Background
Neuroinflammation is a common pathological feature of neurodegenerative diseases such as Alzheimer’s disease (AD), Parkinson’s disease (PD), frontotemporal lobar dementia (FTD), and amyotrophic lateral sclerosis (ALS) and is represented by glial activation and pro-inflammatory cytokine production by the central nervous system (CNS)-resident cells [
1]. Epidemiological studies have revealed that therapeutic use of nonsteroidal anti-inflammatory drugs (NSAIDs) in humans reduces the risk of developing AD and PD [
2‐
4], suggesting that components of the inflammatory pathways may be involved in the pathogenesis of neurodegenerative diseases. Clinical trials designed to slow AD progression by targeting these inflammatory pathways have failed to provide evidence of efficacy [
5]; however, experimental findings from rodents and genome-wide association studies (GWAS) suggest a direct association between neuroinflammation and the development and progression of neurodegeneration [
6‐
12].
Neuroinflammation is a phenomenon that frequently occurs with systemic inflammatory conditions, such as rheumatoid arthritis [
13], sepsis [
14], type 2 diabetes [
15], and obesity [
16]. This suggests a link between the peripheral immune system and the neuroimmune system; however, the underlying mechanisms of this crosstalk remain largely unknown. Previous experimental studies have shown that peripheral administration of lipopolysaccharide (LPS), the major outer membrane component of gram-negative bacteria, or high-fat diet (HFD) feeding induces a profound innate immune response not only in peripheral organs but also in CNS in rodents [
17‐
22]. Exosomes are a group of cell-derived vesicles that are 30–100 nm in size, produced by most cells in the body, and capable of carrying various cargo molecules including mRNA, microRNA, lipids, and proteins. When these vesicles are released into the circulatory system, they are transported to distal sites and internalized into target cells through endocytosis or membrane fusion [
23,
24]. Exosomes are the new mediators in cell-to-cell communication between neighboring cells or distant cells. We hypothesize that the blood-borne exosomes may be involved in the crosstalk between peripheral and CNS immune system.
Methods
Animals
C57BL/6J mice were obtained from Jackson Laboratories. Mice were housed and bred in the animal care facility at the University of Tennessee Health Science Center under a 12/12 h light/dark cycle with ad libitum access to food and had similar body weights unless otherwise specified.
LPS injection
In the LPS model, mice of age 3–4 months were injected intraperitoneally (i.p.) with either LPS (Escherichia coli 055:B5, Sigma-Aldrich, St Louis, MO, USA) or phosphate-buffered saline (PBS); in some experiment, we used saline as control. In previous studies of CNS inflammatory response using the LPS model, an i.p. injected dose of LPS between 0.5–10 mg/kg was generally used to induce a whole body immune response, which included neuroinflammation. In the current study, to study the effects of the temporal- and dose-dependent changes on CNS inflammation, we implemented additional treatment regimens: (1) mice were euthanized 1, 6, and 24 h after a low dose i.p. injection of LPS at 0.5 mg/kg; (2) mice were euthanized after 7 days of daily low dose i.p. injection of LPS at 0.5 mg/kg; (3) mice were euthanized 1 and 6 h after a high dose i.p. injection of LPS at 10 mg/kg. Mice receiving low-dose LPS (0.5 mg/kg) did not show any behavioral changes, while the mice receiving high-dose LPS (10 mg/kg) became severely ill and immobile 3–4 h following injection. To rule out overnight dehydration as a confounding variable, we removed the 24-h time-point in the 10 mg/kg regimen group from our study design. Donor mice receiving i.p. injection of LPS at 5 mg/kg for 24 h were specified in corresponding experiments. Equal numbers of male and female mice were included.
Exosome isolation, quantification, size and zeta potential measurement, and labeling
Exosomes were isolated from the sera by the ExoQuick serum exosome precipitation solution according to the manufacturer’s instructions (EXOQ5A-1, Systems Biosciences, San Francisco, CA, USA). For differential ultracentrifugation as used in Fig.
