cAMP Response Element-Binding ProteinActivation in Ligation Preconditioning in

Neonatal BrainHsueh-Te Lee, BS,1 Ying-Chao Chang, MD, PhD,2 Lin-Yu Wang, MD,3 Shan-Tair Wang, PhD,4

Chao-Ching Huang, MD,5,6 and Chien-Jung Ho, BS6

Perinatal hypoxic-ischemic (HI) brain injury is a major cause of permanent neurological dysfunction in children. Anapproach to study the treatment of neonatal HI encephalopathy that allows for neuroprotection is to investigate thestates of tolerance to HI. Twenty-four-hour carotid-artery ligation preconditioning established by delaying the onset ofhypoxia for 24 hours after permanent unilateral carotid ligation rats markedly diminished the cerebral injury, however,the signaling mechanisms of this carotid-artery ligation preconditioning in neonatal rats remain unknown. Ligation ofthe carotid artery 24 hours before hypoxia provided complete neuroprotection and produced improved performance onthe Morris water maze compared with ligation performed 1 hour before hypoxia. Carotid artery ligation 6 hours beforehypoxia produced intermediate benefit. The 24-hour carotid-artery ligation preconditioning was associated with a robustand sustained activation of a transcription factor, the cAMP response element–binding protein (CREB), on its phos-phorylation site on Ser133. Intracerebroventricular infusions of antisense CREB oligodeoxynucleotides significantly re-duced the 24-hour carotid-artery ligation–induced neuroprotection effects by decreasing CREB expressions. Pharmaco-logical activation of the cAMP-CREB signaling with rolipram 24 hours before hypoxia protected rat pups at behavioraland pathological levels by sustained increased CREB phosphorylation. This study suggests that 24-hour carotid-arteryligation preconditioning provides important mechanisms for potential pharmacological preconditioning against neonatalHI brain injury.

Ann Neurol 2004;56:611–623

Perinatal hypoxic-ischemia (HI) is a major cause ofneonatal mortality and of subsequent neurological dis-abilities among infants who survive it.1–3 Although theimmature brain is relatively protected from HI byadaptive mechanisms, severe insults can trigger self-sustaining neurotoxic cascades lasting several days andresult in prominent neuronal injury.1,4,5 Although ourunderstanding of the pathogenesis of neonatal HI braininjury has increased considerably, there is still no effec-tive treatment.4,5

In adult animal stroke models, preconditioning bynonlethal ischemia exhibits protective effects againstneuronal death in vulnerable brain regions after subse-quent lethal ischemia. The molecules that mediate theinduction of ischemic tolerance include heat shock pro-teins, reactive oxygen species, nitric oxide, nuclearfactor-�B, adenosine A1 receptor, and KATP channels.6–9

Unilateral carotid artery ligation followed by sys-temic exposure to hypoxia is a model widely used toinvestigate HI brain injury in immature rats.10 Hyp-oxia preconditioning produces tolerance, throughhypoxia-inducible factor–1 activation, against HI braininjury in newborn rats.11,12 On the other hand, pre-conditioning established by delaying the onset of hyp-oxia for 24 hours after carotid artery ligation alsomarkedly diminishes the extent of HI injury in the im-mature brain; tissue adaptation occurs between arteryligation and the onset of hypoxia,13,14 but the cellularmechanisms triggered by carotid-artery ligation precon-ditioning remain poorly understood.

Extracellular stimuli can exert long-lasting effects,such as promoting cell growth, proliferation, survival,or death, by triggering signaling cascades that ulti-mately converge onto nuclear transcription factors.15

From the 1Institute of Basic Medical Science, Medical College, Na-tional Cheng Kung University, Tainan, Taiwan; 2Department ofPediatrics, Chang Gung Memorial Hospital Kaohsiung, Taiwan;3Department of Pediatrics, Chi-Mei Hospital, Tainan, Taiwan; In-stitutes of 4Public Health and 5Molecular Medicine; and 6Depart-ment of Pediatrics, Medical College, National Cheng Kung Univer-sity, Tainan, Taiwan.

Received Jan 12, 2004, and in revised form May 28 and Jul 18.Accepted for publication Jul 18, 2004.

Published online Oct 6, 2004 in Wiley InterScience(www.interscience.wiley.com). DOI: 10.1002/ana.20259

Address correspondence to Dr Huang, Department of Pediatrics,National Cheng Kung University Hospital, 138 Sheng-Li Road,Tainan City 704, Taiwan. E-mail: huangped@mail.ncku.edu.tw

ORIGINAL ARTICLES

© 2004 American Neurological Association 611Published by Wiley-Liss, Inc., through Wiley Subscription Services

One transcription factor, cAMP response element–binding protein (CREB), is a key mediator of stimulus-induced nuclear responses that underlie learning andmemory, and the plasticity of the nervous system.15,16

CREB is phosphorylated on Ser133 (pCREB) bycAMP-dependent protein kinase (PKA) or other ki-nases and consequently binds to the cAMP responseelement (CRE) of target genes.15,16 Evidence suggeststhat endogenous CREB activation might provide po-tent survival signals during ischemia,15–18 and duringacquisition of ischemic tolerance in the adult ratbrain.19,20 Whether CREB activation is involved in thesignaling pathway of carotid-artery ligation precondi-tioning against HI in the immature brain, however, re-mains undetermined.

In this study, we used CREB antisense oligode-oxynucleotides (ODNs) to disrupt CREB protein levelsto examine the effect of CREB on carotid-artery liga-tion preconditioning. ODNs are preferentially takenup by neurons in the rodent brain after intracerebraladministration.21,22 Inside the cell, antisense ODNbases pair to their cognate mRNAs to increase turnoveror block translation of targeted mRNA. On the otherhand, pharmacological activation of the cAMP-PKAsignaling pathway by rolipram, a phosphodiesterasetype IV (PDE4) inhibitor,23–25 or forskolin, an adeny-lyl cyclase-activator,18,26 could lead to CREB phos-phorylation. Therefore, this study aimed to test the fol-lowing hypotheses in the neonatal rat brain: (1)preconditioning with carotid-artery ligation protectsagainst HI injury in a time-dependent manner; (2) pre-conditioning with 24-hour carotid-artery ligation en-hances CREB phosphorylation; (3) CREB expressionmediates the preconditioning effects induced by 24-hour carotid-artery ligation; (4) pharmacological pre-conditioning by activation of the cAMP-CREB signal-ing pathway is neuroprotective against HI injury.

