Address correspondence and reprint requests to Amy Brooks-Kayal, Division of Neurology, Abramson Pediatric Research Center, Rm 502, 3615 Civic Center Blvd., Philadelphia, PA 19104, USA. E-mail: firstname.lastname@example.org
Prolonged seizures in early childhood are associated with an increased risk of development of epilepsy in later life. The mechanism(s) behind this susceptibility to later development of epilepsy is unclear. Increased synaptic activity during development has been shown to permanently alter excitatory neurotransmission and could be one of the mechanisms involved in this increased susceptibility to the development of epilepsy. In the present study we determine the effect of status-epilepticus induced by lithium/pilocarpine at postnatal day 10 (P10 SE) on the expression of glutamate receptor and transporter mRNAs in hippocampal dentate granule cells and protein levels in dentate gyrus of these animals in adulthood. The results revealed a decrease in glutamate receptor 2 (GluR2) mRNA expression and protein levels as well as an increase in protein levels for the excitatory amino acid carrier 1 (EAAC1) in P10 SE rats compared to controls. Expression of glutamate receptor 1 (GluR1) mRNA was decreased in both P10 SE rats and identically handled, lithium-injected littermate controls compared to naive animals, and GluR1 protein levels were significantly lower in lithium-controls than in naive rats, suggesting an effect of either the handling or the lithium on GluR1 expression. These changes in EAA receptors and transporters were accompanied by an increased susceptibility to kainic acid induced seizures in P10 SE rats compared to controls. The current data suggest that early-life status-epilepticus can result in permanent alterations in glutamate receptor and transporter gene expression, which may contribute to a lower seizure threshold.
Seizures are more frequent during early childhood than at any other time during life. Animal studies have demonstrated that the immature brain is particularly vulnerable to seizures (Moshe et al. 1983) requiring lower doses of chemo-convulsants to produce status-epilepticus (SE) (Albala et al. 1984; Cavalheiro et al. 1987; Hirsch et al. 1992). Although the long-term consequences of brief seizures in childhood remains uncertain, several epidemiological studies suggest that prolonged seizures or SE during childhood may increase the risk of later development of epilepsy (Annegers et al. 1979; French et al. 1993). The mechanism(s) through which early childhood SE increases the risk of seizures later in life is largely unknown. In animal models, numerous studies have demonstrated that prolonged seizures result in significantly less or no structural damage in the immature brain when compared to the adult brain (Holmes et al. 1988; Stafstrom et al. 1993; Sperber 1996; Sankar et al. 1998; Yang et al. 1998; Raol et al. 2003). This suggests that subcellular or molecular changes, such as alterations in neurotransmitter receptor or transporter expression and/or function may contribute to early-life epileptogenesis.
The subunit expression and composition of glutamate receptors are developmentally regulated (Insel et al. 1990; McDonald et al. 1990; reviewed in Raol et al. 2001). It has been suggested that changes in excitatory synaptic activity during development, caused by either modification in sensory inputs or seizures, can lead to long-term alterations in excitatory neurotransmission. For example, expression of the NR2A subunit of the NMDA receptor is higher in the visual cortices of animals that had visual experience compared to dark-reared animals (Quinlan et al. 1999). Studies in various animal models of developmental seizures have also shown alterations in glutamate receptors (Sanchez et al. 2001) and voltage-gated channels (Chen et al. 2001; Brewster et al. 2002) accompanied by increased seizure susceptibility into adulthood (Jensen et al. 1992; Dube et al. 2000).
We hypothesized that abnormal electrical activity associated with SE at postnatal day 10 (P10) in rats could lead to long-term changes in the expression of glutamate receptors and transporters. To test this hypothesis, we subjected P10 rats to lithium/pilocarpine-induced SE. Three months later, expression of mRNAs for multiple glutamate receptor and transporter genes was assessed in single hippocampal dentate granule cells using antisense-RNA amplification and reverse northern blotting and protein levels were assessed in whole dentate area by western blotting. To determine potential functional consequences associated with observed molecular changes, we monitored animals for development of spontaneous seizures and examined seizure susceptibility to kainic acid.
