Ovariectomy with simulated vaginal delivery to establish a rat model for pelvic organ prolapse

ABSTRACT The widespread prevalence of Pelvic Organ Prolapse (POP) and the paucity of ongoing treatments prompted us to develop a unique rat model combining ovariectomy and simulated vaginal delivery. We hypothesized that the tissue changes caused by low hormone levels and mechanical stretch could complement each other. Thus, the combined model can potentially mimic the collagen metabolism of vaginal wall tissue as well as mechanical stretch properties to complement disease progression in POP. Ovariectomy with sequential simulated vaginal delivery was performed on rats in the modeling group. Sham surgeries were performed as control. At 2, 4, and 12 weeks after modeling, the vaginal tissues of rats were evaluated by Masson’s trichrome staining, Picro-Sirius red staining, immunohistochemistry, western blotting, and uniaxial tensile tests. Compared to the control group, the vaginal tissues of the model rats showed an atrophic epithelial layer and loose collagen fibers. The smooth muscle fibers were ruptured, smaller in diameter, and disorganized. The ratio of collagen type I/III significantly increased, but the contents of both Collagen I and III decreased. The expression of metalloproteinases 2 and 9 in the tissues increased, and the expression of tissue inhibitors of metalloproteinases 1 and 2 decreased. The tangent modulus of the tissues was significantly increased in the model rats. We verified a novel method to establish a pelvic organ prolapse model in rats. This approach combined the advantages of low hormone levels and mechanical stretch effects.


Introduction
Pelvic organ prolapse (POP) significantly affects the quality of life of women. The incidence of POP is as high as 3.40-10.76% in women 1 . Currently, surgery is the predominant treatment for POP. Since animal models are important tools for POP studies, many attempts have been made to develop an ideal animal model.
In terms of animal selection, rabbits are mediumsized but their pelvic floor anatomy is markedly different from that of humans 2 . Large animals, such as sheep, pigs, and non-human primates (NHP), are thought to be better animals for POP research. NHPs are anatomically and histologically similar to humans 3 . However, large animals usually require more space to feed and incur higher costs, which cannot be afforded by many laboratories 4 . Additionally, stricter ethical constraints are in place for large animals.
Despite having distinct pelvic floor anatomies compared to humans, rats and mice still bear histological similarities in the vaginal wall, ligament, and pelvic floor muscle tissues and are common laboratory animals 2 . Further advantages of rodents are their low maintenance costs, ease of operation, and fewer ethical constraints 5 . Compared to mice, rats have a larger body size and a more human-like histological structure 6 . These features make rats an option when animal models merely mimic the vaginal wall tissue characteristics of POP patients histologically.
In humans, childbirth and hormonal changes were two main factors that contributed to POP 7,8 . Hormonal changes, including estrogen and progesterone, resulted in the extracellular matrix (ECM) remodeling in vaginal tissues of POP patients by modulating the expression of matrix metalloproteinases (MMPs), their tissue inhibitors (TIMPs), and the Lysyl oxidase (LOX) 8,9 . The process of vaginal delivery can cause vaginal wall tissue expansion, which directly causes vaginal wall trauma 10,11 .
Two approaches to mimic these two risk factors are commonly used in animal models: ovariectomy (OVX) and simulated vaginal delivery (SVD). Both OVX and SVD can induce tissue responses resembling those of POP patients, including a decrease in collagen, increased expression of MMPs, decreased epithelial and vaginal wall thickness, and higher stiffness [12][13][14][15][16][17][18] . They also have both advantages and disadvantages. Tissue changes after SVD could occur within 0-3 days, but these changes could become normalized as tissue repair progresses 14,16 . However, after OVX, typical changes in tissues take at least two weeks 18 , and, in most studies, even 4 weeks are needed 12,13,17,19,20 . Tissue changes caused by OVX would not recover. As a result, SVD modeling in rats can induce changes faster but not sufficiently enduring. In contrast, OVX causes slower changes which, however, last longer. Furthermore, SVD induces more typical changes of biomechanical and anatomical nature, whereas OVX-induced changes are most typical in terms of protein expression and the structure of the epithelial layer.
The advantages and disadvantages of these two methods in generating a model of vaginal wall prolapse in rats are complementary. In addition to this, we hypothesized that the low hormone levels caused by OVX could prevent the repair of tissue trauma caused by SVD. Therefore, we proposed that the combination of OVX and SVD in rat vaginal wall POP models can shorten the time of modeling and ensure the stability of the model for a long time with typical characteristics.

