All procedures involving animals were conducted with approval from the Institutional Animal Care and Use Committee at the University of Connecticut School of Medicine and in accordance with guidelines set forth by the National Research Council of the National Academies Guide for the Care and Use of Laboratory Animals. Mice used in this study were included wildtype C57BL/6 (strain #000664) and GFAP-Cre mice (JAX strain #024098). To induce experimental autoimmune encephalomyelitis (EAE) wild-type C57BL/6 mice (6–8 weeks old) were immunized with a 1:1 ratio of myelin oligodendrocyte glycoprotein (MOG₃₅₋₅₅, AnaSpec Inc.) dissolved in deionized water and complete Freund's adjuvant (CFA, Sigma) containing 0.5 mg of Mycobacterium tuberculosis H37RA (Difco Laboratories: BD Diagnostics), as previously described (Crocker et al., 2006). The MOG-CFA emulsion was administered subcutaneously (s.c.) into the flanks of the hind-limbs (300 μg/mouse). On days 0 and 2, pertussis toxin (PTX, List Biological) was injected intraperitoneally (i.p.) (500 ng/mouse). Weights and clinical scores were recorded daily. The following grading scheme was used to score clinical signs of disease severity: 0, no clinical signs; 0.5, distal tail limpness; 1, full tail atony; 2, hindlimb paresis; 3, unilateral hindlimb paralysis; 4, bilateral hindlimb paralysis; 5, moribund. Blood was collected at peak clinical illness.

Cultures were generated from cerebral cortices of neonatal C57BL/6 mouse pups (P0–P3) using a neural tissue dissociation kit (Miltenyi Biotec), as previously described (Crocker et al., 2008). Cells were plated into T75 flasks. The purity of each culture was confirmed and consistent with previous reports of 90–97% GFAP+ cells (Crocker et al., 2008), as verified by immunocytochemistry (ICC) for GFAP for astrocytes (1:500, Sigma-Aldrich) and Iba-1 for microglia (1:1,000, WAKO). Cells were grown to 70–80% confluence, washed two times with PBS and incubated for 24 h in serum-free media (Dulbecco's modified eagle medium, 1% Pen-strep, Gibco) or in the presence of 10 ng/mL of IL-1β (Peprotech). Cell culture conditioned media (CCM) was collected after 24 h, spun at 3,000 × g for 15 min at room temperature, and stored at −80°C until use.

Blood was collected from CFA or EAE animals at pre-and peak clinical illness while under deep isoflurane anesthesia using a 1 ml syringe that had been flushed with 0.5 M EDTA (Fisher Scientific). Exosome isolation workflow from blood samples is shown in Figure 3A. Blood volumes collected ranged from 300 to 500 μl/animal. Blood was centrifuged at 2,000 × g for 15 min at room temperature and the upper plasma layer was drawn off. Plasma was centrifuged at 1,500 × g at room temperature and the supernatant drawn off. Plasma samples were then subjected to a differential ultracentrifugation protocol for exosome isolation which were confirmed by electron microscopy, western blotting, and flow cytometry. Briefly, plasma was spun at 12,000 × g for 30 min at 4°C. The supernatant was collected and run through a 0.22 μm spin column (Millipore) at 12,000 × g for 4 min at room temperature. Filtrate was collected and equal amounts were ultracentrifuged at 118,000 × g for 90 min at 4°C. The supernatant was removed and the exosome pellets were re-suspended in 100 μl of sterile, 0.22 μm filtered PBS(−) and stored at −80°C until use. All samples were validated by electron microscopy.

ExoQuick™ precipitation was carried out according the manufacturer's instructions (System Biosciences) and experimental workflow is depicted in Figure 1A. Briefly, 3 mL of conditioned media was mixed with 0.6 mL of ExoQuick-TC™ solution by inverting the tubes several times. The samples were left to incubate overnight at 4°C then centrifuged twice at 1,500 × g for 30 and 5 min, respectively, in order to remove the supernatant. The supernatant was discarded and the pellet was re-suspended in 100 μL of 0.22 μm-filtered PBS(−) and stored at −80°C until use.

