Abstract
Prostaglandins (PGs) are key regulators of reproductive function and associated pathologies. We have established stable endometrial stromal and epithelial cell lines with SV40 large T antigen (TAG) as a model to study PG action in the human endometrium. Two clones for each cell type were selected for rapid growth, PG production and response to interleukin-1β (IL-1β). The resulting stromal (HIESC) and epithelial (HIEEC) cells retain their characteristics for at least 40 population doublings (PDs). The selected clones express progesterone (PR) and estrogen receptor-α (ER-α) at both mRNA and protein levels. By contrast, with the existing known human endometrial cell lines Ishikawa and KLE, HIESC and HIEEC increase their production of PGF2α and PGE2 and cyclooxygenase (COX)-2 protein expression in response to IL-1β. The latter cells also express the main biosynthetic enzymes involved in PG production, cytosolic phospholipase A2 (PLA2), COX-1 and COX-2, PGF synthase and PGE synthase and the corresponding EP2, EP3, EP4 and FP receptors. The selective COX-2 inhibitor NS-398 completely inhibits the increased production of PGs induced by IL-1β in both cell types, whereas dexamethasone (DEX) exerts a stronger inhibition in HIESC than in HIEEC. The latter observation may be related to the higher expression of COX-1 measured in HIEEC. On the basis of the present characterization and previous determination of corresponding primary cell cultures, HIESC and HIEEC appear appropriate to study the contribution of PGs in the regulation of human endometrium function and associated pathologies.
Introduction
The human endometrium undergoes cyclical changes necessary to develop uterine receptivity in response to the ovarian steroid hormones, estradiol (E2) and progesterone. In addition to sex steroids, prostaglandins (PGs) appear as primary regulators of female reproductive function. PGs are involved in the regulation of ovulation, implantation, menstruation and several other aspects of endometrial function (for a review, see Sales and Jabbour, 2003a). PGs are also involved in pathologies of the endometrium including carcinomas, menorrhagia, dysmenorrhoea and endometriosis (for a review, see Sales and Jabbour, 2003b). The biosynthetic pathway leading to PG production begins by the activation of phospholipase A2 (PLA2) or phospholipase C (PLC) liberating arachidonic acid (AA) from membrane phospholipids (for a review, see Kudo and Murakami, 2002). AA is then converted to PGG2 and PGH2 by PG synthases, suicide enzymes also known as cyclooxygenases (COXs) for which there are two isotypes, COX-1 and COX-2 (for a review, see Simmons et al., 2004). PGH2 is the common precursor for all PGs produced through terminal enzymes like PGE synthase for PGE2 and PGF synthase for PGF2α (for a review, see Helliwell et al., 2004). The importance of PGs in reproduction was evidenced by exhibition of multiple reproductive failures in mice where the PG synthase 2 (COX-2) (Lim et al., 1997) as well as FP (Sugimoto et al., 1997) and EP2 (Kennedy et al., 1999) genes were disrupted. We have described the expression and regulation of PG biosynthetic enzymes and receptors in the bovine (Madore et al., 2003; Parent et al., 2003; Arosh et al., 2004a,b; Parent and Fortier, 2005) and human endometrium (Kang et al., 2004).
In the endometrium, PGE2 and PGF2α are the main prostanoids produced (Smith and Kelly, 1988; Sales and Jabbour, 2003a), and their autocrine and paracrine actions are mediated by G-protein-coupled EP and FP receptors. Altered production of PGE2 and PGF2α has been found in uterine fluids in association with gynaecological disorders such as menorrhagia, dysmenorrhoea and endometriosis (Sales and Jabbour, 2003b). Cytokines such as interleukin-1β (IL-1β) are involved in endometrial–embryonic crosstalk at the time of implantation (Simon et al., 1997) and stimulate the production of PGE2 (Tamura et al., 2002) and PGF2α (Huang et al., 1998). The increased production of PG in human decidual stromal cells is associated with an increased number of cells positive for COX-2 in vivo (Ishihara et al., 1995) and increased COX-2 protein expression in vitro (Tamura et al., 2002). We studied the effect of IL-1β on monocyte chemotactic protein (MCP)1, a chemokine involved in uterine receptivity and endometriosis, and PG production using human endometrial stromal and epithelial cells in primary culture (Kang et al., 2004). That study highlighted the difficulties and limitations of biopsies and primary cultures for extensive molecular and cellular studies in the human endometrium. The endometrium is composed mainly of epithelial, stromal and infiltrating haematopoietic cells. Isolated stromal cells in primary culture can go through 10–20 population doublings (PDs) (Rinehart et al., 1991) with minimal morphological changes, whereas epithelial cells do not grow as well in primary culture and exhibit limited ability for subculture (Matthews et al., 1992). Clearly, the difficulty to obtain endometrial biopsies of sufficient size to prepare large populations of cells in primary culture and the inherent variability between such cultures make it difficult to develop a working model for the study of cellular and molecular regulation of PG or related systems.
It has been reported that stable endometrial stromal (Krikun et al., 2004) and glandular epithelial (Kyo et al., 2003) cells could be generated using telomerase to overcome senescence observed particularly in epithelial cells in primary culture (Matthews et al., 1992). Other reports have described the use of SV40 large T antigen (TAG) to extend the lifespan of human endometrial cells in culture (Rinehart et al., 1991; Merviel et al., 1995), but there is only limited availability of normal epithelial cell lines to compare side by side with stromal cells. It has been reported that immortalization of epithelial or other cells with human telomerase reverse transcriptase (hTERT) alone could yield cell lines with limited lifespan (Kyo et al., 2003; Barbier et al., 2005). Therefore, we elected to use TAG as proposed by Merviel et al. (1995).
In this study, we developed a strategy based on the use of a defective retrovirus SSR69 (Westerman and Leboulch, 1996) coding for the TAG protein to extend the lifespan of human endometrial stromal and epithelial cells. We then examined if the resulting HIESC and HIEEC cell lines exhibited the phenotypic characteristics of the primary cultures. Finally, we characterized PG biosynthesis and its modulation using IL-1β.
