The Protein Kinase A Pathway-Regulated Transcriptome of Endometrial Stromal Fibroblasts Reveals Compromised Differentiation and Persistent Proliferative Potential in Endometriosis

Abstract

Intrinsic abnormalities in transplanted eutopic endometrium are believed to contribute to the pathogenesis of pelvic endometriosis. Herein we investigated transcriptomic differences in human endometrial stromal fibroblasts (hESFs) from women with (hESF_endo) vs. without (hESF_nonendo) endometriosis, in response to activation of the protein kinase A (PKA) pathway with 8-bromoadenosine-cAMP (8-Br-cAMP). hESF_nonendo (n = 4) and hESF_endo (n = 4) were isolated from eutopic endometrium and treated ± 0.5 mm 8-Br-cAMP for 96 h. Purified total RNA was subjected to microarray analysis using the whole-genome Gene 1.0 ST Affymetrix platform. A total of 691 genes were regulated in cAMP-treated hESF_nonendo vs. 158 genes in hESF_endo, suggesting a blunted response to cAMP/PKA pathway activation in women with disease. Real-time PCR and ELISA validated the decreased expression of decidualization markers in hESF_endo compared with hESF_nonendo. In the absence of disease, 8-Br-cAMP down-regulated progression through the cell cycle via a decrease in cyclin D1, cyclin-dependent kinase 6, and cell division cycle 2 and an increase in cyclin-dependent kinase inhibitor 1A. However, cell cycle components in hESF_endo were not responsive to 8-Br-cAMP, resulting in persistence of a proliferative phenotype. hESF_endo treated with 8-Br-cAMP exhibited altered expression of immune response, extracellular matrix, cytoskeleton, and apoptosis genes. Changes in phosphodiesterase expression and activity were not different among experimental groups. These data support that eutopic hESF_endo with increased proliferative potential can seed the pelvic cavity via retrograde menstruation and promote establishment, survival, and proliferation of endometriosis lesions, independent of hydrolysis of cAMP and likely due to an inherent abnormality in the PKA pathway.

Endometrial stroma from subjects with endometriosis has increased proliferation potential and a blunted response to cAMP, which is independent of hydrolysis of cAMP by phosphodiesterase.

In endometriosis, endometrium (uterine lining)-like tissue is found primarily on the pelvic peritoneum and ovaries (1). Peritoneal disease is the most common form, occurring in 6–10% of women of reproductive age and is believed to derive mainly from retrograde menstruation and transplantation of eutopic (within the uterus) endometrial tissue fragments and cells to the peritoneum (2). Resulting lesions undergo neoangeogenesis; proliferation; invasion of surrounding structures; infiltration by sensory and sympathetic nerves (3); and elicit a local inflammatory response, adhesion formation, and fibrosis (1,4). Intrinsic abnormalities of the transplanted eutopic endometrium are believed to enable establishing disease in the pelvis (4). Thirty-five to 50% of women with pelvic pain and/or infertility have endometriosis (5), with infertility due to ovulatory dysfunction, poor oocyte quality in response to the inflammatory environment, and abnormal eutopic endometrium with compromised embryonic implantation (1).

Transcriptomic analyses investigating differences in eutopic endometrium from women with and without endometriosis (6,7) reveal abnormalities in the proliferative (estrogen-dominant)-to-secretory [progesterone (P₄) dominant] phase transition (7), normally observed in the menstrual cycle (8). Several studies support resistance to P₄ action in this tissue (9,10). P₄ signaling and the protein kinase A (PKA) pathway intersect and are key to decidualization (differentiation) of human endometrial stromal fibroblasts (hESFs), which are central to successful embryonic implantation, placental function, and pregnancy progression (11,12). Impaired decidualization in vivo has been suggested as an underlying mechanism for infertility in women with endometriosis (10). In vitro, decidualization of hESFs can be induced by P₄ with estradiol priming (13,14) and/or activation of the PKA pathway by relaxin, prostaglandins, and cAMP analogs (13,15,16,17,18). We (8) and others (16) have demonstrated a blunted expression of specific decidual markers in response to activation of the cAMP/PKA pathway in eutopic hESFs from women with endometriosis (hESF_endo) vs. without disease (hESF_nonendo). The molecular bases of these differences are not well understood.

Herein we investigated the transcriptome resulting from activation of the PKA pathway by the cAMP analog, 8-bromoadenosine-cAMP (8-Br-cAMP), in hESFs from women without and with endometriosis. The data demonstrate that activation of the PKA pathway in hESF_nonendo results in down-regulation of genes involved in cell cycle progression and up-regulation of their inhibitors as well as regulation of genes involved in cytoskeletal changes important in hESF differentiation and key signaling pathways and markers of the decidualization process. In contrast, there is dysregulation of the actions of cAMP, in the absence of changes in phosphodiesterases, differentiation of hESF_endo, and persistence of a proliferative phenotype involving key regulators of the cell cycle, including cyclins and cyclin inhibitors in the presence of disease.

Materials and Methods

Subjects

Endometrial tissue biopsies were obtained from 14 reproductive-age women. Six subjects had endometriosis (n = 1 minimal, n = 2 mild, n = 1 moderate-severe, n = 2 severe), based on visualization of lesions during laparoscopy and their histologic evaluation (Table 1). Staging of endometriosis was according to the revised American Fertility Society classification system (19). Participating subjects with endometriosis were 22–46 yr old (mean 33.5 ± 3.0). Controls were eight regularly cycling premenopausal subjects (37–49 yr old, mean 44.4 ± 1.7) undergoing endometrial biopsy or hysterectomy for benign reasons (Table 1), had no history and no evidence of endometriosis at laparoscopy, were documented not to be pregnant, and had not been on hormonal therapies for at least 3 months before tissue sampling.

Table 1.

