Clinical samples

This study was approved by the Medical Ethics Committee of the Fifth Affiliated Hospital of Sun Yat-sen University (approval number: 2019-K330-1) and informed consent was obtained from the individual. Between 2019 and 2021, cervical mucosal samples were collected from 70 healthy physical examinees and 72 patients diagnosed with EMs at the Fifth Affiliated Hospital of Sun Yat-sen University. Diagnosis of EMs was established using B-ultrasound, followed by laparoscopic intervention and histopathological confirmation. Cervical mucosal samples were collected as follows. First, the cervix was exposed using a vaginal speculum. Next, vaginal cleansing was performed using 0.9% saline. Then, cervical mucus was collected from the cervical canal lumen, spanning from the external os to 2 cm inward. All samples were collected between the 7th and 14th day of the menstrual cycle. The cervical mucus was collected as evenly as possible using a sterile cotton swab and placed into tubes containing 1 mL of sterile phosphate-buffered saline (PBS). The tubes were immediately immersed in liquid nitrogen and stored at − 80 °C until further use. Ectopic EMs tissues were either fixed in 4% normal buffered formaldehyde for histological analysis or frozen in liquid nitrogen for subsequent experiments. Patients (1) aged 18–50 years, with a regular menstrual cycle (21–35 days) and menstrual bleeding lasting 2–7 days; (2) with no history of undergoing invasive procedures or using antibiotics or anti-inflammatory drugs in the past 3 months; and (3) with no documented history of sexually transmitted infections (e.g., Chlamydia trachomatis and Neisseria gonorrhoeae), reproductive tract malformations, adenomyosis, endocrine diseases, or tumors were examined. Demographic and clinical data, including age, BMI, relationship status, pregnancy status, and other clinical parameters, were obtained from the medical records.

16S rRNA gene sequence analysis

For 16S rRNA gene sequence analysis, 32 cervical mucus specimens from EMs patients (generation dataset: n = 22; validation dataset: n = 10) and 30 from healthy controls (generation dataset: n = 20; validation dataset: n = 10) were analyzed, as detailed in Table S2. The V4 and V5 hypervariable regions of the bacterial 16S rRNA gene were sequenced using the Illumina MiSeq platform. Raw sequence data were demultiplexed, quality-filtered, and merged to generate clean reads. Operational taxonomic units (OTUs) were clustered at a 97% similarity threshold. Representative sequences were selected for taxonomic annotation to determine species identity and abundance. OTUs abundance, alpha diversity (Shannon and Chao1 indices), and taxonomic composition were assessed to characterize species richness, evenness, and group-specific OTUs. Beta diversity was analyzed using non-metric multidimensional scaling analysis (NMDS) to explore the differences in microbial community structure among different samples or groups. To further investigate microbial differences between the EMs and control groups, multiple statistical and machine learning methods were employed—including EdgR, DESeq2, zero-inflated negative binomial model (ZINB), negative binomial model (NEGBIN), compound Poisson lognormal model (CPLM), Random Forest, and linear discriminant analysis (LDA) effect size (LEfSe). Functional predictions were performed using PICRUST [63], Tax4Fun [64], and FAPROTAX [65].

DNA isolation and qRT-PCR

Cervical mucus specimens were collected from patients with EMs (n = 12) and healthy controls (n = 12). Genomic DNA was extracted using the Bacteria DNA Kit (catalog number: DP118-02, TIANGEN BIOTECH, Beijing) according to the manufacturer’s protocols. Quantitative reverse transcription polymerase chain reaction (qRT–PCR) was performed on a real-time PCR system (ABI7500, Bio-Rad, USA) using ChamQ Universal SYBR qPCR Master Mix (catalog number: Q711-03, Vazyme, Nanjing, China). Each 20 μL reaction included 10 μL of 2 × SYBR Green PCR Master Mix, 1 μL of each primer, and 8 μL of template DNA, with technical duplicates for each sample. The thermal cycling conditions were as follows: initial denaturation at 95 °C for 20 s, followed by 40 cycles of 95 °C for 3 s and 60 °C for 30 s. The S. agalactiae reference strain (ATCC 13813, American Type Culture Collection) served as a positive control, whereas nuclease-free water was used as a negative control. Primers (forward: 5′-AGGAAACCTGCCATTTGCG-3′; Reverse: 5′-CAATCTATTTCTAGATCGTGG-3′) were synthesized by Guangzhou IGE Biotechnology Ltd. For standard curve generation, serial tenfold dilutions of S. agalactiae genomic DNA (107 to 101 copies/μL; BTN14-18,810) were amplified to create a six-point calibration curve. The following quantification formula was used: Cт = a[log(Q)] + b, where Cт is the quantification cycle value, a is the standard curve slope (− 2.364), b is the y-intercept (16.113), and Q is the copy number of S. agalactiae. Bacterial load in samples was calculated using the experimentally derived Cт values and the standard curve equation.