Materials and methods
PG production and IL-1β response in established endometrial cell lines
Ishikawa cells, an estrogen-responsive endometrial cancer line, and KLE cells (ATCC CRL1622) were grown to confluence and treated or not with IL-1β (1 ng/ml) for 24 h. The culture medium was recovered for measurement of PG by enzyme-linked immunosorbent assay (ELISA) as described below. The cells were processed as described for cell lines below, and expression of PG biosynthetic enzymes, COX-1 and COX-2, was evaluated by Western blot. Response to IL-1β treatment was evaluated by measurement of alteration of the above parameters.
Isolation and culture of primary endometrial cells
Endometrial biopsies were taken from women with regular menstrual cycles undergoing gynaecological investigation for benign conditions. Informed consent for donation of anonymous endometrial samples was obtained before tissue collection. Biopsies were classified first according to the stated last menstrual period (LMP) and confirmed by histological examination by the hospital pathologist using standard criteria (Noyes et al., 1950). The pathologist also confirmed the absence of neoplasia and endometritis. The research protocol was approved by Centre Hospitalier Universitaire de Québec (CHUQ) ethics committee on human research. The tissue samples were placed in sterile Hank’s balanced salt solution containing 100 IU/ml penicillin, 100 µg/ml streptomycin and 0.25 µg/ml amphotericin B at 4°C and transported to the laboratory immediately. Primary human endometrial stromal and epithelial cells were prepared as we described recently (Kang et al., 2004). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM/F12) (Sigma, Oakville, ON, Canada) supplemented with 10% fetal bovine serum (FBS) (Wisent Inc, Montreal, QUE, Canada), 10 µg/ml insulin (Invitrogen, Burlington, ON, Canada), 5 µg/ml transferrin, 50 IU penicillin-streptomycin (Wisent Inc, QUE).
The primary cultures used to generate purified stromal and epithelial cells in this study were prepared from biopsies taken respectively from a 40-year-old woman on day 17 of her menstrual cycle and a 37-year-old woman on day 12. The purity of the cultures was assessed morphologically under a phase contrast microscope and immunohistochemically according to the expression of vimentin and cytokeratin as described previously (Kang et al., 2004).
Retrovirus infection and establishment of SV40 TAG cell lines
The retroviral vector SSR69 containing SV40 large TAG and a gene resistant to hygromycin (Figure 1) was transfected with effectene (Qiagen, Mississauga, ON, Canada) in the mouse amphotropic packaging cell line PA 317 (Westerman and Leboulch, 1996). The resulting colonies resistant to hygromycin (800 µg/ml, Roche, Mississauga, ON, Canada) were cultured, and the supernatants containing amphotropic viruses were collected and used to infect, separately, purified stromal and epithelial cells in primary culture. Endometrial cells grown in six-well plates were infected in the presence of polybrene (8 µg/ml, Sigma) for 6 h, and the procedure was repeated 24 h later. The day following the last infection, the cells were trypsinized and seeded in 10 mm dishes in the presence of hygromycin (400 µg/ml). The cultures were grown for 7–8 days until the TAG-infected cells formed colonies while control non-infected cells died in the presence of the antibiotic. A total of 17 clones (17 colonies) for stromal cells and 50 for epithelial cells were picked by clonal selection (cloning o-ring) and grown in 24-well plates until confluency and then seeded in T-25 flasks. The PD for TAG clones was calculated as follows: n(PD) = log(final cells count) – log(inoculation cell count) / 0.301. Because colonies are produced from a single cell, we calculated that at confluency, the initial PD in T-25 flask was 19.2. The TAG clones were maintained in complete culture medium unless specified differently. The clones were then selected according to their growth rate, production of PGs and response to IL1-β. Four cell lines, two of stromal origin (HIESC, clones 2 and 16) and two of epithelial origin (HIEEC clones 18 and 22), were selected and characterized thoroughly in this study.
Schematic representation of the SSR69 retrovirus containing SV40 large T antigen (TAG). Recombinant viruses containing the SV40 TAG and hygromycin resistance genes were packaged in the amphotropic cell line PA-317. The virus released by the cells was used to infect separated human endometrial stromal and epithelial cells in primary culture.
Characterization of HIESC and HIEEC
Stromal and epithelial TAG clones were grown on Lab Tek 4-chamber slides (Nalge Nunc International, Rochester, NY, USA), fixed and processed as previously described (Kang et al., 2004). Monoclonal antibodies against vimentin (Sigma) or cytokeratin (recognizing 4, 5, 6, 8, 13 and 18 isoforms, Sigma) were used as first antibodies and a fluorescent IgG antibody (Alexa Fluor® 488 goat anti-mouse Molecular Probes inc, Eugene, OR, USA) as a second antibody to characterize endometrial cells by immunofluorescence. The fluorescence was visualized using a Zeiss Axiovert 100-Inverted microscope (Zeiss, Germany), and images were captured and integrated using the Northern Exposure program (Empix Imaging Inc., ON, Canada).
Integration of viral SV40 large TAG DNA in the genome of TAG clones was analysed by PCR directly from genomic DNA prepared as described elsewhere (Chapdelaine et al., 1993). The primer pairs were 5′-CAGGTTAAGATCAAGGTCTT-3′ (forward) and 5′-GTGTCGTCCATCACAGTT-3′ (reverse) corresponding respectively to the psi sequence and hygromycin sequence carried by the retrovirus SSR69. Briefly, PCR was performed with 250 ng of genomic DNA and cycle program was the same as described for RT-PCR in the section below. Dystrophin exon 50 amplified with the specific primers 5′-AGGAAGTTAGAAGATCTGAGCTCT-3′ (forward) and 5′‐AGGCTCCAATAGTGGTCAGTCCA-3′ (reverse) were used as internal standard. Protein TAG detection was analysed by immunofluorescence directly on the cells with antibody specific to the protein [SV40 TAG (ab-2)] (Oncogen Research Products, San Diego, CA, USA), and the molecular weight (90 kDa) was determined by Western blot. The karyotype was evaluated for two representative cell lines HIESC-2 (PD 47) and HIEEC-22 (PD 42) using the G-banding procedure to identify the chromosomes (Mitelman, 1995). Sex steroid responses were evaluated either by treatment (24 h) with 17 β-E2 100 nM or up to 10 days with medroxyprogesterone acetate (MPA) and assessment of estrogen receptor (ER) and progesterone receptor (PR) protein and 3H-Thymidine incorporation.
