Neuroprotection against focal ischemic brain injury by the peroxisome proliferator-activated receptor-γ agonist rosiglitazone

Authors

  • Yumin Luo,

    1. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    2. Institute of Neurology, Fudan University Huashan Medical Center, Shanghai, China
    Search for more papers by this author
  • Wei Yin,

    1. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Armando P. Signore,

    1. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    2. Center for Neuroscience, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Feng Zhang,

    1. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Zhen Hong,

    1. Institute of Neurology, Fudan University Huashan Medical Center, Shanghai, China
    Search for more papers by this author
  • Suping Wang,

    1. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Steven H. Graham,

    1. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    2. Department of Pharmacology and
    3. Center for Neuroscience, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    4. Geriatric Research, Educational and Clinical Center, Veterans Affairs Pittsburgh Health Care System, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author
  • Jun Chen

    1. Department of Neurology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    2. Department of Pharmacology and
    3. Center for Neuroscience, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania, USA
    4. Geriatric Research, Educational and Clinical Center, Veterans Affairs Pittsburgh Health Care System, Pittsburgh, Pennsylvania, USA
    Search for more papers by this author

Address correspondence and reprint requests to Dr Jun Chen, Department of Neurology, S507, Biomedical Science Tower, University of Pittsburgh School of Medicine, Pittsburgh, PA 15213, USA.
E-mail: chenj2@upmc.edu

Abstract

Peroxisome proliferator-activated receptor gamma (PPAR-γ) is a nuclear membrane-associated transcription factor that governs the expression of various inflammatory genes. PPAR-γ agonists protect peripheral organs from ischemic injury. In the present study, we investigated whether the PPAR-γ agonist rosiglitazone is neuroprotective against focal ischemic brain injury. C57/B6 mice underwent 1.5-h middle cerebral artery occlusion, and received either vehicle or rosiglitazone treatment of 0.75, 1.5, 3, 6 or 12 mg/kg (n = 9 per group). Cerebral infarct volume, neurological function, expression of pro-inflammatory proteins and neutrophil accumulation were assessed after ischemia and reperfusion. At 48 h after ischemia, infarct volume was significantly decreased with 3–12 mg/kg of rosiglitazone, with a time window of efficacy of 2 h after ischemia at the optimal dose (6 mg/kg). Neutrophil accumulation was significantly decreased in the brain parenchyma of rosiglitazone-treated mice. Ischemia-induced expression of several inflammatory cytokines and chemokines was markedly reduced in rosiglitazone-treated brains, as determined using proteomic-array analysis. Rosiglitazone treatment improved neurological function at 7 days after ischemia. Moreover, in cultured cortical primary microglia, rosiglitazone attenuated inflammatory responses by decreasing lipopolysaccharide-induced release of tumor necrosis factor-α, interleukin (IL)-1β and IL-6. These results suggest that the PPAR-γ agonist rosiglitazone has neuroprotective properties that are at least partially mediated via anti-inflammatory actions, and is thus a potential novel therapeutic agent for stroke.

Abbreviations used
CBF

cerebral flood flow

15D-PGJ2

15-deoxy-d-prostaglandin (12,14)-prostaglandin J2

EMSA

electrophoretic mobility shift assay

GCSF

granulocyte colony-stimulating factor

ICAM-1

intercellular adhesion molecule-1

IL

interleukin

iNOS

inducible nitric oxide synthase

KC

keratinocyte chemoattractant

LPS

lipopolysaccharide

MCA

middle cerebral artery

MCAO

middle cerebral artery occlusion

MCP-1

monocyte chemotactic protein 1

MPO

myeloperoxidase

PBS

phosphate-buffered saline

PPAR-γ

peroxisome proliferator-activated receptor-γ

sTNFR1

soluble tumor necrosis factor receptor 1

TIMP-1

metalloproteinase-1

TNF-α

tumor necrosis factor α

Acute inflammation contributes considerably to the pathogenesis of ischemic brain injury (Barone and Feuerstein 1999; del Zoppo 1999). Inflammatory responses to cerebral ischemia involve a sequential series of processes, including the release of pro-inflammatory cytokines initiated through Toll-like receptors (for review, see Kariko et al. 2004), increased expression of endothelial adhesion molecules and chemotactic factors (Barone and Feuerstein 1999; Carmel et al. 2001; Iadecola and Alexander 2001), activation of microglia and macrophages (Mabuchi et al. 2000), and leukocyte infiltration (Barone and Feuerstein 1999; del Zoppo et al. 2000, 2001). The synergistic actions of inflammatory responses exacerbate brain injury, leading to secondary expansion of ischemic infarction and deterioration of neurological outcomes (Barone and Feuerstein 1999; Dirnagl et al. 1999). Moreover, inflammation associated with post-ischemic reperfusion may limit the efficacy of thrombolytic therapy in acute stroke (Jean et al. 1998). Therefore, interventions aimed at suppressing post-ischemic cerebral inflammatory reactions are an attractive and emerging therapeutic strategy for ischemic stroke (Mori et al. 1992; Chen et al. 1994; Zhang et al. 1994; Chopp et al. 1996; Connolly et al. 1996; Zhang et al. 1996; Goussev et al. 1998; Yenari et al. 1998; Soriano et al. 1999; Suzuki et al. 1999).

