The Role of Postlesion Seizures and Spreading Depression in the Upregulation of Glial Fibrillary Acidic Protein mRNA after Entorhinal Cortex Lesions

EXPERIMENTAL NEUROLOGY ARTICLE NO.

139, 83–94 (1996)

0083

The Role of Postlesion Seizures and Spreading Depression in the Upregulation of Glial Fibrillary Acidic Protein mRNA after Entorhinal Cortex Lesions MELINDA S. KELLEY

AND

OSWALD STEWARD1

Department of Neuroscience, University of Virginia, Charlottesville, Virginia 22908

studied example of synaptic reorganization following injury is that which occurs in the dentate gyrus (DG) after unilateral destruction of the entorhinal cortex (28). Because of the substantial synaptic reorganization that occurs, there has been considerable interest in defining the changes in gene expression that occur in conjunction with this growth. In this regard, several studies have revealed substantial alterations in gene expression by hippocampal astrocytes; these changes in gene expression are thought to be part of the process by which astrocytes become ‘‘reactive’’ in response to the injury. One prominent component of the reactive response is the upregulation of expression of the astrocytic intermediate filament protein glial fibrillary acidic protein (GFAP). Following unilateral lesions of the entorhinal cortex (EC), GFAP mRNA levels increase dramatically in the DG and hippocampus (31). The increases in GFAP mRNA levels are followed by increases in GFA protein levels and astrocyte hypertrophy (5, 26, 30). There appear to be several components of this response that differ in spatial extent and time course. In areas that contain degeneration debris (the molecular layer of the DG and stratum lacunosum moleculare of the hippocampus), GFAP mRNA levels increase rapidly to a peak at 1–2 days postlesion, decrease between 2 and 4 days, and then increase again to a second peak at about 8–10 days postlesion (31). There are also increases in GFAP mRNA levels in nondenervated regions of the hippocampus (stratum radiatum) and in the contralateral hippocampus and DG that contain little degeneration debris (31). In these nondenervated zones, GFAP mRNA levels increase to a peak at 2 days and then return to near baseline. Interestingly, the rapid transient increases in GFAP mRNA levels in nondenervated zones on the side contralateral to the lesion (and presumably also in nondenervated zones ipsilaterally) are seen in some animals but not in all (30). The reasons for the individual differences in the pattern of GFAP induction are not known. However, an important clue is that similar increases in GFAP expression are seen after periods of

Unilateral lesions of the entorhinal cortex have been shown to lead to dramatic increases in GFAP mRNA levels in denervated zones in the hippocampus and dentate gyrus and sometimes (but not always) in nondenervated zones in the contralateral hippocampus and dentate gyrus. The variable distribution of the increases in GFAP mRNA expression suggests that the events which trigger changes in GFAP mRNA levels occur to a variable extent in individual animals. The companion paper characterizes two candidate triggering events: spreading depression (SD) that occurs to a variable extent at the time of the lesion and recurrent seizures that occur during the early postlesion interval. The goal of the present study was to evaluate whether individual differences in the extent or spatial distribution of lesion-induced increases in GFAP mRNA are related to the occurrence of either SD or seizures. We quantified the increases in GFAP mRNA levels in individual animals that had been monitored physiologically to define the incidence of SD and postlesion seizures. The results revealed that the quantitative extent of the increases in GFAP mRNA in denervated zones and was not related to either SD or postlesion seizures. The increases in GFAP mRNA in nondenervated zones also were not related to episodes of spreading depression that occurred at the time of lesion production but were related to the spontaneous seizures that developed during the first 24 h postlesion after the animals had recovered from the surgical anesthesia. Taken together, these data indicate that physiological events that occur during the early postlesion interval can play an important role in determining the pattern and extent of altered cellular gene expression in response to an injury. r 1996 Academic Press, Inc.

INTRODUCTION

The hippocampus and its afferents have served as an important model system for studies of cellular and molecular responses to CNS injury. An especially well1

To whom correspondence should be addressed. 83

0014-4886/96 $18.00 Copyright r 1996 by Academic Press, Inc. All rights of reproduction in any form reserved.

84

KELLEY AND STEWARD

intense neuronal activity, such as hippocampal seizures (11, 32) or episodes of spreading depression (SD) (1, 16). These observations have led to the hypothesis that the increases in GFAP mRNA expression in nondenervated zones may be triggered by episodes of intense neuronal activity, specifically seizures or SD that occur as a result of the lesion. The idea that postlesion physiological events regulate cellular gene expression following injury is of considerable interest because it is generally assumed that injury-induced changes in gene expression are triggered by degenerative processes. However, in the companion paper, we demonstrate that electrolytic lesions of the EC do lead to episodes of SD at the time of the lesion and recurrent seizures during the early postlesion interval. Both SD and seizures have been demonstrated to upregulate the expression of a number of genes that are thought to play a key role in neuronal growth and plasticity. For example, in addition to upregulating GFAP expression (11, 32), seizures upregulate the expression of mRNAs for nerve growth factor (NGF) (6, 18), basic fibroblast growth factor (bFGF) (7), brain-derived neurotrophic factor (BDNF) (15), and various immediate early genes including c-fos and c-jun (23, 27). SD has also been shown to upregulate some of the same genes including GFAP (1, 16), NGF (14), and c-fos (12–14). Because EC lesions (and other injuries which denervate the DG) lead to increases in the expression of many of these genes, including GFAP (31), bFGF (8, 9), c-fos (25, 10), BDNF (10), and NGF (3, 10), the question arises whether these increases are due to degenerative processes or to the physiological events that are induced by the lesions. In the present study, we evaluate whether individual differences in the occurrence of SD or postlesion seizures are related to individual differences in the pattern of GFAP induction following EC lesions. Our results indicate: (a) that the magnitude of the increases in GFAP expression in denervated zones is not related to the physiological events that occur during the immediately after the lesion and (b) that the individual differences in the spatial pattern of GFAP induction are not related to the occurrence of episodes of SD at the time of lesion production, but are related to the seizures that occur during the early postlesion interval. MATERIALS AND METHODS

