Andre G. Douen, *Katsunori Akiyama, Matthew J. Hogan, Fuhu Wang, Li Dong, Ava K. Chow, and Antoine Hakim

Transcription

1 Journal of Neurochemistry Lippincott Williams & Wilkins, Inc., Philadelphia 2000 International Society for Neurochemistry Preconditioning with Cortical Spreading Depression Decreases Intraischemic Cerebral Glutamate Levels and Down-Regulates Excitatory Amino Acid Transporters EAAT1 and EAAT2 from Rat Cerebral Cortex Plasma Membranes Andre G. Douen, *Katsunori Akiyama, Matthew J. Hogan, Fuhu Wang, Li Dong, Ava K. Chow, and Antoine Hakim Neuroscience Research Institute, University of Ottawa, Ottawa, Ontario, Canada; and *Tokai University, Isehara, Japan Abstract: We previously reported a 50% reduction in cortical infarct volume following transient focal cerebral ischemia in rats preconditioned 3 days earlier with cortical spreading depression (CSD). The mechanism of the protective effect of prior CSD remains unknown. Recent studies demonstrate reversal of excitatory amino acid transporters (EAATs) to be a principal cause for elevated extracellular glutamate levels during cerebral ischemia. The present study measured the effect of CSD preconditioning on (a) intraischemic glutamate levels and (b) regulation of glutamate transporters within the ischemic cortex of the rat. Three days following either CSD or sham preconditioning, rats were subjected to 200 min of focal cerebral ischemia, and extracellular glutamate concentration was measured by in vivo microdialysis. Cortical glutamate exposure decreased 70% from 1, ,469.2 M-min in sham-treated (n 8) to M-min in CSD-treated (n 13) rats ( p 0.05). The effect of CSD preconditioning on glutamate transporter levels in plasma membranes (PMs) prepared from rat cerebral cortex was assessed by western blot analysis. Down-regulation of the glial glutamate transporter isoforms EAAT2 and EAAT1 from the PM fraction was observed at 1, 3, and 7 days but not at 0 or 21 days after CSD. Semiquantitative lane analysis showed a maximal decrease of 90% for EAAT2 and 50% for EAAT1 at 3 days post-csd. The neuronal isoform EAAT3 was unaffected by CSD. This period of down-regulation coincides with the time frame reported for induced ischemic tolerance. These data are consistent with reversal of glutamate transporter function contributing to glutamate release during ischemia and suggest that down-regulation of these transporters may contribute to ischemic tolerance induced by CSD. Key Words: Preconditioning Cortical spreading depression Cerebral ischemia Glutamate Neuroprotection Glutamate transporters. J. Neurochem. 75, (2000). Glutamate uptake via sodium-dependent, high-affinity, excitatory amino acid transporters (EAATs) is the primary mechanism for clearance of extracellular glutamate in the cerebral cortex and is essential for terminating the postsynaptic action of glutamate (Rothstein, 1996; Rothstein et al., 1996). During cerebral ischemia there is a rise in extracellular cerebral glutamate content to neurotoxic levels (Choi, 1988; Siesjö, 1994). Although some extracellular glutamate arises from presynaptic vesicular release, this process is ATP-dependent and hence may not be the main source of extracellular glutamate under hypoxic conditions (Nicholls and Attwell, 1990; Szatkowski and Attwell, 1994). In comparison, hypoxia disrupts the ionic gradient across the plasma membrane (PM), leading to a rise in level of intracellular Na, which binds to and induces reversal of the EAAT, thereby facilitating glutamate efflux into the extracellular space (Nicholls and Attwell, 1990; Szatkowski and Attwell, 1994; Rutledge and Kimelberg, 1996). More recently, Li et al. (1999) showed that dihydrokainate, a specific inhibitor of the glial glutamate transporter isoform EAAT2, significantly impaired glutamate release during in vitro anoxic injury of the spinal cord, and Seki et al. (1999) reported that intracerebral infusion of dihydrokainate decreased ischemia-induced glutamate release by 50%. These studies provide strong evidence that EAAT reversal plays a major role in hypoxia-induced glutamate release. There are three different EAAT isoforms within the mammalian cerebrum, designated EAAT1, EAAT2, and EAAT3 (Rothstein, 1996; Rothstein et al., 1996). EAAT1 Resubmitted manuscript received March 3, 2000; accepted March 24, Address correspondence and reprint requests to Dr. M. J. Hogan at Neuroscience Research Institute, University of Ottawa, 451 Smyth Road, Ottawa, Ontario, K1H 8M5 Canada. mhogan@ uottawa.ca Abbreviations used: CBF, cerebral blood flow; CSD, cortical spreading depression; EAAT, excitatory amino acid transporter; ECL, enhanced chemiluminescence; MCAO, middle cerebral artery occlusion; PM, plasma membrane. 812

2 PRECONDITIONING ALTERS GLUTAMATE REGULATION 813 and EAAT2, also known as GLAST and GLT-1, respectively, are primarily astrocytic, whereas EAAT3 is found predominantly in neurons (Rothstein, 1996; Rothstein et al., 1996). Knockout studies have shown that EAAT2 and, to a lesser extent, EAAT1 are responsible for the bulk of glutamate clearance in the rat cerebral cortex (Rothstein et al., 1996). It is interesting that EAAT2 and EAAT1, but not EAAT3, appear to be regulatable proteins (Ginsberg et al., 1995; Rothstein, 1996; Raghavendra Rao et al., 1998). Down-regulation of both EAAT2 and EAAT1 has been observed in crude homogenized membranes following regional deafferentation within the rat cerebrum (Ginsberg et al., 1995). More recently, Raghavendra Rao et al. (1998) showed that controlled cortical impact-induced traumatic brain injury caused