|Home | About | Journals | Submit | Contact Us | Français|
Endometriosis is an incurable gynecological disorder characterized by debilitating pain and the establishment of innervated endometriosis lesions outside the uterus. In a preclinical mouse model of endometriosis we demonstrated overexpression of the PGE2-signaling pathway (including COX-2, EP2, EP4) in endometriosis lesions, dorsal root ganglia (DRG), spinal cord, thalamus and forebrain. TRPV1, a PGE2-regulated channel in nociceptive neurons was also increased in the DRG. These findings support the concept that an amplification process occurs along the pain neuroaxis in endometriosis. We then tested TRPV1, EP2, and EP4 receptor antagonists: The EP2 antagonist was the most efficient analgesic, reducing primary hyperalgesia by 80% and secondary hyperalgesia by 40%. In this study we demonstrate reversible peripheral and central hyperalgesia in mice with induced endometriosis.
Endometriosis is a chronic gynecological disorder affecting 176 million women worldwide1 associated with chronic pain and infertility. Current therapies include invasive surgery, drugs that suppress endogenous hormones2 and non-steroidal anti-inflammatory drugs (NSAIDs) all of which have unwanted side effects. Better treatments for endometriosis-associated pain are needed but their development has been hampered by the lack of a robust preclinical model that fully reproduces the altered pain perception experienced by women with endometriosis.
Endometriosis is caused by the presence of endometrial-like tissue (endometriosis lesions) outside of the uterine cavity3. An association between small nerve fiber infiltration of endometriosis lesions and increased central and peripheral pain has been described (reviewed in ref. 4). Alterations in pain perception in women are thought to involve release of inflammatory mediators and neuropeptides by efferent peripheral nerve endings5, an increase in the sensitivity of nociceptive neurons6, and peripheral hyperalgesia. Continuous input from peripheral afferents can also trigger spinal hyper-excitability (central sensitization)7 resulting in increased pain perception, secondary hyperalgesia and allodynia8. Prostaglandin E2 (PGE2) is a well-established mediator of inflammation and nociception in inflammatory9,10 and neuropathic pain conditions11 and synthesis of prostaglandins within endometriosis lesions has previously been reported12,13,14,15. As the association of PGE2 signaling and endometriosis is well established we chose to validate this pathway as a target for pain attenuation in our mouse model.
A rat model of endometriosis exhibits vaginal hyperalgesia and increased abdominal muscle activity16,17,18. Recently, a study utilizing the same rat model demonstrated increased hyperalgesia in response to von Frey filaments, although an increase was also observed in the sham group19. Another model using autologous transplantation of endometrial tissue onto the gastrocnemius muscle20 has proved useful for the identification of pronociceptive molecules that may be relevant in endometriosis21. Our murine model of endometriosis uses decidualized endometrial tissue injected (rather than sutured) into the peritoneum of recipient mice22. Resultant lesions phenocopy those in women; they are vascularized, innervated, infiltrated by macrophages and also exhibit oestrogen-dependent regulation of vascular-nerve and macrophage-nerve interactions23,24. The aim of our study was to determine if mice with induced endometriosis exhibit peripheral (abdominal) and central (secondary/referred) hyperalgesia and to test compounds targeting prostaglandin receptors to validate this model as a platform for preclinical testing of compounds to treat pain.
