Gp91phox (NOX2) in classically activated microglia exacerbates traumatic brain injury

1Traumatic brain injury (TBI) is a serious condition in emergency medicine, and its pathophysiological profile is varied and complicated. One of the neurotoxic factors thought to be involved is oxidative stress. A large number of studies have reported that oxidative stress, which generates reactive oxygen species (ROS), plays a key role in the development of TBI. Consequently, one of the most obvious ways to manage TBI may be to control ROS generation given that animal experiments have supported the notion that free radical scavengers and antioxidants dramatically reduce cerebral damage. The superoxide anion (O2-) is an important free radical, and is the source of other ROS that lead to lipid peroxidation. Cyclooxygenase, xanthine oxidase, and NADPH oxidases of the NOX family are well known generators of O2- in the brain. However, the main cellular mediator of O2- generation after TBI has not yet been determined. NADPH oxidase, a multiunit enzyme initially discovered in neutrophils, has recently emerged as a major generator of ROS in neurons, glial cells and cerebral blood vessels. NADPH oxidase is composed of membrane-bound (p22phox and gp91phox) and cytoplasmic subunits (p40phox, p47phox, and p67phox). Several homologs of the catalytic subunit of the enzyme, gp91phox, also termed NOX2, exist (NOX1 through NOX5). It has been reported that gp91phox-containing NADPH oxidase produces a large amount of O2- in leukocytes, while numerous papers have reported on the role for gp91phox in various neurodegenerative conditions. However, the source and the roles of gp91phox after TBI have not been established. In this study, we used the gp91phox-/- mouse to investigate the kinetics and the roles of gp91phox following TBI.

2All experimental procedures involving animals were approved by the Institutional Animal Care and Use Committee of Showa University. The gp91phox-/- (C57/B6J) mice are described by Dinauer et al., Wild mice (Wt) were generated from the same chimeric founder, and experiments were performed in age- and weight-matched animals.

3Mice were anesthetized with 2% sevoflurane in 70% N2O and 30% O2. A controlled cortical impact was made using a pneumatically controlled impactor device as described previously.

4We used the BV-2 microglial cell line to investigate which microglia cells express gp91phox. This mouse BV-2 cell line was obtained from Interlab Cell Line Collection (Genova, Italy) and cultured in 10% RPMI1640 (RPMI1640 with 10% fetal calf serum [FCS], 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM L-glutamine [all from GIBCO/BRL, Grand Island, NY]). The cells were grown at 37°C in a humidified 5% CO2 incubator. After harvesting, the cells were washed with PBS twice and resuspended with experimental medium (Dulbecco's modified Eagle's medium [GIBCO/BRL] with 1% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM L-glutamine). The cells were seeded into six-well plates at 1 × 106 cells/well/ml and then exposed to IFNγ (10 ng/ml), IL-4 (20 ng/ml), IL-10 (10 ng/ml, all from Peprotech, Rocky Hill, NJ), or vehicle (n = 3/group). Twenty-four hours later, the cells were collected by centrifugation. The samples were kept at -30°C until analysis.

5The cerebrum was removed from decapitated animals at 0 (sham-operated), 24, and 48 hours after TBI and divided into the ipsilateral and contralateral hemispheres. These samples were then homogenized in lysis buffer (10 mM Tris-HCl [pH 7.4], 0.15 M NaCl and 1% Triton X-100, 1 mM EGTA, 50 mM NaF, 2 mM sodium orthovanadate, 10 mM sodium pyrvate, and protease inhibitor cocktail [Sigma, St. Louis, MO]), and centrifuged at 12,000 × g for 10 minutes on ice. BV-2 samples were sonicated for 10 seconds with lysis buffer to prepare cell suspensions.

6After determination of the protein concentration (BCA protein assay, Thermo Fisher Scientific, Waltham, MA), appropriate amounts of samples were electrophoresed. The separated proteins were then transferred to polyvinylidinene fluoride membranes (Bio-Rad, Hercules, CA). After blocking with 2% Blockace (DS Pharma, Osaka, Japan), the membranes were probed with primary antibodies. After washing, the membrane was probed with horseradish peroxidase (HRP)-conjugated secondary antibodies. The protein bands were detected by chemiluminescence (SuperSignal West Dura Extended Duration Substrate; Pierce, Rockford, IL) and exposed onto X-ray film. The films were scanned, and the signal densities were quantified using the UN-SCAN-IT gel analysis program (Silk Scientific, Orem, UT).

7Mice subjected to TBI were placed under pentobarbital (50 mg/kg, i.p.) anesthesia and perfused with 0.9% NaCl followed by 2% paraformaldehyde (PFA). Brains were removed and processed to frozen blocks that were then cut into 8-μm sections (n = 4-5/group).