Experimental Animals The experimental animals used in this study were 29 240–300 g male Sprague–Dawley rats. Animals were housed two to three per cage and maintained on a standard light–dark cycle. Animals had free access to food and water at all times. The animals that were monitored physiologically were the same animals that were described in the companion paper. Animals were anesthetized with sodium pentobarbi-

tal (50 mg/kg) and positioned in a stereotaxic apparatus. Unilateral electrolytic entorhinal cortex (EC) lesions were produced in a total of 23 animals as previously described (20, see companion paper for additional details). The procedures used to monitor physiological events during the production of the lesions and to monitor EEG activity during the early postlesion period are described in the accompanying paper. The same animals (with lesions) that were monitored physiologically as described in the accompanying paper were also used for GFAP mRNA analysis in the present experiment. An additional group of animals (n 5 5) was added for the present analysis to determine if the blockade of postlesion seizures prevents the lesion-induced upregulation of GFAP mRNA expression. This group of animals received EC lesions in combination with multiple injections of the anticonvulsant phenobarbital (administered before and after the lesion). The details of this injection paradigm are described below. Phenobarbital Treatment In order to develop an injection regimen that produced serum levels of phenobarbital in the anticonvulsive range, an initial experiment was carried out in which two adult rats received four separate intraperitoneal injections of phenobarbital (60 mg/kg) spaced 12 h apart. Twelve hours after the second injection, the animals were anesthetized with pentobarbital (50 mg/ kg) as if they were receiving a lesion, but no lesion was made. The third phenobarbital injection was administered immediately after the animals began to recover from the anesthesia, and the fourth was given approximately 12 h later. Blood was withdrawn from these animals 24 h after the first phenobarbital injection, and again 24 h later (immediately before the animal was euthanized). Levels of serum phenobarbital were determined using a fluorescence polarization assay (Roche Biomedicals) and compared to levels in two naive animals. Assays were performed in the Clinical Laboratory Services Department at the University of Virginia. Animals were euthanized at 48 h after the first phenobarbital injection, and their hippocampi were dissected out for RNA isolation and dot blot hybridization to determine if the multiple injections of phenobarbital affected GFAP mRNA expression. Experimental animals received phenobarbital injections on the same schedule as the animals described above, except that the experimental animals received EC lesions 12 h after the second phenobarbital injection. A total of five animals received EC lesions and phenobarbital injections. cRNA Probe Preparation The preparation of the cRNA probe for GFAP mRNA has been described previously (31). The probe was

NEURONAL ACTIVITY AND GFAP mRNA EXPRESSION

derived from a 2.5-kb cDNA clone for mouse GFAP. The clone was cut with the HindIII restriction enzyme, yielding a 1.26-kb fragment of the 58 region of the clone. This fragment was recloned into the HindIII site of a Bluescript M13 vector by D. Chikaraishi (Tufts University). The plasmid containing the GFAP insert was provided to us as a gift. 35S-Labeled antisense cRNA probes of approximately 1 kb in length were synthesized from the linearized plasmid using the T3 promoter. The specific activity of the probes ranged from 1.8–7.1 3 108 cpm/µg. Specificity of probe binding has been previously confirmed with Northern blot analysis in which it was shown that the cRNA probes produced as described above bind to a single band of approximately 2.6 kb (31). RNA Isolation and Dot Blot Hybridization For dot blot analyses of GFAP mRNA levels, animals were euthanized with halothane and decapitated, and the hippocampi were rapidly dissected out and frozen on dry ice. Total RNA was isolated from the hippocampi using a modification of the technique of Chomczynski and Sacchi (2). The frozen tissue was homogenized in a buffer containing 4 M guanidinium thiocyanate, 5 mM sodium citrate, 0.5% sarkosyl, and 0.1 M b-mercaptoethanol. The homogenate was extracted once with phenol/ chloroform and RNA was precipitated with 2 M sodium acetate and isopropanol at 220°C. The RNA pellet was dissolved in guanidinium buffer and reprecipitated with sodium acetate in isopropanol. The pellet was dissolved in a buffer containing 10 mM Tris–HCl, 5 mM EDTA, and 0.2% SDS and precipitated a final time with sodium acetate in isopropanol. Last, the pellet was washed with ethanol, dried, dissolved in water treated with 0.1% diethylpyrocarbonate and stored at 280°C. Concentrations of RNA were determined by measuring the optical density at 260 nm with a spectrophotometer. The dot blot hybridization protocol has been described previously (31). RNA was dissolved in a 7% formaldehyde, 103 standard saline/citrate (SSC) buffer, and denatured for 10 min at 65°C. Four micrograms of each sample were spotted onto a Nytran membrane in a Bio-Rad manifold. The relationship between the amount of RNA bound to the membrane and the extent of probe binding has been defined previously, and samples containing 4 µg of RNA are nonsaturating (31). After RNA application, the membranes were dried, and the RNA was fixed to the membranes by UV-crosslinking. Membranes were then prehybridized at 65°C in a buffer containing 50% formamide, 53 standard saline/phosphate/EDTA (SSPE), 53 Denhardt’s solution, 1% SDS, 100 µg/ml of tRNA, and 50 µg/ml of salmon sperm DNA. Hybridizations were carried out under the same conditions with the addition of 35S-labeled probe (106 cpm/ml, 4 ng/ml). Following a 20- to 24-h hybridization, the membranes were washed 33 15 min at 65°C in 13