We confirmed lesions in 90% of endometriosis mice (Endo) and those that we did not recover lesions from were excluded from our analysis. The average number of lesions recovered was 1.9 (Supplementary Table 1). Endo mice had significantly higher levels of abdominally-directed licking (Fig. 1a) and decreased exploratory activity (Fig. 1b) compared to controls (p<0.001). Endo mice had significantly lower mechanical withdrawal thresholds for von Frey filaments, not only on the abdomen (Fig. 1c) but also on the hind-paws (Fig. 1d; p<0.05 and p<0.01, respectively). We found no correlation between mechanical allodynia (hypersensitivity) and number of lesions in mice with endometriosis (Supplementary Fig. 1a,b)
Consistent with our previous studies22,23,24 endometriosis-like lesions were recovered from the walls of the parietal peritoneum and the visceral peritoneum covering the uterus, gut, and intestines; mesentery associated with the gut and intestines; adipose tissue associated with the kidney; and underneath the kidneys. EP2, EP4, cyclooxygenase-1 (COX-1) and cyclooxygenase-2 (COX-2) were significantly increased in endometriosis lesions compared to the peritoneum of naive mice (NP) or mice with endometriosis (EP; p<0.01; Supplementary Fig. 2a–d); EP2, EP4, COX-1 and COX-2 proteins were immunolocalised to glandular and stromal cells in lesions and mesothelial cells in the peritoneum (Supplementary Fig. 2). PGE2 concentrations in the peritoneal fluid of Endo mice were significantly increased (p<0.05; Supplementary Fig. 2E). mRNA concentrations of EP1 and EP3 were unchanged (Supplementary Fig. 2f,g). EP2, Cox-1, Scn11a and Trpv1 mRNA concentrations were significantly increased in dorsal root ganglia (DRG; clusters of cell bodies of afferent sensory neurons that transmit noxious stimuli from the periphery to the spinal cord) from endometriosis mice (p<0.05; Fig. 2a–d); EP4 and Cox-2 were unchanged (Supplementary Fig. 3c,d). Trpv1 mRNA was also increased in the DRGs from OVX+E2 mice. Dual immunofluorescence showed that in small DRG cells expressing peripherin (a marker for small unmyelinated sensory neurons; C-fibre nociceptors), the proportion that were immunopositive for TRPV1 was significantly increased in Endo mice compared to naïve and OVX+E2-treated controls (p<0.05; Fig. 2e,f). The proportion of TRPV1-immunopositive neurons in DRG that express EP2 was also increased in these animals (Supplementary Fig. 3e,f). Mice with endometriosis had significantly increased concentrations of COX-2 protein in the spinal dorsal horn (p<0.001), the thalamus (p<0.001) and the anterior cingulate cortex of the brain (p<0.001; Fig. 2f,i). All of these changes would be expected to contribute to pain hypersensitivity.
Figure 3 indicates that injection of the TRPV1 inhibitor JNJ 17203212 (Fig. 3a,b) or the EP4 antagonist L-161982 (Fig. 3c,d), did not reverse abdominal or paw hyperalgesia to a statistically significant extent. Administration of the EP2 antagonist TG6-10-1 resulted in a statistically significant reversal of mechanical allodynia as tested on the abdomen (p<0.05) at 45mins post administration but this did not reach significance for the hind paw (Fig. 3e,f). Time-course graphs are shown in Supplementary Fig. 4. Results were extended using injection (Fig. 3g,h) or oral administration (Fig. 4a,b) of a second EP2 antagonist (PF-04418948)25 and this reduced allodynia in both abdomen and hind-paw tests (p<0.001 using either route). Oral administration resulted in striking, time dependent, and significant impacts (Fig. 4a,b, p<0.001).
The re-purposing of drugs and the development of novel treatments for endometriosis-associated pelvic pain has been limited by the paucity of accessible pre-clinical models. In this study we tested pain responses in a mouse model of endometriosis that phenocopies interactions between endometriosis lesions and peritoneal tissue22,24 identified in humans. Endometriosis (Endo) mice exhibited increased levels of abdominally-directed grooming, a reduction in normal exploratory behavior and reduced mechanical withdrawal thresholds on both the abdomen and plantar hind-paw. The rodent models reported by Berkley, and by Levine and Giudice, both report evidence of hyperalgesia16,20,26,27. These models involve the artificial implantation of uterine tissue (full thickness i.e. endometrium plus myometrium) onto the mesenteric arteries of the small intestine and the gastrocnemius muscle, respectively. Therefore the microenvironment of the endometriosis lesion created in these models may not closely mirror that of the peripheral lesions in women.
PGE2 is increased in the peritoneal fluid of patients with endometriosis28, this up-regulation results from induced expression of cyclooxygenase-2 (COX-2) in endometriotic tissue. PGE2 is thought to be a key player in the pathophysiology of endometriosis and studies have shown that inhibition of COX-2 decreases survival, migration and invasion of endometriotic cells13. The same authors demonstrated that inhibition of EP2 and EP4 can inhibit the epithelial and stromal cell invasion via suppression of matrix metalloproteinases29. Attenuation of PGE2 signaling via lipoxin A4 can also modulate disease progression by attenuation of pro-inflammatory and angiogenic mediators30. A recent study using xenografted endometriotic cell lines in a nude mouse model of endometriosis demonstrated that dual inhibition of EP2 and EP4 could attenuate mechanical hyperlagesia of the pelvic floor via suppression of pro-inflammatory mediators in dorsal root ganglia31, however the authors did not test secondary hyperalgesia or analyze changes in the central nervous system.
