The cells used were macrophages differentiated from THP-1 cells treated with phorbol myristate acetate (PMA, Sigma-Aldrich, USA) for 24 h or from peripheral blood mononuclear cells (PBMCs) treated with GM-CSF (Sargramostim, 50 ng/mL) for 7 days and human umbilical vein endothelial cells (HUVECs). All culture conditions are described in Supplementary Methods .

Cell culture media issued from the different conditions of stimulation were collected and centrifuged at 300g for 10 min, to eliminate cellular debris. Supernatants were centrifuged at 4,500g for 10 min at 4°C to remove apoptotic bodies, then at 70,000g for 90 min to pellet MVs. MVs were washed with phosphate buffered saline (PBS), resuspended in PBS, and stored at −80°C until being used. MVs were analyzed using a Gallios flow cytometer (Beckman Coulter, USA). Protocol settings and gates as forward scatter by side scatter were adjusted by using 0.5-, 0.9-, and 3-µm Megamix beads (Biocytex, France), as previously described (27). Isolated MVs were stained with 250 µg/mL FITC-labeled annexin A5 Kit (Tau Technologies, the Netherlands) for 30 min at room temperature in the dark, before adding 500 µL of annexin-V-binding buffer and flow-counting beads (Biocytex, France) (28).

To detect the presence of IL-1β, MVs were permeabilized in PBS containing 0.2% saponin, which, at this low concentration, guaranteed preservation of the integrity of the MV phospholipid bilayer, and then incubated at 4°C for 1 h with 25 µg/mL Alexa Fluor 647 anti-human IL-1β antibody (BioLegend, clone JK-1B-1, Cat# 508207, RRID: AB_604133, USA) or 25 µg/mL Alexa Fluor 647 mouse IgG1 control isotype (BioLegend Cat# 400155, RRID: AB_2832978). After washing, permeabilized MVs were stained with annexin-V-FITC to separate MVs from background in flow cytometry analysis. The same permeabilization procedure was used for double labeling of IL-1β and P2X7R, with the further use of an anti-P2X7R rabbit monoclonal antibody (Cell Signaling Technology Cat# 13809, RRID: AB_2798319) or a control rabbit IgG (Cell Signaling Technology Cat# 2729, RRID: AB_1031062), followed by addition of Alexa 488-conjugated goat anti-rabbit IgG (Thermo Fisher Scientific Cat# A78953, RRID: AB_2925776).

Cells and MVs were lysed in buffer containing 20 mM tris, pH 7.5,150 µM NaCl, 2 mM EDTA pH 8, 0.1% NP40, and a cocktail of protease inhibitors (Thermo Fisher Scientific, USA). Protein concentrations were measured using the BCA protein assay kit (Thermo Fisher Scientific, USA). For Western blot analysis, equal amounts of proteins (15 to 30 µg) were separated on a 4%–12% gradient Novex gel (Thermo Fisher Scientific, Waltham, United States), and then transferred on nitrocellulose membranes (GE Healthcare, USA). Blots were blocked with TBST (150 mM NaCl, 50 mM Tris-HCl at pH 7.5, and 0.1% Tween 20) containing 5% BSA for 1 h, then probed with the following primary antibodies: mouse anti-human IL-1β 1:1,000 (3A6, Cell Signaling Technology Cat# 12242, RRID : AB_2715503), rabbit anti-caspase-1 p10 1:200 (Santa Cruz Biotechnology Cat# sc-515, RRID : AB_630975), rabbit anti-human caspase-1 1:200 (Santa Cruz Biotechnology Cat# sc-622, RRID : AB_2069053), mouse anti-human NLRP3 1:1,000 (Enzo Life Sciences Cat# ALX-804-819-C100, RRID : AB_2051972), rabbit anti-human P2X7 1:200 (Sigma-Aldrich Cat# P9122, RRID : AB_477418), rabbit anti-human tissue factor (TF) 1:1,000 (Abcam Cat# ab151748, RRID : AB_2814773), rabbit anti-human ICAM-1 1:1,000 (Cell Signaling Technology Cat# 4915, RRID : AB_2280018), rabbit anti-E-selectin 1:200 (Santa Cruz Biotechnology Cat# sc-14011, RRID : AB_2186684), and rabbit anti-human VCAM-1 1:200 (Santa Cruz Biotechnology Cat# sc-8304, RRID : AB_2214058). Membranes were then incubated with horseradish peroxidase-conjugated secondary antibodies (Thermo Fisher Scientific, USA), using ECL or ECL Plus substrates (Thermo Fisher Scientific, USA) and chemiluminescence detection system (G:Box, Synoptics, UK).