85

SSPE/1% SDS, followed by a 30-min wash at 65°C in high stringency buffer (0.13 SSPE/1% SDS). Membranes were treated with RNase A (10 µg/ml) for 15 min at room temperature, dried, and exposed to Kodak X-Omat film for autoradiography. Individual dots were cut out of the membrane for measurement of radioactivity with a scintillation counter. In Situ Hybridization The detailed protocol for in situ hybridization with the GFAP probe has also been described previously (31). Animals were deeply anesthetized with sodium pentobarbital and perfused with 4% paraformaldehyde in 0.1 M phosphate buffer. Brains were postfixed overnight in 4% paraformaldehyde and immersed overnight in 15% sucrose/0.1 M phosphate buffer. Brains were mounted on the chuck of a cryostat with O.C.T. tissue embedding medium and frozen with dry ice. Cryostat sections were cut at 20 µm and collected on polylysine-coated slides. The slides were stored at 280°C. For in situ hybridization, a 1-in-10 series of sections was postfixed in 4% paraformaldehyde, treated with proteinase K [5 µg/ml in a Tris/NaCl buffer (0.5 M NaCl/10 mM Tris, pH 8.0) for 30 min], washed twice in 0.53 SSC, and dried. Slides with sections were placed in humidified Petri dishes, and the sections were covered with 100 µl of hybridization buffer containing 50% formamide, 0.3 M NaCl, 20 mM Tris buffer, 5 mM EDTA, 13 Denhardt’s, 10% Dextran Sulfate, and 10 mM DTT, and prehybridized for 1–3 h at 42°C. Twenty microliters of hybridization buffer, containing 1 3 106 cpm of probe (approximately 1–2 µl) and 2 µg of tRNA, were added to each section and a coverslip was placed over the section. Sections were hybridized overnight at 55°C, treated with RNase A (2 mg/ml in Tris/NaCl buffer) for 30 min at room temperature, and washed for 2 h at 55°C in a high-stringency buffer (0.13 SSC/10 mM b-mercaptoethanol/1 mM EDTA). Slides were exposed to Amersham Betamax Hyperfilm for 1–3 days to obtain film autoradiographs. Data Analysis To determine the extent of hybridization in the in situ hybridizations, the optical density (OD) of film autoradiographs was measured using an MCID Imaging System (Imaging Research, Inc., St. Catherines, Ontario). Relative OD values were standardized by measuring the OD produced by a set of calibration standards. Typically, four to five sections were assessed per animal, and a total of eight brain regions were counted per section, including the DG (dorsal leaf, molecular layer), hippocampus proper (the analysis window was centered over the stratum radiatum, avoiding the stratum lacunosum-moleculare which receives input from the EC), thalamus, and cortex bilaterally. Al-

86

KELLEY AND STEWARD

though some of the quantification areas were not restricted completely to specific layers of the dentate gyrus or hippocampus, these areas were grouped into two general categories: those that were substantially denervated by the lesion (ipsilateral DG) and those that were not substantially denervated by the lesion (stratum radiatum of the ipsilateral and contralateral hippocampus and the contralateral DG). Three (and in a few cases, two) 25-pixel measurements were made in each of these areas, and these values were averaged for each section. The average OD values for identical areas were calculated for each individual animal. Because the analysis required evaluation of the relationship between levels of GFAP mRNA and physiological events in individual animals, the data were analyzed on an animal-by-animal basis. Populationbased statistical analyses were used to compare GFAP mRNA levels between animals that exhibited seizures and those that did not (see below). For these comparisons, unpaired two-tailed t tests were used to determine statistical significance. Comparisons of GFAP mRNA levels assessed with dot blot hybridization were also made using unpaired two-tailed t tests to determine statistical significance. Criteria for Inclusion in the GFAP mRNA Analysis The analysis of GFAP mRNA levels in animals that exhibited spreading depression was limited to animals where SD had been monitored with both DC recordings and recordings of evoked potentials produced by stimulation of the commissural pathway (n 5 8). Two of these animals were excluded because their lesions encroached into a portion of the ventral hippocampus. Thus, a total of six animals was used for analysis of GFAP mRNA levels. For the analysis of GFAP mRNA levels in animals used for postlesion seizure assessment, one animal from the lesion-alone group was excluded because of insufficient fixation of the tissue. Two additional animals in the phenobarbital-treated group were excluded from analysis of GFAP mRNA levels on the side of the brain ipsilateral to the lesion because their lesions encroached into the ventral hippocampus and DG. However, these animals were used for analysis of GFAP mRNA levels in the contralateral hippocampus and DG. Thus, a total of seven animals with lesions and two controls were used for analysis of GFAP mRNA levels in the ipsilateral hippocampus and DG. Nine animals with lesions and two controls were used for analysis of GFAP mRNA levels in the contralateral hippocampus and DG. RESULTS

The pattern of occurrence of SD and postlesion seizures in the animals used for the present study was

described in detail in the companion paper. In the discussion that follows, we describe the relationship between the physiological events and the pattern of GFAP induction in individual animals focusing on two features of the response: (1) the extent of the increases in GFAP mRNA levels in the denervated zone; and (2) the spatial distribution of increases in GFAP mRNA levels in nondenervated zones, especially the hippocampus and DG contralateral to the lesion. Spreading Depression and GFAP Induction Three of the six animals that were monitored during the production of EC lesions exhibited at least one episode of SD. Two animals exhibited several episodes (see Table 1 of companion paper). In all animals, GFAP mRNA levels were substantially increased in the denervated zones in the DG and hippocampus. The extent of the increase in GFAP mRNA levels in the denervated zones was not greater in the animals that exhibited multiple episodes of SD. For example, Fig. 1 illustrates film autoradiographs of sections from three animals with different SD profiles (ranging from 0 to 4) and Fig. 2 illustrates the levels of GFAP mRNA in these animals and the others that were analyzed. Although these animals exhibited different numbers of episodes of SD, the extent of the increases in GFAP mRNA in the denervated zones (ipsilateral side) were comparable. In particular, the animals that exhibited four episodes of SD did not exhibit higher levels of GFAP mRNA than did the three animals that exhibited no episodes of SD. Two of the six animals also exhibited increases in GFAP mRNA levels in nondenervated zones (specifically in the hippocampus and DG on the side contralateral to the lesion). However, there was no relationship between the number of episodes of SD and the increases in GFAP mRNA levels in nondenervated zones. The largest increases on the contralateral side were seen in an animal that exhibited no episodes of SD. Of two other animals that exhibited moderate increases in GFAP mRNA, one exhibited no episodes of SD; the other exhibited three episodes. These results suggest that SD is not the key variable that accounts for the increases in GFAP levels in either denervated or nondenervated zones. The Role of Postlesion Seizures in Upregulating GFAP mRNA Levels As noted in the companion paper, all animals that received EC lesions exhibited postlesion seizure activity on the side ipsilateral to the lesion, and three of five animals exhibited bilateral seizure activity (see accompanying paper for details). To better evaluate the role that postlesion seizures might play in inducing GFAP expression, it was desirable to produce a group of animals in which no seizures occurred. Thus, we at-

