Inflammatory Status Influences Aromatase and Steroid Receptor Expression in Endometriosis

Abstract

Aberrant up-regulation of aromatase in eutopic endometrium and implants from women with endometriosis has been reported. Aromatase induction may be mediated by increased cyclooxygenase-2 (COX-2). Recently, we demonstrated that progesterone receptor (PR)-A and PR-B serve an antiinflammatory role in the uterus by antagonizing nuclear factor κB activation and COX-2 expression. PR-C, which antagonizes PR-B, is up-regulated by inflammation. Although estrogen receptor α (ERα) is implicated in endometriosis, an antiinflammatory role of ERβ has been suggested. We examined stage-specific expression of aromatase, COX-2, ER, and PR isoform expression in eutopic endometrium, implants, peritoneum, and endometrioma samples from endometriosis patients. Endometrial and peritoneal biopsies were obtained from unaffected women and those with fibroids. Aromatase expression in eutopic endometrium from endometriosis patients was significantly increased compared with controls. Aromatase expression in endometriosis implants was markedly increased compared with eutopic endometrium. Aromatase mRNA levels were increased significantly in red implants relative to black implants and endometrioma cyst capsule. Moreover, COX-2 expression was increased in implants and in eutopic endometrium of women with endometriosis as compared with control endometrium. As observed for aromatase mRNA, the highest levels of COX-2 mRNA were found in red implants. The ratio of ERβ/ERα mRNA was significantly elevated in endometriomas compared with endometriosis implants and eutopic endometrium. Expression of PR-C mRNA relative to PR-A and PR-B mRNA was significantly increased in endometriomas compared with eutopic and control endometrium. PR-A protein was barely detectable in endometriomas. Thus, whereas PR-C may enhance disease progression, up-regulation of ERβ may play an antiinflammatory and opposing role.

ENDOMETRIOSIS IS AN estrogen-dependent disease associated with enhanced aromatase expression and local estrogen production in endometriotic tissues (1,2,3). Although retrograde menstruation with subsequent implantation and growth of endometrial cells within the peritoneal cavity is a widely accepted mechanism, multiple lines of evidence suggest that inflammation plays a critical role in the pathogenesis of this disease (4). This is supported by the fact that implanted endometriotic cells and intraperitoneal leukocytes produce proinflammatory cytokines creating a feed-forward regulatory loop in the development and progression of endometriosis (5).

Estrogen synthesis from C₁₉ steroids is catalyzed by aromatase P450, product of the aromatase/CYP19 gene. Human CYP19 is a single-copy gene expressed in a number of tissues, including placenta (6), gonads (7,8), discrete nuclei of brain (9), adipose stromal cells (10,11), and in breast cancer epithelial and stromal cells (12). Expression of aromatase in various tissues is controlled by tissue-specific promoters that lie upstream of tissue-specific first exons encoding the 5′-untranslated regions (UTRs) of aromatase mRNAs. These 5′-UTRs are spliced onto a common junction 38 bp upstream of the translation start site. Thus, the sequence encoding the aromatase P450 protein in each of these tissues is identical (13,14). The alternative use of promoters comprises the basis for differential tissue-specific regulation of aromatase expression by various hormones, growth factors, and cytokines.

Aberrant aromatase induction both in eutopic endometrium and endometriosis implants has been reported to occur predominantly via activation of the strong CYP19 promoter that is active in ovarian granulosa and luteal cells, promoter IIa (15,16,17). Aromatase expression in endometriosis cells is induced via the cyclooxygenase (COX) type 2 (COX-2)-prostaglandin (PG)E₂ pathway, resulting in increased cAMP formation (18). Furthermore, the finding that estradiol-17β (E₂) stimulates COX-2 expression suggests that E₂ and COX-2 exist in a positive feedback loop (19,20).

E₂ plays an important role in controlling the expression of genes involved in wide variety of biological, inflammatory, and neoplastic processes (21). Most biological effects of estrogens are mediated through two distinct and functional estrogen receptors (ER), ERα (22) and ERβ (23). ERα is the dominant receptor in the adult uterus and the major mediator of estrogenic effects [i.e. stimulation of proliferation and induction of progesterone receptor (PR) expression]. Conversely, ERβ has been postulated to oppose the inflammatory and proliferative actions of ERα (24,25). The expression of ERα mRNA was reported to be significantly higher than ERβ in endometriotic lesions and in eutopic endometrium (26).

PR exists as three major isoforms: PR-A (94 kDa), PR-B (114 kDa), and PR-C (60 kDa) (27,28). PR-B and PR-A both bind to progesterone response elements (PREs) in DNA. PR-A lacks one of three transcriptional activation domains that are present in PR-B (29). PR-C is an N-terminally truncated form of PR, which lacks part of the DNA-binding domain and two activation (AF) domains near the amino terminus but contains the ligand-binding domain and nuclear localization signal (28). In endometriosis, preferential expression of PR-A relative to PR-B has been reported (30).

Although many studies have been conducted regarding inflammation and endometriosis, its potential association with aromatase expression and differential expression of ER and PR isoforms has not been determined. The aim of this study was to examine the expression of aromatase and ER and PR isoforms in endometriosis and assess whether these gene products are affected by inflammatory status, as determined by morphological evaluation (red vs. black implants vs. ovarian endometriomas), as well as COX-2 expression.

Materials and Methods

Patients and tissue collection

This is a prospective cross-sectional case-control study of human tissue samples approved by the Institutional Review Board of The University of Texas Southwestern Medical Center at Dallas. All patients included in the study provided informed consent.

The following inclusion criteria were used: age more than 18 yr and no more than 40 yr at the time of the surgical procedure, presence of regular menstrual cycles with the exception of those treated with depot progestin for endometriosis or birth control, absence of any discrete uterine fibroids, and absence of any evidence of past or recent pelvic inflammatory disease. Moreover, patients currently receiving progestin treatment were eligible only if information regarding their medications could be retrieved and recorded. Women with any chronic inflammatory disorders such as rheumatoid arthritis, systemic lupus erythematosus, inflammatory bowel disease, and asthma were excluded. In patients undergoing an elective operative or diagnostic laparoscopy, the diagnosis and staging of endometriosis along with the morphology of the biopsied peritoneal specimens were documented by digital photographs taken during the surgery. The diagnosis of endometriosis was confirmed histologically.

Tissues were obtained from women (n = 14) who had histologically documented endometriosis of various stages according to the revised American Society for Reproductive Medicine criteria (31). During an operative or diagnostic laparoscopy for indications such as pelvic pain, infertility, and adnexal mass consistent with ovarian endometrioma, simultaneous sampling of endometrium (eutopic endometrium), endometriosis implants, ovarian endometrioma capsule, and visually normal peritoneum were carried out. The morphology of the implants (red vs. black) and the stage of the endometriosis for each case were confirmed by digital photographs taken during the surgical procedure. Two of the patients with severe endometriosis had previous hysterectomies; therefore, no eutopic endometrium could be obtained for these cases. The controls (n = 8) consisted of women undergoing laparoscopic tubal ligation or diagnostic laparoscopy with no pelvic findings of endometriosis, inflammatory disease, or uterine fibroids. These women underwent sampling of endometrium and peritoneum.

