Inhibition of neutral sphingomyelinase 2 reduces extracellular vesicle release from neurons, oligodendrocytes, and activated microglial cells following acute brain injury

Abstract

Extracellular Vesicles (EVs) are implicated in the spread of pathogenic proteins in a growing number of neurological diseases. Given this, there is rising interest in developing inhibitors of Neutral Sphingomyelinase 2 (nSMase2), an enzyme critical in EV biogenesis. Our group recently discovered phenyl(R)-(1-(3-(3,4-dimethoxyphenyl)-2,6-dimethylimidazo[1,2-b]pyridazin-8-yl)pyrrolidin-3-yl)carbamate (PDDC), the first potent, selective, orally-available, and brain-penetrable nSMase2 inhibitor, capable of dose-dependently reducing EVs release in vitro and in vivo. Herein, using multiplexed Surface Plasmon Resonance imaging (SPRi), we evaluated which brain cell-derived EVs were affected by PDDC following acute brain injury. Mice were fed PDDC-containing chow at doses which gave steady PDDC brain exposures exceeding its nSMase2 IC₅₀. Mice were then administered an intra-striatal IL-1β injection and two hours later plasma and brain were collected. IL-1β injection significantly increased striatal nSMase2 activity which was completely normalized by PDDC. Using SPRi, we found that IL-1β-induced injury selectively increased plasma levels of CD171+ and PLP1+ EVs; this EV increase was normalized by PDDC. In contrast, GLAST1+ EVs were unchanged by IL-1β or PDDC. IL-1β injection selectively increased EVs released from activated versus non-activated microglia, indicated by the CD11b+/IB4+ ratio. The increase in EVs from CD11b+ microglia was dramatically attenuated with PDDC. Taken together, our data demonstrate that following acute injury, brain nSMase2 activity is elevated. EVs released from neurons, oligodendrocytes, and activated microglial are increased in plasma and inhibition of nSMase2 with PDDC reduced these IL-1β-induced changes implicating nSMase2 inhibition as a therapeutic target for acute brain injury.

Graphical Abstract

1. Introduction

Extracellular Vesicles (EVs) are small, membrane-bound particles released by almost all cells in the body. EVs are subdivided, based on size, into small vesicles (<100–200 nm) and medium/large vesicles (>200 nm)[1]. After their identification more than 40 years ago, despite there being evidence that EVs had broad biological roles[2], EVs were initially thought of as a means for the cell to remove waste. However, over the past decade a growing body of evidence has solidified their role in cell-to-cell communication during both physiological and pathological conditions[3]. Indeed, EV contents, size and membrane composition are highly heterogeneous and dynamic, with variations in EVs physicochemical properties depending on the cellular source, state and environmental conditions. A wide variety of cargo has been isolated from EVs including nucleic acids, proteins, and lipids, enabling EVs to carry out broad functions[3]. EVs have been identified in most body fluids, including plasma, where circulating EVs can be identified as originating from selective tissues and organs, including the central nervous system[4, 5]. Therefore, evaluation of plasma EVs offers the possibility to gain insight into the pathogenic mechanisms occurring during onset and progression of diseases and could also serve as a source of disease or therapeutic biomarkers [6]. For example, EV content has been shown to be altered in patients with Alzheimer’s Disease (AD)[7, 8]. Specifically, the combination of phosphorylated tau and insulin receptor substrate 1 in neuronally-derived plasma EVs were shown to predict AD in patients with high accuracy and specificity about 4 years before symptom onset [7, 8]. miRNA content in brain-derived EV samples from plasma were also shown to correlate with early pathological changes in AD [9]. These findings highlight the potential of plasma EVs being used as a liquid brain biopsy to monitor disease onset and progression.

EVs originating from the multivesicular body (MVB) are thought to be generated via two major pathways, the endosomal sorting complexes required for transport (ESCRT)-dependent and independent pathways. The ESCRT-dependent pathway is the canonical route whereby a series of ESCRT protein complexes are recruited to the membrane which lead to membrane curvature, vesicle formation, and scission of the vesicles into intraluminal vesicles (ILVs) thus creating MVBs [10]. The ESCRT-independent pathway relies on the enrichment of membrane ceramide to induce an inward curvature of the membrane resulting in the formation of MVBs [11]. Regardless of how the MVB is formed, they release EVs into the extracellular space following fusion with the plasma membrane. It is the latter ESCRT-independent pathway that is influenced by neutral sphingomyelinase 2 (nSMase2).

Ceramide can be generated via the breakdown of sphingomyelin (SM) into ceramide and phosphorylcholine by sphingomyelinases[12, 13]. One of the most studied sphingomyelinases is nSMase2 which is concentrated in the brain and has been implicated in AD, Parkinson’s disease (PD), amyotrophic lateral sclerosis (ALS), prion disease, multiple sclerosis (MS), and cerebral ischemia (see reviews [6, 13]). Both genetic and pharmacological inhibition of nSMase2 have shown efficacy in reducing EV release and ameliorating disease progression in these diseases (reviewed in [6]) with AD being one of the more thoroughly studied areas. For example, crossing nSMase2 knockout mice with the 5XFAD AD mouse model led to decreased EV release with reduced amyloid burden and improved cognition[14]. In this same mouse model, pharmacological inhibition of nSMase2 also reduced amyloid burden in the brain[15] and improved cognition[16]. nSMase2 inhibitors also reduced tau pathology in transgenic tauopathy AD mouse models and rapid tau propagation mouse models[17]. To date, several inhibitors of nSMase2 have been developed. However, none have successfully made it to the clinic due to poor potency, low oral bioavailability and/or poor brain penetration. Our group recently discovered phenyl(R)-(1-(3-(3,4-dimethoxyphenyl)-2,6-dimethylimidazo[1,2-b]pyridazin-8-yl)pyrrolidin-3-yl)carbamate (PDDC), the first potent, selective, orally-available, and brain-penetrable nSMase2 inhibitor[16, 18].