Study design
This study has adhered to the laboratory animal welfare regulations and was approved by the ethics committee of Peking Union Medical College Hospital (no. JS-2240, 25/ 02/2020). We used 12-week-old Sprague-Dawley rats (Charles River, Beijing, China) weighing 236 ± 17 g. The animals were raised at a temperature of 20-22 °C, relative humidity of 50-70%, and a 12-h day/night cycle. Food and water were supplied ad libitum. As shown in Figure 1, 24 virgin female rats were used in this study. Eighteen rats underwent ovariectomy (OVX), and six underwent sham surgery. Two weeks after OVX, all 18 rats underwent simulated vaginal delivery (SVD). The vaginal tissue of rats was evaluated at 2, 4, and 12 weeks after SVD. The 6 sham animals were all harvested and evaluated at 4 weeks after sham surgery, which was the same as the 2-week evaluation time of the modeling group.

Surgery procedure
OVX was performed through bilateral abdominal incisions in rats. As shown in Figure 1A, the incision was made on the lateral side of the abdomen about 1.5 cm from the top of the back and 2.0 cm from the groin.
After cutting through the full thickness of the abdominal wall, the white adipose tissue could be found. The adipose tissue was gently pulled out of the abdominal cavity with forceps. Rat ovaries could be found along the adipose tissue. The ovary was bright red in appearance, about 0.5 cm in diameter, and had small nodular processes. After firmly ligating blood vessels and connective tissue at the root of the ovary with sterile sutures, the ovary was removed with a scalpel. The pulled tissue was returned to the abdominal cavity and the incision was then sutured full-thickness. The contralateral side was treated with the same method. In the sham group, rat ovaries were found and observed. Then, the tissue was returned to the abdominal cavity without the procedure of removing ovaries. The rest of the procedure was the same as the modeling group.
SVD was performed using a size 12 Foley catheter ( Figure 1B). The top of the catheter was carefully trimmed, and the balloon, injected with 2.5 mL 0.9% saline solution, was inserted into the rat vagina. SVD lasted for 4 h, followed by the withdrawal of saline solution from the balloon and removal of the catheter.
The rats were sacrificed at the corresponding time points. Full thickness and full length of the vaginal wall was harvested ( Figure 1C). Before harvesting, a suture was stitched at the 12 o 'clock position of the external vaginal opening for marking. Sampling was performed only in the central area of the anterior vaginal wall. Samples were taken only two-thirds of the way down from the top of the vagina. The width of the sampling area was about 5 mm.

Masson's trichrome staining and Picro-Sirius red staining
The samples were fixed using 4% paraformaldehyde, dehydrated, and embedded in paraffin. Sections were prepared by 4 μm thick transection. When different stains were performed on the same individual, serial sections were used to ensure intra-individual consistency. For all individuals, the section region of interest was mainly near the center line. Sections were prepared around the center of the specimen to ensure interindividual reproducibility. After dewaxing, the slices were stained following the typical Masson's trichrome and Picro-Sirius red staining protocols. Briefly, for Masson's trichrome staining, slices were fixed in Bouin's solution and stained sequentially with hematoxylin staining solution, acid ponceau, and aniline blue solution. The slices were observed and photographed using an optical microscope (Nikon, Tokyo, Japan). For Picro-Sirius red staining, the slices were stained with Picro-Sirius Red F3BA (Sigma, St. Louis, MO, USA) according to the manufacturer's instructions. The slices were observed and photographed using a polarization microscope (Nikon). Images of Picro-Sirius Red staining were analyzed using Image-Pro Plus Version 6.0 software (Media Cybernetics, Silver Spring, USA). The red and yellow areas were identified as collagen type I and the green areas as collagen type III. The area ratio of collagen I to collagen III was calculated and compared between the groups.

