Differential COX localization and PG release in Thy-1+ and Thy-1 human female reproductive tract fibroblasts

Laura Koumas, Richard P. Phipps

Abstract

A key role exists for prostaglandins (PGs) in reproductive health, including fertility and parturition. However, the cellular sources and regulation of PG production by cyclooxygenase (COX) in the human female reproductive tract remain poorly understood. We recently reported that human female reproductive tract fibroblasts are divisible into distinct subsets based on their Thy-1 surface expression. Herein, we demonstrate that the expression, induction, and subcellular localization of COX-1 and COX-2 and the downstream PG biosynthesis are markedly different between these subsets. Specifically, Thy-1+ fibroblasts highly express COX-1, which is responsible for high-level PGE2 production, a feature usually attributed to the COX-2 isoenzyme. In contrast, COX-2, generally considered an inducible isoform, is constitutively expressed in the Thy-1 subset, which only minimally produces PGE2. The intracellular signaling pathways for COX regulation also differ between the subsets. Determination of differences in signal transduction, COX expression and localization, and PG production by human reproductive fibroblast subtypes supports the concept of fibroblast heterogeneity and the possibility that these subsets may play unique roles in tissue homeostasis and in inflammation.

  • inflammation
  • myometrium
  • lipid mediators
  • heterogeneity
  • prostaglandin
  • cyclooxygenase

critical events in the human female reproductive tract such as ovulation, implantation, and parturition are characterized by inflammation (9). Close interaction among the different cell layers that form the uterus, namely, the endometrium and myometrium, is required for the successful outcome of these events and is crucial for human existence. Besides cytokines and sex steroid hormones, prostaglandins (PGs) play a central role in the normal functioning of the reproductive tract in processes such as ovulation (7), menstruation (9, 18), implantation (24,) and parturition (25).

PGs are arachidonic acid metabolites, crucial for normal physiology and for inciting and regulating inflammatory responses. Prostanoid biosynthesis is catalyzed by two cyclooxygenases (COXs), COX-1 and COX-2. COX-1 is considered to be a constitutively expressed enzyme that maintains cellular homeostasis, whereas COX-2 is typically induced after stimulation with certain bacterial products and proinflammatory cytokines such as interleukin (IL)-1 and plays a major role in inflammation (40). The various types of PGs are ultimately synthesized by newly characterized synthases that may be differentially expressed among cells and tissues (16).

The study of COXs and PGs comprises a vast area of research in the biology of reproduction. PGs are essential for parturition in mice (12) and are crucial for initiating labor in humans (28, 44). A key PG, called PGE2, is crucial for inflammation, for reproductive tract functions, and for fetal organ maturation (33). COX expression is vital for “cervical ripening,” a process of neutrophil tissue infiltration essential for normal labor (17). COX-1 has also been shown to be critical for normal labor in the mouse, because COX-1-deficient mice demonstrated delayed parturition, resulting in neonatal death (12). COX-1 and COX-2 mRNA is expressed in cow myometrium (11), and COX-1 and COX-2 protein exists at high levels in the human myometrium (41). Increased levels of COX-2 in the myometrium are associated with the onset of labor in women (8).

The concept of fibroblasts as “sentinel cells” has provided a new role for them as initiators of inflammation and regulators of immunity (5, 32, 37). Resident tissue fibroblasts produce proinflammatory cytokines and PGs upon stimulation and thus recruit and communicate with classic immune cells. Fibroblasts themselves, however, remain poorly understood and were previously considered to be only structural cells with little intrinsic heterogeneity. An emerging concept is that fibroblasts are not a homogeneous population but, rather, consist of subsets that have organ-specific functions. For example, fibroblasts are heterogeneous with respect to expression of the surface antigen Thy-1 in several organs, including mouse lung (31) and spleen (4), rat lung (26), and human orbit (39). Little is known, however, about fibroblasts derived from the human female reproductive tract. We recently demonstrated that fibroblasts derived from human myometrium are heterogeneous, and subsets separated on the basis of Thy-1 expression have distinct cytokine and chemokine profiles (21). In the current study, we investigated COX-1 and COX-2 expression by Thy-1+ and Thy-1human myometrial fibroblasts. We demonstrate that myometrial fibroblast subsets have distinct COX expression patterns, subcellular localization, and signal transduction pathways, as well as PG production. The findings in this report support the concept that human myometrial fibroblast subpopulations contribute uniquely to reproductive tract processes associated with inflammation. Our results have important implications for the way in which reproductive tract fibroblasts are viewed. In light of the evidence presented herein, myometrial fibroblasts should now be viewed as key players in PG homeostasis as well as in the initiation of inflammatory events in the reproductive tract.

