Внеклеточный маркер изменяется в связи с болезнью при повторном выпуске нескольких склерозов.

Introduction

Multiple sclerosis (MS), neuromyelitis optica spectrum disorder (NMOSD), and myelin oligodendrocyte glycoprotein antibody–associated disease (MOGAD) are chronic autoimmune demyelinating diseases of the CNS. They present overlapping motor, sensory, and cognitive symptoms, such as vision and walking impairment, respectively, caused by optic nerve and spinal cord lesions. 1 In contrast to MS, where no specific autoantibody profile has been established, current diagnosis of NMOSD and MOGAD heavily relies on serum autoantibodies. 2,3 Although AQP4-IgG and MOG-IgG greatly facilitate diagnosis, their blood concentrations do not correlate with disease activity or treatment effects. By contrast, the serum levels of neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP), reflecting neuronal damage and MS progression, respectively, have been linked to disease activity. 4,5 However, these molecules are not disease-specific. 6,7 Therefore, more sensitive and specific tools are required for improving diagnosis and follow-up of disease course.

In recent years, cell-derived extracellular vesicles (EVs) have emerged as a promising source of disease biomarkers in blood and CSF. 8,9 EVs are small bilayered membranous particles ranging between approximately 50 and 1,000 nm in diameter. By delivering proteins, lipids, and mRNA, EVs contribute to the intercellular communication. 10 In the immune system, cytokines, chemokines, coagulation factors, and antigens enclosed in T-cell–shed EVs are important mediators between the innate and adaptive immune response. 11,12 Notably, high levels of T-cell–derived EVs in serum and CSF are associated with MS relapse. 13,14 Whether immune cells release EVs in different amounts and of different compositions in MS, NMOSD, and MOGAD is not known.

Moreover, CNS cells also secrete EVs in MS; myelin-specific proteins, such as myelin basic protein (MBP) and MOG, have been detected in EVs isolated from the plasma and CSF of patients with MS, suggesting that oligodendrocyte-shed EVs cross the blood-brain barrier (BBB) and reach the circulation. 15 In addition, the level of MBP-positive and MOG-positive EVs correlated with disability measured using Expanded Disability Status Scale (EDSS) scores. 16,17 Very little is known about the protein cargoes of EVs in NMOSD and MOGAD. In NMOSD relapse, astrocytes release specific microRNAs (miRNAs) encapsulated in EVs that are detectable in the blood. 18,19 Importantly, the authors showed that some of these miRNAs induced demyelinating lesions in vitro, ex vivo, and in vivo. 19 At the protein level, it has been suggested that GFAP and complement factor C4b-binding protein α-chain may be enriched in CSF-derived EVs in patients with NMOSD compared with MS, but further investigations are necessary to confirm these observations. 20 Comparative analyses in MS vs NMOSD vs MOGAD are missing to determine whether EV proteins differ in a disease-specific fashion.

The aim of this study was to characterize circulating EVs in patients with MS, NMOSD, and MOGAD in parallel. It is important to underline that EVs are heterogeneous particles, of variable sizes and cellular origins, resulting from different biogenesis mechanisms (cell membrane budding and fusion of multivesicular endosomes). Therefore, throughout this study, the nonrestrictive term “EVs” encompassing all categories of vesicles was preferred, as recommended by the International Society of Extracellular Vesicles and the MISEV2023 Guidelines. 21 In the membrane of EVs, CD9, CD63, and CD81 tetraspanins are considered as specific markers. 22 Nevertheless, the level of tetraspanins varies between cell types, tissues, and diseases. 23 For this reason, tetraspanin-dependent and independent methods were used to examine EVs. An extended analysis of 37 EV epitopes revealed specific enrichments in MS samples by flow cytometry.

Methods

Ethics, Participants, and Plasma Collection

A total of 96 participants were included in this study. Patients diagnosed with relapsing-remitting MS (RRMS) (n = 52), AQP4-positive NMOSD (n = 19), and MOG-positive MOGAD (n = 10) and age-matched and sex-matched healthy controls ([HCs], n = 15) from the Eginition University Hospital were included in this study. The study was approved by the ethical committee of the Eginition University Hospital (#501/30.7.2019). All patients signed an informed consent form for blood sample use for research purposes. Blood was collected into EDTA tubes and centrifuged within 1–5 hours after blood draw, at 3,100 × g, for 7 minutes at room temperature (RT). Plasma was then aliquoted and stored at −80°C until use.

