Background
Tumor necrosis factor (TNF) is an immunomodulatory molecule known to be implicated in central nervous system (CNS) insults such as stroke [
1]. Immune responses within the CNS, as well as systemic inflammatory events, play important roles in the progression, repair and recovery of stroke, offering new immune-based approaches as future treatment strategies in stroke patients. TNF is present in low concentrations in normal brain tissue and upregulated after ischemia [
1]. It exists both as transmembrane (tm)TNF and soluble (sol)TNF. tmTNF acts through cell-to-cell contact to promote juxtacrine signaling and is important for cellular communication in the innate immune system [
2], but also for functional recovery and axonal preservation [
3], whereas solTNF acts in a paracrine manner and is an important mediator of both acute and chronic inflammation [
4].
Anti-TNF therapies such as etanercept, which blocks both solTNF and tmTNF, are currently used to treat chronic inflammatory diseases [
5],[
6], and appear to relieve fatigue and symptoms of depression associated with chronic diseases [
5]. Furthermore, peri-spinal etanercept has been used with success in stroke and traumatic brain injury patients, where treatment resulted in neurological improvement [
7],[
8]. However, their use is hampered by side effects, including increased risk of sepsis, demyelinating disease, neuropathies, heart failure and also infections [
9], which represents a considerable risk for stroke patients. Since etanercept inhibits both solTNF and tmTNF, this raises the possibility that solTNF-specific inhibitors, sparing tmTNF, have the potential to inhibit deleterious inflammation without compromising the immune system’s response to infections. XPro1595, an engineered dominant-negative TNF that inactivates only solTNF [
10], has proven to be effective in animal models of CNS disorders involving increased TNF production [
3],[
11],[
12], and in attenuating experimental arthritis [
13] and endotoxin-induced liver injury [
14], without suppressing the innate immunity to infection, in contrast to etanercept treatment. The ability of XPro1595 to be tmTNF-sparing and solTNF-selective potentially makes XPro1595 a safer clinical drug than etanercept as it ensures that the role of tmTNF in immune function and myelin preservation is not compromised.
In the present study, we used etanercept and XPro1595 to test the effect of systemic administration on functional recovery, infarct volume, and systemic and central inflammatory responses in a murine model of focal cerebral ischemia.
Materials and methods
Animals
Adult male C57BL/6 mice (between seven and eight weeks of age, n = 256) were purchased from Taconic Ltd. (Ry, Denmark) and transferred to the Laboratory of Biomedicine, University of Southern Denmark, where they were allowed to acclimatize for seven days prior to surgery. Animals were housed under diurnal lighting conditions and given free access to food and water. All animal experiments were performed in accordance with the relevant guidelines and regulations approved by the Danish Animal Ethical Committee (numbers 2011/561-1950 and 2013-15-2934-00924).
Induction of permanent middle cerebral artery occlusion
The distal part of the left middle cerebral artery (MCA) was permanently occluded [
15] under Hypnorm and Dormicum anesthesia (fentanyl citrate (0.315 mg/ml; Jansen-Cilag) and fluanisone (10 mg/ml; Jansen-Cilag, Birkerød, Denmark), and midazolam (5 mg/ml; Hoffmann-La Roche, Hvidovre, Denmark)), respectively. After surgery, mice were injected subcutaneously with 1 ml of 0.9% saline and allowed to recover in a 25°C controlled environment. Mice surviving for five days were returned to the conventional animal facility after 24 hours. For post-surgical analgesia, mice were treated with 0.001 mg/20 g buprenorphine hydrochloride (Temgesic, Schering-Plough, Ballerup, Denmark) three times at eight-hour intervals, starting immediately prior to surgery.
Group size and study design
The size of the ischemic infarct was measured in three separate randomized, double-blinded, vehicle-controlled studies in mice allowed to survive for six hours (n = 30), 24 hours (n = 60) and five days (n = 74) after induction of permanent middle cerebral artery occlusion (pMCAO). In order to evaluate the effect of ischemia on functional outcome and the acute phase response (APR), a group of sham-treated mice were included at all time points (total n = 35). Furthermore, un-manipulated controls were included in flow cytometric and microparticle analyses (total n = 27). A total of 12 mice were excluded due to lack of infarct in mice subjected to pMCAO or presence of unintended infarcts in shams. Mortality was 1.8% and there were no differences in mortality between the different treatment groups.
