To recapitulate hormonal priming of the endometrium, we primed human Ishikawa endometrial epithelial cells with E to mimic the proliferative phase, evident from progesterone receptor (PR) A/B upregulation (Supplementary Figure S1). This was followed by sequential stimulation with EP to mimic the secretory phase, evident by down-regulated expression of the PR receptor in a negative feedback loop mechanism as occurs in vivo (Lessey et al., 1988; Lessey et al., 1996) (Supplementary Figure S1). Next, endometrial E-/EP-sEVs were isolated and purified using differential ultracentrifugation coupled to density gradient-based separation (Rai et al., 2021b) (Figure 1A).

In accordance with International Society for Extracellular Vesicles (ISEV) research guidelines (Thery et al., 2018), we characterized sEVs for their biophysical and biochemical properties; sEVs displayed 1.06–1.11 g/ml buoyant density, were positive for stereotypic sEV markers ALIX and TSG101 (Figure 1B) and were ∼140–150 nm diameter based on single particle tracking analysis [below size range for small EVs, as defined by MISEV (Thery et al., 2018)] (Figure 1C) (E-sEVs: mean 142.8 ± 1.1 nm, average particle concentration/ml: 8.21E+08; EP-sEVs: mean 151.3 ± 9.5 nm, average particle concentration/ml: 6.62E + 08). We next performed proteome profiling of sEVs and their parental cells using nLC-MS/MS and data-dependent acquisition (Rai et al., 2021c; Kompa et al., 2021). We found that the proteomes of parental cells versus their derived sEVs (fraction 6–8, 1.06–1.11 g/ml) were distinct (Supplementary Figure S2), with sEVs compared to cells significantly enriched in sEV markers (e.g., CD9/63/81, Alix/PDCD6IP, TSG101) (Greening et al., 2017; Rai et al., 2021d; Mathieu et al., 2021), which includes CD63 as well as proposed universal marker of sEVs/exosomes, SDCBP (Kugeratski et al., 2021) (Figure 1D, Supplementary Table S1) (Rai et al., 2021c; Poh et al., 2021). Further, we show sEVs display lower abundance for non-EV (intracellular/organelle) proteins (e.g., nucleus; HNRNPC, mitochondria; CYSC, endoplasmic reticulum; CALR) (Poh et al., 2021) (Figure 1D, Supplementary Table S1). Using these multiple sEV inclusion proteins to verify the presence of sEVs, is in direct agreement with MISEV guidelines (Thery et al., 2018). These data collectively support successful enrichment of purified sEVs.

We further compared the proteomes of sEVs isolated at different cycle phases (E and EP) to gain insight into their potential role in embryo implantation, particularly trophectoderm/trophoblast invasion. There were striking differences in E- and EP-sEV proteomes (Supplementary Table S2; Supplementary Figure S3) with significantly abundant proteins in EP-sEVs (243 proteins) implicated in cell-cell adhesion, migration, invasion and embryo development (Greening et al., 2016), human embryo implantation (Enciso et al., 2018; Matorras et al., 2018) and endometrial receptivity (Diaz-Gimeno et al., 2011; Altmae et al., 2017; Azkargorta et al., 2018) (Supplementary Figure S3). Additionally, several proteins (DDAH2, PGAM1) were identified in human uterine fluid collected from the secretory phase (Rai et al., 2021a) (Supplementary Figure S3). Further, comparative analysis of Gene Ontology functional enrichment of EP-sEVs (versus E-sEVs) revealed distinct molecular and enzymatic functions associated with ubiquitin binding/interaction, phosphatase activity/binding, molecular adaptor activity, GTPase activity, and ligase binding (Figure 1E, Supplementary Table S3).

Comparative proteome analysis of EP-sEVs (versus E-sEVs) revealed molecular players of epithelial cell migration (BMPR2, DDR1, IGSF8, MST1R), embryo development (COPS3, CUL3, NOTCH1, PLCG1, ADAM10), cell-cell adhesion (SDCBP, EPCAM, NOTCH1), nitric oxide biosynthesis (e.g., DDAH2, AKT1 and SPR) and cell invasion (CSTB, DDR1, RAB25, ST14, TXN) (Supplementary Figure S3; Supplementary Table S2). Importantly, EP-sEVs are enriched in key players shown to remodel endometrium to promote receptivity, or embryo to promote implantation processes (Illera et al., 2000; Haslinger et al., 2013; Zhao et al., 2018; Segura-Benitez et al., 2022); enriched in regulators of endometrial receptivity (CTNNB1, EGFR, EPCAM, NOTCH1 and DDRI), as well as promote implantation processes (BMPR2, BTF3, COPS3, PLCG1, RAB14, ADAM10, CUL3, ITGAV and AKT1) (Table 1). This further supports the successful hormonal priming of endometrial epithelial cells, the distinct molecular composition of sEVs, and highlights the pro-invasive function of EP-primed endometrial cell sEVs. Moreover, similar proteome reprogramming was also observed in sEVs from the human ECC1 endometrial epithelial cells following the same hormonal treatment (Greening et al., 2016) (Supplementary Figure S4).

