Placental expression of gonadotropin-releasing hormone (GnRH)-I and II, as well as their cognate receptor, coincides with a period of extensive remodeling of the maternal-fetal interface, near the end of the first trimester of pregnancy. To further define the role of GnRH in human placentation, we performed a microarray screen of HTR-8/SVneo trophoblasts to identify GnRH-regulated genes and their roles in placentation. This screen revealed that GnRH regulates the expression of four angiogenic chemokines: CXCL2, CXCL3, CXCL6, and CXCL8. The microarray data were subsequently confirmed by an extensive Q-PCR time-course analysis. CXCL8, a representative chemokine, was selected for further analysis and shown to be strongly expressed by trophoblasts at the maternal-fetal interface of the human placenta, as well as to accumulate in a GnRH-dependent manner in trophoblast-conditioned media in culture. Trophoblasts were subsequently shown to recruit lymphocytes (Jurkat T cells and primary peripheral blood T and uterine natural killer cells) in chemotaxis assays and this was shown to be GnRH dependent. Furthermore, this recruitment was shown to occur via the release of CXCR1/CXCR2 interacting chemokines, such as the CXCLs investigated in this study. This novel regulation of chemokines by GnRH signaling demonstrates the role of GnRH in regulating the recruitment of lymphocytes to the decidua and the possibility of a direct effect on spiral artery remodeling via the release of proangiogenic chemokines and secondary effects via release of angiogenic factors by recruited lymphocytes.
- angiogenic chemokine
- extravillous trophoblast
toward the end of the first trimester of human pregnancy a number of changes in cytotrophoblast (CTB) behavior occur at the maternal-fetal interface and these are essential for the healthy progression of a pregnancy to term. In the hypoxic environment of the first-trimester placenta, proliferating cell columns of cytotrophoblasts (CTBs) within the tips of anchored chorionic villi will experience an increase in oxygen tension. Unlike other invasive cell types, CTBs proliferate in hypoxic conditions, and when oxygen tension increases, they differentiate down a highly invasive pathway. This leads to the invasion of CTBs into the uterine wall, where they become known as extravillous trophoblasts (EVTs) (13). These EVTs penetrate the uterine tissue invading interstitially through the decidual wall, as well as endovascularly through the lumen of maternal spiral arteries (32).
Within the uterine wall EVTs participate in several functions vital to healthy fetal development. Most importantly, EVTs play a crucial role in the physiological remodeling of the maternal spiral arteries (12, 35). Here it is observed that interstitially invading EVTs in intimate association with maternal blood vessels first induce the apoptosis of smooth muscle and then endothelial cells that surround the blood vessels. At the same time, endovascularly invading EVTs are doing the same but from the opposite direction (1, 17, 20). Remodeling of maternal spiral arteries is observed to occur as far as one-third into the myometrium leading to the transformation of low-capacitance and high-resistance arteries to those of high capacitance and low resistance (10, 23). The production of angiogenic factors by EVTs and decidual lymphocytes, of which uterine natural killer (uNK) cells (CD56BRIGHTCD16−) comprise the major fraction, is thought to aid in this process (25). Failure to remodel the uterus and maternal spiral arteries can lead to placental insufficiency, where the placenta and consequently the fetus do not receive adequate nutrient supply. Several conditions have been associated with improper placentation, such as preeclampsia and intrauterine growth restriction (11).
Gonadotropin-releasing hormone (GnRH) is a decapeptide hormone with a well-understood role in regulation of the hypothalamic-pituitary-gonadal axis (9, 29). However, the role of GnRH in human placentation is not as clearly defined. Early studies of GnRH in the placenta described its presence and a clear role as a regulator of syncytiotrophoblast (STB) produced human chorionic gonadotropin (21, 27). The expression of GnRH-I and GnRH-II, a second nonhypothalamic form of the decapeptide hormone, and their cognate receptor has been described in placental trophoblasts (4, 37). GnRH-I levels are highest in the first-trimester placenta and are produced by CTBs, STBs, EVTs, and decidual cells (4). GnRH-II shows a constant level throughout pregnancy and is produced by CTB, EVTs, and decidual cells (4, 38). The GnRH receptor (GnRH-RI) shows a temporally regulated expression pattern where its maximal expression correlates with maximal GnRH-I expression (27, 38, 42). This occurs in the late first trimester and early part of the second trimester of pregnancy. Previous studies have demonstrated GnRH-I and GnRH-II are regulators of the proteases matrix metalloproteinase (MMP)-2, MMP-9, and urokinase-type plasminogen activator (uPA) in primary EVTs and decidual stromal cells. The increases in gene and protein expression of the MMPs and uPA were coupled to decreases in gene and protein levels of tissue inhibitor of MMP-1 and plasminogen activator inhibitors (5–8). These factors have been shown to regulate EVT invasiveness (15, 25). Taken together, these observations suggest that GnRH signaling plays an important role in regulating the invasiveness of human trophoblasts.
Since the timing of maximal receptor and ligand expression coincides with the period of extensive tissue remodeling at the maternal-fetal interface, we hypothesize that GnRH signaling regulates the expression of genes that play important roles in placentation. To this end, we have conducted a microarray screen designed to identify genes and their encoded products that are regulated by GnRH signaling in the human placenta.
