Address correspondence and reprint requests to Rong-Yuan Zheng, MD, Department of Neurology, the First Affiliated Hospital and Research Institute of Experimental Neurobiology, Wenzhou Medical College, Wenzhou, Zhejiang, 325000, P.R. China. E-mail: email@example.com or Jiang-Fan Chen MD, PhD., Department of Neurology, Boston University School of Medicine, 715 Albany Street, C329, Boston, MA 02118. E-mail: firstname.lastname@example.org
Studies with multiple sclerosis patients and animal models of experimental autoimmune encephalomyelitis (EAE) implicate adenosine and adenosine receptors in modulation of neuroinflammation and brain injury. Although the involvement of the A1 receptor has been recently demonstrated, the role of the adenosine A2A receptor (A2AR) in development of EAE pathology is largely unknown. Using mice with genetic inactivation of the A2A receptor, we provide direct evidence that loss of the A2AR exacerbates EAE pathology in mice. Compared with wild-type mice, A2AR knockout mice injected with myelin oligodendroglia glycoprotein peptide had a higher incidence of EAE and exhibited higher neurological deficit scores and greater decrease in body weight. A2AR knockout mice displayed increased inflammatory cell infiltration and enhanced microglial cell activation in cortex, brainstem, and spinal cord. In addition, demyelination and axonal damage in brainstem were exacerbated, levels of Th1 cytokines increased, and Th2 cytokines decreased. Collectively, these findings suggest that extracellular adenosine acting at A2ARs triggers an important neuroprotective mechanism. Thus, the A2A receptor is a potential target for therapeutic approaches to multiple sclerosis.
Clinical and experimental evidence indicates that adenosine receptor pathways are involved in modulation of neuroinflammation in MS. MS patients exhibit decreased expression of A1 receptors (A1R) in mononuclear cells in peripheral blood as well as decreased adenosine levels in blood plasma, accompanied by increased tumor necrosis factor-alpha (TNF-alpha) (Mayne et al. 1999; Johnston et al. 2001). Expression of A1Rs on CD45-positive glial cells in the brain also decreased by 50% in MS patients compared with healthy adults and patients with other CNS disease (Johnston et al. 2001). The A1R is known to have anti-inflammatory properties, so the decrease in A1R expression in MS suggests that an increase in pro-inflammatory cytokines may contribute to the pathogenesis of MS. Consistent with this notion, the spinal cord injury induced by myelin oligodendroglia glycoprotein (MOG) in EAE is exacerbated in A1R knockout (A1R-KO) animals compared with their wild-type littermates (Tsutsui et al. 2004). A1R-KO mice had much higher clinical scores for EAE, greater demyelination of spinal cord, and an increase in expression of pro-inflammatory cytokines (IL-1β, iNOS and MMP) compared with wild-type (WT) littermates. In addition, our recent study showed that chronic treatment with caffeine attenuated EAE pathology, as shown by the reduction in EAE incidence, clinical scores, inflammatory cell infiltration, and demyelination in the brain. Chronic caffeine also decreased interferon-γ mRNA and increased TGF-β1 mRNA in spinal cord, an effect correlated with up-regulation of A1Rs in the brain (Chen et al. 2010). Finally, the therapeutic effect of large doses of corticosteroids in EAE is associated with up-regulation of A1Rs (Tsutsui et al. 2008). Thus, activation of A1Rs appears to be an important negative modulator of neuroinflammation in MS and EAE.
However, the modulation of EAE pathology by adenosine may be complex, with distinct effects on different cellular elements, as indicated by a recent study using mice with genetic deficiency in CD73, a molecule responsible for generating extracellular adenosine in various cells including T cells (Mills et al. 2008). In an adoptive transfer experiment using CD73 KO mice, T cells from CD73 KO or WT mice with EAE were transferred to WT T-cell-deficient mice, which were monitored for signs of EAE. Disease pathology was much more severe in recipients of CD73 KO T cells, consistent with the anti-inflammatory effect of adenosine. However, when CD73 KO mice were immunized with MOG directly, the CD73 KO mice were largely resistant to MOG-induced brain/spinal cord injury (Mills et al. 2008). In direct contrast to the result of the transfer experiment, this suggests that extracellular adenosine facilitates the autoimmune response and exacerbates brain injury in the EAE model. This apparently paradoxical effect of adenosine (and CD73) may reflect complex actions of adenosine at more than one adenosine receptor subtype or at the same subtype on different cell populations.
