Introduction

Endometriosis is a disease characterized by the presence of endometrial glands and stroma at ectopic sites, such as the abdominal cavity or the ovaries. This disease occurs in approximately 10% of women of reproductive age, who present symptoms such as pain (including dyspareunia, dysmenorrhea, and chronic pelvic pain) or infertility¹. Endometriosis is treated by surgical procedure, involving removal of active endometriotic lesions, and/or by medical therapy, which, in most of the cases, aims at inducing a hypoestrogenic state in patients. Indeed, it is recognized that estrogen withdrawal causes the involution of endometriotic lesions in patients² and in animals^3,4,5, and consequently endometriosis is generally considered as an estrogen-dependent disease^6,7. Unfortunately, drugs that induce a hypoestrogenic state cannot be used for prolonged duration because of severe side effects (e.g., reduction in bone mineral density). Moreover, current surgical and/or medical therapeutic approaches are not satisfactory as recurrence of the disease is observed in 40% of the patients within 4 years following conservative surgical treatment and/or medical therapy⁸. Hence, there is a definite need for developing new drugs, through the use of animal models, to provide long-term and more efficacious therapeutic alternatives.

Nonhuman primates are the only animal species that spontaneously develop endometriosis⁹. However, these models are costly and do not allow initial screening of compounds. Therefore, establishment of endometriosis in small laboratory animals has been developed and relies on the transplantation of pieces of endometrial tissue to ectopic locations (such as the peritoneal cavity or the subcutis)¹⁰. In rabbit, rat, and some mouse models, endometrium is dissected from one uterine horn and surgically transplanted into the peritoneal cavity of the same animal (autotransplantation)¹⁰. Alternatively, fragments of human endometrium are transferred intraperitoneally or subcutaneously to immunocompromised animals, such as nude¹¹ or SCID¹² mice (xenotransplantation). A fraction of these fragments implant and form endometriotic-like lesions, which resemble lesions found in patients in terms of macroscopic and histological appearance^11,13,14,15, steroid responsiveness^11,16,17, or vascularization¹⁸.

The nude mouse model for endometriosis represents a promising tool for therapeutic testing, mainly because it is based on human endometrium and relies on the use of small rodents. Up to now, drug testing with this model has been performed by the team of Osteen and colleagues and, in most of the cases, their readout has been to compare numbers of lesions, present either intraperitoneally or subcutaneously, in drug-treated versus untreated animals^{15,16,17,18,19}. The use of a more sensitive readout, such as lesion size, would be of great interest to this model. This would allow a more precise follow-up of disease progression and would also be more related to the clinical situation, in which lesion size is taken into account in the staging²⁰ and evaluation of regression of the disease following a treatment²¹.

To perform quantification of lesion size, we adapted the existing nude mouse model for endometriosis to in vivo imaging²². It was reported previously that a fluorescent dye could be added directly to endometrial tissue to allow nonsurgical visualization of human tissue²³. According to our modification of the nude model, human endometriotic-like lesions are rendered fluorescent after adenoviral transfer of the green fluorescent protein (GFP) cDNA into endometrial tissue prior to implantation. This allows repeated visualization of endometriotic-like lesions through the skin of living animals. Hence, this approach would permit one to monitor the effect of a candidate drug on the development of endometriotic lesions in the same animal throughout the treatment period. Herein, we have used this improved animal model for endometriosis to quantify lesion development and regression in ovariectomized mice supplemented or not with estrogens. We also induced regression of established lesions by ganciclovir treatment of animals bearing endometriotic implants expressing the thymidine kinase cDNA. Our findings demonstrate the applicability of the in vivo model for assessment of drug efficacy and for the validation of gene targets and pathways involved in the disease.

Results

Quantification of Endometriotic Lesion Size by in Vivo Imaging

In a previous work, we had demonstrated the feasibility of visualizing fluorescent endometriotic-like lesions in nude mice in a noninvasive manner²². In this study, we wanted to verify whether quantification of lesion size was possible using this fluorescence-based imaging technique. To make a proof of concept for the detection and measurement of human lesion regression in animals by the means of in vivo imaging, we used two approaches. First, we used a well-recognized method for inducing shrinkage of implanted human tissue, i.e., estrogen (E2) deprivation. Second, we wanted to verify whether other means of inducing lesion regression, unrelated to steroid withdrawal, could also be detected and quantified by in vivo imaging. We therefore specifically induced cell death in lesions by expression of the thymidine kinase gene followed by ganciclovir treatment. We chose this gene/prodrug combination because it is associated with an extensive bystander effect, hence allowing high levels of cell death.