2e–g, the serum samples were centrifuged at 20,000×
g at 4 °C for 30 min to remove debris and were then centrifuged at 100,000×
g at 4 °C for 2 h using Sw50.2Ti rotor. The pellets which should be enriched for exosomes were resuspended in 1× PBS. The suspension was centrifuged again at 100,000×
g at 4 °C for 2 h. We purified exosomes from kit unless stated otherwise. For exosome protein quantification, the exosomes were resuspended in PBS (1/5 volume of the input serum) and a portion of the suspension was mixed with 2× radioimmunoprecipitation (RIPA) buffer. The lysates were centrifuged at 12000×
g at 4 °C for 10 min. The supernatant was subsequently quantified by Pierce™ BCA protein assay (23225, Thermo Fisher Scientific, Waltham, MA, USA). For exosome size and zeta potential measurement, the exosome pellets were resuspended in DNase/RNase-free water; size and zeta potential determination of isolated exosomes was performed using Zetasizer Nano-Z (Malvern Instruments, Worcestershire, UK). Exosome labeling was performed using an Exo-Glow Exosome Cargo Labeling Kit according to the manufacturer’s instructions (EXOG200A-1, Systems Biosciences, San Francisco, CA, USA).
Administration of LPS-stimulated blood exosomes from donor to recipient mice
Whole blood (700–800 μl) from mice treated either with or without LPS was collected by cardiac puncture. Blood was centrifuged at 2000×
g for 10 min after sitting undisturbed at room temperature for 30 min to separate the serum. Purified exosomes from the sera were then resuspended in 200 μl of sterilized 1× PBS and passed through a 0.22-μm filter before tail-vein injection to the recipient mice. In our pilot study, we transfused three different doses (500 μg, 1 mg, and 1.5 mg) of exosomes from mice treated with high-dose (10 mg/kg, 6 h) LPS and found that both 1 and 1.5 mg of exosomes induced significant increase of microglia activation in the recipient mice while 500 μg failed to achieve such effect. We therefore used a 1 mg exosome dose throughout the remainder of the study. For the follow-up experiments of infusing LPS-exosomes to recipient mice via the intracerebroventricular (i.c.v.) route, whole blood was collected from the donor mice and the exosomes were isolated from sera using ExoQuick kit and resuspended in sterile PBS (1 mg in 10 μl of saline). Intraventricular infusion of 1 mg of exosomes in 10 μl was performed via implanted cannula, as described in our previous work [
25], over a course of 1 h at the speed of 0.16 μl/min. Mice were all perfused 24 h after receiving exosomes followed up by immunohistochemistry examination of inflammatory markers.
Immunofluorescent staining
After perfusion, mouse brains were dissected and fixed in 4% paraformaldehyde/PBS for 24 h followed by a cryopreservation step in 30% sucrose/PBS for 3 days. Brains were embedded in optimal cutting temperature (OCT) compound (#23-730-571, Fisher Scientific, Hampton, NH, USA) and sectioned coronally on a Leica CM1850 cryostat at a thickness of 20 μm. Sections were attached to microscope slides (#12-550-15, Fisher Scientific, Hampton, NH, USA) and allowed to dry on a slide warmer for an hour at 37 °C. Tissue sections were permeabilized with 0.1% Triton X-100/PBS for 10 min at room temperature. Non-specific bindings were blocked using 5% goat serum/PBS for an hour at room temperature. Primary antibodies diluted in antibody dilution buffer (#25886-05, Electron Microscopy Sciences, Hatfield, PA, USA) were applied to slides for overnight incubation at 4 °C. Antibodies used in this study include: rabbit anti-Iba-1 (#019-19741, Wako Laboratory Chemicals, Richmond, VA, USA; 1:1000), mouse anti-GFAP (G3893, Sigma-Aldrich, St Louis, MO, USA; 1:500), rat anti-CD68 (MCA1957, Bio-Rad Laboratories, Hercules, CA, USA; 1:200), and rabbit anti-S100 (Ab41548, Abcam, Cambridge, MA; 1:1000). On the second day, slides were washed for 5 min × 3 times with PBS at room temperature, followed by the incubation of an Alexa Fluor secondary antibody (Thermo Fisher Scientific, Waltham, MA, USA; 1:200) for an hour at room temperature. After three washes, the sections were counterstained with DAPI, air dried, and mounted with Fluoromount-G (# 0100-01, SouthernBiotech, Birmingham, AL, USA). Our pilot study revealed that the Exo-Green dye faded quickly after use, therefore the brains were rapidly isolated, snap frozen with crushed dry ice. The fresh frozen brains were embedded in OCT compound and sectioned coronally at 20 μm. Sections were mounted on microscope slides and briefly fixed in 4% paraformaldehyde/PBS for 15 min followed by three PBS washes. After tissue permeabilization and blocking, the sections were incubated with primary antibody overnight at 4 °C.