After the lysis of MVs by three freeze–thawing cycles, IL-1β concentrations were measured using a highly sensitive ELISA assay (IL-1 beta Human ELISA Kit, High Sensitivity, Life Technologies, Cat # BMS224HS USA). The same ELISA kit was used for measuring IL-1β concentration in blood samples.

TF activity was measured using a Factor Xa generation assay, as previously described (29). MVs were resuspended in HEPES (Sigma, St Louis, United States), and then incubated in a 96-well plate with anti-TF monoclonal antibody (clone 183 SBTF-1, BioCytex, Marseille, France) or control isotype (clone a-DNP 2H11–2H12, BioCytex, Marseille, France) antibodies for 30 min. A reaction mix containing FVII, Factor X, and CaCl₂ (Sigma, St Louis, United States) was then added to each well to trigger the coagulation cascade and generate FXa. After 2 h of incubation, FXa activity was monitored by the addition of a fluorogenic substrate (factor Xa substrate, Stago 390/450nm, Biocytex, Marseille) (30) and fluorescence was measured, using a fluorimeter (Fluoroskan, Thermo Scientific, United States).

After thawing, MVs were centrifuged at 70,000g for 90 min at 4°C, then fixed in a PBS solution containing 2% paraformaldehyde and 0.2% glutaraldehyde. A 3-µL aliquot was applied on 100-mesh nickel formvar-coated grids (Euromedex, France). After 15 min, MVs were permeabilized with 3 µL of PBS–5% BSA–0.1% saponin for 1 h, then incubated for 1 h with anti-TF rabbit monoclonal antibody (Sigma-Aldrich clone 2L10 ZooMAb^®, Cat# ZRB1811, RRID : AB_2938655) and anti-IL-1β mouse antibody (Cell Signaling Technology Cat# 12242, RRID : AB_2715503) diluted 1:25 in PBS–0.3% BSA. After three washes in PBS, MVs were labeled for 1 h with goat anti-rabbit IgG coupled with 10-nm gold beads (Thermo Fisher Scientific Cat# N-24916, RRID : AB_2539796) and goat anti-mouse IgG coupled with 5-nm gold beads (Tebu-Bio, Cat# 220GA1003, RRID : AB_2938656). Immunogold-labeled MVs were fixed for 5 min in cacodylate buffer containing 2.5% glutaraldehyde, then grids were negatively stained with 0.3% phosphotungstic acid for 1 min, before observation with a transmission electron microscope (JEM 1400 JEOL, United States).

MVs isolated from macrophages were incubated with HUVECs for 4 h at 37°C. Cell pellets were labeled for 1 h with specific anti-ICAM-1/CD54-PE (Beckman Coulter Cat# IM1239U, RRID : AB_131186), anti-VCAM-1/CD106-PE (Beckman, A66085, RRID : AB_2938658), anti-E-selectin/CD62E-PE (BioLegend, HCD62E RRID : AB_2938659), or control isotype antibodies IgG1-PE (Beckman Coulter Cat# A07796, RRID : AB_2832963) and IgG2aκ-PE (BioLegend Cat# 400213, RRID : AB_2800438) (5 µL/reaction) for flow cytometry analysis. For Western blot experiments, cells were washed and scraped in PBS, centrifuged, and lysed in RIPA buffer.

