Burrowing behavior is a potential non-invasive proxy for lesion development in a syngeneic murine model of endometriosis

Endometriosis (EM) is one of the most prevalent reproductive disorders among women. It is characterized by the presence of viable endometrial tissue growing in extra-pelvic regions [1]. This gynecological disorder is prevalent in 10% of women of reproductive age [2], with increased frequencies in those experiencing dysmenorrhea/chronic pelvic pain (CPP) (40–60%), subfertility (30–50%), and pelvic discomfort (71–87%), which are the frequent manifestations of EM [3]. Despite its high prevalence and significant impact on quality of life, the variability in EM severity and lesion traits indicates a complex and not thoroughly understood etiology [4]. Murine models have been widely and effectively exploited to investigate EM. Among these, the syngeneic approach, which involves intraperitoneal (i.p) injection of mouse endometrial tissue into immune-compatible hosts, is extensively utilized for in vivo EM modelling. However, it is limited by an inconsistent lesion induction rate (70–80%) after injecting the donor endometria [5, 6]. Hence, without clear confirmation of lesion existence in all EM-induced animals, the experimental groups risk including lesion-negative animals before the intervention, potentially diluting the treatment effects. Thus, a partial success rate introduces a limitation for intervention research. Further confounding this uncertainty, the EM induction success rate is assessed only upon dissection [6‐10]. The advanced imaging techniques, such as bioluminescent genetic manipulation, contrast agents, high-resolution ultrasound, and MRI, have been explored as non-invasive alternatives to monitor and track the course of EM lesions in real-time [11‐18]. But these approaches are resource-intensive, technically demanding, and not routinely used in many research laboratories. Therefore, there is a pressing need for a simple, cost-effective, reliable approach to identify lesion-positive animals during the experimental timeline.

Burrowing is a conserved, spontaneous behavior in rodents that reflects an animal’s general well-being and motivation to interact with its environment [19]. Decreased burrowing has been observed in various models of prion disease [20], brain lesions [21], complex pain syndrome [22], LPS-induced systemic inflammation [23], Alzheimer’s disease [23], and post-laparotomy pain [24], making it a promising non-evoked behavioral readout. Recent investigations have also demonstrated low burrowing performance in a mouse model of EM [25], showing that lesion-induced pain may affect spontaneous behaviors. However, whether burrowing behavior can be employed as a credible predictor of lesion development and disease progression has not been investigated. In this study, we leveraged a basic experimental setup of BA [19], to evaluate alterations in “non-evoked pain” by monitoring the behavioural alterations of EM mice. Our study aimed to determine if this single assay could provide valuable information regarding lesion burden in a syngeneic EM mouse model.

Statistical analysis and visualization were performed in GraphPad Prism 8.0.1, Inc. (GraphPad Inc., USA). Student’s t-tests or one-way ANOVA (Analysis of Variance) were used to compare the means associated with the study parameters across the groups. For ROC curve analysis and linear regression (LR), SPSS 30.0.0.0 was used. All graphical data are plotted as the means with standard errors of the means (SEMs). Statistical significance for all tests was established at P < 0.05 for data meeting the necessary assumptions for parametric analyses. Significance levels were as follows: * P < 0.05, ** P < 0.01, *** P < 0.001, **** P < 0.0001.

Post-dissection, to validate the impact of burrowing, animals were classified into low burrow score with lesions (LB+) and high burrow score without lesions (HB-) categories based on burrowing score and lesion presence. Among EM-induced mice, every LB mouse had at least one ectopic lesion. Importantly, the burrowing assay demonstrated a near-perfect classification accuracy in differentiating between lesion-positive and lesion-negative EM mice, since none of the EM-induced mice exhibiting burrowing performance above the diagnostic threshold were found to have lesions following dissection. This binary categorization further strengthens the translational potential of burrowing as an early incidence indicator.

To assess systemic neuroinflammation in EM mice, we quantified inflammatory cytokines in the blood and compared the levels to those measured at baseline in control mice. To explore the relationship between systemic cytokine expression, lesion presence, and non-evoked altered behavior, we examined cytokine profiles to burrowing performance across mouse strains. ELISA analysis of serum samples revealed strain-specific changes in inflammatory cytokine levels. IL-6 levels were significantly increased in C57BL/6j EM mice (green panel; t value = 2.786, *p value < 0.05) and showed a marginal increase in BALB/c EM mice (t value = 0.5914) with no significant difference in Swiss albino EM mice (t value = 0.4873) (Fig. 9A). TNF-α levels (blue panel) were significantly elevated in BALB/c EM mice (t value = 3.284, **p value < 0.01, with a slight increase in C57BL/6j EM mice (t value = 1.39) and no change in Swiss albino EM mice (t value = 0.2654) (Fig. 9B). TGF-β levels (orange panel) did not show significant differences among the EM (n = 6) and control groups (n = 6) for any of the three strains (C57BL/6j t value = 0.7291; BALB/c t value = 0.7214; Swiss albino t value = 0.1172) (Fig. 9C). Data reflect mean ± SEM and t value and p value generated by the unpaired t test from control n = 6 and EM n = 6 for all three strains. (*p value < 0.05, **p value < 0.01 represent statistically significant values.