MATERIALS AND METHODS
The human HTR-8/SVneo trophoblast cell line (a kind gift from Dr. Peeyush Lala) was used as a model EVT cell system. This cell line was isolated from explant cultures of first-trimester placenta and immortalized with transfection of the Simian Virus-40 large T-antigen (14). The immortalized cell line shares several properties with primary EVTs (14, 19, 22). Jurkat T lymphocyte cell line used in lymphocyte migration assays was obtained from ATCC. HTR-8/SVneo and Jurkat were cultured in RPMI 1640, 10% FBS, 1% penicillin and streptomycin, 1% nonessential amino acids, 1% sodium pyruvate, and 1% Glutamax-1. HTR-8/SVneo cells stably overexpressing the GnRH-RI coding sequence, subsequently designated HTR-8/SVneo-GnRH-RI, or an empty vector control, both cloned in the pCEP4 plasmid were maintained by hygromycin B selection. BeWo, JEG-3, JAR, and HEK293 cells were also used in this study. BeWo cells were cultured in F-12K nutrient mixture, JEG-3 cells in α-MEM medium JAR cells in RPMI 1640 containing 1% sodium pyruvate, and HEK293 cells in minimum essential medium; all containing 10% FBS and 1% penicillin and streptomycin. (Unless otherwise noted, all materials were purchased from Invitrogen, Burlington, ON, Canada).
Primary cultures of uterine natural killer cells (uNKs) were isolated from first-trimester placental samples as described previously (41). Placentas were obtained from women undergoing elective abortions of healthy pregnancies. Briefly, decidual tissue was separated from placental tissue and finely minced. Minced tissue was suspended in PBS and passed sequentially through nylon mesh filters of decreasing diameter (220 μm, 70 μm, 40 μm) to ensure single cell suspension. These cells were then sorted using a MACS midi kit (Miltenyi Biotec) using anti-CD56 magnetic microbeads to obtain a >97% pure uNK population. Placental tissue samples were acquired with informed consent in accordance with a protocol approved by the University of Western Ontario's Office of Research Ethics. Primary uNKs were cultured in RPMI 1640 containing 1% sodium pyruvate, 0.2% bovine serum albumin (BSA), and 1% human serum.
HTR-8/SVneo-GnRH-RI cells were treated with 100 nM buserelin (Synthetic GnRH-I analogue) (Sigma-Aldrich, Oakville, ON, Canada), 100 nM GnRH-II (Phoenix Pharmaceuticals, Burlingame, CA), 100 nM antide (GnRH-I competitive antagonist; Sigma-Aldrich) for 0 to 24 h in media containing 10% FBS. The concentrations used were based on that reported for similar studies exploring the effect of GnRH on gene expression in the human placenta and uterus (5–8, 33, 34). Additionally, the use of 100 nM of the GnRH-I analogue buserelin falls within the normal physiological range of GnRH produced by placental explants (5–200 nM) in culture (24).Vehicle treatments (PBS) were conducted side by side with hormone treatments to serve as controls.
Western blot analysis of ERK1/2 phosphorylation.
HTR-8/SVneo, HTR-8/SVneo-GnRH-RI, and primary EVTs were serum starved overnight before extracellular signal-regulated kinase (ERK) activation experiments. Buserelin and GnRH-II (both 100 nM) were administered in serum-free RPMI 1640 for 0, 5, 10, 15, 30, and 60 min in 35-mm culture dishes. Cells were then lysed in a lysis buffer (25 mM HEPES, pH 7.5, 300 mM NaCl, 1.5 mM MgCl2, 0.2 mM EDTA, 0.1% Triton X-100) containing protease inhibitors (AEBSF, leupeptin, aprotinin) and phosphatase inhibitors (sodium fluoride, sodium orthovanadate). Fifty micrograms total protein per sample were then separated on a 12.5% polyacrylamide gel, transferred onto 0.4-μm nitrocellulose membrane, and Western blot analyzed for phosphorylated ERK1/2 using a phosphospecific rabbit monoclonal antibody (1:2,000) (Phospho p44/42 MAPK, Thr202/Tyr204) from Cell Signaling Technology (Danvers, MA) and total ERK1/2 using a rabbit monoclonal antibody (1: 2,000) (p44/42 MAPK) from Cell Signalling. The anti-rabbit horseradish peroxidase-conjugated secondary antibody was used at a dilution of 1:2,500 and developed with ECL and ECL Plus Western Blotting Detection Reagents (Amersham, Piscataway, NJ).
Total RNA was prepared from HTR-8/SVneo and HTR-8/SVneo-GnRH-RI cells using an RNeasy Mini Kit (Qiagen, Missassauga, ON, Canada) according to the manufacturer's instructions. Total RNA was subjected to an on column DNase digest to eliminate any possible genomic DNA contamination using the recommended on column RNase-free DNase kit (Qiagen). Total RNA concentration was determined by optical densitometry (260/280 nm) on an N-1000 spectrophotometer (NanoDrop, Wilmington, DE).
Total RNA was isolated from HTR-8/SVneo cells following a 12-h treatment with buserelin, GnRH-II, and antide, and vehicle was subjected to microarray analysis at the London Regional Genomics Centre (London, ON, Canada). RNA was biotinylated and hybridized to U133A 2.0 gene chips (Affymetrics, Santa Clara, CA). Expression profiles were analyzed using Gene Spring software (Agilent Technologies, Missassauga, ON, Canada).