In the study described here, we evaluated behavioral, neurohistological, and immunological characteristics in A2AR-KO mice with MOG-induced EAE. A2AR KO mice developed more severe EAE pathology than their WT littermates, exhibiting a severely demyelinated phenotype with more inflammatory cell infiltration in spinal cord and cerebral cortex, increased in CD4+ and CD8+ T cell populations, and increased expression of pro-inflammatory cytokines and reduced expression of anti-inflammatory cytokines in CNS, blood, and spleen.
Induction of EAE
Homozygous A2AR KO mice (A2AR−/−) and WT littermates (A2AR+/+) mice were generated and characterized as described previously (Chen et al. 1999; Yu et al. 2008), and the parent mice were provided by Dr. Jiang-Fan Chen at Department of Neurology, Boston University School of Medicine. The A2AR KO mice were in congenic C57BL/6 background after back-crossing for more than 12 generations (Yu et al. 2005). All animal experimental procedures of EAE were approved by the Institutional Animal Care and Use Committee at Wenzhou Medical College. They adhered to the NIH Guide for the Care and Use of Laboratory Animals. To induce EAE, 10- to 12-week-old female mice were injected subcutaneously with 200 μg MOG35-55 peptide (AC Scientific, China) emulsified in complete Freund’s adjuvant (CFA, containing 8 mg/ml Mycobacterium tuberculosis H37Ra). Reconstituted lyophilized pertussis toxin (200 ng List Biological Laboratories, Campbell, CA, USA) was injected intraperitoneally immediately after the MOG injection and again 48 h later. Animals were weighed, examined, and graded daily for EAE incidence and severity using a 0–15 rating scale (Weaver et al. 2005) as follows: Tail: 0 reflects no signs, 1 represents a half paralyzed tail, 2 represents a fully paralyzed tail; Limbs: 0 reflects no signs, 1 represents a weak or altered gait, 2 represents paresis, 3 represents a fully paralyzed limb. A fully paralyzed quadriplegic animal would attain a score of 12. Mortality equals a score of 15. Wild-type mice treated with CFA and pertussis toxin but without MOG35-55 peptide were used as control.
Animals were killed after disease signs reached their maximum level and animals were just beginning to show signs of remission (between 18 and 24 days). Asymptomatic EAE mice and control mice were killed on the 25th day post-immunization (dpi). After blood was drawn by retroorbital bleeding, mice were killed and brains and spinal cords were dissected out and processed for routine paraffin embedding and hematoxylin-eosin staining. Semi-quantitative histological evaluation based on the severity of inflammation was performed using the following scale (Okuda et al. 1999): 0, no cellular infiltrates of inflammation; 1, cellular infiltrates only around blood vessel and meninges; 2, mild cellular infiltrates in parenchyma (1–10/section); 3, moderate cellular infiltrates in parenchyma (11–100/section); 4, serious cellular infiltrates in parenchyma (> 100/section).
Myelination in the brain was examined using Luxol fast blue method as described previously (Kim et al. 2006). Briefly, after a defatting and anhydration, brain sections were incubated with Luxol fast blue solution at 56°C oven overnight. After a rinse with 95% ethanol and distilled water to remove excess stain, the slides were differentiated in lithium carbonate solution for 30 s and then in 70% ethylalcohol for 30 s. The slides were then rinsed in distilled water. Differentiation was verified under microscope to ensure that myelin was sharply stained. Sections were then mounted for examination under a light microscope.
To detect axonal injury, we used rabbit anti-mouse β-Amyloid Precursor Protein (β-APP) antibody (ABCAM, UK; 1 : 600) for immunostaining of axonal damage as described previously (Wang et al. 2011). Coronal sections were visualized with ABC reagent and diaminobenzidine. As a negative control, sections were stained without the presence of primary or secondary antibodies. Five randomly chosen fields from each section were counted and averaged.
To detect astrocyte and microglial activation in spinal cord, we used specific antibodies against glial fibrillary acidic protein (GFAP), a marker for astrocytes (Biological Technology Co., LTD, Zhongshan, Beijing, China; 1 : 100) and ionized calcium-binding adaptor molecule 1 (Iba-1), a marker for microglia (ABCAM, UK; 1 : 200). Cerebrum and spinal cord sections (5 μm) were incubated with the antibody overnight, followed by biotin conjugated goat anti-rat IgG-B (1 : 200, Santa Cruz Biotechnology, Santa Cruz, CA, USA) and avidin-biotin complex (Vectastain ABC Elite kit, Vector Laboratories, Burlingame, CA, USA). Peroxidase activity was localized using DAB as a chromogenic substrate (Vector Laboratories). An examiner who was unaware of the experimental condition counted Iba-1- or GFAP-immunopositive cells. Five randomly chosen fields in white-matter tracts, cortex, and cerebellum from each section were counted and averaged.
