{"paper_id":"453befd7-4a0e-4289-99ad-b82de3792f61","body_text":"Endometriosis patient-derived small extracellular vesicles carry unique immune, proteomic and lipidomic \nsignatures associated with mild and severe endometriosis  \n \nFull author list: Jaelis P. Holmes1, Katherine B. Zutautas1, Danielle J. Sisnett1, Donya Hayati1, Olga Bougie 2, \nBruce A. Lessey3, and Chandrakant Tayade1 \n1Department of Biomedical and Molecular Sciences, Queen’s University, Kingston, Ontario, Canada, \nK7L 3N6 \n2Department of Obstetrics and Gynaecology, Kingston Health Sciences Centre, Kingston, Ontario, \nCanada, K7L 2V7 \n3Advocate Health, Wake Forest Baptist, Winston-Salem, NC, 27157, USA \n \nCorresponding author:  \nDr. Chandrakant Tayade1, DVM, PhD \nDepartment of Biomedical and Molecular Sciences \nQueen’s University, Kingston, ON, Canada, K7L 3N6 \nTelephone: 1-613-533-6354 \nFax: 1-613-533-2022 \nEmail: tayadec@queensu.ca \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nAbstract \nEndometriosis (EM) is a heterogeneous, gynecological inflammatory disease affecting over 200 million \nindividuals worldwide, yet the mechanisms underlying lesion establishment, progression, and recurrence remain \nincompletely understood. Small extracellular vesicles (sEVs) mediate intercellular communication through the \ntransfer of proteins, lipids, and nucleic acids reflective of their cellular origin; however, stage- and tissue-\nspecific sEV signatures remain poorly defined. Here, we characterized the molecular and functional landscape \nof EM-derived sEVs across disease stages and biological sources. sEVs isolated from eutopic endometrium, \nectopic lesions, peritoneal fluid, and plasma from mild- and severe-stage EM patients and healthy controls were \nanalyzed by surface marker profiling, proteomics, lipidomics, and integrated multi-omics, with functional \neffects assessed in human uterine microvascular endothelial cells. sEV composition varied by disease stage and \nsample type, with EM lesion-derived sEVs demonstrating stage-dependent loss of epithelial-associated markers \nand enrichment of immune-associated signatures, while EM plasma-derived sEVs exhibited altered adhesion- \nand platelet-associated profiles. Integrated multi-omics identified coordinated programs associated with \nimmune adaptation, extracellular matrix organization, epithelial remodeling, vascular signaling, oxidative \nstress, and metabolic adaptation. Functionally, sEVs derived from severe endometriotic lesions exhibited \nenhanced uptake and mitochondrial localization in endothelial cells and promoted angiogenic activity. Our \nfindings establish sEVs as dynamic mediators of EM disease progression and demonstrate that integrated sEV \nprofiling provides a framework for understanding EM heterogeneity and identifying candidate biomarkers and \ntherapeutic targets. \nKey words: Endometriosis, immune dysfunction, extracellular vesicles, molecular characterization, integrated \nmulti-omics analyses, biomarkers, angiogenesis \n \n1. Introduction \nEndometriosis (EM) is a chronic, estrogen-dependent inflammatory disease characterized by the \npresence of endometrium-like tissue outside the uterine cavity, primarily on the pelvic peritoneum and \novaries.1,2 Affecting approximately 6–10% of reproductive-age women worldwide, EM is associated with \nchronic pelvic pain, dysmenorrhea, dyspareunia, and infertility, substantially reducing quality of life and \nimposing a significant socioeconomic burden.1,2 Disease severity is clinically classified into four stages (I–IV) \naccording to the revised American Society for Reproductive Medicine (rASRM) staging system, with stages I–\nII generally considered mild disease and stages III–IV considered severe disease.1 However, symptom \npresentation and disease progression remain highly heterogeneous between patients, reflecting the complex and \nincompletely understood pathogenesis of EM. Increasing evidence implicates immune dysregulation, chronic \ninflammation, and altered cellular communication in lesion establishment and disease progression.3,4 Notably, \nmild and severe disease stages are thought to reflect distinct pathological states, with mild disease associated \nwith adhesion and survival within the peritoneal environment, and severe disease characterized by persistent \ninflammation, angiogenesis, and progressive tissue remodeling and fibrosis.3,4 Despite advances in diagnosis \nand clinical management, significant diagnostic delays and limited non-invasive biomarkers continue to \nchallenge effective disease detection and characterization, highlighting the need for improved understanding of \nEM-associated molecular and cellular mechanisms.2,5 \n \nExtracellular vesicles (EVs) have emerged as important mediators of intercellular communication in \nboth physiological and pathological conditions.6,7 EVs are lipid membrane-bound particles released by nearly \nall cell types and carry diverse bioactive cargo, including proteins, lipids, and nucleic acids, reflective of their \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\ncell of origin.6,7 They differ in size, biogenesis, and biological function and are classified into distinct subtypes, \nincluding apoptotic bodies, microvesicles, and small extracellular vesicles (sEVs). Among these, sEVs are \nabundant in the uterine microenvironment and are increasingly recognized for their roles in immune \nmodulation, angiogenesis, proliferation, migration, and inflammatory signaling.6–11 These small vesicles, \ntypically ranging from 50–250 nm, encompass populations commonly associated with exosomes and small \nmicrovesicles.6,7 Through receptor-ligand interactions, membrane transfer, and delivery of bioactive cargo, \nsEVs can alter cellular phenotype and function both locally and systemically.7 Importantly, sEVs are emerging \nas carriers of disease-associated molecular signatures, making them attractive candidates for investigating \nmechanisms of disease progression and identifying minimally invasive biomarkers. \n \nIn the context of EM, sEVs exhibit disease-associated alterations to their molecular cargo, including \ndifferential expression of miRNAs, lncRNAs, proteins, and lipids relative to healthy controls.12–14 Among the \nmost extensively studied sEV cargo are miRNAs, including members of the let-7 family, miR-30d-5p, miR-23a, \nmiR-143, and miR-320a, which have been implicated in inflammatory, proliferative, and angiogenic signaling \npathways in EM.13–17 Similar miRNA alterations have also been detected in the plasma of EM patients.13,15 \nComplimenting these observations, recent work using bead-based flow cytometric profiling of sEV surface \nepitopes in a patient-derived endometrial epithelial organoid model demonstrated stage-dependent remodeling \nof sEV phenotypes, with severe-stage EM-derived sEVs exhibiting increased immune-associated surface \nmarkers and reduced stem cell-associated markers relative to control and mild-stage disease.18 Together, these \nfindings support a role for sEVs in EM pathophysiology. However, most studies have focused on individual \nsEV cargo classes or isolated biofluids, leaving the extent to which sEV composition varies across EM disease \nstages and matched tissue compartments poorly understood. Moreover, whether these molecular signatures \nreflect coordinated alterations across the local lesion microenvironment and systemic circulation remains \nunclear. \n \nBuilding on this gap, we aimed to define stage- and tissue-specific sEV signatures in EM through \nintegrated surface marker, proteomic, and importantly lipidomic profiling of sEVs isolated from eutopic \nendometrium, ectopic lesions, plasma, and peritoneal fluid. Integrating lipidomic with proteomic profiling \nenabled characterization of coordinated molecular remodeling across complementary sEV cargo classes, \nproviding insight into immune, metabolic, extracellular matrix (ECM), and vascular remodeling signatures \nassociated with EM progression. Surface marker profiling revealed stage-dependent shifts in immune-, \nadhesive-, and stemness-associated sEV phenotypes, while proteomic and lipidomic analyses uncovered distinct \nmolecular programs linked to lesion establishment and severe disease remodeling. As angiogenesis and vascular \nremodeling are essential features of EM lesion establishment and persistence, and were among the dominant \npathways identified through our integrated analyses, we further evaluated whether stage-specific lesion-derived \nsEVs differentially interact with endometrial endothelial cells by assessing sEV uptake, cytokine secretion, and \nangiogenic capacity. Our findings provide insight into stage- and tissue-dependent molecular signatures and \nfunctional activity of sEVs in EM, supporting their potential role(s) in disease progression and their utility as \nminimally invasive indicators of EM-associated pathophysiology. \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n2. Methods \n \n2.1 Study approval and ethics.  \nThe Queen’s University Health Sciences and Affiliated Teaching Hospitals Research Ethics Board \napproved all methodologies used in this study (OBGY-229-11 and ANAT-029-09). All participants recruited \nfrom gynecology clinics at Kingston Health Sciences Centre (KHSC) provided written informed consent prior \nto sample collection. Samples used for lipidomic analyses were obtained from an independent cohort collected \nthrough Greenville Hospital System (Greenville, South Carolina, USA) and Kingston General Hospital (KGH), \nfollowing institutional research ethics approval and written informed consent from all participants. \nThe primary KHSC cohort was used for extracellular vesicle characterization and functional analyses \nconsisting of matched samples from mild-stage (n=5) and severe-stage EM patients (n=7), including ectopic \nlesions, eutopic endometrium, peritoneal fluid (PF), and plasma, as well as healthy control plasma samples \n(n=8).19 An independent cohort was utilized for lipidomic profiling (Table 1), consisting of EM patient plasma \nsamples (mild, n=8; severe, n=6), eutopic endometrium samples (n=4), and ectopic lesion samples (mild, n=4; \nsevere, n=4) collected from Greenville Hospital System, with healthy control plasma samples (n=6) obtained \nthrough KGH. Across both cohorts, participants met eligibility criteria of being 28–50 years of age, having a \nuterus and at least one ovary, and undergoing scheduled excision surgery within the study period. Individuals \nwith EM received a confirmed diagnosis of EM by histological evidence, suspicious clinical history, and/or \nradiographic evidence consistent with EM, prior to surgery. Inclusion and exclusion criteria, sample processing \nprocedures, and storage conditions were consistent between cohorts. \n2.2 Sample collection from EM patients and control women.  \nFor proteomic profiling and sEV characterization cohort, matched human eutopic endometrium and \nendometriotic lesion biopsies, and PF samples were collected from patients undergoing laparoscopic EM \nexcision surgery. Histopathological analysis of excised lesions was performed to confirm EM diagnosis. \nPeripheral blood was collected prior to surgery. For the independent lipidomic cohort, eutopic endometrial \nsamples were obtained by Pipelle biopsy during surgery, while ectopic lesions and PF were collected during \nlaparoscopic excision procedures. All samples were processed and stored according to standardized procedures, \nwith tissue samples snap-frozen in liquid nitrogen and stored at −80°C until further analysis. PF samples were \nsimilarly stored at −80°C, and plasma was isolated from peripheral blood as described below. EM stage was \ndetermined intraoperatively by the attending surgeon according to the revised American Society for \nReproductive Medicine (rASRM) classification criteria.1 Clinical and demographic characteristics of matched \nEM patient cohort used for sEV characterization, proteomic profiling, and functional analyses, including age, \npathology findings, co-existing pathologies, and disease stage (I–IV), were previously reported by our group. 19 \nControl participants (plasma only) had no clinical indicators of EM or other gynecological conditions, including \ninfertility, pelvic inflammatory disease, or chronic pelvic pain. Clinical characteristics of independent cohort \nused for lipidomic analyses followed similar inclusion criteria (Table 1). \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nTable 1. Clinical characteristics of independent cohort used for lipidomic analyses \nParameters EM (n = 19) Controls (n = 6) \nAge ± SD 28.5 ± 5.7 30.3 ± 6.7 \n     Stage of EM, n (%) \nI 4 (21.05)  \nII 5 (26.32) \nIII 7 (36.84) \nIV 3 (15.79) \n           Sample Type, n \nEutopic endometrium (EU) 4 n/a \nEctopic lesion (EMS) 8 n/a \nPlasma 14 6  \nMatched 7 n/a \n \n \n2.3 Plasma isolation and PF collection \nPeripheral blood samples from control and EM patients were diluted 1:1 with 2% FACS buffer (PBS \nsupplemented with 2% FBS) prior to processing using Lymphoprep density gradient medium (18060; \nSTEMCELL Technologies) and SepMate™-50 tubes (85450; STEMCELL Technologies), according to the \nmanufacturer’s instructions. Following density gradient centrifugation, the plasma fraction was carefully \naspirated and immediately stored at −80°C until further use. PF was processed as per our previous publication.19 \n \n2.4 Tissue protein extract and quantification \nApproximately 50 mg of frozen eutopic endometrium (EU)- or endometriotic lesion (EMS)-derived \ntissue was obtained from each sample while maintaining samples on dry ice to prevent thawing. Tissue samples \nwere transferred into PowerBead tubes (13112-50; Qiagen) preloaded with T-PER™ Tissue Protein Extraction \nReagent (78510; Thermo Fisher Scientific) supplemented with protease inhibitor cocktail (535140-1ML; \nSigma-Aldrich) at a 1:100 ratio. Samples were then homogenized using an Omni Bead Ruptor 24 (Omni \nInternational) for two cycles of 20 s at 5.5 m/s with a 20 s interval between cycles. Homogenization parameters \nwere adjusted as required based on tissue size and fibrous composition. Following homogenization, samples \nwere centrifuged at 10,000 × g for 5 min at 4°C to remove insoluble debris. The resulting protein-containing \nsupernatant was collected on ice prior to protein quantification and storage at −80°C until further analysis. \nBriefly, protein concentration was determined using a bicinchoninic acid (BCA) assay (23227; Thermo Fisher \nScientific), as per manufacturer’s guidelines. \n \n \n \n \n \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n \n2.5 Isolation of EVs, including sEVs, from patient plasma, tissues, and PF.   \nsEVs were isolated from plasma, PF, and tissue samples obtained from EM patients and healthy controls \nusing size exclusion chromatography (SEC). IZON qEV columns (IC10-35 for plasma and PF samples; ICO-35 \nfor tissue samples; IZON Science, Christchurch, New Zealand) were used according to the manufacturer’s \ninstructions. Fractions 7–9, corresponding to sEV-enriched fractions, were collected for downstream analyses. \nAll EV isolation procedures were performed in accordance with the MISEV2025 guidelines.20 Samples \nunderwent additional filtration through a 0.22-μm filter to remove potential contaminants. The quality of sEVs \nisolated using SEC was compared with ultracentrifugation-based isolation methods and was determined to be \nsuperior with respect to purity and yield. Isolated sEV pellets were resuspended in 150 μL of PBS (10010023, \nThermo Fisher Scientific) and either used immediately for downstream applications or stored at –80°C until \nfurther use. sEV samples were subjected to no more than 2 freeze-thaw cycles. The isolated sEV preparations \ndescribed above, derived from matched patient eutopic endometrium, ectopic lesions, plasma, and PF samples \nfrom individuals with mild (n=5) and severe (n=7) EM, as well as plasma-derived sEVs from healthy controls \n(n=8), were used for downstream characterization, MACSPlex surface marker analysis, and proteomic profiling.  \n \n2.6 Transmission electron microscopy analysis of sEVs.  \nIsolated sEVs were characterized by transmission electron microscopy (TEM). sEVs resuspended in \n150μL PBS were fixed in 2.5% glutaraldehyde for 5 min and negatively stained with UranyLess (22409; \nElectron Microscopy Sciences) for 2 min. Samples were transferred onto 200-mesh Formvar-coated copper \ngrids and incubated for 10 min. Grids were analyzed using a Talos F200i transmission electron microscope \noperated at 200 keV (Thermo Fisher Scientific) by trained electron microscopy specialists at the Queen’s \nUniversity Cardiopulmonary Unit. Detected sEVs were subjected to morphometric analysis. \n \n2.7 Nanoparticle tracking analysis of sEVs.  \nIsolated sEVs were analyzed using a ZetaView nanoparticle tracking analyzer (Particle Metrix). \nInstrument alignment was performed using 100 nm reference standard polystyrene beads (3100A; Thermo \nFisher Scientific). sEV samples were diluted 1:100 in PBS and analyzed at a sensitivity setting of 70–80 across \n11 predefined camera positions. According to manufacturer guidelines, samples with fewer than 8/11 valid \nposition reads or fewer than 500 tracked particles were excluded from analysis. Particles measuring <30 nm \nwere considered background artifacts and excluded from quantification. \n \n2.8 Detection and characterization of sEVs with MACsPlex analysis.  \nIsolated sEVs were characterized using the MACSPlex Human EV Kit (130-108-813; Miltenyi Biotec). \nPositive marker expression was defined as fluorescence intensity exceeding the corresponding isotype control \nthreshold. For each sample, 15 μg of EV-associated protein, quantified using a BCA protein assay (A65453; \nThermo Fisher Scientific), was diluted in 120 μL MACSPlex buffer according to the manufacturer’s \nrecommendations. Samples were acquired using a CytoFLEX S flow cytometer (Beckman Coulter) and \nanalyzed using FlowJo™ software (BD Life Sciences, v10). This bead-based flow cytometry platform evaluates \n37 EV surface markers alongside two isotype controls. Relative marker expression was calculated as \nnormalized median fluorescence intensity (nMFI). Plasma-derived sEVs were normalized to the average signal \nof CD63 and CD81, whereas PF- and tissue-derived sEVs were normalized to the average signal of CD9, CD63, \nand CD81. The selection of normalization markers was based on consistent tetraspanin detection patterns within \neach sample matrix. \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n2.9 sEV protein isolation.  \nTotal protein was extracted from isolated sEVs, using a Total Exosome RNA & Protein Isolation Kit \n(4478545; Thermo Fisher Scientific) according to the manufacturer’s instructions. Protein concentration was \nquantified using a BCA protein assay (A65453; Thermo Fisher Scientific). Protein fractions were immediately \nstored at –80°C until downstream analyses.  \n \n2.10 Proteomic profiling of sEVs by mass spectrometry.  \nProteomic profiling was performed by the Proteomics Platform at the Research Institute of the McGill \nUniversity Health Centre (RI-MUHC; Montreal, QC, Canada). Isolated sEV proteins were desalted by dialysis \nagainst 10 mM ammonium bicarbonate and subsequently digested with trypsin (20 μg) at 37°C for 18 h. \nResulting peptide mixtures were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS) \nusing an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific). Sample preparation, data acquisition, \nand initial processing were completed by the RI-MUHC Proteomics Platform according to standard facility \nmethodologies. MS/MS spectra were acquired in positive ion mode and processed for downstream proteomic \nanalysis. MS/MS-derived proteomic datasets were analyzed using DIA-NN and Spectronaut databases (v0.11) \nthrough the Analyst Suites platform.21 Protein abundance values were normalized using variance-stabilizing \nnormalization (VSN), and missing values were not imputed. Differentially expressed proteins were identified \nusing predefined statistical thresholds, including a false discovery rate (FDR) of 1%, adjusted p-value < 0.05, \nand a fold-change cutoff of |log2FC| ≥ 0.5. Functional enrichment and pathway analyses were performed using \nIntegrated Pathway Analyst software, while obaDIA was used for protein functional annotation and Gene \nOntology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. \n2.11 Lipidomic profiling of sEVs by mass spectrometry.  \nLipidomic profiling was performed by Creative Proteomics (Shirley, New York, USA). sEVs isolated \nfrom matched patient eutopic endometrium (n=4), ectopic lesions from patients with mild (n = 4) and severe (n \n= 4) EM, alongside plasma from mild (n=8) and severe (n=6) patients, and healthy controls (n =6), were \nseparated into 50 μL aliquots, followed by the addition of 1.5 mL chloroform:methanol (2:1, v/v) and 0.5 mL \nultrapure water. Samples were vortexed and centrifuged to induce phase separation. The lower organic phase \nwas carefully collected and dried under nitrogen gas. Dried lipid extracts were resuspended in \nisopropanol:methanol (1:1, v/v), and 5 μL lysophosphatidylcholine (LPC )(12:0) internal standard was added \nprior to analysis. Samples were subsequently centrifuged at 12,000 rpm for 10 min at 4°C and the resulting \nsupernatant was collected for LC-MS analysis, which was performed using an ACQUITY UPLC system \ncoupled to a Q Exactive mass spectrometer (Thermo Fisher Scientific). Chromatographic separation was \nachieved using an ACQUITY UPLC BEH C18 column (100 × 2.1 mm, 1.7 μm particle size; Waters). The \nmobile phase consisted of solvent A [60% acetonitrile (ACN), 40% H₂O, and 10 mM ammonium formate] and \nsolvent B [10% ACN, 90% isopropanol, and 10 mM ammonium formate]. Gradient elution was performed as \nfollows: 0–1 min, 30% B; 1–10.5 min, 30–100% B; 10.5–12.5 min, 100% B; 12.5–12.51 min, 100–30% B; and \n12.51–16 min, 30% B. The mobile phase flow rate was maintained at 0.3 mL/min. Mass spectrometry data were \nacquired in both positive and negative electrospray ionization modes using optimized instrument parameters. \n \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n2.12 Integration of proteomic and lipidomic profiling of sEVs.  \nRaw proteomic and lipidomic data files generated from the analyses described above were uploaded to \nthe publicly available OmicsAnalyst platform for multi-omics integration.22,23 Datasets were formatted \naccording to platform requirements and subjected to feature scaling and normalization within the OmicsAnalyst \nworkflow. For tissue-derived sEV analyses, eutopic endometrium samples were designated as the reference \ngroup, whereas healthy control plasma samples served as the reference for plasma-derived sEV analyses. FDR \ncorrection was applied throughout the OmicsAnalyst workflow. Multiple Co-Inertia Analysis (MCIA) was \nperformed to integrate proteomic and lipidomic datasets. Integrated features were further analyzed by Reactome \npathway enrichment and causal discovery analyses to identify coordinated molecular relationships. Hierarchical \nclustering heatmaps were generated using Spectrum clustering, and integrated protein-lipid correlation networks \nwere annotated using the STRING database. \n \n2.13 Cell culture.  \nFor confocal microscopy-based sEV internalization and subcellular localization analyses, pooled sEVs \nisolated from ectopic lesions from individuals with mild (n=3) and severe (n=3) EM were used. For in vitro \nfunctional assays, sEVs isolated from matched patient ectopic lesions and eutopic endometrium from \nindividuals with mild (n=5) and severe (n=7) EM were used as experimental sEV preparations. The same \npooled sEV preparations were maintained across respective experimental assays to ensure consistency between \nexperiments. Human uterine microvascular endothelial cells (HUtMEC; C-12295; PromoCell) were cultured \naccording to manufacturer-provided protocols in Endothelial Cell Growth Medium MV (C-22020; PromoCell), \na low-serum (5% v/v) medium supplemented with fetal calf serum (0.05 ml/ml), endothelial cell growth \nsupplement (0.004 ml/ml), recombinant human epidermal growth factor (10 ng/ml), heparin (90 µg/ml), and \nhydrocortisone (1 µg/ml). Only passages 4-6 were used in experiments. Cellular morphology and proliferative \ncharacteristics were routinely monitored throughout passaging to ensure maintenance of phenotype and cellular \nintegrity. \n \n2.14 Point-scanning confocal microscopy analysis of sEV uptake and subcellular localization \nsEVs were fluorescently labeled with 2 μM MemGlow™ 488 (MG01; Cytoskeleton Inc.) at 37°C for 10 \nmin according to the manufacturer’s protocols Following staining, sEVs were washed using 100 kDa Amicon \nUltra centrifugal filters (UFC5100; Sigma-Aldrich) to remove excess dye. HUtMEC cells were seeded at a \ndensity of 5 × 10⁴ cells/well in μ-Plate 24-well glass-bottom plates (82427; ibidi) and cultured for 24 h prior to \ntreatment. For assessment of sEV uptake, cells were stained with 1 µg/mL Hoechst nuclear dye (62249; Thermo \nFisher Scientific) and 1 μM CellTrace™ BODIPY® TR methyl ester cytoplasmic dye (C34556; Thermo Fisher \nScientific) at 37°C for 30 min according to the manufacturer’s protocols. For assessment of sEV subcellular \nlocalization, cells were stained with 1 µg/mL Hoechst nuclear dye (62249; Thermo Fisher Scientific) and 1 μM \nMitoTracker™ Deep Red FM mitochondrial dye (M22426; Thermo Fisher Scientific) under the same \nconditions. Fluorescently labeled sEVs were resuspended in respective cell culture media and added to \nHUtMECs at a dose of 1 × 10⁴ sEVs/cell for 10 h. \nLive-cell imaging was performed using a MICA point-scanning confocal microscope (Leica \nMicrosystems) maintained at 5% CO₂, 60% humidity, and 37°C. Image acquisition was initiated immediately \nfollowing sEV treatment and continued at 2 h intervals over the 10 h incubation period. Images were acquired at \n60× magnification using a water-immersion objective lens and processed using LAS X Analyst Suite (Leica \nMicrosystems) and ImageJ software (NIH). \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n2.15 Proliferation and apoptosis assays. \nHUtMEC cells were seeded in 96-well plates (163320; Thermo Fisher Scientific) and cultured for 24 h \nprior to sEV treatment. sEVs were resuspended in endothelial cell culture media and added to cells at a dose of \n1 × 10⁴ sEVs/cell for 24 h. To assess cell proliferation and viability, cells were incubated with 10 μL WST-1 \nreagent (5015944001; Sigma-Millipore) for 2 h according to the manufacturer’s instructions. Cellular metabolic \nactivity was quantified by measuring absorbance at 450 nm with a 650 nm reference wavelength using a \nSpectraMax iD3 plate reader (Molecular Devices). To evaluate apoptosis, cells were treated with 100 μL \nCaspase-Glo® 3/7 reagent (G8091; Promega) for 3 h protected from light, and caspase-3/7 activity was \nquantified by luminescence using the SpectraMax iD3 platform according to the manufacturer’s instructions. \nAll conditions were analyzed using 10–15 technical replicates. \n \n2.16 Multiplex cytokine analysis.  \nIn parallel with in vitro functional assays, 100 μL of conditioned media was collected from HUtMEC \ncells 12 h following treatment with pooled sEVs. Cell-only controls were included. Inflammatory cytokine and \nchemokine profiling were performed using a commercially available Human Cytokine/Chemokine Panel A 48-\nPlex Discovery Assay® Array (Eve Technologies; HD48A) on the Luminex xMAP platform (Bio-Rad). All \nconditions were analyzed in three technical replicates. \n \n2.17 Endothelial tube formation assay.  \nEndothelial tube formation assays were performed using 15 well μ-Slide plates (81506; ibidi) according \nto the manufacturer’s instructions. Growth factor-reduced, phenol red-free Matrigel (356221; Corning) was \ndispensed into each well and allowed to polymerize at 37°C for 45 min. HUtMECs were harvested using 0.25% \ntrypsin-EDTA, seeded onto the polymerized Matrigel at a density of 1 × 10⁴ cells/well, and treated with sEVs at \na dose of 1 × 10⁴ sEVs/cell. Controls included PBS vehicle and recombinant human VEGF-A (25 ng/mL; \nMA116629; Thermo Fisher Scientific). Cells were incubated at 37°C and imaged at 12 h using a MICA \nwidefield imaging system (Leica Microsystems). Two images were acquired per well from standardized regions \nalong the well midline to ensure consistent image sampling across conditions. All conditions were analyzed in \nthree technical replicates. Tube formation parameters, including number of tubes, total tube length, and \nbranching points, were analyzed using the automated online platform WimTube (Wimasis GmbH, Munich, \nGermany). \n2.18 Statistics.  \nStatistical analyses were performed using GraphPad Prism software (v11). Data are presented as mean ± \nstandard deviation (SD). Normality and homogeneity of variance were assessed prior to statistical testing. \nOutliers were identified using the ROUT method (Q = 1%) in GraphPad Prism; no outliers were removed unless \nindicated in respective figure captions. Statistical comparisons were performed using one-way analysis of \nvariance (one-way ANOVA), two-way analysis of variance (two-way ANOVA), or repeated-measures one-way \nANOVA, as appropriate based on experimental design, followed by Tukey’s multiple comparisons test for post \nhoc analysis. For experiments involving repeated measurements, repeated-measures analyses were applied. A p-\nvalue < 0.05 was considered statistically significant. \n \n \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n \n3. Results \n \n3.1 EM-derived sEVs display tissue- and stage-specific biomolecular and surface phenotypic profiles  \nTo elucidate stage- and tissue-specific alterations in sEVs in EM, we characterized sEV-enriched \nfractions from matched EM patient samples, and plasma samples from disease-free controls for comparison. \nsEVs were isolated by SEC across all sample types and disease stages (Fig. 1a-n). NTA revealed no significant \ndifferences in median particle size between biological compartments, with plasma-sEVs exhibiting consistent \nparticle size distributions across systemic samples and PF-, EU-, and EMS-sEVs demonstrating comparable size \nprofiles across local lesion-associated compartments (Fig. S1a, b). In contrast, particle concentration varied \nsignificantly by both sample type and disease stage. Plasma from patients with severe EM exhibited \nsignificantly higher particle concentrations compared with mild EM and healthy controls (Fig. 1a), indicating \nincreased circulating vesicle abundance in severe disease. EM patient tissue-derived sEV preparations from \nEMS and EU exhibited significantly higher particle concentrations than PF-derived sEV preparations (Fig. 1b), \nsuggesting tissue compartment-specific differences in sEV abundance within the EM microenvironment. TEM \nanalysis confirmed the presence of vesicles consistent with standard sEV morphology across all sample types, \nwith minimal background contamination (Fig. 1c-h). While all groups exhibited canonical vesicle morphology, \nqualitative differences in structural heterogeneity were observed between compartments and disease stages. \nSpecifically, severe-stage PF- and EMS-derived sEV preparations demonstrated increased morphological \nvariability relative to other sample groups, with some vesicles exhibiting visible intravesicular electron-dense \nstructures, suggesting differences in sEV ultrastructural features associated with disease stage (Fig. 1c, f). \n \nTo further characterize sEV phenotype, canonical tetraspanin markers CD9, CD63, and CD81 were \nassessed using the MACSPlex EV Kit, a bead-based flow cytometry platform (Fig. 1i-k). Tetraspanin \nexpression was detected across all sample groups; however, distinct tissue- and stage-associated differences \nwere observed. Namely, CD9 expression was significantly increased in plasma-sEVs from severe EM patients \ncompared with mild EM and control groups (Fig. 1i), whereas no significant differences in tetraspanin \nabundance were detected among PF-sEVs (Fig. 1j). In contrast, CD63 and CD81 abundance were increased in \nsevere-stage tissue-sEVs from both EU and EMS compared with mild-stage samples (Fig. 1k), suggesting \nstage-associated remodeling of tissue-sEV populations. \n \nBeyond canonical tetraspanin markers, the MACSPlex analysis was used to assess an additional 37 EV-\nassociated surface epitopes across matched patient samples and control plasma (Fig. 1l-q). To visualize broader \npatterns of sEV surface composition across biological compartments and disease stages, detected