We next questioned whether these sEVs could be taken up by human embryo-derived TSCs; for this we employed a previously reported human trophoblast stem cell line (hTSC) established from individual blastomeres of donated human embryos (Zdravkovic et al., 2015; Evans et al., 2019; Poh et al., 2021). Confocal microscopy revealed that sEVs labelled with lipophilic dye DiI (Rai et al., 2021a) were readily taken up by hTSCs within 2 h (Figure 1F, Supplementary Figure S5). To assess their ability to regulate trophectoderm cell invasion, we generated spheroids of hTSCs [∼1200 cells as blastocyst mimetics (Evans et al., 2019; Evans et al., 2020a)], stimulated these with sEVs and assessed their capacity to invade into Matrigel™ matrix. Consistent with previous reports for human ECC1-derived sEVs (Evans et al., 2019; Rai et al., 2021a), we demonstrate EP-sEV treated hTSC spheroids display significantly greater invasive outgrowth in Matrigel™ matrix compared to vehicle control (PBS treated) and E-sEV treated hTSC spheroids (Figure 1G, Supplementary Figure S6).

To provide insights into the molecular changes in trophectoderm/trophoblast cells that support their sEV-driven invasion, we performed proteomic profiling of hTSCs following single-dose sEV treatment (Figure 2A). A total of 2,318 proteins were identified (Supplementary Table S4) of which 108 were either uniquely identified (55) or displayed significantly higher abundance (53, p < .05) in hTSCs treated with EP-sEVs compared to E-sEVs (Supplementary Table S5A/5B). Gene Ontology enrichment analysis (Biological Processes) of proteins distinct between EP-sEVs and E-sEVs identified significantly (p < .05) enriched terms such as “TGF-beta receptor signaling pathway; p < 4.62E-04”, “transmembrane receptor protein serine/threonine kinase signaling pathway; p < 5.11E-04,” “BMP signaling pathway; p < 6.65E-04,” “regulation of cell-substrate adhesion; p < 7.89E-04,” “keratinocyte cell migration; p < 3.59E-05”; processes that are implicated in cell invasion (Figure 2B, Supplementary Table S6). These proteins include migratory factors DPP4, ADAM9, and RALB, cell adhesion/invasive regulators CCN1, COL1A1, and THBS1, and cell motility signaling regulator NRP1 (Supplementary Table S7). Kinase Enrichment Analysis 3 (KEA3) (Kuleshov et al., 2021) was employed to identify upstream kinases whose substrates in hTSCs following EP-sEVs treatment were enriched, including EGFR, AKT1, and MAPK3 (Figure 2D). Human kinome regulatory network was used to highlight the top-ranked kinases with all kinases labelled by WGCNA modules, including various interaction networks associated with MAPK3 (Figures 2E, F).

Of note, 307 (from 2,318) proteins, based on cell surface databases CSPA/SURFY as cell surface proteins (plasma membrane, surface) (e.g., LAMP2, ADAM9, LTBP1, STT3A, DPP4, ITGA2/AV, ERO1A) (Supplementary Table S8) are known to interact with the ECM during invasion. Due to their essential role in ECM remodeling during invasion, we next ascertained their surface localization in hTSCs. For this, we employed cell membrane-impermeant biotin to capture surface proteins that were then identified and quantified using MS, similar to our previous report (Rai et al., 2021c) (Figure 2A, Supplementary Table S9).

A total of 1,346 surface proteins were found in hTSCs (vehicle-treated (UT), E-sEV-treated, EP-sEV-treated; identified in at least 2/3 biological replicates for each group, including 328 identified in CSPA/SURFY (Supplementary Table S10). Of the proteins identified as cell surface from CSPA/SURFY, differential analysis revealed 264 and 220 proteins significantly (p < .05) higher abundance in hTSCs stimulated with EP- or E-sEVs compared to PBS-treated hTSCs (Supplementary Tables S9, S10).