Pharmacological treatment
XPro1595 [
13] or etanercept (Enbrel, Amgen-Wyeth, Thousand Oaks, CA, USA) were administered intravenously once, at a dose of 10 mg/kg, 30 minutes after surgery. Saline was used as the vehicle. Mice subjected to sham surgery were given an intravenous injection of saline 30 minutes after surgery. The peak concentration in serum (C
max) after murine intravenous dosing of XPro1595 at 10 mg/kg was 945.7 μg/ml and the terminal half-life was 19.1 hours (data not shown).
Physiological parameters
Mice were weighed at the time of pre-training, before surgery, and one, three and five days after surgery. Rectal temperature was measured prior to, and 30 minutes and three hours after surgery.
Behavioral tests
Functional outcomes were evaluated one, three and five days after pMCAO using different behavioral tests designed to detect motor deficits. Prior to behavioral testing, mice were allowed to acclimatize in the behavior room.
Grip strength test
The grip strength meter (BIO-GT-3, BIOSEB, Vitrolles, France) was used to study neuromuscular function in mice subjected to pMCAO and sham surgery. The peak amount of force was recorded in five sequential trials and the highest grip value was recorded as the score [
16]. We analyzed the grip strength in individual (left and right) front paws prior to (baseline) and one, three and five days after pMCAO. The unit of force measured is presented as grams (g). Asymmetry between paws in individual mice following pMCAO were calculated and are presented as delta (Δ) grip strength measured in grams (g). Mice that were allowed to survive for 24 hours were tested on day one, and mice that were allowed to survive for five days were tested on day three and five.
Horizontal rod test
In order to test motor coordination, dynamic balance and asymmetry, mice were placed on the centre of a horizontal rod, located 80 cm above the floor. Mice were allowed to explore and walk the rod for three minutes. The frequency of right and left hind limb slips was recorded and the total distance travelled was tracked using the SMART video tracking software (Panlab, Barcelona, Spain)[
17].
In order to evaluate drug-induced differences, motor coordination and performance and balance [
18], we performed the rotarod test (LE8200, Panlab Harvard Apparatus, Barcelona, Spain). The test comprised a pre-training part prior to surgery (30 seconds at four rotations per minute (rpm)) and a trial part consisting of four trials (T1 to T4) 24 hours or five days after surgery. Mice were placed on the rotarod which was set in accelerating mode. The speed of the rotor was accelerated from 4 to 40 rpm over five minutes. Time spent on the rotarod in each trial for each mouse was recorded.
Tissue processing
Fresh frozen tissue
Mice were killed by cervical dislocation, and brains and livers were quickly removed, frozen in CO2 and stored at −80°C until further processing. Blood samples were collected in EDTA-coated Eppendorf tubes, spun twice for 10 minutes at 3,000 g and 4°C, and stored at −80°C until further processing. Brains were cut coronally in six parallel series of 30 μm and liver samples were cut into 30 μm cryostat sections and stored at −80°C until further processing.
Perfusion fixed tissue
Mice were deeply anesthetized with an overdose (0.15 ml) of pentobarbital (200 mg/ml) containing lidocaine (20 mg/ml) (Glostrup Apotek, Glostrup, Denmark) and perfused through the left ventricle using 4% paraformaldehyde (PFA), as previously described [
19]. Brains from mice with 24 hours survival intended for immunohistochemistry for the granulocyte marker Gr1 (see below) were cut coronally in six parallel series as free-floating 60-μm thick sections and stored in a cryoprotective solution at −12°C (n = 6), or were cut coronally into 12 parallel series as 20-μm thick cryostat sections (n = 5) and stored at −20°C, until further processing.
Flow cytometric analysis
Mice were anesthetized intraperitoneally with an overdose of pentobarbital containing lidocaine and perfused through the left ventricle using phosphate-buffered saline (PBS), as previously described [
19]. Prior to perfusion, 80 μl blood was collected from each mouse using EDTA-coated capillaries and placed in Hanks’ balanced salt solution (HBSS: 0.14 M NaCl, 5.4 mM KCl, 0.4 mM MgSO4•7H2O, 0.4 mM Na2HPO4(anhydrate), 1.3 mM CaCl22H2O, 4.2 mM NaHCO3, 0.4 mM KH2PO4, 0.5 mM MgCl26H2O, and 5 mM glucose) as previously described [
19]. Furthermore, spleen and ipsi- and contralateral cortices were quickly removed and processed as previously described [
19].
