Minor cortical injury has previously been shown to improve survival in animals subjected to ischemic insults. Although the mechanism by which an ischemia-tolerant state is achieved is not clear, transient neuronal depolarization is thought to play a central role in the development of the tolerance. One way of producing transient neuronal depolarization is by the induction of cortical spreading depression (CSD). The present study was conducted to evaluate the effect of preischemic transient depolarization, induced by CSD, on postischemic neuronal outcome in rats.
Unilateral CSD was induced by application of KCl to the frontal cortex (CSD hemisphere) in three groups of isoflurane-anesthetized rats (CSD groups; n = 8/group). Sham animals (n = 12) did not undergo CSD. In a fifth group (n = 8), ketamine was administered during KCl application to inhibit CSD. One, three, or seven days after CSD, animals were subjected to forebrain ischemia produced by bilateral carotid artery occlusion. Injury to the striatum, hippocampus, and cortex was evaluated in hematoxylin and eosin-stained brain sections 3 days after ischemia.
Preischemic CSD reduced postischemic injury in the ipsilateral cortex. The ratio of the number of injured neurons in the CSD hemisphere to that in the non-CSD hemisphere was significantly less in the groups subjected to CSD 1 day (0.51 +/- 0.33), 3 days (0.56 /- 0.22), and 7 days (0.40 +/- 0.17) before ischemia than in the sham operated group (1.11 +/- 0.47). In the ketamine group (CSD inhibition), there were no differences in the extent of injury in the two hemispheres (ratio = 0.84 +/- 0.47). Injury to the striatum and hippocampus was similar among the groups. Within each group, injury to these subcortical structures in the CSD hemisphere was not different from that in the non-CSD hemisphere.
The data suggest that preischemic depolarization induced by CSD results in an adaptive response that reduces the vulnerability of cortical neurons to subsequent ischemic injury (ischemic tolerance).
A number of investigators have demonstrated that relatively minor cortical injury (such as needle penetration of the cortex) can make neurons more tolerant to subsequent ischemic insults. [1,2] This tolerance-inducing effect of minor cortical injury is long-lasting and has been shown to provide robust cerebral protection. Although the mechanism by which ischemic tolerance is achieved is not clear, transient neuronal depolarization is common to the various insults that have been used to induce tolerance.  Cortical spreading depression (CSD) is one way of accomplishing this. This phenomenon is characterized by a propagating wave of transient neuronal depolarization.  Recent experimental studies have shown that spreading depression induced neuronal depolarization, when induced 1 day before cerebral ischemia, reduces subsequent ischemic injury. [5–7] This reduction in injury is comparable with that which can be attained by the use of putative neuroprotective agents. However, the duration of the resultant ischemic tolerance state has not been clearly defined. Accordingly, the present study was undertaken to determine the magnitude and duration of the preconditioning effect of CSD in a rat model of incomplete forebrain ischemia. Because NMDA receptor antagonists can inhibit the occurrence of CSD, the effect of the NMDA antagonist ketamine also was assessed to provide additional support for the association of CSD with the preconditioning effect.
The study was approved by the institutional Animal Care and Use Committee. All experimental procedures were performed in accordance with the guidelines established in The PHS Guide for the Care and Use of Laboratory Animals.
Part I-Induction of CSD
Initiation of KCl-induced CSD was monitored with Ag-AgCl electrodes in 10 animals. Six WKY rats were anesthetized with isoflurane (0.8%) and nitrous oxide (70%) in oxygen. After endotracheal intubation, the heads of the animals were secured in a stereotaxic frame. Two 4-mm diameter craniectomies were performed: 3 mm rostral, 3 mm lateral to the bregma; and 3 mm caudal, 3 mm lateral to bregma. Under direct visualization through a stereomicroscope, the underlying dura was incised carefully with a 30-gauge needle. Injury to the cortex was avoided. In the caudal craniectomy, a Ag-AgCl electrode with a tip diameter of 2–3 micro meter was inserted into the cortex to a depth of 200 micro meter. This electrode was referenced to a Ag-AgCl pellet inserted into the neck musculature. After a 15-min stabilization period, gelatin sponge (Upjohn Co., Kalamazoo, WI) soaked with 3 M KCl was applied to the cortex via the rostral craniectomy for 30 min. The ensuing CSD activity was recorded using a DC amplifier (World Precision Instruments, Sarasota, FL). In a separate group of four animals, isoflurane-nitrous oxide anesthesia was discontinued. Simultaneously, ketamine (50 mg/kg bolus and infusion at a rate of 20 mg [center dot] kg sup -1 [center dot] h sup -1) was administered. KCl was applied to the cortex, and the CSD activity was recorded as described previously.