8The sections were incubated with 0.3% H2O2 and then incubated with PBS containing 5% normal horse serum to mask nonspecific reactions. Next, the sections were incubated with antibody raised against gp91phox. One day later, the sections were rinsed and incubated with biotinylated goat anti-rabbit IgG (Santa Cruz Biotechnology, Santa Cruz, CA), and then with an avidin-biotin complex solution (Vector Laboratories, Burlingame, CA) followed by diaminobenzidine (DAB; Vector) as a chromogen.

9A similar procedure was used for multiple immunostaining, except that the sections were not incubated with 0.3% H2O2, and were incubated with Alexa-labeled fluorescence secondary antibodies. Primary and secondary antibodies for multiple-staining are listed in Table 1. 4,6-Diamidine-2-phenylindole dihydrochloride (DAPI, 1:10,000; Roche, Mannheim, Germany) was used for nuclear staining. The fluorescence and immunolabeling were detected using a confocal laser microscope (AX-10, Zeiss; Oberkochen, Germany).

10The areas of injured brain were determined using 2,3,5-triphenyltetrazolium chloride (TTC) staining of tissues 48 hours after TBI. The animals were decapitated, and the brain was sectioned into four 2-mm coronal sections by using a mouse brain matrix. The brain slices were then stained with 2% TTC at 37°C for 30 min and photographed on the anterior surface of each section with a scale bar. The areas of injured brain were delineated by examining differences between the ipsilateral and contralateral regions in the center slice of injured brain and measured by using NIH Image software. http://rsb.info.nih.gov/nih-image/about.html.

11To determine neural apoptosis-like cell death, terminal deoxynucleotidyl transferase-mediated dUTP end-labeling (TUNEL) staining (In Situ Cell Death Detection Kit, POD; Roche) was performed 48 hours after TBI (n = 5) and the number of TUNEL-positive cells in the ipsilateral hemisphere was then counted and compared gp91phox-/- with Wt mice in a similar cortical region (40 × magnification).

12Production of O2- was determined by in situ detection of oxidized hydroethidium (HEt). With the animal placed under anesthesia, the HEt solution was administered (1 mg/mL 0.9% NaCl with 1% DMSO) into the jugular vein (n = 3 per group) 48 hours after TBI. Fifteen minutes later, the brain was removed and frozen in blocks and cryosectioned (8 μm) in the coronal plane. To demonstrate the cellular distribution of Et, the sections were co-stained with antibodies raised against gp91phox, CD11b, GFAP, or NeuN. Fluorescence was detected using confocal laser microscopy (AX-10, Zeiss, Germany)

13Levels of NO and TNFα production are markers of classically activated microglia. NO production was measured using the Griess method (Dojindo, Kumamoto, Japan) as total NO (NO2- and NO3-). TNFα production was measured by enzyme-linked immunosorbent assay using the Duoset ELISA Development System (R&D Systems, Minneapolis, MN).

14Arginase is a marker for alternatively activated microglia. Arginase activity was measured according to a previous paper with minor modification. In brief, the cell homogenate was mixed with equal volumes of prewarmed 50 mM Tris-HCl, pH 7.5 containing 10 mM MnCl2 and incubated for 15 minutes at 55°C. The mixture was incubated in 0.25 M L-arginine for 60 minutes at 37°C to produce urea from arginine and the reactions were stopped by adding Stop solution (H2SO4/H3PO4/H2O, 1:3:7). Then, 1% (final concentration) 1-phenyl-1, 2-propanedione-2-oxime (ISPF) in ethanol was added to the solution, which was heated at 100°C for 45 min. The reaction between urea and ISPF produced a pink color, and absorption was measured at 540 nm.

15Data are expressed as mean ± SE for in vivo experiments. Data are expressed as mean ± SD for in vitro experiments Statistical comparisons were performed using the Student's t tests and two-way analysis of variance (ANOVA) as appropriate. P values less than 0.05 were considered statistically significant.

16Fig 1 shows the results of immunoblotting experiments to describe the expression of gp91phox after TBI. Protein levels of gp91phox were not increased after TBI in the contralateral hemisphere of Wt mice. In the ipsilateral hemisphere of Wt mice, protein levels of gp91phox were greater at 1 and 2 days after TBI relative to the contralateral hemisphere (Fig 1A, B). The data indicate that gp91phox increased in the ipsilateral hemisphere following TBI stress.

17The results of single immunohistochemical detection of gp91phox after TBI are presented Fig 2A. Two days after TBI, gp91phox immunoreactivity was dramatically increased in the peri-contusion region.