Animal experiments

All animal procedures were performed in accordance with the Guidelines for the Ethical Review of Laboratory Animal Welfare (GB/T35892-2018) and approved by the Experimental Animal Ethics Committee of the Fifth Affiliated Hospital of Sun Yat-sen University. A total of 93 8-week-old nulliparous female C57BL/6 mice were purchased from Beijing Charles River Experimental Animal Technical Co., Ltd. (Beijing, China). Of these, 31 were randomly selected as uterine fragment donors, whereas the remaining 62 served as recipients. All animals were housed under standardized conditions at a constant temperature of 26 °C, a 14/10-h light/dark cycle, and free access to food and water.

We established a mouse model of EMs via the intraperitoneal (i.p.) injection of endometrial fragments [66]. After 1 week of acclimatization, each donor mouse received an intramuscular injection of 2 μg/mouse estradiol benzoate (Sigma) to stimulate endometrial proliferation. One week later, the mice were euthanized, and their uteri were harvested. The uterine tissues were placed in a Petri dish containing warm sterile saline and bisected longitudinally with scissors. Both uterine horns from each mouse were minced with scissors into fragments < 1 mm in diameter. These fragments were then injected intraperitoneally into the recipient mice. To minimize potential bias, endometrial tissue fragments from 31 donor mice were pooled, evenly divided into 62 sections, and intraperitoneally into individual recipient mice. This approach reduced inter-donor variability. EMs-like lesions typically form within 3 days following uterine tissue induction [66]; therefore, the control group received 0.1 mL of PBS. To investigate the role of S. agalactiae in lesion progression, 1.0 × 105 colony-forming units (CFU)/mL of S. agalactiae were intraperitoneally injected into the EMs group on days 4, 5, and 6. From days 9 to 15, the mice were treated daily with teicoplanin (Sanofi), a glycopeptide antibiotic commonly used against Gram-positive bacterial infections, particularly Staphylococcus and Streptococcus species [67].

To further investigate whether S. agalactiae promotes EMs development via L-carnitine, we established three additional experimental groups in addition to the control and S. agalactiae groups. The first E. coli group received 1.0 × 105 CFU/mL of Escherichia coli (E. coli) on days 4, 5, and 6 via i.p. injection; the second L-carnitine group received 50 mg/kg L-carnitine (TargetMol, catalog number: T2203) from days 4 to 15 via i.p. injection; and the third S. agalactiae + Mildronate (S. agalactiae-challenged mice) group received a daily i.p. injection of 50 mg/kg Mildronate (TargetMol, catalog number:T6586), an inhibitor of L-carnitine synthesis, from days 4 to 15.

Three weeks after the i.p. injection, the recipient mice were euthanized. The peritoneal cavities were photographed, and ectopic lesions were carefully excised from the surrounding tissue. Based on surface vascularization and color, the ectopic lesions were categorized as either red or white. The ectopic lesions were subsequently fixed in 4% normal buffered formaldehyde for histological evaluation or stored in liquid nitrogen for subsequent experiments. The total injected volume is consistent with the guidelines outlined in the National Institutes of Health Guide for the Care and Use of Laboratory Animals, which recommends an intraperitoneal injection volume of 5–10 mL/kg body weight in mice, equivalent to 125–250 μL for a 25 g mouse.