MATERIALS AND METHODS

Tissue collection and fibroblast strain derivation.

Tissue samples were obtained as previously described (19). In brief, biopsies were collected from myometrium in women undergoing gynecological procedures for benign conditions. All women had regular menstrual cycles (25–35 days), were premenopausal (20–42 y old), and did not receive any form of hormonal treatment during the 3-mo period before the procedure. Written informed consent was received from all patients, and ethical approval was obtained from the University of Rochester research subjects review board.

Tissue was cut into 1-mm3 pieces, and fibroblast cultures were established by standard explant techniques as previously described (10, 35). The cells were immunostained for fibroblast markers as previously described (19, 21). In brief, cells were positive for vimentin (fibroblast marker) but negative for cytokeratin (epithelial cell marker), α-smooth muscle actin (myofibroblasts and smooth muscle cells), CD34 (endothelial cell marker), and CD45 (bone marrow-derived cell marker). The cells that proliferated under these conditions had fibroblast morphology and were not contaminated with other cell types. There was no mycoplasma contamination in the cultures. All cells used in experiments were early passage (passages 4–11). Myometrial fibroblast strains were cultured in RPMI 1640 medium (Life Technologies, Gaithersburg, MD) supplemented with 10% FBS (Hyclone Laboratories, Logan, UT), 0.1 mM nonessential amino acids, and 0.048 mg/ml gentamicin (Life Technologies).

As previously described, fibroblast subset separation was accomplished by first sorting with fluorescence-activated cell sorter according to Thy-1 expression, followed by three to four rounds of magnetic bead selection (21). The myometrial Thy-1+ and Thy-1 subpopulations had a stable Thy-1 phenotype in culture, as was determined by flow cytometry before each experiment, and were >99% Thy-1+ and >99% Thy-1, respectively.

RNA isolation and RNase protection assay.

Total RNA was isolated from myometrial fibroblast subsets at 0, 1, 2, 4, 6, 8, and 18 h after IL-1β treatment by using TRI-reagent (Molecular Research Center, Cincinnati, OH). RNA (8 μg) was used for each sample to hybridize with the probe. COX-1 (human) antisense probe template (nucleotides 379–751 of cDNA sequence accession no.S78220; Cayman Chemical, Ann Arbor, MI), COX-2 (human) antisense probe template (nucleotides 330–752 of cDNA sequence accession no.M90100; Cayman Chemical), and a human β-actin antisense control template (nucleotides 947–704 of cDNA sequence accession no.X00351; Ambion, Austin, TX) were used to prepare labeled RNA probes with the MAXIscript in vitro transcription kit (Ambion). The RNase protection assay was performed by using the RPA II kit (Ambion) according to the manufacturer's protocol.

Immunocytochemistry for COX-1 and COX-2.

Myometrial Thy-1+ and Thy-1 fibroblasts were seeded in chamber slides (Nunc, Naperville, IL) in RPMI with 10% FBS. Cells were serum-starved in 0.5% FBS for 72 h and then stimulated with IL-1β (10 ng/ml; R&D Systems, Minneapolis, MN) or left untreated for 24 h in serum-free medium. Cells were fixed with 2% paraformaldehyde and then stained with a monoclonal mouse anti-human COX-1 or monoclonal mouse anti-human COX-2 antibody (10 μg/ml; Cayman Chemicals) or with isotype control mouse IgG1 or mouse IgG2b (Caltag, Burlingame, CA). Biotinylated horse anti-mouse IgG (heavy and light chain) (1:200; Vector Labs, Burlingame, CA) was added as a secondary antibody, and streptavidin-horseradish peroxidase (1:1000; Jackson ImmunoResearch Labs, West Grove, PA) was added as a substrate. Samples were visualized by adding an aminoethyl-carbachol chromogen (Zymed, South San Francisco, CA) and cover-slipped by using Immu-mount (Shandon, Pittsburgh, PA).

For COX-1 and COX-2 immunohistochemistry, fresh myometrial tissue samples were fixed in neutral buffered formalin and embedded in paraffin, and 5-μm serial sections were cut. Staining for COX-1 and COX-2 was completed as above. Sections were counterstained with hematoxylin. For Western blot analysis, monoclonal mouse anti-human COX-1 and mouse anti-human COX-2 antibodies were used at 50 ng/ml.

PG production.