NEURONAL ACTIVITY AND GFAP mRNA EXPRESSION

87

FIG. 1. Increases in GFAP mRNA in the ipsilateral hippocampus and dentate gyrus in animals with varying numbers of episodes of spreading depression. Levels of GFAP mRNA were assessed with in situ hybridization. These animals are noted on Fig. 2 with an asterisk. The number of episodes of spreading depression exhibited by each of these animals is noted in the lower right corner of each picture. Levels of GFAP mRNA in ipsilateral and contralateral hippocampus and dentate gyrus appeared comparable in animals with 0, 1, and 4 episodes of spreading depression. Arrows indicate hippocampus/dentate gyrus. ECL, entorhinal cortex lesion. Scale bar indicates 1 mm.

tempted to block the seizures that would otherwise occur by treatment with phenobarbital. In an initial experiment involving two animals, it was found that a dosing regimen in which four injections of phenobarbital were delivered at 12-h intervals maintained anticonvulsive levels of the drug for the duration of the experiment. In the two animals that were evaluated, levels of phenobarbital 24 h after the

FIG. 2. Increases in levels of GFAP mRNA in ipsilateral and contralateral dentate gyrus after entorhinal cortex lesions, ranked by the number of episodes of spreading depression that occurred in each animal. Comparable levels of GFAP mRNA were observed in animals that did not exhibit spreading depression and in animals that did exhibit spreading depression. The number at the bottom of each column indicates the number of episodes of spreading depression recorded in that particular animal. Asterisks denote animals shown in Fig. 1. Error bars indicate the SEM; mean values were obtained by averaging values from a single area of tissue over several sections from the same animal.

initial injection (just before the lesion would be made) were 39.8 and 44.9 µg/ml. Levels of phenobarbital 12 h after the last injection were 46.4 and 68.6 µg/ml (in the same two animals). The nominal levels in salinetreated control animals were below 5.0 µg/ml. The levels observed in the phenobarbital-treated animals all fell into or above the range of serum phenobarbital concentrations considered to be anticonvulsive (10–40 µg/ml) (22). To determine if prolonged phenobarbital treatment affected GFAP mRNA levels, GFAP mRNA levels were assessed by dot blot hybridization in the two animals that had received phenobarbital. This analysis revealed that the average GFAP mRNA level in the phenobarbital-treated animals was slightly lower than the average in control animals (161.17 6 35.73 for saline-treated animals, 103.08 6 0.83 for phenobarbitaltreated animals), but these differences were not significant (P 5 0.16). Five animals that received EC lesions plus phenobarbital were used for EEG recordings. Three of these animals did not exhibit any seizure activity during the recording period. However, two of the animals that received multiple injections of phenobarbital did exhibit some seizure activity, and the activity occurred bilaterally. The timing of the seizures in these two phenobarbital-treated animals is shown in Fig. 3A and sample EEG tracings from these two animals are shown in Fig. 3B. Seizures in animals that received EC lesions plus phenobarbital appeared similar to those observed in animals with lesions alone (see accompanying paper for additional details). However, animals treated with phenobarbital did exhibit slightly faster

88

KELLEY AND STEWARD

the spatial distribution of the seizures. Thus, the data below are organized in terms of whether seizures occurred on the side being analyzed (i.e., ipsilateral vs contralateral to the lesion). Figure 4 illustrates film autoradiographs of the distribution of GFAP mRNA in an unlesioned control animal (R1; Fig. 4A), an animal with an EC lesion that exhibited no seizures (P3; Fig. 4B), an animal that exhibited multiple seizures on the side ipsilateral to the lesion (E3; Fig. 4C), an animal that exhibited multiple ipsilateral seizures as well as a single seizure on the contralateral side (E5; Fig. 4D), and an animal that exhibited multiple seizures on both the ipsilateral and contralateral sides (E1; Fig. 4E). The increases in GFAP mRNA in the denervated DG ipsilateral to the lesion appeared comparable regardless of whether or not seizures occurred. However, the extent of the increases in the nondenervated zones and especially in the contralateral hippocampus and DG appeared to be related to the number, timing, and spatial distribution of the seizures, as further described below. These conclusions were confirmed by the quantitative analyses. Seizures and Levels of GFAP mRNA in the Denervated Zones in the Dentate Gyrus

FIG. 3. (A) Time course and duration of ipsilateral and contralateral seizures in animals that received an entorhinal cortex lesion plus phenobarbital. Lines denote ipsilateral seizures, and triangles denote contralateral seizures. One of the two animals that received a lesion plus phenobarbital exhibited seizures that were tightly clustered together into groups. This animal exhibited a smaller increase in GFAP mRNA than would have been expected in terms of the number of seizures that occurred. The timing of the seizures, therefore, may have affected the efficacy of these seizures in upregulating GFAP mRNA. (B) The appearance of the EEG during seizures in animals that received EC lesions plus phenobarbital (P2, P4) and a control animal (ctl). Spike amplitude appeared to be more irregular in animals treated with phenobarbital, and the rate of spike–wave events may have been slightly faster in animals with phenobarbital treatment. Scale bars indicate 10 s, 500 µV.

spike–wave complexes than did untreated animals. In addition, the animal treated with phenobarbital that exhibited multiple seizures showed a much greater degree of clustering of the seizures. Because the phenobarbital treatment did not prevent the development of postlesion seizures in all animals, animals were grouped for data analysis according to whether or not the animals exhibited seizures, regardless of whether or not they received phenobarbital. Relationship between the Pattern of GFAP Induction and Postlesion Seizures In evaluating the relationship between GFAP induction and seizures it is especially important to consider