While we and others have reported that nSMase2 inhibition can reduce the release of EVs into plasma following brain disease or insult [15, 18, 19], the specific population of brain cells affected by nSMase2 inhibition has not been explored. To study this, we took advantage of our optimized biosensor based on Surface Plasmon Resonance imaging (SPRi) technology which allows for the simultaneous tracking of several subpopulations of EVs by spotting numerous ligands on a single chip[5]. SPRi-based biosensors are biophotonics platforms that identify active biomolecular adsorption by measuring the modification of the refractive index on top of a gold-coated surface in real time, without the need for labels. Therefore, sensors based on SPRi technology can be used both to improve the accuracy and reliability of disease diagnosis and monitoring procedures, and to evaluate the efficacy of new therapeutics or rehabilitation protocols. Recently, SPRi has demonstrated its effectiveness for the detection of EVs and for the analysis of molecules expressed on their membranes, including proteins and lipids, whose presence and/or expression levels can be indicative of pathological conditions[20–22].

In this work we show, for the first time, the use of SPRi to evaluate the influence of nSMase2 inhibition on EV release from central nervous system cells into the plasma following an acute brain injury. The array was designed to separate and target general EVs (with anti-CD9 antibody) from EVs released specifically by neurons (nEVs; with anti-CD171 antibody), oligodendrocytes (oEVs; with anti-proteolipid protein 1 (PLP1) antibody), astrocytes (aEVs; with anti-GLutamate and ASpartate Transporter (GLAST) antibody), resting microglia (resting mEVs; with isolectin B4 (IB4)) and activated microglia (activated mEVs; with anti-CD11b antibody). Using this SPRi platform, we evaluated the relative amount of EVs released by each specific cell type circulating in the plasma following an intra-striatal IL-1β injection and the influence of PDDC treatment.

2. Materials and Methods

2.1. Animal studies and PDDC chow dosing

All animal care and experimental procedures complied with the National Institutes of Health guidelines on animal care and were approved by the Johns Hopkins University Institutional Animal Care and Use Committee. Mice were housed in a temperature and humidity-controlled environment under a 12-hr light cycle with food and water available ad libitum and were allowed to acclimate to the colony room for at least 7 days after arrival, before any experimentation. The nSMase2 inhibitor PDDC synthesized in our laboratory using our previously published procedures [16, 18], was formulated into an OpenStandard Diet with 15 kcal% mouse chow at doses that equate to 30mg/kg and 100mg/kg daily based on the assumption that the mice eat 1g chow per 10g body weight as recommended by Research Diets (New Brunswick, NJ; https://researchdiets.com/opensource-diets/custom-diets). To verify this, we weighed the daily food consumption of the mice housed in our facility and found this to be accurate (24 mice tested consumed between 0.91–1.1g/10g body weight/day). Two weeks prior to experimentation, all animals were switched from in-house chow to OpenStandard base vehicle chow to acclimatize to the new chow. Vehicle-treated animals were given the base OpenStandard Diet. Excess fresh chow was exchanged every other day. All the procedures described herein were performed following relevant guidelines and regulations.

2.2. PDDC in vivo pharmacokinetics (PK) and bioanalysis

Male B6C3F1/J mice (Jackson Laboratories, Stock # 100010, Bar Harbor, ME) were enrolled in a PK study quantifying drug exposure following PDDC chow dosing (n=2/time point/group). For the study, mice were switched to vehicle chow for two weeks and then switched to PDDC-containing chow (30mg/kg or 100mg/kg) for two weeks before being sacrificed at four time points throughout a 24h time period (00:00, 7:00, 12:00, 19:00). Times were chosen based on the 12h light cycle where 7:00 is 1h before lights come on and 19:00 was 1h before lights went off. Mice were euthanized by isoflurane overdose and blood was collected via cardiac puncture and placed into iced EDTA-coated BD microtainers (Franklin Lakes, NJ); plasma was harvested from blood by centrifugation at 500 × g for 15 min and stored at −80°C until LC/MS/MS bioanalysis. Brain tissues were harvested following blood collection and immediately snap frozen in liquid nitrogen and stored at −80°C until LC/MS/MS bioanalysis.

For the bioanalysis, calibration standards were prepared using naïve mouse plasma or brain with additions of known amounts of PDDC. PDDC standards and samples were extracted from plasma and brain by a one-step protein precipitation using acetonitrile (Sigma-Aldrich, St. Louis, MO) (100% v/v) containing the internal standard (losartan 500 nM; Tocris, Minneapolis, MN). The samples were vortexed (30 sec) and centrifuged at 10,000 × g (10 min at 4 °C). An aliquot of the supernatant (50 μl) was diluted with water (50 μl) and transferred to a 250-μl polypropylene vial sealed with a Teflon cap and analyzed via LC/MS/MS as our lab has described previously[18]. Plasma concentrations (nmol/ml) as well as tissue concentrations (nmol/g) were determined and plots of mean plasma concentration versus time were constructed. Non-compartmental analysis modules in Phoenix WinNonlin version 7.0 (Certara USA, Inc., Princeton, NJ) were used to quantify exposures (AUC_0–t).