Endometrium samples from reproductive-age women who underwent hysterectomy for uterine fibroids (n = 7) without any evidence of adenomyosis, endometriosis, adnexal mass, or pelvic inflammatory disease were also included. These endometrial samples served as additional controls for endometrial expression of various mRNA transcripts and proteins. All diagnoses and endometrial phases were verified by pathology reports. The distribution of patient population and tissue samples are shown in Table 1.

Table 1.

Tissues of origin collected for the study

Parameter	Endometriosis (n = 14)	Controls (n = 8)	Endometrium from patients with uterine fibroids (n = 7)
Age range (yr)	28–40	23–37	33–42
Menstrual phase: proliferative/secretory/medical suppressiona	7/2/5	4/1/3	3/4/0
Endometrial biopsy	12	8	7
Endometriosis implant biopsy	16
Endometrioma capsule	4
Peritoneal biopsy	10	7

^aD-MPA. 

A portion of all prospectively collected samples was placed in formalin fixative, and the remainder of the tissues were kept in RNAlater (Ambion Inc., Austin, TX) solution and stored at −80 C for future analysis. Furthermore, if submitted by the primary surgeon, all pathology reports of resected tissues were obtained, and the diagnoses were verified.

Histology and immunohistochemistry

All specimens fixed in 10% paraformaldehyde solution were embedded in paraffin blocks. Sections were stained with hematoxylin and eosin for histological evaluation of the biopsied tissues. This confirmed the histological diagnoses and the phase of endometrium as proliferative or secretory. Paraffin sections also were subjected to immunohistochemical analysis for aromatase, PR, and ERβ. All immunohistochemistry photomicrographs were taken at ×156.2 magnification with 50% digital zoom.

Aromatase P450.

Sections (8 μm) were cut from paraffin-embedded tissues and mounted on silane-coated slides. Sections were deparaffinized and rehydrated in decreasing concentrations of ethanol, washed in PBS (0.01 m, pH 7.2), and incubated in hydrogen peroxide in methanol (3%, vol/vol) for 30 min to block endogenous peroxidase. After washing in PBS, the sections were incubated in normal goat serum diluted in 2% PBS for 20 min at room temperature. For aromatase P450 immunostaining, the tissues were incubated with a rabbit antihuman aromatase polyclonal antibody (Biovision Inc., Mountain View, CA) diluted 1:50 in PBS overnight at 4 C. A Vector Nova Red Detection Kit (Vector Laboratories, Burlingame, CA) was employed to identify immunoreactivity (Vectostain Elite ABC kit; Vector). The immunoreactive proteins appeared as a red end-product. Slides were counterstained with hematoxylin (blue). Sections of human placenta were used as a positive control, and a section of uterine visceral peritoneum was used as negative control.

ERβ.

ERβ immunostaining was performed by the University of Texas Southwestern Pathology Immunohistochemistry Laboratory using a mouse monoclonal ERβ antibody (clone 14C8; GeneTex Inc., San Antonio, TX). All immunostaining was performed at room temperature on a BenchMarkXT automated immunostainer using the UltraVIEW systems with horseradish peroxidase and diaminobenzidine (DAB) chromogen (Ventana Medical Systems, Tucson, AZ). Optimal antibody dilutions were predetermined using human prostate as a positive control. Human prostate sections were included in each immunostaining procedure to assure appropriate staining.

Three-micron sections were mounted on positively charged glass slides and air dried overnight. Sections were then placed onto a BenchMarkXT where the deparaffinization and heat retrieval were performed. Sections were then incubated for 1 h with either primary antibody (1:50 vol/vol) diluted in ChemMate buffer (Ventana Medical Systems) or with buffer alone as a negative reagent control. After washing in buffer, sections were incubated with a freshly prepared mixture of DAB and hydrogen peroxide in buffer for 8 min, followed by washing buffer and water. Sections were then counterstained with hematoxylin and eosin, dehydrated in a graded series of ethanol and xylene, and coverslipped. Slides were reviewed by light microscopy. Positive reactions with DAB were identified as a dark brown reaction product.

PR isoforms.

PR-A, PR-B, and total PR immunostaining was performed as described for ERβ. For detection of PR-A, a human PR-A-specific mouse monoclonal antibody purchased from Novocastra (NCL-L-PGR-312; Vision BioSystems Inc., Norwell, MA) was used (32,33). For PR-B immunostaining, mouse monoclonal antibody Ab-6 (NeoMarkers, Fremont, CA) was used (32,34). Total PR immunostaining was performed using the rabbit polyclonal antibody sc-539 (C-20; Santa Cruz, Biotechnology, Santa Cruz, CA). For PR-A and PR-B immunostaining, the sections were pretreated with Target Retrieval Solution using a modified pressure cooker (Dako Co., Carpinteria CA). Optimal primary antibody dilutions were predetermined using known positive control tissues (e.g. fresh endomyometrium sections from hysterectomy specimens, in which myometrium is negative for PR-A, whereas endometrial stroma is positive). The dilutions (vol/vol) used for PR-A, PR-B, and total PR were 1:200, 1:50, and 1:100, respectively. Immunoreactive products were identified with DAB and H₂O₂, as above. Sections were counterstained with hematoxylin.

Quantitative real-time RT-PCR (qRT-PCR)

Total RNA from tissue samples stored in RNAlater solution was extracted by the one-step method of Chomczynski and Sacchi (35) using TRIzol reagent (Invitrogen, Carlsbad, CA). The isolated RNA was quantified by measuring the OD of the samples at a wavelength of 260 nm. The quality of RNA was ascertained by the presence of ratios between 1.6 and 2.0 at 260/280 nm. RNA was treated with deoxyribonuclease to remove any contaminating DNA, and then 4 μg RNA was reverse-transcribed using random primers and Superscript II RNaseH-reverse transcriptase (Invitrogen). The relative abundance of each mRNA product in the tissue samples was determined by qRT-PCR using a modification of previously published methods (36).

Primer sets directed against human CYP19 exons IIa, I.3, I.4, I.1, ERα, ERβ, PR, COX-2, and h36B4 (for normalization) mRNA transcripts were designed using Primer Express software (PE Applied Biosystems, Foster City, CA) based on published sequences for these mRNAs (Table 2). The unique, tissue-specific UTRs in hCYP19 transcripts were demonstrated in our laboratory to be selectively amplified using the well-validated primer sets used in this study (37).

Table 2.

