17β-Estradiol and Lipopolysaccharide Additively Promote Pelvic Inflammation and Growth of Endometriosis

Abstract

Endometriosis is a multifactorial disease mostly affecting women of reproductive age. An additive effect between inflammation and stress reaction on the growth of endometriosis has been demonstrated. Here we investigated the combined effect between 17β-estradiol (E₂) and lipopolysaccharide (LPS) on pelvic inflammation and growth of endometriotic cells. Peritoneal fluid was collected from 46 women with endometriosis and 30 control women during laparoscopy. Peritoneal macrophages (Mφ) and stromal cells from eutopic/ectopic endometrial stromal cells (ESCs) were isolated from 10 women each with and without endometriosis in primary culture. Changes in cytokine secretion (interleukin 6 [IL-6] and tumor necrosis factor α [TNF-α]) by Mφ and proliferation of ESCs in response to single and combined treatment with E₂ and LPS were measured by enzyme-linked immunosorbent assay and by bromodeoxyuridine incorporation assay, respectively. A significantly increased secretion of IL-6 and TNF-α in Mφ culture media was found in response to E₂ (10⁻⁸ mol/L) compared to nontreated Mφ. This effect of E₂ was abrogated after pretreatment of cells with ICI 182720 (10⁻⁶ mol/L; an estrogen receptor [ER] antagonist). Combined treatment with E₂ and LPS (10 ng/mL) additively promoted IL-6 and TNF-α secretion by peritoneal Mφ and growth of eutopic/ectopic ESCs. The additive effects of E₂ + LPS on cytokine secretion and growth of ESCs were effectively suppressed after combined blocking of ER and Toll-like receptor 4. An additive effect was observed between E₂ and LPS on promoting proinflammatory response in pelvis and growth of endometriosis.

Introduction

Endometriosis is an estrogen-dependent chronic inflammatory disease mostly affecting women of reproductive age. The exact pathogenesis of endometriosis is still unclear. There are some established hypotheses and regulatory factors supporting the development or maintenance of this disease.^1,2 However, it is difficult to uniformly explain the pathogenesis of endometriosis by a single factor. Hormonal factors and inflammation are commonly involved in the regulation of endometriosis.³ As an initial inflammatory mediator, bacterial endotoxin or lipopolysaccharide (LPS) has been recently reported to regulate Toll-like receptor 4 (TLR4)-mediated growth of endometriosis.⁴ An additive effect between inflammation and stress reaction on TLR4-mediated growth of endometriosis has been demonstrated.⁵ However, information regarding a possible additive effect between ovarian steroids and LPS in pelvic environment of women with endometriosis is limited.

Besides a role in reproduction, the ovarian steroid hormones such as 17β-estradiol (E₂) and progesterone (P) have been recognized to influence numerous immune and inflammatory responses.⁶ The immunomodulating actions of E₂ are thought to mainly result from their specific effects on the different cellular components of the immune system because most of them express estrogen receptors (ERs).^6,7 Among these cellular targets, macrophages (Mφ) may play a pivotal role in the modulation of immune responses by E₂ or by other initial/secondary inflammatory mediators.^7,8

The expression of ERs and TLR4 in macrophages and in eutopic/ectopic endometrial cells has been described previously.^4,7 We recently demonstrated production of a number of macromolecules in the culture media of Mφ derived from women with and without endometriosis and in individual response to LPS and E₂.^4,7 We previously observed that a variable amount of E₂ and LPS is available in the pelvis of women with and without endometriosis across the phases of the menstrual cycle.^3,4 Menstrual phase-dependent pattern of E₂ and endotoxin level in peritoneal fluid (PF) has been described elsewhere.^4,9 Since endometriotic lesions are constantly exposed to a combined milieu of E₂ and LPS in pelvic environment, we speculated that E₂ and LPS might act together as an additive promoter in the growth of endometriosis.

To address this, we first measured concentrations of E₂ and endotoxin (LPS) in different body fluids of women with and without endometriosis. Second, we investigated the pattern of interleukin (IL) 6 and tumor necrosis factor α (TNF-α) secretion by peritoneal Mφ and growth of eutopic/ectopic endometrial stromal cells (ESCs) in response to single and combined treatment with E₂ and P or LPS. As a marker of active inflammation in pelvis, we selected these 2 cytokines (IL-6/TNF-α) for our current study. Third, we extended our experiment to examine the role of ER and TLR4 in E₂ and LPS function. Finally, considering a hypoestrogenic environment, we examined pattern of changes in IL-6/TNF-α levels in the PF after GnRH agonist treatment in women with endometriosis.