Infarct volumetric analysis
Every sixth section was stained with toluidine blue solution (TB: 0.08 M Na
2HPO
4•2H
2O, 0.07 M citric acid, and 0.01% TB (Merck Millipore, Hellerup, Denmark)) for direct infarct volume estimation using the Cavalieri principle, as previously described [
15],[
16]. In addition, in order to correct for edema, the volume of the contralateral and the nonischemic ipsilateral cortex and the volume of injury spanning from 1,080 μm anterior to 1,080 μm posterior of the anterior commissure was compared using an indirect method of infarct volume estimation [
16].
Quantitative PCR
Liver and brain mRNAs were extracted using the RNeasy Mini Kit (Qiagen, Manchester, UK)) according to the manufacturer’s instructions. cDNA was prepared as previously described [
16],[
20] and qPCR analysis was performed using the following conditions: five minutes primer extension at 25°C, followed by 25 minutes reverse transcription at 55°C and finally five minutes enzyme inactivation at 95°C, as previously described [
16],[
20]. Samples were run against standard curves generated from serially diluted cDNA from liver samples obtained from mice subjected to pMCAO. Primer sets were designed by PrimerDesign Ltd. (Southampton, UK) and analyzed using SYBR green as previously described [
20]. Primer sets were: serum amyloid A2 (
SAA2) (forward: TTCATTTATTGGGGAGGCTT and reverse: GCCAGCTTCCTTCATGTCAG), serum amyloid P-component (
SAP) (forward: CAAGGCGGCAGAGTTCAC and reverse: GGAGAGGATTTTTATTTGGC), Chemokine (C-C motif) ligand 2 (
CCL2) (forward: TGAAGTTGACCCGTAAATCTGAA and reverse: AGGCATCACAGTCCGAGTC), interleukin (
IL)
-1β (forward: TGTAATGAAGACGGCACAC and reverse: TCTTCTTTGGGTATTGCTTGG), Chemokine (C-X-C motif) ligand (
CXCL1) (forward: GCTGGGATTCACCTCAAGAAC and reverse: TGTGGCTATGACTTCGGTTTG),
CXCL10 (forward: CATCCCGAGCCAACCTTCC and reverse: CACTCAGACCCAGCAGGAT),
IL-10 (forward: AGGACTTTAAGGGTTACT and reverse: AATGCTCCTTGATTTCTG),
iNOS (forward: GGACAGCACAGAATGTTCCAGAA and reverse: CAAAATCTCTCCACTGCCCCAG), and
TNF (forward: GCCTCCCTCTCATCAGTTCTAT and reverse: TTTGCTACGACGTGGGCTA). Arg1 primers (Mm00475988_m1) were purchased from Life technologies (Nærum, Denmark). Liver results were reported relative to the expression of the housekeeping gene glyceraldehyde phosphate dehydrogenase (
GAPDH) [
20]. All data were normalized to the corresponding sham group, which at all time points represented a mean value of 1. Brain
TNF,
IL-1β and
CD11b mRNA qPCR analyses were performed as previously described [
16].
Immunohistochemistry
Immunohistochemical staining for TNF was performed using the alkaline phosphatase-conjugated rabbit anti-TNF antibody (Sigma-Aldrich, Brøndby, Denmark) as described in Lambertsen
et al. [
21]. Visualization of the Mac-1 antigen (CD11b; AbDSerotec, Copenhagen, Denmark) on fresh frozen sections and the Gr1 antigen (Ly-6G and Ly-6C, BD Biosciences, Albertslund, Denmark) on free-floating vibratome and perfusion fixated sections was performed with the streptavidin and horseradish peroxidase technique [
15],[
22]. Substitution of the primary antibody with serum immunoglobulin (IgG: DakoCytomation, Glostrup, Denmark) or specific isotype controls gave no signal.