Overview of the Workflow for Plasma Analyses

Circulating NfL and GFAP were assayed in plasma samples using single-molecule array (SIMOA) ( Figure 1A ). We then used orthogonal methods to analyze plasma EVs, in accordance with current recommendations (MISEV2023), and evaluated EV characteristics with different methods. We assessed EV size, density, and their subcategories based on tetraspanin expression with a single-particle interferometric resonance imaging sensor (SP-IRIS) using the novel ExoView R100 system (Unchained Labs, Pleasanton, CA) ( Figure 1B ). EVs isolated with a size-exclusion chromatography-based technique were characterized by cryo-electron microscopy (cryo-EM), Western blotting, and nanoparticle-tracking analysis (NTA) ( Figure 1C ). The multiplex antibody bead–based flow cytometry assay allowed simultaneous analysis of 37 cell-specific EV surface epitopes using the MACSPlex EV kit (Miltenyi Biotec, Bergisch Gladbach, Germany) (Figure 1C).

Schematic Illustration of Plasma Analyses

(A) Ultrasensitive detection of neurofilament light chain (NfL) and glial fibrillary acidic protein (GFAP) by single-molecule array (SIMOA) in plasma samples. (B) EVs were analyzed using the single-particle interferometric resonance imaging sensor (SP-IRIS, ExoView, Unchained Labs) after immunocapture with CD9, CD81, and CD63 antibodies directly in plasma. (C) EV isolation with size-exclusion chromatography (SmartSEC, System Biosciences) followed by the characterization of isolated EVs by Western blotting (WB), nanoparticle-tracking analysis (NTA), and cryo-electron microscopy (cryo-EM). Extended EV-specific marker analysis by bead-based flow cytometry (MACSPlex EV kit, Miltenyi) after size-exclusion chromatography-based EV isolation. The levels of 37 EV surface markers plus 2 isotype controls (MIgG, human IgG1 isotype control antibody [REA293]) were simultaneously monitored. Each type of EV capture bead was labeled with 1 of the 39 antibodies. Bound EVs were detected with a cocktail of APC-labeled CD9, CD81, and CD63 antibodies. Illustrations created in BioRender Falk, V. (2025) BioRender.com/uqfyfpg and Adobe Photoshop (v. 2024). EV = extracellular vesicle.

Ultrasensitive Detection of NfL and GFAP With SIMOA

NfL and GFAP were measured in plasma in duplicates with a SIMOA, using the Neurology 2-plex B assay (Quanterix, Billerica, MA). Because the blood levels of NfL and GFAP are influenced by age and sex, 4 these factors were taken into consideration by calculating adjusted z-scores. 4,5 NfL z-scores were adjusted for age, based on a large reference cohort, 5 whereas preliminary GFAP z-scores were adjusted for age and sex based on a reference database of 4,943 samples from healthy individuals. 4

Single-Particle Interferometric Resonance Imaging Sensor for EV Characterization

Single-particle interferometric resonance imaging sensor (SP-IRIS) with ExoView (Unchained Labs, Pleasanton, CA) enables size and density measurement of EVs exhibiting membrane-bound tetraspanins CD9, CD63, and CD81 in crude plasma, without previous EV isolation. Plasma samples were thawed at RT and centrifuged at 5,000 × g for 20 minutes. Samples were diluted 1:50 in ExoView incubation solution II, to obtain an optimal number of around 3,000 counts per spot, and 50 µL was applied to prescanned chips overnight at RT. EVs were captured by anti-CD9, anti-CD63, and anti-CD81 antibodies as well as an isotype control (anti-MIgG). Vesicle surface immunofluorescence was performed according to the manufacturer's protocol, using antibodies against CD9, CD63, and CD81 provided by the manufacturer. Incubated chips were washed with the CW100 Chip Washer. Antibodies were diluted 1:500 in the provided blocking solution and incubated on the chips for 1 hour. Chips were then washed and scanned using the ExoView® Automated Imager. Results were analyzed with the ExoView Analyzer. Cutoffs were set to obtain the average count of <50 counts for CD63 and CD81 (red and green channel), with an average total detected of 5%, and an average count of <100 counts for CD9 (blue channel) with an average total detected of 1%. Sizes were measured by interferometric microscopy (IM) on every EV captured using the tetraspanins. Captured EVs giving off a fluorescent signal were counted for calculation of particle concentration, and the fluorescence intensity was measured per particle. Size distribution data were preprocessed in Python 3.9.7 using the numpy (version 1.21.5) and pandas (version 1.4.4) packages.

Size-Exclusion Chromatography-Based EV Isolation

Plasma samples were thawed at RT, and EVs were isolated through size-exclusion chromatography using the SmartSEC kit (System Biosciences, Palo Alto, CA) according to the manufacturer's protocol. In brief, plasma samples were centrifuged twice to eliminate larger vesicles, and 250 µL of plasma was added onto the washed column. To ensure proper elution, 250 µL of 0.1-µm filtered phosphate-buffered saline (PBS) was added to the column and it was incubated for 30 minutes while rolling. Columns were centrifuged at 500 × g for 30 seconds to elute EVs. Protease and phosphatase inhibitors were immediately added to the EV elution. Intact EVs were stored at 4°C for storage up to 1 week, or at −20°C for longer storage. For Western blotting, EVs were directly lysed in 0.5% CHAPS (PanReac AppliChem, Darmstadt, Germany) and stored at −20°C until use.

