Bioluminescent imaging in induced mouse models of endometriosis reveals differences in four model variations

Abstract

Our understanding of the aetiology and pathophysiology of endometriosis remains limited. Disease modelling in the field is problematic as many versions of induced mouse models of endometriosis exist. We integrated bioluminescent imaging of 'lesions' generated using luciferase-expressing donor mice. We compared longitudinal bioluminescence and histology of lesions, sensory behaviour of mice with induced endometriosis and the impact of the gonadotropin-releasing hormone antagonist Cetrorelix on lesion regression and sensory behaviour. Four models of endometriosis were tested. We found that the nature of the donor uterine material was a key determinant of how chronic the lesions were, as well as their cellular composition. The severity of pain-like behaviour also varied across models. Although Cetrorelix significantly reduced lesion bioluminescence in all models, it had varying impacts on pain-like behaviour. Collectively, our results demonstrate key differences in the progression of the 'disease' across different mouse models of endometriosis. We propose that validation and testing in multiple models, each of which may be representative of the different subtypes/heterogeneity observed in women, should become a standard approach to discovery science in the field of endometriosis.

Keywords: Endometriosis; GnRH antagonist; Lesion; Pain.

Fig. 1.

Establishment of non-invasive bioluminescent imaging in the ‘menses’ model of induced endometriosis. (A) Left to right: Cag-luc-eGFP mouse injected s.c. (1.5 mg in 100 µl volume) with luciferin; Cag-luc-eGFP mouse with no injection; and wild-type FVB/N mouse injected s.c. with luciferin. Whole-body imaging (top panels) and dissected uteri from corresponding mice (bottom panels). (B) Schematic representation of the DO model of endometriosis. ‘Menses’ endometrium from Cag-luc-eGFP mice was injected i.p. into ovariectomized (ovx; with add-back oestradiol) wild-type FVB/N recipient mice. Whole-body bioluminescent imaging was performed 21 days post tissue injection. (C) Whole-body images showing that bioluminescent focal lesions were localized to the abdominal region of the mice. (D) Whole-body images demonstrating the difference between i.p. (left panel) and s.c. (right panel) administration of luciferin at day 7 post tissue inoculation (in the same mouse), indicating that s.c. administration can differentiate between attached explants and unattached floating endometrial tissue. (E) At dissection, lesions were observed attached to the abdominal wall, fat and organs. (F) Paraformaldehyde (4%)-fixed lesion tissue was stained for luciferase (red), with nuclei stained with DAPI (blue). (G) Endogenous GFP (green) expression by lesions was visualized by live fluorescence microscopy prior to fixation. Surrounding GFP⁻ tissue can be observed (phase white). (H) Representative H&E image of a lesion with stromal cells with or without glandular epithelia and the presence of hemosiderin. Bar on left hand side of A (also applies to C,D) is lookup table (LUT) representing photon counts. Set to 0.0653 (minimum) and 0.46 (maximum) ×10^-3 counts.

Fig. 2.

Comparison of lesion luminescence and longevity across different models of endometriosis. (A) Schematic representation of the DI mouse model of endometriosis. ‘Menses’-like endometrium from Cag-luc-eGFP was injected i.p. into intact wild-type FVB/N recipient mice. (B) Schematic depicting the minimally invasive mouse models of endometriosis (NI and MI). Naïve endometrium from intact Cag-luc-eGFP was injected i.p. into wild-type FVB/N recipient mice or full thickness uterus (including myometrium) was injected into intact wild-type FVB/N recipient mice. (C) At 21 days post tissue injection the MI model exhibited significantly higher bioluminescent signal compared to all other models (DO, n=9; DI, n=10; NI, n=29; MI, n=20). Sample sizes are representative of mice (not lesions) and were achieved by performing experiments 2-4 times. Boxes, interquartile range; whiskers, minimum to maximum. Statistical analysis performed using a one-way ANOVA with Tukey's multiple comparison test (**P<0.01, ***P<0.001). (D) The percentage of mice with focal bioluminescent lesions was quantified at 7, 21 and 42 days post tissue injection. This showed a progressive decline in the number of mice that had detectable lesions in all four models of induced endometriosis. (E-H) Representative longitudinal images of individual mice from day 10-42 post tissue injection. DO model (E); DI model (F); two individual mice from the minimally invasive models (G, NI; H, MI). In G, two mice are presented to illustrate the variation across individual recipients in these two groups. The mouse on the left had a large lesion at day 10 that was maintained at day 42. The mouse on the right had four focal lesions at day 10 that were spontaneously resolved by day 42. In H, two mice are presented. The mouse on the left had one lesion at day 10 that resolved at day 42, whereas the mouse on the right had two lesions on day 10 that progressively increased in size. Bar on left hand side of E (also applies to F-H) is LUT representing photon counts set to 0.0653 and 0.46.

