Asingle episode of neonatal seizures permanently alters glutamatergic synapses.код для вставкиСкачать
A Single Episode of Neonatal Seizures Permanently Alters Glutamatergic Synapses Brandon J. Cornejo, BS,1,2 Michael H. Mesches, PhD,1,3,4 Steven Coultrap, PhD,1 Michael D. Browning, PhD,1,5 and Timothy A. Benke, PhD, MD1,3,5,6 Objective: The contribution of seizures to cognitive changes remains controversial. We tested the hypothesis that a single episode of neonatal seizures (sNS) on rat postnatal day (P) 7 permanently impairs hippocampal-dependent function in mature (P60) rats because of long-lasting changes at the synaptic level. Methods: sNS was induced with subcutaneously injected kainate on P7. Learning, memory, mossy fiber sprouting, spine density, hippocampal synaptic plasticity, and glutamate receptor expression and subcellular distribution were measured at P60. Results: sNS selectively impaired working memory in a hippocampal-dependent radial arm water-maze task without inducing mossy fiber sprouting or altering spine density. sNS impaired CA1 hippocampal long-term potentiation and enhanced long-term depression. Subcellular fractionation and cross-linking, used to determine whether glutamate receptor trafficking underlies the alterations of memory and synaptic plasticity, demonstrated that sNS induced a selective reduction in the membrane pool of glutamate receptor 1 subunits. sNS induced a decrease in the total amount of N-methyl-D-aspartate receptor 2A and an increase in the primary subsynaptic scaffold, PSD-95. Interpretation: These molecular consequences are consistent with the alterations in plasticity and memory caused by sNS at the synaptic level. Our data demonstrate the cognitive impact of sNS and associate memory deficits with specific alterations in glutamatergic synaptic function. Ann Neurol 2007;61:411– 426 Approximately 3 in 1,000 infants suffer from neonatal seizures (seizures occurring in the first month of life); 16% of these children develop learning disabilities speculatively mediated in part by the seizures themselves.1 Neonatal seizures are often repetitive and prolonged.2 Severe neonatal seizures, or status epilepticus, have been independently associated with an adverse developmental outcome.3 The cost of educating children with learning disabilities can be two to five times that of their peers.4 Treating the impact of neonatal seizures necessitates an understanding of the mechanisms involved. Models of neonatal seizures eliminate factors such as concurrent illnesses, prior brain insult, medication effects, and behavioral interactions that impact learning ability. In developing rats, multiple episodes (ie, over several days) of early-life seizures result in later life learning impairment that is correlated with hippocampal cell loss and synaptic reorganization.5–7 In direct contrast, immature rats experiencing a single episode (ie, over a single day) of early-life seizures do not suffer later-life behavioral alterations, cell loss, or synaptic reorganization in many studies.8 –12 However, recent work demonstrates that a single episode of early-life seizure may impair hippocampal-dependent memory and synaptic plasticity through alterations in inhibitory synaptic transmission.13 Most excitatory synaptic transmission in the central nervous system is mediated by glutamate receptorchannels classified according to their preferred agonists: kainate, ␣-amino-3-hydroxy-5-methyl-4-isoxazole propionate (AMPA), or N-methyl-D-aspartate (NMDA).14 Cloned subunit subtypes for AMPA (GluR1-4) and NMDA receptors (NR1, NR2A-D) have been identified.14 Ca2⫹ influx through NRs is thought to mediate long-term potentiation and depression (LTP and LTD), the in vitro assays of rodent learning and memory.15 Postsynaptic changes in GluR subunit numbers16 or properties17 are thought to underlie synaptic From the Department of 1Pharmacology, 2Medical Scientist Training Program, and 3Department of Pediatrics, University of Colorado, School of Medicine; 4Veterans Affairs Hospital; 5Neuroscience Program and 6Department of Neurology, University of Colorado, School of Medicine, Denver, CO. Published online Feb 23, 2007 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ana.21071 Received Jul 15, 2006, and in revised form Nov 6. Accepted for publication Dec 1, 2006. Address correspondence to Dr Benke, Departments of Pediatrics, Neurology, and Pharmacology, University of Colorado, School of Medicine, Box B-182, 4200 East 9th Avenue, Denver, CO 80262. E-mail: email@example.com This article includes supplementary materials available via the Internet at http://www.interscience.wiley.com/jpages/0364-5134/suppmat Published 2007 by Wiley-Liss, Inc., through Wiley Subscription Services 411 modification. Scaffolding proteins such as PSD-95 regulate these changes.18 We expand earlier work investigating the effect of sNS on inhibitory synapses.13 In this report, we have associated the effects of a single episode of neonatal seizures (sNS) induced by kainate acid (KA) at postnatal day 7 (P7) with alterations of hippocampal excitatory synapses at the behavioral, cellular, and molecular levels. sNS caused permanent alterations in memory after P60. Measurements in vitro demonstrated a permanent adaptation of “plasticity of plasticity” or metaplasticity19 caused by sNS. GluR1 was shifted to intracellular pools, whereas expression of PSD-95 was enhanced and NR2A was decreased. We present an excitatory single-synaptic model incorporating our molecular findings and using the subunit “rules” for synaptic plasticity20 that explains the alterations in plasticity. Our findings advance the understanding of how neonatal seizures affect memory and glutamatergic, excitatory synaptic function. Materials and Methods Animals All studies conformed to the requirements of the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Institutional Animal Care and Use subcommittees of the Denver VA Medical Center and the University of Colorado Health Sciences Center. Timed-pregnant Sprague–Dawley rats (Charles Rivers Labs, Wilmington, MA) gave birth in-house. All rats were housed in microisolator cages with water and chow available ad libitum. Seizure Induction KA, a fixed glutamate analog,14 was used to induce temporal-lobe seizures.21 After injection, it subsequently leads to excess glutamate, and thus represents any insult leading to an excess of glutamate including hypoxia.22 Male rat pups were subcutaneously injected with KA (1–2mg/kg; Tocris, Ellisville, MO) on P7 (P0 defined as the date of birth), resulting in discontinuous behavioral and electrical seizure activity lasting up to 3 hours. P7 was chosen as the most sensitive time in development to the long-term effects of sNS,13 and it correlates with human neonatal seizures clinically,2 biochemically23 and electrographically.24 Onset of seizure activity occurred within 30 minutes of injection and was characterized by intermittent myoclonic jerks, generalized tonic-clonic jerks, scratching, “swimming,” and “wetdog shakes.” Mortality was less than 3%. Ictal bursts were typically less than 10 minutes, separated by 5 to 10 minutes,24 and not consistent with prior definitions of status epilepticus (ictal bursts ⬍ 20 minutes) in this model, which can be associated with eight times greater mortality.11 In contrast, ictal bursts longer than 20 minutes were observed after KA only in animals older than P12 (data not shown). Because of discrepancies in the definition of status epilepticus, which have included ictal discharges as brief as 5 minutes in adult humans,25 we have not applied this term for clarity. 