delayed, transient down-regulation of both EAAT2 and EAAT1 from a membrane fraction isolated from rat cerebral cortex. In addition, chronic downregulation of the EAAT2 isoform has been linked to neuronal cell death in some patients with amyotrophic lateral sclerosis (Rothstein et al., 1996). Taken together, these observations suggest that glial EAATs are regulatable proteins that may influence cell survival through modulation of extracellular cerebral glutamate levels. We and others have shown that preconditioning the brain with recurrent waves of cortical spreading depression (CSD) will reduce neuronal injury caused by subsequent cerebral ischemia (Kawahara et al., 1994; Matsushima et al., 1996a,b). Whereas Kawahara et al. (1994) reported less ischemic damage to hippocampal CA1 neurons in animals pretreated with CSD, we demonstrated that induction of CSD 3 days before middle cerebral artery occlusion (MCAO) in rats reduced cortical infarct volume by 50% compared with sham-treated controls (Matsushima et al., 1996a). However, the mechanism underlying this CSD-mediated neuroprotection is unclear. Given our present understanding of glutamate excitotoxicity, we considered whether preconditioning with CSD could attenuate glutamate release during ischemia. The studies were undertaken to measure intraischemic cerebral glutamate levels following CSD preconditioning and to determine the influence of CSD on the presence of EAATs in PM extracts from the rat cerebral cortex. MATERIALS AND METHODS Reagents Phenylmethanesulfonyl fluoride, Tris-HCl, sodium dodecyl sulfate, glycerol, and 2-mercaptoethanol were purchased from Sigma (St. Louis, MO, U.S.A.). TEMED (N,N,N,N -tetramethylenediamine), Tween 20, acrylamide, bisacrylamide, EDTA, glycine, nitrocellulose paper, and protein assay kit were obtained from Bio-Rad (Hercules, CA, U.S.A.). Enhanced chemiluminescence (ECL) supplies were obtained from Amersham (Little Chalfont, Bucks, U.K.). All other bench chemicals were obtained from Sigma and were of analytical grade. Monoclonal antibodies to the Na,K -ATPase 1 subunit (110 kda) and polyclonal antibodies to EAAT2/GLT-1 (75 kda) were obtained from Affinity Bioreagents (Golden, CO, U.S.A.). Polyclonal antibodies to EAAT1/GLAST (65 kda) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, U.S.A.). Polyclonal antibodies to EAAT1/GLAST and EAAT3 were also obtained as a gift from Dr. Jeffery D. Rothstein (Department of Neurology, Johns Hopkins University). Secondary antibodies were obtained from Transduction Laboratories (Lexington, KY, U.S.A.). Animals and CSD Male Sprague Dawley rats weighing 300 g were obtained from Charles River (Montreal, QC, Canada). All surgical manipulations, CSD, and decapitation were performed with the animal under halothane anesthesia (4% induction, 0.75% maintenance), and all procedures followed the guidelines of the Canadian Council of Animal Care. Animals were assigned to either a CSD or a sham treatment group. With the rat under anesthesia the tail artery was cannulated with a polyethylene catheter to monitor mean arterial blood pressure and obtain samples for arterial blood gas and blood glucose measurements during recurrent CSD. Body temperature was monitored with a rectal probe and controlled between 36.5 and 37.5 C. Rats were placed on a stereotactic frame (Stoelting Co., Wood Dale, IL, U.S.A.), and 2- and 0.5-mm-diameter burr holes were made over the left occipital cortex and left frontal cortex, respectively, as previously described (Matsushima et al., 1998). A platinum electrode was inserted 2 mm into the frontal cortex, fixed to the skull with dental cement, and used to record CSD. Recurrent CSDs were induced in the left hemisphere by placing a cotton pledget soaked with 0.5 M KCl onto the intact dura over the left occipital cortex; pledgets soaked in 0.5 M NaCl were used in the sham-treated group. Pledgets were replaced every 15 min for 2 h and then removed. The wound was rinsed with saline, the skin was sutured over the skull, the tail cannula was removed, and the rats were allowed to recover from anesthesia before returning to their cages. Animals were then subjected to MCAO and microdialysis 3 days later or killed for extraction of the PM fraction from the cerebral cortex and subsequent western blot analysis at specific intervals (0, 1, 3, 7, and 21 days) post-csd. MCAO Rats were anesthetized 3 days after CSD (n 13) or sham pretreatment (n 8), and the left femoral artery was cannulated with a polyethylene catheter to monitor physiological parameters. Animals were placed in a stereotactic frame. A microdialysis probe (diameter, 0.24 mm; active length, 2.0 mm; CMA-11; CMA/Microdialysis, AB, Carnegie Medecin, Stockholm, Sweden) was inserted stereotactically into the left parietal cortex (bregma, 2.0 mm; lateral, 5.0 mm; depth, 5.5 mm; angle, 30 ), and a second probe was placed in the caudate (bregma, 2.0 mm; lateral, 3.5 mm; depth, 7.0 mm). Platinum wire electrodes for subsequent measurement of cerebral blood flow (CBF) were fixed to the microdialysis probes. Sixty minutes later focal ischemia was induced by occluding the middle cerebral artery using a 3-0 nylon monofilament thread (Zea Longa et al., 1989; Matsushima et al., 1996a,b). The thread was introduced through the common carotid artery, and the tip of the thread was advanced 18.0 mm distal to the carotid bifurcation. The occlusion of the middle cerebral artery was confirmed by measurement of CBF using the hydrogen clearance method (Farrar, 1987). Studies were terminated by decapitation of the animal after 200 min of MCAO.

