Complement component 3 inhibition by an antioxidant is neuroprotective after cerebral ischemia and reperfusion in mice


  • Jiwon Yang,

    1. Department of Pharmacology, College of Pharmacy, Sookmyung Women's University, Seoul, Republic of Korea
    Search for more papers by this author
  • Hye-na Ahn,

    1. Department of Pharmacology, College of Pharmacy, Sookmyung Women's University, Seoul, Republic of Korea
    Search for more papers by this author
  • Minsun Chang,

    1. Department of Medical and Pharmaceutical Science, College of Science, Sookmyung Women's University, Seoul, Republic of Korea
    Search for more papers by this author
  • Purnima Narasimhan,

    1. Department of Neurosurgery, Department of Neurology and Neurological Sciences, and Program in Neurosciences, Stanford University School of Medicine, Stanford, California, USA
    Search for more papers by this author
  • Pak H. Chan,

    1. Department of Neurosurgery, Department of Neurology and Neurological Sciences, and Program in Neurosciences, Stanford University School of Medicine, Stanford, California, USA
    Search for more papers by this author
  • Yun Seon Song

    Corresponding author
    1. Department of Neurosurgery, Department of Neurology and Neurological Sciences, and Program in Neurosciences, Stanford University School of Medicine, Stanford, California, USA
    • Department of Pharmacology, College of Pharmacy, Sookmyung Women's University, Seoul, Republic of Korea
    Search for more papers by this author

Address correspondence and reprint requests to Dr. Y. S. Song, College of Pharmacy, Sookmyung Women's University, Chungpa-ro 47 gil 100, Yongsan-ku, Seoul, 140-742, Korea. E-mail:


Oxidative stress after stroke is associated with the inflammatory system activation in the brain. The complement cascade, especially the degradation products of complement component 3, is a key inflammatory mediator of cerebral ischemia. We have shown that pro-inflammatory complement component 3 is increased by oxidative stress after ischemic stroke in mice using DNA array. In this study, we investigated whether up-regulation of complement component 3 is directly related to oxidative stress after transient focal cerebral ischemia in mice and oxygen-glucose deprivation in brain cells. Persistent up-regulation of complement component 3 expression was reduced in copper/zinc-superoxide dismutase transgenic mice, and manganese-superoxide dismutase knock-out mice showed highly increased complement component 3 levels after transient focal cerebral ischemia. Antioxidant N-tert-butyl-α-phenylnitrone treatment suppressed complement component 3 expression after transient focal cerebral ischemia. Accumulation of complement component 3 in neurons and microglia was decreased by N-tert-butyl-α-phenylnitrone, which reduced infarct volume and impaired neurological deficiency after cerebral ischemia and reperfusion in mice. Small interfering RNA specific for complement component 3 transfection showed a significant increase in brain cells viability after oxygen-glucose deprivation. Our study suggests that the neuroprotective effect of antioxidants through complement component 3 suppression is a new strategy for potential therapeutic approaches in stroke.

Abbreviations used



CCAAT/enhancer-binding protein delta


complement component 3


cerebral blood flow




2′,7′-dichlotofluorescin diacetate


Enzyme-linked immunosorbent assay


Glyceraldehyde-3-phosphate dehydrogenase


ischemia and reperfusion


ionized calcium-binding adaptor molecule 1


inducible nitric oxide synthase




lactate dehydrogenase


neuron-specific nuclear protein


oxygen-glucose deprivation




phosphate-buffered saline


quantitative polymerase chain reaction


reactive oxygen species


small interfering RNA


copper/zinc-superoxide dismutase


manganese-superoxide dismutase


transient focal cerebral ischemia


transcription factor II D






2,3,5-triphenyltetrazolium chloride



Stroke is the rapidly developing loss of brain function because of failure of cerebral blood flow (CBF). It is the third leading cause of death worldwide, and the mortality rate has not improved during the past two decades. Reperfusion after cerebral ischemia generates additional overproduction of free radicals, which leads to secondary brain injury (Taskapilioglu et al. 2009). The brain is a vulnerable target for reactive oxygen species (ROS)–induced damage for many reasons, such as high oxygen consumption, a low level of protective antioxidants, and a high concentration of peroxidizable lipids. Oxygen deprivation causes nerve cells in the affected area to die within minutes.

Complement is a host defense system that identifies pathogens and injured cells, recruits inflammatory cells, and induces cell lysis (del Zoppo 1999). The complement cascade has three different pathways: the classical pathway, the mannose-binding lectin pathway, and the alternative pathway. The initiators of the three pathways are different, but all three ultimately lead to activation of complement component 3 (C3). This is the crucial step for the biological activity of the complement system because C3-cleavage byproducts lead to further downstream activation, which generates responses such as anaphylaxis, chemotaxis, and phagocytosis (Daha 2010). It is also reported that there is a fourth extrinsic protease pathway, which generates C5a in the absence of C3 (Huber-Lang et al. 2006).