412 Annals of Neurology Vol 61 No 5 May 2007 Control male rat pups were injected with an equivalent volume of 0.9% saline. Male pups were chosen to eliminate the effects of hormonal cycles on behavior. Rats were then tagged using a commercially available microchip tagging system (Avid Identification Systems, Norco, CA) so that experimenters remained blinded to the treatment. Offspring were returned to their dam after observable seizure activity ceased. Rats were weaned and separated according to sex at P22. At P60 to P90 behavioral, histological, electrophysiological, and biochemical analyses were undertaken with male rats. Four litters (average 10 pups/litter, 50% male) were used for the Morris water maze (MWM) and 5 separate litters for the radial arm water maze (RAWM); one litter was used in both. Timm’s staining was performed on four litters used for MWM and RAWM testing. Two separate litters were used for Golgi staining. Six separate litters were used for electrophysiology. Six separate litters were used for biochemistry. Morris Water Maze Rats were trained in the standard hippocampal-dependent MWM (1.5m diameter, 10cm diameter platform submerged 1cm below the surface of 21°C water obscured with black tempura paint [Prang, Heathrow, FL], as described previously26). Two blocks of four trials each occurred over 5 days. During each training trial, the amount of time spent in each quadrant, distance traveled, and latency to target platform were measured. If the rat did not find the target platform within 180 seconds, it was guided there and allowed to stay for 30 seconds. During the probe trials, the amount of time spent in each quadrant, distance traveled, latency to target platform, time spent in the target zone (a smaller region approximately 13cm in diameter around the target location), and number of platform crossings were recorded. Automated data collection using WaterMaze (Actimetrics, Evanston, IL) incorporated a video camera connected to a personal computer. Four-Trial Radial Arm Water Maze We used the four-trial radial arm water maze (4T-RAWM) to confirm MWM findings because it is a more difficult task27,28 and to train rats for episodic-like memory testing (see later). The 4T-RAWM27,28 consisted of 12 arms (15cm wide ⫻ 43cm long) emanating from a circular choice area (60cm in diameter) in a 1.5m diameter tank of 21°C water. An escape platform (10 ⫻ 15cm, with a black surface to match the tank background) was situated at the end of one of the arms, approximately 2cm below the surface of opaque water. Rats were pretrained in the maze for 5 days. Pretraining consisted of shaping the rats to find the target arm by initially preventing entry into the nontarget arms and gradually increasing the number of available arms until all 12 were open. Testing began on day 6, and the rats were trained over 4 days. The start arm for each trial was determined in a pseudorandom fashion with a given arm used once per day. The start and goal arms were different for each rat within a group on a given day, but equivocal across groups, to avoid place and position preferences. The goal arm for a given rat remained the same each day. Four trials were administered per day (maximum, 180 seconds) with a 30-second intertrial interval. If the rat did not find the escape platform within 180 seconds, it was guided to the correct arm and allowed to stay for 30 seconds. Two-Trial Radial Arm Water Maze with Delay We used the two-trial RAWM with delay (2T-RAWM) to isolate episodic-like memory deficits that can be seen with normal performance in the MWM and are critically dependent on hippocampal NMDA receptors.29 Testing in the 2T-RAWM followed the 4T-RAWM on day 14 over 4 days. One training trial was administered (maximum, 180 seconds). If the rat did not find the target platform within 180 seconds, it was guided to the correct arm and allowed to stay for 30 seconds. A 4-hour delay was inserted after the training trial. In the delay trial, the target platform was in the same location as the training trial for a given rat. The start location was novel for each rat during the delay trial. If the rat did not find the target platform within 180 seconds, it was guided to the correct arm and allowed to stay for 30 seconds. During RAWM trials, each rat was assessed for total errors (ie, entry into an arm that did not contain the escape platform), latency to the target platform, and repeat errors (reentry into an arm previously visited). Data were collected in a similar fashion to the MWM. Data are presented as the mean ⫾ standard error of the mean for each trial across 4 days. The results for each rat were pooled from each scorer and region, averaged, and reported as a Timm’s staining score for both the DG and CA3 regions. Slides with poor staining or artifact were not scored. Golgi Staining After anesthesia with an injection of pentobarbital (120mg/kg intraperitoneally), rats were transcardially perfused with 0.1M phosphate buffer (pH 7.4) followed by 10% formalin in 0.1M phosphate buffer. The brain was rapidly removed and rinsed in 0.1M phosphate buffer. The FD Neurotechnologies Rapid Golgi Stain Kit (FD Neurotechnologies, Ellicot City, MD) was used according to the manufacturer’s proprietary directions. Coronal sections (90m) were made through the hippocampus using a sliding microtome and mounted onto gelatin- and chrom alum–coated slides and stained. Sections were coverslipped using DPX mountant (BioChemika). The CA1 stratum radiatum region of representative sections was visualized at 100 to 1,000 times and scored for spine density (density calculated as the number of spines/total length of dendrite) and branching number along the primary apical dendrite of pyramidalshaped neurons.32 Typically, 391 ⫾ 10m/dendrite was scored (N ⫽ 123 neurons; range, 146 – 693m). Slides with poor staining or artifact were not scored. Timm’s Staining Hippocampal Slice Preparation and Electrophysiology Timm’s staining was performed as described previously.30 After anesthesia with an injection of pentobarbital (120mg/kg intraperitoneally), rats were transcardially perfused with a sodium sulfide solution followed by 10% formalin in 0.1M phosphate buffer. Brains were removed, cryoprotected for 2 days in 10% formalin with 20% glycerol in 0.1M phosphate buffer, then rapidly frozen on dry ice/acetone. Coronal sections (30m) were made through the hippocampus using a sliding microtome and mounted onto gelatin- (Fisher, Hampton, NH) and chrom alum–coated (Fisher) slides. In the dark, mounted sections were placed in a solution of gum arabic, citrate buffer, and hydroquinone for 30 to 45 minutes or until the molecular layer in the dentate gyrus (DG) was clearly stained. Sections were defatted, counterstained with cresyl violet (Standard Fluka, St. Louis, MO), and coverslipped with DPX mountant (BioChemika, St. Louis, MO). Scoring by three separate individuals at a total magnification of 400 to 1,000 times was performed as described elsewhere31: The CA3 and DG regions were scored on a scale of 0 (no granules in the supragranular region of DG and no granules in the stratum pyramidal or stratum oriens along any portion of the CA3 subregion), 1 (occasional granules in the stratum pyramidal or stratum oriens occurring in discrete bundles), 2 (occasional to moderate granules in the stratum pyramidal or stratum oriens), 3 (prominent but discontinuous granules in the stratum pyramidal or stratum oriens), 4 (prominent granules in the stratum pyramidal or stratum oriens occurring in nearcontinuous distribution along the entire CA3 region), or 5 (continuous or near-continuous dense laminar band of granules in the supragranular region of DG and continuous or near-continuous dense laminar band of granules stratum pyramidal and stratum oriens along the entire CA3 region). After rapid decapitation and removal of the brain, sagittal hippocampal slices (400m) were made using a vibratome (Vibratome, St. Louis, MO) in ice-cold sucrose artificial cerebrospinal fluid (saCSF; 206mM sucrose, 2.8mM KCl, 1mM CaCl2, 3mM MgSO4, 1.25mM NaH2PO4, 26mM NaHCO3, 10mM D-glucose and bubbled with 95%/5% O2/ CO2).33 Slices were recovered in a submersion-type chamber perfused with oxygenated artificial cerebrospinal fluid (aCSF; 124mM NaCl, 3mM KCl, 1mM MgSO4, 2mM CaCl2, 1.2mM NaH2PO4, 26mM NaHCO3, 10mM D-glucose and bubbled with 95%/5% O2/CO2) at room temperature for at least 60 minutes, and then submerged in a recording chamber perfused with aCSF. All electrophysiology was performed in the CA1 region. Two twisted-tungsten bipolar stimulating electrodes were offset in the CA1 to stimulate the two independent Schaffer collateral-commissural pathways using a constant current source (WPI, Sarasota, FL) with a fixed duration (20 microseconds), each at a rate of 0.033Hz. Field excitatory postsynaptic potentials (fEPSPs) were recorded from the stratum radiatum region of CA1 using a borosilicate glass (WPI) microelectrode (pulled to 6 to 9M⍀ when filled with 3M NaCl; Sutter, Novato, CA,), amplified 1,000 times (WPI and Warner, Hamden, CT), and digitized (CIODAS08/JR-A0; Measurement Computing, Middleboro, MA) at 10kHz using LTP-version 2.4.34 Input-output curves, as a measure of basal synaptic function, were generated at four stimulus intensities: 10, 20, 50, and 95% of maximal slope value (in msec/mV). Paired-pulse facilitation, an index of functional presynaptic CA1 glutamate release, was measured by two pulses 50 milliseconds apart. Paired-pulse facilitation associated with presynaptic function is maximal at this interpulse interval.35 Input-output curves and paired-pulse facilitation were measured after obtaining a stable baseline of 30 Cornejo et al: Neonatal Seizures