Primers were designed to amplify 150–425 base pair segments of each gene. All primers were synthesized by Sigma Genosys Custom Oligonucleotides (Oakville, ON, Canada). CXCL2 (expected product size: 161 bp): F-5′GCAGGGAATTCACCTCAAGA3′, R-5′TTTTCAGCATCTTTTCGATGA3′; CXCL3 (expected product size: 150 bp): F-5′GCAGGGAATTCACCTCAAGA3′, R-5′TTTTCGATGATTTTCTGAACCA3′; CXCL6 (expected product size: 158 bp): F-5′GTCCTGTCTCTGCTGTGCTG3′, R-5′AACTTGCTTCCCGTTCTTCA3′; CXCL8 (expected product size: 196 bp): F-5′GTGCAGTTTTGCCAAGGAGT3′, R-5′CTCTGCACCCAGTTTTCCTT3′; β-actin (expected product size: 275 bp): F-5′GGGAAATCGTGCGTGACATTAAG3′, R-5′TGTGTTGGCGTACAGGTCTTTG3′; GnRH-I (expected product size 380 bp): F-5′ATTCTACTGACTTGGTGCGTG3′, R-5′GGAATATGTGCAACTTGGTGT3′; GnRH-II (expected product size: 413 bp): F-5′CTGAAGGAGCCATCTCATCCA3′, R-5′CGGAGAACCTCACACTTTATT3′.
cDNA generation and quantitative real-time PCR analysis.
Total RNA was reverse transcribed into cDNA with Superscript RT-II (Invitrogen, Burlington, ON, Canada) using 1 μg of RNA. Reactions were performed according to manufacturer's recommendations using random hexamer primers (Amersham, Piscataway, NJ). IQ SYBR green supermix (Bio-Rad, Missassauga, ON, Canada) was used to perform quantitative real-time PCR (Q-PCR) analysis. All sample reactions were performed as five independent trials with accompanying serial dilution standard curves and water blanks for each gene and treatment time. Each gene was normalized to β-actin as an internal control and was presented relative to vehicle-treated samples for that time point using the Pfaffl method (31). Reaction efficiencies used in these calculations were derived from the standard curves.
Human placental sections between 12 and 14 wk gestation were obtained with informed consent in accordance with a protocol approved by the University of Western Ontario's Office of Research Ethics. Sections containing chorionic villi attached to the decidua were processed for indirect double immunostaining. Specimens were fixed in 4% formaldehyde in PBS, pH 7.4, for 12–24 h. After ethanol dehydration, tissues were embedded in paraffin and sectioned in 7-μm sections. Localization of CXCL8 was determined by dual staining using fluorescent secondary antibodies. Goat anti-human CXCL8 antibody (R&D systems, Minneapolis, MN) was used at 1:250 dilution, and mouse anti-human CK7 (DAKO, Missassauga, ON, Canada) antibody was used at 1:100 dilution. Alexa-488 anti-goat and Alexa-546 anti-mouse secondary antibodies were used at 1:200 dilution (Molecular Probes, Invitrogen, Burlington, ON, Canada).
HTR-8/SVneo GnRH-RI cells were plated at a confluency of 300,000 cells per 35-mm dish. Twenty-four hours later, cells were treated with buserelin, GnRH-II, antide, or vehicle in 3 ml of media for 12 h. The conditioned medium was then collected and centrifuged at 5,000 g for 10 min to remove any cellular debris and then stored at −20°C until used. An ELISA kit specific to CXCL8 (D8000C) (R&D systems, Minneapolis, MN) was used to quantify the secreted CXCL8 according to the manufacturer's recommended protocols.
Lymphocyte chemotaxis assays.
To assess the ability of primary EVTs and various cell lines to chemoattract lymphocytes, a transwell migration system was developed. Jurkat T cells and primary T and uNK cells were used to determine the effect of different cell types on lymphocyte migration. HTR-8/SVneo cells, or other adherent cell types, were seeded into 24-well culture dishes at a density of 100,000 cells per well, and cells were cultured for 24 h before the chemotaxis assay. Jurkat T cells were labeled with PKH67 dye (Sigma-Aldrich, Oakville, ON, Canada), a green fluorescent general membrane dye, according to the manufacturer's protocol, and then washed twice in serum-free media and resuspended in serum-free media containing BSA (20 μg/ml) to a final concentration of 1,000,000 cells/ml. This labeling technique did not affect cell viability (data not shown). The 24-h cultured HTR-8/SVneo cells were washed twice with serum-free media and covered with 500 μl of serum-free media containing BSA. Transwell chambers (6.5 mm diameter, 8 μm pore; Corning Life Sciences, Lowell, MA) were added to each well, and 200 μl (200,000 cells) of PKH67 labeled Jurkat T cells were added to the upper chamber. Cells were incubated for 12 h before the transwell chambers were removed and discarded. Lower chamber media containing Jurkat T cells was collected following which cells in the lower chamber (HTR-8/SVneo and associated Jurkat T cells) were trypsinized for 5 min at room temperature, collected, and pooled with the media. This cell suspension was centrifuged for 10 min at 400 g. Medium was aspirated and cells were resuspended in PBS. After another round of centrifugation and aspiration, the final cell pellet was resuspended in 500 μl of PBS containing 1% formaldehyde. The cell suspension was transferred to appropriate containers for flow cytometry and green fluorescently labeled Jukat T cells were quantified by flow cytometry.
Antide treatment was administered at 100 nM (5–8) when HTR-8/SVneo cells were plated. Antide was again added to serum-free media when conditions were switched to serum-free conditions. Transwell chambers and Jurkat T cells were used exactly as described above.
Repertaxin was used at 1 μM to inhibit CXC receptors (CXCR) 1 and 2 on the surface of Jurkat T cells. The concentration used was based on a study by Cassili et al. (2). Jurkat T cells were incubated with repertaxin or vehicle at room temperature for 20 min in serum-free media (containing BSA) following PKH67 labeling and before addition to the transwell chambers. Samples were subsequently processed as described above.