It was once thought that the blood–brain barrier blocks the brain from the immune system. However, astrocytes, microglia, and neurons produce complement to maintain immunosurveillance in the brain (Gasque et al. 2000). We know that ischemic stroke increases complement levels and results in neurological disability (Cojocaru et al. 2008). C3 deficiency is associated with a better outcome after acute stroke (Cervera et al. 2010). C3a and C5a are anaphylatoxins that are bioactive fragments of C3 and C5. These anaphylatoxins induce release of various mediators from mast cells and phagocytes that amplify inflammatory responses. Inhibition of the complement cascade produces a better outcome in heart, liver, and brain after ischemic damage (Mocco et al. 2006; Ducruet et al. 2009). Although numerous approaches have been tried to reduce ischemic damage, including inhibition of C1, C3, C5, and membrane attack complex, and using cobra venom factor for complement depletion, none has reached the market (Arumugam et al. 2009; Qu et al. 2009). Although many studies have been done to understand the relationship between complement components and stroke, the role of the complement cascade in cerebral ischemia is still not fully understood.

Here, we studied the relationship between C3 up-regulation and oxidative stress after transient focal cerebral ischemia (tFCI). To examine how complement components depend on oxidative stress after ischemic stroke, we used copper/zinc-superoxide dismutase transgenic (SOD1 Tg) mice and manganese-superoxide dismutase knock-out (SOD2 KO) mice. There were no significant differences in glutathione and ascorbate levels between SOD1 Tg and wild-type (WT) mice after focal cerebral ischemia (Kinouchi et al. 1991). As glutathione and catalase system are intact in SOD1 Tg and SOD2 KO mice, the total ROS levels were increased not because the compensatory reaction but because oxidative stress from ischemic damage. We pharmacologically studied the inhibitory role of the antioxidant N-tert-butyl-α-phenylnitrone (PBN) in complement activation after tFCI in mice. We also specifically knocked-down C3 level to figure out specific role of C3 after ischemic injury, and measured C3 plasma level. The goal of this study was to elucidate the interactions between complement cascade activation and oxidative stress after cerebral ischemia.

Materials and methods

All experiments with animals were performed in accordance with the Association for Assessment and Accreditation of Laboratory Animal Care of Sookmyung Women's University. All animals were killed with an overdose of anesthetic after the experiments to minimize pain or discomfort.


Eight-week-old CD1 male mice were purchased from Samtaco (Osan, Korea). Tg mice of the TgHS/SF-218 strain, which carries the SOD1 gene with a CD1 background, were derived from the founder stock (Epstein et al. 1987). There were no observable phenotype differences between the Tg mice and their WT normal littermates. SOD2 KO mutants (heterozygous) with a CD1 background were backcrossed with CD1 mice for at least 10 generations and genotypes were determined as previously described (Li et al. 1998). Littermate WT mice with an identical genetic background were used in each experiment. The physiological parameters showed no significant difference between SOD1 Tg or SOD2 KO and WT (Kinouchi et al. 1991; Kim et al. 2002). All animals were housed in a purpose-built facility with a controlled environment and maintained in an isolator in which the control was set to keep temperature and relative humidity at 23 ± 2°C and 50 ± 10%, respectively. Artificial lighting provided a 24 h cycle of 12 h light:12 h dark.


Male CD1 mice (38 to 40 g) were anesthetized with 2.0% isoflurane in 30% oxygen and 70% nitrous oxide using a face mask. Rectal temperature was controlled at 37°C using a homeothermic blanket and physiological parameters were monitored throughout the surgeries. The mice were subjected to 45 min of tFCI as described previously with minor modifications (Song et al. 2007). Briefly, an 11.0-mm 5-0 surgical monofilament nylon suture, blunted at tip, was introduced through the left external carotid artery stump. After middle cerebral artery occlusion, blood flow was restored by withdrawal of the nylon suture, allowing reperfusion. In sham-operated mice, the same procedure was performed except for middle cerebral artery occlusion. Regional CBF was measured continuously in the middle cerebral artery territory from the temporal bone surface with a laser-Doppler probe (Omegawave, Tokyo, Japan) during animal surgery. Changes in CBF were expressed as a percentage of the baseline value. No significant differences were observed among different groups. Animal exclusion criteria were as follows: neurological behavior score below 2 immediately after reperfusion, bleeding during tFCI, dead during reperfusion, or intra-arterial thrombolysis. Altogether, 11 mice were excluded out of 266 mice according to the exclusion criteria.

Behavioral analysis

An analysis was done to determine the degree of behavioral and neurological deficits in the mice after 30 min, 1 h, 24 h, 3 days, and 7 days of reperfusion. Circling and forelimb behavior was tested and scored according to a 5-score scale system. Circling 5-score scale: Grade 0, no observable deficit; Grade 1, flexion of contralateral torso; Grade 2, circling clockwise; Grade 3, spinning clockwise longitudinally; and Grade 4, unresponsive. Forelimb 5-score scale: Grade 0, normal function; Grade 1, occasional asymmetric forelimb flexion; Grade 2, asymmetric forelimb flexion; Grade 3, asymmetric forelimb and torso flexion; and Grade 4, no spontaneous motor activity.


PBN is a commonly used free-radical spin-trap agent that inhibits the induction of nitric oxide synthase, thereby preventing overproduction of nitric oxide. We injected PBN (40 and 60 mg/kg) (Sigma-Aldrich, St. Louis, MO, USA) intraperitoneally1 h before onset of tFCI and every 24 h after tFCI to reduce ROS for PBN pre-treatment group, according to Lancelot et al.(1997), with minor modifications. For PBN post-treatment group, PBN was injected 1 h after reperfusion and every 24 h for 7 days.

