Cyclooxygenase-2 mediates the sensitizing effects of systemic IL-1-beta on excitotoxic brain lesions in newborn mice

wwwelseviercomlocateynbdi

Neurobiology of Disease 25 (2007) 496ndash505

Cyclooxygenase-2 mediates the sensitizing effects of systemicIL-1-beta on excitotoxic brain lesions in newborn mice

Geacuteraldine Favraisab Leslie Schwendimannab Pierre Gressensabc and Vincent Leliegravevreab

aInserm U676 Paris FrancebUniversiteacute Paris 7 Faculteacute de Meacutedecine Denis Diderot IFR02 Paris FrancecAP HP Hocircpital Robert Debreacute Service de Neurologie Peacutediatrique Paris France

Received 20 July 2006 revised 23 October 2006 accepted 29 October 2006Available online 12 December 2006

Epidemiological and experimental data implicate maternalndashfetalinfection and an associated increase in circulating cytokines in theetiology of cerebral palsy We have previously shown that pretreatmentof newborn mice with systemic interleukin-1-beta exacerbatesibotenate-induced excitotoxic brain lesions Such lesions are consistentwith those observed in cerebral palsy The present study builds on thismurine model to assess the role of cyclooxygenase in interleukin-1-beta-induced brain toxicity Pups pretreated with interleukin-1-betadeveloped greater ibotenate-induced brain damage than controls aneffect blocked by the co-administration of nimesulide (cyclooxygenase-2 inhibitor) or indomethacin (cyclooxygenase-1 and -2 inhibitor)Cyclooxygenase inhibitor administration prevented the interleukin-1-beta-induced increase in the production of brain prostaglandin E2 (acyclooxygenase metabolite) and changes in the expression of braininterleukin-6 interleukin-18 tumor necrosis factor-alpha and brain-derived neurotrophic factor It also stimulated the expression of braininterleukin-10 Our data suggest that the sensitizing effects ofcirculating inflammatory cytokines on the brain are mediated by theinducible isoform cyclooxygenase-2 which generates excess prosta-glandin E2 Some of these deleterious effects could involve anautocrineparacrine loop leading to a disruption of the balancebetween pro- and anti-inflammatory cytokines in the braincopy 2006 Elsevier Inc All rights reserved

Keywords Cerebral palsy Pro-inflammatory cytokine ExcitotoxicityIbotenate Nimesulide Indomethacin

Introduction

Despite a reduction in neonatal mortality and morbidity in thelast 40 years rates of cerebral palsy (CP) remain significant inWestern countries (Hagberg et al 1996 Himmelmann et al2005) Recently hypothesized etiologies of CP have gone beyondhypoxicndashischemic mechanisms to include multiple preconceptional

Corresponding author Inserm U676 Hocircpital Robert Debreacute 48 BlvdSeacuterurier F-75019 Paris France Fax +33 1 40 03 1995

E-mail address gressensrdebreinsermfr (P Gressens)Available online on ScienceDirect (wwwsciencedirectcom)

0969-9961$ - see front matter copy 2006 Elsevier Inc All rights reserveddoi101016jnbd200610012

and prenatal factors such as hypoxiaperfusion failure geneticcomponents growth-factor deficiency and maternal infection andinflammation leading to the production of excess cytokines(Nelson and Willoughby 2000 Volpe 2001 Dammann et al2002 Gressens et al 2002)

The potential deleterious role of perinatal inflammation hasbeen proposed for both preterm and full term neonates at risk forthe development of brain lesions and CP (Murphy et al 1995Zupan et al 1996 Nelson and Willoughby 2000 Volpe 2001Dammann et al 2002 Gressens et al 2002) According to themechanism hypothesized in utero infection andor inflammationinduce a transplacental inflammatory response associated with theproduction of excess circulating cytokines capable of harming thedeveloping brain Findings from several studies support anassociation between maternalndashfetal infection circulating cytokines(such as interleukin or IL-1-beta IL-6 and tumor necrosis factor orTNF-alpha) and periventricular white matter damage (PWMD) inpreterm infants (Romero et al 1990 Greig et al 1993 Singh etal 1996 Yoon et al 1996 Yoon et al 1997 Martinez et al1998) In addition a striking association between increased levelsof perinatal circulating cytokines including IL-1-beta IL-6 IL-8IL-9 and TNF-alpha and the subsequent occurrence of CP in full-term infants has been reported (Nelson et al 1998)

Using a murine model of neonatal excitotoxic brain lesionsbased on the intracerebral administration of ibotenate a glutamateanalog acting on N-methyl-D-aspartate (NMDA) and metabotropicreceptors we have previously shown that pups pretreated with IL-1-beta IL-6 or TNF-alpha develop significantly greater ibotenate-induced cortical and white matter damage than controls (Dom-mergues et al 2000) The precise molecular mechanisms by whichcirculating mediators of inflammation have a deleterious effect onperinatal brain lesions remain a matter for debate (Hagberg andMallard 2005) Circulating cytokines do not seem to cross theintact bloodndashbrain barrier (BBB) However three alternativepathways have been proposed to link serum cytokine levels withbrain lesion sizes Firstly circulating cytokines could alter thepermeability of the BBB to inflammatory mediators and cellsSecondly circulating cytokines could act directly on parts of the