Materials and MethodsThis study was approved by the Animal Care Committee atNational Cheng Kung University. Ten to 12 pups of eithersex per dam were used and housed with a 12-hour-light/12-hour-dark schedule in a temperature- and humidity-controlled colony room. The pups were housed with theirdams until weaning on postnatal (P) day 21 and then housedin groups of four to five per cage. The male and female ratswas equally distributed between the experimental groups.

In P7 rat pups, unilateral common carotid artery ligationfollowed 1 hour later by 8% oxygen hypoxia for 2 hoursproduces selective damage in the hemisphere ipsilateral to theartery occlusion that resembles HI damage to the humanneonatal brain.10

Carotid-Artery Ligation PreconditioningTo examine whether the carotid-artery ligation precondition-ing could be established in a time-dependent manner, weperformed carotid artery ligation in rat pups 24 hours (24-

hour ligation group), 6 hours (6-hour ligation group), or 1hour (1-hour ligation group) before exposing them to 2hours of 8% oxygen hypoxia on P7. The animals were anes-thetized with 2.5% halothane (balance, room air), and theright common carotid artery was surgically exposed and per-manently ligated with 5-0 surgical silk. After surgery, thepups were returned to the dam for a 1-, 6-, or 24-hour re-covery before hypoxia. On P7, the three groups were placedin airtight 500ml containers partially submerged in a 37°Cwater bath through which humidified 8% oxygen (balance,nitrogen) was maintained at a flow rate of 3L/min for 2hours. After completion of hypoxia, the rat pups were re-turned to their cage. There were two control groups: shamcontrols (sham-operated P6 rats without hypoxia) and liga-tion controls (P6 rats with permanent artery ligation butwithout hypoxia).

cAMP Response Element-Binding Protein AntisenseOligodeoxynucleotide InfusionsThe sequence of the CREB antisense and scrambled ODNwere: CREB antisense, 5�-TGGTCATCTAGTCACCG-GTG-3�; and scrambled, 5�-GTCTGCAGTCGATCTA-CGGT-3�.21 The antisense sequence perfectly matched therat CREB gene corresponding to nucleotides 27 to 46.21,22

The scrambled sequence showed no significant matches inthe GenBank database. P6 rats, anesthetized with 2.5% halo-thane, were intracerebroventricularly injected with antisenseor scrambled ODN (2nmol in 1�l) in the right cerebralhemisphere using a 30-gauge needle on a 10�l Hamiltonsyringe. The location of each injection in relation to thebregma was 2.0mm posterior, 1.5mm lateral, and 2.0mmdeep to the skull surface.27 In the 24-hour ligation group,the first ODN infusion was given 3 hours before ligation,and the second and third were given at 6 and 12 hours afterligation. Twenty-four hours after ligation (12 hours after thelast ODN infusion), both groups underwent 8% oxygenhypoxia for 2 hours. An HI control group received three in-tracerebroventricular injections of scrambled ODN underthe same schedule and underwent HI on P7. On P21, hemi-spheric weight reduction was measured.

Autoradiographic Determination of Cerebral BloodFlowCerebral blood flow (CBF) was determined in the immaturerats by the iodo-[14C] antipyrine (IAP) autoradiographictechnique.28 Immediately after hypoxia, each rat was injectedsubcutaneously with 5�Ci IAP in 0.1ml normal saline intothe midline back of the rat. The brains of the rats killed 2minutes after IAP injection were immediately removed, fro-zen and processed for autoradiography.28 In brief, braincoronal sections 20�m thick were cut in cryostat, mountedon glass slides, dried at 55°C on a hot plate, and subjected tocarbon-14 autoradiography.

Forskolin PreconditioningP6 rats received a 5�l intracerebroventricular injection of ve-hicle [dimethyl sulfoxide (DMSO)] or forskolin (25�g inDMSO) as described above.27 Twenty-four hours after injec-tion, the rats underwent HI. On P21, hemispheric weightreduction was measured. To determine whether DMSO in-

612 Annals of Neurology Vol 56 No 5 November 2004

duced a preconditioning effect, we injected P6 rats similarlywith DMSO or saline 24 hours before HI and hemisphericweight reduction was compared.

Rolipram PreconditioningP6 rats were administered a single intraperitoneal injection of0.1ml of rolipram (1 or 3mg/kg) or DMSO; 24 hours later,they underwent HI. Rectal temperature was measured at 1,3, and 24 hours after injection and immediately after HI. Toavoid temperature changes, we kept pups and their dam inan infant incubator maintained at 30 to 31°C during thefirst 24 hours after injection.

Outcome MeasuresMORRIS WATER MAZE. In brief, a circular pool was filledwith water, and an 8 � 8cm platform was positioned in thecenter of one of the quadrants, 1cm below the water surface.From Days 1 to 4, P32 rats were given 24 training sessions(six per day) to locate the submerged platform. On Day 5,the platform was removed, and rats were placed back intothe pool and allowed 60 seconds of free swimming. Escapelatency, escape distance, and swimming patterns of the ratswere monitored by a camera mounted above the pool(EthoVision, Wageningen, The Netherlands). After theprobe test, rats were submitted to four trials of a visuallycued learning task to locate a green escape platform 2cmabove the water.29

HEMISPHERIC WEIGHT REDUCTION. On P36, the brainswere removed. After removal of the brainstem and cerebel-lum, the forebrain was sectioned at the midline, and left andright hemispheric weights were determined. The percentageof hemispheric weight reduction measured as (left hemi-sphere weight � right hemisphere weight)/left hemisphereweight was used as the measure of cerebral injury in thisstudy. Our preliminary data from P7 rats with HI injuryshowed that the changes in the hemispheric weight reductionwere highly correlated not only to the changes in the infarctbrain areas (r � 0.87; p � 0.0001; n � 27) according toplates 17, 20, 23, 28, 31, and 34 in a rat brain atlas,30 butalso to the hemispheric volume changes (r � 0.91; p �0.0001; n � 27) measured on P36. This finding was con-sistent with another report.31 Therefore, only hemisphericweight data are shown in this study.