Western Blotting

Western blotting was used to detect a common EV marker for characterization. Protein concentration of EV solutions was determined with a detergent compatible protein assay (BioRad, Hercules, CA) of lysed EVs. 10 µg of EVs was loaded onto 4%–12% Bis-Tris NuPAGE (Invitrogen, Waltham, MA) gels and run at 120 V for 80 minutes. Resolved proteins were transferred to a nitrocellulose membrane in a transfer system at 400 mA for approximately 70 minutes. Membranes were then washed with tris-buffered saline 0.2% Tween (TBS-T) 3 times for 10 minutes before blocking in 5% bovine serum albumin in TBS-T for 1 hour at RT. Thereafter, membranes were incubated with a primary antibody against Alix (1:500, NBP1-90201, Novus Biologicals, Centennial, CO) in a rolling shaker overnight at 4°C. Because tetraspanins were measured with both SP-IRIS and flow cytometry, Alix, an intravesicular marker, was used for Western blotting. Membranes were then washed and incubated with a secondary antibody conjugated with streptavidin-HRP for 1 hour at RT before detection with WesternSure Chemiluminescent Substrate (LI-COR, Lincoln, NE) on a C-DiGit Blot Scanner (LI-COR).

Cryo-EM

For cryo-EM, 4 µL of isolated EVs was applied to a glow-discharged Lacey grid (30 seconds 10 mA Baltzers CTA 010). Vitrification was performed using a FEI Vitrobot Mark IV. The samples were applied to the grids at 4°C for 10–30 seconds at 100% humidity followed by 4 seconds of blotting. Vitrified grids were stored in liquid nitrogen before acquisition. Imaging was performed on a FEI Tecnai F20 equipped with a Falcon III camera with a total dose not exceeding 60 e−/Å2.

NTA

Fresh EVs were examined by NTA. The concentration and size of EVs were determined using the NanoSight NS300 and the NanoSight 3.2 software programs (Malvern Panalytical, Malvern, United Kingdom). EVs were diluted 1:100–300, to reach a concentration of 10–100 particles per frame, in 0.1- µm filtered PBS. The camera level was set to 12, and a 488-nm laser was used. Each sample went through 5 repeated measures of 1 minute.

Multiplex Analysis of EV Epitopes by Flow Cytometry

Isolated EVs were subjected to flow cytometry using the MACSPlex EV Kit IO (Miltenyi Biotec, Bergisch Gladbach, Germany), enabling the detection of 39 EV surface immune cell–related epitopes (including 2 isotype controls). The kit consists of a cocktail of 39 different beads, each coated with antibodies specific to EV epitopes. The bead populations can be distinguished by the different fluorescent intensities of the fluorophores phycoerythrin (PE) and fluorescein isothiocyanate (FITC). EVs bound to these antibody-coated beads were detected with an additional cocktail comprising allophycocyanin (APC)-labeled anti-CD9, anti-CD63, and anti-CD81, forming a sandwich complex. Positive signals indicate the presence of the respective surface epitope, colocalized by the typical EV tetraspanins. Detection of EV surface markers was performed in accordance with the manufacturer's protocol. In these experiments, 20 µg of EVs was incubated with the beads overnight at RT in an orbital shaker (450 rpm). Subsequently, samples were centrifuged at 3,000 × g for 5 minutes, and the supernatant was carefully removed. A cocktail of APC-labeled anti-CD9, anti-CD63, and anti-CD81 antibodies (0.5-µg antibody/reaction) was mixed with the pellet, and it was incubated for 1 hour at RT under agitation (450 rpm). Furthermore, samples were washed 2 times with the MACSPlex Buffer before transfer to 5-mL round bottom tubes. Captured EVs were measured with the BD LSRFortessa (BD Biosciences, Franklin Lakes, NJ). Results were analyzed in FlowJo (v. 10.10.0) to obtain median signal intensities for all EV surface markers. Semiquantitative comparisons were allowed by subtracting the median signal intensities of the blank sample (buffer only) from the median signal intensity measured in each sample. Then, the mean of the measured median APC signal intensities of the bead populations coated with CD9, CD63, and CD81 antibodies was used as a normalizing factor for the rest of the markers.

Statistical Analysis

Raw data were assembled in Microsoft Excel 365. Statistical analyses were performed in GraphPad Prism 9. Normally distributed data were analyzed with 1-way analysis of variance (ANOVA) and Pearson correlation while the Kruskal-Wallis test and Spearman correlation were applied for non-normally distributed data. For disease comparisons, 1-way ANOVA tests were followed by the Tukey post hoc test, and a p value < 0.05 was considered as significant. For multiple comparisons of 39 EV markers by flow cytometry, a p value < 0.001 was considered as significant, using Bonferroni correction for multiple testing.

