Activation of cerebral peroxisome proliferator-activatedreceptors gamma promotes neuroprotectionby attenuation of neuronal cyclooxygenase-2overexpression after focal cerebral ischemia in rats
Yi Zhao, Andreas Patzer, Thomas Herdegen, Peter Gohlke, and Juraj Culman1
Institute of Pharmacology, University Hospital of Schleswig-Holstein, Campus Kiel,Kiel, Germany
ABSTRACT Up-regulation of cyclooxygenase (COX)-2exacerbates neuronal injury after cerebral ischemia andcontributes to neuronal cell death. The present studyclarifies the function of cerebral peroxisome-prolifera-tor-activated receptor(s) gamma (PPAR�) in the ex-pression of COX-2 in neurons of the rat brain aftermiddle cerebral artery occlusion (MCAO) with reper-fusion by immunohistochemistry, Western blot, andimmunofluorescence staining. In peri-infarct corticalareas the PPAR� was located in both microglia andneurons, whereas COX-2 was almost exclusively ex-pressed in neurons. PPAR� immunolabeling reachedthe peak 12 h after MCAO, whereas the number ofCOX-2 immunostained cells gradually rose and reachedits peak at 48 h. Intracerebroventricular infusion ofpioglitazone, an agonist of the PPAR�, over a 5-dayperiod before and 2 days after MCAO, reduced theinfarct size, the expression of tumor necrosis factor �(TNF-�), COX-2, and the number of cells positivelystained for COX-1 and COX-2 in the peri-infarct corti-cal regions. COX-2 induction was also attenuated in theipsilateral but not in the contralateral hippocampus. Inprimary cortical neurons expressing the PPAR�, piogli-tazone suppressed COX-2 expression in response tooxidative stress. This protective effect was reversedafter cotreatment with GW 9662, a selective antagonistof the PPAR�, clearly demonstrating a PPAR�-depen-dent mechanism. Our data provide evidence that acti-vation of neuronal PPAR� considerably contributes toneuroprotection by prevention of COX-2 up-regulationin vitro and in peri-infarct brain areas.—Zhao, Y.,Patzer, A., Herdegen, T., Gohlke, P., Culma, J. Activa-tion of cerebral peroxisome proliferator-activated re-ceptors gamma (PPAR�) promotes neuroprotection byattenuation of neuronal cyclooxygenase-2 overexpres-sion after focal cerebral ischemia in rats. FASEB J. 20,1162–1175 (2006)
Ischemic brain injury caused by a transient or per-manent reduction of cerebral blood flow results from a
complex of pathological events. Neurons localized inthe ischemic core, where the blood flow is severelyreduced, die within minutes after the ischemic episode(1, 2). A number of studies have demonstrated thecrucial role of inflammation in the progression ofneuronal loss and brain injury. Postischemic inflamma-tion is triggered by activation of proinflammatory genesand expression of adhesion molecules and cytokines,such as tumor necrosis factor � (TNF-�) and interleu-kin-1� (IL-1�), promoting accumulation of neutro-phils, macrophages and activated microglia (1). Inflam-matory reactions result in an expansion of the tissueinjury, which gradually develops at the periphery of theischemic core over hours and days after ischemic event.Ischemic neurons increasingly express cyclooxygenase(COX)-2, an enzyme that worsens the ischemic injuryby producing superoxide and prostaglandins (1). Suchreactions are deleterious to the damaged brain tissue asanti-inflammatory interventions result in alleviation ofneurological deficits and reduction of the infarct vol-ume (3).
All isoforms of COX, COX-1, COX-2, and COX-3, thelatter being encoded by the COX-1 gene, are expressedin the brain (4, 5). Although the chemical properties ofCOX-1 and COX-2 are similar, these two enzymes maysubserve different functions after ischemic injury.COX-2 has attracted more attention as it can directlydamage neurons (5, 6). COX-2, which is up-regulatedin neurons after ischemic injury, has been associatedwith excitotoxicity mediated by N-methyl-D-aspartate(NMDA) receptors as well as with neuronal cell death(7). Prostaglandin synthesis and free radical-mediatedlipid peroxidation induced by in vivo activation ofNMDA receptors in the hippocampus are also depen-dent on COX-2 activity (8). Selective inhibition ofCOX-2 attenuated the postischemic prostaglandin ac-
1 Correspondence: Institute of Pharmacology, UniversityHospital of Schleswig-Holstein, Campus Kiel Hospitalstrasse 424105 Kiel, Germany. E-mail: [email protected]
cumulation and reduced ischemic damage (9, 10).Similarly, COX-2-deficient mice showed reduced sus-ceptibility to ischemic brain injury and the NMDA-mediated neurotoxicity (11, 12). While COX-2 directlycontributes to the delayed neuronal cell death afterischemic injury, the function of COX-1 appears to berestricted to inflammatory processes mediated by infil-trated monocytes and glial cells (6, 13).
The peroxisome proliferator-activated receptor(s)gamma (PPAR�) has been primarily implicated inanti-inflammatory processes. The PPAR� acts as a neg-ative regulator of macrophage activation and PPAR�agonists inhibit the production of inflammatory cyto-kines in monocytes (14, 15). In cortical neuron-gliacocultures, PPAR� agonists abolished both LPS-stimu-lated expression of the inducible NOS (iNOS) inmicroglia and the NO release and COX-2 expression inneurons (16). Similarly, PPAR� ligands reduced theiNOS expression and attenuated cell death in cerebel-lar granule cells (17, 18).
Systemic treatment with PPAR� agonists, such aspioglitazone or troglitazone, improves the recoveryfrom cerebral ischemia (19, 20). We recently demon-strated that it is the intracerebral action of pioglitazonethat provides neuroprotection (21). The PPAR� inmicroglia/macrophages has been suggested to mediatethese beneficial effects, as treatment with PPAR� ago-nists attenuated microglia/macrophage accumulationand reduced the expression of proinflammatory medi-ators (20, 21). However, the role of neuronal PPAR� inneuroprotection after ischemic injury and their effectson COX-2 expression in neurons have not yet beenaddressed. Therefore, we investigated the role of thePPAR� in modulation of COX-2 expression in ischemicneurons and the impact on the progression of cerebraldamage after ischemic insult. In the present study, weassessed the localization of the PPAR� and COX-2 inthe peri-infarct cortical area after transient unilateralocclusion of the middle cerebral artery (MCA). Wedemonstrate that intracerebroventricular (ICV) treat-ment with pioglitazone, an agonist of the PPAR�,confers neuroprotection by suppression of COX-2 in-duction in neurons after cerebral ischemia. Experi-ments carried out in cortical cell culture provide evi-dence that the neuroprotective effects of pioglitazoneare directly mediated by the neuronal PPAR�.
MATERIALS AND METHODS
Induction of cerebral ischemia and implantation of osmoticminipumps
Focal cerebral ischemia was induced in male, normotensiveWistar rats (body wt 200–220 g; Charles River, Sulzfeld,Germany) by occlusion of the MCA for 90 min with subse-quent reperfusion as described previously (22, 54). Regionalcerebral blood flow (rCBF) was continuously monitored atone point (1 mm posterior to the bregma, 6 mm from themidline) on the surface of each hemisphere by laser-Doppler-flowmetry (Periflux system 5000, PERIMED) (23). Abrupt
reduction in rCBF by �75% to 90% indicated a successfulocclusion of the MCA. Body temperature was maintained at37°C with a heating pad. For ICV infusions of pioglitazone orvehicle, osmotic minipumps (ALZET Model No.2002, CharlesRiver, Sulzfeld, Germany), delivering solutions at a rate of 0.5�l/h, were implanted subcutaneously (s.c.) as described indetail elsewhere (22).
Measurement of infarct volume
Rats were deeply anesthetized and transcardially perfusedwith ice-cold PBS, pH 7.4, followed by 4% paraformaldehyde.The whole brains were removed, postfixed in 4% paraformal-dehyde overnight and cryoprotected in 30% sucrose at 4°Cfor 72 h. Serial coronal sections (40 �m) were cut in a cryostatfrom the level bregma �3.7 mm to the concentration bregma–6.7 mm. The sections were used to determine the infarctvolume and immunohistochemical evaluation of PPAR�,COX-1, and COX-2. To measure the infarct size, 15 sec-tions from different brain levels were stained with cresylviolet. Slice images were digitalized, and the area of theinfarct was determined in each slice (Leica QWin imageanalysis system) (24).
