The control group consisted of women with proven fertility, all parous (parity 1–5) and with confirmed ovulation in the tested cycles. These patients were recruited from general gynaecology clinics, having presented for sterilisation with no confirmed diagnosis of PCOS. The infertile group consisted of women diagnosed with PCOS. Endometrial biopsies were obtained from women in a natural menstrual cycle whose phase was confirmed by ultrasound, hormonal and histological criteria. A urinary LH test was also used to confirm ovulation, and in post-ovulatory cycles, the sample was taken on day 6–8 post LH surge. All patients were receiving no exogenous hormonal treatment for at least 2 months prior to the procedure. PCOS was defined using the Rotterdam criteria by the presence of two or more features of clinical and biochemical hyperandrogenism, oligo-anovulation and/or polycystic ovaries (ultrasound) . Related disorders such as tumours producing androgens or patients with Cushing’s syndrome were excluded.
Ovulatory PCOS patients had confirmed polycystic ovaries on ultrasound and hyperandrogenism; ovulated spontaneously, with serum progesterone levels measured at LH+7 at least 30 nm/L; and were infertile despite regular ovulatory cycles in the presence of patent tubes and normal sperm parameters. Tubal patency was confirmed by either a HyCoSy scan or laparoscopy and dye test. The patients in control and study groups were matched for body mass index and smoking habits.
Samples for histological evaluation and immunohistochemical studies were taken with a sterile Pipelle endometrial suction for sampling of the functional layer of the endometrium . Blood samples were collected from patients at mid proliferative (days 5 to 8) for routine fertility assessment of the hormonal panel and at secretory phase (day 21, LH+7) for progesterone, CD44, OPN and inflammatory cytokine levels. Biopsies were taken at proliferative and secretory phases timed with the blood sampling. Proliferative phase circulating androgen levels were also assessed in the fertile and PCOS patients. Patients with diagnosed coincident uterine pathology that could modify the endometrial structure and function (as hyperplasia, endometrial polyp or endometritis) were excluded from the study. Ethical approval was obtained from the South Wales Local Research Ethics Committee (Wales 6 reference 05/WMW02/103 and 12/WA/0289); written consent was obtained from all patients at the time of recruitment. The study duration was 2 years from patient recruitment to data collection and analysis.
Assessment of hormone levels in serum
Electrochemiluminescence immunoassays (ECLIA, Roche) were used to measure the serum levels of testosterone (Elecsys® Testosterone II assay), sex hormone–binding globulin (SHBG, Elecsys® SHBG assay), dehydroepiandrosterone sulfate (DHEAS Elecsys® assay), progesterone (Elecsys® Progesterone II), LH (Elecsys® LH) and FSH (Elecsys® FSH) using an Elecsys 2010 immunoassay analyser (Roche). Androstenedione was assayed using RIA (Beckman Coulter). Samples with total testosterone levels over 1.8 nmol/l were also analysed by LC-MS to confirm values, as per NHS Wales policy; these values are reported in Table 1.
Samples were fixed in 10% buffered formaldehyde for 24 h and embedded in paraffin wax, and 3–4-μm-thick sections prepared on positively charged slides for immunohistochemical studies. The sections were de-waxed using a Roche dewaxing solution. The tissue sections were incubated with rabbit anti-human OPN polyclonal antibody (AB1870; Millipore) and rabbit anti-CD44 monoclonal antibody (clone EPR1013Y; Millipore), both diluted 1:100. Rabbit IgG was used as negative control, and adjacent sections were cut and stained in parallel using identical procedures. Renal cell cancer and tonsil positive control were used for OPN and CD44 respectively.
For antigen retrieval, the slides were incubated in CC1 buffer (Ventana Biotek Solutions, Tucson, AZ) for an hour on heated plates at 100 °C on a Benchmark XT processor. Primary antibody incubation was for 36 min at dilution 1:100 at 37 °C. Positive immunostaining was detected through interaction of avidin-biotin peroxidase (ABC) complex with biotin conjugated secondary antibody using a Ventana I View DAB detection kit (Ventana Biotek Solutions, Tucson, AZ) and avidin-biotin blocker. The slides were subsequently counterstained with haematoxylin, dehydrated, cleared and mounted in DPX mountant to be examined under light microscopy. We used an immunohistochemical scoring system (IHC) in which the observers perform a thorough examination of all the immunohistochemical sections of the tissue slide using a multi-headed microscope . The endometrial epithelium was assessed separately for the lumen and glands and scored for intensity and distribution of staining. The intensity of staining was scored from (0)—absent to (4)—strong. The distribution of staining was assessed as follows: (0)—absent, (1)—less than 30%, (2)—30 to 60%, (3)—more than 60% and (4)—100% of the tissue surface stained (H-score). The observers were blinded to the patients’ diagnosis, demographics and timing in the cycle of endometrial biopsy.