Peroxisome proliferator-activated receptor gamma (PPAR-γ) is a nuclear membrane-associated transcription factor. Upon ligand-specific activation, PPAR-γ negatively regulates the expression of various pro-inflammatory genes (Jiang et al. 1998; Ricote et al. 1998). A member of the nuclear hormone receptor superfamily, PPAR-γ was originally reported to be highly expressed in adipocytes and to play an important role in their differentiation, as well as in lipid biosynthesis and glucose homeostasis (Tontonoz et al. 1994; Lehmann et al. 1995). In addition, evidence for an important role in immunomodulation is also emerging. PPAR-γ is activated by its natural ligand 15-deoxy-d-prostaglandin (12,14)-prostaglandin J2 (15D-PGJ2) and thiazolidinediones, a class of synthetic chemical ligands. Thiazolidinediones are used clinically as insulin sensitizers to treat type 2 diabetes. However, it is known that various thiazolidinediones provide a potent protective effect against ischemic/reperfusion injury of the heart and other peripheral organs by attenuating the inflammatory response (Nakajima et al. 2001; Yue et al. 2001; Okada et al. 2002; Wayman et al. 2002). Recent studies indicate that PPAR-γ is also expressed at appreciable levels in the CNS, especially in microglia and astrocytes, important mediators of the inflammatory response in the brain (Bernardo et al. 2000; Cristiano et al. 2001). Indeed, thiazolidinediones have shown promising beneficial effects in several cellular and animal models of CNS disorders where an inflammatory component is strongly implicated, such as experimental autoimmune encephalomyelitis (Niino et al. 2001; Diab et al. 2002; Feinstein et al. 2002), Parkinson's disease (Dehmer et al. 2004), amyotrophic lateral sclerosis (Kiaei et al. 2005), and Alzheimer's disease (Landreth and Heneka 2001).

The objective of the present study was to evaluate the effect of systemic administration of rosiglitazone, the most potent thiazolidinedione for PPAR-γ activation, in a murine model of focal cerebral ischemia and reperfusion. We also investigated the effect of rosiglitazone treatment on post-ischemic inflammatory responses in the brain. Our results suggest that rosiglitazone markedly decreased brain injury after focal ischemia and reperfusion, and this neuroprotective effect of rosiglitazone is attributed at least in part to its anti-inflammatory actions.

Materials and methods

Murine model of transient focal ischemia

Animal surgery

All animal experiments were approved by the University of Pittsburgh Institutional Animal Care and Use Committee and carried out in accordance with the NIH Guide for the Care and Use of Laboratory Animals. Focal cerebral ischemia was produced by intraluminal occlusion of the left middle cerebral artery (MCA) with a nylon monofilament suture as originally described (Yang et al. 1994) with slight modifications (Cao et al. 2002). Male 2- to 3-month-old C57/B6 mice (25–30 g each, Jackson Immuno-Research, The Jackson Laboratory, Bar Harbor, ME, USA) were anesthetized with 1.5% isoflurane in a 30% O2/68.5% N2O mixture under spontaneous breathing. Rectal temperature was controlled at ∼37.0°C during and after surgery via a temperature-regulated heating pad. Systolic arterial blood pressure was monitored through a tail cuff (XBP1000 Systems, Kent Scientific Corporation, Torrington, CT, USA), and arterial blood gas was analyzed at 15 min before and 30 min after the onset of ischemia. The animals underwent left MCA occlusion for 90 min and then reperfusion for the indicated duration. Changes in regional cerebral blood flow (rCBF) before, during and after MCA occlusion were evaluated in animals using laser Doppler flowmetry (Cao et al. 2002). After recovering from anesthesia, the animals were maintained in an air-conditioned room at 20°C.

Drug administration

Rosiglitazone (Cayman Chemical, Ann Arbor, MI, USA) was dissolved in dimethylsulfoxide and then further diluted in physiological saline (1 : 3 ratio), and injected intraperitoneally at 1 h before MCA occlusion. For long-term survival studies (48 h to 7 days), the animals received a second injection of rosiglitazone at 24 h after MCA occlusion. The animals that received administration of rosiglitazone or vehicle (dimethylsulfoxide) were assigned randomly to experimental groups consisting of nine animals each. To determine the time window of efficacy of rosiglitazone in the MCA occlusion model, the optimal dose of the compound was administered as described above, beginning at 0, 0.5, 1, 2, 3 or 6 h after the onset of reperfusion, and all animals received a second dose at 24 h after ischemia.

Behavioral tests

Two types of behavioral functional tests were performed in animals by an observer who was blinded to the experimental conditions. Neurological deficits were scored on a 0–5 scale (Murakami et al. 1998): no neurological deficit (0); failure to extend the right forepaw fully (1); circling to the right (2); falling to the right (3); unable to walk spontaneously (4); dead (5). The second test, the corner test, which is a sensorimotor functional assessment, was performed in ischemic animals as described previously (Zhang et al. 2002; Gao et al. 2005). In this test, the ischemic mouse turns preferentially toward the non-impaired (left) side. The turns in one direction versus the other were recorded from 10 trials for each test.

Measurement of infarct volume

At 48 h after MCA occlusion, brains were removed and the forebrain was sliced into seven coronal sections 1-mm thick. Sections were stained with 3% 2, 3, 5-triphenyltetrazolium in saline for 20 min, and then fixed in 4% paraformaldehyde in phosphate-buffered saline (PBS; pH 7.4). Infarct volume was determined using the MCID (Imaging Research Inc., St. Catherine's, Ontario, Canada) image analysis as previously described (Cao et al. 2002).