Western Blot AnalysisTissue was homogenized in cold lysis buffers as describedpreviously.29,32 Fifty-microgram samples were resolved in10% sodium dodecyl sulfate polyacrylamide gel electrophore-sis gels and blotted electrophoretically to nitrocellulose mem-branes. Membranes were incubated with primary antibodies,and immunoreactivity was detected by horseradish-conjugated secondary antibody and visualized with enhancedchemiluminescence. The following primary antibodies (Up-state Biotechnology, Charlottesville, VA) were used: anti–phospho-Ser133-CREB (pCREB) (1:1,000 dilution), anti-CREB (1:1,500), and brain-derived neurotrophic factor(BDNF; 1:1,000).

cAMP Response Element-Binding ProteinPhosphorylated on Ser133 ImmunohistochemistryTwenty-four hours after hypoxia, 24- and 1-hour ligationgroups were perfused with 4% paraformaldehyde, and thebrains were postfixed for 24 hours and then cryoprotected in30% sucrose solutions. Coronal sections (20�m) were incu-bated with anti–pCREB antibody (1:100) and visualizedwith an avidin-biotin system (Vector Laboratories, Burlin-game, CA).29 In each pCREB immunohistochemistry stain-ing, negative control was also performed with normal rabbitserum as a substitute for the primary antibody.

Reverse Transcription Polymerase Chain Reaction ofBrain-Derived Neurotrophic FactorTotal RNA was extracted from each hemisphere according tothe TRIzol protocol (Invitrogen, San Diego, CA). A 5�gportion of total RNA and 1.5�g oligo-dT primer were in-cubated at 70°C for 10 minutes and gradually cooled toroom temperature. Each RT mixture, containing 25 units ofM-MLV reverse transcriptase (Promega, Madison, WI), 10�l5 � reaction buffer, 0.5mM dNTP, and nuclease-free dis-tilled water, was added to a final volume of 50�l. The sam-ples were incubated at 37°C for 90 minutes followed by de-naturation at 95°C for 10 minutes. Each PCR (20�l)contained 2�l of RT product, 1 unit of Taq DNA polymer-ase (Viogene, Taiwan), 2�l 10 � PCR buffer plus MgCl2,0.2mM dNTP, and 0.5�M gene-specific primers (BDNF:forward 5�-GACAAGGCAACTTGGCCTAC-3�, reverse 5�-CTGTCACACACGCTCAGCTC-3�, product size: 356bp;GAPDH: forward 5�-ACATTGTTGCCATCAACGAC-3�,reverse 5�-ACGCCAGTAGACTCCACGAC-3�, productsize: 216bp). Amplified reaction was performed with a ther-mocycler for a single 3-minute initial denaturation at 94°Cfollowed 33 cycles (BDNF) or 26 cycles (GAPDH) underthe conditions: 94°C (20 seconds), 55°C (20 seconds), and72°C (20 seconds) and final extension at 72°C for 4 min-utes. The PCR products were separated on 1.5% agarose gelscontaining ethidium bromide and quantified by densitome-try. The BDNF PCR product was normalized to that ofGAPDH PCR product in each sample.

StatisticsStatistical significance (p � 0.05) was determined using one-way analysis of variance (ANOVA) to compare hemisphericweight reduction and probe test and visual motor test resultsof the water maze between experimental groups. Multivariateanalysis of variance (MANOVA) was used to compare escapetime over the learning phase of the water maze. Post hoccomparisons using Tukey’s method were used in one-wayANOVA and Bonferroni’s method in MANOVA. Continu-ous data were represented as mean � SEM, unless indicatedotherwise.

Results

Carotid-Artery Ligation–Induced PreconditioningEffects of Ligation PreconditioningDuring water-maze training, there was a significantgroup difference between the sham and ligation control

Lee et al: pCREB in Brain Preconditioning 613

groups, and the 1-, 6-, and 24-hour ligation groups(p � 0.001). Post hoc multiple comparisons showedthat 1-hour ligation group spent significantly moretime finding the submerged platform (p � 0.001), lesstime in the target quadrant (p � 0.05), and more timereaching the visible platform (p � 0.001), than theother groups (Fig 1A–C). There were no significantdifferences in time spent finding the submerged plat-form, staying in the target quadrant, or reaching thevisible platform between the sham control, ligationcontrol, 6-hour, and 24-hour ligation groups.

The degree of brain injury, as measured by the de-gree of cerebral hemispheric weight reduction, was also

significantly greater in the 1-hour ligation group thanin the other groups (all p � 0.001; see Fig 1D). Therewas no significant difference in the degree of brain in-jury among the 6-, 24-hour ligation groups, the shamand ligation-control groups, although the neuroprotec-tion in the 6-hour ligation group was not as completeas the other three groups.

Immediately after hypoxia, CBF was reduced in thecortex and lateral parts of the striatum ipsilateral to thecarotid artery ligation compared with the contralateralhemisphere in the 1-hour ligation group. In contrast,no hemispheric difference in CBF was seen in the 24-hour ligation group (see Fig 1E).

Fig 1. Carotid-artery ligation preconditioning induced neuroprotection against hypoxic-ischemic brain injury at behavior and patho-logical levels in a time-dependent manner. (A) During water-maze training, the 1-hour ligation group spent significantly more timefinding the platform than the other groups (all p � 0.001). There were no significant differences between the 6-, 24-hour ligation,sham and ligation control groups. (B) On the probe test, the 1-hour ligation group spent significantly less time in the target quad-rant than each of the other groups (all p � 0.05). There were no significant differences between the 6-, 24-hour ligation, shamand ligation control groups. (C) In the visual motor performance, the 1-hour ligation group spent significantly more time than eachof the other groups (all p � 0.001). In contrast, there was no significant difference among the 6-, 24-hour ligation, sham andligation control groups. (D) The degree of brain injury, measured by hemispheric weight reduction, in the 1-hour ligation groupwas significantly greater than in the other groups (all p � 0.001). There was no significant difference among the sham, ligationcontrol groups, 6-, and 24-hour ligation groups. (#p � 0.001, ✽p � 0.05). (E) Posthypoxia cerebral blood flow (CBF) examina-tion by iodo-[14C] antipyrine autoradiography showed reduced CBF in the hemisphere ipsilateral to artery ligation compared withthe contralateral hemisphere in the 1-hour ligation group. In contrast, no hemispheric CBF difference was seen in the 24-hour liga-tion group (R � right hemisphere, L � left hemisphere).