Characteristics of patients donated endometrial biopsy samples for the study

Patient	Cycle phase	hESF used in experiments	Diagnosis at laparoscopy	Age (yr)	Ethnicity
Endometriosis
233	MSE	Microarray, ELISA, validation	Mild endometriosis, pelvic pain	31	Caucasian
243	ESE	Microarray, ELISA, validation	Minimal endometriosis, bilateral ovarian cyst, intramural myoma	46	Caucasian
267	LSE	Microarray, ELISA, validation	Mild endometriosis, pelvic pain	32	Caucasian
288	PE	Microarray, ELISA, validation, PDE assays	Severe endometriosis	22	Caucasian
279	PE	PDE assays	Moderate-severe endometriosis, pelvic pain	30	Unspecified
381	MSE	PDE assays	Severe endometriosis, pelvic pain, myoma	40	Asian
No endometriosis
229	LSE	Microarray, ELISA, validation	Pelvic pain (no endometriosis at laparoscopy)	47	Caucasian
236	PE	Microarray, ELISA, validation	Symptomatic pelvic prolapse	47	Caucasian
237	ESE	Microarray, ELISA, validation	Intramural myoma, left paratubal cyst	39	Caucasian
238	ESE	Microarray, ELISA, validation	Endometrial polyp	41	Black
285	PE	PDE assays	Intramural myoma, pelvic adhesions	37	Caucasian
293	PE	PDE assays	Intramural myoma	49	Caucasian
298	ESE	PDE assays	Endometrial polyp	46	Asian
326	PE	PDE assays	Intramural myoma	49	Asian

PE, Proliferative endometrium, MSE, midsecretory endometrium; LSE, late secretory endometrium. 

The University of California, San Francisco, Committee on Human Research and the Stanford University Committee on the Use of Human Subjects in Research approved the study. Written informed consent was obtained from subjects. Samples were also obtained through the University of California, San Francisco, National Institutes of Health Human Endometrial Tissue and DNA Bank with appropriate institutional review, approvals, and informed consent from all subjects.

Isolation, culture, and decidualization of endometrial stromal cells

All samples were collected, transported, and processed as described (13,20). Decidualization was performed by treating hESFs cultured in low-serum medium with 0.5 mm 8-Br-cAMP for 96 h and confirmed by measuring decidual markers, IGF binding protein (IGFBP)-1, and prolactin (PRL), in conditioned media by ELISA (13). Time zero (t-0) control samples were collected before initiating treatment. Cells cultured for the corresponding time periods without treatment served as additional controls. hESF cell morphology before and after decidualization was documented. Cells lysates in RLT lysis buffer (QIAGEN, Valencia, CA) containing β-mercaptoethanol and conditioned media were collected after 96 h of incubation.

Total RNA isolation, microarray hybridization, and real-time RT-PCR

Total RNA was isolated from hESFs from subjects without (n = 4) and with (n = 4) endometriosis and purified using RNeasy Plus minikit (QIAGEN). Samples were quantified by spectroscopy, and purity was analyzed by the 260:280 absorbance ratio. RNA quality and integrity were assessed using Bioanalyzer 2100 (Agilent Technologies, Santa Clara, CA) with all samples having high-quality RNA [RNA integrity number (RIN) = 9.7–10]. Hybridization was performed with Human Gene 1.0 ST arrays (Affymetrix, Inc, Santa Clara, CA). Briefly, 100 ng of total RNA from each sample were reverse transcribed to cDNA, followed by overnight in vitro transcription to generate cRNA, which was reverse transcribed, and the 5.5 μg of sense cDNA were fragmented and labeled. The quality of cDNA and fragmented cDNA was assessed in the Agilent bioanalyzer. Microarrays were hybridized, washed, stained, and scanned according to the protocol described in the WT sense target labeling assay manual from Affymetrix (version 4; FS450_0007).

For quantitative (Q) RT-PCR analysis, 1 μg of RNA was converted to cDNA using the iScript cDNA synthesis kit (Bio-Rad Laboratories, Hercules, CA). The real-time RT-PCR reaction was carried out for 40 cycles with the primers listed in Table 2.

Table 2.

Primer sequences used in real-time RT-PCR experiments

Gene	Sense primer 5′–3′	Antisense primer 5′–3′
IGFBP1	CTATGATGGCTCGAAGGCTC	TTCTTGTTGCAGTTTGGCAG
PRL	CATCAACAGCTGCCACACTT	CGTTTGGTTTGCTCCTCAAT
FOXO1A	AAGAGCGTGCCCTACTTCAA	CTGTTGTTGTCCATGGATGC
CXCR4	GGGCCTGAGTGCTCCAGTAG	GGGTAGAAGCGGTCACAGAT
SST	CCCAGACTCCGTCAGTTTCT	ATCATTCTCCGTCTGGTTGG
IL13AR2	CCTTGAAAACAACAAATGAAACC	TAGTTATATTTGTAACCGGTCTGCTTT
DKK1	CATCAGACTGTGCCTCAGGA	CCACAGTAACAACGCTGGAA
DKK3	ATGAGTATGAAGTTGGCAGC	TATTGCACATCTACCCACAG
CCND1	GTGGGTGTGCAAGCCAGGT	TTCCTGTCCTACTACCGCCT
CDC2	GCTTATGCAGGATTCCAGGTT	CAATCCCCTGTAGGATTTGGT
CDKN1A (p21)	GTCCGTCAGAACCCATGC	GCTTCCTCTTGGAGAAGATCA
CDK6	GCCTCTTTTTCGTGGAAGTT	AATTGGTTGGGCAGATTTTG
RPL19	GCAGATAATGGGAGGAGCC	GCCCATCTTTGATGAGCTTC

Microarray gene expression data analysis and statistical analysis

To minimize technical (nonbiological) variability among arrays, densitometry values between arrays were normalized using the Robust Multichip Average function and further transformed to the logarithmic scale (log2). Probes with a known GenBank accession ID correspondence were selected for functional analysis. Statistically significant differences between groups were determined using statistical analysis of microarrays (21) and RankProd (22) methods, using the Bioconductor (http://www.bioconductor.org/) packages Siggene and RankProd, respectively, both run under R software (http://www.r-project.org/). We used two different statistical tests to identify the most robustly differentially expressed genes. Functional annotations were carried out using the Ingenuity Pathway Analysis (IPA; Ingenuity Systems, Redwood City, CA, http://www.ingenuity.com/), in which gene symbols and fold changes of the up- and down-regulated genes were imported.

Hierarchical clustering

Hierarchical clustering is an unsupervised way of grouping samples based only on their gene expression similarities (23). Herein we conducted hierarchical cluster analysis of differentially expressed genes from all samples (the combined gene list) using the smooth correlation for distance measure algorithm (GeneSpring 7.3) to identify samples with similar patterns of gene expression. The output data are also displayed graphically as a dendrogram of hESF_nonendo and hESF_endo samples. The raw data files are stored at the National Center for Biotechnology Information gene Expression Omnibus database, number GSE17504.