Cell preparation, culture, and flow cytometry
The spleen was isolated on a sterile wire sieve over a Petri dish filled with phosphate-buffered saline. Single cell suspensions were prepared by splitting blood cells using red blood cell lysing buffer (Beyotime Institute of Biotechnology, Shanghai, China). Spleen cells were plated in 24-well culture plates at 3 × 106/well in RPMI 1640 (Gibco, Rockville, MD, USA) supplemented with 10% fetal bovine serum (obtained from Hangzhou Sijiqing Biological Engineering Matericals Co., Ltd, Hangzhou, China), 1% mycillin (Gibco), with or without 20 μg/ml MOG35-55 for 3–4 days. After incubation, they were collected and subjected to flow cytometric assay. Single cell suspensions were resuspended in fluorescence-activated cell sorting (FACS) buffer, and cells (about 2 × 105 cells) were incubated with rat anti-mouse monoclonal antibodies for cell surface staining (anti-CD3R-PE, anti-CD8aR-FITC, anti-CD4-APC, all obtained from BD Biosciences San Jose CA USA) for 30 min in the dark, and then washed with phosphate-buffered saline. As a negative control, cells were stained with mouse IgG1 antibodies according to the manufacturer’s instructions. Stained cells were separated on a FACSC Calibur (Becton-Dickinson, Franklin Lakes, NJ, USA). Data were analyzed with Cell QuestV312 software (Becton Dickson).
RNA isolation and RT-PCR detection
Blood mononuclear cells were isolated by density gradient centrifugation (Lymphocyte Separation Medium was obtained from Sigma, St Louis, MO, USA). RNA was isolated using Trizol (Invitrogen, Carlsbad, CA, USA), according to the manufacturer’s protocol, and dissolved in diethylpyrocarbonate-treated water. RNA was quantitated by UV spectrophotometer (Beckman, LA, USA) to ensure that the ratio of the absorbances at 260 nm and 280 nm wavelength (A260/A280) was between 1.8 and 2.0; 50–100 mg tissue was used to isolate brain mRNA. For spleen cells, RNA was isolated from 3 × 106 cells that had been incubated with or without 25 μg/ml MOG35-55 for 3–4 days; 1 μg of RNA was used for the synthesis of complementary DNA (TaKaRa Bio Inc, BcaBESTTMRNA PCR Kit, Dalian China), and PCRs were performed as described in manufacturer’s guidelines (Bio-Rad, CA, USA). Mouse primer sequences were used as follows (***designed and synthesized by Shanghai Genecore Biotechnologies Co., Ltd): β-actin1, forward 5′-ACCGTGAAAAGATGACCCAGAT-3′ and reverse 5′-GGATTCCATACCCAAGAAGGAA-3′; β-actin2, forward 5′-CCCACTCCTAAGAGGAGGATG-3′ and reverse -5′-AGGGAGACCAAAGCCTTCAT-3′; IFN-γ, forward 5′-TCCTGCAGAGCCAGATTATCTCT-3′ and reverse 5′-ACCTCAAACTTGGCAATAC TCATG-3′; TGF-β, forward 5′- TGGACCGCAACAACGCCATCTATG-3′ and reverse 5′- TGGA GCTGAAGCAATAGTTGGTAT-3′; IL-10, forward 5′-TTACCTGGTAGAAGTGATGCCC -3′ and reverse 5′-ACACCTTGGTCTTGGAGCTTATG-3′; IL-17, forward 5′-TCTGTGTCTCTGATGCT GTTGC-3′ and reverse 5′- TTCATTGCGGTGGAGAGTC-3′; IL-6,forward 5′- TTCACAAGTCCG GACAGGAG-3′ and reverse 5′- TGGTCTTGGTCCTTAGCCAC-3′; A1R, forward 5′- CATCCTGGCT CTGCTTGCTATT-3′ and reverse 5′- TTGGCTATCCAGGCTTGTTCC-3′; PCR products were separated by electrophoresis through a 2.0% agarose gel and visualized by ethidium bromide staining. The relative abundance of the target genes was normalized to β-actin.
ELISA analyses of cytokine protein levels
To determine the effect of A2AR KO on the basal level of cytokines in the brain, cortex of normal WT and A2AR KO mice that did not receive MOG injection (n = 5 per group) were dissected out and homogenized in cell-lysis buffer supplemented with phenyl-methyl-sulphonyl fluoride. After centrifugation, the supernatants were used for determination of IFN-γ, IL-17, IL-6, TGF-β1, and IL-10 levels by ELISA method according to the manufacture’s procedure (R&D Systems, Minneapolis, MN, USA). Optical densities were measured using a Model 680 micro-plate reader (Bio-RAD Laboratories, Hercules, CA, USA) at 450 nm. Total protein was determined using the bicinchoninic acid assay (Bio-RAD Laboratories) (Li et al. 2010).