Fibroblasts were treated as described above. PGE2production was assayed by using an enzyme immunoassay (Cayman Chemicals). At 2 h before the addition of IL-1β, cultures were treated with ethanol (vehicle control), indomethacin (10 μM; Sigma, St. Louis, MO), SC-58125 (5 μM) or SC-58560 (0.5 μM). SC-58560 and SC-58125 were kind gifts from Dr. Peter Isakson (Searle, Skokie, IL).

Western blot analysis.

Cells were treated as above for COX-1 and COX-2 analysis or treated with IL-1β (10 ng/ml) and harvested in a time course at 1, 2, 5, 15, and 30 min, 1 h, and 6 h for extracellular signal-regulated kinase (ERK)1 (p42), ERK2 (p44), and p38 mitogen-activated protein (MAP) kinase analysis. The ERK1/2 inhibitor PD-98059 (Calbiochem, San Diego, CA) inhibits MAP kinase/ERK kinase (MEK1) activation and thus phosphorylation of ERK1/2 and was added at 50 μM at the indicated time points or overnight. The p38 inhibitor SB-203580 (Calbiochem) was also added at 50 μM overnight. DMSO was added as vehicle control to untreated cells. Equal amounts of fibroblast protein lysates were subjected to Western blot analysis as previously described (36). Mouse anti-human COX-1 and mouse anti-human COX-2 antibodies were added in blocking buffer. The phospho-p44/42 MAP kinase (Thr-202/Tyr-204) E10 monoclonal antibody in mouse (1:2,000) and phospho-p38 MAP kinase (Thr-180/Tyr-182) antibody in rabbit (1:1,000) were purchased from Cell Signaling Technologies (Beverly, MA). Positive controls included phosphorylated p42 and p38 MAP kinase cell extracts (5 and 10 μl).

Statistical analysis.

Statistical significance was determined by using Student's paired two-tailed t-test or a generalized linear model analysis, where P < 0.05 indicates statistical significance between the samples tested. Log transformation was performed to correct for variance.

RESULTS

COX-1 and COX-2 mRNA expression by myometrial fibroblast subsets.

Fibroblasts are one of the main cell types responsible for PG production via expression of COX-1 and COX-2 (5, 46). COXs are also central for homeostasis in the reproductive tract (17,29). For these reasons, we first examined whether or not Thy-1+ and Thy-1 myometrial fibroblasts express COX-1 and/or COX-2 mRNA. Thy-1+ and Thy-1 fibroblasts were left untreated or stimulated in a time course with IL-1β, a cytokine abundant in the reproductive tract and a known inducer of COX-2 (15). In particular, IL-1 is present in the uterus throughout the menstrual cycle and during pregnancy as a product of reproductive tract macrophages (18). COX-1 and COX-2 mRNA levels were quantitated by normalizing to β-actin by using an RNase protection assay. Thy-1+ myometrial fibroblasts expressed COX-1 mRNA constitutively, whereas COX-2 mRNA was absent in untreated cells but was greatly upregulated after IL-1β treatment (10 ng/ml) (Fig.1, A and B). COX-1 mRNA levels in myometrial Thy-1 fibroblasts were relatively unchanged throughout the IL-1β time course. Interestingly, constitutive COX-2 mRNA levels were detected in Thy-1myometrial fibroblasts and, surprisingly, were not affected by IL-1β treatment (Fig. 1, C and D). This is an intriguing finding because COX-2 is thought of as an inducible enzyme, associated with inflammatory processes.

Fig. 1.

Cyclooxygenase (COX)-1 and COX-2 mRNA expression by Thy-1+ and Thy-1 myometrial fibroblast subsets. Thy-1+ and Thy-1 fibroblasts were treated with interleukin (IL)-1β (10 ng/ml) for 1, 2, 4, 6, 8, and 18 h, and mRNA transcripts for COX-1 and COX-2 were determined by RNase protection assay. Thy-1+ myometrial fibroblasts expressed constitutive COX-1 mRNA, whereas COX-2 mRNA was absent in untreated cells and upregulated with IL-1β with maximum expression at 4 and 6 h (A and B). Thy-1myometrial fibroblasts expressed constitutive mRNA transcripts for COX-1 as well as for COX-2 (C and D). Bands were quantitated by normalizing to β-actin mRNA, and densitometry analysis was performed by using a Kodak Digital analysis system (data not shown). Blots are representative of 2 experiments with similar results.

COX-1 and COX-2 protein expression by Western blot.

