We investigated the involvement and roles of the ionotropic purinergic receptor P2X7R in microglia in mediating lipopolysaccharide (LPS)-induced inflammatory responses and neuronal damage in rat striatum. A detailed in vivo study showed that LPS injection into striatum markedly increased the expression of P2X7R in microglia compared with control (saline)-injected animals. Additionally, LPS injection upregulated a broad spectrum of proinflammatory mediators, including inducible nitric oxide synthase (nitric oxide production marker), 3-nitrotyrosine (peroxynitrite-mediated nitration marker), 4-hydroxynonenal (lipid peroxidation marker), and 8-hydroxy-2′-deoxyguanosine (oxidative DNA damage marker), and reduced neuronal viability. The P2X7R antagonist oxidized ATP (oxATP) was effective in attenuating expressions of all inflammatory mediators and in addition inhibited LPS-induced activation of the cellular signaling factors p38 mitogen-activated protein kinase and transcriptional factor nuclear factor κB. Most importantly, in vivo, oxATP blockade of P2X7R also reduced numbers of caspase-3-positive neurons and increased neuronal survival in LPS-injected brain. In vitro, LPS stimulation of cultured human microglia enhanced cellular expressions of a host of proinflammatory factors, including cyclooxygenase-2, interleukin-1β (IL-1β), IL-6, IL-12, and tumor necrosis factor-α; all factors were inhibited by oxATP. A novel finding was that LPS potentiated intracellular [Ca2+]i mobilization induced by the P2X7R ligand 2′,3′-O-(4-benzoyl-benzoyl) ATP, which could serve as a mechanistic link for P2X7R amplification of inflammatory responses. Our results suggest critical roles for P2X7R in mediating inflammation and inhibition of this subtype purinergic receptor as a novel therapeutic approach to reduce microglial activation and confer neuroprotection in inflamed and diseased brain.
Inflammation is a critical component in the progressive neurodegeneration manifest in neurological disorders. The activation of glial cells, microglia, and astrocytes is a characteristic finding in brain inflammation. Microglia, as the immunocompetent resident cells of the brain, possess properties particularly suitable in mediating cellular inflammatory responses. The release of proinflammatory mediators from activated microglia has been suggested as a contributing factor to pathology in disease (McGeer and McGeer, 1998; Lue et al., 2001; Walker et al., 2001), with anti-inflammatory agents conferring some degree of neuroprotection (Suzuki et al., 2004; Klegeris and McGeer, 2005). An important aspect in the induction of an inflammatory response may be a signal from damaged neurons to activate microglia. In this regard, evidence is available to indicate that increased levels of extracellular ATP can mobilize and activate microglia under pathological conditions associated with spinal cord injury and ischemia (Le Feuvre et al., 2002; Franke et al., 2004; Wang et al., 2004).
Microglia (and astrocytes) express purinergic receptors for the two receptor subfamilies, ionotropic (P2XR) and metabotropic (P2YR), which mediate mobilization of intracellular Ca2+ [Ca2+]i (Moller, 2002; McLarnon, 2005). A particularly interesting member of the P2XR family is P2X7R, principally expressed by macrophage-like cells of the brain and linked with a high conductance pore. Activation of P2X7R generally requires elevated levels of extracellular ATP (in excess of 1 mm), suggesting an involvement in pathophysiological conditions. During activation, P2X7R mediates influx of Na+ and Ca2+ and concomitant efflux of K+, which in microglia helps shape and modulate a diversity of cellular inflammatory responses (Gudipaty et al., 2003; Witting et al., 2004).
Functionally, P2X7R plays a critical role in microglial release of the proinflammatory cytokine interleukin-1β (IL-1β) by reducing intracellular levels of K+ (Ferrari et al., 1999; Humphreys et al., 2000; Rothwell and Luheshi, 2000). Increased levels of IL-1β are implicated in the pathogenesis of neurodegenerative diseases such as Alzheimer's disease (AD), multiple sclerosis, and Parkinson's disease (Griffin and Mrak, 2002; Li et al., 2003; Ferrari et al., 2006). Recent work has reported that activation of P2X7R in microglia led to an increased production of superoxide anion (O2−) in a transgenic Tg2576 mouse model of AD (Parvathenani et al., 2003). Application of the P2X7R ligand 2′,3′-O-(4-benzoyl-benzoyl) ATP (BzATP) increases the production of proinflammatory cytokines from macrophages and microglial cells (Rampe et al., 2004), with the specific P2X7R antagonist oxidized ATP (oxATP) reported effective against neuronal damage (Wang et al., 2004).
In this study, we investigated the roles of P2X7R in linking inflammatory responses with neuronal damage in a lipopolysaccharide (LPS) animal model of inflammation and in vitro in LPS-stimulated human microglia. The overall findings suggest critical functions for microglial P2X7R-mediated oxidative stress and proinflammatory responses in causing neuronal loss in inflamed striatal brain. Importantly, inhibition of P2X7R in microglia has been found effective in reducing a broad spectrum of inflammatory responses and conferring neuroprotection in inflamed brain. Furthermore, our results suggest that pharmacological modulation of [Ca2+]i mobilization mediated by P2X7R in activated microglia may underlie the neuroprotective effects.