Materials and methods
Sprague–Dawley rats obtained from Charles-River (Kingston, PA, USA) were used for the study. The Institutional Animal Care and Use Committee of the Children's Hospital of Philadelphia approved all the protocols used in this study. At P9, male rat pups were briefly separated from their mother and injected with either lithium chloride (3 mEq/kg, i. p., Sigma, St Louis, MO, USA) or saline (saline control). On P10, pups were again separated from their mother and received either i.p. injection of 60 mg/kg pilocarpine (pilocarpine treated group, Sigma) or saline (lithium and saline control groups). The pilocarpine-treated pups uniformly developed prolonged SE within 15–30 min of pilocarpine injection and seizures recurred intermittently over a period of 5–6 h. The behavioral seizures included almost continuous head shaking and tail stiffening without a return to baseline behavior between events. All three groups of pups, pilocarpine-, lithium- and saline-treated remained separated from their mother for identical periods of time until seizures had resolved in all the pups. All rats were monitored behaviorally for the appearance of seizures. In addition, one pilocarpine-treated rat pup from each litter was implanted with bipolar electrodes (Plastic One, Roanoke, UA, USA) in CA1 area of hippocampus to confirm the appearance of electrographic seizures. The following stereotaxic coordinates were used: AP, −3.1 mm from bregma, ML, 2.0 mm; DV, 3.0 mm from the skull surface (Paxinos and Watson 1986). A separate set of control pups (naive control) was not separated from their mother until weaning. All the pups were weaned at P21. Rats were weighed at weaning and in adulthood and no difference in the average weight was seen between the different groups.
Isolation of neurons
Once the animals were adults (> 90 days), neurons were acutely isolated using a modification of published protocols (Brooks-Kayal et al. 2001). The rats were anaesthetized with halothane (Sigma) and decapitated using guillotine. Brains were dissected in chilled, oxygenated (100%) HEPES medium containing (in mm(NaCl, 155; KCl, 3; MgCl2, 1; CaCl2, 3; glucose, 25 and HEPES, 10; pH value adjusted to 7.35 with NaOH). Longitudinal hippocampal slices 500-µm in thickness were cut on a tissue chopper and the dentate gyrus was identified, dissected and cubed under an anatomical microscope. Tissue cubes then were enzymatically digested at 30°C for 30–45 min under a oxygen-infused PIPES-buffered saline containing (in mM): NaCl, 120; KCl, 5; MgCl2, 1; CaCl2, 1; d-glucose, 25; PIPES, 20 and protease XXIII (Sigma), 3 mg/mL (pH value was adjusted to 7.2 with NaOH). After digestion, tissue cubes were rinsed thoroughly and incubated in enzyme-free medium for 1 h before use. Neurons were isolated by trituration in 1 mL of oxygenated HEPES buffer using a series of descending diameter of fire-polished glass pipette. The cell suspension was then plated onto a 60-mm Petri dish, and the cells were allowed to settle 10–15 min before the buffer was replaced with extracellular buffer. A total of 79 neurons isolated randomly from whole dentate gyrus were amplified for the study: 24 neurons for naive control (8 rats); 17 neurons for lithium control (7 rats), 18 neurons for saline control (5 rats) and 20 neurons for pilocarpine treated group (8 rats).
Relative expression of mRNAs within individual acutely isolated dentate granule cells (DGCs) were measured using the technique of single-cell antisense RNA (aRNA) amplification according to a protocol published previously (Brooks-Kayal et al. 1998, 2001). DGCs were identified based on their structural morphology (≈ 10 µm in diameter, round or oval cell body with a single, stubby process) and were aspirated in their entirety into the micropipette. Samples of extracellular contents were also aspirated, and processed in parallel with cellular aspirates to assess for potential mRNA contamination of medium from dying cells. The contents of each microelectrode were expelled into a microcentrifuge tube and first strand cDNA synthesis was performed using avian myeloblastosis virus reverse transcriptase (AMVRT; Seikagaku America, Ijamsville, MD, USA) and a T7 oligonucleotide primer that selectively primes only mRNA for reverse transcription. Double stranded DNA was made by incubation with dNTPs, T4 DNA polymerase and the Klenow fragment of DNA Polymerase I. The single stranded hairpin loop was removed with S1 nuclease, the ends of the double-stranded template blunted with T4 DNA polymerase and the Klenow fragment of DNA polymerase I for 2 h, then the cDNA used for synthesis of amplified aRNA using NTPs and T7 RNA polymerase. aRNA was then again synthesized into a single-stranded cDNA template for a second round of amplification. The final aRNA synthesis includes 25 pmol of [32P]CTP[αP].