Western blot analysis was next performed on untreated and IL-1β-stimulated (24 h) myometrial parental fibroblasts, as well as on Thy-1+ and Thy-1 subsets (Fig.2). A strain of human lung fibroblasts treated with IL-1β was used as a strong positive control (lane 1) for both COX-1 and COX-2 (46). COX-1 is seen as a 68-kDa band on a Western blot. Our results demonstrate that COX-1 protein was highly expressed in both unstimulated and IL-1β-treated Thy-1+ fibroblasts. The parental fibroblast strain showed similar results but with less intense bands. COX-1 protein expression in Thy-1 cells was barely detectable, seen as very faint bands. COX-2 is typically observed in fibroblasts as a 72-kDa band. Interestingly, the unstimulated myometrial parental fibroblasts possessed a band of a mass somewhat smaller than 72 kDa, which remained constant with IL-1β treatment. The typical 72-kDa COX-2 band was expressed only after IL-1β stimulation. The isolated Thy-1+ fibroblasts expressed the 72-kDa band, which was absent in unstimulated cells but was upregulated after addition of IL-1β. In contrast, the Thy-1 population displayed only the smaller COX-2 band in both unstimulated and IL-1β-treated cells, supporting the data showing constitutive mRNA for COX-2 in this myometrial fibroblast subset (Fig. 1 D). Interestingly, human myometrial tissue expresses only the 70-kDa band for COX-2 (data not shown).

Fig. 2.

COX-1 and COX-2 protein expression by parental, Thy-1+, and Thy-1 myometrial fibroblasts. COX-1 and COX-2 protein expression was determined by Western blot. Cells were treated with IL-1β (10 ng/ml) for 24 h or left unstimulated. Protein (10 μg) was separated via electrophoresis and blotted onto nitrocellulose. COX-1 and COX-2 bands were visualized by using chemiluminescence. Pos.Ctl. indicates positive control human lung fibroblasts treated with IL-1β. Lane 1, parental fibroblasts untreated; lane 2, parental fibroblasts and IL-1β; lane 3, Thy-1+ subset untreated;lane 4, Thy-1+ subset and IL-1β; lane 5, Thy-1 subset untreated; lane 6, Thy-1 subset and IL-1β. A: COX-1 expression is indicated at 68 kDa. Thy-1+ fibroblasts express high levels of COX-1 compared with parental or Thy-1fibroblasts. B: typical COX-2 expression is indicated at 72 kDa. Parental fibroblasts display a lower band specific for COX-2 (∼70 kDa) that is only expressed by the Thy-1 subset. Thy-1+ fibroblasts only express the typical 72-kDa band when stimulated with IL-1β. Blots are representative of 3 experiments with similar results. Densitometry analysis of COX-1 and COX-2 expression was analyzed by a Kodak Digital analysis system (data not shown).

Immunocytochemistry for COX-1 and COX-2 in Thy-1+ and Thy-1 myometrial fibroblasts was determined by using immunocytochemistry. High cytoplasmic expression of COX-1 was observed in the Thy-1+ subset whether left untreated or stimulated with IL-1β for 24 h (10 ng/ml) (Fig.3 A). The Thy-1subset revealed an interesting pattern of COX-1 expression, where IL-1β-stimulated cells show intense nuclear staining (Fig.3 A). COX-2 expression in the Thy-1+subpopulation was almost nonexistent in the untreated cells but was upregulated in a heterogeneous manner after IL-1β stimulation (Fig.3 A). This induction of COX-2 after IL-1β is typical for fibroblastic cells (Fig. 3 B). COX-2 staining in the Thy-1 cells revealed a cytoplasmic localization, which, interestingly, was not affected by IL-1β. This apparent constitutive COX-2 expression (Fig. 3 A) is unexpected but correlates with the Western blot COX-2 protein expression in Fig. 2. For comparison, human lung fibroblasts showing “typical” COX-1 and COX-2 staining patterns in untreated and IL-1β-stimulated cells are shown in Fig. 3 B.

Fig. 3.

Localization of COX-1 and COX-2 in fibroblast subsets. COX-1 and COX-2 expression was determined by immunocytochemistry as described in materials and methods.Isotype controls were also used at 10 μg/ml and showed no staining (data not shown). A: Thy-1+ myometrial fibroblasts exhibited high constitutive COX-1 expression, whereas COX-2 was low in untreated cells but upregulated with IL-1β (original magnification ×400). COX-1 expression in Thy-1 untreated fibroblasts was low but was associated with the nucleus after IL-1β stimulation. COX-2 expression in Thy-1 fibroblasts was constitutive and was not affected by IL-1β (original magnification ×200). B: COX-1 and COX-2 expression in human lung fibroblasts. COX-1 was constitutively expressed at low levels in unstimulated and IL-1β-treated cells. COX-2 was expressed at low levels in unstimulated cells and was upregulated with IL-1β. Data are representative of 4 experiments with similar results.