Western blotting
Total protein was extracted in 1% lysis buffer (RIPA, Merck Millipore, Hellerup, Denmark) containing a soluble protease inhibitor cocktail (Roche Diagnostics, Hvidovre, Denmark) according to Lambertsen
et al. [
16]. Protein concentrations were estimated using the Bradford Protein Quantification method.
Western blotting analysis for TNF (Abcam, Cambridge, UK, 1:2,000) was performed using 20 μg protein extract separated on bis/tris 4-12% SDS-PAGE gels (Nupage™, Invitrogen, Tåstrup, Denmark) essentially as previously described [
16]. SeeBlue Plus2 pre-stained standard (Invitrogen) was used as a molecular weight marker and 0.5 ng 17 kDa murine recombinant TNF (Sigma Aldrich) was included as a positive control. Densitometry was performed using Image J analysis software (version 1.47, National Institutes of Health (NIH), Bethesda, Maryland, USA) following recommendations of the Image J developers. Analysis was performed on two independent gels with two mice per group.
Flow cytometry
Flow cytometry was performed essentially as previously described [
16],[
19] using FACSVerse (BD Biosciences) and data analyzed using the FACSuite software. TNF
+ microglia (CD11b
+CD45
dim), TNF
+ macrophages (CD11b
+CD45
highGr1
−) and TNF
+ granulocytes (CD11b
+CD45
highGr1
+) were identified as previously detailed [
16],[
19]. Control mice and mice allowed to survive for six and 24 hours after pMCAO were treated intravenously with either saline, XPro1595 or etanercept 30 minutes after surgery.
Prior to fixation, cells were stained for live/dead cells for 30 minutes at 4°C using a Fixable Viability Dye eFluoro 506 (eBioscience, Hatfield, UK) diluted in PBS. A total of 1,000,000 events were collected using forward scatter (FSC) and side scatter (SSC) and analysis of the live/dead gate revealed comparable numbers of dead cells in all the samples. In addition, blood and spleen samples were collected and analyzed for CD45, CD11b, Gr1, and CD3 expression.
Positive staining for TNF (Biolegend, Copenhagen, Denmark), CD11b, CD45, Gr1 and CD3 (BD Pharmingen, Albertslund, Denmark) was determined based on fluorescence levels of the respective isotype controls (Biolegend and BD Pharmingen). The mean fluorescence intensity (MFI) was calculated as the geometric mean of each population in the TNF, CD45 and CD11b positive gates, respectively.
Estimation of polymorphonucleated cells within the infarcted cortex
The number of polymorphonuclear cells/mm2 as a measure for granulocyte infiltration six and 24 hours after pMCAO was estimated based on nuclear morphology using TB-stained sections. In practice, calibrated high-power fields (40×) located within the infarct area, and spanning from 1,080 μm anterior to 1,080 μm posterior of the anterior commissure, were photographed and manually counted by a blinded observer on a minimum of 10 frames from each mouse.
Microvesicle analysis
In total, 100 μl of plasma was diluted 1:10 in Dulbecco’s sterile filtered PBS (Sigma Aldrich) and centrifuged at 30,000 g for one hour to remove interfering lipoprotein particles. The pellet was resuspended in 100 μl PBS containing 0.1% bovine serum albumin (Sigma Aldrich) diluted 1:10 in PBS immediately prior to analysis. Microvesicle size and concentration were determined by Nanoparticle Tracking Analysis (NTA) using a NS500 analyser equipped with a 488 nm laser and NTA software (Nanosight Ltd, Espoo, Finland) as previously described [
23]. Analysis settings were standardized using 100 nm colloidal silica microspheres (100, 150, 300 and 400 nm; Polysciences, Eppelheim, Germany) and these data were used to verify size measurements and calibrate concentration measurements. Five 30-second videos were made for each sample. The sample was advanced with a five-second delay between each recording using the script control facility. The videos were analyzed in
batch process mode using automatic blur and minimum expected particle size with an automatic detection threshold level 10, after visually checking that the five size profiles on the screen were in concordance.