A1 receptor ligand binding assay
Total membranes were prepared from striatum, hippocampus, and cortex, and single-point saturation binding assays were performed in duplicate to quantify total A1R levels as described before (Wei et al. 2011). Each 300 μl binding assay, consisting of assay buffer (50 μl) and membranes (200 μl, 100–200 μg), was incubated with 3H-DPCPX (s.a. 111.6 Ci/mmol, GE Healthcare, Amersham Pharmacia Biotech, Piscataway, NJ, USA) at final concentration of 2 nM, at 26°C for 2 h. Non-specific binding was determined using 2 μM xanthine amine congener.
Statistical analyses were performed with the SPSS 13.0 statistical program (Statistical Package of Social Science Software, Chicago, IL, USA). RT-PCR data were displayed as means ± SEM unless otherwise stated, and significance was determined using a 2-tailed Student’s t-test. The EAE incidence analysis was performed using the Fisher’s exact test, AMD clinical neurologic scores and histopathologic scores were assessed by LSD post hoc comparison and Mann–Whitney U-test. Comparison between multiple groups was analyzed by One-way anova, whereas comparison between two groups was by Student t-test. Differences were considered significant at p < 0.05.
A2AR inactivation increased morbidity and exacerbated neurobehavioral deficits following EAE induction in mice
Beginning on the 11th day post-immunization (dpi) with MOG-CFA, some mice displayed decreased appetite, weight loss, rougher hair, and less exploratory activity. Body weight loss was more evident in A2AR KO mice than in WT mice (Fig. 1a). By the 13th dpi, several mice in the MOG-induced groups showed the signs of EAE onset, including weak crawl or altered gait, flaccid tail, and hind leg paresis. These signs of neurological disability gradually worsened and by 18–24 dpi, mice were quadriplegic and frequently moribund. There were no visible behavioral symptoms in the CFA control group, which exhibited a gradual increase in body weight. Furthermore, 13 of 16 A2AR KO mice and 10 of 16 WT mice displayed neurological signs of EAE (Fig. 1b), indicating a tendency toward increased EAE incidence in the A2AR KO group. Compared with WT mice, A2AR KO mice tended to display shorter latency to first appearance of behavioral signs and increased maximal neurologic deficit score (Fig. 1c). These data indicate that activation of the A2AR decreases incidence of disease and alleviates the severity of MOG-induced EAE in mice.
A2AR inactivation increased inflammatory cell infiltration, demyelination, and axonal injury in the spinal cord following EAE induction in mice
Hematoxylin-eosin staining revealed characteristic EAE-related neuropathological features, including parenchymal inflammatory cell infiltration and perivascular cuffing with mononuclear cells in CNS tissues (Fig. 2a–c). Thirteen of 16 A2AR KO mice displayed the characteristic EAE histopathologic alteration, consistent with the morbidity detected by clinical behavioral scoring (Fig. 1c). Ten of 16 WT mice showed the EAE histopathologic alteration, largely in accordance with the morbidity determined by neurological disabilities, with the exception of one mouse that displayed obvious inflammatory cell infiltration and perivascular cuffing, but no clinical signs (score 0). We considered this a subclinical case of EAE. There was no detectable inflammatory cell infiltration in the CFA control mice. Although the difference between A2AR KO and WT mice in incidence of EAE histopathology did not reach significance, semi-quantitative histopathologic scores were significantly higher in A2AR KO mice than in WT mice (p <0.05, Mann–Whitney U-test, Fig. 2d).
The severity of demyelination in brainstem was evaluated with Luxol fast blue staining (Fig. 3a–c). Representative brain sections showed extensive demyelination in the EAE-WT and EAE-KO group: Luxol fast blue staining was most extensive in the injured brainstem. The intensity of Luxol fast blue staining was significantly lower in EAE-KO mice than in EAE-WT mice due to loss of myelin content in brainstem sections (p <0.01, Fig. 3d).
To assess axonal injury in brain, we determined expression of β-APP in brainstem sections of EAE-WT and EAE-KO mice by immunohistochemistry. β-APP deposits increase in axons after focal blockage of axonal transport. Thus, the increased expression of β-APP has been widely used as a marker for axonal injury (Wang et al. 2011). β-APP staining was detected in neuron-like cells in the brainstem section from the EAE-KO group (Fig. 4a–c). The number of β-APP positive cells in the brainstem section was significantly higher in EAE-KO mice than in EAE-WT mice (p <0.01, Fig. 4d).
