Journal of Neuroscience Methods 94 (2000) 187–192

Spectrophotometric measurement of experimental brain injury

Edward Preston *, Jacqueline WebsterInstitute for Biological Sciences, National Research Council of Canada, Bldg M54, Montreal Road, Ottawa, Ont., Canada K1A OR6

Received 22 March 1999; received in revised form 23 August 1999; accepted 2 September 1999

Abstract

Freshly sampled brain tissue exposed to 2,3,5-triphenyltetrazolium chloride (TTC) acquires a red color because mitochondrialenzymes reduce the colorless TTC to a red, water-insoluble formazan deposit. Pan-necrotic areas remain uncolored, which enablesquantitation of experimental brain injury by optical scanning and image analysis of serial slices to determine the relative volumeof red versus infarcted, non-stained, tissue. The accuracy of this method can be challenged, however, when infarction isaccompanied by areas of partial, scattered injury where differences in coloration are difficult to see or quantify. We tested thefeasibility of measuring scattered injury using a principle which underlies standard assays for in vitro cell survival, namelyextracting deposited formazan with a solvent and measuring its level by spectrophotometry. Anesthetized, adult Sprague–Dawleyrats were subjected to 12 min of cerebral ischemia to produce selective, delayed neuronal death in hippocampus, striatum andcortex. Some rats also received 6 h of whole-body hypothermia treatment (31.5–32.5°C) immediately after ischemia. Ischemia ratsand non-operated controls were sacrificed 1 week later. Hippocampus and portions of cerebrum were incubated 90 min in a 2%TTC solution and then soaked in a measured volume of 50:50 ethanol and dimethylsulfoxide to extract the red formazan product.Spectrophotometric measurements of the extract showed a diminished formazan coloration (absorbance/g brain) in all samplesfrom the untreated ischemia group compared to non-operated controls. This apparent brain injury was attenuated in the groupof ischemia rats that received hypothermia treatment. We conclude that solvent extraction and spectrophotometric quantitationof formazan has potential utility as an objective way to index experimental brain injury even if this is diffuse in nature and notamenable to measurement by conventional image analysis techniques. © 2000 Elsevier Science B.V. All rights reserved.

Keywords: Tetrazolium; Brain ischemia; Hypothermia; Formazan; Neuronal injury; Cerebral infarction; Stroke

www.elsevier.com/locate/jneumeth

1. Introduction

The accurate quantitation of brain damage producedby experimental trauma, ischemia or other cerebralinsult in laboratory animals is highly important to theinvestigation of injury mechanisms and therapies. Thehistological approach to this typically involves hema-toxylin–eosin or similar staining of paraffin or frozensections taken serially through the injury region andeither counting the relative number of dead or dyingneurones (reviewed by Corbett and Nurse, 1998), orestimating infarct volume through optical scanning andimage analysis (Osborne et al., 1987; Swanson et al.,1990). Infarct volume is also frequently measured byincubating serial coronal slices of fresh brain e.g. 2 mm

thick, in 2,3,5-triphenyltetrazolium chloride (TTC)which, normally colorless, is reduced by succinate dehy-drogenase in mitochondria of surviving tissue to a redformazan product. Some combination of photographyor scanning and image analysis is used to measure thearea of normal (red) and infarcted (uncolored) tissue ineach slice face, and thereby a figure for infarct volumeor percentage tissue loss (Bederson et al., 1986; Isayamaet al., 1991; Goldlust et al., 1996; Yang et al., 1998). Asignificant challenge lies in delineating the margins be-tween damaged and normal tissue. Done manually, thisrequires an observer to visually outline the infarctedregions on an image of each section, which may bedifficult to accomplish accurately if there are areaswhere damage is patchy and not clearly indicated by avisible difference in TTC staining (e.g. Sirimanne et al.,1994). Recognition of this problem has led to design ofautomated methods whereby the injury is defined bycomputer-based acquisition and analysis of pixel distri-bution (e.g. Goldlust et al., 1996).

* Corresponding author. Tel.: +1-613-993-9329; fax: +1-613-941-4475.