markers were \ngrouped into biological categories, including adhesion-associated, immune-associated, antigen-presentation, \nepithelial-associated, stemness-associated, and platelet-associated markers (Fig. 1l). Source-specific heatmap \nanalyses further illustrated compartment- and disease-associated variation in surface marker expression across \nplasma-, PF-, and tissue-derived sEVs (Fig. S1c-e). Together, these analyses demonstrated distinct \ncompartment-specific patterns of sEV surface composition across sample sources. \n \nPrincipal component analysis (PCA) demonstrated partial separation of sEV surface marker profiles \naccording to disease stage in plasma-sEVs (Fig. 1m) and EU- and EMS-sEVs from mild- and severe-stage EM \npatients (Fig. 1n), indicating disease-associated variation in overall sEV surface composition. At the individual \nmarker level, severe-stage plasma-sEVs demonstrated significantly elevated expression of adhesion- and \nplatelet-associated markers (CD29, CD41b, CD42a, and CD62P) relative to healthy controls, consistent with \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\naltered vascular- and platelet-associated signaling in severe disease (Fig. 1o). PF-sEVs from mild-stage disease \nexhibited increased expression of stemness-associated (CD133/1) and immune-modulatory (CD24) compared \nwith severe-stage disease (Fig. 1p). In tissue-sEVs, CD133/1 and CD326/EPCAM expression was reduced in \nEMS compared with matched EU, indicating differences in stemness- and epithelial-associated sEV surface \nprofiles (Fig. 1q). Moreover, when comparing expression across disease stages, HLA-II expression was \nincreased in severe-stage tissues, which is consistent with enhanced immune activation in severe lesions. \n \n3.2 Proteomic profiling of lesion microenvironment-derived sEVs reveals stage-associated remodeling in \nEM \nGiven the disease stage- and tissue-specific differences observed in sEV surface marker profiles, we \nnext investigated whether these phenotypic alterations were accompanied by broader changes in sEV-associated \nprotein cargo. Proteomic profiling of sEVs isolated from matched EU, EMS, and PF revealed pronounced stage- \nand tissue-dependent remodeling of the sEV proteome across the local EM microenvironment. PCA \ndemonstrated clear separation between mild- and severe-stage tissue-derived sEV samples, consistent with \nstage-associated proteomic signatures (Fig. 2a). Differential protein overlap analyses further supported this \ndivergence, with severe-EMS-sEVs exhibiting substantial expansion of uniquely detected proteins (1012, 16% \nunique analytes) relative to mild-stage disease. In contrast, EU- and EMS-sEVs within each disease stage \nshared the majority of detected proteins and exhibited minimal unique representation (<2% unique analytes; \nFig. 2b-d), indicating that proteomic differences were driven predominantly by disease stage rather than tissue \nsource. \n \nConsistent with these findings, unsupervised hierarchical clustering segregated mild- and severe-stage \ntissue samples into distinct clusters (Fig. 2e). Although, one severe-stage EU sample clustered more closely \nwith mild-stage tissues, suggesting partial retention of eutopic-like molecular characteristics in select severe-\nstage cases. A similar, though less pronounced, pattern was observed within the peritoneal microenvironment, \nwhere PF-derived sEV proteomes demonstrated partial stage-associated clustering alongside increased protein \ndiversity in severe-stage disease (Fig. S2a-c), consistent with remodeling of the sEV proteome during EM \nprogression.  \n \nTo define the biological programs underlying these stage-associated proteomic shifts, pathway \nenrichment analyses were performed. Comparisons between mild-stage EU and severe-stage EMS demonstrated \nbroad enrichment of inflammatory and immune-associated signaling networks, including cytokine signaling, \nantigen presentation, innate immune activation, cellular stress responses, and apoptosis-related pathways (Fig. \n2f, g), consistent with establishment of an increasingly immune-active lesion microenvironment in advanced \ndisease. In parallel, comparisons between mild- and severe-EMS-sEV proteomes identified enrichment of \npathways linked to epithelial remodeling, cytoskeletal organization, adhesion dynamics, and immune regulation \n(Fig. 2h, i), supporting progressive restructuring and influence on shaping the dynamic lesion \nmicroenvironment during disease progression. At the protein level, these alterations reflected coordinated shifts \nbetween epithelial maintenance-associated and inflammatory remodeling-associated programs. Mild-stage \ndisease was characterized by enrichment of proteins linked to epithelial integrity and junctional stability, \nincluding galectin-7 (LGALS7) and junction plakoglobin (JUP), which were preferentially enriched in mild-\nstage EMS and EU tissues relative to severe-stage samples (Fig. 2j, k). In contrast, severe-stage disease was \ncharacterized by enrichment of proteins associated with immune modulation, including galectin-3 (LGALS3) \nand HLA-DRA, as well as ECM remodeling, including prostacyclin synthase (PTGIS) and versican (VCAN) \n(Fig. 2l-o).  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n \nConsistent with tissue-derived findings, PF-derived sEV proteomes demonstrated stage-associated \nalterations linked to cell adhesion and tissue remodeling in mild-stage disease and altered metabolic and ECM-\nassociated processes in severe-stage disease (Fig. S2d-k). Biological process and pathway enrichment analyses \nrevealed that severe-stage PF-derived sEVs were enriched for proteins associated with membrane remodeling, \nECM biosynthesis, metabolic reprogramming, and inflammatory stress adaptation, consistent with progression \ntoward a chronic inflammatory and lesion permissive state (Fig. S2d, e). Notably, several proteins demonstrated \ncompartment-specific enrichment patterns between tissue- and PF-derived sEVs, suggesting selective \nextracellular distribution of stage-associated signaling programs within the lesion microenvironment (Fig. S3). \nProteins linked to remodeling (PROM1 or CD133, MGAT1) and proliferative-associated processes (MYOF, \nTGFBR3) were preferentially enriched in mild-stage PF-derived sEVs despite reduced abundance in \ncorresponding tissue-derived populations (Fig. S3e,f), whereas proteins associated with inflammatory stress \nadaptation and proteostatic regulation (XRCC5, PSMD9) demonstrated concordant enrichment across severe-\nPF- and tissue-sEVs (Fig. S3g). PF-derived sEV proteomes demonstrated partial overlap with tissue-derived \nstage-associated signatures, while also exhibiting distinct enrichment patterns associated with the peritoneal \nmicroenvironment. Collectively, these findings identify coordinated but compartment-specific stage-associated \nsEV proteomic programs in EM. \n \n3.3 Plasma-derived sEV proteomes reflect progressive systemic immune, vascular and ECM-remodeling \nprograms in EM \nTo determine whether the stage-associated molecular remodeling observed within lesion compartments \nwas reflected systemically, we performed proteomic profiling of plasma-sEVs from EM patients and healthy \ncontrols. PCA demonstrated clustering of EM patient samples relative to healthy controls. In contrast to the \nstage-associated separation observed in tissue-sEV proteomes (Fig. 2a), plasma-derived sEVs exhibited less \ndistinct separation between mild- and severe-stage disease, consistent with a more conserved systemic disease-\nassociated sEV signature across EM stages (Fig. 3a). Differential protein overlap analyses identified both \nshared and stage-specific systemic signatures, including a subset of uniquely detected proteins in severe-stage \nplasma-sEVs (117, 9% unique analytes), consistent with progressive systemic proteomic remodeling in severe \ndisease (Fig. 3b, c). Unsupervised hierarchical clustering largely separated patient and control plasma samples. \nThough, partial clustering of mild-stage patients with controls suggested that mild disease retains a more \nhomeostatic systemic sEV profile, which becomes lost in a stage-dependent manner (Fig. 3d). \n \nTo define the biological programs underlying systemic sEV remodeling, pathway enrichment analyses \nwere performed. Comparisons between mild-stage patients and healthy controls revealed enrichment of \ninflammatory signaling and membrane remodeling pathways, including innate immune activation, cytokine-\nassociated signaling, complement activation, and lipid metabolic processes, consistent with systemic immune \nactivation and membrane lipid remodeling in mild EM (Fig. 3e, f). In contrast, severe plasma-sEVs \ndemonstrated enrichment of epithelial remodeling programs coupled with stress- and inflammation-associated \nsignaling pathways, including Rho GTPase-linked cytoskeletal regulation and keratinization-associated \nprocesses (Fig. 3g, h). \n \nAt the protein level, systemic alterations reflected coordinated changes in epithelial integrity, immune \nregulation, and remodeling-associated signaling. Proteins associated with epithelial maintenance and junctional \nstability, including LGALS7 and JUP, were preferentially enriched in control plasma-sEVs relative to both \nmild- and severe-stage patient samples (Fig. 3i, j), consistent with a loss of epithelial homeostasis in severe \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\ndisease. Arginase-1 (ARG1) was similarly elevated in controls, supporting altered systemic immune regulatory \nand myeloid-associated signaling during EM progression (Fig. 3k). In contrast, severe plasma-sEVs \ndemonstrated enrichment of proteins associated with inflammatory persistence, vascular remodeling, and pro-\nfibrotic signaling, including NRAS, TGFB1, and coagulation factor XIII A chain (F13A1) (Fig. 3l-n), \nconsistent with enhanced systemic inflammatory signaling, vascular activation, and ECM stabilization in severe \ndisease. \n \nComparisons of plasma- and tissue-derived sEV proteomes revealed both conserved and compartment-\nspecific signatures. Proteins associated with epithelial adhesion (DSC1, PKP1) and homeostasis (CSTA, \nPOF1B) were enriched in control plasma and mild-stage tissues (Fig. S3a), whereas proteins linked to \ninflammatory and remodeling-associated programs (TGFB1, FN1, ICAM1, and F13A1) demonstrated \nconcordant enrichment across severe-stage plasma and EMS-derived sEVs (Fig. S3b). These findings support \npartial reflection of lesion-associated remodeling programs within the systemic circulation. Immune-associated \nproteins displayed both shared and compartment-specific regulation across disease states, with antigen-\npresentation and inflammatory activation markers enriched in severe-stage tissues and plasma-derived sEVs \n(HLA-related proteins, B2M) (Fig. S3c). Proteins associated with immune homeostasis and