Noninvasive Follow-up of Lesion Development in Animals Supplemented or Not with Estrogens

Because it is well known that estrogens are required for the growth of endometriotic lesions in humans and animals, we implanted fluorescent human endometrial tissue in mice supplemented (N = 8) or not (N = 8) with E2. We then compared the development of endometriotic-like lesions in treated and untreated animals by noninvasive imaging. Fig. 1 shows typical images obtained for lesions developing in ovariectomized nude mice with or without an E2 pellet. As we previously reported²², GFP fluorescence emitted by the human endometrial transplants decays with time. This is due to the transient GFP expression achieved by adenoviral infection of endometrial fragments. Hence the number of GFP cDNA copies and the level of transcription decrease with cell division and time. Because maintenance and growth of endometriotic lesions are estrogen-dependent, we hypothesized that mice subjected to a hypoestrogenic state would display a faster decrease in GFP fluorescence, corresponding to accelerated lesion shrinkage. This was indeed the case, and Fig. 1 shows a representative experiment in which a lesion is detecTable up to day 29 in an E2-treated mouse while it is hardly discernible at day 16 in an untreated mouse, despite the fact that lesions were of similar sizes at the beginning of the experiment.

Figure 1.

Noninvasive in vivo imaging of fluorescent endometrial implants allows dynamic monitoring of lesion development. Endometrial fragments were incubated with AdGFP viral particles for 20 h. After washings, five fragments were injected subcutaneously into ovariectomized nude mice supplemented (N = 8) or not (N = 8) with E2-releasing pellets. Each mouse was repeatedly imaged in triplicate, and images acquired at days 4, 6, 8, 12, 14, 16, 19, 23, 27, and 29 posttransplantation are shown for one mouse in each group.

Full figure and legend (207K)

For each mouse and each time point, the size of lesions was quantified and results of a typical experiment are presented in Fig. 2. The size of lesions, as measured by GFP fluorescence and expressed as a number of pixels, varied from 8720 to 67,270 pixels at day 4 postimplantation, showing that the five endometrial fragments initially injected can generate lesions of variable sizes in different mice. However, the mean lesion size was similar between the groups with and without E2 at days 4 and 6. Starting from day 12, the mean lesion size becomes significantly smaller in the group without E2, and a clear segregation between the two groups in terms of lesion size occurs at later stages (Fig. 2).

Figure 2.

Quantification of endometriotic-like lesion size in ovariectomized mice supplemented or not with E2. Ovariectomized nude mice having received or not an E2 pellet (N = 8 for each group) were periodically imaged in triplicate, and endometriotic-like lesion size (number of pixels) was quantified by image analysis. Each symbol represents the mean of three measures for each animal at each time point. Bars indicate the mean surface of endometriotic implants for each group. Comparison of lesion size in estradiol-supplemented versus nonsupplemented animals was performed by a nonparametric Mann–Whitney test. A significant difference in size of lesions is observed between the two groups starting from day 12 (P 0.05 at days 12, 27, and 29; P 0.005 at days 13, 14, 16, 19, 21, and 23). These results are representative of six independent experiments.

Full figure and legend (58K)

Another way of analyzing these data is to consider as an endpoint the time at which each lesion presents a 75% reduction from its initial size. This allows the generation of Kaplan–Meier survival curves for each group of mice (Fig. 3A). The mean time to observe a 75% reduction in size was significantly higher (P = 0.0001) for the E2 group (26 1 days) compared to the group without E2 (14 1 days). A similar analysis can be performed for a 50% reduction in lesion size (Fig. 3B), and a significant difference (P = 0.0002) is also found between E2-treated and untreated animals (means of 21 2 and 10 1 days, respectively).

Figure 3.