E-mail address: ed.preston@nrc.ca (E. Preston)

0165-0270/00/$ - see front matter © 2000 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 5 -0270 (99 )00146 -6

E. Preston, J. Webster / Journal of Neuroscience Methods 94 (2000) 187–192188

Tetrazolium salts have also been widely usedfor assays of cell survival and growth in in vitro condi-tions (Denizot and Lang, 1986; Hussain et al., 1993)including neuronal tissue culturing (Varming et al.,1996). To index cell survival from the degree of tetra-zolium reduction, a suitable solvent is used to ex-tract the colored formazan product, which is thenmeasured by spectrophotometry. It appeared to usthat potentially this could also be a relativelysimple, objective approach to assess tissue loss incurredin vivo after experimental brain injury even if this wasprimarily diffuse in nature. We tested this idea in a ratmodel of cerebral ischemia, modified after Smith et al.(1984), which produces selective, delayed neu-ronal death in regions of hippocampus, striatum andcortex.

2. Methods

2.1. Cerebral ischemia produced by two-6essel (carotid)occlusion plus hypotension (2VO model)

Male Sprague–Dawley rats (346–400 g at time ofsurgery) were purchased from Charles River, Montreal,and housed in pairs in standard plastic rodent boxeswith unrestricted access to Purina chow and water.Experiments were approved by a local committee forthe Canadian Council on Animal Care. For effecting2VO ischemia, rats were anesthetized with sodium pen-tobarbital (65 mg/kg i.p.), intubated and mechanicallyventilated with a 30:70% mix of O2 and N2 at a rateadjusted to produce normal blood gases and pH.Tympanic and colonic temperatures were main-tained between 37.5 and 38.0°C throughout surgicalprocedures and ischemia. Cerebral ischemia wasproduced by withdrawal of blood from the cannulatedtail artery into a heparinized syringe and main-tenance of arterial pressure between 42 and 47 mmHgfor 12 min. During this time, both common carotidarteries were temporarily clamped. Followingblood volume restoration and unclamping of thecarotids the wounds were sutured; the rat wasremoved from ventilation and maintained normother-mic until it recovered from anesthesia. For hypothermiatreatment after the 12-min ischemia, the rat was cooledby placing it supine on towelling overlying a bag ofcrushed ice. When colonic temperature had reached32°C (within 25 min) towelling thickness was adjustedas necessary to maintain temperature at 31.5–32.5°Cfor 6 h. Between 2.5 and 3 h into this hypothermiaperiod, rats received supplementary pentobarbital (20mg/kg i.p.) to maintain somnolence and inhibit shiver-ing. After the 6-h treatment the rat was warmed underan infrared lamp to normothermia, and returned tohousing.

2.2. TTC incubation and measurement of tissueformazan

Rats to be assessed for brain TTC staining wereanesthetized with 3% fluothane and then decapitated.The head was chilled in an icebath for 30 s; the brainwas removed and placed in a zinc matrix (HarvardApparatus Canada, St Laurent, Quebec). Using therostral margin of the pons as a landmark, razor bladeswere positioned at locations 6, 8 and 10 mm forwardand pressed downward to produce two 2-mm thickcoronal sections consisting mainly of striatum andfronto-parietal cortex. These samples, designated as‘forebrain’ were placed in a tared liquid scintillationvial and weighed. The remaining brain (pons-medullaand cerebellum discarded) was placed on a chilledsurface and dissected into two more samples: bothhippocampi (pooled), and the remainder of the cere-brum (mainly parieto-occipital cortex and midbrain). Abathing solution containing (in mM): 140 NaCl, 5 KCl,1 CaCl2, 10 Hepes, and 3 glucose (Mealing et al., 1997),was used to freshly dissolve TTC (2% solution) and 5ml was added to each tissue vial. These were laid ontheir side in a covered waterbath shaker set at 37°C andallowed to incubate for 90 min. The vials were periodi-cally checked to ensure that portions of tissue did notadhere to the vial walls or to one another. Afterwardsthe TTC was removed and the tissues rinsed twice byaddition and withdrawal of a few ml saline. The vialwas placed on a pan balance and a 50:50 mixture ofethanol/dimethylsulfoxide (Sladowski et al., 1993) wasthen added to solubilize the formazan (5 g forhippocampus and the forebrain slices, 10 g for theremainder of the cerebrum). The vials were tightlycapped and placed in a dark cupboard for a 24-hperiod, which preliminary tests showed sufficient todissolve and redistribute the tissue formazan through-out the contents of the vial. For analysis, each vial wasbriefly shaken, and four 100-ml aliquots of the redsolvent extract were placed in cuvettes and each dilutedwith 1900 ml of fresh ethanol/DMSO solvent. Averageabsorbance of the cuvettes was read at 485 l in aspectrophotometer, and absorbance per g tissue wascalculated for each sample. Each of six analysis sessionsinvolved the tissues from three rats: two ischemia rats(one hypothermia-treated and one untreated) and onenon-operated control (18 rats total). Percentage loss inbrain TTC staining (and apparent tissue injury) in eachischemia-treated rat, compared to the unoperated con-trol, was calculated from the following equation:

% Loss=100× (1−absorbance per g brainischemic/absorbance per g braincontrol)

For hippocampus, apparent injury was also calculatedon the basis of reduction in absolute absorbance, with-out normalization for tissue weight differences:

E. Preston, J. Webster / Journal of Neuroscience Methods 94 (2000) 187–192 189

% Loss=

100× (1−absorbanceischemic/absorbancecontrol)

Prior to the above experiments, brains freshly re-moved from three normal (non-ischemic) rats were usedto approximate the length of time tissues should beincubated in TTC to achieve adequate coloration. Us-ing the matrix, four 2-mm coronal slices were freshlyobtained from each brain. These were sagittally bi-sected on a cold plate and individual halves were placedin labeled, tared liquid scintillation vials, weighed andincubated in 2% TTC. Incubations were organized sothat each pair of symmetrical slices would be assignedtwo incubation times, e.g. 25 vs. 50, 50 vs. 75, 75 vs.100, or 100 vs. 125 min prior to solvent extraction todetermine the relative increments in formazanproduction.

3. Results

The experiments with symmetrical coronal slicesshowed that increasing incubation time from 25 to 50min caused a large increase in the amount of TTCreduction and coloration with the formazan product(Fig. 1). Increasing incubation time from 50 to 75 minalso improved staining, but there was little incrementproduced by increasing incubation time from 75 to 100min or 100 to 125 min. Tissues incubated for 75 min orlonger exhibited red formazan coloration extendingthroughout their thickness. We arbitrarily chose an

Table 1Comparison of TTC staining in ischemia-injured and normal ratbrain as measured by solvent extraction and spectrophotometrya

Treatment Absorbance per g braingroup

Hippocampus RemainderForebraincerebrum

100.293.8A. Control 124.094.4 51.291.6B. Ischemia+ 46.791.5d86.095.0e 113.293.6

hypothermia100.599.1b58.795.7c 36.593.7cC. Ischemia

a Data are from 18 rats (six per treatment group) with each analysisinvolving three rats, one from each group. Within groups, values forremainder cerebrum are markedly lower than hippocampus or fore-brain because twice the volume of solvent was used for formazanextraction (see Section 2). Values are mean9S.E.M. for absorbancereadings (×100) of regional formazan color product generated afterincubation in 2% TTC for 90 min. Ischemia rats underwent 12-minglobal ischemia by carotid occlusion plus hypotension, followed bynormothermic recovery (no treatment, row C) or 6-h hypothermiatreatment (row B). Rats were sacrificed for measurements 1 weeklater along with non-operated controls (A). All significant differences(PB0.05, ANOVA plus Tukey–Kramer test) were as follows (be-tween treatment groups, within regions):

b PB0.01 comparing both ischemia groups against control.c PB0.001 comparing both ischemia groups against control.d PB0.05 comparing ischemia+hypothermia (B) to ischemia (C).e PB0.01 comparing ischemia+hypothermia (B) to ischemia (C).

incubation time of 90 min as appropriate for furtherexperiments.