3 814 A. G. DOUEN ET AL. Microdialysis and HPLC measurement Microdialysis probes were continuously perfused with Krebs Ringer s-bicarbonate solution (ph 7.4) at a rate of 2 l/min, starting at 20 min before implantation and continuing to the end of the experiment, using a microinjection pump (CMA/100; CMA/Microdialysis AB). Dialysates were collected in 20- l fractions at 10-min intervals using a microfraction collector (CMA/140; CMA/Microdialysis AB, Carnegie Medicin). Samples were stored at 80 C for subsequent analysis. The glutamate concentration in the dialysate was measured by HPLC using baseline software (Waters Chromatography Division, Millipore Corp., Milford, MA, U.S.A.) as previously described (Osuga and Hakim, 1994). In brief, 10 l of dialysate was mixed with 10 l of50 M norvaline internal standard solution. Ten microliters of this solution was automatically reacted with 20 l ofo-phthaldialdehyde for 60 s just before injection. A glutamate calibration line based on the ratio of glutamate/norvaline peak areas was obtained by measurement of four amino acid standard solutions (Sigma). The glutamate concentration in the dialysate was calculated by reference to the standard dose response line. The mobile phase gradient was controlled from 37.5 to 55% methanol in 0.05 M sodium monophosphate buffer. Index of glutamate exposure The glutamate concentration in the dialysates obtained at timed intervals from each animal during MCAO was determined and plotted to yield a glutamate concentration curve over a 200-min interval. The integrated area under the curve was defined as the index of exposure to glutamate ( M-min) and represents a measure of the total amount of extracellular glutamate that the ischemic cerebral cortex is exposed to during the 200 min of focal ischemia (Osuga and Hakim, 1994). Membrane preparation Extraction of a PM-enriched fraction from cerebral cortex was achieved through modification of subcellular fractionation techniques previously established in adipocytes (Douen et al., 1988) and skeletal muscles (Douen et al., 1990). Rats were killed at specific intervals following KCl-induced CSD: day 0 (immediately following CSD) and days 1, 3, 7, and 21. Shamoperated animals were treated with 0.5 M NaCl instead of KCl (described above). Rats were decapitated, and the brains were rapidly removed and placed in ice-cold saline. All remaining steps were conducted at 4 C. Visible pial vessels were removed, and left and right cerebral cortices were carefully dissected from subcortical structures and processed separately. Cortices were homogenized at 500 rpm in 10 ml of 10 mm NaHCO 3 /0.1 mm phenylmethanesulfonyl fluoride buffer at ph 7.0 (buffer A) with 10 strokes of a Teflon pestle in a Wheaton glass homogenizer. The homogenizer was rinsed with buffer A, and the washings were combined with the homogenized sample to yield a final volume of 20 ml. After removing a 1-ml aliquot for subsequent analysis, the crude homogenate was spun at 30,000 g for 30 min to yield a pellet (P 1 ) and a supernatant. P 1 was resuspended in 10 ml of buffer A and layered onto 5.5 ml of ice-cold 35% sucrose solution (35 g of sucrose/100 ml of buffer A) in 12-ml Beckman ultracentrifuge tubes. Samples were centrifuged at 150,000 g av for2hinabeckman titanium SW 41 swing bucket rotor. The membrane fraction at the interface was carefully harvested, washed, resuspended in 10 ml of buffer A, and precipitated by centrifugation at 200,000 g max for 1 h. The latter membranes were resuspended in buffer A and stored at 80 C. Protein content was determined by a Coomassie Blue method using the Bio-Rad protein determination kit. Detection of PM Western blot analysis using monoclonal antibodies to the PM-specific Na,K -ATPase 1 subunit were used to identify the presence of PMs as previously reported (Marette et al., 1992). Western blot analysis Equivalent amounts of membrane protein were subjected to sodium dodecyl sulfate polyacrylamide gel electrophoresis on 10% polyacrylamide gels and electrophoretically transferred to nitrocellulose filter membranes for 2 h as previously described (Douen et al., 1990; Marette et al., 1992). The nitrocellulose membranes were incubated for 1 h at room temperature with phosphate-buffered saline (ph 7.2) containing 0.1% Tween 20 and then blocked for 3 h with phosphate-buffered saline (ph 7.2) containing 0.1% Tween 20 and 5% skimmed milk. Nitrocellulose membranes were then incubated overnight with the primary antibody at 4 C followed by four washings with phosphate-buffered saline (ph 7.2) containing 0.1% Tween 20 and incubation with the secondary antibody for 3 h at room temperature. After two washings with phosphate-buffered saline the nitrocellulose membranes were exposed to the luminescent substrate (Amersham) for 