Slotblot preparation and hybridization
NMDA receptor subunit cDNAs (NR1, NR2A−2D), α-amino-3-hydroxy-5-methylisoxazole-4-propionate (AMPA) receptor subunit cDNAs (GluR1–4), kainic acid (KA) receptor subunit cDNAs (GluR5–7, KA2), glutamate transporter cDNAs (EAAC1, GLT1), β-actin (internal reference), glial fibrillary acidic protein (GFAP, control for glial contamination), neurofilament L (NF-L, marker for neuronal phenotype) and pBluescript plasmid (background) cDNAs were included on each slot blot. NMDA, AMPA and KA receptor subunits clones were provided by Drs D. Lynch (University of Pennsylvania, Philadelphia, PA, USA) and S. Heinemann (Salk Institute, La Jolla, CA, USA). EAAC1 cDNA was provided by Dr M. Hediger (Harvard Medical School, Boston, MA, USA), GLT1 cDNA was provided by Dr B. Kanner (Hebrew University, Jerusalem, Israel). Drs J. Eberwine and V. Lee at the University of Pennsylvania provided GFAP and NF-L clones, respectively. The value for the slot containing plasmid cDNA (pBluescript) was considered as background; NF-L served as a marker for neuronal phenotype; GFAP acted as a control for glial contamination and β-actin acted as an internal reference value. Each blot was prehybridized for 12 h at 42°C in 5 mL of prehybridization solution (50% formamide, 5 × saline sodium citrate buffer (SSC; pH 7.0), 5 × Denhardt's solution, 0.1% sodium dodecyl sulfate (SDS), 1 mm sodium pyrophosphate, and 100 µg/mL salmon sperm DNA), and then hybridized with the radiolabeled aRNA probe from an individual cell for 60 h (42°C). The blots were washed to a final concentration of 0.2 × SSC at 42°C for 30 min, then directly exposed for 4 h to a Molecular Dynamics Phosphor-Image screen with a linear dynamic range over five orders of magnitude. All hybridization signals fell well within this dynamic range.
The intensity of the autoradiographic signal was measured by three-dimensional laser scanning densitometry utilizing image-quant software from Molecular Dynamics (Sunnyvale, CA, USA). For each blot, the relative abundance of each subunit mRNA was calculated as the hybridization signal for that subunit cDNA divided by the hybridization signal for β-actin cDNA on the blot for that cell. Presence of a subunit mRNA was defined as hybridization signal above background by greater than or equal to 1% of the β-actin cDNA signal on the blot. This value was selected because it represents ± 1 SD unit of the estimated variability in background noise (based on differences in hybridization signal for bluescript plasmid cDNA and GFAP cDNA).
Western blot analysis
The dentate gyrus was dissected from naive control (n = 3), lithium control (n = 3) and pilocarpine-treated rats (n = 2) and sonicated in 50 mm Tris/HCl, 140 mm NaCl, 10 mm EDTA, 2% SDS, 0.5 mm, 10 mm NaF, 30 mm sodium pyrophosphate, 50 mmβ-glycerophosphate and the following protease inhibitors: 0.2 mm phenylmethylsulfonyl fluoride (PMSF), 5 mm 3-isobutyl-1-methylxanthine (IBMX), 1 mm sodium orthovandate, and a 1 : 1000 dilution of protease inhibitor cocktail (Sigma, St. Louis, MO, USA). Protein concentrations of samples were analyzed using the Bio-Rad RC/DC reagent kit (Bio-Rad Laboratories, Hercules, CA, USA). Identical quantities of protein homogenate from each animal were loaded onto 10% SDS/polyacrylamide gels and run for 1 h at 100 V then transferred to nitrocellulose by electroblotting (100 V, 1 h). Protein (70 µg) was loaded for GluR 1, GluR 5, and EAAC1 gels and 50 µg was loaded for the GluR 2 gels. Nitrocellulose membranes were blocked in 5% non-fat dry milk made with 0.05% Tween 20 in Tris-buffered saline (pH 7.4) for 2 h at room temperature. The blots were then incubated with primary antibodies diluted in 5% non-fat dry milk overnight at 4°C followed by incubation with secondary antibody for 2 h at room temperature. Protein bands were visualized using Super Signal West Pico chemiluminiscent substrate kit (Pierce, Rockford, IL, USA). Primary antibodies were used as follows: rabbit anti-GluR1 polyclonal Ig, 1 µg/mL dilution (Chemicon International, Temecula, CA, USA); rabbit anti-GluR2 polyclonal Ig, 0.5 µg/mL dilution (Chemicon International); rabbit anti-EAAC1 polyclonal Ig, 1 : 75 dilution (Chemicon International) and rabbit anti-GluR5 polyclonal Ig, 2 µg/mL (Upstate Biotechnology, Lake Placid, NY, USA). Anti-rabbit IgG conjugated with horseradish peroxidase (HRP) (Amersham, Buckinghamshire, UK) was used as the secondary antibody. Bands of the appropriate size were quantified using nih image software (National Institutes of Health, Bethesda, MD, USA). To assess for potential variability in loading, all the blots were secondarily reacted with mouse anti-neuronal nuclei (NeuN) polyclonal Ig, 1 : 100 dilution (Chemicon International) and anti-mouse IgG secondary Ig conjugated with HRP (Jackson ImmunoResearch Laboratories, West Grove, PA, USA). No differences in loading concentration were identified between samples, and the NeuN signal was not used for normalization.