Alter Synapses 413 minutes. After baseline stabilization of fEPSP slope at approximately 50% of maximal slope for at least 20 minutes, LTP was induced (100Hz ⫻ 1 second) in one Schaffer collateral-commissural pathway, whereas LTD was induced 60 minutes after LTP in the other pathway (900 paired-pulse stimuli at 1Hz with 50-millisecond interpulse interval). Slices with unstable baseline were discarded. Subcellular Fractionation After rapid decapitation, brains were removed and placed in ice-cold aCSF. The hippocampus was unrolled by inserting a thin instrument into the hippocampal fissure separating CA1 from the DG. The DG and CA3 regions were then rolled back and the connection between CA1 and CA3 was cut, removing CA3 and DG. The subiculum was then cut away from CA1. From the isolated CA1, 400m slices were prepared with a McIlwain tissue chopper. CA1 minislices were visually inspected to ensure that the other regions had been more than 99% effectively removed. This method has been found to produce viable slices.36,37 Homogenization (in 320mM sucrose, 10mM tris[hydroxymethyl]aminomethane [Tris], pH 7.4) was followed by subcellular fractionation with slight modification.38 Homogenates were centrifuged at 100 g to remove nuclei and large debris (P1). Crude synaptosomal membranes (P2) were then prepared from supernatants (S1) by centrifugation at 10,000 g. Resulting supernatants (S2) were spun at 100,000 g to obtain the light membrane fractions (P3). Final pellets were resuspended in STE buffer (10% sodium dodecyl sulfate, 10mM EDTA, 100mM Tris, pH 8) and boiled for 5 minutes. Protein concentrations were determined using a BCA protein assay kit (Pierce, Rockford, IL) with BSA (bovine serum albumin) as a standard. Samples were diluted to equal protein concentrations in sample loading buffer (2.3% sodium dodecyl sulfate, 67.5mM Tris, 10% glycerol, 5% ␤-mercaptoethanol, 0.017% bromophenol blue) and assayed by semiquantitative Western blot. Cross-linking Hippocampal slices that were prepared and recovered as for electrophysiology were placed into either ice-cold aCSF (to measure total protein concentration) or ice-cold aCSF containing 1mg/ml BS3 (BIS-[sulfosuccinimidyl] Suberate; Pierce) (to measure intracellular protein concentration) for 40 minutes at 4°C. To quench remaining BS,3 slices were washed three times in cold aCSF containing 20mM Tris pH 7.6. Slices were suspended in STE buffer, sonicated, and samples prepared for Western blotting as with subcellular fractionation. Semiquantitative Western Blotting Western blotting was performed as described previously.39 Samples were loaded in duplicate on 7.5% polyacrylamide gels and a six-point dilution series of rat hippocampal homogenate was included on each gel as a standard curve. Blots, after blocking with either 5% Carnation nonfat dry milk or 3% BSA (Sigma, St. Louis, MO) in TTBS (140mM NaCl, 20mM Tris, pH 7.6, 0.1% Tween 20 [Sigma]), were probed overnight with primary antibody at 4°C. Primary antibodies to GluR1 and GluR2/3 (Chemicon, Temecula, CA) 414 Annals of Neurology Vol 61 No 5 May 2007 were diluted 1:3,000 in 1% Carnation nonfat dry milk, NR1 (BD Biosciences, San Diego, CA) was diluted 1:3,000 in 1% BSA, NR2A and NR2B40 were diluted 1:1,000 and 1:3,000, respectively in 1% BSA, and PSD-95 (Affinity Bioreagents, Golden, CO) was diluted 1:5,000 in 1% BSA. Blots were washed three times with TTBS, probed for 1 hour at room temperature with horseradish peroxidase–conjugated goat anti–mouse or goat anti–rabbit secondary antibodies (BioRad, Temecula, CA), then washed three more times with TTBS. Immunodetection was effected using a chemiluminescent substrate (SuperSignal West Femto Maximum Sensitivity Substrate; Pierce) and an Alpha Innotech imaging system; digital images were quantified using AlphaEase software (Alpha Innotech, San Leandro, CA). Immunoreactivity was reported as the density of sample bands relative to the standard curve. Only values falling within the standard curve generated from the dilution series included on each gel were incorporated into the final analysis. Statistics Data are expressed as mean ⫾ standard error of the mean, and n ⫽ number of rats for a given treatment. Two-way repeated-measures analyses of variance (ANOVAs) with Holm–Sidak post hoc analysis, two-way ANOVAs, Mann– Whitney rank-sum, and Student’s t tests were used, where indicated, for statistical comparisons for electrophysiological, anatomical, and biochemical data using Origin 7.5 S5R (Origin Lab Corporation, Northampton, MA) or SigmaStat (Systat, Point Richmond, CA). Significance was reported at p ⬍ 0.05. Results Morris Water Maze and Four-Trial Radial Arm Water Maze Showed No Effect of Single Episode of Neonatal Seizures on Spatial Learning and Recall Previous work has described the utility of the MWM in assessing hippocampal-dependent spatial learning.26,41 In developing rats, multiple episodes of seizures cause hippocampal-dependent spatial learning deficits as measured with the MWM.5–7 Recent work has demonstrated that a single episode of early-life seizures has a negative impact on behavior and hippocampal-dependent spatial learning using a fooddependent radial arm maze and other behavioral tests.13 Radial arm mazes require a greater memory load, and thus are capable of potentially demonstrating subtle learning and memory deficits.27,28 Our utilization of the RAWM omits the necessity of a food reward and is therefore hippocampal dependent28 and more comparable with the MWM. We initially used the MWM in adult rats to assess potential deficits induced by sNS to directly compare our results with prior studies. With respect to distance traveled, two-way repeated-measures ANOVA demonstrated no overall effect of sNS. There was a nonsignificant difference in block 3 (the first block of day 2) in which control rats traveled less distance to find the platform than sNS rats (Fig 1A). Rats performed equally well during the probe trial, indicating that hippocampal-dependent spatial recall after learning was grossly undisturbed after sNS (see Fig 1B). Although no overall effect of training or treatment was observed in the 4T-RAWM, a pairwise Holm–Sidak analysis following two-way repeated-measures ANOVA showed that sNS rats (13.15 ⫾ 0.78 total errors; n ⫽ 12) made significantly more total errors on trial 1 compared with control rats (11.32 ⫾ 0.65 total errors; n ⫽ 8) ( p ⫽ 0.004), suggesting a subtle defect in learning. However, both sNS and control rats acquired the task equally (see Fig 1C). There were no differences in swim speed after sNS (25.4 ⫾ 0.5cm/sec; n ⫽ 10) compared with the control group (25.7 ⫾ 0.5cm/sec; n ⫽ 11; Student’s t test, p ⫽ 0.68). Fig 1. Behavioral effects of a single episode of neonatal seizures (sNS) in adult rats were isolated to a deficit in episodic-like or working memory in the two-trial radial arm water maze (2T-RAWM). (A) Adult male rats that experienced sNS (solid circles) or saline-injected control rats (open circles; treated on postnatal day 7 [P7] and matured to P60) were trained on the Morris water maze (MWM). Both sNS (n ⫽ 11) and control rats (n ⫽ 10) performed equally across all days of training regarding distance traveled. No differences were observed with respect to swim speed (see Results). (B) No differences in percentage time spent in each quadrant (NE, Target, SW, SE) were detected during the probe trial. (C) In the four-trial (4T)-RAWM, total errors were defined as an incorrect arm entry and were used to measure the extent of task acquisition and recall. The task was learned equally by both sNS (n ⫽ 12) and control rats (n ⫽ 8) because both had similar total errors for trials 1 through 4 (T1-T4); however, a pairwise Holm–Sidak analysis after two-way repeated-measures analysis of variance (ANOVA) demonstrated that sNS rats made slightly but significantly more total errors on T1 compared with control rats (see Results). (D) For testing in the 2T-RAWM, a 4-hour delay was inserted between the training trial and the delay trial over the following 4 days (see Methods). After a 4-hour delay, sNS rats (solid bars) demonstrated a significant increase in the number of total errors made compared with control rats (open bars; 6.58 ⫾ 0.75 vs 3.73 ⫾ 0.44) (see Results). (E) Latency to target platform also was used to characterize working memory deficits. sNS rats had a significantly longer latency to platform during the delay trial compared with control rats (52.38 ⫾ 7.34 vs 29.82 ⫾ 2.97) (see Results). (F) Repeat errors, a measure of working memory, were defined as an entry into an arm that had been previously visited. sNS rats made significantly more repeat errors than control