2.3. Striatal IL1-β injection ± PDDC

IL1-β striatal injections were performed as previously described by our group [18, 23, 24]. Briefly, male (2–3 month) GFAP–GFP transgenic mice (Strain # 003257, Jackson Laboratories, n = 3/group) were anaesthetized with 3% isoflurane (Baxter, Deerfield, IL) in oxygen (Airgas) and placed in a stereotaxic frame (Stoelting Co, Wood Dale, IL). We chose male mice to avoid effect of female specific hormones (i.e., oestrogen) that are known to have anti-inflammatory properties[25]. A small burr hole was drilled in the skull over the left striatum using a dental drill (Fine Scientific Tools, Foster City, CA). IL-1β (0.1 ng/3 μl, EMD Millipore, Burlington, MA) was injected (total volume of 3 μl) at the rate 0.5 μl·min⁻¹ via a pulled glass capillary (tip diameter < 50 μm)[24]. The stereotaxic coordinates based on bregma were A/P + 1; M/L −2; and −3 D/V[23]. Saline injection was used as the control. For the dose response target engagement study, mice were randomly separated into 4 groups (n=3–4/group): 1) saline + vehicle chow; 2) IL-1β + vehicle chow; 3) IL-1β + 30mg/kg PDDC chow; 4) IL-1β + 100mg/kg PDDC chow. For the SPRi analysis, mice were randomly separated into 3 groups (n=3/group): 1) saline + vehicle chow; 2) IL-1β + vehicle chow; and 3) IL-1β + 100mg/kg PDDC chow. Mice were pretreated with vehicle for two weeks then either continued on vehicle or switched to PDDC chow 6 days prior to IL1-β or saline injection. Mice subjected to intra-striatal injection of IL-1β were also injected with the NSAID carprofen (rimadyl, 5 mg·kg⁻¹, i.p., Zoetis, Parsippany-Troy Hills, NJ) and closely monitored during recovery; no adverse reactions were observed. Following infusion, the capillary was held in place for 5 min to allow for the solution to diffuse into the tissue. Mice were euthanized at 2 hr post-IL-1β treatment with overdose of anaesthetic (isoflurane, Baxter). Blood samples were taken by cardiac puncture with heparin (Sigma) coated syringes and EDTA tubes (BD). Blood was immediately centrifuged at 2,700 × g for 15 min (20°C) to obtain plasma, that was used for EVs analysis (below). Striatal brain tissue was collected for nSMase2 enzymatic activity assessment (below).

2.4. nSMase2 enzymatic activity assay

The nSMase2 enzymatic activity assay has been described by our group previously[19, 26]. Briefly, lysates of striatal brain tissue derived from the 4 different experimental mice groups were used as the enzyme source to catalyze the hydrolysis of exogenous sphingomyelin (20μM; Thermo Fisher, Waltham, MA) to ceramide and phosphorylcholine. Phosphorylcholine undergoes dephosphorylation by alkaline phosphatase (4 U/ml; Sigma-Aldrich) to produce choline which in turn is oxidized by choline oxidase (0.1 U/ml; Sigma-Aldrich) to betaine and hydrogen peroxide. Hydrogen peroxide reacts with Amplex Red (50 μM, Thermo Fisher) in the presence of peroxidase (HRP, 1 U/ml; Worthington Biochemical Corporation, Lakewood, NJ) to generate the fluorescent molecule resorufin. Generation of fluorescence was monitored by measuring relative florescence units with excitation at 530 nm and emission at 590 nm. Reactions were carried out for 1 hr at 37°C in 100-mM Tris–HCl pH 7.4, 10-mM MgCl₂, and 0.1% Triton X-100 (Sigma-Aldrich).

2.5. Analysis of EVs in plasma

Detailed protocols and data regarding EV isolation and characterization have been uploaded in the EV-TRACK knowledgebase (EV-TRACK ID: EV210004)[27]. In brief, plasma collected from the intra-striatal IL1-β injection study was centrifuged at 10,000 × g for 15 min (4°C) to generate platelet-free plasma and to remove large particles such as apoptotic bodies. The number of plasma EVs were quantified using ZetaView Nanoparticle Tracker (Particle Metrix GmBH, Meerbusch, Germany) and the corresponding ZetaView software (8.03.04.01) as we have previously described[18]. The instrument pre-acquisition parameters were set to 23°C, a sensitivity of 65, a frame rate of 30 frames per second, a shutter speed of 100, and laser pulse duration equal to that of shutter duration. Post-acquisition parameters were set to a minimum brightness of 25, a maximum size of 200 pixels, and a minimum size of 10 pixels. For each sample, 1 ml of the supernatant was injected into the sample-carrier cell and the particle count measured at five positions, with two cycles of reading per position. The cell was washed with 1X phosphate buffered saline (PBS; Life Technologies, Carlsbad, CA) after every sampling. Concentration of EVs per ml (±SEM) was calculated from three independent experiments performed on each sample. In parallel, EVs were also isolated from plasma by size exclusion chromatography (qEVsingle, Izon, Christchurch, New Zealand), following the manufacturer’s instruction. Briefly, 100 μl of sample were loaded in the column and freshly filtered PBS was used as eluent; fractions containing EVs, from 6 to 11, were collected and stored at −20°C with protease inhibitors until use in SPRi experiments.