1 min. Amersham Hyperfilm was then exposed to the membranes for 1, 5, 15, or 30 s. Immunoblotting of each membrane sample was conducted in duplicate or triplicate. Statistical analysis All data are expressed as mean SD values. The measured physiological parameters, CBF, and the number of CSD depolarization waves were compared between groups using twosample t tests. The difference in glutamate exposure between sham- and CSD-treated rats was analyzed using two-sample t tests assuming heterogeneity of variance. The Hyperfilm images of immunoreactive bands obtained from the ECL procedure were analyzed in a semiquantitative manner using a computer-based imaging device (MCID; Imaging Research, St. Catharines, ON, Canada). Images were digitized, and the band intensities were calculated as the product of increased relative optical density within the band area of the band. The ratio of band intensities from samples obtained from the left (CSDtreated) cortex over the right (untreated) cortex was calculated. Data were analyzed using independent t tests and ANOVA followed by pairwise comparisons using Tukey s HSD (Systat, SPSS, Chicago, IL, U.S.A.). RESULTS Typically, 11 2 CSD depolarizations were observed during the 2 h of 0.5 M KCl application, whereas none were observed during application of 0.5 M NaCl. The average duration of the CSD waves was 70 4 s, consistent with previous reports of CSD (Matsushima et al., 1996a,b). There was no statistically significant difference in physiological measurements (mean arterial blood pressure, arterial ph, PCO 2,PO 2, blood glucose, and rectal temperature) between the sham- and CSD-treated groups during these experiments. The extracellular glutamate concentration in cortex during focal cerebral ischemia in CSD- and sham-treated rats is shown in Fig. 1. The cortical glutamate concen-

4 PRECONDITIONING ALTERS GLUTAMATE REGULATION 815 FIG. 1. Extracellular glutamate levels during cerebral ischemia from CSD- and sham-treated rats. A: Glutamate concentration ( M) in dialysate of cerebral cortex at 10-min intervals during 200 min of cerebral ischemia in selected CSD- and sham-treated rats. B: Comparison of total glutamate exposure ( M-min) during ischemia following CSD (n 11) or sham (n 8) treatment. Mean values are represented by the horizontal lines and are different at *p 0.05 (one-tailed t test with correction for unequal sample size and variance). tration was measured in dialysates obtained at timed intervals over a 200-min period, and the results from selected CSD- and sham-treated rats are shown in Fig. 1A. The index of exposure to glutamate in the cerebral cortex for each individual animal is shown in Fig. 1B. In the CSD treatment group only 23% (three of 13 rats) had a glutamate exposure index of 500 M-min, compared with 87% (seven of eight rats) in the sham treatment group. The mean SEM index of cortical glutamate exposure during ischemia was M-min in the CSD treatment group (n 13) compared with 1,770 1,470 M-min in the sham treatment group (n 8) (Fig. 1B). This represents a 70% decrease in the exposure of cortex to extracellular glutamate following CSD treatment ( p 0.05). CBF at the same cortical location did not differ between the CSD- ( ml/100 g/min) and sham-treated ( ml/100 g/min) rats. Within the caudate the index of glutamate exposure was 1,248 1,100 M-min in the CSD treatment group and did not differ from 940 1,000 M-min in the sham treatment group. CBF at the same location within the caudate did not differ between the CSD- ( ml/100 g/min) and sham-treated ( ml/100 g/min) rats. We next considered the possibility that this CSDmediated decrease in intraischemic cortical glutamate may be linked to regulation of EAATs in the cerebral cortex. PM-enriched fractions were prepared from rat cerebral cortex and identified by immunodetection of the PM-specific Na,K -ATPase 1 subunit, which has been reported to be at least three times more sensitive than biochemical techniques (Marette et al., 1992). Na,K -ATPase 1 subunit monoclonal antibodies readily detected the 110-kDa band of this protein in the 35% sucrose fraction as compared with absent or minimally detectable levels in the crude homogenate (Fig. 2), confirming the ability of the fractionation technique to isolate PM-enriched fraction. There was no significant difference in the left:right cortex ratio of Na,K - ATPase 1 subunit levels in CSD-treated ( ) versus sham-treated ( ) rats, indicating that CSD did not alter the enrichment of the PM fractions. PMs prepared from rat cerebral cortex following CSD or sham treatment were used for immunodetection of EAAT isoforms. We initially explored the presence of EAAT in PMs prepared 3 days post-csd using the same experimental paradigm previously shown to impart neuroprotection against focal