Data Availability

Anonymized patient data not published within this article will be made available by request from any qualified investigator.

Results

Patient Demographics

This study included patients diagnosed with RRMS (n = 52), AQP4-IgG seropositive NMOSD (n = 19), and MOG-IgG seropositive MOGAD (n = 10) and healthy donors (HC, n = 15) ( Table ). Patients with NMOSD were significantly older than those with RRMS and consisted of women only. In comparison, the RRMS and MOGAD groups comprised approximately two-thirds of women. The age at disease onset is in general later in the NMOSD group than in MS and MOGAD groups, and the predominance in women is stronger in the NMOSD group than in MS and MOGAD groups. 24 Thus, the heterogeneity of the cohort is inherent to the diseases. Patients with NMOSD also exhibited more severe symptoms than patients with RRMS and MOGAD, as determined using the EDSS. At the time of blood/plasma collection, only a few patients experienced a clinical relapse. Most of the patients with RRMS was treated with natalizumab (NTZ), whereas patients with NMOSD received rituximab (RTX) and patients with MOGAD were given heterogeneous treatments ( Table ). The time elapsed since initiation of current treatment did not differ significantly between patient groups (data not shown). Potential treatment effects on the composition of EVs were evaluated.

Table Patient Demographics RRMS (n = 52) NMOSD (n = 19) MOGAD (n = 10) HC (n = 15) p Value Mean age (range) 38.7 (19.0–66.0) 56.3 (35.0–72.0) 47.9 (24.0–76.0) 37.4 (7.4) A: <0.0001 B: 0.1469 C: 0.4360 D: 0.9775 E: 0.0004 F: 0.1593 Women/men (% women) 37/18 (67) 19/0 (100) 7/3 (70) 9/6 (60) N/A Median EDSS score (range) 1.5 (0–4.0) 3.5 (1.0–7.0) 1.0 (0–4.5) N/A A: 0.0004 B: 0.2383 C: 0.0018 Current relapse (%) 4 (0.7) 4 (21) 1 (10) N/A N/A Treatment (number) NTZ: 33 RTX: 12 RTX: 2 N/A N/A DMF: 10 AZA: 2 AZA: 1 GA: 2 MP: 3 MP: 3 None: 7 PLEX: 1 MMF: 3 None: 1 None: 1 Abbreviations: AZA = azathioprine; DMF = dimethylfumarate; EDSS = Expanded Disability Status Scale; GA = glatiramer acetate; HC = healthy control; MMF = mycophenolate mofetil; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; MP = methylprednisolone; N/A = not applicable; NMOSD = neuromyelitis optica spectrum disorder; NTZ = natalizumab; PLEX = plasma exchange; RRMS = relapsing-remitting multiple sclerosis; RTX = rituximab. A: RRMS vs NMOSD, B: RRMS vs MOGAD, C: NMOSD vs MOGAD, D: RRMS vs HC, E: NMOSD vs HC, F: MOGAD vs HC. Statistics: 1-way ANOVA for age and Kruskal-Wallis test for EDSS scores.

Patient Demographics

Abbreviations: AZA = azathioprine; DMF = dimethylfumarate; EDSS = Expanded Disability Status Scale; GA = glatiramer acetate; HC = healthy control; MMF = mycophenolate mofetil; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; MP = methylprednisolone; N/A = not applicable; NMOSD = neuromyelitis optica spectrum disorder; NTZ = natalizumab; PLEX = plasma exchange; RRMS = relapsing-remitting multiple sclerosis; RTX = rituximab.

A: RRMS vs NMOSD, B: RRMS vs MOGAD, C: NMOSD vs MOGAD, D: RRMS vs HC, E: NMOSD vs HC, F: MOGAD vs HC. Statistics: 1-way ANOVA for age and Kruskal-Wallis test for EDSS scores.

Detection of Plasma NfL and GFAP by SIMOA

Previous studies showed that the plasma increase in GFAP concentration correlates with the astrocytic injury response in the CNS of patients with NMOSD and MS, but not in those with MOGAD. 4,25 In addition, NfL, whose release in blood results from CNS neuron damage, is considered as a disease activity marker in MS. 5 Therefore, we compared the levels of NfL and GFAP in the plasma of patients with RRMS, NMOSD, and MOGAD ( Table ) by SIMOA ( Figure 2 ). Patients with RRMS had higher levels of NfL and GFAP than HCs ( Figure 2, A and B ). No significant difference was found for either NfL z-score or GFAP z-score between the 3 disease groups ( Figure 2A ). However, patients with NMOSD displayed a significant increase in GFAP compared with HCs ( Figure 2B ).