Microarray validation by real-time PCR

The most highly up- or down-regulated genes, as well as some genes of interest, based on previous transcriptomic analyses (4), were chosen for validation by QRT-PCR. QRT-PCR was performed in duplicate according to the manufacturer’s instructions and as described above, with the above primers.

ELISA

Conditioned media from cultured hESFs were subjected to ELISA to determine IGFBP1 and PRL concentrations (Alpha Diagnostic International, San Antonio, TX, and Diagnostic Systems Labs, Webster, TX, respectively). Levels of IGFBP1 and PRL for each sample were normalized to total RNA.

Phosphodiesterase (PDE) activity

Reagents were purchased from Sigma (St. Louis, MO) unless otherwise specified. Protease and phosphatase inhibitor were from Roche Diagnostics (Indianapolis, IN). Isolated hESF cells from subjects (n = 4 with and n = 3 without endometriosis) were cultured as described above with or without subsequent treatment for 96 h with 0.5 mm 8-Br-cAMP. Cell pellets were lysed in 500 μl lysis buffer [50 mm Tris-HCl, 1% Nonidet P-40, 150 mm NaCl, 1 mm EDTA, and protease and phosphatase inhibitors (pH 7.4)] and then gently rotated at 4 C for 30 min. Lysates were then spun for 20 min at 15,000 × g at 4 C. The supernatant was removed and protein concentration determined by the BCA protein assay. The supernatant was then diluted 5-fold in 10 mm potassium phosphate (pH 6.8), 25 mm β-mercaptoethanol, and 1 mg/ml BSA, and 10 μl were assayed for cAMP PDE activity in a final volume of 100 μl, as described (24), with 100 nm [³H]cAMP as substrate and 2 mm EGTA. [³H]cAMP was from Amersham Healthcare (Piscataway, NJ). The activity of the PDE4 family was determined in the presence of 1 μm rolipram. The cAMP PDE assay was also performed in the presence of 100 μm isobutylmethylxanthine (IBMX) to determine the apparent contribution of the PDE8 family, which is insensitive to IBMX (25,26).

Statistical evaluation

Statistical analysis of data generated in the ELISA was performed using a two-tailed type 3 Student’s t test. For QRT-PCR and PDE assays, the nonparametric Mann-Whitney test was used. Significance was determined at P ≤ 0.05.

Results

Samples cluster by treatment (not disease)

hESFs from women with and without endometriosis have different responses to 8-Br-cAMP with regard to secretion of the decidualization markers, IGFBP1 and PRL (Fig. 1A). To determine hESF sample clustering patterns, hierarchical clustering analysis of the expression profiles of hESF_endo and hESF_nonendo was conducted, using all gene sets (Fig. 1B). As demonstrated in Fig. 1B, samples clustered based on treatment, and not on disease, subject age, or how the tissue was obtained, consistent with our previous observations on whole endometrial tissue (7). Figure 1C demonstrates the Venn diagram of common and unique genes regulated in hESF_endo and hESF_nonendo by 8-Br-cAMP treatment (supplemental Tables S1 and S2, published as supplemental data on The Endocrine Society’s Journals Online web site at http://endo.endojournals.org). Interestingly, the total number of genes regulated by cAMP was much lower in hESF_endo, only 135 genes being common between the two groups.

Figure 1.

A, IGFBP1 and PRL protein secretion in conditioned medium from hESFs treated with 0.5 mm cAMP for 96 h, normalized to total RNA, n = 4. *, Significance accepted at P ≤ 0.05. Error bars, ±sem. B, Hierarchical clustering analysis of hESF_nonendo (no endometriosis) and hESF_endo (endometriosis) samples treated with (+) or without (−) 0.5 mm cAMP for 96 h. C, Venn diagram of shared and unique genes between hESF_endo and hESF_nonendo response to cAMP treatment (gene lists in supplemental Tables S1 and S2).

The PKA pathway-regulated transcriptomes of hESF_nonendo and hESF_endo

Gene expression profiles were derived from hESF_nonendo and hESF_endo in response to 8-Br-cAMP or vehicle for 96 h, and four comparison groups were generated: A) hESF_nonendo treated with cAMP vs. vehicle control (i.e. no cAMP); B) hESF_endo treated with cAMP vs. vehicle control; C) vehicle control hESF_endo vs. vehicle control hESF_nonendo; and D) cAMP-treated hESF_endo vs. cAMP-treated hESF_nonendo (supplemental Tables S1–S3 and Table 3). Group A presented with the highest number of regulated genes (319 up- and 414 down-regulated) after 1.5-fold cutoff. Remarkably, only 158 genes were regulated in hESF_endo by cAMP, with 77 up- and 95 down-regulated genes (supplemental Tables S1 and S2). Analysis of microarray data demonstrated that mRNA levels of IGFBP1 and PRL in response to 8-Br-cAMP were significantly higher in hESF_nonendo in response to cAMP (group A) (12.25- and 30.8-fold, respectively), compared with hESF_endo (group B) (3.0- and 8.8-fold, respectively) (supplemental Tables S1 and S2). Moreover, analysis of group D showed that IGFBP1 and PRL levels were lower in cAMP-treated hESF_endo (4.2- and 4-fold, respectively) (Table 3) but not in vehicle controls (supplemental Table S3). There was a general trend of diminished response in hESFs from women with endometriosis to cAMP (as seen from lower expression of the cAMP-exclusively regulated somatostatin gene, SST) and overall a more limited response of cAMP/P₄ coregulated genes characteristic of decidualization (i.e. PRL, IGFBP1, FOXO1A, SFRP1, TNFAIP5, MMPs, IL-8, IL-6, IL13RA2, IL-1B) (supplemental Table S2). IGFBP1, PRL, SST, and FOXO1A gene expression was further validated by QPCR (Fig. 2).

Table 3.