Elastic Van Gieson (EVG) staining
An EVG staining kit (Jinhong Biotech, Wuhan, China) was used to evaluate elastic fibers. Slices were prepared as described above and the staining was performed following the instructions from the manufacturers. Briefly, the slices were dewaxed and incubated with the elastin staining solution overnight at room temperature. Van Gieson staining solution was then incubated for 1 minute. The slices were observed and Note: Ovariectomy (OVX) was performed first (A), followed by simulated vaginal delivery (SVD) 2 weeks later (B). Sampling was performed only in the central area of the anterior vaginal wall (C). The evaluation was performed at 2, 4, and 12 weeks after modeling. Tissue morphology, collagen expression, and biomechanics (D) were evaluated photographed using an optical microscope (Nikon). Elastic fibers were stained black.

Immunohistochemistry
Slices were prepared as described above. After dewaxing, the antigens were exposed using citrate buffer and blocked with serum. Slices were then incubated with specific antibodies at 4 °C for at least 12 h. The antibodies used in this study were ACTA2 (actin alpha 2, smooth muscle, also known as smooth muscle actin alpha, αSMA) 1:200, MMP2 1:100, MMP9 1:100, and TIMP2 1:200 (ABclonal; Wuhan, China); and TIMP1 1:200 (Invitrogen; Carlsbad, CA, USA). The slices were then incubated with horseradish peroxidase-conjugated secondary antibodies (Solarbio, Beijing, China) at room temperature (20-25 °C) for 1 h. A DAB substrate kit (Solarbio) was used to visualize the sections. Slices were observed and photographed using an optical microscope (Nikon). Images were analyzed using the ImageJ software (National Institute of Health (NIH), Bethesda, MD, USA). When analyzing, the positive area was manually selected. The threshold was manually adjusted to include the positive region as completely as possible. Then, the integrated density relative to the selected area was calculated for comparison between groups. Positive areas of the vascular muscularis were excluded when analyzing ACTA2.

Uniaxial tensile test
Samples were collected and maintained wet throughout the assay using a 0.9% saline solution. A mechanical testing machine (MTS Systems, Minnesota, USA) was used to generate the load-elongation curve. We trimmed the anterior vaginal wall along the longitudinal axis to a length of 12 mm and a width of 2 mm. The vaginal wall was 1 mm thick; hence, the cross-sectional area was 2 mm 2 . The sample was loaded with a pair of clamps spaced 1 cm apart. Before stretching, a force of 1 N was applied. The sample was then stretched at a speed of 12 mm/min, and the data were recorded every 30 ms. The Y-axis of the loadelongation curve displays the stress in MPa, and the X-axis is the percentage of elongation. The maximum strain was defined as the strain rate (%) at fracture. The slope of the curve provides the tangent modulus of the sample. Because the vaginal wall tissue stretch is not linear, we used the tangent modulus at C0 and C1 to describe tissue biomechanical properties. C0 was determined under low deformation (less than 5%) and C1 under moderate deformation (approximately 30-50%). The slopes of C0 and C1 were calculated from the region of the curve that could be best approximated by a straight line.

Statistical analysis
Statistical analyses and graphical representation were performed using GraphPad Prism 8 (GraphPad Software Inc., USA). Data are presented as mean±standard deviation. A two-tailed Student's t-test was used to assess differences between the two groups. The one-way analysis of variance was used to evaluate the differences among the three groups. The modeling group rats at different time points were compared with the control group rats at 4 weeks after sham surgery. Statistical significance was defined as P < 0.05.