ischemia (Matsushima et al., 1996a,b, 1998). There was no difference in EAAT1, EAAT2, and EAAT3 in PMs isolated from the right and left cortices of the sham-treated rats. In contrast, in the CSD-treated rats we observed a marked reduction in EAAT2 and EAAT1, but not EAAT3, from PMs isolated from the left (CSD) cortex compared with the right (untreated) cortex (Fig. 3a). Comparison of the left:right cortex ratio of EAAT isoforms using a semiquantitative lane analysis showed that CSD treatment resulted in a 90 and 50% reduction, respectively, in EAAT2 and EAAT1 (Fig. 3B; p 0.01) 3 days later. We subsequently determined the temporal profile for CSD-mediated down-regulation of PM EAAT2 and EAAT1 (Fig. 4). Immediately following CSD treatment there was no change in PM EAAT2 and EAAT1. However, downregulation of both EAAT2 and EAAT1 was evident by 24 h post-csd and persisted for 1 week, with the maximal effect occurring at 3 days. DISCUSSION Although we and others have reported increased tolerance to ischemia following preconditioning of the mammalian brain with CSD, the mechanisms underlying this neuroprotection are unclear. CSD induces several metabolic and molecular changes, including activation of voltage-sensitive calcium channels (Osuga et al., 1997), up-regulation of immediate early genes (Herrera and Robertson, 1990; Herdegen et al., 1993), increased FIG. 2. Comparison of the distribution of the PM-specific Na,K -ATPase 1 subunit in crude homogenates (H) and membranes (PM) isolated on a 35% sucrose cushion. Lanes 1 6 contain samples prepared from the cortex of a sham-treated rat: 35% membrane fraction (lanes 1 3) and crude homogenate (lanes 4 6). Lanes 7 12 show samples prepared from the cortex of a CSD-treated rat: 35% membrane fraction (lanes 7 9) and crude homogenate (lanes 10 12).

5 816 A. G. DOUEN ET AL. Our measurements of glutamate exposure are consistent with our previous observations of neuroprotection in this model of CSD-induced tolerance. We have previously demonstrated that CSD elicited by KCl infusion into the occipital cortex results in widespread cortical depolarization but does not involve the striatum (Osuga et al., 1997). We have observed a similar pattern of protection with reduced infarct volume following focal cerebral ischemia in the CSD preconditioned cortex but no change in infarct volume within the striatum (Matsushima et al., 1996a). The finding that preconditioning with CSD modulated intraischemic glutamate, coupled with the important role of EAAT in regulating extracellular glutamate levels and increasing reports that glial EAAT2 and EAAT1 are regulatable proteins (Ginsberg et al., 1995; Rothstein, 1996; Raghavendra Rao et al., 1998), prompted further investigation into the effect of CSD on the presence of FIG. 3. Detection of EAAT2, EAAT1, and EAAT3 in PMs prepared from cerebral cortex of rats 3 days after either CSD or sham treatment. A: Western blot analysis. Samples were electrophoresed in duplicate, and lanes were loaded with equivalent amounts of protein. Lanes 1 4 represent a CSD-treated rat: Lanes 1 and 2 show EAAT in PM of the left (L) CSD-treated cortex, and lanes 3 and 4 show EAAT in PM of the right (R) untreated cortex. Lanes 5 8 represent a sham-treated rat: Lanes 5 and 6 show EAAT in PM of the left (L) sham-treated cortex, and lanes 7 and 8 show EAAT in PM of the right (R) untreated cortex. B: Semiquantitative lane analysis of the ratio of the left (ipsilateral):right (contralateral) cortex band intensities of EAAT2, EAAT1, and EAAT3 detected in the PM of CSD- (o; n 3) and sham-treated ( ; n 3) rats. There was marked and significant down-regulation of EAAT2 and EAAT1 but not EAAT3, corroborating the western blot analysis. Data are mean SD (bars) values. **p growth factor expression (Kokaia et al., 1993; Matsushima et al., 1998), and increased levels of glial fibrillary acidic protein (Kraig et al., 1991; Matsushima et al., 1998), but the relevance of these changes to induced neuroprotection remains to be determined. Because neuronal injury following cerebral ischemia is mediated in part through elevation of extracellular cerebral glutamate levels, we considered whether CSD preconditioning could influence intraischemic cerebral glutamate levels. Indeed, we observed that, relative to sham treatment, induction of CSD 3 days before MCAO resulted in a significant 70% decrease in the exposure of cortex to extracellular glutamate levels during the ischemic insult. This was not a consequence of altered CBF as both the CSD- and sham-treated rats experienced the same ischemic insult. We did not observe a similar reduction in glutamate exposure within the caudate. Our intraischemic glutamate concentrations are comparable to those in previous reports from our laboratory (Osuga and