COX-1 and COX-2 expression in human myometrium.

In situ staining for COX-1 and COX-2 was performed on human myometrium tissue sections. Immunohistochemistry revealed both COX-1 and COX-2 to be expressed in human myometrium (Fig.4). COX-1 was observed throughout the tissue and was also localized to single fibroblastic cells embedded in a field of collagen, as indicated by arrows in Fig. 4 A. COX-1 localization was also apparent in the structural cells surrounding blood vessels throughout the tissue (data not shown). COX-1 was additionally expressed on smooth muscle cell bundles of the myometrium, as shown by the arrowhead (Fig. 4 A). COX-2 was highly expressed by individual fibroblasts in the stroma of the myometrium. Single fibroblasts set in a collagen matrix stained positive for COX-2 as indicated by arrows (Fig. 4 B). This localization in the tissue correlates with the in vitro data of COX-2 constitutive expression by Thy-1 myometrial fibroblasts.

Fig. 4.

Immunohistochemical localization of COX-1 and COX-2 in human myometrium. Fresh myometrial tissue samples were fixed in formalin and paraffin-embedded, and 5-μm-thick sections were cut. Immunohistochemistry was performed as described in materials and methods. A: COX-1 was expressed throughout the tissue and was also localized to single stromal cells (arrows). Arrowhead indicates expression on smooth muscle cell bundles of the myometrium. B: COX-2 was highly expressed by individual fibroblasts (arrows) in the stroma of the myometrium. Original magnification ×400. Data are representative of 3 experiments with similar results.

PG production by myometrial Thy-1+and Thy-1 fibroblasts.

PGE2 is a key proinflammatory mediator and is potentially critical in the reproductive tract for sustaining successful pregnancy and taking part in cervical ripening (17, 45). We evaluated PGE2 release by myometrial fibroblast subsets after the cells were left untreated or stimulated for 24 h with IL-1β. PGE2 production by Thy-1 fibroblasts was very low (but detectable) in unstimulated cells and, surprisingly, was not affected by IL-1β treatment. In contrast, basal PGE2 production by Thy-1+ fibroblasts was high and was increased twofold by IL-1β treatment (Fig.5 A). Thy-1+ and Thy-1 fibroblasts were also screened for ability to produce PGF2α, PGD2, and PG6-keto-F1α. Whereas Thy-1+ fibroblasts produced PGF2α and PGD2 in response to IL-1β, the Thy-1 fibroblasts produced only very low basal levels of these prostanoids, which were not upregulated by IL-1β (data not shown).

Fig. 5.

PGE2 was produced only by the Thy-1+ fibroblasts and was derived mainly from COX-1. PGE2 production was assayed by enzyme immunoassay.A: PGE2 production by Thy-1+ and Thy-1 myometrial fibroblasts. Only the Thy-1+fibroblasts produced PGE2 when stimulated with IL-1β. Statistical analysis was done by using Student's paired two-tailedt-test, where * P < 0.05 indicates statistical significance when comparing IL-1β-treated cells with untreated cells. B: PGE2 production by the Thy-1+ subset after COX inhibitor administration. Cells were treated with the COX-1 and COX-2 general inhibitor indomethacin (Indo; 10 μM), the COX-2-selective inhibitor SC-58125 (5 μM), or the COX-1-selective inhibitor SC-58560 (0.5 μM), alone or together with IL-1β. The COX-1-selective inhibitor SC-58560 effectively blocked total PGE2 production. The COX-2-selective inhibitor SC-58125 completely blocked IL-1β-induced PGE2production in a lung fibroblast strain (data not shown). A generalized linear model was used to determine statistical significance. * P < 0.01 indicates statistical significance when comparing sample with untreated or IL-1β-stimulated cells, as is respective in each case. * Statistically significant decrease after inhibitor addition for comparison of the effect of each inhibitor in reducing PGE2 from untreated cells or cells stimulated with IL-1β, as appropriate. Experiments were done in quadruplicate and repeated 3 times.

After establishing basal and induced PGE2 production in the Thy-1+ myometrial fibroblast subpopulation, we next determined the COX isoform responsible for the PGE2synthesis. Thy-1+ myometrial fibroblasts were treated with the drug indomethacin, which blocks both COX-1 and COX-2. Some groups of cells were also treated with SC-58125 (5 μM), a selective COX-2 inhibitor, or SC-58560 (0.5 μM), a selective COX-1 inhibitor (Fig.5 B). The inhibitors were added to unstimulated cells or to cells activated with IL-1β. The COX-1-selective inhibitor SC-58560, but not the COX-2-selective inhibitor SC-58125, completely abolished PGE2 production to the same level as indomethacin in the Thy-1+ subset. The COX-2 inhibitor reduced most of the IL-1β-stimulated PGE2 production.