Tissue morphology and structure
As shown in Figure 2A-D, Masson staining showed that the epithelium layer atrophied significantly after SVD modeling. In the lamina propria, compared with the control group, the collagen fiber of the vaginal wall was loose, and there was an obvious separation among collagen fibers. In EVG staining, the elastic fibers had a greater tortuosity and shorter length in the modeling group. As shown in Figure 2E-H, the areas where the differences were more pronounced were marked with arrows.

Smooth muscle histomorphology and content
In immunohistochemistry (IHC) for ACTA2 ( Figure 2I-L), the muscle fibers in the control group were properly structured and thick. However, in the model group, the muscle fibers were ruptured, smaller in diameter, disorganized, and more widely distributed. The difference was  Figure 2L). The regions of muscular layer were labeled with red lines in Figure 2I-L. There was no significant difference in the ACTA2 content among the different groups in both the size of the IHC-positive area and the band thickness and intensity in western blot assays ( Figure 2M-N).

Collagen content and composition
Using Picro-Sirius red staining for rat vaginal wall tissue, we found that the ratio of collagen type I/III significantly increased in the modeling group at weeks 2, 4, and 12 when compared to the control group (Figure 3A-M; P = 0.002, P < 0.001, and P = 0.002, respectively). The collagen type I/III ratio also significantly increased from week 2 to week 4 (P = 0.002). In western blots, collagen I was significantly decreased at weeks 2, 4, and 12 compared with the control group ( Figure 3N, P = 0.011, P = 0.001, and P = 0.003, respectively). Collagen I levels at week 4 significantly decreased from week 2 (P = 0.018). Collagen III also significantly decreased at weeks 2, 4, and 12 compared to that in the control group ( Figure 3O, P = 0.007, P = 0.006, and P = 0.009, respectively). There was no significant difference in the modeling group among different time points (P = 0.646).

Matrix metalloproteinases (MMPs) expression
IHC and western blotting were also performed to compare the expression of MMP2 and MMP9 in rat vaginal wall tissues (Figure 4). In IHC, the expression of MMP2 at weeks 2, 4, and 12 was significantly higher than that in the control group ( Figure 4A-D, I, P = 0.003, P = 0.013, and P = 0.031, respectively), and the modeling group did not show a significant difference at different time points (P = 0.728). The expression of MMP9 at weeks 2, 4, and 12 was also significantly higher than that in the control group ( Figure 4E-H, I; P = 0.015, P = 0.012, and P = 0.008, respectively), and the modeling group at different time points showed no significant difference (P = 0.606). The typical regions were labeled with arrows in Figure 4.
Western blot expression of MMP2 at weeks 2, 4, and 12 was significantly increased ( Figure 4J, P = 0.005, P = 0.014, and P = 0.002, respectively), and there was no significant difference within the modeling group (P = 0.796). The expression of MMP9 at weeks 2, 4, and 12 also increased significantly ( Figure 4K, P = 0.020, P = 0.007, and P = 0.005, respectively) but there was no significant difference between the groups (P = 0.823).

Tissue inhibitor of metalloproteinases expression
In IHC for Tissue inhibitor of metalloproteinases 1 (TIMP1) in rat vaginal wall tissues ( Figure 5), quantitative analysis showed that the expression of TIMP1 was significantly decreased in the modeling group at weeks 2, 4, and 12 ( Figure 5A-D, I, P = 0.005, P = 0.001, and P = 0.002, respectively), and no significant difference was found within the modeling group (P = 0.191). The expression of TIMP2 was significantly decreased also in the modeling group at weeks 2, 4, and 12 ( Figure 5E-H, I, P = 0.003, P = 0.001, and P = 0.001, respectively), and no significant difference was found within the modeling group (P = 0.104). The typical regions were labeled with arrows in Figure 5.
Western blot analysis showed that the expression of TIMP1 at weeks 2, 4, and 12 were significantly decreased ( Figure 5J, P = 0.011, 0.011, and 0.034, respectively), and there was no significant difference within the modeling group (P = 0.597). TIMP2 expression at weeks 2, 4, and 12 also significantly decreased ( Figure 5K, P = 0.005, 0.002, and 0.005, respectively), and there was also no significant difference within the modeling group at different time points (P = 0.349).