2,3,5-Triphenyltetrazolium chloride staining

After 45 min of tFCI and 3 days of reperfusion, the brains were carefully removed and sliced into 1-mm coronal sections. The slices were then incubated in 2% 2,3,5-triphenyltetrazolium chloride (TTC) solution (Sigma-Aldrich) for 10 min at 37°C followed by fixation with 4% paraformaldehyde. The infarct volumes were measured using Image J software (NIH, Bethesda, MD, USA). Direct infarct volume was obtained by integrating measured infracted areas, and indirect infarct volume was obtained by deducting the non-infarcted volume of the ipsilateral hemisphere from the contralateral hemisphere.

Preparation of brains

Animals were decapitated after reperfusion. The brains were removed and the forebrain was cut coronally 3-mm to 8-mm distal from the frontal pole by using the mouse brain matrix. Samples were obtained from the middle cerebral artery territory brain tissue on the ischemic side, including the striatum and cortex, and were kept at −80°C until use.

Quantitative real-time polymerase chain reaction

The brain samples were removed and the middle cerebral artery territory was obtained at each time point. TRIzol reagent (Invitrogen, Carlsbad, CA, USA) was used to extract total RNA according to the instructions of the manufacturer. All quantitative polymerase chain reactions (QPCRs) were performed using the 7500 Real-Time PCR system (Applied Biosystems, Foster City, CA, USA). A total of 1 μg RNA was reverse-transcribed using iscript™ cDNA Synthesis Kit (Bio-Rad, Hercules, CA, USA). QPCR was performed using SYBR Green Realtime PCR master mix (Toyobo, Osaka, Japan). Data analysis was done using 7500 software (Applied Biosystems). Standardization was performed using human RNase P gene according to the manufacturer's instructions. The primers sequences were as follows: C3 forward GAGAAAAGC CCAACACCAGCTA, reverse GCTCCACCCACGTGTCCTT; Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) forward GAACGGGAAGCTTGTCATCA, reverse TGACCTTGCCCACAGCCTTG.

Western blot analysis

Cell lysis buffer (Cell Signaling Technology, Beverly, MA, USA) and ProteoExtract Subcellular Proteome Extraction Kit (EMD Chemicals, Gibbstown, NJ, USA) were used to obtain protein samples. Protein concentrations were measured using the bicinchoninic acid method. Equal amount of proteins were separated on a NuPAGE® 4–12% Bis-Tris Gel (Invitrogen) and immunoblotted with the specific antibodies as follows: anti-human C3 (EMD Chemicals), anti-mouse C3 (Hycult Biotechnology, Plymouth Meeting, PA, USA), anti-human C3a and anti-human C5a (EMD Chemicals), 4-hydroxynoneal (4-HNE; Oxis international Inc., Beverly Hills, CA, USA), cyclooxygenase-2 (COX-2; Cayman Chemical, Ann Arbor, MI, USA), inducible nitric oxide synthase (iNOS; Becton Dickinson, San Jose, CA, USA), β-actin (Sigma-Aldrich), and transcription factor II D (TFIID; Santa Cruz Biotechnology, Santa Cruz, CA, USA). The signals were then detected with an enhanced chemiluminescent kit (Thermo Scientific Dharmacon, Lafayette, CO, USA).

C3 level measurement in mouse plasma by ELISA

A 96-well immunoplate was coated with 50 μL of 2 μg/mL goat anti-human C3 (EMD Chemicals) in binding solution (0.1 M Na2HPO4, pH 9.0) for overnight at 4°C. Then the wells were blocked with 200 μL of blocking solution [10% Fetal Bovine Serum in phosphate-buffered saline (PBS)] for 1.5 h at 20°C. The C3 peptide (EMD Chemicals) was diluted in blocking buffer and both C3 peptide standard and mouse plasma were added to the wells. The bound C3 was detected with avidin-horseradish peroxidase. The reaction was developed by adding 3,3′,5,5′-tetramethylbenzidine (TMB) substrate (Becton Dickinson) and stopping solution (2M H2SO4). The optical density was measured in an Enzyme-linked immunosorbent assay (ELISA) reader at 405 nm. All unbound proteins were removed by washing with PBS containing 0.05% tween-20.

Double immunofluorescent staining

The mice were cardioperfused with heparin (10 U/mL), followed by fixation with 4% paraformaldehyde after tFCI, after 24 h of reperfusion. The brains were post-fixed with 4% paraformaldehyde overnight at 4°C and sliced into 40–μm coronal sections using a vibratome (Leica Microsystems, Buffalo Grove, IL, USA). The sections were blocked with normal horse serum and then probed with goat anti-human C3 diluted 1 : 50 (EMD Chemicals). For immunofluorescent double staining, neuron-specific nuclear protein (NeuN; Millipore, Billerica, MA, USA) or ionized calcium-binding adaptor molecule 1 (Iba-1; Wako Pure Chemical Industries, Osaka, Japan) was used. The stained sections were then mounted using fluorescence mounting medium and detected by confocal microscopy (LSM510; Carl Zeiss, Thornwood, NJ, USA). Heparin was used for only immunofluorescent staining samples.

Primary cortical neuron culture

Mouse cerebral cortices at E16 were isolated, then the tissue was minced and trypsinized with 0.25% trypsin. Cortical neurons were plated on poly-d-lysine-coated plates and cultured in minimum essential medium (Invitrogen). The medium was changed to Neurobasal medium containing B-27 (Invitrogen) after 1 day and every other day. The neurons were transfected with 25 nM of C3 specific small interfering RNA (siRNA; Ambion, Austin, TX, USA) 5 days after isolation and 24 h later the cells were subjected to oxygen-glucose deprivation (OGD). All experiments were completed within 10 days.