T cell purification.
Peripheral blood mononuclear cells (PBMCs) were isolated from anticoagulant citrate dextrose-anticoagulated blood of 12 healthy blood donors using standard procedures. CD4+ T cells were isolated from PBMCs using the magnetic-based CD4 MultiSort Kit (Miltenyi Biotec). CD4+CD8− T cells were then isolated from the CD4+ population with CD8 microbeads (Miltenyi Biotec).
uNK chemotaxis assay.
Because of the incompatibility of the primary uNK cells and PKH67 membrane labeling system, PKH67 was not used to assess uNK migration. Alternatively, uNK cells were labeled using a green fluorescent nuclei stain YOYO1 (Invitrogen). This labeling technique did not affect cell viability (data not shown). After staining was completed, cells were washed twice in growth medium and adjusted to a concentration of 500,000 cells/ml. To avoid possible allogenic reactions, different placental samples were not pooled. As a result of this, the number of lymphocytes in each well was limited to 100,000. Labeled uNKs were also cultured in growth medium until the end of the experiment at which time they were counted and used to generate standard curves of cell number against fluorescence intensity. The transwell migration part of these experiments was identical to that described for the Jurkat T cell migration assay. After trypsinization and collection of the contents of the lower chamber, the cells were resuspended in 100 μl PBS containing 1% formaldehyde and loaded into 96-well fluorescence microplates. The accompanying sample matched serial dilution standard curves were measured at the same time. Plates were read on an Infinite M200 plate reader (Tecan US) set to excite at 480 nm and measure emission at 525 nm. Standard curves were then used to determine the number of cells per well.
HTR-8/SVneo, HTR-8/SVneo GnRH-RI stables, and primary EVTs are responsive to GnRH-I and GnRH-II.
To validate the use of the HTR-8/SVneo and HTR-8/SVneo-GnRH-RI as models for investigating GnRH signaling in the invasive trophoblast, it was necessary to demonstrate an active GnRH signaling pathway in these cells and compare it with what is observed in primary cultures of EVTs. A functional GnRH-RI signaling pathway was assayed for by looking at changes in ERK activation following GnRH treatment and Western blot analysis. Cells were serum starved overnight to reduce basal ERK activation and then treated with 100 nM buserelin (GnRH-I analogue) or 100 nM GnRH-II for 0, 5, 10, 15, 30, or 60 min. ERK1/2 analysis revealed that both untransfected and transfected HTR-8/SVneo cell lines showed increases in the phosphorylated forms of ERK1 (44 kDa) and ERK2 (42 kDa) in response to both buserelin and GnRH-II stimulation. This increase in activated ERK in both cell lines was observed most frequently between 5 and 15 min of GnRH stimulation, and this was very similar to that observed in the primary EVTs (Fig. 1A). Untreated cells at 0 min showed a small amount of ERK activation that could have resulted from the basal activity of a number of signaling systems converging on mitogen-activate protein kinase (MAPK). One such signaling system would also include the GnRH-RI system since we showed that HTR-8/SVneo cells express mRNAs to both GnRH-I and GnRH-II, which presumably results in endogenous production of the ligands that can act back on the endogenously expressed receptor (Fig. 1B) Data shown are representative Western blots from four independent experiments.
Microarray analysis of GnRH-regulated gene expression.
To identify genes that are regulated by GnRH in the human placenta, HTR-8/SVneo-GnRH-RI cells were treated for 12 h with 100 nM buserelin, GnRH-II, or antide before RNA isolation. Gene expression profiles were determined by comparison of GnRH treatments to a vehicle-treated (PBS) control. Candidate genes were identified by selecting genes that showed an increased expression of at least 1.5-fold over vehicle following either buserelin or GnRH-II treatment, yet remained unchanged or downregulated following antide treatment. Several candidate genes were identified and grouped according to function. One functional group consisted of genes with established proangiogenic function and potential roles in remodeling of the decidua and maternal spiral arteries. Within this group, we further identified a subclass of genes representing most of the CXC motif ligand (CXCL) chemotactic cytokines (or chemokines) (Table 1) and these (CXCL2, CXCL3, CXCL6, and CXCL8) were selected for further study. Upon identifying the angiogenic chemokines as putative GnRH-regulated genes, we next examined the microarray data to determine whether angiostatic CXC chemokines genes (CXCL4, CXCL9, and CXCL10) showed a change in expression, but none was seen and this was subsequently confirmed by Q-PCR (data not shown).
Time-course analysis of GnRH-regulated chemokine gene expression.