Primers used in the Q-PCR analysis

Gene	Primer (5′–3′)	Reference No.
PR-B
Forward	ACA CCT TGC CTG AAG TTT CG	NM000926
Reverse	CTG TCC TTT TCTGGG GGA CT
PR-AB
Forward	GAG GAT AGC TCT GAG TCC GAG GA	NM000926
Reverse	TTT GCC CTT CAG AAG CGG
PR-ABC
Forward	TCA GTG GGC AGA TGC TGT ATT T	NM000926
Reverse	GCC ACA TGG TAA GGC ATA ATG A
H36B4
Forward	TGC ATC AGT ACC CCA TTC TAT CA	XR017813
Reverse	AAG GTG TAA TCC GTC TCC ACA GA
CYP19I.1
Forward	ACG GAA GGT CCT GTG CTC G	NM000103
Reverse	GTA TCG GGT TCA GCA TTT CCA
CYP19I.4
Forward	CTG ACA GGA GGT CCC TGG C	NM031226
Reverse	CGG GTT CAG CAT TTC CAA AA
CYP19 I.3
Forward	CAC TCT ACC CAC TCA AGG GCA	M74714
Reverse	TTG GCT TGA ATT GCA GCA TTT
CYP19 IIa
Forward	CAG GAG CTA TAG ATG AAC CTT TTA GGG	S85356
Reverse	CTT GTG TTC CTT GAC CTC AGA GG
COX-2
Forward	TTC CAG ATC CAG AGC TCA TTA AA	AY462100
Reverse	CCG GAG CGG GAA GAA CT
ERα
Forward	AGA GAA GTA TTC AAG GAC ATA ACG ACT ATA T	NM00125
Reverse	TCT TCC TCC TGT TTT TAT CAA TGG
ERβ
Forward	AAG TTG GCC GAC AAG GAG TT	NM001040276
Reverse	ACA GGC TGA GCT CCA CAA AG

To delineate PR isoform expression, the first human primer set, termed PR-B (38), was designed to amplify sequences specific for PR-B (upstream of the second AUG translation initiation site), whereas the second human primer set, PR-AB, was designed to amplify sequences downstream of the second AUG translation initiation site. Finally, the human primer set for PR-ABC (39) was designed to amplify sequences within the ligand-binding domain of the PR, a region common to all PR isoforms. None of these primer sets corresponded to sequences in any of the other steroid hormone receptors.

For the quantitative analysis of mRNA expression, the ABI Prism 7700 Detection System (Applied Biosystems) was employed using the DNA binding dye SYBR Green (PE Applied Biosystems) for detection of PCR products. Thermocycling was done in a final volume of 10 μl containing 2 μl cDNA sample, 0.6 ml primer, 3.4 ml sterile water, and 6 μl SYBR Green I. The cycling conditions were 50 C for 2 min and 95 C for 10 min, followed by 40 cycles at 95 C for 15 sec and 60 C for 1 min. The cycle threshold was set at a level where the exponential increase in PCR amplification was approximately parallel among all samples. All primer sets produced amplicons of the expected size and sequence.

We calculated the relative fold changes using the comparative cycle times (Ct) method with human ribosomal protein h36B4 mRNA as the reference guide. Over a wide range of known cDNA concentrations, PR primer sets were demonstrated to have good linear correlation (slope = −3.4) and equal priming efficiency for the different dilutions compared with their Ct values (data not shown). Given that all PR primer sets had equal priming efficiency, the ΔCt values (PR primer − internal control) for each PR primer set were calibrated to the samples with the lowest PR abundance (highest Ct value), and the relative abundance of each primer set compared with calibrator was determined by the formula, 2^−ΔΔCt, whereby ΔΔCt is the calibrated Ct value. The relative abundance of PR-A could then be calculated by subtracting the relative abundance of PR-B from that of PR-AB, whereas the relative amount of PR-C was calculated by subtracting the relative abundance of PR-AB from PR-ABC.

Immunoblotting

Immunoblots were performed on select groups of tissue sets to verify the presence of corresponding proteins for aromatase and PR isoforms tested. Tissue lysates and nuclear extracts prepared as described (40) were electrophoresed using a precast Novex gel electrophoresis system with 3–8% Tris acetate gels or 4–12% Bis-Tris gels (Invitrogen). Proteins were then electrophoretically transferred onto polyvinylidene fluoride membranes, which were incubated for 1 h at room temperature with rabbit antibodies directed against human aromatase P450 (1:500) (Biovision) and PR (1:500) (Santa Cruz Biotechnology; sc-539/C-20). Membranes were incubated with horseradish peroxidase-conjugated antirabbit IgG secondary antibodies, and immunoreactive bands were visualized for aromatase (58 kDa), PR-B (120 kDa), and PR-A (94 kDa).

Statistics

qRT-PCR arbitrary values were expressed as mean ± sem. The nonparametric tests used for comparisons included Kruskal-Wallis test, median test, and the Mann-Whitney U test. Pearson correlation was used to investigate the potential associations among various parameters tested in pooled results for endometrium or endometriosis implants. All tests were two sided with a significance level of P < 0.05.

Results

Aromatase expression is increased in eutopic endometrium of women with endometriosis

In all tissue sets, aromatase mRNA transcripts predominantly contained the gonad-specific exon IIa at their 5′-ends. Adipose- and placenta-specific mRNA transcripts were not detectable, and cancer-related exon I.3-containing mRNA transcripts were relatively low. In general, expression of exon I.3 transcripts paralleled those of the exon IIa-containing transcripts (data not shown).

Aromatase expression was increased significantly in eutopic endometrium of endometriosis patients compared with control endometrium or endometrium of patients with fibroids (Fig. 1A, P = 0.029 by median test). Endometrial expression of aromatase was examined further in subgroups according to phases of the menstrual cycle or suppressive treatment [three control women and three women with endometriosis receiving depot-medroxyprogesterone acetate (D-MPA)]. In proliferative phase, eutopic endometrium of endometriosis patients showed higher aromatase expression compared with luteal phase (Fig. 1B). This finding suggested that aromatase expression in endometrium of endometriosis patients was affected by progesterone, although differences between phases of the menstrual cycle did not reach statistical significance. In endometriosis patients receiving D-MPA, aromatase expression in eutopic endometrium was comparable to secretory-phase expression levels of control subjects.

Figure 1.

Aromatase/CYP19 mRNA transcripts containing untranslated exon IIa (CYP19 IIa) are up-regulated in eutopic endometrium of women with endometriosis, reaching highest levels in the proliferative phase of the menstrual cycle. Aromatase expression is manyfold higher in endometriosis implants than in eutopic endometrium among women with minimal-mild (I-II) and moderate-severe (III-IV) stage disease. A, CYP19 IIa mRNA levels were analyzed in endometrium from unaffected women (control), women with endometriosis, and those with uterine fibroids using qRT-PCR. Relative levels of CYP19 IIa mRNA were calculated by normalizing against h36B4 mRNA. Data, expressed as arbitrary units, are the mean ± sem of values from eight control samples, 12 eutopic endometrial samples, and seven fibroid samples. *, Significantly different from the others at P = 0.029. B, CYP19 IIa mRNA levels were analyzed in endometrium from unaffected women (control), women with endometriosis, and those with uterine fibroids at proliferative and secretory phases of the menstrual cycle and after medical suppression with D-MPA treatment. Relative levels of CYP19 IIa mRNA were calculated by normalizing against h36B4 mRNA. Data, expressed as arbitrary units, are the mean ± sem of values from the number of subjects indicated on the abscissa. C, CYP19 IIa mRNA levels were analyzed in eutopic endometrium and endometriosis implants using qRT-PCR, as described above. The ratios of mRNA levels in implants relative to eutopic endometrium are plotted according to the endometriosis revised American Society of Reproductive Medicine stage. Data are the mean ± sem of values from the number of subjects indicated on the abscissa.

