Female BALB/c mice aged between 6 – 8 weeks utilized in this study were purchased from The Jackson Laboratory and acclimatized for at least one week after delivery. Randomized grouping of animals was performed and every 4 mice were housed in same cage till the end of experiments. All study mice were exposed to same pathogen-free housing conditions with freely accessible food (standard chow) and water, 18 – 25°C temperature, and alternating 12 light/12 dark. Approval from the Institutional Animal Care and Use Committee (IUCAC) and from the University of South Carolina was obtained before performance of any study experiments (AUP2374).

Surgical instruments including Castroviejo scissors, uniband LA-1 micro point scissors, serrated jaw MF-2 micro forceps, fully curved micro forceps MF-3, straight micro forceps soldering tweezers, insertion/extraction tweezers and anti-wicking tweezers were purchased from Cedarlane labs – Canada. Surgical sutures including 5-0 Perma Hand Silk Black 1X18’’ PS-3, 5-0 Perma Hand Silk Black 2X60’’ no needle and 4-0 Perma Hand Silk Black 1X18’’ G-3 were purchased from the Ethicon – USA. Diethylstilbestrol (DES) was purchased from Sigma – USA and prepared in mineral oil (Sigma-Aldrich, USA). GIMA 2mm diameter biopsy punch (GIMA, UK) was used to divide the endometrial layer into 2mm pieces. An 18G syringe needle (BD, USA) attached to a tuberculin syringe (BD, USA) was used to deliver endometrial specimens into the peritoneal cavity. Banamine (Merck, USA) solution was used every 12 hours for 2 two days to alleviate the pain during post-surgical care period. DMEM/F12 medium was used for preparation of endometrial transplant tissue (Sigma-Aldrich, USA). Skin incisions were closed with a Reflex Clip Applier (World Precision Instrument, USA) and clips were removed using Reflex Clip Removing Forceps (World Precision Instrument, USA).

As the endometriosis is an estrogen-dependent disorder, mouse model was designed to mimic this disorder by using ovariectomized mice treated with estradiol (37). Briefly, donor and recipient mice were anaesthetized with inhaled isoflurane before being surgically prepared. The right and left ovaries with attached salpinx were ligated and removed. Muscular layers were sutured with silk, while the skin was closed with metal clips. Systemic analgesia, Benamine, was given every 12 hours post-surgery for 48 hours. Seven days later, the clips were removed and OVX mice were used.

The endometrial tissue was used to induce the experimental endometriosis in mice as described (38, 39). The schematic details are also provided in Figure 1 . Briefly, BALB/c female mice were assigned to one of four experimental groups ( Figure 1 ): OVX+END, ovariectomized mice receiving transplanted endometrial tissue; OVX+VEH, ovariectomized mice injected intraperitoneally with PBS but without transplant; Naïve (N) mice, non-ovariectomized mice with no treatment as control group and lastly, Naïve+END (NE), non-ovariectomized mice receiving transplanted endometrial tissue. All DES treatments were performed as 100µg/kg of DES injected subcutaneously. In the OVX+END and OVX+VEH groups (collectively considered ovariectomized-OVX), a dose of DES was delivered and ovariectomy was performed at day 0, while N and NE groups were housed without intervention. On day 5, OVX mice received another injection of DES in order to stimulate endometrial growth while N and NE groups were still under the influence of endogenous estrogen from their ovaries. Here, N group received no further intervention until they were euthanized. On day 7, OVX mice were split into OVX+VEH group, endometrium transplant donor group (END), and the OVX+END group described above to receive the transplanted tissue. One donor endometrium (END) provided enough tissue to seed the peritoneum of two recipient mice.

The donor mice were euthanized via overdose of isoflurane inhaled anesthesia. In a petri dish containing DMEM/F12 medium, the uterine horns of donor mice were longitudinally opened, and the endometrial layer was pealed from the myometrium and serosa layers using micro forceps and then divided into approximately 2mm pieces. This tissue was then transferred to another petri dish containing phosphate buffer saline (PBS). Approximately 35µg (wet weight) of the endometrial tissue samples were loaded into a 1ml syringe via an 18g needle in a total volume of 500 µl PBS. Loaded endometrial tissues were then transplanted to the peritoneal cavity through abdominal wall of either OVX+END or Naïve+END recipients to complete the transplant procedure.

At days 10, 15, and 20, the OVX+END and OVX+VEH groups received injections of DES, post ovariectomy procedure, while Naïve+END and Naïve did not receive DES still under hormonal influence of intact ovaries. The experimental endpoint was 14 days post endometrial transplant, which was 21 days post OVX procedure.