rats (2.58 ⫾ 0.42 vs 1.02 ⫾ 0.21). Asterisk indicates statistically significant difference by two-way repeated-measures ANOVA, p ⱕ 0.05. Cornejo et al: Neonatal Seizures Alter Synapses 415 Two-Trial Radial Arm Water Maze with Delay Showed Specific Effect of Single Episode of Neonatal Seizures on Memory Recent studies suggest that the hippocampus underlies the “automatic recording” of episodic-like memory in humans and in rodents.29 Episodic-like memory has been defined as recall for a single event that involves spatial, nonspatial, and temporal cues.29 Isolated, episodic-like memory deficits can be seen with normal performance in the MWM and are critically dependent on NMDA receptors.29 The 2T-RAWM directly tested the effects of sNS on memory. Adult sNS rats made more (6.58 ⫾ 0.75; n ⫽ 28) total errors than control rats (3.73 ⫾ 0.44; n ⫽ 29) during the delay trial (see Fig 1D; two-way repeated-measures ANOVA, F (1, 55) ⫽ 7.928; p ⫽ 0.007). sNS rats took longer (52.38 ⫾ 7.34 seconds; n ⫽ 29) during the delay trial to find the platform than littermate control rats (29.82 ⫾ 2.97 seconds; n ⫽ 28) (see Fig 1E; two-way repeated-measures ANOVA, F (1, 55) ⫽ 4.874; p ⫽ 0.031). The number of repeat errors (defined as reentries into arms that were already visited) was also examined. These errors are considered a measure of working memory and reflect impairment in the mechanisms responsible for maintaining short-term information processing.29 sNS rats demonstrated more than twice as many repeat errors (2.58 ⫾ 0.42; n ⫽ 29) during the delay trial compared with control rats (1.02 ⫾ 0.21; n ⫽ 28) (see Fig 1F; two-way repeated-measures ANOVA, F (1, 55) ⫽ 6.708; p ⬍ 0.001). Thus, although learning (and recall after learning) appears grossly normal, this demonstrates a specific alteration of episodic-like or working memory, or both. A Single Episode of Neonatal Seizures Did Not Cause Mossy Fiber Sprouting We next examined two aspects of hippocampal morphology that have been associated with deficits in learning and memory. Aberrant mossy fiber sprouting after multiple episodes of early-life seizures31,42 and early-life stress43 has been linked with poor performance on the MWM. We found no aberrant mossy fiber sprouting after sNS (see Supplementary Figs 1A1 and 1A2) (CA3: Mann–Whitney rank-sum, p ⫽ 0.230, n ⫽ 19 control rats, n ⫽ 18 sNS rats; DG: Mann– Whitney rank-sum, p ⫽ 0.681, n ⫽ 18 control rats, s ⫽ 18 sNS rats). As a positive control, seizures were induced in adult rats by injection with KA (10mg/kg) on P42. Based on previous data, it was expected that these rats would have aberrant mossy fiber sprouting in the hippocampus compared with saline-injected control rats.31 In confirmation of this, rats that seized as adults (n ⫽ 2) developed easily detectable mossy fiber sprouting in the DG and CA3 region of the hippocampus 416 Annals of Neurology Vol 61 No 5 May 2007 compared with control rats (n ⫽ 2) (see Supplementary Figs 1B1 and 1B2). A Single Episode of Neonatal Seizures Did Not Alter Spine Density or Dendritic Branching Reduction in spine density after multiple early-life seizures has been linked with poor performance on the MWM32,44 and in other rodent models of learning and memory deficits.45 Reduction in spine density is also found in humans with Down’s syndrome and other metabolic disorders.46 We examined spine density and branching by bright-field imaging of Golgi-stained CA1 hippocampal neurons. There were no changes between sNS (n ⫽ 5; 52 neurons) and control rats (n ⫽ 7; 71 neurons) for either spine density (Figs 2A, B) or branching along primary apical dendrites (see Figs 2A, C). A Single Episode of Neonatal Seizures Altered Hippocampal CA1 Plasticity The lack of evidence for histological changes that could underlie the sNS-induced alterations in memory led us to examine the possibility that sNS might alter CA1 hippocampal function or synaptic plasticity in vitro. We found no significant differences between groups for paired-pulse facilitation (Fig 3A), suggesting sNS did not alter presynaptic function. By comparing presynaptic fiber volley size with fEPSP slope at different stimulus intensities (see Fig 3B), we found no significant difference between groups, suggesting that sNS did not change overall excitability. LTP and LTD are thought to reflect the processes involved in learning and memory,15 and consideration of the balance between the two has been referred to as metaplasticity.19 When a 100Hz ⫻ 1-second stimulus was used to induce LTP, slices from sNS rats demonstrated a significant and profound loss of LTP compared with control rats as measured by percentage change in fEPSP slope (Fig 4A, asterisk indicates p ⱕ 0.05 by Student’s t test). We next induced LTD using a robust47 stimulus protocol (900 paired pulses at 1Hz, 50-millisecond interpulse interval). There was a significant enhancement of LTD in slices from rats that experienced sNS (see Fig 4B, asterisk indicates p ⱕ 0.05 by Student’s t test). Thus, sNS could have been a metaplastic event that altered the balance between LTP and LTD. Effects of Single Episodes of Neonatal Seizures on Glutamate Receptors and PSD-95 Alterations in membrane expression of glutamate receptors,15 either caused by increases or decreases in total numbers, or shifts between membrane and intracellular pools, could play key roles in determining the “set point” for LTP versus LTD, that is, metaplasticity. To test the effects of sNS on membrane expression of glutamate receptors, we used both subcellular fraction- Fig 2. Memory deficits caused by a single episode of neonatal seizures (sNS) were not associated with changes in spine density or spine branching measured in adult rats. (A) No observable differences in branching pattern or spine density were seen after Golgi staining (see Materials and Methods) at low (100⫻, left panels; scale bar ⫽ 20m) and higher magnification (1,000⫻, right panels; scale bar ⫽ 10m) after sNS (n ⫽ 5; 51 neurons) or for saline control rats (n ⫽ 7; 72 neurons) after postnatal day (P60). Quantification (B) showed no differences for spine density. Spine densities were calculated as the number of spines per micron along the apical dendrite. Quantification (C) showed no differences for the number of branches on the apical dendrite. The total number of branches on the selected apical dendrite used for spine density was counted. ation and cross-linking techniques to separate and measure the concentration of glutamate receptor subunits in both the membrane and intracellular compartments of neurons in the CA1 region of the hippocampus. In the first set of experiments, we prepared both intracellular and surface membrane fractions using classical subcellular fractionation from control and sNS tissue, and then probed the fractions with antibodies against GluR1, GluR2/3, and NR1. We found that there was a significant (ie, 52%) increase in intracellular GluR1 in CA1 slices prepared from rats that experienced sNS (2.35 ⫾ 0.26 immunoreactivity/g protein; n ⫽ 11) compared with control rats (1.54 ⫾ 0.21 immunoreactivity/g protein; n ⫽ 8) (Student’s t test, p ⫽ 0.03) (Figs 5A, D). This effect appeared to be specific for GluR1, because we saw no effect of sNS on levels of GluR2/3 or NR1 in either the membrane fraction or intracellular fraction (see Figs 5A–C). To be certain that the change in GluR1 levels was not simply an artifact of the subcellular fractionation assay, we confirmed our results using a cross-linking methodology. This technique incorporated the use of the membrane-impermeable cross-linker BS3 that specifically cross-links extracellular domains of membrane proteins and allows their subsequent separation from intracellular proteins through gel electrophoresis. The concentration of total protein is compared with the measured concentration of intracellular protein and expressed as a percentage of the total. While using this assay, we detected no change in the total amount of GluR1. However, we detected a significant increase in intracellular GluR1 in rats that experienced sNS (n ⫽ 8) compared with control rats (n ⫽ 6), as the intracellular percentage increased from 5.8 ⫾ 0.5 to 10.8 ⫾ 1.1% (Fig 6A) (Student’s t test, p ⱕ 0.0008). This percentage change (100*[10.8 ⫺ 5.8]/5.8 ⫽ 85%) was comparable with that seen with fractionation techniques (52%) given the differences in methodology. We speculate that a comparable change in the membrane fraction would be difficult to detect because most GluR1 (100 –5.8% ⫽ 94.2% in control rats to 100 –10.2% ⫽ 89.2% after sNS) is on the membrane surface. Such a percentage change (100*[94.2 ⫺ 89.2]/ 94.2 ⫽ 5.3%) was within the standard error for the Cornejo