Data analysis
Quantitative data are presented as means ± standard error of mean (SEM). Weight and temperature analyses were performed using two-way repeated measures (RM) analysis of variance (ANOVA). Infarct volumetric analysis, qPCR, flow cytometry, grip strength and microvesicle analyses were performed using one-way ANOVA. Grip strength asymmetry and horizontal rod analyses were performed using paired t-tests. Pearson correlation analysis was used to analyze correlations between liver chemokines and between microvesicle counts and infarct volumes. All statistical analyses were followed by the appropriate post-hoc test and performed using Prism 6 software for Macintosh (GraphPad software, La Jolla, CA, USA) and considered significant at P ≤0.05.
Discussion
In the present study, we found that in a mouse model of focal cerebral ischemia, two different TNF inhibitors improved functional outcome, modified the hepatic APR, changed microglial CD45 expression in the neocortex and affected microvesicle numbers in the serum, without affecting lesion volume. Selective inhibition of solTNF using XPro1595 had comparable effects with non-selective inhibition of both solTNF and tmTNF using etanercept, suggesting that solTNF plays an important role in the observed effects.
The finding that anti-TNF therapy did not have an effect on lesion volume, but improved behavioral outcome in the present study, has previously been shown in an animal model of hemorrhagic stroke where they used etanercept [
30]. Also, in a model of transient focal cerebral ischemia in mice, Sumbria
et al. [
31] showed that systemic injection of etanercept had no effect on lesion volume. We also recently showed that anti-TNF therapy had no effect on lesion size after spinal cord injury when administered systemically, however when anti-TNF therapies were administered epidurally for three consecutive days using mini-osmotic pumps, we observed both a reduction in lesion size and an improvement in functional outcome in XPro1595-treated mice, but not in etanercept-treated mice [
12]. These findings suggest that anti-TNF therapies have to be administered directly to the lesioned CNS in order to affect lesion size. Recent findings that systemically administered XPro1595 (10 mg/kg) does indeed cross the blood-brain barrier (BBB) [
32] and the findings in the present study that CNS TNF levels are reduced six hours after pMCAO in anti-TNF-treated mice suggest that both etanercept and XPro1595 may have crossed the BBB, however this could also be the result of endothelial dysfunction rather than transport across the BBB [
33]. Future studies using mini-osmotic pumps following experimental stroke are needed in order to clarify whether XPro1595 can also reduce lesion size after experimental stroke.
One-way etanercept has been shown to improve behavior is through anti-nociceptive effects. Boettger
et al. showed improved locomotor and pain-related behavior after etanercept treatment in a rat model of chronic antigen-induced arthritis, even with no resolution in joint swelling and inflammation, and suggested that reduction of the effect of peripheral TNF on pain fibers contributed to pain relief [
34]. This could also be the case in the present study, however, this will require further investigation. The suppression of granulocyte recruitment to experimental stroke lesions has previously been shown to reduce infarct volume and reduce cell death, but this has usually been linked to associated reductions in infarct volume [
35]. A reduction of granulocyte infiltration into the lesioned brain in animals treated with etanercept has been suggested to be mediated via etanercept’s effect on the APR in the liver [
28]. In the present study, we found that anti-TNF therapy altered the APR in the liver, and more specifically the mRNA expression of chemokines associated with granulocyte recruitment, such as CXCL10 and CXCL1.
Systemic injections of etanercept have also previously been found to attenuate traumatic brain injury by ameliorating neurological and motor dysfunction and by initially reducing brain TNF protein levels [
36],[
37]. Despite unaltered brain TNF mRNA levels and comparable numbers of TNF
+ microglia and leukocytes at 24 hours in the different experimental groups, we found that brain TNF protein levels were decreased six hours after pMCAO in mice treated with anti-TNF therapy, which is in line with the mechanism of action of both XPro1595 and etanercept neutralizing TNF at the protein level. In the present study, TNF
+ cells were located in the infarct and peri-infarct at six hours in all three experimental groups, but to a much lesser extent in XPro1595- and etanercept-treated mice than in saline-treated mice. Furthermore, the morphology of TNF
+ cells in XPro1595- and etanercept-treated mice were more glial-like, whereas the morphology in saline-treated mice were mixed glial-like and macrophage-like, suggesting that the reduced levels of TNF in the anti-TNF-treated mice could be due to a peripheral reduction in TNF produced by infiltrating macrophages.