Determination of seizure susceptibility and spontaneous seizures in adult rats
Once the animals were adults (> 90 days) a subset of pilocarpine-treated (n = 3) and lithium control animals (n = 3) were kept in separate coded cages and video recorded for 24 h/day for 14 days. The animals were monitored specifically for any discrete alterations in behavior such as wild-running, head-bobbing, facial-clonus, forelimb-clonus, or tail-stiffening that might indicate behavioral seizures. Pilocarpine-treated rats were then implanted with electrodes in the CA1 area of the dorsal hippocampus and frontal cortex for EEG recording. The coordinates were adopted from the rat brain atlas of Paxinos and Watson (1986). For CA1: anterior/posterior (AP), −3.0 mm from bregma; medial/lateral (ML), 2.0 mm and DV, 3.0 mm. To record from cortex, two stainless steel screws were placed over the frontal cortex (AP, 2.0 mm from bregma; ML, 4.00 mm). After a minimum of 3 days of recovery from surgical stress, rats were recorded both behaviorally and electrographically for 7 days (6 h/day) to monitor for behavioral (as defined above) or electrographic seizures. Electrographic seizures were defined as any repetitive sharp or spike discharges that were clearly different than the state-appropriate background activity, evolved over time, lasted for ≥ 8 s and were not associated with a scratching artefact. Inter-ictal discharges were defined as sharp or spike discharges (single or repetitive) that persisted < 8 s. These animals, along with lithium and naive controls, were processed for Timm's and cresyl violet staining.
A separate subset of lithium control (n = 3), naive control (n = 3) and pilocarpine-treated (n = 5) rats were implanted in CA1 and frontal cortex to examine the seizure threshold to KA. The rats were placed in a restrainer with continuous EEG monitoring (Nihon-Kohden, Shinjuku-ku, Tokyo, Japan) and KA (dissolved in sterile distilled water; Sigma) at a rate of 630 µg/min, was infused continuously into the lateral tail vein using a Harvard infusion pump (South Natick, MA, USA). The infusion was continued until the appearance of the first electrographic seizure in the hippocampus, defined as repetitive sharp or spike-wave discharges lasting for 8 s or more (see Results). All electrographic seizures began in hippocampus and spread rapidly to the cortex. All rats were monitored behaviorally until the appearance of a generalized seizure (stage V). The behavioral seizures included facial clonus, forelimb clonus and generalized tonic-clonic seizures. Most of the rats went immediately into stage V seizures at the end of infusion. The dose of KA required for each rat was calculated as the total dose required to induce a seizure divided by the animal's weight. The average weight of the rats at the time of KA injections did not differ between the groups.
Timmi's and cresyl violet staining
In adulthood (> 90 days), separate subsets of animals from lithium control (n = 3), naive control (n = 3), and pilocarpine-treated groups (n = 6) were perfused intracardially using a gravity fed system under deep anesthesia (ketamine-xylazine, 100 mg/kg body weight; Sigma). Rats that had previously received KA for the seizure susceptibility study were excluded from the histology studies. The perfusion procedure was similar to that described by Sloviter (1982) and involved the following: (i) saline (0.9% sodium chloride), 10 min; (ii) 0.37% sulfide solution, 12 min; (iii) 4.0% para-formaldehyde, 15 min. Brains remained in situ overnight, and were then stored in fixative for at least 1 day and then 100 mm phosphate buffer for 1 day before 40-µm thick vibrotome sections were taken (Vibrotome Company, St. Louis, MO, USA). The sections were mounted on slides and were stained with a modification of the Timm's procedure (Sloviter 1982) and counterstained with cresyl violet. The Timm's stained sections were scored visually by an observer blinded to the identity of the section, using the Tauck and Nadler (1985) scoring system. The presence of cell loss was assessed visually using a Nikon TE300 light microscope.