Kaplan–Meier analysis of endometriotic-like lesion regression in mice lacking estradiol supplementation. Ovariectomized nude mice having received or not an E2 pellet (N = 8 for each group) were periodically imaged in triplicate, and endometriotic-like lesion size (number of pixels) was quantified with image analysis software. The time at which each mouse had a lesion with (A) a 75% or (B) a 50% reduction in its initial size was used to generate survival curves (Kaplan–Meier analysis). A significant difference was observed between E2-supplemented and nonsupplemented groups (log rank test, P = 0.0001 and P = 0.0002 for A and B, respectively).

Full figure and legend (74K)

To verify whether findings obtained by in vivo imaging correlated with the actual size of lesions, mice were sacrificed at the end of the experiment (day 25) and lesions were measured with a caliper. Results presented in Table 1 show that a significant difference is observed in the size of lesions in E2-treated versus untreated animals, using either in vivo imaging or a caliper. Altogether, these results demonstrate that in vivo imaging can reveal a significant difference in endometriotic lesion maintenance in animals treated or not with E2, as expected.

Table 1 - Comparison of lesion size determined either by in vivo imaging or with a caliper at the end of the experiment (day 25).

Full table

Induction of Lesion Regression by Gene Suicide Therapy and Quantification by in Vivo Imaging

The experiments described above demonstrated that lesion implantation and maintenance could be followed in a quantitative manner in two groups of mice harboring different steroid environments. We next wanted to verify whether regression of established lesions could be induced following a treatment and measured in living animals. This would establish the usefulness of our model to measure the efficacy of a candidate drug or yet to evaluate the impact of expressing a given gene within the implanted endometrial tissue. To induce massive cell death in human endometriotic lesions, we used the thymidine kinase (TK) system: the prodrug ganciclovir (GCV) becomes toxic in cells expressing the TK enzyme and in neighboring cells, a phenomenon known as the bystander effect²⁴.

First, we wanted to assess whether human endometrial cells were sensitive to GCV treatment following infection by an adenovirus containing the TK cDNA. We infected a human endometrial cell line, HEC-1A, with either AdGFP or AdTK and treated infected cells with increasing doses of GCV. Infection efficiencies were 76% for AdGFP and 24% for AdTK, as measured by GFP expression. Fig. 4 shows that cells infected with AdGFP were resistant, whereas TK-expressing cells became sensitive to GCV treatment. Although only 24% of the cells were transduced by AdTK, 70% of cells underwent cell death, most certainly reflecting the bystander effect. This result led us to expect that a large amount of cell death would occur within endometrial implants expressing TK in vivo.

Figure 4.

Ganciclovir treatment induces cell death in an endometrial cell line expressing the thymidine kinase gene. HEC-1A cells were infected with either AdGFP or AdTK viral particles at an m.o.i. of 1. The proportions of cells expressing either GFP or TK 1 day prior to GCV treatment were 76 and 24%, respectively. Treatment with increasing doses of ganciclovir (GCV) was initiated 5 days postinfection and was pursued for 7 days. At the end of GCV treatment, cells were harvested and cell death was assessed by propidium iodide (PI) staining and flow cytometry. The percentage of apoptotic cells for each GCV dose is shown.

Full figure and legend (38K)

Because the GFP encoded by the AdTK virus is less bright and expressed at lower levels than that encoded by the AdGFP virus, it was not suiTable for in vivo imaging. Hence, human endometrial fragments were co-infected with both AdGFP and AdTK viruses. Control fragments were infected with AdGFP alone. Infection efficiencies were 40% for AdGFP alone and 51% for AdGFP + AdTK. Thirty-two ovariectomized mice supplemented with E2 were separated into two groups, one receiving human endometrial fragments infected with AdGFP only and the other receiving fragments infected with both AdGFP and AdTK. Half of the mice in each group were treated with GCV and the other half received PBS, starting from day 4 after implantation. Fig. 5A shows the size of lesions in mice with lesions expressing both GFP and TK. Mice treated with GCV had significantly smaller lesions starting from day 8, compared to mice receiving PBS. Notably, the group of mice with lesions expressing GFP alone, whether treated or not with GCV, behaved like the control (PBS) mice in Fig. 5A (data not shown). A faster regression of endometriotic-like lesions expressing TK and treated with GCV is also observed in a survival-type analysis (Fig. 5B). Mean time for a regression of 75% in size was 18 2 days for PBS-treated mice and 12 1 days for GCV-treated mice (P = 0.004). For a regression of 50% in size (Fig. 5C), mean time was 15 2 days for PBS-treated mice and 10 0 days for GCV-treated mice (P = 0.006). These results show that cell death due to GCV is able to induce the regression of TK-expressing endometriotic lesions and that this regression can be measured by in vivo imaging.

