In this study, we developed and optimized a translationally relevant preclinical murine model of endometriosis by analyzing, adapting, and modifying previously published methods [21, 22, 27]. Currently, there is no standardized preclinical animal model for endometriosis. A murine model is ideal due to size, fecundity, and amenability to genetic manipulation. We used immunocompetent, intact mice with implantation of menstrual phase endometrium that spontaneously attached. Additionally, we felt implantation into menstruating mice improved model fidelity.
We used soy-free diet and paper bedding to limit exposure to exogenous estrogens (i.e., the mycotoxin zearalenone in corncob bedding, and/or phytoestrogens in soy based rodent diets)[23–25] because supplemental estrogen has been shown to promote endometrial lesion growth in rats and mice [19, 28]. Future studies examining the effects of exogenous estrogens or endocrine disrupters should consider feed, bedding, and caging choice.
We selected mCherry donor mice to allow longitudinal, non-invasive monitoring of developing endometrial lesions in recipient B6 females using fluorescent imaging (IVIS Lumina XR®), but resolution was inadequate with a low signal to background ratio. This differed from the success reported using non-invasive fluorescent monitoring with endometrial tissue labeled with mCherry adenoviral vectors (AAV) [29]. However, in that study, the strongest, most reliable signal was generated from endometrial tissue implanted subcutaneously on the ventral abdomen compared to intraperitoneal implantation, the more translationally relevant location. Additionally, the endpoint was only 20 days, and the use of the AAV vector would increase signal compared to the transgenic mouse alone. Immunohistochemical staining of harvested endometrial lesions demonstrated angiogenesis (CD31) and proliferation (Ki67). The average number of days of disease was 34.85, with confirmed endometriotic lesions also harvested at 60 and 90 days post implantation indicating prolonged survival with this model. With optimization of imaging and the use of transgenic strains expressing fluorophores, non-invasive longitudinal assessment is possible.
All menstruating recipient mice returned to normal estrus cyclicity. This allowed the implanted menstrual phase endometrium to undergo cyclical hormonal exposure to estrogen and progesterone, a more translational model than supplementation with supraphysiologic doses of estrogen and/or progesterone in ovariectomized mice. All mice were sacrificed in proestrus, to decrease variability of lesion size and appearance. Infertility is a common complaint of women suffering from endometriosis and 30–50% of women with endometriosis seek care due to infertility alone [30]. With the use of intact mice, future studies can examine fertility in recipient mice.
Pathologists have classically defined endometriosis in women as the presence of endometroid glands and stroma in biopsied lesions. Paradigm 1 had 20% (2/10) of mice and paradigm 5 had 50% (5/10) of mice with confirmed lesions containing glands. Paradigms 2, 3 and 4 had mice with lesions that only contained stroma and hemosiderin. Although there was no significant difference detected between paradigm 1 and 5 for lesions with glands or uterine lesions, increased group size may allow for detection of differences. Stromal endometriosis in women is characterized as small, microscopic nodules or plaques of endometrioid-type stroma often containing arteriole-like vascular channels, hemosiderin pigment, inflammatory cells, microcalcifications and reactive mesothelial proliferation. It is a common form of endometriosis and can occur with or without the presence of typical endometriosis [31]. Many of the confirmed lesions in this study demonstrated one or more of these characteristics. Additionally, we evaluated lesions at a much earlier timepoint of disease in the mouse (approximately 3 human equivalent years), compared to histological evaluation of suspected lesions in women, as women are typically diagnosed an average of 8 years after symptoms appear. There is tremendous heterogeneity in endometriosis lesions, and evidence of ordered progression is lacking. It is possible that stromal endometriosis is an earlier phenotype of the disease or is a result of a type of metastatic process. Use of an appropriate mouse model, displaying the potential for prolonged lesion survival (90 days) would allow for further evaluation of lesion progression, the role mesothelial inflammation and proliferation plays, and better characterization of endometriosis phenotypes.
In addition to mCherry donor + B6 recipient mice, we have successfully induced endometriosis using decidualized/menstruating B6(Cg)-Tyrc−2J/J or B6.129S6-Gper1tm1Cwan/J females as both donors and recipients (unpublished). It is critical that this model is evaluated in other widely used strains of mice, as differences in endometrial lesion phenotype have been demonstrated between C57BL/6 and BALB/c mice [32] and inbred mouse lines can differ in their immunologic and inflammatory pathways. Immune cell dysfunction in endometriosis is associated with impaired phagocytosis/clearance of menstrual debris, increased survival/proliferation of ectopic tissue, neuro-angiogenesis, and pain. Macrophages (Mφ) are the most abundant immune cells present in endometriotic lesions, and, in response to estrogen, adopt a ‘wound-healing’ (anti-inflammatory) M2 phenotype compatible with survival and growth of ectopic tissue [33]. Mast cells (MC) are also present, and the MC stabilizer ketotifen effectively reduces hyperalgesia in a rat model of endometriosis, indicating MC may play a role in endometriosis pain [34]. Ovarian cycle, estradiol, neuroimmune/neuroendocrine factors, and a nervous system responsive to hormones all play a role in the pain a woman with endometriosis experiences [35]. The use of an ovariectomized and/or immunodeficient mouse would preclude the study of these factors in a mouse model. Immunohistochemical staining of confirmed endometriotic lesions revealed the presence of both macrophages (Cd68) and mast cells (toluidine blue) within or in close proximity to ectopic endometrial tissue, validating this model for use in future studies of these immune cells and for evaluation of pain and pain modulation.