Materials and Methods
All animal experiments were performed in accordance with the guidelines set by the Committee on Animal Research at the University of British Columbia. Male Sprague Dawley rats (240–260 g) were anesthetized with sodium pentobarbital (50 mg/kg, i.p.) and then mounted in a stereotaxic apparatus (David Kopf Instruments, Tujunga, CA). Animals received unilateral stereotaxic injection of 5 μg of LPS (Sigma, St. Louis, MO) dissolved in PBS over 4 min at the following coordinates: anteroposterior (AP), +1.0 mm; mediolateral (ML), −2.6 mm; dorsoventral (DV), −5.0 mm (Paxinos and Watson, 1998). oxATP (Sigma) was dissolved in water and further diluted to PBS, and animals received intracerebroventricular injections (coordinates: AP, +0.48 mm; ML, −1.0 mm; DV, −3.8 mm) 30 min before LPS injection. Control animals received intracerebroventricular injections of PBS; another group received intracerebroventricular injections of oxATP alone. The animals were deeply anesthetized at 6 h, 12 h, 24 h, or 3 d after LPS injection and transcardially perfused with heparinized saline, followed by 4% paraformaldehyde for additional analysis. For systemic administration of LPS, animals intraperitoneally received two concentrations of LPS (0.5 or 5 mg/kg). Animals that received LPS were killed at 1 d after LPS injection.
The methods used in the isolation and identification of microglia have been described previously (Nagai et al., 2001). In brief, human embryonic brain tissues (12–18 weeks of gestation) were dissected into small blocks, incubated in PBS containing 0.25% trypsin and 40 μg/ml DNase for 30 min at 37 °C, and dissociated into single cells by repeated pipetting. Dissociated cells were plated in T75 flasks in a medium consisting of DMEM with high glucose containing 5% horse serum, 25 μg/ml gentamicin, and 2.5 μg/ml amphotericin B. Freely floating microglia were harvested from a medium of mixed cell cultures after 7–10 d of growth in culture flasks. Immunostaining with CD11b and Ricinus communis agglutinin, specific markers for microglia, ensured that the purity of cultures was in excess of 98%. The use of embryonic human tissues was approved by the Clinical Screening Committee for Human Subjects of the University of British Columbia.
Free-floating sections (40 μm serial coronal sections) were processed for immunostaining as described previously (Ryu et al., 2003). To ensure striatal sections were matched between groups, anatomical landmarks provided by the brain atlas (Paxinos and Watson, 1998) were used for tissue selection. The primary antibodies used for immunostaining were as follows: neuronal-specific nuclear protein (NeuN) for neurons (1:500; Chemicon, Temecula, CA), OX-42 for microglia (1:400; Serotec, Oxford, UK), GFAP for astrocytes (1:500; Sigma), P2X7R (1:200; Alomone Labs, Jerusalem, Israel), 3-nitrotyrosine (3-NT) (1:500; Upstate Biotechnology, Lake Placid, NY), 4-hydroxynonenal (4-HNE) (1:500; Jaica, Shizuoka, Japan), 8-hydroxy-2′-deoxyguanosine (8-OHdG) (1:500; Jaica), inducible nitric oxide synthase (iNOS) (1:100; Upstate Biotechnology), and cleaved caspase-3 (1:200; Cell Signaling Technology, Beverly, MA). Phospho-p38 mitogen-activated protein kinase (MAPK) (1:250; Cell Signaling Technology) and p65 subunit of nuclear factor κB (NFκB) (1:250; Santa Cruz Biotechnology, Santa Cruz, CA) were used for intracellular signaling markers. Sections were incubated at room temperature (20–22°C) for 2 h in biotinylated anti-mouse/rabbit IgG (1:200; Vector Laboratories, Burlingame, CA) followed by an ABC elite system (1:200; Vector Laboratories) for 1 h. The reaction products were developed using a DAB kit (Sigma). Alexa Fluor 594 anti-rabbit and Alexa Fluor 488 anti-mouse IgG (1:100) secondary antibodies (Invitrogen, Carlsbad, CA) were used for immunofluorescent staining. For each immunohistochemical staining, the primary antibody was omitted to assess for nonspecific binding of the secondary antibody. For 3-NT immunohistochemistry, positive (use of 24 mm peroxynitrite) and negative (use of 10 mm nitrotyrosine) controls were performed (Ayata et al., 1997).
For in vivo experiments, brains were removed and the striatum was dissected onto a cold metal tissue matrix (Harvard Apparatus, Montreal, Quebec, Canada). Total RNA was then extracted from striatal tissue using TRIzol reagent (Invitrogen) subjected to DNase I treatment, and complementary DNA synthesis was performed using Moloney murine leukemia virus reverse transcriptase (RT) (Invitrogen). Reverse transcriptase was omitted as a negative control. For in vitro experiments, human microglia were incubated in serum-free medium for 48 h. This procedure yielded a homogeneous population of cells with ramified, process-bearing morphology. PCR primers are all intron spanning, and sequences and expected product sizes are listed in Table 1. PCR conditions were as follows: initial denaturation at 95°C for 6 min followed by 35 cycles of denaturation at 95°C for 45 s, annealing at 55–60°C for 1 min, and extension at 72°C for 1 min. A final extension was performed at 72°C for 10 min. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-actin were used as reaction standards for human and rat samples, respectively. The amplified DNAs were identified using 1.5% agarose gels containing ethidium bromide and visualized under UV light. The band intensities of PCR products in control and with stimuli were measured using NIH Image J 1.24 software and expressed as relative mRNA levels (mRNA values normalized to GAPDH or β-actin).