18To identify the cell types expressing gp91phox in the peri-contusional area, cells were co-labelled with antibodies raised against gp91phox and markers of microglial (CD11b), astroglial (GFAP), and neuronal (NeuN) cells (Fig 2B). The immunoreactivity for gp91phox was mainly co-localized to microglia that had amoeboid-like morphological features with round cell body and short processes, suggesting cytotoxic-type classical activation. Immunoreactive staining for gp91phox was also co-localized to a few astrocytes and neurons.

19TTC staining was used to evaluate the role of gp91phox on the severity of TBI in gp91phox-/- and Wt mice (Fig 3A, B). Images of the TTC-stained anterior surface of coronal sections demonstrated that the injured brain area in gp91phox-/- mice was significantly smaller (P < 0.01) than that of the Wt mice (Fig. 3B). We verified the prevention of the cortical injury with TUNEL staining, which was used to identify apoptotic-like cell death at 48 hours after TBI (Fig 3C, D). The number of TUNEL-positive cells of gp91phox-/- mice was significantly less (p < 0.05) compared with Wt mice (Fig 3D).

20High intensity fluorescence staining by the oxidation product Et was observed, indicating high levels of O2- production. The peri-contusional area of Wt mice accounted for much of the Et signal after TBI (Fig 4A). Et fluorescence was lower in gp91phox-/- mice (Fig 4A) compared to Wt mice (p < 0.05) (Fig 4B). The Et fluorescence was associated with p91phox-positive cells (Fig. 4C), and CD11b-positive microglial cells (Fig. 4D). Some Et fluorescence was slightly detected in a few neurons (Fig. 4D). These results suggest that O2- is mainly produced by microglia expressing gp91phox.

21Peroxynitrite (ONOO-) is an oxidant and nitrating agent and is synthesized by the reaction between O2- and nitric oxide (NO). ONOO- can damage a wide array of cellular molecules, including DNA and proteins. 3-NT is an oxidized metabolite of ONOO- in vivo. We performed multiple-immunostaining for 3-NT and cell marker antibodies in the peri-contusional area at 48 hours. Immunoreactivity for 3-NT in Wt mice clearly co-localized with microglia, astrocytes, and degenerated neurons (Fig 5; upper panel). In gp91phox-/- mice, however, 3-NT immunoreactions were dramatically suppressed in the microglial and astroglial cells (Fig 5). While 3-NT immunoreactivity was observed in gp91phox-/- mouse neurons, the integrity of these cells was preserved (Fig 5). Taken together these findings suggest that production of O2- and ONOO- by gp91phox may influence the severity of TBI.

22From the results presented in Fig. 2 we suggested that gp91phox co-localized with classical activated microglia. We then tried to identify the microglia that expressed gp91phox by using the mouse microglial cell line BV-2 and by activating these cells with IFNγ, IL-4, or IL-10. As shown in Fig 6, the media from the BV-2 cells showed significantly increased total NO after exposure to IFNγ, but not to IL-4 or IL-10 at 24 hours. The IFNγ-exposed BV-2 cells had significantly increased TNFα in their media and expressed iNOS as well. All of these factors indicate that the BV-2 cells were activated according to the classically activated phenotype. On the other hand, the IL-4-exposed BV-2 cells showed increased arginase activity and levels of Ym-1, indicating an alternatively activated phenotype. IL-10-exposed BV-2 cells did not show any changes in levels of these factors. Immunoblotting experiments showed that gp91phox levels increased in the IFNγ-exposed BV-2. Moreover, immunoblotting for a different NADPH oxidase subunit, p22phox, mirrored that of gp91phox, suggesting that not only gp91phox but also NADPH oxidase expression was induced in the classically activated microglia.

23We demonstrate here that gp91phox is increased in the ipsilateral hemisphere after TBI and specifically in amoeboid-shaped microglial cells. Mice that are gene-deficient for gp91phox exhibit reduced primary cortical damage, as evidenced by reduced areas of contusion, and reduced secondary damage as detected by TUNEL staining. Moreover, we have shown that the gene-deficient mice have lower levels of ROS at the injury site and widespread oxidative damage after TBI. Finally, we demonstrate in a BV-2 microglial cell line that gp91phox and/or NADPH oxidase are increased in classically activated microglial cells which are activated by IFNγ.