List of up- and down-regulated genes in hESF_endo treated with 0.5 mm cAMP for 96 h (hESF_endo cAMP) vs. hESF_nonendo cAMP, expressed as fold change (FC)

Gene symbol	FC (hESFendo cAMP/hESFnonendo cAMP)	Description
CST1	3.53	Cystatin SN
ANGPT2	3.28	Angiopoietin 2
AK5	2.78	Adenylate kinase 5
DSC3	2.66	Desmocollin 3
ATP1B1	2.58	ATPase, Na+
ITGA4	2.56	Integrin, α4 (antigen CD49D, α4 subunit of VLA-4 receptor)
ARHGDIB	2.50	ρ-GDP dissociation inhibitor (GDI)-β
MYCN	2.45	v-myc Myelocytomatosis viral related oncogene, neuroblastoma derived (avian)
PDE8B	2.41	Phosphodiesterase 8B
HSD17B2	2.40	Hydroxysteroid (17-β) dehydrogenase 2
DHRS3	2.23	Dehydrogenase
HEY1	2.21	Hairy
CHRM2	2.20	Cholinergic receptor, muscarinic 2
CCND2	2.19	Cyclin D2
KCNMB4	2.18	Potassium large conductance calcium-activated channel, subfamily M, β-member 4
PSG4	2.08	Pregnancy specific β-1-glycoprotein 4
DOCK9	2.06	Dedicator of cytokinesis 9
TNFSF15	2.05	TNF (ligand) superfamily, member 15
MXRA5	2.05	Matrix-remodeling associated 5
FAM46C	1.93	Family with sequence similarity 46, member C
IL1B	−1.45	IL-1, β
CHI3L1	−1.69	Chitinase 3-like 1 (cartilage glycoprotein-39)
TNFAIP3	−1.84	tumor necrosis factor, α-induced protein 3
KLHL23	−1.85	Kelch-like 23 (Drosophila)
SLIT2	−1.86	Slit homolog 2 (Drosophila)
POSTN	−1.88	Periostin, osteoblast specific factor
REN	−1.89	Renin
CLIC2	−1.90	Chloride intracellular channel 2
PDGFRL	−1.91	Platelet-derived growth factor receptor-like
C13orf33	−1.92	Chromosome 13 open reading frame 33
IL33	−1.94	IL-33
TMEM154	−1.94	Transmembrane protein 154
MYC	−1.95	v-myc Myelocytomatosis viral oncogene homolog (avian)
NOG	−1.95	Noggin
TLR4	−1.96	Toll-like receptor 4
OLFML1	−1.98	Olfactomedin-like 1
CGA	−1.98	Glycoprotein hormones, α polypeptide
RXFP1	−1.98	Relaxin
BDKRB1	−1.99	Bradykinin receptor B1
DKK1	−1.99	Dickkopf homolog 1 (Xenopus laevis)
RDH10	−1.99	Retinol dehydrogenase 10 (all-trans)
KYNU	−2.01	Kynureninase (l-kynurenine hydrolase)
PEG10	−2.05	Paternally expressed 10
CXCL12	−2.05	Chemokine (C-X-C motif) ligand 12 (stromal cell-derived factor 1)
SLC7A8	−2.06	Solute carrier family 7 (cationic amino acid transporter, y+ system), member 8
IL1R1	−2.07	IL-1 receptor, type I
HSD11B1	−2.08	Hydroxysteroid (11-β) dehydrogenase 1
BMP3	−2.09	Bone morphogenetic protein 3 (osteogenic)
IGFBP2	−2.09	insulin-like growth factor binding protein 2, 36 kDa
TWIST1	−2.11	Twist homolog 1 (acrocephalosyndactyly 3
ABCC9	−2.11	ATP-binding cassette, sub-family C (CFTR
EFNB2	−2.13	Ephrin-B2
EMILIN2	−2.16	Elastin microfibril interfacer 2
LIF	−2.16	Leukemia inhibitory factor (cholinergic differentiation factor)
MFAP4	−2.16	Microfibrillar-associated protein 4
CDH13	−2.16	Cadherin 13, H-cadherin (heart)
GNG11	−2.16	Guanine nucleotide binding protein (G protein), γ 11
PTGS2	−2.17	Prostaglandin-endoperoxide synthase 2 (prostaglandin G
MOBKL2B	−2.18	MOB1, Mps one binder kinase activator-like 2B (yeast)
(Continued)

Table 3A.

Continued

Gene symbol	FC (hESFendo cAMP/hESFnonendo cAMP)	Description
IL13RA2	−2.19	IL-13 receptor, α 2
CD68	−2.19	CD68 molecule
IL6	−2.20	IL-6 (interferon, β 2)
NOX4	−2.22	NADPH oxidase 4
IL24	−2.29	IL-24
SNCA	−2.31	Synuclein, α (non A4 component of amyloid precursor)
ERAP2	−2.34	Endoplasmic reticulum aminopeptidase 2
PTGES	−2.37	Prostaglandin E synthase
ABCA9	−2.38	ATP-binding cassette, sub-family A (ABC1), member 9
IL8	−2.43	IL-8
EDNRA	−2.47	Endothelin receptor type A
GADD45G	−2.53	Growth arrest and DNA-damage-inducible, γ
ROR2	−2.55	Receptor tyrosine kinase-like orphan receptor 2
ABCA8	−2.56	ATP-binding cassette, sub-family A (ABC1), member 8
IL11	−2.57	IL-11
HGF	−2.57	Hepatocyte growth factor (hepapoietin A
ALDH1A1	−2.59	Aldehyde dehydrogenase 1 family, member A1
ENPP1	−2.60	Ectonucleotide pyrophosphatase
ABCA6	−2.62	ATP-binding cassette, sub-family A (ABC1), member 6
NOV	−2.63	Nephroblastoma overexpressed gene
ALDH1A2	−2.64	Aldehyde dehydrogenase 1 family, member A2 (ALDH1A2), transcript variant 3, mRNA
IGFBP3	−2.75	IGF binding protein 3
NR4A2	−2.78	Nuclear receptor subfamily 4, group A, member 2
THSD7A	−2.84	Thrombospondin, type I, domain containing 7A
ENPEP	−2.87	Glutamyl aminopeptidase (aminopeptidase A)
C4orf31	−2.89	Chromosome 4 open reading frame 31
EDG2	−3.06	Endothelial differentiation, lysophosphatidic acid G-protein-coupled receptor, 2
SFRP1	−3.28	Secreted frizzled-related protein 1
PTHLH	−3.32	PTH-like hormone
EDNRB	−3.42	Endothelin receptor type B (EDNRB), transcript variant 2, mRNA
FST	−3.57	Follistatin
KGFLP1	−3.69	Keratinocyte growth factor-like protein 1
KGFLP1	−3.69	Keratinocyte growth factor-like protein 1
KGFLP1	−3.69	Keratinocyte growth factor-like protein 1
CCL8	−3.88	Chemokine (C-C motif) ligand 8
FGF7	−3.91	Fibroblast growth factor 7 (keratinocyte growth factor)
TNFAIP6	−3.91	TNF, α-induced protein 6
ENPP2	−3.95	Ectonucleotide pyrophosphatase
PRL	−3.99	Prolactin
LTBP1	−4.08	Latent TGF-β binding protein 1
COL15A1	−4.13	Collagen, type XV, α1
CCL2	−4.16	Chemokine (C-C motif) ligand 2
IGFBP1	−4.17	IGFBP1
MMP8	−5.35	Matrix metallopeptidase 8 (neutrophil collagenase)
SST	−5.36	Somatostatin
TFPI2	−7.12	Tissue factor pathway inhibitor 2