The mean arbitrary values ± sem for aromatase expression in eutopic endometrium of patients with minimal to mild (stage I-II; n = 6) and moderate to severe (stage III-IV; n = 6) endometriosis were 525.5 ± 455.4 and 593.2 ± 513.3, respectively, which were comparable (data not shown). This finding indicates that increased aromatase expression in eutopic endometrium of women with endometriosis does not vary with advanced-stage disease.

Aromatase mRNA expression is further up-regulated in peritoneal implants and in endometrioma

Expression of aromatase also was studied in peritoneal implants of women with endometriosis. Aromatase expression in endometriosis implants was manyfold higher than that of the eutopic endometrium samples. This is reflected in the calculation of the ratios of aromatase mRNA expression in endometriosis implant/eutopic endometrium by stage of the disease (Fig. 1C). Although aromatase expression was found to be up-regulated in all endometriosis tissues, the magnitude of up-regulation varied considerably according to the visual appearance of the peritoneal implants (e.g. red vs. black) and ovarian endometrioma capsules (Fig. 2A, median test P = 0.014). Aromatase mRNA levels were increased significantly in red implants relative to black implants and endometrioma cyst capsule. Aromatase expression in the implants did not differ according to the phase of the cycle (Fig. 2B). Although the implants showed suppressed aromatase expression in women treated with D-MPA, the mean arbitrary value was still much higher than that observed in the corresponding eutopic endometrium (621.4 ± 367.4 vs. 116.3 ± 113.8, respectively). These data suggest a limited effect of progesterone in endometriosis implants compared with eutopic endometrium.

Figure 2.

Aromatase expression is higher in red endometriosis implants than in black implants and endometriomas, does not change according to menstrual phase, and is suppressed by D-MPA treatment. Visually normal peritoneum exhibits increased aromatase expression in endometriosis. A, Gonad-specific exon IIa-containing (CYP19 IIa) mRNA levels were analyzed in red and black implants and in endometrioma cyst wall determined to be free of ovarian tissue. B, CYP19 IIa mRNA levels were analyzed in endometriosis implants according to the menstrual cycle phase or with D-MPA treatment. C, CYP19 IIa mRNA levels were analyzed in biopsies of visually normal peritoneum from unaffected subjects (control) and from women with stage I-II and stage III-IV disease. Control value was 52.4 ± 16.5 arbitrary units, which cannot be visualized on this scale. For all tissues studied, CYP19 IIa mRNA was analyzed by qRT-PCR. Relative levels of CYP19 IIa mRNA were calculated by normalizing against h36B4 mRNA. Data, expressed as arbitrary units, are the mean ± sem of values from the number of subjects indicated on the abscissa.

Aromatase expression in biopsies of peritoneum that were visually and histologically free of endometriosis was higher in samples from patients with moderate to severe endometriosis compared with those from women with minimal to mild disease (Fig. 2C, P > 0.05).

Aromatase protein levels also are increased in eutopic endometrium and in peritoneal and ovarian implants from women with endometriosis

Immunoblotting was performed on a eutopic endometrium (proliferative phase) and a red endometriosis implant sample from a patient with minimal-mild (stage I-II) endometriosis and on three control proliferative phase endometrium samples from patients without endometriosis. Aromatase P450 protein was highly expressed in both the eutopic endometrium and the red endometriosis implant, whereas aromatase was not detected in control endometrium (Fig. 3A).

Figure 3.

Aromatase protein expression is up-regulated in eutopic endometrium and implants from women with endometriosis. A, Immunoblot of aromatase protein in three samples of control endometrial biopsies (EMB) (3635, 3712, and 3352) and in eutopic endometrium (EU) and a red implant (IMP) from a patient with stage I endometriosis. C–E, Aromatase immunostaining of red implant (C), eutopic endometrium (D), and black implant (E) from a patient with dysmenorrhea and stage I endometriosis. B, Nonimmune IgG control staining of the red implant section in C, Note immunostaining (dark red) of both stromal and epithelial cells in red and black endometriosis implants. Eutopic endometrium in late proliferative phase (D) shows intense epithelial staining, whereas the immunostaining in stroma is less intense. Aromatase immunostaining of a eutopic endometrial sample in secretory phase (F) shows lack of immunostaining in the epithelial cells with positive staining in the stroma. In an endometrioma cyst wall (G), diffuse immunostaining (dark red) of the stromal and epithelial cells can be seen. In all of these sections, epithelial (glandular) elements are indicated by a thick blue arrow, and stromal cells are indicated by a thin green arrow. Normal uterine visceral peritoneal sample (H) stained for aromatase serves as a negative control. Human placental sections were used as a positive control (G) (thick blue arrow indicates the syncytiotrophoblast). Aromatase immunostaining is indicated by the dark red color, and counterstaining with hematoxylin is in blue.

Immunohistochemistry of a proliferative-phase eutopic endometrium sample from a patient with minimal to mild endometriosis demonstrated intense aromatase immunostaining in the cytoplasm of glandular (epithelial) cells, whereas the stromal immunostaining was faint (Fig. 3D). Red (Fig. 3C) and black (Fig. 3E) endometriosis implants from the same patient demonstrated diffuse cytoplasmic immunostaining both in stromal and glandular elements. In the tissue sections presented, aromatase immunostaining was somewhat more intense in the red implant than the black implant (Fig. 3, C and E). Eutopic endometrium from secretory phase showed diminished immunostaining in glandular cells, whereas stromal immunostaining was still evident (Fig. 3F).

In an endometrioma section, diffuse immunostaining of stroma with more intense staining of epithelial cells was seen (Fig. 3G). In stromal cells, aromatase immunostaining was often not uniform, and its intensity was variable. Aromatase immunostaining in many tissue sections was focally intense rather than diffuse. A uterine visceral peritoneal section used as negative control manifested negative aromatase immunostaining (Fig. 3H). Placenta was used as a positive control for aromatase immunostaining (Fig. 3I).

COX-2 mRNA expression is increased in endometrium and implants of women with endometriosis

Expression of COX-2 mRNA was increased in implants and in eutopic endometrium of women with endometriosis, compared with control endometrium, although the differences among the groups did not reach significance. As we observed for aromatase mRNA expression, the highest levels of COX-2 mRNA were observed in red implants. COX-2 mRNA levels in red implants were greater than in endometrioma cyst capsule, which was greater than black implants, which was greater than eutopic endometrium of endometriosis patients, which was greater than fibroid endometrium, which was greater than control endometrium (Fig. 4).

Figure 4.

COX-2 mRNA expression is highest in red implants and reflects the pattern of aromatase expression. COX-2 mRNA levels were analyzed by qRT-PCR in endometrial biopsies (EMB) from unaffected subjects (control EMB), from women with uterine fibroids (fibroid EMB), and endometriosis (eutopic EMB), in red and black endometriotic implants and in endometrioma. Relative levels of COX-2 mRNA were calculated by normalizing against h36B4 mRNA. Data, expressed as arbitrary units, are the mean ± sem of values from the number of subjects indicated on the abscissa.

ERα and ERβ expression

qRT-PCR analysis.