et al: Neonatal Seizures Alter Synapses 417 Fig 3. A single episode of neonatal seizures (sNS) did not alter synaptic excitability measured in the CA1 region of adult hippocampal slices. Field excitatory postsynaptic potentials (fEPSPs) were measured from the stratum radiatum of the CA1 region of adult hippocampal slices in response to stimulation of the Schaffer collateral-commisural pathway (see Materials and Methods). (A) sNS (n ⫽ 11) did not significantly alter paired-pulse (50-millisecond interpulse interval) facilitation compared with saline-injected control rats (n ⫽ 15). No differences were observed between groups. P1 ⫽ normalized fEPSP slope on first pulse; P2 ⫽ normalized fEPSP slope after 50-millisecond interpulse interval. (B) Fiber volley amplitude and fEPSP slope were measured at four different stimulus intensities (see Materials and Methods). Data points for each measurement (averaged from 16 sweeps) from sNS (solid circles; n ⫽ 8) and saline-injected control rats (open circles; n ⫽ 9) are plotted. The slope of the regression line for sNS (dotted) and control rats (solid) were similar. Sample traces at 10, 20, 50, and 95% of maximal slope value are shown from control (thick line) and sNS animals. measurement of GluR1 in the membrane fraction determined by subcellular fractionation (see Fig 5D). As shown previously,37 the specificity of BS3 for extracellular domains was confirmed by the absence of any effect of the cross-linker on the intracellular protein synapsin (data not shown). Control animals demonstrated expected expression patterns for mature animals.48 We found no differences in either totals or localization of GluR2/3 or NR1 between control rats and rats that experienced sNS (see Figs 6B, C), as with subcellular fractionation. We detected a significant 28% loss of total NR2A in rats that experienced sNS (0.534 ⫾ 0.042 immunoreactivity/g protein; n ⫽ 9) compared with control rats (0.745 ⫾ 0.097 immunoreactivity/g pro- 418 Annals of Neurology Vol 61 No 5 May 2007 tein; n ⫽ 9) (Student’s t test, p ⬍ 0.03) (see Fig 6D). However, the percentage of this total located intracellularly was unchanged by sNS. We found no differences in either totals or localization of NR2B between control rats and rats that had experienced sNS (see Fig 6E). PSD-95, one of the most abundant scaffolding proteins, directly interacts with NRs and indirectly with GluRs.49 PSD-95 is almost completely expressed in the intracellular compartment of only excitatory synapses.50 Therefore, we used immunoblotting to investigate the intracellular concentration of PSD-95 (Figs 7A, B). We observed a significant 43% increase in the total concentration of PSD-95 in rats that experienced Fig 4. A single episode of neonatal seizures (sNS) caused a metaplastic shift that impaired long-term potentiation (LTP) and enhanced long-term depression (LTD) measured in the CA1 region of adult hippocampal slices. Field excitatory postsynaptic potentials (fEPSPs) were measured from the stratum radiatum of the CA1 region of adult hippocampal slices in response to stimulation of the Schaffer collateral-commisural pathway (see Materials and Methods). (A) After LTP induction (100Hz ⫻ 1 second), sNS (filled circles; n ⫽ 10) caused a near-total loss of LTP compared with controls (solid circles; n ⫽ 14). *p ⱕ 0.05, Student’s t test. Sample averaged sweeps (arrowhead at stimulus artifact, average of four sweeps) show pre- (thin lines) and post-LTP (thick lines) traces. (B) After an extended baseline period demonstrating stability, sNS rats (n ⫽ 11) displayed significantly enhanced LTD (900 ⫻ 50-millisecond paired pulses at 1Hz) compared with their littermate controls (n ⫽ 15). *p ⱕ 0.05, Student’s t test. Sample averaged sweeps show pre- (thin lines) and post-LTD (thick lines) traces. sNS (1.084 ⫾ 0.127 immunoreactivity/g protein; n ⫽ 8) compared with control rats (0.757 ⫾ 0.076 immunoreactivity/g protein; n ⫽ 11; Student’s t test, p ⱕ 0.03). Given that the number of synapses did not appear to be altered by sNS (see Fig 2), this suggests that the amount of PSD-95 in each excitatory synapse was increased after sNS. Discussion Our results establish using multiple approaches that sNS disrupted hippocampal-dependent memory and plasticity that were correlated with alterations at excitatory, glutamatergic synapses. Although much early data argued that immature rodents tend to be resistant to a single episode of seizure-induced changes in physiology, behavior, or histology (reviewed in Stafstrom51), recent work13 has challenged this and has demonstrated altered inhibitory synaptic function. Now, by linking crucial glutamatergic receptor and scaffolding changes with plasticity and behavior, we are mechanistically beginning to understand the harmful issues underlying just a single episode of relatively mild neonatal seizures. Standard behavioral testing (MWM) and a more difficult test (4T-RAWM) determined hippocampaldependent learning and recall were grossly normal in adult rats after sNS. The 2T-RAWM demonstrated that sNS specifically impaired episodic-like or working memory, or both. This mixture of normal learning and recall with an episodic-like memory deficit is also seen Cornejo et al: Neonatal Seizures Alter Synapses 419 Fig 5. A single episode of neonatal seizures (sNS) caused an isolated, long-term increase in intracellular glutamate receptor 1 (GluR1) without changes in GluR2/3 or N-methyl-D-aspartate (NMDA) receptor 1 (NR1) localization in adult CA1 hippocampus. Membrane and intracellular fractions were generated using subcellular fractionation and analyzed on Western blots using a known standard curve to generate a relative concentration (immunoreactivity/g protein) (see Materials and Methods). (A) Example blots of GluR1, GluR2/3, and NR1. Greater density is observed in the intracellular fraction stained by GluR1 after sNS compared with control; all other blots are nearly equal. (B) There was no effect of sNS on NR1 concentrations in the membrane (sNS: n ⫽ 9; control: n ⫽ 8) or intracellular pools (sNS: n ⫽ 8; control: n ⫽ 7). (C) There was no effect of sNS on GluR2/3 concentrations in the membrane (sNS: n ⫽ 8; control: n ⫽ 9) or intracellular pools (sNS: n ⫽ 7; control: n ⫽ 7). (D) sNS rats resulted in an increase in intracellular GluR1 concentrations (2.35 ⫾ 0.26 immunoreactivity/g protein; n ⫽ 11) compared with their littermate control rats (1.54 ⫾ 0.21 immunoreactivity/g protein; n ⫽ 8; p ⫽ 0.03, Student’s t test;). There was no effect of sNS on GluR1 concentrations in the membrane (sNS: 1.54 ⫾ 0.14 immunoreactivity/g protein; n ⫽ 12; control: 1.54 ⫾ 0.15 immunoreactivity/g protein; n ⫽ 10). in GluR1 knock-out mice52 and in mice in which specific GluR1-modulating phosphorylation sites have been altered.53 In these knock-out studies, both LTP and LTD were obliterated. In contrast, we found that there was an overall shift in plasticity that favored the expression of greater LTD at the expense of a total loss of LTP; thus, other plasticity mechanisms emerged here, as in prior knock-out studies, to sustain MWM performance. These findings were specifically associated with an increase in the intracellular pool of GluR1, a net loss of NR2A, and an increase in PSD95. We did not find any changes in GluR2/3, NR1, or NR2B regarding their total amounts or their distribution. Prior studies by others8,13,54 have not reported any evidence of cell loss caused by sNS. Therefore, because sNS did not change overall excitability or spine numbers, we conclude that sNS caused alterations at the level of individual excitatory, glutamatergic synapses. The underlying derangements in GluR1, NR2A, and 420 Annals of Neurology Vol 61 No 5 May 2007 PSD-95 that result in altered plasticity at an excitatory synapse can be explained by saturated or impaired receptor expression, or both, and altered induction mechanisms (Fig 8). Given the dramatic increase in PSD-95, the stability in spine numbers, and the use of field recording techniques that sample many synapses at a time, we speculate that the majority of synapses in the CA1 region of the hippocampus have been affected. Scaffolding proteins such as PSD-95 bring GluRs to synapses and upregulate them in LTP55,56 and remove them in LTD.55,57 In cultured neurons, overexpression of PSD-95 is associated with sustained synaptic enhancement,58 occlusion of LTP,59,60 and enhanced LTD.59 We speculate that the 43% increase in PSD-95 expression that we observed in adult synapses after sNS drives a sustained synaptic enhancement or saturation and partially underlies the profound loss of LTP. Therefore, it is consistent with prior findings that we Fig 6. A single episode