Figure 2.

QPCR validation of microarray data of gene expression in hESF_nonendo and hESF_endo decidualized in vitro with 0.5 mm with 8Br-cAMP for 96 h, expressed as fold change to the expression in 96 h vehicle controls. *, Significance accepted at P ≤ 0.05 (Mann-Whitney test). Error bars, ±sem. FC, Fold change; SST, somatostatin; CXCR4, C-X-C chemokine receptor type 4.

Functional network analysis

The four study groups (A–D) were subjected to canonical pathway analysis and generation of networks. Genes were selected based on the fold change cutoff (1.5-fold) and P < 0.05. In group A, 691 genes were identified as regulated after applying both parametric statistical analysis of microarrays and nonparametric (RankProd) tests, from which 351 molecules were eligible for generating networks and pathways. In group B, 158 genes were regulated, with 103 molecules eligible for generating networks and pathways. In group C, of 245 molecules identified as different between control hESF_endo vs. control hESF_nonendo, 202 were eligible for generating networks. In group D, 105 genes were different between cAMP-treated hESF_endo vs. hESF_nonendo, with 96 eligible for generating networks and pathways. Although numerous pathways were identified as regulated, many canonical pathways are connected with each other because many genes participate in multiple pathways.

In group A, the five leading pathways with a high ratio of regulated molecules per total number of molecules in the pathway were (ratio; P value): hepatic fibrosis (12 of 135; 0.089), cell cycle G₂/M DNA damage checkpoint regulation (five of 43; 0.116), caveolar-mediated endocytosis (seven of 81; 0.086), cAMP mediated signaling (10 of 159, 0.063), and G protein-coupled signaling (11 of 202, 0.054) (supplemental Table S4). The activation of other pathways in this group is worth mentioning, including vascular endothelial growth factor, tight junction, notch, bone morphogenetic protein, TGF-β, integrin, and sonic hedgehog signaling pathways and others known to be actively involved in endometrial physiology.

There is a blunted response to cAMP in hESF_endo vs. hESF_nonendo [Fig. 1A, (13)], and therefore, few genes were anticipated to be regulated per total number of molecules in each pathway. The top five pathways in group B (hESF_endo+cAMP vs. vehicle control) were (ratio; P value): cell cycle G₂/M DNA damage checkpoint regulation (three of 43; 0.07), lipopolysaccharide/IL-1-mediated inhibition of retinoid X receptor function (five of 198; 0.025), caveolar-mediated endocytosis (three of 81; 0.037), tight junction signaling (four of 164; 0.024), calcium signaling (four of 202; 0.02) (supplemental Table S5). The complete lists of regulated genes and canonical pathways are presented as supplemental data (supplemental Tables S1–S6) and Table 3. Of special interest is the list of canonical pathways differentially regulated in cAMP treatment of hESF_endo vs. hESF_nonendo. Pathways, uniquely activated in hESF_endo response to cAMP (group B) and in hESF_nonendo response to cAMP (group A) are presented in supplemental Table S6.

Cell cycle G₁/S and G₂/M DNA damage checkpoint regulation pathways

Microarray analysis of the hESF_nonendo treated with vs. without cAMP (group A) demonstrated down-regulation of cell cycle pathways due to significant down-regulation of cyclin (CCN)-D1 (−2.68-fold) and cyclin-dependent kinase (CDK)-6 (−2.56-fold), up-regulation of cyclin-dependent kinase inhibitor (CDKN)-1A (p21) (+1.7-fold), decrease in CDKN3 (−1.9-fold) and cell division cycle 2 (−2.6). These data, indicating a decrease in cell cycling in normal hESF on decidualization, support previous reports (18,27). However, in contrast, microarray analysis revealed that in hESF_endo down-regulation of cell cycle pathway members in response to cAMP (group B) was not observed, and there was no change in CCND1, CDK6, CDKN1A, and cell division cycle 2, with only CDKN3 decreased 2.87-fold. In groups C and D, CCND2 was up-regulated 1.68- and 2.19-fold, respectively, indicating that even in vehicle-treated hESF_endo, the basal level of this cell cycle regulator is higher than in hESF_nonendo (supplemental Tables S1–S4). Expression of mRNA for some of the cell cycle genes was validated by QRT-PCR and comparison with the microarray data are presented in Fig. 3. Very recently, we demonstrated that 5-bromo-2′-deoxyuridine incorporation is significantly decreased in cAMP-treated hESF_nonendo vs. hESF_endo (27). These data suggest that hESF_endo exhibits higher proliferation potential under the decidualizing stimulus (cAMP) and are consistent with the transcriptomic analyses herein.

Figure 3.

All hESFs were treated with or without 0.5 mm cAMP for 96 h. Validation by real-time QPCR expression of cell cycle genes’ mRNA. *, Significance accepted as P ≤ 0.05. Error bars, ±sem. FC, Fold change.

Differentiation

Several canonical signaling pathways were regulated in hESFs after cAMP treatment, as determined by IPA analysis, and some of them exhibited differences between hESF_endo vs. hESF_nonendo.