Oxygen-glucose deprivation

Ischemia was induced using an anaerobic chamber (Plas Labs, Lansing, MI, USA). Mouse primary cortical neurons were subjected to 3 h of OGD by replacing the medium without glucose and placing the cells in an anaerobic chamber (Plas Labs). After OGD, the medium was replaced with normal culture medium and the cells were returned to 5% CO2/95% air incubation for reoxygenation periods.

Cell cytotoxicity assay

Cell cytotoxicity was evaluated using lactate dehydrogenase (LDH) assay kit (BioVision, Milpitas, CA, USA). After OGD, media was collected and incubated with LDH catalyst and dye mixture at 37°C for 30 min in dark. The absorbance was measured at 490 nm using a microplate ELISA reader (Molecular Devices, Sunnyvale, CA, USA). The reference wavelength was measured at 600 nm.

Measurement of ROS generation

Intracellular ROS were measured using the oxidation-sensitive fluoroprobe 2′,7′-dichlorofluorescin diacetate (DCFH-DA; Sigma-Aldrich). Primary neurons were then incubated with 10 mM DCFH-DA dissolved in serum-free medium at 37°C for 30 min. 3 h of OGD was induced and the cells were lysed after 30 min of reperfusion. ROS levels were quantified using Victor 3 (PerkinElmer, Waltham, MA, USA) at an excitation wavelength of 488 nm and an emission wavelength of 525 nm. During whole procedure after DCFH-DA treatment, the cells were kept out of light to minimize the quenching of the fluoroprobe.

Luciferase assay

The C3-luciferase plasmid containing a 1.1-kb fragment of the C3 promoter (−1030 to +58) was used (Fan et al.1996). Transient transfection was performed with 500 ng of reporter plasmid and endogenous control β-galactosidase using Lipofectamin® LTX and Plus Reagent (Invitrogen). After 24 h, OGD were induced and then the primary neurons were lysed in 1× Passive lysis buffer (Promega, Madison, WI, USA) after reperfusion. The extracts were prepared for luciferase reporter assay. Luciferase activity was assessed using Victor 3 (PerkinElmer).

Statistical analysis

The data are expressed as mean ± SEM. Comparisons between two or multiple groups were performed by one-way anova followed by the Neuman–Keuls multiple comparison test (Prism; GraphPad, San Diego, CA, USA). < 0.05 was considered significant.


Higher oxidative stress enhanced C3 expression after tFCI

Our previous temporal profile of a DNA array showed that C3 expression was down-regulated in SOD1 Tg mice that had lower superoxide levels compared with WT mice after tFCI (Song et al. 2007). Differential expression of the C3 gene was further confirmed and compared by QPCR. WT and SOD1 Tg mice were subjected to tFCI followed by 6 h, 24 h, and 7 days of reperfusion. In the WT mice, C3 mRNA levels were increased up to 7 days. At 7 days of reperfusion, the WT mice showed highly increased C3 mRNA levels compared with sham mice, with a statistical significance. However, the SOD1 Tg mice showed a significant reduction in C3 mRNA levels compared with the WT mice (Fig. 1a). The mRNA levels were normalized to GAPDH for each sample. To further investigate the close relationship between C3 up-regulation and oxidative stress after tFCI, we used SOD2 KO mice, which have higher oxidative stress levels than WT mice. The level of 4-HNE, an α,β-unsaturated hydroxyalkenal which is produced by lipid peroxidation in cells, was highly increased in SOD2 KO mice compared with WT mice (Fig. 1b). The C3-related protein levels were significantly increased after tFCI in the SOD2 KO mice compared with the WT mice (Fig. 1b). These results confirm that higher levels of oxidative stress results in increased levels of C3-related proteins after tFCI.

Figure 1.

Higher oxidative stress increased C3 after transient focal cerebral ischemia (tFCI). (a) Relative C3 mRNA levels were measured by QPCR using total RNA extracted from the ischemic brains of mice. mRNA expression was detected using the C3 primer sequences described in the Materials and methods. Copper/zinc-superoxide dismutase (SOD1) Tg mice had reduced C3 mRNA levels after tFCI compared with WT mice (***< 0.001, = 4, comparison between WT and SOD1 Tg mice at the same time points, ###< 0.001, = 4, compared to WT sham). (b) Relative C3 protein levels were measured by western blotting after tFCI. β-Actin was used as an internal control. In manganese-superoxide dismutase (SOD2) KO mice, the amounts of C3 protein were increased up to 3 days and were higher than in the WT mice after tFCI. 4-HNE protein levels were also up-regulated in SOD2 KO mice. The graphs illustrate that C3 proteins, including C3 α-chain and β-chain, were significantly increased in the SOD2 KO mice after tFCI (***< 0.001, = 4, comparison between WT and SOD2 KO mice at the same time points, ###< 0.001, = 4, compared to WT sham).