614 Annals of Neurology Vol 56 No 5 November 2004

Ligation Preconditioning and cAMP ResponseElement-Binding Protein Phosphorylated on Ser133We performed Western blot analyses to determine thetemporal profile of pCREB in the cortex and hip-pocampus ipsilateral to the artery ligation. Hippocam-pal pCREB levels increased and reached a peak at 6hours after ligation preconditioning; those levels de-creased but remained elevated 24 hours after precondi-tioning. Cortical pCREB levels increased, with onepeak at 3 hours and another at 24 hours after precon-ditioning. Total CREB levels remained unchanged (Fig2A, B).

We next examined the temporal profile of pCREBposthypoxia. In the 1-hour ligation group, hippocam-pal pCREB levels were already high 3 hours afterhypoxia, when first measured, and then progressively

decreased to control level by 24 hours after hypoxia.In the 24-hour ligation group, hippocampal pCREBlevels progressively increased 6 to 24 hours after hyp-oxia. Cortical pCREB levels in the 1-hour ligationgroup increased 3 to 12 hours after hypoxia and de-creased to control level 24 hours after hypoxia. In the24-hour ligation group, increased cortical pCREB lev-els persisted 24 hours after hypoxia. Total CREB re-mained unchanged (see Fig 2C, D).

Immunohistochemistry of cAMP Response Element-Binding Protein Phosphorylated on Ser133There were few pCREB-immunopositive cells in thecortex and hippocampus in the sham control (Fig 3A,D). Twenty-four hours after hypoxia in the 24-hour

Fig 2. The temporal profile of cAMP response element–binding protein phosphorylated on Ser133 (pCREB) levels in the ipsilateralhippocampus and cortex before and after hypoxia. The pCREB levels was robustly increased and sustained in the hippocampus (A)and cortex (B) during the 24 hours of ligation preconditioning. Hippocampal pCREB levels increased and reached a peak at 6hours after preconditioning, and the levels decreased but remained elevated 24 hours after preconditioning. Cortical pCREB levelsincreased, with one peak at 3 hours and another at 24 hours after preconditioning. Total CREB remained unchanged in bothgroups. (C) In the 1-hour ligation group, hippocampal pCREB levels were already high 3 hours after hypoxia and then progressivelydecreased to control level by 24 hours after hypoxia. In the 24-hour ligation group, hippocampal pCREB levels progressively in-creased 6 to 24 hours after hypoxia. (D) Cortical pCREB levels in the 1-hour ligation group increased 3 to 12 hours after hypoxiaand decreased to control level 24 hours after hypoxia. In the 24-hour ligation group, increased cortical pCREB levels persisted 24hours after hypoxia. Total CREB remained unchanged.

Lee et al: pCREB in Brain Preconditioning 615

ligation group, there was a marked increase ofpCREB immunoreactivities in the cortex and hip-pocampus on the lesioned hemisphere (see Fig 3B,E), which contrasted with the very weak pCREB im-munoreactivity in the 1-hour ligation group (see Fig3C, F).

Effect of Oligodeoxynucleotides on LigationPreconditioning and cAMP Response Element-Binding ProteinIn the 24-hour ligation group, we used antisense ODNspecific for CREB to diminish CREB expression to seewhether CREB expression was necessary for the neuro-protective effects in that group. Rats were unilaterallyinfused with 5� biotinylated antisense ODN 3 hoursbefore being killed to examine the localization of ODNdelivery. Dark staining showed ODN in the ipsilateralstriatum, cortex, and hippocampus (Fig 4A). The de-gree of brain injury was significantly different betweenthe three groups, including the 1-hour ligation grouppretreated with scrambled ODN and the 24-hour liga-tion group with scrambled or CREB antisense ODNinfusions (p � 0.05; Fig 4B). Post hoc multiple com-parisons showed that the antisense ODN-infused ratshad significantly greater brain injury than the scram-bled ODN-infused rats in the 24-hour ligation group

(p � 0.001). The 1-hour ligation group pretreatedwith scrambled ODN had significantly greater braininjury than did the antisense ODN-infused rats (p �0.001) or scrambled ODN-infused rats in the 24-hourligation group (p � 0.001). The antisense ODN infu-sions reduced CREB protein expression by 60% 6hours after the last antisense ODN infusion; CREBlevels returned to control level 12 hours after the lastinfusion (at the time of hypoxia; see Fig 4C). Therewas no significant difference in the degree of brain in-jury between CREB-antisense-ODN–treated (0.4 �0.2%, n � 19) and scrambled-ODN–treated (0.7 �0.4%, n � 18) rat pups not exposed to HI, whichexcluded the potentially deleterious effects of repeatedantisense ODN infusions in the developing brain.

Effects of Forskolin Pharmacological PreconditioningThe degree of brain injury was significantly lesser in theforskolin-pretreated rats than in the DMSO-pretreatedrats (p � 0.05; see Fig 4D). Although DMSO has manybiological effects that could be neuroprotective,33 therewas no significant difference in the degree of brain in-jury between DMSO- (38.8 � 2.2%, n � 26) andsaline- (35.5 � 4.1%, n � 17) pretreated groups. Com-pared with vehicle, forskolin injection caused sustainedincreases of pCREB in the hippocampus and cortex 6 to

Fig 3. Anatomical distribution of cAMP response element–binding protein phosphorylated on Ser133 (pCREB) immunoreactivity inthe cortex and hippocampus 24 hours after hypoxia. There were few pCREB-immunopositive cells in the cortex (A) and hippocam-pus (D) in the sham control. A marked increase of pCREB-immunoreactive cells was seen in the cortex (B) and hippocampus (E)on the lesioned hemisphere in the 24-hour ligation group. Inset shows high-power image of cornu ammonis 1 region of the hip-pocampus (CA1) (arrow) with increased pCREB immunoreactivities mainly located in the nucleus of the CA1 cells. In contrast,very weak pCREB immunoreactivity was found in the cortex (C) and hippocampus (F) in the 1-hour ligation group (n � 4 pergroup). Scale bars � 200�m in A–C, 500�m in D–F, and 50�m in inset of E.