Immunohistochemical detection and morphometricevaluation of PPAR�, COX-1, and COX-2
Coronal free-floating brain sections (40 �m) were incubatedin PBS containing 0.1% Triton X-100 (PBST) for 5 min,followed by incubation in 0.03% H2O2 in methanol for 10min at room temperature to quench the endogenous perox-idase and in PBST with 5% normal goat serum (NGS) for 1 hto block unspecific binding. After additional washing, thesections were incubated with primary antibodies specific forPPAR� (rabbit anti-rat PPAR� polyclonal antibody (pAb), 1:100, Santa Cruz Biotechnology, Santa Cruz, CA, USA), COX-1(rabbit anti-rat COX-1 pAb, 1:1000, Cayman Chemical, AnnArbor, MI, USA) or COX-2 (rabbit anti-rat COX-2 pAb, 1:600,Cayman Chemical) overnight at 4°C. The slices were thenwashed and incubated with a biotinylated secondary antibody(Ab) (goat anti-rabbit Ab, 1:200, Vector Laboratories, Burlin-game, CA, USA) at 37°C for 1 h, followed by incubation withavidin-conjugated peroxidase at 37°C for 45 min (Vectastainavidin-biotin complex peroxidase kit). Immunolabeling wasvisualized using 3, 3�-diaminobenzidine as chromogen. Forthe morphometric evaluations, three random and nonover-lapping areas (0.125 mm2 per area) were chosen in theboundary zone of the ischemic core in the frontoparietalcortex as described previously (25). The quantification ofpositively stained cells was carried out using Leica imageanalyzing software (Leica Qwin).
Immunoflorescence staining
Serial 8 �m coronal sections (corresponding to coronalcoordinates bregma �0 to –2 mm) were fixed with coldacetone and incubated in PBST for 5 min. Afterward, thesections were incubated in 5% NGS in PBST at room temper-ature for 30 min. Cultured primary neuronal cells, grown onglass cover slips, were fixed with 4% paraformaldehyde in PBSfor 1 h at 4°C, washed, and permeabilized with 0.3% TritonX-100 in PBS for 5 min. Subsequently brain sections orprimary neuronal cells were incubated with the first primaryAb (rabbit anti-rat COX-2, 1:600 or rabbit anti-rat PPAR�,1:50 at 4°C) overnight, followed by incubation with thesecond primary Ab at 4°C for 4 h [mouse monoclonalantibody (mAb) against neuronal nuclei (NeuN), 1:100(Chemicon, Temecula, USA), mouse mAb against rat CD 68,
1163NEURONAL PPAR� ATTENUATES COX-2 AFTER CEREBRAL ISCHEMIA
1:100 (Serotec Ltd, Serotec, Oxford, UK), or goat pAb againstglial fibrillary acidic protein (GFAP), 1:800, Santa Cruz Bio-technology]. After washing with PBST, slices were incubatedwith the corresponding secondary Ab in 1% BSA at 37°C for1 h (Alexa Fluor®546-conjugated goat anti-rabbit Ab, AlexaFluor®488-conjugated goat antimouse Ab or Alexa Flu-or®546-conjugated donkey anti-goat Ab, Molecular Probes,Leiden, The Netherlands). After washing in PBS, the sectionswere mounted in Slow Fade® Light antifade reagent (Molec-ular Probes, Leiden, The Netherlands). For qualitative anal-ysis, stained sections or primary neuronal cells were analyzedwith a Leica DMR fluorescence microscope.
Western blot analysis
The dissected brain areas (cortical and hippocampal tissues;see the protocols) were homogenized in liquid nitrogen witha denaturing lysis buffer containing 20 mM Tris (pH 7.4), 2%SDS supplemented with 1% phosphatase inhibitor Cocktail II(Sigma-Aldrich, St. Louis, MO, USA) and protease inhibitormixture (Roche, Indianapolis, IN, USA). Cultured primaryneuronal cells were directly lysed in the lysis buffer. After ashort incubation (5 min at 95°C), the lysates were brieflysonicated and centrifuged (15,000 g at 4°C for 15 min) toremove insoluble materials. The protein concentration in thesupernatant was measured using the BCA protein assay Kit(Pierce, Rockford, IL, USA). Equivalent amounts of protein(15 �g per sample for COX-2 and 40 �g per sample forTNF-�) were separated on 12% SDS-polyacrylamide gels andtransferred to polyvinylidene difluoride transfer membranes(Millipore Corporation, Bedford, MA, USA). The membraneswere blocked with 4% nonfatty dry milk in Tris-bufferedsaline containing Tween-20 (TTBS) and incubated with theprimary Ab (rabbit anti-COX-2 pAb, 1:10 000 or goat anti-ratTNF-� pAb, 1:1 000) at 4°C overnight. After three washingsteps with TTBS, the membranes were incubated with horse-radish peroxidase-conjugated donkey anti-rabbit secondaryAb (1:3000, Amersham, Piscataway, NJ, USA) for 1 h at roomtemperature. The signal was visualized using the enhancedchemiluminescence (ECL) detection system and ECL hyper-film (Amersham). All membranes were finally stained withPonceau S (Sigma-Aldrich) to verify equal protein loading.The TNF-� and COX-2 bands were scanned and analyzedusing the quantification software (Quantity one, Bio-Radlaboratories, Hercules, CA, USA). Protein levels were ex-pressed as arbitrary units. �-Actin was used as a loadingcontrol and the bands were scanned. Since the obtainedvalues were very consistent, we did not normalize the TNF-�and COX-2 signals to the �-actin.
Primary neuronal cell culture
Mixed cortical cultures containing both neurons and astro-cytes were prepared as described recently (26). Briefly, thecerebral cortices from 0 to 1 day neonatal Wistar rats weredissected and dispersed using 0.25% trypsin, followed bygentle trituration to release cells. After washing with neuro-basal medium (Invitrogen, Carlsbad, CA, USA), cells werecounted and plated onto poly-L-lysine- (0.1 mg/ml) pre-coated cover slips in 4-well plates or Petri culture dishes (35mm) at a density of 1.5 105 cells for immunofluorescencestaining, or 2 106 cells for Western immunoblotting. Cellswere maintained in the neurobasal medium containing 2%B27 supplement, 0.5 mm L-glutamine, 100 U/ ml penicillin,and 100 �g/ml streptomycin (Invitrogen) in a humidifiedatmosphere of 5% CO2-95% air, at 37°C. Every 3 days, half ofthe culture medium was changed. Seven days after plating,the cultures consisted of 70–80% neurons and 20–30%
astrocytes as evaluated using a mAb against NeuN (Chemi-con, Temecula, CA, USA) and a goat polyclonal antiserumagainst GFAP (Santa Cruz Biotechnology), respectively.
Chemicals
Pioglitazone (Calbiochem, Merck Biosciences GmbH, Schwal-bach an Taunus, Germany) (6 mm) for ICV infusion wasprepared as described (21). For in vitro experiments (primarycell cultures), pioglitazone was dissolved in DMSO (1/4 of thefinal volume) and an equal volume of PBS was added toobtain the final concentration. The selective PPAR� antago-nist GW 9662 (Cayman Chemical) was dissolved in DMSO(10% of the final vol) and PBS was added to obtain the finalconcentration.
Experimental protocols
PPAR�- and COX-2-positive cells in the peri-infarctfrontoparietal cortex after cerebral ischemia
Rats subjected to MCAO for 90 min were deeply anesthetizedand transcardially perfused (see above) at the following timesafter the onset of reperfusion: 12 h, 1, 2, 3, 5, and 7 days(n4–5 rats/group). Sham-operated rats (n4) were treatedICV with vehicle and underwent the same surgical proceduresexcept for the occlusion of the MCA. Coronal brain sections(40 �m) were used for immunohistochemical detection ofthe PPAR� and COX-2 and serial 8 �m coronal sections(corresponding to coronal coordinates bregma 0 to –2 mm)to detect the PPAR� in neurons, microglia, and astrocytes bydouble-immunoflorescence staining.
Effect of pioglitazone on TNF-�, COX-1, and COX-2expression in response to cerebral ischemia
Vehicle (control group, n15) or pioglitazone (n14) weredelivered into the lateral brain ventricle by means of osmoticminipumps, which were implanted s.c. 5 days before MCAO.On day 6, the MCA was occluded for 90 min in all rats.Without interruption, ICV infusions of vehicle and pioglita-zone were continued for 2 consecutive days. Sham-operatedrats were treated ICV with vehicle and underwent the samesurgical procedures except for the occlusion of the MCA.Forty-eight hours after MCAO, the brains were removed,postfixed, and cryoprotected as described above. Coronalbrain sections (40 �m) were used to measure infarct volume(all brains) and to detect COX-1 and COX-2 immunostainedcells. Their quantification (sham: n4; vehicle: n6, piogli-tazone: n6) was carried out on three consecutive slicesobtained at the following brain levels: 1) bregma �0 mm, 2)bregma –1.3 mm, and 3) bregma –2.3 mm. Serial 8 �mcoronal sections were used for immunofluorescence stainingfor COX-2 in cells localized at the border of the infarctregion.