To evaluate the extent of GFAP induction in the denervated zones in the DG ipsilateral to the lesion, quantitative analyses were carried out on seven animals (four animals with EC lesions alone, three animals with EC lesions that had been treated with phenobarbital). Of these animals, two exhibited no seizures; the remaining five animals exhibited two or more seizures (ranging from 2 to 14). As illustrated in Fig. 5A, the extent of the increase in GFAP mRNA levels in the denervated zone was comparable in animals that exhibited no seizures (mean OD level 5 0.861 6 0.016) and in animals that exhibited numerous seizures (mean OD level 5 0.881 6 0.083). It should be noted that one of the animals (with seizures) used for analysis had a limited number of sampling areas in the DG where the density levels did not exceed the linear range of the film. Although this animal was included for the analysis, the mean for this group was also calculated excluding this animal (mean OD level 5 0.801 6 0.024). The inclusion or exclusion of this animal does not substantially affect the mean for this group. In both groups, GFAP mRNA levels were about fourfold higher than the control values (mean OD level in control animals 5 0.214 6 0.003). Seizures and Levels of GFAP mRNA in Nondenervated Zones The enigmatic individual differences in the pattern of GFAP induction occur in nondenervated zones in the hippocampus ipsilateral to the lesion (for example,

NEURONAL ACTIVITY AND GFAP mRNA EXPRESSION

stratum radiatum) and in the hippocampus and DG on the contralateral side. Thus, we were especially interested in evaluating whether the individual differences in GFAP induction in these zones were related to the pattern of postlesion seizures. Figure 5B illustrates a comparison of the levels of GFAP mRNA in nondenervated zones in the ipsilateral hippocampus in animals that exhibited seizures on the ipsilateral side (n 5 5) and animals that did not (n 5 2). GFAP mRNA levels were clearly greater in the nondenervated zones in animals that exhibited at least one seizure on the ipsilateral side (mean OD level 5 0.606 6 0.056) in comparison to animals that did not exhibit seizures (mean OD level 5 0.331 6 0.002; P 5 0.036). The mean OD level in the hippocampus of control animals was 0.207 6 0.001. However, there was no obvious relationship between the extent of GFAP induction and the timing and number of seizures, as was the case for the GFAP induction in the contralateral hippocampus and DG (discussed below). Figure 6 illustrates the levels of GFAP mRNA in the DG (Fig. 6A) and hippocampus (Fig. 6B) contralateral to the lesion as a function of whether or not seizures occurred on the contralateral side. A total of nine animals (four animals with lesions alone, five animals with lesions and phenobarbital) was used for this analysis. As noted above, two of these animals were excluded from the previous analysis because the lesions encroached upon the ipsilateral DG. Again, GFAP mRNA levels were substantially higher in animals that exhibited seizures on the contralateral side than in animals that did not exhibit seizures, although these differences were not significant (P 5 0.070 for DG, and 0.097 for hippocampus). Closer inspection of the data revealed, however, that the extent of the increases in GFAP mRNA in individual animals appeared to be closely related to both the number and timing of the seizures. Figure 7 directly compares GFAP mRNA levels in the DG on the side contralateral to the lesion and the seizure profiles in individual animals. An animal that exhibited five seizures beginning 11 h before euthanasia (E1) exhibited higher levels of GFAP mRNA than an animal that exhibited three seizures beginning 7 h before euthanasia (E4). An animal that exhibited a single seizure that occurred 3 h before sacrifice (E5, refer to accompanying paper for seizure time course) exhibited a level of GFAP mRNA on the contralateral side that was comparable to the levels observed in control animals. These results are understandable based on the fact that the increases in GFAP mRNA levels develop over a period of hours and do not reach a peak until 24–48 h following a seizure (32). Thus, in the animals in which seizures occurred only a few hours prior to sacrifice, there would not have been sufficient time for GFAP mRNA to be upregulated.

89

Levels of GFAP mRNA in the contralateral hippocampus also appeared to be directly related to the timing and number of seizures on the side contralateral to the lesion (data not shown). Levels of GFAP mRNA were not as high in the hippocampus as they were in the DG, but the same trend in GFAP mRNA levels existed. Animals that exhibited greater numbers of seizures, and seizures that began earlier in the recording period, exhibited higher levels of GFAP mRNA than animals that exhibited fewer seizures or seizures that occurred closer to the time of euthanasia. Interestingly, the animal that received phenobarbital and exhibited multiple seizures (P2) showed extensive clustering of its seizures (see Fig. 3A). This animal also exhibited a slightly lower level of GFAP mRNA expression than did animals that experienced multiple seizures that were not clustered (i.e., E1). Although this difference could have been due to the fact that this particular animal was treated with phenobarbital, it also could have been due to the close spacing of the seizures. DISCUSSION

This study was undertaken to evaluate whether physiological events that occur at the time of an experimental EC lesion or during the early postlesion period played a role in upregulating GFAP expression in the DG and hippocampus. The results revealed that induction of GFAP mRNA in the denervated zones of the DG is not regulated by lesion-induced changes in neuronal activity. The extent of GFAP induction in the DG appeared to be comparable regardless of the occurrence of episodes of SD at the time of the lesion or seizures during the early postlesion period. These data are consistent with the hypothesis that very early changes associated with degeneration are the primary signal that leads to the astrocyte response in the denervated zones. In contrast, the increases in GFAP mRNA that occur in nondenervated zones in some, but not all animals were found to be related to physiological events, specifically the seizures that occur during the early postlesion period. In the following discussion, we consider some limitations that constrain our interpretations and then the implications of the present results for understanding injury-induced alterations in cellular gene expression. Regulation of GFAP mRNA Expression in the Nondenervated Zones The upregulation of GFAP mRNA levels in nondenervated zones in the ipsilateral hippocampus as well as the contralateral hippocampus and DG were greatest in animals that exhibited multiple seizures and seizures that began early in the postlesion period. This is