At the endpoint of the experiments (14 days post-transplantation experiments), all study mice were anesthetized and euthanized under isoflurane for tissue sampling and downstream analysis. Blood samples were collected for complete blood count by VetScan analyzer system (Abaxis, USA) and for serological analysis. Single cell suspension of peritoneal fluid (PF) was prepared by using RBCs-lysis buffer (Sigma-Aldrich, USA) and strained with the 70-micron strainer (ThermoFisher, USA) to be utilized in either real-time metabolism analyzer Seahorse (Agilent, USA) or FACS-Celesta flowcytometry sorting system (BD, USA). Colon flushes (CF) were collected from the colon under aseptic conditions in sterile PBS for microbiome analysis. CF for metabolomic and short chain fatty acids analysis were collected in sterile distilled water. Uterine horns were excised for histopathological investigation with hematoxylin and eosin (H&E) staining and to collect uterine horn lavage fluid (ULF).

PF was collected by injecting the mice under deep anesthesia by isoflurane, then the skin of abdominal wall was opened to expose the peritoneal sac. Five milliliters of sterile PBS were intraperitoneally injected and the mouse then rolled thoroughly for 3 – 5 minutes to elute all the peritoneal traces and cells. ULF was collected by slow-passing 2 – 3 ml of PBS and collected in centrifuge tubes. Collected PF and ULF was then transferred to conical tubes for centrifugation and separation of the supernatant and stored at -80°C for analysis. While the pelleted cells were treated with RBC lysis buffer for 30 – 60 seconds and then blocked by using cold 10%-FBS buffer and washed with cold FACS for cell counting. Cells were resuspended in FACS buffer for counting by using Trypan blue staining and TC20 automated cell counter (40).

Briefly, peritoneal fluid was collected in cold FACS and processed into single cell suspension by using RBC lysis buffer (Sigma-Aldrich, USA) for about 60 seconds and then washed with cold FACS. Then all samples were filtered with 70µm strainer (ThermoFisher, USA), centrifuged at 4°C, 1000 RPM for 10 minutes and then the cell pellets were resuspended in cold FACS. To determine the T cell and natural killer (NK) subset, isolated cells were stained with anti-CD3-APC and NK1.1-BB515 antibodies (Biolegend, USA), respectively. Finally, the stained samples were analyzed by using BD-FACS Celesta flow cytometry system (BD, USA) and acquired data were visualized by using built-in Diva software (BD, USA) as described before (41).

Whole blood was obtained via retroorbital vein rupture by using heparinized capillary tubes inserted into medial canthus of isoflurane-anesthetized mice. The collected blood samples were kept in heparinized tube before being transferred to be analyzed by blood-autoanalyzer system, Vetscan HM5 hematology analyzer (Abaxis, USA) for complete blood count assessment.

To evaluate the proliferative capacity of PF inflammatory cells, 10⁵ cells per well were seeded in DMEM/F12 medium for 12 hours incubated with 1µCui/well of ³H-thymidine isotope at 37°C and 5% CO2, then radioactivity was measured using MicroBeta Trilux liquid-scintillation counter to (36, 42).

Harvested uterine horns were fixed in 4% paraformaldehyde overnight and then stored in PBS at 4°C till the time of sectioning and staining with hematoxylin and eosin (43).

The collected colonic flushes were prepared for 16s rRNA as described by previous publications (44, 45).

Briefly, colons of study mice were opened and about 100 mg of luminal solid contents were removed and then suspended with distilled water. Collected flushes were centrifuged to separate the supernatant and stored at -80°C for future use. At time of SCFA analysis, all samples were thawed gradually in ice box and then acidified by hydrochloric acid (HCL) before adding 4-methylvaleric acid as internal standard. Gas chromograph CP-3800 (Varian) and spectrometry mass (GC-MS) system was used to quantify the SCFAs. Varian MS Workstation (version 6.9.2) was used to collect and analyze acquired data. Finally, the linear equation was used to calculate the concentrations of SCFAs in each sample (41, 46).

Serum was separated from study mice and used for metabolic extraction by using liquid chromatograph mass spectrometry (LC-MS) as described in our previous study (36). Briefly, metabolites derived from the TCA and glycolysis were characterized by using 5mM of ammonium acetate (pH9.9) and other buffers in assistance of Luna 3µM NH2 100 A° chromatography column (Phenomenex, CA). All identified metabolites were normalized to internal standard (47).

These two metabolic pathways were evaluated by using real-time metabolism analyzer Seahorse (Agilent, USA) according to the manufacturer’s protocols. The principle of measurements was based on the calculation of oxygen consumption rate (OCR) and extra cellular acidification rate (ECAR). Cells (3X10⁵) taken from single cell suspension of PF was seeded into XFp miniplates by assistance of Cell-Tak biological adhesive (Corning, USA). At the end of analyzer run, Hoechst 33342 dye (Invitrogen, USA) was applied to the plate wells to count the remaining live cells in order to normalize the acquired data based on the number of living cells used in this run. Counting of live cells was performed by using automated imaging microscope, Cytation5 Imaging System (BioTek, USA). Generated data by Seahorse were transferred to be interpreted by Seahorse Wave Desktop Software (Agilent, USA) to calculate the kinetic energy (36).