For Western blot analysis of TNF-� and COX-2, additionalgroups of rats were used. Rats were pretreated ICV with eithervehicle or pioglitazone and exposed to MCAO (see above).TNF-� was analyzed in brains isolated 24 h after MCAO(vehicle: n6; pioglitazone: n5). Brain samples for COX-2determination were obtained 48 h after MCAO (vehicle: n5;pioglitazone: n5). The brains were placed on their dorsalsurface in a plastic rat brain matrix with a coronal sliceorientation (World Precision Instruments, Inc., Sarasota, FL,USA) and cut into 7 serial 2 mm thick coronal sectionsbetween � 4 and –8 mm relative to the bregma. Western blotanalysis of TNF-� was carried out in the frontoparietal cortexadjacent to the ischemic core. The following tissue samples
1164 Vol. 20 June 2006 ZHAO ET AL.The FASEB Journal
were isolated and used for Western blot analysis of COX-2: 1)frontoparietal cortical tissue adjacent to the area of damage,2) the corresponding cortical area in the contralateral hemi-sphere, and 3) the hippocampus from both sides.
Effect of pioglitazone on COX-2 in primary neuronsexposed to oxidative damage
All experiments on mixed cortical cultures were carried outon day 7 after plating. The PPAR� in primary neuronal cellsand astrocytes was detected by double-immunofluorescencestaining carried out as described above. The optimal concen-tration and the duration of the exposure of cells to H2O2 toinduce oxidative neuronal damage were tested in preliminaryexperiments. Cellular toxicity was assessed by measuring thelactate dehydrogenase (LDH) release into the culture me-dium after exposure of cells to 3 different concentrations ofH2O2 (100, 200, and 400 �M) for 2, 4, 8, or 24 h. Exposure to100 �M H2O2 for 24 h produced a 4-fold increase in LDH(data not shown). To test the effects of PPAR� ligands onCOX-2 expression, primary neurons were exposed to 100 �MH2O2 for 24 h. Cells were treated with vehicle or pioglitazone(1 �M) in the presence or absence of the PPAR� antagonist,GW 9662 (1 �M), 30 min prior to H2O2 exposure (100 �M).Twenty-four hours after the treatment, cells were either fixedwith 4% paraformaldehyde in PBS and used for immunoflu-orescence staining or lysed to obtain protein extracts forWestern blot. A 24 h exposure of cells to pioglitazone or GW9662 alone did not induce any toxic effects as revealed bymeasurement of LDH release into the culture medium.
The animal protocols were approved by the GovernmentalCommittee for the Ethical Use of Animals in the GermanFederal State Schleswig-Holstein.
Statistical analyses
All values are expressed as means � se. Comparisons ofinfarct areas between the vehicle-treated and the pioglita-zone-treated groups of rats were analyzed by repeated mea-sures of ANOVA with two independent groups of subjects,followed by a post hoc Bonferroni test for pairwise compari-sons. The effects of ischemia on the time courses of PPAR�and COX-2 immunostained cells in the frontoparietal cortexwere analyzed by 1-way ANOVA, followed by a post hocDunnett test (comparisons vs. sham, 0 days). Comparisons ofthe numbers of cells positively stained for COX-1 and COX-2in the vehicle- and pioglitazone-treated groups of rats werecarried out by 1-way ANOVA, followed by a post hoc Bonfer-roni test. Student’s t test for unpaired samples was used tocompare the data on TNF-� and COX-2 expression in thecortex and hippocampus between the vehicle- and pioglita-zone-treated groups of rats. Statistical analysis of COX-2expression in primary neuronal cell culture (Western blotanalysis) was carried out by 1-way ANOVA, followed by a posthoc Bonferroni test. Statistical significance was accepted atP � 0.05.
RESULTS
Regional cerebral blood flow
A reduction of rCBF by �75% of the baseline indicatesa successful occlusion of the middle cerebral artery.rCBF reductions during MCAO and rCBF values duringreperfusion were identical in both groups of rats (vehi-cle-treated: n15; pioglitazone-treated: n14) at any
time during MCAO and reperfusion (10 min MCAO:vehicle-treated: –83%, pioglitazone-treated: –83%;30 min MCAO: vehicle-treated: – 80%; pioglitazone-treated: –81%; 60 min MCAO: vehicle-treated: –79%,pioglitazone-treated: –80%; 30 min reperfusion: vehi-cle-treated: –34%, pioglitazone-treated: –31%).
PPAR�- and COX-2-positive cells in the frontoparietalcortex after focal cerebral ischemia
We first examined the time course of PPAR�- andCOX-2 expression in cells after cerebral ischemia. Thehighest accumulation of cells positively stained for thePPAR� was observed in the frontoparietal cortex of theischemic side. MCAO with reperfusion significantlychanged the density of these cells (F6,186.735; P �0.001). The number of PPAR� immunoreactive cellsdramatically increased 12 h after MCAO (P�0.01),then returned to basal values and remained unchangeduntil the end of the observation period of 7 days (Fig.1). Double immunofluorescence staining showed thatalmost all microglial cells displayed an intense PPAR�immunoreactivity (Fig. 2). However, the PPAR� was
Figure 1. Time-dependent expression of the PPAR� (upperpanel) and cyclooxygenase-2 (COX-2, lower panel) in cellslocalized in the peri-infarct cortical area. Rats (n4–5) werekilled at the indicated times after middle cerebral arteryocclusion with reperfusion. Data are expressed as themeans � se. Statistical comparison with sham-operated ani-mals (0 days): *P � 0.05; **P � 0.01, calculated by ANOVA,followed by a post hoc Dunnett test.
1165NEURONAL PPAR� ATTENUATES COX-2 AFTER CEREBRAL ISCHEMIA
also localized in neurons as identified by the neuron-specific marker NeuN (Fig. 3), whereas only few astro-cytes (GFAP-positive cells) were immunopositive forPPAR� (data not shown).
Basal COX-2 immunostaining was present in thefrontoparietal cortex of sham-operated animals (Fig.
1). In rats exposed to cerebral ischemia, the highestaccumulation of COX-2-positive cells was detected atvery close vicinity to the ischemic core. The density ofthese cells significantly changed after MCAO withreperfusion (F6,3728.843; P�0.001). The number ofCOX-2 immunostained cells had already increased at12 h, reached a peak at 48 h after MCAO, and re-
Figure 2. Expression of the PPAR� in microglial cells local-ized in the peri-infarct cortical areas. Representative immu-nofluorescence staining on the same sections against A) CD68 as a marker for microglial cells (green), and B) PPAR�(red). C) Overlapping of PPAR� immunoreactivity in micro-glial cells (yellow) shows that a large number of microglialcells in this area express the PPAR� (arrows).
Figure 3. Expression of the PPAR� in neurons localized in theperi-infarct cortical areas. Representative immunofluores-cence staining on the same sections against A) NeuN as amarker for neurons (green), and B) PPAR� (red). C) Over-lapping of PPAR� immunoreactivity in neuronal cells (yel-low) shows that about half of the neurons in this area werepositively stained for the PPAR� (arrows).
1166 Vol. 20 June 2006 ZHAO ET AL.The FASEB Journal
mained elevated until the end of the observation pe-riod of 7 days (see Figs. 1, 5). Immunofluorescencestaining of brain sections revealed that COX-2 waslocalized mainly in neurons (see Fig. 7, below).
Activation of the PPAR� reduces the expression ofTNF-�, COX-1, and COX-2 and protects brain tissueafter focal cerebral ischemia
Since pioglitazone reaches very low concentrations inthe brain soon after its systemic application, a long-term ICV infusion of pioglitazone was chosen to ensurean effective activation of brain PPAR� and to excludeany contribution of peripheral effects of the PPAR�ligand.
Injured brain cells increasingly produce IL-1� andTNF-� early after ischemic insult. Both cytokines con-tribute to the ischemic injury (1). Twenty-four hoursafter MCAO, the expression of IL-1� was inhibited inischemic brains of rats treated systemically with trogli-tazone (20). In the present study, pioglitazone down-regulated TNF-� in the frontoparietal cortex adjacentto the ischemic core 24 h after MCAO when comparedwith vehicle-treated rats. (Fig. 4).