497G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

brain lacking the BBB such as the circumventricular organsmeninges and choroid plexus or as demonstrated in the adult brainindirectly through the activation of the vagal nerve Thirdlycytokine effects could be mediated by cyclooxygenase (Cox)located on the BBB In particular cytokines could activate theinducible isoform Cox-2 to enhance the local production ofprostaglandin E2 (PGE2) that could have deleterious effects on thedeveloping brain Some of these deleterious effects could involvean autocrineparacrine loop leading to the excess production ofinflammatory cytokines by brain cells

In the present study using the murine model of neonatalexcitotoxic brain lesions described above we explored themechanisms by which systemically injected inflammatory cyto-kines sensitize the developing brain We investigated the impact ofsystemically administered IL-1-beta on the production of cytokinesby brain cells focusing on the potential role of Cox-2 in thisprocess To do this we used indomethacin and nimesulide twonon-steroidal anti-inflammatory drugs (NSAIDs) that inhibit Coxactivity Our data support the previously reported epidemiologicalassociation between high levels of circulating pro-inflammatorycytokines and an elevated risk of developing CP andor PWMD

Materials and methods

Experimental protocols were approved by the institutionalreview committee and meet Inserm guidelines as well as the Guidefor the Care and use of Laboratory Animals as promulgated andadopted by the National Institutes of Health USA

Animals and drug administration

Pregnant Swiss mice were carefully monitored twice a day todetermine the exact time of delivery (P0) Newborn pups were

Table 1Sequences of primers used in the quantitative PCR

Genes Primer sequences (sense and reverse respectively)

Cox-2 5prime-TGCAGAATTGAAAGCCCTCTA-3primeOH5prime-CCCCAAAGATAGCATCTGGAC-3primeOH

IL-1-beta 5prime-GAAGATGGAAAAGCGGTTTG-3primeOH5prime-GTACCAGTTGGGGAACTCTGC-3primeOH

IL-18 5prime-TTCGTTGACAAAAGACAGCC-3primeOH5prime-TATCAGTCTGGTCTGGGGTTC-3primeOH

IL-6 5prime-ACACATGTTCTCTGGGAAATC-3primeOH5prime-AGTGCATCATCGTTGTTCATA-3primeOH

IL-10 5prime-CTCCCCTGTGAAAATAAGAGC-3primeOH5prime-GCCTTGTAGACACCTTGGTC-3primeOH

TNF-alpha 5prime-CCGATGGGTTGTACCTTGTCT-3primeOH5prime-GTGGGTGAGGAGCACGTAGT-3primeOH

BDNF 5prime-GGACTCTGGAGAGCGTGAATG-3primeOH5prime-GCTCTTCGATGACGTGCTCA-3primeOH(amplicon common to all the describedsplice variants for BDNF)

Beta2-microglobulin 5prime-CCGGCTTGTATGCTATC-3primeOH5prime-AGTTCATGTTCGGCTTC-3primeOH

GAPDH 5prime-GGCCTTCCGTGTTCCTAC-3primeOH5prime-TGTCATCATACTTGGCAGGTT-3primeOH

Beta-glucuronidase 5prime-CCTGCGGTTGTGATGTGGTCT-3primeOH5prime-GGGTGAGGTCCAGGGCTTTG-3primeOH

HPRT 5prime-TGGTGAAAAGGACCTCTCGAA-3primeOH5prime-TCAAGGGCATATCCAACAACA-3primeOH

injected intraperitoneally (ip) twice a day (between 8 and 10 AMand again between 6 and 8 PM) on days P1 to P4 and once(between 8 and 10 AM) on P5 with 5 μl of PBS or DMSO(controls) or IL-1-beta (Serotec 40 to 160 ng diluted in 5 μlPBS) Certain animals from the three groups described above wereco-injected with the Cox-1 inhibitor nimesulide (Sigma 001 to1 mgkg diluted in DMSO) or the Cox-1 and -2 inhibitorindomethacin (Sigma 00125 to 0125 mgkg diluted in PBS)(Riendeau et al 1997) A control group was injected withnimesulide alone (1 mgkg)

Excitotoxic brain lesion and lesion size determination

At P5 2 h after the last ip injection 10 μg of ibotenate(Sigma) diluted in PBS containing 002 acetic acid wasinjected intracerebrally to 7 to 16 pups in each experimentalgroup as previously described (Marret et al 1995 Gressens etal 1997 Dommergues et al 2000) Injections were performedunder a warming lamp on pups anesthetized with isofluraneusing a 26-gauge needle and a 50-μl Hamilton syringe mountedon a calibrated microdispenser The needle was inserted to a depthof 2 mm from the external surface of the skin of the scalp in thefrontoparietal area of the right hemisphere 1 mm from the midlinein the mediolateral plane and 15 mm anterior to the junctionbetween the sagittal and lambdoid sutures in the rostrocaudalplane Two 1 μl boluses each containing 5 μg ibotenate wereinjected at an interval of 30 s In all cases the tip of the needlereached the periventricular white matter as verified histologicallypost-mortem After injection pups were returned to their damsuntil sacrifice