The procedures used in Ca2+-sensitive fluorescence microscopy in this laboratory have been described previously (Khoo et al., 2001; McLarnon et al., 2001). In brief, the fluorescent Ca2+ indicator fura 2-AM was added with pluronic acid (both agents at 1 μm) to cultured human microglia bathed in physiological saline solution (PSS) at room temperature for 30 min. PSS contained the following (in mm): 126 NaCl, 5 KCl, 1 CaCl2, 1.2 MgCl2, 10 HEPES, and 10 d-glucose, pH 7.4. Ca2+-free PSS contains the same ionic composition as PSS with the exception that it included EGTA (at 1 mm) with no added CaCl2. Cells were washed for 10 min in dye-free PSS. Coverslips were placed on the stage of an inverted microscope (Axiovert with 40× quartz objective; Zeiss, Oberkochen, Germany), and alternating 340 and 380 nm UV light was used for excitation. Fluorescent signals were recorded by passing 510 nm of emission light (bandwidth of 40 nm) to a digital CCD camera (Retiga 1300i; QImaging, Burnaby, British Columbia, Canada). An imaging system (Empix, Mississauga, Ontario, Canada) was used to process and analyze fluorescent signals based on the wavelength ratios (340/380 nm), and ratio images were processed every 6 s. In Mn2+ quenching studies, the divalent cation Mn2+ (at 25 μm) was added to the Ca2+-free PSS. In the quenching experiments, an excitation wavelength of 360 nm was used because this wavelength corresponds closely to the isosbestic point of the fura 2-AM spectrum. All imaging experiments were done at room temperature.
The four immunostained sections (AP, +1.4, +1.2, +1.0, and +0.8) were digitized and analyzed using image analysis program NIH version 1.57 (Wayne Rasband, National Institutes of Health, Bethesda, MD) as described previously (Ryu et al., 2004). Quantification of neuronal damage in the striatum was performed by counting neurons double stained with NeuN and caspase-3 using a Zeiss Axioplan 2 fluorescent microscope (40× objective). The number of NeuN-positive (NeuN+) and caspase-3+ cells was counted and expressed as caspase-3-immunoreactive neurons per square millimeter. The number of NeuN+ cells were counted for treated and control groups to determine neuronal loss. OX-42-immunoreactive cells were counted for the measurement of microglial activation. For quantification of p38 MAPK, double staining of OX-42+ microglia with phospho-p38 MAPK was measured. The number of cells expressing phospho-p38 MAPK was then presented as cells per square millimeter. For NFκB, sections were immunostained with the p65 subunit of the transcription factor and counterstained with 4′,6-diamidino-2-phenylindole for cellular nuclei. Microglia expressing NFκB were counted for quantification of the transcription factor and presented as cells per square millimeter. For the measurement of oxidative stress, areas containing 4-HNE- or 8-OHdG-immunostained cells were sharply delineated, measured in four sections of the striatum, and expressed as 4-HNE- or 8-OHdG-immunoreactive areas. For the quantification of iNOS and 3-NT, immunoreactive cells were counted from four different affected areas of the striatum. Quantitative analyses were done in a blinded manner.
Statistics and analysis.
All results are expressed as means ± SEM. Statistical significance of differences for among group comparisons was evaluated using one-way ANOVA followed by Newman–Keuls post hoc multiple comparison test (Prism 3.0; GraphPad Software, San Diego, CA). Significance levels were set at p < 0.05 for all tests.
Effects of LPS on P2X7R expression and colocalization in rat brain
We initially examined the effects of LPS on P2X7R expression in rat brain using RT-PCR and immunohistochemical analysis. LPS (5 μg) was injected into rat striatum, and expression of P2X7R was determined at different postinjection time points (1, 4, 8, 12, 24, 48 h). Representative RT-PCR results (Fig. 1A) showed no evident P2X7R in control (12 h after injection of PBS). However, P2X7R increased in a time-dependent manner to a maximum level at 12 h after injection of LPS. Little or no expression of the purinergic receptor was found at the longest time of 48 h after LPS injection.
Significant upregulation of P2X7R expression was found at 24 h after LPS injection by immunohistochemical analysis (Fig. 1B). In control striatum, PBS (at 24 h) injection induced little or no P2X7R expression (Fig. 1B, left). However, LPS injection (Fig. 1B, right, green staining) markedly upregulated expression of P2X7R compared with PBS-injected striatum. Overall, immunoreactivities (IR) to P2X7R for PBS and LPS-injected groups were 1.0 ± 0.5 and 36.0 ± 3.8/mm2, respectively (n = 4 animals). These results show marked upregulation of P2X7R in the presence of LPS.
The next set of experiments examined cellular expression of P2X7R at 24 h after injection of LPS (or PBS as control) using double immunostaining for cell-specific markers (NeuN for neurons, GFAP for astrocytes, OX-42 for microglia; red cellular staining) with P2X7R (green staining). A representative staining pattern is presented in Figure 1C showing that LPS induced a predominant colocalization of P2X7R with microglia (right panel) and to a lesser extent with astrocytes (middle panel); no evident immunostaining for P2X7R was found in neurons (left panel).
Quantification of cellular P2X7R IR is shown in Figure 1D (n = 4 animals). With LPS injection, the maximum P2X7R IR was found in microglia (42.7% of OX-42+ microglia) with a lesser IR measured in astrocytes (22.5% of GFAP+ astrocytes); the corresponding values in PBS-injected animals were 10.3% (microglia) and 8.5% (astrocytes). Neurons showed minimal IR for P2X7R with respective values of 1% (LPS injection) and 0.7% (PBS injection). Thus, expression of P2X7R is primarily upregulated in microglia and to a lesser extent in astrocytes in LPS-injected rat striatum.