In preliminary ischemia experiments hippocampusremoved from six rats 1 week after 10 min of 2VOischemia and incubated 90 min in TTC exhibited adecrease in formazan coloration (total absorbance) av-eraging 21.895.8% (9S.E.M.) compared to non-oper-ated rats. The striatum and parietal cortex (bothsampled without the matrix) exhibited mean decreasesin absorbance per g averaging 17.495.6 and 11.594.5% respectively. For the following experiments theischemia duration was extended to 12 min, with theexpectation this would produce a more substantiveinjury, against which the protective effect of hypother-mia could be measured. Matrix-guided sampling ofcortex and striatum was employed to minimize anatom-ical sampling variation.

The group of six rats which underwent 12 min ofuntreated 2VO ischemia showed TTC staining whichwas significantly lower in all three brain regions com-pared to that in control tissues from non-operated rats,as indicated by the absorbance figures in Table 1 (rowC versus A). This was not the case for brain sampledfrom ischemia rats that underwent hypothermia treat-ment, for which decreases in staining were less markedand not statistically significant (row B vs. A). Afterhypothermia-treated ischemia, regional TTC stainingand absorbance values were higher than for untreated

Fig. 1. Upward increments in formazan colour product with prolon-gation of incubation time in 2,3,5-triphenyltetrazolium chloride(TTC). Coronal 2-mm slices of rat cerebrum were bisected andsymmetrical halves incubated in 2% TTC for two different incubationtimes, e.g. 25 vs. 50, 50 vs. 75 min. The height of each vertical barindicates the mean (9S.E.M.) percent increase in TTC stainingproduced by extending the incubation time in three slice pairs.Prolongation of incubation time beyond 75 min had a relativelyminor effect. Staining was assessed by a standardized solvent extrac-tion for 24 h, followed by dilution of the colored formazan productand its measurement by spectrophotometry.

E. Preston, J. Webster / Journal of Neuroscience Methods 94 (2000) 187–192190

ischemia, i.e. specifically for hippocampus and the re-mainder of the cerebrum (row B versus C).

Heights of the vertical bars in Fig. 2 represent thepercentage of apparent tissue injury calculated from thedecreases in formazan coloration in the foregoing ex-periments. After untreated ischemia the largest ‘injury’was seen in hippocampus. The hypothermia treatmentsignificantly reduced injury in both hippocampus andthe remainder of cerebrum. The percentage of injury inhippocampus calculated on the basis of total formazanproduction (without normalization for sample weight)was 41.098.8 (S.E.M.) for untreated ischemia versus15.696.7% for ischemia plus hypothermia (PB0.05,Student’s t-test), values comparable to those graphed inFig. 2.

The average weights (mg) of dissected tissues ana-lyzed in the foregoing experiments were as follows forthe control, ischemia plus hypothermia, and untreatedischemia groups respectively: hippocampus, 11594(S.E.M.), 11396, 11295; forebrain, 28698, 287911,272911; remainder, 943939, 924945, 911944. Ede-matous swelling, which would lead to increased sampleweight, was not evident in ischemia groups.

Blood parameters monitored during ischemia did notindicate any significant difference between groups (P\0.05, Students t-test) to account for differences in is-chemic injury. Values (mean9S.E.M.) were as follows(ischemia versus ischemia+hypothermia): glucose,

9.490.6 vs. 9.190.9 mM; pH, 7.4490.01 vs. 7.4390.01; PCO2, 40.891.5 vs. 40.891.7 (mmHg); PO2,12894 vs. 13895.