Circulating NfL and GFAP in Plasma

NfL values were adjusted for age, based on a large reference cohort of healthy individuals to calculate z-scores. GFAP values were adjusted for age and sex to calculate preliminary GFAP z-scores, based on an ongoing collection of a healthy control reference cohort. Dashed lines indicate mean values in HCs (n = 15, i.e., z-score = −0.6317 for NfL and z-score = −0.5643 for GFAP), and significances above indicate whether observed values differ from the mean HC z-score. (A) Patients with RRMS, but not patients with NMOSD or MOGAD, presented significantly higher levels of NfL compared with healthy controls. (B) Only patients with RRMS and NMOSD showed a significantly higher level of GFAP compared with healthy controls. No significant differences were found between disease groups. Statistics: Wilcoxon signed rank test, *** p < 0.001, **** p < 0.0001. GFAP = glial fibrillary acidic protein; HC = healthy control; NMOSD = neuromyelitis optica spectrum disorder; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease.

Tetraspanin-Dependent EV Capture and Characterization

A single-particle interferometric resonance imaging sensor (SP-IRIS) with the ExoView system was used for plasma EV immunocapture and morphometric analyses in 22 RRMS, 19 NMOSD, and 10 MOGAD patient plasma samples without previous EV isolation (eFigure 1). The size of tetraspanin-labeled EVs did not differ significantly between disease groups (eFigure 1, A and B). MIgG provided very low unspecific EV detection in comparison with the tetraspanins (eFigure 2A; p < 0.001 or p < 0.0001). Fluorescence intensity of the CD81 signal for EVs captured with CD63 and CD9 was lower in patients with RRMS than in the 2 other groups, suggesting differential enrichment in this tetraspanin (eFigure 1C). The concentration of EVs was similar between disease groups (eFigure 1D). In summary, SP-IRIS analyses with ExoView did not reveal major disease-specific differences in EV size or concentration.

Isolated EV Characterization by NTA

In a more classical approach, EVs were isolated by size-exclusion chromatography before EV marker detection by Western blotting ( Figure 3A ), cryo-EM observation ( Figure 3B ), and NTA. The relatively low level of albumin found in EV samples in comparison with plasma suggests that most of the plasma proteins are eliminated in the isolation process ( Figure 3A ). Accordingly, enrichment in intravesicular protein Alix was observed in the EV lysate compared with plasma. Cryo-EM images of isolated EVs allowed us to distinguish typical lipid bilayer membranes in small vesicles ( Figure 3B ), thus confirming the preservation of EV integrity and morphology.

Characterization of EVs After Size-Exclusion Chromatography-Based Isolation

EVs were isolated by size-exclusion chromatography (SEC, SmartSEC, System Biosciences). (A) Analysis of the classical EV marker Alix by Western blotting after isolation. Equal amounts of EVs and plasma samples (10 µg) were used. A weak level of albumin was detected in the EV fraction. (B) Freshly isolated EVs were subjected to cryo-electron microscopy (cryo-EM). Cryo-electron microscopy observations showed EVs with typical lipid bilayer membrane. EV = extracellular vesicle.

We then sought to determine the size and concentrations of isolated EVs by NTA. NTA is commonly used to determine the density and size of EVs. 26 Strikingly, NTA measurement revealed lower concentrations of EVs in the RRMS group than in HCs ( Figure 4B ). The mean EV size did not vary between the 3 disease groups (123.1 nm in RRMS; 114.3 nm in NMOSD, and 117.2 nm in MOGAD) but was significantly bigger than that of HCs (96.3 nm) ( Figure 4C ). Detailed analysis of EV size distribution showed a peak at 76.5 nm in patients with RRMS, 77.5 nm in patients with NMOSD, 76.5 nm in patients with MOGAD, and 72.5 nm in HCs, suggesting that small EVs represent the bulk of isolated vesicles ( Figure 4D ), consistent with cryo-EM observations. The distribution of EV sizes normalized to the total number of EVs did not reveal significant variations between disease groups ( Figure 4D ). Potential treatment effects on EV size and density were examined by comparing subgroups of patients with RRMS receiving NTZ (n = 33), dimethylfumarate/glatiramer acetate (n = 12), or no treatment (NT = 7) (eFigure 3). No significant differences were observed. In summary, NTA measurements suggest that RRMS is characterized by smaller plasma EVs independent of the treatments administered while the density of EVs increased in the 3 disease groups compared with HCs.

Changes in EV Density and Size Revealed by NTA

After isolation ( Figure 1 ), the density and size of EV particles were measured by NTA. (A) Representative image of particles detected with NTA. (B) The healthy control (HC) group displayed a significantly higher EV concentration than the RRMS group. (C) The diameters of the EVs were similar for all 3 patient groups with a mean of 110.5 nm in RRMS, 114.3 nm in NMOSD, and 117.2 nm in MOGAD, but significantly smaller in HC (mean = 96.3 nm). (D) The EV size distributions presented a peak at 76.5 nm in RRMS, 77.5 nm in NMOSD, 76.5 nm in MOGAD, and 72.5 nm in HC samples, suggesting that small EVs are predominant in all groups. Consistent with the higher concentration of EVs in HCs shown in (A), the number of EVs was increased in HC samples. Normalized EV size distributions showed very similar size profiles between the groups. Three patients with NMOSD were excluded because of low amount of material. Statistics: 1-way ANOVA test, * p < 0.05, ** p < 0.01. EV = extracellular vesicle; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; NMOSD = neuromyelitis optica spectrum disorder; NTA = nanoparticle-tracking analysis; RRMS = relapsing-remitting MS.