Hakim, 1994, 1996). In the sham pretreatment group our levels of elevated glutamate were similar within cortex and caudate, as was the degree of blood flow reduction. FIG. 4. Temporal profile of EAAT2 and EAAT1 down-regulation. A: Western blot analysis shows EAAT2 detection in PMs prepared from CSD-treated rats immediately following CSD treatment (0 day) and at days 1, 3, 7, and 21 following CSD treatment (n 3 rats for each time point). Samples were assayed in duplicate. Lanes were loaded with equivalent amounts of protein. Lanes 1 and 2 show EAAT2 detection in PM from the left (L) CSD-treated cortex, and lanes 3 and 4 show EAAT2 detection in PM of the right (R) untreated cortex. B: Semiquantitative lane analysis of the ratio of the left (ipsilateral):right (contralateral) cortex band intensities of EAAT2 detected in the PM of CSDtreated rats. Data are mean SD (bars) values. *p 0.05, **p 0.01 by ANOVA with Tukey s HSD. C: Semiquantitative lane analysis of the ratio of the left (ipsilateral):right (contralateral) cortex band intensities of EAAT1 detected in the PM of CSDtreated rats. Data are mean SD (bars) values. *p 0.05, **p 0.01 by ANOVA with Tukey s HSD. Three rats were studied at each time except 7 days, where n 2.

6 PRECONDITIONING ALTERS GLUTAMATE REGULATION 817 EAATs in the cerebral cortex. We observed that CSD treatment caused a delayed and transient down-regulation of EAAT2 and EAAT1 from PMs isolated from the cerebral cortex. This down-regulation of EAAT2 and EAAT1 was most evident at 3 days post-csd: 90 and 50%, respectively. In contrast, CSD did not alter EAAT3 (Fig. 3) or Na,K -ATPase (Fig. 2) within the PM, indicating that CSD-mediated down-regulation of EAAT2 and EAAT1 was not the result of nonspecific down-regulation of PM proteins. Moreover, the temporal profile of EAAT2 and EAAT1 down-regulation is coincident with both the observed decrease in intraischemic extracellular glutamate concentrations in CSD-treated rats (Fig. 1) and the temporal profile of both ischemia- (Kato et al., 1991) and CSD-induced (Taga et al., 1997) neuroprotection. Although uptake through glial EAAT is the primary mechanism of cerebral glutamate clearance under normal conditions, these facilitative transporters are sensitive to ionic perturbations that occur during ischemia (Nicholls and Attwell, 1990; Szatkowski and Attwell, 1994). Clearance of glutamate from the extracellular space via the EAAT relies on the simultaneous uptake and cotransport of sodium, which moves into the cell down a concentration gradient. However, this sodium ionic gradient is reversed during ischemia such that sodium, and concomitantly glutamate, is extruded from the cells via the EAAT (Nicholls and Attwell, 1990; Szatkowski and Attwell, 1994; Rutledge and Kimelberg, 1996). This mechanism of hypoxia-induced reversal of the EAAT is further supported by recent studies showing that inhibition of EAAT2 with dihydrokainate during cerebral ischemia (Seki et al., 1999) or anoxic injury to the spinal cord (Li et al., 1999) significantly decreases glutamate release, thereby implicating glutamate flux through EAAT2 as an important source of extracellular cerebral glutamate during ischemic conditions. In addition, Seki et al. (1999) showed that 80% of ischemia-induced glutamate release could be impaired through combined blockade of EAAT2 and anion channels, with the latter mediating glutamate release through activation of regulatory cell volume pathways. Hence, these observations suggest only a modest role of presynaptic release of glutamate during ischemia and imply the potential for amelioration of glutamate excitotoxicity by inhibition of EAATs (Robinson, 1999). Consequently, reduction of EAAT2 and EAAT1 from PMs following CSD preconditioning would mean less EAAT available for glutamate efflux during subsequent ischemia and would be consistent with decreased intraischemic glutamate levels (Fig. 1) and hence neuroprotection. Although down-regulation of EAAT2 is thought to be associated with decreased glutamate clearance and neurotoxicity in patients with amyotrophic lateral sclerosis (Rothstein, 1996), this pathologic condition is likely mediated by a chronic decrease in EAAT2 content over months to years. We have demonstrated here that CSDmediated down-regulation of EAAT2 is a transient phenomenon with a maximal effect occurring at 3 days post-csd, consistent with the reported period of CSDinduced neuroprotection (Fig. 4). Because reversal of glutamate transport by the EAAT exacerbates excitotoxicity during ischemia, transient blockade of EAAT2 during an ischemic insult may be neuroprotective (Robinson, 1999; Seki et al., 1999). We cannot at this time entirely rule out the possibility that other mechanisms participate in the induction of neuroprotection. First, it