ERK1, ERK2, and p38 MAP kinase phosphorylation in myometrial fibroblast subsets.

Previous studies demonstrated a role for ERK1 (p42) and ERK2 (p44) MAP kinases in response to IL-1β, as well as in the signaling pathway for COX-2 but not COX-1 (1, 23). The MAP kinase p38 is also involved in inflammation-type responses and can participate in COX-2 induction (14). We first determined phosphorylation of ERK1, ERK2, and p38 in the myometrial fibroblast subsets after an IL-1β time course. The MEK1 inhibitor PD-98059 (50 μM) was included at each time point to ascertain its blocking effect on phosphorylation of ERK1/2. Interestingly, the two subsets displayed differential regulation of ERK1 and ERK2. Thy-1+ fibroblasts showed sustained phosphorylation of ERK1 and ERK2, which peaked at 5–30 min after IL-1β treatment and subsided by 6 h (Fig.6 A). Thy-1fibroblasts had a more transient ERK phosphorylation that peaked by 5 min after IL-1β stimulation and disappeared by 15 min (Fig.6 A). Also, in the case of Thy-1 cells, ERK1 was phosphorylated first at the 1 and 2 min time points, whereas ERK2 only peaked at 5 min together with ERK1. PD-98059 inhibited IL-1β-induced ERK phosphorylation in both subsets. Both myometrial fibroblast subpopulations showed IL-1β-induced p38 phosphorylation, even though the p38 time course also varied between the subsets (Fig.6 B). In Thy-1+ fibroblasts, peak p38 phosphorylation was at 15 and 30 min of IL-1β treatment, whereas Thy-1 fibroblasts showed an earlier and more sustained p38 phosphorylation, beginning at 1 min of IL-1β stimulation.

Fig. 6.

Extracellular signal-regulated kinase (ERK)1, ERK2, and p38 phosphorylation by Thy-1+ and Thy-1myometrial fibroblast subsets. Phospho-p42, -p44 and -p38 were determined by Western blot as described in materials and methods. Protein (20 μg) was separated via electrophoresis and blotted onto nitrocellulose. A: ERK1 (p42) and ERK2 (p44) phosphorylation was assessed in myometrial fibroblast subsets after an IL-1β time course. The MEK1 inhibitor PD-98059 was included at each timepoint as an inhibitor of ERK1/2 phosphorylation. Thy-1+fibroblasts show sustained phosphorylation of ERK1 and ERK2 that peaked at 5–30 min of IL-1β treatment and subsided by 6 h. Thy-1 fibroblasts show a more transient ERK phosphorylation that peaked by 5 min of IL-1β treatment and diminished by 15 min. PD-98059 inhibited IL-1β-induced ERK phosphorylation in both subsets. B: Thy-1+fibroblasts show peak p38 phosphorylation at 15 and 30 min of IL-1β treatment. Thy-1 fibroblasts show a more sustained p38 phosphorylation, beginning at 1 min of IL-1β stimulation. Results are representative of 3 experiments. ctl, Control.

Effect of ERK1, ERK2, and p38 MAP kinase signaling pathways on COX-1 and COX-2 expression in Thy-1+ and Thy-1 myometrial fibroblasts.

We next examined whether ERK1/2 or p38 MAP kinases were involved in the signaling pathway of COX-1 or COX-2 expressed by the myometrial fibroblast subpopulations. Myometrial fibroblast subsets were treated with IL-1β (10 ng/ml) for 18 h, with or without the MEK1 inhibitor PD-98059 (50 μM), to determine the effect of ERKs on COX-1 and COX-2 expression. To determine the effect of p38 signaling on COX-1 and COX-2, the downstream p38 inhibitor SB-203580 (50 μM) was employed. As anticipated, COX-1 expression was not affected by PD-98059 or SB-203580 in either subset (Fig. 7). IL-1β-induced COX-2 expression was completely abolished by addition of PD-98059 in Thy-1+ fibroblasts, implicating ERK1 and ERK2 in its regulation (Fig. 7). However, PD-98059 had no effect on COX-2 expression in Thy-1 fibroblasts. Therefore, ERKs do not appear to contribute to constitutive COX-2 expression in this fibroblast population. SB-203580 also blocked COX-2 expression in Thy-1+ fibroblasts; thus p38 is also a signaling pathway utilized for COX-2 induction in this subset. Similar to PD-98059, the p38 inhibitor SB-203580 had no effect on COX-2 expression in Thy-1 fibroblasts, also excluding this MAP kinase from maintaining constitutive COX-2 in Thy-1 cells. The same results described above on the effect of ERKs and p38 MAP kinase pathways on COX-1 and COX-2 were observed by using immunocytochemistry (data not shown).