24Gp91phox is expressed constitutively in neurons but not in glial cells, and O2- production might play a role in neuronal homeostasis. However, in neuropathological conditions including neurodegenerative diseases and stroke, the gp91phox-containing NADPH oxidase has been observed in glial cells, neurons, fibroblasts and vascular endothelial cells, and seems to be involved in ROS formation. Microglial cell gp91phox appears to be involved in the induction of neuronal damage in Parkinson's disease, Alzheimer's disease, and ischemic stroke. In the present study, we demonstrated that gp91phox is mainly expressed in microglia, and at lower levels in neurons and astrocytes. These microglia exhibit features of being classically activated. We then demonstrated that the microglial phenotypes expressed both gp91phox and p22phox, which probably reflects the induction of NADPH oxidase in the BV-2 mouse microglial cell line that we used. Our results reveal that NADPH oxidase is induced in INFγ-stimulated, classically activated microglial cells characterized by increased NO, iNOS, and TNFα.

25In gp91phox-/- mice, in situ generation of O2-and an ONOO- metabolite, 3-NT, were suppressed in microglia after TBI. ONOO- produced by O2- and NO after TBI induces oxidative damage, secondary brain damage and neuroinflammation after TBI. We have verified that TUNEL-positive apoptotic-like cells in the peri-contusional area and contusion area are suppressed in gp91phox-/- mice. These results suggest that gp91phox in classically activated microglia-like cells has a harmful role in primary and secondary brain damage after TBI.

26The question of whether microglial cells play harmful or beneficial roles in CNS injures has been widely debated and reviewed over several decades. The roles of activated microglia in neuroinflammation are thought to be complex. Classical activation is induced by IFNγ and is related to the production of proinflammatory mediators in the innate immune response. Another form of activation, called "alternative activation," is induced by IL-4 and IL-13 and, compared to classical activation, does not result in high levels of expression of proinflammatory mediators such as cytokines and NO. The roles of alternatively activated microglia during inflammatory process are thought to involve tissue repair, the production of anti-inflammatory cytokines, fibrosis, and extracellular matrix reconstruction. Recently, Ohtaki et al. reported that the injection of human mesenchymal stromal cells protects against ischemic brain injury by modulating inflammatory and immune responses through the alternative activation of microglia and/or macrophages. These studies and our data suggest that controlling microglial activation and understanding its mechanism and functional significance following TBI may open exciting new therapeutic avenues.

27The suppression of free radical generation and the scavenging of free radicals after brain damage are important therapies. Animal experiments have supported the notion that free radical scavengers and antioxidants dramatically reduce TBI. Excessive O2- may produce destructive hydroxyl radicals (OH-) and alkoxyl radicals (OR-) by the iron-catalyzed Haber-Weiss reaction. The brain is especially prone to radical damage because it is highly enriched in easily peroxidizable unsaturated fatty acid side chains and iron. Many studies investigating ischemic injury suggest that inhibition of NADPH oxidase or gp91 phox is an important therapeutic target for neuroprotection. Some recent studies have demonstrated that expression of gp91phox increases in brain after intracerebral hemorrhage, resulting in enhanced lipid peroxidation. These studies also reported that hemorrhage volume, brain edema, and neurological function are reduced in gp91phox-/- mice, while Lo et al. reported that neurological outcomes are improved in gp91phox-/- mice. On the other hand, Liu et al. reported that there are no significant differences in mortality rate, brain water content and intensity of oxidative stress between gp91phox-/- and wild type mice in a mouse model of subarachnoid hemorrhage (SAH). In the present study, we have shown that, in gp91phox-/- mice, gp91phox expressed in classically-activated microglial-like cells plays a key role in O2- production after TBI, and that gp91phox-derived O2- is a key signal for contusion and cell death after TBI. Our findings indicate that gp91phox inhibition and control of microglial classically-activation might provide a new therapeutic option by suppressing ROS generation after TBI. However, the indirect influence with TBI by which ROS production by other NOX families and immune cells such as endothelial cell or leukocytes is currently unclear and will be an important question for future TBI studies.

28In conclusion, we have shown that gp91phox is expressed in classical activated microglial-like cells mainly in the peri-contusional area after TBI. An important generator of O2- is gp91phox during the acute phase of TBI. As ROS derived from gp91phox play an important role in primary and secondary brain damage after TBI, modulation of gp91phox in classically activated microglia and gp91phox-derived ROS may provide a new therapeutic strategy in combating post-traumatic brain injury.

29The authors declare that they have no competing interests.

30KD performed the majority of experiments and data analysis, and wrote the initial version of the manuscript. HO was involved in evaluation of microglia using BV-2 cell. TN, SY, DS and KM were substantial contributions to western blotting assay and immunohistochemistry. ST provided gp91 knockout mice. KS, SS and TA supervised all experimental procedures. All of the authors have read and approved the final version of the manuscripts.