Regulation of OPN and CD44 was assessed in the Ishikawa endometrial epithelial cell line (ECACC 99040201, STR authentication, Public Health England, UK), a well-differentiated human endometrial adenocarcinoma cell line, expressing both ERα and PR A & B receptors, regulated in a manner similar to that of normal endometrium . Ishikawa cell lines at low passage number (passage number ≤ 22) were cultured at 37 °C with 5% CO2; at least 24 h prior to experiments, the medium was changed to phenol red-free media with 10% charcoal stripped FCS. Cells were stimulated with pro-inflammatory cytokines TNF-α (25 ng/ml, 4 h) (Miltenyi Biotec UK, Cat no. 130-094-014) and IFN-γ (2 IU, 24 h) (Miltenyi Biotec UK, Cat no. 130-096-872) separately, or steroid hormones 17-β-oestradiol (E2; 10 nM, 48 h) (Sigma-Aldrich, UK Cat-No E8875) and progesterone (P4; 1 μM, 48 h) (Sigma-Aldrich UK, Cat-No M1629) alone or in combination as previously described [4, 24]. Following stimulation, cell pellets were collected for mRNA and chromatin immunoprecipitation analysis. The presence of ERα, PRA and PRB receptors in Ishikawa cells was confirmed by immunoblots (data not shown).
Proteins were quantified (Bradford assay) and equal amounts (20 μg) resolved by SDS-PAGE, transferred to PVDF membranes and blocked overnight with 10% BSA, in 0.1% Tween-20-TBS (TTBS). Membranes were subsequently incubated at 4 °C with rabbit anti-human ERα rabbit polyclonal antibody (HC-20 Santa Cruz Biotechnology, USA) (diluted 1/500 in 5% BSA-TTBS buffer) or anti-human PR A/B rabbit polyclonal antibody (H-190 Santa Cruz Biotechnology, USA) (diluted 1/1000 in 5% BSA-TTBS buffer). Blots were then incubated for 1 h with IgG horseradish peroxidase secondary antibody diluted 1/2000 in 5% BSA-TTBS buffer. Between incubation steps, membranes were washed several times with TTBS. Blots were analysed for GAPDH levels (GAPDH rabbit polyclonal antibody (FL-335, Santa Cruz Biotechnology, USA)) to normalise protein loading in each well. Immunoreactive bands were visualised using a ChemiDoc System Bio-Rad Imager (Bio-Rad) and quantified by Quantity One® Imaging software (Bio-Rad), as described previously .
RNA isolation and qPCR
Endometrial biopsies were snap-frozen in liquid nitrogen immediately following excision during the surgical procedure and further stored in liquid nitrogen until subsequent analyses. Total RNA was isolated from snap frozen biopsies and cell line using a RNeasy mini Kit (Qiagen, UK). DNase-I-treated RNA was reverse transcribed into cDNA (high-capacity cDNA conversion; Applied Biosciences, UK) before assessing CD44 and OPN expression using specific primer pairs available on request (Beacon Design 2.0; Premier Biosoft International, USA). qPCR amplification was performed in triplicate in 96-well plates in a Bio-Rad IQ iCycler. Serial dilutions of cDNA were used to plot a calibration curve, and gene expression was quantified by plotting threshold cycle values. Expression levels were normalised to values obtained for the reference gene Ribosomal Protein 60S L 19 (RPL-19) .
Enzyme-linked immunosorbent assay
Serum samples (50 μl) were collected following centrifugation of blood samples at 1500 g for 10 min and stored at −20 °C. Biopsy media was collected, centrifuged and stored as described above prior to enzyme-linked immunosorbent assay (ELISA) tests. The levels of CD44 (CD44 Human ELISA Kit, Abcam, UK), OPN (Human OPN DuoSet ELISA R&D Systems, UK), TNF-α (Human TNF-alpha DuoSet ELISA, R&D Systems, UK) and IFN-γ (Human IFN-gamma DuoSet ELISA, R&D Systems, UK) were measured in serum and in biopsy media by ELISA following the manufacturer’s protocol. Samples were analysed in triplicate and data obtained for in vitro experiments is representative of four independent experiments.
Chromatin immunoprecipitation analysis
Ishikawa cell sample fixation, DNA shearing and chromatin immunoprecipitation (ChIP) were performed following the manufacturer’s instructions (Porvair Sciences, UK). Chromatin was quantified using a Nanodrop spectrophotometer and visualised using agarose gel electrophoresis to ensure correct distribution of fragment sizes prior to immunoprecipitation. Four hundred nanograms of each chromatin sample was used per ChIP with 0.8 μg of the relevant antibody; NF-κB anti-p65 (C-20, Santa Cruz Biotechnology, USA), STAT1 (phospho Y701, ab30645, AbCam, UK), anti-ERα (HC-20, Santa Cruz Biotechnology, USA) and non-specific rabbit IgG (Rockland Inc., USA) antibodies were used. Enriched fragments were reverse cross-linked, and protein removed by protease digestion prior to qPCR. Specific primer sets (available on request) for predicted NF-κB, STAT1 and ERα binding sites in the OPN and CD44 promoters were used to assess promoter occupancy by qPCR, which was performed as described .
Data distributions were assessed for normality using the Kolmogorov-Smirnov tests. Non-normally distributed data were analysed with the Mann-Whitney U test applied post hoc to determine statistical significance. For normally distributed data, an ANOVA test followed by a t test was used to determine significant differences between groups. The test statistic and corresponding P value were reported. Correlations between CD44 and OPN immunostaining and OPN and CD44 serum levels in each group were assessed by the nonparametric Spearman test. Correlations between TNF and IFN serum levels with OPN and CD44 levels in media and serum were performed via Pearson correlation coefficient. All data analysis was performed using SPSS version 16.0 (SPSS, Chicago, IL).