Figure 5.

Ganciclovir treatment induces a significant regression of endometrial implants expressing the TK gene in nude mice. Endometrial fragments were incubated with AdGFP (m.o.i. 500) and AdTK (m.o.i. 25) viral particles for 20 h. After washings, five fragments were injected subcutaneously into 16 ovariectomized nude mice supplemented with E2-releasing pellets. Eight mice received an ip injection of GCV (50 mg/kg/day) on each weekday for a 2-week period, starting 4 days after tissue transplantation. Mice in the control group (N = 8) received PBS injections. Each mouse was imaged in triplicate at days 1, 4, 6, 8, 11, 13, 15, 18, 20, and 22 posttransplantation, and lesion size was quantified. (A) Each symbol represents the mean of three measures for each animal at each time point. Bars indicate the mean surface of endometriotic implants for each group. Comparison of lesion size (in pixels) in treated versus untreated animals was performed by a nonparametric Mann–Whitney test. A significant difference between GCV- and PBS-treated groups was observed (P 0.05 at days 8, 15, 18, 20, and 22; P 0.005 at days 11 and 13). (B and C) The time at which each mouse had a lesion with (B) a 75% or (C) a 50% reduction from its initial size was used to generate survival curves (Kaplan–Meier analysis). A significant difference was observed between GCV- and PBS-treated groups (log rank test, P = 0.004 and P = 0.006 for B and C, respectively). These results are representative of two independent experiments.

Full figure and legend (98K)

Discussion

Animal models of endometriosis based on the xenotransplantation of human endometrium into nude mice are gaining interest^11,13,14,25 and have permitted the elucidation of mechanisms involved in the establishment of endometriosis as well as the testing of potential therapeutic compounds^16,17,18,19. Despite the great advantage of being based on human endometrial tissue, the nude mouse model still presents some limitations. For instance, the number of endometriotic-like lesions that will effectively develop, following the injection of human endometrial fragments either intraperitoneally or subcutaneously, is variable from one animal to another. Hence, a difference in the number or size of lesions in two groups of animals could in fact be due to different "take rates" at the beginning of the experiment rather than to the effect of a drug. It is therefore of great interest to be able to monitor the initial number, and size, of lesions established in each mouse, which is possible with the improvement of the noninvasive nude mouse model proposed by our group²². In addition, the fact that only implanted human tissue expresses GFP helps in the identification of endometriotic lesions during dissection and renders histological confirmation of host tissue origin unnecessary. Finally, GFP-emitting cells have to remain viable to express the transgene; therefore only active lesions are monitored by noninvasive imaging.

To verify whether noninvasive optical imaging was suiTable to quantify reduction in lesion size adequately, we used two different approaches: estrogen deprivation in the animal or induction of cell death in established lesions by expression of thymidine kinase in endometrial tissue followed by GCV treatment. Quantification of lesions by in vivo imaging demonstrated that up to 6 days after implantation, there was no difference in the number and mean size of lesions between groups of ovariectomized mice having been supplemented or not with E2. This result suggests either that the presence of E2 is not necessary for the initiation of lesion implantation or that the residual E2 present in endometrial fragments is sufficient for the initial steps of lesion formation. On the other hand, our results clearly demonstrated that E2 became necessary for lesion maintenance starting from day 12 postimplantation, a time at which a significant reduction in endometrial implant size became apparent in the ovariectomized group without E2. This could be observed with endometrial tissue collected during either the proliferative or the secretory phase of the menstrual cycle (data not shown). These results are in agreement with previous studies in monkeys and mice showing that implantation and initial growth of ectopic endometrium does not require estrogens, while this steroid is necessary for the long-term maintenance of lesions^3,26. We were also able to induce and measure lesion regression in established lesions expressing the thymidine kinase cDNA by treating animals with GCV. Together, our results make the proof of concept that local induction of apoptosis leads to lesion regression. Thus, nonhormonal approaches for endometriosis, based on the use of proapoptotic or antiangiogenic genes²⁷, for instance, would represent interesting avenues for future gene therapy management of endometriosis.