Immunohistochemistry

Animals were killed in a carbon dioxide chamber at the indicated time points after ischemia, and the brains were prepared for freshly frozen coronal sectioning (15 μm of thickness) as described previously (Chen et al. 1997). After air-drying, the sections were fixed with 4% paraformaldehyde in PBS, pH 7.4, for 10 min at 25°C and then permeabilized with 1% Triton X-100 in PBS for 30 min. Endogenous peroxidases were inhibited by incubating sections for 20 min in 3% H2O2 and 30% methanol in PBS, pH 7.4, and non-specific antigen binding sites were blocked with 2% bovine serum albumin in PBS, pH 7.4, for 20 min. Sections were then incubated overnight at 4°C with the goat polyclonal anti intercellular adhesion molecule-1 (ICAM-1) antibody at a dilution of 1 : 400 (Serotec INC, Raleigh, NC, USA), the goat polyclonal anti-myeloperoxidase (MPO) antibody at a dilution of 1 : 100 (Santa Cruz Biotechnology, Santa Cruz, CA, USA), or biotinylated tomato lectin (for the detection of activated microglia) at a dilution of 1 : 100 (Vector Laboratories, Burlingame, CA, USA). Sections were washed in PBS three times and then incubated with biotinylated horse anti-mouse IgG (for ICAM-1 experiments) or biotinylated donkey anti-goat IgG (for MPO experiments) secondary antibody at a dilution of 1 : 1000 (Santa Cruz Biotechnology). After extensive washes, the secondary antibodies or biotinylated tomato lectin were detected using the Elite Vectastain ABC kit and the diaminobenzidine substrate kit (Vector Laboratories).

Myeloperoxidase activity measurement

MPO activity, an indicator of polymorphonuclear cell accumulation, was determined in brain tissues as described previously (Yang et al. 1998). Animals were killed in a carbon dioxide chamber and then perfused transcardially with ice-cold saline to flush all blood components from the vasculature. The brains were removed, dissected to separate the forebrain hemispheres, and immediately frozen at −80°C. Tissues were thawed on ice, weighed, and homogenized at 4°C in 1 mL of 50 mmol/L Tris-HCl, pH 7.4. Samples were added to 5 mL of 5-mmol/L phosphate buffer, pH 6.0, homogenized, and centrifuged at 30 000 g for 30 min at 4°C. The supernatant was discarded, and the pellet was re-suspended in 0.5% hexadecyltrimethylammonium bromide (Sigma, St Louis, MO, USA) in 50 mmol/L potassium phosphate buffer at 25°C for 2 min (wet tissue weight/volume ratio 1 : 2). The samples were frozen on dry ice. Freezing and thawing were then performed three times with 10-s sonications between cycles. After the last sonication, the samples were incubated at 4°C for 15 min and centrifuged at 12 500 g for 15 min. The supernatant (100 μL) was mixed with 100 μL of 50 mmol/L phosphate buffer, pH 6.0, which contained 0.167 mg/mL o-dianisidine dihydrochloride (Sigma) and hydrogen peroxide (0.005%). The rate of change in absorbance was measured spectrophotometrically at 460 nm. MPO activity was expressed in milliunits per gram weight of wet tissue (one milliunit was defined as the activity of enzyme degrading 1 pmol of peroxide/min at 25°C).

Electrophoretic mobility shift assay

One hour after injection of rosiglitazone at the indicated doses, the mice were killed, and their brains were rapidly removed, dissected, and frozen in liquid nitrogen. Nuclear proteins were extracted from brain tissues using our previously published protocol (Lan et al. 2003). Protein content in nuclear fractions was measured using the Bradford method (Bio-Rad Bulletin 1177, Bio-Rad Laboratories, Hercules, CA, USA). Electrophoretic mobility shift assay (EMSA) was performed to determine PPAR-γ-specific DNA binding activity in nuclear extracts as described previously (Fuenzalida et al. 2005) with slight modifications. In brief, a 25-bp oligonucleotide was synthesized (Gibco-BRL, Carlsbad, CA, USA) according to the sequence of consensus binding site for the PPAR response element (5′GGAACTAGGTCAAAGGTCATCCCCT3′) and then end-labeled with γ-32P–ATP using T4 polynucleotide kinase. The DNA–protein binding reaction was performed in a total volume of 30 μL containing the binding assay buffer (10 mm Tris-HCl, 50 mm NaCl, 0.5 mm DTT, 0.5 mm EDTA, 1 mm MgCl2, 5% glycerol, pH 7.6), 0.0175 pmol of labeled double-stranded probe (> 10 000 cpm), 30 μg of nuclear protein, and 1 μg of poly(dI-dC). After incubation for 20 min at room temperature, the reaction mixture was subjected to electrophoresis for 2 h on a non-denaturing 4% polyacrylamide gel in 0.25 × TBE (22.5 mm Tris borate and 0.5 mm EDTA). The gel was dried and subjected to autoradiography. The specificity of PPAR-γ binding activity was determined by performing both competition assays and supershift assays. For the competition assays, 10- or 100-fold molar excess of unlabeled double-stranded oligonucleotide of the PPRE binding site was used as a specific competitor. For supershift assays, 2.5 or 5 μg of antibody against PPAR-γ (Affinity BioReagents, Golden, CO, USA) was added 30 min before the addition of the oligonucleotide probe and incubated at room temperature.

Proteomic-array analysis

Proteomic-array analysis was conducted using the ChemiArrayTM Mouse Antibody Arrays kit (Chemicon, Temecula, CA, USA) according to the manufacturer's instructions. Compared with ELISA, this array system allows simultaneous detection of 32 cytokines at once, with greater sensitivity and detection range, and fewer variations (Lin et al. 2003). Briefly, each antibody-array membrane was placed in the provided tray. Blocking buffer was added to the membrane and incubated at room temperature for 30 min. The membrane was then incubated with brain protein samples at room temperature for 1.5 h. After washing, the membrane was incubated with diluted biotin-conjugated anti-cytokines primary antibodies at room temperature for 1.5 h. The membrane was washed and then incubated with HRP-conjugated streptavidin at room temperature for 2 h. After a final wash, the membrane signals were detected directly using a chemiluminescence imaging system. The signal intensities were quantified by densitometry. In each array assay, the validity of the experiment was confirmed by the positive staining of the three pairs of positive controls as well as the lack of staining in the two pairs of negative controls included in the array membranes.