Detection of steroid and PG receptors by RT-PCR
The expression of ERs (ERα and ERβ) and PR mRNAs was analysed by RT-PCR. The primer pairs used were 5′-GCAGACAGGGAGCTGGTTCA-3′ (forward) and 5′-AGAGATGCTCCATGCCTTTG-3′ (reverse) for ERα (524 bp) (accession number X03635), 5′-TCACATCTGTATGCGGAACC-3′ (forward) and 5′-CGTAACACTTCCGAAGTCGG-3′ (reverse) for ERβ (345 bp) (accession number BC024181), 5′-AACATGTCAGTGGGCAGATG-3′ (forward) and 5′-GCAGCAATAACTTCAGACATC-3′ (reverse) for PR (438 bp) (accession number NM_000926). Total RNA was extracted from cells grown in T-75 flasks and 1 ml of TRIZOL reagent (Invitrogen) was added to isolate RNA according to the manufacturer’s instructions. The first strand of cDNA was synthesized at 42°C (50 min) from 2.5 µg of RNA using superscript II reverse transcriptase RNAse H (Invitrogen) as described by the manufacturer with random hexamer primers (250 ng) (Amersham Biosciences, Piscataway, NJ, USA). Usually, 2 µl aliquots of the first-strand cDNA were amplified with the primers (20 pmol) previously described and PCR was performed using Taq DNA polymerase (Amersham Biosciences) for 35 cycles: 95°C (1 min) as denaturing temperature, 55°C (1 min) as annealing temperature, 72°C (1.5 min) as elongating temperature and a final step at 72°C (10 min). The PCR products were analysed on 1.5 % agarose gels stained with ethidium bromide (Sigma). For the PG receptors, the primer pairs used were 5′-TTGTCGGTATCATGGTGGTG-3′ (forward) and 5′-ATGTACACCCAAGGGTCCAG-3′ (reverse) for EP1 (159 bp) (accession number NM_000955), 5′-GTCTGCTCCTTGCCTTTCAC-3′ (forward) and 5′-CGACAACAGAGGACTGAACG-3′ (reverse) for EP2 (175 bp) (accession number NM_000956), 5′-AGCTTATGGGGATCATGTGC-3′ (forward) and 5′-ACAGCAGGTAAACCCAAGGA-3′ (reverse) for EP3 (197 bp) (accession number NM_000957), 5′-CTGGTGGTGCTCATCTGCT-3′ (sense) and 5′-TATCCAGGGGTCTAGGATGG-3′ (reverse) for EP4 (149 bp) (accession number NM_000958), 5′-TCTGGTTACAATGGCCAACA-3′ (forward) and 5′-ATGCACTCCACAGCATTGAC-3′ (reverse) for FP (183 bp) (accession number NM_000959). Human β-actin was used as the positive control and the primers used were 5′-ACGGCTGCTTCCAGCTCCTCC-3′ (forward) and 5′-AGCCATGCCAATCTCATCTTGT-3′ (reverse) for a PCR product of 524 bp (accession number BC013835). RNA preparation and cDNA transcription procedures were the same as described above for steroid receptors.
In vitro decidualization of HIESC and prolactin assay
Induction of decidualization was performed as described for primary stromal cell cultures by Brosens et al. (1999). Briefly, confluent HIESC were grown in six-well plates in RPMI without phenol-red [Gibco-BRL (Invitrogen), Mississauga, Ontario, Canada] containing 2% dextran-coated charcoal-treated FBS (DCC-FBS) and 1% penicillin-streptomycin in the presence or absence of 0.5 mM 8-bromo-cAMP (Sigma, St. Louis, MO, USA) and 10–6 M MPA (Pharmacia Canada, Mississauga, ON, Canada). As a control, the epithelial cell line HIEEC was subjected to the same treatment. The culture medium was changed every other day and supernatants were collected at days 1, 6 and 10 of treatment for prolactin (PRL) assay. PRL measurement was done by ELISA using the ADVIA Centaur® immunoassay system (Bayer HealthCare LLC, Tarrytown, NY, USA). The lower detection limit was 0.8 µg/l. The intra- and inter-assay coefficients of variation (n = 144) were 1.9–4.4 and 2–5.3%, respectively.
Propagation and treatment of TAG clones
The culture and propagation of HIESC 2 and 16 and HIEEC 18 and 22 was done in RPMI 1640 (Invitrogen, ON, Canada) without phenol red, containing 10% FBS (Wisent Inc, QUE) and 50 UI penicillin-streptomycin (Wisent Inc, QUE). Briefly, cells were grown at 37°C under a humidified atmosphere of 5% CO2 and 95% air. The cells at PD 40 were first cultured in T-flasks (75 cm2) until early confluency and then plated at a 1/2 split ratio in 24-well plates. After 24 h, the culture medium was replaced with fresh medium supplemented with 10 % FBS depleted of steroids by dextran-charcoal (DC) extraction and cells grown for 4 days with a medium change every other day. After this period, the culture medium was replaced for RPMI without serum and cells were stimulated or not with IL-1β (1 ng/ml) for 24 h. Cells were stimulated in the presence or absence of a specific COX-2 inhibitor NS-398 (1 µM) or an other known inhibitor of PG biosynthesis, dexamethasone (DEX) (from 0 to 1 µM). At the end of the treatment period, the culture medium was recovered and stored at –20°C until protein (Chapdelaine et al., 2001) and PG analysis.