From the technical point of view, the subcutaneous location was chosen for lesion implantation because this site has been shown to allow the proper development of endometriotic-like lesions in nude mice^11,15,18, because the tissue is located at the same position from one day to another and because emitted fluorescence is minimally absorbed by tissues, thus allowing more reproducible imaging. Ultimately, monitoring of lesions located intraperitoneally is desirable. This will require the use of red-emitting fluorescent proteins, whose fluorescence is less absorbed by tissues, and more sensitive imaging equipment, which demands further development. In addition, the possibility of expressing the fluorescent reporter in lesions in a sTable manner would certainly represent an improvement of the model. Indeed, although adenoviruses are efficient vectors for introducing genetic material into endometrial fragments, they provide only transient expression. Other vectors such as lentiviruses, which combine high infection efficiency, broad target cell specificity, and sTable expression, represent potentially interesting candidates. Additionally, it has been recently shown that mixed cultures of stromal and epithelial cells adhered to the mouse peritoneum and led to the formation of endometrial implants²⁸. In vitro transduction of monolayers of endometrial primary cells would certainly lead to increased infection efficiencies, therefore generating endometriotic-like lesions in which the majority of cells would express the gene of interest, permitting their long-term follow-up.

The noninvasive nude mouse model for endometriosis presented in this study allows monitoring of lesion localization, viability, remodeling, and size at different time points in the same animal. This model will help in better understanding the disease evolution in the living animal and will also permit faster, more precise, and accurate characterization of a drug's effect on experimental endometriosis.

Materials and Methods

Endometrial tissue collection

Endometrial tissue was obtained from women undergoing laparoscopy or laparotomy for diagnostic purposes or tubal ligation/reanastomosis. All participants had regular menstrual cycles (between 21 and 35 days), had received no hormonal treatment, and had not been pregnant, breast-feeding, or using an intrauterine device for the past 3 months. All patients were diagnosed as endometriosis-free by visual inspection of the peritoneal cavity. Endometrial biopsies were obtained with a suction curette (Pipelle; Millex, Montreal, QC, Canada) and immediately transferred into RPMI 1640 medium supplemented with 2% fetal calf serum (FCS), 100 g/ml penicillin, and 100 IU streptomycin (all from Wisent, Inc., St-Bruno, QC, Canada). Four patients were in the proliferative and four in the secretory phase of the menstrual cycle phase, as determined by histological dating²⁹. All patients gave their informed written consent prior to tissue collection. This study was approved by Procrea's Ethics Review Board and the Internal Review Board of each participating clinical institution.

Adenoviruses

Replication-defective adenoviral particles (adenovirus serotype 5, with the E1 and E3 regions deleted) were used. The first virus carried the GFP cDNA driven by the strong CMV5 promoter (AdenoExpress; Qbiogene, Montreal, QC, Canada; herein named AdGFP). A second virus, with the E4 region deleted, and carrying both EGFP and TK genes under individual CMV promoters (AdCMVEGFP/CMVTKdelE4, herein named AdTK) was used for gene suicide experiments³⁰.

Preparation of endometrial fragments and transduction with adenoviral particles

Endometrial tissue was washed with PBS, transferred to DMEM/F12 culture medium supplemented with 10% FCS (Wisent, Inc.) and 10 nM 17- estradiol (Sigma–Aldrich, Oakville, ON, Canada), and cut into 1- to 2-mm³ pieces. Five tissue fragments were distributed into each well of a 48-well plate with 100 l of culture medium containing AdGFP particles (m.o.i. 500). For gene suicide experiments, endometrial fragments were co-infected with AdGFP (m.o.i. 500) and AdTK (m.o.i. 25) particles. AdGFP was used for in vivo imaging, whereas AdTK was used for inducing cell death. Infection was performed by incubation for 20 h at 37°C, 5% CO₂ under agitation.