The sex of the pups was determined by visual inspection atbirth and was confirmed by post-mortem anatomical inspectionand PCR for SRY

Amplicon sizes (pb) Spanning regions NBCI access

94 1522ndash1616 NM01198

85 697ndash782 NM008361

80 330ndash410 NM008360

82 197ndash279 NM031168

82 463ndash545 NM010548

76 548ndash624 NM013693

72 219ndash291 MMBDNFA

86 99ndash185 MM2BMR

80 1093ndash1173 XM111622

108 1321ndash1429 NM010368

90 552ndash642 BC00468

498 G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

Five days after intracerebral injection of ibotenate (ie onP10) pups were killed by decapitation and their brains fixed informaldehyde for 5 days After embedding in paraffin brainswere serially sectioned at 15 μm intervals from the frontal tothe occipital pole and every third section stained with cresylviolet This permits an accurate and reproducible determinationof the maximal diameter of the lesion in the sagittalfrontoparietal axis as a function of the number of sectionscontaining the lesion and the thickness per section (Marret etal 1995 Gressens et al 1997 Husson et al 2002 Husson etal 2005) We also used this measure as an index of lesionvolume Throughout the study lesion size was determined by

Fig 1 Systemic administration of IL-1-beta (40 ng) exacerbates excitotoxic braiviolet-stained sections showing brain lesions induced by ibotenate injected on Pvehicle (A) IL-1-beta+vehicle (B) or IL-1-beta+1 mgkg nimesulide (C) showinlesion () LV lateral ventricle Scale bar=40 μm (DndashE) Quantification of thTreatment group is indicated on the X axis (Nime nimesulide Indo indomethacthe sagittal fronto-occipital axis Numbers in brackets are the number (n) ofsignificant differences from black (D) or hatched (E) bars plt005 plt001

two independent investigators blind with respect to thetreatment status of the animal from which tissue had beentaken

Measurement of serum levels of IL-1-beta by ELISA

Five animals in each experimental group were injected withIL-1-beta or vehicle in combination with nimesulide (1 mgkg)or vehicle as described above These animals were not injectedwith ibotenate At P5 pups were anesthetized with isofluraneand underwent a thoracotomy The pulmonary arteries weresectioned and the blood that filled the thoracic cavity was

n lesions and Cox inhibitors abolish this deleterious effect (AndashC) Cresyl5 and studied on P10 Brains from a pup treated between P1 and P5 withg typical neuronal loss in layers IIndashVI (arrow) and the white matter cystice size of brain lesions induced by ibotenate on P5 and studied on P10in IL-1 IL-1-beta) Bars represent mean lengthplusmnSEM of the lesion alonganimals used in each experimental group Asterisks indicate statisticallyplt0001 in ANOVA with Bonferronis multiple comparison tests

Fig 2 Systemic administration of IL-1-beta (40 ng) increases circulatinglevels of IL-1-beta (A) ELISA quantification of serum IL-1-betaconcentration within the first 12 h following ip administration of IL-1-beta on P5 (B) ELISA quantification of serum IL-1-beta concentration onP5 6 h after treatment indicated on the X axis Bars represent meanconcentrationplusmnSEM of IL-1-beta Asterisks indicate statistically significantdifferences from black () or hatched (sect) bars sectplt005 plt001 usingANOVAwith Bonferronis multiple comparison tests

499G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

carefully aspirated with a syringe About 150 μl of blood wascollected for each pup Blood samples were allowed to sedimentovernight at 4degC then centrifuged at 2000timesg and the seracollected for long-term storage at minus80degC The level of IL-1-betain the samples was estimated using an ELISA kit (RampDSystems) according to the manufacturerrsquos recommendations Inbrief 50 μl of standard control buffer or sample was combinedwith 50 μl of assay buffer in IL-1-beta antibody-coated wells onthe ELISA plate and incubated at room temperature for 2 hWells were carefully washed five times prior to the addition of100 μl of the appropriate horseradish peroxidase (HRP)conjugate and incubated for 2 more hours After a secondwash cycle 100 μl of hydrogen peroxidetetramethylbenzidinesubstrate solution was added per well and the plate incubatedfor 30 min at room temperature in the dark The reaction wasstopped by addition of the hydrochloric acid solution providedin the kit The absorbance at 450 nm was measured with amicroreader (Spectracount Packard) with wavelength correctionat 570 nm

RNA extraction and quantification of gene expression by real-timePCR

Six animals in each experimental group were injected withIL-1-beta or vehicle in combination with nimesulide (1 mgkg)or vehicle as described above These animals were not injectedwith ibotenate Pups were sacrificed by decapitation on P5Total RNA was extracted according to a protocol derived fromthe original procedure of Chomczynski and Sacchi (1987)consisting of two independent total RNA extractions separatedby a DNAseI treatment (DNA-freetrade kit Ambion) aspreviously described in detail (Lelievre et al 2002) RNAquality and concentration were assessed by spectrophotometryand capillary electrophoresis on RNAstdsens biochips usingExperiontrade apparatus according to the manufacturerrsquos instruc-tions (Biorad) Total RNA (600 ng) was subjected to reversetranscription using the Iscripttrade kit from Biorad Negativecontrols (samples in which reverse transcriptase was omitted)were individually amplified by PCR using the different primersets used in the present study to ensure the absence ofgenomic DNA contamination To specifically amplify mRNAencoding various mouse proteins we designed the specificprimer sets (sense and antisense respectively) using Oligo60and M-fold software for Cox-2 IL-1-beta IL-6 IL-10 IL-18TNF-alpha and BDNF (Zuker 2003) Primer sequences aregiven in Table 1