Effects of oxATP on LPS-mediated microglial activation
We next examined the number of OX-42+ microglia and also morphological changes in the presence of oxATP, an inhibitor of P2X7R, injected intracerebroventricularly (at 1 μm) 30 min before LPS injection. Representative immunostaining for OX-42 is presented in Figure 2A and shows that PBS injection (left panel) was mainly associated with a ramified morphology indicative of resting cells. LPS injection (at 5 μg for 3 d) induced an evident shift to an ameboid morphology with cells displaying roundish cell bodies and shortened processes (middle panel) suggestive of an activated state. Additionally, numbers of OX-42+ cells were increased in LPS-injected striatum. oxATP treatment (at 1 μm) was effective in diminishing the levels of OX-42+ microglia in LPS-injected striatum (right panel). oxATP alone did not induce morphological changes (data not shown).
OX-42-immunoreactive cells per square millimeter were counted for each treatment with overall results presented in Figure 2B. The numbers of microglia were almost tripled with LPS injection relative to PBS-injected striatum (140.7 ± 4.1 and 51.2 ± 2.0/mm2, respectively). In the presence of oxATP (at 1 μm) with LPS, the corresponding numbers of OX-42+ cells were 92.3 ± 3.3/mm2 (n = 5 animals), representing a significant decrease (by 34.4%) compared with LPS alone. These results suggest involvement of P2X7R in mediating LPS-induced microglial proliferative responses.
Effects of oxATP on LPS-induced activation of p38 MAPK and NFκB
We then investigated the expression of p38 MAPK dependence on P2X7R in LPS-injected striatum (at 5 μg for 6, 12, 24, and 72 h) in the absence and presence of oxATP (at 1 μm) using double immunostaining. Little or no activation of p38 MAPK was observed in PBS-injected striatum (Fig. 3A, for 24 h, top left). However, LPS injection (5 μg for 24 h) elicited a large degree of phosphorylation of p38 MAPK (green staining) in OX-42+ microglia (red staining) (top middle panel). The injection of oxATP effectively attenuated phosphorylation of p38 MAPK (right top panel).
The bar graph in Figure 3B (n = 4 animals) shows the time-dependent changes in p38 MAPK phosphorylation after LPS injection and subsequent inhibition in the presence of oxATP. Overall, at 24 h, 42.2 ± 14.2 cells/mm2 expressed phosphorylated p38 MAPK in PBS-injected striatum. The corresponding values in LPS-injected and LPS plus oxATP-injected striatum were 184.5 ± 17.4 and 112.4 ±16.3 cells/mm2, respectively. The number of p38 MAPK+ cells was significantly decreased in the presence of oxATP by 39.1% compared with LPS alone. The corresponding decreases in p38 MAPK with oxATP at 12 and 72 h were 32.7 and 57.2%, respectively; all effects were significant. These results suggest that activation of P2X7R is linked with downstream p38 MAPK activity in microglia.
The translocation of NFκB to the nucleus initiates transcription of a number of proinflammatory mediators such as tumor necrosis factor-α (TNF-α) (Combs et al., 2001). We next studied activation of NFκB in microglia with LPS stimulation (at 5 μg for 6, 12, 24, and 72 h) in the absence and presence of oxATP. Representative immunostaining showed little activation of NFκB at 24 h in PBS-injected striatum (Fig. 3A, bottom left panel). Intrastriatal injection of LPS markedly upregulated nuclear translocation of the p65 subunit, a component of the NFκB p65/p50 heterodimer (bottom middle panel, green staining) localized with OX-42+ microglia (red staining). The activation of NFκB was considerably reduced in the presence of oxATP (bottom right panel).
The bar graph in Figure 3C (n = 4 animals) shows LPS-induced time-dependent changes in NFκB and inhibition in the presence of oxATP. Overall, at 24 h, 73.4 ± 19.8 cells/mm2 expressed nuclear translocation of the p65 subunit in PBS-injected striatum. The corresponding values in LPS-injected and LPS plus oxATP-injected striatum were 403.2 ± 42.1 and 264.1 ± 35.2 cells/mm2, respectively. The number of p65+ cells were significantly decreased in the presence of oxATP (by 34.5%) compared with LPS alone. At the time points of 12 and 72 h, oxATP reduced NFκB by 33.8 and 46.1%, respectively, with both values reaching significance. Our results show that inhibition of P2X7R attenuates LPS-induced activation of NFκB in microglia.
Effects of oxATP on LPS-induced iNOS expression
The involvement of P2X7R in iNOS signaling was next investigated using RT-PCR and immunohistochemical analysis. Production of iNOS was observed in a small population of cells in PBS-injected brain (Fig. 4A, left) but was considerably elevated with LPS injection (5 μg at 24 h after injection) (middle panel). The inset of Figure 4A (middle) shows colocalization of iNOS (green) with OX-42+ microglia (red). In the presence of oxATP (at 1 μm), LPS-induced production of iNOS was attenuated (right panel).
The time-dependent changes in numbers of iNOS+ cells were measured at different time points (6, 12, 24, and 72 h) with PBS injection or with LPS injection in the absence and presence of oxATP (at 1 μm). As shown in Figure 4B (n = 4 animals), little or no iNOS production was measured at any time point in PBS-injected rat brain. With LPS injection, the number of iNOS+ cells reached a maximum at 24 h and returned to a basal level at 72 h after injection. The numbers of iNOS+ cells were significantly reduced at 12 h (43.3% reduction) and 24 h (56.3% reduction) in the presence of oxATP (at 1 μm) compared with LPS alone.