In our mouse model we confirmed that the prostaglandin E signaling pathway was over-expressed in the pelvic cavity of our Endo mice and concentrations of PGE2 were increased in their peritoneal fluid. Release of PGE2 at sites of peripheral inflammation can contribute to pain hypersensitivity by lowering the threshold and enhancing the excitability of nociceptor sensory fibers32. This occurs at least in part via EP receptor-mediated activation of intracellular kinases in the nociceptor terminal that causes phosphorylation of the nociceptive ion channel TRPV133,34 and an up-regulation of Nav1.9 voltage-gated sodium channel (SCN11A)35, which is then transported to peripheral nerve terminals to contribute to increased excitability7. We detected a significant increase in expression of EP2, COX-1 and both TRPV1 and SCN11A ion channels in the DRGs of mice with endometriosis. This parallels observations of increased COX-1 and COX-2 in TRPV1-positive DRG cells in other models of inflammatory hyperalgesia36. TRPV1-immunoreactivity was increased in small, peripherin-positive, nociceptive neurons in DRG of mice with endometriosis and EP2 expression was further increased in these cells. All of these findings are consistent with the development of sensory neuron hyperexcitability in our mice. The possibility of some additional role of prostaglandin signaling in non-neuronal cells, as seen in some other pain models37 cannot be excluded. Elevated TRPV1 expression in small DRG cells innervating pelvic regions in mice with endometriosis is consistent with our previous findings that TRPV1 mRNA is elevated in peritoneal lesions from women with endometriosis38.
In this study, assessment of COX-2 expression in CNS regions within the pain-processing pathway revealed striking increases in expression at spinal, thalamic and cortical levels. COX-2 expression in the CNS is established as a sensitive and responsive biomarker of centralized inflammatory pain39,40,41 and is an important finding consistent with the inflammatory pain and widespread central sensitization as experienced by women with endometriosis. The phenomenon of central sensitization42 has been postulated as a key contributor to the co-morbid pain syndromes experienced by women with the condition4. Central sensitization is described as a maladaptation of the CNS resulting from continued or repetitive input from nociceptors, in endometriosis it is likely that this input is provided by afferent nerve fibers innervating endometriosis lesions. One of the first steps in this process of central sensitization is an increase in expression of genes encoding neurotrophins, neuropeptides and ion channels critical in sensing and detecting noxious stimuli8. In the model of endometriosis generated by Berkley et al, rats exhibit vaginal hyperalgesia26 and it is argued by the authors that this finding suggests central sensitization as an underlying factor because spinal segments associated with the induced endometriosis cysts are distant from spinal segments receiving input from the vagina43,44. Using this same model a decrease in μ-opioid and NMDA receptor immunoreactivity in the periaqueductal gray area of the brain was detected in rats with endometriosis compared to controls. Torres-Reverón et al suggested that a decrease in NMDA receptor expression could be an attempt to homeostatically regulate pain perception, whilst a decrease in μ-opioid receptor expression suggests decreased modulatory activity of opioid receptors that could contribute to hyperalgesia in the condition45. We have also documented molecular alterations in the CNS of mice with endometriosis and we believe this as an important step change in our understanding of endometriosis-associated pain.
Having demonstrated amplified pain behaviors, we tested therapeutic strategies for reversal of mechanical allodynia. The effects of TRPV1 inhibition were modest, whilst antagonism of EP2, particularly following oral administration of the highly selective antagonist PF-04418948 had pronounced effects on peripheral and secondary hyperalgesia. This is consistent with reports that EP2 null mice do not develop spinal hyperalgesia following induced peripheral inflammation46. In a recent study, the effect of a mixture of EP2/EP4 antagonists had a modest impact on pelvic floor hyperalgesia in mice which may either reflect the limited specificity of the antagonists tested or use of human endometriotic cell lines (not intact tissue fragments) in recipient mice lacking a full complement of immune cells31. In our hands, the selective EP4 antagonist L-161982, did not reverse the sensitivity to mechanical stimulation.
In summary, we show that induction of endometriosis in mice with an intact immune system is associated with maladaptation of the CNS, consistent with central sensitization4. We have demonstrated striking reversal of both peripheral and secondary hyperalgesia via EP2 antagonism. In conclusion, we present evidence that a murine model of endometriosis displaying local and central sensitization can be used for pre-clinical testing of therapeutics for endometriosis-associated pain.