Gene Ontology enrichment analysis (Biological Processes) of proteins distinct between each cluster (EP-sEVs vs. UT; 264 proteins, E-sEVs vs. UT; 220 proteins, and EP-sEVs vs. E-sEVs; 185 proteins; Supplementary Table S10) identified significantly (p < .05) enriched terms such as “wound healing,” “cell-substrate adhesion,” and “substrate adhesion-dependent cell spreading” for both EP-sEV and E-sEVs (vs. UT) groups (Figure 2C). For EP-sEV treatment (vs. E-sEVs) processes significantly enriched include “mRNA processing”; p < 3.21E-05, and “extracellular matrix organization”; p < 4.92E-05” (Figure 2C). For EP-sEV treatment (vs. E-sEVs and UT) specific proteins were shown implicated in regulation of antigen presentation (PSMA5, PSMD11, PSMA2), crosslinking of collagen fibrils (BMP1, LOXL1), NOTCH and VEGF signalling (NOTCH2, PSMD11, PSMA2/NCKAP1, NRAS, ROCK2, PXN), and EPHB-mediated forward signalling (ROCK2, ARPC3, EPHB4), important processes regulating cell invasion (Supplementary Table S9/11). These findings suggest EP-sEVs promote remodeling of cell surface proteome in hTSCs, with changes in proteome landscape associated with regulators of implantation and cell invasion (Table 2).

Kinase Enrichment Analysis 3 (KEA3) (Kuleshov et al., 2021) further identified upstream kinases whose substrates from cell surface hTSCs following EP-sEVs treatment were enriched, including cell invasive regulators MAPK3 and AKT1 (Figures 2G–I). We further identify specific growth factor and kinase receptors, including EPHB4/B2, ERBB2, STRAP, EGFR, and PDGFRA, and upregulated expression of pro-invasive regulators associated with focal adhesions (NRP1, PTPRK, ROCK2, TEK) and embryo implantation (FBLN1, NIBAN2, BSG) (Supplementary Table S9) in response to EP-sEV treatment on hTSC surfaceome remodelling. Of note, we report “positive regulation of MAPK cascade”; p < 4.68E-04 (Table 2). This highlights proteome reprogramming of trophoblasts by sEVs towards an invasive phenotype.

The functional response in target cells is dictated by various signaling pathways activated by EVs, which can be identified by studying changes in the phosphoproteome landscape of target cells. Indeed, interrogation of dysregulated proteins in EP-sEVs-treated hTSCs (global or surface proteomes) identified MAPK3 as upstream kinase implicated in proteome reprogramming (Figures 2D, G). Because of the central role MAPK3 plays in MAPK activation in cellular invasion, we investigated whether sEVs activate MAPK pathway in hTSCs. Thus, we performed phosphoproteome analysis of hTSCs treated with EP-sEVs (compared to untreated hTSCs). We identified 2,543 phosphosites (localization probability >0.75) and 933 phosphoproteins across treatment groups (n = 4 per group) (Figures 3A–D, Supplementary Table S12). Of these, 1,252 phosphosites corresponding to 585 proteins were significantly enriched in EP-sEV treated hTSCs compared to untreated hTSCs (Figure 3E, Supplementary Table S12). Their functional enrichment analysis highlighted their involvement in “activation of MAPK signaling, p < 2.34E-02,” “regulation of stress-activated MAPK cascade, p < 1.70E-03,” “protein phosphorylation, p < 4.11E-03” and “ERK1 and ERK2 cascade, p < 4.17E-02,” as well as upstream regulators of MAPK signaling, including MAPK3, and various cyclin-dependent and serine/threonine-protein kinase regulators based on kinase Z-score (Figure 3G, Supplementary Table S13). At a phosphosite level, we pinpoint MAPK3/ERK1 T202 and Y204 and MAPK1/ERK2 T185 and Y187 as significantly higher abundance peptides in hTSCs treated with EP-sEVs compared to untreated hTSCs (Figure 3F, H, 4A, Supplementary Table S12).

We demonstrate EP-sEV-mediated significant activation of MAPK by Western blotting using site-specific antibody (p < .05) (Figure 4B). Further, to ascertain their direct role in hTSC invasion, we stimulated hTSC spheroids with EP-sEVs that was challenged with selumetinib, a pharmacological inhibitor of MAPK3/ERK1 T202 and Y204 and MAPK1/ERK2 T185 and Y187 phosphosites (Figure 4B). We found that selumetinib significantly attenuated MAPK activation in hTSC cells (p < .01) and also abrogated hTSC spheroid invasion into Matrigel matrix to basal levels (p < .0001) (Figures 4C, D). Thus, our data show that endometrial sEVs promote hTSC invasion through MAPK activation.

