Ž .Brain Research 794 1998 119–126

Research report

Competing processes of cell death and recovery of function followingischemic preconditioning

Paul Dooley, Dale Corbett )

DiÕision of Basic Medical Sciences, Faculty of Medicine, Memorial UniÕersity of Newfoundland, St. John’s, NF, Canada A1B 3V6

Accepted 17 February 1998

Abstract

The goal of the present study was to determine the neuroprotective efficacy of ischemic preconditioning using behavioral,electrophysiological and histological endpoints at various time points up to 90 days postischemia. Gerbils were exposed to a brief,

Ž .non-injurious episode of forebrain ischemia 1.5 min on each of 2 consecutive days. Three days following this preconditioningprocedure, the animals received a 5 min occlusion. Other animals underwent sham surgery or a 5 min occlusion without preconditioning.

Ž .Ischemic preconditioning appeared to provide striking histological protection at both rostral ;80% and ;67% of sham and posteriorŽ .levels of hippocampus ;94% and ;78% of sham at 3 and 10 days survival, respectively. However, in spite of the near normal number

of CA1 neurons, animals displayed marked impairments in an open field test of habituation as well as reduced dendritic field potentials inthe CA1 area. Additionally, in ischemic animals the basal and apical dendritic regions of CA1 were nearly devoid of the cytoskeletal

Ž .protein microtubule associated protein 2 MAP2 . Staining levels of MAP2 in preconditioned and sham animals were similar. Withincreasing survival time, open field behavior as well as CA1 field potential amplitude recovered. Nonetheless, CA1 cell death in ischemic

Ž .preconditioned animals continued over the 90-day survival period P-0.05, vs. sham levels . Ischemic preconditioning provides aŽsignificant degree of neuroprotection characterized by a complex interplay of protracted cell death and neuroplasticity recovery of

.function . These competing processes are best elucidated using a combination of functional and histological endpoints as well as multipleŽ .and extended survival times i.e., greater than 7–10 days . q 1998 Elsevier Science B.V. All rights reserved.

Keywords: Cerebral ischemia; MAP2; CA1; Recovery of function; Neuroplasticity; Neuroprotection

1. Introduction

Following global ischemia, hippocampal CA1 cells de-generate over 2–4 days, a process termed delayed neuronal

w xdeath 28,44 . This time frame provides a window ofopportunity for rescuing these cells and thereby preventing

w xensuing learningrmemory impairments 52,56,57 . Whilefunctional deficits are of primary clinical importance, mostresearchers have relied solely upon the degree of CA1 cellloss as their outcome measure. It is essential that func-tional as well as histological endpoints be used since afterischemia CA1 cells may be functionally incompetent while

w xmaintaining normal morphology 6,22,23 . A number ofŽ .behavioral tests e.g., T-maze and open field have been

found effective in determining the level of hippocampalw xfunction 13,52,53 following ischemic insult. However,

such tests may lose power with extended survival times

) Corresponding author. Fax: q1-709-737-7010; E-mail:corbett@morgan.ucs.mun.ca

due to functional compensation by undamaged brain re-w x Žgions 17 . Thus, electrophysiological evaluation e.g., CA1

.field potentials can be used in addition to behavioralendpoints to more directly ascertain the functional compe-tence of a particular neuronal system following ischemic

w xinsult, especially at extended survival times 40 .Equally problematic to the near exclusive reliance upon

Ž .histological endpoints is the use of short e.g., 3–7 dayssurvival times. It is possible that a particular treatment mayhave merely extended the interval over which vulnerablecells die. For example, in both gerbil and rat models oftransient forebrain ischemia, the AMPA antagonist NBQX,as well as the omega conotoxin, SNX-111, are neuropro-

Ž .tective at relatively short survival times e.g., 1 to 7 daysw x8,9,34,46,47 . However, it has recently been shown thatNBQX and SNX-111 provide little CA1 protection at 10w x w x41 or 28 days 33 postischemia. Similar findings havebeen noted with focal ischemia where the NMDA antago-nist MK-801 appears to delay, rather than prevent cell

w xdeath 51 .

0006-8993r98r$19.00 q 1998 Elsevier Science B.V. All rights reserved.Ž .PII S0006-8993 98 00220-0

( )P. Dooley, D. CorbettrBrain Research 794 1998 119–126120

While hypothermia appears to be the gold standard ofw xneuroprotection against ischemic insult 12,13,21,40 , other

interventions also offer robust neuroprotection. For exam-ple, ischemic preconditioning which consists of exposing

Ž .the animal to brief 1.5–2 min , non-injurious global is-chemic episodes prior to a subsequent severe insult pro-

w xtects CA1 neurons 26,27,31 and reduces infarct size inw xfocal ischemia 48 . However, at present it is unknown

whether this protection includes functional preservation orif the protection is even permanent since survival timeshave typically been 7 days.

Recent evidence from our laboratory has indicated thatischemic preconditioning, while providing substantial CA1protection at 10 days, does not provide any appreciablefunctional preservation as determined by behavioral assess-

w xment 16 . Furthermore, following ischemic precondition-ing, CA1 cell counts declined significantly from 10 to 30

w xdays survival 16 . The purpose of the present study there-fore was to investigate the nature of the early dissociationbetween histological and behavioral integrity using indica-

Žtors of functional viability e.g., behavior, electrophysio-.logical analysis that have been validated in previous

w xstudies 13,40 . In addition, we utilized immunocytochemi-cal techniques to demonstrate ischemia-induced changes in

Ž .microtubule-associated protein 2 MAP2 which is local-w xized primarily in dendrites 11 and is a sensitive indicator

w xof ischemic injury 2,30 . Finally, we sought to determinewhether the protracted cell loss observed following is-chemic preconditioning will continue when extended sur-

Ž .vival times e.g., 90 days are used.