Transient occlusion of the MCA induced COX-2 asevidenced by accumulation of COX-2-positive cells inthe peri-infarct cortical areas 2 days after MCAO(bregma �0.0 mm: F2,1227.16; P�0.001; bregma –1.3mm: F2,1241.08, P�0.001, and bregma –2.3 mm:F2,1329.45, P�0.001). Pioglitazone significantly re-duced the number of COX-2 immunostained cells at allthree levels of the brain (Fig. 5). Western blot analysisof COX-2 expression in the frontoparietal cortex con-firmed the immunohistochemical data. Pioglitazone
suppressed the COX-2 induction in the peri-infarctcortical area and in the ipsilateral hippocampus, whichwas not directly damaged by a reduction in blood flowduring ischemia (Fig. 6). In contrast, pioglitazone didnot affect COX-2 levels in the contralateral cortex orhippocampus (Fig. 6).
The majority of cells in the boundary zone to theischemic core expressed COX-2. Double immunofluo-rescence staining revealed that COX-2 was localizedmainly in neurons (Fig. 7) and only sparsely in micro-glial cells (Fig. 8), whereas astrocytes were not stained(data not shown). In vehicle-treated rats exposed toMCAO, COX-2 was induced in the majority of neuronslocalized in the peri-infarct cortex. In contrast, activa-tion of cerebral PPAR� considerably reduced the num-ber of COX-2 immunoreactive cortical neurons, indi-cating that pioglitazone prevented neuronal damage(Fig. 7).
Compared to COX-2, the density of cells expressing
Figure 4. The expression of tumor necrosis factor � (TNF-�)is reduced in ischemic brain of rats treated with pioglitazone.Western blot analysis of TNF-� in the frontoparietal cortexipsilateral to the ischemic injury. A representative blot isshown in the upper part. The density analysis was performedon the TNF-� bands. �-Actin was used as a loading control.The histogram in the lower part shows a reduction in TNF-�expression in pioglitazone-treated rats (hatched columns)when compared to the vehicle-treated group (solid columns).Results are expressed as the means � se. Statistical compari-son with the vehicle-treated group: **P � 0.01 calculated byStudent’s t test for unpaired samples.
Figure 5. The PPAR� reduces the expression of cyclooxygen-ase-2 (COX-2) and cyclooxygenase-1 (COX-1) in the peri-infarct frontoparietal cortex. Depicted are the numbers ofcells positively stained for COX-2 and COX-1 in sham-oper-ated rats (n4, empty columns) and rats treated intracere-broventricularly with vehicle (n6, solid columns) or withpioglitazone (n6, hatched columns). Pioglitazone substan-tially reduced the number of cells that stained positively forCOX-2 (upper panel) and for COX-1 (lower panel), 48 hafter middle cerebral artery occlusion for 90 min. Results areexpressed as the means � se. Statistical comparison withsham-operated animals: *P � 0.05, **P � 0.01, and ***P �0.001 and with the vehicle-treated group: †P � 0.05 and ††P �0.01, calculated by ANOVA, followed by a post hoc Bonfer-roni test.
1167NEURONAL PPAR� ATTENUATES COX-2 AFTER CEREBRAL ISCHEMIA
COX-1 in the boundary zone to the infarct core wasmuch lower (Fig. 5). Transient cerebral ischemia didnot significantly change the number of COX-1-positivecells at the level of the bregma (P0.21), whereas anincrease was observed at two brain levels caudally to thebregma (bregma –1.3: F2,139.14, P�0.01; bregma–2.3: F2,133.84, P�0.05). As with COX-2, pioglitazonereduced the density of COX-1-positive cells in theseareas (Fig. 5).
Two days after focal cerebral ischemia, severe unilat-eral injury was clearly detected as area of pallor that wassharply demarcated from the adjacent tissue. The totalinfarct volume in pioglitazone-treated rats was 154 � 19mm3, which was significantly lower than that in vehicle-treated rats (244�17 mm3, P�0.05). The decrease in
infarct size was significant in 7 of 15 brain sectionsexamined (F9.36; P�0.01) (Fig. 9).
The PPAR� reduces COX-2 of primary neurons andsupports their survival after oxidative damage
Experimental data indicate that PPAR� ligands protectneurons rather indirectly, through effects on mono-cytes and microglia (16). In the last set of experiments,we investigated whether a direct activation of thePPAR� in neurons can modify the expression of COX-2triggered by oxidative damage. The culture conditionsresulted in a mixed cell population with �30% ofastroglia and �70% neurons (data not shown). Doubleimmunofluorescence staining showed that the PPAR�was localized in almost all neurons (Fig. 10), but onlysparsely in astrocytes (�25%) (Fig. 11). A 24 h expo-sure of neurons to H2O2 induced a robust expression ofCOX-2. Double immunofluorescence staining revealedthat the majority of neurons were positively stained forCOX-2 (Fig. 12 and Fig. 13). Pioglitazone added toprimary neuronal cells prior to H2O2 significantly re-duced the number of COX-2-positive neurons. Block-ade of PPAR� by cotreatment with the selective PPAR�antagonist, GW 9662, reversed the effect of pioglita-zone (Fig. 13). Western blot analysis confirmed theresults obtained in immunohistochemical experiments.The changes in COX-2 expression in primary neuronalcell culture exposed to H2O2 with or without pioglita-zone/GW 9662 were highly significant (F3,327.165;P�0.001). H2O2-induced COX-2 expression was sup-pressed in the presence of pioglitazone, and GW 9662reversed the pioglitazone-induced reduction in COX-2expression indicating that the effects of pioglitazonewere indeed mediated by PPAR� (Fig. 13).
DISCUSSION
The results of the present study provide novel insightinto the mechanisms of the PPAR�-mediated neuropro-tection. The PPAR� has been detected in several areasof the adult brain and spinal cord and its presence wasobserved in both neuronal and glial cells (27–31). Theexpression of PPAR� in microglia is tightly regulatedand dependent on microglial functional state (28).Consistent with the up-regulation of PPAR� immuno-reactivity and mRNA in the ischemic brain (32 andreferences therein), few hours after MCAO, PPAR�immunolabeling increased in the cortical peri-infarctareas mainly in microglial cells and, to a lesser extent,in neurons. The number of PPAR�-positive cells thendramatically declined and the lowest values werereached 2 days after ischemic injury, when the postisch-emic inflammation had already developed and theperi-infarct region is strongly infiltrated with invadingmacrophages and activated microglia (24). This obser-vation is consistent with the finding of a down-regula-tion of the PPAR expression during microglia activa-tion (28). TNF-� is a potent suppressor of the PPAR�
Figure 6. Cyclooxygenase-2 (COX-2) protein expression isdecreased in ischemic brains of rats treated with pioglitazone.Western blot analysis of COX-2 in the frontoparietal cortex(upper panel) and in the hippocampus (lower panel) ipsilat-eral (ips) and contralateral (ctr) to the ischemic injury.Representative blots are shown on the right upper side;histograms are on the lower left side of each panel. Thedensity analysis was performed on the COX-2 bands. �-Actinwas used as a loading control. Pioglitazone (hatched col-umns) attenuated the COX-2 expression in the ipsilateralfrontoparietal cortex and in the hippocampus when com-pared to the vehicle-treated group (solid columns). There wasno significant difference in COX-2 induction in the contralat-eral frontoparietal cortex and hippocampus. Results areexpressed as the means � se. Statistical comparison with thevehicle-treated group: ***P � 0.001 calculated by Student’s ttest for unpaired samples.
1168 Vol. 20 June 2006 ZHAO ET AL.The FASEB Journal
expression in adipocytes, and such antagonism alsoappears to occur in inflammatory cells (33). Conse-quently, the dramatic drop in the initially increasednumber of PPAR�-positive cells was most likely inducedby a rapid down-regulation of PPAR� during microgliaactivation initiated by TNF-� or by other mediators ofinflammation.
We demonstrate that activation of the PPAR� sup-presses the COX-2 induction. COX-2, which is consti-tutively expressed throughout the brain in discretepopulations of neurons (34), plays a crucial role in theprogression of brain injury after ischemia (6). COX-2mRNA levels in the cortical peri-infarct area reach theirhighest levels 12 h after ischemic injury, and the extentof COX-2 mRNA induction correlates with the severityof tissue damage (9, 35, 36). Here we report that thenumber of COX-2-positive cells in the peri-infarct cor-tex gradually increased, reaching the maximum 48 hafter the ischemic insult. In line with previous findings,COX- 2 expression occurred mainly in neurons (9),although a few positively stained microglial cells werealso detected. A robust induction of COX-2 in peri-infarct neurons and infiltrating neutrophils was re-ported in humans who died 1–2 days after ischemicinsult in the territory supplied by the MCA (37, 38).