A2AR inactivation enhanced expression of proinflammatory cytokines and reduced anti-inflammatory cytokine mRNA expression in CNS and periphery following EAE immunization
Since an altered ratio of pro- and anti-inflammatory cytokines contributes to the development of EAE, we analyzed the expression of critical cytokines related to the pathogenesis or recovery of EAE. At the peak of EAE clinical signs, both WT and A2AR KO mice showed higher expression of IFN-γ, IL-17, IL-6, TGF-β1, and IL-10 mRNAs in cerebral cortex and cervical cord compared with CFA control mice. Furthermore, expression of IFN-γ, IL-17, and IL-6 mRNA was significantly higher in the brain and cervical cord of A2AR KO mice compared with WT mice, whereas expression of TGF-β1 and IL-10 mRNA was significantly lower (p <0.01, Fig. 5a and b). Similarly, IFN-γ was significantly up-regulated in peripheral blood mononuclear cells and spleen of A2AR KO mice, especially in splenocytes following MOG35-55 immunifaction (p <0.05, data not shown). These findings suggest that absence of A2AR expression increased the release of pro-inflammatory cytokines and depressed anti-inflammatory cytokines. To examine the effect of genetic inactivation of A2ARs on the basal levels of cytokine protein, we used ELISA to determine the protein levels of IFN-γ, IL-17, IL-6, TGF-β1, and IL-10 in cerebral cortex of untreated WT and A2AR KO mice. We found no differences between WT and KO mice in the levels of these cytokines in the brain (p >0.05, Table 1). Moreover, receptor binding assay using 3H-DPCPX as a ligand showed no significant change in A1R binding in cortex and hippocampus, but did reveal a small (∼10%) reduction in binding in the striatum (see Table 2, n = 4 per group, p > 0.05), a result consistent with our earlier report that A1R binding was reduced by 8% in the striatum of A2AR global KO (Chen et al. 2001). Thus, A1R binding in cortex was not affected in A2AR KO despite the apparently increase in A1R mRNA level by PCR analysis (data not shown), suggesting that cortical EAE pathology in A2AR KO mice cannot be accounted for by compensatory changes in A1R protein expression.
Table 1. The basal level of five pro- and anti-inflammatory cytokines in the brain of naive WT and A2AR Ko mice
WT (n = 5)
5.35 ± 0.75
6.46 ± 0.44
7.24 ± 0.77
36.26 ± 9.92
6.88 ± 0.92
KO (n = 5)
5.09 ± 1.76
6.20 ± 0.36
5.01 ± 0.69
45.40 ± 4.50
5.22 ± 0.56
Table 2. The A1 receptor binding densities in cortex, hippocampus, and striatum of naive WT and A2AR KO
*p < 0.05, comparing WT with KO.
WT (n = 4)
1471 ± 11.89
1060 ± 6.385
1213 ± 34.55
WT (n = 4)
1450 ± 69.77
1112 ± 45.04
1069 ± 23.80*
A2AR inactivation enhanced activation of microglia cells and astrocytes in cerebral cortex and spinal cord
Ionized calcium-binding adaptor molecule 1 (Iba-1) is a Ca2 + -binding peptide expressed selectively in microglia/macrophages and induced in activated monocytes and microglial cells. In CFA-WT mice, immunohistochemistry revealed few Iba-1 immunoreactive (ir) cells, and these cells exhibited the morphological features of resting microglial cells. However, in EAE mice of both genotypes, there was a marked increase in Iba-1-ir cells with morphological features of activated microglial cells (i.e. retracted, coarse branches, and enlarged soma) in cerebral cortex, white matter, and cerebellum (p <0.01). Importantly, after induction of EAE, both the number and intensity of Iba-1-ir cells in the CNS were significantly higher in A2AR KO mice than in WT mice (p <0.01, Fig. 6a–d).
GFAP is expressed selectively in astrocyte endochylema and serves as a marker of mature astrocytes. Under the light microscope, GFAP immunostaining revealed just a small number of GFAP-ir cells in cerebral cortex, white matter and cerebellum of CFA-treated control mice. GFAP-ir was markedly increased in EAE mice compared with control (p <0.01), regardless of the A2AR genotype. However, the increase in both number and intensity of GFAP-ir cells in the CNS was significantly greater in A2AR KO mice than in WT mice (p <0.01, Fig. 7a–d).