2. Materials and methods

2.1. Animals

These experiments were carried out in accordance withthe guidelines established by the Canadian Council onAnimal Care and were also approved by the MemorialUniversity of Newfoundland Animal Care Committee.

ŽFifty-three female, Mongolian gerbils High Oak Ranch,.Baden, ON ranging from 12 to 16 weeks old and weigh-

ing 50 to 60 g were used in the study.

2.2. Brain temperature measurement

Brain temperature measurement was carried out as pre-w xviously described 12 . Briefly, 4 days prior to surgery,

Žgerbils were anesthetized with sodium pentobarbital 65.mgrkg and a 5 mm, 20 gauge guide cannula was im-

planted overlying the dorsomedial striatum at the duralsurface. Two days later, wireless temperature probesŽ .model XM-FH, Mini-Mitter, Sunriver, OR were insertedinto the striatum at a level approximating the depth of thehippocampus and brain temperature was monitored contin-

uously for a 3-h period to establish a baseline temperatureprofile.

2.3. Cerebral ischemia

Anesthesia was induced using a mixture of 2%Halothane in 30% oxygen and 70% nitrous oxide andanimals were subsequently maintained using 1.5%Halothane. Brain temperature was measured as previously

w x Ždescribed 12 and maintained at normothermia i.e., 36.0–.36.58C . Gerbils were subjected to either 5 min of cerebral

Ž . Ž .ischemia I , sham surgery S , or ischemic precondition-Ž .ing IP . None of the animals were subjected to precondi-

tioning ‘only’ insults because previous data from thislaboratory have indicated that this procedure does notresult in histological or functional deficits, even with longŽ . w x30 day survival times 16 . Cerebral ischemia was in-duced by isolation of the common carotid arteries througha ventral midline incision in the neck followed by bilateralocclusion of the arteries using vascular clamps for thespecified duration. Sham operated animals underwent ex-actly the same procedure except there was no arterialocclusion. Ischemic preconditioning consisted of two brief,

Ž .non-injurious ischemic insults 1.5 min each on consecu-tive days followed 3 days later by a 5 min occlusion.During all occlusions brain temperature was maintained at

Ž36.0–36.58C by wrapping a hot water blanket model.TP-3E, Gaymar Industries, Orchard Park, NY around the

head. Likewise, rectal temperature was maintained at37.0–38.08C by a separate homeothermic blanket systemŽ .Harvard Apparatus, South Natick, MA . Following is-chemia or sham surgery, all animals were maintained atnormothermia using an overhead lamp until they were able

Ž .to regulate their own temperature 15 to 30 min . Shamand I animals survived for 30 days postsurgery. This singlesurvival time was used since it has previously been shown

Žthat near complete loss of CA1 neurons and therefore the.associated CA1 dendritic field potentials has taken place

w xby 3 days postischemia 28,44 . The 30 day survival wasused such that behavioral data, for comparison purposes,could be acquired. Furthermore, we have previously shownw x16 that the behavioral habituation profiles and CA1 countsof sham operated gerbils between 10 and 30 days post-surgery are indistinguishable from normal unoperated ani-mals. This is also the case with animals subjected to only

w xthe two episodes of ischemic preconditioning 16 . IPŽ .animals were subdivided into groups surviving for 3 IP 3 ,

Ž . Ž . Ž . Ž10 IP 10 , 30 IP 30 and 90 IP 90 days after the final 5.min carotid artery occlusion.

2.4. BehaÕior testing

On days 3, 7, 10 and 30, animals were tested for 10 minŽ 3.in an open field apparatus 72=76=57 cm which was

divided into 25 equal squares. A visual tracking system

( )P. Dooley, D. CorbettrBrain Research 794 1998 119–126 121

Ž .HVS Systems, Kingston, UK recorded the number ofsquares entered per minute. Testing was carried out in asound proof room in which conditions were maintainedconstant throughout the experiment. The open field test hasbeen shown to be a sensitive indicator of an animal’s

w xability to habituate to a novel environment 3,5,53 .

2.5. Electrophysiological assessment

Following the last day of behavioral testing, gerbilswere re-anesthetized and brain temperature was reduced to308C with a cold water blanket. Following decapitation,brains were removed and bisected, with one hemisphere

Ž .being immersed in cold 48C formalin for later histologi-cal analysis. The hippocampus was dissected free from theremaining hemisphere and 500 mm transverse sectionswere cut using a tissue chopper. Slices were placed in a

w x Ž .modified 1,40,43 artificial cerebrospinal fluid ACSFŽ .solution containing in mmolrl : sucrose, 215.8; KCl, 3.5;

CaCl , 2.0; NaHCO , 25.0; MgCl , 1.3; glucose, 11.0;2 3 2Ž .that was bubbled with 95% O r5% CO pH 7.3–7.4 at2 2

room temperature for 10 min. Slices were then transferredto an ACSF solution which differed from the modifiedACSF only in that the sucrose was replaced by NaClŽ .126.0 mmolrl . Slices were incubated for at least 1 hprior to experimentation.