90

KELLEY AND STEWARD

FIG. 4. Increases in GFAP mRNA expression in animals that have received entorhinal cortex lesions, as assessed with in situ hybridization. The animal in (A) [referenced as R1 in this paper and in the accompanying paper] was a control animal, that was used for EEG recording but did not receive an entorhinal cortex lesion. Levels of GFAP mRNA expression are low throughout the section. The animal in (B) [P3] received a lesion, but did not exhibit any seizures. Despite the absence of seizures, levels of GFAP mRNA are high in the dentate gyrus ipsilateral to the lesion. The animal in (C) [E3] received a lesion and exhibited multiple ipsilateral seizures, but no contralateral seizures. In this animal, levels of GFAP mRNA appeared to increase only in the ipsilateral hippocampus and dentate gyrus. The animal in (D) [E5] received a lesion and exhibited multiple ipsilateral seizures, but only a single contralateral seizure late in the recording period. It appears that a single

NEURONAL ACTIVITY AND GFAP mRNA EXPRESSION

FIG. 5. Mean increases in GFAP mRNA expression in the ipsilateral dentate gyrus (A) and hippocampus (B) after entorhinal cortex lesions. Levels of GFAP mRNA were assessed from X-ray films apposed to hybridized sections. (A) Levels of GFAP mRNA were not significantly different in the ipsilateral dentate gyrus of animals with ipsilateral seizures than in animals without ipsilateral seizures (P 5 0.889). (B) However, levels of GFAP mRNA in the ipsilateral hippocampus were significantly different from levels in animals without ipsilateral seizures (P 5 0.036). No relationship appeared to exist between the number and timing of ipsilateral seizures and the increases in GFAP mRNA in the ipsilateral hippocampus (values for individual animals not shown). rec ctl, recording control animals; sz, seizures. Error bars indicate SEM; mean values were determined by averaging levels of GFAP mRNA in a particular region in each group (with seizures vs without seizures). The number at the bottom of each bar indicates the number of animals in each treatment group. (*P , 0.05)

consistent with previous studies which have shown that multiple seizures upregulate levels of GFAP mRNA to a greater extent than single seizures (32). In addition, the time of seizure onset also appeared to be directly related to the extent of the increases in GFAP mRNA expression. This is consistent with the fact that it takes some time after a seizure for GFAP mRNA levels to increase. Thus, if seizures do not begin until a few hours before euthanasia, there is insufficient time for GFAP mRNA levels to rise. The spatial pattern of GFAP mRNA upregulation was also consistent with the pattern of seizures. Animals that did not exhibit seizures on the contralateral side exhibited increases in GFAP mRNA expression on the ipsilateral side only. However, it is important to recall that these conclusions are based on recordings

91

FIG. 6. Mean increases in GFAP mRNA expression in the contralateral dentate gyrus (A) and hippocampus (B) after entorhinal cortex lesions. Levels of GFAP mRNA were assessed from X-ray films apposed to hybridized sections. Levels of GFAP mRNA were substantially lower in animals without seizures than levels in animals with seizures, although these differences were not significant (P 5 0.070 for the dentate gyrus, 0.097 for the hippocampus). rec ctl, recording control animals; sz, seizures. Error bars indicate SEM; mean values were determined by averaging levels of GFAP mRNA in a particular region in each group (with seizures vs without seizures). The number at the bottom of each bar indicates the number of animals in each treatment group.

made via cortical screw electrodes, not via electrodes implanted in the structures of interest (the hippocampus and DG). It is possible that some of the seizure activity reflected events in the cortex rather than in the hippocampus or that there were spatially restricted hippocampal seizures that were not detected. It is not possible to resolve these concerns because the placement of recording electrodes into the hippocampus appeared to dramatically increase seizure frequency. However, the close relationship between the pattern of GFAP mRNA induction in the hippocampus and the occurrence of activity lends support to the proposed relationship between the two. The present results provide a plausible explanation for the individual differences in the extent of GFAP mRNA induction in the hippocampus contralateral to an EC lesion—a phenomenon that until now has been enigmatic (30). In particular, the variable increases in GFAP mRNA levels on the side contralateral to the lesion are explainable by the variable occurrence of

contralateral seizure was not sufficient to increase levels of GFAP mRNA in the contralateral hippocampus and dentate gyrus. Last, the animal in (E) [E1] received a lesion and exhibited multiple ipsilateral and contralateral seizures. The levels of GFAP mRNA are high in both the ipsilateral and contralateral hippocampi and dentate gyri. The presence of low levels of GFAP mRNA in the contralateral hippocampi and dentate gyri of animals (B), (C), and (D) suggests that seizures may be involved in the upregulation of GFAP mRNA in the contralateral hippocampus and dentate gyrus. However, levels of GFAP mRNA are still high in the ipsilateral dentate gyrus of animal (B), even though the animal did not exhibit any ipsilateral seizures. These data suggest that seizures are not required for the upregulation of GFAP mRNA in this region. ECL, entorhinal cortex lesion. The filled arrow indicates the hippocampus and dentate gyrus ipsilateral to the lesion, the open arrow indicates the hippocampus and dentate gyrus contralateral to the lesion. Scale bar indicates 5 mm.

92

KELLEY AND STEWARD

FIG. 7. Increases in GFAP mRNA expression in the contralateral dentate gyrus 24 h after entorhinal cortex lesions. Levels of GFAP mRNA were assessed from X-ray films apposed to hybridized sections. Levels of GFAP mRNA appeared to be related to both the number of contralateral seizures, as well as to the onset of seizure activity. The number at the top of each bar indicates the total number of contralateral seizures exhibited by each animal. The number at the bottom of each bar indicates the time of seizure onset in number of hours before sacrifice. Animals that exhibited greater numbers of seizures and seizures beginning earlier in the recording period exhibited higher levels of GFAP mRNA than animals with fewer seizures that started later in the recording period. In addition, animals that did not exhibit contralateral seizures did not exhibit increases in contralateral GFAP mRNA. Individual bars labeled with an E represent animals that received an entorhinal cortex lesion only; bars labeled with a P represent animals that received an entorhinal cortex lesion plus phenobarbital; bars labeled with an R represent recording control animals. Error bars represent SEM; mean values were obtained by averaging values from a single area of tissue over several sections from the same animal.