neutrophil-\nassociated responses were relatively enriched in healthy controls (CD37, IL36G), suggesting systemic immune \nremodeling during disease progression (Fig. S3c). Additionally, proteins associated with immune modulation \nand tissue adaptation (ATG1 and S100A7) were enriched in mild-stage EMS-sEVs relative to severe-stage \ndisease samples, potentially supporting a role for these processes in lesion development (Fig. S3d). \nCollectively, these findings support the emergence of coordinated systemic and lesion-associated sEV \nproteomic programs in EM. \n \n \n3.4 Lipidomic profiling of lesion-derived sEVs reveals stage-associated remodeling in EM \nWhile proteomic and surface marker profiling provide important insight into the molecular cargo and \nimmunological landscape of sEVs, they do not fully capture alterations in membrane lipid composition that \naccompany vesicle structure and function. As major structural components of sEVs, lipids regulate membrane \norganization, cargo packaging, and recipient cell interactions. Although metabolic dysregulation is increasingly \nrecognized as an important feature of EM pathophysiology, the lipid composition of disease-associated sEVs \nremains poorly understood. We therefore incorporated lipidomic profiling to identify stage- and tissue-specific \nmetabolic signatures and to complement proteomic characterization of EM-derived sEVs. \n \nLipidomic profiling of tissue-derived sEVs from a separate patient cohort containing pooled EU samples \n(n = 4; combined mild- and severe-stage due to limited sample availability) and EMS from patients with mild (n \n= 4) and severe (n = 4) EM revealed pronounced stage- and tissue-specific alterations in lipid composition, \npredominantly in EMS-derived sEVs. Lipid species were analyzed using LC-MS in both positive and negative \nelectrospray ionization modes to improve lipidome coverage, as complementary ionization strategies enable \ndetection of distinct subsets of lipid species based on their physical and chemical properties. PCA revealed clear \nseparation between EU and severe-stage EMS-sEVs, and most notably between mild- and severe-EMS-sEVs \n(Fig. 4a, d, j, m). Differentially expressed lipids were visualized in an unsupervised heatmap, revealing distinct \nclustering patterns among disease groups based on their lipid composition (Fig. 4b, e, k, n). To further define \nthe biological programs underlying these lipidomic alterations, metaboanalyst enrichment was performed. \nComparisons between severe-EMS and EU sEVs demonstrated enrichment of multiple membrane-associated \nlipid classes, including sphingomyelins (SM), glycerophosphocholines (GPC), phosphatidylethanolamines (PE), \nphosphatidylserines (PS), phosphatidylinositols (PI), and ceramides (Cer), together with increased abundance of \nneutral glycerolipids (DG, MAG, and TG; Fig. 4c, f). Similarly, comparisons between mild- and severe-EMS-\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nsEVs identified enrichment of membrane phospholipids and sphingolipids alongside neutral lipid classes, \nincluding SM, PI, PG, DG, MAG, TG, and monogalactosyldiacylglycerols (MGDG) in severe-EMS, supporting \nmembrane remodeling and altered lipid metabolism associated with severe EM (Fig. 4l, o). \n \nAt the individual lipid level, severe-EMS-sEVs demonstrated increased abundance of \nlysophosphatidylcholine (LPC 16:0), oxidized phosphatidylethanolamine (PEt 37:2 + 4O), and oxidized \nceramide species (Cer d44:1 + O, Cer t42:0 + O) relative to both EU samples and mild-stage EMS (Fig. 4g-i, p-\nr). These alterations are consistent with coordinated remodeling of membrane architecture together with \nincreased oxidative lipid signaling in severe disease. In contrast, comparisons between mild-EMS- and EU-\nsEVs demonstrated modest lipidomic remodeling (Fig. S4). Mild-stage lesions were enriched in neutral lipid \nclasses, including MAG, DG, and TG, together with select fatty acid, MGDG, Cer, and hexosylceramide \n(HexCer) species. These findings suggest that mild-EMS-sEVs likely undergo selective membrane remodeling, \nalthough to a lesser extent than the oxidative and membrane-associated lipid alterations observed in severe \ndisease. \n3.5 Plasma-derived sEV lipidomic profiling identifies disease-associated systemic lipid remodeling in EM \nHaving identified disease stage-dependent remodeling of EMS-derived sEV lipidome, we next examined \nwhether circulating plasma-sEVs from a separate patient cohort containing mild-stage (n=6), severe-stage (n=6) \nand healthy controls (n=8), would reveal disease-specific enrichment. Indeed, PCA and hierarchical clustering \nanalyses clearly demonstrated separation between severe-stage EM and control plasma samples, indicating \nsubstantial systemic remodeling of the circulating sEV lipidome (Fig. 5a, b, g, h). Metaboanalyst enrichment \nrevealed increased representation of multiple membrane-associated lipid classes, including GPC, PE, SM, Cer, \nand neutral glycerolipids (TG and MAG), consistent with widespread remodeling of circulating membrane \nlipids during advanced disease (Fig. 5c, i). At the individual lipid level, severe plasma-sEVs demonstrated \nincreased abundance of oxidized triglycerides (TG 54:5 + O), oxidized phosphatidylcholines (PC 36:6 + OO), \nlysophosphatidylcholines (LPC 16:0), and glycosphingolipid species (Hex2Cer d30:0, Hex2Cer d29:0 + 2O) \ntogether with reduced CoQ10, collectively supporting increased oxidative stress, altered membrane remodeling, \nand mitochondrial dysfunction in severe EM (Fig. 5d-l). Comparisons between mild-EM and controls similarly \ndemonstrated distinct lipidomic differences, with PCA, heatmap visualization, and lipid class enrichment \nanalyses revealing separation between groups and altered lipid composition patterns (Fig. S4), indicating that \nsystemic alterations in the circulating sEV lipidome are detectable in mild-stage disease. \nDirect comparison of mild- and severe-plasma-sEVs demonstrated comparatively modest stage-\nassociated differences, with partial overlap observed by PCA despite selective enrichment of \nglycerophospholipid (GP), fatty acid (FA), and ceramide (Hex1Cer(d40:4), Cer(t45:0 + O)) classes in severe \ndisease (Fig. 5m-u). Plasma-sEVs demonstrated fewer stage-dependent differences compared with EMS-sEVs. \nSeveral lipid species, including LPC(16:0), oxidized triglycerides, and glycosphingolipids, were consistently \naltered across tissue- and plasma-derived sEV populations. \n \n3.6 Integrated proteome and lipidome signatures identify coordinated molecular remodeling of severe-\nEMS-derived sEVs \nIntegrated multi-omics analyses were performed to determine whether stage- and tissue-dependent \nalterations identified across individual proteomic and lipidomic datasets represented coordinated molecular \nprograms. Following preprocessing and integration of proteomic and lipidomic features, intra- and inter-omics \ncorrelation analyses demonstrated covariance within and between molecular datasets in both tissue- and plasma-\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nderived sEVs, supporting relationships between complementary molecular cargo classes (Fig. S5a, d; S6a, d). \nProteome-lipidome covariance was greatest within lesion-derived sEVs (positive ion RV = 0.66), whereas \nplasma-derived sEVs demonstrated moderate integration across positive (RV = 0.40) and negative ion datasets \n(RV = 0.63). Unsupervised hierarchical clustering of integrated protein and lipid features (displayed in the \nupper and lower portions of the heatmaps, respectively) demonstrated stage-associated grouping of tissue- and \nplasma-sEV samples, with more pronounced separation observed between mild- and severe-tissue-sEVs \ncompared with circulating plasma-sEVs (Fig. S5b, e; S6b, e). Correlation network analyses further identified \nextensive positive and negative associations between protein and lipid features across both tissue- and plasma-\nderived sEVs, revealing interconnected molecular relationships between distinct cargo classes (Fig. S5c, f; S6c, \nf). \nBuilding on these findings, multiple co-inertia analysis (MCIA) was performed to evaluate concordant \nvariation between proteomic and lipidomic profiles at the sample level. Integrated tissue-derived sEV profiles \ndemonstrated separation between severe-EMS and EU samples, while mild-EMS samples exhibited greater \nsimilarity to EU-derived profiles, suggesting progressive molecular divergence in severe EM (Fig. 6a,d). \nPlasma-derived sEV profiles showed partial separation between severe-plasma and controls, whereas mild-\nplasma-sEVs displayed greater overlap with control profiles, suggesting progressive systemic molecular \nremodeling in severe EM that remains less pronounced than the signatures observed in tissue-derived sEVs \n(Fig. 6g, j).  \nTo identify the biological programs associated with the integrated molecular features distinguishing \ntissue compartments and disease stages, Reactome pathway enrichment analysis was performed. In tissue-\nderived sEVs, comparisons between severe-EMS and EU samples revealed enrichment of metabolic, \nmembrane-associated, immune, and cellular organization pathways in positive ion mode (Fig. 6b), whereas \nnegative ion mode analysis identified additional enrichment of endothelial-, immune-, and membrane \nremodeling-associated pathways (Fig. 6e). In plasma-derived sEVs, integrated analysis of severe-stage patients \nrelative to healthy controls identified enrichment of pathways associated with immune and intercellular \ncommunication, cell junction and ECM organization, mitochondrial metabolism, and platelet/hemostatic \nsignaling in positive ion mode (Fig. 6h). Negative ion plasma analysis similarly demonstrated enrichment of \nendothelial-, platelet-associated, immune, oxidative stress-, and apoptosis-related pathways in severe-stage \npatients (Fig. 6k). These coordinated pathway enrichments suggest that the integrated molecular signatures \nidentified by multi-omic analyses reflect interconnected biological processes rather than isolated protein or lipid \nalterations, prompting further investigation of feature-level molecular relationships. \nTo further investigate relationships between individual molecular features, causal discovery analysis was \nperformed using integrated proteomic and lipidomic datasets to infer potential regulatory connections within \nsEV-associated molecular networks. These analyses identified relationships between molecular features that \nextended beyond individual protein or lipid alterations. Tissue-derived sEV networks revealed interconnected \nmodules associated with mitochondrial bioenergetics, membrane lipid remodeling, cellular stress responses, \ncytoskeletal organization, epithelial remodeling, vascular-associated signaling, and oxidative stress responses \n(Fig. 6c,f). Plasma-derived sEV networks demonstrated interconnected modules