For negative staining, 15 μl drops of exosomes (in PBS) were adsorbed onto activated copper grids with carbon coating (Electron Microscopy Sciences) for 15 min, washed by dabbing the grid onto three drops of deionized water, and stained with 1% uranyl acetate (Electron Microscopy Sciences) for 1 min. Grids were imaged under a Hitachi H-7650 transmission electron microscope. For immunogold labeling, 15 μl of isolated exosomes were adsorbed onto activated nickel grids with carbon coating for 15 min, Grids were then transferred to 100 μl drops of PBS and washed twice for 3 min each. Following the PBS wash, grids were transferred to 100 μl drops of a PBS/50 mM glycine solution and washed four times for 3 min each. Grids were then blocked for 10 min on 100 μl drops of blocking buffer (1% BSA in PBS). Grids were then incubated for 30 min at room temperature on 30 μl drops of primary antibody against TSG101 (1:10 dilution in 1% BSA; Genetex) and GFAP (1:10 dilution in 1% BSA; Millipore and Novus Biologicals). Grids were washed on 100 μl droplets of PBS three times for 5 min, blocked in 1% BSA for 5 min, and incubated with the appropriate secondary antibody conjugated with 10 nm gold particles or 15 nm gold particles (Electron Microscopy Sciences; 1:15 dilution in blocking solution) for 30 min. Grids were washed three times for 5 min in PBS followed by successive deionized water washes and counter-stained with 15 μl of 1% uranyl-acetate for 1 min. Excess uranyl-acetate was removed by gently blotting the grids and left to air dry before imaging under a transmission electron microscope. Scanning electron micrographs were performed on astrocyte in cultures that were fixed in 2.5% gluteraldehyde, post-fixed using osmium tetroxide and then dehydrated using ethanol and processed by critical point drying. Cells were then spatter-coated with gold palladium and visualized using a JOEL Scanning EM (JOEL USA Inc.).

Exosome preparations were lysed in RIPA buffer (with protease inhibitor cocktail; Sigma) and separated by SDS-PAGE. Proteins were transferred to nitrocellulose and immunoblotted using unconjugated antisera against CD63 (GeneTex), GFAP (Sigma), or TSG101 (GeneTex) that were then visualized by HRP-conjugated secondary antisera using chemiluminescence (ECL; Amersham).

Flow cytometric analysis on exosome surface markers was performed at the UConn Health flow cytometry core. CD63 (PE anti-mouse; 1:100; Biolegend), GFAP (Alexa Fluor 647 anti-mouse; 1:50; BD Biosciences), and TSG101 (FITC anti-mouse; 1:50; Lifespan Biosciences) were added to sterile, 0.22 μm-filtered PBS (1x, pH 7.4) and spun at 10,000 × g at 4°C for 30 min to remove protein aggregates from the staining material. Supernatants were transferred to fresh 0.2 mL PCR tubes and a 10 μl aliquot of exosome suspension was added. Tubes were vortexed before being placed in a 37°C incubator for 60 min. Stained samples were then transferred to round-bottom polystyrene tubes and analyzed using a Becton Dickinson (BD) FACS Aria-II with a 130 μm nozzle and 10 PSI. Samples were run at the lowest flow rate to ensure the most focused core stream until approximately 10,000 single events [based on a 1:1 ratio of side scatter (SSC) pulse height to pulse area] were captured. Threshold was set based upon SSC at the minimum value of 200 in order to remain unbiased for fluorescence particle detection. Noise was determined by running the diluent (0.22 μm-filtered PBS) until ~500–1,000 “noise events” were captured. This was considered the background signal for the instrument. For analysis of exosomes, a minimum of 10,000 events were captured and sorted for subsequent electron microscopy analysis. The detector voltage for Alexa Fluor 647 was set such that unlabeled EVs gave a signal that was 2x the robust standard deviation of the “noise events” to clearly distinguish the sample contents from the diluent and background signal from the instrument. This was determined by the cytometer baseline report from BD Cytometer Setup and Tracking software. Reference beads of known sizes (Spherotech, Nano Fluorescent Size Standard Kit; #NFPPS-52-4K) were also used to determine instrument performance the day's experiment as outline in section 3.1. Analysis and gating was performed using BD FACS Diva software and FlowJo V10.2.

Experiments were performed in quadruplicate technical replicates for each of at least three biological replicates per condition. Comparisons between treatments were made using Student's t-test or repeated measures ANOVA with significance indicated where appropriate. Data are presented as mean ± SEM using scatter plots to demonstrate data distribution within each treatment group. The null hypothesis for all experiments was P < 0.05.