4. Discussion

We have previously confirmed by histological meth-ods that the experimental ischemia procedure, as wecarry it out in the rat (12-min 2VO), causes CA1neuronal death in hippocampus (MacManus et al.,1995). The key finding in the present experiments wasthat 12-min 2VO also caused a decrease in TTC stain-ing that was measurable by solvent extraction andspectrophotometry. This decrease was pronounced inhippocampus as might be anticipated from histologicalstudies which have already established that hippocam-pus is the most sensitive of tissues to cerebral ischemia(Pulsinelli and Brierley, 1979; Smith et al., 1984). Histo-logical assessments of ischemic hippocampal injury anddrug protection are often based on counting neuronallosses in standard coronal sections taken from therostral hippocampus. Assessments through the middleor posterior regions are often omitted, perhaps becauseof the labour involved, yet ideally should be performedgiven that these regions are less sensitive to injury(Smith et al., 1984; see also review by Corbett andNurse, 1998) and may reveal experiment-relatedchanges not evident in the rostral hippocampus. Thespectrophotometric measurements presented here in-dexed whole hippocampus in groups of rats with aneffort that seems to us a small fraction of that whichwould have been required using histological techniques,which ideally would involve assessment of all cells inserial sections through the entire hippocampus. In factour spectrophotometric evaluation of the scattered in-jury produced by global ischemia seems an atypicalapplication of TTC staining methodology which hastraditionally focussed on the measurement of infarctvolume in focal injury models (e.g. Liszczak et al., 1984;Bederson et al., 1986; Isayama et al., 1991; Goldlust etal., 1996). That diffuse injury can be measured suggestsa potential advantage in applying the formazan extrac-tion principle in focal injury models that prove difficultto assess by conventional scanning methods. Such anapproach should provide an index not only of pan-ne-crosis or infarction, but should incorporate into thisany scattered or penumbral damage. The potential util-ity of the spectrophotometric assay was also indicatedby the results involving post-ischemic hypothermia,which has already been established to afford a pro-longed protection of selectively vulnerable tissues fromdelayed ischemic damage (Colbourne and Corbett,1994). Accordingly the data showed that tissues ofischemic rats that had undergone hypothermia treat-ment exhibited significantly higher TTC staining and

Fig. 2. Percentage decrease in TTC staining, indicating apparentdegree of brain injury, measured by solvent extraction and spec-trophotometry 1 week after brain ischemia (2VO). Rats underwent12-min cerebral ischemia which was left untreated or followed by 6-hwhole-body hypothermia (31.5–32.5°C). Bar height is mean9S.E.M.for decrease in TTC stain (absorbance per g tissue) compared tovalues for non-injured control tissue measured in the same analysis,i.e. whole hippocampus, coronal forebrain slices incorporating stria-tum and cortex, and remainder of cerebrum (cerebellum and pon-medulla discarded). Within brain region: *PB0.05, **PB0.01comparing bar height for 2VO+hypothermia group versus untreated2VO (Student’s 2-tailed t-tests). Within treatment groups, betweenregions: PB0.05 only for forebrain versus hippocampus, untreated2VO group (ANOVA plus Tukey–Kramer test).

E. Preston, J. Webster / Journal of Neuroscience Methods 94 (2000) 187–192 191

less apparent damage than did untreated rats. Usingspectrophotometry we have confirmed in separate ex-periments (unpublished) that the statistically significantprotective effect of hypothermia is still present 4 weeksafter 2VO ischemia.

Certain factors bear particular mention as they mayaffect the magnitude of injury data and its interpreta-tion, based on formazan absorbance measurements. Itis proposed that invading leukocytes and macrophagesassociated with the cleanup and healing process cansignificantly augment TTC staining (Liszczak et al.,1984). It is also known that ischemia suppresses proteinsynthesis, mitochondrial enzymes, and glucosemetabolism, effects which can last hours or days afterinsult (Jaspers et al., 1990; Sims, 1992; Krause andTiffany, 1993). Possibly mitochondrial enzyme suppres-sion in surviving cells, along with outright death ofcells, would explain the magnitude of the apparentinjury in hippocampus after untreated ischemia, i.e.�40%, Fig. 2. We raise this point because this figuremight seem rather high to be attributed solely to selec-tive hippocampal degeneration, which in our experienceis largely restricted to the CA1 and hilar nuclei after12-min 2VO. The potential for spectrophotometricmeasurements to incorporate scattered losses of all cells(neurones and glia) plus changes in the biochemistry ofsurviving cells as it affects TTC reduction, has impor-tant implications for the comparative evaluation of thismethod relative to conventional approaches. For exam-ple, the question arises whether percentage neuronallosses as assessed by histology can be expected to agreewith formazan absorbance measurements other than ina directional manner, i.e. a certain intensity of injury orefficiency of treatment may change histological andspectrophotometric assessments in the same directionbut not necessarily by the same amount. Likewise theencompassing nature of spectrophotometric measure-ments could be expected to elevate figures for % hemi-spheric infarct damage compared to an area scanningmethod insensitive to subtle decreases in formazan col-oration surrounding the infarct.