RT-PCR was used to determine iNOS mRNA expression in PBS, LPS, and LPS plus oxATP-injected rat striatum. A representative RT-PCR result is presented in Figure 4C (top), showing no expression of iNOS in PBS-injected rat striatum (lane 1, 12 h after injection). LPS (5 μg, 12 h after injection) induced a high level of iNOS expression (lane 2) that was attenuated with oxATP treatment (lane 3). Semiquantification of RT-PCR results (n = 3 experiments) are presented in Figure 4C (bottom). LPS injection induced a sixfold increase of iNOS expression relative to PBS-injected rat striatum, and LPS plus oxATP significantly reduced iNOS expression (by 67%) from that measured with LPS alone. We conclude that inhibition of P2X7R reduces LPS-induced expression and production of iNOS in rat striatum.
Effects of oxATP on LPS-mediated nitration
The enhancement of iNOS with LPS injection prompted us to examine the involvement of P2X7R on LPS-induced nitration of proteins using 3-NT immunostaining. Representative immunostaining results (Fig. 5A) indicate that little or no 3-NT+ cells were evident in PBS-injected control (top left panel). However, LPS injection (5 μg, 3 d after injection) increased the number of 3-NT+ cells (top middle panel). In the presence of oxATP with LPS, the number of 3-NT+ cells was markedly attenuated (top right panel).
Double-staining immunofluorescence was then used to determine association of 3-NT with microglia. The results (Fig. 5A, bottom row) showed a high degree of colocalization of OX-42+ microglia with 3-NT+ cells in LPS-injected rat striatum (bottom middle panel). The presence of oxATP with LPS inhibited both OX-42 and 3-NT IR (bottom right panel).
The bar graph summarizing results is presented in Figure 5B (n = 5 animals). Overall, a low level of 3-NT+ cells were colocalized with microglia in PBS-injected striatum (1.1 ± 0.6/mm2 cells). The corresponding values in LPS-injected and LPS plus oxATP-injected striatum were 40.1 ± 4.8 and 18.9 ± 2.1/mm2 cells, respectively. The number of 3-NT+ cells was decreased in the presence of oxATP by 53% compared with LPS alone. These findings suggest a role for P2X7R in microglia in mediating cellular production of peroxynitrite and oxidative stress in LPS-injected striatum.
We also examined extent of the colocalization of 3-NT with neurons. Representative double staining for 3-NT with striatal neurons (marker NeuN) is presented in Figure 5C. At 24 h after LPS injection, only a relatively small proportion of NeuN+ cells appear associated with 3-NT. Examples of 3-NT with colocalization to neurons are indicated by arrows (Fig. 5C, middle). In the presence of oxATP, we estimate a twofold reduction in the number of 3-NT/NeuN cells.
Effects of oxATP on LPS-mediated lipid peroxidation and oxidative DNA damage
Representative staining for effect of P2X7R on LPS-induced lipid peroxidation (4-HNE marker) in the rat striatum is shown (Fig. 6A, top row). LPS injection (5 μg for 24 h) induced extensive 4-HNE IR indicating lipid peroxidation in the striatum compared with PBS-injected striatum (Fig. 6A, top row). oxATP was effective in reducing the 4-HNE+ areas compared with LPS-injected striatum. Quantification in levels of 4-HNE is presented in Figure 6B, with LPS causing a considerable increase in 4-HNE IR (41-fold increase vs PBS). Treatment with oxATP was effective in reducing 4-HNE (by 50% vs LPS alone; n = 5 animals).
Reactive oxygen species can act on DNA, causing breakdown of DNA strands and base modification. 8-OHdG is an oxidized form of deoxyguanosine, one of the constituents of DNA used as a marker for oxidative DNA damage (Kasai et al., 1987). Representative results are presented showing that LPS (5 μg for 24 h), but not PBS, injection was associated with 8-OHdG IR and oxidative DNA damage (Fig. 6A, bottom left and middle). In the presence of oxATP (at 1 μm), LPS-induced oxidative DNA damage was reduced (Fig. 6A, bottom right).
Quantification of data are presented in Figure 6C, with LPS injection eliciting a large increase in 8-OHdG IR (52-fold increase vs PBS). LPS plus oxATP treatment significantly reduced 8-OHdG+ areas by 54% compared with LPS-injected striatum (n = 5 animals). Our results suggest that P2X7R is linked with oxidative stress and that oxATP inhibition of this purinergic subtype receptor reduces lipid peroxidation and oxidative DNA damage in LPS-injected rat striatum.
Effects of oxATP on LPS-induced neuronal loss and caspase-3 activation
An important aspect of this work was to determine effects of P2X7R modulation on extents of neuronal damage. We studied this point by measuring neuronal viability in LPS-injected striatum (5 μg for 3 d) in the absence and presence of oxATP. Neuronal viability was assessed by counting immunostained neurons using NeuN antibody (Dou et al., 2003; Ryu et al., 2004). A typical result is shown in Figure 7A. The number of NeuN+ cells was considerably decreased in LPS-injected striatum (Fig. 7A, middle panel) compared with PBS-injected control (left panel). However, in the presence of oxATP treatment, numbers of NeuN+ cells were increased (right panel) compared with LPS-injected striatum.