To establish the identity of exosomes from astrocytes in more complex biological fluids, we first collected serum-free media from primary astrocytes in culture in order to benchmark the approach and reagents under these defined conditions. Exosomes were collected and then analyzed by transmission electron microscopy, which revealed negatively stained exosome-like double membrane vesicles with size ranges between 50 and 150 nm, with a stereotypical cup-shaped morphology (Figure 1B). The identity of these extracellular vesicles as exosomes was then confirmed by immunogold labeling with the exosome marker TSG101, in conjunction with the astrocyte marker, glial fibrillary acidic protein (GFAP) (Figure 1C). Together, these histomorphological features confirmed the collection of exosomes from the astrocyte-conditioned media.

We next sought to determine whether flow cytometry could also be adopted as a means to reliably identify exosomes in media samples conditioned by primary astrocytes in culture. Flow cytometry analysis of exosomes is presently considered a highly desirable, widely accessible technology by which to assess the presence of small vesicles, however, the standardization of this method for small particle analysis is still in development with no final consensus on the exact method to use. A significant limitation of this analysis remains the resolution of size using forward angle light scatter (FALS) to trigger a particle “event” on the flow cytometer. Due to their small size, extracellular vesicles cannot be distinguished using a traditional forward scatter signal detector or photodiode. Some customized and specialized instrumentation uses a more sensitive photomultiplier tube for FALS (Stoner et al., 2016), however not all labs have access to such instrumentation. On the other hand, side angle light scatter (SSC) and fluorescence are both detected using the more sensitive photomultiplier tube (PMT) which allows for resolution of low signals such as those produced by small vesicles with respect to the background electronic “noise” of the instrument. Taking advantage of these detectors, we set the cytometer to trigger events based on the lowest possible setting for SSC and subsequently examined the fluorescence of any particle that registered above the noise of the instrument for the given fluorescence detector. Noise was determined by acquiring a sample of the media used to dilute the vesicles (Figure 1D). The instrument performance for this detection method was determined using a set of yellow-fluorescent nano-calibration beads on each experimental day to ensure detection of a 200 nm particle (Figure 1E). Although polystyrene bead particles do not have the same refractive index as lipid vesicles for the generation of SSC, this test ensured the instrument was at least capable of detecting small particles above noise. Another caveat to the use of flow cytometry for small particle detection is the possibility of larger aggregates of particles (“swarm detection”), where several vesicles were detected simultaneously (van der Pol et al., 2014). To reduce this possibility, all exosome samples were run both undiluted and diluted to ensure the fluorescence intensity of the signal remained constant. In addition to the dilution of the sample, each exosome sample was run simultaneously with an unstained sample, a sample allowed to react with an isotype control Ab, and a sample which only contained antibody with dilution media. This strategy then enabled us to discern the labeled exosome signal from aberrant signal from unbound antisera and also non-specific binding of Ab to exosomes (Figure 1F). Using this step-wise approach we identified exosomes bound to Abs specific for the astrocyte marker GFAP as well as canonical exosome markers TSG101 and the tetraspanin protein, CD63 (Figure 1F).

We next tested whether our methodology could be used to identify changes in exosome release. Previous work has shown that IL-1β is a potent inducer of exosome release. We began by examining astrocytes using scanning electron microscopy to determine if membrane perturbations/projections, associated with extracellular vesicle release in other cell types could be observed in astrocytes. In untreated astrocytes (Figure 2A), the plasma membrane was observed to have many long, sinewy projections. In contrast, shortly after treatment with IL-1β, the cell surface of astrocytes was noticeably different with numerous bulbous protrusions which resemble microblebs (Figure 2B), which reflect intracellular budding events consistent with the process of exosome release from exocytosis of MVBs (Gyorgy et al., 2011). This subcellular structural response was then associated with a measurable increase in the abundance of exosomes into the cell culture media (Figures 2C–G). To test if our flow cytometry approach could be used to identify exosomes release from astrocytes activated by IL-1β we assayed the proportion of exosomes released into the conditioned media collected 24 h after treatment. IL-1β treatment significantly increased the abundance of CD63+ particles smaller than 220 nm (exosomes) (Figure 2C, P < 0.001). The proportion of exosomes released by astrocytes was also increased (Figure 2D, P < 0.01), and we confirmed the exosomal origin of these extracellular vesicles using a second marker Tsg101, which also identified a significant increase in response to IL-1β treatment (Figure 2E, P < 0.05). We further validated the identity of increased exosomes as having been derived from astrocytes by determining the relative increase in the proportion of GFAP+/CD63+ (Figure 2F, P < 0.01) and GFAP+/TSG101+ exosomes (Figure 2G, P < 0.05). Thus, we confirmed the utility of all three markers of astrocyte-derived exosomes (GFAP, CD63 and TSG101) and determined that these markers exhibited remarkable concordance between each marker in terms of its magnitude of increase in response to IL-1β treatment. Once again, for each exosome isolate tested, we verified by TEM the appropriate size class of extracellular vesicles collected from each sample (data not shown).