616 Annals of Neurology Vol 56 No 5 November 2004

24 hours after injection (see Fig 4E). There was no dif-ference in total CREB between the two groups.

Effects of RolipramDuring water-maze training, the escape times were sig-nificantly different between the controls and rats pre-conditioned with rolipram or vehicle (p � 0.001; Fig

5A). The vehicle-pretreated rats spent significantlymore time finding the platform than did rolipram-preconditioned rats (p � 0.01) and controls (p �0.001), and that controls spent less time than didrolipram-preconditioned rats (all p � 0.05). On theprobe test, there were significant differences (p �0.005) between the four groups in time spent in the

Fig 4. cAMP response element–binding protein (CREB) antisense oligodeoxynucleotide (ODN) infusions diminished 24-hour liga-tion preconditioning effect by reducing CREB protein; in contrast, forskolin preconditioning 24 hours before hypoxia induced neuro-protection by CREB activation. (A) Antisense ODN (2nmol in 2�l) with an added 5� biotinylated group was infused into theright hemisphere to demonstrate the location of ODN delivery. Representative brain sections (30�m) showed dark stained biotinyl-ated ODN in the ipsilateral striatum, cortex, and hippocampus. (B) The degree of brain injury was significantly different betweenthe three groups, including the 1-hour ligation group treated with scrambled ODN, and the 24-hour ligation group with scrambledor CREB antisense ODN infusions (p � 0.05). In the 24-hour ligation group, three times of CREB antisense or scrambled ODNinfusions were performed with the first ODN infusion given 3 hours before ligation and the second and third infusion at 6 and 12hours after ligation. Twenty-four hours after ligation, both groups underwent 8% oxygen hypoxia for 2 hours. Post hoc comparisonsshowed that the CREB antisense ODN-infused rats had significantly greater brain injury than scrambled ODN-infused rats(21.2 � 2.8% vs 0.7 � 0.6%, p � 0.001). The 1-hour ligation animals pretreated with scrambled ODN had significantlygreater brain injury (43.9 � 3.1%) than did the 24-hour ligation group pretreated with scrambled ODN (p � 0.001) or withantisense ODN (p � 0.001). (C) The CREB protein expression was determined 6 and 12 hours after the last ODN infusion. TheCREB protein expression was reduced by 60% 6 hours after the last antisense ODN infusion, and the levels returned to basal level12 hours after the last infusion (at the time of hypoxia). (D) The degree of brain injury was significantly lesser in the forskolin-pretreated rats than in the vehicle-pretreated rats (25.3 � 3.0% vs 35.8 � 2.7%, p � 0.05). (E) Compared with vehicle, fors-kolin injection caused sustained increases of pCREB levels in the hippocampus and cortex 6 to 24 hours after injection. There wasno difference in total CREB between the two groups. CREB � cAMP response element–binding protein; pCREB � CREB phos-phorylated on Ser133; ICV � intracerebroventricular.

Lee et al: pCREB in Brain Preconditioning 617

target quadrant (see Fig 5B). Vehicle-pretreated ratsspent significantly less time than did the 3mg/kg roli-pram–preconditioned (p � 0.05) and control rats(p � 0.001). There was no significant difference be-tween 1mg/kg rolipram–preconditioned and vehicle-pretreated rats, or between 1 or 3mg/kg rolipram–pre-conditioned rats and controls. In the visual motor test,there were significant differences (p � 0.001) betweenthe four groups in time spent reaching the visible plat-form. Vehicle-pretreated rats spent significantly moretime reaching the platform than 1mg/kg rolipram–pre-conditioned (p � 0.05), 3mg/kg rolipram–precondi-tioned (p � 0.001), and control rats (p � 0.001; seeFig 5C). There was no significant difference between

1mg/kg rolipram–preconditioned, 3mg/kg rolipram–preconditioned, and control rats.

The degree of brain injury between the four groupsshowed significant differences (p � 0.0001): it wasgreater in the vehicle-pretreated group than in theother groups (all p � 0.001; see Fig 5D). There was nodifference between 1mg/kg rolipram–preconditioned,3mg/kg rolipram–preconditioned, and control rats, al-though a greater degree of brain injury was observed inthe 1mg/kg rolipram–preconditioned group.

The neuroprotective effects of rolipram observedhere could not be attributed to hypothermia: the tem-perature of the 1 and 3mg/kg rolipram–preconditionedanimals did not differ from that of vehicle-pretreated

Fig 5. Rolipram preconditioning induced neuroprotection at behavioral and pathological levels. (A) During water-maze training,the vehicle-pretreated rats spent significantly more time finding the submerged platform than did rolipram-preconditioned rats (all p� 0.01) and control (p � 0.001). The controls also spent less time than did rolipram-preconditioned rats (all p � 0.05). (B) Onthe probe test, vehicle-pretreated rats spent significantly less time in the target quadrant than did the 3mg/kg rolipram–precondi-tioned (p � 0.05) and control rats (p � 0.001). There was no significant difference between 1mg/kg rolipram–preconditioned andvehicle-pretreated rats, 1mg/kg rolipram–preconditioned and controls, or between 3mg/kg rolipram–preconditioned rats and controls.(C) In the visual motor test, vehicle-pretreated rats spent significantly more time than 1mg/kg rolipram–preconditioned (p � 0.05),3mg/kg rolipram–preconditioned (p � 0.001), or control rats (p � 0.001). In contrast, there was no significant difference between1mg/kg rolipram–preconditioned, 3mg/kg rolipram–preconditioned, and control groups. (D) The degree of brain injury in thevehicle-pretreated group was significantly greater than the other groups (each p � 0.001). There was no difference between 1mg/kgrolipram–preconditioned, 3mg/kg rolipram–preconditioned, and control groups, although a greater degree of brain injury was ob-served in the 1mg/kg rolipram–preconditioned group (#p � 0.001, ✽p � 0.01, p � 0.05).

618 Annals of Neurology Vol 56 No 5 November 2004

animals before (3 hours after injection: 34.4 � 0.2°C,34.2 � 0.3°C, and 34.3 � 0.3°C; 24 hours after in-jection: 34.6 � 0.2°C, 34.3 � 0.3°C, and 34.6 �0.2°C) or after hypoxia (34.8 � 0.3°C, 34.6 � 0.4°C,and 34.6 � 0.3°C).