Cell lines and cell culture

The hEM15A cell line, a transformed eutopic endometrium stromal cell line [68], was kindly provided by Professor Xiaohong Chang of the Peking University People’s Hospital. The cells were cultured at 37 °C in 5% CO2 using minimum essential medium (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 15% fetal bovine serum (FBS) (Gibco, Thermo Fisher Scientific, Waltham, MA, USA). Human umbilical vein endothelial cells (HUVECs) were purchased from ScienCell (catalog number: 8000) and cultured in Endothelial Cell Medium (ECM) (ScienCell, catalog number: 1001) supplemented with 10% of FBS. For the S. agalactiae metabolic supernatant assay, hEM15A cells and HUVECs were co-cultured using conditioned media. S. agalactiae (ATCC 13813) and S. anginosus (ATCC 33397) were cultured at 1 × 105 CFU/mL in brain heart infusion (BHI) broth (Genbank Biosciences Inc., China), whereas E. coli (ATCC 11775) was cultured at the same concentration in the Luria–Bertani broth (Beyotime Biotechnology, Shanghai, China). S. agalactiae at 1 × 105 CFU/mL was heat-inactivated by incubation at 75 °C for 30 min and subsequently cultured in BHI broth. After 12 h of incubation, the supernatants of live S. agalactiae and E. coli and heat-inactivated S. agalactiae were filtered twice using a 0.22-μm membrane filter. The co-cultured medium was prepared by mixing the filtered bacterial supernatant with the cell culture medium at a 1:1 ratio.

Subsequently, 5 × 105 of hEM15A cells or HUVECs were cultured for 24 h in the co-culture media. The cells were divided into six groups: control (PBS), incubated with cultural supernatant of E. coli, incubated with culture supernatant of heat-inactivated S. agalactiae, incubated with cultural supernatant of S. agalactiae, incubated with L-carnitine (TargetMol, T2203, 40 µM), and incubated with culture supernatant of S. agalactiae and Mildronate (TargetMol, T6586, 40 µM).

Migration/invasion assay

The migration and invasion abilities of hEM15A cells and HUVECs were evaluated using a modified Boyden chamber migration assay [69]. Briefly, Transwell chambers (Corning, Corning, NY, USA) with 8-μm pore inserts were used. Next, cells were starved for 12 h in a serum-free medium. For both assays, the cells were harvested, resuspended in serum-free medium at a concentration of 1 × 105 cells/100 μL, and seeded into the upper chambers of Transwell inserts (Corning, USA) either pre-coated with 100 μL of 1:8-diluted Matrigel (BD Biosciences) (for invasion assays) or left uncoated (for migration assays). The medium supplemented with 10% FBS was added to the lower chamber as a chemoattractant. After 20 h of incubation, the Transwell inserts were collected and fixed in 75% ethanol. Cells that migrated to the bottom of the top chamber were stained with crystal violet and counted in five randomly selected fields per insert under a light microscope (200 × magnification). Cell counts were analyzed using the ImageJ software (National Institutes of Health). Data obtained from three separate chambers were expressed as the mean values.

Cell proliferation analysis

Cell proliferation was monitored in real time using the IncuCyte S3 platform (Sartorius, Göttingen, Germany). Overall, 5000 hEM15A cells in 100 μL/well were seeded into 96-well plates and treated with the co-culture media prepared by mixing filtered bacterial supernatant and cell culture medium in a 1:1 ratio. Subsequently, the cells were imaged using a phase-contrast channel on an IncuCyte S3 platform (Sartorius, Göttingen, Germany). Phase-contrast images were captured from five distinct regions within each well every 12 h using a 10 × objective. Cell confluence was quantified as a percentage using the IncuCyte S3 image analysis software, which detects cell edges automatically.

