Tissue processing and fluorescence-activated cell sorting (FACS) of endometrial cell populations

Tissue biopsies were divided into two fresh tissue samples processed separately for FACS and for histological examination in formalin-fixed, paraffin-embedded tissue. Tissue processing for viable cell isolation and FACS analysis were performed as previously described (15). Briefly, enzymatically dissociated endometrial cells were incubated in blocking buffer [PBS with 40% human serum and 1% BSA] for 30 minutes and then labeled with the following fluorochrome-conjugated antibodies (BD Biosciences) in PBS containing 10% human serum and 1% BSA: cluster of differentiation (CD)-45 (phycoerythrin-Cy7 anti-CD45) at 1:20 dilution to label contaminating leukocytes for their removal; epithelial cell adhesion molecule (EPCAM; allophycocyanin anti-EPCAM) at 1:20 dilution to label eEP; cluster of differentiation 146 [CD146 or melanoma cell adhesion molecule (MCAM), CD146, fluorescein isothiocyanate anti-MCAM] at 1:5 dilution to label eEN/perivascular cells; β-type platelet-derived growth factor receptor (PDGFRB; phycoerythrin anti-PDGFRB) at 1:5 dilution to label eSFs. eMSCs were sorted using double labeling for CD146 and PDGFRB antibodies, respectively, both at 1:5 dilutions. The cell suspension was sorted using a FACS Aria II with FACS Diva software (BD Biosciences). The FACS-sorted cell pellets were stored at −80°C until RNA extraction.

RNA and cDNA preparation for microarray analysis and quantitative real-time PCR (Q-RT-PCR)

Total RNA was isolated form FACS-sorted cell populations and purified using the Arcturus PicoPure RNA isolation kit (Applied Biosystems, Life Technologies Corporation) following the manufacturer's instructions. An additional deoxyribonuclease treatment was performed using the ribonuclease-free deoxyribonuclease set (Qiagen). Reverse transcription and amplification of isolated RNA into cDNA was performed using NuGEN WT-Ovation Exon FFPE System V2 (NuGen). The integrity of resultant cDNA was assessed using the Agilent 2100 bioanalyzer (Agilent Technologies), and individual samples meeting yield and quality standards were further processed and hybridized to Affymetrix Human Gene 1.0 ST arrays (Affymetrix), probing 21 014 genes. Arrays were scanned according to the protocol described in the wild-type sense target labeling assay manual from Affymetrix (version 4; FS450_0007).

Validation of microarray data by fluidigm-based Q-RT-PCR

Eighty-one genes were chosen from the differentially expressed genes (≥2.0 fold changes [FCs]) to validate lineage-specific gene profiles and cell type-specific differences between the study groups. A total of 29 cDNA samples from FACS-sorted endometrial cell populations and 6 internal controls were analyzed in duplicate by Q-RT-PCR using the Fluidigm 96.96 dynamic array with integrated fluidic circuits and the BioMark HD system (Fluidigm) as previously described (15). The updated Fluidigm protocol 37, with updated modifications to volume/concentration of reagents and timing of reactions for preamplification and Biomark quantitative PCR, was used for all procedures. Briefly, cDNA was preamplified to generate a pool of target genes in 5-μL reactions using Taq-Man Pre-Amp master mix (Applied Biosystems), 200 ng cDNA, and 500 nM for each primer pair. Samples were then exonuclease treated (Exonuclease I; New England BioLabs). Using previously generated optimal dilution curves, samples were diluted 1:5 in a Tris-EDTA dilution buffer (TEKnova). Q-RT-PCR was performed using SsoFast Evagreen supermix with low ROX binding dye (Biotium Inc) with a final primer concentration of 5 μM. Data were processed by user-detected threshold settings and linear baseline correction using Biomark real-time PCR Analysis Software (version 3.0.4). Melt curves were assessed using the melting temperature threshold.