The overall results are shown in the bar graph in Figure 7B (n = 5 animals). Numbers of NeuN+ cells were counted in PBS-injected striatum and used as normalized control (100%) to analyze neuronal viability. NeuN+ cells were significantly decreased to 39.5 ± 1.8% in LPS-injected striatum compared with PBS-injected striatum. However, oxATP pretreatment significantly increased numbers of striatal neurons to a level of 62%, representing a significant degree of neuroprotection. Thus, activation of P2X7R in microglia in inflammatory conditions is associated with enhancement of neuronal damage, an effect that is partially inhibited in the presence of oxATP.
We also investigated the effects of oxATP on LPS-mediated neuronal caspase-3 activation at 24 h after LPS injection. Representative staining results for caspase-3 (green) and neurons (NeuN; red) are presented in Figure 7C. The results show little or no caspase-3 IR at 24 h in PBS-injected striatum (Fig. 7C, left panel) and elevated caspase-3 in LPS-injected striatum (middle panel). Pretreatment with oxATP reduced caspase-3 activation in LPS-injected brain (right panel). We examined caspase-3 IR at different time points (6, 12, 24, and 72 h) of LPS after injection and found a maximum caspase-3 activation at 24 h with reduced levels at 72 h (data not shown).
The effects of oxATP on LPS-mediated neuronal caspase-3 activation are presented in Figure 7D. At 24 h after LPS injection, 18.1 ± 3.4 cells/mm2 were caspase-3+ and oxATP at 1 μm was effective in inhibiting (50% vs LPS alone) the activation of neuronal caspase-3 (n = 4 animals). These results suggest that LPS-induced activation of P2X7R in microglia may be involved in activation of neuronal caspase-3, leading to neuronal loss in LPS-injected animals.
Effects of LPS intraperitoneal injection on expression of P2X7R
An interesting question was whether application of peripheral LPS would also cause changes in striatal expression of P2X7R. This point was examined using intraperitoneal injection of the endotoxin (single intraperitoneal injections at 0.5 and 5 mg/kg) with analysis for P2X7R expression at 24 h after intraperitoneal injection. Representative results are presented in Figure 8 and show no P2X7R IR with control (PBS) administration (Fig. 8, left) and a small increase in the ionotropic receptor with the lower dose of LPS (0.5 mg/kg, middle). At the higher LPS dose of 5 mg/kg, a marked upregulation in P2X7R expression was observed (right panel). These data are consistent with recent reports on effects of peripheral LPS to increase brain inflammation (see Discussion).
Effects of oxATP on LPS-induced gene expressions in cultured human microglia
We examined the involvement of P2X7R in LPS-induced gene expressions for proinflammatory and anti-inflammatory mediators and chemokines in human fetal microglia using RT-PCR. We first determined that exposure of cultured microglia to LPS (100 ng/ml) increased expression of P2X7R in a time-dependent manner (Fig. 9A). A low constitutive expression of receptor was detected in unstimulated microglia with maximum levels found at 8 h after LPS treatment. In the presence of oxATP (at 300 μm, applied 2 h before LPS treatment), LPS-induced upregulation of P2X7R was attenuated to near basal level (Fig. 9B).
Microglia were treated with LPS for 8 h, in the absence and presence of oxATP (at 300 μm), to examine expressions of proinflammatory cytokines and inducible enzyme cyclooxygenase-2 (COX-2). A representative result is presented in Figure 9C. The results show no basal expressions of TNF-α, IL-6, IL-1β, IL-12, or COX-2 in control (Fig. 9C, lane 1). All inflammatory mediators were highly expressed after human microglia exposure to LPS (Fig. 9C, lane 2). LPS in the presence of oxATP abolished expressions of all factors (Fig. 9C, lane 3). oxATP treatment alone had no effect to alter expressions of the proinflammatory mediators (Fig. 9C, lane 4).
We next examined effects of LPS and oxATP on expression of the chemokines IL-8 and monocyte chemoattractant protein-1 (MCP-1) and the anti-inflammatory cytokines IL-10 and TGF-β1. LPS stimulation of human microglia induced expression of both chemokines, which were absent in unstimulated cells (Fig. 9D). oxATP treatment partially reduced expression of MCP-1 but not IL-8. Both TGF-β1 and IL-10 were constitutively expressed in unstimulated microglia, and exposure to LPS did not alter the expression of these anti-inflammatory cytokines.
Semiquantitative RT-PCR results for proinflammatory and anti-inflammatory cytokines, enzyme COX-2, and the chemokines are presented in the bar graphs in Figure 9E (n = 3 experiments). oxATP was effective in reducing band intensities to below detectable levels for all proinflammatory mediators. Block of P2X7R reduced LPS induction of MCP-1 expression by 50% with no effects on IL-8 or anti-inflammatory cytokines. The RT-PCR results indicate that inhibition of P2X7R using oxATP abolishes expression of proinflammatory cytokines and enzyme COX-2 and attenuates expression of MCP-1 in LPS-stimulated human microglia.
Effects of oxATP on LPS-modulated [Ca2+]i in cultured human microglia
We initially measured effects of the selective P2X7R agonist BzATP to mobilize [Ca2+]i in human microglia. BzATP (300 μm) application (control, top trace) induced a prolonged [Ca2+]i increase that was abolished with oxATP pretreatment (at 300 μm, applied 2 h before BzATP application) (Fig. 10A, middle trace). In the absence of extracellular Ca2+, BzATP had no effect to alter [Ca2+]i (Fig. 10A, bottom trace). The importance of Ca2+ entry was also demonstrated using Mn2+ quenching experiments. As shown in Figure 10B, administration of BzATP in the presence of Mn2+ caused a rapid loss of fluorescence (measured at 360 nm), indicating the nature of an influx pathway. The Ca2+ ionophore ionomycin elicited a small additional quenching of fluorescence.