of neonatal seizures (sNS) caused a long-term shift favoring intracellular glutamate receptor 1 (GluR1) and a loss of total N-methyl-D-aspartate (NMDA) receptor 2A (NR2A) in adult CA1 hippocampus. Total hippocampal CA1 protein concentration (immunoreactivity/g protein [Imm/g protein]) and intracellular fractions (percentage expression [% Intra Exp]) were generated using cross-linking techniques (see Materials and Methods). (A) There was no effect of sNS on total GluR1 concentration (sNS: 0.52 ⫾ 0.05 immunoreactivity/g protein, n ⫽ 11; control: 0.57 ⫾ 0.07 immunoreactivity/g protein, n ⫽ 8). There was a significant increase in the intracellular fraction of GluR1 after sNS (10.83 ⫾ 1.1%; n ⫽ 8) compared with saline controls (C; 5.84 ⫾ 0.54%; n ⫽ 6; p ⫽ 0.008, Student’s t test). (B) There was no effect of sNS on total GluR2/3 concentration (sNS: n ⫽ 8; control: n ⫽ 9) or intracellular pools (sNS: n ⫽ 8; control: n ⫽ 6). (C) There was no effect of sNS on total NR1 concentration (sNS: n ⫽ 9; control: n ⫽ 8) or intracellular pools (sNS: n ⫽ 9; control: n ⫽ 9). (D) sNS caused a decrease (0.534 ⫾ 0.042 immunoreactivity/g protein; n ⫽ 9) in total NR2A concentration compared with saline controls (0.745 ⫾ 0.097 immunoreactivity/g protein; n ⫽ 9; p ⫽ 0.03, Student’s t test). However, there were no differences in NR2A concentrations in intracellular fractions between sNS (n ⫽8) and control rats (n ⫽ 8). (E) There was no effect of sNS on total NR2B concentration (sNS: n ⫽ 8; control: n ⫽ 10) or intracellular pools (sNS: n ⫽ 8; control: n ⫽ 10). observed complete occlusion of LTP and greater LTD in adult rats that experienced sNS. After sNS, adult synapses likely have decreased membrane expression of GluR1, because of a 52 to 85% increase in the amount of GluR1 in the intracellular pool. GluR1 and GluR2/3 are the predominant GluRs in adult synapses.61 We predict that if there are fewer total functional GluR receptors after sNS, overall excitability has to be balanced. This is supported by our findings of unchanged excitability (see Fig 3). For this to occur, some individual GluRs must be potentiated or enhanced. Limitations with the biochemical assays used do not permit a differentiation between those membrane-bound receptors that are synaptic versus extrasynaptic. However, our data do not suggest that sNS has caused any such shift in GluRs (see Fig 3). Cornejo et al: Neonatal Seizures Alter Synapses 421 Fig 7. A single episode of neonatal seizures (sNS) increased the concentration of PSD-95 in adult CA1 hippocampus. Total hippocampal CA1 protein concentration (immunoreactivity/g protein) was generated by probing whole CA1 homogenates with antibody against PSD-95 (see Materials and Methods). (A) Typical anti–PSD-95 immunoblot shows marked increase in density after sNS. (B) There was a significant increase in the total concentration of PSD-95 after sNS (1.084 ⫾ 0.127 immunoreactivity/g protein; n ⫽ 8) compared with saline-injected control rats (0.757 ⫾ 0.076 immunoreactivity/g protein; n ⫽ 11; p ⫽ 0.008, Student’s t test). Fig 8. Schematic of proposed mechanisms causing altered plasticity after a single episode of neonatal seizures (sNS). Postsynaptic changes in glutamate receptor (GluR) subunit numbers16 or properties17 are thought to underlie synaptic modification, which has resulted in postulated “subunit rules”: (1) for GluR1-4, synaptic removal of GluR2 drags along GluR1 and GluR3 which causes long-term depression (LTD); (2) GluR1 or GluR3 not associated with GluR2 (“homomers”) act independently; (3) insertion, modification, or both of GluR1 underlies LTP.20 Insertion of GluR1 predominates in long-term potentiation (LTP) at younger ages62 but is supplanted by modification at later ages.48 Preferential activation of different N-methyl-D-aspartate (NMDA) receptor subtypes is thought to underlie the induction of LTP (NR2A) versus LTD (NR2B).63– 65 GluR channels at excitatory CA1 hippocampal spiny synapses are formed by GluR1 (red bars) self-associating into homomers or in association with GluR2 (black bars) into heteromers. After sNS, adult CA1 synapses have less of their total GluR1 expressed on the surface. As a result of this proportional shift after sNS, GluR1s are more likely to be associated with GluR2s, but less total receptors are suggested to be present. Because of equal excitability compared with naive synapses, it is postulated that potentiated GluR1s (asterisk) are stabilized by more relative PSD-95 (filled spheres). By applying subunit rules (GluR1s mediate LTP and GluR2s drag along GluR1s for LTD) and noting the loss of NR2A (activation favors LTP) and proportional gain of NR2B (activation favors LTD) after sNS, these synapses have less LTP and more LTD. 422 Annals of Neurology Vol 61 No 5 May 2007 Alternatively, one might expect greater potential movement of GluR1 to the membrane with LTP (see Fig 8, “rule 3”); however, this is a feature of only immature synapses62 and not adult synapses.48 Even after sNS the intracellular pool size is only 10% of the total (see Fig 6A), thus permitting only limited LTP from this pool, if it were involved. Nevertheless, any further movement of GluR1s into the membrane may be prevented by the overexpression of PSD-95.59 Decreased activation of NR2A because of its 28% loss also likely contributes to the inability to induce LTP after sNS based on pharmacological63,64 and biochemical data.65 Because of the intracellular shift of GluR1, remaining membrane-bound GluR1s are more likely to be associated with GluR2 after sNS. Because GluR2s drag along GluR1s with LTD, greater LTD would be predicted in adult rats that experienced sNS (see Fig 8, “rule 1”). Conversely, in naive synapses, some GluR1s are likely to be unassociated with GluR2, and as a result (see Fig 8, “rule 2”), these independent GluRs will not be removed, resulting in less LTD in naive synapses. PSD-95 overexpression favors greater LTD59 after sNS. It is also possible that there is a greater activation of NR2B both due to the 28% loss of NR2A yet stable expression of NR1 and the increased expression of PSD-95 (which favors synaptic clustering of NR2B50) after sNS. Greater activation of NR2B would favor more LTD induction63,64 as a result of sNS. Our data are consistent with enhanced LTD and absent LTP underlying impaired hippocampal-dependent memory and supported by multiple lines of evidence. Saturated LTP has been shown to impair hippocampaldependent learning.66 The RAWM has been shown to be hippocampal dependent and capable of testing memory.67 NMDA-R–mediated hippocampal LTP underlies spatial learning.41,68 However, when NMDA-R were blocked and episodic-like memory was specifically impaired, other plasticity mechanisms substituted (as they likely did here) to allow MWM performance.29 Under certain conditions, NMDA-R–independent forms of LTP induction are observed to coexist.69 In future studies, it will be important to determine whether these can be activated or enhanced to improve learning and memory. We did not find any evidence of synaptic reorganization or spine loss. It is possible that more subtle alterations in spine morphology occurred that could not be detected with the techniques used. Severe cognitive dysfunction, altered histology, and impaired plasticity became apparent only after multiple episodes of earlylife seizure activity.5–7 Similarly, in other mouse models of developmental syndromes with learning impairment, altered spines were associated with either abnormal postsynaptic responsiveness45 or more severe learning impairment in the MWM,70 neither of which we observed. In a previous study of a single episode of early-life seizures, increased inhibitory synaptic transmission was demonstrated to mediate impaired LTP in the perforant path and DG.13 Similarly, in the Ts65Dn mouse model of Down’s syndrome, increased inhibition has been shown to mediate impaired LTP.71 Notably, this was found primarily in the DG, whereas the effect was much less pronounced in CA1,71 as studied here. Altered inhibition in CA1 after a single episode of early-life seizures could contribute and partially explain some of our findings. However, the pronounced changes in glutamate receptors and PSD-95 (which is not found in inhibitory synapses50) argue that the altered plasticity and memory that we observe are substantially mediated by alterations at glutamatergic synapses. The effects of a single episode of early-life seizures in rodent models are interrelated to the severity of the seizures, developmental age, and technique. More severe and more prolonged seizures caused by bicuculline at P4,72 direct intrahippocampal injection of kainate at P7,73 