is interesting that our observation of transient down-regulation of EAAT1 and EAAT2 is comparable to that recently observed following focal brain injury (Raghavendra Rao et al., 1998). Although we have shown that extradural application of KCl to induce CSD does not result in widespread neuronal injury (Matsushima et al., 1998), we cannot exclude the possibility that minimal focal cortical damage occurs following KCl application, which might influence EAAT regulation. The role of localized injury in neuroprotection is unknown but has been reported following other focal preconditioning stimuli (Glazier et al., 1994; Chen et al., 1996). Second, it is possible that down-regulation of EAATs may play a role in modulation of feedback mechanisms involved in glutamate regulation and release. It has been previously suggested that a tonic rise in synaptic glutamate levels, potentiated through the blockade of glutamate transporters using L-trans-2,4-pyrrolidine dicarboxylic acid, leads to feedback regulation of presynaptic glutamate release via activation of metabotropic glutamate receptors (Maki et al., 1994). Excitatory postsynaptic potentials can be manipulated by agonists and antagonists of specific presynaptic metabotropic glutamate receptors, suggesting a role of metabotropic glutamate receptors as autoreceptors controlling glutamate release (Dube and Marshall, 1997). Hence, it is possible that CSD-mediated down-regulation of EAAT2 and EAAT1 from PMs might represent a physiologic blockade of glutamate uptake that may facilitate an increase in ambient glutamate levels. This may in turn serve to autoregulate presynaptic metabotropic glutamate receptors and ultimately glutamate release during ischemic conditions. However, this remains speculative. In conclusion, it is likely that multiple factors contribute to neuroprotection following CSD preconditioning. We have previously reported a close association between growth factor expression and activation of astrocytes with CSD-induced tolerance to cortical ischemia (Matsushima et al., 1998). However, a specific final mechanism underlying neuroprotection has not been described. EAAT down-regulation may be such a mechanism. In this study we show that preconditioning with CSD 3 days before focal ischemia decreases intraischemic glutamate levels by 70%, compared with the nonpreconditioned animal. This decrease in intraischemic glutamate content correlates with a simultaneous 90 and 50% down-regulation, respectively, of EAAT2 and EAAT1 from the PM and supports our previous study showing a 50% reduction in infarction volume following CSD preconditioning. Our data are consistent with the view that reversal of the glial EAAT is a major contributor to extracellular

7 818 A. G. DOUEN ET AL. glutamate excess during ischemia. CSD-mediated reduction in PM EAAT2 and EAAT1 levels may reduce glutamate efflux during subsequent ischemia and contribute to the neuroprotective effect of CSD preconditioning. Acknowledgment: This work was supported by the Heart and Stroke Foundation of Ontario and the Medical Research Council of Canada. A.G.D. is the recipient of a Heart and Stroke Foundation of Ontario Research Fellowship. The technical assistance of Ms. Georgette Roy and secretarial assistance of Ms. Bea Robertson and Mrs. Elaine Boudreau are acknowledged. REFERENCES Chen J., Graham S. H., Zhu R. L., and Simon R. P. (1996) Stress proteins and tolerance to focal cerebral ischemia. J. Cereb. Blood Flow Metab. 16, Choi D. W. (1988) Glutamate neurotoxicity and diseases of the nervous system. Neuron 1, Douen A. G., Kacem R., and Jones M. N. (1988) Direct interaction of phenylarsine oxide with hexose transporters in isolated rat adipocytes. Biochim. Biophys. Acta 994, Douen A. G., Ramlal T., Rastogi S., Bilan P. J., Cartee G., Vranic M., Holloszy J. O., and Klip A. (1990) Exercise induces recruitment of the insulin responsive glucose transporter. Evidence for distinct intracellular insulin- and exercise-recruitable transporter pools in skeletal muscles. J. Biol. Chem. 265, Dube G. R. and Marshall K. C. (1997) Modulation of excitatory synaptic transmission in locus coeruleus by multiple presynaptic metabotropic glutamate receptors. Neuroscience 80, Farrar J. K. (1987) Hydrogen clearance technique, in Cerebral Blood Flow Physiologic and Clinical Aspects (Wood J. H., ed), pp McGraw-Hill, New York. Ginsberg S. D., Martin L. J., and Rothstein J. D. (1995) Regional deafferentiation down-regulates subtypes of glutamate transporter proteins. J. Neurochem. 65, Glazier S., O Rourke D. M., Graham D. I., and Welsh F. A. (1994) Induction of ischemic tolerance following brief focal ischemia in rat brain. J. Cereb. Blood Flow Metab. 