Microglial cultures and ELISA for cytokines

Microglia were isolated and purified from Sprague–Dawley rat brains at post-natal day 1 as described previously (Li et al. 2004). Briefly, cerebral cortical tissues were minced mechanically and subsequently digested with trypsin (0.25%) and Dnase I (0.01%). Dissociated cells were re-suspended in Dulbecco's modified Eagle's medium that was supplemented with 10% fetal bovine serum, and then seeded in 75-cm2 flasks at a density of 5 × 106 cells per flask. The cells were maintained in an incubator with 5% CO2 at 37°C. The medium was changed every 3 days. Two weeks after seeding, the flasks were shaken at 180 r.p.m. in an orbit shaker at 37°C for 5 h, and the floating cells were collected and transferred to a new flask to allow adherence for 1 h. The attached cells were collected and plated in 24-well plates coated with poly-l-lysine. The cells were maintained in Dulbecco's modified Eagle's medium supplemented with 2% fetal bovine serum for 24 h. Microglial purity was determined using DiI-Ac-LDL labeling (Li et al. 2004), a cell marker for microglia, which showed > 98% purity in cultures.

To activate the microglia, 500 ng/mL lipopolysaccharide (LPS; Escherichia coli 055 : B5; Sigma) was added to the cultures. In some wells, rosiglitazone (10 μm) was added 30 min before LPS treatment to evaluate its effect on microglial activation. After 4 or 16 h of LPS incubation, culture media were centrifuged to remove the detached cells, and the supernatants were collected and stored at −80°C until assayed for tumor necrosis factor (TNF)-α, interleukin (IL)-1β, and IL-6 secretion. Levels of TNF-α, IL-1β, and IL-6 secreted by microglia in the media were determined using an ELISA assay kit (BD Biosciences, Franklin Lakes, NJ, USA) according to the manufacturer's instructions. To determine if the effect of rosiglitazone on cytokine release is dependent on PPAR-γ activation, in some experiments the PPAR-γ antagonist GW9662 was added (10 μm) to the cultures 30 min before rosiglitazone incubation.

Statistical analysis

Results are reported as mean ± SEM. The difference between means was assessed by the Student's t-test (single comparisons) or by anova and post-hoc Bonferroni/Dunn tests (for multiple comparisons), with p < 0.05 considered statistically significant.

Results

Rosiglitazone reduces infarct volume in mice following focal ischemia/reperfusion

The first objective of this study was to determine whether the PPAR-γ agonist rosiglitazone mitigates ischemic brain damage. Infarct volume was measured based on the regional loss of 2,3,5-triphenyltetrazolium chloride (TTC)-staining in brain sections 48 h post-reperfusion following 90 min of ischemia. MCA occlusion produced ipsilateral cerebral infarcts averaging ∼58 mm3 in volume. Administration of rosiglitazone 1 h prior to MCA occlusion, and a second dose at 24 h after ischemia reduced the infarct volume in a dose-dependent manner (Fig. 1a). A significant reduction averaging 45 and 64% was achieved by two doses of 3 and 6 mg/kg (p < 0.05 vs. vehicle controls), respectively. An additional increase in the dose to 12 mg/kg did not result in an additional decrease in infarct volume. Reduction of infarct size by rosiglitazone was not accompanied by significant alterations in arterial blood pressure and blood gases compared with animals that received vehicle alone (Table 1). To rule out the possibility that rosiglitazone might alter cerebral blood flow (CBF), laser Doppler flowmetry was used to monitor cortical CBF in both vehicle- and rosiglitazone-treated (6 mg/kg) animals. No difference in CBF was found between animals that received rosiglitazone or vehicle during or after MCA occlusion (Fig. 1b). Moreover, as determined in six normal mice, administration of rosiglitazone did not significantly alter serum glucose levels (89.2 ± 3.6 mg/dL before injection vs. 87.7 ± 3.8 mg/dL 30 min after injection, p > 0.05).

Figure 1.

Rosiglitazone treatment decreases infarct volume after focal ischemia/reperfusion. (a) Dose-dependent protection by rosiglitazone treatment. Rosiglitazone or vehicle was infused 1 h prior to and 24 h after 1.5-h MCA occlusion at the indicated doses, and the infarct volume was measured at 48 h after ischemia. Data are means ± SEM, n = 9 per group, *p < 0.05; **p < 0.01 relative to controls (analysis of variance and post-hoc Bonferroni/Dunn tests). (b) Changes in cortical blood flow, as determined using laser Doppler flowmetry at 15, 30, 45, 60, 75, 95 and 110 min after MCA occlusion, are not different between rosiglitazone-treated (6 mg/kg) and vehicle-treated brains (n = 7 per group) during or after MCA occlusion. (c) Time window of efficacy. Therapeutic window determination was made by infusing rosiglitazone (6 mg/kg) at 0.5, 1, 2, 3, or 6 h after the completion of MCA occlusion and then 24 h after the first infusion (n = 9 per group). The infarct volume was measured at 48 h after ischemia.

Table 1.  Physiological variables
  Body temperatures (°C)pH PaO2 (mmHg)PaCO2 (mmHg)Systolic arterial blood pressure (mmHg)
  1. All data are means ± SEM from 12 rosiglitazone-treated mice and nine vehicle-treated mice, respectively. There were no statistically significant differences in any of the parameters between the vehicle and rosiglitazone groups at a specific period (before ischemia or during ischemia) or within a group at different periods (before ischemia vs. during ischemia). Pre-ischemic, 15 min before ischemia; Ischemic, during ischemia (30 min).