In a baboon model, the ratio of ERα to ERβ was reported to be decreased in endometriosis implants compared with normal endometrium (41). In ovarian endometriomas, ERβ expression was reported to be repressed compared with endometrium (26). In this study, we extended this analysis in endometriosis patients to include comparison of the ERβ to ERα ratio in normal endometrium vs. eutopic endometrium, red and black implants, and ovarian endometriomas. The ratios of mean arbitrary units of ERβ to ERα mRNA in control endometrium samples (0.0098 ± 0.0028), eutopic endometrium of patients with endometriosis (0.0184 ± 0.0053), endometrium of patients with fibroids (0.0367 ± 0.0091), and endometriosis implant samples (0.8030 ± 0.5259) were all less than unity (Fig. 5). Thus, these tissues were dominant in terms of ERα expression. By contrast, in ovarian endometrioma cyst capsule samples, the ratio of ERβ to ERα increased dramatically to 6.2 ± 2.4, demonstrating that these tissues were rich in ERβ mRNA transcripts.

Figure 5.

The ERβ to ERα mRNA expression ratio is increased in endometriosis implants as compared with control and eutopic endometrium and is even further up-regulated in endometriomas. ERα and ERβ mRNA levels were analyzed by qRT-PCR in endometrial biopsies (EMB) from unaffected subjects (control EMB) and from women with endometriosis (eutopic EMB) and with fibroids (fibroid EMB), in red and black peritoneal implants, and in endometriomas. Relative levels of ERα and ERβ mRNA were calculated by normalizing against h36B4 mRNA, and the ratios of ERβ to ERα mRNA were calculated. Data are the mean ± sem of ERβ/ERα in tissues from the number of subjects indicated on the abscissa. *, Significantly different (P = 0.008) from all others; +, significantly different (P = 0.013) from each other.

The Kruskal-Wallis test revealed significant differences among the tissue groups (Fig. 5, P = 0.008). Pairwise comparisons with Mann-Whitney U test showed significant differences between control endometrium vs. fibroid endometrium (P = 0.013) and endometriosis implants vs. endometrioma (P = 0.011). As expected, the pairwise comparisons between endometrioma cyst capsule samples vs. control (P = 0.007), fibroid (P = 0.014), and eutopic endometrium (P = 0.004) samples all were significant. However, the comparisons between endometriosis implants and the endometrium samples from these three groups were not significant. ERβ/ERα mRNA expression ratios appeared higher in black lesions than in the red implants, although the difference between these groups did not reach statistical significance (Fig. 5).

Immunohistochemistry for ERβ.

For ERβ immunostaining, human prostate sections were used as nonimmune IgG negative and positive controls (Fig. 6, A and B). In the prostate, immunostaining was observed mostly within the epithelial cells; immunoreactivity was mainly nuclear, but some cytoplasmic staining was observed, as well (Fig. 6B). The same pattern of nuclear and cytoplasmic immunostaining also was noted in some of the collected endometriosis tissue samples. Generally, when detected, the ERβ staining was focal rather than diffuse within any examined tissue section.

Figure 6.

Immunostaining of ERβ is increased in endometriosis. A, Human prostate counterstained with hematoxylin after incubation with nonimmune IgG (negative control); B, human prostate ERβ immunostaining in brown is mostly in the epithelial cells and is both nuclear and cytoplasmic; C, secretory endometrium from a control patient with mostly negative epithelial and stromal ERβ immunostaining; only some light cytoplasmic staining can be seen; D, eutopic endometrium from a patient with endometriosis at late proliferative phase with intense cytoplasmic and faint nuclear staining in the epithelial cells and some faint staining in stroma as well; E, endometrioma cyst capsule showing intense nuclear immunostaining of stromal cells and epithelial cells; immunostaining in the epithelial cells is both nuclear and cytoplasmic; F, endometrioma cyst capsule from another patient, with intense staining present in nuclei and cytoplasm of the glandular elements and some sporadic staining also observed in the stroma; G, red endometriosis implant, showing diffuse and intense staining of the epithelial and immediate subepithelial layers and sporadic and faint immunostaining observed in the stromal cells; H, black endometriosis implant, with intense ERβ immunostaining seen in both epithelial and stromal cells; I, visually and histologically normal peritoneum from a patient with stage II endometriosis, with some mild ERβ immunostaining evident only in the endothelial cells (black arrow), as expected. In all sections, thick blue arrows indicate epithelial compartment, and thin green arrows indicate the stromal cells.

In a secretory endometrium sample from a control patient without any evidence of endometriosis, both epithelial and stromal ERβ immunostaining was barely detectable. Very faint cytoplasmic immunostaining was seen in both epithelial and stromal compartments (Fig. 6C). However, eutopic endometrium from a patient with severe endometriosis at the late proliferative phase showed intense cytoplasmic and faint nuclear staining, especially in the epithelial cells (Fig. 6D).

Endometrioma cyst samples demonstrated intense immunostaining for ERβ (Fig. 6, E and F). In one of these samples, intense cytoplasmic and nuclear staining was seen in the epithelial cells, whereas staining in stroma was more sporadic (Fig. 6F).

Endometriosis implants demonstrated a varying intensity of staining, both in stromal and glandular elements. A red implant from a patient with mild endometriosis demonstrated immunostaining of the epithelial and immediate subepithelial cells with faint staining of the stroma (Fig. 6G). A black endometriosis implant from another patient with moderate to severe disease showed intense staining of the epithelial component and sporadic but intense nuclear immunostaining of the stromal cells (Fig. 6H).

Peritoneal sections from an endometriosis patient revealed faint ERβ immunostaining, localized in the endothelial cells of blood vessels, as expected (Fig. 6I). Visually normal peritoneum from unaffected subjects without any evidence of endometriosis demonstrated a similar pattern of mild ERβ immunostaining within the endothelial cells of the blood vessels (data not shown).

Expression of PR isoforms

After the relative expression for each PR isoform mRNA transcript was determined for each tissue studied, the relative ratio of each isoform was calculated. For example, to compute the relative abundance of PR-C mRNA, the following formula was used: percent PR-C mRNA = [PR-C mRNA/(PR-A + PR-B + PR-C mRNA)] × 100. This approach was then used for further comparisons.

Total PR (PR-ABC) transcript mean arbitrary values ± sem for control endometrium, eutopic endometrium, red implant, black implant, and endometrioma tissue samples were 43,139 ± 9,646, 57,798 ± 13,795, 92,846 ± 45,211, 19,816 ± 4,902 and 38,126 ± 17,446, respectively. These were not significantly different from each other (P = 0.25, Kruskal-Wallis test). However, analysis of the relative ratios of each PR isoform demonstrated that the percent PR-C mRNA expression was increased in peritoneal implants and ovarian endometrioma tissue samples in association with decreased PR-A and PR-B expression compared with the control and eutopic endometrium (Fig. 7A). When comparisons were made among all tissues for each PR isoform, the differences between the percent PR-A (P = 0.03), PR-C (P = 0.04), and PR-B (P = 0.03) mRNA for samples of endometrioma, red and black endometriosis implants, and eutopic, fibroid, and control endometrium were found to be significant (Kruskal-Wallis test).

Figure 7.

