Circulating levels and subcutaneous adipose tissue gene expression of pigment epithelium-derived factor in polycystic ovary syndrome and normal women: a case control study

This case–control study was carried out with women of reproductive age seen at the Gynecological Endocrinology Unit at Hospital de Clínicas de Porto Alegre, Brazil. Women with a body mass index (BMI) ranging from 18.5 to 39.9 kg/m² were selected for the study. Twenty-two hirsute oligo-amenorrheic women (<9 cycles/year), presenting with increased levels of serum testosterone (>0.8 ng/mL) and/or free androgen index (FAI) (>6.1) and/or polycystic ovaries in the absence of other disorders causing hirsutism [21, 22], were included. A control group was set up with 14 non-hirsute ovulatory women (regular cycles and luteal phase progesterone levels higher than 3.8 ng/mL) with normal androgen levels and normal ovaries on ultrasound, recruited through public advertisement at the same Gynecological Endocrinology Unit. None of the women from either group had received any drugs known to interfere with hormone levels, blood pressure, or metabolic variables for at least 3 months before the study. Women with diabetes, liver or kidney disease, thyroid dysfunction or pregnancy were excluded.

Twenty-two additional volunteers without laparoscopic evidence of pelvic disease were recruited for collection of subcutaneous and omental adipose tissue samples. This was done to ensure that PEDF gene expression could be reliably measured in scAT. The study protocol was approved by the Research and PostGraduate Group of Hospital de Clinicas de Porto Alegre (GPPG/HCPA) and the local Ethics Committee, and written informed consent was obtained from all participants.

Medical interview and physical examination were performed as previously described [23, 24]. Hirsutism was defined as a modified Ferriman-Gallwey score of 8 or more [25]. Blood pressure (BP) was measured in the sitting position after a 10-minute rest [26]. Anthropometric measurements included body weight, height, body mass index (BMI), waist circumference and waist/hip ratio (WHR) [23, 27, 28]. Hormonal and metabolic assessment as well as abdominal/transvaginal ultrasound were performed in all PCOS and control participants between days 2 and 10 of the menstrual cycle or on any day if the patient was amenorrheic. Polycystic ovary appearance (PCO) was defined as previously reported [5, 29]. After an overnight 12-hour fast, blood samples were drawn from an antecubital vein between 8 a.m. and 10 a.m. for determination of serum PEDF and lipid profile at baseline, and glucose and insulin before and 2 hours after ingestion of 75 g oral anhydrous glucose (oral glucose tolerance test). Blood samples were also assessed for measurement of luteinizing hormone (LH), sex hormone–binding globulin (SHBG) and total testosterone (TT). Free androgen index was estimated by dividing TT (nmol/L) by SHBG (nmol/L) × 100. Homeostasis model assessment index to estimate insulin resistance (HOMA-IR) was calculated by multiplying insulin (mIU/mL) by glucose (mmol/L) and dividing this product by 22.5 [30]. Low-density lipoprotein (LDL) cholesterol was estimated indirectly with the Friedewald formula [31].

Total cholesterol, high-density lipoprotein (HDL) cholesterol, triglycerides, and glucose were determined by colorimetric-enzymatic methods (Siemens Advia System, Deerfield, IL, USA). Enzyme-linked immunosorbent assay (Chemicon International, Temecula, CA, USA) was used to measure total serum PEDF, with a sensitivity (S) of 0.9 ng/mL, intra-assay coefficient of variation (CV) of <5.3% and interassay CV <16%. As recommended by the manufacturer, to prevent PEDF from associating with other circulating proteins that may interfere with quantification of total serum concentration of this adipokine, samples were pre-treated with urea (8 M final concentration) for 60 minutes in ice and diluted 450 times in dilution buffer before being applied in duplicate to ELISA plates. Serum insulin (S = 0.2 mIU/mL), LH (S = 0.1 mIU/mL) and SHBG (S = 0.35 nmol/L) levels were measured with electrochemiluminescent immunoassays (Roche Diagnostics, Mannhein, Germany), with intra-assay CV <3% and interassay CV <5%. Total serum TT levels were measured with radioimmunoassay (Diagnostics Systems Laboratories Inc., Webster, TX), with S <0.1 ng/mL and CV <9.6%.

scAT biopsy was performed in all participants (22 PCOS and 14 control women) to obtain a 250 mg adipose tissue sample from the periumbilical region. To assess scAT and omental adipose tissue (omAT), another 22 normal women were recruited and samples were obtained through laparoscopy. Procedures were performed between days 2 and 10 of the menstrual cycle, or on any day if the patient was amenorrheic, by surgeons from the Plastic Surgery Service at Hospital de Clínicas de Porto Alegre or at the Human Reproductive Unit of Hospital Fêmina. Adipose tissue fragments were immediately frozen in liquid nitrogen and stored at −80°C until total RNA isolation.