Digital image quantification
Images with a resolution of 4080 × 3072 pixels were captured with an Olympus IX50 microscope (Olympus Corporation, Shinjuku, Tokyo, Japan). Images were acquired within the hippocampal and neocortical regions. A total of ~ 50 coronal sections (20 μm per section) through the hippocampal region of one animal brain were cut, and one section from every eight sections were collected to constitute a six-section serial set from which we quantified the staining intensity. Before image acquisition, we generally looked through each section to see if the hippocampal and neocortical structures were intact and if the section was flattened but not folded up. Six to eight images per structure per animal were captured, covering as many sections as possible from the six-section serial set. Exposure time was manually adjusted to minimize saturation while maintaining adequate signal-to-background contrast. Images from the same antibodies were acquired using the same exposure time. The images were converted to 680 × 512 pixels and analyzed using ImageJ software (National Institutes of Health, Bethesda, MD;
http://imagej.nih.gov/ij/). Image files were converted to RGB stacks and color-inverted, converting the color images to grayscale mode. Then, we adjusted the threshold of the images to define the positive signals from the surrounding background. The total staining intensity were expressed by integrated intensity (i.e., mean gray value × area) using the ROI manager function of ImageJ. Areas of positive aggregates were identified between 3 to 500 pixel^2 with circularity of 0 to 1. Areas of false-positive aggregates with no cell morphology were manually excluded. The average value from the six to eight images was used to represent the value from one mouse brain. The number of CD68+ cells was counted manually.
Quantitative real-time polymerase chain reaction (qRT-PCR)
Either 100 μl of whole blood, 100 mg of liver, or half forebrain was used as input. Total RNA was isolated using Trizol reagent (Invitrogen, Carlsbad, CA, USA). For mRNA analysis, the cDNAs were synthesized using High-Capacity cDNA Reverse Transcription Kit (4368814, Thermo Fisher Scientific, Waltham, MA, USA). For brain microRNA analysis, the cDNAs were synthesized by miScript II RT Kit (218,161, Qiagen, Germantown, MD, USA). For exosomal microRNA analysis, the cDNAs were synthesized by Complete SeraMir Exosome RNA Amplification kit (RA800A-1, Systems Biosciences, San Francisco, CA, USA) according to the manufacture’s protocol. An equal volume of input serum was used for the same batch of samples. The expression of microRNA was normalized to a synthetic external spike-in small RNA control (RA805A-1, Systems Biosciences, San Francisco, CA, USA). Detection of mRNA or microRNA were performed using with SsoAdvanced™ Universal SYBR® Green Supermix (Bio-Rad Laboratories, Hercules, CA, USA) and an Eppendorf Mastercycler realplex Real-Time PCR system. Primers used were the following: Tnf, forward 5′-CCCTCACACTCAGATCATCTTCT-3′, reverse 5′-GCTACGACGTGGGCTACAG-3′; Il6, forward 5′-AGTTGCCTTCTTGGGACTGA-3′, reverse 5′-TCCACGATTTCCCAGAGAAC-3′. Their expressions were normalized to β-actin: Actb, forward 5′-CTAAGGCCAACCGTGAAAAG-3′, reverse 5′-ACCAGAGGCATACAGGGACA-3′. Primers for mature microRNAs were designed using the mouse sequence from miRBase. Their expressions were normalized either to the spike-in small RNA control in exosomal microRNA analysis, or to 5S rRNA: 5S, forward 5′-GCCCGATCTCGTCTGATCT-3′; reverse 5′-GCCTACAGCACCCGGTATC-3′ in brain microRNA analysis.