The antioxidant PBN suppressed C3 levels after tFCI in mice

To further investigate the role of ROS in the C3 system after tFCI, we intraperitoneally injected a radical-trapping agent, PBN (40 and 60 mg/kg), 1 h before onset of tFCI and every 24 h after tFCI for PBN pre-treated group. Total RNA was prepared from brain after tFCI followed by 24 h or 3 days of reperfusion. C3 mRNA levels were highly increased after tFCI up to 3 days, and both 40 and 60 mg/kg PBN lowered C3 mRNA levels (Fig. 2a). As 40 mg/kg of PBN significantly decreased C3 mRNA levels, we used that dose for all subsequent experiments. PBN post-treated group significantly down-regulated C3 mRNA levels up to 7 days of reperfusion compared with vehicle-treated group (Fig. 2b). In accordance with the QPCR data, the C3 protein increased up to 3 days in the vehicle group. However, in the PBN group, it did not show any significant increases, and was considerably lower than the vehicle group. Also, PBN-treated mice showed lower 4-HNE protein level as well as lower C3 protein levels (Fig. 2c). After 45 min of tFCI followed by 24 h or 3 days of reperfusion, blood was collected and plasma was isolated by centrifugation. C3 plasma level was significantly increased after tFCI, but was suppressed by PBN (Fig. 2d). These results confirmed that C3 levels were increased after cerebral ischemia and were decreased by antioxidant.

Figure 2.

The antioxidant N-tert-butyl-α-phenylnitrone (PBN) suppressed C3 levels after transient focal cerebral ischemia (tFCI). (a) The vehicle and PBN were injected 1 h prior to onset of tFCI and every 24 h after tFCI. tFCI increased C3 mRNA levels in the vehicle-treated group. In the PBN pre-treated group, the relative C3 mRNA level was reduced compared with the vehicle-treated group (***< 0.001, = 6, compared to PBN 0 mg/kg-treated mice at the same time points; -fold to sham 24 h) (b) Post-treatment of PBN also significantly decreased C3 mRNA levels after tFCI compared to vehicle-treatment (***< 0.001, = 4, compared to vehicle-treated mice at the same time points; -fold to vehicle-sham). Veh indicates vehicle. (c) Western blot analysis showed that the increase in C3 was markedly suppressed by PBN after tFCI. 4-HNE levels were down-regulated by PBN (***< 0.001, = 6, comparison between vehicle– and PBN-treated mice at the same time points; -fold to vehicle-sham). (d) C3 blood levels increased after tFCI and were reduced by antioxidant. Blood from mice were collected after tFCI followed by 24 h and 3 days reperfusion, plasma isolated, and C3 levels were measured by ELISA (***< 0.001, = 4, compared to PBN 0 mg/kg-treated mice at the same time point; -fold to sham 24 h).

Neuronal and microglial distribution of C3 was reduced in the PBN-treated mice after tFCI

Up-regulation of C3 by oxidative stress was identified by QPCR and western blotting after cerebral ischemia in the mouse brains. To confirm the distribution of C3 after ischemia and reperfusion (I/R) injury, brain slices were co-stained with C3 (red) and brain cell-type markers (green). The activated neurons and microglia were changed their morphology. Neurons and microglia were activated after tFCI, and C3 was highly deposited in neurons and microglia in the ipsilateral region of the brains and accumulated in the nuclei after tFCI. This C3 expression in neurons and microglia was reduced by the antioxidant PBN (Fig. 3a). To confirm the subcellular localization of C3, nuclear extracts obtained from the ipsilateral region of the brains after tFCI were immunoblotted. In the vehicle-treated mice, tFCI resulted in significant nuclear accumulation of the C3 protein. The PBN-treated mice had reduced protein levels in C3 and in C3a and C5a, two strong anaphylatoxins, compared with the vehicle-treated mice. Nucleic C3a and C5a levels were 1.38-fold and 1.23-fold lower in the PBN-treated group than in the vehicle-treated group 3 days after tFCI (Fig. 3b). These results showed that nuclear accumulation of the C3 protein was inhibited by antioxidant PBN in neurons and microglia after cerebral I/R in mice.

Figure 3.

Distribution of C3 was suppressed by N-tert-butyl-α-phenylnitrone (PBN) in neurons and microglia after transient focal cerebral ischemia (tFCI) in mice. (a) Positive staining for C3 (red) was highly revealed in the ipsilateral region compared with the sham region after tFCI in the WT mice. PBN-treated brains showed less C3 compared with vehicle-treated brains. Co-staining with neuronal and microglial cell-type markers (green) distinctly revealed C3 in neurons and microglia after cerebral I/R injury. Neurons and microglia were also highly activated after tFCI in the vehicle group compared with the PBN group. C3 co-localization with neurons or microglia after tFCI was significantly reduced by PBN treatment. Yellow indicates co-localization of anti-C3 with anti-NeuN or anti-Iba-1 (***< 0.001, = 20, compared to vehicle-treatment group). Scale bar, 40 μm. (b) The C3 protein level in the nucleus was significantly suppressed by PBN compared with the vehicle after tFCI. C3a and C5a protein levels were markedly decreased by PBN compared with the vehicle. TFIID was used as an internal control for nucleus (*< 0.05 and ***< 0.001, = 6, comparison between vehicle– and PBN-treated mice at the same time points). TFIID indicates transcription factor II D.