Mice experiments were conducted according to the institutional ethical committees for animal care from the local animal ethics committee of Marseille. C57BL/6J mice received intraperitoneal injections of PBS or IL-1 receptor antagonist IL-1Ra (30 mg/kg), with or without subsequent one peritoneal injection of 25×10⁶ MVs in PBS. Peripheral blood and peritoneal fluids were collected. The details of experimental conditions are described in Supplementary Methods .

Ten patients with JIA according to the International League of Associations for Rheumatology (ILAR) definition were recruited from the Department of Paediatrics at Marseille and Nimes hospitals, during 2015–2016. Parents (or patient’s legal guardians) and patients received information and consent of parents (or patient’s legal guardians) was obtained for all individual participants included in the study. Clinical presentations consisted mainly in refractory polyarticular and systemic forms of the disease, but all patients presented active disease defined as AJI flare according to Juvenile Arthritis Disease Activity Score 10 (JADAS10). The main demographic, clinical, and biological characteristics are shown in Table 1 . Ten healthy adult donors were also included.

Sampling was justified by the flare of the disease. A fasting blood sample was obtained by venipuncture in citrate-containing tubes, which were centrifuged twice at 2,500g for 15 min at room temperature in order to obtain platelet-poor plasma (PPP), within a maximum of 2 h after blood collection, as previously described (31). The resulting plasma was divided into aliquots and stored at −80°C until analysis.

PPP aliquots were thawed at room temperature and centrifuged at 20,000g for 90 min to pellet MVs. Resuspended MVs (30 µL) were transferred into different tubes for MV labeling. Annexin V-FITC (Tau Technologies, the Netherlands) and conjugated antibodies (Ab) were added to identify specific MV subsets: CD31-Phycoerythrin (PE) Ab (Beckman Coulter Cat# IM2409, RRID : AB_131205), CD41-PC7 Ab (Beckman Coulter Cat# 6607115, RRID : AB_2800448), CD11b-APC Ab (Beckman Coulter Cat#A87782, RRID : AB_2938661), CD235a-Alexa 750 Ab (Beckman Coulter Cat# A89314, RRID : AB_2938662), CD14-APC Ab (Beckman Coulter Cat# IM2580, RRID : AB_2800451), CD66b-APC Ab (Beckman Coulter Cat# B08756, RRID : AB_2893284), and CD11b-PE Ab (Beckman Coulter Cat# IM2581U, RRID : AB_131334). The numeration of total MVs and MVs derived from specific cell subsets (absolute number/mL plasma) were calculated according to the acquired counts of MVs and microbeads and the dilution performed during sample preparation.

Data represent the cumulative results of all experiments and are expressed as median with interquartile range for the indicated number (n) of experiments. Statistical analysis was performed with Prism version 9 software. Significant differences were determined by using nonparametric Mann–Whitney and Wilcoxon test; p-values < 0.05 were considered statistically significant.

According to the Minimal Information for Studies of Extracellular Vesicles (MISEV) 2018 guidelines (32), MVs issued from THP-1 differentiated into macrophages by PMA treatment were then purified by successive centrifugations and ultracentrifugations ( Figure 1A ). MVs from unstimulated and stimulated macrophages were characterized using several methods. First, using flow cytometry, the classical gating strategy was determined ( Supplementary Figure S1 ). Then, the presence of typical membranous and cytosolic MV-associated molecules was confirmed by detecting CD18 (integrin subunit β2) and Annexin-V binding on purified MVs ( Figure 1B ). In addition, the presence of integrin subunit β3 and CD81 was confirmed by Western blotting ( Figure 1C , upper panel), with no major contamination by soluble proteins such as albumin ( Figure 1C , lower panel). Then, tunable resistive pulse sensing assays showed that vesicles had similar average sizes (respectively 299, 215, and 205 nm for conditions CTRL, LPS, and LPS+ATP), which correspond to the range of large MVs ( Figure 1D ). Taken together, molecular and size features of MV preparations were consistent with the definition of MV according to the MISEV 2018.