Enzyme immunoassays of PGs
PGE2 and PGF2α were assayed by enzyme immunoassays using acetylcholinesterase-linked PG tracers (Cayman, Ann Arbor, MI, USA) as described previously (Asselin et al., 1996). Rabbit anti-PGE2 (kindly provided by Dr TG Kennedy) (Evans et al., 1982) and sheep anti-PGF2α (Bio-Quant, Ann Arbor, MI, USA) were used as selective antibodies. The inter- and intra-assay coefficient of variations (n = 12) were 16 and 10% respectively.
Measurement of protein expression by Western blot analysis
HIESC and HIEEC were washed twice in phosphate-buffered saline (PBS 1×), and proteins were extracted and measured directly in SDS-PAGE loading buffer as previously described (Chapdelaine et al., 2001). Aliquots of 10 µg protein were loaded in each well and separated on 10 or 12.5% SDS-PAGE and then transferred onto nitrocellulose membranes (Bio-Rad, Mississauga, Merk Montreal, QUE, ON, Canada). The membranes were blocked overnight at 4°C in 5% (w/v) non-fat dried milk (BLOTTO) in PBS containing 0.05% Tween-20 (PBS-T) and then incubated with first antibodies for 1 h at room temperature. The first antibody dilutions were as follows: anti-ERα (sc-538, Santa-Cruz biotechnology, CA, USA) dilution 1/1000, anti-PR (sc-543, Santa-Cruz biotechnology) dilution 1/500, anti-Phospho-cPLA2 (Ser505, Santa-Cruz biotechnology) dilution 1/1000, anti-microsomal PGE synthase 1 (anti-mPGES1) (Cayman) dilution 1/500, anti-COX-2 (Merck 243) and anti-COX-1 (Merck 241) dilution 1/3000 were kindly provided by Dr S Kargman (Merck, QUE, Canada), anti-AKR1B1, a polyclonal antiserum raised from recombinant protein in our laboratory (Madore et al., 2002), was used at a 1/2000 dilution, and anti-β-Actin (Sigma) was used as an internal standard at a 1/5000 dilution. After incubation of the membranes with first antibodies in PBS-T containing 5% BLOTTO, three washings were performed at room temperature in PBS-T (10 min). Then, the membranes were incubated for 1 h at room temperature with the second antibodies, goat anti-rabbit or anti-mouse, conjugated with horseradish peroxidase (Jackson Immunoresearch Laboratories, Cedarlane, ON, Canada) diluted 1/10 000 in PBS-T with 5% BLOTTO followed by three washes (10 min) in PBS-T. The chemiluminescence signal was analysed with an autoradiography film (2 min exposure) after treatment of the membrane with Renaissance reagent (NEN, Perkins Elmer, Boston, MA, USA).
Statistical analysis
Data are presented as the mean ± SEM. Statistical analysis was performed using ANOVA followed by Fischer’s Protected LSD, Duncan New Multiple Range and Student-Newman-Keuls multiple comparison tests (Super ANOVA; Abacus Concepts, Berkeley, CA, USA).
Results
PG production and IL-1β response in established endometrial cell lines
The production of PGE2 and PGF2α by Ishikawa and KLE cells is presented in Table I and compared with that of primary stromal and epithelial cell cultures. The production of the two PGs is much higher in primary endometrial cells than in the two cell lines both in the absence and presence of IL-1β (1 ng/ml). Western analysis showed that low production of PGs in Ishikawa and KLE occurs in spite of high expression of COX-2 and significant expression of COX-1 (Figure 2). However, in agreement with the measured production of PGs, IL-1β does not stimulate COX-1 or COX-2 protein expression. Low production of PGs by comparison with primary endometrial cell cultures (Table I) and absence of effect on COX-1 and COX-2 expression (Figure 2) show that these cell lines do not represent a working model to study the PG system of endometrial cells.
Comparison of prostaglandin (PG) production between primary endometrial and endometrial adenocarcinoma cells
| Primary endometrial cells | Cell lines | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Stromal | Epithelial | Ishikawa | KLE | |||||||||||
| PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | |||||||
| Control | 550 ± 150 | 330 ± 80 | 560 ± 80 | 250 ± 50 | <35 | <35 | <35 | <35 | ||||||
| IL-1β (1 ng/ml) | 3100 ± 70 | 3600 ± 110 | 1100 ± 250 | 1090 ± 30 | <35 | 100 | <35 | <35 | ||||||
| Primary endometrial cells | Cell lines | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Stromal | Epithelial | Ishikawa | KLE | |||||||||||
| PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | |||||||
| Control | 550 ± 150 | 330 ± 80 | 560 ± 80 | 250 ± 50 | <35 | <35 | <35 | <35 | ||||||
| IL-1β (1 ng/ml) | 3100 ± 70 | 3600 ± 110 | 1100 ± 250 | 1090 ± 30 | <35 | 100 | <35 | <35 | ||||||
PGs were measured in triplicate wells from three separate experiments (n = 3). PGE2 and PGF2α levels are expressed as mean ± SE pg/ml of culture medium from confluent cultures following treatment or not with IL-1β for 24 h.
Comparison of prostaglandin (PG) production between primary endometrial and endometrial adenocarcinoma cells
| Primary endometrial cells | Cell lines | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Stromal | Epithelial | Ishikawa | KLE | |||||||||||
| PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | |||||||
| Control | 550 ± 150 | 330 ± 80 | 560 ± 80 | 250 ± 50 | <35 | <35 | <35 | <35 | ||||||
| IL-1β (1 ng/ml) | 3100 ± 70 | 3600 ± 110 | 1100 ± 250 | 1090 ± 30 | <35 | 100 | <35 | <35 | ||||||
| Primary endometrial cells | Cell lines | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Stromal | Epithelial | Ishikawa | KLE | |||||||||||
| PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | PGE2 | PGF2α | |||||||
| Control | 550 ± 150 | 330 ± 80 | 560 ± 80 | 250 ± 50 | <35 | <35 | <35 | <35 | ||||||
| IL-1β (1 ng/ml) | 3100 ± 70 | 3600 ± 110 | 1100 ± 250 | 1090 ± 30 | <35 | 100 | <35 | <35 | ||||||
PGs were measured in triplicate wells from three separate experiments (n = 3). PGE2 and PGF2α levels are expressed as mean ± SE pg/ml of culture medium from confluent cultures following treatment or not with IL-1β for 24 h.