To standardize the experiments four classic housekeepinggenes were tested ie beta2-microglobulin glyceraldehyde-3-phosphate dehydrogenase (GAPDH) beta-glucuronidase andhypoxanthine guanine phosphoribosyl transferase (HPRT) Theprimer sequences are given in Table 1

Preliminary experiments showed that GAPDH levels re-mained highly stable among the different samples and treatmentconditions This housekeeping gene was therefore chosen tostandardize all the quantitative experiments presented here Real-time PCR was set up using SYBR green-containing supermixtradefrom Biorad for 45 cycles of a three-step procedure including a20-second denaturation step at 96degC a 20-second annealing stepat 60degC followed by a 20-second extension step at 72degCAmplification specificity was assessed by melting curve andsubsequent amplicon sequencing after subcloning into the

TOPO-II vector (Invitrogen) Quantification was carried outusing standard curves made from serial dilutions of controlRNA sample or of the corresponding cDNA cloned into thePCRII vector The differences between samples were calculatedon the basis of the specific ratios (gene of interesthousekeepinggene)

Measurement of cerebral prostaglandin PGE2 production byradioimmunoassay

Animals (6ndash9 in each experimental group) were injected withIL-1-beta or vehicle in combination with indomethacin(0125 mgkg) nimesulide (1 mgkg) or vehicle as describedabove These animals were not injected with ibotenate Pupswere sacrificed by decapitation on P5 Brain samples wereextracted and assayed for PGE2 concentration using a com-mercially available radioimmunoassay kit (NEK020 from NENPerkin Elmer) Following the manufacturerrsquos instructions anindomethacin-containing Tris buffer was used to prepare thebrain extracts to prevent non-specific PGE2 production during

500 G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

this process Samples (8 per group) were first diluted to fitwithin the range of the PGE2 standard curve and then measuredin triplicate In every sample protein content was determinedusing the Bradford method and data were expressed as thePGE2 content per mg of total protein

Statistical analyses

Quantitative data are expressed as meanplusmnSEM for eachtreatment group Results were compared using Studentrsquos t-tests oranalysis of variance (ANOVA) with Bonferronirsquos multiple compar-ison of means test (GraphPad Prism version 401 for WindowsGraphPad Software)

Fig 3 Effects of systemic administration of IL-1-beta (40 ng) and Cox inhibitors oquantification of Cox-2 mRNA in brain extracts of P5 mice treated between P1 anratiosplusmnSEM Asterisks indicate statistically significant differences from black ocomparison tests (D) Radioimmunoassay quantification of PGE2 in brain extracts ocontentplusmnSEM Asterisks indicate statistically significant differences from black ()comparison tests

Results

Effects of IL-1-beta on ibotenate-induced lesions reversal by Coxinhibitors

Control pups injected ip with vehicle between P1 and P5 andintracerebrally with ibotenate on P5 developed cortical lesions andperiventricular white matter cysts (Figs 1A and D) The corticallesion was typical of ibotenate-induced lesions with severeneuronal loss in all neocortical layers and the almost completedisappearance of neuronal cell bodies along the axis of ibotenateinjection There was no significant effect of gender on the size ofthe lesions (Fig 1D)

n brain Cox-2 expression and brain PGE2 production (AndashC) Real-time PCRd P5 as indicated on the X axis Data are presented as mean Cox-2GAPDHr hatched (sect) bars sectsectplt001 in ANOVA with Bonferronis multiplef P5 mice treated as indicated on the X axis Data are presented as mean PGE2or hatched (sect) bars sectsectsectplt0001 in ANOVAwith Bonferronis multiple

501G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

IL-1-beta (40 ng) exposure between P1 and P5 significantlyincreased the size of both cortical plate lesions (up to 112) and theunderlying white matter lesions (up to 270) (Figs 1B and D)Again there was no significant effect of gender Accordingly allsubsequent analyses were carried out using data from male andfemale pups combined

Co-injection of nimesulide with vehicle induced a dose-dependent inhibition of the sensitizing effect of IL-1-beta onexcitotoxic brain lesions (Figs 1C and E) Nimesulide alone had nodetectable effect on ibotenate-induced lesion (Fig 1E) Co-injection

Fig 4 Effects of systemic administration of IL-1-beta and nimesulide on brain cy(CndashD) IL-10 (EndashF) IL-18 (GndashH) TNF-alpha (IndashJ) and BDNF (KndashL)mRNA in brmean cytokineGAPDH ratiosplusmnSEM Asterisks indicate statistically significantplt0001 in ANOVAwith Bonferronis multiple comparison tests

of indomethacin mimicked the protective effects of nimesulide onIL-1-beta sensitization (Fig 1E)