Twenty-eight rats were used for the mRNA analysis, and were assigned to four different groups (naive control, saline control, lithium control, and pilocarpine treatment). Relative expression of mRNAs was measured in 79 different cells from the 28 individual rats. The effect of group on the various outcomes as measured repeatedly across different cells was analyzed based on a longitudinal mixed effects approach, the SAS Proc Mixed models (SAS Institute Inc. 1999). The SAS Proc Mixed procedure fits a variety of mixed linear models to data and the data are permitted to exhibit correlation and non-constant variability, allowing modeling of the means of the data as well as their variances and covariances. In this study, using measurements from different cells from the 28 individual rats, the compound symmetry covariance matrix structure was assumed. Covariance parameters were needed because repeated measurements are taken from each animal, and these repeated measurements may be correlated or exhibit variability that changes. The estimated least-squares means and standard errors for the various outcome measures were calculated for each of the four groups. The least-squares means are predicted population margins; that is, they estimate the marginal means over a balanced population, and the estimated standard errors are adjusted for the covariance parameters in the model. For clarity of presentation, the least-squares means and standard errors of these outcome measures are based on nontransformed data; however, the reported p-values are p-values from models using logarithmic-transformed data.
For statistical comparison of the mean protein levels and seizure susceptibility between the three groups examined (pilocarpine-treated, lithium control and naive), one-way anova analysis was used followed by post hoc unpaired t-test when the anova showed a significant effect of group.
Alterations in mRNA expression of glutamate receptors and transporters
We examined the effect of prolonged early-life seizures on the expression of glutamate receptor and transporter mRNAs in hippocampal dentate granule cells (DGCs) when animals reached adulthood. The relative expression of mRNA for the GluR2 subunit of AMPA receptors (relative to β-actin) was significantly lower in DGCs from pilocarpine-treated rats compared to naive rats (0.189 ± 0.046 for pilocarpine vs. 0.364 ± 0.049 for naive; F3.82, p = 0.044 for effect of group; p = 0.016 for pairwise comparison) and showed a trend toward being lower than lithium control rats (0.306 ± 0.049; p = 0.090 for pairwise comparison; Fig. 1B). There was no difference in GluR2 expression between DGCs from naive and lithium control rats (p = 0.403 for pairwise comparison). Relative expression of the GluR1 subunit of AMPA receptors was lower in DGCs from both lithium control and pilocarpine treated rats compared to naïve animals (0.459 ± 0.057 for lithium control, 0.397 ± 0.055 for pilocarpine and 0.678 ± 0.058 for naive; F7.88, p = 0.004 for effect of group; p < 0.02 for pairwise comparisons; Fig. 1B). There was no statistical difference between GluR1 subunit expression in DGCs from lithium control and pilocarpine treated rats (p = 0.414 for pairwise comparison). Also, no statistical difference between any of the groups was found for GluR3 (F1.83, p = 0.193) and GluR4 (F0.54, p = 0.596) subunits.
Contrary to changes in AMPA receptors, there was no statistically significant effect of group on the relative expression of mRNA for any of the subunits of the KA receptors we studied (GluR5, GluR6, GluR7 and KA2; Fig. 1C). Note that for these subunits, four groups were analyzed. A saline-injected control group was included to control for possible effects of lithium injection. No effects attributable to lithium injection were seen in any of the analyses in which the saline group was included (the saline control group was not included in the AMPA receptor studies due to technical difficulties). There were also no significant differences in mRNA expression between the groups for any of the NMDA receptor subunits (data not shown).