The comparative cycle threshold (Ct) method was used to obtain relative expression for each grouping comparison, in which the amount of target was normalized to the hypoxanthine phosphoribosyltransferase 1 represented by ΔCt (15). Expression was normalized to an internal calibrator for sorted cells ΔΔCt and total FC calculated by 2^−ΔΔCt (ABI. http://hcgs.unh.edu/protocol/realtime/userbulletin2.pdf). If one or more samples in eEP_PCOS/Ctrl, eEN_PCOS/Ctrl, eSF_PCOS/Ctrl, or eMSC_PCOS/Ctrl groups failed to produce a ΔCt value leaving less than 3 samples for the analysis, the FC calculation was not conducted [not calculated (NC), Tables 2 and ​and33].

Table 2.

Microfluidic Q-RT-PCR Validation of Differentially Expressed Genes Between Different Endometrial Cell Populations

Select genes	Microarraya	Q-RT-PCRb	P Valuec
eMSC vs eEP (FC)
eMSC-EP
MMP2	24.21	18.47	.009
MCAM	11.41	103.93	<.001
ITGB1	−6.14	−72.68	.028
DEFB4	−6.39	−28.30	.045
MUC1	−6.40	−71.06	.001
MUC16	−6.96	−51.37	.014
LAMC2	−15.61	−45.24	.001
CDH1	−16.34	−29.40	.034
EPCAM	−16.94	−218.26	.010
MMP7	−20.74	34.03	.021
MMP26	−24.21	−61.87	.005
eMSC vs eEN (FC)
PDGFRB	18.64	285.28	.005
COL3A1	11.79	NC	NC
MMP16	7.40	81.19	.028
MCAM	1.70	1.00	.982
FLT1	−5.92	−17.09	.001
CDH5	−11.09	−36.21	.013
CD34	−11.11	−28.65	.021
VWF	−18.52	−53.15	.007
PECAM1	−19.03	−52.99	.006
SELE	−21.44	−59.51	.029
EMCN	−22.55	−60.02	.001
MMRN1	−28.11	98.98	.020
eMSC vs eSF (FC)
RGS5	39.86	203.99	.001
ANGPT2	25.62	99.01	<.001
SLC38A11	17.57	189.63	<.001
CDH6	16.53	175.16	.015
MCAM	10.22	177.96	<.001
JAG1	7.54	62.89	<.001
PDGFRB	6.41	9.96	.007
HEY2	4.84	34.73	.034
NOTCH3	2.50	162.62	.045
NOTCH1	2.31	94.15	.053
PDGFRA	−9.46	−25.18	.033

Open in a separate window

Abbreviations: EP, endometrial epithelium; EN, endometrial endothelium/perivascular cells.

^aDifferentially expressed genes (ANOVA; P < .05); cutoff for fold change set as 2.0.

^bComparative Ct method, FC.

^cIndependent-samples t test (PASW) comparing the fold change between different cell types in quantitative PCR (P < .05, bolded).

Table 3.

Microfluidic Q-RT-PCR Validation of Select Differentially Expressed Genes Between Overweight/Obese Women With PCOS and Overweight/Obese Controls