VehiclePre-ischemic37.1 ± 0.27.36 ± 0.04127.6 ± 8.640.7 ± 1.796.7 ± 4.5
Ischemic36.9 ± 0.17.33 ± 0.03123.3 ± 7.739.3 ± 2.193.4 ± 5.2
RosiglitazonePre-ischemic36.8 ± 0.17.37 ± 0.03131.4 ± 1.939.8 ± 1.994.2 ± 4.1
Ischemic36.9 ± 0.17.38 ± 0.05130.2 ± 8.441.5 ± 2.192.7 ± 4.7

To determine whether delaying administration of rosiglitazone could also decrease infarct volume, injection of rosiglitazone was initiated at various time points after the completion of MCA occlusion. Delayed treatment with rosiglitazone continued to significantly reduce infarct volume when the treatment was initiated at 0.5, 1, or 2 h after ischemia, but the effect was lost when the administration was delayed by 3 or 6 h (Fig. 1c). Nevertheless, as compared with the pretreatment regimen, delayed administration of rosiglitazone was less effective in reducing infarct volume.

To determine whether rosiglitazone treatment indeed spared brain tissue, rather than delayed tissue loss, infarct volume was measured at 7 days after ischemia in two additional experimental groups (n = 9/group). The rosiglitazone-treated [6 mg/kg, 1 h before and 24 h after middle cerebral artery occlusion (MCAO)] group showed significantly reduced (41%) infarct volume compared with the vehicle group (Fig. 2a). Behavioral functional tests, including neurological deficit scores and corner tests (sensorimotor functions), were also conducted in these two groups. The results revealed that the animals treated with rosiglitazone showed significantly improved functional outcomes compared with the vehicle-treated animals (Figs 2b and c).

Figure 2.

Rosiglitazone treatment improves functional outcomes after focal ischemia/reperfusion. (a) The long-term effect of rosiglitazone on infarct volume. Rosiglitazone (6 mg/kg) or vehicle was infused 1 h prior to and then 24 h after 1.5-h MCA occlusion, and the infarct volume was measured at 7 days after ischemia. Data are means ± SEM, n = 9 per group, *p < 0.05 relative to vehicle-treated group (two-tail Student's t-test). (b, c) Behavioral functional tests in rosiglitazone-treated and vehicle-treated mice after MCA occlusion (n = 9 per group). The animals were subjected to neurological deficit scores (b) 7 days after MCA occlusion and corner tests (c) 7 days after MCA occlusion or sham operation, and then killed for infarct volume measurement. Data are means ± SEM. (b) *p < 0.05 relative to vehicle-treated group; (c) *p < 0.05 relative to vehicle-treated sham group, †p < 0.05 relative to vehicle-treated MCAO group (analysis of variance and post-hoc Bonferroni/Dunn tests).

Rosiglitazone increases PPAR-γ-specific DNA-binding activity after ischemia

Although rosiglitazone is a PPAR-γ agonist, it has not been investigated previously whether rosiglitazone can indeed activate the PPAR-γ receptors in the brain. Furthermore, it is unknown whether the neuroprotective effects of rosiglitazone in the MCAO model are mediated via the activation of PPAR-γ. To address these issues, we performed the EMSA to examine PPAR-γ-specific DNA-binding activity in cortical nuclear extracts from brains treated with rosiglitazone (6 mg/kg) or vehicle. In sham control cortex, very low levels of PPAR-γ DNA-binding activity were present (Fig. 3a, lane 1). After 0.5, 1, or 2 h post-ischemic reperfusion, PPAR-γ DNA-binding activity in vehicle-treated brains was similar to that in sham control (Fig. 3a, lanes 2–4). However, PPAR-γ DNA-binding activity was markedly enhanced in rosiglitazone-treated brains during the same time frame after ischemia (Fig. 3a, lanes 5–7). The specificity of the EMSA for PPAR-γ binding was confirmed by supershift assay, in which the addition of an anti-PPAR-γ antibody to the DNA-binding reaction resulted in the shift of the band to larger molecular weight (Fig. 3b, lanes 3–4). The specificity of the PPAR-γ DNA-binding activity was also tested by adding excess (10- or 100-fold) of unlabeled PPAR-γ oligonucleotide (cold probe) to the assay, which eliminated the signals (Fig. 3b, lanes 5–6).

Figure 3.

Rosiglitazone enhances PPAR-γ DNA-binding activity in the brain after focal ischemia/reperfusion. (a) Electrophoresis mobility shift assay (EMSA) demonstrates rosiglitazone-induced elevation of PPAR-γ DNA binding activity in the brain after transient focal ischemia. EMSA was performed using cortical nuclear extracts from mice treated with either vehicle (lanes 2–4) or 6 mg/kg of rosiglitazone (lanes 5–7) and mice were killed at 0.5, 1, or 2 h after MCA occlusion. Sham-operated mice serve as controls (lane 1). The autoradiograph is the representative of three independent experiments with similar results. (b) The specificity of PPAR-γ DNA binding activity was confirmed by two different control assays. In the gel supershift assay, the addition of an anti-PPAR-γ antibody (lane 3, 2.5 μg; lane 4, 5 μg) to the DNA-binding reaction results in the shift of the band to larger molecular weight (super-shift indicated by an arrow). In the competition assay, an excess amount (10-fold, lane 5 or 100-fold, lane 6) of unlabeled PPAR-γ oligonucleotide (cold probe) was added to the reactions, which diminished the signals. Representative data from two independent experiments with similar results are presented.

To determine if PPAR-γ activation is critical for the neuroprotective effect of rosiglitazone in the cerebral ischemia model, the PPAR-γ antagonist GW9662 (5 μg) was administered (i.c.v.) to the mice 30 min prior to rosiglitazone treatment (6 mg/kg). EMSA showed that GW9662 treatment abolished rosiglitazone-induced enhancement of PPAR-γ DNA-binding activity (Fig. 4a). Infarct volume was also measured at 48 h after ischemia, and the results revealed that the neuroprotective effect of rosiglitazone was significantly reduced when mice were pretreated with GW9662 (Fig. 4b). These results suggest that a PPAR-γ-dependent mechanism may contribute significantly to the neuroprotection by rosiglitazone. In addition, GW9662 treatment alone did not significantly alter the infarct volume, suggesting that endogenous PPAR-γ activity, if without enhancement by rosiglitazone treatment, is inadequate to confer neuroprotection.