Expression of the neuronal glutamate transporter, EAAC1, was also examined in the four groups (Fig. 1D). The mean relative expression of EAAC1 in DGCs from pilocarpine treated rats was over threefold higher than the mean in DGCs from either the lithium or saline controls and 1.5-fold higher than in DGCs from naive controls, however, this difference was not significant due to the high degree of variability within groups (0.571 ± 0.168 for pilocarpine, 0.140 ± 0.178 for lithium, 0.168 ± 0.199 for saline and 0.381 ± 0.164 for naive; F1.14, p = 0.351 for effects of group). There was no significant difference between the groups in relative expression of mRNA for the glutamate transporter GLT-1 (0.069 ± 0.027 for pilocarpine, 0.022 ± 0.028 for lithium, 0.003 ± 0.027 for saline and 0.014 ± 0.024 for naive; F1.07, p = 0.379 for effect of group; Fig. 1D).
Alterations in protein levels of glutamate receptors and transporters
To analyze whether protein levels of glutamate receptors and transporters were also changing following lithium/pilocarpine induced SE at P10, we performed western blot analysis on protein samples from microdissected dentate gyrus for GluR1, GluR2, GluR5 and EAAC1 in the 3 groups (Fig. 2A; saline controls were not included for protein studies). In parallel with the mRNA changes seen, GluR2 subunit protein levels differed significantly among the three groups (F10.97, p < 0.01 for effect of groups; anova). GluR2 subunit levels were lower in pilocarpine-treated rats than in either naive control rats or lithium control rats (p < 0.05 for pairwise comparisons; Fig. 2B). GluR2 protein levels did not differ between naive and lithium control groups. The protein levels of GluR1 were also significantly different among the groups (anova, F10.12; p < 0.04 for effects of group). GluR1 levels were significantly lower in the lithium control group compared to either naive or pilocarpine-treated groups (p < 0.05 for pairwise comparisons; Fig. 2B). GluR1 protein levels were not different between naive and pilocarpine treated groups. Similar to the mRNA expression, protein levels of GluR5 subunit were not significantly different between groups. Finally, in general agreement with the mRNA levels, EAAC1 levels were increased significantly in pilocarpine-treated rats compared to either naive control rats or lithium control rats (anova, F9.254, p < 0.02 for effect of group; p < 0.05 for pairwise comparisons). The levels of EAAC1 protein in dentate from lithium control rats were not significantly different from naive control rats.
Early-life seizures increase susceptibility to seizures in adulthood
To determine whether changes in glutamate receptors and transporters have any functional consequences, we determined seizure threshold to KA in rats from each of the three groups. There was increased seizure susceptibility to KA in the rats that had experienced SE at P10. The dosage for KA (mg/kg body weight) required to induce a seizure in adult rats who had experienced prolonged SE at P10 (7.96 ± 0.95, n = 5) was reduced to approximately 50% of control values (15.50 ± 2.75 for lithium controls and 14.72 ± 3.00 for naive controls, n = 3 rats/group; Fig. 3C). Despite the lowered seizure threshold, no spontaneous behavioral or electrographic seizures or interictal epileptiform discharges were observed in any of the rats during prolonged video EEG recording. Also, there was no mossy fiber sprouting or visible cell loss in CA1, CA3, hilus or dentate gyrus in any of the groups studied (Fig. 4).
The results of the current study demonstrate that pilocarpine-induced prolonged SE at P10 causes long-term changes in both glutamate receptors and transporters in hippocampal dentate gyrus. These changes are associated with increased susceptibility to KA-induced seizures. These rats, however, do not develop epilepsy as adults and do not have any major structural changes in the hippocampus.
The changes in levels of the GluR2 subunit of the AMPA receptor seen following P10 status may contribute to the observed increased seizure susceptibility in adulthood. We demonstrated a decrease in GluR2 mRNA expression in DGCs and decreased GluR2 protein in dentate gyrus of adult rats that had experienced prolonged SE early in life. Glutamate receptors, especially AMPA receptors, have been suggested to be critical mediators of excitotoxicity in the immature brain. In rat, AMPA receptor density peaks in the second postnatal week (Insel et al. 1990) and thus SE during this developmental window may be particularly likely to result in chronic changes in these receptors as observed in our study. A number of studies performed in the perinatal hypoxia–ischemia model (Jensen and Wang 1996; Koh and Jensen 2001; Jensen et al. 1995) support the role of AMPA receptor changes in epileptogenesis following early-life seizures. Jensen and Wang (1996) reported that rats exposed to hypoxia at P10 have reduced seizure thresholds to chemo-convulsants into adulthood and that the acute and long-term epileptogenic effects of hypoxia can be inhibited by AMPA receptor antagonists but not by NMDA receptor antagonists. Similarly, topiramate, which selectively blocks AMPA and KA currents but not NMDA-evoked currents (Shank et al. 2000) suppresses seizures during hypoxia at P10 and also prevents long-term changes in seizure susceptibility (Koh and Jensen 2001).