Circulating levels of IL-1-beta effects of exogenous IL-1-beta andnimesulide

In order to evaluate the impact of exogenous IL-1-beta onsystemic inflammation we measured blood levels of IL-1-betafollowing ip injection of exogenous IL-1-beta (40 ng) As shownin Fig 2A there was a peak of circulating IL-1-beta 1 h after

tokine expression Real-time PCR quantification of IL-1-beta (AndashB) IL-6ain extracts of P5mice treated as indicated on theX axis Data are presented asdifferences from black (x) or hatched (sect) bars plt005 sectsectplt001

Fig 4 (continued)

502 G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

injection followed by a rapid decrease IL-1-beta levels were backto normal within 6 h after ip administration These results suggestthat in these conditions exogenously administered IL-1-beta iseliminated within 6 h

Injection of nimesulide (1 mgkg) alone had no detectableeffect on circulating levels of IL-1-beta measured 6 h afterinjection (Fig 2B) In contrast the co-injection of nimesulideand IL-1-beta significantly increased blood levels of IL-1-beta(Fig 2B) Although the precise mechanism by which nimesulideenhances the half-life of exogenously administered IL-1-beta isunclear these results suggest that the neuroprotective effects ofCox inhibitors against the IL-1-beta-induced sensitization are notlinked to a peripheral degradation of exogenously administeredIL-1-beta

Cerebral Cox-2 expression and PGE2 production in response toIL-1-beta

In order to test the hypothesis that Cox-2 plays a role in thesensitizing effects of IL-1-beta on excitotoxic brain lesions and thatthe neuroprotective effects of Cox inhibitors in this model aremediated by the blockade of this Cox-2 we measured Cox-2expression and the Cox-2-mediated production of PGE2 in thebrain

Quantitative PCR analysis showed that IL-1-beta (40 ng) didnot significantly modify the expression of Cox-2 in the brain (Fig3A) Similarly administration of higher doses of IL-1-beta (80 or160 ng) had no effect on the expression of Cox-2 mRNA (data notshown) As previously described (Moalic et al 2001 Tanaka et

503G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

al 2002 Takeuchi et al 2004) the administration of nimesulidealone (1 mgkg) but not indomethacin (0125 mgkg) induced asignificant increase in Cox-2 expression (Figs 3BndashC) The co-injection of IL-1-beta and Cox inhibitors had no significant effecton Cox-2 expression when compared to the administration of IL-1-beta alone (Figs 3BndashC)

In contrast IL-1-beta (40 ng) treatment significantlyincreased PGE2 levels as demonstrated by RIA (Fig 3D) mostlikely reflecting an increase in Cox activity This IL-1-beta-induced increase in the production of PGE2 was blocked by co-treatment with indomethacin (0125 mgkg) or nimesulide(1 mgkg) while the administration of indomethacin ornimesulide alone had no detectable effect on PGE2 production(Fig 3D)

Cerebral expression of cytokines in response to IL-1-beta and theeffects of nimesulide

To test the hypothesis that systemically administered IL-1-betadisrupts the cerebral expression of various pro- or anti-inflamma-tory cytokines through Cox-2 activation we used quantitativePCR to measure mRNA levels in brain extracts for a panel ofcytokines

Treatment with IL-1-beta alone induced a dose-dependentincrease of the expression of IL-1-beta (Fig 4A) Treatment withIL-1-beta alone also increased the expression of IL-6 and TNF-alpha but this effect was not dose-dependent within the range oftested doses of IL-1-beta (Figs 4C and I) The effect of treatmentwith IL-1-beta alone on IL-18 and BDNF expression wasbiphasic with a decreased expression with 40 ng IL-1-beta andan increased expression with the highest dose (160 ng) of IL-1-beta (Figs 4G and K) Treatment with IL-1-beta alone had nodetectable effect on IL-10 expression (Fig 4E)

The administration of nimesulide alone (1 mgkg) signifi-cantly increased the cerebral expression of IL-10 and BDNFwhile it had no detectable effect on the other measured cytokines(Fig 4)

Co-treatment with IL-1-beta (40 ng) and nimesulideabolished the IL-1-beta-induced changes in the expression ofIL-6 IL-18 TNF-alpha and BDNF but failed to affect theincreased expression of IL-1-beta transcripts (Fig 4) Theincreased expression of IL-10 observed with nimesulide wasalso observed with co-treatment with nimesulide and IL-1-beta(Fig 4)

Discussion

The most salient finding of this study is that the deleteriouseffects of pre-treatment with systemic IL-1-beta on excitotoxicbrain lesions in newborn mice were totally abolished by the co-administration of nimesulide or indomethacin two Coxinhibitors In addition the IL-1-beta-induced increase in theproduction of brain PGE2 was totally abolished by the Coxinhibitor indomethacin The blockade of Cox activity alsoreversed some of the effects of the systemic administration ofIL-1-beta on brain production of various cytokines includingIL-6 IL-18 and TNF-alpha

Although gender has been shown to affect several parametersrelating to neonatal brain lesions (Hagberg et al 2004 Nijboeret al in press) we found that the IL-1-beta-induced exacerba-tion of excitotoxic brain lesions is gender-independent

Role of Cox in IL-1-beta-induced exacerbation of neonatalexcitotoxic brain lesions