A portion of the freshly isolated EVs were concentrated by ultracentrifugation (100,000 × g, 70 min, 4°C; Rotor SW60; Beckman Coulter, Brea, CA, USA) for analysis via Transmission Electron Microscopy (TEM) to confirm the effective isolation of vesicles with cup-shape and size around 100–200 nm. In particular, 5 μl of EVs were absorbed on Formvar carbon-coated grids (Sigma-Aldrich) for 10 min and then negatively stained with 2% uranyl acetate (Thermo Fisher) for 10 min. The excess of uranyl was removed with a filter paper, and the dried grids were examined with a transmission electron microscope (Leo 812AB, Zeiss, Oberkochen, Germany) at 80 kV.

2.6. SPRi array preparation

The SPRi-array was prepared on SPRi-bare biochips by means of the SPRi-Arrayer (Horiba Scientific SAS, Palaiseau, FR) working at 70% humidity with a metal-ceramic capillary pin of 0.7 mm, following the optimized protocol described in our recent publication[5]. In particular, the activated chip surface was functionalized with different antibody ligands at 0.5 μg/ml: anti-CD9 (14–0098, eBioscience, Inc, San Diego, CA, USA)), anti-CD171 (14-1719-82, eBioscience), anti-proteolipid protein 1 (PLP1; HBM-PLP-50, HansaBioMed, Tallinn, EE), anti-GLutamate and ASpartate Transporter (GLAST; EAAT1/GLAST-1/SLC1A3 Antibody, NB100–1869SS, Novus Biologicals LLC, Centennial, CO, USA), anti-CD11b (553311, BD Biosciences, San Jose, CA, USA), as well as the ligand isolectin B4 (IB4) from Bandeiraea simplicifolia (L3019, Merck). An anti-rat IgG1 antibody (407402, BioLegend Inc, San Diego, CA, USA) was used as a negative control. We designed 4 separated spots for each ligand family. After ligand immobilization, the chip was blocked in a solution of ethanolamine (1M; pH 9) for 30 minutes and washed with water.

2.7. SPRi measurements

The simultaneous detection of multiple populations of EVs was performed through SPRi measurements with XelPlex instrument from Horiba Scientific SAS (Palaiseau, FR). Freshly prepared HBS-ET (1.5 M NaCl, 100 mM HEPES, 30 mM EDTA, Tween 0.5%, pH 7.4; Sigma Aldrich) was used as a running buffer for each experiment. 500 μl of EVs (pool of samples from the same experimental group) were injected into the flow cell of the instrument, flushing at 25 μl/min on the surface of the chip where ligands have been immobilized. Real time evaluation of interactions between EVs and each ligand was performed through the analysis of SPRi signal intensities reached at the end of the association, for each sample. Glycine (50 mM, pH 2; Sigma Aldrich) was used as regeneration buffer, between EV samples injections (200 μl, 50 μl/min).

2.8. SPRi data analysis

We used EzSuite (Horiba, France) and Origin2018 (OriginLab, USA) software for the analysis of SPRi data. The signal obtained on antibody anti-IgG spots was subtracted from the signals obtained on all the other ligands present on the same chip. Then, we normalized the signal intensities on each ligand for the SPRi intensity collected on anti-CD9 spots (a marker of general EVs) in order to decrease the pre-analytical variability due to the experimental steps.

2.9. Statistical analysis

During data collection and analysis, the experimenter was blinded to the treatment groups using a separate key. Total EV concentrations and nSMase2 activity levels were statistically analyzed using GraphPad Prism 7 [RRID: SCR_002798] (GraphPad Software; San Diego, CA). A one-way ANOVA with Tukey’s post hoc test was used to determine statistical significance. The SPRi statistical analysis was performed using GraphPad Prism and Origin2018 (OriginLab, USA). To determine statistical differences between SPRi signals of the experimental groups, one-way ANOVA with Tukey’s multiple Comparison post-test with 95% confidence intervals was performed. Results were deemed significant when p<0.05.

3. Results

3.1. Oral PDDC results in brain concentrations of drug exceeding its >IC₅₀ for nSMase2

PDDC (Fig 1A) has been previously shown to have excellent oral bioavailability (%F = 88) when administered via intragastric gavage[18]. In this study we incorporated PDDC into mouse chow and conducted a PK study to quantitate PDDC levels in the plasma and brain over a 24h period (Fig. 1B). Time points for drug level measurements were determined based on a 12h light-cycle. Plasma and brain levels of PDDC were determined at 4 time points (0h, 7h, 12h, and 19h). The midnight time-point (0h) was 4h following lights out while the noon time-point (12h) was 4h following lights on. We tested doses of 30mg/kg and 100mg/kg. For the 30mg/kg dose, the total PDDC plasma area under the curve (AUC_0–19h) was calculated to be 43.31 with a brain AUC_0–19h of 31.41 giving a brain to plasma ratio (AUC_brain/AUC_plasma) of 0.73 (Fig. C). For the 100mg/kg dose, the total PDDC plasma AUC_0–19h was calculated to be 251.15 nmol/hr·mL; while the brain AUC_0–19h was 197.86 nmol/hr·mL, resulting in the brain to plasma ratio of 0.79. Both doses gave brain to plasma ratios consistent with the prior report[18] (Fig. 1C). The sustained exposures of free (unbound) PDDC in both brain and plasma exceeded the PDDC IC₅₀ of 300 nM for nSMase2 during the entire 24-hour period for the 100mg/kg dose while the 30mg/kg dose delivered levels below the IC₅₀ (Fig 1D). Therefore, 100mg/kg was chosen for the IL-1β studies below.