Immune genes

The most highly regulated gene (88.5-fold) on hESF_nonendo decidualization with cAMP was IL13RA2, high-affinity IL-13 receptor-α2, which was not regulated in response to cAMP in hESF_endo (supplemental Tables S1 and S2), indicating marked dysregulation in the setting of endometriosis. CXCR4 (C-X-C chemokine receptor type 4), the receptor for CXCL12 (stromal cell derived factor 1), was significantly up-regulated 58.3- and 33.5-fold in hESF_nonendo and hESF_endo, respectively, in response to cAMP treatment. QRT-PCR validation of IL13RA2 and CXCR4 expression in hESFs before and after cAMP treatment confirmed the microarray data and underscored diminished responsiveness of hESF_endo to cAMP (Fig. 2).

Wnt signaling pathway

The Wnt inhibitor Dickkopf-1, exclusively regulated by P₄ (20), was significantly lower in hESF_endo compared with hESF_nonendo, and treatment with or without cAMP had no effect on its expression. SFRP1 (an inhibitor of the Wnt receptor Frizzled) was 3.27-fold decreased in hESF_endo. Furthermore, increased expression of nemo-like kinase protein (a negative regulator of Wnt signaling that translocates from the cytoplasm to nucleus and decreases transcription of CCND1) was found in hESF_nonendo in response to cAMP but not in hESF_endo. Overall, IPA suggests activation of the Wnt canonical and noncanonical pathways in endometriosis.

Forkhead box O (FOXO) signaling pathway

FOXO1A is important in the decidualization process (28,29), and, interestingly, the FOXO1A transcript was increased in hESF_nonendo, but not hESF_endo, in response to cAMP. FOXO transcription factors up-regulated Bcl-2-interacting mediator of cell death (BIM) expression, as reported in other cell types (30), which is an initiator of apoptosis (31,32).

ERK/MAPK signaling pathway

We and others recently found that the MAPK pathway is constitutively active in hESF_endo vs. hESF_nonendo (27,33). Herein dual-specificity phosphatases 1 and 4, which are known inhibitors of ERK1/2, were up-regulated 4.4- and 3.26-fold, respectively, in hESF_nonendo treated with cAMP. This can lead to increased E-twenty six (ETS) transcription factors, such as ETS2 [which was up-regulated 2.7-fold (supplemental Table S1)] that are involved in stem cell development, cell senescence and death, and tumorogenesis (34). ERK/MAPK signaling did not change significantly in hESF_endo on cAMP treatment.

Cytoskeleton and extracellular matrix (ECM) genes

Changes in the cytoskeleton and cell shape are hallmarks of hESF decidualization. Several cytoskeletal structural genes and some ECM genes were down-regulated in hESF_nonendo in response to cAMP. α2-Smooth muscle actin and γ2-smooth muscle actin were significantly down-regulated during the normal decidualization process (6.7- and 6.0-fold, respectively) (35), whereas in cAMP-treated hESF_endo, only α2-smooth muscle actin was decreased, consistent with increased migration/motility of stromal cells in endometriosis. Filamins A, B, and C, which are involved in remodeling the cytoskeleton and participate in anchoring membrane proteins for the actin cytoskeleton, were greater than 2-fold decreased in response to cAMP in hESF_nonendo, whereas only filamin B was (down)regulated in hESF_endo (supplemental Tables S1 and S2). Myosin light-chain kinase, myosin heavy-chain 10, and myosin light-chain 9 regulatory unit were all greater than 2-fold down-regulated in decidualized hESF_nonendo but not in cAMP-treated hESF_endo (except for myosin light-chain 9 regulatory unit). Fibronectin 1, anilin, laminin-α4, matrix-remodeling associated 5, and some kinesin family members were significantly down-regulated in hESF_nonendo, but not hESF_endo, after treatment with cAMP (supplemental Tables S1 and S2).

Transgelin (TAGLN, SM22), a smooth muscle actin-binding protein specific for fibroblasts and smooth muscle cells and a repressor of matrix metalloproteinase (MMP)-9 (36), was down-regulated (15.5- and 11.3-fold) after decidualization with cAMP in hESF_nonendo and hESF_endo, respectively. MMP10 and MMP8 were markedly (15.6- and 15.8-fold, respectively) up-regulated in hESF_nonendo in response to cAMP, supporting earlier reports (15,37,38) and their role in matrix remodeling during the normal decidualization process and preparation for implantation. Of note is that the MMP8 transcript was 5.35-fold lower in hESF_endo, whereas MMP1, MMP3, MMP10, and MMP12 were greater than 2-fold up-regulated in hESF_endo vs. hESF_nonendo. Collagen type VIα1, collagen type VIα3, and collagen α-1(XIV) chain (COL14A1; undulin) transcripts were significantly decreased in hESF_nonendo response to cAMP and were not regulated in hESF_endo in the present study.

Tenascin C (hexabrachion), an extracellular matrix protein that binds collagen, fibronectin, and integrins, which can act as an inhibitor of cell adhesion to fibronectin (39), was the only gene in our study that was 8.15-fold down-regulated in cAMP-treated hESF_endo but not regulated by cAMP in hESF_nonendo. Levels of the tenascin C transcript were slightly (1.6-fold) lower in untreated hESF_endo vs. hESF_nonendo.

Validation and functional analysis of PDE regulation

Because of a blunted response of hESF_endo vs. hESF_nonendo to 8-Br-cAMP, we investigated whether the resistance to activation of the PKA pathway is due to hydrolysis of the cAMP analog 8-Br-cAMP by specific PDEs. The PDE4 family is the largest PDE family in the mammalian PDE superfamily and encompasses four genes, PDE4A–D (40). In the current microarray study, PDE4B and PDE4D were up-regulated 3.2- and 1.85-fold, respectively, by cAMP in hESF_nonendo but not hESF_endo (supplemental Tables S1 and S2), although QRT-PCR validation did not confirm these differences (Fig. 4A). Our microarray analysis also identified PDE8B, a cAMP-specific PDE, whose transcript decreased significantly (−4.0-fold) in hESF_nonendo, but not hESF_endo, after treatment with cAMP. Moreover, levels of PDE8B by microarray analysis were significantly higher in hESF_endo compared with hESF_nonendo, both untreated and treated with cAMP (2.2- and 2.41-fold, respectively, groups C and D). The analysis of PDE8B transcripts levels by QRT-PCR between cAMP-treated hESF_nonendo and hESF_endo did not detect a statistically significant difference but did confirm the trend observed by microarray analysis (Fig. 4A).

Figure 4.