Part II-Forebrain Ischemia Studies
Non-fasted WKY rats were anesthetized with isoflurane. After endotracheal intubation, mechanical ventilation of the lungs was initiated with an inspired gas mixture of 2% isoflurane in 30% O2, balance N2O. A temperature probe was inserted into the temporalis muscle, and pericranial temperature was servoregulated to 37 [degree sign] Celsius. The animal's head was secured in a stereotaxic frame. After infiltration with 0.25% bupivacaine, the scalp was incised and reflected. The designated CSD hemisphere was randomly chosen. A 4-mm diameter craniectomy was made 2.5 mm lateral and 2.5 mm rostral to the bregma. Under a stereo microscope, the dura was incised and reflected. The end-tidal concentration of isoflurane was then reduced to 0.8%. Cortical spreading depression was induced by the application of gelatin sponge, soaked with 3 M KCl, onto the cortex for a period of 30 min.  Thereafter, the gelatin sponge was removed; the scalp was sutured, and the animal was allowed to awaken. On recovery of consciousness, the trachea was extubated. After an observation period of 2 h, the animal was returned to the animal care facility. In a separate group of animals, the NMDA antagonist ketamine (50 mg/kg bolus followed by a continuous infusion at a rate of 15 mg [center dot] kg sup -1 [center dot] h sup -1) was administered before KCl application.
Sham animals underwent all of the above procedures except craniectomy and KCl application. In these animals, one hemisphere was randomly designated the “CSD hemisphere.”
Ischemia was induced 1, 3, or 7 days after CSD induction (n = 8/group). Sham animals (n = 12) were rendered ischemic 1 and 3 days after CSD. Animals in the ketamine group (n = 8) were subjected to ischemia 3 days after CSD.
Fasted rats were anesthetized with isoflurane, and their tracheas were intubated. Mechanical ventilation was then instituted as described previously. Ventilatory parameters were adjusted to maintain normocapnia (PaCO2, 35–40 mmHg). A thermistor (Mon-A-Therm, Mallinckrodt, St. Louis, MO) was inserted between the temporalis muscle, and the skull and pericranial temperature was servoregulated to 37 [degree sign] Celsius. Needle electroencephalographic (EEG) electrodes were inserted in a biparietal configuration, and the EEG was monitored continuously (Grass Instruments, Quincy, MA). The tail artery and external jugular vein were cannulated. Both carotid arteries were exposed via a mid-line pretracheal incision and were encircled loosely with ligatures. The inspired concentration of isoflurane was reduced to 0.8%, and the animals were left undisturbed for an equilibration period of 15 min. Arterial pressure, heart rate, PaO2, PaCO2, pH, hematocrit, and serum glucose were measured during this period.
The animals' heads were then placed in a humidified plexiglass chamber as described previously.  Five minutes after the administration of 50 U heparin, 2.5 mg of trimethaphan was administered intravenously. Simultaneously, blood was withdrawn through the external jugular vein catheter into a syringe that was placed in a heated water bath (the temperature of the withdrawn blood was maintained at approximately 37 [degree sign] Celsius). At a MAP of 35 mmHg, both carotid arteries were occluded with vascular clips for 15 min. Ischemia was confirmed by the observation of an isoelectric EEG. Thereafter, the clips were removed, and reperfusion of the brain was established. The withdrawn blood was reinfused. The vascular catheters were removed, and the wounds were closed. All wounds were infiltrated with 0.25% bupivacaine. Anesthetic administration was then discontinued. On resumption of spontaneous ventilation, the endotracheal tube was removed, and the animals were placed in a humidified and heated chamber into which oxygen was flushed continuously. After an observation period of 2 h, the animals were returned to the animal care facility.
Three days after forebrain ischemia, the animals were anesthetized with pentobarbital. They were killed by transcardiac perfusion with heparinized saline, followed by 200 ml of phosphate buffered paraformaldehyde. The animals were decapitated, and the brains were refrigerated in situ at 4 [degree sign] Celsius for 24–48 h. The brains were then removed carefully. After dehydration in graded concentrations of ethanol and butanol, the brains were embedded in paraffin. Eight- micron thick coronal sections were prepared and stained with hematoxylin and eosin. Injury to the dorsolateral striatum, rostral hippocampus (CA1 and CA3 sectors), and ventral hippocampus (CA1 and CA3 sectors) was evaluated in coronal planes 600 micro meter, 3,300 micro meter, and 6,000 micro meter posterior to the bregma, respectively, according to the atlas of the rat brain of Palkovits and Brownstein.  Injury to these structures was graded on a four-point scale as described previously: 0 = no injury; 1 = less than 10% of neurons injured; 2 = between 10% and 50% neurons injured; 3 = greater than 50% neurons injured. Injury to the parietal and temporal-occipital cortex was evaluated in coronal planes 3,300 micro meter and 6,000 micro meter posterior to the bregma. Within each hemisphere, injured cortical neurons were counted in eight successive high power fields (400 x) that spanned the entire depth of the cortex, from the entorhinal fissure to the interhemispheric fissure. Histologic evaluation was performed by two observers who had no knowledge of the experimental group assignment.