Western blot analysis of cyclooxygenase (COX)-1 and COX-2 in Ishikawa and KLE. Ishikawa and KLE cells were grown to confluency and treated or not with increasing IL-1β concentrations (0.01, 0.1 and 1 ng/ml). COX-1 (lower panel) and COX–2 (upper panel) expressions were evaluated by Western analysis of cell protein extracted as described in Material and methods.
Phenotypic characteristics of endometrial stromal and epithelial TAG clones
Endometrial stromal and epithelial clones were expanded and maintained in culture for at least 40 PDs with minimal morphological changes. Although HIESC and HIEEC could be maintained in continuous culture for over 40 PD without apparent senescence, non-transformed stromal cells exhibited a change in shape at passage 15, whereas epithelial cells could not be passaged. Figure 3 illustrates the morphological characteristics of primary stromal (A) and epithelial (B) cells compared with the corresponding stromal (C, E) and epithelial (D, F) TAG clones. Among the selected clones, HIESC-2 and HIEEC-22 were fast growing lines reaching confluency after 3–4 days, whereas HIESC-16 and HIEEC-18 were slightly slower, taking 5 days to reach confluency following a standard 1:3 split ratio.
Morphological and phenotypic characteristics of endometrial cells in culture. Endometrial stromal (A) and epithelial (B) cells were isolated and grown to confluency before infection with the SV40 large T antigen described in Figure 1. The resulting stromal HIESC-2 (C, E) and epithelial HIEEC-22 (D, F) shown at PD 40 expressed morphological characteristics comparable with the original primary cultures (A, B). The expression of the cytoskeleton filaments vimentin and cytokeratin was studied by immunofluorescence as described in Materials and methods. Vimentin was expressed by both HIESC (C) and HIEEC (D) but cytokeratin (E, F) was expressed in HIEEC (F) only. Clones 2 and 22 are shown, and other clones tested showed same properties (magnification 100×).
Immunofluorescence was used to confirm intermediate filament protein expression in HIESC and HIEEC at PD 40. In Figure 3, panels C and D illustrate that vimentin is expressed in both stromal HIESC-2 and epithelial HIEEC-22 clones, whereas panels E and F illustrate that cytokeratin is present only in HIEEC-22. These observations on TAG-transformed cells correlate well with other (Matthews et al., 1992; Nisolle et al., 1995) and our own characterization of primary endometrial cells (Kang et al., 2004). The same patterns of expression were found for the other stromal and epithelial clones tested (results not shown).
Karyotype analysis of the representative HIESC-2 and HIEEC-22 cell lines is summarized in Table II. The stromal HIESC-2 cell line has an intact 46XX karyotype, whereas the epithelial HIEEC-22 has two deletions, one X and one 5 chromosome.
Karyotype analysis of TAG cells
| Cells | Karyotype |
|---|---|
| HIESC-2 | 46XX |
| HIEEC-22 | 45X –X, –5, +mar |
| Cells | Karyotype |
|---|---|
| HIESC-2 | 46XX |
| HIEEC-22 | 45X –X, –5, +mar |
Karyotype analysis of TAG cells
| Cells | Karyotype |
|---|---|
| HIESC-2 | 46XX |
| HIEEC-22 | 45X –X, –5, +mar |
| Cells | Karyotype |
|---|---|
| HIESC-2 | 46XX |
| HIEEC-22 | 45X –X, –5, +mar |
Presence of SV40 large TAG in infected endometrial cell lines
The expression of SV40 large TAG was evaluated by immunofluorescence with an antibody specific to TAG protein. Figure 4A shows that both HIESC-2 and HIEEC-22 at PD 40 express a strong nuclear fluorescence in all cells. The same was observed in all HIESC and HIEEC clones tested (results not shown). Moreover, viral integration was confirmed by amplification of a specific DNA fragment (expected size of 750 bp) from genomic DNA extracted from TAG clones and analysed by RT-PCR (Figure 4B). Similarly, a positive signal was found at the expected molecular weight (90 kDa) following Western analysis of the same cell lines with a specific antibody against SV40 large TAG protein (Figure 4C).
Integration of the recombinant retroviral SV40 large T antigen (TAG) within the genome of HIESC and HIEEC. TAG expression was analysed by immunofluorescence (A) in HIESC and HIEEC, clones 2 and 22 shown respectively, magnification 100×. Panel B represents the PCR analysis where viral integration results in an amplified product at 750 bp. HIESC clones 2 and 16 and HIEEC clones 18 and 22 are represented. Genomic DNA extracted from stromal cells (SC) in primary culture or from retroviral plasmid (P) was used as negative and positive controls, respectively. Amplification of dystrophin (exon 50) giving a PCR product of 100 bp was performed as an internal standard. Panel C represents the analysis of TAG protein expression by Western analysis. A band at the expected MW of 90 kDa can be seen in HIESC (clones 2 and 16) and HIEEC (clones 18 and 22) but not in non-infected SC.