Fig. 3.

Identification of endometriosis ‘hallmarks’ in lesions derived from different mouse models of endometriosis. (A-D) Representative H&E stains from each model. (E-H) Immunodetection of Vimentin to visualize stromal fibroblasts. Inset in E is a negative control section of whole uterus (primary antibody omitted). (I-L) Immunodetection of cytokeratin to visualize epithelial cells. Inset in I is a negative control section of gut (primary antibody omitted). (M-P) PSR stain to identify areas of collagen deposition as a marker of fibrosis. Red stain indicates presence of collagen fibres. (Q-T) Immunodetection of α-smooth muscle actin, a marker of myofibroblasts. Inset in Q is a negative control section of gut (primary antibody omitted). Scale bars: 200 µm (A-D); 50 µm (E-T).

Fig. 4.

Quantification of histological structures in lesions recovered at day 42 post tissue injection. (A) Percentage of lesions that were immunopositive for cytokeratin (irrespective of glandular structure; positive (+ve stain) and the percentage of lesions that exhibited cytokeratin positive glandular structure (DO, n=7 lesions; DI, n=5 lesions; NI, n=5 lesions; MI, n=23 lesions). (B) Area of collagen deposition (as a measure of fibrosis) in lesions from the different models. Error bars indicate s.d.; individual data points represent individual lesions. (C) Percentage of lesions that were immunopositive for α-smooth muscle actin (α-sma). (D) Location of lesions recovered from the different models. (E) Area of lesions measured using ImageJ. Error bars indicate s.e.m.; individual data points represent individual lesions.

Fig. 5.

Changes in sensory behaviour are evident in different mouse models of endometriosis. (A,B) Mechanical hyperalgesia measured using von Frey filaments applied to the abdomen. Sham-ovx, n=10; DO, n=17; Sham intact, n=7; DI, n=10; NI, n=29. (C,D) Mechanical hyperalgesia measured using von Frey filaments applied to the hindpaw. Values plotted are an average of measurements taken over 3 days from days 40, 41 and 42. Boxes, interquartile range with mean shown; whiskers, minimum and maximum data points. Statistical significance was determined using Mann–Whitney or Kruskal–Wallis tests (*P<0.05, **P<0.01, ***P<0.001).

Fig. 6.

Response of intact endometriosis models to Cetrorelix. (A) Schematic showing the treatment and imaging schedule of DI, NI and MI mice. (B) Percentage change in lesion bioluminescent signal calculated from vehicle- (H₂O) and Cetrorelix (Cet)-treated mice with endometriosis (DI group). Percentage change=(day 21 signal − day 7 signal)/day 7 signal×100. Vehicle, n=22; Cetrorelix n=17. (C) Representative images from day 7 (pre-drug) and day 21 (end of experiment, DI group). (D) Mechanical hyperalgesia measured using von Frey filaments applied to the abdomen. (E) Mechanical hyperalgesia measured using von Frey filaments applied to the hindpaw. (F) Impact of Cetrorelix treatment on lesion bioluminescent signal intensity (P<0.05) in the NI group. Vehicle, n=13; Cetrorelix, n=10. (G) Representative images from day 7 (pre-drug) and day 21 (end of experiment, NI group). (H,I) Impact of Cetrorelix on mechanical hyperalgesia at the abdomen (H) and hindpaw (I) in the NI model. (J) Impact of Cetrorelix on lesion signal intensity in the MI group. Vehicle, n=17; Cetrorelix, n=17. (K) Representative images from day 7 and day 21 (MI group). (L,M) Impact of Cetrorelix on mechanical hyperalgesia when von Frey filaments were applied to abdomen (L) and hindpaw (M). Boxes, interquartile range with mean shown; whiskers, minimum and maximum data points. Statistical analysis was performed using a Mann–Whitney test or Kruskal–Wallis test (*P<0.05, **P<0.01, ***P<0.001). Bar on left hand side of C, G and K is LUT representing photon counts set to 0.0653 and 0.46.

Fig. 7.

Blueprint of models compared in the study. We have used a traffic light scoring system to denote whether a model performs well (green), partially (orange) or not at all (red) in modelling a particular feature or outcome. We have scored the intact models as green for measuring fertility outcomes because they have the ability to be used in this context; however, we have not yet demonstrated any differences in pregnancy outcome in the models.