Figure 4.

Activation of PPAR-γ receptor contributes to the neuroprotective effect of rosiglitazone. (a) EMSA shows that the PPAR-γ receptor antagonist GW9662 is effective in reducing rosiglitazone-induced PPAR-γ DNA-binding activity in the brain. GW9662 (5 μg) was infused i.c.v. into the brain 30 min before rosiglitazone administration (6 mg/kg), and cortical nuclear extracts were prepared for EMSA at 1 h after MCA occlusion. The graph in the right panel illustrates the semiquantitative results of EMSA from four animals per group. **p < 0.01 versus sham non-ischemic control brains (first column); †p < 0.05 versus rosiglitazone-treated brains (second column). (b) Representative TTC-stained coronal sections from brains 48 h after 1.5-h MCA occlusion, showing that GW9662 reduces the neuroprotective effect of rosiglitazone after ischemia. The graph on the right illustrates the infarct volume in four groups of mice. Data are means ± SEM, n = 9 per group, **p < 0.01 versus vehicle-treated group (first column); †p < 0.05 versus the rosiglitazone-treated (third column) group (analysis of variance and post-hoc Bonferroni/Dunn tests).

Rosiglitazone attenuates inflammatory reactions after ischemia

Western blot analysis was performed to determine the effect of rosiglitazone treatment on the expression levels of the pro-inflammatory proteins ICAM-1 and MPO in the brain after ischemia and reperfusion. The expression of both ICAM-1 (24 h after MCAO) and MPO (48 h after MCAO) was markedly increased in ischemic brains. In contrast, rosiglitazone treatment significantly attenuated ischemia-induced expression of both pro-inflammatory factors, based on the semiquantitative measurements of immunoblots from four animals per experimental condition (Fig. 5a).

Figure 5.

Rosiglitazone-treated brains show reduced inflammatory reactions after focal ischemia/reperfusion. (a) Representative western blots for myeloperoxidase (MPO) and ICAM-1 show that both protein levels were attenuated in brains treated with rosiglitazone compared with the vehicle controls. The semiquantitative data (fold increases vs. sham non-ischemic brains) based on four brains per condition are presented below the blots. *p < 0.05 versus vehicle treatment. ICAM-1 and MPO immunoblotting was performed at 24 and 48 h after ischemia, respectively. (b) Ischemia-induced MPO activity, determined at 24 and 48 h after MCA occlusion, was reduced in rosiglitazone-treated brains. Data are means ± SEM, n = 5 per group, *p < 0.05; **p < 0.01 relative to vehicle-treated group (analysis of variance and post-hoc Bonferroni/Dunn tests). (c) Immunohistochemistry for ICAM-1 (panels i–iii, 24 h after ischemia), MPO (panels iv–vi, 48 h after ischemia), and lectin (panels vii–iv, 48 h after ischemia) was performed in sham non-ischemic brains (left) and in vehicle- (middle) or rosiglitazone-treated (right) ischemic brains. Images are taken from cortical areas relevant to the infarct border. Note that a large number of inflammatorily activated cells (indicated by arrows) are present in the vehicle-treated ischemic brains, whereas the active cells are relatively rare in rosiglitazone-treated brains. Scale bars: 100 μm (iii, vi); 50 μm (ix).

Elevated levels of MPO in ischemic brain are indicative of infiltration of neutrophils (Kaczorowski et al. 1995). According to the results of the MPO activity assay (n = 5 per group), MPO activity was increased about 2.0- and 4.3-fold in vehicle-treated brains at 24 and 48 h after ischemia, respectively, as compared with sham-operated non-ischemic brains. Rosiglitazone treatment significantly attenuated MPO activity at both time points after ischemia (Fig. 5b).

The inhibitory effect of rosiglitazone treatment on the ischemia-induced expression of ICAM-1 and MPO was confirmed at the cellular level using immunohistochemistry (Fig. 5c). ICAM-1 immunoreactivity was observed predominantly in small vessels in regions that bordered the infarct core at 24 h after MCA occlusion (Fig. 5a, panels i–iii). Very little or no ICAM-1 immunoreactivity was observed in the contralateral hemisphere (panel i). Compared with vehicle-treated ischemic brains (panel ii), treatment with rosiglitazone reduced numbers of ICAM-1-positive vessels after ischemia (panel iii). Similar to ICAM-1 staining, MPO-positive neutrophils were not detected in the contralateral hemispheres, but markedly increased staining was readily observed in both the infarct core and border at 48 h after MCAO (Fig. 5c, panels iv–vi). Compared with vehicle-treated controls (panel v), treatment with rosiglitazone decreased MPO staining after ischemia (panel vi). For both MPO and ICAM-1 immunohistochemistry experiments, incubation without the primary antibody resulted in the complete lack of staining (data not shown).

Tomato lectin immunohistochemistry was used to detect microglial cells in brains as described previously (Acarin et al. 1994). Lectin binding occurred in both resting and activated microglial cells in the brain (Fig. 5c, panels vii–ix). In the non-ischemic hemisphere, only resting microglia were present (panel vii); whereas in the ischemic brain (48 h after MCAO), large amounts of activated microglial cells (intensely stained cell body and fattening of processes) were detectable throughout the hemisphere, but robustly concentrated in ischemic regions that bordered the infarct core (panel viii). Compared with vehicle-treated ischemic brains, rosiglitazone-treated brains showed markedly reduced numbers of activated microglial cells after ischemia (panel ix).