Adipose tissue total RNA extraction was carried out in phenol-guanidine isothiocyanate (Trizol^®, Invitrogen™ Life Technologies, Foster City, CA, USA) as previously described [32, 33]. Concentration and quality of total RNA were assessed using a GeneQuant spectro-photometer (Pharmacia Biotech, Cambridge, England).

Reverse transcription of 1 μg of total RNA into cDNA was carried out using the Superscript II First-Strand Synthesis System for RT-PCR (Invitrogen™ Life Technologies, Foster City, CA, USA), according to manufacturer instructions, in a PCT-100™ Programmable Thermal Controller (MJ Research Inc., Watertown, MA, USA). Real-time PCR was performed in triplicate in a 7500 Fast Real-Time PCR System thermal cycler with 7500 Fast System Sequence Detection 1.4 Software (Applied Biosystems, Foster City, CA, USA). SYBR^® Green dye fluorescence was detected by real-time monitoring experiments [34‐36]. Primers were designed using the Primer Express 3.0 Software for Real-Time PCR (Applied Biosystems, Foster City, CA, USA), and acquired from Invitrogen™ Life Technologies (Foster City, CA, USA). Primer sequences were designed to target two exons of mRNA with respect to known splice variants and single-nucleotide polymorphism positions. The forward and reverse primer sequences designed for pigment epithelium-derived factor (NM_002615.5) were (5′ to 3′) CATCATTCACCGGGCTCTCTAC and GGCCTGGTCCCATATGACTTTT, respectively. These primers anneal between residues 463 to 484 (forward) and 654 to 633 (reverse), producing a 192-bp amplicon. mRNA quantitation was normalized to glyceraldehyde-3-phosphate dehydrogenase (NM_002046.3). The forward and reverse GAPDH primer sequences, ACCCACTCCTCCACCTTTG and CTCTTGTGCTCTTGCTGGG (5′ to 3′), respectively, anneal between residues 970 to 988 and 1,147 to 1,129, resulting in an amplicon of 178 bp. Complementary DNA samples (0.87 ng/μL) were mixed with a predetermined forward and reverse primer volume (0.5 and 0.7 μL for PEDF; 0.9 and 0.9 μL for GAPDH) and 12.5 μL of 2X Fast SYBR^® Green Master Mix (Applied Biosystems, Foster City, CA, USA) for a total volume of 25 μL. Protocol conditions consisted of denaturation at 94°C for 2 min followed by 50 cycles (30 sec, 94°C and 30 sec, 60°C). Amplicons produced single sharp peaks during melting curve analysis.

Data were analyzed by relative quantitation using the comparative C_T method [37]. Validation assays were performed by amplification of the target and reference genes, separately, using serial dilutions of an mRNA sample. Both target and reference mRNAs presented equal efficiencies of amplification. The ΔΔC_T method calculates changes in gene expression as relative fold difference between an experimental and calibrator sample, correcting for nonideal amplification efficiencies [38].

Data were described as mean ± standard deviation (SD) or median [interquartile range]. The Wilcoxon two-related-samples test or the unpaired two-tailed Student’s t-test was used to compare group means for data with Gaussian distribution. The Mann–Whitney U test was used to compare group medians for data with non-Gaussian distribution. Pearson’s or Spearman’s rank correlation coefficients were calculated between variables using a two-tailed test for significance. A forward stepwise multiple regression model was also calculated using serum PEDF as a dependent variable, and fasting insulin and TT as independent variables. Log₁₀ transformation was used to normalize the distribution of non-Gaussian variables, and these variables were back-transformed into their original units for presentation. Data were considered statistically significant at p < 0.05. The Statistical Package for the Social Sciences v. 18 (SPSS, Chicago, IL) was used for all statistical analyses.