Data from the sham group animals (made ischemic 1 and 3 days after sham CSD) were summated. Physiologic data were analyzed by a one-way analysis of variance (ANOVA). Post hoc differences among groups were identified by Scheffe's test. Histologic scores were analyzed by the Kruskal-Wallis test. Cortical injury to the CSD hemisphere, presented as a percent of that in the non-CSD hemisphere, was also analyzed by the Kruskal-Wallis test. Differences among groups were identified by the Mann-Whitney U tests with an appropriate Bonferroni correction factor for multiple comparisons. A P value of less than 0.05 was considered to be statistically significant. All data are presented as mean +/- SD.
Part I-Induction of CSD
The application of 3 M KCl to the cortical surface reliably produced CSD activity in the animals anesthetized with isoflurane and nitrous oxide (Figure 1). In 30 min, approximately 8–10 CSD waves were detected (median, 8; range, 8–10). In the four ketamine anesthetized animals, CSD waves were not detected, confirming that the dose of ketamine administered was effective in preventing CSD.
Part II-Forebrain Ischemia Studies
The physiologic parameters in the animals that were subjected to ischemia after CSD are presented in Table 1. Physiologic parameters among the experimental groups were not different.
Cortex. The number of injured neurons in the CSD and non-CSD hemispheres in each animal in the five experimental groups is presented in Table 2. In Figure 2, injury to the CSD hemisphere is presented as a percent of the injury in the non-CSD hemisphere. Preischemic CSD significantly reduced cortical injury in the hemisphere ipsilateral to CSD. This protective effect of CSD was observed 24 h after CSD and was still evident 3 and 7 days after CSD. Ketamine, which inhibited CSD, abolished the protective effect of previous KCl application, i.e., there was no interhemispheric difference in the extent of injury in this group.
In the animals that were subjected to CSD, cortical injury at the site of KCl application was evident. The injury was limited to the craniectomy site and extended into the cortex to a depth of 1–1.5 mm.
Subcortical Structures. The histologic scores in the striatum and hippocampus are presented in Table 3. Within each group, there were no differences between the CSD hemisphere and the non-CSD hemisphere. There were also no differences among the groups in the extent of injury. In all groups, severe injury to the striatum and rostral CA1 sector of the hippocampus was observed. In most animals, more than 50% of the neurons were injured (grade 3) in these structures. Injury to the rostral CA3 sector was not as severe. The ventral CA1 sector of the hippocampus was injured the least; the median grade of injury was 1.
The present study was conducted to evaluate the influence of preischemic transient neuronal depolarization, induced by cortical spreading depression (CSD), on postischemic neuronal outcome. The results demonstrate that preischemic transient depolarization produced by spreading depression can reduce ischemic neuronal injury, but only in the cortex. In the cortex, this ischemic preconditioning effect was seen 24 h after CSD induction and was still evident 7 days later. Two observations suggest that ischemic preconditioning was a result of depolarization induced by CSD rather than other effects of cortical injury produced by KCl application. First, the reduction in injury was seen in a cortical region that was remote from the site of surgical craniectomy. In the striatum and hippocampus, preischemic CSD did not attenuate injury after forebrain ischemia. This observation is not inconsistent as CSD waves, when elicited in the cortex, do not depolarize subcortical structures.  Second, ketamine, an NMDA antagonist that inhibited CSD, abolished the ischemic preconditioning.
The observation that the protective effects of previous CSD-induced depolarization were evident as late as 7 days after CSD indicate that tolerance induced by CSD is long-lasting. The time course of this preconditioning effect in our study is different from that reported recently by Kawahara et al.  In a rat model of cardiac arrest ischemia, these investigators demonstrated that the protective effect of previous spreading depression was observed only 3 days after spreading depression but not 1 day and 7 days after spreading depression. The reason for these varying results may lie in the differing approach to the evaluation of neuronal injury. In the study of Kawahara et al., spreading depression was induced in the cortex and in the hippocampus, although neuronal injury was evaluated in only the hippocampus. By contrast, in the present study, neuronal injury was specifically evaluated within the cortex. It is entirely possible that the preconditioning effect of CSD varies from structure to structure and also may vary, depending on the nature of the preconditioning stress.