PBN administration reduced neuronal damage after I/R in mice

Our data reveal that the PBN-treated mice showed decreased inflammatory C3 levels in neurons and microglia after tFCI. To clarify the functional role of PBN in cerebral ischemia, we studied whether pre- or post-injection of PBN would reduce brain infarct volume and improve neurological outcome in mice after tFCI. The PBN-treated brains showed smaller infarct volume than the vehicle-treated brains in measurement of the direct and indirect ischemic infarct volume after tFCI and 3 days of reperfusion (Fig. 4a). The mice underwent neurobehavioral testing after 30 min, 1 h, 24 h, 3 days, and 7 days reperfusion and were scored according to procedure mentioned in Materials and methods. In the sham-operated mice, no behavioral or neurological deficits appeared. PBN pre-treated mice showed significantly improved neurological outcome (Fig. 4a). PBN post-treated mice also showed reduced infarct volume (Fig. 4b). Neurological behavior was ameliorated after 24 h of reperfusion and up to 7 days in PBN post-treated mice group (Fig. 4b). Both pre- and post-treated PBN administration significantly reduced brain infarct volume and improved neurological scores after cerebral ischemic injury (Fig. 4a and b). PBN administration did not significantly change regional CBF in middle cerebral artery territory during and after tFCI compared with vehicle (0 mg/kg) administration (Fig. 4c). Our results revealed that better outcomes for infarct volume and neurological deficiency after cerebral ischemia occurred in the PBN-treated mice.

Figure 4.

N-tert-butyl-α-phenylnitrone (PBN) reduced infarct volume and improved neurological behavior after transient focal cerebral ischemia (tFCI). (a) The brains of vehicle– and PBN pre-treated mice were sliced into 1-mm coronal sections and stained with TTC followed by 4% paraformaldehyde. Infarct volume was significantly reduced by treatment with PBN compared with the vehicle (***< 0.001, = 12, compared with vehicle-treated mice). Using circling and forelimb tests, the PBN pre-treated mice showed significant neurological improvement compared with the vehicle-treated mice after tFCI up to 7 days of reperfusion. The highest score represents the worse neurological outcome (*< 0.05, **< 0.01 and ***< 0.001, = 12, comparison between vehicle-treated mice and PBN-treated mice). Veh indicates vehicle. (b) PBN post-treated mice showed significantly reduced infarct volume and ameliorated neurological behavior (**< 0.01 and ***< 0.001, = 13, compared with vehicle-treated mice). (c) Regional CBF in percent of baseline was assessed by laser Doppler flowmetry before, during, and after tFCI. There were no significant regional CBF differences among vehicle-, PBN 40 mg/kg-, and PBN 60 mg/kg-administrated group (= 3, compared to PBN 0 mg/kg-treated mice).

Down-regulation of the C3 gene improved cell viability and reduced intracellular ROS level after OGD

To clarify whether activation of C3 by ROS is involved in cell death after ischemic injury, mouse primary cortical neurons were pre-treated with 25 nM of C3 siRNA and subjected to 3 h of OGD, which increased cytotoxicity in scrambled siRNA-treated primary neurons (Fig. 5a). We measured intracellular ROS level by using DCFH-DA after 3 h OGD. The intracellular ROS level was increased after OGD followed by 30 min of reperfusion. C3 siRNA-transfection showed significantly decreased ROS level compared to scrambled siRNA-transfection (Fig. 5b). Increased COX-2 and iNOS protein levels after 3 h of OGD were also decreased by C3 siRNA-transfection in primary neurons (Fig. 5c). This implies that C3 plays a deleterious role via oxidative stress after OGD in mouse primary cortical neurons.

Figure 5.

Inhibition of C3 increased cell viability and reduces intracellular reactive oxygen species (ROS) levels after oxygen-glucose deprivation (OGD). (a) Mouse primary cortical neurons were transfected with C3 siRNA and subjected to 3 h of OGD, followed by 6 and 24 h of reperfusion. Cell cytotoxicity was assessed by lactate dehydrogenase (LDH) assay at 490 nm. C3 knock-down neurons showed decreased cell cytotoxicity after OGD (***< 0.001, = 4, comparison between scrambled siRNA and C3 siRNA; -fold to normoxia scrambled). (b) Intracellular ROS levels were measured by DCFH-DA staining. C3 knock-down neurons showed decreased intracellular ROS levels after 3 h of OGD (***< 0.001, = 4, comparison between scrambled siRNA and C3 siRNA; -fold to normoxia scrambled). (c) C3 knock-down neurons also showed down-regulated cyclooxygenase-2 (COX-2) and inducible nitric oxide synthase (iNOS) protein levels (**< 0.01 and ***< 0.001, = 4, comparison between scrambled siRNA and C3 siRNA; -fold to normoxia scrambled). Nor scr indicates normoxia scrambled.

C3 promoter activity was increased after OGD in mouse primary cortical neurons

Our results show that C3 level was increased after tFCI and up-regulated C3 level was related to higher oxidative stress. To determine whether C3 activation is directly related to oxidative stress, we transiently transfected C3-luc plasmid, which contains a 1.1-kb fragment of the C3 promoter, to primary cortical neurons (Fig. 6a). Twenty-four hours after transfection, 3 h of OGD was induced and C3 promoter activity was measured after 6 h, 24 h, 48 h, and 72 h of reperfusion. The promoter was highly activated after 3 h OGD up to 72 h of reperfusion (Fig. 6b). These results showed that ROS generated by OGD directly activated transcription of the C3 promoter, not only at an early time point but also at a late time point up to 72 h after reperfusion in mouse primary cortical neurons.

Figure 6.

C3 was transcriptionally increased after oxygen-glucose deprivation (OGD) primary cortical neurons. (a) Eleven hundred base pairs from the C3 promoter region was inserted into backbone vector (Fan et al.1996). (b) Mouse primary cortical neurons were transfected with the C3-luc plasmid and were induced 3 h OGD followed by 6 h, 24 h, 48 h, and 72 h of reperfusion. The luciferase activity was quantified. C3 promoter was significantly activated after 3 h OGD up to 72 h reperfusion in primary neurons (***< 0.001, = 6, compared to normoxia; -fold to normoxia). Nor indicates normoxia.