Fig. 7.

Effect of ERK1, ERK2, and p38 mitogen-activated protein (MAP) kinase signaling pathways on COX-1 and COX-2 expression in Thy-1+ and Thy-1 myometrial fibroblasts. Myometrial fibroblast subsets were treated with IL-1β (10 ng/ml) for 18 h, with or without PD-98059 (50 μM) or SB-203580 (50 μM) as described in materials and methods. A: COX-1 expression was not affected by PD-98059 or SB-203580 in Thy-1+ fibroblasts. IL-1β-induced COX-2 expression was completely abolished by addition of PD-98059 or SB-203580 in the Thy-1+ subset. B: COX-1 expression was unaffected by PD-98059 or SB-203580 in Thy-1 fibroblasts. PD-98059 and SB-203580 did not inhibit COX-2 expression by Thy-1 myometrial fibroblasts. Blots are representative of 3 experiments with similar results.

DISCUSSION

Fibroblasts are dynamic cells whose functions and characteristics differ according to their anatomic location and the environment to which they are exposed. Variations in fibroblast responses from tissue to tissue and within a single tissue can now be explained by the existence of subsets within a fibroblast population. The data contained herein support the concept that two types of fibroblasts inhabit the human myometrium and that these subsets have unique roles in producing PGs, seminal mediators of inflammatory responses, and reproductive tract functions. Table 1 summarizes the characteristics of the Thy-1+ and Thy-1subsets and their COX and PG profiles. We have established that only Thy-1+ human myometrial fibroblasts are capable of producing substantial levels of PGE2, a PG that has dramatic effects on processes of the reproductive tract. PGE2 is a vasoactive mediator, participating in leukocyte recruitment at the site of its production. A role for invading leukocytes has been postulated in parturition (17). Another important role of PGE2 is its synergy with IL-8 for neutrophil recruitment (17, 18), which is involved in cervical ripening, a prerequisite for safe labor (22). Infiltrating neutrophils are the main source of the collagenase involved in digesting collagen bundles, leading to the ripening or softening of the cervix (17). Thy-1+ and Thy-1 myometrial fibroblasts were recently shown to produce IL-8 upon stimulation with IL-1β (21). Thy-1+ myometrial fibroblasts can thus contribute to processes such as cervical ripening, acting as effectors in the initiation of the inflammatory response and neutrophil recruitment at the time preceding parturition.

View this table:
Table 1.

Summary of key cyclooxygenase and prostaglandin expression patterns and signal transduction conduits in Thy-1+ and Thy-1 human myometrial fibroblasts

We demonstrate that high basal levels of PGE2 in the Thy-1+ subset were mediated through COX-1. This is an extremely interesting finding, because high PGE2 production is usually a consequence of COX-2 upregulation (6). The data reported herein are the first to demonstrate that PGE2production is mainly mediated by COX-1 in human reproductive fibroblasts. We argue that human myometrial fibroblasts are unique in this respect, further establishing the concept of fibroblast heterogeneity across tissues. Because PGE2 is so critical in the reproductive tract in sustaining pregnancy by maintaining a type 2 cytokine environment, it makes sense that it is mainly synthesized by COX-1, a largely constitutive enzyme. The high COX-1 expression in Thy-1+ fibroblasts allows sustained PGE2synthesis and designates the Thy-1+ myometrial fibroblast subset as a critical component in maintaining necessary PGE2 homeostasis in the myometrium.

COX-1 and COX-2 expression in Thy-1 myometrial fibroblasts is also of great interest. The expression pattern of COX-1 and COX-2 in Thy-1 fibroblasts does not follow that typical of fibroblasts. COX-1 is expressed at barely detectable levels in unstimulated cells but becomes associated with the nucleus upon IL-1β treatment. This is a novel and exciting finding, because it is the first time that we are aware of that COX-1 expression was shown to localize to the nucleus after stimulation. Its significance, however, is unclear at present. We speculate that a function of nuclear COX-1 is the production of PGs that can interact with nuclear receptors, such as peroxisome proliferator-activated receptors (PPARs). Furthermore, PGE2 released at the nuclear site could interact with nuclear PGE2 receptors (2, 3) and directly modulate gene transcription. Further studies are needed to determine the function of nuclear COX-1-mediated prostanoids and ascertain their role in the human female reproductive system.