Figure 1. PDDC chow provides constant drug exposures in the brain which exceeds its >IC50 for nSMase2.

A) Chemical structure of PDDC [(Phenyl(R)-(1-(3-(3,4-Dimethoxyphenyl)-2,6-Dimethylimidazo[1,2-b]pyridazin-8-yl)pyrrolidin-3-yl)-Carbamate]. B) Schedule of the pharmacokinetic study. Mice were pretreated with vehicle chow to acclimate for two weeks prior to PDDC-containing chow administration. The first time point was taken at midnight (denoted as 0h) with subsequent time points 7h, 12h, and 19h hours later. C) Pharmacokinetic parameters for total PDDC. D) Pharmacokinetic profile of free, unbound PDDC levels. Points represent mean ± SD.

3.2. PDDC normalized the IL-1β-induced increase in brain nSMase2 activity

To evaluate the regulation of nSMase2 activity following intra-striatal IL-1β, nSMase2 activity was measured in the striatum at 2 h post injection. We have previously shown that IL-1β injection causes robust neuroinflammation and a significant increase in the release of brain-derived EVs into the plasma[18, 23]. The schedule of procedures is outlined in Figure 2A. As previously described, mice were pretreated with vehicle chow for two weeks prior to being placed into groups: 1) saline + vehicle chow; 2) IL-1β + vehicle chow; 3) IL-1β + 30mg/kg PDDC chow; 4) IL-1β + 100mg/kg PDDC chow. Six days before either the saline or IL-1β injection, mice were fed either vehicle or PDDC chow. Two hours following striatal injections of saline or IL-1β, mice were sacrificed, and the brains were collected. We observed a significant increase in brain nSMase2 activity in the vehicle treated IL1β-injected mice compared with the saline control group (Figure 2B. Saline = 1.23 × 10⁵ RFU/mg/h; Vehicle = 1.93 × 10⁵ RFU/mg/h; p < 0.0001). We did not observe any decrease in brain nSMase2 activity in the 30mg/kg treated group (Figure 2B. 30mg/kg = 1.97 × 10⁵ RFU/mg/h; p = 0.976). This enhanced nSMase2 activity level was not observed in the 100mg/kg PDDC treated group (Figure 2B. p = 0.993 versus saline; p < 0.0001 versus Vehicle).

Figure 2. Intra-striatal IL-1β increases brain nSMase2 activity and the total number of EVs in plasma; these changes are normalized by PDDC treatment.

A) Experimental schedule. Mice were pretreated with vehicle chow for 2 weeks prior to dosing with either additional vehicle or PDDC chow. Six days later, mice were given intra-striatal injections of saline or IL-1β (0 h). Two hours later, mice were sacrificed, and plasma was collected for EV analysis and the striatum dissected for nSMase2 activity analysis. B) Ex vivo brain nSMase2 activity was significantly increased following IL-1β injection and was completely normalized with 100mg/kg PDDC treatment while 30mg/kg PDDC was unchanged. n = 3/group; assays performed in triplicate. C) Total plasma EVs were significantly increased following IL-1β injection; PDDC treatment completely inhibited the IL-1β-induced increase. Bars represent mean ± SD. ** denotes p < 0.01. *** denotes p < 0.001. **** denotes p < 0.0001. Statistical analysis was done using a one-way ANOVA with Tukey’s post hoc test.

3.3. PDDC normalized the IL-1β-induced increase in plasma EVs

We next assessed whether the PDDC treatment could inhibit EV release in vivo following a striatal injection of IL-1β. As the 100mg/kg dose provided levels of free PDDC above its IC₅₀ in brain and significantly inhibited the IL-1β-induced increase in nSMase2 activity, we conducted the IL-1βexperiments with this dose. Plasma was isolated two hours following striatal injections of saline or IL-1β and the number of EVs was determined. The vehicle group showed a significant increase in total EVs compared to the saline group (Figure 2C. Saline = 0.917 × 10¹² EVs/mL; Vehicle = 3.32 × 10¹² EVs/mL; p = 0.0068). PDDC treatment completely normalized the IL1β-induced EV increase (Figure 2C. PDDC = 0.291 × 10¹² EVs/mL; p = 0.0016 vs Vehicle).