ROS are a fundamental mechanism of brain damage and the excessive production of ROS is associated with oxidative stress (Allen and Bayraktutan 2009). Growing evidence suggests that stroke, the third leading cause of death in the world, is related to immunological inflammation. The complement system is an integral part of the immune defense mechanism and is also a primary mediator of the inflammatory process (Frank and Fries 1991). The complement cascade is activated in the brain after cerebral I/R, and there is significant evidence that complement components are deposited after cerebral ischemic injury (Cowell et al. 2003). Previous studies have shown that complement levels were increased after ischemic stroke and that complement pathway inhibition by C1q and C3 inhibition and mannose-binding lectin deficiency lead to a better neurological outcome (Cojocaru et al. 2008; Huang et al. 2008; Cervera et al. 2010). A recent study identified C3 as the most influential mediator of ischemic injury in the complement cascade (Mocco et al. 2006). C3 activation was increased at 72 h, and complement depletion significantly reduced brain edema 72 h after intracerebral hemorrhage (Vakeva et al. 1994; Xi et al. 2001). Consistent with earlier reports, we found that C3 levels were increased after tFCI up to 7 days in mice. However, conflicting views have emerged about the role of complement components. Complement has a role in CNS regeneration, and C3a and C5a are crucial for hepatocyte proliferation and liver regeneration in mammalian systems (Mastellos et al. 2001; Strey et al. 2003; Daveau et al. 2004; Rahpeymai et al. 2006). However, excessive complement activation leads to severe damage after stroke, and reoxygenation of anoxic endothelial cells has also been shown to increase complement activation and deposition (Väkevä and Meri 1998). This implies that complement activation has both protective and deleterious effects, so it should be modulated rather than blunted. In this study, we used SOD2 KO and SOD1 Tg mice to find the close relationship between C3 activation and excessive superoxide radicals. Our results showed that SOD2 KO mice with higher ROS levels showed higher C3 level and SOD1 Tg mice brains had decreased C3 level after I/R. The natural free-radical scavenger, SOD, reduced ischemic brain injury and simultaneously reduced C3 level. These data imply that complement activation after cerebral I/R is directly related to oxidative stress.

Previous studies have shown the relationship between the complement cascade and antioxidants in heart I/R models. The mechanisms of cardioprotection after ischemia appear to be antioxidant activity and direct scavenging of superoxide anions, as well as a reduction in the levels of the C-reactive protein and membrane attack complex in infarcted tissue (Lockwood and Gross 2005). In a gastrointestinal I/R model, anti-complement treatment preserved SOD1 activity and decreased oxidative stress (Montalto et al. 2003). However, how complement components and antioxidants are related in brain I/R injury has not yet been discovered. Thus, we studied the effect of an antioxidant along with the complement cascade activation in a brain ischemia model to investigate whether oxidative stress is directly related to complement component levels after tFCI.

Over the last few years, free-radical scavengers, such as ebselen and the radical-trapping agent NXY-059, have gained considerable attention in the field of stroke research (Allen and Bayraktutan 2009). However, in a clinical trial, the use of ebselen had a disappointing outcome, compared with a placebo, at 3 months. Another phase III study was slated to begin in 2001, but no report has appeared (Yamaguchi et al. 1998; Ginsberg 2008). NXY-059 scores mirrored those of the placebo group (Ginsberg 2008). There have been no commonly used antioxidant drugs for stroke treatment yet. In this study, we used the antioxidant PBN to examine whether it can reduce C3 activation for neuroprotection after cerebral ischemia. PBN is a spin-trapping agent that shows neuroprotection against brain injury and it was previously reported that PBN did not change CBF as we confirmed (Miyajima and Kotake 1995; Liu et al. 2003). Earlier studies revealed that PBN shows neuroprotective effects by improving mitochondrial function and recovery of the cerebral energy state after tFCI, and by attenuating lactate formation, in addition to free-radical scavenging (Folbergrová et al. 1995; Kuroda et al. 1996; Lewén and Hillered 1998). PBN also activates extracellular signal-regulated kinase, suppresses stress-activated protein kinase/c-Jun N-terminal kinase and p38 activation, and increases expression of heat-shock proteins 27 and 70 (Tsuji et al. 2000). The anti-inflammatory mechanism of PBN in the immune system after cerebral ischemia has not been examined nor extensively researched. Our results showed that PBN reduced C3 mRNA level which was increased by tFCI, and nuclear accumulation of C3-related proteins, which suggests that PBN regulates the C3 level during cerebral I/R in mice. The neurological deficit gradually recovered as previously reported, but PBN significantly improved behavioral recovery compared to vehicle (Yang et al. 2009). PBN also reduced infarct volume.

Our study also found that the anaphylatoxins C3a and C5a were increased after tFCI, but were reduced by PBN administration. These results correspond with previously reported studies that C3a and C5a also play a deleterious role in ischemic stroke (Ducruet et al. 2012; Pavlovski et al. 2012). Many biological activities are associated with complement activation, including formation of anaphylatoxins such as C3a and C5a (Chakraborti et al. 2000). Other downstream factors of C3, such as C3b, C3d, C5 and membrane attack complex, also induce neuronal death after ischemic stroke (Ducruet et al. 2012; Pavlovski et al. 2012; Arumugam et al. 2007). Administration of C3d reduces primary neural progenitor cells proliferation, and antagonism of the C3a receptor promotes proliferation of migrating neuroblasts after stroke (Ducruet et al. 2012). Endogenously generated C5a also exacerbated neuronal apoptosis under ischemic condition (Pavlovski et al. 2012).