14, Herdegen T., Sandkuhler J., Gass P., Kiessling M., Bravo R., and Zimmermann M. (1993) JUN, FOS, KROX and CREB transcription factor proteins in the rat cortex: basal expression and induction by spreading depression and epileptic seizures. J. Comp. Neurol. 333, Herrera D. G. and Robertson H. A. (1990) Application of potassium chloride to the brain surface induces the c-fos proto-oncogene: reversal by MK-801. Brain Res. 510, Kato H., Liu Y., Araki T., and Kogure K. (1991) Temporal profile of the effects of pretreatment with brief cerebral ischemia on the neuronal damage following secondary ischemic insult in the gerbil: cumulative damage and protective effects. Brain Res. 553, Kawahara N., Ruetzler C. A., and Klatzo I. (1995) Protective effect of spreading depression. Neurol. Res. 17, Kokaia Z., Gido G., Ringstedt G., Bengzon J., Kokaia M., Siesjö B. K., Persson H., and Lindvall O. (1993) Rapid increase of BDNF mrna levels in cortical neurons following spreading depression: regulation by glutamatergic mechanisms independent of seizure activity. Mol. Brain Res. 19, Kraig R. P., Dong L., Thisted R., and Jaeger C. B. (1991) Spreading depression increases immunohistochemical staining of glial fibrillary acidic protein. J. Neurosci. 11, Li S., Mealing G. A. R., Morley P., and Stys P. K. (1999) Novel injury mechanism in anoxia and trauma of spinal cord white matter: glutamate release via reverse Na -dependent glutamate transport. J. Neurosci. 19, RC16, 1 9. Maki R., Robinson M. B., and Dichter M. A. (1994) The glutamate uptake inhibitor L-trans-pyrrolidine-2,4-dicarboxylate depresses excitatory synaptic transmission via a presynaptic mechanism in cultured hippocampal neurons. J. Neurosci. 14, Marette A., Burdette E., Douen A. G., Vranic M., and Klip A. (1992) Insulin induces the translocation of GLUT-4 glucose transporters from a unique intracellular organelle to transverse tubules in rat skeletal muscles. Diabetes 41, Matsushima K., Hogan M. J., and Hakim A. M. (1996a) Cortical spreading depression protects against subsequent focal cerebral ischemia in rats. J. Cereb. Blood Flow Metab. 16, Matsushima K., Schmidt-Kastner R., and Hakim A. M. (1996b) Spreading depression may protect from subsequent ischemic damage by increasing expression of neurotrophic factors. Brain Res. 807, Matsushima K., Schmidt-Kastner R., Hogan M. J., and Hakim A. M. (1998) Cortical spreading depression activates trophic factor expression in neurons and astrocytes and protects against subsequent focal brain ischemia. Brain Res. 807, Nicholls D. and Attwell D. (1990) The release and uptake of excitatory amino acids. Trends Pharmacol. Sci. 11, Osuga H. and Hakim A. M. (1994) Relevance of interstitial glutamate to selective vulnerability in focal cerebral ischemia. J. Cereb. Blood Flow Metab. 14, Osuga H. and Hakim A. M. (1996) Relationship between extracellular glutamate concentration and voltage-sensitive calcium channel function in focal cerebral ischemia in the rat. J. Cereb. Blood Flow Metab. 16, Osuga H., Hakim A. M., Osuga H., and Hogan M. (1997) In vivo uptake of [ 3 H]nimodipine into brain during cortical spreading depression. J. Cereb. Blood Flow Metab. 17, Raghavendra Rao V. L., Baskaya M. K., Doğan A., Rothstein J. D., and Dempsey R. J. (1998) Traumatic brain injury down-regulates glial glutamate transporter (GLT-1 and GLAST) proteins in rat brain. J. Neurochem. 70, Robinson S. E. (1999) Inhibition of ischemia-induced glutamate release in rat striatum by dihydrokainate and anion channel blocker. Stroke 30, 440. Rothstein J. D. (1996) Excitotoxic hypothesis. Neurology 47 (Suppl.), S19 S26. Rothstein J. D., Dykes-Hoberg M., Pardo C. A., Bristol L. A., Jin L., Kuncl R. W., Kanai Y., Hediger M. A., Wang Y., Schielke J. P., and Welty D. F. (1996) Knockout of glutamate transporters reveals a major role for astroglial transport in excitotoxicity and clearance of glutamate. Neuron 16, Rutledge E. M. and Kimelberg H. K. (1996) Release of [ 3 H]-Daspartate from primary astrocyte cultures in response to raised external potassium. J. Neurosci. 16, Seki Y., Feustel P. J., Keller R. W., Tranmer B. I., and Kimelberg H. K. (1999) Inhibition of ischemia-induced glutamate release in rat striatum by dihydrokainate and anion channel blocker. Stroke 30, Siesjö B. K. (1994) Calcium-mediated processes in neurodegeneration. Ann. NY Acad. Sci. 747, Szatkowski M. and Attwell D. (1994) Triggering an execution of neuronal death in brain ischemia: two phases of glutamate release by different mechanisms. Trends Neurosci. 17, Taga K., Patel P. M., Drummond J. C., Cole D. J., and Kelly P. J. (1997) Transient neuronal depolarization induces tolerance to subsequent forebrain ischemia in rats. Anesthesiology 87, Zea Longa E. Z., Weinstein P. R., Carlson S., and Cummins R. (1989) Reversible middle cerebral artery occlusion without craniectomy in rats. Stroke 20,