All experiments of this study were repeated 3 independent times. Sample sizes varied between 3 – 5 mice per group and the numbers are stated in the figure legend. Metabolomes were normalized and analyzed as Log2 examined by t-test by using R-studio (R Studio Inc, USA) as described previously (36). One-Way ANOVA test was applied whenever there were more than two groups to compare with multiple Tukey’s correction. The t-test was applied with Holm-Sidak correction method to compare two groups. p<0.05 considered as significant threshold and the levels were depicted as: *p<0.05, **p<0.01, ***p<0.001 and ^#p<0.0001. When statistical comparison includes more than three groups, different lowercase letters were used to depict significant differences among these groups. For example, groups having the same letters ‘a’ would be insignificant while groups depicting ‘a’ and ‘b’ would be statistically significant.

At fourteen days post-receiving endometrial transplantation, local and systemic immunologic responses were evident ( Figures 2 , 3 ). Histopathological investigation of uterine horns excised from the four groups of mice using H&E staining showed that in the OVX+END group there was an increase in the inflammatory cells (yellow arrows) to the endometrium as well as myometrium ( Figures 2A, B ) when compared to the control OVX+Veh or Naive groups. By using ImageJ software, the statistical analysis showed significant increase in the number of inflammatory cells (mostly neutrophils) in the endometrium (endometritis) as well as in myometrium (myometritis) of OVX+END group in comparison with all other study groups ( Figure 2C ). There was also significant augmentation (p<0.05) in the inflammatory cells in the peritoneal cavity of OVX+END group when compared to OVX+VEH or Naïve groups ( Figure 3A , left panel). Interestingly, the Naïve+END group also behaved similar to OVX+END group. The cell counts in the uterine lumen were not significantly altered in all groups tested ( Figure 2A , right panel). The total WBC counts in the peripheral blood of the OVX+END group was significantly higher (p<0.05) than the OVX+VEH and naïve mice ( Figure 3B , right panel). The percentage of NK cells were higher in the OVX+END group when compared to other groups ( Figure 3C ). The total number of CD3⁺ T cells and CD3⁺NK⁺ cells in the peritoneal cavity were found to be significantly increased (p<0.05) in the OVX+END group when compared to Naïve or OVX+VEH groups ( Figures 3D, E ). The percentage of lymphocytes and neutrophils in the peripheral blood were studied too. When we tested the percentage of lymphocytes in the blood, it was decreased in all groups when compared to the naïve group ( Figure 3F ). The neutrophil percentage in the peripheral blood of mice transplanted with endometrial tissue (Naïve+END, OVX+END) was significantly higher (P<0.05) than in the non-transplanted mice ( Figure 3G ). ³H-thymidine incorporation assay was performed to estimate the proliferative capacity of the inflammatory cells in the peritoneal fluid and to assess the ability of these cells to proliferate. Interestingly, the results showed that there was significant increase (p<0.05) in proliferation capacity of inflammatory cells isolated from OVX+END group when compared to Naïve or OVX+VEH group. The Naïve+END mice also demonstrated a similar response as the OVX+END mice ( Figure 3H ). Overall, the results indicated that the endometrial transplants in OVX mice were associated with an exacerbated inflammatory response in the peritoneal cavity and peripheral blood. It was interesting to note that naïve mice that received the endometrial transplant (Naïve+END), also exhibited some immunological changes like OVX+END group such as increase in the percentages of neutrophils while with respect to most of the immunological profiles studied, they behaved like the naïve controls.