Multiplex Analysis of EV Epitopes by Flow Cytometry

Surface markers were studied on isolated EVs by flow cytometry with the MACSPlex EV kit (Miltenyi Biotec, Figure 1C ). EV subpopulations were isolated with capture beads coated with antibodies directed against 37 distinct EV epitopes ( Figure 5A ). The capture beads were discriminated according to PE and FITC fluorescence intensities ( Figure 5B ), and captured EVs were detected by the APC signal from the CD9, CD81, and CD63 antibodies. The APC intensity of each bead population was quantified. To different extents, the levels of CD29, CD31, CD40, CD44, and CD69, which are T-cell–associated and leukocyte-associated markers, significantly changed in patients with RRMS along with the time elapsed since the last relapse and NfL or GFAP z-scores ( Figure 5C ). CD29, CD31, and CD69 were positively associated with the time elapsed since the last relapse. A negative correlation was established between the levels of CD31, CD40, CD44, CD49c, and CD69 and plasma NfL in the RRMS sample in contrast to NMOSD, MOGAD, and HC samples, suggesting that increased expression of specific EV markers may be associated with low disease activity in RRMS ( Figure 5C ). A similar relationship was observed between CD44 and CD69 and plasma GFAP ( Figure 5C ). Of note, the levels of EV markers did not significantly differ between the whole groups (eFigure 4). In addition, no significant differences were observed between patients with RRMS treated with NTZ (n = 33) and those receiving dimethylfumarate/glatiramer acetate (n = 12) or left untreated (NT, n = 7) (eFigure 5), suggesting that EV molecular composition changes are not attributable to treatment. In conclusion, our multiplex data suggest that the increase in a subset of EV markers may specifically be associated with low disease activity in RRMS.

Detection of Immune Cell Markers in EVs in Relationship With Disease Activity

EV epitope measurements of 37 markers and 2 isotype controls after SEC-based EV isolation ( Figure 1C for explanation). (A) Localization of the 37 EV surface markers on different cell types. (B) Strategy of bead gating and analyses by front scatter (FSC) and side scatter (SSC). Doublets and no bead events were excluded. The 39 different bead subpopulations are differentiated by the intensity of the fluorophores PE and FITC. Differently labeled beads were discriminated in clusters, confining subgates. The APC intensity within each subgate was quantified. (C) CD29, CD31, and CD69 were positively associated with the number of years elapsed since the last relapse in patients with RRMS, whereas CD31, CD40, CD44, CD49c, and CD69 displayed a negative correlation with NfL. CD44 and CD69 were also negatively associated with GFAP in patients with RRMS. Other groups (NMOSD, MOGAD, and HC) did not exhibit any significant associations with the time elapsed since the last relapse, NfL or GFAP, and specific EV markers. For each marker and sample, background was subtracted from the median APC signal. The mean median signal obtained from CD9, CD81, and CD63 for each sample was used for normalization. Twenty µg of proteins was used for each EV sample analysis. Statistics: Spearman correlation, 95% CI reported, p and r values reported for statistical significant findings. APC = allophycocyanin; FITC = fluorescein isothiocyanate; GFAP = glial fibrillary acidic protein; HC = healthy control; MOGAD = myelin oligodendrocyte glycoprotein antibody–associated disease; NfL = neurofilament light chain; NMOSD = neuromyelitis optica spectrum disorder; PE = phycoerythrin; RRMS = relapsing-remitting MS.

Discussion

In this study, we analyzed the concentration, size, and specific molecular markers of EVs in RRMS, NMOSD, and MOGAD plasma samples to potentially identify disease-specific biomarkers. Compared with HCs, we observed a decrease in EV density and an increase in size in the RRMS group, whereas only an increase in mean EV size was detected in NMOSD and MOGAD groups. The level of a subset of immune cell–related EV markers was negatively correlated with the blood concentration of NfL and GFAP in RRMS, suggesting an association with disease activity in RRMS, but not in NMOSD and MOGAD. In addition, treatments did not significantly influence the amount and composition of RRMS-EVs. Our results suggest that some EV markers may specifically be associated with neuronal damage in RRMS.