seizure activity on the contralateral side. It is important to note, however, that there may be other factors that influence whether or not GFAP mRNA will be induced on the contralateral side after unilateral EC lesions. For example, Laping et al. (17) have demonstrated that the probability of induction of GFAP mRNA on the contralateral side is determined in part by circulating levels of corticosterone. In their study, both ipsilateral and contralateral increases in GFAP mRNA were observed after EC lesions. Contralateral increases were smaller than ipsilateral increases, unless the animals were adrenalectomized (ADX) prior to the EC lesions in which case GFAP mRNA was strongly upregulated on the side contralateral to the lesion. When ADX/EC lesion animals were supplemented with corticosterone, there was no upregulation of GFAP mRNA on the contralateral side. These data suggest that larger contralateral increases in GFAP mRNA are more likely to occur in animals with low levels of circulating corticosterone, and smaller contralateral increases in GFAP mRNA are more likely to occur in animals with normal levels of corticosterone. One interpretation of these results is that there is a relationship between postlesion seizures and circulat-

ing corticosterone. In this regard, there are two possible interpretations of the combined results: (1) that seizures regulate GFAP mRNA levels by decreasing levels of circulating corticosterone or (2) that low levels of corticosterone increase the probability of postlesion seizures. However, data from several other studies are not consistent with either interpretation. Young et al. (33) have shown that electroconvulsive seizures (ECS) lead to increased levels of plasma corticosterone. ECS also increases levels of GFAP mRNA in the DG (29). Together, these studies suggest that high levels of corticosterone would be associated with high levels of GFAP mRNA expression. Furthermore, both Cottrell et al. (4) and Lee et al. (19) have demonstrated that animals that have been adrenalectomized prior to seizure induction (with either kainic acid or kindling) exhibit shorter periods of afterdischarge (4) and fewer seizures (19) than do non-ADX animals. In these studies, low levels of corticosterone were associated with decreased seizure severity. These data suggest that in the Laping et al. (17) study, the group of ADX/EC lesion animals should have exhibited less seizure activity than the animals with EC lesions alone. Yet ADX animals exhibited bilateral increases in GFAP mRNA levels. It is important to note that no assessments of seizure activity were made in the Laping et al. (17) study, so it is not known if seizures produced by EC lesions were attenuated by ADX. However the combined results do suggest that circulating corticosterone and seizures operate independently to upregulate GFAP mRNA expression. At present, the intercellular signalling mechanisms involved in the seizure-induced upregulation of GFAP mRNA expression are not known. Possible mechanisms include release of ions, neurotransmitters, and growth factors. In terms of ions, an obvious candidate signalling mechanism is extracellular K1 which increases dramatically following seizures (21). Seizures also lead to a massive release of neurotransmitters for which astrocytes have receptors (24). Finally, seizures cause an upregulation of expression of several peptides and growth factors, any of which could trigger GFAP expression in astrocytes (see Introduction). The present data do not provide any new insights that would help to distinguish between these possibilities. Spreading Depression and GFAP mRNA Expression The results from the first part of the present study indicate that although SD can occur in association with EC lesions, SD does not appear to be involved in postlesion increases in GFAP mRNA expression. Animals that received EC lesions, but did not exhibit SD showed increases in GFAP mRNA expression comparable to those that occurred in animals that did exhibit SD. Although these data indicate that SD does not

NEURONAL ACTIVITY AND GFAP mRNA EXPRESSION

appear to account for individual differences in GFAP mRNA induction, the results do not exclude the possibility that lesion-induced SD alters the expression of other genes in the hippocampus and DG.

9.

10.

Postlesion Seizures and Gene Expression In broader terms, the present study provides strong support for the possibility that seizures associated with brain injury can produce dramatic changes in gene expression outside the area of direct damage. These seizures may alter neuronal, as well as glial gene expression. If posttraumatic seizures are capable of altering the expression of many different genes, these events could contribute to either the severity of posttraumatic brain dysfunction and the spatial extent of brain damage or could enhance repair processes. In particular, it is thought that increases in the expression of various neurotrophic substances may support long term modifications of cell metabolism and function, which could improve recovery of function following brain damage. In future studies, it will be important to determine the incidence of posttraumatic seizures in different models of brain injury and evaluate the effects of these seizures on neuronal and glial gene expression.

11.

12.

13.

14.

15.

16.

ACKNOWLEDGMENTS 17. This work was supported by NIH Grant NS29875. The authors thank Dr. Dona Chikaraishi for her generous gift of the GFAP cDNA.

REFERENCES 1.

2.

3.

4.

5.

6.

7.

8.

BONTHIUS, D. J., AND O. STEWARD. 1993. Induction of cortical spreading depression with potassium chloride upregulates levels of messenger RNA for glial fibrillary acidic protein in cortex and hippocampus: Inhibition by MK-801. Brain Res. 618: 83–94. CHOMCZYNSKI, P., AND N. SACCHI. 1987. Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol chloroform extraction. Anal. Biochem. 162: 156–159. CONNER, J. M., B. FASS-HOLMES, AND S. VARON. 1994. Changes in nerve growth factor immunoreactivity following entorhinal cortex lesions: Possible molecular mechanism regulating cholinergic sprouting. J. Comp. Neurol. 345: 409–418. COTTRELL, G. A., C. NYAKAS, E. R. DEKLOET, AND B. BOHUS. 1984. Hippocampal kindling: Corticosterone modulation of induced seizures. Brain Res. 309: 377–381. GAGE, F. H., P. OLEJNICZAK, AND D. M. ARMSTRONG. 1988. Astrocytes are important for sprouting in the septohippocampal circuit. Exp. Neurol. 102: 2–13. GALL, C. M., AND P. J. ISACKSON. 1989. Limbic seizures increase neuronal production of messenger RNA for nerve-growth factor. Science 245: 758–761. GALL, C. M., R. BERSCHAUER, AND P. J. ISACKSON. 1994. Seizures increase basic fibroblast growth factor mRNA in adult rat forebrain neurons and glia. Mol. Brain Res. 21: 190–205. GOMEZ-PINILLA, F., J. W.-K. LEE, AND C. COTMAN. 1992. Basic FGF in adult rat brain: cellular distribution and response to entorhinal lesion and fimbria-fornix transection. J. Neurosci. 12: 345–355.