We found confirmed lesions in translationally relevant locations such as the bladder, uterine ligaments, and serosal surface of the uterus (5/10 recipients in paradigm 5). Lesions in women are generally found in gravity dependent areas such as the posterior cul de sac [2] and consistent with the quadruped nature of the mouse, most lesions were located on the ventral aspect of serosal surfaces. Although the ovaries are a common site of endometriosis in women, no lesions were found on the ovaries in recipient mice. This is not surprising as mice have an ovarian bursa and closed tubo-ovarial junction and there is no bursa surrounding the ovaries in women [36]. The closed tubo-ovarial junction also precludes the study of spontaneous endometriosis development due to retrograde menstruation in a decidualized mouse, thus requiring the use of both a donor and recipient mouse and a surgical or injectable method of implantation. Many lesions were attached to adipose tissue. The significance of this needs further exploration, as there is a strong relationship between estrogen and adipose tissue in women [37] and endometriosis is associated with lean body mass index and low waist-to-hip ratio [1]. Additionally, higher levels of gene expression characteristic of M2 phenotype macrophages are associated with adipose tissue in lean mice [38], and endometriosis alters adipose stem cell population and metabolic gene expression [39].
Using laparoscopy, we implanted discrete biopsies of menstrual endometrium into immunocompetent, intact menstruating mice which resulted in a 100% incidence of endometriosis and mean confirmed lesion counts of more than 2 per mouse. Implantation of discrete biopsies via laparoscopy was successful in a previous study with the same incidence of endometriosis (100%) and higher reported take rate (60%); however, uterine biopsies included non-menstrual endometrium and myometrium versus menstrual phase endometrium in our study [22]. Additionally, in paradigm 5 we utilized menstruating recipient mice that were time matched to menstruating donor mice ensuring that both mice transitioned from the proliferative to the secretory phase simultaneously and were in the menstrual phase of the cycle at the time of transplant surgery. Consequently, both mice were subject to the same hormonal milieu and the recipient mice were in the same phase physiologically as women experiencing retrograde menstruation. (Fig. 8). In paradigm 5, we optimized date of sacrifice of donor and implantation into recipient mice to day 9.5 post induction of pseudopregnancy and 4.5 days post decidualization. In menstruating mice, day 9.5, correlates with highest vaginal bleeding scores, largest decrease in systemic progesterone levels, and maximum expression of mRNA profiles. Notably, prostaglandin-endoperoxide synthase 2 (Ptgs2) is 10-fold higher on day 9.5 and vascular endothelial growth factor (Vegfa) and platelet endothelial cell adhesion molecule-1 (PECAM-1) expression are also upregulated. [21] Ptgs2 and Vegfa expression are elevated in ectopic lesions in women suffering from endometriosis and PECAM-1 is highly expressed in pelvic endometriotic lesions [40–42]. The use of decidualized menstrual endometrium in animal models is critical to attempt to replicate expression of potentially important factors that could be therapeutic targets or participate in the etiology and pathogenesis of endometriosis.
Injection of decidualized endometrial slurry was largely unsuccessful when compared to placement of discrete biopsies and compared to reported results injecting decidualized endometrium [27]. However, to have similar conditions between groups, we made a 5 mm incision in mice injected with slurry, comparable to the incision made for placement of laparoscope. This differs from other reported models that perform an intraperitoneal injection without laparotomy. The physiologic effects of anesthesia and laparotomy cannot be ignored, and although the laparotomy incision was small and should decrease stress responses compared to larger incisions [43] this could have negatively affected the development of endometriotic lesions with slurry injection model. All three paradigms utilizing discrete endometrial biopsies had higher group incidences of endometriosis and higher mean confirmed lesions counts compared to the slurry injection paradigms. It has been previously reported in a chicken chorioallantoic membrane (CAM) model of implantation that larger biopsies (> 1mm3) resulted in lesion formation compared to no lesion development when endometrial cells were transplanted [44]. This suggests that intact tissue architecture is crucial to implantation, survival, and proliferation, and could explain the lack of lesions when endometrial slurry was used. However, further exploration of injected menstrual phase endometrial slurry into menstruating mice is warranted, as this could be a non-invasive model allowing for iterative transplantation.
Paradigms 1 and 5, (discrete biopsies with laparoscopy) resulted in 100% incidence of endometriosis, while paradigm 2 (discrete biopsies with laparotomy) had an 80% incidence. Additionally, paradigms 1 and 5 had mean lesion counts of 2.7 and 2.6 respectively, with paradigm 2 having 1.1. The discrete endometrial biopsies adhered to abdominal tissues without the use of suture or glue. The spontaneous attachment of menstrual phase endometrium allows for early study of disease progression including angiogenesis, apoptosis, proliferation, inflammation, and chemotactic homing response. There are reported differences in the morphology of mouse peritoneum and post-operative adhesion formation post laparoscopy or laparotomy [45, 46]. Bulging of mesothelial cells with the presence of intercellular clefts occurs 1–2 hours after CO2 insufflation [47]. This phenomenon has been hypothesized to play a role in peritoneal metastasis and infiltration of tumor cells into the sub mesothelial connective tissue matrix [48]. The biological response of the peritoneum to laparoscopy aids in endometrial biopsy attachment and proliferation, as a compromised mesothelial barrier may play a role in the pathogenesis of the disease [49]. Surgical treatment has also been reported as a potential risk factor for recurrence of ovarian endometriomas [50]. Further investigation into the role diagnostic and therapeutic laparoscopy potentially plays in recurrence and progression of endometriosis is warranted and would be possible using this model.
Continued validation and use of this close approximation of “best fit” animal model of endometriosis will allow for translational exploration of the etiology and pathogenesis of endometriosis and future studies in fertility by using a menstruating, sexually intact and immunocompetent mouse.