Human Ishikawa endometrial epithelial cells (Nishida et al., 1985) were cultured and maintained in a 1:1 mix of Dulbecco’s Modified Eagle’s Medium/Nutrient Mixture F-12 (DMEM/F-12) (Invitrogen-Gibco, Carlsbad, United States) supplemented with 1% (v/v) Penicillin-Streptomycin (Pen/Strep) (Life Technologies) and 5% (v/v) Fetal Bovine Serum (FBS) (Life Technologies). Cells were maintained at 37°C, 5% CO₂ and routinely passaged using 0.5% Trypsin-EDTA (Invitrogen-Gibco). For generation of endometrial sEVs, Ishikawa cells were cultured in CELLine™ AD-1000 Bioreactor Classic Flasks (Integra Biosciences) as previously described (Ji et al., 2014). Briefly, Ishikawa cells (∼20 × 10⁶) were seeded into the cultivation chamber, with 500 ml of FBS-supplemented DMEM/F-12 added to the nutrient supply chamber. Cells were cultured for 5 days at 37°C, 5% CO₂ to expand. Thereafter, cells in the cultivation chamber were gently washed to remove FBS, and media replaced with DMEM/F-12 supplemented with 0.6% (v/v) Insulin-Transferrin-Selenium solution (ITS) (Invitrogen-Gibco) and 1% (v/v) Pen/Strep. Culture media in nutrient supply chamber was replaced weekly, while media in cultivation chambers were collected and replaced according to a cyclic hormonal treatment regime to recapitulate proliferative and secretory phases of the menstrual cycle (Greening et al., 2016).

Ishikawa cells were treated with β-oestradiol (E) (10⁻⁸ M, E8875; Sigma-Aldrich) supplemented in DMEM/F-12 serum-free media (E-media), and cultured at 37°C, 5% CO₂ for 48 h to mimic the proliferative phase. E-primed Ishikawa conditioned media (CM) was collected and cells were subsequently treated with β-estradiol (10⁻⁸ M) and medroxyprogesterone 17-acetate (10⁻⁷ M, M1629; Sigma-Aldrich) (EP) supplemented in DMEM/F-12 serum-free media (EP-media), and cultured at 37°C, 5% CO₂ for 48 h to mimic the secretory phase as described (Greening et al., 2016; Poh et al., 2021). The ECC1 human endometrial luminal epithelial cell line was fully validated and cultured as previously described (Greening et al., 2016) with experimental details as for the Ishikawa cells.

Ishikawa endometrial epithelial cells were obtained from Professor Masato Nishida (via Prof. Guiying Nie with permission). ECC1 human endometrial luminal epithelial cells were from Professor Lois Salamonsen [American Type Culture Collection (ATCC)].

The human trophectodermal stem cells T3-TSC (hTSC) derived from individual blastomeres of donated human embryos (Zdravkovic et al., 2015) were maintained as described (Poh et al., 2021). Briefly, cells were cultured in DMEM/F-12 supplemented with 1% (v/v) Pen/Strep, 10% (v/v) FBS, 10 nM fibroblast growth factor (FGF) (bFGF, R&D Systems) and 10 nM SB431542 (#1614, Tocris Bioscience), in flasks pre-coated with 0.5% (w/v) gelatin, and maintained at 37°C, 5% CO₂.

hTSC spheroids were generated as described (Evans et al., 2019) with modifications. hTSCs (∼1500 cells) in 100 µl of hTSC media were cultured in round bottom ultra-low attachment 96-well plates (Costar) for 72 h at 37°C, 5% CO₂ (1 spheroid/well). Spheroids were observed using bright field microscopy and deformed spheroids were excluded from functional use.