Gene	Microarraya	Q-RT-PCRb	P Valuec
eEP
LGR5	3.87	1.80	.427
ORM1	3.15	NC	NC
ORM2	2.88	5.75	.384
LICAM	2.64	NC	NC
SEMA3E	2.48	1.33	.765
UBD	2.38	4.34	.436
IL6	2.35	6.64	.182d
SERPINE2	2.25	1.63	.504
CCL2	2.14	NC	NC
TNFAIP6	2.01	NC	NC
SPP1	−2.28	NC	NC
TGFB1	−2.48	NC	NC
CLDN4	−2.81	−11.30	.042
IGF1	−3.24	−86.99	.026
OLFM4	−3.46	−5.49	.263
ENDRA	−4.37	−269.42	.083
DCN	−5.57	−163.30	.057
SPARCL1	−5.10	−19.68	.021
SPARC	−6.73	−108.15	.028
SFRP4	−9.53	−33.18	.028
eEN
MAL2	3.14	NC	NC
TTN	2.29	18.58	.068
APOD	2.16	4.03	.003
TGFB2	2.16	1.99	.386
INSR	−2.02	−1.58	.360
IL6	−2.05	NC	NC
RHEB	−2.06	−1.28	.791
AKTIP	−2.15	−11.80	.140
SLC16A6	−4.65	NC	NC
eSF
LRP2	3.62	1.65	.456
CCL2	3.31	9.63	.008
ICAM 1	2.87	NC	NC
TNFAIP3	2.51	9.65	.003
RHOJ	2.49	18.79	.003
VLDLR	2.43	1.38	.269
C4A/B	2.20	34.41	.002
HHIP	−2.21	−3.97	.016
RSPO3	−2.35	−3.21	.174
ADAMTS5	−2.56	−3.16	.102
ANLN	−2.69	−3.09	.089
KIF23	−3.15	−4.44	.038
FGF7	−3.74	NC	NC
eMSC
SPRR3	6.65	NC	NC
LCN2	5.06	100.14	.020
CEACAM7	3.62	NC	NC
MAL	2.26	NC	NC
ICAM1	2.00	NC	NC
IL8	2.24	3.98	.039
SMAD2	−2.31	−1.45	.078
OR51E2	−2.37	−7.71	.055
SERPINI1	−2.48	−1.44	.639
TAGLN	−2.65	−2.67	.306
ACTG2	−3.12	−4.56	.085

Open in a separate window

^aDifferentially expressed genes (ANOVA; P < .05); cutoff for FC set as 2.0.

^bComparative Ct method.

^cIndependent-samples t test (PASW), comparing the FC in each cell type between the study groups in quantitative PCR (P < .05, bolded).

^dAfter excluding an outlier (FC = 76.72, P = .044).

Immunohistochemistry

Chemokine (C-C motif) ligand 2 (CCL2) and IL-6 proteins were localized in paraffin-embedded human endometrial tissue sections using the IMPRESS Universal Polymer detection kit (Vector Laboratories) and rabbit polyclonal antibodies for CCL2 (ab9669; Abcam) at 1:100 and IL-6 (ab6672; Abcam) at 1:600. Normal rabbit IgG was used as a negative control. Sections were deparaffinized in xylene, rehydrated through graded ethanols, and washed with PBS. Antigen retrieval was performed at 100°C in trisodium citrate (10 mM, pH 6) for 20 minutes followed by quenching endogenous peroxidase (3% H₂O₂ in methanol) for 10 minutes and blocking at 22°C for 30 minutes in Ready-to-Use (R.T.U.) horse serum (Vectastain universal quick kit, R.T.U.; Vector Laboratories). Primary antibodies were incubated at 4°C overnight and secondary antibody (ImmPRESS universal antibody polymer detection kit; Vector Laboratories) was added for 30 minutes at 22°C. The immunoreactive antigen was visualized using 3,3′-diaminobenzidine (DAB peroxidase substrate kit; Vector Laboratories) and counterstained with hematoxylin blue (Vector Laboratories). Slides were viewed on a Leica DM 5000 microscope equipped with a Leica 350DX camera (Leica Microsystems, Inc), and the staining in each section was evaluated independently by several observers. Additional protein validation was limited by tissue availability.

To quantify and statistically analyze the intensity of immunoreactive cells, a modified version of the H-score system, the Quick Score, was used (18). Here the intensity and the proportion of nuclear and cytoplasmic brown staining (for CCL2 and IL-6) throughout each image were termed category A and were assigned scores from 1 to 6 (1, 0%–4%; 2, 5%–19%; 3, 20%–39%; 4, 40%–59%; 5, 60%–79%; 6, 80%–100%). The whole image was scanned at ×200 magnification to gauge the general level of intensity throughout. The average intensity, corresponding to each of the categories, labeled negative, weak, intermediate, and strong staining, was given a score from 0 to 3, respectively, and termed category B. A × B was used as a multiplicative Quick Score, and thus, the maximum score achievable is 18. Blinded scorers (n = 4) were instructed to judge images taken for each sample (control vs PCOS), and these scores were subsequently used for statistical analysis.