Rosiglitazone attenuates the levels of pro-inflammatory cytokines after ischemia

To evaluate the effect of rosiglitazone treatment on the levels of various inflammatory cytokines after ischemia, ChemiArrayTM mouse cytokine antibody arrays were used to examine the protein profiles of a panel of 32 cytokines in brain protein extracts. In the present study, we tested three experimental groups (six animals per group), including the sham-operation control group, ischemic group with vehicle treatment, and ischemic group with rosiglitazone treatment (6 mg/kg). Each protein sample was pooled from two brains of the same experimental conditions; thus, the array experiments were repeated three times using three different sets of samples.

Figure 6(a) shows a representative set of arrays, which show six cytokines that were significantly up-regulated in the ischemic brain. The increases in cytokines in vehicle-treated ischemic brains were 4-fold for monocyte chemotactic protein 1 (MCP-1), 5.9-fold for metalloproteinase-1 (TIMP-1), 11.7-fold for keratinocyte chemoattractant (KC), 13.2-fold for granulocyte colony-stimulating factor (GCSF), 14.4-fold for soluble tumor necrosis factor receptor 1 (sTNFR1), and 41.2-fold for interleukin-6 (IL-6). In contrast, rosiglitazone treatment significantly attenuated the ischemia-induced increases of all six of these cytokines (Fig. 6b). The magnitudes of attenuation in cytokine levels by rosiglitazone ranged from 40% (TIMP-1) to 95% (IL-6) compared with the ischemic group with vehicle treatment.

Figure 6.

Rosiglitazone-treated brains show reduced cytokine expression after focal ischemia/reperfusion. (a) A representative set of ChemiArrayTM mouse cytokine antibody arrays show that the levels of six cytokines (indicated by the white boxes) are up-regulated in vehicle-treated ischemic brains, as compared with the sham non-ischemic control and the rosiglitazone-treated ischemic brain. The brain extracts were prepared at 24 h after MCA occlusion or sham operation. H1-2, granulocyte colony-stimulating factor (GCSF); B3-4, interleukin-6 (IL-6), J3-4, keratinocyte chemoattractant (KC), L3-4, monocyte chemotactic protein 1 (MCP-1); G5-6, soluble tumor necrosis factor receptor 1 (sTNFR1); I5-6, metalloproteinase-1 (TIMP-1). (b) Semi-quantitative results of six cytokines from optical density measurement of three sets of antibody arrays. Three experimental groups (six animals per group) were tested, including the sham control group, ischemic group with vehicle treatment, and ischemic group with rosiglitazone treatment (6 mg/kg). Each protein sample was pooled from two brains of the same experimental conditions. *p < 0.05; **p < 0.01 versus the vehicle-treated group (analysis of variance and post-hoc Bonferroni/Dunn tests).

Rosiglitazone attenuates LPS-induced cytokine release from cultured microglia

To further confirm that rosiglitazone is capable of directly inhibiting cytokine production and release from inflammatory cells, the effect of rosiglitazone on LPS-induced release of TNF-α, IL-1β, and IL-6 was examined in cultured rat brain microglia. Consistent with previous reports (Li et al. 2004; Nolan et al. 2004; Jin et al. 2005), LPS triggered the activation of microglia and massive release of TNF-α(Fig. 7a), IL-1β (Fig. 7b), and IL-6 (Fig. 7c) at 4 and 16 h after toxin incubation. Pre-incubation with rosiglitazone significantly attenuated LPS-induced release of all three cytokines examined, whereas the inhibitory effect of rosiglitazone disappeared in the presence of the PPAR-γ antagonist GW9662 (Fig. 7).

Figure 7.

Rosiglitazone inhibits the release of TNF-α, IL-1β, and IL-6 from cultured microglial cells. Unprovoked microglia (vehicle) produced minimal levels of TNF-α, IL-1β, and IL-6. Incubation with 500 ng/mL LPS for either 4 or 16 h induced a marked elevation of TNF-α (a), IL-1β (b), and IL-6 (c) in the media. Pre-incubation with rosiglitazone significantly inhibited the LPS-induced elevations of TNF-α, IL-1β, and IL-6. In the presence of the PPAR-γ antagonist GW9662, the inhibitory effect of rosiglitazone on cytokine releases is no longer detectable. *p < 0.05; †p < 0.01 versus the LPS group (analysis of variance and post-hoc Bonferroni/Dunn tests) from three independent experiments.

Discussion

The data presented in this report demonstrate the neuroprotective effect of rosiglitazone in a murine model of focal cerebral ischemia and reperfusion. Rosiglitazone, when administered 1 h prior to MCA occlusion, and again at 24 h after MCA occlusion, reduced infarct volume up to 64%, independent of any alterations in cerebral blood flow and other physiological parameters. Importantly, the anti-infarction effect of rosiglitazone was associated with improved functional outcomes assessed 7 days after ischemia. These results are consistent with two recently reported studies, which also showed neuroprotective effect of pre-ischemia treatment of two different PPAR-γ agonists, pioglitazone and troglitazone, in the rat model of transient focal ischemia (Shimazu et al. 2005; Sundararajan et al. 2005). Our study has further extended these studies to show that rosiglitazone has a therapeutic window of up to 2 h after post-ischemic reperfusion. Thus, the current study highlights rosiglitazone and other thiazolidinediones as a promising class of preventive and therapeutic agents for the treatment of stroke.