Ultra-high performance liquid chromatography-tandem mass spectrometry analysis

For metabolite extraction, 1 mL of ice-cold methanol was added to 200 μL ectopic endometrial homogenate. The mixture was vortexed for 1 min, incubated at 4 °C for 10 min, and centrifuged at 16,000 g for 10 min at 4 °C. The supernatants were evaporated to dryness and reconstituted in 100 μL of 0.1% formic acid in 5% acetonitrile. A 5 μL aliquot of each sample was then injected and separated on an ACQUITY UPLC HSS T3 analytical column (2.1 × 150 mm, 1.8 μm, 100 Å, Waters) maintained at 35 °C using a Thermo Scientific Dionex UltiMate 3000 Rapid Separation LC system. The mobile phases consisted of 0.1% formic acid in water (a) and acetonitrile (b), with a flow rate of 0.3 mL/min. Data were acquired using an Orbitrap Fusion Lumos Tribrid mass spectrometer (Thermo Scientific, San Jose, CA, USA) equipped with a heated electrospray ionization source operating in positive ion mode with a spray voltage of + 3500 V. The ion transfer tube temperature, vaporized temperature, sheath gas flow, auxiliary gas flow, and sweep gas were maintained at 300 °C, 350 °C, 40 units, 15 arbitrary units, and 1 unit, respectively. Metabolite profiling was performed in a data-dependent acquisition mode over a mass range of 100–1000 m/z, with resolutions set at 60,000 for MS1 and 15,000 for MS/MS. The automatic gain control target and maximum injection time were configured at 5 × 104 and 50 ms, respectively. Finally, the metabolomic features were analyzed using Compound Discoverer (v3.1, Thermo Fisher Scientific).

L-carnitine colorimetric assay

The carnitine standard curve was constructed according to the manufacturer’s protocol (BioVision, K642-100). Briefly, ectopic lesions were homogenized in 100 µL of assay buffer and centrifuged at 13,000 g for 10 min to remove insoluble materials. Subsequently, 50 µL of the supernatant was directly diluted with assay buffer. The sample wells were adjusted to a final volume of 50 µL/well in a 96-well plate. Thereafter, 40 µL of reaction buffer—containing L-carnitine converting enzyme, development mix, substrate mix, and a probe—was added to each well. The reaction mixture was incubated in the dark for 30 min at room temperature. Finally, the optical density (OD) was measured at 570 nm, and the L-carnitine content was calculated using the standard curve.

Immunohistochemical staining

Immunohistochemical (IHC) staining was performed to assess the protein expression of CD34 and anti-Streptococcus Group B in EMs tissue. Briefly, formalin-fixed, paraffin-embedded tissue Sects. (4 µm) were blocked with 10% goat serum for 30 min at room temperature. The sections were then incubated with the primary antibody against CD34 (Invitrogen, catalog number: MA1-10,202, 1:500) or anti-Streptococcus Group B (Abcam, catalog number: ab53584, 1:200) for 15 h at 4 °C. After rinsing with PBS, the sections were incubated for 30 min at 37 °C with a horseradish peroxidase polymer conjugate (ZSGB-BIO). Hemangioma tissues and tissues from S. agalactiae-challenged mouse model were employed as positive controls for CD34 and Streptococcus Group B. PBS was used as a negative control in place of the primary antibody. Quantitative immunohistology analysis was performed by two independent pathologists (J.H. and Z.L.) using the ImageJ software (National Institutes of Health). The protein expression scores were calculated based on OD, yielding continuous values ranging from 0 to 300. Any discrepancies between observers were resolved by consensus.

Enzyme-linked immunosorbent assay

The VEGF concentrations in cell supernatants and tissue samples were measured using enzyme-linked immunosorbent assay (ELISA). For the supernatants of hEM15A cells or HUVECs, the VEGF concentration was quantified using Quantikine ELISA Human VEGF Immunoassay (catalog number: DVE00, R&D Systems) following the manufacturer’s instructions. For the mouse samples, the VEGF concentrations in tissue homogenates (prepared in cold PBS using an electric homogenizer) or serum were determined using the Mouse VEGF Simplestep ELISA Kit (catalog number: ab209882, Abcam) according to the manufacturer’s instructions. Ectopic lesions were rapidly frozen in liquid nitrogen, homogenized, and lysed with the extraction buffer provided in the kit (100 mg of wet tissue in 500 µL of cell extraction buffer PTR). Following centrifugation, the supernatants were analyzed using the Mouse VEGF ELISA Kit (Abcam, ab209882). The total VEGF content was calculated based on the standard curve and normalized to the tissue wet weight (g) to derive the VEGF concentration (pg/g tissue).