3.4. PDDC normalized the IL-1β-induced increase in neuronal- and oligodendroglial-derived EVs in plasma

The SPRi-based biosensor previously optimized in the Laboratory of Nanomedicine and Clinical Biophotonics of IRCCS Fondazione Don Carlo Gnocchi (Italy) was adapted for the analysis of murine plasma EVs to capture different populations of EVs according to the expression of cell-specific surface antigens with a good sensitivity and specificity [5]. Figure 3A shows examples of the SPRi sensorgrams obtained after the interaction of EVs with anti-CD11b and anti-CD171 antibodies, together with the respective CCD differential images where the lighting spots confirmed the effective absorption of EVs on the chip. Indeed, SPRi is able to detect variations in the refractive index in the immediate vicinity of the chip surface that are related to the mass adsorbed on the chip. These variations occur when ligands interact specifically with injected EVs. The SPRi sensor allows to monitor in real time the interactions across the entire chip surface providing images of the chip surface with a video CDD camera and collecting SPRi signals simultaneously from every active sites of ligands. TEM analysis of the EVs used in these experiments confirmed their preserved integrity, showing the typical cup-shaped morphology and dimensions (~150 nm; Figure 3B).

Figure 3. Plasma EVs detection on the SPRi chip.

A) SPRi sensorgram following the injection of 500 μL of EVs (25 μL/min) on the chip with CD11b and CD171 as ligands (4 spots of each family). Each curve is the average of the signal collected on the spots of the same ligand. CCD differential images of the spots (reflectivity variation) after the interaction with EVs on the SPRi chip. B) TEM images of EVs isolated by size-exclusion chromatography and concentrated with ultracentrifugation from plasma mice involved in the study. Scale bars 500 nm.

Utilizing SPRi analysis we set out to characterize which neural cell type populations were affected by intra-striatal IL-1β and nSMase2 inhibition. The SPRi array used for these experiments was prepared by immobilizing antibodies against CD171, PLP1 and GLAST on a defined area of the chip to capture EVs coming from neurons, oligodendrocytes and astrocytes, respectively. The SPRi signal intensities collected at the end of association, related to the specific interaction between EVs and each ligand, showed that IL-1β induced an increase in the amount of nEVs (CD171+) and oEVs (PLP1+) that circulated in the peripheral blood, while aEVs (GLAST+) were unaltered (Figure 4). PDDC treatment in IL-1β injected mice reverted the amount of CD171+ EVs to control levels (p = 0.007 vs. Vehicle). Indeed, the levels of nEVs in saline and treated groups were not statistically different (Figure 4A. Saline: 1.734 ReS; PDDC: 1.005 ReS). In addition to increasing nEVs, intra-striatal IL1β also increased oEVs and this increase was reduced with PDDC treatment (Figure 4B. Vehicle: 8.281 ReS; PDDC: 4.551 ReS; p = 0.00003). In contrast, neither IL-1β or PDDC affected the number of aEVs (Figure 4C).

Figure 4. Intra-striatal IL-1β increases the circulating neuronal- and oligodendroglial-derived EVs in plasma; this increase is normalized by PDDC treatment.

SPRi signal intensities related to the interaction between EVs and anti-CD171 antibody (A), anti-PLP1 antibody (B) and anti-GLAST antibody (C) spotted on the same chip. The signals are the average of 3 or 4 spots/family with standard deviations indicated. Significant differences were determined performing one-way ANOVA with Tukey’s post hoc test. * (p<0.05), ** (p<0.01) or *** (p<0.001).

3.5. PDDC normalized the IL-1β-induced increase in CD11b+ activated microglial-derived EVs in plasma

Next, we investigated whether IL-1β injection and PDDC altered EVs released from microglia using an SPRi chip where markers of activated microglia (CD11b) and non-activated microglia (IB4) were immobilized. We found that the amount of non-activated microglial-derived EVs (IB4+) was lower in the IL1β vehicle group compared to saline and were further reduced by PDDC treatment (Figure 5A). In contrast, the IL1β vehicle group had significantly increased levels of activated CD11b+ EVs compared to saline animals (p = 0.015). This increase was reduced by PDDC (Figure 5B. Vehicle: 16.410 ReS; PDDC: 5.360 ReS; p = 0.008).

Figure 5. Intra-striatal IL-1β increases the EVs released from activated CD11b+ microglia but not resting microglia; the increase in EVs from activated CD11b+ microglia is normalized by PDDC treatment.

SPRi signal intensities related to the interaction between EVs and IB4 (A) and anti-CD11b antibody (B) spotted on the same chip. C) Ratio between SPRi signal intensity on CD11b spots and on IB4 spots as an indicator of neuroinflammation condition. The signals are the ratio between the average of 3 spots/family with standard deviations indicated. Significant differences were determined performing one-way ANOVA with Tukey’s post hoc test. * (p<0.05), ** (p<0.01) or *** (p<0.001).

As we recently reported, the ratio between the SPRi signal obtained on anti-CD11b and IB4 ligands is informative of the neuroinflammation condition and may serve as a biomarker of pathogenic processes [20]. Interestingly, IL-1β injection caused an increase in neuroinflammation-associated EVs compared to saline mice, where CD11b+ EVs were more abundant compared to IB4+ EVs (Figure 5C. Saline: 5.003; Vehicle: 14.753; p = 0.000066). PDDC treatment reduced this inflammation index, causing a significant reduction of CD11b/IB4 ratio (Figure 5C. Vehicle: 14.753; PDDC: 6.476; p = 0.000096).