It is known that channel properties of receptors depend, in part, on the subunit composition. A number of studies have suggested that changes in glutamate receptor subunits may contribute to the development of chronic seizures and cognitive dysfunction in immature and adult epilepsy models (Mathern et al. 1998; Doi et al. 2001; Sanchez et al. 2001; Sogawa et al. 2001). For example, the presence of the GluR2 subunit in AMPA receptors makes the channel impermeable to calcium influx (Hollmann et al. 1991) and decreases the channel conductance (Swanson et al. 1997). Thus, in the current study, decreased GluR2 expression in DGCs of adult animals who experienced prolonged SE at P10 may cause an increase in calcium influx and higher channel conductance leading to greater seizure susceptibility. Sanchez et al. (2001) found increased seizure susceptibility into adulthood and decreased GluR2 subunit expression in the CA1 region of the hippocampus 96 h after global hypoxia induced seizures at P10. Further, GluR2 knockdown in the CA3 region of the hippocampus at P13 has been shown to produce age-dependent, seizure-like behavioral manifestations when the rats reach adulthood (Friedman and Veliskova 1998). These results in combination with ours suggest that decreased GluR2 expression occurs in multiple different principle cell types in the hippocampus after early-life SE and that GluR2 reduction in any of these cell types may be associated with increased seizure susceptibility.
The decreased GluR1 subunit mRNA expression in both lithium control and pilocarpine-treated rats compared to naive rats and the significant decrease in GluR1 subunit protein in lithium control rats compared to naive rats suggest that changes in GluR1 expression may be related to the effects of the handling and/or lithium injection the rats received as pups. We can not differentiate clearly whether GluR1 mRNA changes are related to the lithium injection, the handling procedure or both, as we were unable to separately analyze GluR1 levels in handled animals that did not receive lithium (i.e. saline controls) due to technical problems. Analysis of GluR5, GluR6, KA1, KA2 and EAAC1 mRNA levels in saline control animals, however, showed no difference between lithium injected and saline injected controls, suggesting that lithium injection did not alter expression of these genes. Additional studies of GluR1 expression in saline control rats will be needed to determine whether the decreased GluR1 levels in lithium control rats are more related to the effects of handling, the effects of lithium or a combination of these factors. Effects of early-life handling on glucocorticoid receptor levels (Meaney et al. 1988) and GABA receptors (reviewed in Caldji et al. 2000) have been observed. To our knowledge effects of early-life handling on AMPA receptor expression have not previously been reported.
Kainate receptors assembled from GluR5–7, KA1 and KA2, are believed to modulate synaptic transmission by both pre- and postsynaptic mechanisms (Chittajallu et al. 1999). Substantial evidence indicates that alterations in KA receptors may contribute to temporal lobe seizures in adult models (Mulle et al. 1998; Bernard et al. 1999). Knockout studies have demonstrated that GluR6-deficient mice are less susceptible to systemic administration of KA (Mulle et al. 1998) and over-expression of GluR6 in rat hippocampus can produce seizures (Telfeian et al. 2000). Further, selective antagonists of GluR5 subunit-containing KA receptors can prevent pilocarpine-induced limbic seizures (Smolders et al. 2002). These phenomena indicate that kainate receptors containing the GluR5/GluR6 subunit play an important role in synaptic transmission as well as in epileptogenesis in the adult brain. In the current study, however, we did not observe a significant change in KA receptor subunit mRNAs or protein in dentate after lithium-pilocarpine induced seizures at P10. Thus, despite the strong evidence in adult epilepsy models for a role of KA receptors, it remains unclear what role, if any, changes in KA receptors may play in epileptogenesis in the developing brain.