Differential expression analysis of FACS-isolated endometrial cell populations

The transcriptomes of isolated cells types were analyzed by comparing the gene expression of the mesenchymal stem cell population with the individual profiles of other cell types (Supplemental Tables 1, A–C, Journals Online web site at http://jcem.endojournals.org). Some of these differentially expressed genes were chosen for further validation with microfluidic Q-RT-PCR (Table 2).

The endometrial epithelial cells highly expressed genes characteristic for the epithelial lineage, including epithelial specific integrins, matrix metalloproteinases, defensins, tight junction proteins, mucinsm and lamnins compared with the eMSC population (Supplemental Table 1A). Comparison of eMSCs vs eENs and vs eSFs revealed the eEN and eSF populations expressing genes related to endothelial and fibroblast functions, respectively (Supplemental Tables 1, B and C). The eMSC population clustered close to the eSF population in PCA and HC (Figure 2) and had a gene expression profile characteristic of adult endometrial mesenchymal stem/progenitor cells (Supplemental Table 1C) (15). Cell type-specific gene expression was validated by Q-RT-PCR (Table 2).

Endometrial cell type-specific differential gene expression between PCOS women and controls

No statistical differences among participants were observed in age (P = .734) or BMI (P = .054) between the 2 study groups (Table 1). To assess cell type-specific differences in gene expression of each cell type between women with PCOS and controls, 4 major comparisons were performed: epithelial (eEP_PCOS vs eEP_Ctrl), endoethelial (eEN_PCOS vs eEN_Ctrl)_, fibroblast (eSF_PCOS vs eSF_Ctrl,), and mesenchymal stem cell(eMSC_PCOS vs eMSC_Ctrl). The corresponding lists of differentially expressed genes (P < .05, ≥ 2.0 FC, Supplemental Tables 2, A–D) comprised 216 eEP, 168 eEN, 113 eSF, and 69 eMSC differentially expressed genes in PCOS vs controls. Select differentially expressed genes were also analyzed by microfluidic Q-RT-PCR, which largely validated the microarray approach (Table 3).

The comparison eEP_PCOS vs eEP_Ctrl revealed increased expression of inflammation-related genes, eg, CCL2, IL-6, orosomucoid 1 (ORM1), TNF, and α-induced protein 6 (TNFAIP6). In addition, several genes implicated in oncogenesis showed increased expression: cell adhesion molecule with homology to L1CAM (CHL1) or decreased expression: claudin 4 (CLDN4), secreted protein, acidic secreted frizzled-related protein 4 (SFRP4), and secreted protein acidic and rich in cysteine (SPARC, osteonectin). Interestingly, the expression of insulin growth factor 1 (IGF-1) was decreased in PCOS endometrium compared with controls. Of note in eEN_PCOS vs eEN_Ctrl, there was up-regulation of apolipoprotein D (APOD). Similar to eEP_PCOS, eSF_PCOS showed increased expression of inflammatory genes vs eSF_Ctrl, eg, complement component 4A and 4B (C4A/B), CCL2, intercellular adhesion molecule 1 (ICAM1), and TNAIFP3, whereas the expression of genes related to cell growth and proliferation [fibroblast growth factor 7 (FGF7); Hedgehog-interacting protein (HHIP); kinesin-like protein (KIF23)] were decreased. Among the most highly differentially expressed genes between eMSC_PCOS and eMSC_Ctrl were genes related to inflammatory processes (IL-8 and ICAM1) and cancer [proline rich protein 3 (SPRR3) and lipocalin 2 (LCN2)].