It has been shown that brain cells themselves were able to produce complement components (Lévi-Strauss and Mallat 1987). Among the numerous brain cells, microglia and neurons produce almost all complement proteins (Morgan and Gasque 1996; Gasque et al. 2000). Our double immunofluorescent staining results showed that neurons and microglia changed their morphology to activated form after tFCI and that C3 was co-localized in them. Previous studies reported that complement component binds to nucleic acid in lupus and inner-nuclear layers of retina (Papp et al. 2010; Amadi-Obi et al. 2012). On the basis of these, we hypothesize that normally, C3 is mainly located in the cytosol and is translocated to the nucleus when the complement cascade is activated by oxidative stress and has a deleterious role. However, further study is needed to figure out the specific role and mechanism of activated C3 in nuclear. In mouse brain primary cortical neurons, inhibition of C3 by treatment with C3 siRNA decreased cell cytotoxicity after OGD. Furthermore, intracellular ROS levels, which were highly up-regulated by OGD, were significantly decreased by C3 siRNA-transfection. These results indicate that excessive C3 produced by ROS has a deleterious role in neurons after I/R injury in mice.

Despite these findings, the mechanisms governing the relationship between oxidative stress and C3 activation have been unknown. To further investigate the redox-sensitive mechanisms of C3 activation, promoter research is much needed. The C3 promoter has many oxidative stress– and inflammation-related factors such as activator protein-1, interleukin-1β, p38, and CCAAT/enhancer-binding protein delta (C/EBPδ) (Juan et al. 1993; Schaefer et al. 2005; Maranto et al. 2008). Interleukin-1β regulates the C3 gene in astrocytes and C/EBPδ is a pivotal transcription factor involved in brain inflammation (Cardinaux et al. 2000; Maranto et al. 2008). Also, activator protein-1 plays a role in C3a receptor expression (Schaefer et al. 2005). In this study, we transiently transfected C3 plasmid which contains C3 promoter region to mouse primary cortical neurons, subjected 3 h of OGD and reperfused 6 h, 24 h, 48 h, and 72 h. The promoter was highly activated after OGD up to 72 h. It has been known that oxidative stress peaks at a much earlier time point, but an inflammatory reaction occurs because of secondary ROS (Spychalowicz et al. 2012; Yang et al. 2012). Thus, C3 might elevate because of inflammation after stroke. As C3 promoter is activated, interleukin-1 and interleukin-6 levels are also elevated and these lead to secondary ROS production (Won and Baumann 1990; Vik et al. 1991; Jendrysik et al. 2011). Hence, our results suggest that down-regulation of C3 also lowers ROS level. To clarify the relationship between C3 and oxidative stress, further research of C3 promoter and its transcriptional factors are needed.

As the peripheral and the brain immune system are not separated, but connected; the immune response after tFCI is systemic including the brain (Yong and Rivest 2009; Crehan et al. 2012). Thus, cerebral C3 might come from both CNS cells and peripheral cells. Another possibility that brain and peripheral system are separated and activates C3 independently has been suggested (van Beek et al. 2003). The specific mechanism between the CNS and the peripheral system needs to be further studied. Growing evidence suggests that inflammatory markers can predict stroke since inflammation plays an important role in pathophysiology of brain ischemia (Vibo et al. 2007; Whiteley et al. 2012). Developing blood markers to diagnose and prevent stroke has been done in many research groups. It has been previously reported that glial fibrillary acidic protein, tau, lipoprotein-associated phospholipase A2, and granulocyte colony-stimulating factor levels in blood were increased in stroke patients (Nylen et al. 2007; Hu et al. 2012; Schiff et al. 2012; Tsai et al. 2012; Yu et al. 2012). Also, plasma C3 levels were elevated in cryptogenic and large-vessel disease stroke, and high plasma C3 levels were related with unfavorable outcomes only in large-vessel disease stroke (Stokowska et al. 2011). Our results showed that C3 plasma levels were increased after tFCI and were down-regulated by antioxidant PBN administration. C3 plasma levels also showed co-relation with behavioral change. Thus, plasma C3 levels measurement has possibility to diagnose and prevent stroke in a clinical setting.

The complement cascade is activated after cerebral ischemia and plays a deleterious role, and oxidative stress is closely associated with C3 activation. Mice with lower ROS levels showed a better neurological outcome with lower C3 levels after tFCI. In this study, we identified not only a causal relationship between ROS and C3 activation in stroke but also a new anti-inflammatory effect of PBN through suppression of C3 activation in mice. An antioxidant that is able to inhibit complement C3 activation may offer the new strategy for therapeutic approaches in stroke patients.


The authors declare no conflict of interest. This study was supported by National Research Foundation of Korea grant #KRF-2008-313-E00359 and SRC program #20090063046 to Y.S.S. and by grants PO1 NS014543, RO1 NS025372, and RO1 NS038653, from the National Institutes of Health, and by the James R. Doty Endowment to P.H.C. The authors thank Cheryl Christensen for her editorial assistance and Elizabeth Hoyte for assistance with the figures.