To determine the biological activity of IL-1β conveyed by MVs, we used MVs purified from supernatants of THP-1 cells differentiated into macrophages by PMA treatment and stimulated with LPS or LPS+ATP. MV numeration showed that the same number of MVs were issued from LPS-stimulated or non-stimulated macrophages (ns, Figure 1E ). Further addition of ATP tended to increase the total number of MVs compared to non-stimulated macrophages, despite heterogeneous results (58% increase, p < 0.05, Figure 1E ). Similar results were obtained using MVs from PBMC differentiated into macrophages (p < 0.005, Supplementary Figure S2 ).

We then studied NLRP3 inflammasome components carried by MVs from THP-1 treated with PMA and differentiated in macrophages. Using ELISA, we observed that IL-1β concentration was significantly higher in MV lysates issued from LPS- and LPS+ATP-stimulated macrophages compared to MV lysates from non-stimulated macrophages (p < 0.05, Figure 2A ). We used flow cytometry to detect IL-1β in MVs after membrane permeabilization by saponin, as previously reported (33) ( Figure 2B ) and observed a significant increase of IL-1β-positive MV population shed from macrophages stimulated with LPS+ATP as compared to non-stimulated ones [8.1 (2–13.9) vs. 0.115 (0.03–0.475)%, p < 0.01, Figure 2C ].

Western blot experiments were performed only with MVs purified from the supernatant of PMA-treated THP-1 cells, as large amounts of MVs were required for protein expression analysis. MVs from non-stimulated macrophages contained small concentrations of the IL-1β precursor (33 kDa), likely resulting from THP-1 cell activation by PMA. After LPS stimulation, increased levels of IL-1β precursor (pro-IL-1β) and very low amounts of mature IL-1β (17 kDa) as well as intermediary IL-1β (28 kDa) form were observed, suggesting that priming with LPS induced pro-IL-1β synthesis by macrophages. With the addition of exogenous ATP, IL-1β mature form increased in MVs, consistent with ATP-induced pro-IL-1β maturation ( Figure 2D ).

We then investigated the presence of NLRP3 and caspase-1 in MVs from THP-1 treated with PMA, differentiated in macrophages. Using Western blot, NLRP3 ( Figure 2E ) and the 40-kDa procaspase-1 and the 10-kDa mature and 20-kDa mature caspase-1 forms ( Figure 2F ) were clearly detected in MVs secreted by stimulated macrophages, indicating that MVs issued from activated macrophages contained all the machinery necessary for the processing of IL-1β. In addition, when PMA-treated THP-1 cells were pre-treated by caspase-1-inhibitor YVAD then stimulated with LPS+ATP, MV content was modified showing less IL-1β mature form and increased concentrations of pro-IL-1β ( Figure 2G ), demonstrating that IL-1β maturation in MVs was caspase-1-dependent.

Next, we studied the pro-inflammatory and pro-coagulant properties of IL-1β-positive MVs. Since MVs contain all the molecules necessary to generate the 17-kDa mature IL-1β, we asked whether they could process and release IL-1β in vitro. First, using flow cytometry analysis and gating on large permeabilized Annexin-V-positive MVs, in order to reduce exosome contamination, we observed co-expression of IL-1β and P2X7 receptor in and on the same MV. More than 80% of MVs issued from non-stimulated macrophages expressed only P2X7R, whereas permeabilized MVs issued from LPS+ATP-stimulated macrophages expressed both P2X7R and IL-1β ( Figure 3A ). Western blot analysis confirmed the presence of P2X7R in MV lysates ( Figure 3B ). MVs were purified from LPS-stimulated macrophages before being incubated with PBS or ATP. After ultracentrifugation to eliminate MVs, IL-1β concentration was measured in MV-free supernatant and a threefold increase of IL-1β concentrations was observed in the supernatant of MVs stimulated with ATP, compared to the supernatant of MVs stimulated with PBS alone (p < 0.05, Figure 3C ).