To confirm and expand on the results of the microarray analysis, an extensive time-course analysis of GnRH-regulated chemokine gene expression was conducted by real-time Q-PCR. Time-course analysis of GnRH-regulated gene expression was determined following 1, 3, 6, 12, and 24 h treatment of HTR-8/SVneo-GnRH-RI cells with buserelin, GnRH-II, antide (all at 100 nM), or vehicle. Changes in gene expression of CXCL2, CXCL3, CXCL6, and CXCL8 were determined by SYBR green Q-PCR. Changes in expression were determined by normalization of each sample to β-actin and expressed relative to vehicle-treated cells for each time point. CXCL2 expression was significantly greater than antide following 1, 3, 6, and 24 h of buserelin treatment (2.2, 2.3, 2.1, and 2.0-fold increases in expression relative to vehicle treatments) (P < 0.05) (Fig. 2, A–C, E). GnRH-II treatment was significantly greater at 3- and 24-h treatment times (P < 0.05) (Fig. 2, B and E). CXCL2 expression following antide treatment was very similar to vehicle treatment and not significantly different at any time point (Fig. 2, A–E). However, at 12 h, antide treatment showed a slight increase in CXCL2 mRNA expression, but the difference remained insignificant (Fig. 2D). CXCL3 also showed increased gene expression in response to GnRH agonist treatments. Buserelin treatment yielded gene expression levels that were significantly elevated above antagonist treatment at 1, 3, 6, and 12 h (P < 0.05) (Fig. 2, F–I) with the peak expression occurring at 3 h posttreatment (3.6-fold greater than vehicle) (Fig. 2G). GnRH-II behaved similarly showing peak expression at 3 h of treatment (4.3-fold increase) (Fig. 2G) with significant increases existing at 1, 3, 6, and 24 h treatment (P < 0.05) (Fig. 2, F–H, J). Once again, CXCL3 expression following antide treatment closely resembled vehicle with little change in expression over 24 h (Fig. 2, F–J).
CXCL6 expression was significantly greater than antagonist treatment at 3 and 12 h treatment (P < 0.05) (Fig. 2, L and N). GnRH-II treatment showed an early response (Fig. 2K), which was absent at 3 h (Fig. 2L) and was strongest again after 6 h (8.0-fold increase) (Fig. 2M). GnRH-II induced CXCL6 expression was significantly greater than antide at 1, 6, 12, and 24 h (P < 0.05) (Fig. 3, K, M–O). Once again, CXCL6 expression in response to antide treatment closely resembled vehicle treatment with little change in expression over 24 h (Fig. 2, K–O).
Buserelin treatments of 6, 12, and 24 h significantly increased the expression of CXCL8 over antagonist treatment (P < 0.05) (Fig. 2, R–T). This increase was greatest at 12 h, which showed a greater than fivefold increase over vehicle-treated cells (Fig. 2S). GnRH-II behaved somewhat differently showing significant differences at the earlier time points of 1 and 3 h posttreatment (P < 0.05) (Fig. 2, P and Q). The greatest difference was observed after only 1 h of treatment with GnRH-II (3.4-fold increase) (Fig. 2P). Relative mRNA expression levels are the mean of five independent drug treatments and Q-PCR experiments. For each of five independent experiments, the pattern of response to all treatments was similar. Data are expressed as means ± SE and analyzed by one-way ANOVA (Dunnet's posttest comparing each treatment to antagonist).
Immunofluorescence analysis of CXCL8 protein expression in the human placenta.
To determine whether expression of the chemokines was detectable at the protein level in the first-trimester human placenta, CXCL8 was selected as a representative CXC chemokine to conduct an immunofluorescence study. An analysis of the maternal-fetal interface clearly identified numerous villi with at least one that was anchored to the decidua (confirmed by examining a series of adjacent tissue sections to the section shown in Fig. 3). Immunostaining using an anti-cytokeratin 7 (CK7) antibody clearly identified cytotrophoblasts in the cell column and rest of the villi (Fig. 3A, red). Immunostaining of the same tissue section with anti-CXCL8 antibody revealed that CK7-positive cells all highly expressed CXCL8 (Fig. 3B, green). CK7 and CXCL8 show strong colocalization indicating that trophoblast cells produce CXCL8 at the maternal-fetal interface (Fig. 3D, yellow). Nuclei of cells were stained with Hoechst and appear blue (Fig. 3C).
Quantification of CXCL8 in HTR-8/SVneo-conditioned media following GnRH treatment.
To determine whether the changes in chemokine gene expression observed translated into differences in chemokine protein levels in the conditioned media of HTR-8/SVneo-GnRH-RI-cells, CXCL8 was selected as a representative CXC cytokine due to its strong gene expression and the availability of quality antibodies for protein analysis. The concentration of immunoreactive CXCL8 in conditioned media following 12 h of GnRH treatment was determined by ELISA. The data revealed that both buserelin and GnRH-II significantly increased (P < 0.05, P < 0.001) CXCL8 concentrations to 157.0 and 193.3 pg/ml, respectively, compared with the conditioned media of vehicle-treated cells, which accumulated CXCL8 at a concentration of 117.3 pg/ml. Treatment with the receptor antagonist antide resulted in a significant reduction (P < 0.05) in CXCL8 concentration to 82.6 pg/ml compared with vehicle and agonist-treated cells (Fig. 4). CXCL8 protein levels are expressed as means ± SE of three independent experiments. Significance determined by one way ANOVA with Bonferroni's posttest.
Effect of placental and nonplacental cells on lymphocyte chemotaxis.