Beneficial effects of systemically administered PPAR�agonists in ischemic stroke have been recently demon-strated (19, 20). Pioglitazone may improve the recoveryfrom ischemic stroke by reduction of oxidative stressand increased production of NO in endothelial cells,which in turn can modulate cerebral blood flow duringischemia and reperfusion period (39). Activation ofPPAR� in the brain may improve the recovery fromstroke by blocking of pathophysiological processes andevents, such as inflammatory reactions and apoptosis.Thus, the beneficial effects of systemically administeredPPAR� agonists in stroke result from activation of bothperipheral and central PPAR�. In the present study, wesought to investigate the effect of an activation of brain
PPAR� on postischemic expression of COX-1 andCOX-2 in neuronal and glial cells. To achieve anexclusive and selective activation of brain PPAR�, pio-glitazone was continuously infused ICV over a 5-dayperiod before and during 1 or 2 consecutive days afterMCAO with reperfusion. We have reported that piogli-tazone infused ICV at the given dose attenuated accu-mulation of inflammatory cells in the peri-infarct areaand improved the recovery from ischemic stroke. Asimilar dose of another PPAR� agonist, troglitazone,reduced the neuronal death in response to bacterialLPS when microinjected into the cerebellum (21 andreferences therein).
Pioglitazone substantially reduced the number ofCOX-2-positive cells and the COX-2 expression in peri-infarct cortical tissue. In vitro studies and ex vivoexperiments in animals have conclusively demonstratedthat augmented COX-2 activity within ischemic neu-rons exacerbates ischemic injury and promotes neuro-nal cells death (6). In primary neuronal cell cultures,inhibition of COX-2 ameliorated the NMDA- inducedcell injury and protected neurons from the LPS-in-duced neuronal death (40, 41). Selective COX-2 inhib-itors reduced the infarct volume after focal cerebralischemia and the COX-2-deficient mice showed re-duced susceptibility to ischemic brain injury (9, 11, 12).In view of these findings, the observed reduction in theCOX-2 expression in pioglitazone-treated rats stronglyindicates that activation of cerebral PPAR� protectsneurons against ischemic injury induced by excitotox-icity and anoxia, and prevents neuronal loss in theperiphery of the infarct, which is the potential targetfor therapeutic intervention.
The decrease in COX-2 expression in peri-infarctcortical tissue in pioglitazone-treated rats may resultfrom the inhibition of macrophage accumulation andmicroglia activation. There is now compelling evidencethat proinflammatory cytokines IL-1� and TNF-� play acentral role in the progression of postischemic brain
Figure 7. Expression of cyclooxygenase-2(COX-2) in neurons localized in the peri-in-farct frontoparietal cortex in vehicle-treatedand pioglitazone-treated rats. Double immuno-fluorescence staining on the same sectionsagainst A) NeuN as a marker for neurons(green), and B) COX-2 (red). C) Overlappingof COX-2 immunoreactivity in neurons (yellow)in rats treated intracerebroventricularly withvehicle shows that almost all neurons at theedge of the infarct core were positively stainedfor COX-2. D) Double immunofluorescencestaining for NeuN/COX-2 (yellow) in ratstreated ICV with pioglitazone demonstrates thatless than half of the neurons in the peri-infarctfrontoparietal cortex express COX-2.
1169NEURONAL PPAR� ATTENUATES COX-2 AFTER CEREBRAL ISCHEMIA
injury. Increased production of IL-1� and TNF-� byinjured brain cells stimulates the expression of adhe-sion molecules including intercellular adhesion mole-cule 1 (ICAM 1) and vascular adhesion molecule(VCAM) on the endothelial cell surface, which medi-
ates the adhesion and migration of neutrophils andmacrophages into the ischemic brain parenchyma.Infiltrating cells and activated microglia increasinglyproduce inflammatory cytokines that further exacer-bate postischemic inflammation and contribute to neu-ronal damage (1, 2). PPAR� agonists can regulate theexpression of adhesion molecules, cytokines, and che-mokines and other inflammatory mediators throughmultiple mechanisms. For instance, PPAR� ligandsinhibit the increase in ICAM 1 and VCAM and suppressthe production of inflammatory cytokines, such asIL-1� and TNF-� (see ref 32 for a review). Indeed,systemically applied troglitazone decreased IL-1� levelsin the ischemic brains (20). Pioglitazone in the presentstudy was infused ICV, therefore, we do not assume thatthe PPAR� ligand could interfere with the productionof adhesion molecules in endothelial cells. However,pioglitazone reduced the expression of TNF-� in peri-infarct cortical tissue that, together with the decreasedlevels of IL-1� (20), may account for the lower densitiesof reactive microglial cells and macrophages in thisarea observed in pioglitazone-treated rats after MCAO(21). The down-regulation of the neutrophilic andreactive microglia infiltration and production of toxicmediators may in turn reduce the expression of COX-2in injured neurons and prevent neuronal cell death.
Pioglitazone also attenuated the induction of COX-2in the hippocampus. The blood flow in the hippocam-pus is usually not reduced during MCAO, and neuronalCOX-2 is induced by activation of NMDA receptors (7).In rodents, COX-2 was particularly expressed in thevulnerable hippocampal CA1 neurons and contributedto the delayed neuronal death after cerebral ischemia(12, 42, 43). Our data indicate that activation of brainPPAR� reduces COX-2 expression and amelioratesneuronal injury and cell death in both peri-focal areas,
Figure 8. Expression of cyclooxygenase-2 (COX-2) in micro-glia in the peri-infarct frontoparietal cortex in vehicle-treatedrats. Double immunofluorescence staining on the same sec-tions against A) CD 68 as a marker for microglial cells(green), and B) COX-2 (red). C) Overlapping of COX-2immunoreactivity in microglial cells (yellow) shows that onlyfew of these cells are positively stained for COX-2.
Figure 9. Activation of cerebral PPAR� reduces the infarct sizeafter focal cerebral ischemia. Rats were treated intracerebrov-entricularly with vehicle (n15, solid circles) or with piogli-tazone (n14, empty circles) over a 5-day period before and2 days after middle cerebral artery occlusion (MCAO) for 90min. Total (cortical and subcortical) areas of injury areshown. Pioglitazone significantly reduced the infarct size. Thex axis shows the anterior (A) -posterior (P) distance from thebregma, Results are expressed as the means � se. Statisticalcomparison with the vehicle-treated group: *P � 0.05 and**P � 0.01, calculated by 2-way ANOVA, followed by a posthoc Bonferroni test.
1170 Vol. 20 June 2006 ZHAO ET AL.The FASEB Journal
in which neurons are at a high risk to die, andextrafocal sites, where neurons are not directly dam-aged by a local reduction or an arrest of blood flow.Profound beneficial effects of pioglitazone on theneuronal survival most likely contribute to the im-proved outcome of the ischemic damage, such as thereduction of the infarct size. Systemic treatment with
PPAR� ligands confers neuroprotection after ischemicstroke (19, 20). Intracerebral application of pioglita-zone exerts beneficial effects similar to those observedafter systemic treatment (21). Taken together, ourfindings provide evidence that the neuroprotective
Figure 10. Primary neuronal cells express the PPAR�. Doubleimmunofluorescence staining for A) neurons (green), and B)PPAR� (red). C) Overlapping of PPAR� immunoreactivity inprimary neuronal cells (yellow) demonstrates that almost allneuronal cells express the PPAR�.
Figure 11. Expression of the PPAR� in astrocytes in mixedcortical cell cultures. Double immunofluorescence stainingfor A) astrocytes (green) and B) PPAR� (red). C) Overlappingof PPAR� immunoreactivity in astrocytes (yellow) shows thatthe PPAR� is only sparsely localized in astrocytes (arrow).
1171NEURONAL PPAR� ATTENUATES COX-2 AFTER CEREBRAL ISCHEMIA
actions of PPAR� agonists are mediated by intracere-bral PPAR� and are independent of peripheral mech-anisms.