Two separate findings support a key role for Cox in the IL-1-beta-induced exacerbation of excitotoxic brain lesions (i)nimesulide and indomethacin two Cox inhibitors completelyabrogated the effects of IL-1-beta on excitotoxic lesions (ii)despite the increase in Cox-2 mRNA expression by Cox inhibitoradministration indomethacin and nimesulide blocked the IL-1-beta-induced production of PGE2 a major product of Coxenzymatic activity

The increased expression of Cox-2 mRNA induced by Coxinhibitors has been previously reported in other models (Moalic etal 2001 Tanaka et al 2002 Takeuchi et al 2004) and mostlikely reflects a feedback loop aimed at compensating for reducedCox activity with Cox inhibitors acting at a post-transcriptionallevel

Indomethacin blocks both Cox-1 and Cox-2 while nimesulideis a specific Cox-2 inhibitor The fact that both drugs blocked theIL-1-beta sensitization argues in favor of a key role for theinducible isoform Cox-2 in the present model In addition to itseffects on Cox-2 nimesulide may also interfere with othermechanisms involved in inflammation such as the release ofhistamine from mast cells and basophils the formation ofhydroxyl-radicals and superoxide radicals by activated polymor-phonuclear neutrophils (PMNs) the phagocytosis of PMNsneutrophil adherence or the production of platelet activatingfactor from activated platelets (Rainsford et al 2001) Thepotential contribution of these effects of nimesulide in the presentmodel needs to be further evaluated

The excess production of PGE2 has been shown to havedeleterious effects on neural cells (Kawano et al 2006) and mightrepresent a downstream mechanism by which systemic IL-1-betaexacerbates neonatal excitotoxic brain lesions in the present modelOther prostaglandin metabolites of Cox-2 some of which havepotent inflammatory activity (Burian and Geisslinger 2005 Firuziand Pratico 2006) may also participate in this neural toxicityFurther studies are necessary to determine the precise contributionof Cox metabolites to neural toxicity

Cox-2 is located on the BBB (Mark et al 2001) but also inbrain cells such as astrocytes (Maslinska et al 1999) Furtherstudies will be necessary to determine the relative contribution ofthese differentially distributed Cox-2 in the present model

Role of brain cytokine production in the effects of systemicadministration of IL-1-beta

Some of the sensitizing effects of IL-1-beta-induced Coxactivation could involve an autocrineparacrine loop leading to adisruption in the balance between pro- and anti-inflammatorycytokines produced by brain cells Indeed the systemic adminis-tration of IL-1-beta induced significant changes in the brainexpression of various cytokines an increase in the expression ofIL-1-beta IL-6 and TNF-alpha three pro-inflammatory cytokinesThe effect on IL-18 another pro-inflammatory cytokine was morecomplex as a biphasic curve was observed with a decrease in theexpression at low dose and an increased expression at higher dosesThe precise mechanism underlying this biphasic effect remainsunclear The systemic administration of IL-1-beta had no detectableeffect on the production of mRNA coding for the anti-inflammatory cytokine IL-10 However this apparent lack of

504 G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

effect might be due to very low basal expression of this cytokinewith any further decrease in expression occurring below the limitsof detectability Despite the decreased expression of IL-18 at lowdose it is tempting to make the hypothesis that the systemicadministration of IL-1-beta tilts the brain cytokine balance towardsa pro-inflammatory state

These effects of systemic IL-1-beta on IL-6 IL-18 and TNF-alpha brain expression seem to be mediated by Cox activation assupported by the blocking effect of nimesulide

In addition nimesulide induced a detectable and significantincrease in IL-10 mRNA expression further favoring a switch inthe cytokine balance towards an anti-inflammatory state after Coxinhibition This effect of nimesulide on IL-10 mRNA suggests thatIL-10 expression is down-regulated by the basal production ofPGE2

Increased brain expression of IL-1-beta was not modified bynimesulide This could suggest that some effects of systemic IL-1-beta administration on the newborn brain are Cox-independentHowever nimesulide by increasing plasmatic levels of IL-1-beta(Fig 2B) can potentially increase brain expression of IL-1-beta(Fig 4A) which would mask its inhibitory effect through Cox-2blockade

Further explaining the neuroprotective effects of Cox inhi-bition in the present model nimesulide significantly increasedthe expression of this cytokine with trophic and neuroprotectiveproperties (Husson et al 2005 Bemelmans et al 2006) Thiseffect of nimesulide on BDNF mRNA suggests that BDNFexpression is partially controlled by the basal production ofPGE2

Of note nimesulide significantly increased blood concentra-tions of IL-1-beta suggesting that the key effects of nimesulideare not due to a peripheral degradation of exogenouslyadministered IL-1-beta but rather to a direct inhibition of Coxon the BBB The precise mechanism by which nimesulideincreases plasmatic IL-1-beta concentration remains unclearCox-2 inhibitors have been shown to alter renal function insome patients (Zhang et al 2006) This effect could potentiallyreduce the clearance of IL-1-beta and enhance its plasmaticconcentration

Potential implications for human neonates

In human neonates high levels of several pro-inflammatorymarkers in amniotic fluid or in cord blood have been associatedwith an increased risk for the subsequent development of CP (Yoonet al 1996 Yoon et al 1997 Martinez et al 1998 Nelson et al1998) This suggests a role for immune factors in the pathophy-siology of brain lesions associated with CP