Western blot analysis
Around 20 μg of protein per sample was used for western blot analyses performed as described previously [
26]. Antibodies used includes anti-TSG101 antibody (SC-7964, Santa Cruz Biotechnology, Dallas, TX, USA) and anti-β-Actin antibody (#A2228, Sigma-Aldrich, St Louis, MO, USA).
Enzyme-linked immunosorbent assay (ELISA)
TNF-α levels were measured using Mouse TNF-α Quantikine ELISA Kit according to the manufacturer’s instructions (MTA00B, R&D Systems, Minneapolis, MN, USA).
HFD feeding
In the diet-induced obesity model, 6-week-old male mice were fed either a HFD containing 60% calories from fat (TD. 06414, Harlan Teklad, Indianapolis, IN, USA) or a standard chow diet for 16 weeks. Mouse body weights were measured before and after the 16 weeks of diet feeding.
Statistics
A Kolmogorov–Smirnov test was used to determine if the values were normally distributed. Statistical analysis was performed using one-way ANOVA analysis followed by a Bonferroni post hoc test for multiple comparisons. Student’s t test was performed when comparing two means. P values of < 0.05 were considered significant. Graphs were analyzed and prepared using GraphPad Prism 5.0 (La Jolla, CA, USA).
Discussion
In this study, we investigated a specific mechanism of peripheral contribution to neuroinflammation. Using a LPS model, we show that the microglial cells and astrocytes in mouse brain are activated in a dose- and time-dependent manner by peripherally administered LPS. This observation is consistent with many other reports [
33]. A higher dose LPS (> 3 mg/kg) challenge has been known to induce blood-brain barrier (BBB) disruption, followed by leukocyte recruitment in the brain and CNS inflammatory response [
34‐
36]. While most research in general utilize doses between 0.5–10 mg/kg, a dose of LPS as low as 0.1 mg/kg administered intraperitoneally is sufficient to induce microglial activation in mice without compromising the BBB components [
37]. Using different systemic LPS treatment regimens: single low- (0.5 mg/kg) and single high (10 mg/kg) dose, as well as multiple injections (0.5 mg/kg daily for 7 days), we compared the effect of dose, treatment time, and multiple challenges on glial activation. LPS can activate the CNS immune pathways and processes independently of BBB disruption [
36]. A similar conundrum has been present in the field of AD [
38]. Efforts to unravel this mysterious tangle have been substantially undermined in part due to the historical concept of “CNS immune privilege.” Here, we present novel findings that the serum-derived exosomes, when transfused from LPS-challenged mice into C57BL/6J mice via tail-vein enhance microglial and astrocytic activation and increase the expression of inflammatory cytokines in the brain. In addition, we compared the effects of serum-derived exosomes from LPS-challenged donor mice in recipient mice through i.v. and i.c.v. routes. Of note, the i.c.v. infused LPS exosomes induced comparable level of microglial activation in the hippocampal regions with i.v. injection (Figs.
5c and
9b, c and Additional file
2: Figure S2). Our observations suggest that exosomes may be one factor mediating the activation of neuroinflammatory process during systemic peripheral inflammation. Our results may also shed light on mechanisms of communication between the brain immune surveillance and the peripheral immune system.