Some of the anti-inflammatory and neuroprotective
effects of PPAR� ligands do not require activation ofthe PPAR� (32, 44–46). To define more precisely therole of PPAR� in the reduction of COX-2 expression inischemic tissue, we examined the effects of pioglitazoneand the selective PPAR� antagonist, GW 9662, on theexpression of COX-2 in neonatal rat primary neuronsin response to reactive oxygen species (ROS). Oxida-tive damage mediated by ROS during the reperfusionfurther exacerbates ischemic injury and results in aseries of molecular events leading to neuronal celldeath (1, 2). H2O2 is a major ROS generated as aby-product of cellular metabolic events. The beneficialeffects of pioglitazone in primary cortical neuronsexposed to H2O2 namely, the attenuation of COX-2expression and the prevention of neuronal cell death,were completely reversed by GW 9662, clearly indicat-ing a PPAR�-dependent mechanism.
Data from experimental studies on neuroinflamma-tion suggest that PPAR� ligands protect neurons indi-rectly, through effects on monocytes/microglia (46,47). In neuron-microglia cocultures, PPAR� ligandsinhibited LPS-induced neuronal death by blocking theproduction of neurotoxic molecules in microglia (16).We hypothesize that in addition to the anti-inflamma-tory actions, activation of PPAR� in neurons is decisivefor the neuroprotective actions of PPAR� ligands, as 1)primary neuronal cells express PPAR� and 2) pioglita-zone substantially prevents neuronal damage and celldeath in response to oxidative injury in the absence ofmicroglia as demonstrated in the present experiments.Similarly, other PPAR� agonists, such as thiadiazolidi-nones, protected primary cortical neurons from apo-ptosis induced by cell-free medium from LPS-activatedmicroglial cultures (48). These findings convincinglydemonstrate that, apart from the PPAR�-mediated in-hibition of microglia activation, it is the activation ofthe PPAR� in neurons that inhibits COX-2 expressionand confers neuroprotection against cerebral ischemiaand noxious stimuli. Recent findings have demon-strated that activation of the PPAR� promotes neuro-regeneration and neurite outgrowth in cultured cells ofneuronal origin (49). These neurotrophic effects mayalso contribute to neuroprotective actions of thiazo-lidinediones in ischemic brain.
Pioglitazone significantly reduced the number ofcells in the peri-infarct cortical areas that stained posi-tively for COX-1. In contrast to COX-2, the role ofCOX-1 in ischemic stroke is controversially discussed(6, 13). COX-1 plays a critical role in the maintenanceof resting CBF and in the vasodilation induced byendothelium-dependent vasodilators (50). The COX-1-deficient mice were more susceptible to brain damageafter cerebral ischemia due to a more pronounced CBFreduction during ischemia (51). Moreover, COX-1gene transfer augments prostacyclin PGI2 and reducesleukotriene productions (52). In apparent contrast tothese studies, COX-1 has been demonstrated to exac-erbate neuronal damage and to promote neuronaldeath following global ischemia (53). Thus, the ob-served inhibition of COX-1 induction by pioglitazone
Figure 12. Cyclooxygenase-2 (COX-2) is induced in primaryneuronal cells exposed to oxidative damage. Seven days afterplating, primary cortical cells were exposed to hydrogenperoxide (H2O2) (100 �M) to induce oxidative damage.Representative immunofluorescence staining for A) neurons(green), B) COX-2 (red), and C) overlapping of COX-2immunoreactivity in primary neuronal cells (yellow) shows arobust induction of COX-2 in all neurons.
1172 Vol. 20 June 2006 ZHAO ET AL.The FASEB Journal
may confer additional protection against ischemic neu-ronal damage.
The present data favor the view that besides thePPAR�-mediated suppression of inflammation and in-vasion of microglia/macrophages (20, 21), activation ofthe PPAR� in neurons promotes neuroprotection. Ac-tivation of PPAR� attenuated TNF-� production, whichmay contribute considerably to the observed inhibitionof COX-2 expression and reduction of ischemic injury.Hosomi et al. (55) recently demonstrated that blockadeof TNF-� by neutralizing antibodies reduces the infarctsize and edema after transient focal cerebral ischemia.On the other hand, activation of PPAR� in neurons candirectly decrease COX-2 production, as PPAR� ligandswere shown to suppress transcriptional activation ofCOX-2 in various cell types (56–58).
Our findings also address an important issue aboutthe potential therapeutic use of PPAR� agonists in theprevention and treatment of ischemic stroke. Thiazo-lidinediones lower glucose (Glc) concentrations byameliorating insulin resistance, improve endothelialdysfunction, reduce blood pressure, and amelioratedyslipidemia (59). A recent clinical study has demon-strated that pioglitazone reduces the composite ofall-cause mortality, nonfatal myocardial infarction, and
stroke by �16% in patients with type 2 diabetes whohad extensive evidence of macrovascular disease (60).Therefore, patients with type 2 diabetes may benefitfrom treatment with thiazolidinedione, since thesedrugs lower Glc concentrations and prevent seriouscardiovascular events, including stroke. On the otherside, nondiabetic patients who have a high risk of strokeshould not necessarily be treated with pioglitazone orsimilarly acting drugs, because such treatment has beenassociated with serious side effects, including higherincidence of heart failure, edema not attributable toheart failure and increases in body weight (60). Resultsof the present and other studies (19, 20) lead one toassume that PPAR� agonists might be beneficial whengiven after ischemic injury. The most serious damagingevents in ischemic brain tissue develop within few hoursor days after ischemic insult (1). Treatment of patientswith PPAR� agonists during this short period mayimprove the outcome of ischemic stroke without seri-ous side effects, such as heart failure, which are gener-ally observed after a lengthy treatment. In a recentexperimental study, troglitazone proved beneficial ef-fects in rats when injected 1 h after the onset of MCAO(20). Unfortunately, the rats were examined only once,24 h after MCAO, so the effects of a long-lasting
Figure 13. Activation of PPAR� in primary neuro-nal cells reduced cyclooxygenase-2 (COX-2) in-duction in response to oxidative damage. Sevendays after plating, primary cortical cells wereexposed to A) vehicle or B) incubated with hydro-gen peroxide (H2O2) (100 �M) alone, or C) inthe presence of pioglitazone (1 �M), or (D)pioglitazone � GW 9662 (1 �M). Pioglitazoneand the PPAR� antagonist, GW 9662, were added30 min prior to H2O2. Upper panel: A robustinduction of COX-2 (fluorescent red) was ob-served in neurons exposed to H2O2. Pioglitazoneadded to primary neuronal cells prior to H2O2reduced the number of COX-2-positive neuronsand this effect was reversed by cotreatment withGW 9662. Lower panel: Western blot analysis ofCOX-2 in primary neuronal cells. Pioglitazonesuppressed the H2O2- induced COX-2 expressionand GW 9662 reversed the pioglitazone-inducedreduction in COX-2 expression. The density anal-ysis was performed on the COX-2 bands. �-Actinwas used as a loading control. A representativeblot is shown on the right lower side of the figure.Results are expressed as the means � se. *P �0.05, **P � 0.01, statistical comparisons to vehi-cle-treated cells, #P � 0.05 and †P � 0.05, statisti-cal comparisons to H2O2-treated and H2O2 �pioglitazone � GW 9662-treated cells, respec-tively, calculated by ANOVA, followed by a posthoc Bonferroni test.
1173NEURONAL PPAR� ATTENUATES COX-2 AFTER CEREBRAL ISCHEMIA
activation of PPAR� after stroke could not be ascer-tained. New synthetic PPAR� agonists with higher avail-ability in the brain and follow-up studies in experimen-tal animals treated after ischemic insult may providethe rationale for the treatment of patients sufferingfrom stroke with PPAR� agonists.
The authors thank Jan Brdon for his excellent technicalassistance. Supported by the Deutsche Forschungsgemein-schaft, SFB 415 (project A12).