Ž .Excitatory postsynaptic potentials EPSPs wererecorded within stratum radiatum in a fluid interface cham-

Ž .ber Fine Science Tools, Vancouver, BC perfused withoxygenated ACSF at a rate of 2 ml per minute andmaintained at 338C. Ultrasmall, concentric bipolar stimu-

Žlating electrodes 100 mm, Frederick Haer, Brunswick,.ME were positioned for orthodromic stimulation of the

Schaffer collaterals. Stimulation parameters consisted of0.1 ms constant current pulses delivered at 0.05 Hz. Glass,micropipette recording electrodes filled with 2 M NaCland having a tip diameter of approximately 20 mm and aresistance of 1.0 to 1.5 MV were positioned in stratumradiatum. Responses were amplified, digitized and storedfor later analysis as well as being displayed on an oscillo-scope. A slice was considered viable and recordings usedfor analysis if a population spike could be elicited from thedentate granule cells in response to stimulation of the

Ž .perforant path 0.1 ms pulses at 0.05 Hz . Dentate granulecells are not injured by the duration of ischemia employedin the present experiments.

2.6. Histology

The hemisphere that was immersion fixed in cold for-malin was paraffin embedded and two series of 6 mmsections were cut. One series was stained with haema-toxylin and eosin and neurons exhibiting a distinct nucleusand lack of shrinkage or eosinophilia were counted inmedial, middle and lateral regions of CA1 in sections

Ž w xtaken from two levels of hippocampus 1.7 level A andw x .2.2 mm level B posterior to bregma by placing a 200

mm long grid over these regions and counting the cellsw xwithin the grid 13 . Similarly, at 2.7 mm posterior to

Ž .bregma level C , cells were counted in the middle regiononly. The second series of sections was immunostained

Ž .against MAP2 Sigma, St. Louis, MO using the avidin–Ž .biotin–peroxidase complex procedure Vectastain ABC .

The specificity of this antibody has previously been estab-w xlished 55 . Furthermore, repeated tests for non-specific

staining were negative. The intensity of MAP2 stainingwas quantified by calculating the relative optical densityŽ .ROD in CA1 using NIH Image software running on aMacintosh 7600 computer. Briefly, the ROD of the CA1dendritic field was normalized to the ROD of the corpus

Ž .callosum CC using the following formula: CA1 RODyCC RODrCA1 ROD. Assessment of MAP2 ROD was

Ž .carried out in stratum radiatum of rostral level A sectionsonly without knowledge of treatment condition. Previouspublications from this laboratory have confirmed that thepattern of CA1 cell loss following ischemia in the gerbilw x13,40 and in particular following ischemic precondition-

w xing 16 , does not differ between hemispheres. Thus, wefeel it appropriate to compare physiological measures fromone hemisphere with histological measures from the other.

2.7. Statistics

Repeated measures analysis of variance was used toanalyze the open field data. Based on our previous resultsw x16 , specific comparisons between treatment means werecarried out on individual test days using planned compari-son analysis. CA1 fEPSPs, CA1 cell counts and MAP2ROD were analyzed using a simple ANOVA. Individualpost-hoc comparisons were evaluated using the Newman–Keuls test. Behavioral and histological data were obtainedfrom the same animals. Electrophysiological and MAP2measurements were derived from subsets of these animals.

3. Results

3.1. Temperature

Mean preischemic brain temperature for all animals was36.48C during a 3-h baseline monitoring period. Meanintraischemic brain temperature ranged from 36.28C to36.48C in all groups. All animals, except shams, showed a

Ž .slight rise ;0.7–0.98C in brain temperature upon reper-w xfusion which is characteristic of this animal model 14 .

However, postischemic brain temperature recorded over a24-h period following either 5 min occlusion or sham

Žsurgery did not differ between groups Fs1.680, Ps.0.1623 . No evidence of seizure activity was observed in

any of the animals following ischemia.

( )P. Dooley, D. CorbettrBrain Research 794 1998 119–126122

3.2. BehaÕior

Mean open field activity levels for each group onpostischemic days 3, 7, 10, and 30 are presented in Fig. 1.Repeated measures analysis of variance indicated a signifi-

Ž .cant day effect Fs24.574, P-0.01 , likely the result ofhabituation to the open field as indicated in Fig. 1. Both Iand IP animals were significantly more active than S on

Ž .test days 3 and 7 P-0.01 and on Day 10 except for theŽ .IP 30 group. This was also the case on day 30 P-0.05

though the differences between these groups was consider-ably less at this time. While obvious similarities in open

Žfield behavior exist between the IP and I groups slow rate.of habituation , animals in the I group were significantly

Ž .more active on all test days P-0.05 .

3.3. Electrophysiological assessment

ŽFollowing the last day of behavioral testing dependent.on group orthodromic field EPSPs were recorded from

CA1 pyramidal cells in hippocampal slices taken from asubgroup of animals. Maximal amplitudes from three stim-ulation trials were recorded and averaged from slices atanterior, middle and posterior CA1. These means wereaveraged to yield an overall mean of CA1 fEPSP ampli-

Žtude for each animal. As expected, the magnitude maxi-

Ž .Fig. 1. Total open field activity scores mean"S.D. within each 10 minŽ . Ž .test session for the ischemia I , ischemic preconditioned IP and sham

Ž .S groups on postischemic test days 3, 7, 10 and 30. Locomotor activityscores for ischemic preconditioned animals did not differ across survivaltimes and were thus pooled for analysis. The number of ischemicpreconditioned animals on test days 3, 7, 10 and 30 were 37, 26, 25 and

Ž . Ž .18, respectively. Sham ns7 and ischemic ns9 animals were in-cluded on all test days. x s P -0.01; Is P -0.05 with respect tosham operated animals.