Up-regulation of proinflammatory genes in PCOS endometrium and reproductive function

Cytokines and infiltrating leukocytes participate in endometrial cyclic changes, modulating endometrial structural and functional components required for embryo attachment (21) CCL2 (also known as monocyte chemoattractant protein-1, MCP-1), a potent chemoattractant/chemokine, which induces monocytes to leave the bloodstream and enter tissues to become resident macrophages during tissue repair (22), was increased in eEP_PCOS and eSF_PCOS in anovulatory PCOS women, compared with controls. CCL2 protein increases in cycling endometrial eEP during the window of implantation (WOI), suggesting a role in embryo attachment (23, 24). The observed increase in CCL2 in eEP_PCOS in the absence of P₄ suggests an aberrancy, which could theoretically lead to enhanced macrophage influx in the endometrium of women with PCOS and warrants further investigation (25). The data regarding eSF CCL2 expression is interesting because some studies report decreased/absent secretion of CCL2 during WOI and in vitro E₂ and P₄ administration decreases eSF CCL2 secretion (23, 26). The increased expression of CCL2 could result in negative consequences for the endometrium in women with PCOS, especially if the altered secretion pattern prevails during the secretory phase.

IL-6, increased in eEP_PCOS, is a multifunctional cytokine with a wide range of biological activities inducing cytokine production and recruitment of macrophages and megakaryocytes and cell growth (27). Similar to CCL2, IL-6 is secreted by endometrial epithelium and is also expressed in endometrial stromal cells (28, 29). IL-6 is present during the WOI and has an important role coordinating placental morphogenesis and trophoblast invasion (29, 30). IL-6-deficient mice have reduced fertility and fewer implantation sites, whereas, in vitro IL-6 exposure decreases embryo attachment and growth (29, 31). Women experiencing recurrent miscarriage have decreased endometrial IL-6 production, whereas elevated IL-6 levels in plasma and cervical mucus are associated with unexplained infertility (29, 32, 33). PCOS endometrium demonstrates progesterone resistance and impaired decidualization (3) by unclear mechanisms, and the role of IL-6 in this abnormality awaits further clarification. Although we are not aware of previous studies on IL-6 expression in PCOS endometrium, other experimental approaches suggest an aberrant immune response in PCOS endometrium in the secretory phase (3, 4). That inflammation is pronounced in overweight/obese PCOS (proliferative phase) endometrium compared with overweight/obese controls suggests it is independent of high BMI. An increased or otherwise altered inflammatory network as part of the endometrial disease phenotype could result in compromised endometrial cellular differentiation and endometrial receptivity, impaired embryonic implantation, and subsequent poor pregnancy outcomes in women with PCOS.

Proinflammatory changes may promote cancer in PCOS endometrium

In addition to normal reproductive processes, cytokines and a proinflammatory milieu are involved in tumorigenesis, including in the endometrium (12). Up-regulated inflammatory genes in eEP_PCOS and/or eSF_PCOS may contribute to the increased risk of endometrial cancer in women with PCOS. CCL2 increases the expression of survivin, an apoptosis inhibitor, and stimulates tumor cell proliferation, migration, and invasiveness by activating the phosphatidylinositol 3-kinase/Akt and MAPK/ERK1/2 signaling pathways that are also activated in endometrial cancer (34, 35). IL-6 has also been linked to several cancers, including endometrial adenocarcinoma, by promoting a proinflammatory environment, metastasis, and growth, presumably through signal transducer and activator of transcription-3 (STAT3) activation (36, 37). Metformin and statin treatments decrease the inflammatory response in endometrial stromal cells in vitro (38, 39). The fact that 2 PCOS women were on metformin [samples used in microarray and in immunohistochemistry (IHC)] and 1 control woman was using simvastatin (sample used in IHC) could have blunted some of the differences in inflammatory gene expression between the study groups.