REFERENCES
1. Dirnagl, U., Iadecola, C., and Moskowitz, M. A. (1999) Pathobi-ology of ischaemic stroke: an integrated view. Trends Neurosci.22, 391–397
2. Lee, J. M., Grabb, M. C., Zipfel, G. J., and Choi, D. W. (2000)Brain tissue responses to ischemia. J. Clin. Invest. 106, 723–731
3. Iadecola, C., and Alexander, M. (2001) Cerebral ischemia andinflammation. Curr. Opin. Neurol. 14, 89–94
4. Chandrasekharan, N. V., Dai, H., Roos, K. L., Evanson, N. K.,Tomsik, J., Elton, T. S., and Simmons, D. L. (2002) COX-3, acyclooxygenase-1 variant inhibited by acetaminophen and otheranalgesic/antipyretic drugs: cloning, structure, and expression.Proc. Natl. Acad. Sci. U. S. A. 99, 13926–13931
5. Phillis, J. W., and O’Regan, M. H. (2003) The role of phospho-lipases, cyclooxygenases, and lipoxygenases in cerebral isch-emic/traumatic injuries. Crypt. Rev. Neurobiol. 15, 61–90
6. Graham, S. H., and Hickey, R. W. (2003) Cyclooxygenases incentral nervous system diseases: a special role for cyclooxygen-ase 2 in neuronal cell death. Arch. Neurol. 60, 628–630
7. Miettinen, S., Fusco, F. R., Yrjanheikki, J., Keinanen, R., Hir-vonen, T., Roivainen, R., Narhi, M., Hokfelt, T., and Koistinaho,J. (1997) Spreading depression and focal brain ischemia inducecyclooxygenase-2 in cortical neurons through N-methyl-D-aspar-tic acid-receptors and phospholipase A2. Proc. Natl. Acad. Sci.U. S. A. 94, 6500–6505
8. Pepicelli, O., Fedele, E., Bonanno, G., Raiteri, M., Ajmone-Cat,M. A., Greco, A., Levi, G., and Minghetti, L. (2002) In vivoactivation of N-methyl-D-aspartate receptors in the rat hip-pocampus increases prostaglandin E(2) extracellular levels andtriggers lipid peroxidation through cyclooxygenase-mediatedmechanisms. J. Neurochem. 81, 1028–1034
9. Nogawa, S., Zhang, F., Ross, M. E., and Iadecola, C. (1997)Cyclo-oxygenase-2 gene expression in neurons contributes toischemic brain damage. J. Neurosci. 17, 2746–2755
10. Candelario-Jalil, E., Mhadu, N. H., Gonzalez-Falcon, A., Garcia-Cabrera, M., Munoz, E., Leon, O. S., and Fiebich, B. L. (2005)Effects of the cyclooxygenase-2 inhibitor nimesulide on cerebralinfarction and neurological deficits induced by permanentmiddle cerebral artery occlusion in the rat. J. Neuroinflammation2, 3
11. Iadecola, C., Niwa, K., Nogawa, S., Zhao, X., Nagayama, M.,Araki, E., Morham, S., and Ross, M. E. (2001) Reduced suscep-tibility to ischemic brain injury and N-methyl-D-aspartate-medi-ated neurotoxicity in cyclooxygenase-2-deficient mice. Proc. Natl.Acad. Sci. U. S. A. 98, 1294–1299
12. Sasaki, T., Kitagawa, K., Yamagata, K., Takemiya, T., Tanaka, S.,Omura-Matsuoka, E., Sugiura, S., Matsumoto, M., and Hori, M.(2004) Amelioration of hippocampal neuronal damage aftertransient forebrain ischemia in cyclooxygenase-2-deficient mice.J. Cereb. Blood Flow Metab. 24, 107–113
13. Schwab, J. M., and Schluesener, H. J. (2003) Cyclooxygenasesand central nervous system inflammation: conceptual neglect ofcyclooxygenase 1. Arch Neurol. 60, 630–632
14. Ricote, M., Li, A. C., Willson, T. M., Kelly, C. J., and Glass, C. K.(1998) The peroxisome proliferator-activated receptor-gammais a negative regulator of macrophage activation. Nature 391,79–82
15. Jiang, C., Ting, A. T., and Seed, B. (1998) PPAR-gamma agonistsinhibit production of monocyte inflammatory cytokines. Nature391, 82–86
16. Kim, E. J., Kwon, K. J., Park, J. Y., Lee, S. H., Moon, C. H., andBaik, E. J. (2002) Effects of peroxisome proliferator-activatedreceptor agonists on LPS-induced neuronal death in mixedcortical neurons: associated with iNOS and COX-2. Brain Res.941, 1–10
17. Heneka, M. T., Feinstein, D. L., Galea, E., Gleichmann, M.,Wullner, U., and Klockgether, T. (1999) Peroxisome prolifera-tor-activated receptor gamma agonists protect cerebellar gran-ule cells from cytokine-induced apoptotic cell death by inhibi-tion of inducible nitric oxide synthase. J. Neuroimmunol. 100,156–168
18. Heneka, M. T., Klockgether, T., and Feinstein, D. L. (2000)Peroxisome proliferator-activated receptor-gamma ligands re-duce neuronal inducible nitric oxide synthase expression andcell death in vivo. J. Neurosci. 20, 6862–6867
19. Shiraz, T., Inoue, I., Araki, N., Asano, Y., Sawada, M., Fuquay, D.,Nagoya, H., and Greenberg, J. H. (2005) A peroxisome prolif-erator-activated receptor-gamma agonist reduces infarct size intransient but not in permanent ischemia. Stroke 36, 353–359
20. Sundararajan, S., Gamboa, J. L., Victor, N. A., Wanderi, E. W.,Lust, W. D., and Landreth, G. E. (2005) Peroxisome prolifera-tor-activated receptor-gamma ligands reduce inflammation andinfarction size in transient focal ischemia. Neuroscience 130,685–696
21. Zhao, Y., Patzer, A., Gohlke, P., Herdegen, T., and Culman, J.(2005) The intracerebral application of the PPARgamma-ligandpioglitazone confers neuroprotection against focal ischaemia inthe rat brain. Eur. J. Neurosci. 22, 278–282
22. Dai, W. J., Funk, A., Herdegen, T., Unger, T., and Culman, J.(1999) Blockade of central angiotensin AT(1) receptors im-proves neurological outcome and reduces expression of AP-1transcription factors after focal brain ischemia in rats. Stroke 30,2391–2398; discussion 2398–2399
23. Schmid-Elsaesser, R., Zausinger, S., Hungerhuber, E., Baeth-mann, A., and Reulen, H. J. (1998) A critical reevaluation of theintraluminal thread model of focal cerebral ischemia: evidenceof inadvertent premature reperfusion and subarachnoid hem-orrhage in rats by laser-Doppler flowmetry. Stroke 29, 2162–2170
24. Lou, M., Blume, A., Zhao, Y., Gohlke, P., Deuschl, G., Herd-egen, T., and Culman, J. (2004) Sustained blockade of brainAT1 receptors before and after focal cerebral ischemia alleviatesneurologic deficits and reduces neuronal injury, apoptosis, andinflammatory responses in the rat. J. Cereb. Blood Flow Metab. 24,536–547
25. Xu, J., Culman, J., Blume, A., Brecht, S., and Gohlke, P. (2003)Chronic treatment with a low dose of lithium protects the brainagainst ischemic injury by reducing apoptotic death. Stroke 34,1287–1292
26. Brewer, G. J. (1995) Serum-free B27/neurobasal medium sup-ports differentiated growth of neurons from the striatum,substantia nigra, septum, cerebral cortex, cerebellum, anddentate gyrus. J. Neurosci. Res. 42, 674–683
27. Cullingford, T. E., Bhakoo, K., Peuchen, S., Dolphin, C. T.,Patel, R., and Clark, J. B. (1998) Distribution of mRNAsencoding the peroxisome proliferator-activated receptor alpha,beta, and gamma and the retinoid X receptor alpha, beta, andgamma in rat central nervous system. J. Neurochem. 70, 1366–1375
28. Bernardo, A., Levi, G., and Minghetti, L. (2000) Role of theperoxisome proliferator-activated receptor-gamma (PPAR-gamma) and its natural ligand 15-deoxy-Delta12, 14-prostaglan-din J2 in the regulation of microglial functions. Eur. J. Neurosci.12, 2215–2223
29. Cristiano, L., Bernardo, A., and Ceru, M. P. (2001) Peroxisomeproliferator-activated receptors (PPARs) and peroxisomes in ratcortical and cerebellar astrocytes. J. Neurocytol. 30, 671–683
30. Moreno, S., Farioli-Vecchioli, S., and Ceru, M. P. (2004) Immu-nolocalization of peroxisome proliferator-activated receptorsand retinoid X receptors in the adult rat CNS. Neuroscience 123,131–145