MVs issued from macrophages have been reported to carry TF (34); we thus asked whether MVs contained both IL-1β and TF. MVs were isolated from supernatants of macrophages derived from THP-1 and MV lysates were subjected to Western blot analysis. Both IL-1β and TF could be detected in the same MV issued from LPS+ATP-stimulated macrophages ( Figure 3D left and right panels, respectively). Moreover, using transmission electronic microscopy (TEM) on MV double-labeled for both IL-1β and TF, we observed that some MVs contained both IL-1β and TF ( Figure 3E ). To confirm TF procoagulant activity on MVs, an in vitro Factor-Xa-generation-assay was performed. TF activity was significantly increased with MVs issued from LPS+ATP-stimulated macrophages, when compared with MVs issued from LPS-stimulated macrophages and non-stimulated macrophages (p < 0.05 respectively, Figure 3F ).

To determine whether IL-1β-positive MVs were biologically active, we first studied the ability of MVs issued from LPS+ATP-stimulated macrophages to activate endothelial cells in vitro. Using Western blotting, we observed that MVs significantly increased ICAM-1 and VCAM-1 expression (p < 0.01 and p < 0.001, respectively, Figure 4A , upper panel a and middle panel b) but not E-selectin expression on HUVECs (ns, Figure 4A , lower panel c). Interestingly, MV pre-incubation with IL-1Ra did not reduce either ICAM-1 or VCAM-1 expression on endothelial cells (ns, Figure 4A ). Endothelial cell activation was also confirmed by cytometry analysis on HUVECs using the same conditions ( Supplementary Figure S3 ; Supplementary Table 1 ).

Then, we studied the ability of MVs issued from LPS+ATP-stimulated macrophages to induce leukocyte recruitment in mice. Fifteen hours after intra-peritoneal injection of MVs, leukocyte recruitment (CD45+ cells) into peritoneal cavity was significantly increased (p < 0.01, Figure 4B ), a process that was significantly reduced in the presence of IL-1 receptor antagonist (p < 0.05, Figure 4B ).

To support the existence of these MVs in human disease, we studied the presence of circulating MVs in the plasma of patients with JIA or from healthy donors. Ten patients (sex ratio = 1) with a mean age of 11.1 years [5–16 years], were studied ( Table 1 ). All patients had active disease according to JADAS10, which motivate a therapeutic increment. Among them, nine patients presented a polyarticular form including three with systemic clinical and biological symptoms consistent with a systemic form (sJIA: patients #4, #9, and #10). No patient was treated with anakinra at the time of the flare. The absolute number of total MVs and of leukocyte-derived MVs were not significantly different in plasma of JIA patients compared to healthy controls, possibly due to the relatively low number of patients (data not shown). However, if the concentration of CD11b⁺MVs derived from myeloid cells was not significantly higher in JIA patients compared to healthy controls (ns, Figure 5A ), CD45⁺CD66^-MVs issued from myeloid cells after exclusion of neutrophils were significantly increased in JIA patients (p < 0.05, Figure 5B ). Importantly, using flow cytometry analysis, we could detect higher concentrations of circulating IL-1β+Annexin-V-positive MVs in JIA patients compared to healthy controls (p < 0.05, Figure 5C ). Circulating IL-1β+Annexin-V-positive MVs represented a small proportion of total MVs ( Figure 5D ) but were detectable in all patients, especially in those with an active polyarticular disease (patient#1: 0.47%, patient#2: 0.31%) and with biological systemic inflammation (patient#9: 0.27%, and patient #10: 1.06% respectively).