COX-2 expression in Thy-1 myometrial fibroblasts was also unusual in that it is constitutively expressed in both unstimulated and IL-1β-treated cells. Recent evidence, however, indicates that some tissues appear to have constitutive COX-2, such as human thyroid epithelium (38), rat kidney (13), and stromal cells in the mouse small intestine lamina propria (30). Interestingly, constitutive COX-2 expression was also reported in the rat and human male reproductive tract (20, 27). It appears that constitutive COX-2 plays a key role in the reproductive system. Our current findings indicate that COX-2 maintenance in Thy-1 fibroblasts is ERK1/2 and p38 independent. Thus the cause of basal COX-2 protein levels in Thy-1 cells remains to be discovered. A possibility could be an autocrine or paracrine stimulus that adjusts COX-2 levels in these cells. Future studies need to assess the importance of constitutive COX-2 in the human female reproductive tract and determine its regulation in Thy-1 myometrial fibroblasts.

The lower molecular weight band detected for COX-2 in the Thy-1 subset (as revealed by Western blot) compared with the Thy-1+ cells is of interest. Human myometrial tissue expresses this lower 70-kDa band for COX-2 (data not shown). The lower band isoform could arise due to a glycosylation change, making it less capable of metabolizing arachidonic acid. In experiments where exogenous arachidonic acid was added, the Thy-1 subset still produced very low levels of PGE2 (data not shown). The Thy-1 myometrial fibroblast subset can, however, respond to IL-1β by upregulating IL-6 and IL-8 production (21). PG synthesis also depends on the presence of PG synthases, whose expression may vary during different events (for example, pregnancy) in the reproductive tract. Low activity of PGE2 synthase in Thy-1 fibroblasts could account for the low PGE2 production. Another possibility is that this form of COX-2 acts as a preformed “precursor” proenzyme that awaits a stimulus for its activation and a robust synthesis of PGs.

Labor is associated with increased COX-2 expression in uterine tissues (29, 43). COX-2 expression is suggested to be responsible for induction of PGF2α (33), which is a vasoconstrictor and responsible for labor induction in mice (42). Specifically, the primary role of the myometrium is to provide contractions during labor. In our studies, the Thy-1+ myometrial subset was capable of increased PGF2α production with IL-1β stimulation, and this was mainly via COX-2, as determined by using COX-selective inhibitors (data not shown). Given that the COX general inhibitor indomethacin is widely and successfully used to prevent preterm delivery in humans, a role implicating COXs is suggested. Furthermore, studies administering the COX-2 selective drug nimesulide in sheep showed a delay in delivery when labor was induced in ewes (33). COX-2 selective inhibitors were also shown to inhibit spontaneous uterine contractions in the rat (34). Our results provide insight to these reports and suggest the potential use of a COX-2-selective drug to prevent preterm labor. In this way, fetal well-being is less likely to be compromised, because inhibition of COX-1-derived PGE2will not occur. Alternatively, premature contractions may be better inhibited by selectively blocking COX-2 and PGF2α. However, care should be taken in this assessment because the function of constitutive COX-2 in the myometrium remains to be elucidated.

In conclusion, Thy-1+ and Thy-1 fibroblast subsets in the female human myometrium are dynamic cell populations with distinct characteristics of COX subcellular distribution, regulation, and PG production. The identification of unique fibroblast subpopulations in the human reproductive tract that have distinct profiles with respect to inflammatory phenotype underscores the importance of considering alternate roles of fibroblast subpopulations in reproductive tract processes. It is likely that each myometrial fibroblast subpopulation contributes differently to both the physiological functions and the development of inflammation in the human reproductive system. Further understanding of fibroblast subset physiology is imperative, because the overproliferation of one subpopulation may render a patient more susceptible to a condition involving an inflammatory process in the human reproductive tract.

Acknowledgments

We thank Dr. Kerry O'Banion and Dr. Patricia Sime for critical reading of the manuscript.

Footnotes

  • This research was supported by United States Public Health Service Grants DE-11390, HL-56002, EY-08976, and ES-01247, the Burroughs-Wellcome Foundation, and the Pepper Center.

  • Address for reprint requests and other correspondence: R. P. Phipps, Univ. of Rochester School of Medicine and Dentistry, 601 Elmwood Ave, Box 850, MRB-X Rm. 3-11001, Rochester, NY 14642 (E-mail:Richard_Phipps{at}urmc.rochester.edu).

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • April 24, 2002;10.1152/ajpcell.00065.2002

REFERENCES

View Abstract