We normalized measured absorbances to weight ofthe tissue being analyzed, i.e. absorbance per g, tocompensate for variations in the amount of tissue dis-sected and analyzed. However this approach requiresthat tissue edema not be present because an increase inthe proportion of H2O per g ischemic brain relative tonon-ischemic controls would spuriously augment per-cent injury calculated from absorbance/g figures. Tissueedema in ischemia groups would elevate average tissueweights compared to controls; this was not seen how-ever and therefore edema was unlikely a source of errorin this study. Moreover, we would not expect edema 1week after 12-min 2VO on the basis of previous experi-ments showing that blood–brain barrier opening andedema was reliably effected 24 h after prolonged is-

chemia, e.g. 25 min of 2VO, but not 10 min 2VO(Preston et al., 1993). If it were necessary to estimatebrain injury at early times after global ischemia whenedema is present, preferably this should involve com-parison of total absorbances, without tissue weightcorrection, of formazan extracted from samples thatcan be consistently dissected, e.g. hippocampus orwhole cerebrum from ischemia versus non-ischemiacontrols.

In their measurements of unilateral (focal) ischemicinjury using image analysis, Swanson et al. (1990) ad-dressed the problem that edematous swelling of theinfarct can lead to overestimation of its size. The infarctwas therefore estimated indirectly by subtracting vol-ume of the non-lesioned tissue in the infarcted hemi-sphere from volume of the contralateral, intacthemisphere. However, if vasogenic edema were to ex-tend into healthy tissue surrounding an infarct, suchvolume calculations could underestimate infarct vol-ume. It should be possible to evaluate focal injuryspectrophotometrically even when substantial edema ispresent simply by comparing total absorbances of for-mazan extracted from the injured versus contralateralhemisphere, assuming precise division down the sagittalmidline. With either focal or global injury, use of totalabsorbance to calculate injury would also be advisableif there was a significant loss of tissue mass, e.g. severalweeks after severe ischemia.

We conclude that, in principle, tetrazolium salt stain-ing of fresh brain tissue followed by solvent extractionand spectrophotometric measurement of formazan canprovide an objective index of experimental brain injuryeven if this is diffuse or involves tissue swelling and isnot easily assessed with conventional methods. Wepropose that the spectrophotometric measurementswould encompass all cells, i.e. neurones, glia and infl-ammatory cells, reflecting on the absence of those de-stroyed, but also on changes in the mitochondrial state(as it affects tetrazolium reduction) of surviving anddying cells as influenced by time after injury, experi-mental therapies, etc. Further research into factorsaffecting the tetrazolium reducing activity of tissue inthe analysis procedure, e.g. choice of tetrazolium saltand incubation conditions, should help optimize poten-tial application of this measurement principle for re-search or drug screening purposes involving brain orother organs.

References

Bederson JB, Pitts LH, Germano SM, Nishimura MC, Davis RL,Bartkowski HM. Evaluation of 2,3,5-triphenyltetrazolium chlo-ride as a stain for detection and quantification of experimentalcerebral infarction in rats. Stroke 1986;17:1304–8.

Colbourne F, Corbett D. Delayed and prolonged post-ischemic hy-pothermia is neuroprotective in the gerbil. Brain Res1994;654:265–72.