The ability of several placental cell lines, as well as primary EVTs, to stimulate the chemoattraction of lymphocytes was assayed by a transwell chemotaxis assay using both Jurkat T cells and freshly isolated T cells (CD4+CD8−) from peripheral blood. Four different placental trophoblast model cell lines, one nonplacental cell line (HEK293), and an empty chamber control were investigated for their ability to stimulate the migration of lymphocytes through 8-μm pores to the lower transwell chamber after 12 h of coincubation. In this study, since we were comparing unmodified cell lines to one another, we used the nontransfected HTR-8/SVneo cells. Overall, we observed that the pattern of recruitment was very similar for both Jurkat cells and primary T cells isolated from peripheral blood (Fig. 5, A and B). HTR-8/SVneo cells stimulated the highest level of lymphocyte migration (10,715 ± 1,897 Jurkat T cells vs. 13,915 ± 633 primary T cells) among all cell lines, and this was very similar to that observed with primary EVTs (12,640 ± 914 Jurkat T cells vs. 16,336 ± 1,198 primary T cells). The number of Jurkat and primary T cells recruited by HTR-8/SVneo cells and primary EVTs was significantly greater (P < 0.01) than all the other placental cell lines, the nonplacental HEK293 cells (716 ± 185 Jurkat cells vs. 502 ± 143 primary T cells), and empty well controls (254 ± 151 Jurkat cells vs. 121 ± 21 primary T cells). BeWo cells had significantly greater (P < 0.05) lymphocyte recruitment (4,862 ± 405 cells) than those of the empty well control. There were no significant differences among BeWo, JEG3 (2,081 ± 222 Jurkat cells vs. 3,477 ± 377 primary T cells), and JAR (2,019 ± 309 Jurkat cells vs. 2,839 ± 671 primary T cells). Similarly, there were no significant differences between JEG3, JAR, and HEK293 in their capacity to attract lymphocytes nor were they significantly different from empty well controls (Fig. 5, A and B). Next, we looked at the effect of HTR-8/SVneo cells to recruit primary uNK cells. The data show that compared with the empty well control (1,152 ± 234), HTR-8/SVneo and primary EVTs recruited a significantly (P < 0.01) larger number (4,081 ± 588 for HTR-8/SVneo and 5,093 ± 695 for EVTs) of uNKs. Lymphocyte migration assays are expressed as the mean number of cells ± SE of four independent experiments performed in triplicate.
Because of the low number of uNK cell numbers isolated from decidual tissue coupled to the aim to minimize possible allogenic reactions that could result from pooling multiple donor uNKs, only HTR-8/SVneo cells were used for this and subsequent experiments investigating the role of GnRH as a regulator of lymphocyte recruitment.
Effect of GnRH-RI signaling on lymphocyte chemotaxis.
HTR-8/SVneo-GnRH-RI cells or HTR-8/SVneo cells expressing an empty vector (control) were investigated for their ability to stimulate the migration of Jurkat T-cells through 8-μm pores to the lower transwell chamber after 12 h of coincubation. Results show that HTR-8/SVneo-GnRH-RI cells (8,316 ± 1,207 cells) relative to vector-expressing vehicle-treated cells (4,784 ± 792 cells) exhibited a greater trend of lymphocyte recruitment. To determine whether this effect was specifically mediated by GnRH/GnRH-RI signaling, the receptor antagonist antide was used to block GnRH-RI activation. HTR-8/SVneo cells were pretreated with antide for 24 h before lymphocyte coincubation to block the effect of any endogenously produced GnRH-I or II. Antide treatment significantly reduced (aP < 0.01) lymphocyte migration in GnRH-RI overexpressing cells (2,436 ± 602 cells) relative to vehicle-treated cells. To determine whether the lymphocyte migration is facilitated by changes in the levels of the chemokines under investigation repertaxin (1 μM) was used to inhibit their cognate receptors (CXCR1 and CXCR2) on the lymphocytes (2). A 20-min treatment of Jurkat T cells with 1 μM repertaxin before migration showed a trend of reduced lymphocyte migration, but this difference was not significantly different from vehicle treatment (Fig. 6A).
Primary uNK cells were also investigated for GnRH-RI-regulated lymphocyte chemotaxis. Like Jurkat T cells, uNK cells showed a trend of greater lymphocyte recruitment in cells overexpressing GnRH-RI relative to the matched vector controls (3,953 ± 749 cells and 2,882 ± 313 cells, respectively). Again, like Jurkat T cells, when uNK cells were treated with antide, the average number of migrated uNKs was lower, but this difference was only significantly different from vehicle treatment in GnRH-RI overexpressing cells (2,028 ± 460 cells; bP < 0.05). Vector-transfected HTR-8/SVneo also had a reduced mean number of migrated uNKs after repertaxin treatment, but the difference was not significant (1,764 ± 145 cells). However, treatment of uNK cells with repertaxin resulted in a significant reduction in uNK cell migration when HTR-8/SVneo cells were overexpressing GnRH-RI (1,660 ± 448 cells; bP < 0.05) (Fig. 6B).
Lymphocyte migration assays are expressed as the means ± SE of at least three individual experiments each performed in triplicate.
In an attempt to further define the function of GnRH signaling in human placentation, we performed a microarray analysis to globally screen for the GnRH-regulated genes in a human placental trophoblast cell line HTR-8/SVneo that stably overexpressed the GnRH-RI coding sequences. To justify the use of HTR-8/SVneo cell system, we first demonstrated that these cells express a functional GnRH-RI signaling pathway, which behaves similarly to primary EVTs. This was done by showing responsiveness to buserelin, a synthetic analogue of GnRH-I, and GnRH-II, determined by Western blot analysis of ERK phosphorylation levels. We also demonstrated that HTR-8/SVneo cells expressed mRNAs for both GnRH-I and GnRH-II and that the GnRH agonist concentration used (100 nM), which was within the normal physiological range produced by placental explants (5–200 nM) in culture, was sufficient to activate GnRH-RI signaling in this cell type (24).