Expression of steroid receptors in endometrial cell lines
The expression ofsteroid receptors was evaluated by RT-PCR (Figure 5). ERα and PR mRNAs were present in every cell line tested (Figure 5A), whereas ERβ was expressed minimally and not detected in every cell line (result not shown). At the protein level, both isoforms of PR, A (approximately 80–90 kDa) and B (approximately 110 kDa), are recognized by the sc-543 antibody and could be detected in both epithelial and stromal cell lines by Western blot analysis (Figure 5B and C). ERα is also present in the two cell types. In HIESC, following treatment with MPA and cAMP to induce decidualization, there is a time-dependent decrease in the expression of PR and to a lesser extent ERα. Treatment of the cells with 17β-E2 did not influence expression of ER or PR in HIEEC (Figure 5C) or HIESC (not shown) and did not stimulate cell proliferation as assessed by 3H-Thymidine incorporation (result not shown).
Expression of steroid receptors in HIESC and HIEEC. The expression of steroid receptors ERα and PR mRNAs was analysed by RT-PCR (A). The corresponding bands were found at the expected molecular weight (MW) for all clones and primary stromal cells (SC). ERα and PR (isoforms a and b) proteins were evaluated by Western analysis. In HIESC-2, all three forms are expressed and down-regulated during the decidualization process induced by medroxyprogesterone acetate (MPA) and cAMP (B). In HIEEC cells, all steroid receptors tested are expressed but not influenced by treatment with 17-β estradiol (E2) (C). In panels B and C, arrows on the left of gel show the MW markers and on the right, the apparent MW of PR isoforms a and b. Data represent analysis of protein extracts from two independent experiments.
Expression of prostanoid receptors in endometrial cell lines
The expression of prostanoid receptor mRNAs was evaluated by RT-PCR. Every cell line, including stromal cells in primary culture used as a reference, expressed EP2, EP3, EP4 and FP whereas EP1 (not shown) was not found (Figure 6). These observations could be reproduced for the transformed cell lines at PD 40 and PD 50 without any change in the pattern of expression.
Expression of prostaglandin receptors in HIESC and HIEEC. Expressions of prostanoid receptors EP2, EP3, EP4 and FP mRNAs were analysed by RT-PCR. The corresponding bands were found at the expected molecular weight (MW) for all clones and primary stromal cells (SC).
Decidualization of HIESC and PRL production
Treatment of confluent HIESC with 8-bromo-cAMP (0.5 mM) in combination with MPA (1 µM) induced a transformation of morphology from spindle-shaped to ovoid cells with abundant cytoplasm characteristic of decidual cells (left, Figure 7A). By comparison, the same treatment did not alter significantly the morphology of HIEEC-22 (right, Figure 7A). PRL levels remained at or below the lower detection limit in epithelial (HIEEC-22) and non-treated stromal (HIESC-2) cell lines. Treatment stimulated PRL to its maximum on day 6 (4.3 ± 1 µg/l) and PRL decreased slowly to reach 2.6 ± 0.6 µg/l on day 10 (Figure 7B) in HIESC-2 but had no effect in HIEEC-22.
Induction of decidualization following treatment with c-AMP and medroxyprogesterone acetate (MPA). Treatment with c-AMP (0.5 mM) and MPA (10–6 M) for 10 days induced a morphological change in HIESC–2 from a spindle to an ovoid shape (A, left) but had no effect in HIEEC-22 (A, right), magnification 100×. Induction of PRL secretion by the same treatment in HIESC-2 and HIEEC-22 is shown in B, at different days of culture (1, 6 and 10). A significant increase was found at days 6 and 10 with maximal levels observed at day 6 (P < 0.05). The data represent the mean ± SEM from three independent experiments run in duplicate.
Correlation between PG E2 and F2α production and expression of biosynthetic enzymes in HIESC and HIEEC
We examined the effect of IL-1β (1 ng/ml) on PGE2 and PGF2α production in HIESC and HIEEC cells. Figure 8 illustrates that PGE2 (A and C) and PGF2α (B and D) levels were highly stimulated by IL-1β in HIESC-2 (A and B) and HIEEC-22 (C and D) and that inhibition of COX-2 by the selective inhibitor NS-398 blocked the stimulation. The levels of PGE2 produced in the presence of IL-1β were always higher (5–10 times) than those of PGF2α. Similar data were obtained for HIESC-16 and HIEEC-18 (results not shown).
Correlation between prostaglandin (PG) production and expression of biosynthetic enzymes. HIESC-2 (A, C, E) and HIEEC-22 (B, D, F) were grown to confluency and treated with IL-1β (1 ng/ml) in the presence or absence of the COX-2 inhibitor NS-398 (1 µM). PGE2 (A, C) and PGF2α (B, D) production is shown for HIESC (A, B) and HIEEC (C, D) at basal (C) and IL-1β stimulated levels in the absence (IL) and presence (IL+NS-398) of NS-398. A–D results are the mean ± SEM of three experiments run in quadruplicate. Western analysis of the expression of the different biosynthetic enzymes is shown for HIESC-2 (E) and HIEEC-22 (F) after treatment in the absence or presence of IL-1β (results are representative of one out of two experiments where lanes 1, 2 and 3, 4 are duplicates). Panels G (HIESC-2) and H (HIEEC-22) represent the quantitation of the autoradiography signal [integrated density value (IDV)] for the Western blot analysis shown in panels E and F and are the mean ± SEM of two experiments run in duplicate.
The expression of the 110 kDa, phosphorylated cytosolic PLA2 (cPLA2) (Kudo and Murakami, 2002), 72 kDa COX-1, 70–72 kDa COX-2, 36 kDa endometrial PGF synthase (AKR1B1) (Madore et al., 2002), 16 kDa microsomal PGE synthase (mPGES1) and 45 kDa β-actin used as internal standard was evaluated by Western analysis. Analysis was performed on whole-cell proteins extracted from confluent cultures of HIESC and HIEEC treated or not with IL-1β. First, as illustrated in Figure 8(E and F), treatment of the cells with IL-1β for 24 h stimulated the expression of all the enzymes studied. The most striking increase was observed in HIESC-2 (E), for all but COX-1 enzymes. Interestingly, COX-1 protein was barely detected in HIESC-2 whereas a good basal expression level, stimulated by IL-1β, was observed in HIEEC-22 (Figure 8F). As illustrated in G and H, all the PG biosynthetic enzymes (PLA2, COX-2, AKR1B1 and mPGES1) were significantly increased by IL-1β treatment in both HIESC and HIEEC. COX-2 expression was not affected by treatment with NS-398 (results not shown).