4. Discussion

Mice fed PDDC-containing chow achieved sustained levels of PDDC in the brain and plasma over a 24h period, which was well tolerated and exceeded its IC₅₀ for nSMase2, ensuring target engagement (Fig 1). Target engagement was also confirmed by the significant inhibition of endogenous brain nSMase2 enzymatic activity following PDDC-containing chow (Fig 2). Using an intra-striatal IL-1β injection model of acute brain injury, we observed a significant increase in brain nSMase2 enzymatic activity that was completely normalized in mice fed PDDC chow (Fig 2). This increase was not surprising as increased levels of nSMase2 and ceramide have been reported in multiple models of brain injury, neuroinflammation, and neurological disease [6, 13, 28]. Furthermore, IL-1β has been shown to be an activator of nSMase2 in vitro[29] and in brain tissue[30]. IL-1β is also a potent activator of microglia, causing the release of pro-inflammatory cytokines such as TNFα, which has been shown to further elevate nSMase2 activity [31, 32]. Increased nSMase2 activity has been shown to increase ceramide production and EV biogenesis and release[11]. Increased EV release has important implications in neuroinflammation as mEVs[33, 34] and aEVs[23] influence peripheral immune cell infiltration into the brain following acute injury[35]. The role of peripheral immune cells in exacerbating brain injury is a growing area of interest. In response to injury, astrocytes and microglia shed EVs into circulation. aEVs have been shown to elicit an acute-cytokine response (ACR) in peripheral organs, most notably in the liver, which primes peripheral leukocytes to transmigrate into the injury site in brain leading to further tissue damage and behavioral changes [36, 37]. Recently, our group demonstrated that aEVs were mediators of this ACR. Adoptively transferring plasma EVs isolated from mice administered intra-striatal IL-1β increased the expression of cytokines in liver and evoked leukocyte infiltration in the brain. Co-injecting the nSMase2 inhibitor altenusin with IL-1β prevented the adoptively transferred EVs from promoting an ACR [23]. Likewise, a selective knock-down of nSMase2 expression in astrocytes before IL-1β injection prevented the adoptively transferred plasma EVs from eliciting an ACR. EVs isolated from saline injected animals or EVs stripped of protein and miRNA did not induce ACR, indicating that the content of EVs is critical in activating the ACR. Importantly, many immunomodulatory cytokines and chemokines have been reported as EV cargo[38]. nSMase2 inhibitors, including PDDC[18], have been shown to significantly decrease pro-inflammatory cytokines, including IL-1β, IL-6, and TNF-α, in acute brain injury mouse models[19, 23, 33].Therefore, EVs released in response to injury can carry pro-inflammatory cytokines and exacerbate the elevated immune response. Reducing the elevated release of these EVs may therefore be an effective way to minimize any potential damage caused by an overactive immune response both centrally and peripherally.

Confirming previous reports [18, 19, 39], intra-striatal IL-1β injections induced a significant increase in the total number of EVs released into the plasma. We show that this injury-induced increase in EV release was completely normalized by PDDC treatment (Fig 2). While we and others have reported that nSMase2 inhibition can reduce the release of EVs into the plasma following brain disease or insult [15, 17–19, 39], the specific population of all the brain cells affected by nSMase2 inhibition has not yet been explored. Inhibition of microglia- [17] and astrocyte- [18, 19] derived EV release by nSMase2 inhibitors have been studied, however these experiments only focused on a single cell type. Using our SPRi-based sensor and its multiplexing properties [40], we were able to analyze, with good specificity and sensitivity, all the brain cell populations susceptible to intra-striatal IL-1β and PDDC treatment in a single experiment. We observed selective increases in plasma nEVs and oEVs, which were both attenuated by PDDC treatment. In contrast, aEVs remained unchanged by IL1β or PDDC. Interestingly, IL-1β also induced a selective increase in activated (CD11b+) versus resting (IB4+) mEVs, resulting in an increase in the ratio of CD11b+ to IB4+ mEVs. This increase was significantly reduced by PDDC, indicating a potential immune modulatory role. Indeed, the value of CD11b/IB4 ratio has been recently applied in a pilot study about the analysis of multiple circulating EVs and hippocampal volume in AD patients, demonstrating a correlation between the ratio and the total hippocampal volume of patients [20]. This result has shown the potentiality in using this new index to study the severity of neurodegeneration related to inflammatory processes [20]. Although it has not yet been validated and tested in other studies, these preliminary results suggest that the EV related CD11b/IB4 ratio could become a new marker of inflammation for disease monitoring and for the evaluation of the effects of anti-inflammatory drugs. Compared to other studies using CD11b as a single marker of systemic anti-inflammatory effect of drugs [41], we propose the CD11b/IB4 ratio on the EV surface as a more specific parameter for monitoring inflammation processes occurring in the brain.

In addition to acute brain injury, reducing EV release from activated microglia could have broader therapeutic implications including chronic neuroinflammation and neurodegenerative diseases such as AD[42], Parkinson’s disease (PD)[43], and amyotrophic lateral sclerosis (ALS)[44]. There has been a growing interest in targeting microglial activation as a therapeutic strategy, with many studies demonstrating that inhibiting microglial activation [45] or depleting microglia [17] has positive effects on AD pathology in mouse models. Similar results have been observed in PD where reducing microglial activation[46, 47] or depleting microglia[48, 49] led to improved behavioral outcomes and neuronal survival in PD mouse models. Altering microglial activation in ALS is also becoming a major therapeutic target with studies showing positive effects in mouse models where microglia and macrophage activation is reduced either pharmacologically [50] or by carrying out a bone marrow transplant with genetically altered less reactive macrophage [51].