Ž .Fig. 2. Mean CA1 fEPSP amplitudes "S.D. recorded in hippocampalŽ . Žslices from sham S; ns6 and ischemic without preconditioning I;

.ns9 animals at 30 days survival and ischemic preconditioned animals atŽ . Ž . Ž . Ž3 IP3; ns6 , 10 IP10; ns4 , 30 IP30; ns5 and 90 days IP 90;

.ns9 postischemia. x s P -0.01; Is P -0.05 with respect to shamoperated animals.

.mum amplitude of the elicited responses from animalsŽ . Žexposed to the 5 min ns9 ischemia no precondition-

.ing was greatly reduced in comparison with those recordedŽ . Žfrom sham operated ns6 animals y1.06 vs. y4.68

.mV, respectively, P-0.01 . However, the maximum am-plitude of the fEPSPs elicited from ischemic precondi-tioned animals increased as a function of survival timeŽ .Fig. 2 from y2.36 mV and y3.06 mV on postischemic

Ž . Ž . Ždays 3 ns6 and 10 ns4 , respectively P-0.05 vs.. Ž .sham to y3.98 mV at 30 days ns5 survival which did

not differ significantly from sham. However, this recoverytrend appeared to be transient as responses recorded fromischemic preconditioned animals surviving for 90 daysŽ .ns9 were reduced to y3.4 mV which was significantly

Ž .lower than sham responses P-0.05 .

3.4. Histology

Consistent with functional endpoints, CA1 cell countsconfirmed severe neuronal loss at all rostral–caudal levelsin the 5-min ischemic group vs. sham operated animalsŽ .P-0.01 . Cell counts in preconditioned gerbils variedaccording to the rostral–caudal level of the hippocampus

Ž .and with survival time. At the most rostral level level A ,cell counts from ischemic preconditioned animals at 3 dayssurvival did not differ significantly from those of shams

Ž .but were significantly lower P-0.05 at 10 and 30 daysŽ .survival Fig. 3 . Similar to the decline in fEPSP amplitude

at 90 days, ischemic preconditioned animals showed asignificant reduction in the number of viable CA1 neurons

Ž .at 90 days survival vs. shams, P-0.05 . At middle and

( )P. Dooley, D. CorbettrBrain Research 794 1998 119–126 123

Ž .Fig. 3. Number of CA1 cells present in one hemisphere at 1.7 level A ,Ž . Ž . Ž2.2 level B and 2.7 mm level C posterior to bregma. Sham Sham;. Ž .ns7 and ischemic without preconditioning I; ns9 animals were

sacrificed at 30 days survival. Ischemic preconditioned animals wereŽ . Ž . Ž .sacrificed at 3 IP3; ns11 , 10 IP10; ns8 , 30 IP30; ns9 and 90

Ž .days IP90; ns9 postischemia. Each symbol represents the total numberof CA1 cells counted for one gerbil. Horizontal lines indicate the group

Ž .mean. The I group had significantly P -0.05 fewer CA1 cells than anyŽ .of the other groups irrespective of the level i.e., A, B or C assessed. The

IP90 group had significantly fewer cells than sham animals at both levelsŽ .A and B P -0.05 whereas animals allowed to survive for 10 or 30 days

Ž .IP10 and IP30 groups were found to have fewer CA1 cells than shamsŽ .P -0.05 only at level A. Cell counts in the IP3 animals were notdifferent from shams.

Ž .posterior hippocampal levels levels B and C only theCA1 cell counts from level B of the IP 90 group weresignificantly less than sham cell counts. At all survivaltimes assessed, the ischemic preconditioned groups hadsignificantly more intact CA1 neurons than the 5 min

Ž .ischemic animals P-0.01 .A semi-quantitative assessment of the immunocyto-

chemical localization of MAP2 in rostral hippocampusŽ .level A indicated that after ischemia there was a neartotal depletion of MAP2 in the CA1 dendritic regionŽ .P-0.01, vs. sham , while in the ischemia resistant CA3

Ž .region it remained well preserved Table 1 . While MAP2staining in preconditioned animals was somewhat lowerthan that observed in shams, the differences were notsignificant. Like the cell counts, at the most caudal level

Table 1Ž .CA1 MAP2 relative optical density ROD measurements

Ž .Group CA1 MAP2 ROD mean"S.D.aI 0.02"0.04

S 0.69"0.08IP3 0.44"0.28IP10 0.42"0.28IP90 0.47"0.24

Ž .MAP2 staining of rostral CA1 level A is expressed as the mean"S.D.ROD of the CA1 apical dendritic field was normalized to the ROD of the

Ž .corpus callosum CC according to the formula: CA1 RODyCCRODrCA1 ROD.a Indicates P -0.01 vs. all other groups.

Ž .assessed level C , the intensity of the MAP2 signal wasnot greatly affected by the ischemic insult.

4. Discussion

Ischemic preconditioning has repeatedly been shown toconfer a substantial degree of histological preservation to

w xthe vulnerable CA1 region against ischemia 26,27,31,35 .However, it has not been determined whether this protec-tion is permanent or more importantly, whether it extendsinto the functional domain. The present results demonstrate

Ž .that cell counts derived from Nissl stained cell bodies donot necessarily reflect the degree of functional integrity. At3 and 10 days survival, IP animals, while exhibiting

Žexceptional cellular preservation 79.7 and 66.9%, respec-.tively when levels A and B are combined were signifi-

cantly impaired on both behavioral and electrophysiologi-cal measures.