E- and EP-primed CM from Ishikawa cells were centrifuged at 500 g, 5 min and 2,000 g, 10 min at 4°C, with supernatant then centrifuged at 10,000 g, 30 min, 4°C (SW28 rotor; Optima L-90K Ultracentrifuge). The resulting supernatant was centrifuged at 100,000 g, 1 h, 4°C to pellet crude sEVs which were washed in 1 ml PBS at 100,000 g, 1 h, 4°C. sEVs from Ishikawa cells were purified using OptiPrep™ density gradient-based separation as described (Rai et al., 2021c). Briefly, 1 ml volumes of 40%, 20%, 10%, and 0.6 ml of 5% iodixanol solution were layered sequentially in a polypropylene tube (11 × 60 mm). Dilutions of iodixanol solution were made in 0.2 M sucrose/1× PBS solution. sEVs (200 µl) were overlaid and centrifuged at 100,000 g, 18 h, 4°C (SW60 rotor, Optima L-90K Ultracentrifuge). Twelve 300 µl fractions were collected, diluted in 1 ml PBS and subjected to a wash spin at 100,000 g, 1 h, 4°C. Pellets were re-suspended in 20–100 µl PBS; fractions six to eight were pooled (termed purified E-sEVs or EP-sEVs) and stored at −80°C. The density of each fraction was determined using a control OptiPrep™ gradient overlaid with 200 µl PBS; each fraction was diluted 1:10,000 in MilliQ and absorbance measured at 244 nm using Nanodrop 2000c; absorbance of fractions were converted to density as described (Rai et al., 2021c).

sEV particle size was determined using nanoparticle tracking analysis (NTA, NS300 NanoSight system) as described (Rai et al., 2021c). Syringe pump speed was set to 100 camera detection threshold was at least 10 and temperature set to 25°C. Three technical replicates (60 s video) were recorded and analyzed using NTA software (3.1.45).

Experimental parameters are submitted to EV-TRACK knowledgebase (EV-TRACK ID: EV220403) (Consortium et al., 2017).

Samples were lysed in western blot buffer [4% (w/v) SDS, 20% (v/v) glycerol and 0.01% bromophenol blue, 0.125 M Tris-hydrochloride, pH 6.8] with 100 mM dithiothreitol (DTT, Thermo Fisher Scientific) as described (Rai et al., 2021c). Membranes were incubated with primary mouse or rabbit antibodies against ALIX (1:1,000 dilution) (3A9; Cell Signaling Technology), TSG101 (1:1,000 dilution) (622696; BD Biosciences), GAPDH (1:1,000 dilution) (D4C6R; Cell Signaling Technology), estrogen receptor alpha (ERα) (1:500 dilution) (D6R2W; Cell Signaling Technology), progesterone receptor A/B (PR A/B) (1:500 dilution) (D8Q2J; Cell Signaling Technology), p-MAPK (9101; Cell signaling), MAPK (9102; Cell signaling), in Tween-PBS (TPBS), overnight at 4°C. Membranes were rinsed with TPBS and incubated with secondary antibodies (1:15,000); IRDye 800CW goat anti-mouse antibody or IRDye 680RD goat anti-rabbit antibody (Li-COR Biosciences), for 1 h while shaking at RT. Membranes were rinsed with TPBS and imaged using Odyssey Infrared Imaging System (Li-COR Biosciences, Nebraska United States), measuring 700 nm and 800 nm wavelengths. Protein-based densitometry to quantify p-MAPK/MAPK relative expression was performed using ImageJ (v1.53c).

E- and EP-sEVs (10 µg) were incubated with 1 µM DiI (Invitrogen) staining solution at RT, 10 min sEVs (and PBS dye control) were overlaid on a 100 µl cushion of 10% OptiPrep™ (in PBS) and centrifuged at 100,000 g, 1 h, 4°C. Supernatants were removed and pellets resuspended in PBS. hTSCs were cultured to ∼70% confluency as droplets in 8-well glass chamber slides (Sarstedt) pre-coated with 0.5% (w/v) gelatin, at 37°C, 5% CO₂. Cells were incubated with 100 μg/ml E- or EP-sEVs (or PBS dye control) at 37°C, 2 h. Cells were washed in DMEM/F-12 media. Nuclei were stained with Hoechst 33,342 stain (Thermo Fisher Scientific) (10 μg/ml) for 10 min, washed with DMEM/F-12 and fixed with 4% formaldehyde at RT, prior to imaging by Nikon A1R confocal microscope equipped with resonant scanner, using a Plan Fluor 20× MImm (DIC N2, Nikon, Tokyo, Japan). Images were sequentially acquired and are representative of three biological replicates.