Effect of DEX on PG production and COX-2 expression in HIESC and HIEEC treated with IL-1β
We evaluated the effect of DEX, a synthetic anti-inflammatory steroid analogous to cortisol, on the accumulation of PGE2 and PGF2α in the presence of IL-1β (Figure 9). We observed a dose-dependent decrease of PGE2 and PGF2α levels in both HIESC-2 (A) and HIEEC-22 (B) in response to DEX. However, the inhibition at 10–6 M was more complete in HIESC-2 (>90%) than in HIEEC-22 (<50%). We also studied the effect at the level of COX-2, the rate-limiting enzyme in PG production. As shown in Figure 9(C and E; D and F), DEX used at 10–7 and 10–6 M decreased significantly COX-2 protein levels in HIESC-2 but not HIEEC-22. We observed an almost complete inhibition of COX-2 in HIESC-2, whereas there was a minimal effect in HIEEC-22. These observations correlate well with the level of inhibition of PG production shown in panels A and B and suggest that DEX may act at both protein expression and activity levels. Moreover, it suggests that epithelial and stromal cells have distinct sites of regulation of PG biosynthesis.
Effects of dexamethasone (DEX) on prostaglandin (PG) production and COX-2 expression. PGE2 and PGF2α production (ng/ml) was measured in the culture medium of HIESC-2 (A) and HIEEC-22 (B) treated with IL-1β (1 ng/ml) in the absence or presence of DEX at the indicated concentrations. The results represent the mean ± SEM of two experiments run in duplicate. Western analysis of COX-2 expression in HIESC-2 (C) and HIEEC-22 (D) treated with IL-1β (1 ng/ml) in the absence or presence of DEX at the indicated concentrations (one representative out of two experiments). E and G represent respectively the relative integrated density value (IDV) for Cox-2 and β-actin from panel C and a replicate experiment (HIESC-2), whereas F and H represent the same for panel F and a replicate experiment. Values are the mean ± SEM of two experiments run in duplicate.
Discussion
PGs are primary regulators of female reproductive function. This has been highlighted by disruption of COX-1 and COX-2 genes in the mouse where it induced parturition and multiple reproductive failures, respectively (Langenbach et al., 1997; Lim et al., 1997; Morham et al., 1997). In animals, endometrial PGs regulate uterine contractility, fertilization, implantation and luteal function (Doualla-Bell et al., 1998; Arosh et al., 2004a,b). In humans, PGs appear to regulate menstruation and abnormal production of PGE2, and PGF2α is associated with menstrual disorders such as menorrhagia and dysmenorrhoea (Sales and Jabbour, 2003a,b). Regulation of PG production with non-steroidal anti-inflammatory drugs (NSAIDs), functional inhibitors of COXs, is commonly used to help relieve the symptoms of menstrual disorders.
The cellular and molecular mechanisms associated with human disorders, especially their dynamic regulation, are best studied in vitro using cell cultures. Primary cultures appear as the ideal in vitro model because of the relative proximity with the corresponding cells in situ. Some studies have used endometrial cell cultures where stromal and epithelial cells were grown together (Arnold et al., 2001), but most studies were done with endometrial stromal cells alone (Huang et al., 1998; Tamura et al., 2002). Indeed, fibroblastic cells are typically more viable and easier to grow in culture, whereas epithelial cells such as those found in endometrial glands exhibit limited viability and barely survive 2 weeks in culture (Matthews et al., 1992; Kyo et al., 2003). Finally, the biological material is limited and difficult to obtain, and primary cultures are plagued by limited viability and inherent variability. Therefore, the aim of this work was to develop stable cell lines derived from normal human endometrium to establish a model to study PG biosynthesis and receptor systems in this tissue.
Some groups have succeeded in expanding human endometrial cells in culture using SV40 large TAG (Rinehart et al., 1991; Merviel et al., 1995) or hTERT (Kyo et al., 2003; Krikun et al., 2004), and the resulting cell lines expressed some characteristics of the original phenotype. One group has developed immortalized human endometrial epithelial and stromal cells by liposome transfection with SV40 TAG, but only a few clones were analysed and PG analysis was not reported (Merviel et al., 1995). Unfortunately, none of the previously described cell lines were characterized in relation with the PG system, neither available commercially. We present here the characteristics of endometrial stromal and epithelial cell lines derived following infection with a defective retrovirus containing the protein SV40 large TAG. The introduction of the viral oncoprotein within the genome of the endometrial cells extends their lifespan until at least PD 60. The SV40 large TAG is known to interact with tumour suppressor protein p53 preventing apoptosis and Rb proteins leading to a proliferative state (Klawitz et al., 2001). SV40 large TAG introduced in the cells may modify normal structure and functional characteristics of the original cells (Ray, 1995).
The karyotype of HIESC-2 is normal at 46XX, and overall morphological and functional properties appear identical to primary stromal cells. The HIEEC-22 epithelial line has a slightly altered karyotype where one X and one 5 chromosome were deleted. Nevertheless, a copy of each of the 23 chromosomes is represented in the karyotype. Although less than ideal, chromosome alteration is common in cell lines and in the present case it did not appear to affect the PG system. In the case of chromosome X, deletion of one copy is expected to exert minor impact because it is reported that one of these is normally inactivated in somatic cells (Chow et al., 2005). This chromosome deletion probably results from the transformation of primary cells with TAG. However, HIEEC-22 cells have maintained a PG system comparable with that of primary cultures especially with respect to their response to IL-1β.