The recent work of Matsushima et al.  and of Kobayashi et al.  also is consistent with our observation CSD that induced preconditioning may be apparent within 24 h. These investigators observed a reduction in cortical injury in animals that were subjected to CSD 24 h before an episode of focal ischemia. Accordingly, the available data therefore clearly indicate that a preconditioning effect may be apparent as early as 24 h after CSD. However, the precise duration of the preconditioning effect has yet to be determined. In that connection, the present study has extended the findings from other investigations further by the demonstration that the tolerance-inducing effect of preischemic spreading depression lasts at least 7 days. The possibility exists that this effect might persist for a period greater than 7 days, and the total duration remains to be defined.
Ischemic preconditioning has also been shown to occur after short episodes of sublethal ischemia (ischemia of a brief duration that does not itself result in ischemic neuronal injury). In gerbils, sublethal ischemia produced by bilateral carotid artery occlusion for 2 min has been demonstrated to result in the induction of tolerance to subsequent episodes of ischemia. [13–15] In preconditioned gerbils subjected to ischemia from 1 to 7 days after the sublethal ischemia, ischemic injury to the hippocampus, striatum, and neocortex was reduced substantially.  This tolerance-inducing effect of previous sublethal ischemia lasted at least 7 days but dissipated after 2 weeks. MK-801, when administered during the sublethal ischemia, abolished the preconditioning effect.  Accordingly, tolerance induced by CSD and by sublethal ischemia is similar in at least two respects: the duration of tolerance persists for at least 7 days and the induction of tolerance can be abolished by NMDA antagonists. However, the extent of tolerance induced by these two techniques appears to be different. In the study of Kato et al.,  sublethal ischemia provided almost complete protection against subsequent ischemia in the neocortex. By contrast, in the present study, CSD-induced tolerance was not as complete. Within the CSD hemisphere, considerable neuronal necrosis in the cortex was evident, although it was substantially less than in the non-CSD hemisphere.
The mechanism by which CSD produced ischemic preconditioning is not clear at present. One possibility is that CSD causes the synthesis of substances that can reduce the vulnerability of neurons to ischemic injury. CSD has been shown to lead to the synthesis of mRNA for nerve growth factory (NGF) and brain-derived neurotrophic factor (BDNF).  In vitro, NGF has been found to protect hippocampal and cortical neurons from hypoglycemic injury [18–20] and striatal neurons from excitotoxic injury.  Similarly, BDNF has been shown to reduce hypoglycemic injury to dentate granule cells  and excitotoxic injury to cerebellar granule cells.  In addition, the widespread cortical synthesis of basic fibroblast growth factor (bFGF) mRNA that occurs after cortical ablation injury has been attributed to the occurrence of CSD.  In cultured rat hippocampal and human cortical neurons, bFGF reduces hypoglycemic injury.  Collectively, these data suggest that CSD can lead to the synthesis of several endogenous and potentially neuroprotective factors  that might contribute to the preconditioning effect observed in the present study.
It is also possible that CSD reduces the rate at which energy depletion occurs during ischemia. The data of Kawahara et al. indicate that CSD reduces CMR glucose as long as 3 days after CSD is elicited.  Ischemia-induced anoxic depolarization is thought to occur when the available ATP stores are depleted.  One can therefore speculate that the onset of anoxic depolarization might have been delayed in the CSD hemisphere because of CMR reduction. If so, then the reduction of injury in the CSD hemisphere may have been due to a reduced effective duration of ischemia (as measured by the duration of anoxic depolarization). However, this concept remains a matter of speculation and experimental clarification is needed.
In the present study, injury to the CSD hemisphere was compared with that in the non-CSD hemisphere. The major advantage of this approach was that in each animal, the non-CSD hemisphere served as the control. The effect of interanimal variability in the severity of the injury was reduced significantly. However, inherent to this experimental design is the assumption that the severity of ischemia is similar in both hemispheres in each animal. Gionet et al. have shown that there may be a considerable amount of interhemispheric variability in the severity of injury in the forebrain ischemia model. However, they demonstrated that this variability can be reduced by decreasing the absolute level of blood pressure that is maintained during ischemia.  A target level of 35 mmHg (in contrast to the more commonly used intraischemic pressure of 50 mmHg ) during ischemia was therefore chosen to minimize this variability. In addition, the injury to the subcortical structures was similar in the CSD and non-CSD hemisphere. These data provide support for the validity of a comparison of injury in the CSD hemisphere to that in the non-CSD hemisphere.
In summary, the results of this study show that previous CSD induces tolerance against subsequent ischemic injury. This CSD-mediated ischemic preconditioning is evident 24 h after CSD and persists for at least 7 days. Ketamine, an NMDA antagonist and proven inhibitor of CSD, abolished this preconditioning effect. The mechanisms by which CSD induced ischemic tolerance are not clear at present. Identification of these mechanisms may provide insight into potentially novel therapeutic approaches that might be used to reduce postischemic neuronal injury.