Transwell-Matrigel™ invasion assays were performed using growth factor-reduced Matrigel™ matrix (Corning), as previously described (Rai et al., 2019). Briefly, 8-well microscopy chambers (Corning) coated with Matrigel™ were overlaid with hTSC spheroids (harvested after 72 h growth, ∼30 spheroids/tube) that were pre-treated with sEVs or PBS alone for 2 h at 37°C. The inserts were nested onto 24-well companion plate (Corning) containing DMEM/F-12 with 1% (v/v) Pen/Strep, 10 nM FGF, 10 nM SB431542 and either E-sEVs or EP-sEVs (50 μg/ml) or PBS alone and incubated for 24 h at 37°C. Subsequently, 50 µl media was removed and mixed 1:1 with Matrigel™ and gently overlaid on spheroids in wells, Matrigel™ allowed to solidify for 30 min at 37°C followed by addition of 200 μl of DMEM/F-12 [10% (v/v) FBS, 1% (v/v) Pen/Strep] to each well. After 48 h, spheroids were imaged using Olympus FSX100 [n = 13–20 spheroids analyzed, with four measurements (mm) taken per spheroid]. The extent (%) of invasion was assessed by calculating [(outer diameter—inner diameter)/(inner diameter) × 100] using a digital ruler. Data are presented as average ±SEM. Where indicated, hTSC spheroids were pre-treated with 14 nM Selumetinib (Sellekchem, AZD6244) (Rai et al., 2019) or DMSO for 30 min at 37°C before treatment with EP-sEVs (100 μg/ml) or PBS for 45 min hTSC spheroids were overlaid onto solidified Matrigel™ to invade for 48 h and imaged as described (n = 7–13 spheroids analyzed).

Global mass spectrometry-based proteomics of Ishikawa cells or derived sEVs (n = 3), or hTSCs following sEV treatment (n = 5) (10 μg) was performed as described (Claridge et al., 2021b; Lozano et al., 2022) using single-pot solid-phase-enhanced sample preparation (SP3) method (Hughes et al., 2019). Briefly, Ishikawa cell and sEV derivatives were solubilized in 1% (v/v) SDS, 50 mM tetraethylammonium bromide (TEAB) pH 8.0, while hTSCs were solubilized in 1% (v/v) SDS, 50 mM HEPES pH 8.0. Samples were incubated at 95°C for 5 min, with cell lysates tip probe sonicated (10 s, 23 amplitude) (Misonix—S-4000 Ultrasonic Liquid Processor). Samples were reduced with 10 mM DTT for 50 min at 25°C followed by alkylation with 20 mM iodoacetamide (IAA) for 30 min at 25°C in the dark. The reaction was quenched to a final concentration of 20 mM DTT. Magnetic beads were prepared by mixing SpeedBeads™ magnetic carboxylate modified particles (65152105050250, 45152105050250, Cytiva) at 1:1 (v:v) ratio and washing twice with 200 µl MS-water. Magnetic beads were reconstituted to a final concentration of 100 μg/μl. Magnetic beads were added to the samples at 10:1 beads-to-protein ratio and 100% ethanol (EA043, ChemSupply) added for a final concentration of 50% ethanol (v/v). Protein-bound magnetic beads were washed three times with 200 µl of 80% ethanol and reconstituted in 50 mM TEAB and digested with trypsin (Promega, V5111) at a 1:50 enzyme-to-substrate ratio for 16 h at 37°C with constant shaking (1,000 rpm). The peptide mixture was acidified to a final concentration of 2% formic acid (FA) (pH ∼1–2) and centrifuged at 20,000 g for 1 min. The peptide digests were frozen at −80°C and dried by vacuum centrifugation (Savant SPD121P, Thermo Fisher Scientific), reconstituted in 0.07% trifluoroacetic acid (TFA), and quantified by Fluorometric Peptide Assay (23,290, Thermo Fisher Scientific) as per manufacturer’s instruction.

Peptides were analyzed on a Dionex UltiMate NCS-3500RS nanoUHPLC coupled to a Q-Exactive HF-X hybrid quadrupole-Orbitrap mass spectrometer equipped with nanospray ion source in positive mode as described (Greening et al., 2021; Kompa et al., 2021). Peptides were loaded (Acclaim PepMap100 C18 3 μm beads with 100 Å pore-size, Thermo Fisher Scientific) and separated (1.9 µm particle size C18, 0.075 × 250 mm, Nikkyo Technos Co. Ltd.,) with a gradient of 2%–28% acetonitrile containing 0.1% FA over 95 min at 300 nl·min⁻¹ followed by 28–80% from 95–98 min at 300 nL·min⁻¹ at 55°C (butterfly portfolio heater, Phoenix S&T). An MS1 scan was acquired from 350 to 1,650 m/z [60,000 resolution, 3 × 10⁶ automatic gain control (AGC), 128 m injection time] followed by MS/MS data-dependent acquisition (top 25) with collision-induced dissociation and detection in the ion trap (30,000 resolution, 1 × 10⁵ AGC, 60 m injection time, 28% normalized collision energy, 1.3 m/z quadrupole isolation width). Unassigned, 1, six to eight precursor ions charge states were rejected and peptide match disabled. Selected sequenced ions were dynamically excluded for 30 s. Data was acquired using Xcalibur software v4.0 (Thermo Fisher Scientific). All MS-based proteomics data (cellular, sEV, phosphoproteome, surfaceome) have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository and are available via ProteomeXchange with identifier (PXD027642). Therefore, for this study, biological replicates were performed as indicated: global proteome analysis (n = 5) and cell surfaceome analysis (n = 3).