Expression of PR-C mRNA relative to PR-A and PR-B mRNAs is markedly increased in endometriosis implants and endometrioma cyst capsule, as compared with normal and eutopic endometrium. PR-A protein expression is absent in an endometrioma biopsy sample. A, Relative levels of PR-A, PR-B, and PR-C mRNAs were analyzed in control endometrial biopsies (EMB), eutopic EMB, endometriosis implants, and endometriomas by qRT-PCR using specific primers for PR-B, PR-A plus PR-B, and PR-A plus PR-B plus PR-C. The percentage of each PR isoform mRNA relative to total PR expression was calculated. Data are the mean ± sem of percentage of PR isoform mRNA for the number of subjects indicated on the abscissa. Statistical differences are discussed in the text. B, Immunoblot of PR protein expression in an endometrial biopsy from eutopic endometrium and endometrioma cyst wall from the same subject. Note lack of PR-A band in endometrioma sample.

Pairwise comparisons (Mann-Whitney U test) did not show any significant differences between control endometrium vs. eutopic endometrium, control endometrium vs. fibroid endometrium, and control endometrium vs. red and black endometriosis implants. However, control endometrium samples showed a significantly higher percent PR-A (P = 0.01) and significantly lower percent PR-C (P = 0.01) mRNA than those of endometrioma samples. Endometrioma samples also demonstrated a significantly lower percent PR-A (P = 0.006) and significantly higher percent PR-C (0.008) mRNA than the eutopic endometrium samples (Fig. 7A). Comparisons between eutopic endometrium and red endometriosis implants for all three PR isoforms tested were not significant. However, black implants demonstrated significantly lower percent PR-B (P = 0.04) than eutopic endometrium. When only red and black implants were compared with each other, there were no significant differences. The percent PR-A mRNA was significantly lower (P = 0.006), and the percent PR-C mRNA was significantly higher in endometriomas as compared with eutopic endometrium. Although a pattern of lower percent PR-A and higher percent PR-C mRNA emerged when eutopic endometrium, red implant, black implant, and endometrioma were evaluated together (Fig. 7A), the differences between black implant and endometrioma was not significant. When red implants were compared with endometrioma, the percent PR-C mRNA was significantly higher in endometrioma (P = 0.04), whereas the percent PR-A mRNA was comparable (P = 0.05).

Immunoblotting performed on a sample set of eutopic endometrium and endometrioma capsule from the same patient demonstrated PR-A, PR-B, and a truncated (∼60-kDa) isoform in eutopic endometrium. By contrast, the endometrioma capsule manifested robust immunoreactivity for PR-B and for the truncated isoform, whereas PR-A immunoreactivity was barely detectable (Fig. 7B). It should be noted that whereas PR-B mRNA levels were relatively low compared with the other PR isoforms in all tissue samples, PR-B protein expression was roughly equivalent to that of the truncated isoform in eutopic endometrium and endometrioma samples.

Immunohistochemistry for PR isoforms.

Immunostaining for PR isoforms revealed primarily nuclear distribution for PR-A and nuclear and cytoplasmic staining for PR-B and total PR (Fig. 8), as has been reported previously (33,34). In a control endometrium sample from late proliferative phase, PR-A immunostaining was localized primarily in nuclei of epithelial and stromal cells (Fig. 8A), whereas PR-B staining was both nuclear and cytoplasmic in epithelial cells (Fig. 8B). Total PR immunohistochemistry revealed intense stain in the cytoplasm of the epithelial cells and some scattered nuclear and cytoplasmic staining in the stroma (Fig. 8C).

Figure 8.

Immunostaining for PR isoforms reveals absence of PR-A protein in endometrioma. A–C, Control endometrium from late proliferative phase. A, PR-A immunostaining (in brown) is localized primarily in nucleus. Both epithelial and stromal cells stain positively for PR-A; B, PR-B staining is both nuclear and cytoplasmic and somewhat more intense in cytoplasm of epithelial cells; C, immunostaining with sc-539 (total PR) antibody reveals intense cytoplasmic staining in the epithelial cells and some scattered nuclear and cytoplasmic staining in stroma. D–F, Secretory-phase eutopic endometrium from a patient with endometriosis. D, PR-A immunostaining is more intense in the nuclei of the stromal cells compared with the epithelium; E, PR-B is present in both stromal and epithelial cell nuclei, with more intense staining observed in stromal cells; F, total PR immunostaining with sc-539 reveals intense nuclear staining in both stroma and epithelia. Intense cytoplasmic staining is observed in the epithelial cells. G–I, Red endometriosis implant. G, PR-A immunostaining is localized within the stromal cells of the implant; H, PR-B staining, although mostly in the stroma, also can be observed in the epithelial cells; I, total PR immunostaining is intense, both in the stromal and epithelial cells. J–L, Endometrioma cyst capsule. J, PR-A immunostaining is barely detectable, both in epithelial and the stromal cells of the endometrioma; K, epithelial cells and some stromal cells in the same endometrioma stain well for PR-B; L, nuclear and cytoplasmic total PR immunostaining are present both in epithelial and stromal cells.

In a secretory-phase eutopic endometrium, PR-A staining was more intense in the nuclei of the stromal cells compared with the epithelium (Fig. 8D). However, PR-B immunostaining was detectable in both stromal and epithelial cell nuclei with more intense staining observed in the stroma (Fig. 8E). Total PR immunostaining showed intense nuclear staining in both stroma and the epithelium (Fig. 8F). There also was intense staining in the cytoplasm of the epithelial cells.

In a red endometriosis implant, PR-A immunostaining was mostly localized in the stroma (Fig. 8G). PR-B immunostaining was more intense in the stroma, with some scattered staining in the epithelial cells (Fig. 8H). However, total PR immunostaining was evident in both stroma and epithelium (Fig. 8I).

Endometrioma was characterized by a near absence of PR-A immunostaining (Fig. 8J). However, PR-B and total PR immunostaining were detectable in both epithelial and stromal cells (Fig. 8, K and L).

Correlation analysis

To study the potential association between mRNA expression of the nuclear receptors, aromatase and COX-2 in this cross-sectional study of human tissues, Pearson correlation analysis was performed. Among a total of 27 endometrial tissue samples collected from histologically normal controls and from women with endometriosis or fibroids, significant positive correlation was found between aromatase expression and ERβ expression (P = 0.012; r = 0.48). In endometriosis implants, the ERβ/ERα expression ratio was significantly correlated with the percent PR-C (P = 0.03; r = 0.62). Again in implants, ERα mRNA expression was significantly correlated with PR-A mRNA expression (P = 0.0015; r = 0.7).

Discussion

Endometriosis is an estrogen-dependent disease, because the condition is rarely encountered before puberty and its symptoms respond to therapeutic measures to inhibit estrogen production or action (42,43,44). Endometriosis implants manifest differential hormonal responsiveness compared with eutopic endometrium, and this has been suggested to be secondary to differences in histochemical characteristics and hormone receptor distribution and to interference by inflammatory cytokines (45,46).

In this study, we investigated expression of aromatase, COX-2, and ER and PR isoforms in matched peritoneal tissue, eutopic endometrium samples, endometriosis implants, and endometrioma cyst capsules from patients with endometriosis. These findings were compared with those for endometrium and peritoneal samples of women without endometriosis. This study was unique in that each patient was prospectively identified and carefully staged, and matched tissues from the same patient were obtained and compared.