involving phospholipid and \nsphingolipid remodeling, epithelial and ECM organization, innate immune signaling, oxidative stress, lipid \ntransport, proteostasis, and metabolic adaptation (Fig. 6i,l). \nCollectively, integrated proteomic and lipidomic analyses reveal coordinated stage- and tissue-\nassociated sEV molecular programs during EM progression. Tissue-derived sEVs demonstrated increasingly \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\ndistinct lesion-associated molecular signatures involving immune activation, ECM organization, vascular \nsignaling, and metabolic adaptation, whereas plasma-derived sEVs reflected parallel systemic inflammatory, \nvascular, and metabolic remodeling that became more apparent with advancing disease despite greater overlap \nbetween integrated profiles. \n3.7 EM-derived sEVs induce stage- and tissue-specific functional alterations in human uterine \nmicrovascular endothelial cells \nTo determine whether stage-specific molecular differences in EMS-derived sEVs were translated into \nfunctional effects, we assessed their impact on endothelial uptake and angiogenic activity in human uterine \nmicrovascular endothelial cells (HUtMECs). Cells were treated with MemGlow™-labeled sEVs derived from \nmild or severe EMS lesions and monitored by live-cell imaging over a 10 h time course. HUtMEC uptake of \nboth mild- and severe-EMS-sEVs was visualized over time, with minimal intracellular fluorescence \nimmediately following treatment and increased signal observed at 4 h post-incubation (Fig. 7a,b). EMS-derived \nsEV-associated fluorescence overlapped with the cytoplasmic stain BODIPY TR, indicating intracellular \naccumulation within the cytoplasm following endothelial cell uptake (Fig. 7c, d). In contrast, sEV-associated \nfluorescence did not overlap with Hoechst-labeled nuclei at any time point, suggesting that internalized sEVs \nremained predominantly excluded from the nuclear compartment (Fig. 7a, b).  \n \nFollowing sEV uptake, the intracellular fluorescence pattern observed within HUtMECs demonstrated a \nmorphology resembling mitochondrial structures, suggesting potential mitochondrial association of internalized \nEMS-sEVs. Given the enrichment of mitochondrial-associated proteins identified through proteomic profiling, \n(ATP5F1A-C, NDUF6,8,10, COX4I1, COX6, and SDHB), together with increased abundance of oxidative lipid \nspecies in severe-EMS-sEVs (Fig. S7a-i; Fig. 4h, i, q, r), we next investigated whether EMS-sEVs specifically \nlocalize to the mitochondria following uptake by HUtMECs. Using the same live-cell imaging workflow \ndescribed above, cells were treated with MemGlow™-labeled mild- or severe-EMS-sEVs and monitored over a \n10 h time course. To capture early intracellular uptake dynamics, additional early timepoints were included for \nassessment of sEV localization relative to mitochondria (Fig. 7g, h). To specifically assess potential \nmitochondrial association of internalized sEVs, BODIPY TR staining was replaced with a mitochondria-\ntargeted fluorescent stain. Both treatment groups demonstrated progressive co-localization of MemGlow™ \nfluorescence with MitoTracker™ Deep Red staining over time (Fig. 7e, f), enabling assessment of sEV-\nassociated fluorescence relative to mitochondrial structures. Quantitative co-localization analysis demonstrated \nsignificantly greater mitochondrial-associated fluorescence in severe EMS-sEV-treated cells compared with \nmild-EMS-sEV-treated cells between 4 and 10 h post-treatment (T2–T5; Fig. 7g, h). Together, these findings \ndemonstrate efficient uptake and mitochondrial localization of EMS-sEVs by HUtMECs, with enhanced \nmitochondrial association of severe-EMS-sEVs, reflecting increased capacity to modulate endothelial cell \nfunction. \n \nTo determine whether uptake of lesion-derived sEVs translated into functional alterations in endothelial \ncell behaviour, proliferation, apoptosis, and cytokine secretion were assessed following treatment with the same \ncohort of stage- and tissue-specific sEV populations. Mild-EMS-sEVs induced a significant increase in \nHUtMEC proliferation following 12 h incubation relative to control conditions (Fig. 8a), consistent with \nenhanced endothelial metabolic activity and proliferative potential. In contrast, no significant differences in \napoptosis (measured via Caspase 3/7 activity) were observed across treatment groups during the same timeline \n(Fig. 8b), suggesting that sEV exposure did not substantially alter apoptotic signaling under these experimental \nconditions. \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nAnalysis of conditioned media further revealed distinct cytokine release profiles following treatment \nwith stage- and tissue-specific sEV populations. Severe-EMS-sEVs induced significantly greater secretion of \nchemokines, and growth- and tissue repair-associated factors compared with other treatment groups (Fig. 8c). \nMCP-1 secretion was significantly increased relative to cell-only controls (P < 0.05), whereas MCP-3 (P < \n0.05), FGF-2 and VEGF-A (P < 0.05), and M-CSF (P < 0.01) were significantly elevated compared with cell-\nonly controls, EU-sEVs, and mild-EMS-sEVs. (Fig. 8c). In contrast, mild-EMS- and severe-EU-sEVs induced \nsignificantly greater secretion of pro-inflammatory cytokines and chemokines (IL-6, IL-8; P < 0.05, GROα; P < \n0.01), with a concomitant increase in PDGF-AA (P < 0.01) relative to other treatment groups, suggesting \npotential involvement in endothelial and environmental remodeling programs linked to lesion establishment and \nvascular adaptation (Fig. 8c). \n \nBuilding on the increased VEGF-A secretion observed following severe-EMS-sEV treatment (Fig. 8d), \nan angiogenic tube formation assay was performed to determine whether sEV-induced endothelial signaling \ntranslated into functional changes in angiogenic capacity. Treatment with mild-EMS- and severe-EMS-, as well \nas matched EU-sEVs, induced endothelial tubulogenesis following 12 h incubation (Fig. 8e-j). Live-cell \nimaging revealed more extensive vascular network formation in severe-EMS sEV-treated HUtMECs (Fig. 8j) \nrelative to vehicle control, mild sEV- and severe-EU sEV-treated HUtMECs(Fig. 8e, g-i), with a morphology \nthat appeared comparable to VEGF-A-treated positive control conditions (Fig. 8f). These qualitative \nobservations were supported by quantitative WimTube analysis demonstrating significant increases in total tube \nnumber and branching points (Fig. 8k-m). These results demonstrate that sEVs derived from distinct disease \nstages and tissue sources differentially modulate endothelial cell uptake, cytokine secretion, and angiogenic \nbehaviour.  \n \n4. Discussion \nEM remains a heterogeneous disease with substantial impacts on quality of life, reproductive health, and \nclinical management in over ~200M women worldwide.24 Current therapeutic strategies, including hormonal \nsuppression and surgical intervention, primarily focus on symptom management and lesion removal, but do not \nprevent disease recurrence and require careful consideration in individuals seeking to preserve or achieve \nfertility.5,25 Despite its considerable clinical burden, the molecular mechanisms governing lesion establishment, \nectopic tissue survival, and progression toward chronic inflammatory, vascular, and fibrotic states remain \nincompletely understood. In particular, the early molecular events that enable refluxed endometrial tissue to \nimplant and persist within the peritoneal cavity remain poorly defined.1,2 This knowledge gap is further \ncomplicated by the substantial biological heterogeneity of EM, where distinct lesion subtypes exhibit diverse \nmolecular and cellular features despite frequent classification by rASRM stage. Together, these challenges \ncontribute to delayed diagnosis, limited non-invasive biomarkers, and a lack of mechanism-directed therapeutic \nstrategies. Previous studies have shown that sEVs are altered in EM and contribute to intercellular \ncommunication through the transfer of proteins, lipids, and nucleic acids that reflect the physiological state of \ntheir cells of origin. 6,7 These findings have highlighted sEVs as both mediators of disease biology and \npromising sources of non-invasive biomarkers. However, most studies have examined individual cargo classes \nor isolated biological compartments, limiting our understanding of stage- and tissue-specific sEV signatures.  \n \nIn the present study, we performed a comprehensive stage- and tissue-specific characterization of EM-\nderived sEVs isolated from eutopic endometrium, ectopic lesions, PF, and plasma using complementary surface \nimmune phenotyping, proteomic, lipidomic, and functional analyses. Together, these approaches identified \ncoordinated molecular programs associated with immune remodeling, epithelial plasticity, ECM organization, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nmetabolic adaptation, and vascular signaling that evolve throughout disease progression. Importantly, functional \nstudies demonstrated that disease-associated sEV populations actively influence endothelial cell behaviour in \nvitro, supporting a role for sEV-mediated intercellular communication in shaping the endometriotic lesion \nmicroenvironment. Collectively, these findings provide new insight into the molecular mechanisms underlying \ndisease heterogeneity and establish an integrated framework for understanding how stage- and tissue-specific \nalterations in sEV composition may contribute to EM pathogenesis. \nCharacterization of sEV populations confirmed expected morphology and size distributions across all \nsample types, while revealing significant stage- and source-specific differences in vesicle properties. TEM \nanalysis revealed increased structural heterogeneity within severe-PF- and EMS-derived sEV preparations, \nincluding distinct intravesicular electron-dense structures. While the composition of these structures remains \nunknown, their increased prevalence in severe disease-associated sEV populations suggests that disease stage is \naccompanied by alterations in vesicle organization beyond changes in size or abundance. Severe-stage plasma \ncontained increased concentrations of circulating sEVs together with selective enrichment of CD9 expression, \nconsistent with previous reports identifying CD9 as a predominant marker of circulating biofluid-derived EV \npopulations.26 In contrast, both severe-stage EMS and EU tissue-derived sEVs demonstrated increased \nexpression of CD63 and CD81, supporting stage-dependent remodeling of local vesicle populations and \naligning with previous studies reporting preferential enrichment of these tetraspanins in tissue-derived EV \nsamples.27 Together, these findings highlight that canonical EV markers are highly context-dependent and \nemphasize the importance of considering sample origin when interpreting sEV-associated signatures in EM.  \nSurface phenotyping further demonstrated that sEV populations undergo pronounced stage- and tissue-\ndependent phenotypic remodeling, indicating that severe disease is accompanied not only by quantitative \nchanges in vesicle abundance but also by alterations in molecular composition that may influence recipient cell \ninteractions. While circulating plasma-derived sEVs retained relatively conserved surface profiles, lesion-\nderived vesicles exhibited disease stage-associated enrichment of immune-, adhesion-, and antigen presentation-\nassociated markers, suggesting increasing specialization of local sEV populations within the inflammatory \nlesion microenvironment. Importantly, these surface signatures may be interpreted as indicators of altered sEV \ncomposition within the lesion microenvironment, which could arise from changes in cargo sorting, cellular \ncomposition, or the activation state of EV-producing cells.  \nSevere-stage lesions were accompanied by reduced epithelial- and stemness-associated markers, \nincluding CD133/1 and EpCAM (CD326), together with increased HLA-II expression. Given the established \nroles of EpCAM in epithelial organization, these findings suggest a shift in lesion-derived sEV surface \ncomposition away from epithelial-associated features toward immune-associated signatures in severe EM. \nHowever, these changes may reflect both altered molecular sorting into sEVs and broader remodeling of the \nlesion cellular landscape, where increased stromal activation, ECM remodeling, and immune infiltration may \nalter the relative contribution of epithelial-, stromal-, and immune-derived vesicle populations.28–30 Although \nCD133 expression varies depending on biological source and analytical approach, CD133/prominin-1 has been \nassociated with epithelial and stemness features in endometrial and endometriotic tissues, and recent organoid-\nderived EM studies demonstrate incorporation of CD133 into epithelial-derived sEV populations.18,27 Our \nfindings extend these observations by demonstrating stage-dependent reductions in sEV-associated CD133/1 \nacross both lesion- and PF-derived sEVs, suggesting that alterations in epithelial-associated sEV signatures \nwithin the local lesion microenvironment are associated with severe disease. Although systemic surface \nphenotypes were comparatively subtle, severe-plasma-derived sEVs demonstrated increased expression of \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nadhesion- and platelet-associated markers (CD29, CD41b, CD42a, and CD62P), consistent with growing \nevidence linking platelet activation, vascular remodeling, and coagulation pathways to EM \npathophysiology.28,31,32 Together with the observed enrichment of CD9 in severe-plasma-sEVs, these alterations \nhighlight the context-dependent nature of circulating sEV surface signatures and support the utility of plasma-\nderived vesicles as minimally invasive indicators of EM-associated remodeling, providing a potential \nframework for future biomarker development and patient stratification. \nImportantly, these stage- and tissue-specific surface phenotypes closely paralleled the molecular \nprograms identified through proteomic and lipidomic profiling, revealing coordinated yet compartment-specific \nremodeling of sEV composition across EM. Tissue-derived sEV proteomes demonstrated distinct stage-\nassociated molecular programs, with mild-stage sEVs enriched for signatures associated with epithelial \nmaintenance, adhesion stability, and proliferative remodeling, while severe-stage sEVs exhibited increased \nrepresentation of immune activation, ECM remodeling, vascular signaling, and inflammatory stress adaptation. \nThe increased structural heterogeneity observed by TEM in severe disease-associated sEV populations further \nsupports the concept that severe EM is accompanied by coordinated remodeling of sEV composition and \norganization. These structural differences coincided with distinct lipidomic signatures in EMS-derived sEVs, \nsuggesting that alterations in membrane composition may influence vesicle organization, cargo distribution, or \nbiophysical properties. In contrast to EMS-derived sEVs, plasma-sEVs exhibited a largely shared disease-\nassociated lipid signature across mild- and severe-stage EM relative to healthy controls, with additional stage-\nassociated lipid alterations observed in severe disease. Notably, LPC(16:0), oxidized triglycerides, and \nglycosphingolipids were recurrently altered across multiple biological compartments, suggesting that \ncoordinated membrane remodeling and oxidative lipid metabolism represent conserved features of EM-derived \nsEVs despite broader tissue- and stage-specific differences. Together, alterations in surface markers, structural \nfeatures, and molecular cargo demonstrate that disease-associated sEV remodeling extends beyond individual \nmarkers or cargo classes, reflecting broader restructuring of vesicle biology across disease compartments.  \n \nTo our knowledge, this represents the first comprehensive integration of sEV-associated proteomic and \nlipidomic profiles across distinct sample types and disease stages in EM. Integrated multi-omic analyses \nrevealed coordinated relationships between complementary molecular cargo classes, demonstrating that stage- \nand tissue-dependent alterations identified through individual datasets converge on shared patterns of sEV \nremodeling. These findings suggest that local lesion-associated vesicles undergo more tightly coupled \nmolecular remodeling, while circulating sEV populations retain broader disease-associated signatures shaped by \ndiverse cellular contributions. Notably, these multi-omic signatures further align with previously identified \nregulatory RNA alterations in EM-derived sEVs, suggesting that disease-associated remodeling extends across \nmultiple cargo classes rather than representing isolated changes within individual molecular layers. Lesion-\nderived vesicles exhibited altered expression of let-7 family members, miR-23a, miR-206, and miR-320a, \ntogether with dysregulation of the long non-coding RNAs H19 and NEAT1, while plasma-derived sEVs \ndisplayed distinct circulating miRNA profiles relative to healthy controls.13 These regulatory RNA programs \nhave been implicated in biological processes central to EM progression, including inflammatory signaling, \ncellular plasticity, invasive phenotypes, and tissue remodeling.33–36 The convergence of RNA, protein, and lipid \nsignatures supports a model in which distinct sEV cargo classes collectively contribute to disease-associated \nextracellular signaling and highlights the value of integrated molecular profiling for defining conserved and \ntissue-specific mechanisms underlying EM pathogenesis. \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nTo determine whether these coordinated molecular programs translated into biologically meaningful \neffects on recipient cells, we evaluated the functional effects of tissue-derived sEVs on HUtMECs, an \nendothelial cell model representative of vascular remodeling during lesion establishment and persistence. Given \nthe dependence of ectopic lesions on vascularization, endothelial cells represent a key cellular target through \nwhich EM-derived sEVs may influence lesion development. Although both mild- and severe-EMS-sEVs were \nefficiently internalized, severe-stage vesicles exhibited significantly greater intracellular accumulation and \npreferential mitochondrial localization, indicating that disease stage influences not only sEV molecular cargo \nbut also the dynamics of recipient cell interactions. This observation is of particular interest given the \nenrichment of mitochondrial-associated proteins (ATP5F1A-C, NDUF6,8,10, COX4I1, COX6, and SDHB) and \noxidative lipid species (PEt37:2+4O, Cer d44:1+O, Cer t42:0+O) within severe-EMS-derived sEVs. \nAdditionally, the reduced abundance of CoQ10 in plasma-derived sEVs from individuals with severe EM \nsuggests that mitochondrial- and oxidative stress-associated alterations may extend beyond the lesion \nmicroenvironment into the systemic circulation. Metabolic dysregulation and oxidative stress are increasingly \nrecognized features of EM pathophysiology, with previous metabolomic studies identifying alterations in \nenergy metabolism and systemic metabolic profiles, alongside disruption of lipid metabolic pathways, including \nphospholipid-associated remodeling, in individuals with EM.37,38 Together, these findings suggest that sEV-\nassociated lipid and protein remodeling reflects broader metabolic and oxidative adaptations associated with \ndisease stage occurring both within the lesion microenvironment and systemically through circulating sEV \npopulations. Moreover, the enhanced mitochondrial localization of severe-EMS-derived sEVs in recipient \nendothelial cells raises the possibility that EM-derived sEVs may actively influence mitochondrial-associated \nprocesses through the transfer of bioactive cargo. This concept is supported by growing evidence that \nextracellular vesicles selectively package and transfer mitochondrial-associated proteins capable of influencing \nrecipient cell metabolism and bioenergetic homeostasis.39 Future studies will be required to determine whether \nEM-derived sEVs directly alter mitochondrial function and metabolic activity within recipient cells. \n \nAlthough circulating plasma-sEVs provide valuable insight into systemic disease-associated alterations, \nlesion-derived sEVs were selected for functional assessment because they directly reflect the local cellular \nenvironment in which ectopic lesions establish and undergo vascular remodeling. Importantly, the distinct \nendothelial responses induced by EMS-derived sEVs were not explained by differences in vesicle \ninternalization across disease stages, highlighting the importance of stage-specific sEV cargo composition in \nshaping endothelial functional responses. The stage-specific effects observed in HUtMECs were consistent with \nthe distinct molecular signatures identified within mild- and severe-stage sEV populations. Mild-EMS-derived \nsEVs preferentially promoted endothelial proliferation and induced cytokine programs associated with \ninflammatory activation (IL-6, IL-8, and GROa) and vascular adaptation (PDGF-AA) in mild EM, aligning with \nthe enrichment of proteins associated with epithelial maintenance, remodeling, and proliferative signaling \nidentified in mild-stage sEVs. In contrast, severe-EMS-derived sEVs promoted a more pronounced pro-\nangiogenic phenotype characterized by increased VEGF-A secretion, enhanced tube formation, and elevated \nexpression of chemokines involved in immune recruitment and tissue remodeling (MCP-1, MCP-3, M-CSF, and \nFGF-2), consistent with the enrichment of inflammatory, vascular, and ECM-associated molecular programs \nidentified in severe EM. Together, these findings suggest that stage-dependent sEV cargo composition \ncontributes to distinct endothelial responses, highlighting the functional relevance of sEV molecular remodeling \nin EM pathogenesis.  \n \nCollectively, the functional phenotypes induced by lesion-derived sEVs closely mirrored the coordinated \nmolecular programs identified through integrated surface phenotyping, proteomic, and lipidomic analyses, \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\ndemonstrating that these molecular signatures reflect biologically relevant disease-associated processes rather \nthan descriptive differences alone. These findings support a model in which ectopic lesion establishment and \npersistence are shaped by coordinated communication within a lesion-supportive microenvironment,1–4 with \nstage- and tissue-specific sEV populations coordinating immune activation, epithelial plasticity, vascular \nadaptation, ECM remodeling, and metabolic reprogramming in EM. By integrating complementary molecular \nlayers with functional validation, this study provides a novel systems-level framework for understanding EM \nbiology that would not be captured through individual omic approaches alone. Importantly, this work identifies \nlesion-derived sEVs as a previously underappreciated component of the EM microenvironment and highlights \ntheir potential as mechanistic mediators and sources of clinically relevant molecular signatures Although \ntranslation of patient-derived EM findings remains challenging due to substantial clinical and biological \nheterogeneity, including variation in lesion subtype, anatomical location, hormonal status, symptom severity, \nand frequence of comorbid conditions, larger clinically stratified cohorts will be essential to validate these \nsignatures and define their utility across the diverse spectrum of EM. Ultimately, these findings provide a \nfoundation for future studies aimed at leveraging integrated sEV-based molecular profiling to improve disease \nclassification, uncover clinically relevant biomarkers, and advance precision approaches for EM management. \nAuthor Contributions \nJ.P.H conceived and conducted experiments, analyzed data, and wrote the manuscript. K.B.Z. and D.J.S. \nassisted with experiments and processing human patient samples. D.H. isolated sEV samples for lipidomic \nanalyses. O.B. and B.A.L. contributed human patient samples. C.T. conceived experiments, provided reagents \nand financial support. All authors read, edited, and approved the manuscript. \n \nAcknowledgements \nWe thank Oliver Jones for assistance with TEM sample processing and imaging, and Jeffrey Mewburn \nfor valuable microscopy expertise and guidance. We also thank the Abraham laboratory for training and support \nwith NTA and for access to their instrumentation. We thank Kira King and Jessica Pudwell for their assistance \nwith patient sample coordination. We thank Creative Proteomics and the Proteomics and Molecular Analysis \nCore at the Research Institute of the McGill University Health Centre (RI-MUHC) for their sequencing \nservices. \n \nFunding Information \n J.P.H. is a recipient of the Canada Graduate Research Scholarship from the Canadian Institutes of \nHealth Research (CIHR). This research is supported by funding from the CIHR (CIHR 394 570 & 394 022) and \nNatural Sciences and Engineering Research Council (388 772; C.T.). \n \nDeclaration of Interest Statement \nThe authors declare that they have no competing financial interests or personal relationships that could \nhave influenced the work reported in this manuscript. No additional funding or support was received for this \nstudy beyond that disclosed in Funding Information. \n \n \n \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\nREFERENCES:  \n1. Zondervan, K. T., Becker, C. M. & Missmer, S. A. Endometriosis. New England Journal of Medicine \n382, 1244–1256 (2020). \n2. As-Sanie, S. et al. Endometriosis. JAMA 334, 64 (2025). \n3. Horne, A. W. & Missmer, S. A. Pathophysiology, diagnosis, and management of endometriosis. BMJ \n379, e070750 (2022). \n4. Symons, L. K. et al. The Immunopathophysiology of Endometriosis. Trends Mol. Med. 24, 748–762 \n(2018). \n5. Agarwal, S. K. et al. Clinical diagnosis of endometriosis: a call to action. Am. J. Obstet. Gynecol. 220, \n354.e1-354.e12 (2019). \n6. van Niel, G. et al. Challenges and directions in studying cell–cell communication by extracellular \nvesicles. Nat. Rev. Mol. Cell Biol. 23, 369–382 (2022). \n7. Yáñez‐Mó, M. et al. Biological properties of extracellular vesicles and their physiological functions. J. \nExtracell. Vesicles 4, (2015). \n8. Beal, J. R., Ma, Q., Bagchi, I. C. & Bagchi, M. K. Role of Endometrial Extracellular Vesicles in \nMediating Cell-to-Cell Communication in the Uterus: A Review. Cells 12, 2584 (2023). \n9. Matsuura, Y. et al. Exosomal miR-155 Derived from Hepatocellular Carcinoma Cells Under Hypoxia \nPromotes Angiogenesis in Endothelial Cells. Dig. Dis. Sci. 64, 792–802 (2019). \n10. Bertolini, I. et al. Small Extracellular Vesicle Regulation of Mitochondrial Dynamics Reprograms a \nHypoxic Tumor Microenvironment. Dev. Cell 55, 163-177.e6 (2020). \n11. Zhao, G. et al. Exosomal Sonic Hedgehog derived from cancer‐associated fibroblasts promotes \nproliferation and migration of esophageal squamous cell carcinoma. Cancer Med. 9, 2500–2513 (2020). \n12. Chu, X. et al. Extracellular vesicles in endometriosis: role and potential. Front. Endocrinol. (Lausanne). \n15, (2024). \n13. Khalaj, K. et al. Extracellular vesicles from endometriosis patients are characterized by a unique \nmiRNA-lncRNA signature. JCI Insight 4, (2019). \n14. Dai, L., Gu, L. & Di, W. MiR-199a attenuates endometrial stromal cell invasiveness through suppression \nof the IKK /NF- B pathway and reduced interleukin-8 expression. Mol. Hum. Reprod. 18, 136–145 \n(2012). \n15. Hawkins, S. M. et al. Functional MicroRNA Involved in Endometriosis. Molecular Endocrinology 25, \n821–832 (2011). \n16. Haikalis, M. E., Wessels, J. M., Leyland, N. A., Agarwal, S. K. & Foster, W. G. MicroRNA expression \npattern differs depending on endometriosis lesion type†. Biol. Reprod. 98, 623–633 (2018). \n17. Saare, M. et al. Challenges in endometriosis miRNA studies — From tissue heterogeneity to disease \nspecific miRNAs. Biochimica et Biophysica Acta (BBA) - Molecular Basis of Disease 1863, 2282–2292 \n(2017). \n18. Wang, Y. et al. Human Endometriotic Lesion‐Derived Small Extracellular Vesicles Impair Macrophage \nFunction in the Peritoneal Microenvironment. J. Extracell. Vesicles 15, (2026). \n19. Sisnett, D. J. et al. An investigation of the IL-23/Th17 axis and transcriptomic profiles of T helper \nsubsets in endometriosis. Preprint at https://doi.org/10.64898/2026.06.11.731688 (2026). \n20. Saint-Pol, J. & Culot, M. Minimum information for studies of extracellular vesicles (MISEV) as toolbox \nfor rigorous, reproducible and homogeneous studies on extracellular vesicles. Toxicology in Vitro 106, \n106049 (2025). \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n21. Monash Proteomics and Metabolomics Platform & Monash Bioinformatics Platform, Monash University. \nDIA-Analyst: An interactive web-platform to analyze and visualize proteomics data preprocessed with \nSpectronaut and DIA-NN.  https://analyst-suites.org/apps/dia-analyst/. \n22. Zhou, G., Ewald, J. & Xia, J. OmicsAnalyst: a comprehensive web-based platform for visual analytics of \nmulti-omics data. Nucleic Acids Res. 49, W476–W482 (2021). \n23. Ewald, J. D. et al. Web-based multi-omics integration using the Analyst software suite. Nat. Protoc. 19, \n1467–1497 (2024). \n24. World Health Organization (WHO). Endometriosis. Fact sheets. https://www.who.int/news-room/fact-\nsheets/detail/endometriosis. (2023). \n25. Giudice, L. C. & Kao, L. C. Endometriosis. The Lancet 364, 1789–1799 (2004). \n26. Karimi, N., Dalirfardouei, R., Dias, T., Lötvall, J. & Lässer, C. Tetraspanins distinguish separate \nextracellular vesicle subpopulations in human serum and plasma – Contributions of platelet extracellular \nvesicles in plasma samples. J. Extracell. Vesicles 11, (2022). \n27. Garcia-Martin, R., Brandao, B. B., Thomou, T., Altindis, E. & Kahn, C. R. Tissue differences in the \nexosomal/small extracellular vesicle proteome and their potential as indicators of altered tissue \nmetabolism. Cell Rep. 38, 110277 (2022). \n28. Zhang, Q., Duan, J., Liu, X. & Guo, S.-W. Platelets drive smooth muscle metaplasia and fibrogenesis in \nendometriosis through epithelial–mesenchymal transition and fibroblast-to-myofibroblast \ntransdifferentiation. Mol. Cell. Endocrinol. 428, 1–16 (2016). \n29. Vissers, G., Giacomozzi, M., Verdurmen, W., Peek, R. & Nap, A. The role of fibrosis in endometriosis: a \nsystematic review. Hum. Reprod. Update 30, 706–750 (2024). \n30. Symons, L. K. et al. The Immunopathophysiology of Endometriosis. Trends Mol. Med. 24, 748–762 \n(2018). \n31. Bortot, B. et al. Platelets as key cells in endometriosis patients: Insights from small extracellular vesicles \nin peritoneal fluid and endometriotic lesions analysis. The FASEB Journal 38, (2024). \n32. Ding, D., Liu, X., Duan, J. & Guo, S.-W. Platelets are an unindicted culprit in the development of \nendometriosis: clinical and experimental evidence. Human Reproduction 30, 812–832 (2015). \n33. Farzaneh, M. et al. Emerging roles of the long non-coding RNA NEAT1 in gynecologic cancers. J. Cell \nCommun. Signal. 17, 531–547 (2023). \n34. Yuan, D. et al. Expression of lncRNA NEAT1 in endometriosis and its biological functions in ectopic \nendometrial cells as mediated via miR-124-3p. Genes Genomics 44, 527–537 (2022). \n35. Ghazal, S. et al. H19 lncRNA alters stromal cell growth via IGF signaling in the endometrium of women \nwith endometriosis. EMBO Mol. Med. 7, 996–1003 (2015). \n36. Raveh, E., Matouk, I. J., Gilon, M. & Hochberg, A. The H19 Long non-coding RNA in cancer initiation, \nprogression and metastasis – a proposed unifying theory. Mol. Cancer 14, 184 (2015). \n37. Dai, Y. et al. Integrative analysis of transcriptomic and metabolomic profiles reveals abnormal \nphosphatidylinositol metabolism in follicles from endometriosis‐associated infertility patients. J. Pathol. \n260, 248–260 (2023). \n38. Dutta, M. et al. Metabolomics reveals perturbations in endometrium and serum of minimal and mild \nendometriosis. Sci. Rep. 8, 6466 (2018). \n39. Todkar, K. et al. Selective packaging of mitochondrial proteins into extracellular vesicles prevents the \nrelease of mitochondrial DAMPs. Nat. Commun. 12, 1971 (2021). \n  \n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint \n\n.CC-BY 4.0 International licenseavailable under a \n(which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made \nThe copyright holder for this preprintthis version posted August 14, 2026. ; https://doi.org/10.64898/2026.08.13.744471doi: bioRxiv preprint","source_license":"CC-BY-4.0","license_restricted":false}