The present study also underscores the necessity ofusing extended survival times. Ischemic preconditioningwas found to greatly extend the interval over which CA1cells die following ischemia, particularly in the rostral

Ž .hippocampus i.e., level A . At 3 days survival, cell countsfrom IP animals were not significantly different fromsham, however, by 90 days the number of viable neurons

Ž .declined significantly P-0.05 from 77.4% in the 3-daysurvival group to 46.8% of sham values. Neuroprotection

Ž .was more persistent at more caudal levels levels B and Cwhere the CA1 cell counts were not different from shams

Ž .except at level B for the 90 day survival group P-0.05 .Delayed cell loss with increasing survival time has alsobeen observed after several other intervention strategiesŽ . w xe.g., the AMPA antagonist NBQX 19,33,41 . The in-crease in cell loss and attenuated field potential amplitudein the IP 90 animals cannot be attributed to differencesbetween it and the other IP groups incurred during induc-tion of ischemia since intraischemic brain temperature didnot differ between groups. Furthermore, there were nobehavioral differences between the IP 90 and other IPgroups that would have indicated variations in the efficacyof the preconditioning insults.

( )P. Dooley, D. CorbettrBrain Research 794 1998 119–126124

The observed dissociation between histological preser-vation and function is not unique to ischemic precondition-ing since similar results have been reported elsewhere

w xusing other intervention approaches 6,18,24,32,36 . Ani-mals receiving 5 min of ischemia without preconditioningwere significantly more active than sham animals on all

w xtest days 3,7,10,30 . Over the course of testing, however,the I group, like IP animals, exhibited a gradual decline inactivity scores. At 30 days, however, behavioral scores forIP animals approximated those of sham. It is thereforepossible that behavioral recovery results from compensa-

Ž .tion by uninjured structures within e.g., caudal CA1 ordistal to the hippocampus.

Previous studies have used changes in neuronal trans-mission, assessed electrophysiologically, to evaluate thechanges which occur following an ischemic insultw x10,25,29,40,50 . This approach provides a more directmeasure of hippocampal function than behavioral testing.In the present study, CA1 fEPSPs recorded in slices from Ianimals were significantly reduced compared to shams at

Žall survival times. At early survival times 3 and 10 days.postischemia IP animals had fEPSPs which were signifi-

Žcantly attenuated relative to sham maximum amplitudes.were 51.3 and 65.4% of sham, respectively, P-0.05 .

However, by 30 days, fEPSP amplitudes had recovered tosham levels suggesting true neuroprotection as a result ofischemic preconditioning. Similar evidence of a protractedfunctional recovery has been observed following inductionof epileptiform activity in rat hippocampal slice culturew x39 . However, when the efficacy of ischemic precondi-tioning was assessed at 90 days, not only had CA1 cellcounts continued to decline but the amplitude of fEPSPswere once again found to be significantly attenuated rela-

Ž .tive to sham values P-0.05 . It would appear that whileinjured cells recover some degree of functional capacityfollowing ischemic preconditioning, the recovery processis not completely sustained, again illustrating the necessityof assessing neuroprotection at extended survival times. Itis possible that the observed delayed neuronal death re-flects these cells’ inability to withstand even low levelstimulation. Such enhanced susceptibility of CA1 neuronsto excitatory input has previously been observed followingischemic insult, though at a much shorter survival timew x49 .

At present it is unknown whether this trend of contin-ued CA1 neuronal death would continue at still greater

Ž .survival times e.g., 6 months or 1 year or if the loss willcease at some point. For example, long duration postis-chemic hypothermia has been shown to result in protectionof CA1 neurons which decreases from 90% at 1 month

w xpostischemia to 70% at 6 months 13 but does not con-tinue to decline since the protection observed at 12 months

w xis also 70% 15 . Nonetheless, ischemic preconditioningappears to provide a far greater degree of protection than

w xany currently available drug therapy 7,8,34 .In an attempt to identify the origin of the early dissocia-

tion between histological and functional preservation,changes in dendritic morphology were assessed since CA1dendrites and their spines are particularly susceptible to

w xischemic injury 37 . Furthermore, it is widely acceptedthat processes mediating memory and habituation occur at

w xthe synaptic level 4,20 . Brain sections were processed forimmunocytochemical localization of the structural proteinMAP2 which is believed to be involved in the crosslinking of microtubules necessary for stabilizing dendritic

w xstructure and regulating plasticity 38 . This protein isnormally present in large quantities in the CA1 dendriticregion and is degraded following severe ischemic insult

w xprior to cell death 2,37 . Thirty days after the 5 minocclusion in the I group when most CA1 neurons havedied, we observed a total loss of MAP2 in the apicaldendritic region of CA1. In contrast, tissue from shamoperated animals exhibited an intense localization of MAP2in this area. The pattern of MAP2 localization in IPanimals, though reduced was similar to that observed insham animals at 30 days. Thus, it would appear thatdisruption of dendritic function, at least as revealed byMAP2 staining, cannot account for the behavioral andelectrophysiological abnormalities observed in the precon-ditioned animals. Previous evidence has shown that is-chemia induced dendritic abnormalities result in impairedelectrophysiological function within the hippocampus, in-cluding reduced CA1 field potential amplitude and loss of

w xLTP 22 . It may be that some subtle disruption in den-Ž .dritic morphology e.g., altered spine density not detected

using the present techniques is responsible for the attenu-Žated fEPSPs in the IP group at 3 and 10 days postis-

.chemia . Furthermore, given the importance of synapticintegrity to the induction of LTP and the postulated rela-tionship of the latter to learning and memory, it is possiblethat such dendritic abnormalities in the IP animals mightalso be responsible for the open field impairments in thepreconditioned animals.