Determination of infection efficiencies

GFP fluorescence was used to determine infection efficiencies by flow cytometry on single-cell suspensions obtained from five infected fragments. After infection, fragments were harvested, washed three times with PBS, incubated for 40 min in 0.05% trypsin–0.53 mM EDTA (Wisent, Inc.) supplemented with 0.1% collagenase (Sigma–Aldrich), and mechanically disrupted. Single-cell suspensions were washed twice and resuspended in PBS. Propidium iodide (PI) (Sigma–Aldrich) was added at a final concentration of 10 g/ml to assess cell viability. Analysis of GFP fluorescence and detection of dead cells were performed on a Coulter XL flow cytometer (Beckman Coulter, Ville St-Laurent, QC, Canada) using 525 and 670 nm detection filters, respectively.

Transplantation of GFP-expressing endometrial fragments into nude mice

Six- to 8-week-old ovariectomized female nude mice (nu/nu) were purchased from Harlan Sprague–Dawley Laboratories (Indianapolis, IN, USA) and housed under specific-pathogen-free conditions. When necessary, sterile 60-day release pellets, containing 0.36 mg of 17- estradiol (Innovative Research of America, Sarasota, FL, USA), were implanted subcutaneously into mice 24 h prior to endometrial tissue injection. Blood levels of E2 achieved with these pellets ranged from 150 to 200 pg/ml, as determined by the manufacturer. The combination of ovariectomized mice plus pellets was chosen over cyclic mice with no estradiol supplementation because it gave consistently higher take rates of human endometrial tissue (data not shown). After adenoviral infection, endometrial fragments were harvested, washed three times with PBS–2% FCS, and resuspended in 300 l of PBS–2% FCS. Each animal received a single subcutaneous injection, on the ventral midline, of five GFP-expressing endometrial tissue fragments, using an 18-gauge needle. Adenovirus infection did not affect the ability of endometrial fragments to implant²². All animal studies were approved and conducted in accordance with Institutional Animal Care Committee guidelines.

Noninvasive in vivo imaging of GFP-expressing human endometrial tissue

For noninvasive imaging of GFP-expressing endometrial tissue, mice were anesthetized with 1.5% v/v isoflurane (Abbott Laboratories, Montreal, QC, Canada) and illuminated by a light-emitting source (Illumatool TLS; Montreal Biotech, Kirkland, QC, Canada) equipped with a 470 nm filter. Images were recorded with a Nikon CoolPix 990 digital camera (Nikon Canada, Montreal, QC, Canada), mounted with a 515 nm viewing filter. Animals were imaged repeatedly for at least 3 weeks after tissue injection. For each time point, three images were acquired for each mouse. Image analysis was used to determine the size of endometriotic-like lesions. An imaging software especially designed for this application (SIF software; Centre de Recherche Informatique de Montréal, CRIM) was used. This software identifies for each image the number and intensity of pixels corresponding to the spectral signature of GFP (present only in regions of the image where a lesion appears). When more that one lesion was present, the calculated surface was the sum of each individual lesion. In addition, three images were acquired for each mouse at each time point, and the mean lesion size and associated coefficient of variation (CV) for each triplicate were calculated by the imaging software. This allowed a more robust determination of lesion size and potential sources of variation. The mean CV among all triplicates obtained from eight different experiments (1629 triplicates) was 12, and 75% of all CV values were under 12%. To minimize variations due to changes either in the amount of excitation light reaching the animal or in the absorption of fluorescence by the tissue surrounding lesions, lesions were implanted subcutaneously and the positioning of animals and lesions was carefully reproduced by means of contention and marks from day to day.

In vitro determination of ganciclovir toxicity in TK-expressing endometrial cells

HEC-1A cells (endometrial carcinoma, ATCC HTB112) were infected with either AdGFP or AdTK viral particles at an m.o.i. of 1, and infection efficiency was determined by flow cytometry. Treatment with increasing doses of GCV (Sigma–Aldrich) started 5 days postinfection and was conducted for 7 days. At the end of the experiment, the proportion of dead cells was determined by flow cytometry on single-cell suspensions, using PI as a vital dye.