A2AR inactivation increased CD4+ T-cell populations in the spleen
To determine the effect of A2AR inactivation on peripheral CD4+ and CD8+ T-cell populations in the spleen during development of EAE, we isolated splenocytes from EAE mice at 18–24 dpi, the peak of EAE clinical signs. For asymptomatic EAE mice and CFA control mice, splenocytes were isolated on the 25th dpi. Cells were incubated with or without 20 μg/ml MOG35-55 for 3–4 days, at which point we determined the number of CD4+ and CD8+ T cells using flow cytometry. There was no statistical difference between CFA-treated WT (CFA-WT) and CFA-treated A2AR KO (CFA-KO) mice in the mean percentage (± s.e.m.) of CD4+ (CFA-WT 18.03 ± 1.54 vs. CFA-KO 18.37 ± 1.27, p > 0.05, n = 5, t-test) or CD8+ cells (CFA-WT 10.13 ± 1.10 vs. CFA-KO 11.79 ± 0.93, p > 0.05, n = 5, t-test). However, following induction of EAE, A2AR KO mice had more CD4+ T cells than WT mice (Fig. 8a and b). In vitro immunostimulation with MOG35-55 induced a further significant increase in the number of CD4+ T cells in both EAE-KO and EAE-WT mice. However, two-way ANOVA revealed no genotype treatment interaction after MOG stimulation. In contrast, there was no statistical difference in the number of CD8+ T cells between EAE mice (WT and A2AR KO) compared with CFA-mice (A2AR KO and WT) (data not shown). Similarly, EAE-KO mice had slightly higher numbers of CD8+ T cells than EAE-WT, but this did not reach statistical significance (data not shown). These data indicate that A2AR inactivation selectively increased the number of CD4+ cells induced following EAE and suggest that, in WT mice, activation of the A2AR counteracted development of EAE in part by suppressing CD4+ cells.
To examine whether genetic inactivation of A2ARs affect CD4+ in naive mice, we used flow cytometry to determine the number of CD4+ and CD8+ T cells in naive WT and A2AR-KO mice and found no statistical differences (p >0.05, data not shown).
Genetic deletion of A2AR exacerbates EAE by activating microglia in the CNS and CD4 cells in peripheral tissues
Our studies with A2AR KO mice provide direct evidence that genetic inactivation of the A2AR exacerbates EAE pathology induced by MOG immunization in mice. This exacerbation of EAE is evident in several measures of EAE pathology: greater body weight loss, increased neurological deficit score, increased incidence of EAE, increased maximal behavioral score, and exacerbated inflammatory cell infiltration, demyelination and axonal injury in spinal cord, and other brain regions. These results suggest that A2AR activation by extracellular adenosine inhibits MOG-induced neuroinflammation and demyelination in the brain and neurobehavioral deficits.
The mechanisms underlying the exacerbation of EAE pathology in A2AR KO mice are not clear. Loss of A2AR-mediated modulation of microglial cells is an important mechanism by which A2AR inactivation exacerbates EAE pathology. The activation of microglia is closely associated with the development of histopathologic lesions and progression of EAE in many different ways: (a) when microglia cells are activated, some become microglia-derived brain macrophages, which phagocytize cells through direct cell-to-cell contact; (b) microglia release many kinds of cytotoxic molecules, including H2O2, NO, IL-1, IL-6, TNF-α, IFN-γ, and TGF-β. Release of these substances can activate microglial cells, leading to feed-forward regulation of inflammation; (c) activated microglia express MHC-II antigen, and function as antigen presenting cells (APC). We found reduced microglial activation (evident by reduced Iba-1-ir) with tiny cell morphology in the CFA control group, while the classical-activated microglia present after EAE induction had a number of retracted, coarse branches, and enlarged soma, and some had an ameba-like activity. Interestingly, both the number and intensity of Iba-1-ir cells in the CNS increased significantly in the EAE-KO group compared with the EAE-WT group. This suggests that genetic deletion of A2AR exacerbates EAE by activating microglia in the CNS, which may be an important inflammatory step in the destruction of myelin and axons.
Astrocytes are essential components of the blood brain barrier and the normal white matter in the brain. They produce anti-inflammatory cytokines (such as IL-10 and TGF-β) and neurotrophic factors [such as GDNF and Brain-derived neurotrophic factor (BDNF)] to support and nourish the neurons and to promote regeneration and repair of injured myelin by acting on oligodendrocytes. Consistent with this notion, GFAP null mice had severe EAE compared with WT littermates (Liedtke et al. 1998). Similarly, our recent study showed that idazoxan attenuated EAE pathology by increasing activation of astrocytes (Wang et al. 2009). Thus, enhanced expression of GFAP in A2AR KO likely reflects a compensatory response to the exacerbated tissue damage that occurs in the absence of the A2AR.
