Molecular and preclinical basis to inhibit PGE₂ receptors EP2 and EP4 as a novel nonsteroidal therapy for endometriosis

Significance

Endometriosis is an inflammatory gynecological disease of reproductive age women associated with chronic pelvic pain and infertility. Endometriosis remains as the single major cause for hysterectomy in reproductive age women in the United States. The pathogenesis of endometriosis remains an enigma in reproductive medicine. Current hormonal therapies cause undesirable side effects, reproductive health concerns, and are unable to prevent recurrence of disease. Results of the present study indicate that selective inhibition of prostaglandin E2 (PGE₂) receptors EP2 and EP4 suppresses the growth and survival of endometriosis lesions, decreases pelvic pain, and restores endometrial functional receptivity through multiple mechanisms. Our novel findings provide a molecular and preclinical basis to formulate long-term nonhormonal therapy for endometriosis.

Abstract

Endometriosis is a debilitating, estrogen-dependent, progesterone-resistant, inflammatory gynecological disease of reproductive age women. Two major clinical symptoms of endometriosis are chronic intolerable pelvic pain and subfertility or infertility, which profoundly affect the quality of life in women. Current hormonal therapies to induce a hypoestrogenic state are unsuccessful because of undesirable side effects, reproductive health concerns, and failure to prevent recurrence of disease. There is a fundamental need to identify nonestrogen or nonsteroidal targets for the treatment of endometriosis. Peritoneal fluid concentrations of prostaglandin E₂ (PGE₂) are higher in women with endometriosis, and this increased PGE₂ plays important role in survival and growth of endometriosis lesions. The objective of the present study was to determine the effects of pharmacological inhibition of PGE₂ receptors, EP2 and EP4, on molecular and cellular aspects of the pathogenesis of endometriosis and associated clinical symptoms. Using human fluorescent endometriotic cell lines and chimeric mouse model as preclinical testing platform, our results, to our knowledge for the first time, indicate that selective inhibition of EP2/EP4: (i) decreases growth and survival of endometriosis lesions; (ii) decreases angiogenesis and innervation of endometriosis lesions; (iii) suppresses proinflammatory state of dorsal root ganglia neurons to decrease pelvic pain; (iv) decreases proinflammatory, estrogen-dominant, and progesterone-resistant molecular environment of the endometrium and endometriosis lesions; and (v) restores endometrial functional receptivity through multiple mechanisms. Our novel findings provide a molecular and preclinical basis to formulate long-term nonestrogen or nonsteroidal therapy for endometriosis.

Endometriosis is a debilitating, chronic inflammatory gynecological disease of reproductive age women. Two major clinical symptoms of endometriosis are chronic intolerable pelvic pain and subfertility or infertility, which profoundly affect the quality life in women (1–3). The prevalence of the disease is 5–10% in reproductive age women, increases to 20–30% in women with subfertility, and to 40–60% in women with pain and infertility (1–3). Endometriosis remains as the single major cause for hysterectomy in reproductive age women in the United States, with an annual estimated societal cost of ∼$69.4 billion (3). These significant individual and public health concerns underscore the importance of understanding the pathogenesis of endometriosis.

Although endometriosis has been traditionally viewed as an estrogen (E₂)-dependent and progesterone (P₄)-resistant disease (1–3), its pathogenesis remains an enigma in reproductive medicine. The most widely accepted hypothesis first advanced by Sampson in 1927 is that viable endometrial tissue fragments move in a retrograde fashion through the fallopian tubes into the pelvic cavity during menstruation (4). These ectopic endometrial cells invade the underlying peritoneum, survive for considerable time (∼7–8 y), and establish peritoneal endometriosis characterized by heterogeneous lesions/phenotypes.

Current treatment strategies include surgical intervention, medical therapy, or a combination of both. After surgical removal of endometriosis lesions, the disease reestablishes within 3–5 y in ∼30–50% of women. Surprisingly, the disease reoccurs in ∼10% of women who have had the uterus and both ovaries removed (5). Hormonal therapy to induce a hypoestrogenic state through the use of oral contraceptives, progestagens, androgenic agents, and gonadotropin releasing hormone analogs can be prescribed only for a short time due to undesirable side-effects, pseudomenopause, and bone density loss in reproductive age women (1–3, 5). Nevertheless, the recurrence rate is ∼50–60% after cessation of therapy within a year (5). Together, existing treatment modalities fail to prevent recurrence of disease, and affect pregnancy and reproductive health of women. There is a fundamental need to identify potential cell signaling pathways for nonestrogen or nonsteroidal targets for endometriosis.

Peritoneal fluid concentrations of prostaglandin E₂ (PGE₂) are higher in women with endometriosis, and this increased PGE₂ plays an important role in survival and growth of endometriosis lesions (6–9). Inhibition of PGE₂ biosynthesis impedes growth of endometriosis (9) and chronic pelvic pain in women (7), and decreases growth of endometriosis lesions in animal models (8). COX-2 is the rate limiting enzyme that regulates biosynthesis of PGE₂. The biological actions of PGE₂ are mediated via G protein receptors EP1, EP2, EP3, and EP4 by integrating multiple cell signaling pathways (10). Recent studies from our laboratory indicate that selective inhibition of EP2 and EP4 inhibits adhesion, invasion, growth, and survival of human endometriotic epithelial and stromal cells by modulating integrins, matrix metalloproteinases (MMPs) and tissue inhibitors of metalloproteinases (TIMPs), cell cycle, survival, and apoptosis pathways in vitro (11–14).

The objectives of the present study were to determine the pharmacological effects of selective inhibition of EP2/EP4 on: (i) growth and survival of endometriosis lesions; (ii) molecular and cellular aspects of the pathogenesis of endometriosis; (iii) pelvic pain and molecular pain pathways; and (iv) endocrine and immune modulation in the endometrium, using chimeric mouse model of endometriosis as preclinical testing platform. Our novel results collectively indicate that inhibition of PGE₂-EP2/EP4 signaling could emerge as a potential nonestrogen or nonsteroidal therapy for endometriosis.

Results

Therapeutic Effects of Selective Inhibition of EP2/EP4 on Growth of Endometriosis Lesions.

Xenograft of mixed populations of endometriotic epithelial cells 12Z-GFP and stromal cells 22B-RFP in ovariectomized and estrogen-treated nude mice induced peritoneal endometriosis characterized by heterogeneous white and bleeding red lesional phenotypes which is similar to the human peritoneal endometriosis (Fig. 1). Histomorphology of the endometriosis lesions showed that epithelial cells 12Z-GFP formed endometrial glands and stromal cells 22B-RFP formed stroma and thus established functional communicating endometrial glands and surrounding stroma similar to the human peritoneal endometriosis. Under gross examination, 8–10 endometriosis lesions/mouse were visible. Importantly, under fluorescence dissection microscopy, 20–25 clusters of endometriosis lesions/mouse were detectable. Among the visible peritoneal endometriosis lesions, ∼60% were white lesions and 40% were bleeding red lesions (Fig. 1 A–D). Selective inhibition of EP2/EP4 from days 15–28 of xenograft decreased (P < 0.05) the growth of endometriosis lesions up to 60% dose-dependently, and the maximal effect was observed at 25 mg/kg. Inhibition of EP2/EP4 decreased (P < 0.05) the growth of epithelial cells 12Z-GFP and stromal cells 22B-RFP up to 70% and 60% respectively (Fig. 1 E–J). Real-time in vivo bioimaging, fluorescence dissection microscopy imaging, and morphometry data together indicated that pharmacological inhibition of EP2/EP4 decreased (P < 0.05) growth (size and volume) of endometriosis lesions. In addition, EP2-I/EP4-I (25 mg/kg) did not affect functions of liver and kidney (Fig. 1K), and the treated mice were apparently healthy.

Fig. 1.

Effects of inhibition of EP2 and EP4 (EP2/4-I) on growth and survival of endometriosis lesions. Fluorescence microscopy of (A1) human endometriotic epithelial cells 12Z-GFP and (A2) stromal cells 22B-RFP grown in coculture and imaged with red or green channels and (A3) overlay of both channels. (B) A mixture of epithelial cells 12Z-GFP and stromal cells 22B-RFP suspension was injected into the peritoneal cavity of nude mice, peritoneal endometriosis was induced (day 1), and necropsied on day 29 as detailed in Materials and Methods. Gross examination of white (1) and bleeding red (2) endometriosis lesions phenotypes with adhesions (ad). (C1–C3) Fluorescence zoomstereo microscopy examination of dissemination of 12Z-GFP and 22B-RFP cells of endometriosis lesions in the peritoneal cavity. Histomorphology of endometriosis lesions, (D1) 12Z-GFP cells formed the glands (GLE), (D2) 22B-RFP formed the stroma (STR), and (D3) established communicating glands and stroma. (E) Endometriosis nude mice were treated with EP2 (AH6809) and EP4 (AH23848) inhibitors (EP2/4-I) at 5, 10, or 25 mg/kg, i.p, at 24h intervals from days 15–28 of xenograft. Growth of endometriosis lesions was imaged in vivo real-time on days 1 (before), 7, 14, 21, and 28 of xenograft. Representative in vivo images for (F) control and (G) EP2/4-I at 25 mg/kg on day 28 are shown. (H) Gross number and volume of endometriosis lesions. (I) Fluorescence zoomstereo microscopy examination of peritoneal endometriosis lesions and (J) quantity of 12Z-GFP and 22B-RFP cells in these lesions. (K) Plasma biochemical parameters. *Control vs. EP2/4-I, P < 0.05, n = 6.

Inhibition of EP2/EP4-Induced Apoptosis of Endometriosis Lesions.

Inhibition of EP2/EP4 increased (P < 0.05) TUNEL positive epithelial as well as stromal cells in endometriosis lesions (Fig. 2 A1–A3). Next, we determined epithelial-stromal cell specific apoptosis. Inhibition of EP2/EP4 induced (P < 0.05) expression of cl-caspase-3 (Fig. 2 B1–B5) and cl-PARP (Fig. 2 C1–C5) proteins in both epithelial and stromal cells of endometriosis lesions. We examined eight lesions per mouse, among them ∼60% lesions showed increased expression of cl-caspase-3 and cl-PARP proteins and increased TUNEL-positive cells. These lesions were classified as regressing lesions and selected for further study to investigate the underlying molecular pathways through which inhibition of EP2/EP4 induced apoptosis of endometriosis lesions.

Fig. 2.

Effects of inhibition of EP2 and EP4 (EP2/4-I) on induction of apoptosis and activation of caspase-3 and PARP proteins in endometriosis lesions. (A) TUNNEL assay. (B) The 22B-RFP stromal cells (STR) were labeled with Alexa 594 (red) and cl-caspase3 protein was labeled with Alexa 488 (green) antibodies. (C) The 12Z-GFP epithelial cells (GLE) were labeled with Alexa 488 (green), and cl-PARP protein was labeled with Alexa 594 (red) antibodies. Nuclei were stained with DAPI (blue). (D) IgG controls. Peritoneal endometriosis was induced in nude mice, treated with EP2 (AH6809) and EP4 (AH23848) inhibitors (EP2/4-I) at 25 mg/kg body weight, and necropsied as detailed in Fig. 1. *Control vs. EP2/4-I, P < 0.05, n = 8.

Inhibition of EP2/EP4 Decreased Survival, Invasion, and Proinflammation Machinery Proteins in Endometriosis Lesions.

Inhibition of EP2/EP4 decreased (P < 0.05) expression of PGE₂ biosynthesis and signaling proteins COX-2, and EP2 and EP4 respectively (Fig. 3 A1–C3); proinflammatory cytokine proteins IL1β, TNFα, and IL6 (Fig. 3 D1–F3); important intracellular survival pathway proteins p-AKT, p-ERK1/2, and active β-catenin (Fig. 3 G1–I3); and invasion pathway proteins MMP2 and MMP9 (Fig. 3 J1–K3) in an epithelial-stromal cell specific as well as protein-specific pattern in endometriosis lesions.

Fig. 3.

Effects of inhibition of EP2 and EP4 (EP2/4-I) on regulation of key proteins involved in inflammation, survival, invasion, angiogenesis, and biosynthesis and signaling of PGE₂, E₂, and P₄ in endometriosis lesions. (A1–A3) COX-2. (B1–B3) EP2. (C1–C3) EP4. (D1–D3) IL1β. (E1–E3) TNFα. (F1–F3) IL6. (G1–G3) p-AKT. (H1–H3) p-ERK1/2. (I1–I3) β-catenin. (J1–J3) MMP2. (K1–K3) MMP9. (L1–L3) p450 aromatase. (M1–M3) ERα. (N1–N3) ERβ. (O1–O3) PR. (P1–P3) SF-1. (Q1–Q3) VEGF. (R1–R3) vWF proteins. (S1–S2) IgG controls. Nuclei were stained with DAPI (blue), and each protein was stained using Alexa 488 (green) or Alexa 594 (red) secondary antibodies. Peritoneal endometriosis was induced in nude mice, treated with EP2 (AH6809) and EP4 (AH23848) inhibitors (EP2/4-I) at 25 mg/kg, and necropsied as detailed in Fig. 1. *Control vs. EP2/4-I, P < 0.05, n = 8.

Inhibition of EP2/EP4 Decreased Estrogen Biosynthesis and Signaling and Increased Progesterone Signaling Machinery Proteins in Endometriosis Lesions.

Cytochrome p450 aromatase, ERα, and ERβ proteins were abundantly expressed in epithelial and stromal cells of endometriosis lesions. Inhibition of EP2/EP4 decreased (P < 0.05) p450 aromatase, ERα, and ERβ proteins (Fig. 3 L1–N3) in both cell types of endometriosis lesions. Interestingly, PR protein was not expressed in epithelial cells and expressed at very low levels in the stromal cells, which supports the typical P4-resistance state of endometriosis. Inhibition of EP2/EP4 increased (P < 0.05) expression of PR both in epithelial and stromal cells of endometriosis lesions and restored P₄-responsive state (Fig. 3 O1–O3). Surprisingly, expression of one of the important transcription factors SF1 in epithelial and stromal cells of endometriosis lesions was not modulated by EP2/EP4 inhibition (Fig. 3 P1–P3), suggesting posttranslational mechanisms.

Inhibition of EP2/EP4 Decreased Angiogenesis of Endometriosis Lesions.

The primary angiogenesis signal vascular endothelial growth factor (VEGF) was highly expressed in the stromal cells but not in epithelial cells of endometriosis lesions. Inhibition of EP2/EP4 decreased (P < 0.05) expression of VEGF protein (Fig. 3 Q1–Q3) in stromal cells. Expression pattern of the endothelial cell marker protein Von Willebrand factor (vWF) indicated the density of endothelial cells around the endometriosis lesions. Inhibition of EP2/EP4 decreased (P < 0.05) expression of vWF, and in turn, decreased existing and newly developing angiogenesis network of endometriosis lesions (Fig. 3 R1–R3).

Inhibition of EP2/EP4 Decreased Pelvic Pain in Endometriosis.

We used chimeric Rag2g(c) intact mouse model of endometriosis to investigate pain mechanisms. Xenograft 12Z-GFP and 22B-RFP cells induced peritoneal endometriosis characterized by heterogeneous white and red lesional phenotypes similar to that of nude mouse model as shown (Fig. 1 B and I). Vaginal cytology confirmed that peritoneal endometriosis progressively prolonged the E₂ phase of the estrous cycle after 3 wk of disease, suggesting a compromised estrous cycle. Inhibition of EP2/EP4 decreased (P < 0.05) the growth of endometriosis up to 60% (Fig. 4A) irrespective of E₂ or P₄ phase of the estrous cycle. Importantly, inhibition of EP2/EP4 decreased growth, survival, and dissemination of endometriosis lesions in such an E₂-dominant state. As shown in Fig. 3 above, the epithelial cells 12Z and stromal cells 22B of endometriosis lesions have inherent capacity to produce E₂. Therefore, the source of E₂-dominance appears to be the peritoneal endometriosis lesions rather than mouse ovaries in this model.

Fig. 4.

Effects of inhibition of EP2 and EP4 (EP2/4-I) on innervation of endometriosis lesions, regulation of proinflammatory machinery proteins in DRG, and nociception of pelvic pain in endometriosis. (A) Growth of endometriosis lesions measured by gross examination and morphometry (A and B) and fluorescence zoomstereo microscopy (C and D). (B) Pelvic floor referred hyperalgesia using von-Frey test. (C) Expression of neuronal markers PGP9.5, CGRP, SP, TRPV1, and VMAT proteins in endometriosis lesions. Expression of COX-2 (D), EP2 (E), EP4 (F), ILβ (G), TNFα (H), and IL6 (I) proteins in DRG neurons L1, L2, L3, L4, L5, and S1. (J) IgG controls. GLE, glandular epithelium; STR, stromal cells. Nuclei stained with DAPI (blue), each protein stained with Alexa 488 (green) or Alexa 594 (red) secondary antibodies. *Control vs. EP2/4-I, P < 0.05, n = 8. After 3 wk of disease, endometriosis Rag2g(c) mice were treated with EP2 (AH6809) and EP4 (AH23848) inhibitors (EP2/4-I) at 25 mg/kg for 2 wk. At the end of 5 wk, the mice were necropsied at the E₂ versus P₄ phase of the estrous cycle as confirmed by vaginal cytology, and data from E₂ phase shown.

Mechanical hyperalgesia was assessed by stimulating the pelvic floor with von-Frey filaments and the threshold force required to elicit a behavioral withdrawal response was determined. Peritoneal endometriosis decreased (P < 0.05) pelvic floor withdrawal threshold (reflecting increased pain to von-Frey stimulus). Inhibition of EP2/EP4 increased (P < 0.05) pelvic floor pain threshold or decreased mechanical hyperalgesia irrespective of E₂ versus P₄ phase of the estrous cycle (Fig. 4B, data from E₂ phase is shown). Interestingly, the observed pelvic floor mechanical hyperalgesia is correlated with growth of endometriosis lesions.

We examined innervation (formation of new nerve fibers) of endometriosis lesions by determining the expression of PGP9.5 (pan neuronal marker), TRPV1 (afferent nerve marker), CGRP (C-sensory nerve fiber marker), SP (Aδ-sensory nerve fiber marker), and VMAT (sympathetic nerve fiber marker) proteins around the endometriosis lesions during E₂ phase of the estrous cycle. Results (Fig. 4C) indicated that endometriosis lesions developed their own innervations (PGP9.5-positive neurites), which include sensory/afferent nerve fibers (CGRP-, SP-, TRPV1-positive neurites) and sympathetic/efferent nerve fibers (VMAT-positive neurites). Inhibition of EP2/EP4 decreased (P < 0.05) the expression of PGP9.5, TRPV1, CGRP, SP, and VMAT proteins and in turn decreased C and Aδ sensory fiber innervation of endometriosis.

Visceral primary afferent neurons (nociceptors) detect and transmit information from the pelvic region into the spinal cord. The cell bodies of these primary afferent neurons are located in the lumbosacral dorsal root ganglia (DRG). Inflammation contributes to the sensitization of the primary afferent neurons leading to the enhancement of visceral sensitivity to mechanical stimulation. Pelvic organs are primarily innervated from L1, L2, L3, L4, L5, and S1 of the spinal cord. Therefore, we next examined regulation of proinflammatory molecular markers in the DRG neurons from L1 to S1. Results indicated that PGE₂ biosynthetic and signaling machinery proteins COX-2, EP2 and EP4 (Fig. 4 D–F) and proinflammatory cytokine proteins IL1β, TNFα, and IL6 (Fig. 4 G–I) were abundantly expressed in L1, L2, L3, L4, L5, and S1 DRG neurons. Importantly, inhibition of EP2/EP4 decreased (P < 0.05) expression of these proteins in DRG neurons L1-S1 selectively at different levels.

Inhibition of EP2/EP4 Decreased E₂-Dominance and P₄-Resistance in Eutopic Endometrium in Endometriosis.

We used chimeric Rag2g(c) intact mouse model of endometriosis to investigate endocrine-immune modulations in the eutopic endometrium. Inhibition of EP2/EP4 decreased (P < 0.05) the expression of PGE₂ biosynthetic and signaling proteins COX-2, EP2, and EP4 (Fig. 5 A1–C3) and proinflammatory cytokine proteins IL1β, TNFα, and IL6 (Fig. 5 D1–F3) proteins in epithelial and stromal cells of endometrium. The key E₂ biosynthesis enzyme p450 aromatase protein was only expressed in stromal cells but not in the epithelial cells of endometrium. Interestingly, inhibition of EP2/EP4 decreased (P < 0.05) expression of p450 aromatase protein (Fig. 5 G1–G3) in stromal cells and concomitantly increased its expression in epithelial cells of endometrium. ERα protein (Fig. 5 H1–H3) was expressed in both epithelial and stromal cells, and inhibition of EP2 and EP4 decreased (P < 0.05) its expression in both cell types of endometrium. ERβ protein (Fig. 5 I1–I3) predominantly expressed in epithelial cells compared with stromal cells, and inhibition of EP2 and EP4 decreased (P < 0.05) its expression in both cell types of endometrium. PR protein (Fig. 5 J1–J3) was not expressed in stromal cells but was expressed in glandular epithelial cells. Importantly, inhibition of EP2/EP4 increased/restored (P < 0.05) its expression in stromal cells and concomitantly decreased its expression in glandular epithelial cells of endometrium.

Fig. 5.

Effects of inhibition of EP2 and EP4 (EP2/4-I) on regulation of key proteins involved in biosynthesis and signaling of PGE₂, E₂, and P₄, and proinflammation in eutopic endometrium in endometriosis. (A1–A3) COX-2. (B1–B3) EP2. (C1–C3) EP4. (D1–D3) ILβ. (E1–E3) TNFα. (F1–F3) IL6. (G1–G3) p450 aromatase (p450arom). (H1–H3) ERα. (I1–I3) ERβ. (J1–J2) PR protein. (K1–K2) IgG controls. Nuclei were stained with DAPI (blue), each protein was stained with Alexa 488 (green) secondary antibody. GLE, glandular epithelium; LE, luminal epithelium; STR, stromal cells. *Control vs. EP2/4-I, P < 0.05, n = 8. Peritoneal endometriosis was induced in Rag2g(c) mice, treated with EP2 (AH6809) and EP4 (AH23848) inhibitors (EP2/4-I) at 25 mg/kg, necropsied as detailed in Fig. 4, and data from E₂ phase shown.

Discussion

PGE₂ plays an important role in the pathogenesis of endometriosis (6–9). COX-2, EP2, EP4, p-ERK1/2, p-AKT, β-catenin, MMP2, and MMP9 proteins are highly expressed/activated in endometriosis lesions in women (11–15). PGE₂ increases SF1, StAR, and p450 aromatase genes and stimulates de novo biosynthesis of E₂ which in turn increases PGE₂ biosynthesis in endometriosis lesions (1). PGE₂ and E₂ (ERα and ERβ) interactive pathways appear to suppress PR expression leading to loss of P₄ action and P₄-resistance in endometriosis lesions (1, 16). In addition, endometriosis lesions secrete proinflammatory cytokines (17). COX-2/PGE₂ induces VEGF and promotes angiogenesis of endometriosis lesions (8).

Results of the present study indicate that inhibition of EP2/EP4 suppresses survival, invasion, biosynthesis, and signaling of PGE₂ and E₂, production of proinflammatory cytokines, and increases P₄ signaling machinery proteins in an epithelial and stromal cell-specific pattern in endometriosis lesions. These results together establish that inhibition of EP2/EP4 restores appropriate cross-talk among PGE₂, E₂, and P₄ pathways and converts the proinflammatory, E₂-dominant, and P₄-resistant state into a P₄-responsive state and thus decreases growth, survival, and dissemination of peritoneal endometriosis lesions. An inverse relationship among EP2, EP4, p450 aromatase, ERα, ERβ, and PR proteins in epithelial and stromal cells suggests that PGE₂ and E₂ interactive pathways are the important regulators of PR signaling in endometriosis lesions. However, the underlying molecular interactive mechanisms among EP2/EP4, ER, and PR in endometriosis lesions in an epithelial-stromal cell specific pattern are largely unknown.

The mechanisms underlying the detection and transmission of nociceptive signals from endometriosis lesions are also largely unknown. i.p. concentrations of PGE₂ (18), IL1β, and TNFα (19) are higher in women with endometriosis. PGE₂ acts on the peripheral nociceptors and increases responsiveness to peripheral stimuli through TRPV1 and induces chronic inflammatory pain through EP2 and EP4 (20, 21). Localized peripheral inflammation increases expression of COX-2 protein in the spinal cord (22) and DRG neurons (23) and EP4 protein in L5 DRG neurons (20, 21). Pharmacological and genomic inhibition of COX-2 (22) or EP4 (20, 21) decreases inflammatory pain hypersensitivity. Furthermore, activation of EP2 or EP4 increases TRPV1 activity in DRG neurons (20, 21). Inhibition of EP4 decreases PGE₂-induced sensitization of DRG neurons and release of neuropeptides SP and CGRP (20, 21). Advanced active endometriosis lesions are innervated by sympathetic and sensory C and Aδ fibers and they express CGRP and TRPV1 proteins in a rat allograft model (24).

Results of the present study indicate that endometriosis lesions induce innervation of C and Aδ afferent sensory and efferent nerve fibers around the endometriosis lesions, promote proinflammatory microenvironment of DRG neurons from L1-S1 which innervate pelvis and pelvic organs, and increase pelvic floor hyperalgesia. Inhibition of EP2/EP4 decreases innervation of endometriosis lesions, decreases the proinflammatory state of DRG neurons, and suppresses peripheral nociception. These results together establish that the inflammatory response of endometriosis lesions initiates a cascade of events resulting in enhancement of the responsiveness of nociceptors and DRG neurons, and inhibition of EP2/EP4 suppresses these pain pathways and decreases endometriosis-induced pelvic pain. In women with endometriosis, the pelvic pain can be assessed by bilateral hyperalgesia and allodynia from T9-S1 spinal segments (25) and mechanical hyperalgesia of the pelvis, where pain is reported in response to applied pressure. Although the von-Frey DRG mice models are widely used to measure hyperalgesia in pain models, we used this interactive approach, to our knowledge for the first time, to measure the pelvic floor pain threshold in endometriosis chimeric mouse model with translational relevance.

The underlying causes of endometriosis-induced infertility remain unclear and are likely multifactorial. In women with endometriosis, the microenvironment of the endometrium becomes proinflammatory, E₂-dominant, and P₄-resistant, and thereby impairs receptivity for establishment of pregnancy (1, 2, 26, 27). The endometrium first becomes hyperresponsive to E₂ and then resistant to P₄ (1, 16, 28). The loss of normal endometrial function becomes more evident with progression of disease (28). The disruption of these signaling processes in endometrial epithelial and stromal cells appears to be a consequence of growth of peritoneal endometriosis lesions, which might be orchestrated by progressive deregulated interactions among PGE₂, E₂, and P₄.

Results of the present study indicate that inhibition of EP2/EP4 decreases PGE₂, E₂ biosynthesis and signaling, proinflammatory cytokine production, and increases P₄ signaling machinery proteins in an epithelial-stromal cell-specific pattern in the eutopic endometrium. These results together demonstrate that inhibition of EP2/EP4 suppresses E₂-dominant state and concomitantly increases P₄-responsive state of the endometrium. Thus, it restores the ability of the endometrium respond appropriately to PGE₂, E₂, and P₄ and its functional receptivity for establishment of pregnancy. Specific inverse regulatory relationship among COX-2, EP2/EP4, p450 aromatase, ERα, ERβ, and PR in epithelial and stromal cells of the endometrium suggest that PGE₂-E₂ interactive pathways are the important regulators of PR and p450 aromatase. However, the underlying molecular and cellular mechanisms that lead to restoration of PR and suppression of ER expression in an epithelial-stromal cell specific pattern are largely unknown.

In conclusion, results of the present study, to our knowledge for the first time, indicate that selective pharmacological inhibition of EP2/EP4: (i) suppresses the growth and survival of endometriosis lesions; (ii) decreases endometriosis-induced innervation and pelvic pain; (iii) decreases proinflammatory, E₂-dominant, and P₄-resistant molecular environment of the endometrium and endometriosis lesions; and (iv) restores endometrial functional receptivity through multiple mechanisms (Fig. S1). Our novel results provide molecular and preclinical basis to formulate personalized phenotype-based long-term nonestrogen or nonsteroidal therapy through inhibition of EP2/EP4 as a more effective alternative treatment to existing hormonal therapies for endometriosis. At present, no therapeutic agents are available to inhibit EP2/EP4 receptors for the treatment of endometriosis or other inflammatory diseases in human medicine. Discovery of specific targeted drugs, injectable antibodies or small molecules to inhibit EP2/EP4 is expected to provide new clinical trials and open new area of research in endometriosis. Findings of the present study provide new knowledge that fills an important gap in the current understanding of the pathogenesis, diagnosis, and treatment of endometriosis.

Fig. S1.

Proposed mechanism through which selective inhibition of EP2/EP4 induces apoptosis of endometriosis lesions, decreases pelvic pain, and improves endometrial receptivity in endometriosis. Endometriosis Lesions: (1) Inhibition of EP2/EP4 inhibits PGE₂-EP2/EP4 signaling in (2) epithelial and stromal cells of endometriosis lesions; (3) suppresses survival and invasion and increases intrinsic apoptosis pathways; (4) inhibits endometriosis-induced neoangiogenesis; (5) decreases production of proinflammatory mediators; (6) decreases E₂ biosynthesis and signaling and suppresses E₂-dominance; (7) increases PR signaling, decreases P₄ resistance, and restores P₄-responsive state; and together lead to (8) apoptosis of epithelial and stromal cells and regression of endometriosis lesions. Pelvic Pain: Inhibition of EP2/EP4 by either (9) suppressing the growth of peritoneal endometriosis lesions and their secretory products or (10) directly acting on the nociceptors and DRG neurons (11) decreases afferent/sensory and efferent/sympathetic innervation of endometriosis lesions; (12) decreases proinflammatory microenvironment of DRG (L1-S1); and (13) suppresses peripheral nociception pain pathways and decreases endometriosis-induced pelvic pain. Eutopic Endometrium: Inhibition of EP2/EP4 by either (9) suppressing the growth of peritoneal endometriosis lesions and their secretory products or (14) directly acting on the endometrial epithelial and stromal cells (15) inhibits production of proinflammatory mediators; (16) decreases E₂ biosynthesis and signaling and suppresses E₂ dominance; (17) increases PR signaling and decreases P₄ resistance and restores P₄-responsive state; and (18) improves endometrial receptivity machinery.

Materials and Methods

Human fluorescent endometriotic epithelial cell line 12Z and stromal cell line 22B were xenografted and peritoneal endometriosis was induced in immunocompromised mice. The experimental mice were treated with EP2 and EP4 inhibitors. Effects of treatment on growth, survival, pain, and infertility and the underlying molecular mechanisms were determined (29–37). All procedures were approved by the Institutional Animal Care and Use Committee at Texas A&M University. Detail methods (Fig. S2 and Table S1) are provided in the SI Materials and Methods.

Fig. S2.

GFP and RFP vector map.

Table S1.

Details of antibodies used

Antibodies	Manufacturer	Catalog no.	Concentration used, ICC
Anti-human monoclonal rabbit pAKT	Cell Signaling	4060	1:100
Anti-human monoclonal rabbit pERK1/2	Cell Signaling	4370	1:100
Anti-human polyclonal rabbit β-catenin	Cell Signaling	9562	1:100
Anti-human polyclonal rabbit MMP2	Santa Cruz	sc-10736	1:200
Anti-human polyclonal goat MMP9	Santa Cruz	sc-6840	1:200
Anti-human polyclonal rabbit IL1β	Abcam	ab2105	1:250
Anti-human polyclonal rabbit TNFα	Abcam	ab6671	1:100
Anti-human polyclonal rabbit IL6	Abcam	ab6672	1:250
Anti-human polyclonal rabbit COX-2	Cayman	160107	1:250
Anti-human polyclonal rabbit EP2	Cayman	158441	1:250
Anti-human polyclonal rabbit EP4	Cayman	101775	1:250
Anti-human polyclonal rabbit p450arom	Santa Cruz	sc-20772	1:200
Anti-human polyclonal rabbit ERα	Santa Cruz	sc-542	1:50
Anti-human polyclonal rabbit ERβ	Thermo Scientific	PA1-311	1:100
Anti-human monoclonal rabbit PR A-B	Cell Signaling	3153	1:200
Anti-human polyclonal rabbit SF1	Sigma	HPA018883	1:200
Anti-human monoclonal mouse VEGF	Santa Cruz	sc-7269	1:50
Anti-human polyclonal rabbit Von Willebrand factor (vWF)	Dako	A0082	1:200
Anti-human monoclonal mouse PGP9.5	Abcam	ab1189	1:50
Anti-rat polyclonal rabbit CGRP	Abcam	ab47027	1:1,000
Anti-human monoclonal mouse SP	Abcam	ab14184	1:1,000
Anti-human polyclonal rabbit TRPV1	Abcam	ab74813	1:250
Anti-human polyclonal rabbit VMAT	Abcam	ab87589	1:300
Anti-human polyclonal rabbit cl-Caspase3	Cell Signaling	9661	1:250
Anti-human monoclonal mouse cl-PARP	Abcam	ab110315	1:500
Anti-human polyclonal rabbit ZsGreen	Clontech	632474	1:250
Anti-human monoclonal mouse dsRED	Clontech	632393	1:1,000
Anti-human polyclonal rabbit dsRED	Clontech	632496	1:2,000
Anti-Mouse goat polyclonal IgG1 Secondary Antibody, Alexa Fluor 488 conjugate	Invitrogen	A21121	1:250
Anti-Rabbit goat polyclonal IgG (H+L) Secondary Antibody, Alexa Fluor 488 conjugate	Invitrogen	A11008	1:250
Anti-Mouse goat polyclonal IgG (H+L) Secondary Antibody, Alexa Fluor 594 conjugate	Invitrogen	A11032	1:500
Anti-Rabbit goat polyclonal IgG (H+L) Secondary Antibody, Alexa Fluor 594 conjugate	Invitrogen	A11037	1:500

SI Materials and Methods

Materials.

General chemicals and reagents used in the study were molecular and cell biological grade from Sigma-Aldrich, Fisher Scientific, VWR International, LLC, or Invitrogen Life Technologies.

Human Endometriotic Cell Lines.

Immortalized endometriotic epithelial cell line 12Z and stromal cell line 22B used in this study were derived from active red peritoneal endometriosis lesions from woman suffering from endometriosis for more than 8 y during the proliferative phase of the menstrual cycle (29). These 12Z and 22B cells share several phenotypic and molecular characteristics of primary cultured endometriotic cells (29). Accumulating information from our and other laboratories indicate that 12Z and 22B cells mimic the active/progressive phase of endometriosis (11, 13, 14, 29). Importantly, xenograft of a mixed population of these 12Z and 22B cells into the peritoneal cavity of nude mice are able to proliferate, attach, invade, reorganize, and establish peritoneal endometriosis-like lesions and that histomorphology is similar to that of spontaneous peritoneal endometriosis in women (30). We have shown that 12Z and 22B cells produce large amounts of PGE₂ at basal conditions. Therefore, inhibition of EP2 and EP4 is the best approach rather than treating the cells with PGE₂ to investigate PGE2-EP2/EP4 signaling. As expected, inhibition of EP2 and EP4 decreased their adhesion, invasion, growth, and survival of 12Z and 22B cells in vitro (11–14).

Lentiviral Transduction and Transgenic 12Z-GFP and 22B-RFP Cells.

The backbone for the construct described here was based on the Neomycin - EF1a mCherry (NEC) vector provided by Michael Golding as described (31). The NEC vector was cut with Nhe1 and Not1, and GFP was removed and replaced with either ZSGreen or dsRED to produce NEF-Green and NEF-Red, respectively (Fig. S2). The endometriotic epithelial cell line 12Z was transduced with lentivirus containing NEF-Green plasmid and established stable 12Z-GFP cell line. The endometriotic stromal cell line 22B was transduced with lentivirus NEF-Red and established stable 22B-RFP cell line. Recombinant lentivirus was produced as described (32). Briefly, lentiviral plasmid (NEF-Green or NEF-Red) was cotransfected into HEK 293T cells along with pCMV R8.91 (packaging plasmid) and pMDG (vesicular stomatitis virus glycoprotein plasmid) via calcium phosphate transfection, as described (31). Media was changed 24 h posttransfection, and supernatant was collected 48 h later. Supernatants were centrifuged (2,000 × g for 20 min) to remove cellular material and subsequently filtered through a 0.45-μm filter. Supernatant containing the recombinant lentiviral particles was either stored at 4 °C until use or frozen in aliquots at −80 °C. The 12Z and 22B cell lines were exposed to recombinant lentivirus during overnight incubation and selected using neomycin starting 3 d posttransduction. Expression of the fluorescent marker and resistance to neomycin confirmed expression of the integrated transgene.

Culture of 12Z-GFP and 22B-RFP Cells.

Human endometriotic epithelial cells 12Z-GFP and stromal cells 22B-RFP were cultured in DMEM/F12 without special steroid treatment containing 10% FBS (HyClone) and penicillin (100 U/mL), streptomycin (100 µg/mL), and amphotericin-B 2.5 µg/mL in a humidified 5% CO₂ and 95% air at 37 °C as we described (11–14). At 70% confluency, the 12Z-GFP and 22B-RFP cells were processed for xenograft as we described (30).

Chimeric Nude Mouse Model of Endometriosis.

Five-week old bilateral ovariectomized nude mice (HSD-athymic nude-foxn1^Nu) were purchased from Harlan. Nude mice were housed in microisolator cages in a pathogen-free facility at a monitored ambient temperature of 23 °C in the Laboratory Animal Research and Resources (LARR) Core Facility, Comparative Medicine Program, Texas A&M University. Nude mice were maintained under a standard 12-h photoperiod and fed with mouse diet 2819 and free access to water. Animal husbandry was carried out under laminar-flow hoods. All procedures were approved by the Institutional Animal Care and Use Committee at Texas A&M University. At 5 wk of age, sterile 60-d–release pellets containing 0.72 mg of 17β-E₂ (Innovative Research of America) were implanted s.c. on the back of each ovariectomized nude mouse. After 2 d, the human endometriotic cells were xenografted as we described (30). Stable 12Z-GFP cells (3 × 10⁶) and 22B-RFP (0.5 × 10⁶) cells were mixed in 250 μL of DMEM/F12 and 50 μL of matrigel, to maintain the physical contact and communication among epithelial and stromal cells. The mixture cell suspension was injected into the peritoneal cavity of nude mice to mimic the condition of retrograde menstruation in women and peritoneal endometriosis was induced, as we reported previously (30). Day of xenograft was considered as day 1. From days 15–28 of xenograft, the experimental endometriosis nude mice (n = 6 per dose) were treated with inhibitors for EP2 (AH6809) and EP4 (AH23848) at 0, 5, 10, and 25 mg/kg body weight. Growth of endometriosis lesions were monitored on days 1 (before), 7, 14, 21, and 28 by NightOWL LB 983 in-vivo Imaging System as described below. The experimental endometriosis nude mice were necropsied on day 29 of xenograft. The growth, volume and dissemination of peritoneal endometriosis of lesions were measured by gross visual examination and fluorescence dissection microscopy as described below.

Chimeric Rag2g(c) Mouse Model of Endometriosis.

Chronic pelvic pain and infertility are the two major clinical symptoms of endometriosis in women. We used chimeric Rag2g(c) intact-mouse model to investigate the association between endometriosis and these symptoms. Rationale is that xenografted human endometriotic epithelial cells 12Z-GFP and stromal cells 22B-RFP can be maintained for more than four months in intact Rag2g(c) mice compared with 5 wk (5–10 wk) in ovariectomized-estrogenized nude mice. Rag2g(c) intact-mouse model allows more therapeutic interventions and assessment of more end–points during estrous cycle or adult ages. Rag2g(c) mice were purchased (Taconic Biosciences) and then breeding colony was established, housed, and maintained at LARR, Texas A&M University as described above. At 8 wk of age, peritoneal endometriosis was induced by xenograft of 12Z-GFP and 22B-RFP cells, as described above. Day of xenograft was considered as day 1. From days 21–35 of xenograft, the experimental endometriosis Rag2g(c) mice (n = 8 per group) were treated with inhibitors for EP2 (AH6809) and EP4 (AH23848) at 25 mg/kg body weight (effective dose chosen from dose–response experiment above); and necropsied on day 36 of xenograft at E₂ phase confirmed by vaginal cytology. The growth, volume, and dissemination of peritoneal endometriosis of lesions were measured by gross visual examination and fluorescence dissection microscopy as described below. Vaginal cytology (33) was performed every day from 21 to +36 (xenograft day 1). Mechanical hyperalgesia was assessed by stimulating the pelvic floor with von-Frey filaments on days 0 (WK0), 7 (WK1), 14 (WK2), 21 (WK3), 28 (WK4), and 35 (WK5) as described below.

Evaluation of Endometriosis Lesions and Collection of Tissues.

Nude and Rag2g(c) experimental endometriosis mice were euthanized as we described (30). The abdominal cavity was opened and presence of endometriosis lesions was examined by gross visual examination. The color (red or white), number, and anatomical location of the endometriosis lesions were recorded and tracked. Grossly, the experimental endometriosis lesions were measured in two dimensions, the larger denoted “a” and the smaller denoted “b,” and total volume, calculated using the formula V = a × b² × 0.5 (34). Then, the entire abdominal cavity was examined under fluorescence zoomstereo dissection microscope (Nikon AZ100) to determine the dissemination of 12Z-GFP and 22B-RFP clusters of endometriosis lesions, imaged, and tracked as described below. All of the lesions were dissected under the fluorescence zoomstereo dissection microscope and care was taken not to include the underlying peritoneal tissues. Then, uterus was collected from each mouse. Portions of endometriosis lesions and uterus were embedded in Optimal Cutting Temperature (OCT) compound and cryopreserved.

Dissection of Dorsal Root Ganglia (DRGs).

The DRGs were isolated as described (35) with modifications. Rag2g(c) experimental endometriosis mice were euthanized as we described (30). The skin and abdominal organs were removed. The spinal cord was cut from T10-S5. Under the dissection microscope, the muscles around the spinal cord were trimmed off for the entire length. Then, spinal cord was cut more precisely from L1 to S1. The bone around dorsal aspect of the spinal cord was cut using sharp scissors at 2 and 10 clock positions from L1 to S1, the bone was lifted and removed, and thus spinal cord was opened. During this microsurgery, care was taken not to destroy the bilateral DRG and spinal cord. The exposed spinal cord was placed in a labeled cassette and fixed in 2% paraformaldehyde (PFA) overnight (16–18 h) at 4 °C. Next day under the discretion microscope, the remaining bone at basal aspect of the spinal cord was removed. The specific DRGs were identified from L1 to S1, dissected, labeled, embedded in OCT, and cryopreserved.

Mechanical Referred Hyperalgesia.

Withdrawal responses to the application of von Frey filaments (Stoelting) to the pelvic floor/pelvic abdomen were examined as a measure of referred hyperalgesia. The xenografted Rag2g(c) mice (8–10 mice as group) were habituated in a Plexiglas chamber fitted with a raised wire mesh (5 × 5 mm apertures) under a clear Pyrex container/mouse (inner diameter 8.75 cm, height 4.5 cm) for 60 min before the von Frey testing. The von Frey filaments were applied through the wire mesh to the pelvic abdomen of the freely moving animals and care was taken to avoid external genitalia. The von Frey filament force was given to each mouse in ascending order (0.008, 0.02, 0.04, 0.07, 0.16, 0.4 g) three times each for 5 s with an interstimulus interval of 10 min. During von Frey testing, care was taken not to stimulate the same point twice in succession and to avoid desensitization or “windup” effects. A withdrawal response to a von Frey filament force was defined as (i) sharp retraction of abdomen, (ii) immediate licking or scratching of site of application of von Frey filament force, or (iii) jumping. The threshold was defined as the force (grams) that was represented by the von Frey filament that elicited three consecutive responses.

Immunocytochemistry (ICC).

Immunocytochemistry was performed according to the protocol provided by Cell Signaling Technology. The endometriosis lesions and uterus cryosections (10 μm) were fixed in 2% PFA for 15 min at room temperature and followed by fixed in methanol for 10 min at 4 °C. The DRG sections were by fixed in methanol for 10 min at 4 °C. The tissue sections were incubated with primary antibodies for overnight at 4 °C. The sections were further incubated with Alexa Fluor 488 and Alexa Fluor 594 conjugated secondary antibodies for 60 min at room temperature. Nuclei were stained with DAPI (ProLong Gold antifade, Molecular Probes). Details of the antibodies and concentrations used are given in Table S1. For the negative control, serum or IgG from respective species with reference to the primary antibody at the respective dilution was used.

Digital images were captured using a Zeiss Axioplan 2 Research Microscope (Carl Zeiss) with an Axiocam HR digital color camera. The intensity of staining for each protein was quantified using Image-Pro Plus 6.3 image processing and analysis software according to the manufacturer’s instructions (Media Cybernetics). The detailed methods for quantification are given in the instruction guide: “The Image-Pro Plus: The proven solution for image analysis.” In brief: a minimum three images of at 400× magnification were captured randomly without hot-spot bias in each tissue section per animal. The integrated optical intensity (IOD) of immunostaining was quantified under RGB mode and expressed in 1OD × 10³ unit. Numerical data were expressed as least square mean ± SEM. This technique is more quantitative than conventional blind scoring systems, and the validity of quantification was reported by our group (36).

Real-Time In Vivo Fluorescence Bioimaging- NightOWL II LB 983.

The 12Z-GFP and 22B-RFP xenografted experimental endometriosis nude mice were anesthetized using Isoflurane. The anesthetized mice were placed abdomen up in the chamber and imaged for GFP fluorescence at appropriate excitation and emission using a NightOWL II LB 983 NC100 (Berthold Technologies) to assess growth of endometriosis at days 1 (before), 7, 14, 21, 28 of postxenograft. The excitation source is a ring light used for epi-illumination, mounted 12 cm above the mice. Using the IndiGO software (Berthold Technologies), background and autofluorescence were corrected as per the manufacturer’s procedures. Automated peaksearch function was used to delineate the area of the fluorescence of 12Z-GFP in endometriosis lesions in the peritoneal cavity. Average brightness of the areas was measured and subtracted from the background value. Mean average brightness in control mice was considered as 100%. Based on control values, effects of EP2/EP4 inhibition (treatment) was calculated, and expressed in numerical data. More details on imaging and quantification methods are available in the operating manual LB983 NightOWL II with indiGO software and have been reported (37).

Fluorescence Stereo Microscopy Imaging.

The 12Z-GFP and 22B-RFP xenografted experimental endometriosis nude and Rag2g(c) mice were euthanized, skin removed, and abdominal cavity opened. The entire abdominal cavity was examined under fluorescence zoomstereo dissection microscope (Nikon AZ100) to determine the disseminated clusters of 12Z-GFP and 22B-RFP cells of endometriosis lesions. The fluorescent endometriosis lesions were recorded, tracked, and images captured under GFP and RFP filters at 1× magnification. Intensity of GFP and RFP in each image (clusters of lesions) was quantified using Image-Pro Plus as described above and expressed in numerical data. The Nikon AZ100 Fluorescence stereomicroscope is equipped with AZ100 Plan Fluor Objectives 1×, 2× and 5×, fluorescent light source-excite series 120 PC, Nikon DS QiMc digital camera, and Nikon NIS Elements BR 3.22 software.

Statistical Analyses.

Statistical analyses were performed using general linear models of Statistical Analysis System (SAS). Effects of inhibition of EP2 and EP4 on growth and volume endometriosis and expression levels of different proteins in endometriosis lesions, DRG, endometrium were analyzed by one-way analysis of variance (ANOVA) followed by Tukey–Kramer HSD test. Effects of inhibition of EP2 and EP4 (within treatment) and dose and time interactions on growth of endometriosis, and treatment and time interactions on hyperalgesia were analyzed by repeated measures for multivariate ANOVA and followed by F-test and Wilks’ Lambda and/or Hotelling-Lawley tests. The numerical data are expressed as the mean ± SEM. Statistical significance was considered at P < 0.05.