NfL and GFAP are considered as potential surrogate biomarkers of neuronal damage and disease activity in demyelinating diseases of the CNS. 4 These 2 proteins are released in the blood circulation after neuronal lesion and astrogliosis. Therefore, the increased levels of GFAP and NfL that we found in this study suggest prominent axonal damage and gliosis in RRMS, consistent with what has been reported by others. 4,5,25 By contrast, NMOSD samples only showed an increase in GFAP z-scores, attributable to the astrocytopathy characterizing this disease. 25 No significant differences were observed in the MOGAD group relative to HCs. Other studies showed higher levels of NfL in the NMOSD group and NfL and GFAP in patients with MOGAD than in HCs. 7 These differences with us may come from the sample size; the relatively smaller sample size of our study, in particular for the NMOSD and MOGAD groups, may limit the observation of statistical differences. In any case, the detection of NfL and GFAP in plasma seems useful to follow disease activity but does not allow us to discriminate diseases from each other. 6,7 For this reason, plasma NfL and GFAP were used as biomarkers of disease activity to understand the changes occurring for EVs.

EVs isolated from RRMS plasma were less abundant but bigger on average than from HC plasma. The reduction in EV number seems to affect the whole spectrum of particle sizes, suggesting the involvement of mechanisms controlling nonselectively microparticle and exosome shedding. Of interest, NMOSD and MOGAD samples only showed a size increase. Particle size distribution analysis suggests that EVs larger than 150 nm may predominantly contribute to this effect. From this observation, it is tempting to propose that exosomes, whose diameters comprised between 50 nm and 150 nm, are less likely to account for this change. However, the overlapping sizes of exosomes and microparticles (50–500 nm) do not allow to draw such a clear conclusion. 22 Morphometric measurements are not informative on the subtypes of EVs released and on the underlying biogenesis mechanisms that may be altered. Moreover, although more quantitative, tetraspanin-based ExoView measurements do not allow us to differentiate exosomes from microparticles. Indeed, CD81, CD63, and CD9 tetraspanins are no longer considered as strict exosomal markers. The surface of microparticles can exhibit the same tetraspanins. 27 Therefore, the limitations of NTA and ExoView prompted us to undertake an extended analysis of EV markers by MACSPlex/flow cytometry.

The examination of immune cell–specific EV markers revealed significant correlations with the levels of NfL and GFAP present in plasma and the time elapsed since the last relapse. In fact, a limited number of EV markers decreased as the plasma concentration of NfL or GFAP increased, suggesting a negative correlation with disease activity in patients with RRMS but not in other groups. In particular, a decline in the levels of CD31, CD40, CD44, CD49c, and CD69 was observed for increasing concentrations of NfL. Similarly, lower levels of CD44 and CD69 were associated with higher concentrations of GFAP in plasma. At low NfL and GFAP concentrations, the increased levels of some EV markers (e.g., CD31 and CD44) in patients with RRMS, above those detected in other patient groups, suggest that the correlations described in RRMS are attributable to EV marker enrichment when disease activity is weak. The function of these EVs enriched in specific markers is not clear. Nevertheless, we speculate that the release of CD31-positive and CD69-positive EVs may be associated with remission because they increased with the time elapsed since the last relapse and they were higher when NfL and/or GFAP levels were relatively low. Still, the biological function of CD31-positive or CD69-positive EVs in the pathologic mechanisms of RRMS is elusive. CD31 has been involved in leukocyte extravasation and in BBB permeability. 28,29 The upregulation of this protein might reflect an attempt at restoring the BBB integrity. 29 However, it is important to distinguish the effect that plasma membrane proteins may have in cells from the physiologic consequences of their increase in EVs. In fact, the release of EVs is a mechanism that can induce a cell loss of function by reducing the amount of plasma membrane proteins. Indeed, for example, the release of EVs mediates the complete elimination of transferrin receptor 1 from the plasma membrane of reticulocytes during their maturation into red blood cells. 30 This mechanism may explain why increased levels of CD31 and other EV markers are associated with low concentrations of NfL and GFAP. One may thus speculate that the release of EVs with CD29, CD31, and CD49c, which are proteins involved in leukocyte extravasation, might coincide with a reduction in CNS infiltration by immune cells. 28,31 The same reasoning may apply to CD69 and CD44 that are highly expressed in the T cells of active MS lesions. 32 The potential removal of CD40 at the surface of antigen-presenting cells may dampen cell differentiation and regulate the immune response. 33,34 Inversely, low levels of CD29, CD31, and CD44 in EVs have been associated with systemic inflammation in polytrauma patients. 35 Whether the secretion of EVs enriched in CD29, CD31, CD40, CD44, CD49c, and CD69 is a mechanism reducing leukocyte activation in the periphery and protecting the BBB in MS deserves further investigation.