Aromatase and COX-2 expression

In eutopic endometrium of patients with endometriosis, we observed increased aromatase expression compared with the endometrium of women without endometriosis. Moreover, aromatase expression in endometriosis implants was severalfold higher than that of eutopic endometrium, with the highest expression observed in red implants. Interestingly, the pattern of COX-2 expression was similar to the pattern of aromatase expression, with the highest levels present in red implants (Fig. 4). COX-2 expression is up-regulated in the acute stages of inflammation, and its induction has been shown to play an important role in inflammation-related aberrant aromatase expression (18). Up-regulation of endometrial aromatase also has been observed in other inflammatory and proliferative conditions, including adenomyosis, leiomyomas, and endometrial carcinoma (47,48,49).

Endometriosis implants are known to elicit an inflammatory response mediated by invading leukocytes and their cytokines (50). It has been demonstrated in stromal cells from endometriomas that PGE₂ and cAMP stimulate expression and activity of aromatase (19). PGE₂ has been observed to be a potent stimulator of estrogen biosynthesis in endometriotic stromal cells by increasing cAMP levels via binding to EP2 and EP4 receptors (2,19). Prostaglandin levels were found to be higher in endometriosis tissue than in normal endometrium, which does not manifest enhanced aromatase expression (51,52). Furthermore, increased COX-2 immunostaining was observed in eutopic endometrium and in endometriotic lesions of women with endometriosis compared with endometrium from women without the disease (53). It also was observed in endometrial stromal cell cultures that COX-2 expression was induced by IL-1β, resulting in increased production of PGE₂ (54). PGE₂, in turn, may also further up-regulate COX-2 expression in endometrial stromal cells (55). The increased local estradiol formed in endometriosis further induces COX-2 expression, creating a positive feedback loop for aromatase induction (56). In primary human uterine microvascular endothelial cells, estradiol was observed to increase COX-2 mRNA expression and PGE₂ production. These effects were fully reversed by a nonselective ER antagonist, ICI 182,780 (20). Because these cells mainly expressed ERβ, it was suggested that this effect might be mediated via ERβ.

Aromatase mRNA transcripts in gonads, brain, adipose, breast cancer tissue, and placenta contain different first exons (IIa, If, I.4, I.3/IIa, and I.1, respectively), which are alternatively spliced onto a common site just upstream of the translation initiation codon in exon II (10,57). In this study, we observed that the aromatase transcripts in endometrium from control subjects and aromatase transcripts up-regulated in eutopic endometrium, implants, and endometrioma of endometriosis patients contained untranslated exon IIa at their 5′-ends; transcripts containing untranslated exon I.3 also were detected, albeit at lower levels. This reflects increased activation of the gonad- and cancer-specific promoters.

Aberrant expression of aromatase in endometriosis has previously been reported (16,17). The local estrogen produced, in turn, plays a paracrine and intracrine role as is also observed in breast cancer and uterine leiomyomas (58,59). It also has been found that 17β-hydroxysteroid dehydrogenase (17β-HSD) type 2, which converts E₂ to estrone, is deficient in endometriosis (60). This undoubtedly would further lead to higher local levels of E₂ within these tissues. Our study took a further step to examine the expression of aromatase in eutopic endometrium and endometriosis lesions according to their morphology and the surgical stage of the disease. The highest aromatase expression was detected in red implants of endometriosis when compared with black implants and in ovarian endometrioma cyst capsule, which represents a chronic inflammatory stage of the disease.

The expression of aromatase in eutopic endometrium of endometriosis patients was higher than that of fibroid and control endometrium in the cases studied, and the expression was highly influenced by the phase of the menstrual cycle and D-MPA treatment (Fig. 1, A and B). Furthermore, it was observed that aromatase expression in endometriosis implants was markedly elevated compared with eutopic endometrium (Fig. 1C). Additionally, eutopic endometrium manifests aberrantly enhanced proliferative-phase aromatase expression compared with normal endometrium, whereas luteal-phase levels of aromatase expression in eutopic endometrium were greatly reduced. This suggests that aromatase expression in eutopic endometrium remains sensitive to suppression by luteal-phase levels of progesterone. On the other hand, endometriosis implants manifested elevated aromatase expression in both proliferative and secretory phases of the menstrual cycle, suggesting a lack of sensitivity to luteal-phase levels of progesterone (Fig. 2B). Whereas D-MPA treatment down-regulated aromatase expression in the implants, the expression levels remained severalfold higher than those of eutopic endometrium (arbitrary values in Fig. 2B vs. Fig. 1B, respectively), further supporting the existence of progesterone resistance in endometriotic lesions, as suggested previously (30,51,52).

ERα and ERβ

ERα and ERβ can regulate gene expression in opposing ways. This regulation occurs either via the classical pathway through direct ER binding to estrogen response elements (EREs) or via nonclassical pathways through protein-protein interactions with other transcription factors, including activating protein-1 (61), nuclear factor-κΒ (NF-κΒ), and stimulating protein-1 (62). Although the amino acid sequence of ERβ ligand-binding domain is about 60% identical to that of ERα (23), E₂ is a potent endogenous ligand for ERβ and binds equally well to ERα and ERβ (63). In addition to their classical proliferative effects on the reproductive tract via EREs, nonselective estrogens such as 17β-estradiol also demonstrate antiinflammatory and antiproliferative activity (64,65,66,67), which has been observed in disease models including atherosclerosis, sepsis, uveitis, arthritis, and inflammatory bowel disease (68,69,70,71,72,73). The antiinflammatory activity of estrogen has been attributed to inhibition of NF-κΒ activity and DNA binding via direct protein-protein interactions (74,75,76,77), induction of IκBα expression (78), or competition for essential coactivators (79,80).

The actions of ERα and ERβ at EREs can oppose each other, depending on the cellular context. For example, when coexpressed with ERα, ERβ caused a concentration-dependent reduction in ERα-mediated transcriptional activation of the cyclin D1 gene, which mediates estrogen-related proliferation (61,81,82). Importantly, from studies with ERβ knockout mice, it appears that ERβ plays an inhibitory role in the expression of IGF-I and vascular endothelial growth factor in the endometrial stroma (83). E₂ acting through ERα is known to induce PR expression in the uterus (84). However, the finding that E₂ increased PR expression in uterine stromal cells of ERα knockout mice suggests that ERβ may mediate some of the effects of E₂ in uterus as well (85).

In normal endometrium, ERα was found to be highly expressed in the epithelium, although the mitogenic effects of E₂ may be through growth factors secreted from endometrial stromal cells (86). In ERβ-deficient mice, the uterus was found to be larger than in wild-type mice, and the proliferative response to E₂ was found to be stronger (83), suggesting an antiproliferative role of ERβ.

When human endometrial biopsies were implanted into nude mice to establish endometriotic lesions, treatment with an ERβ agonist resulted in complete regression of the lesions in the majority of animals (87). However, because only ERα was detected in the recovered lesions, it was speculated that the selective agonist might have exerted its antiinflammatory effects indirectly on immune cells. Selective ERβ agonists have also been demonstrated to exert antiinflammatory actions in rat disease models (88).