The pathological process of EAE is believed to be mediated by CD4+ T helper (Th)1 cells with some contribution from CD8+ T cells, as indicated by the presence of myelin antigen-specific CD8+ T cells in EAE mice and patients with MS (Crawford et al. 2004; Fletcher et al. 2010; Ji and Goverman 2007, Keegan and Noseworthy 2002; McFarland and Martin 2007). In addition, adoptive transfer of MBP- (Huseby et al. 2001) or MOG- (Sun et al. 2001) specific CD8+ T cells induces EAE in recipient mice. In keeping with this notion, we showed that the percentage of CD4+ T cells in spleen (with or without in vitro MOG stimulation) increased in EAE-KO mice compared with EAE-WT mice. Thus, the stronger cellular immunologic response to MOG and exacerbated brain damage in EAE-KO mice may be in part associated with increased CD4+ T cells from peripheral tissues.
Our finding of enhanced activation of microglial cells in the brain and increased CD4+ T cells in the spleen in A2AR EAE-KO mice compared with EAE-WT littermates suggests that A2AR-mediated modulation of these inflammatory components is critical to the phenotypes observed in A2AR KO mice. In addition to CD4+ T cells and microglial cells, A2AR activity may also act on neutrophils to influence infiltration of inflammatory cells from peripheral blood through the blood brain barrier (BBB) into the CNS. Caffeine, a non-selective adenosine receptor antagonist, was recently shown to improve BBB integrity in animal models of Parkinson’s disease and Alzheimer’s Disease (Chen et al. 2008a,b). This suggests that genetic inactivation of A2ARs might increase BBB integrity. Thus, A2AR KO effects on the BBB unlikely to explain the exacerbation of EAE pathology.
Finally, the exacerbation of brain damage in MOG-treated A2AR KO mice may be because of the reduction of BDNF level and/or activity. This notion is strongly supported by two lines of evidence: (i) Linker and colleagues used a similar MOG-induced animal model of EAE and demonstrated a protective role of endogenous BDNF in the chronically inflamed nervous system (Linker et al. 2010); (ii) genetic deletion of the A2AR abolished the excitatory effects of BDNF on synaptic transmission. This reduced function of BDNF in A2AR KO mice was correlated with a reduction in BDNF protein (Tebano et al. 2008).
Genetic and pharmacological inactivation of the A2AR may target different cellular elements to produce opposite effects on EAE pathology
Intriguingly, the A2AR antagonist SCH58261 was recently shown to exert neuroprotective effects against MOG-induced pathology, and this neuroprotective effect was attributed to reduced neutrophil attachment to endothelium and the resulting decrease in infiltration of inflammatory cells into the CNS (Mills et al. 2008). Thus, A2AR activity can impact EAE development through multiple molecular targets on different cellular elements, including CD4+ T cells, neutrophils, microglial cells, and probably other cell types, including endothelium and neurons. This array of potential targets may explain why genetic and pharmacological inactivation of the A2AR appears to have opposite effects on EAE. Although we have direct evidence for exacerbation of EAE pathology in A2AR KO mice, previous studies have shown that pharmacological blockade of the A2AR offers neuroprotection against EAE (Mills et al. 2008). Global genetic deletion inactivates A2AR activity in all cellular components, while SCH58261 may preferentially affect neutrophils to reduce neutrophil infiltration into CNS and exert neuroprotective effects. Additional studies using selective inactivation of the A2AR in distinct cellular elements are needed to clarify this issue.
The exacerbated EAE-induced brain damage we observed in A2AR KO mice is not the phenotype predicted based on the recent finding that genetic deficiency in extracellular nucleotidase CD73, a molecule critical for generating extracellular adenosine, protects mice against MOG-induced brain damage (Mills et al. 2008). We speculate that genetic deletion of CD73 not only reduced the extracellular adenosine level but also increased levels of AMP, ADP, and ATP because AMP is not converted to adenosine in the absence of CD73. Recent reports support this possibility by showing that pharmacological blockade of CD73 with deoxycoformycin induced a long-lasting accumulation of AMP/ADP/ATP (Bekar et al. 2008; Goldman et al. 2010).