Rolipram and cAMP Response Element-BindingProtein Phosphorylated on Ser133Compared with vehicle, a single rolipram precondi-tioning, especially in the 3mg/kg group, caused robustand sustained increases in pCREB for 24 hours in thehippocampus (Fig 6A, B) and cortex (see Fig 6D, E).Total CREB remained unchanged by rolipram treat-ment.

Effects of Rolipram on Posthypoxia cAMP ResponseElement-Binding Protein Phosphorylated on Ser133and Brain-Derived Neurotrophic Factor ExpressionCompared to vehicle, 3mg/kg rolipram precondition-ing caused robust and sustained increases of pCREB in

the bilateral hippocampus (see Fig 6C) and cortex (seeFig 6F) at 24 and 72 hours after hypoxia. Total CREBwas unchanged between the two groups. SustainedCREB phosphorylation may activate downstream tar-gets such as BDNF by binding to CRE located in thepromoter region of the BDNF gene.15,17 We next ex-amined whether posthypoxia rolipram-induced sus-tained CREB phosphorylation could increase BDNFexpression at the transcriptional and translational lev-els. Compared with vehicle, 3mg/kg rolipram precon-ditioning caused enhanced BDNF mRNA (Fig 7A, C)and protein expression (see Fig 7B, D) in the bilateralhippocampus and cortex 24 and 72 hours after hyp-oxia.

DiscussionThis study showed that carotid-artery ligation precon-ditioning can time-dependently provide neuroprotec-tion against hypoxic-ischemic brain injury in neonatal

Fig 6. Effects of rolipram preconditioning on the temporal expression of pCREB levels before and after hypoxia. Compared withvehicle, a single rolipram preconditioning, especially in the 3mg/kg group, caused robust and sustained increases in pCREB levels for24 hours in the hippocampus (A, B) and cortex (D, E). Total CREB remained unchanged by rolipram treatment. Compared withvehicle, 3mg/kg rolipram preconditioning caused robust and sustained increases of pCREB expression in the bilateral hippocampus(C) and cortex (F) 24 hours and 72 hours after hypoxia. Total CREB was unchanged between the two groups. L � left hemi-sphere; R � right hemisphere; V � vehicle; DMSO � dimethylsulfoxide; CREB � cAMP response element–binding protein;pCREB � CREB phosphorylated on Ser133.

Lee et al: pCREB in Brain Preconditioning 619

rats, and that complete behavioral and pathologicalneuroprotection requires 24-hour ligation precondi-tioning. The neuroprotective mechanism of 24-hourcarotid-artery ligation preconditioning is associatedwith robust and sustained CREB phosphorylation.Pharmacological activation of cAMP-CREB signalingcascades by sustained CREB phosphorylation, espe-cially with rolipram, 24 hours before hypoxia protectedrat pups against brain injury. These findings provideevidence that persistent CREB activation is an impor-tant step in the signal transduction that underliescarotid-artery ligation preconditioning against HI in-jury in the immature brain.

In rats, occlusion of one common carotid artery inthe absence of subsequent hypoxia causes only minoralteration in cerebral blood flow in the ipsilateral cere-bral hemisphere.14 When combined with systemic hyp-oxia, however, unilateral carotid artery ligation is asso-ciated with decreased CBF and ischemic brain damage

in the ipsilateral hemisphere.10,13,14,28,34 In addition,delaying the onset of hypoxia for 24 hours after carotidartery ligation prevents the depletion of ATP andmarkedly diminishes the extent of brain injury in theneonatal rats.13,14 Our findings suggest that the neuro-protection induced in the 24-hour ligation precondi-tioning is associated with compensatory cerebrovascularadaptation in the ipsilateral hemisphere during a sub-sequent episode of hypoxia.

Synapse-to-nucleus signaling leading to CREB-mediated transcription is important for neuronal syn-aptic plasticity, memory function, regeneration, andsurvival in response to various stresses.15,16 Because ofits responsiveness to multiple intracellular cascades andto multiple external signals, CREB is well positioned toparticipate in intricate nuclear computations.15 Ser133of CREB is a convergence point for phosphorylationby many kinase-signaling cascades and is necessary fora molecular switch that controls gene expressions. In

Fig 7. Effects of rolipram preconditioning on the temporal expression of brain-derived neurotrophic factor (BDNF) mRNA and pro-tein posthypoxia. Compared with vehicle (DMSO), 3mg/kg rolipram preconditioning caused increased BDNF mRNA (A, C) andprotein (B, D) expression in the bilateral hippocampus and cortex 24 hours and 72 hours after hypoxia. L � left hemisphere; R �right hemisphere. DMSO � dimethylsulfoxide; GAPDH glyceraldehyde 3-phosphate dehydrogenase.

620 Annals of Neurology Vol 56 No 5 November 2004

neurons, CREB is phosphorylated under hypoxia oroxidative stress, suggesting that activation of CREB-dependent survival programs in response to stimulimight represent a cellular form of defense.15,17–20 Dif-ferential susceptibility to hypoxia-induced neuronaldeath correlates well with the ability to sustain CREBphosphorylation. In the hippocampus, temporaryglobal ischemia induces CREB phosphorylation that istransient in cornu ammonis 1 region of the hippocam-pus (CA1) neurons but prolonged in dentate gyrusneurons.17,18 Remarkably, whereas CA1 neurons aredrastically depleted after the ischemic insult, dentategyrus neurons are substantially spared.17,20 In addition,in an adult rat focal ischemia model, persistent CREBphosphorylation was found in the periischemic area ofthe cortex, where neurons remained viable, whereasonly a transient increase in CREB phosphorylation wasnoted in the ischemic core with subsequent neuronaldeath.18 These findings suggest that neuronal survivalis closely associated with persistent rather than tran-sient CREB phosphorylation in ischemia-vulnerable re-gions.