The fact that the levels of EV markers did not significantly vary in patients receiving natalizumab compared with other treatments suggests that EV changes are independent of treatments, even if potential treatment effects cannot be completely ruled out. For example, by inhibiting T-cell infiltration into the CNS, natalizumab may induce peripheral T-cell accumulation and increased EV shedding. 36 Indeed, we observed that the number of lymphocytes was doubled in the blood of patients with MS receiving natalizumab in comparison with those treated with dimethylfumarate or left untreated (data not shown). In addition, in the NMOSD and MOGAD groups, cell-depleting treatments may result in reduced T-cell–derived EV release. However, our current data rather support a disease-specific enrichment of T-cell EV markers in RRMS. 37

In this study, we use a combination of standard methods such as NTA and advanced EV marker-based analytic methods such as SP-IRIS (ExoView). Of interest, a lower level of EVs was shown in patients with RRMS by NTA but not by SP-IRIS. Therefore, further studies are necessary to confirm the EV density difference. In addition, technical limitations and differences between NTA and SP-IRIS should be taken into consideration. Particle levels reported by NTA were approximately 1–2-fold higher than those of SP-IRIS. In addition, NTA allows us to measure a wider range of EV diameters than SP-IRIS. NTA tends to overestimate both concentration and size, in comparison with SP-IRIS and transmission electron microscopy, 38 because clusters of contaminants or EVs are also measured by NTA. 39 The 2 methods are technologically different: NTA is based on the movement of particles in solution 40 while SP-IRIS allows fluorescence measurement of immunocaptured EVs. 41 Consequently, NTA detects all particles in the solution within the range of measurement (approximately 60–1,000 nm) while SP-IRIS only enables measurement of particles with CD9, CD63, or CD81. In addition, EVs were isolated by SEC from plasma before NTA, whereas crude plasma was used for SP-IRIS, an approach that may limit the loss of EVs. 42 In summary, in the growing field of EVs, caution should be exerted with the interpretation of results showing variations in the composition or density of EVs when using a limited number of methods. 21,39

In this study, the analyses of EVs in the plasma of patients revealed specific changes in the levels of the EV surface markers CD29, CD31, CD40, CD44, CD49c, and CD60 in association with the time elapsed since the last relapse and the concentrations of NfL and GFAP in patients with RRMS. We hypothesize that the secretion of EVs may influence the activation state of immune cells in the periphery, although a different cell origin (e.g., CNS origin) of EVs cannot be ruled out. EVs represent a source of potential markers whose functional characterization may allow us to improve the monitoring of MS.

Acknowledgment

The authors thank Dr. Sandrine Joly (Department of Neurology, University of Bern), for technical support, the Microscopy Imaging Center of the University of Bern and Dr. Mircea-Ioan Iacovache for cryo-EM examinations. NTA measurements were performed at the Flow Cytometry and Cell Sorting Facility of the Department for BioMedical Research (DBMR) of the University of Bern.

Glossary

blood-brain barrier

cryo-electron microscopy

Expanded Disability Status Scale

extracellular vesicle

glial fibrillary acidic protein

healthy control

myelin basic protein

isotype control IgG

microRNA

myelin oligodendrocyte glycoprotein antibody–associated disease

multiple sclerosis

neurofilament light chain

neuromyelitis optica spectrum disorder

nanoparticle-tracking analysis

phosphate-buffered saline

relapsing-remitting MS

room temperature

rituximab

single-molecule array

single-particle interferometric resonance imaging sensor

Author Contributions

V. Lim Falk: drafting/revision of the manuscript for content, including medical writing for content; major role in the acquisition of data; analysis or interpretation of data. N. Mueller-Wirth: major role in the acquisition of data. D. Karathanasis: major role in the acquisition of data. M.E. Evangelopoulos: major role in the acquisition of data. A. Maleska Maceski: major role in the acquisition of data; analysis or interpretation of data. A. Zadic: analysis or interpretation of data. J. Kuhle: major role in the acquisition of data; analysis or interpretation of data. C. Schlup: major role in the acquisition of data. S. Marti: analysis or interpretation of data. K. Guse: analysis or interpretation of data. A. Chan: drafting/revision of the manuscript for content, including medical writing for content; analysis or interpretation of data. V. Pernet: drafting/revision of the manuscript for content, including medical writing for content; study concept or design; analysis or interpretation of data.

Study Funding

This project was supported by CSL Behring.

Disclosure

Andrew Chan has served on advisory boards for and received funding for travel or speaker honoraria from Actelion-Janssen, Alexion, Almirall, Biogen, BMS, Celgene, Dresden International Uni, Horizon, Merck, Mylan (Viatris), Novartis, Roche, Sanofi-Genzyme, Teva, UCB, Uni Leipzig, Wiley, all for hospital research funds; received research support from Biogen, CSL Behring, Genzyme, Roche, UCB and European Union, and the Swiss National Research Foundation; is an associate editor of the European Journal of Neurology; and serves on the editorial board for Clinical and Translational Neuroscience and as topic editor for the Journal of International Medical Research. The other authors report no relevant disclosures. Go to Neurology.org/N for full disclosures.

References

Associated Data

Data Availability Statement

Anonymized patient data not published within this article will be made available by request from any qualified investigator.