E. Preston, J. Webster / Journal of Neuroscience Methods 94 (2000) 187–192192

Corbett D, Nurse S. The problem of assessing effective neuroprotec-tion in experimental cerebral ischemia. Prog Neurobiol1998;54:531–48.

Denizot F, Lang R. Rapid colorimetric assay for cell growth andsurvival. Modifications to the tetrazolium dye procedure givingimproved sensitivity and reliability. J Immunol Methods1986;89:271–7.

Goldlust EJ, Paczynski RP, He YY, Hsu CY, Goldberg MP. Auto-mated measurement of infarct size with scanned images oftriphenyltetrazolium chloride-stained rat brains. Stroke1996;27:1657–62.

Hussain RF, Nouri AM, Oliver RT. A new approach for measure-ment of cytotoxicity using colorimetric assay. J Immunol Methods1993;160:89–96.

Isayama K, Pitts LH, Nishimura MC. Evaluation of 2,3,5-triphenyltetrazolium chloride staining to delineate rat brain in-farcts. Stroke 1991;22:1394–8.

Jaspers RMA, Berkelbach van der Sprenkel JW, Tulleken CAF,Cools AR. Local as well as remote functional and metabolicchanges after focal ischemia in rats. Brain Res Bull 1990;24:23–32.

Krause GS, Tiffany BR. Suppression of protein synthesis in thereperfused brain. Stroke 1993;24:747–56.

Liszczak TM, Hedley-Whyte ET, Adams JF, Han DH, Kolluri VS,Vacanti FX, Heros RC, Zervas NT. Limitations of tetrazoliumsalts in delineating infarcted brain. Acta Neuropathol1984;65:150–7.

MacManus JP, Hill IE, Preston E, Rasquinha I, Walker T, BuchanAM. Differences in DNA fragmentation following transient cere-bral or decapitation ischemia in rats. J Cereb Blood Flow Metab1995;15:728–37.

Mealing GAR, Lanthorn TH, Small DL, Black MA, Laferriere NB,Morley P. Antagonism of N-methyl-D-aspartate-evoked currents

in rat cortical cultures by ARL 15896AR. J Pharmacol Exp Ther1997;281:376–83.

Osborne KA, Shigeno T, Balarsky AM, Ford I, McCulloch J, Teas-dale GM, Graham DI. Quantitative assessment of early braindamage in a rat model of focal cerebral ischaemia. J NeurolNeurosurg Psychiatry 1987;50:402–10.

Preston E, Sutherland G, Finsten A. Three openings of the blood–brain barrier produced by forebrain ischemia in the rat. NeurosciLett 1993;149:75–8.

Pulsinelli WA, Brierley JB. A new model of bilateral hemisphericischemia in the unanesthetized rat. Stroke 1979;10:267–72.

Sims NR. Energy metabolism and selective neuronal vulnerabilityfollowing global cerebral ischemia. Neurochem Res 1992;17:923–31.

Sirimanne ES, Guan J, Williams CE, Gluckman PD. Two models fordetermining the mechanisms of damage and repair after hypoxic–ischaemic injury in the developing rat brain. J Neurosci Methods1994;55:7–14.

Sladowski D, Steer SJ, Clothier RH, Balls M. An improved MTTassay. J Immunol Methods 1993;157:203–7.

Smith ML, Auer RN, Siesjo BK. The density and distribution ofischemic brain injury in the rat following 2–10 min forebrainischemia. Acta Neuropathol 1984;64:319–32.

Swanson RA, Morton MT, Tsao-Wu G, Savalos RA, Davidson C,Sharp FC. A semiautomated method for measuring brain infarctvolume. J Cereb Blood Flow Metab 1990;10:290–3.

Varming T, Drejer J, Frandsen A, Schousboe A. Characterization ofa chemical anoxia model in cerebellar granule neurons usingsodium azide: protection by nifedipine and MK-801. J NeurosciRes 1996;44:40–6.

Yang Y, Shuaib A, Li Q. Quantification of infarct size on focalcerebral ischemia model of rats using a simple and economicalmethod. J Neurosci Methods 1998;84:9–16.

.