The mechanisms mediating delayed neuronal death fol-lowing ischemia have yet to be determined. Likewise, theprocesses responsible for the neuroprotective effect ofischemic preconditioning remain elusive. Based on recentevidence it is possible that mitochondria may play animportant role. It has previously been observed that in

Žresponse to high levels of neuronal activation e.g., gluta-. 2qmate stimulation the resulting increase in free Ca is at

least partially buffered by mitochondria which sequester2q w xthe Ca 54 . However, if this stimulation is of sufficient

strength, levels of Ca2q may exceed the buffering capacitywhich in turn results in degradation of the mitochondrial

w xmembrane potential 45 which renders the organelle non-functional. Recent evidence has indicated that ischemicpreconditioning may prevent this malfunction by enhanc-ing the buffering capacity of the mitochondria in response

2q w xto a high Ca load 42 thereby potentially sparing thefunction of this structure which is essential to the mainte-nance of cellular viability.

( )P. Dooley, D. CorbettrBrain Research 794 1998 119–126 125

The present study has clearly demonstrated that is-chemic preconditioning may not be as efficacious a neuro-protectant as has typically been reported. Nonetheless,elucidation of the mechanisms mediating the neuroprotec-

Žtive effect of ischemic preconditioning which is more.effective than current drug therapies may enable develop-

ment of pharmacological agents which provide a beneficialand practical treatment strategy. Most importantly, thepresent results provide compelling evidence for using bothmultiple indices of neuroprotection as well as extendedsurvival times in evaluating putative neuroprotectiveagents. Only when results of these methods of assessmentconverge can clear cut conclusions be drawn regarding thetrue efficacy of a particular neuroprotective interventionstrategy.

Acknowledgements

Ž . Ž .Research DC and studentship PD support was pro-vided by the Medical Research Council of Canada. Theauthors thank Suzanne Evans and Kathy McKay for tech-nical assistance and Dr. Suzanne Nurse and Jennifer Dow-den for helpful comments on the manuscript. Dr. JohnEvans provided helpful statistical advice.

References

w x1 G.K. Aghajanian, K. Rasmussen, Intracellular studies in the facialnucleus illustrating a simple new method for obtaining viable mo-

Ž .toneurons in adult rat brain slices, Synapse 3 1989 331–338.w x2 M. Aoki, K. Abe, T. Yoshida, A. Hattori, A. Kogure, Y. Itoyama,

Early immunohistochemical changes of microtubule based motorproteins in gerbil hippocampus after transient ischemia, Brain Res.

Ž .669 1995 189–196.w x3 A.M. Babcock, D.A. Baker, R. Lovec, Locomotor activity in the

Ž .ischemic gerbil, Brain Res. 625 1993 351–354.w x4 T.V.P. Bliss, G.L. Collingridge, A synaptic model of memory:

Ž .long-term potentiation in the hippocampus, Nature 361 1993 31–38.w x5 C.A. Boast, S.C. Gerhardt, G. Pastor, J. Lehmann, P.E. Etienne,

J.M. Liebman, The N-methyl-D-aspartate antagonists CGS-19755and CPP reduce ischemic brain damage in gerbils, Brain Res. 442Ž .1988 345–348.

w x6 H.-W. Bothe, H.-J. Bosma, H. Hofer, K.-A. Hossmann, W.F. Anger-meier, Selective vulnerability of hippocampus and disturbances ofmemory storage after mild unilateral ischemia of gerbil brain, Stroke

Ž .17 1986 1160–1163.w x7 A. Buchan, H. Li, W.A. Pulsinelli, The N-methyl-D-aspartate antag-

onist, MK-801, fails to protect against neuronal damage caused bytransient, severe forebrain ischemia in adult rats, J. Neurosci. 11Ž .1991 1049–1056.

w x8 A.M. Buchan, H. Li, S. Cho, W.A. Pulsinelli, Blockade of theAMPA receptor prevents CA1 hippocampal injury following severebut transient forebrain ischemia in adult rats, Neurosci. Lett. 132Ž .1991 255–258.

w x9 A.M. Buchan, D. Xue, Z.-G. Huang, K.H. Smith, H. Lesiuk, De-layed AMPA receptor blockade reduces cerebral infarction induced

Ž .by focal ischemia, NeuroReport 2 1991 473–476.w x10 G. Busaki, T.F. Freund, F. Bayardo, P. Somogyi, Ischemia-induced

changes in electrical activity of the hippocampus, Exp. Brain Res. 78Ž .1989 268–278.

w x11 A. Caceres, G. Banker, O. Steward, L. Binder, M. Payne, MAP2 is

localized in the dendrites of hippocampal neurons which develop inŽ .culture, Dev. Brain Res. 13 1984 314–318.

w x12 F. Colbourne, D. Corbett, Delayed and prolonged post-ischemicŽ .hypothermia is neuroprotective in the gerbil, Brain Res. 654 1994

265–272.w x13 F. Colbourne, D. Corbett, Delayed postischemic hypothermia: a six

month survival study using behavioral and histological assessmentsŽ . Ž .of neuroprotection, J. Neurosci. 15 11 1995 7250–7260.

w x14 F. Colbourne, S.M. Nurse, D. Corbett, Temperature changes associ-Ž .ated with forebrain ischemia in the gerbil, Brain Res. 602 1993

264–267.w x15 F. Colbourne, G. Sutherland, D. Corbett, Postischemic hypothermia:

a critical appraisal with implications for clinical treatment, Mol.Ž .Neurobiol. 14 1997 55–85.

w x16 D. Corbett, P. Crooks, Ischemic preconditioning: a long term sur-vival study using behavioural and histological endpoints, Brain Res.