Bild vergrößern

In recent years, a massive amount of published research has been conducted using mouse models of EM to find novel therapeutics. However, a translational gap often exists between successful preclinical treatment strategies and real-world efficacy [32]. The main cause of this discrepancy is the lack of thorough verification of EM lesion formation in preclinical studies, resulting in interventions being tested on animals that might not have even developed the condition. Moreover, EM does not occur spontaneously in rodents, likely due to the absence of endometrial shedding during the estrous cycle. To utilize this paradigm, homologous or heterologous UF must be surgically implanted or s.c/i.p injected into the peritoneal cavity of mice [33]. Regarding the preclinical models of EM, the syngeneic model is the second most common model, where the main method of delivery is intraperitoneal injection of UFs (62% of publications) without causing immunological rejection. Additionally, it does not prime the original immune system, unlike surgical administration [34]. Therefore, we chose to implement the syngeneic model of EM for our research. The inherent variability in the lesion establishment rate may be due to factors such as estrogen, the immune system, genetic background, and varied inflammatory responses [35]. In our study, we observed lesion development rates of 84.61% in C57BL/6j (n = 26), 84.37% in BALB/c (n = 32), and 67.85% in Swiss albino mice (n = 28), highlighting this inherent variability. The Swiss albino strain is generally known to have a different immunological profile compared to the C57BL/6j and BALB/c strains [36], which likely contributes to this reduced lesion formation success. Overall, our recent work has demonstrated that among different strains of syngeneic mice, the Swiss albino strain fared poorly in terms of phenotypes of fibrosis and pain-associated symptoms [26]. The substantially reduced lesion incidence in Swiss albino mice accounts for their behavioral profile differences across behavioral experiments. This variation emphasizes the importance of identifying precise behavioral markers of lesion presence, since not all EM-induced animals developed ectopic lesions (Fig. 10).

Bild vergrößern

Moreover, there are inconsistencies in how disease burden is captured in animal models [37]. The behavioral assays in animal models rely primarily on evoked responses to external stimuli, whereas clinical evaluations emphasize spontaneous and ongoing symptoms reported by patients [38]. Existing non-surgical EM models often fail to incorporate reliable behavioral markers to evaluate ongoing spontaneous pain and lesion incidence. This created a significant gap in replicating patient symptomatology in vivo [39]. Thus, there is a pressing need for a murine model that can closely mirror the clinical heterogeneity of EM by integrating lesion incidence with behavioral readouts, thereby bridging the gap between animal and clinical research.