Neurons have also been reported to release EVs in vitro[52, 53], however, the levels of nEVs released into the plasma in response to acute brain injury in vivo have not been well explored. One study quantified the levels of nEVs in mouse brain 24 hrs following a cerebral ischemia/reperfusion injury and reported no change [54]. In contrast, we observed a significant increase in plasma nEVs 2 hrs following IL-1β injection. While appearing to be conflicting, it is possible that neurons release EVs only during the acute phase of neuronal injury as the brains analyzed from the ischemia study was collected 24h following reperfusion versus 2h following injury in the present study. Interestingly, in a clinical traumatic brain injury (TBI) study, the plasma nEVs isolated from patients were toxic to cultured neuron-like cells in vitro, indicating that the nEV cargo may also have important disease implications[55].

Analyzing EV cargo rather than EV number is also a growing area of study as specific cargo have been implicated in propagating neurodegenerative diseases such as hyperphosphorylated tau in AD[17] or α-synuclein in PD[56]. With respect to AD, both nEVs[57] and mEVs[17] have been found to contain seed-competent hyperphosphorylated tau. Interestingly, chronic traumatic encephalopathy (CTE) resulting from repeated TBI exposure overlaps with AD in that toxic accumulation of hyperphosphorylated tau is present in brain [58] and plasma [59, 60] samples from CTE patients. Treating patients with nSMase2 inhibitors therefore may provide benefit in both acute and chronic injury states.

Another brain cell that has been observed to release EVs are oligodendrocytes [61, 62]. While oEVs have demonstrated neuroprotective effects[61, 62], there is also evidence that they contribute to pathological conditions. For example, oEVs isolated following cytokine stimulation were found to exacerbate the cytotoxic effects of IFN-γ on naïve oligodendrocytes in vitro [63]. Additionally, antigen transfer from oligodendrocytes to microglia via oEVs has been proposed as an initial step in inducing myelin sheath destruction in multiple sclerosis (MS) [64]. In further support, oEVs have been shown to inhibit the differentiation of oligodendrocyte precursor cells and myelin formation[65]. Recently, our team showed that inhibiting nSMase2 in a MS mouse model led to improved myelination, although brain or plasma oEVs were not measured[66]. Given these negative roles oEVs play in MS and that PDDC decreased plasma oEVs, nSMase2 inhibition could potentially have dual effects on improving MS via reducing oEV ceramide-mediated apoptosis and myelin sheath destruction.

We did not observe any differences in the levels of aEVs in IL-1β treated mice despite previous observations of the opposite [18]. While perplexing, the experiments differ in aEVs isolation techniques. In our previous study, aEVs were identified by counting GFP+ particles from mice which expressed GFP under the astrocyte-specific promoter glial fibrillary acidic protein (GFAP). In this study, we utilized an antibody against GLAST-1/EAAT1 to quantify aEVs. While both GFAP and GLAST1 are considered markers of astrocytes, neither completely marks all populations. GFAP is an abundant intermediate filament protein expressed in mature astrocytes and is highly upregulated during reactive astrogliosis[67], however, not all cells are GFAP positive and expressing GFP under the GFAP promoter does not completely overlap with GFAP immunostaining, indicating heterogeneity in expression [68]. GLAST expression is found to be higher in immature astrocytes with the overall brain levels decreasing during development[69]. Despite this overall reduction in GLAST expression in the adult brain, certain regions have significant levels in adult brains, such as the cerebellum and striatum[69, 70]. Therefore, the GFAP-GFP reporter may be identifying a larger and stronger signal, especially from reactive astrocytes responding to the neuroinflammation, leading to larger increase in detected aEVs indicating that different subpopulations of astrocytes may release EVs differentially in response to acute brain injury.

Since different brain EV populations have implications in many neurodegenerative disease areas, the ability to monitor these populations easily in patients is increasingly important and can facilitate understanding the mechanisms of action of drugs, monitoring their effectiveness and move to a more personalized medicine. SPRi is emerging as a technique to detect and characterize EVs, allowing vesicle subtype screening on specific microarrays in addition to demonstrating EV presence with CCD differential images on the SPRi biochip. Thanks to its multiplexing capability, it allows the phenotyping of vesicles and the simultaneous tracking of multiple markers on EV membranes all while using limited amounts of sample [71]. Moreover, compared to other analytical methods, SPRi requires limited pre-purification steps as it relies on the specific antigen-antibody interaction, with limited interference of contaminants like lipoproteins. Therefore, the analysis of EV samples through the SPRi technology is highly sensitive and high-throughput, allowing to estimate the relative amount of each specific subtype that could potentially be influenced by the pharmacological treatment.

In summary, we show that brain nSMase2 activity is increased following an acute IL1β-induced brain insult leading to an increased number of EVs in plasma. Using multiplexed SPRi, we demonstrated that the increased EVs in plasma were specifically released from neurons, oligodendrocytes, and activated microglial cells. The recently discovered nSMase2 inhibitor PDDC was capable of completely normalizing these IL-1β-induced changes implicating inhibition of nSMase2 as a therapeutic target for acute brain injury. The presented data also raises the possibility of utilizing SPRi-based biosensors as an innovative approach to quantitate circulating brain-derived EVs as an indicator of neuroinflammation and a liquid biomarker to evaluate the effectiveness of new therapeutics.