Ž .760 1997 129–136.w x17 D. Corbett, S.J. Evans, S.M. Nurse, Impaired acquisition of the

Morris water maze following global ischemic damage in the gerbil,Ž .NeuroReport 3 1992 204–206.

w x18 J.C. de la Torre, T. Fortin, G.A.S. Park, K.S. Butler, P. Kozlowski,B.A. Pappas, H. de Socarraz, M.T. Richard, Chronic cerebrovascularinsufficiency induces dementia-like deficits in aged rats, Brain Res.

Ž .582 1992 186–195.w x19 W.D. Dietrich, R. Busto, O. Alonso, M.Y.-T. Globus, M.D. Gins-

berg, Intraischemic but not postischemic brain hypothermia protectschronically following global forebrain ischemia in rats, J. Cereb.

Ž .Blood Flow Metab. 13 1993 541–549.w x20 F.A. Edwards, LTP—a structural model to explain the inconsisten-

Ž . Ž .cies, Trends Neurosci. 18 6 1995 250–255.w x21 E.J. Green, W.D. Dietrich, F. van Dijk, R. Busto, C.G. Markgraf,

P.M. McCabe, M.D. Ginsberg, N. Schneiderman, Protective effectsof brain hypothermia on behavior and histopathology following

Ž .global cerebral ischemia in rats, Brain Res. 580 1992 197–204.w x22 N. Hori, D.O. Carpenter, Functional and morphological changes

Ž .induced by transient in vivo ischemia, Exp. Neurol. 129 1994279–289.

w x23 H. Ishimaru, A. Takahashi, Y. Ikarashi, Y. Maruyama, Pentobarbitalprotects against CA1 pyramidal cell death but not dysfunction ofhippocampal cholinergic neurons following transient ischemia, Brain

Ž .Res. 673 1995 112–118.w x24 R.M.A. Jaspers, F. Block, C. Heim, K.-H. Sontag, Spatial learning is

Ž .affected by transient occlusion of common carotid arteries 2VO :comparison of behavioural and histopathological changes after ‘2VO’

Ž .and ‘four-vessel-occlusion’ in rats, Neurosci. Lett. 117 1990 149–153.

w x25 M.S. Jensen, J.D.C. Lambert, F.F. Johansen, Electrophysiologicalrecordings from rat hippocampal slices following in vivo brain

Ž .ischemia, Brain Res. 554 1991 166–175.w x26 H. Kato, Y. Liu, T. Araki, K. Kogure, Temporal profile of the

effects of pretreatment with brief cerebral ischemia on the neuronaldamage following secondary ischemic insult in the gerbil: cumula-

Ž .tive damage and protective effects, Brain Res. 553 1991 238–242.w x27 K. Kirino, Y. Tsujita, A. Tamura, Induced tolerance to ischemia in

Ž .gerbil hippocampal neurons, J. Cereb. Blood Flow Metab. 11 1991299–307.

w x28 T. Kirino, Delayed neuronal death in the gerbil hippocampus follow-Ž .ing ischemia, Brain Res. 237 1982 57–69.

w x29 T. Kirino, H.P.C. Robinson, A. Miwa, A. Tamura, N. Kawai,Disturbance of membrane function preceding delayed neuronal death

Ž .in gerbil hippocampus, J. Cereb. Blood Flow Metab. 12 1992408–417.

w x30 K. Kitagawa, M. Matsumoto, M. Niinobe, K. Mikoshiba, R. Hata,H. Ueda, N. Handa, R. Fukunaga, Y. Isaka, K. Kimura, T. Kamada,Microtubule-associated protein 2 as a sensitive marker for cerebralischemic damage-immunohistochemical investigation of dendritic

Ž .damage, Neuroscience 31 1989 401–411.

( )P. Dooley, D. CorbettrBrain Research 794 1998 119–126126

w x31 K. Kitagawa, M. Matsumoto, M. Tagaya, R. Hata, H. Ueda, M.Niinobe, N. Handa, R. Fukunaga, K. Kimura, K. Mikoshiba, T.Kamada, Ischemic tolerance phenomenon found in the brain, Brain

Ž .Res. 528 1990 21–24.w x32 T. Kudo, K. Tada, M. Takeda, T. Nishimura, Learning impairment

and microtubule-associated protein 2 decrease in gerbils underŽ .chronic cerebral hypoperfusion, Stroke 21 1990 1205–1209.

w x33 H. Li, A. Buchan, Delayed treatment with NBQX or SNX-111following transient forebrain ischemia, Can. J. Neurol. Sci. 22Ž .1995 46.

w x34 H. Li, A.M. Buchan, Treatment with an AMPA antagonist 12 hoursfollowing severe normothermic forebrain ischemia prevents CA1

Ž .neuronal injury, J. Cereb. Blood Flow Metab. 13 1993 933–939.w x35 Y. Liu, H. Kato, N. Nakata, K. Kogure, Protection of rat hippocam-

pus against ischemic neuronal damage by pretreatment with sub-Ž .lethal ischemia, Brain Res. 586 1992 121–124.