Analysis of PDE4 and PDE8 in hESF_nonendo and hESF_endo. A, Validation by real-time QPCR expression of PDE8B, PDE4B, and PDE4D mRNAs. All hESFs were treated with or without 0.5 mm cAMP for 96 h. B, Total and IBMX-insensitive cAMP PDE activity in hESFs was determined as described in Materials and Methods. Shown is the cAMP PDE activity of hESF lysates (n = 4 with and n = 3 without endometriosis) using 100 nm [³H] cAMP as substrate in the absence of added inhibitors (Total) or in the presence of 100 μm IBMX to measure the apparent PDE8 component (+IBMX). cAMP-PDE activity is expressed as picomoles of cAMP hydrolyzed per minute per milligram of total cell protein. Error bars, ±sem. C, PDE4 activity in hESF lysates. The PDE4 cAMP PDE activity in hESFs was determined as described in Materials and Methods by calculating the activity that was sensitive to 1 μm of the PDE4 selective inhibitor rolipram. The data represent n = 3 hESF_nonendo or hESF_endo assayed in duplicate. Error bars, ±sem. No endo, hESF_nonendo; endo, hESF_endo. *, Significant difference (P ≤ 0.05) compared with respective control. FC, Fold change.

To investigate whether the observed microarray analysis changes in the PDE4 and PDE8 families, translated into significant functional effects, we investigated the cAMP PDE activity of cultured hESF_nonendo and hESF_endo. The contribution of PDE4 to the cAMP hydrolytic activity in hESF_nonendo and hESF_endo was delineated by the use of rolipram, a potent and selective inhibitor of all PDE4 isoenzymes. Because there is no available specific PDE8 inhibitor available, the PDE8 component was measured in the presence of IBMX. IBMX is a nonselective PDE inhibitor, which inhibits all families of cAMP-hydrolyzing PDEs except PDE8 (25,26); therefore, the cAMP hydrolytic activity that remained in the presence of IBMX was taken as the putative PDE8 component. There were no observed changes in the putative PDE8 component identified as the cAMP hydrolytic activity resistant to IBMX when comparing nondecidualized hESF_endo vs. hESF_nonendo or when comparing cAMP-decidualized hESFs from the same subjects (Fig. 4B). Also, no changes between total cAMP hydrolytic activities were observed between nondecidualized hESF_nonendo and hESF_endo as well as between cAMP-decidualized hESF_nonendo and hESF_endo (Fig. 4B). Furthermore, no significant difference in PDE4 activity in lysates of hESF_endo vs. hESF_nonendo was observed (Fig. 4C). Decidualization with 8-Br-cAMP produced a robust 6.7-fold increase in PDE4 activity in both hESF_nonendo and hESF_endo, with no difference between them (Fig. 4C).

Discussion

General comments

The current study presents two important concepts. The first focuses on genes, pathways, and biological processes activated or repressed in response to the decidualizing stimulus/activation of the PKA pathway in hESF_nonendo. The second is the difference in response to the decidualizing stimulus/activation of the PKA pathway of hESF_endo compared with hESF_nonendo.

In general, our data support evidence reported by us and others regarding decreased expression of cell cycle-regulating genes on decidualization in vivo and in vitro in hESFs from women without endometriosis (8,15,17,41,42,43,44). This is further supported by early studies on histology and mitotic figures and rates of DNA synthesis in secretory endometrium (45). The cell cytoskeleton and extracellular matrix genes, such as actins, fibulins, kinesins, myosins, and MMPs, were significantly regulated in cAMP-treated hESF_nonendo, consistent with the characteristic changes in endometrial stromal cell morphology during the decidualization process (46). Characterization of global gene expression in response to cAMP confirmed earlier findings of our group using a more limited microarray platform (15). Importantly, in the current study, we identified new genes and pathways that are important for the process of decidualization, including vascular endothelial growth factor, Notch, tight junction, G protein-coupled receptor signaling, axonal guidance signaling, and others (supplemental Table S4).

Differentiation vs. proliferation in hESF_nonendo vs. hESF_endo

Overall, our study demonstrated that hESF_endo displays a blunted response to cAMP/PKA pathway activation with a decrease in the total number of genes regulated, compared with hESF_nonendo. Subsequently hESF_endo demonstrated a diminished differentiation capacity, reflected in decreased expression of decidualization markers, cytoskeletal components, matrix-degrading enzymes, immune genes, and key signaling pathways involved in the differentiation process, known and revealed herein.

In a comprehensive microarray analysis of endometrial tissue biopsies from women with and without endometriosis, we demonstrated that there is a delayed transition from the proliferative to secretory phase, with persistent expression of genes involved in mitosis and proliferation in early secretory endometrium (ESE) of women with endometriosis but not in those without disease (7). Even though ESE samples clustered according to their cycle phase, genes involved in cell proliferation maintained a fingerprint consistent with the proliferative phase (7). Johnson et al. (47) reported that eutopic endometrium from women with endometriosis demonstrated increased proliferation and decreased apoptosis, compared with endometrial tissue from women without disease (47). These observations are consistent with the present microarray analysis and validated by QRT-PCR (Fig. 3 and supplemental tables). Taken together, the data strongly support an abnormal transition from the estrogen-dominant proliferative phase to the progesterone (and PKA)-dominant secretory phase in endometrium in the setting of endometriosis (47,48,49).

Because cell cycle progression is inhibited in the hESF_nonendo in response to cAMP, with differentiation/decidualization biomarkers being up-regulated, the data support decreased proliferation and increased differentiation as hallmarks of the endometrial stromal fibroblast in the secretory phase of menstrual cycle. In addition, the observed blunted response of hESF_endo to cAMP inhibition of proliferation and induction of decidualization biomarkers may account for the increased survival of endometrial cells and the persistent proliferative phenotype of endometrial tissue during the proliferative-secretory transition in women with vs. without disease (7). Interestingly, the cell cycle pathway in human endometrium may be regulated in part by microRNAs (miRNAs), and recently we demonstrated that of seven miRNAs significantly down-regulated in women with vs. without endometriosis, five are involved in regulation of cell cycle (50). These observations support increased transitioning through the cell cycle in endometrium from women with endometriosis and that miRNAs are potential mediators in the delayed proliferative to secretory transition of endometrium from women with endometriosis (7).