The microarray screen compared gene expression profiles in cells following a 12-h treatment with vehicle, buserelin, GnRH-II, and antide. This analysis revealed that 37 genes, following agonist or antagonist treatment, showed 1.5-fold change in expression (increase or decrease) relative to the vehicle-treated control. These genes were subsequently organized into functional groups, one of which comprised several members of the CXC motif ligand chemokine family. CXC chemokines are a family of small molecules that regulate cell chemotaxis through their interactions with the G protein-coupled CXC chemokine receptors (CXCR) (39). This group of molecules is further subdivided by their receptor interactions. An additional ELR motif adjacent to the CXC motif determines receptor specificity for the majority of these ligands. CXC ligands possessing this additional conserved motif interact with CXCR1 and/or CXCR2, whereas ligands lacking the sequence interact with other receptors, most frequently CXCR3. Although all CXC signaling regulates chemotaxis, CXCR1/2 interacting ligands are potent angiogenic factors, whereas CXCR3 ligands are angiostatic molecules (40). The CXC ligands we have determined to be upregulated by GnRH treatment all contain the proangiogenic ELR motif. Because of the proangiogenic nature of the molecules encoded by these genes and the arterial remodeling that occurs during the period of maximal GnRH signaling in the human placenta, CXCL2, CXCL3, CXCL6, and CXCL8 were chosen for further analysis.
To verify the microarray data, Q-PCR analyses of GnRH-induced changes in CXCL gene expressions were conducted and these indeed confirmed the microarray data, showing increased gene expression of CXCL2, CXCL3, CXCL6, and CXCL8. In addition, the Q-PCR time-course expression analysis revealed that most of these chemokine genes were upregulated as early as 1 and 3 h poststimulation (CXCL2, CXCL3, and CXCL6) in response to either GnRH-I (buserelin) or GnRH-II. It must be noted though that buserelin treatment did not induce a significant change in CXCL8 expression until 6 h after treatment, suggesting the possibility that unlike CXCL8, the expression of CXCL2, CXCL3, and CXCL6 might be directly regulated by GnRH-RI upon buserelin stimulation, although secondary effects cannot be conclusively ruled out. Expression remained elevated to varying degrees for each gene until 12 h poststimulation. After 24 h of treatment, the differences observed at earlier time points had become less pronounced for all genes. This result is comparable to previous studies looking at the changes in gene expression of MMP2, MMP9, and uPA levels in trophoblasts in response to GnRH. These studies showed reduced changes in gene expression later than 24 h (7, 8). Antagonist treatment of cells generally did not elicit a change in gene expression. Infrequently, however, there was a slight increase in chemokine mRNA expression in response to antide treatment. This result is a possible effect of the endogenous production of GnRH-I and GnRH-II by the cell line. This phenomenon was only observed after 12 h of treatment, which may be the result of accumulation of endogenously produced GnRH ligands competing with the antagonist for receptor binding.
Next, we chose CXCL8 as a representative CXC ligand for protein expression analysis due to the availability of high-quality antibodies and its strong level of gene expression. Immunofluorescence analysis demonstrated that CXCL8 is abundant at the maternal-fetal interface. Dual fluorescence labeling of CXCL8 and CK7, a protein specific to trophoblast cells, showed a strong colocalization (30). This indicates that trophoblasts are producing CXCL8 in vivo, thus confirming the in vitro data obtained with the HTR-8/SVneo cell line. Further support that trophoblasts produce CXCL8 is derived from ELISA-based analysis of CXCL8 levels in trophoblast-conditioned media. We have determined that cultured HTR-8/ SVneo-GnRH-RI cells secrete readily detectable quantities of CXCL8 that accumulate in culture media after 12 h of incubation. This accumulation of CXCL8 was significantly increased following buserelin or GnRH-II treatment. Furthermore, the accumulation of CXCL8 was significantly decreased following 12 h of treatment with the GnRH antagonist antide. The increased transcription and accumulation of CXCL8 in the conditioned media of cells treated with GnRH agonists coupled to the lack of its detection in lysates (data not shown) indicates that these cells do not store this protein but instead secrete it as it is produced. To date, the only record of CXCL8 production by trophoblasts showed that STB cells increase CXCL8 gene expression and protein release in response to contact with malaria-infected red blood cells. This study did find similar levels of CXCL8 accumulation in STB cells as we did for HTR-8/SVneo-GnRH-RI cells, 175 and 117 pg/ml, respectively (28). Our study, however, is the first to demonstrate modulation of trophoblast CXCL8 production in nonpathological scenarios, as well as by GnRH-RI signaling.
We have shown that GnRH regulates the expression of four CXC chemokines in the HTR-8/SVneo model EVT cell line. The angiogenic function of these chemokines implies they may be players in remodeling of the maternal spiral arteries at the end of the first trimester of pregnancy. This is supported by the strong expression of CXCL8 observed in trophoblasts, especially those within the cell column of anchoring chorionic villi in the first trimester. In culture, treatment of trophoblasts with GnRH agonists and antagonist show a respective increase and decrease in CXCL8 accumulation. Because of the chemotactic nature of these proteins and the abundance of lymphocytes in the pregnant uterus it was necessary to investigate the possibility of an interaction between GnRH signaling and lymphocyte traffic in the decidua.
The relationship between GnRH-RI signaling and lymphocyte migration was examined in depth by the development of a transwell chemotaxis assay utilizing fluorescently labeled lymphocytes, multiple placental cell lines, primary EVTs, GnRH-RI receptor overexpression, and antagonist drug treatments.