Investigation of gut microbial communities via sequencing the 16s rRNA V3-V4 regions showed that there was dysbiosis in the microbial environment of the colon in all experimental groups compared to the Naïve group ( Figure 4 ). Bioinformatic analysis of our collected data showed that the commensal bacteria of individual mice were clustered closely within their own group, while the 4 groups tested showed clear segregation ( Figure 4A ). A cladogram generated from linear discriminant analysis effect size (LEfSe) demonstrated that Clostridiales_order, Tenericutes phylum and its down tree ancestry, Mollicutes class, Anaerplasmatales order, Anaeroplasmaceae family, and Anaeroplasma genus constituted the detectable commensal bacteria in the guts of the Naïve group ( Figure 4B ). The Naïve+END and OVX+END mice showed simultaneous deviation from Naive and OVX+VEH gut microbiotas and showed similarities between each other with the enrichment of Phylum Tenericutes, Class Mollicutes, Order Aneroplasmatales, and Genus Aneroplasma. Lacnospiraceae family and its Coprococcus genus, Ruminococcus genus, and Ruminococcus gnavus were commensal biomarkers in the Naïve+END group that were not present in the Naïve group ( Figure 4B ). Also, Clostridiales order was the biomarker bacteria in the colons of OVX+VEH ( Figure 4B ). Statistical analysis of occupational taxonomy units (OTUs) of bacteria showed significant alterations (p<0.05) among the study groups ( Figure 4C ). Furthermore, linear discriminant analysis (LDA) scores set at >2-fold change among the study groups (Naïve, Naive +END, OVX+VEH and OVX+END), using PICRUST-generated level 3 KEGG pathways of 16s rRNA data revealed that there was no biomarker functional pathway of the bacteria for Naïve+END group ( Figure 4D ). The other groups showed different pathways involved in the bacterial metabolism that are related to the host metabolism ( Figure 4D ). Together, these data suggested that each treatment group had a distinct population of bacteria ( Figure 4A ) and that Naïve or OVX mice that received endometrial transplants exhibited enhanced gus-enriched bacteria like Ruminococcus spp. which in turn could improve the E2 metabolism as a consequence ( Figure 4B ).

Next, we studied the bacterial end-product concentrations in the guts represented by the short chain fatty acids (SCFAs) by using Gas Chromatography-Mass Spectrometry (GC-MS) showed in Figure 5A . Upon examination of the 4 groups of mice, the overall trend indicated that significant (p<0.05) decreased levels of acetic acid ( Figure 5B ), propionic acid ( Figure 5C ), butyric acid ( Figure 5D ) and valeric acid ( Figure 5E ) in OVX+END group when compared to the Naïve group. Similar changes were also seen in OVX+VEH group suggesting that ovariectomy and DES treatment had significant impact on colonic SCFAs. The Naïve+END group also exhibited both similarities and differences in SCFAs when compared to the Naïve group thereby suggesting that endometrial transplants in naïve mice also induced some changes in SCFAs. Together, these data suggested that bacterial metabolites, SCFAs, exhibit differential expression following endometrial transplants and because SCFAs also regulate immune cells, the SCFAs may play a role in immune response during endometriosis.

We next studied the interplay between estrobolome dysregulation of microbiota and their potential influence on the immune cells’ metabolomic response in the endometriosis model using cells from the peritoneal cavity. These cells were seeded into specific culture plates of XFp Seahorse real-time metabolism analyzer before being incubated and analyzed based on calculation of oxygen consumption rate and extracellular acidification rate to estimate the ATP production rate ( Figures 6 , 7 ). Real-time metabolism analysis revealed that the basal and stressed mitochondrial respiration of immune cells were significantly (p<0.05) higher in the Naïve+END ( Figures 6A, B ) and OVX+END groups ( Figures 7A, B ) in comparison to non-transplanted mice of Naïve and OVX+VEH groups, respectively. Furthermore, the main source of ATP production in the Naïve group was via mitochondrial respiration ( Figures 6C, D ). However, transplantation of endometrial tissues led to significantly (p<0.05) accelerated mitochondrial ATP production which in turn significantly (p<0.05) increased the total ATP molecules in cells from Naïve+END group ( Figure 6D ). In the OVX groups, endometrial transplantation modified the respiratory pathways of inflammatory cells ( Figures 7B, C ). Stimulating the inflammatory cells in the peritoneal cavity of OVX+END mice led to further significant (p<0.05) increase in the rate of ATP production via mitochondrial and glycolytic processes in comparison with OVX-VEH group ( Figure 7D ). Together, these data suggested that endometrial transplantation-induced immune cells exhibit accelerated metabolic rate.

TCA-related metabolites levels, such as fumarate, glutamate, malate and succinate were significantly (p<0.05) higher in Naïve+END group than in Naïve mice ( Figure 6E ). Interestingly, glycolysis-related metabolites such as 3-phosphoglycerate and 2-phosphglycerate (3PG and 2PG), glucose-6-phosphate and fructose-6-phosphate (G6P/F6P) and phosphoenolpyruvate (PEP) were significantly (p<0.05) elevated in the plasma of Naïve+END mice in comparison to Naïve group ( Figure 6E ). Interestingly, when such metabolites were studied in OVX+END mice, some of the metabolites showed opposite results. For example, the levels of 3PG, 2PG, lactate, malate, PEP were decreased while some such as glucose/fructose, glyceraldehyde, ketoglutarate, oxalate, and pyruvate were increased in OVX+END when compared to the OVX+VEH group ( Figure 7E ). It was noteworthy that several of the metabolite expression profile was different between OVX+END vs Naïve+END group ( Figure 6E vs Figure 7E ), thereby suggesting that ovariectomy did affect the metabolomic profile in addition to endometrial transplant.