Etanercept has been suggested to ameliorate microglial activation [
38], however, in the present study, we found no effect of anti-TNF therapy on CD11b expression, whereas microglial, but not macrophage, CD45 expression was increased in the ipsilateral hemisphere 24 hours after anti-TNF treatment. This, combined with the increased number of CD11b
+CD45
dim microglia 24 hours after pMCAO, suggests that anti-TNF therapy either increases microglial proliferation in the brain or increases the surface expression of the CD45 marker on microglia as a response to treatment in the ischemic brain. Previous studies have shown that increased CD45 expression is involved in ‘microglial alertness’ and activation following injury to the CNS, and that an increase in this surface protein likely reflects a response to ongoing neuroinflammation [
39], suggesting that anti-TNF treatment may induce increased activation of microglia.
Microvesicles have been described as important mediators of intercellular communication and are emerging as potential biomarkers of tissue damage. Interestingly, microvesicles have been found to propagate inflammatory signals [
40], and the subtypes of endothelial microvesicles from stroke patients have been shown to correlate with lesion volume and functional outcome [
41]. Importantly, in the present study, microvesicle numbers in saline-treated mice 24 hours after pMCAO were found to correlate significantly with infarct volume, which was not the case in either etanercept- or XPro1595-treated mice, suggesting an altered response due to anti-TNF therapy. In line with this, we observed increased numbers of microvesicles in anti-TNF-treated mice five days after pMCAO. The microvesicles showed characteristics of shed microvesicles due their mean diameter of 200 nm [
42]. As the increase in microvesicle number does not appear to correlate to the infarct volume in anti-TNF-treated groups, we speculate that the vesicles contribute to an altered inflammatory response in the brain, as microvesicles represent an important means of intercellular communication between cells, serving as transfer vehicles for proteins, lipids, RNA and microRNA [
42]. In addition, microvesicles are known to exchange information with endothelium, thereby actively regulating vascular function or participating in vascular rearrangement [
43]. However, their precise functions are still not fully understood.
Based on the data presented here, we suggest that XPro1595 and etanercept improve functional outcome in mice subjected to focal cerebral ischemia by altering the peripheral immune response leading to decreased infiltration of granulocytes into the infarct, potential altered microglial alertness and, most likely, an improvement in motivational state. The finding that XPro1595 was just as efficient as etanercept in improving functional outcome and altering the APR suggests that solTNF (and not tmTNF) is principally involved in peripheral inflammation after stroke. Finally, previous studies have indicated a direct correlation between TNF and blood pressure in hypertensive humans [
44] and have reported a decrease in blood pressure in an experimental model of systemic lupus erythematosis, a chronic inflammatory disorder with prevalent hypertension, following etanercept administration weekly for a duration of four weeks [
45]. In the present study, we only administered etanercept once, 30 minutes after induction of experimental stroke, which did not appear to affect either the mortality or health status of the mice, however, whether etanercept did indeed reduce blood pressure in the present study remains to be elucidated in future studies.
All together, these findings may have important implications for future treatments of solTNF-mediated diseases, where anti-TNF therapy targeting both solTNF and tmTNF can be substituted with drugs only targeting solTNF, potentially resulting in less severe side effects for the patient, including demyelinating diseases and infections.
Acknowledgements
We acknowledge skilled technical help from technicians Louise Lykkemark, Dorte Lyholmer, Signe Marie Andersen and Sussanne Petersen.
This work was supported by research grants from the Lundbeck Foundation, Copenhagen, Denmark (KLL (R54-A5539) and BHC (R67-A6383)), the Novo Nordisk Foundation, Hellerup, Denmark (R153-A-12550 and R168-A14120) and the Carlsberg Foundation, Copenhagen, Denmark (KLL, 2007_01_0176). Fonden til Lægevidenskabens Fremme is also acknowledged for their financial support.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made.
The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Competing interests
DES is an employee of Xencor and holds stock and stock options in the company. All other authors have no financial conflicts of interests.
Authors’ contributions
KLL and BHC conceived the studies, designed experiments, performed experiments, data analysis and wrote the paper. MD, NAM, YC, LK, MO, M-LBM, HBG, CG, HGP, TD, DCA performed experiments, analysed and interpreted data. DES provided the XPro1595 and etanercept and provided useful input to the drafting of the paper. IILS assisted with microvesicle experiments and BF with brain qPCR analysis. All authors read and approved the final manuscript.