Lymphocyte chemotaxis assays showed that all of the cell lines investigated released chemoattractants, which induced the directional migration of both Jurkat T cells and primary peripheral blood T (CD4+CD8−) and uNK cells, and that lymphocytes had the greatest migration response to HTR-8/SVneo cells and primary EVTs. This assay also demonstrated that with respect to recruiting lymphocytes, the HTR-8/SVneo cell line produced a response almost identical to that seen with primary cultures of EVTs suggesting that for these purposes, with the appropriate controls, the HTR-8/SVneo cell line is an appropriate model cell line for studying trophoblast-lymphocyte interactions. Additionally, we found that the number of Jurkat T cells and freshly isolated CD4+CD8− T cells (peripheral blood) that migrated in response to the trophoblast cell lines was also very similar. This too suggests that for these puposes, with the appropriate controls, the Jurkat T cell line is an appropriate model cell line for studying lymphocyte-trophoblast interactions.
Investigation of the role that GnRH signaling plays in this lymphocyte recruitment phenomenon using the HTR-8/SVneo cell lines showed that GnRH plays an important role in regulating this process. Antide treatment of both vector transfected and GnRH-RI overexpressing HTR-8/SVneo cells reduced the number of migrated lymphocytes to nearly identical levels. Although the difference was only statistically significant for cells expressing elevated levels of GnRH-RI, it is important to note that in the presence of antide, when receptor signaling is blocked, lymphocyte migration levels are identical, regardless of receptor expression levels. The ability of this GnRH antagonist to reduce recruitment of lymphocytes places it as an integral regulator of this trophoblast function.
This result is in agreement with our earlier observation that CXCL8 levels in culture media, and presumably the levels of other chemokines as well, are significantly altered by GnRH or GnRH antagonist treatment. The migration assays, however, are not sufficient to determine whether this GnRH-dependent effect is specifically due to changes in levels of the CXCLs under investigation. To address this, we employed the use of repertaxin, an allosteric noncompetitive inhibitor of CXCR1 and CXCR2, the cognate receptors of the CXCL2, CXCL3, CXCL6, and CXCL8. Treatment of uNK cells with 1 μM repertaxin significantly reduced the migration of these cells in response to HTR-8/SVneo-GnRH-RI cells relative to vehicle treatments. Once again, as with antide treatment, there was a trend of reduced lymphocyte migration for all cell types and GnRH-RI expression levels. This result indicates lymphocyte migration in response to factors produced by HTR-8/SVneo cells is a result of ligands interacting with CXCR1/2, which CXCL2, CXCL3, CXCL6, and CXCL8 are examples. The discrepancy in lymphocyte migratory response in the presence of repertaxin between Jurkat T cells and uNKs remains unknown but is likely attributed to these cells being two completely different lymphocyte lineages each of which responds differently to the chemokine set being produced by the trophoblasts.
Trophoblast cells inducing lymphocyte chemotaxis have been described relatively recently in the literature (18, 43). Studies utilizing primary placental trophoblasts and primary lymphocytes have documented similar findings to the ones that we describe here, being that trophoblasts recruit lymphocytes. This study (18) specifically found that this recruitment or migration was in part dependent on CXCL16 signals emanating from trophoblasts. CXCL16 does not signal through CXCR1 or CXCR2 and interestingly, in our study when these two receptors were blocked, the migration of lymphocytes was never reduced to zero. CXCL16 therefore most likely accounts for a portion of the migration that remained after GnRH antagonist treatment or repertaxin treatment, which blocks CXCR1 and CXCR2. It has also been shown that uNK cells are recruited by trophoblasts via the production of CXCL12 and its interaction with CXCR4 on uNKs. It is proposed to be a mechanism that underlies the establishment of the large uNK populations of the decidua by recruitment of CD56brightCD16− NK cells (uNK) from the peripheral blood (18, 43). This same study clearly demonstrated that decidual lymphocytes express several chemokine receptors (most importantly to our work, CXCR1 and CXCR2) and that trophoblasts do not produce measurable levels of CXCR3 (angiostatic) interacting chemokines (43). With this strong supporting evidence coupled to our results, we propose that GnRH signaling acts as a positive regulator of lymphocyte migration into the decidual tissue early in pregnancy.
Absent or low numbers of EVTs and decidual leukocytes, specifically uNK cells, have been associated with incomplete spiral artery remodeling (36). It has been previously shown that there exists extensive crosstalk between placental trophoblasts and decidual lymphocyte, which results in the production of angiogenic factors in close proximity to the maternal spiral arteries (16, 26). We have demonstrated that GnRH regulates the expression of several chemokines in placental trophoblasts. GnRH agonist and antagonist treatment, as well as GnRH-RI overexpression studies in invasive trophoblasts, demonstrate that chemokine expression is modulated by this signaling cascade, which in turn regulates the interaction with nearby lymphocyte populations. Taken together, it seems that in the late first-trimester placenta, which exhibits maximal expression of GnRH-RI and GnRH-I, GnRH is an important regulator of placentation. More specifically, GnRH seems to be a regulator of maternal spiral artery remodeling, through the production of proangiogenic chemokines and the recruitment of decidual lymphocytes to the maternal spiral arteries.
The following are recipients of salary awards: A. V. Babwah was given the CIHR/CIHR's Institute of Gender and Health/Ontario Women's Health Council New Investigator Award and M. Bhattacharya was given the NSERC University Faculty Award.
Research reported in this study was supported by a grant from the Canadian Institutes of Health Research (CIHR) MOP-81383 and the Academic Development Fund, University of Western Ontario (SG05-53).
The authors thank Aleah Hazan for technical assistance and Dr. Peeyush Lala for the gracious donation of the HTR-8/SVneo cell line. We thank the clinical staff in the Department of Obstetrics and Gynecology, University of Western Ontario, for kind assistance in providing access to human placentas.
- Copyright © 2009 the American Physiological Society