Additional studies investigated possible interactions between BzATP and LPS in modulating mobilization of [Ca2+]i in human microglia. A typical sustained BzATP (300 μm) response is presented in Figure 10C (labeled as control, middle trace). In the presence of LPS pretreatment (100 ng/ml for 2 h), the BzATP-induced increase in [Ca2+]i was considerably enhanced relative to control (Fig. 10C, top trace). The potentiation by LPS of the BzATP response was blocked by oxATP (at 300 μm) (Fig. 10C, bottom trace).
The bar graph summarizing results from experiments using procedures of Figure 10C is presented in Figure 10D, showing a significant increase of peak amplitudes (by 274%) of [Ca2+]i with LPS pretreatment compared with BzATP response in the absence of LPS (n = 5 separate experiments). BzATP-induced enhancement of [Ca2+]i in the presence of LPS (100 ng/ml for 2 h) was reduced with pretreatment of oxATP (by 89%). These results suggest that LPS potentiation of P2X7R-mediated [Ca2+]i mobilization may lead to modulation of functional responses in human microglia.
An important finding from this work is that activation of P2X7R in LPS-injected rat striatum is intimately linked with enhancement of a spectrum of inflammatory processes and a significant loss in numbers of striatal neurons. Most importantly, oxATP inhibition of P2X7R was effective in the inhibition of LPS-induced inflammation and was neuroprotective in inflamed brains. As discussed below, the overall results from in vivo and in vitro studies implicate P2X7R in microglia as a critical factor in mediating inflammatory responses.
Double-immunofluorescence-staining measurements showed P2X7R to be primarily associated with microglia, with a considerably smaller proportion of purinergic receptors colocalized with astrocytes; NeuN+ cells showed no expression of P2X7R (Fig. 1). Previous work has documented expression of P2X7R in glia (Collo et al., 1997), with some reports for the presence of this purinergic receptor subtype in neurons (Deuchars et al., 2001; Franke et al., 2004). Immunostaining patterns for OX-42+ microglia (Fig. 2A) indicated a predominantly ramified morphology for PBS injection, with cells showing small bodies and fine processes. In contrast, LPS intrastriatal injection was associated with ameboid morphology, with cells exhibiting swollen cell bodies and retracted processes (Fig. 2B). oxATP treatment before LPS injection was highly effective in reducing numbers of activated microglia, but remaining OX-42+ cells still showed an ameboid morphology (Fig. 2C). This result would suggest that, although oxATP reduced the number of activated microglia, the P2X7R antagonist did not prevent cellular activation. Supranigral LPS administration has been reported to increase numbers of OX-42+ microglia and to induce a shift in morphology from resting to activated state (measured at 24 h after LPS injection) (Iravani et al., 2005). At a longer time of 30 d after LPS injection, numbers of OX-42+ microglia were then reduced, with cells showing a predominant ramified morphology.
LPS induced similar patterns of time-dependent changes in the intracellular signaling pathway p38 MAPK and the nuclear transcription factor NFκB in microglia (Fig. 3). Treatment with oxATP effectively reduced IR for both p38 MAPK and NFκB for all but the earliest time point of 6 h after LPS injection. The similarity in time-dependent changes and pharmacological actions of oxATP indicate coupling between the two intracellular inflammatory mediators. Our results suggest that p38 MAPK and NFκB regulate a proliferative microglial response rather than cellular activation. Proliferation of microglia has been associated with phosphorylation of p38 MAPK (Tikka et al., 2001) and activation of NFκB (Ryu et al., 2002; Suo et al., 2003).
Because nitrogen species contribute to responses in inflamed brain (Brown and Bal-Price, 2003; Lee et al., 2005), we examined levels of inducible enzyme iNOS and 3-NT. Marked upregulation of iNOS was evident in LPS-injected striatum (at 24 h) with colocalization to microglia (Fig. 4); expression of the enzyme was significantly reduced with oxATP. RT-PCR results (at 12 h after LPS injection) also indicated upregulation of iNOS relative to control (PBS injection). Double-immunofluorescence staining, performed at the time of maximum iNOS response (24 h after LPS injection), indicated colocalization of 3-NT with OX-42+ microglia (Fig. 5A). The expressions of 3-NT and OX-42 were attenuated by oxATP. Evidence for colocalization of 3-NT with neurons was found (Fig. 5C) but comprised a minority of the neuronal population. In the presence of oxATP, we estimate a twofold reduction in the number of 3-NT/NeuN cells.
Previous work has reported involvement of 3-NT in LPS-induced brain damage (Tomas-Camardiel et al., 2004) and effects of oxATP to downregulate iNOS expression and nitric oxide production in LPS-primed macrophages (Hu et al., 1998). The inhibition of iNOS has a reported neuroprotective effect in an animal model of LPS-mediated inflammation (Arimoto and Bing, 2003). Recent work has also reported that glial-derived nitrogen species mediate enhanced blood–brain barrier permeability in amyloid-β-injected rat hippocampus (Ryu and McLarnon, 2006).