Experiments were performed in accordance with the Guidelines of the Committee for Research and Ethical Issues of the International Association for the Study of Pain. Experiments were performed under licensed approval from the UK Home Office (London). C57BL/6 mice (Harlan Laboratories; Derby, UK) were given access to food and water ad libitum, ambient temperature and humidity were 21°C and 50% respectively. Endometriosis was induced in the mice as previously described22. In brief, the endometrium of syngeneic donor mice underwent hormonal manipulation and induction of decidualization using an in-house protocol to model endometrial differentiation, breakdown and repair47: detailed analysis of tissue samples has shown that progesterone withdrawal (removal of P4 pellet) results in rapid induction of hypoxia, tissue breakdown and induction of angiogenic genes47,48. Endometrial tissue (~6hours post-progesterone withdrawal) was recovered by opening the horn and scraping with a scalpel. Approximately 40mg tissue (equivalent to one decidualized horn) was suspended in 0.2ml PBS and injected into the peritoneal cavity of ovariectomised recipient mice that were supplemented with 500ng Estradiol Valerate (EV). This supplementation was maintained by subcutaneous injection of 500ng EV every 3 days (modification of previously published model). After allowing lesions to form over 21 days behavioral assessments were performed.
All behavioral tests were performed starting 21 days after endometriosis induction on 2 consecutive days (day 21 and 22). Mechanical allodynia was measured using calibrated Semmes-Weinstein von Frey filaments (Stoelting, Wood Vale, IL), according to the manufacturer’s instructions. Von Frey filaments were applied to the skin perpendicular to the plantar surface of the hindpaws or to the lower abdomen, as in refs 49, 50, 51. Filaments were applied to the abdomen or hind-paw ten times, force in grams (g) of the filament evoking a withdrawal response in 50% of cases was recorded. Initial testing of the abdomen in naïve mice indicated lower thresholds in more caudal regions, in agreement with a previous report50. For the following spontaneous behavior tests mice were placed in observation boxes for two 5min periods and manually observed by two independent investigators (one blinded to experimental group). Periods of spontaneous abdominally directed licking (an element of normal grooming behavior52) were recorded and an average generated. Excessive abdominally-directed licking has been reported to represent a useful biomarker of abdominal visceral pain46. Each time a mouse exhibited abdominal grooming was recorded as an event. The matrix of brain regions activated during pain includes areas impacting on affective behaviors, corresponding to the anxiety- and depression-like signs associated with chronic pain states53. Paradigms of altered affective sate in rodent pain models include reduced exploratory behavior54. Exploratory activity here was recorded in a modification of the open-field setting55,56, with mice retained within their home box with a cardboard tunnel in the centre of the enclosure; open-field tunnel entries were manually recorded by two independent investigators (one blinded to experimental group).
Four groups of mice were analysed; (i) naïve controls (no surgical procedures; n=9), (ii) OVX+E2 controls (n=9), for which, mice were ovariectomised and given E2 valerate (Sigma, UK), s.c. 500ng in sesame oil every 2 days to mirror the surgical and hormonal status of the Endo mice; (iii) OVX+E2+PBS controls (n=6); as in group ii plus i.p injection of PBS, to mirror injection of tissue as in Endo mice (iv) endometriosis mice (Endo mice; as group iii, with ‘menstrual’ donor material in PBS injected i.p (n=18)). On day 23 mice were culled and the following samples recovered: peritoneal fluid (PF, recovered as in ref. 57 by injecting 3ml ice cold PBS into the peritoneal cavity followed by gentle massage and recovery (approximately 2ml was recovered from the injected 3ml). PF was then centrifuged and frozen), peritoneal biopsy, endometriotic lesions, L5-L6 DRGs, lumbar spinal cord, thalamus and anterior cingulate cortex. Samples were collected into RNAlater (Applied Biosystems, Warrington, UK) and frozen, neutral-buffered formalin prior to paraffin embedding for immunohistochemical analysis (uterus, peritoneum, and endometriosis lesions) or frozen on dry ice prior to sectioning and immunofluorescence staining (DRG) or protein extraction for Western blot analysis (spinal cord and brain). Endometriotic lesions were recognized as red, brown or white tissue deposits on the visceral or parietal peritoneum and were carefully dissected away from any surrounding fat or peritoneum. The presence of glands plus stroma in suspected lesions were confirmed by haematoxylin/eosin staining. Biopsies that did not contain both glands and stroma were not included in further analysis.