intraperitoneal injection of lithium pilocarpine at P1274 and up,75 prolonged electrical stimulation after P14,76 or kainate at older ages51 are each associated with overt cellular damage. Thus, our current demonstration that a single episode of relatively mild seizures caused substantial, permanent functional alterations without cellular damage now establishes with prior studies13 the insidious nature of early-life seizures, but uniquely implicates the damage to dysfunctional glutamatergic synapses. It is important to distinguish whether we were inducing status epilepticus. Neonatal status epilepticus has been found to be associated with an adverse clinical outcome.3 Although this has been challenged,77 recent studies suggest that outcome may be poor regardless of classification.78 We do not think that the term status epilepticus should be applied here for several reasons. First, defining status epilepticus has been frequently debated. In neonatal humans, the current definition is for either: (1) electrical seizures lasting longer than 30 minutes, or (2) greater than 50% of the recording period occupied by ictal discharges.1,79,80 In rodents, discharges longer than 20 minutes have been taken as evidence of status epilepticus.11 In contrast, in adult humans, discharges as brief as 5 minutes have been considered to be a sign of status epilepticus.25 Second, many resist applying the term status epilepticus to neonatal seizures, because in humans during this developmental time period, the level of consciousness cannot be determined interictally.1,2 Even when human neonatal seizures are prolonged, as is often the case clinically, the term status epilepticus “should not be applied” and a series of electrical seizures should be called a single clinical seizure.2 We suggest that our findings represent a rodent model of a single episode of relatively Cornejo et al: Neonatal Seizures Alter Synapses 423 mild human neonatal seizures because the seizure duration in our hands does not experimentally fit a criterion of neonatal status epilepticus. A better understanding of the developmental mechanisms of excitatory synaptic function that are both influenced and subsequently altered by sNS is necessary. Recent studies suggest that our results at P7 could be uniquely applicable to a developmental window equivalent to human neonatal seizures clinically,2 biochemically,23 and electrographically.24 As monitoring studies have suggested that sNS induced by kainate does not result in an epileptic state,9,11 we assume that our findings resulted from a single episodic insult. As we have shown, novel approaches are required to begin to address effective treatment24 because conventional therapy24,81 is also inadequate clinically.82 Additional studies might be necessary to clarify the “dose–response” relation between seizure severity and long-term outcome. Nevertheless, given the insidious problem of neonatal seizure detection,78 our findings might suggest that neonates at-risk for seizures be prophylactically treated. However, defining risk and balancing outcome with the potentially adverse consequences of treatment83 requires further study. Through these additional characterizations, we expect that clinically relevant interventions to prevent the effects of sNS will be forthcoming. This work was supported by the NIH (National Institute of Mental Health/APA Diversity Program in Neuroscience, MH18882-17, B.J.C.; National Institute of Neurological Disorders and Stroke, NS041267, T.A.B.). We acknowledge E. Stubblefield, K. Manning, C. Algeier, Drs K. Staley, M. Dell’Acqua, V. Dzhala, C. Adams, and J. Yonchek for their contributions. References 1. Mizrahi E. Acute and chronic effects of seizures in the developing brain: lessons from clinical experience. Epilepsia 1999; 40(suppl 1):S42–S50. 2. Arzimanoglou A, Guerrini R, Aicardi J. Neonatal seizures. Aicardi’s epilepsy in children. 2nd ed. Philadelphia: Lippincott Williams & Wilkins, 2004:191–192. 3. Ortibus EL, Sum JM, Hahn JS. Predictive value of EEG for outcome and epilepsy following neonatal seizures. Electroencephalogr Clin Neurophysiol 1996;99:175–185. 4. Burke M. Education commission of the States, report on Special Education-Finance. 2003. www.ecs.org/clearinghouse/48/ 92/4892.htm. 5. Sarkisian MR, Tandon P, Liu Z, et al. Multiple kainic acid seizures in the immature and adult brain: ictal manifestations and long-term effects on learning and memory. Epilepsia 1997; 38:1157–1166. 6. Chang YC, Kuo YM, Huang AM, Huang CC. Repetitive febrile seizures in rat pups cause long-lasting deficits in synaptic plasticity and NR2A tyrosine phosphorylation. Neurobiol Dis 2005;18:466 – 475. 7. Swann JW. The effects of seizures on the connectivity and circuitry of the developing brain. Ment Retard Dev Disabil Res Rev 2004;10:96 –100. 424 Annals of Neurology Vol 61 No 5 May 2007 8. Nitecka L, Tremblay E, Charton G, et al. Maturation of kainic acid seizure-brain damage in the rat. II. Histopathological sequelae. Neuroscience 1984;13:1073–1094. 9. Stafstrom CE, Chronopoulos A, Thurber S, et al. Agedependent cognitive and behavioral deficits after kainic acid seizures. Epilepsia 1993;34:420 – 432. 10. Holmes GL, Gairsa J-L, Chevassus-Au-Louis N, Ben-Ari Y. Consequences of neonatal seizures in the rat: morphological and behavioral effects. Ann Neurol 1998;44:845– 857. 11. Stafstrom CE, Thompson JL, Holmes GL. Kaininc acid seizures in the developing brain: status epilepticus and spontaneous recurrent seizures. Brain Res Dev Brain Res 1992;21: 227–236. 12. Haas KZ, Sperber EF, Opanashuk LA, et al. Resistance of immature hippocampus to morphological and physiological alterations following status epilepticus and kindling. Hippocampus 2001;11:615– 625. 13. Lynch M, Sayin U, Bownds J, et al. Long-term consequences of early postnatal seizures on hippocampal learning and plasticity. Eur J Neurosci 2000;12:2252–2264. 14. Dingledine R, Borges K, Bowie D, Traynelis SF. The glutamate receptor ion channel. Pharmacol Rev 1999;51:7– 61. 15. Bliss TV, Collingridge GL. A synaptic model of memory: longterm potentiation in the hippocampus. Nature 1993;361: 31–39. 16. Kullmann DM, Nicoll RA. Long-term potentiation is associated with increases in quantal content and quantal amplitude. Nature 1992;357:240 –244. 17. Benke TA, Luthi A, Isaac JTR, Collingridge GL. Modulation of AMPA receptor unitary conductance by synaptic activity. Nature 1998;395:793–797. 18. Collingridge GL, Isaac JTR, Wang YT. Receptor trafficking and synaptic plasticity. Nat Rev Neurosci 2004;5:952–962. 19. Abraham WC, Bear MF. Metaplasticity: the plasticity of plasticity. Trends Neurosci 2006;19:126 –130. 20. Lee SH, Simonetta A, Sheng M. Subunit rules governing the sorting of internalized AMPA receptors in hippocampal neurons. Neuron 2004;43:221–236. 21. Tremblay E, Nitecka L, Berger ML, Ben-Ari Y. Maturation of kainic acid seizure-brain damage syndrome in the rat. I. Clinical, electrographic and metabolic observations. Neuroscience 1984;13:1051–1072. 22. Yager JY, Armstrong EA, Miyashita H, Wirrell EC. Prolonged neonatal seizures exacerbate hypoxic-ischemic brain damage: correlation with cerebral energy metabolism and excitatory amino acid release. Dev Neurosci 2002;24:367–381. 23. Talos DM, Follet PL, Folkerth RD, et al. Developmental regulation of AMPA receptor subunit expression in forebrain and relationship to regional susceptibility to hypoxic/ischemic injury: Part II. Human cerebral white matter and cortex. J Comp Neurol 2006;497:61–77. 24. Dzhala VI, Talos DM, Sdrulla DA, et al. NKCC1 transporter facilitates seizures in the developing brain. Nat Med 2005;11: 1205–1213. 25. Lowenstein DH. Status epilepticus: an overview of the clinical problem. Epilepsia 1999;40(suppl 1):S3–S8. 26. Morris RG, Garrud P, Rawlins JN, et al. Place navigation impaired in rats with hippocampal lesions. Nature 1982;297: 681– 683. 27. Hyde LA, Hoplight BJ, Denenberg VH. Water version of the radial-arm maze: learning in three inbred strains of mice. Brain Res 1998;785:236 –244. 28. Mesches MH, Gemma C, Veng LM, et al. Sulindac improves memory and increases NMDA receptor submints in aged Fischer 344 rats. Neurobiol Aging 2004;25:315–324. 29. Morris RGM, Frey U. Hippocampal synaptic plasticity: role in spatial learning or the automatic recording of attended experience. Philos Trans R Soc Lond B Biol Sci 1997;352: 1489 –1503. 30. Sloviter RS. A simplified Timm’s stain procedure compatible with formaldehyde fixation and routine paraffin embedding of rat brain. Br Res Bull 1982;8:771–774. 31. Holmes GL, Sarkisian M, Ben-Ari Y, Chevassus-Au-Louis N. Mossy fiber sprouting after recurrent seizures during early development in rats. J Comp Neurol 1999;404:537–553. 