18.

19.

20.

21.

22. 23.

24. 25.

26.

27.

93

GUTHRIE, K. M., T. NGUYEN, AND C. M. GALL. 1995. Insulin-like growth factor-1 mRNA is increased in deafferented hippocampus: Spatiotemporal correlation of a trophic event with axon sprouting. J. Comp. Neurol. 352: 147–160. GWAG, B. J., F. SESSLER, K. KIMMERER, AND J. E. SPRINGER. 1994. Neurotrophic factor mRNA expression in dentate gyrus is increased following angular bundle transection. Brain Res. 647: 23–29. HANSEN, A., O. S. JORGENSEN, T. G. BOLWIG, AND D. I. BARRY. 1991. Hippocampal kindling in the rat is associated with time-dependent increases in the concentration of glial fibrillary acidic protein. J. Neurochem. 57: 1716–1720. HERDEGEN, T., J. SANDKUHLER, P. GASS, M. KIESSLING, R. BRAVO, AND M. ZIMMERMAN. 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: 271–288. HERRERA, D. G., AND H. A. ROBERTSON. 1990. Application of potassium chloride to the brain surface induces the c-fos protooncogene: Reversal by MK-801. Brain Res. 510: 166–190. HERRERA, D. G., D. MAYSINGER, R. GADIENT, C. BOECKH, U. OTTEN, AND A. C. CUELLO. 1993. Spreading depression induced c-fos-like immunoreactivity and NGF mRNA in the rat cerebral cortex. Brain Res. 602: 99–103. ISACKSON, P. J., M. M. HUNTSMAN, K. D. MURRAY, AND C. M. GALL. 1991. BDNF mRNA expression is increased in adult rat forebrain after limbic seizures: Temporal patterns of induction distinct from NGF. Neuron 6: 937–948. KRAIG, R. P., L. DONG, R. THISTED, AND C. B. JAEGER. 1991. Spreading depression increases immunohistochemical staining of glial fibrillary acidic protein. J. Neurosci. 11: 2187–2198. LAPING, N. J., N. R. NICHOLS, J. R. DAY, AND C. E. FINCH. 1991. Corticosterone differentially regulates the bilateral response of astrocyte mRNAs in the hippocampus to entorhinal cortex lesions in male rats. Mol. Brain Res. 10: 291–297. LAUTERBORN, J. C., P. J. ISACKSON, AND C. M. GALL. 1994. Seizure-induced increases in NGF mRNA exhibit different time courses across forebrain regions and are biphasic in hippocampus. Exp. Neurol. 125: 22–40. LEE, P. H. K., L. GRIMES, AND J. S. HONG. 1989. Glucocorticoids potentiate kainic acid-induced seizures and wet dog shakes. Brain Res. 480: 322–325. LOESCHE, J., AND O. STEWARD. 1977. Behavioral correlates of denervation and reinnervation of the hippocampal formation of the rat: Recovery of alternation performance following unilateral entorhinal cortex lesions. Brain Res. Bull. 2: 31–39. LOTHMAN, E. W., AND G. G. SOMJEN. 1976. Functions of primary afferents and responses of extracellular K1 during spinal epileptiform seizures. Electroencephalogr. Clin. Neurophysiol. 41: 253–267. MACDONALD, R. L., AND M. J. MCLEAN. 1986. Anticonvulsant drugs: Mechanisms of action. Adv. Neurol. 44: 713–736. MORGAN, J. I., D. R. COHEN, J. L. HEMPSTEAD, AND T. CURRAN. 1987. Mapping patterns of c-fos expression in the central nervous system after seizure. Science 237: 192–197. MURPHY, S., AND B. PEARCE. 1987. Functional receptors for neurotransmitters on astroglial cells. Neuroscience 22: 381–394. PHILLIPS, L. L., AND E. T. BELARDO. 1994. Increase of c-fos and ras oncoproteins in the denervated neuropil of the rat dentate gyrus. Neuroscience 58: 503–514. ROSE, G., G. LYNCH, AND C. W. COTMAN. 1976. Hypertrophy and redistribution of astrocytes in the deafferented dentate gyrus. Brain Res. Bull. 1: 87–92. SAFFEN, D. W., A. J. COLE, P. F. WORLEY, B. A. CHRISTY, K. RYDER,

94

KELLEY AND STEWARD

J. M. BARABAN. 1988. Convulsant-induced increase in transcription factor messenger RNAs in rat brain. Proc. Natl. Acad. Sci. USA 85: 7795–7799. 28. STEWARD, O. 1986. Lesion-induced synapse growth in the hippocampus: In search of cellular and molecular mechanisms. In The Hippocampus (R. L. Isaacson and K. H. Pribram, Eds.), Vol. 3, pp. 65–111. Plenum Publishing, New York. 29. STEWARD, O. 1994. Electroconvulsive seizures upregulate astroglial gene expression selectively in the dentate gyrus. Mol. Brain Res. 25: 217–224. 30. STEWARD, O., M. S. KELLEY, AND E. R. TORRE. 1993. The process of reinnervation in the dentate gyrus of adult rats: Temporal relationship between changes in the levels of glial fibrillary AND

acidic protein (GFAP) and GFAP mRNA in reactive astrocytes. Exp. Neurol. 124: 167–183. 31. STEWARD, O., E. R. TORRE, L. L. PHILLIPS, AND P. A. TRIMMER. 1990. The process of reinnervation in the dentate gyrus of adult rats: Time course of increases in mRNA for glial fibrillary acidic protein. J. Neurosci. 10: 2373–2384. 32. STEWARD, O., E. R. TORRE, R. TOMASULO, AND E. LOTHMAN. 1991. Neuronal activity up-regulates astroglial gene expression. Proc. Natl. Acad. Sci. USA 88: 6819–6823. 33. YOUNG, E. A., R. L. SPENCER, AND B. S. MCEWEN. 1990. Changes at multiple levels of the hypothalamo-pituitary adrenal axis following repeated electrically induced seizures. Psychoneuroendocrinology 15: 165–172.