Materials and Methods

Reagents

Culture media

Roswell Park Memorial Institute medium (RPMI)-1640 medium for macrophages and Dulbecco modified essential medium (DMEM):Hams F12 medium for stromal cells were supplemented with 100 IU/mL of penicillin G, 50 mg/mL of streptromycin, and 2.5 μg/mL of amphotericin B (GIBCO, Grand Island, New York). Fetal bovine serum (FBS), dimethylsulfoxide (DMSO), serum/phenol red-free RPMI/DMEM media, E₂, P, ICI 182720 (an ER antagonist), polymyxin B (an LPS antagonist), and LPS (derived from Escherichia coli, serotype 0111: B4) were purchased from Sigma Chemical Co (St Louis, Missouri). Anti-TLR4 neutralizing antibody (HTA-125) was purchased from HyCult Biotechnology, Pennsylvania).

Patients

During the period of April 2011 and June 2012, PF was collected from 46 women with endometriosis and 30 women without endometriosis during laparoscopy. Women with endometriosis aged between 20 and 42 years were recruited by either elective laparoscopy for infertility or diagnostic laparoscopy for dysmenorrhea and subsequently confirmed by histology. The control group, between 18 and 32 years old, consisted of fertile women without any evidence of endometriosis and were operated on for dermoid cyst, serous cyst adenoma, or mucinous cyst adenoma. The staging and the morphological distribution of peritoneal lesions were based on the revised classification of the American Society of Reproductive Medicine (r-ASRM).¹⁰ Sera and biopsy specimens were collected from a proportion of these women before and during laparoscopy. Menstrual fluid (MF) was collected from a separate population of women.

Neither the study group nor the endometriosis-free group had been on hormonal medication in 3 months prior to the surgical procedure. All control women and women with endometriosis had regular menstrual cycles (28-32 days). The phases of the menstrual cycle was determined by histological dating of eutopic endometrial samples taken simultaneously with pathological lesions during laparoscopy.

A separate group of 20 women with endometriosis, treated with gonadotropin-releasing hormone agonist (GnRHa) for a period of 4 to 6 months,were also enrolled for this study. All these women belonged to r-ASRM stage III to IV endometriosis and 15 of them had coexisting peritoneal lesions in pelvis. Peritoneal fluid was also collected from these GnRHa-treated women.

All body fluids and biopsy specimens were collected in accordance with the guidelines of the Declaration of Helsinki and were approved by the institutional review board of Nagasaki University. An informed consent was obtained from all women.

Collection of Menstrual Blood

Under strict aseptic measure, we collected menstrual blood from a separate population of 20 women with endometriosis and 15 women without endometriosis on day 1 to 3 of the menstrual cycle as described previously.^4,11 The materials obtained were transferred into heparinized endotoxin-free plastic containers. After serial processing and centrifugation, MF was collected and stored. All patients underwent a laparoscopy to confirm presence or absence of endometriosis before collecting menstrual blood.

All samples of sera, MF, and PF were collected prospectively and stored at −80°C for subsequent analysis. The concentration of E₂ was measured in sera/MF/PF by a modified immulize-enzyme amplified luminescence system as described previously.¹² The detection limit of E₂ was 20 pg/mL, and the intra- and interassay coefficients of variation were <10% for E₂.

The limulus amoebocyte lysate test (Endotoxin-Single Test; Wako-Jun-Yaku Co Ltd, Tokyo, Japan) was used to analyze bacterial endotoxin in sera, MF, and PF. In this assay, a chromogenic substance was converted by the endotoxin-activated limulus amoebocyte enzyme system, resulting in a color reaction.¹¹ The sensitivity of the assay to endotoxin was 5.0 pg/mL, and it gave a positive result at a level of >10.0 pg/mL.

Isolation of Mφ and ESCs

Macrophages from the PF and stromal cells from the eutopic and ectopic (peritoneal lesions) endometria were collected from 10 women (5 in the proliferative phase and 5 in the secretory phase) each with or without endometriosis during laparoscopy. All 10 women with endometriosis belonged to r-ASRM stage I to II endometriosis and all 10 women in each group belonged to 46 and 30 women with and without endometriosis from whom PF was collected.

The detailed procedures of the isolation of Mφ and ESCs were described previously.^5,7 Briefly, PF samples were centrifuged at 400g for 10 minutes, and the cellular pellet was underlayered with Lymphocyte Separation Medium (ICN, Aurora, Ohio) and centrifuged at 400g for 10 minutes. Macrophages were collected from the interface and cultured in RPMI-1640 medium (GIBCO). The macrophages were allowed to adhere to the culture plate for 2 hours, after which the nonadherent cells were removed by washing the plates 3 times with RPMI medium, and adherent cells were cultured further with RPMI medium + 5% FBS. The adherent cells remaining on the plates were estimated by their morphology and by immunocytochemical staining using CD68 (KP1, 1:50), a mouse monoclonal antibody from Dako, Denmark. The purity of Mφ was more than 95% as judged by positive cellular staining for CD68.

Glandular epithelial cells were separated from stromal cells and debris by filtration through narrow gauge sieves. The characteristics of the cultured stromal cells (ESCs) were determined by morphological and immunocytochemical staining using CD10 (56C6, 1:40), a mouse monoclonal antibody (Dako). The purity of eutopic/ectopic ESCs preparation was more than 95%, as judged by positive cellular staining for CD10 and negative staining for CD45 (a pan-leukocyte marker), cytokeratin (an epithelial cell marker), von Willebrand factor (a microvessel marker), and α-smooth muscle actin (a marker of myofibroblasts).

Activation and Ovarian Steroid Stimulation of Mφ and ESCs

Peritoneal Mφ and ESCs were plated in 96-well microplates and allowed to attach for 24 hours in phenol red-free media plus 5% FBS. After this time, the cells were washed with phosphate-buffered saline and incubated in serum-free and phenol red-free RPMI-1640 media (for Mφ) or DMEM/F-12 media (for ESCs) containing either E₂ (10^{− 8} mol/L)/P (10^{− 6} mol/L) or LPS (10ng/mL) or a combination for another 24 hours. These doses of E₂ and P were used according to a previous study protocol.¹³ Control cells were incubated in just phenol red-free RPMI-1640 media or DMEM/F-12 media. Ovarian steroids were dissolved in DMSO and steroid-free media used for control cultures containing equal concentrations of DMSO.

Activation studies involved the addition of 10 ng/mL of LPS just before commencement of the 24-hour incubation with E₂ and P. This dose of LPS was selected from a previous dose-dependent study from our laboratory that showed a maximum activation of Mφ with a dose of 10 to 100 ng/mL and at an incubation period of 24 to 48 hours.^4,7 In order to examine the blocking effect of ER and TLR4 on the secretion of different macromolecules in Mφ culture media, Mφ were pretreated with ICI (10^{− 6} mol/L) and anti-TLR4 antibody (10 μg/mL) 20 minutes before E₂, E₂ + P, LPS, and E₂ + LPS treatment and further incubated for 24 hours. These doses were used as described elsewhere.^4,14 The culture media was collected in triplicate, pooled, and frozen at −70°C until testing.

Cytokine Assays in the Media

The concentrations of IL-6 and TNF-α in the culture media of treated- and nontreated Mφ and in PF were measured in duplicate using a commercially available sandwich enzyme-linked immunosorbent assay obtained from R&D System (Minneapolis, Minnesota) in a blind fashion (Quantikine). The antibodies used in IL-6 and TNF-α determination do not cross-react with other cytokines. The limits of detection were 0.70 pg/mL for IL-6 and 4.4 pg/mL for TNF-α. Both the intra-assay and interassay coefficients of variation were <10% for all these assays.

Cell Proliferation Assays

5-Bromo-2-deoxyuridine (BrdU) labeling and detection kit measures ESCs proliferation by quantitating BrdU incorporated into the newly synthesized DNA of replicating cells.¹⁵ The incorporated BrdU can be detected by a quantitative cellular enzyme immunoassay (Biotrak; Amersham Pharmacia Biotech Ltd, United Kingdom) using monoclonal antibodies directed against BrdU. The detail procedure of BrdU incorporation assay was described previously.^4,7 We examined the proliferation of eutopic/ectopic ESCs in response to LPS, E₂, LPS + E₂, anti-TLR4 antibody, and ICI. The differences in cell proliferation were expressed as the percentage of controls. The absorbance values correlated directly with the amount of DNA synthesis and thereby to the number of proliferating cells in culture.

Possible contamination of endotoxin with E₂/P-treated Mφ and ESCs was examined by measuring endotoxin levels in the culture media by the limulus amoebocyte lysate test (Endotoxin-Single Test; Wako-Jun-Yaku Co Ltd) and by pretreatment of cells with polymyxin B (1 μg/mL), an LPS antagonist.

Statistical Analysis

All results are expressed as either mean ± standard error of the mean or mean ± standard deviation. The distribution of each result was initially analyzed by F-test. When F-test indicated skewed distribution, we applied nonparametric statistical analysis such as Mann-Whitney U test. When F-test indicated normal distribution of the results between groups, we applied parametric statistical analysis such as Student t test. Since concentrations of IL-6 and TNF-α in the Mφ culture media and in the PF were not normally distributed and in order to remain consistent with sample size, all 2-group comparisons were performed with nonparametric test. Other continuous variables between groups were compared with Student t-test. For comparisons among groups, the Kruskal-Wallis test was used. A box plot analysis of E₂/endotoxin/IL-6/TNF-α levels in body fluids was performed using the medians and interquartile range. A P value <.05 was considered statistically significant.

Results

There were no significant differences in clinical characteristics between women with or without endometriosis (Table 1). As an initial study, we also examined 6 women with endometriosis but without infertility. We did not find any difference in cytokine profile or cell growth in response to LPS and E₂ in these 2 groups of women with endometriosis with and without infertility.

Table 1.

Clinical Profiles of Patients With and Without Endometriosis.^a

	Control (n = 30)	Endometriosis (n = 46)
Age, mean ± SD, years	28.6 ± 3.8	29.8 ± 4.6
Range of age, years	18-32	20-42
r-ASRM staging: I-II/III-IV, n		24/22
Menstrual cycle: P/S/M, n	10/20/0	20/23/3
Serum study for E2 and endotoxin levels, n	20	30
MF study for E2 and endotoxin levels, n	15	20
PF study for E2 and endotoxin levels, n	30	46
Macrophage study, n (P/S)	10 (5/5)	10 (5/5)
ESCs study, n (P/S)	10 (5/5)	10 (5/5)

Abbreviations: r-ASRM, revised classification of the American Society of Reproductive Medicine; P, proliferative phase; S, secretory phase; M, menstrual phase; MF, menstrual fluid; PF, peritoneal fluid; ESCs, endometrial stromal cells; SD, standard deviation.

^aThe results are expressed as mean ± SD.

17β-Estradiol/P/LPS-stimulated cytokine secretion and ESCs proliferation were similar between cells collected from the proliferative phase and the secretory phase of the menstrual cycle. Therefore, here we represented our combined data irrespective of the phases of the menstrual cycle.

17β-Estradiol/Endotoxin Concentrations in Body Fluids

No significant difference in E₂ levels in the sera, MF, and PF was observed between women with endometriosis and control women (Figure 1A). Except in sera, endotoxin (LPS) level in MF and PF was significantly higher in women with endometriosis than in control women (P < .05 for each, Figure 1B). We did not find any difference in E₂/endotoxin levels in PF between r-ASRM stage I to II and III to IV endometriosis (data not shown).

Figure 1.

Concentrations of 17β-estradiol (E₂; A) and endotoxin (lipopolysaccharide [LPS]; B) in sera, menstrual fluid (MF), and peritoneal fluid (PF) derived from control women (white box) and women with endometriosis (hatched box). B, *P < .05 versus controls for either MF or PF. Boxes represent the distance (interquartile range) between the first (25%) and the third (75%) quartiles, and horizontal lines in the boxes represent median values.

Time-Dependent Production of IL-6/TNF-α by Mφ in Response to E₂

In a time-dependent study, we found that short-term exposure (3-12 hours) of Mφ with E₂ (10^{− 8} mol/L) failed to increase IL-6/TNF-α secretion in culture media of Mφ derived from women with and without endometriosis. In contrast, long-term exposure (24-72 hours) with E₂ significantly increased production of IL-6/TNF-α in Mφ culture media in women with endometriosis over control women (P < .05 vs control at 24, 48, and 72 hours; Figure 2A and B).

Figure 2.

Time-dependent changes in the secretion of interleukin (IL) 6 (A) and tumor necrosis factor α (TNF-α; B) in the culture media of macrophages (Mφ) derived from the peritoneal fluid of 10 women with endometriosis (black bar) and 10 control women (white bar) and in response to 17β-estradiol (E₂; 10⁻⁸ mol/L). *P < .05 versus controls for each indicated incubation time. The results are expressed as mean ± standard error of the mean (SEM) of triplicate experiments with Mφ derived from each woman.

Production of IL-6/TNF-α by Mφ in Response to E₂ and P

Direct stimulation of Mφ in culture with E₂ and P resulted in a variable increase in the secretion of IL-6 and TNF-α by peritoneal Mφ. The production of IL-6 and TNF-α was significantly higher by E₂ in women with endometriosis than in control women (P < .05 for each of IL-6 and TNF-α) or when compared to E₂ nontreated Mφ (P < .05 for each, Figure 3A and B). An additive effect between E₂ and P on IL-6/TNF-α secretion was not observed. A higher tendency in the secretion of TNF-α in Mφ culture media was observed in women with endometriosis than in control women in response to P without displaying any significant difference between them (P = .07, Figure 3B). Since P failed to show a significant increase in the secretion of IL-6/TNF-α by Mφ, we performed blocking experiments on E₂ using ICI, an ER antagonist. The ICI + E₂ and ICI + E₂ +P significantly reversed the secretion of IL-6 and TNF-α by Mφ compared to ICI-nontreated cells (P < .05 for each of IL-6 and TNF-α; Figure 3A and B). The suppressive effect of ICI on IL-6/TNF-α secretion was observed in women with endometriosis but not in control women.

Figure 3.

Production of interleukin (IL) 6 (A) and tumor necrosis factor α (TNF-α; B) in the culture media of peritoneal fluid macrophages (Mφ) derived from 6 women each with endometriosis and control women and in response to 17β-estradiol (E₂), progesterone (P), a combination of E₂ and P (E₂ + P), and ICI, an estrogen receptor (ER) antagonist + E₂, and ICI + E₂ + P. *P < .05 versus E₂-untreated Mφ (A and B). The results are expressed as mean ± standard error of the mean (SEM) of triplicate experiments for each treatment.

17β-Estradiol Modulation of IL-6 and TNF-α Production by LPS-Activated Mφ

Peritoneal Mφ from control women was used in order to eliminate any bias in the basal activation status of these cells. The production of IL-6 and TNF-α in the culture media by LPS-activated Mφ was significantly higher than by LPS nonactivated Mφ (P < .05; Figure 4A and B). We observed that activation of basal Mφ further enhanced the response of these cells to E₂. In fact, exogenous treatment with E₂ was able to further increase the amount of both IL-6 (P < .05) and TNF-α (P < .01) secretion by Mφ when these cells were preactivated with LPS (Figure 4A and B). The response of E₂ on LPS-activated Mφ in the secretion of IL-6/TNF-α was significantly stronger than in LPS-treated and E₂-untreated Mφ (P < .05). The stimulating effect on cytokine secretion was not observed for P in LPS-activated Mφ. We also found that LPS has the similar additive effect on E₂ pretreated Mφ (data not shown).

Figure 4.

Production of interleukin (IL) 6 (A) and tumor necrosis factor α (TNF-α; B) in the culture media by lipopolysaccharide (LPS)-activated (10 ng/mL) and/or 17β-estradiol (E₂; 10⁻⁸ mol/L)- and progesterone (P; 10⁻⁶ mol/L)-treated macrophages (Mφ) that were derived from the peritoneal fluid of 10 women without endometriosis. The results are expressed as mean ± standard error of the mean (SEM) of triplicate experiments with Mφ.

17β-Estradiol/LPS-Stimulated and ER/TLR4-Mediated Production of IL-6/TNF-α

Individual treatment with E₂ and LPS significantly increased the secretion of IL-6 and TNF-α in the culture media of Mφ derived from women with endometriosis than in control women (P < .05, Figure 5A and B). A significant additive effect between E₂ and LPS was observed on IL-6 and TNF-α secretion in both controls and women with endometriosis compared to individual treatment with E₂ or LPS. Mann-Whitney U test indicated a significant difference in secretion of both IL-6 and TNF-α between E₂-treated and E₂ + LPS-treated (P < .05 for both IL-6 and TNF-α) and also between LPS-treated and E₂ + LPS-treated Mφ (P < .05 for both IL-6 and TNF-α; Figure 5A and B). A significant decrease in IL-6/TNF-α secretion by LPS-treated Mφ was observed after blocking of ER (P < .05 vs LPS-treated cells). Although individual blocking effect failed, the additive effect of E₂ + LPS on IL-6/TNF-α secretion was effectively suppressed after combined blocking of ER and TLR4 (P < .05 vs ICI- or anti-TLR4 antibody-treated Mφ; Figure 5A and B). We also performed basal effects of ICI and anti-TLR4 antibody on LPS- and E₂-nontreated Mφ, and we did not find any difference between treated and untreated Mφ (data not shown). Using a similar treatment protocol, we observed that eutopic ESCs derived from women with and without endometriosis were able to secrete IL-6 and TNF-α in the culture media in response to E₂/LPS and a similar additive effect was observed with E₂ + LPS (data not shown).

Figure 5.

The exogenous single or combined effect of 17β-estradiol (E₂) and LPS on the production of interleukin (IL) 6 (A) and tumor necrosis factor α (TNF-α; B) in the culture media of Mφ derived from the peritoneal fluid of 10 women each with endometriosis (black bar) and control women (white bar). *P < .05 versus control; ^† P < .05 versus individual LPS- or E₂-treated Mφ; ^‡ P < .05 versus ICI-untreated Mφ; ^¶ P < .05 versus single treatment with ICI or anti-Toll-like receptor 4 (anti-TLR4) antibody. The results are expressed as mean ± standard error of the mean (SEM) of triplicate experiments for each treatment. LPS indicates lipopolysaccharide.

Exogenous Effects of LPS and E₂ on ESCs Proliferation

The effects of LPS and E₂ on BrdU incorporation into eutopic and ectopic ESCs were significantly higher in women with endometriosis than in control women or untreated cells (P < .05; Figure 6A and B). A significant additive effect on BrdU incorporation into eutopic/ectopic ESCs was observed between LPS and E₂ (P < .05 vs individual LPS or E₂ treatment; Figure 6A and B). Pretreatment of eutopic/ectopic ESCs with either anti-TLR4 antibody or ICI significantly suppressed BrdU incorporation compared to either LPS- or E₂-treated cells (P < .05 for each). A significant decrease in BrdU incorporation by LPS-treated eutopic/ectopic ESCs was observed after blocking of ER (P < .05 vs LPS-treated ESCs for Figure 6A, and P < .05 for Figure 6B). In contrast to eutopic ESCs (Figure 6A), individual pretreatment of cells with ICI and anti-TLR4 antibody significantly decreased E₂ + LPS-stimulated BrdU incorporation into ectopic ESCs (P < .05 for each, Figure 6B). The additive effect of E₂ + LPS on BrdU incorporation into eutopic/ectopic ESCs was further decreased after combined treatment with ICI and anti-TLR4 antibody (P < .05 vs individual ICI- or anti-TLR4 antibody-treated cells; Figure 6A and B).

Figure 6.

The exogenous single or combined effect of LPS and 17β-estradiol (E₂) on the bromodeoxyuridine (BrdU) incorporation into eutopic (A) and ectopic (B) endometrial stromal cells (ESCs) derived from 10 women each with endometriosis (black bar) and control women (white bar). Results of BrdU incorporation are expressed as a percentage of control. The BrdU incorporation of nontreated cells equals 100%. Like cytokine secretion as shown in Figure 5, a similar pattern of BrdU incorporation in eutopic/ectopic ESCs was observed in response to treatment with E₂, LPS, LPS + E₂, ICI (an estrogen receptor [ER] antagonist), and anti-TLR4 antibody (A and B). *P < .05 versus control ESCs (A); *P < .05 versus nontreated ectopic ESCs (B); ^† P < .05 versus single LPS or E₂ treatment (A and B); ^‡ P < .05 in (A) and ^# P < .05 in (B) versus LPS-treated ESCs; ^‡ P < .05 versus LPS + E₂-treated cells (B); ^¶ P < .05 versus single treatment with ICI or anti-TLR4 antibody (A and B). The results are expressed as mean ± standard error of the mean (SEM) of triplicate experiments for each treatment. LPS indicates lipopolysaccharide; TLR4, Toll-like receptor 4.

Since BrdU incorporation study represents the simple incorporation of BrdU into the proliferated DNA of ESCs and does not reflect the actual cell growth as accounted by increased cell number, therefore, we also examined the cell growth of eutopic/ectopic ESCs by the cell number (initial plating 10⁵ cells/well) using the similar treatment protocol of E₂, LPS, and E₂ + LPS. We found a parallel and significantly increased cell growth under the similar stimulation for both eutopic and ectopic ESCs (data not shown).

Concentrations of IL-6 and TNF-α in PF After GnRHa Treatment

We measured concentrations of IL-6 and TNF-α in the PF of control women and GnRHa-treated and -untreated women with endometriosis. We found that both IL-6 and TNF-α were significantly higher in the PF of women with endometriosis than in control women (P < .05 for IL-6; P < .01 for TNF-α; Figure 7). The PF levels of IL-6 and TNF-α were significantly decreased in GnRHa-treated women when compared to GnRHa-untreated women with endometriosis (P < .05 for IL-6; P < .001 for TNF-α; Figure 7).

Figure 7.

Concentrations of interleukin (IL) 6 (white box) and tumor necrosis factor α (TNF-α; hatched box) in the peritoneal fluid derived from 30 control women, 46 women with endometriosis and 20 gonadotropin-releasing hormone agonist (GnRHa)-treated women. Boxes represent the distance (interquartile range) between the first (25%) and third (75%) quartiles, and horizontal lines in the boxes represent median values.

Exclusion of Endotoxin Contamination With E₂/P-Treated Cells

In order to exclude the possible contamination of endotoxin with E₂/P-treated cells, we repeatedly measured endotoxin level in the culture media. We could not detect any endotoxin in the culture media of treated cells. Pretreatment of Mφ with polymyxin B (1 μg/mL) failed to decrease the levels of any of these macromolecules in the culture media of treated cells (data not shown).

Discussion

We demonstrated here that in addition to individual action of E₂ and LPS in promoting inflammatory response and growth of endometriosis, an additive effect between E₂ and LPS might be involved in further worsening of pelvic inflammation and growth of endometriosis. If we consider the internal milieu of pelvic environment, a substantial amount of E₂ and endotoxin in the pelvis of women with endometriosis^4,9 may jointly trigger this detrimental effect. Our current findings are consistent with the increased production of different macromolecules by the activated Mφ as described previously.^16,17

We found that peritoneal Mφ were directly stimulated to secrete IL-6 and TNF-α by E₂, a response that was blocked by ER antagonist and was enhanced when preactivated with LPS. In our separate experiment (data not shown), we also found that LPS has similar additive effect on E₂ pretreated macrophages. This means that E₂ and LPS may, in either way, exhibit an additive effect on cytokine secretion and cell proliferation. Recent studies on immune–endocrine interaction in endometriosis detailed individual roles of inflammatory mediators and ovarian steroids without mentioning any additive effect between them.^3,18 We learned from our current findings that E₂ and LPS might function either alone or in combination to regulate ER- and TLR4-mediated promotion of proinflammatory response and growth of endometriosis. Decreased secretion of cytokines and BrdU incorporation into ESCs after blocking of ER and in response to LPS indicates a role of E₂ in promoting LPS function.

Our findings were further supported by our extended study with GnRHa-treated and -untreated women with endometriosis. With the concept in mind that GnRHa treatment creates a state of hypoestrogenic environment, we found that higher PF levels of IL-6 and TNF-α were significantly decreased after GnRHa treatment compared to GnRHa-untreated women with endometriosis. Our findings coincided with the recent report indicating regression of the inflammatory microenvironment in the pelvis of women with endometriosis after GnRHa treatment.¹⁹

To date it was assumed that ovarian steroid hormones exhibit direct actions on the endometrial or endometriotic tissues.¹⁸ Indirect actions of ovarian steroids on immune cells are scarcely described. Since the major cellular constituents of PF are Mφ, comprising between 82% and 99% of the total cell population,²⁰ it is quite reasonable to speculate that these cells may be responsive to ovarian steroids. Our current study is a further piece of evidence to describe this interaction between Mφ and ovarian steroid hormones.

We previously demonstrated that Mφ retain the messenger RNAs encoding both ER and PR and showed nuclear staining for both ER and PR in isolated Mφ and in eutopic/ectopic endometria of women with and without endometriosis.⁷ These results are in accordance with the results of McLaren et al.¹³ We recently proposed a novel concept of “bacterial contamination hypothesis” for the development of endometriosis via LPS/TLR4-mediated engagement of innate immune response.⁴ The gene and protein expression of TLR4 in Mφ and in eutopic/ectopic ESCs has already been described.^4,21 Here, we further established that as a marker of initial inflammatory mediator in pelvis, LPS together with E₂ additively promoted proinflammatory response in pelvis as well as growth of eutopic and ectopic ESCs. We presume that different macromolecules in PF including IL-6 and TNF-α may be the result of combined action of E₂ and LPS on eutopic/ectopic ESCs as well as on peritoneal Mφ.

It has previously been reported that ERs are expressed in significantly higher concentrations in Mφ of women with endometriosis with parallel higher expressions of IL-6 and TNF-α.²² Therefore, the question still remains whether ERs might be involved in Mφ differentiation. However, in our previous study,⁷ we found that ER expression did not differ in CD68-immunostained Mφ between women with and without endometriosis. Again, in our initial study, we found that CD68 expression was not different between LPS-treated or LPS-untreated Mφ. Our another study showed that TLR4 expression was similar in Mφ derived from women with and without endometriosis.⁴ These findings support that secretion of cytokines (IL-6/TNF-α) by Mφ in response to E₂/LPS does not depend on the amount of ER/TLR4 expression rather depends on time-dependent differentiation of Mφ and also on receptor/ligand binding affinity of Mφ between women with and without endometriosis. Our findings of increased secretion of IL-6 and TNF-α in the culture media after long-term exposure (24-72 hours) of human peritoneal Mφ to E₂ coincided with previous findings.^8,23

The pattern of biological function of E₂ and ER expression depends on the status of immune cell differentiation. In undifferentiated monocyte cell line U937, E₂ induces apoptosis and exhibits anti-inflammatory function. In contrast, E₂ exhibits proinflammatory response when U937 cells are differentiated with phorbol ester to a Mφ phenotype.²⁴ This was correlated with differential expression of ERs, undifferentiated monocytes express ERβ while differentiated Mφ express both ERα and ERβ.^8,24 In fact, we used CD68 antigen (clone KP1) as a marker to detect highly differentiated peritoneal Mφ in our current study which express both ERα and ERβ.

The main limitation of our current study is the lack of information regarding intracellular mechanistic basis of the single and combined effects of E₂ and LPS on cytokine production and cell growth. After LPS/TLR4 binding, a sequence of activation of intracellular adaptor molecules, phosphorylation of IκB, and activation of nuclear factor κB (NF-κB) in the upregulation of target genes is described in women with endometriosis.²⁵ The NF-κB signaling controls cyclin D1 activity, a positive regulator of G1-to-S-phase progression, leading to cell proliferation.²⁵ Estrogen is known to influence target gene expression in 2 ways. E₂ ligand-activated ERs can regulate gene expression directly via an ERE or indirectly by interacting with other transcription factors including AP-1 and Sp1 as demonstrated in the promoter of the target gene.^26,27 This may explain a mechanistic basis of E₂/ER-mediated IL-6/TNF-α secretion by Mφ. Nongenomic effect of E₂ on the endometrium has been reported elsewhere.²⁸ For example, treatment of EECs/ESCs with E₂ results in Ca²⁺ influx with consequent changes in number and morphology of endometrial cells. Treatment with ICI failed to antagonize this nongenomic effect of E₂. The effective inhibitory effect of ICI on E₂-mediated cytokine secretion in our study may support a genomic effect of E₂.

An additive interaction between NF-κB and ER signaling pathways in human endometrial cells is reported elsewhere.¹⁴ Estrogen can both activate and inhibit the NF-κB. The ERα and ERβ can associate with NF-κB and with steroid hormone coactivators at the promoter region of NF-κB-regulated gene. It is possible that cofactors, previously described as coactivators or corepressors, can switch functionality depending on the promoter context.^14,28 Recently, King and coworkers described a positive interaction between ERs and NF-κB signaling pathway in an endometrial epithelial cell line.¹⁴ Estradiol and IL-1β treatment of EECs enhance ERE activity by NF-κB- and ER-dependent mechanism. A similar mechanism may also operate in the combined action of E₂ and LPS in our study.

There is some diverse information on whether E₂ directly or indirectly regulates TLR4 expression. The E₂-stimulated downregulation of PI3K/Akt pathway,⁸ a negative regulator of TLR4 activation, in mice Mφ may explain E₂-promoted Mφ activation. A similar interaction may be involved in the additive effect between E₂ and LPS in our current study. There are indications that ovarian steroids can affect the expression levels of TLR4 in immune competent cells. For example, E₂ enhances TLR4 expression in Mφ, while progesterone decreases its expression in the brain of mice with experimental autoimmune encephalomyelitis.²⁹ These bimodal effects of E₂ on TLR4 may further promote LPS-activated cytokine production in our study. We do not know exactly whether LPS might enhance ER expression and/or the responsive downstream factors. One possibility of LPS-mediated E₂/ER expression may be related to LPS-promoted aromatase gene expression. This is still an open question for future investigation.

Our results have some biological implications: (1) ovarian steroids may influence the autocrine regulation of macrophage or stormal cell functions and (2) an inflammatory response in pelvic environment and ovarian steroid hormone may function independently or in an orchestrated manner, which may be involved in the growth or persistence of endometriosis. Based on our current findings, we propose that in addition to conventional estrogen-suppressing agent, combined targeting of LPS/E₂ or TLR4/ER could be useful to more effectively suppress pelvic inflammation and growth of endometriosis.

Considering endometriosis as a multifactorial disease,^2,3 our findings may provide some new evidence to understand the role of more than a single factor in the pelvis of women with endometriosis. Besides combined effect between inflammation and stress reaction on the growth of endometriosis,⁵ our current findings demonstrated involvement of a mutual action between initial inflammatory mediator (LPS) and ovarian steroid (E₂) in further promoting pelvic inflammation and growth of endometriosis. Further studies are needed to strengthen our current findings.