The protective effect of rosiglitazone against ischemic brain injury is likely attributable, at least in part, to its anti-inflammatory properties through the activation of PPAR-γ. In accordance with a previous report (Sanguino et al. 2003), we were able to detect PPAR-γ-specific DNA-binding activity in cerebral nuclear extracts, and found that this activity was markedly enhanced by rosiglitazone administration but reduced by the PPAR-γ antagonist GW9662. It is known that activation of PPAR-γ results in its binding to specific PPAR response elements in the promoter regions of various specific target genes, which in turn leads to either activation or suppression of target genes (Desvergne and Wahli 1999). In several cell types, including cultured cerebral astrocytes and microglia, thiazolidinediones have been shown to inhibit the transcriptional induction of various pro-inflammatory genes, such as TNF-α, IL-1, IL-6, major histocompatibility complex class II (MHCC-II), inducible nitric oxide synthase (iNOS), and cyclooxygenase-2 (Jiang et al. 1998; Ricote et al. 1998; Bernardo et al. 2000; Heneka et al. 2000; Fahmi et al. 2001; Luna-Medina et al. 2005). A number of studies have explored the role of the anti-inflammatory properties of thiazolidinediones in animal models of acute digestive or periarticular inflammation and chronic polyarthritis (Kawahito et al. 2000; Saubermann et al. 2002; Cuzzocrea et al. 2003a; Cuzzocrea et al. 2003b). The beneficial effect of thiazolidinediones in models of experimental autoimmune encephalomyelitis and the MPTP model of Parkinson's disease has also been linked to their anti-inflammatory actions, such as through blockage of NFκB activation and induction of iNOS (Feinstein et al. 2002; Dehmer et al. 2004).

In the present study, we demonstrate that rosiglitazone-treated ischemic brains had markedly reduced levels of several pro-inflammatory cytokines that are thought to contribute to ischemic brain injury. We also show that rosiglitazone treatment attenuated ischemia-induced activation of microglia, expression of ICAM-1, and neutrophil infiltration. Moreover, we found that rosiglitazone has a potent inhibitory effect on LPS-induced cytokine release from cultured brain microglia. The precise mechanism underlying the latter effect is unclear, but may occur via the PPAR-γ-mediated repression of STAT-1 and NF-κB (Bernardo et al. 2000), key factors known to influence LPS cell signaling. The capability of rosiglitazone to modulate microglial inflammatory responses may be particularly relevant to its beneficial effect on ischemic brain injury. Activated microglia are believed to be a major source for cytokine production during post-ischemic reperfusion, and their persistent activation may contribute to the secondary expansion of cerebral infarction (Mabuchi et al. 2000; Carmel et al. 2001; Iadecola and Alexander 2001). In addition, by releasing reactive oxygen species, activated microglia may also contribute to brain injury by damaging endothelial cells, disrupting the blood–brain barrier, and enhancing leukocyte infiltration, which further exacerbates brain injury (Barone and Feuerstein 1999; del Zoppo et al. 2000, 2001). Our observations are in agreement with the recent study by Sundararajan et al. (2005), who showed that another PPAR-γ agonist, troglitazone, attenuated the activation of microglia and reduced the expression of reactive oxygen species-producing cyclooxygenase-2 and iNOS following transient focal ischemia. Similar findings have also been reported in animal models of peripheral organ ischemia, such as myocardial infarction (Nakajima et al. 2001; Yue et al. 2001; Wayman et al. 2002), renal ischemia/reperfusion injury (Sivarajah et al. 2003), and ischemic lung injury (Ito et al. 2003). Taken together, the results suggest that the anti-inflammatory actions of rosiglitazone may play a key role in its protective effect against ischemic/reperfusion injury.

Despite the overwhelming evidence reported in the literature concerning the anti-inflammation effect of PPAR-γ activation, the data presented here cannot rule out the possibility that other mechanisms may also contribute to the beneficial effect of rosiglitazone treatment against ischemic brain injury. Indeed, several studies have suggested that certain PPAR-γ agonists may offer direct neuroprotective effects in cultured neurons. For instance, Uryu et al. (2002) reported that troglitazone inhibits both glutamate neurotoxicity and low-potassium-induced apoptosis in cerebellar granule neurons. A similar effect by troglitazone and 15D-PGJ2, a natural ligand for PPAR-γ, was also observed in retinal ganglion cells (Aoun et al. 2003). Several thiazolidinediones, including rosiglitazone, were found to inhibit β-amyloid-induced neurodegeneration in hippocampal neurons, presumably through the modulation of the Wnt signaling pathway (Inestrosa et al. 2005). Moreover, recently Shimazu et al. (2005) reported that treatment with pioglitazone in the rat model of transient focal ischemia resulted in increased levels of Copper-Zinc superoxide dismutase, suggesting that alterations in the expression of free radical scavengers could contribute to the neuroprotection. Therefore, further investigation may be warranted to determine to what degree the mechanisms independent of inflammation suppression underlie the neuroprotective effect of thiazolidinediones in the brain.

In summary, this study demonstrates the marked neuroprotective effect of rosiglitazone in the murine model of focal ischemia and reperfusion. The neuroprotection by rosiglitazone is attributed at least partially to its anti-inflammatory actions in the brain. The demonstrated efficacy of rosiglitazone in both pre- and post-ischemia administrative regimens supports the prospect that it may be a good candidate for stroke therapy. Several thiazolidinediones, including rosiglitazone, have been used successfully to treat type 2 diabetes without notable cytotoxicity or complications (Spiegelman 1998; Day 1999). In addition, thiazolidinediones do not appear to alter serum glucose levels in the non-diabetic human or rodents (Plutzky 2003; Shimazu et al. 2005; Sundararajan et al. 2005). These features make thiazolidinediones attractive as a clinical therapeutic for stroke, especially for patients with diabetes, which is considered a major risk factor for stroke and its clinical consequences.

Acknowledgements

This project was supported by NIH/NINDS grants NS43802, NS45048 and NS36736. JC and SHG were also supported in part by the Geriatric Research, Education and Clinical Center, Veterans Affairs Pittsburgh Health Care System, Pittsburgh, Pennsylvania. We thank Liping Sheng and Cristine O'Horo for their excellent technical assistance, Carol Culver for editorial assistance, and Pat Strickler for secretarial support.

Ancillary