One of the continuing obstacles in preclinical EM research is the lack of a 100% induction success rate. One of the primary reasons for the discrepancy is that EM mice used in pre-clinical research often with unconfirmed lesion formation. The available induction procedures, such as surgical implantation or intraperitoneal injection of UF, failed to ensure consistent lesion establishment since the lesion formation is influenced by various other factors, such as estrogen, the immune system, genetic background, and varied inflammatory responses [40]. Several studies show promising effects in lesion regression or pain reduction without accounting for whether all experimental animals effectively developed EM before intervention. This diversity hampers the evaluation of therapy outcomes, as it remains unclear whether the absence of lesions at the study endpoint signifies the treatment efficacy or merely unsuccessful model induction [6‐9]. These limitations highlight the necessity for fully verified EM models that provide consistent evaluation of lesion existence before and post treatment. The correlation between pain severity and EM presentations remains inadequately defined [39]. However, epidemiological evidence suggests that the relationship between pelvic pain intensity and EM is independent of the lesions’ size, anatomical locations, or number of lesions and EM stage [41, 42]. Additionally, the sole approach to validate EM formation is through endpoint dissection, which poses a considerable obstacle for translating findings to clinical therapeutic applications. While EM pain may be provoked by physical stimuli (e.g., dysuria, dyspareunia, dyschezia), a significant portion of the chronic pain lacks a definitive mechanical trigger [43]. Thus, measuring non-evoked behavioral responses in mouse models may better reflect patient experiences and facilitate translational studies. Hence, we wanted to study whether altered burrowing behavior can be a biomarker of EM incidence. Then we investigated pain or general discomfort using both spontaneous and evoked behaviors. Burrowing, one of the social habits of rodents, is significantly conserved in laboratory rodents [19]. Bashir et al. have recently demonstrated that EM mice showed reduced burrowing compared to control animals [25]. Their findings align with our hypothesis, and accordingly, we observed a significant and consistent decrease in burrowing performance of EM mice compared to the control mice. This demonstrates traits similar to those of symptomatic women with EM, who reported a diminished quality of life and heightened psychological distress. The behavioural shift was also complied with the presence of EM lesions post-sacrifice. The combined analysis of burrowing data from all three strains using ROC curve analysis yielded a combined AUC of 0.883 (model quality 0.95), demonstrating that a BA is a highly sensitive test for the presence of lesions within this experimental context. We observed a weak correlation between burrow score and lesion number(r² = 0.008). This lack of a strong correlation closely mirrors the clinical presentation of EM in human patients. In women, the severity of chronic pelvic pain is often independent of the lesion size, anatomical location, or the number of lesions. Thus, BA effectively predicts the presence of the disease and associated discomfort, but the resulting low burrowing is not solely governed by the quantity of the ectopic lesions.

A drawback of most previous investigations is the absence of additional validation to verify decreased burrowing activity as a pain-specific outcome. So, it is essential to compare non-evoked burrowing behaviour-related outcomes with other pain metrics to better understand their interrelation. For instance, in a study examining neuropathic pain in rats, burrowing performance exhibited no correlation with evoked mechanosensory thresholds, concluding decreased burrowing may signify elements of pain, which are not directly associated with increased sensory responses [44]. This constraint leaves the relationship between burrowing behavior and pain largely unknown, encouraging further investigation. Given that EM frequently causes significant pelvic pain that affects both the evoked and non-evoked responses, this finding has clinical significance [45]. Also, it has been hypothesized that the pain reported in burrowing tests is spontaneous, rather than evoked pain, as earlier studies revealed no association between the quantity of material burrowed and an evoked paw withdrawal measurement [44, 46]. In contrast, we showed a correlation between lower burrowing activity and pain-response assays. Hence, these correlations enhance the utility of burrowing as a behavioral predictor of pain and disease incidence [47]. To our knowledge, this is the first study to show combined evoked and non-evoked pain responses and their correlation with burrowing behavior in a syngeneic mouse model of EM. Compared to the control mice, which showed a distinctly disrupted, scattered exploring pattern, characterized by an increase in the number of central entries, the EM mice exhibited decreased exploratory behavior and increased anxiety-like behavior, spending more time in the peripheral zones [27]. EM mice also exhibited hyperalgesia. Given that EM frequently causes significant pelvic pain that affects both evoked and non-evoked responses, this finding has clinical significance [45]. It has been previously hypothesized that the pain reported in BA is more akin to spontaneous rather than evoked [44, 46]. However, our findings reveal a clear correlation between poor burrowing activity and heightened sensitivity in evoked behavioral assays, further strengthening the validity of burrowing as a behavioral predictor of lesion incidence in this context [47].

Our preclinical mouse model also confirms its significance by assessing the levels of a subset of pro-inflammatory markers in the serum of EM mice to determine the presence of a local inflammatory milieu (51). Previous findings demonstrated that local inflammation is a crucial mediator in adhesion formation, with IL-β, IL-6, and TNF-α identified among major participants [48, 49]. TNF-α has been involved in early phases of EM progression in mice, and its rise has been linked with chronic pelvic pain [50‐52]. We discovered increased pro-inflammatory cytokines (IL-6, TNF-α, and TGF-β) in the serum of C57BL/6j and BALB/c EM mice, but not in Swiss albino mice. This strain-specific variance underlines the heterogeneity for strain-specific changes in cytokine responses. The elevated cytokine levels we observed presumably contribute to both lesion persistence and pain onset.

Our research demonstrates the feasibility of utilizing an inexpensive burrow tube as a cost-effective, non-invasive instrument for assessing lesion burden in EM mice before endpoint analysis. In conclusion, when used as a biomarker, classical burrowing behavior serves as a reliable biomarker, yielding a high-confidence assessment of EM incidence. This finding underscores the potential of BA as a vital tool in designing and evaluating preclinical studies, facilitating the distinction between animals with lesions and those without.