Exosomes are taken up by the acceptor cells mainly through endocytosis and membrane fusing [
39]. It has been reported that the blood-borne exosomes are internalized by dendritic cells or macrophages through phagocytosis, a specific form of endocytosis [
24]. Using the fluorescently labeled exosomes, we are able to observe an accumulation of green fluorescence in brain cells 6–24 h after intravenous injection, and mainly in the microglia. However, the number of cells with noticeable fluorescent signal in brain parenchyma was surprisingly low; the green fluorescent intensity was strongest in the areas around the third and the lateral ventricles, especially in the ependymal cells. Our findings are consistent with several reports on exosome biodistribution, in which the majority of fluorescently labeled or radiolabeled exosomes injected intravenously are accumulated in the liver, spleen, and lung within less than 1 h, while barely detectable in brain [
40‐
43]. In one of the reports, using an in vivo nanoparticle imaging system, the authors show that brain accumulates detectable level of fluorescently labeled exosomes around the third ventricle and lateral ventricle areas at 24 h after intravenous injection, but to a much lesser extent in comparison with that in the liver, spleen, and lung [
42]. Considering serum-derived exosomes are heterogeneous in size and source, exosomes derived from different cell types may have distinct tissue-specific homing effects [
44]. The observed low efficiency of brain exosome uptake may be attributed to a small population of exosome subtypes which preferentially target to the brain in the variety of serum-derived exosomes. Moreover, the prominent microglial uptake may also imply a predominant source of donor exosomes secreted from peripheral monocyte-macrophage-lineage cells, which would be consistent with the common knowledge that LPS response is mediated by the myeloid cells. While the level of astrocytic exosome uptake was low, the observation of astrocytic activation by exosome infusion is clear. We speculate that the astrocytic activation was a secondary event of the activated microglia, as recently reported [
45]. It should be noted that the fluorescent or radioactive tracer, which is a lipophilic or membrane permeable chemical, may also alter the nature and tissue distribution of exosomes. A more suitable and reliable/stable labeling strategy is under development to further investigate the exosome homing cell types and mechanisms. Another possible explanation for the low percentage of green fluorescence incorporated neural cells after the i.v. injection of labeled exosomes might be that the peripheral exosomes need to be first packaged by a certain cell type such as epithelial cells or ependymal cells and released again before they can be taken up by the neural cells. Although we cannot rule out possible involvement of other intermediate cell types such as the epithelium lining up the blood CSF brain barrier to be the major sites for the uptake and repackage of peripheral exosomal contents, the observation that i.c.v. infused LPS exosomes induce similar degrees of neuroinflammation implies a direct causative role of peripheral exosomes.
In addition to the LPS model, we also employed the HFD-induced obesity model. HFD has long been known to induce gliosis in the hypothalamus [
18,
29‐
32]. Gliosis was increased as early as 1 day after the feeding, when there was no leukocyte recruitment to the brain [
46]. The effect of HFD on microglial activation in the hippocampus areas has also been reported after chronic feeding [
19]. In line with these previous studies, we showed that HFD feeding for 16 weeks in C57BL/6J mice increased microglial activation in the hippocampus and neocortex. Similarly, the exosomes purified from HFD-fed sera acutely increased the microglial response in the two brain regions after 24 h, providing a proof-of-concept that exosomes mediate the neuroinflammation in HFD-induced obesity. One limitation of the study was that the amounts of exosomes applied may overestimate the pathophysiological amounts. Considering that development of neuroinflammation during HFD feeding is a chronic long-term process, one can infer that serum exosome uptake by the brain cells is also a continuous process along the feeding period. It is possible that the brain accepts negligible or minimal amounts of serum-derived exosomes on a daily basis. Future work will be needed to examine the effect of a long-term treatment regimen (e.g., repeated injections) with low-dose exosomes in order to more closely approximate the physiological conditions.
It is also unclear what the sources of these brain-targeting serum-derived exosomes are, as exosomes can be secreted by almost all types of cells. Identifying the sources is a challenging feat, but not an impossible one. Cell type specific markers either on the exosome membrane or within the exosomes can be used as a way to identify the host cells; however, current research in this field is primitive [
47]. The observations that microglia being the major subtype of glial cells affected by LPS-exosomes strongly suggest that the effective exosomes are from a myeloid source, consistent with the notion that mononuclear and dendritic cells are the major mediators for LPS-induced systemic inflammation.
In our study, exosome abundance in the sera was not affected by systemic inflammation, as their protein concentrations in serum-derived exosomes were unaltered, suggesting that it was the identity changes in the contents (i.e., mRNA, microRNA, protein, or lipid) in the LPS-treated exosomes that caused systemic and CNS inflammation. Our study pioneered research that characterizes detailed time-course response of the circulating exosomal microRNAs under LPS challenge. We show that the serum-derived exosomes purified from mice with LPS challenge (< 1 h) contained elevated expression of inflammation-related microRNAs, including the miR-21, miR-125a, miR-146a, and miR-155, all of which have been involved in regulating Toll-like receptor (TLR) signaling [
48].