Tube formation assay

Briefly, HUVECs were serum-starved in ECM supplemented with 1% FBS for 6 h, harvested using trypsin, and resuspended in the same medium. Next, 2.5 × 104 of HUVECs were seeded into a 96-well plate pre-coated with growth factor-reduced Matrigel and incubated with E. coli, heat-inactivated S. agalactiae, S. agalactiae, L-carnitine, and S. agalactiae combined with the L-carnitine synthesis inhibitor Mildronate (4 h after S. agalactiae treatment). Tube formation was evaluated after 8 h, and images were captured from five random fields of view per well using the IncuCyte S3 system (ESSEN Bioscience, Sartorius, Germany). The total length of tubular segments was quantified using the Angiogenesis Analyzer plugin for Image J, available at the NIH website (https://imagej.nih.gov/ij/macros/toolsets/Angiogenesis%20Analyzer.txt).

Fluorescence in situ hybridization

FISH was performed on formalin-fixed paraffin-embedded endometrial and EMs specimens. Tissue sections were hybridized with a 5′ FAM-labeled S. agalactiae-specific probe, Saga [70]. The probe sequence (5′-GTAAACACC AAACMTCAGCG -3′) was obtained from probeBase (http://probebase.csb.univie.ac.at/). A scrambled probe (5′-CAATTGGGCCCGCTTTAAC CCAATCTC-3′) was used as the nonspecific negative control. Following deparaffinization and rehydration, sections were treated sequentially with 0.2 N HCl and proteinase K for 10 min each. After blocking with buffer at 55 °C for 2 h, the FAM-labeled probe (diluted 1:50 in 25% hybridization buffer and pre-heated at 88 °C for 3 min) was applied and hybridized overnight in a dark, humid chamber at 42 °C. Specimens were then washed in wash solution (20 mM Tris–HCl, pH: 7.2; 40 mM NaCl) and mounted with DAPI-antifade solution. Fluorescent images were acquired using a fluorescence microscope (Leica, DM6B).

Identification of differentially expressed mRNAs and gene set enrichment analysis

Differentially expressed mRNAs were identified, and Gene Set Enrichment Analysis (GSEA) was performed to compare endometrial tissue samples with a high abundance of S. agalactiae with those from uninfected controls. The abundance of S. agalactiae was determined by 16S rRNA sequencing, and tissue samples were collected from the Department of Pathology at the Fifth Affiliated Hospital, Sun Yat-sen University.

Total RNA was extracted using RNeasy Micro Kit (Qiagen, catalog number: 74,004), and paired-end libraries were prepared using the TruSeq RNA Sample Preparation Kit (Illumina, USA) following the TruSeq RNA Sample Preparation Guide. Gene expression data were obtained using the Illumina NovaSeq 6000 platform. Quality control and preprocessing of the raw sequencing data were performed using FastQC and TrimGalore. Gene-level expression counts were calculated utilizing the STAR aligner. Differential gene expression analysis was conducted using the DESeq2 package in R. In this study, differentially expressed genes related to EMs were defined by an absolute log fold change greater than 1 (|log2FC|> 1) and a P value of < 0.05. Furthermore, GSEA was performed to identify the key pathways associated with S. agalactiae infection in EMs.

Statistical analysis

Statistical analyses were performed using SPSS version 18.0, and results were expressed as the mean ± standard deviation. Student’s t-test was used for comparisons between two groups. For comparisons among multiple groups, the homogeneity of variance was first assessed; a one-way analysis of variance or non-parametric independent sample t-test (Mann–Whitney U test) was used when was confirmed. Pearson’s correlation analysis was performed to evaluate the correlation between protein expression levels. All tests were two-tailed, and a P value of < 0.05 was considered significant.