Tissue biomechanical properties
The maximum strain at weeks 2 and 12 was significantly decreased compared with the control group ( Figure 6B, P = 0.034, and P = 0.028, respectively). C0 and C1 were calculated to compare the biomechanical properties in different groups ( Figure 6). The C0 value was significantly higher in the modeling group at weeks 2, 4, and 12 than in the control group ( Figure 6C, P = 0.002, P = 0.012, and P = 0.005, respectively), as well as the C1 value at weeks 2, 4, and 12 ( Figure 6D, P = 0.014, P = 0.030, and P = 0.024, respectively). Both CO and C1 showed no significant differences within the modeling group at the different time points (P = 0.513 and P = 0.880, respectively).

Discussion
We established a rat model of POP with typical POP tissue and biomechanical characteristics by using OVX combined with SVD.
In POP patients, collagen composition and content were significantly altered in the vaginal wall tissue compared to non-POP individuals. Collagen types I and III were the main subtypes in vaginal wall tissues. Previous studies have revealed that the content of collagen types I and III decreased in the vaginal wall tissues of POP patients compared to those without POP [21][22][23][24] . Type I collagen is associated with tissue strength and type III with tissue elasticity. The ratio of collagen type I to type III was reported to be increased in the vaginal wall of POP patients 25,26 . Collagen catabolism in the vaginal wall tissue has been reported to be more active in patients with POP. The high catabolic activity was mainly manifested by an increase in MMPs and a decrease in TIMPs. The expression of MMPs, including MMP1, MMP2, MMP3, MMP8, MMP9, and MMP12, was reported to be elevated in the vaginal wall tissue of POP patients 24,[27][28][29][30] . Accordingly, the expression level of TIMPs, mainly TIMP1 and TIMP2, decreased in the POP group 24,27,29,31 . In this study, the changes in collagen and TIMPs in our rat model were consistent with those in POP patients reported in the literature. Among MMPs, only MMP2 and MMP9 increased in our rat model.
The biomechanical properties of the human vaginal wall are difficult to measure because it is difficult to obtain large and complete pieces of vaginal wall tissue. Several studies have investigated the stiffness of the vaginal wall in cadavers and tissues from patients with POP. The results showed that under low tension, the stiffness of the non-POP group was between 0.11 and 0.49, while that of the POP group was between 0.27 and 5.51. When measured under moderate tension, the stiffness of the non-POP group was between 0.55 and 4.01, while that of the POP group is between 8.81 and 17.1 [32][33][34][35] . This indicates that the stiffness of the vaginal wall tissue significantly increased in patients with POP.
An increase in the stiffness was also observed in our rat model.
Multiple parturition and menopause are two important risk factors for developing POP in women 36,37 . Multiple parturitions may be the strongest predictor of POP. Vaginal delivery causes extensive damage to the pelvic floor supporting tissue and causes the so-called "balloon effect" 37 . However, rat fetuses are too small; thus, it was difficult to induce POP through multiple parturitions 38 . Menopause is an independent risk factor of POP. Menopause contributes to POP because hormonal changes disrupt the balance between collagen anabolism and catabolism in pelvic tissues. SVD and OVX are the most frequently used methods for studying POP in rats which Note: Immunohistochemistry evaluation was performed for MMP2 (A-D) and MMP9 (E-H), and the positive area percentage was analyzed (I). Western blotting was performed to evaluate the expression of MMP2 (J) and MMP9 (K). *, P < 0.05. Scale bar = 50 μm. Data are presented as mean ± standard deviation simulated the two risk factors of vaginal delivery and menopause.
Significant changes in serum estradiol, progesterone, LH, and FSH levels after OVX do not occur until 3 weeks later 39 . Therefore, histological and biomechanical alterations following OVX do not occur early in most studies. In previous studies, as shown in Table 1, changes in tissue morphology, collagen, MMPs, and biomechanics usually appeared 8-16 weeks after OVX 12,[17][18][19] .