Prooncogenic gene expression of PCOS endometrial cells

In addition to normal endometrial function, increased WNT/β-catenin signaling is involved in endometrial hyperplasia and cancer (40). Down-regulation of SFRP4 in PCOS endometrium indicates abnormal proliferative phase development in PCOS epithelium (41). Because SFRP4 serves as WNT signaling antagonist, EP_PCOS compared with EP_Ctrl is of note because SFRP4 inhibits endometrial cancer cell proliferation through inhibition of WNT7A signaling in vitro (42). Furthermore, SFRP4 expression correlates with poor outcome in ovarian cancer, supporting its role as a tumor suppressor gene (43). SPARC, which has tumor suppressor properties, was also down-regulated in eEP_PCOS vs eEP_Ctrl. SPARC functions as a key regulator of matrix proteinase-associated tissue remodeling; however, its function in endometrial epithelium is not well understood (44). Previous studies have implicated its involvement with TGFB signaling both restricting growth and proliferation and also presenting with immunomodulatory properties, attenuating the mitogenic and proinvasive effects of CCL2 in ovarian cancer cells (45). Interestingly, CLD4 and CHL1, which are aberrantly expressed in several cancers, including endometrial cancer, were decreased in eEP_PCOS (46, 47). The fact that genes regulating cellular growth and proliferation (IGF1, FGF7, HHIP, KIF23) were down-regulated in EP_PCOS and eSF_PCOS compared with controls suggests that the changes in PCOS endometrium may occur prior to or even without excessive proliferation. Altogether the data support that a proinflammatory and procancerous milieu promotes an endometrial disease phenotype in PCOS, which likely contributes to the predisposition to endometrial cancer in PCOS women in the long term.

Altered genotype in eMSCs of the PCOS endometrium

Mesenchymal stem cells participate in wound healing relevant to endometrial regeneration responding to changes in their microenvironment (48). Herein we found 69 differentially expressed genes in eMSC_PCOS vs eMSC_Ctrl with up-regulation of the proinflammatory chemokine IL-8 and adhesion molecule ICAM1. Mesenchymal stem cells secrete IL-8, which has a paracrine function in the stem cell niche (49) and is linked to an abnormal microenvironment related to several malignancies (50). In PCOS endometrium IL-8 may promote a proinflammatory milieu with ICAM1 and an abnormal footprint in downstream progeny, having a potential effect on lineage cell development and differentiation, resulting in altered endometrial function. SPRR3, whose function is not well defined, and oncogene LCN2, which is a potent inflammatory mediator (51), were the most up-regulated genes in eMSC_PCOS vs. eMSC_ctrl. Increased expression of both genes has been observed in several malignancies including high LCN2 expression in endometrial cancer (52).

Given that the transcriptome of eMSC was altered in PCOS endometrium together with other cell types suggests that the endocrine and metabolic PCOS environment may contribute to differences in gene expression between the study groups. Moreover, several studies have shown that proper maintenance and regulation of the stem cell microenvironment is crucial to sustain normal stem cell development (53). It should be noted that the eMSC_ctrl were from cycling endometrium, whereas eMSC_PCOS were from anovulatory women. It is unknown at this time whether eMSCs differ in the cycling vs noncycling endometrium, although a limited study revealed no menstrual cycle phase dependence of the eMSC transcriptome (15), and turnover of eMSC in quiescent endometrium is not known. Whether the eMSC transcriptome found herein is due to the anovulatory state or to metabolic disturbances or hyperandrogenemia in PCOS remains to be determined. Altogether the altered eMSC_PCOS genotype is a novel finding and may give new insights into endometrial abnormalities related to PCOS.

We acknowledge John Tamaresis, PhD, and Risto Bloigu, PhD, for statistical advice; Amy Hamilton and Florence Ng for technical support with the Fluidigm analysis and the Biomark system; Ny Jieng for assistance with sample collection: Kim Chi Vo and Brittni Johnson for tissue collection and processing, and the University of California, San Francisco National Institutes of Health Human Endometrial Tissue and DNA Bank for tissue acquisition. We also acknowledge Tara Rombaldo for technical support with FACS analysis and Linda Ta and Yanxia Hao (Genomics Core at the Gladstone Institute) for their technical skills regarding microarray processing.

This work was supported by the Sigrid Juselius Foundation, the Academy of Finland, the Finnish Medical Foundation, the Orion-Farmos Research Foundation, the Maud Kuistila Foundation (to T.T.P.), the Ruth L. Kirschstein National Research Service Award 1F32HD074423-01 (to J.C.C.), and the National Institutes of Health Eunice Kennedy Shriver National Institute of Child Health and Human Development Grant U54HD 055764-06 Specialized Cooperative Centers Program in Reproduction and Infertility Research (to LCG).

Disclosure Summary: The authors have nothing to disclose.