Our result of increased EAAC1 levels in the hippocampus following prolonged SE is similar to findings in earlier studies carried out in adult epilepsy models (Ueda and Willmore 2000; Crino et al. 2002). Glutamate transporters are believed to be crucial for preventing accumulation of neurotoxic levels of extracellular glutamate. EAAC1 is the primary neuronal glutamate transporter in the forebrain and GLT-1 is primarily distributed in astrocytes (Furuta et al. 1997). Reduction of EAAC1 expression using antisense oligonucleotides has been shown to result in seizures in rats, possibly by reducing intracellular glutamate stores needed for GABA synthesis (Sepkuty et al. 2002). Thus, increased EAAC1 levels observed in the present study might be an important compensatory change that both enhances clearance of glutamate from the synaptic cleft and may serve to increase inhibitory synaptic transmission by increasing GABA production. In kindled epileptic rats it has been found that seizures induce fast inhibition in the mossy fiber terminals projecting from dentate to CA3 (Gutierrez 2000) and increase the endogenous GABA content in mossy fiber terminals (Gomez-Lira et al. 2002). This suggests that increased EAAC1 expression may augment inhibition of excitatory CA3 pyramidal cells and potentially help prevent spontaneous seizures from occurring in this developmental model. However, it is also possible that EAAC1 increases in DGCs may result in enhanced inhibition of inhibitory interneurons, effectively enhancing the excitability of CA3 and contributing to the increased susceptibility to seizures. Ultimately, recordings of the entire hippocampal circuit in slices isolated from pilocarpine treated rats will be required to determine what, if any, changes in hippocampal excitability result from the seizure-induced EAAC1 changes we currently report.
Our observation that prolonged SE at P10 does not lead to development of spontaneous seizures later in life is in agreement with studies by several other groups (Cavalheiro et al. 1987; Priel et al. 1996; Nehlig et al. 2002). Our results regarding the increased susceptibility to KA-induced seizures following lithium/pilocarpine-induced SE are consistent with other developmental seizure models such as hyperthermia (Dube et al. 2000), hypoxia–ischemia (Jensen et al. 1992) and kindling (Moshe and Albala 1982). This observation however, is in contrast to that of Nehlig et al. (2002) who did not find a change in seizure susceptibility in an immature pilocarpine model. This discrepancy could be due to a difference in methods, as Nehlig and colleagues did not record from hippocampus and injected a fixed amount of kainic acid intraperitonally, whereas, we infused continuously through lateral tail vein. These differences may have permitted us to assess more subtle differences in seizure onset within the limbic system. The change in seizure susceptibility occurred in the absence of any visible hippocampal structural damage, however, we did not perform detailed neuroanatomical analysis such as stereology, so subtle cell loss could have been missed in the current study. Subtle cell loss has been reported by Sankar et al. (1998) using the lithium/pilocarpine model in slightly older rat pups (P14) than were used in the current study. A lack of chronic structural changes have also been described following hypoxia–ischemia (Jensen et al. 1992), kindling (Moshe and Albala 1982) and hyperthermia (Toth et al. 1998) induced seizures during postnatal development.
There are several limitations in the present study that must be considered when interpreting the results. A significant limitation of any study in immature animals is the potential effects of handling itself on the outcome of the experiment. A second limitation is that the present study cannot differentiate the muscarinic effects of pilocarpine itself from the effects of the seizures that the pilocarpine induces. Immature rats are so exquisitely sensitive to the convulsant effects of pilocarpine that it is nearly impossible to administer a truly ‘subconvulsant’ dose that would permit this differentiation. Additional studies utilizing different SE models that do not involve pilocarpine will therefore be required to fully address this question. An additional limitation is that we examined mRNA levels only to dentate granule cells. Although this approach has the advantage of cellular specificity and a high sensitivity for detecting subtle changes in mRNAs occurring only in a single cell type, it has the disadvantage of not giving a broader picture of all changes in the hippocampus. To fully understand the effects of P10 seizures on excitatory receptor and transporter changes in the hippocampus, future studies will need to examine the expression of these genes in other principle cells and interneuron subtypes. Finally, as discussed above, we can only infer the possible functional consequences of the DGC specific changes in mRNA and regional changes in protein shown here. Electrophysiological studies to assess receptor function within individual neurons as well as excitability of the entire hippocampal circuit will be needed to fully assess the functional impact of the observed molecular changes.
In summary, our data suggest that prolonged early-life seizures lead to long-lasting changes in expression of glutamate receptors and transporters in the dentate gyrus of the hippocampus. These molecular changes are accompanied by permanent changes in seizure susceptibility, supporting the idea that early-life seizures may lead to persistent effects on neuronal excitability in the limbic system.
The authors thank Eric Bolisay for his technical assistance with the studies. This study was supported by funding from the National Institutes of Health (NS RO1-38595, A.R.B-K). Statistical support was provided by the CHOP MRDDRC (P30 HD 26979).