The effect of oxATP to inhibit lipid peroxidation is a novel finding in this study. Both oxidative stress markers 4-HNE and 8-OHdG were absent in PBS-injected striatum, considerably elevated with LPS injection, and reduced with oxATP pretreatment of LPS-injected striatum (Fig. 6). In the latter case, similar effects of oxATP to block 4-HNE (by 56%) and 8-OHdG (by 60%), relative to LPS alone, were found. Inhibition of oxidative stress may be linked with effects on neuronal viability. Double immunostaining (NeuN/caspase-3) showed upregulation of neuronal caspase-3 in LPS-injected striatum with suppression by oxATP. To translate caspase-3 activity to neuronal viability, numbers of NeuN+ striatal neurons were counted. At 3 d after LPS injection, numbers of neurons were reduced relative to PBS with partial recovery for oxATP pretreatment before LPS application. The involvement of caspase-3 in neuronal damage (Fig. 7) is consistent with findings of upregulation of this factor in LPS-injected (Cunningham et al., 2005) and quinolinic acid-injected (Ryu et al., 2005) rat brain. In the latter study, pharmacological inhibition of microglial activation reduced neuronal caspase-3 activation and was neuroprotective.
An ancillary question addressed in this work was the effect of application of peripheral LPS on striatal P2X7R. The results (Fig. 8) showed that both low and high intraperitoneal injections of LPS (0.5 and 5 mg/kg) induced an increased P2X7R expression relative to PBS control. Although results were not quantified, the higher dose of LPS yielded marked increases in levels of the purinergic ionotropic receptor. These data are consistent with recent reports on actions of peripheral LPS to induce inflammatory responses in brain. In one study (Qin et al., 2007), a single systemic dose of LPS (5 mg/kg, i.p.) caused a sustained increase in brain levels of the inflammatory factors TNF-α, IL-1β, MCP-1, and NFκB p65 and also damage to dopaminergic neurons. The authors suggested that LPS effects were attributable to actions of TNF-α to transfer information from periphery to brain. Effects of intraperitoneal injection of LPS on inflammation in wild-type and prion disease mice have been examined (Cunningham et al., 2005). The results showed increased levels of proinflammatory cytokines and other inflammatory factors in both groups of animals compared with controls, with diseased mice showing enhanced inflammation relative to wild-type animals.
Results from in vitro experiments supported in vivo findings and suggested a novel mechanism underlying inflammatory responses. LPS induced significant P2X7R expression in human microglia with partial block by oxATP (Fig. 9B), indicating that oxATP interferes with a signal to increase mRNA. LPS stimulation of microglia elevated expressions of proinflammatory factors (COX-2, TNF-α, IL-6, IL-1β, and IL-12); all factors were inhibited by oxATP pretreatment. Expressions of the chemokines IL-8 and MCP-1 were also increased in stimulated cells, but only MCP-1 was sensitive to oxATP. Interestingly, LPS stimulation had no effect to alter expressions of the anti-inflammatory cytokines IL-10 and TGF-β1. Thus, activation of P2X7R in microglia leads to enhanced cellular expression of an assemblage of proinflammatory mediators that, cumulatively, cause neuronal damage.
A putative mechanism to account for P2X7R-dependent inflammatory processes is LPS amplification of the entry of Ca2+ through the associated pore. Stimulation of human microglia with the specific P2X7R ligand BzATP caused a sustained increase in [Ca2+]i that was blocked by oxATP and abolished in Ca2+-free PSS. A novel finding, however, was that LPS pretreatment was highly effective in the potentiation of [Ca2+]i induced by BzATP (Fig. 10C,D). This result would be consistent with direct LPS enhancement in numbers of P2X7R. Another possibility is that LPS stimulated microglial release of ATP, which then activated additional P2X7 receptors (Ferrari et al., 1997). In any event, relatively high levels of ATP (in excess of 1 mm) seem generally required to activate P2X7R (Hide et al., 2000; McLarnon, 2005). Such elevated concentrations of ATP could be expected under pathological situations whereby endogenous cells were damaged and chronically leaky. However, to our knowledge, no evidence for such high levels of ATP have been reported in brain. In this case, the measurement of transient localized ATP may be confounded by the rapid enzymatic activity of nucleotidases. It is also possible that lower concentrations of ATP may be effective in activating P2X7R in vivo.
An important consideration is the possible limitation in the use of oxATP as a selective P2X7R antagonist. Although oxATP has often been characterized in this manner, other work has reported nonselectivity in actions of this compound. In a study of three different cell types not expressing P2X7R, oxATP was effective in attenuating LPS induction of inflammatory mediators (Beigi et al., 2003). Furthermore, this work also reported oxATP inhibitory actions against LPS-induced nitric oxide production when applied to P2X7R knock-out mice. Thus, oxATP could act to inhibit proinflammatory responses in a manner independent of effects on P2X7R.
In conclusion, the present work has documented functional roles for P2X7R in inducing inflammatory responses in an animal model of inflamed brain. Supporting evidence for LPS-induced potentiation of Ca2+ influx through this purinergic receptor subtype has been derived from in vitro study. Importantly, inhibition of P2X7R-mediated signaling by oxATP has been found effective in limiting inflammatory responses and conferring neuroprotection. Overall, our results suggest modulation of P2X7R as a plausible strategy for therapeutic intervention to reduce the damaging effects of inflammation in neurodegenerative diseases.
This work was supported by a doctoral research award from the Heart and Stroke Foundation of Canada (H.B.C.) and grants from the Pacific Alzheimer Research Foundation (J.G.M.), Alzheimer's Association USA (J.G.M.), and Korea Science and Engineering Foundation/Brain Disease Research Center (S.U.K.).
- Correspondence should be addressed to Dr. James G. McLarnon, Department of Anesthesiology, Pharmacology, and Therapeutics, University of British Columbia, 2176 Health Sciences Mall, Vancouver, British Columbia, Canada V6T 1Z3.