hTSCs (∼20,000) were seeded in 96-well plates pre-coated with 0.5% (w/v) gelatin and cultured to 100% confluency. Cells were serum-starved for 24 h prior to treatment with 100 μg/ml purified EP-sEVs or PBS control for 15 min at 37°C, 5% CO₂ (n = 4). Cells were immediately washed 2× in ice-cold PBS and lysed in-well with MS lysis buffer [1% (v/v) SDS, 1:100 HALT protease phosphatase inhibitor cocktail (Thermo Fisher Scientific, 78,442), 50 mM HEPES pH 8] on ice for 5 min, heat treated at 95°C for 5 min prior to SP3 sample preparation. Cell lysates were reduced, alkylated, quenched and trypsin digested as described, with the resulting peptide mixture acidified to a final concentration of 2% FA, and centrifuged at 20,000 g for 1 min. For phosphoproteomic analysis by tandem mass tag (TMT) multiplexing, peptides were labelled with 10-plex TMT labels according to the manufacturer’s instructions (Thermo Fisher Scientific, 90406/A34807, lot UG287488/278919). A list of the sample labelling strategy is available in PRIDE proteomeXchange (PXD027642). Peptides were labelled in a final concentration of 50% acetonitrile for 90 min at RT followed by de-acylation with 0.25% hydroxylamine for 15 min at RT and quenching with 0.1% TFA. Peptides from each 10-plex experiment were pooled and digests lyophilised by vacuum centrifugation and reconstituted in Binding/Equilibration Buffer for phosphopeptide enrichment, using the High-Select™ TiO2 Phosphopeptide Enrichment workflow, as per manufacter’s instructions (Thermo Fisher Scientific, A32993). Briefly, TMT-labelled peptide digests were transferred to a pre-equilibrated TiO₂ spin tip and centrifuged twice at 1,000 g, 5 min. The column was washed twice with binding/equilibration buffer and subsequent wash buffer at 3,000 g, 2 min, followed by a wash with MS-grade water at 3,000 g, 2 min. Phosphopeptides were eluted in 100 µl phosphopeptide elution buffer by centrifugation at 1,000 g, 5 min, dried by vacuum centrifugation and reconstituted in 0.07% TFA, before quantification by Colorimetric Peptide Assay (Thermo Fisher Scientific, 23,275) as per manufacturer’s instructions.

MS-based proteomic analysis of TMT-labelled phosphopeptides was performed on a Dionex UltiMate NCS-3500RS nanoUHPLC coupled to a Q-Exactive HF-X hybrid quadrupole-Orbitrap mass spectrometer equipped with nanospray ion source in positive mode. TMT labelled peptides were separated (1.9 µm particle size C18, 0.075 × 250 mm, Nikkyo Technos Co. Ltd.,) with a gradient of 4%–28% ACN containing 0.1% FA over 224 min at 300 nl/min at 55°C (butterfly portfolio heater, Phoenix S&T). The MS1 scan was acquired from 300 to 1650 m/z (60,000 resolution, 3e6 AGC, 128 m injection time) followed by MS/MS data-dependent acquisition of the top 15 ions with HCD (30,000 resolution, 1e5 AGC, 60 ms injection time, 33 NCE, 0.8 m/z isolation width). Unassigned, 1, six to eight precursor ions charge states were rejected and peptide match disabled. Selected sequenced ions were dynamically excluded for 30 s. Data was acquired using Xcalibur software v4.0. Therefore, for this study, biological replicates of the phosphoproteome analysis (n = 4).