32. Jiang M, Lee CL, Smith KL, Swann JW. Spine loss and other persistent alterations of hippocampal pyramidal cell dendrites in a model of early-onset epilepsy. J Neurosci 1998;18: 8356 – 8368. 33. Kuenzi FM, Fitzjohn SM, Morton RM, et al. Reduced longterm potentiation in hippocampal slices prepared using sucrosebased artificial cerebrospinal fluid. J Neurosci Methods 2000; 100:117–122. 34. The LTP program: a data acquisition program for on-line analysis of long-term potentiation and other synaptic events. J Neurosci Methods 2001;108:71– 83. 35. Clark KA, Randall AD, Collinridge GL. A comparison of paired-pulse facilitation of AMPA and NMDA receptormediated excitatory postsynaptic currents in the hippocampus. Exp Brain Res 1994;101:272–278. 36. Alvestad RM, Grosshans DR, Coultrap SJ, et al. Tyrosine dephosphorylation and ethanol inhibition of NMDA receptor function. J Biol Chem 2003;278:11020 –11025. 37. Grosshans DR, Clayton DA, Coultrap SJ, Browning MD. Analysis of glutamate receptor surface expression in acute hippocampal slices. Sci STKE 2002;137:8. 38. Dunah AW, Standaert DG. Dopamine D1 receptor dependent trafficking of striatal NMDA glutamate receptors to the postsynaptic membrane. J Neurosci 2001;21:5546 –5558. 39. Grosshans DR, Browning MD. Protein kinase C activation induces tyrosine phosphorylation of the NR2A and NR2B subunits of the NMDA receptor. J Neurochem 2001;76:737–744. 40. Snell LD, Nunley KR, Lickteig R, et al. Regional and subunit changes in NMDA receptor mRNA and immunoreactivity in mouse brain following chronic ethanol ingestion. Brain Res Mol Brain Res 1996;40:71–78. 41. Morris RGM, Anderson E, Lynch GS, et al. Selective impairment of learning and blockade of long-term potentiation by an N-methyl-D-aspartate receptor antagonist, AP5. Nature 1986; 319:774 –776. 42. Anderson AE, Hrachovy RA, Antalffy BA, et al. A chronic focal epilepsy with mossy fiber sprouting follows recurrent seizures induced by intrahippocampal tetanus toxin injection in infant rats. Neuroscience 1999;92:73– 82. 43. Brunson KL, Kramar E, Lin B, et al. Mechanisms of late-onset cognitive decline after early-life stress. J Neurosci 2005;25: 9328 –9338. 44. Isokawa M. Remodeling dendritic spines of dentate granule cells in temporal lobe epilepsy patients and the rat pilocarpine model. Epilepsia 2000;41(suppl):S14 –S17. 45. Moretti P, Levenson JM, Battaglia F, et al. Learning and memory and synaptic plasticity are impaired in a mouse model of Rett syndrome. J Neurosci 2006;26:319 –327. 46. Purpura DP. Dendritic spine “dysgenesis” and mental retardation. Science 1974;186:1126 –1128. 47. Bashir ZI, Jane DE, Sunter DC, et al. Metabotropic glutamate receptors contribute to the induction of long-term depression in the CA1 region of the hippocampus. Eur J Pharmacol 1993; 239:265–266. 48. Grosshans DR, Clayton DA, Coultrap SJ, Browning MD. LTP leads to rapid surface expression of NMDA but no AMPA receptors in adult rat CA1. Nat Neurosci 2002;5:27–33. 49. Allison DM, Gelfand VI, Spector I, Craig AM. Role of actin in anchoring postsynaptic receptors in cultured hippocampal neurons: differential attachment of NMDA versus AMPA receptors. J Neurosci 1998;18:2423–2436. 50. Kennedy MB. The postsynaptic density at glutamatergic synapses. Trends Neurosci 1997;20:264 –268. 51. Stafstrom CE. Assessing the behavioral and cognitive effects of seizures on the developing brain. Prog Brain Res 2002;135: 377–390. 52. Reisel D, Bannerman DM, Schmitt WB, et al. Spatial memory dissociations in mice lacking GluR1. Nat Neurosci 2002;5: 868 – 873. 53. Lee HK, Takamiya K, Han J-S, et al. Phosphorylation of the AMPA receptor GluR1 subunit is required for synaptic plasticity and retention of spatial memory. Cell 2003;112:631– 643. 54. Wirrell EC, Armstrong EA, Osman LD, Yager JY. Prolonged seizures exacerbate perinatal hypoxic-ischemic brain damage. Pediatr Res 2001;50:445– 454. 55. Ehlers MD. Reinsertion or degradation of AMPA receptors determined by activity-dependent endocytic sorting. Neuron 2000;28:511–525. 56. Hayashi Y, Shi S-H, Esteban JA, et al. Driving AMPA receptors into synapses by LTP and CaMKII: requirement for GluR1 and PDZ domain interaction. Science 2000;287:2262–2267. 57. Snyder EM, Colledge M, Crozier RA, et al. Role of A kinaseanchoring proteins (AKAPs) in glutamate receptor trafficking and long term synaptic depression. J Biol Chem 2005;280: 16962–16968. 58. Beique J-C, Andrade R. PSD-95 regulates synaptic transmission and plasticity in rat cerebral cortex. J Physiol 2003;546.3: 859 – 867. 59. Stein V, House DRC, Bredt DS, Nicoll RA. Postsynaptic density-95 mimics and occludes hippocampal long-term potentiation and enhances long-term depression. J Neurosci 2003;23: 5503–5506. 60. Ehrlich I, Malinow R. Postsynaptic density 95 controls AMPA receptor incorporation during long-term potentiation and experience driven synaptic plasticity. J Neurosci 2004;24:916 –927. 61. Ritter LM, Vazquez DM, Meador-Woodruff JH. Ontogeny of ionotropic glutamate receptor subunit expression in the rat hippocampus. Dev Brain Res 2002;139:227–236. 62. Palmer MJ, Isaac JTR, Collinridge GL. Multiple, developmentally regulated expression mechanisms of long-term potentiation at CA1 synapses. J Neurosci 2004;24:4903– 4911. 63. Liu L, Wong TP, Pozza MF, et al. Role of NMDA receptor subtypes in governing the direction of hippocampal plasticity. Science 2004;304:1021–1024. 64. Massey PV, Johnson BE, Moult PR, et al. Differential roles of NR2A and NR2B-containing NMDA receptors in cortical long-term potentiation and long-term depression. J Neurosci 2004;24:7821–7828. 65. Kim MJ, Dunah AW, Wang YT, Sheng M. Differential roles of NR2A- and NR2B-containing NMDA receptors in Ras-ERK signaling and AMPA receptor trafficking. Neuron 2005;46: 745–760. 66. Moser EI, Krobert KA, Moser MB, et al. Impaired spatial learning after saturation of long-term potentiation. Science 1998;281:2038 –2042. 67. Hodges H. Maze procedures: the radial-arm and water maze compared. Cogn Brain Res 1996;3:167–181. 68. Davis S, Butcher SP, Morris RGM. The NMDA receptor antagonist D-2-amino-5-phosphonopentanoate (D-AP5) impairs spatial learning and LTP in vivo at intracerebral concentrations comparable to those that block LTP in vitro. J Neurosci 1992; 12:21–34. Cornejo et al: Neonatal Seizures Alter Synapses 425 69. Raymond CR, Redman SJ. Spatial segregation of neuronal calcium signals encodes different forms of LTP in rat hippocampus. J Physiol 2006;570.1:97–111. 70. Stasko MR, Costa ACS. Experimental parameters affecting the Morris water maze performance of a mouse model of Down syndrome. Behav Brain Res 2004;154:1–17. 71. Kleschevnikov AM, Belichenko PV, Villar AJ, et al. Hippocampal long-term potentiation suppressed by increased inhibition in the Ts65Dn mouse, a genetic model of Down syndrome. J Neurosci 2004;24:8153– 8160. 72. Wasterlain C. Effects of neonatal status epilepticus on rat brain development. Neurology 1976;26:975–986. 73. Leite JP, Babb TL, Pretorius JK, et al. Neuron loss, mossy fiber sprouting, and interictal spikes after intrahippocampal kainate in developing rats. Epilepsy Res 1996;26:219 –231. 74. Kubova H, Mares P, Suchomelova L, et al. Status epilepticus in immature rats leads to behavioral and cognitive impairment and epileptogenesis. Eur J Neurosci 2004;19:3255–3265. 75. Sankar R, Shin DH, Liu H, et al. Patterns of status epilepticusinduced neuronal injury during development and long-term consequences. J Neurosci 1998;18:8382– 8393. 426 Annals of Neurology Vol 61 No 5 May 2007 76. Thompson K, Holm AM, Schousboe A, et al. Hippocampal stimulation produces neuronal death in the immature brain. Neuroscience 1998;82:337–348. 77. Bye AM, Cunningham CA, Chee KY, et al. Outcome of neonates with electrographically identified seizures or at risk of seizures. Pediatr Neurol 1997;57:225–231. 78. McBride MC, Laroia N, Guillet R. Electrographic seizures in neonates correlate with poor neurodevelopmental outcome. Neurology 2000;55:506 –513. 79. Wical BS. Neonatal seizures and electrographic analysis: evaluation and outcomes. Pediatr Neurol 2006;10:271–275. 80. Volpe JJ. Neonatal seizures. Neurology of the newborn. 4th ed. Philadelphia: W.B. Saunders, 2000:178 –214. 81. Mares P, Folgergrova J, Kubova H. Excitatory amino acids and epileptic seizures. Physiol Res 2004;53(suppl 1):S115–S124. 82. Painter MJ, Scher MS, Stein AD, et al. Phenobarbital compared with phenytoin for the treatment of neonatal seizures. N Engl J Med 1999;341:485– 489. 83. Bittigau P, Sifringer M, Ikonomidou C. Antiepileptic drugs and apoptosis in the developing brain. Ann N Y Acad Sci 2003;993: 103–114.