The objective of developing a validated approach to identify exosomes from primary astrocytes conditioned media was to be able to apply this approach to more complex biological fluids. Interestingly, the astrocyte marker protein GFAP has been reported to be found in blood from patients suffering from a number of neurodegenerative diseases including Alzheimer's, ALS, and multiple sclerosis (Malmestrom et al., 2003; Mayer et al., 2013). Therefore, we hypothesized that GFAP identified in blood samples may indicate the presence of astrocyte-derived exosomes. This would suggest that astrocyte-derived exosomes are present in peripheral circulation.

To test this possibility, we collected blood samples from naïve C57BL/6 mice and isolated exosomes from blood plasma using a differential centrifugation protocol (Witwer et al., 2013). Exosomes isolated by this method (Figure 3A) were analyzed by TEM to determine if the vesicles present had all the characteristics of exosomes we observed in our conditioned media samples. Negative staining confirmed a distinctive cup-shaped morphology of the double membrane vesicles and size class consistent with exosomes, ranging from 50 to 150 nm size range, in these blood plasma samples (Figure 3B). Immunogold labeling of exosomes with the astrocyte marker GFAP identified astrocyte-derived exosomes in these samples from peripheral blood (Figure 3C).

To address whether these GFAP-positive exosomes in blood were indeed derived from astrocytes in the CNS and not from other peripheral tissues that have been suggested to be immunoreactive for GFAP (Malmestrom et al., 2003), we collected blood samples from CRE recombinase mice in where the transgene was under the control of the mouse GFAP promoter [B6.Cg-Tg(Gfap-cre)73.12Mvs/J] (Garcia et al., 2004). CRE mRNA expression in these mice is limited to the brain and spinal cord whereas no CRE expression is detected in peripheral tissues (data not shown). Blood plasma samples were collected and the exosome pellet subjected to western blot analysis for the exosome marker CD63, as well as astrocyte specific markers GFAP and the transgene product, CRE protein. This western blot analysis confirmed both expression of exosome marker, CD63 (Figure 3D), as well as the astrocyte markers GFAP (Figure 3D) and the gene product of the transgene, CRE (Figure 3E). Together, these findings confirmed that GFAP+ exosomes present in peripheral blood circulation of naÏve adult mice were of astrocytic origin.

We next examined whether the amount of astrocyte exosomes in peripheral blood changed during course of clinical disease in a model of a neurological disease. To test this, we used the same strain of C57BL/6 mice for which we derived our primary glial cultures, and immunized adult mice to induce EAE using myelin oligodendrocyte glycoprotein (MOG) peptide as described above. This model of inflammatory demyelination induces a robust, reproducible, and predictable CNS demyelination that is used to model the immune-mediated pathology of multiple sclerosis. In this set of experiments, mice immunized with MOG₃₅₋₅₅ peptide, or given an equal volume of CFA, were monitored daily for signs of clinical illness and at the peak clinical disability (Figure 3F), Blood was collected from mice either 1 week following immunization with MOG, or at the peak of clinical disability (Day 16) post immunization (Figure 3F). Blood samples from CFA-treatment groups were time-matched to either 7 or 16 day time points (Figures 3G,H) Exosome samples were then analyzed using flow cytometry to determine the abundance of GFAP+ exosomes in the EAE group when compared to CFA controls. Flow cytometry identified increased a low abundance of GFAP+ exosomes in all samples and the relative abundance in CFA-treated animals did not differ from MOG-immunized animals 7 days post injection (Figure 3G; n = 4–6/group). In contrast, analysis of GFAP+ exosomes from blood of EAE mice taken during peak clinical disability (Day 16) revealed a significant increase in the relative abundance of GFAP+_exosomes in EAE mice when compared with time-matched CFA-only treated controls (Figure 3H). Taken together, these data indicate that neuroinflammation associated with EAE was associated with increased presence of astrocyte-derived exosomes in the blood.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.