Balloon SVD in rats mimics the damage to the tissue structure during natural delivery. Changes in the vaginal wall in rats were observed from 2 days to 4 weeks after SVD [14][15][16] , which indicates that they occurred earlier but recovered shortly after SVD. In a study by Downing et al., rats at 2 days after SVD had higher compliance and greater tortuosity in elastic fibers than virgins, but they returned to normal at 2 weeks. The rationale of our hypothesis was that the OVX would reduce the ability of tissue recovery. The trauma caused by SVD can therefore last longer.
Similar attempts had been made in previous studies in other pelvic floor dysfunction diseases. Aranha et al. investigated the effects of OVX plus SVD on the urethral tissue 41 . They found that a combination of OVX and SVD decreased both collagen I and III while SVD decreased only collagen III. Previous studies have reported that combining vaginal distention with pudendal nerve crush could prolong the anatomical and functional recovery of the vaginal wall 15,42 . These studies Western blotting was performed to evaluate the expression of TIMP1 (J) and TIMP2 (K). *, P < 0.05. Scale bar = 50 μm. Data are presented as mean ± standard deviation indicated that the combination of two methods might overcome the shortcomings of SVD. The model could be established quickly and maintained for a relatively long time. This may overcome the limitations of using these two methods alone. Our results showed that a rat model with typical POP tissue characteristics was established. The model remained stable for 2-12 weeks after construction. Moreover, changes in the stiffness of the vaginal wall were inconsistent in previous studies (Table 1). Our results showed an increase in tangent modulus which was consistent with that in patients.
The strengths of this study are as follows. First, this two-step method of generating a rat model resulted in significant changes both in terms of histopathology and biomechanics. Second, our method established a usable model within the relatively short time of 4 weeks compared to OVX alone, and alterations in the model remained stable for a relatively long time of 12 weeks compared to SVD alone.
The main limitation of our work was that the rats are small in size and do not walk upright like humans which precluded the observation of bulging symptoms in our model. This model bore only histopathological POP features, but no anatomical features. For those studies involving the effects of therapies on bulging symptoms, largesize animals should be more suitable. Given the strengths and limitations, some specific types of studies were potentially suitable for the application of this model. The effect of hormones and drugs on the vaginal tissue might be observed in this model. The biocompatibility of exogenous implants, including artificially synthesized and biological mesh, might be evaluated using this model. The evaluation of stem cell transplantation or immunotherapy might also be performed on this model. Another limitation was that Note: The biomechanical evaluation was performed by the uniaxial tensile test. Tangent modulus was calculated at low strain as C0 and moderate strain as C1 (A). The maximum strain was recorded and compared (B). The tangent modulus of tissues at C0 (C) and C1 (D) was compared between different groups. *, P < 0.05. Data are presented as mean ± standard deviation we set up the OVX+SVD and control groups but lacked OVX or SVD alone. Thus, our sole comparison between the two models had been by querying the literature. This might have contributed to a potential bias in our study. Our research was only a preliminary exploration of this model. When applied in the future, researchers should specifically verify the corresponding indicators according to their research objectives. When verifying, the combined method, two single methods, and the control group should all be tested to achieve effectiveness and economy.
According to our results, the combination of the two methods was a feasible strategy. The effects of a combination of SVD with pudendal nerve crush were previously evaluated on urethral tissue and presented a superiority compared with pudendal nerve crush alone 15,42 . SVD plus pudendal nerve crush should also be tried in POP animal models in further studies.
In conclusion, we demonstrated the feasibility of constructing a rat model of POP by performing SVD after OVX. The rat model showed typical characteristics of POP in biomechanics, collagen content, composition, and metabolism.

Disclosure statement
No potential conflict of interest was reported by the author(s).