Assuming that data obtained for the present murine model canbe extrapolated to human neonates it is possible that the activationof Cox-2 and the subsequent excess production of PGE2 couldmediate some of the sensitizing effects of systemic inflammationon human neonates at risk for the development of CP If confirmedthis Cox-2PGE2 physiopathological pathway could represent anew target for neuroprotection

In the present model some effects of Cox inhibitors arepotentially deleterious including increased levels of circulating IL-1-beta and increased brain expression of COX-2 mRNA and IL-18However the net effect of Cox inhibition is significantlyneuroprotective against excitotoxic insults in newborn miceexposed to a systemic inflammatory context

Conclusion

Our data support the hypothesis that in newborn mice thesensitizing effects of circulating inflammatory cytokines onexcitotoxic brain lesions are mediated by the Cox located on theBBB In particular cytokines activate the inducible isoform Cox-2resulting in the enhanced local production of PGE2 and otherderivatives that exacerbate excitotoxic brain lesions Some of thesedeleterious effects could involve an autocrineparacrine loopleading to a disruption in the balance between inflammatory andanti-inflammatory cytokines produced by brain cells Targeting thisCox-2-mediated pathway might represent a novel neuroprotectiveavenue for neonates at risk for CP

Acknowledgments

We thank Marc Laburthersquos laboratory for their help with pros-taglandin measurement This work was supported by the INSERMUniversiteacute Paris 7 the Fondation pour la Recherche Meacutedicale andthe Fondation Grace de Monaco This article was prepared witheditorial help from Gap Junction wwwgap-junctioncom

References

Bemelmans AP Husson I Jaquet M Mallet J Kosofsky BEGressens P 2006 Lentiviral-mediated gene transfer of brain-derivedneurotrophic factor is neuroprotective in a mouse model of neonatalexcitotoxic challenge J Neurosci Res 83 50ndash60

Burian M Geisslinger G 2005 COX-dependent mechanisms involved inthe antinociceptive action of NSAIDs at central and peripheral sitesPharmacol Ther 107 139ndash154

Chomczynski P Sacchi N 1987 Single-step method of RNA isolation byacid guanidinium thiocyanate-phenol-chloroform extraction AnalBiochem 162 156ndash159

Dammann O Kuban KC Leviton A 2002 Perinatal infection fetalinflammatory response white matter damage and cognitive limitationsin children born preterm Ment Retard Dev Disabil Res Rev 846ndash50

Dommergues MA Patkai J Renauld JC Evrard P Gressens P 2000Proinflammatory cytokines and interleukin-9 exacerbate excitotoxiclesions of the newborn murine neopallium Ann Neurol 47 54ndash63

Firuzi O Pratico D 2006 Coxibs and Alzheimerrsquos disease should theystay or should they go Ann Neurol 59 219ndash228

Greig PC Ernest JM Teot L Erikson M Talley R 1993 Amnioticfluid interleukin-6 levels correlate with histologic chorioamnionitis andamniotic fluid cultures in patients in premature labor with intactmembranes Am J Obstet Gynecol 169 1035ndash1044

Gressens P Marret S Hill JM Brenneman DE Gozes I Fridkin MEvrard P 1997 Vasoactive intestinal peptide prevents excitotoxic celldeath in the murine developing brain J Clin Invest 100 390ndash397

Gressens P Rogido M Paindaveine B Sola A 2002 The impact ofneonatal intensive care practices on the developing brain J Pediatr 140646ndash653

Hagberg H Mallard C 2005 Effect of inflammation on central nervoussystem development and vulnerability Curr Opin Neurol 18 117ndash123

Hagberg B Hagberg G Olow I van Wendt L 1996 The changingpanorama of cerebral palsy in Sweden VII Prevalence and origin in thebirth year period 1987ndash90 Acta Paediatr 85 954ndash960

Hagberg H Wilson MA Matsushita H Zhu C Lange M GustavssonM Poitras MF Dawson TM Dawson VL Northington FJohnston MV 2004 PARP-1 gene disruption in mice preferentiallyprotects males from perinatal brain injury J Neurochem 90 1068ndash1075

Himmelmann K Hagberg G Beckung E Hagberg B Uvebrant P2005 The changing panorama of cerebral palsy in Sweden IX

505G Favrais et al Neurobiology of Disease 25 (2007) 496ndash505

Prevalence and origin in the birth-year period 1995ndash1998 Acta Paediatr94 287ndash294

Husson I Mesples B Bac P Vamecq J Evrard P Gressens P2002 Melatoninergic neuroprotection of the murine periventricularwhite matter against neonatal excitotoxic challenge Ann Neurol 5182ndash92

Husson I Rangon CM Lelievre V Bemelmans AP Sachs P MalletJ Kosofsky BE Gressens P 2005 BDNF-induced white matterneuroprotection and stage-dependent neuronal survival following aneonatal excitotoxic challenge Cereb Cortex 15 250ndash261

Kawano T Anrather J Zhou P Park L Wang G Frys KA Kunz ACho S Orio M Iadecola C 2006 Prostaglandin E2 EP1 receptorsdownstream effectors of COX-2 neurotoxicity Nat Med 12 225ndash229