In vivo drug treatment

GCV treatment (50 mg/kg/day) was started 4 days after tissue transplantation. Mice in the treated group received a single ip injection on each weekday for a 2-week period (for a total of 10 injections). Mice in the control group received PBS injections.

Statistical analysis

Comparison of lesion size in treated versus untreated animals was performed by a nonparametric Mann–Whitney test. Alternatively, the time at which lesions presented a 50 or 75% reduction in their size compared to the size at the beginning of treatment was computed and survival Tables were generated (Kaplan–Meier). Mean survival times were compared with the log rank test. Results were considered statistically different when P 0.05.

References

1. Gosselin, D., et al. (1999). Insights into the physiopathology of endometriosis—novel diagnostic and therapeutic approaches. Curr. Opin. Oncol. Endocr. Metab. Invest. Drugs. 1: 31–43.
2. Wheeler, J. M., Knittle, J. D. and Miller, J. D. (1992). Depot leuprolide versus danazol in treatment of women with symptomatic endometriosis. I. Efficacy results. Am. J. Obstet. Gynecol. 167: 1367–1371. | PubMed | ChemPort |
3. Dizerega, G. S., Barber, D. L. and Hodgen, G. D. (1980). Endometriosis: role of ovarian steroids in initiation, maintenance, and suppression. Fertil. Steril. 33: 649–653. | PubMed | ChemPort |
4. Rajkumar, K., Schott, P. W. and Simpson, C. W. (1990). The rat as an animal model for endometriosis to examine recurrence of ectopic endometrial tissue after regression. Fertil. Steril. 53: 921–925. | PubMed | ChemPort |
5. Cummings, A. M. and Metcalf, J. L. (1995). Effects of estrogen, progesterone, and methoxychlor on surgically induced endometriosis in rats. Fundam. Appl. Toxicol. 27: 287–290. | Article | PubMed | ISI | ChemPort |
6. Barbieri, R. L. (1998). Endometriosis and the estrogen threshold theory. Relation to surgical and medical treatment. J. Reprod. Med. 43: 287–292. | PubMed | ChemPort |
7. Kitawaki, J., et al. (2002). Endometriosis: the pathophysiology as an estrogen-dependent disease. J. Steroid Biochem. Mol. Biol. 83: 149–155. | Article | PubMed | ChemPort |
8. Rice, V. M. (2002). Conventional medical therapies for endometriosis. Ann. N. Y. Acad. Sci. 955: 343–352. | PubMed |
9. D'Hooghe, T. M. (1997). Clinical relevance of the baboon as a model for the study of endometriosis. Fertil. Steril. 68: 613–625. | Article | PubMed | ChemPort |
10. Vernon, M. W. (1990). Experimental endometriosis in laboratory animals as a research model. Prog. Clin. Biol. Res. 323: 49–60. | PubMed | ChemPort |
11. Zamah, N. M., et al. (1984). Transplantation of normal and ectopic human endometrial tissue into athymic nude mice. Am. J. Obstet. Gynecol. 149: 591–597. | PubMed | ChemPort |
12. Aoki, D., Katsuki, Y., Shimizu, A., Kakinuma, C. and Nozawa, S. (1994). Successful heterotransplantation of human endometrium in SCID mice. Obstet. Gynecol. 83: 220–228. | PubMed | ChemPort |
13. Nisolle, M., Casanas-Roux, F., Marbaix, E., Jadoul, P. and Donnez, J. (2000). Transplantation of cultured explants of human endometrium into nude mice. Hum. Reprod. 15: 572–577. | Article | PubMed | ISI | ChemPort |
14. Grummer, R., et al. (2001). Peritoneal endometriosis: validation of an in-vivo model. Hum. Reprod. 16: 1736–1743. | Article | PubMed | ChemPort |
15. Bruner-Tran, K. L., Webster-Clair, D. and Osteen, K. G. (2002). Experimental endometriosis: the nude mouse as a xenographic host. Ann. N. Y. Acad. Sci. 955: 328–339. | PubMed |