In this study, the stromal and epithelial clones were used between PD 40 and 50 and cells were still expressing the nuclear SV40 TAG (detected by immunofluorescence). Under the phase contrast microscope, stromal cells exhibit a characteristic elongated fibroblast shape, whereas epithelial cells are polygonal. The original phenotypic characteristics of epithelial cells expressing cytokeratin and vimentin and stromal cells expressing vimentin only are also conserved. Interestingly, the pattern of cytokeratin expression with variable immunofluorescence intensity among the HIEEC cells was also observed in endometrial glandular cells immortalized with telomerase (Kyo et al., 2003). We found that the mRNAs of main prostanoid receptors EP2, EP3, EP4 and FP are present in both TAG clones and stromal cells in primary culture. HIESC and HIEEC cells express steroid receptors ERα and PR at both the mRNA and protein levels, but we failed to detect a measurable response in terms of proliferation, steroid receptor expression or PG production following treatment with 17β-E2. We cannot tell at the present time whether ERα is non-functional, partially functional or constitutively activated because we did not conduct a thorough study on every estrogen-responsive gene. Alteration in ER function in the presence of SV40 large TAG was reported to be consequent to interaction with the CREB-binding protein (CBP), an essential component of the ER complex (Ali and Decaprio, 2001; Barbier et al., 2005). Functional PRs are present in stromal HIESC-2 because treatment with cAMP and MPA, a synthetic progesterone, induced decidualization, evidenced by altered morphology and increased PRL secretion. As it was reported elsewhere (Brosens et al., 1999; Tessier et al., 2000), we observed a PR and ERα protein decrease during decidualization of HIESC. The same treatment was ineffective in the epithelial HIEEC-22 line. In vitro decidualization was previously reported following immortalization of endometrial stromal cells with SV40 large TAG, but the PG system was not studied (Brosens et al., 1996).
Our results on the regulation of PG production are particularly interesting. First, it is noticeable that both HIESC and HIEEC are highly responsive to IL-1β as was found by us and others in human endometrial cells in primary culture (Huang et al., 1998; Tamura et al., 2002; Kang et al., 2004). Similarly, the inhibition of PG production with NS-398 and DEX is comparable with what was observed for endometrial stromal cells in primary culture (Huang et al., 1998). Together, these observations suggest that the original phenotype of stromal cells with regard to PG production was maintained in HIESC. Interestingly, the HIEEC clones appear to produce higher levels of both PGE2 and PGF2α especially in the presence of IL-1β. This increased production of PG in epithelial clones is associated with higher basal expression of all biosynthetic enzymes. Also, although NS-398 achieves a similar level of inhibition in HIESC and HIEEC clones, DEX is much more efficient in stromal HIESC (>90%) than in epithelial HIEEC (40–50%). As mentioned earlier, the human endometrial epithelial cells have not been studied extensively, but it has been reported that glucocorticoid receptors are primarily found in the stromal compartment of the human endometrium (Bamberger et al., 2001). This can explain the difference between HIEEC and HIESC following treatment with DEX. The high levels of PGE2 and PGF2α produced by HIESC and HIEEC and its stimulation by IL-1β is comparable with what was observed in primary endometrial cells (Kang et al., 2004). By contrast, established endometrial cell lines such as KLE and Ishikawa, in spite of being oestrogen responsive (Holinka et al., 1986) and expressing high basal levels of COX-2 protein, do not possess a functional PG system comparable with primary endometrial cells. These observations confirm that any cell line should be validated before being used for the study of specific, physiologically relevant systems.
Analysis of the biosynthetic enzymes involved in PGF2α and PGE2 production in HIESC and HIEEC reveals that cPLA2, COX-2 and the two terminal synthases mPGES-1 and AKR1B1 are expressed and increased in response to IL-1β. Also, COX-1 is barely detectable in HIESC cells by comparison with HIEEC cells thus highlighting an additional distinction between the two types of cells. The last observation correlates well with the pattern of expression of COX-1 and COX-2 in epithelial and stromal cells studied by immunohistochemistry in the human endometrium (Stavreus-Evers et al., 2005). In this study, COX-2 protein level was not affected by NS-398, demonstrating that the inhibition of PG production induced by this inhibitor was on enzyme activity rather than transcription or translation of the gene product. By contrast, DEX induced a dose-dependent reduction in COX-2 protein, but only in HIESC stimulated with IL-1β.
In summary, we successfully established human endometrial stromal (HIEEC) and epithelial (HIEEC) cells by infection with the SV40 retrovirus large TAG directly on purified stromal or epithelial cells in primary culture. The cell lines produced maintain many morphological and functional characteristics of endometrial cells and tissue. A major advantage of these cell lines is first to document the properties of endometrial epithelial cells which have only been marginally studied due to limited viability in primary culture (Matthews et al., 1992) and difficulty in establishing an immortalized phenotype (Kyo et al., 2003). In addition, the availability of the two endometrial cell types allowed highlighting functional differences in the pattern of expression of PG biosynthetic enzymes and their inhibition by glucocorticoids. HIESC and HIEEC globally appear as very good models for in vitro study of the PG synthesis and receptor systems, but as with any other cell culture model, expression of the endometrial cell properties to be tested must be checked and data generated utilized with caution until validation in vivo. HIESC and HIEEC cells appear ready for the study of the major transcriptional factors responsible for the expression of PG biosynthetic and catabolic enzymes, the PG receptors and the PGT transporter that we have identified in the human endometrium (Kang et al., 2005). Ultimately, this will contribute to the understanding of the contribution and the management of the PG system in menstrual disorders.
Acknowledgements
This study was supported by a grant from the Canadian Institute of Health Research (CIHR) to MAF. We thank Dr Philippe Y Laberge for provision of endometrial biopsies for the generation of endometrial cell lines and Dr Martin Lemay for histological examination and dating of endometrial samples.
References
Author notes
1Unité de Recherche en Ontogénie et Reproduction and 2Unité de Recherche en Génétique Humaine, Centre de Recherche du CHUL, CHUQ, Faculté de Médecine, Université Laval, Sainte-Foy, Québec, Canada