31. Cimini, A., Benedetti, E., Cristiano, L., Sebastiani, P., D’Amico,M. A., D’Angelo, B., and Di Loreto, S. (2005) Expression ofperoxisome proliferator-activated receptors (PPARs) and reti-noic acid receptors (RXRs) in rat cortical neurons. Neuroscience130, 325–337
1174 Vol. 20 June 2006 ZHAO ET AL.The FASEB Journal
32. Sundararajan, S., and Landreth, G. E. (2004) Antiinflammatoryproperties of PPARgamma agonists following ischemia. DrugNews Perspect. 17, 229–236
33. Desvergne, B., and Wahli, W. (1999) Peroxisome proliferator-activated receptors: nuclear control of metabolism. Endocr. Rev.20, 649–688
34. Yamagata, K., Andreasson, K. I., Kaufmann, W. E., Barnes, C. A.,and Worley, P. F. (1993) Expression of a mitogen-induciblecyclooxygenase in brain neurons: regulation by synaptic activityand glucocorticoids. Neuron 11, 371–386
35. Collaco-Moraes, Y., Aspey, B., Harrison, M., and de Belleroche,J. (1996) Cyclo-oxygenase-2 messenger RNA induction in focalcerebral ischemia. J. Cereb. Blood Flow Metab. 16, 1366–1372
36. Yokota, C., Kaji, T., Kuge, Y., Inoue, H., Tamaki, N., andMinematsu, K. (2004) Temporal and topographic profiles ofcyclooxygenase-2 expression during 24 h of focal brain ischemiain rats. Neurosci. Lett. 357, 219–222
37. Iadecola, C., Forster, C., Nogawa, S., Clark, H. B., and Ross, M. E.(1999) Cyclooxygenase-2 immunoreactivity in the human brainfollowing cerebral ischemia. Acta Neuropathol. (Berlin) 98, 9–14
38. Tomimoto, H., Shibata, M., Ihara, M., Akiguchi, I., Ohtani, R.,and Budka, H. (2002) A comparative study on the expression ofcyclooxygenase and 5-lipoxygenase during cerebral ischemia inhumans. Acta Neuropathol. (Berlin) 104, 601–607
39. Dobrian, A. D., Schriver, S. D., Khraibi, A. A., and Prewitt, R. L.(2004) Pioglitazone prevents hypertension and reduces oxida-tive stress in diet-induced obesity. Hypertension 43, 48–56
40. Hewett, S. J., Uliasz, T. F., Vidwans, A. S., and Hewett, J. A.(2000) Cyclooxygenase-2 contributes to N-methyl-D-aspartate-mediated neuronal cell death in primary cortical cell culture.J. Pharmacol. Exp. Ther. 293, 417–425
41. Araki, E., Forster, C., Dubinsky, J. M., Ross, M. E., and Iadecola,C. (2001) Cyclooxygenase-2 inhibitor ns-398 protects neuronalcultures from lipopolysaccharide-induced neurotoxicity. Stroke32, 2370–2375
42. Nakayama, M., Uchimura, K., Zhu, R. L., Nagayama, T., Rose,M. E., Stetler, R. A., Isakson, P. C., Chen, J., and Graham, S. H.(1998) Cyclooxygenase-2 inhibition prevents delayed death ofCA1 hippocampal neurons following global ischemia. Proc. Natl.Acad. Sci. U. S. A. 95, 10954–10959
43. Koistinaho, J., Koponen, S., and Chan, P. H. (1999) Expressionof cyclooxygenase-2 mRNA after global ischemia is regulated byAMPA receptors and glucocorticoids. Stroke 30, 1900–1905;discussion 1905–1906
44. Chawla, A., Barak, Y., Nagy, L., Liao, D., Tontonoz, P., andEvans, R. M. (2001) PPAR-gamma dependent and independenteffects on macrophage-gene expression in lipid metabolism andinflammation. Nat. Med. 7, 48–52
45. Smith, S. A., Monteith, G. R., Holman, N. A., Robinson, J. A.,May, F. J., and Roberts-Thomson, S. J. (2003) Effects of perox-isome proliferator-activated receptor gamma ligands ciglitazoneand 15-deoxy-delta 12,14-prostaglandin J2 on rat cultured cere-bellar granule neuronal viability. J. Neurosci. Res. 72, 747–755
46. Drew, P. D., Storer, P. D., Xu, J., and Chavis, J. A. (2005)Hormone regulation of microglial cell activation: relevance tomultiple sclerosis. Brain Res. Rev. 48, 322–327
47. Kielian, T., and Drew, P. D. (2003) Effects of peroxisomeproliferator-activated receptor-gamma agonists on central ner-vous system inflammation. J. Neurosci. Res. 71, 315–325
48. Luna-Medina, R., Cortes-Canteli, M., Alonso, M., Santos, A.,Martinez, A., and Perez-Castillo, A. (2005) Regulation of inflam-
matory response in neural cells in vitro by thiadiazolidinonesderivatives through peroxisome proliferator-activated receptorgamma activation. J. Biol. Chem. 280, 21453–21462
49. Zhao, Y., Foryst-Ludwig, A., Bruemmer, D., Culman, J., Bader,M., Unger, T., and Kintscher, U. (2005) Angiotensin II inducesperoxisome proliferator-activated receptor gamma in PC12Wcells via angiotensin type 2 receptor activation. J. Neurochem. 94,1395–1401
50. Niwa, K., Haensel, C., Ross, M. E., and Iadecola, C. (2001)Cyclooxygenase-1 participates in selected vasodilator responsesof the cerebral circulation. Circ. Res. 88, 600–608
51. Iadecola, C., Sugimoto, K., Niwa, K., Kazama, K., and Ross, M. E.(2001) Increased susceptibility to ischemic brain injury incyclooxigenase-1-deficient mice. J. Cereb. Blood Flow Metab. 21,1436–1441
52. Lin, H., Lin, T. N., Cheung, W. M., Nian, G. M., Tseng, P. H.,Chen, S. F., Chen, J. J., Shyue, S. K., Liou, J. Y., Wu, C. W., andWu, K. K. (2002) Cyclooxygenase-1 and bicistronic cyclooxygen-ase-1/prostacyclin synthase gene transfer protect against isch-emic cerebral infarction. Circulation 105, 1962–1969
53. Candelario-Jalil, E., Gonzalez-Falcon, A., Garcia-Cabrera, M.,Alvarez, D., Al-Dalain, S., Martinez, G., Leon, O. S., andSpringer, J. E. (2003) Assessment of the relative contribution ofCOX-1 and COX-2 isoforms to ischemia-induced oxidativedamage and neurodegeneration following transient global ce-rebral ischemia. J. Neurochem. 86, 545–555
54. Koizumi, J., Yoshida, Y., Nakazawa, T., and Ooneda, G. (1986)Experimental studies of ischemic brain edema, I: a newexperimental model of cerebral embolism in rats in whichrecirculation can be introduced in the ischemic area. Jon. J.Stroke 8, 1– 8
55. Hosomi, N., Ban, C.R., Nay, T., Takahashi, T., GAO, P., Song,X-y. R, and Kohno, M. (2005) Tumor necrosis factor-� neutral-ization reduced cerebral edema through inhibition of matrixmetalloproteinase production after transient focal cerebral isch-emia. J. Cereb. Blood Flow Metab. 25, 959–967
56. Subbaramaiah K., Lin, D. T., Hart, J. C., and Dannenberg, A. J.(2001) Peroxisome proliferator-activated receptor � ligandssuppress the transcriptional activation of cyclooxygenase-2.J. Biol. Chem. 276, 1440–12448
57. Han, S., Inoue, H., Flowers, L. C., and Sidell, N. (2003) Controlof COX-2 gene expression through peroxisome proliferator-activated receptor � in human cervical cancer cells. Clin. CancerRes. 9, 4627–4635
58. Mendes, M., and LaPointe, M. C. (2003) PPAR� inhibition ofcyclooxygenase-2, PGE2 synthase, and inducible nitric oxidesynthase in cardiac myocytes. Hypertension 42, 844–850
59. Vasudevan, A. R., and Balasubramanyam, A. (2004) Thiazo-lidinediones: a review of their mechanisms of insulin sensitisa-tion, therapeutic potential, clinical efficacy, and tolerability.Diabetes Technol. Therap. 6, 850–863
60. Dormandy, J. A, Charbonnel, B., Eckland, D. J. A., Erdmann, E,Massi-Benedetti, M., Moules, I. K., Skene, A. M., Tan, M. H.,Lefebvre, P. J., Murray, G. D., et al. (2005) Secondary preven-tion of macrovascular events in patients with type 2 diabetes inthe PROactive Study (PROspective pioglitAzone Clinical Trial inmacroVascular Events): a randomised controlled trial. Lancet366, 1279–1289
Received for publication October 6, 2005.Accepted for publication February 10, 2006.
1175NEURONAL PPAR� ATTENUATES COX-2 AFTER CEREBRAL ISCHEMIA