cAMP hydrolysis

Because 8-Br-cAMP was used to induce decidualization, a G protein-coupled-receptor/G protein/cyclase defect in hESF_endo has been excluded (40). Thus, hydrolysis of 8-Br-cAMP by PDEs (40,51) was considered as a possibility for the blunted response to cAMP treatment in hESF_endo vs. hESF_nonendo. Of the 11 mammalian PDE families (PDEs1–11), PDE4, PDE7, and PDE8 are the cAMP-specific PDEs (26). Hydrolysis of cyclic nucleotides by the PDEs provides the major cellular mechanism for the dampening of cellular cyclic nucleotide signaling (26,52). Our results demonstrate that cAMP treatment of hESF_nonendo and hESF_endo leads to a robust induction of PDE4 activity, confirming that intracellular cAMP is involved in regulating its own synthesis and degradation (53,54). Despite changes in PDE4 and PDE8 mRNA transcripts detected in our microarray studies of hESF_endo vs. hESF_nonendo, no apparent changes were observed in functional characterization of these cAMP-PDE activities in these cells under the conditions tested. It should be noted that the PDE8 family is poorly understood and no specific inhibitor exists for this enzyme family. Therefore, our functional evidence for assessing PDE8 activity is based on the noted insensitivity of this PDE to the PDE inhibitor IBMX (25,26). Because the cAMP PDE activities in hESF_nonendo and hESF_endo were not significantly different, we conclude that hESF_nonendo and hESF_endo do not have dysregulated cAMP homeostasis. However, with dysregulation of the cAMP/PKA pathway in hESF_endo not being attributed to altered hydrolysis of cAMP, there may still be a difference in the expression of downstream mediators, such as cAMP response element modulator (supplemental Tables S1 and S2). Because PDEs are regulated by receptors coupled to cAMP (55,56), the fact that the observed abnormality in the PKA signaling pathway in women with vs. without disease did not affect PDE4 induction suggests that there are differences in the sensitivity of different genes to cAMP/PKA activation or that there are specific coregulators that alter transcription of a subgroup of genes (e.g. PDE4D vs. IGFBP1 and CCND1). This characterization represents the first comparison of cAMP hydrolysis/PDE activity in eutopic endometrium from women with and without endometriosis.

Other pathway involvement in the pathogenesis of endometriosis

Of note is the involvement of the Wnt pathway in the pathogenesis of endometriosis (44,57). The data herein further support this and provide important insight into these observations. For example, the Wnt inhibitor secreted frizzled related protein (SFRP)-1 is highly expressed in endometrial stroma in the proliferative phase, compared with the secretory phase, in normal endometrium (57,58) as well as in endometriotic tissue compared with eutopic endometrium, and direct regulation of SFRP1 by estrogen has been suggested (57). Transcripts for both Wnt inhibitors, Dickkopf-1 and SFRP1, were decreased in untreated hESF_endo in the present study, suggesting a potential contribution of the Wnt pathway to development and/or maintenance of endometriosis with its persistent proliferative signature that is amplified in response to steroid hormones and activation of the PKA pathway.

Dysregulation of immune genes in the setting of endometriosis is well documented (7,58). In the present study, IL13RA2 was the most highly up-regulated gene in hESF_nonendo but not hESF_endo in response to cAMP. The function of IL13RA2 in endometrial physiology or pathology is largely unknown. Chen et al. (59) demonstrated that IL13RA2 is cleaved and solubilized in vitro by MMP8, which was increased in our array data 15.8-fold on decidualization of hESF_nonendo, suggesting a contribution of that mechanism to the normal decidualization process in human endometrium.

Many genes involved in cell shape and cytoskeleton functioning were decreased in hESF_nonendo on decidualization; whereas these changes were blunted in hESF_endo. Collagens VI are present in endometrial stroma in proliferative and ESE as a component of the intercellular collagenous network, with a decrease in expression during the midsecretory phase (Refs. 60 and 61 and herein). No alteration in collagen, smooth muscle actin, fibronectin, kinesins, and other ECM molecule expression in hESF_endo on cAMP treatment is consistent with impaired decidualization of hESF from women with endometriosis, which is accompanied by impaired morphological transformation to the decidualized phenotype. Earlier studies demonstrated down-regulation of tenascin C by P₄ in the stroma of normal endometrium and dysregulation of it in endometriosis (62). Interestingly, tenascin C was one of the most highly down-regulated genes (0.16-fold) in the proliferative-to-early secretory transition in women without endometriosis (8), and whether that is due to progesterone and/or activation of the PKA pathway in hESFs is uncertain. Studies of tenascin C up-regulation in tumor ECM by epithelial growth factor and down-regulation by steroid hormones (63) underscore the complexity of regulation of this ECM glycoprotein.

Summary and conclusions

The results presented herein demonstrate that eutopic endometrial stromal fibroblasts from women with endometriosis differ from those obtained from women without disease. These differences lie in processes that are related to cell survival, cell proliferation, neoangiogenesis, immune function, cell shape, and ECM modification that may account for why a subgroup of women develops endometriosis, despite most women experiencing retrograde menstruation (64,65). The data support the postulate that in endometriosis, eutopic endometrial cells with increased proliferative potential, as well as decreased immunogenicity, when seeded into the pelvic cavity after retrograde menstruation, are endowed with properties to establish endometriotic lesions. Altered immune response, impaired ECM degradation, cell cytoskeleton changes, and establishment of a blood supply and accompanying innervation likely promote further survival and growth of transplanted/implanted stromal cells from the eutopic endometrium.

From a clinical perspective, in view of the role of decidual cells in tissue hemostasis (66,67), an impaired decidual response of hESFs may also lead to premenstrual uterine spotting in patients with endometriosis, although this remains to be determined.

The current and earlier studies (13,16,68) are convincing in that whereas the endometrium from women with endometriosis exhibits a molecular signature of P₄ resistance, there is an apparent impairment of the PKA pathway in hESFs. The observations that the decidualization process involves activation of the PKA- and progesterone receptor-dependent signaling pathways that interface with and are influenced by a multitude of other pathways are important in understanding basic mechanisms underlying differentiation of this tissue in response to steroid hormones. Likewise, investigation into the effects of progesterone on hESF_endo vs. hESF_nonendo is currently underway in our laboratory. Furthermore and of equal importance, abnormal signaling pathways in endometrium (described herein and Refs. 27,33 and 68) and endometriosis lesions (9,69) of women with disease offer new opportunities to develop novel, targeted therapies to treat endometriosis-related pain and/or infertility.