16. Bruner, K. L., Matrisian, L. M., Rodgers, W. H., Gorstein, F. and Osteen, K. G. (1997). Suppression of matrix metalloproteinases inhibits establishment of ectopic lesions by human endometrium in nude mice. J. Clin. Invest. 99: 2851–2857. | PubMed | ChemPort |
17. Bruner, K. L., Eisenberg, E., Gorstein, F. and Osteen, K. G. (1999). Progesterone and transforming growth factor-beta coordinately regulate suppression of endometrial matrix metalloproteinases in a model of experimental endometriosis. Steroids. 64: 648–653. | Article | PubMed | ChemPort |
18. Hull, M. L., et al. (2003). Antiangiogenic agents are effective inhibitors of endometriosis. J. Clin. Endocrinol. Metab. 88: 2889–2899. | Article | PubMed | ChemPort |
19. Bruner-Tran, K. L., et al. (2002). Steroid and cytokine regulation of matrix metalloproteinase expression in endometriosis and the establishment of experimental endometriosis in nude mice. J. Clin. Endocrinol. Metab. 87: 4782–4791. | Article | PubMed | ChemPort |
20. The American Society for Reproductive Medicine, (1997). Revised American Society for Reproductive Medicine classification of endometriosis: 1996. Fertil. Steril. 67: 817–821. | PubMed |
21. Murphy, A. A. (2002). Clinical aspects of endometriosis. Ann. N. Y. Acad. Sci. 955: 1–10. | PubMed |
22. Fortin, M., et al. (2003). An improved mouse model for endometriosis allowing the non-invasive assessment of lesion implantation and development. Fertil. Steril. 80: 832–838. | Article | PubMed |
23. Tabibzadeh, S., Miller, S., Dodson, W. C. and Satyaswaroop, P. G. (1999). An experimental model for the endometriosis in athymic mice. Front. Biosci. 4: C4–C9. | PubMed | ChemPort |
24. Mesnil, M. and Yamasaki, H. (2000). Bystander effect in herpes simplex virus-thymidine kinase/ganciclovir cancer gene therapy: role of gap-junctional intercellular communication. Cancer Res. 60: 3989–3999. | PubMed | ISI | ChemPort |
25. Osteen, K. G., Bruner, K. L. and Sharpe-Timms, K. L. (1996). Steroid and growth factor regulation of matrix metalloproteinase expression and endometriosis. Semin. Reprod. Endocrinol. 14: 247–255. | PubMed | ChemPort |
26. Rossi, G., et al. (2000). Dynamic aspects of endometriosis in a mouse model through analysis of implantation and progression. Arch. Gynecol. Obstet. 263: 102–107. | Article | PubMed | ChemPort |
27. Dabrosin, C., Gyorffy, S., Margetts, P., Ross, C. and Gauldie, J. (2002). Therapeutic effect of angiostatin gene transfer in a murine model of endometriosis. Am. J. Pathol. 161: 909–918. | PubMed | ChemPort |
28. Beliard, A., Noel, A., Goffin, F., Frankenne, F. and Foidart, J. M. (2002). Role of endocrine status and cell type in adhesion of human endometrial cells to the peritoneum in nude mice. Fertil. Steril. 78: 973–978. | Article | PubMed |
29. Noyes, R. W., Hertig, A. T. and Rock, J. (1975). Dating the endometrial biopsy. Am. J. Obstet. Gynecol. 122: 262–263. | PubMed | ChemPort |
30. Rancort, C., Bergeron, C., Lane, D., Garon, G. and Piché, A. (2003). Delivery of herpes simplex virus thymidine Kinase bystander effect by engineered human mesothelial cells for the treatment of ovarian cancer. Cytotherapy. 5: 509–522. | Article | PubMed | ChemPort |

Acknowledgements

We thank the clinicians and the clinical network team for their precious collaboration in endometrial sample collection. We thank Jean-Benot Racine and Pascal Croteau (MetrioGene BioSciences, Montreal, Canada) for helpful advice. We are grateful to François Bouthillier (Biotechnology Research Institute, Montreal, Canada) and Carole Bergeron (Université de Sherbrooke, Sherbrooke, Canada) for the preparation of adenoviral particles.