Agents for i.p. injections were dissolved in 10% dimethylsulphoxide, 50% PEG-400, 40% de-ionised H20 and injected in a volume of 100μl/25g using 30mg/kg JNJ 17203212 (TRPV1 antagonist), 10mg/kg TG6-10-1 (EP2 antagonist; Calbiochem)58, 10mg/kg PF-0441894825 (EP2 antagonist; Abcam),10mg/kg L-161982 (EP4 antagonist; Abcam). Von Frey testing was performed at 30, 45 and 60minutes post injection: for PF-0418948, testing was also carried out at 75minutes). PF-04418948 (10mg/kg in 0.5% w/v methylcellulose +0.1% V/V Tween-20 in purified water) was also administered as an oral gavage and the von Frey test performed every 15mins starting at 30mins.
RNA was extracted from control uterine biopsies recovered from naïve mice, peritoneal biopsies from naïve and endometriosis mice, endometriosis lesions and dorsal root ganglia using an RNeasy kit (QIAGEN) according to the manufacturers instructions. RNA was quantified using a NanoDrop ND 1000. Quantitative PCR was performed as detailed in refs 59,60; briefly, cDNA was synthesized using SuperScript VILO enzyme (Invitrogen) with 100ng starting template. PCRs were performed using Roche Universal Probe Library (Roche Applied Science) using primer sequences detailed in Supplementary Table 2. 18S was used as a reference gene. Thermal cycling was performed on a 7900 Fast real-time PCR machine. Data was analysed with RQ manager software (Applied Biosystems) using the Ct method; samples were normalised to a uterine control sample.
Dorsal root ganglia (DRGs) were embedded in OCT (CellPath) and frozen on dry ice. Sixteen μm cryostat sections were blocked for 1hour at room temperature then incubated with primary antibodies (Supplementary Table 3). Specificity of the TRPV1 antibody has been previously established62. Secondary antibodies, from Molecular Probes or Sigma-Aldrich, goat anti-chicken Alexafluor 488 (1:1000), goat anti-guinea pig AlexaFluor 568 (1:1000) or goat anti-guinea-pig CF405A (1:1000) were applied for 1hour. Sections were mounted in ProLong® Gold Antifade (Life Technologies). Confocal images were acquired at x20 magnification using a Nikon A1R microscope and ImageJ software was used to quantify co-staining. Standard controls omitting primary antibodies were immunonegative.
Approximately 2ml of the PF was collected into tubes containing indomethacin (10μM) to prevent ex vivo PGE2 metabolism. PGE2 levels were analyzed using DetectX® prostaglandin E2 enzyme immunoassay kit (Arbor Assays, MI, USA).
Tissue samples were collected into sealable tubes and frozen on dry ice, and then subsequently homogenized in Laemmli buffer, heated to 80°C for 5min and centrifuged. Aliquots of lysate supernatant were analysed using the NuPage XCell SureLockTM Minicell gel electrophoresis system (Invitrogen) with approximately 12μg protein loaded per lane. Membranes were incubated overnight at 4°C in 2% non-fat dried milk in 0.1M PBS with 0.1% Tween-20, containing anti-COX-2 antibody (Supplementary Table 3)63. Membranes were washed and incubated for 50min at room temperature with peroxidase-conjugated donkey anti-rabbit antibody (Chemicon, 1:20,000) and detected by peroxidase-linked enhanced chemiluminescence. Membranes were re-probed with mouse monoclonal anti-GAPDH (Supplementary Table 2). Films were scanned and band intensities were quantified by densitometry using ImageJ.
Statistical analysis used a one-way ANOVA with a Newman Keuls or Tukey’s test, or a Kruskal Wallis with a Dunn’s multiple comparison test. A p value of less than 0.05 was considered significant. *p<0.05, **p<0.01, ***p<0.001.
How to cite this article: Greaves, E. et al. EP2 receptor antagonism reduces peripheral and central hyperalgesia in a preclinical mouse model of endometriosis. Sci. Rep. 7, 44169; doi: 10.1038/srep44169 (2017).
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Ronnie Grant for preparation of figures and Dr Alexis Laux-Biehlmann for critical comments on the figures. These studies were supported by a MRC Programme Grant G1100356/1 awarded to PTKS and an MRC Career Development Award MRC MR/M009238/1 awarded to EG.
The authors declare no competing financial interests.
Author Contributions E.G. and S.F.W. conceived manuscript, performed experimental work, analysed results and wrote manuscript. H.J., M.M. and L.H. performed experimental work. R.M. performed experimental work and contributed critical feedback on manuscript. P.T.K.S. and A.H. conceived and wrote manuscript.