It has been suggested that heterodimers of ERα and ERβ could associate with ERE in vitro (89). Hence it is conceivable that the ERβ to ERα ratio within a particular tissue may affect local gene expression. In the present study, a graded increase in the ERβ to ERα ratio was observed from endometrium samples to endometriosis implants, with dominance of ERβ in samples of ovarian endometriomas (Fig. 5). We considered that the elevated expression levels of ERβ in endometrioma cyst capsule might possibly be due to contamination from ovarian granulosa cells and endeavored to isolate the cyst wall from follicular units. To assess possible contamination, we performed conventional histology and determined that the specimens analyzed were devoid of ovarian follicles, although a focal presence could not be ruled out. The aromatase expression pattern also did not support a potential granulosa cell contamination because the highest aromatase expression was detected in peritoneal implants. Furthermore, immunohistochemistry for ERβ clearly demonstrated immunostaining within the stromal cells of the cyst capsule and endothelial cells and within the endometrial glandular cells. Of note, ERβ immunostaining was both nuclear and cytoplasmic (Fig. 6), which agrees with previous reports (90,91,92,93).

Given the antiinflammatory and ERα-antagonizing effects of ERβ, we postulate that up-regulation of ERβ expression in the progression from eutopic endometrium to ovarian endometrioma cyst wall may result from the chronic inflammation that ultimately promotes local containment of endometriotic lesions.

PR isoform expression in endometriosis

Recent findings from our laboratory suggest that PR serves an antiinflammatory role in the uterus during most of pregnancy and that, at term, PR function declines through decreased expression of PR coactivators (94) and increased expression of the truncated PR isoform PR-C (95). PR-C inhibits PR-B transcriptional activity in transfected cells (95). Because PR-C lacks the capacity to bind DNA but binds progesterone (96), PR-C may inhibit PR function by sequestering progesterone and coactivators away from PR-A and PR-B isoforms. Additionally, PR-C can form heterodimers with PR-B, reducing the capacity of PR-B to interact with PREs controlling progesterone-responsive genes (97). In the present study, we observed that PR-C mRNA expression relative to PR-A and PR-B was dramatically up-regulated in peritoneal implants and in ovarian endometriomas. This was associated with a concomitant decline in the relative expression of PR-A and PR-B mRNA isoforms (Fig. 8).

We previously observed that inflammatory cytokines and NF-κB activation were associated with increased PR-C expression in T47D breast cancer cells (95). Hence, PR-C dominance in endometrioma and to a lesser extent in implants may result from persistent NF-κB activation. It is important to note that relative PR isoform mRNA levels may not be reflective of the relative expression levels of PR isoform protein. In fact, in the present study, immunoblot analysis of eutopic endometrium and endometrioma from a patient with endometriosis revealed roughly equivalent expression of PR-A, PR-B, and a truncated PR isoform in the eutopic endometrial sample. By contrast, PR-A protein was markedly decreased relative to PR-B and the truncated isoform in the endometrioma (Fig. 8B).

As mentioned above, endometriosis tissue also manifests a relative deficiency in 17β-HSD type-2 enzyme, which converts estradiol to estrone (60). This enzyme is normally induced by progesterone in endometrium (98). Thus, the apparent deficiency of 17β-HSD type 2 has been attributed to the progesterone resistance of endometriosis implants.

Our findings are in contrast to a previous study in which it was observed that that PR-A was the only PR isoform detected in endometriosis implants, both at the protein and mRNA levels (30). Because PR-A has been reported to act in a gene- and cell type-specific manner as a repressor of PR-B function (29,99), it was postulated that this may explain the apparent progesterone resistance of endometriosis implants. This view regarding the potentially antagonistic role of PR-A in the uterus is not supported by the phenotype of PR-B knockout mice (100), which exclusively express PR-A in progesterone target tissues. In that study, it was found that PR-A was sufficient to elicit normal uterine and ovarian responses to progesterone; the major phenotype observed was a defect in mammary gland morphogenesis.

Our findings regarding localization of PR-A and PR-B in endometrium are in agreement with other published studies. For example, PR-A immunostaining has been reported both in glandular and stromal cells of the endometrium with highest intensity in late proliferative phase (101). By contrast, in all of the endometrioma samples we examined, PR-A immunostaining was barely detectable (Fig. 8J), whereas PR-A was predominantly located in the subepithelial stromal compartment in an endometriosis implant (Fig. 8G). In breast cancer, increased PR-A expression has been associated with a more aggressive phenotype, in which cells are more adherent to the extracellular matrix and have increased migratory capacity (102). Although ovarian endometrioma cyst capsule is composed mostly of stromal cells, it manifests low levels of PR-A expression. Interestingly, endometriomas do not invade the ovarian tissue and remain as isolated cystic structures surrounded by a fibrotic capsule (103). In this regard, endometriomas, which lack PR-A, are less invasive than peritoneal implants, which express relatively high levels of PR-A.

Conclusions

In this study, we prospectively and cross-sectionally analyzed the expression of a number of potentially interacting regulatory factors in tissues from women with endometriosis to further understand the natural history and pathogenesis of this disease. Our findings suggest that the up-regulation of aromatase in eutopic endometrium and in implants is associated with induction of COX-2 and up-regulation of ERβ. Up-regulation of aromatase and COX-2 expression was most pronounced in red implants. Red lesions represent an acute and active phase of endometriosis associated with neovascularization (104,105,106). The enhanced synthesis of estrogens within red implants may subsequently act in a positive feed-forward loop to promote further vascularization by stimulating expression of angiogenic growth factors (107).

The increased inflammatory response that we observed in endometriotic lesions also was associated with enhanced mRNA expression levels of the inhibitory PR isoform PR-C relative to the PR-A and PR-B isoforms. This, in turn, may further exacerbate inflammation and aromatase induction by antagonizing ligand-dependent and -independent antiinflammatory actions of PR-B and PR-A (57). Interestingly, the expression levels of ERβ relative to ERα and of PR-C relative to other PR isoforms were highest in endometriomas. We suggest that, although aromatase and COX-2 may be induced by acute inflammation, PR-C and ERβ may serve as markers of a chronic inflammatory response. Although increased PR-C may enhance the disease process, up-regulation of ERβ may play an antiinflammatory and opposing role in its progression. Immunohistochemical analysis of aromatase, ERβ, and PR isoforms revealed that expression of these proteins was limited to specific cell types, especially within the stromal compartment. We suggest that variability in the cellular composition of different endometriosis lesions may be caused by the inflammatory status of the lesion. For example, in acutely inflamed peritoneal red implants of endometriosis, more epithelial cells and inflammatory infiltrate are evident, whereas in chronically inflamed ovarian endometriomas, stromal cells, histiocytes, and fibroblasts may predominate. We consider that such changes in tissue composition induced by inflammatory status may contribute to the differential receptor expression pattern in the various lesions of endometriosis that we have observed.

Studies are in progress using cultured cells and tissues to further define the mechanisms for the regulation of aromatase in endometriosis and the roles of regulatory hormones and transcription factors in the pathogenesis, progression, and resolution of this disease.