In the adult rat model of mild focal ischemia, en-hanced CREB phosphorylation was observed in thehippocampal CA1 neurons that were subjected tomild ischemia but did not show apparent histologicaldamage. These findings suggest that mild ischemiawas sufficient to trigger CREB phosphorylation in re-sponse to intracellular cAMP or Ca2 in-flux.15,16,18,20,35 Our study shows that unilateral liga-tion of the carotid artery itself in newborn rats doesnot cause significant long-term behavioral or morpho-logical changes. Instead, this form of artery ligationevoked rapid and robust sustained CREB phosphory-lation in the ipsilateral cortex and hippocampus. Ourfindings confirm that the time interval between liga-tion and hypoxic insult is important for the degree ofcerebral damage in the neonatal rat brain.14,34 Theinterval of 1 to 2 hours was established in the classicalLevine rat pup model.10 Although rapid CREB phos-phorylation could be elicited 1 to 3 hours after liga-tion, it afforded no neuroprotection to hypoxic injuryduring this interval. In contrast, we showed signifi-cant neuroprotection at 6 hours and complete neuro-protection at 24 hours. The degree of neuroprotec-tion induced by 24-hour ligation correlated with theduration of CREB phosphorylation. These findingsstrongly suggest that the signaling pathways leadingto sustained CREB phosphorylation and CREB-mediated downstream gene expression constitute animportant neuroprotection mechanism against HI inthe neonatal brain.

Our study also demonstrated that the duration ofCREB phosphorylation determined the degree of neu-roprotection posthypoxia. Marked and persistentpCREB expression occurred 24 hours after hypoxia in

the 24-hour ligation group that later showed completeneuroprotection; in contrast, pCREB was back to con-trol levels 24 hours after hypoxia in the 1-hour ligationgroup that showed obvious brain damage. Our resultsare consistent with study findings that CREB phos-phorylation may not always be sufficient for transcrip-tion activation, and that the duration of CREB phos-phorylation after stimulation is one key determinant ofwhether a stimulus can activate CREB-mediated neu-roprotective gene transcription.15,17–19,35,36

Because reduction of CREB expression during HIpotentially could aggravate the degree of brain dam-age, the duration of decreased CREB by antisenseODN infusions for the 24-hour ligation group waslimited to 6 hours before hypoxia. This might explainwhy the neuroprotection induced by 24-hour ligationpreconditioning is not totally eliminated by blockingCREB. We found that, in the 24-hour ligation group,neuroprotection was significantly reduced in the anti-sense ODN-infused rats compared to the scrambledODN-infused rats, and that antisense ODN infusionalone in the P6 rat pups did not cause significanthemispheric weight reduction compared with scram-bled ODN infusions. The latter strongly arguesagainst the short course of antisense ODN initiating asequence-specific cytotoxic effect in the developingbrain. Taken together, these findings provide evi-dence that CREB is potentially required for the pre-conditioning effect induced in the 24-hour ligationgroup.

Studies have shown that cAMP-mediated signaltransduction is closely associated with neuronal survivalin acute cerebral ischemia.18 Our finding that forskolininjection 24 hours before HI induced significant neu-roprotection by persistently increasing CREB phos-phorylation suggests that activation of cAMP-CREBsignaling is an important neuroprotective cascade inthe immature brain. Furthermore, our outcome studyshowed that a single 3mg/kg rolipram injection 24hours before hypoxia provides profound neuroprotec-tion at both morphological and behavioral levels. Inaddition, compared with the vehicle-pretreated rats,rolipram-preconditioned rats showed higher and sus-tained pCREB expression and its downstream gene(BDNF mRNA) and protein, in the hippocampus andcortex 24 and 72 hours after hypoxia. In the centralnervous system, BDNF plays an important role in reg-ulating the survival and differentiation of neurons dur-ing development.15–17 BDNF is also markedly neuro-protective against neonatal HI injury.27 The BDNFgene is a CREB family target whose protein productfunctions at synapses to control adaptive neuronal re-sponses. Taken together, these findings strongly suggestthat preconditioning by activating the cAMP-CREBsignaling pathway is sufficient to provide neuroprotec-tion in vivo against HI in the immature brain. Further

Lee et al: pCREB in Brain Preconditioning 621

study will be needed to determine whether the cAMP-CREB signaling pathway is neuroprotective when acti-vated after hypoxia.

Acquiring tolerance to HI through pharmacologicalpreconditioning, we believe, has clinical implicationsin perinatal medicine. Drug treatment that acceleratesthe induction of neonatal hypoxic tolerance in high-risk pregnancy may be a potential procedure, but itshould be strong enough to show robust protectionwithout long-term side effects. We found that 24-hour ligation preconditioning induced sustainedCREB phosphorylation, completely protected theneonatal rat brain against HI, and concurrently pre-served brain function. Furthermore, preconditioningby pharmacologically activating cAMP-CREB signal-ing cascades with rolipram also provided marked neu-roprotection at pathological and behavioral levels.This study suggests that CREB activation in vivo isan important event in neuroprotection against HI in-jury in the immature brain. More important, drugsthat mimic the beneficial effects of carotid-artery li-gation preconditioning by activating cAMP-CREBsignaling cascades potentially would provide noveltherapeutic approaches for the treatment of HI braininjury in high-risk newborns.

This study was supported by grants from the Taiwan National Sci-ence Counsel (NSC: 92-2314-B-006-068, 92-2314-B-006-069,C.C.H.), National Health Research Institute (NHRI-EX92-9131NN, C.C.H.), Chi-Mei Medical Research Foundation(L.Y.W.), and National Cheng Kung University Hospital (90-05,C.C.H.).

References1. Volpe JJ. Neurology of the newborn. 4th ed. Philadelphia:

Saunders, 2001.2. Huang CC, Wang ST, Chang YC, et al. Measurement of the

urinary lactate:creatinine ratio for the early identification ofnewborn infants at risk for hypoxic-ischemic encephalopathy.N Engl J Med 1999;34:328–335.

3. Whitelaw A. Systemic review of therapy after hypoxic-ischemicbrain injury in the perinatal period. Semin Neonatol 2000;5:33–40.

4. Berger R, Garnier Y. Pathophysiology of perinatal brain dam-age. Brain Res Rev 1999;30:107–134.

5. Johnston MV, Trescher WH, Ishida A, Nakajima W. Neuro-biology of hypoxic-ischemic injury in the developing brain. Pe-diatr Res 2001;49:735–741.