hTSCs (∼80,000) were seeded in 24-well plates pre-coated with 0.5% (w/v) gelatin and cultured to ∼80% confluency, serum-starved for 24 h prior to treatment with E- or EP-sEVs (50 μg/ml) or PBS for 48 h (n = 3). Cell surface proteins were biotinylated using Pierce™ Cell Surface Biotinylation (A44390, Thermo Fisher Scientific) as described (Rai et al., 2021c). Briefly, cells were washed twice with PBS and labeled with 0.25 mg/ml EZ-Link Sulfo-NHS-SS-Biotin for 10 min at RT. Cells were washed twice with ice-cold Tris-buffered saline and lysed (30 min on ice) as per manufacturer’s instructions. Biotin-labelled proteins were captured onto NeutrAvidin Agarose slurry (30 min at RT with end-over-end mixing on a rotor). Samples were loaded onto epTIPS (Eppendorf, 200 μl) fitted with 20 μm nylon net (NY2004700, Merck Millipore), washed 3× as per manufacturers instruction, and reduced in 10 mM DTT in 100 mM TEAB for 45 min at 25°C. Eluted proteins were alkylated, quenched and trypsin digested as described, with resultant peptides acidified (pH ∼2.0) to a final concentration of 2% FA, peptides desalted using SDB-RPS Stage-Tips (Rai et al., 2021a) followed by elution with 30%–80% acetonitrile, 0.1% TFA, and dried by vacuum centrifugation. Peptides were reconstituted in 0.07% TFA and quantified by Fluorometric Peptide Assay. Peptides were analysed on a Dionex UltiMate NCS-3500RS nanoUHPLC coupled to a Q-Exactive HF-X hybrid quadrupole-Orbitrap mass spectrometer equipped with nanospray ion source in positive data-dependent acquisition mode over 95 min gradient, as described (Rai et al., 2021c).

Raw mass spectrometric data were processed in MaxQuant (v1.6.14.0) with its built-in search engine Andromeda (Cox and Mann, 2008) to perform peptide identification and quantification as described (Kompa et al., 2021; Lozano et al., 2022). Database search was with the Andromeda search engine against the Human-only canonical sequence database (2020) with a contaminants database employed. Cysteine carbamidomethylation was set as a fixed modification and acetyl (Protein N-term) and methionine oxidations as variable modifications. Differential search parameters included: TMT tags on peptide N terminus/lysine and carbamidomethylation of cysteine set as a fixed modification; for phosphoproteome analysis phosphorylation of serine, threonine and tyrosine were set as variable modifications; for biotin surface proteome analysis additional Thioacyl (DSP) {[C (3)H (4)OS]} (Rai et al., 2021c) was employed. Enzyme specificity was set as C-terminal to arginine and lysine using trypsin protease, and a maximum of two missed cleavages allowed. Match between runs options were enabled, with the matching time window set to 1 min, with label-free protein quantitation (LFQ) performed where applicable. False discovery rate (FDR) at the PSM, protein, and site levels were each 1%. Peptides were identified with an initial precursor mass deviation of up to 7 ppm and a fragment mass deviation of 20 ppm. A maximum of two missed cleavages were allowed. Protein group or phosphorylation site tables were imported into Perseus (v1.6.7) for analysis, with contaminants and reverse peptides removed.

Data analysis was performed using Perseus of the MaxQuant computational platform (Tyanova et al., 2016), phosphomatics (Leeming et al., 2021), Kinase Enrichment Analysis 3 (KEA3) (Kuleshov et al., 2021) and R programming language (Notaras et al., 2021a; Notaras et al., 2021b). Protein intensities were log2 transformed and subjected to principal component analysis (PCA) with missing values imputed from normal distribution (width 0.3, downshift 1.8) and Student’s t-test. Phosphosites quantified in at least one sample were log2 transformed, median centred, and scaled within each treatment/condition. Missing values were then imputed for each phosphosite based on its pattern of missingness in the EP-sEV or untreated groups. Specifically, phosphopeptides were quantified in at least two out of three biological replicates in each group. Hierarchical clustering was performed using Euclidian distance and average linkage clustering. The volcano plot of student’s t-test p-value versus log2 fold change was generated. g: Profiler and Reactome databases were utilized for functional enrichment and network/pathway analysis, significance p < .05 as described (Rai et al., 2021c; Notaras et al., 2022). To further annotate function, phosphoproteome profiles were analysed with the PhosphoSitePlus database (Hornbeck et al., 2019). Surfaceome data analyzed includes proteins previously experimentally verified as a cell-surface (CSPA) (Bausch-Fluck et al., 2015) or predicted as surfaceome proteins based on SURFY (Bausch-Fluck et al., 2018). Cytoscape (v3.8.1) (Shannon et al., 2003) was used to generate STRING based protein–protein interaction network analysis. Bar plots were generated using GraphPad Prism (v9.0.0). Statistical analyses were performed using Perseus and GraphPad Prism, with unpaired two-sample Student’s T. test or one-way ANOVA performed (statistical significance defined at p < .05).