Lelievre V Hu Z Byun JY Ioffe Y Waschek JA 2002 Fibroblastgrowth factor-2 converts PACAP growth action on embryonic hind-brain precursors from stimulation to inhibition J Neurosci Res 67566ndash573

Mark KS Trickler WJ Miller DW 2001 Tumor necrosis factor-alphainduces cyclooxygenase-2 expression and prostaglandin release in brainmicrovessel endothelial cells J Pharmacol Exp Ther 297 1051ndash1058

Marret S Mukendi R Gadisseux JF Gressens P Evrard P 1995Effect of ibotenate on brain development an excitotoxic mouse model ofmicrogyria and posthypoxic-like lesions J Neuropathol Exp Neurol54 358ndash370

Martinez E Figueroa R Garry D Visintainer P Patel K Verma USehgal PB Tejani N 1998 Elevated amniotic fluid interleukin-6 as apredictor of neonatal periventricular leukomalacia and intraventricularhemorrhage J MaternndashFetal Investig 8 101ndash107

Maslinska D Wozniak R Kaliszek A Modelska I 1999 Expression ofcyclooxygenase-2 in astrocytes of human brain after global ischemiaFolia Neuropathol 37 75ndash79

Moalic S Liagre B LeBail JC Beneytout JL 2001 Dose-dependentmodulation of apoptosis and cyclooxygenase-2 expression in human1547 osteosarcoma cells by NS-398 a selective cyclooxygenase-2inhibitor Int J Oncol 18 533ndash540

Murphy DJ Sellers S MacKenzie IZ Yudkin PL Johnson AM1995 Casendashcontrol study of antenatal and intrapartum risk factors forcerebral palsy in very preterm singleton babies Lancet 346 1449ndash1454

Nelson KB Willoughby RE 2000 Infection inflammation and the riskof cerebral palsy Curr Opin Neurol 13 133ndash139

Nelson KB Dambrosia JM Grether JK Phillips TM 1998 Neonatalcytokines and coagulation factors in children with cerebral palsy AnnNeurol 44 665ndash675

Nijboer CH Groenendaal F Kavelaars A Hagberg HH vanBel FHeijnen CJ in press Gender-specific neuroprotection by 2-

iminobiotin after hypoxiandashischemia in the neonatal rat via a nitricoxide independent pathway J Cereb Blood Flow Metab (Electronicpublication ahead of print) doi101038sjjcbfm9600342

Rainsford KD Seabrook RW Spencer S Hewson AT 2001 Effectsof nimesulide and its metabolites or manufacturing intermediates on theviability and growth of the human hepatoma HepG2 cell line Life Sci69 2965ndash2973

Riendeau D Charleson S Cromlish W Mancini JA Wong EGuay J 1997 Comparison of the cyclooxygenase-1 inhibitoryproperties of nonsteroidal anti-inflammatory drugs (NSAIDs) andselective COX-2 inhibitors using sensitive microsomal and plateletassays Can J Physiol Pharmacol 75 1088ndash1095

Romero R Avila C Santhanam U Sehgal PB 1990 Amniotic fluidinterleukin 6 in preterm labor Association with infection J Clin Invest85 1392ndash1400

Singh B Merchant P Walker CR Kryworuchko M Diaz-Mitoma F1996 Interleukin-6 expression in cord blood of patients with clinicalchorioamnionitis Pediatr Res 39 976ndash979

Takeuchi K Tanaka A Hayashi Y Kubo Y 2004 Functionalmechanism underlying COX-2 expression following administration ofindomethacin in rat stomachs importance of gastric hypermotility DigDis Sci 49 180ndash187

Tanaka A Araki H Hase S Komoike Y Takeuchi K 2002 Up-regulation of COX-2 by inhibition of COX-1 in the rat a key to NSAID-induced gastric injury Aliment Pharmacol Ther 16 (Suppl 2) 90ndash101

Volpe JJ 2001 Perinatal brain injury from pathogenesis toneuroprotection Ment Retard Dev Disabil Res Rev 7 56ndash64

Yoon BH Romero R Yang SH Jun JK Kim IO Choi JHSyn HC 1996 Interleukin-6 concentrations in umbilical cordplasma are elevated in neonates with white matter lesions associatedwith periventricular leukomalacia Am J Obstet Gynecol 1741433ndash1440

Yoon BH Jun JK Romero R Park KH Gomez R Choi JHKim IO 1997 Amniotic fluid inflammatory cytokines (interleukin-6interleukin-1beta and tumor necrosis factor-alpha) neonatal brainwhite matter lesions and cerebral palsy Am J Obstet Gynecol 17719ndash26

Zhang J Ding EL Song Y 2006 Adverse effects of cyclooxygenase 2inhibitors on renal and arrhythmia events meta-analysis of randomizedtrials JAMA 296 1619ndash1632

Zuker M 2003 Mfold web server for nucleic acid folding andhybridization prediction Nucleic Acids Res 31 3406ndash3415

Zupan V Gonzalez P Lacaze-Masmonteil T Boithias C drsquoAllest AMDehan M Gabilan JC 1996 Periventricular leukomalacia riskfactors revisited Dev Med Child Neurol 38 1061ndash1067