6. Kirino T. Ischemic tolerance. J Cereb Blood Flow Metab 2002;22:1283–1296.

7. Chen J, Simon R. Ischemic tolerance in the brain. Neurology1997;48:306–311.

8. Gonzalez-Zulueta M, Feldman AB, Klesse LJ, et al. Require-ment of nitric oxide activation of p21ras/extracellular regulatedkinase in neuronal ischemic preconditioning. Proc Natl AcadSci USA 2000;97:436–441.

9. Dirnagl U, Simon RP, Hallenbeck JM. Ischemic tolerance andendogenous neuroprotection. Trends Neurosci 2003;26:248–254.

10. Rice J, Vanucci R, Brierley J. The influence of immaturity onhypoxic-ischemic brain damage in the rat. Ann Neurol 1981;9:131–141.

11. Gidday JM, Fitzgibbons JC, Shah AR, Park TS. Neuroprotec-tion from ischemic brain injury by hypoxic preconditioning inthe neonatal rat. Neurosci Lett 1994;168:221–224.

12. Bergeron M, Gidday JM, Yu AY, et al. Role of hypoxia-inducible factor-1 in hypoxia-induced ischemic tolerance inneonatal rat brain. Ann Neurol 2000;48:285–296.

13. Hylton CM, Pesenson MA, Welsh FA. Adaptive preservation ofATP and tolerance to hypoxia following carotid artery ligationin the immature rat. J Cereb Blood Flow Metab 1995;15:1137–1140.

14. Bronner G, Mitchell K, Welsh FA. Cerebrovascular adaptationafter unilateral carotid artery ligation in the rat: preservation ofblood flow and ATP during forebrain ischemia. J Cereb BloodFlow Metab 1998;18:118–121.

15. Lonze BE, Ginty DD. Function and regulation of CREB familytranscription factors in the nervous system. Neuron 2002;35:605–623.

16. Shaywitz AJ, Greenberg ME. CREB: a stimulus-induced tran-scription factor activated by a diverse array of extracellular sig-nals. Annu Rev Biochem 1999;68:821–861.

17. Walton MR, Dragunow I. Is CREB a key to neuronal survival?Trends Neurosci 2000;23:48–53.

18. Tanaka K. Alteration of second messengers during acute cere-bral ischemia—adenylate cyclase, cyclic AMP-dependent pro-tein kinase, and cyclic AMP response element binding protein.Prog Neurobiol 2001;65:173–207.

19. Mabuchi T, Kitagawa K, Kuwabara K, et al. Phosphorylation ofcAMP response element-binding protein in hippocampal neu-rons as a protective response after exposure to glutamate invitro and ischemia in vivo. J Neurosci 2001;21:9204–9213.

20. Hara T, Hamada J, Yano S, et al. CREB is required for acqui-sition of ischemic tolerance in gerbil hippocampal CA1 region.J Neurochem 2003;86:805–814.

21. Guzowski JF, McGaugh JL. Antisense oligodeoxynucleotide-mediated disruption of hippocampal cAMP response elementbinding protein levels impairs consolidation of memory for wa-ter maze training. Proc Natl Acad Sci USA 1997;94:2693–2698.

22. Zhang JJ, Okutani F, Inoue S, Kaba H. Activation of the cyclicAMP response element-binding protein signaling pathway inthe olfactory bulb is required for the acquisition of olfactoryaversive learning in young rats. Neuroscience 2003;117:707–713.

23. Conti M, Jin SL. The molecular biology of cyclic nucleotidephosphodiesterases. Prog Nucleic Acid Res Mol Biol 1999;63:1–38.

24. Barad M, Bourtchouladze R, Winder DG, et al. Rolipram, atype IV-specific phosphodiesterase inhibitor, facilitates the es-tablishment of long-term potentiation and improves memory.Proc Natl Acad Sci USA 1998;95:15020–15025.

25. Nagakura A, Niimura M, Takeo S. Effects of a phosphodiesteraseIV inhibitor rolipram on microsphere embolism-induced defectsin memory function and cerebral cyclic AMP signal transductionsystem in rats. Br J Pharmacol 2002;135:1783–1793.

26. Wei F, Qiu CS, Kim SJ, et al. Genetic elimination of behav-ioral sensitization in mice lacking calmodulin-stimulated adeny-lyl cyclases. Neuron 2002;36:713–726.

27. Han BH, Holtzman DM. BDNF protects the neonatal brainfrom hypoxic-ischemic injury in vivo via the ERK pathway.J Neurosci 2000;20:5775–5781.

622 Annals of Neurology Vol 56 No 5 November 2004

28. Vannucci RC, Lyons DT, Vasta F. Regional cerebral bloodflow during hypoxia-ischemia in immature rats. Stroke 1988;19:245–250.

29. Chang YC, Huang AM, Kuo YM, et al. Febrile seizures impairmemory and cAMP response-element binding protein activa-tion. Ann Neurol 2003;54:706–718.

30. Paxinos G, Watson C. The rat brain in stereotaxic coordinates.2nd ed. San Diego: Academic Press, 1986.

31. Gidday JM, Fitzgibbons JC, Shah AR, et al. Reduction in ce-rebral ischemic injury in newborn rat by potentiation of endog-enous adenosine. Pediatr Res 1995;38:306–311.

32. Viola H, Furman M, Izquierdo LA, et al. PhosphorylatedcAMP response element-binding protein as a molecular markerof memory processing in rat hippocampus: effect of novelty.J Neurosci 2000;20:RC112.

33. Phillis JW, Estevez AY, O’Regan MH. Protective effects ofthe free radical scavengers, dimethyl sulfoxide and ethanol, incerebral ischemia in gerbils. Neurosci Lett 1998;13:244:109 –111.

34. Dwyer B, Nishimura R, Fujikawa DG. Cerebral hypoxia-ischemia in immature rats: methodological considerations. ExpNeurol 1988;99:772–777.

35. Tanaka K, Nogawa S, Nagata E, et al. Persistent CREB phos-phorylation with protection of hippocampal CA1 pyramidalneurons following temporary occlusion of the middle cerebralartery in the rat. Exp Neurol 2000;161:462–471.

36. Ahn S, Riccio A, Ginty DD. Spatial consideration for stimulus-dependent transcription in neurons. Annu Rev Physiol 2000;62:803–823.

Lee et al: pCREB in Brain Preconditioning 623