2{alpha} Accumulation by Interferon-{tau} in a Bovine Endometrial Epithelial Cell Line Endocrinology, February 1, 2009; 150(2): 897 - 905. [Abstract] [Full Text] [PDF] |
||||
![]() |
N. Krishnaswamy, P. Chapdelaine, J. P. Tremblay, and M. A. Fortier Development and Characterization of a Simian Virus 40 Immortalized Bovine Endometrial Stromal Cell Line Endocrinology, January 1, 2009; 150(1): 485 - 491. [Abstract] [Full Text] [PDF] |
||||
![]() |
T. E Spencer, O. Sandra, and E. Wolf Genes involved in conceptus-endometrial interactions in ruminants: insights from reductionism and thoughts on holistic approaches Reproduction, February 1, 2008; 135(2): 165 - 179. [Abstract] [Full Text] [PDF] |
||||
![]() |
T. G Kennedy, C. Gillio-Meina, and S. H. Phang Prostaglandins and the initiation of blastocyst implantation and decidualization Reproduction, November 1, 2007; 134(5): 635 - 643. [Abstract] [Full Text] [PDF] |
||||
![]() |
Y.-T. Tee, C.-P. Han, J.-L. Ko, G.-D. Chen, S.-F. Yang, S.-C. Chen, H.-J. Tsai, L.-Y. Lin, and P.-H. Wang Evaluation of Matrix Metalloproteinase 2 Expression in Cervical Carcinogenesis Using Tissue Array and Integrated Optical Density for Immunoreactivity Reproductive Sciences, October 1, 2007; 14(7): 719 - 726. [Abstract] [PDF] |
||||
![]() |
Y. Chen, E. Antoniou, Z. Liu, L. B Hearne, and R M. Roberts A microarray analysis for genes regulated by interferon-{tau} in ovine luminal epithelial cells Reproduction, July 1, 2007; 134(1): 123 - 135. [Abstract] [Full Text] [PDF] |
||||
![]() |
Z Cheng, E L Sheldrick, E Marshall, D C Wathes, D R E Abayasekara, and A P F Flint Control of cyclic AMP concentration in bovine endometrial stromal cells by arachidonic acid Reproduction, May 1, 2007; 133(5): 1017 - 1026. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. Lapointe, M. Roy, I. St-Pierre, S. Kimmins, D. Gauvreau, L. A. MacLaren, and J.-F. Bilodeau Hormonal and Spatial Regulation of Nitric Oxide Synthases (NOS) (Neuronal NOS, Inducible NOS, and Endothelial NOS) in the Oviducts Endocrinology, December 1, 2006; 147(12): 5600 - 5610. [Abstract] [Full Text] [PDF] |
||||
![]() |
T. R. Bilby, A. Guzeloglu, L. A. MacLaren, C. R. Staples, and W. W. Thatcher Pregnancy, Bovine Somatotropin, and Dietary n-3 Fatty Acids in Lactating Dairy Cows: II. Endometrial Gene Expression Related to Maintenance of Pregnancy. J Dairy Sci, September 1, 2006; 89(9): 3375 - 3385. [Abstract] [Full Text] [PDF] |
||||
![]() |
L Cammas, P Reinaud, N Bordas, O Dubois, G Germain, and G Charpigny Developmental regulation of prostacyclin synthase and prostacyclin receptors in the ovine uterus and conceptus during the peri-implantation period. Reproduction, May 1, 2006; 131(5): 917 - 927. [Abstract] [Full Text] [PDF] |
||||
![]() |
Y. Chen, J. A. Green, E. Antoniou, A. D. Ealy, N. Mathialagan, A. M. Walker, M. P. Avalle, C. S. Rosenfeld, L. B. Hearne, and R. M. Roberts Effect of Interferon-{tau} Administration on Endometrium of Nonpregnant Ewes: A Comparison with Pregnant Ewes Endocrinology, May 1, 2006; 147(5): 2127 - 2137. [Abstract] [Full Text] [PDF] |
||||
![]() |
M. Parent, E. Madore, L. A MacLaren, and M. A Fortier 15-Hydroxyprostaglandin dehydrogenase in the bovine endometrium during the oestrous cycle and early pregnancy. Reproduction, March 1, 2006; 131(3): 573 - 582. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. Parent and M. A. Fortier Expression and Contribution of Three Different Isoforms of Prostaglandin E Synthase in the Bovine Endometrium Biol Reprod, July 1, 2005; 73(1): 36 - 44. [Abstract] [Full Text] [PDF] |
||||
![]() |
C. Sjoblom, C. T. Roberts, M. Wikland, and S. A. Robertson Granulocyte-Macrophage Colony-Stimulating Factor Alleviates Adverse Consequences of Embryo Culture on Fetal Growth Trajectory and Placental Morphogenesis Endocrinology, May 1, 2005; 146(5): 2142 - 2153. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. A. Arosh, S. K. Banu, S. Kimmins, P. Chapdelaine, L. A. MacLaren, and M. A. Fortier Effect of Interferon-{tau} on Prostaglandin Biosynthesis, Transport, and Signaling at the Time of Maternal Recognition of Pregnancy in Cattle: Evidence of Polycrine Actions of Prostaglandin E2 Endocrinology, November 1, 2004; 145(11): 5280 - 5293. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. Guzeloglu, T. R. Bilby, A. Meikle, S. Kamimura, A. Kowalski, F. Michel, L. A. MacLaren, and W. W. Thatcher Pregnancy and Bovine Somatotropin in Nonlactating Dairy Cows: II. Endometrial Gene Expression Related to Maintenance of Pregnancy J Dairy Sci, October 1, 2004; 87(10): 3268 - 3279. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. Guzeloglu, F. Michel, and W. W. Thatcher Differential Effects of Interferon-{tau} on the Prostaglandin Synthetic Pathway in Bovine Endometrial Cells Treated with Phorbol Ester J Dairy Sci, July 1, 2004; 87(7): 2032 - 2041. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. Guzeloglu, P. Subramaniam, F. Michel, and W. W. Thatcher Interferon-{tau} Induces Degradation of Prostaglandin H Synthase-2 Messenger RNA in Bovine Endometrial Cells Through a Transcription-Dependent Mechanism Biol Reprod, July 1, 2004; 71(1): 170 - 176. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |