Polycystic ovary syndrome: possible involvement of androgen-induced, chemerin-mediated ovarian recruitment of monocytes/macrophages^†

Animals

Female Sprague Dawley rats (Charles River, Montreal, Canada) were maintained on 12 h cycle (light and dark) and given food and water ad libitum. All procedures were carried out in accordance with the Guidelines for the Care and Use of Laboratory Animals, Canadian Council on Animal Care, and were approved by University of Ottawa Animal Care Committee.

In vivo DHT treatment

Female rats at 21 days of age were implanted subcutaneously with silicone capsules without (sham control) or with DHT (Steraloids Inc., Newport, USA), as previously described [27–29] and continuously releasing (83 μg DHT/day) for 3, 7, 15, and 28 days. Animals were weighed prior to the DHT implant and at the termination of experiments. Sham control rats were sacrificed at the diestrous stage of the estrous cycle.

Serum and ovary collection in DHT-treated rats

The rats were anesthetized by isoflurane (3%–4%; Baxter Corporation, Mississauga, Canada), and blood was collected by heart puncture. Serum was separated by centrifugation (2060 g, 5 min, 4°C) and stored at –80°C pending analysis. Whole ovaries were fixed in 4% paraformaldehyde (PFA; Sigma Aldrich, Oakville, Canada) for 24 h at 4°C and embedded in paraffin. Ninety serial sections (5 μm/section; 450 μm of tissue) were obtained per ovary. Histological sections were used to quantify ovarian follicles (H&E staining), to determine in situ apoptosis (TUNEL assay) and in the quantification of ovarian macrophages (immunofluorescence). At least three animals and nine sections per animal were used in each experiment, and a range of 5–20 ovarian structures were analyzed per section (45–120 ovarian structures per rat).

Human samples

All patient samples were selected according to the Rotterdam criteria by which two of the three phenotypes were met: (1) oligo- or anovulation, (2) clinical and/or biochemical signs of hyperandrogenism, or (3) polycystic ovaries. Exclusion criteria include the presence of congenital adrenal hyperplasia, Cushing's syndrome, androgen-secreting tumors, 21-hydroxylase deficient nonclassic adrenal hyperplasia, hypogonadotropic hypogonadism, premature ovarian failure, or syndromes of severe insulin resistance [2]. Ovarian sections (5 μm) from 9 non-PCOS and 16 PCOS subjects were obtained from the Department of Pathology, Queen Mary Hospital, University of Hong Kong. The use of archived paraffin-embedded tissue blocks for research purpose was approved by the Institutional Review Board of the University of Hong Kong/Hospital Authority Hong Kong West Cluster (HKU/HA HKW IRB UW UW 14–469). The requirement for individual consent was waived since residual formalin fixed paraffin-embedded samples after diagnosis was used.

Serum and follicular fluid from 55 non-PCOS and 25 PCOS subjects were obtained at Department of Obstetrics and Gynaecology, Taipei Medical University Hospital, Taipei Medical University, Taipei, Taiwan in accordance with Institutional Review Board of Taipei Medical University, Taipei (#: TMU-JIRB 201410033). All subjects (non-PCOS and PCOS) had BMI ≤ 30 and they were age-matched (non-PCOS: 35.47 ± 0.47 vs PCOS: 34.79 ± 0.67). Hormonal information for the non-PCOS and PCOS subjects includes serum levels of estradiol-17β (3051 ± 269.1 pg/ml vs 3589 ± 546.5 pg/ml), progesterone (0.98 ± 0.11 ng/ml vs 0.95 ± 0.11 ng/ml), and anti-Mullerian hormone (4.11 ± 0.34 vs 9.65 ± 1.08 ng/ml; P < 0.001). Testosterone-related measurements were also taken and were significantly different between non-PCOS and PCOS subjects: total testosterone (95.92 ± 5.35 ng/dL vs 127.40 ± 8.97 ng/dL; P < 0.01), free testosterone (0.46 ± 0.03 ng/dL vs 0.73 ± 0.10 ng/dL; P < 0.05), bioavailable testosterone (10.93 ± 0.75 ng/dL vs 17.49 ± 2.63 ng/dL; P < 0.05), and free androgen index (FAI; 1.87 ± 0.14 vs 3.04 ± 0.53 ng/dL; P < 0.05), where FAI = total testosterone/sex hormone binding globulin × 100.

In situ quantification of ovarian M1 and M2 macrophages in rats and humans

Ovarian sections were deparaffinized and hydrated. Antigen retrieval in rat sections was performed using 0.1% trypsin (Sigma-Aldrich, Oakville, Canada) in 0.1% of CaCl₂ (Sigma-Aldrich) solution pH 7.8 (30 min, 37°C). Sections were washed in phosphate-buffered saline (PBS; Sigma-Aldrich) with 0.05% tween-20 (Sigma-Aldrich) (PBS-T), and nonspecific bindings were blocked with nonfat dry milk 5% (30 min). The slides were incubated (overnight, 4°C) with mouse anti-rat CD163 or mouse IgG1 (isotype control had the same concentration as the antibody) followed by anti-mouse conjugated with alexa-fluor 594 [1 h, room temperature (RT)]. Sections were incubated with biotinylated anti-rat CD68 or mouse IgG1 as negative control (2 h, RT; isotype control had the same concentrations as antibody) followed by FITC-conjugated streptoavidin (1 h, RT).

Antigen retrieval in human ovarian sections was achieved with citrate buffer (Sodium citrate-Fisher, Nepean, Canada) (10 mM; pH 6.0, 95°C, 20 min). Nonspecific binding was blocked with nonfat dry milk (5%, 30 min). M1 and M2 macrophage phenotypes were confirmed by the expression of two receptors: CD86 and MHC II for M1 macrophages and CD206 and CD163 for M2 macrophages (Supplemental Figure S1). Quantitative analyses were conducted using anti-MHCII and anti-CD163 antibodies. Mouse anti-human MHC II and rabbit anti-human CD163, and their respective isotype controls (matched concentration with antibodies) were incubated overnight at 4°C. Slides were washed with PBS-T and incubated with anti-mouse conjugated with alexa-fluor 594 and anti-rabbit conjugated with alexa-fluor 488 (1 h, RT).

Rat and human sections were washed and mounted with ProLong Gold Antifade Mountant with DAPI (Thermo Fisher Scientific, Waltham, USA). Ovarian sections were observed by fluorescence microcopy (Zeiss Axioplan 2, Zeiss, North York, Canada), and images were recorded and analyzed using the Axion Vision program (Axion Vision software, Zeiss). In rats, all ovarian follicles with an oocyte in a section were considered. In human biopsies, 4 to 50 images of the stroma were taken per section (number of images was dependent on the size of the biopsy). Antibodies used in this study are described in Supplemental Table S1.

TUNEL assay

In situ localization of apoptosis was achieved using the TUNEL assay, according to manufacturer's instructions (Sigma-Aldrich; cat# 11767291910).

Flow cytometric analysis of ovarian, splenic, and peritoneal macrophages in rats

Ovarian and splenic cell suspensions were obtained by purging the tissue through a 70-μm cell strainer (Fisher Scientific, Ottawa, Canada). Cell suspension was washed with PBS (320 g, 5 min, 4°C) and leukocytes were isolated by percoll gradient centrifugation (40% and 80%; Sigma-Aldrich, St Louis, USA; 350 g, 25 min, RT), as reported by Liu et al [30]. To prevent blood leukocytes contamination, abdominal artery was perfused with PBS prior to the tissue harvest. Total peritoneal leukocytes were obtained by peritoneal lavage (PBS/EDTA 50 mM; RT). Total peritoneal cells were filtered through a 70-μm cell strainer, washed with PBS, and re-suspended in 1–2 ml of RBC lysis buffer (2 min, RT; eBioscience Thermo Fisher) and immediately washed with PBS.

Ovarian, splenic, and peritoneal leukocytes (10⁵ cells) were stained with viability dye eFluor 450 in PBS (Biolegend, San Diego, USA; 30 min, 4°C) and washed with FACS buffer (PBS with 1% BSA, 0.1% NaN₃ and 2% FBS; Gibco- Thermo Fisher). Cell surface staining (anti-CD163 and anti-CMKLR1) was performed (45 min, 4°C) followed by the intracellular staining (anti-CD68) using leucoperm kit (Bio-Rad Laboratories, Inc, Mississauga, Canada) as per the manufacturer's instructions. Cells were washed and re-suspended in 1% of PFA. Flow cytometry acquisition was performed using the Cyan ADP 9 analyzer (Beckman Coulter Inc., Mississauga, Canada) and the data were analyzed using Kaluza software (Beckman Coulter Inc.). Antibodies and isotypes used in this study are described in Supplemental Table S1. Three experimental replicates were performed (6 rats per replicate; total of 18 rats per group).

Flow cytometric analysis of blood monocytes in rats

Blood was collected through heart puncture, and K2EDTA-coated blood collection tubes (BD Vacutainer Blood collection, Mississauga, Canada) were used. Blood mononuclear cells were separated using Histopaque- 1077 (Sigma-Aldrich), washed (PBS- 320 g, 10 min, 4°C), and stained with viability dye eFluor 450 (30 min, 4°C). The cells were then stained with mouse anti-rat CD3, mouse anti-rat CD4, mouse anti-rat CD11b, mouse anti-rat CD43, and rabbit anti-rat CMKLR1, and their respective isotype controls (45 min, 4°C; Supplemental Table S1), washed, and resuspended in 1% of PFA for flow cytometry analysis.

Assessment of ovarian and serum levels of chemerin by western blotting in rats and humans

Ovaries (10 rats per group) were homogenized using tissue extraction reagent I (Thermo Fisher) supplemented with the protease inhibitors cocktail mini cOmplete (Roche, Mississauga, Canada) and centrifuged (12,000 g; 30 min; 4°C). Ovarian (20 μg of protein) and serum (5 μl) proteins were separated electrophoretically by SDS-PAGE and electrotransferred onto a nitrocellulose membrane. Nonspecific binding was blocked with nonfat dry milk (5%; 1 h, RT). Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was assayed as a loading control in ovarian lysates. The membranes were washed and incubated with goat anti-mouse chemerin (1:1000) overnight at 4°C, and with rabbit anti-goat (1 h, RT). Signal intensity was assessed using Pierce ECL Western Blotting Substrate kit (Thermo Fisher) and Molecular Imager VersaDoc MP imaging system (Bio-Rad Laboratories, Inc). Blots were densitometrically analyzed using the ImageJ software. Chemerin levels in human follicular fluid and serum (5 μl) from 20 non-PCOS and 16 PCOS subjects were similarly assessed, except by the use of goat anti-human chemerin (1:1000).

Single cultures and co-cultures of granulosa cells and macrophages in rats

Peritoneal leukocytes were collected from six noncycling female rats (35 day old; 6 rats per replicate). One million cells/well (12-well plate) were plated for 4 h in 10% FBS M199 medium (Gibco Thermo Fisher, Burlington, Canada). Nonadherent immune cells were excluded by washing with fresh M199 medium. Adherent macrophages were co-cultured with granulosa cells. Of the 1 × 10⁶ leukocytes plated, 0.4 × 10⁶ cells remained attached on the plate after washes, 93% of which were macrophages [M1 (39%), M2 (54%)]) (Supplemental Figure S2A).

Granulosa cells were isolated from eCG (Sigma, St Louis)-treated female rats (21 day old; 10 rats per replicate) 48 h prior to sacrifice, as previously described [31]. 0.6 × 10⁶ (single culture) or 0.2 × 10⁶ granulosa cells (co-cultured with macrophages) were cultured in 12-well plates. Granulosa cells were stained with live cell tracker (CellTracker Orange CMRA Thermo Fisher; 2.5 μM; 30 min) prior to the co-culture, as per the manufacturer’s recommendation, washed, and laid onto macrophages (four macrophages: two granulosa cells). Cells were cultured in 10% FBS M199 (24 h) following by FBS-free M199 (16 h) prior to DHT treatment (1 μM in FBS-free M199; 12 and 24 h). Macrophage contamination in granulosa cell preparations was <0.37% (Supplemental Figure S2B).

Early and late apoptosis in rat granulosa cells: annexin V and 7AAD staining

Granulosa cells from single culture or co-culture were stained with annexin V and 7-AAD, according to the manufacturer's instructions (Biolegend cat# 640911). A total of 10⁵ cells were washed in cell staining buffer, re-suspended in annexin V binding buffer, and incubated (15 min, RT, in dark) with annexin V conjugated with alexa Fluor 647 and 7-AAD.

Blood mononuclear chemotaxia assay

Transwell permeable supports (polycarbonate membrane pore size of 3 μm; Corning, New York, USA) were used to assess blood mononuclear cell migration toward chemerin gradient. Enriched blood mononuclear cells (Histopaque; 1077 separation) were resuspended in 10% FBS RPMI and added to the upper chamber (10⁵ cells in 100 μl) while 600 μl of chemerin-containing medium (100 ng/ml) was placed in the lower chamber. Cells were cultured (37°C, 5% CO₂) for 6, 12, and 24 h. Migrated cells were immediately fixed with 4% PFA (30 min, RT). Nonmigrating cells (top of the membrane) were removed using cotton swabs, while migrating cells (bottom of the membrane) were stained (10 min, RT) with Hoechst 33342 dye (1:1000; Thermo Fisher). The membranes were washed in PBS-T and placed on glass slides for microscopic analysis. Using ×40 magnification, 20–25 photos were taken in random areas of each membrane (n = 6 replicates; 4 animals per replicate). Hoechst-positive cells (nucleus DNA) were quantified and the results expressed as mean number of cells per 0.3795 mm².

Statistical analysis

Parametric unpaired t-test was used to determine significant differences between two experimental groups in rats (control and DHT) in regard to (1) ovarian and body weight; (2) ovarian and serum chemerin in rats; (3) ovarian, splenic and peritoneal macrophages, and blood monocytes expressing CMKLR1 (flow cytometry). One-way ANOVA was used to compare cell migration in response to chemerin (three time points). Two-way ANOVA was used to compare (1) ovarian follicle quantification (time and DHT treatment), (2) in situ quantification of ovarian M1 and M2 macrophages (ovarian follicle stage and DHT treatment), (3) the in situ TUNEL assay in the ovarian sections (ovarian follicle stage and DHT treatment), and (4) the in vitro apoptosis rate in rats (annexin 5 and 7-AAD assay; time and DHT treatment). In humans, the continuous variables were tested for normality using the Kolmogorov–Smirnov test and formed the basis for the choice of nonparametric test. The parametric unpaired t-test was used to compare follicular fluid and serum chemerin (normally distributed), while the nonparametric Mann–Whitney U-test was used to compare the human stromal macrophages (M1 and M2 quantification and CMKLR1+ M1 macrophages by immunofluorescence) in non-PCO and PCO ovaries (data not normally distributed). Data are expressed as mean ± SEM, unless otherwise stated. Significant differences were considered at P ≤ 0.05.

The ovarian changes following chronic DHT treatment in vivo

The ovaries were primarily affected by DHT during early phase of treatment. Ovarian weight was reduced by DHT after 7 days of treatment (Figure 1A; P < 0.001 and P < 0.001, for 7 and 15 days respectively), while changes in body weight were only observed after 28 days (P < 0.001; Figure 1B).

Folliculogenesis was affected with 15 days after DHT treatment (Figure 1C and D). The number of early antral follicles (Figure 1C; P < 0.05) and unhealthy follicle (P < 0.001) was significantly elevated in the DHT-treated ovaries compared to controls. Unhealthy follicles were morphologically distinctive from the atretic ones (also present in control ovaries) presenting reduced layer of granulosa cells with infiltrated macrophages but intact theca cell layer. A representative healthy antral follicle and an unhealthy follicle are shown in Figure 1E (i–ii). The frequency of pre-antral, antral, or pre-ovulatory follicles did not differ between control and DHT-treated ovaries after 15 days (Figure 1C).

Ovarian follicle stage-specific M1 and M2 macrophages in DHT-treated rats and human PCOS ovaries

Ovarian M1 and M2 macrophages were assessed in the ovaries from androgenized rats and in humans (non-PCOS or PCOS subjects). In rats, M1 (CD68+CD163–) and M2 (CD68+CD163+) macrophages were quantified in a follicle stage-specific manner (Figure 2Ai–v), including theca, granulosa cell layers, and the antrum of individual ovarian follicles following 3, 7, and 15 days of DHT treatment. The number of M1 and M2 macrophages present was dependent on the stage of ovarian follicular development, as well as was regulated by DHT (Table 1). Three days of DHT treatment failed to influence ovarian M1 macrophages frequency in any of the follicle stages (Table 1), while significantly reduced M2 macrophages in pre-ovulatory follicles (P < 0.05; Table 1). Seven and 15 days of DHT treatment significantly increased M1 macrophages in antral follicles (P < 0.05). While 7 days of DHT significant increased the number of M1 macrophages in pre-ovulatory follicles (P < 0.001; Table 1), rats treated for 15 days with DHT had the number of pre-ovulatory follicles markedly reduced preventing the statistical comparison of macrophage phenotypes in these follicles. The frequency of M2 macrophages in antral and pre-ovulatory follicles from 7 and 15 days DHT-treated ovaries was slightly but not significantly reduced (Table 1). Unhealthy follicles were predominant in DHT-treated ovaries and rich in M1 macrophages. A single structure resembling unhealthy follicle was found in control ovaries, which prevented comparisons between control and DHT-treated ovaries. The most prominent change in M1: M2 macrophage ratio was evident in pre-ovulatory follicles at 7 days after DHT treatment (Figure 2B), as pre-ovulatory follicles were not longer observed in the ovaries from rats treated for 15 days.

In humans, M1 (MHC II+CD86+ cells) and M2 (CD163+CD206+ cells) macrophages were identified (liver as positive control, Supplemental Figure S1A–C) and quantified in the stroma of ovaries from PCOS and non-PCOS subjects (Figure 2Bi–ii). A lower number of M2 (P < 0.001; Figure 2D) but not M1 (Figure 2C) macrophages were observed in the stroma from PCOS ovaries compared to control. The M1: M2 ratio was higher in PCOS subjects than controls (Figure 2E; P < 0.01), suggesting an altered immunological balance in the ovary.

Granulosa cells apoptosis in mature ovarian follicles under androgenic stimulus is dependent on the presence of macrophages

To elucidate the outcomes of the abnormal increase of macrophages in mature ovarian follicles, we determined whether macrophages are involved in the androgenic induction of apoptosis in the ovarian follicles. In situ apoptosis assay (TUNEL) was performed in ovaries of rats treated with DHT for 3 and 7 days (Figure 3A), and ovarian follicles with a large number of apoptotic granulosa cells were considered apoptotic. ANOVA indicates that DHT treatment (P = 0.001) and ovarian follicle stage (P = 0.001) significantly affected the follicle apoptotic rate but with no apparent significant interaction (P > 0.05). Seven days of DHT treatment significantly increased the incidence of TUNEL + pre-ovulatory follicles compared to control (P < 0.05; Figure 3B). No significant differences were observed in pre-antral, early antral, or antral follicles. Moreover, to investigate whether granulosa cell apoptosis under androgenic influence is dependent on the presence of macrophages, the effect of DHT (12 and 24 h) in vitro was studied in single culture of granulosa cells derivate from mature follicles (isolated from rats treated with eCG for 48 h) and in granulosa cells co-cultured with macrophages, and apoptosis was evaluated (Figure 3C). Although DHT was ineffective in inducing apoptosis in granulosa cells in single culture (Figure 3D–F), granulosa cells co-cultured with macrophages in the presence of DHT exhibited significant increase in apoptosis at early stage (12 and 24 h, P < 0.05) and late stage (24 h, P < 0.05) apoptosis when compared to control (Figure 3G and H). Necrosis (7AAD + cells) was not affected by DHT (Figure 2I).

Chemerin is upregulated in ovaries from PCOS women and DHT-treated rats

Chemerin levels in the follicular fluid (P < 0.05) but not in the serum were significantly higher in lean PCOS patients (BMI ≤ 30) compared to their matched control (Figure 4A and B) suggesting that ovarian and serum chemerin may have distinct cellular origins and are differentially regulated. Similar to these observations in the human, serum and ovarian chemerin also appeared to be differently regulated in rats treated with DHT without (15 days-normal weight) or with symptoms of metabolic changes (28 days-overweight rats). Only ovarian chemerin level was significantly higher in rats treated with DHT for 15 days (P < 0.001; Figure 5A), and no differences were detected at the serum level (Figure 5B). In contrast, in rats treated with DHT for 28 days, which exhibit an overweight phenotype, both ovarian and serum chemerin levels were significantly higher (P < 0.001 and P < 0.05, respectively; Figure 5A and B).

A transient influx of ovarian CMKLR1+ M1 macrophages in unhealthy follicles after 15 days of DHT treatment

Chemerin may be involved in the recruitment of M1 macrophages to the ovaries under androgenic conditions. To test this hypothesis, we first investigated whether the number of ovarian CMKLR1+ macrophages was upregulated in vivo by DHT. The frequency of ovarian macrophages (M1 and M2) expressing CMKLR1 was measured by flow cytometry and the gating strategy is demonstrated in Figure 5C. Fifteen days of DHT treatment significant increased ovarian CMKLR1+ M1 macrophages (P < 0.05; Figure 5D) without significant changes in ovarian CMKLR1+ M2 macrophage numbers (Figure 5E). The frequency of ovarian CMKLR1+ M1 macrophages returned to levels comparable to control with longer duration of DHT treatment (28 days of treatment; P > 0.05; Figure 5D). In situ analysis demonstrated that CMKLR1+CD68+ cells were mainly localized in unhealthy follicles (Figure 5F).

Human stromal M1 macrophages expressing CMKLR1 (MHC II+CMKLR1+ cells) in the ovaries were also quantified by immunofluorescence but no differences between PCOS and non-PCOS patients were evident (Figure 4C and D).

Increased splenic M1 macrophages after 28 days of DHT is independent of CMKLR1

In contrast to ovarian macrophages, changes in the frequency of splenic M1 macrophages were noted only after 28 days of DHT treatment (P < 0.05; Figure 6A and B) and were independent of CMKLR1 expression (Figure 6D). The frequency of splenic M2 macrophages was not altered by DHT (Figure 6C). Very few or no splenic M1 macrophages expressed CMKLR1 compared to M2 macrophages and were not altered by DHT (Figure 6D and E). Indeed, peritoneal macrophages from rats treated for 15 days with DHT were also evaluated (Supplemental Figure S3A). DHT failed to influence the incidence of peritoneal M1 or M2 macrophages (Supplemental Figure S3B).

Transient decrease of blood monocytes expressing CMKLR1 at 15 days

Our aforementioned results suggest that the incidence of ovarian CMKLR1+ M1 macrophages is tissue specific and regulated by androgen in a time-dependent manner. Therefore, we investigated whether the recruitment of blood monocytes expressing CMKLR1 would be involved in the increase of ovarian CMKLR1+ M1 macrophages after DHT treatment. Blood monocytes were identified as CD3–CD11b/c+CD43– (classical monocytes) or as CD3–CD11b/c+CD43+ (nonclassical monocytes), and both monocyte subsets expressed CMKLR1 (Figure 6F). Lower incidence of classical monocytes expressing CMKLR1 was detected within 15 days of DHT treatment (P < 0.05; Figure 6G), while nonclassical monocytes expressing CMKLR1 were lower at 7 and 15 days after DHT treatment (P < 0.05; Figure 6G). The number of monocytes expressing CMKLR1 return to levels similar to control rats by 28 days of DHT.

Blood monocytes migrate toward chemerin gradient

To test the function of CMKLR1 on monocytes, a migration assay was performed using transwell plate inserts. Mononuclear cells challenged with chemerin for 6, 12, and 24 h exhibited higher migration rate over time compared to control groups (Figure 6H and I). Migrating mononuclear cells were significant higher after 24 h compared to control (P < 0.05; Figure 6I). No differences were found at 6 and 12 h.