Bidirectional regulation of brain injury by A2AR in EAE and other kinds of brain injury and peripheral tissue damage may be dictated by local glutamate level
In contrast to the consistent A2AR-mediated inhibition of inflammation observed in peripheral tissues, the effects of A2ARs on neuroinflammation and neuronal damage in the brain are complex and apparently paradoxical. Activation of A2ARs inhibits the production of proinflammatory cytokines such as TNF-α, IL-6, and IL-12 (Hasko et al. 2000; Mayne et al. 2001; Ohta and Sitkovsky 2001; Day et al. 2003, 2004; Sitkovsky et al. 2004). Largely based on the effects of A2AR action in peripheral tissues, extracellular adenosine acting at A2ARs has been proposed as an endogenous “breaker” that inhibits inflammation and limits extensive inflammatory tissue damage (Ohta and Sitkovsky 2001). However, both activation and inactivation of the A2AR have been shown to reduce neuroinflammation and brain damage in various animal models of neurological disorders (Chen et al. 2007). In many pathological conditions including stroke (Von Lubitz et al. 1995; Melani et al. 2003), excitotoxicity (Cunha 2005; Cognato et al. 2010), Parkinson’s disease model (Richardson et al. 1999; Schwarzschild et al. 2006), and traumatic brain injury (Dai et al. 2010), A2AR antagonism has been shown to attenuate brain injury. On the other hand, A2AR activation can afford neuroprotection against cerebral hemorrhage (Mayne et al. 2001), spinal cord injuries (Cassada et al. 2002), and EAE-induced spinal cord injury, as shown here. For example, we showed recently that genetic inactivation or pharmacological blockade of A2ARs can exert opposite effects on brain injury in the same cortical impact model of traumatic brain injury dependent on the local extracellular glutamate level in the brain (Dai et al. 2010). It will be important to investigate whether glutamate levels affect A2AR action on EAE-induced pathology. Furthermore, A2AR antagonists may exert different effects dependant on their relative access to different populations of A2AR-expressing cells in peripheral tissues (including bone marrow derived cells in the blood) and in the brain. Given the multiple cellular processes involved in EAE models, we speculate that A2AR antagonists acting at peripheral immune cells (such as bone marrow derived cells) and at neurons have different effects on control MOG-induced brain injury. Further studies are needed to clarify the effect of A2AR in distinct cellular elements in EAE and in different brain injury models.
Activated A1 and A2A receptors coordinately inhibit neuroinflammation and EAE pathology
Our findings suggest that extracellular adenosine acting at A1 and A2A receptors can exert an important control mechanism that is protective against EAE-induced neuroinflammation and brain damage. This notion is supported by the following observations: (1) adenosine and A1R levels are reduced in blood cells and CSF of MS patients (Mayne et al. 1999; Johnston et al. 2001); (2) A1R deletion exacerbates EAE pathology in mice while A1R activation attenuates it (Tsutsui et al. 2004); (3) following up-regulation by caffeine, A1R activation suppresses pro-inflammatory responses and augments anti-inflammatory responses, diminishing oligodendrocyte cytotoxicity (Chen et al. 2010); (4) as described here, inactivation of A2AR also exacerbates EAE in mice. A1Rs and A2ARs are coupled to Gi and Gs, respectively, to inhibit and stimulate cAMP levels in inflammatory cells. Coordinated modulation of neuroinflammation by A1Rs and A2ARs in EAE may indicate the involvement of additional mechanisms (such as PKA and PKC pathways) to produce a synergistic effect on neuroinflammation. Finally, 3H-A1R ligand binding assay showed that A1R protein levels in cortex and hippocampus were not affected despite apparent increase in A1R mRNA by PCR analysis (data not shown) and A1R binding in the striatum was slightly (∼10%) but significantly reduced in A2AR KO mice, consistent with our early finding (Chen et al. 2001). As A1R binding in cortex was not affected in A2AR KO, the EAE pathology in cortex of A2AR KO mice cannot be accounted for by compensatory changes in A1R protein despite the apparent change in A1R mRNA in the cortex of A2AR KO mice. Thus, A2AR exacerbates brain damage by A1R-independent mechanisms in the cortex.
In summary, our data suggest that the exacerbation of EAE seen in A2AR KO mice is mediated by loss of A2AR modulation of microglial activation in the brain in parallel with an increase in CD4+ T cells in the spleen, resulting in increased expression of Th1 cytokines and decreased expression of Th2 cytokines. Along with previous data from our lab and others, these results suggest that extracellular adenosine acting at A2ARs and A1Rs activates important anti-inflammatory and neuroprotective mechanisms to limit EAE-induced neuroinflammation and brain damage. These results identify the A1R–A2AR pathway as a potential drug target for modification of disease course of multiple sclerosis.
This work was supported by grants from Wenzhou Medical College Key Research Project (No.30328015), the Joint Found of China Ministry of Public Health-Zhejiang Bureau of Health (WKJ 2005-2-041), Wenzhou Science &Technology bureau (H20100014), the Building Funding of Zhejiang Key Subject (Pharmacology and Biochemical Pharmaceutics), and by US public health grant (NIH) NS41083. The authors declare that they have no conflicts of interest.