w x36 B.G. Lyeth, L.W. Jenkins, R.J. Hamm, C.E. Dixon, L.L. Phillips,G.L. Clifton, H.F. Young, R.L. Hayes, Prolonged memory impair-ment in the absence of hippocampal cell death following traumatic

Ž .brain injury in the rat, Brain Res. 526 1990 249–258.w x37 D.F. Matesic, R.C.S. Lin, Microtubule-associated protein 2 as an

early indicator of ischemia-induced neurodegeneration in the gerbilŽ .forebrain, J. Neurochem. 63 1994 1012–1020.

w x38 A. Matus, Microtubule-associated proteins: their potential role inŽ .determining neuronal morphology, Annu. Rev. Neurosci. 11 1988

29–44.w x39 M. Muller, B.H. Gahwiler, L. Rietschin, S.M. Thompson, Reversible

loss of dendritic spines and altered excitability after chronic epilepsyin hippocampal slice culture, Proc. Natl. Acad. Sci. U.S.A. 90Ž .1993 257–261.

w x40 S. Nurse, D. Corbett, Direct measurement of brain temperatureduring and after intraischemic hypothermia: correlation with behav-

Ž .ioral, physiological and histological endpoints, J. Neurosci. 14 19947726–7734.

w x41 S. Nurse, D. Corbett, Neuroprotection following several days ofmild, drug-induced hypothermia, J. Cereb. Blood Flow Metab. 16Ž .1996 474–480.

w x42 S. Ohta, S. Furuta, I. Matsubara, K. Kohno, Y. Kumon, S. Sakaki,Calcium movement in ischemia-tolerant CA1 neurons after transientforebrain ischemia in gerbils, J. Cereb. Blood Flow Metab. 16Ž .1995 915–922.

w x43 G.W. Payne, R.S. Neuman, Effects of hypomagnesia on histamineH receptor-mediated facilitation of NMDA responses, Brit. J. Phar-1

Ž .macol. 121 1997 199–204.w x44 W.A. Pulsinelli, J.B. Brierly, F. Plum, Temporal profile of neuronal

damage in a model of transient forebrain ischemia, Ann. Neurol. 11Ž .1982 491–499.

w x45 A.F. Schnider, E.C. Olson, N.C. Spitzer, M. Montal, Mitochondrialdysfunction is a primary event in glutamate neurotoxicity, J. Neu-

Ž .rosci. 16 1996 6125–6133.w x46 M.J. Sheardown, E.O. Nielson, A.J. Hansen, P. Jacobsen, T. Hon-

Ž .ore, 2,3-Dihydroxy-6-nitro-7-sulfamoyl-benzo f quinoxaline: a neu-Ž .roprotectant for cerebral ischemia, Science 247 1990 571–574.

w x47 M.J. Sheardown, P.D. Suzdak, L. Nordholm, AMPA, but not NMDA,receptor antagonism is neuroprotective in gerbil global ischaemia,

Ž .even when delayed 24 h, Eur. J. Pharmacol. 236 1993 347–353.w x48 R.P. Simon, M. Niiro, R. Gwinn, Prior ischemic stress protects

Ž .against experimental stroke, Neurosci. Lett. 163 1993 135–137.w x49 H. Tsubokawa, K. Oguro, H.P.C. Robinson, T. Masuzawa, T.

Kirino, N. Kawai, Abnormal Ca2q homeostasis before cell deathrevealed by whole cell recording of ischemic CA1 hippocampal

Ž .neurons, Neuroscience 49 1992 807–817.w x50 L. Urban, K.H. Neill, B.J. Crain, J.V. Nadler, G.G. Somjen, Postis-

chemic synaptic physiology in area CA1 of the gerbil hippocampusŽ .studied in vitro, J. Neurosci. 9 1989 3966–3975.

w x51 J. Valtysson, L. Hillered, P. Andine, H. Hagberg, L. Persson,Neuropathological end points in experimental stroke pharmacother-apy: the importance of both early and late evaluation, Acta Neu-

Ž .rochir. 129 1994 58–63.w x52 B.T. Volpe, H.P. Davis, A. Towle, W.P. Dunlap, Loss of hippocam-

pal CA1 pyramidal neurons correlates with memory impairment inrats with ischemic or neurotoxin lesions, Behav. Neurosci. 106Ž .1992 457–464.

w x53 D. Wang, D. Corbett, Cerebral ischemia, locomotor activity andŽ .spatial mapping, Brain Res. 533 1990 78–82.

w x q 2q54 R.J. White, I.J. Reynolds, Mitochondria and Na rCa exchangebuffer glutamate-induced loads in cultured cortical neurons, J. Neu-

Ž .rosci. 15 1995 1318–1328.w x55 K. Yoshimi, M. Takeda, T. Nishimura, T. Kudo, Y. Nakamura, K.

Tada, N. Iwata, An immunohistochemical study of MAP2 andclathrin in gerbil hippocampus after cerebral ischemia, Brain Res.

Ž .560 1991 149–158.w x56 S. Zola-Morgan, L.R. Squire, D.G. Amaral, Human amnesia and the

medial temporal region: enduring memory impairment following abilateral lesion limited to field CA1 of the hippocampus, J. Neurosci.Ž .6 1986 2950–2967.

w x57 S. Zola-Morgan, L.R. Squire, D.G. Amaral, Memory impairment inmonkeys following lesions limited to the hippocampus, Behav.

Ž .Neurosci. 100 1986 155–160.