Depressed responses to applied and synaptically-released GABA in CA1 pyramidal cells, but not in CA1 interneurons, after transient forebrain ischemia

Depressed responses to applied andsynaptically-released GABA in CA1 pyramidalcells, but not in CA1 interneurons, after transientforebrain ischemia

Ren-Zhi Zhan, J Victor Nadler and Rochelle D Schwartz-Bloom

Department of Pharmacology and Cancer Biology, Duke University Medical Center, Durham, North Carolina,USA

Transient cerebral ischemia kills CA1 pyramidal cells of the hippocampus, whereas most CA1interneurons survive. It has been proposed that calcium-binding proteins, neurotrophins, and/orinhibitory neuropeptides protect interneurons from ischemia. However, different synapticresponses early after reperfusion could also underlie the relative vulnerabilities to ischemia ofpyramidal cells and interneurons. In this study, we used gramicidin perforated patch recording in exvivo slices to investigate c-aminobutyric acid (GABA) synaptic function in CA1 pyramidal cells andinterneurons 4 h after a bilateral carotid occlusion accompanied by hypovolemic hypotension. Atthis survival time, the amplitudes of both miniature inhibitory postsynaptic currents (mIPSCs) andGABA-evoked currents were reduced in CA1 pyramidal cells, but not in CA1 interneurons. Inaddition, the mean rise time of mIPSCs was reduced in pyramidal cells. The reversal potential for theGABA current (EGABA) did not shift toward depolarizing values in either cell type, indicating that thedriving force for chloride was unchanged at this survival time. We conclude that early duringreperfusion GABAergic neurotransmission is attenuated exclusively in pyramidal neurons. This islikely explained by reduced GABAA receptor sensitivity or clustering and possibly also reducedGABA release, rather than by an elevation of intracellular chloride. Impaired GABA function maycontribute to ischemic neuronal death by enhancing the excitability of CA1 pyramidal cells andfacilitating N-methyl-D-aspartic acid channel opening. Therefore, normalizing GABAergic functionmight be a useful pharmacological approach to counter excessive, and potentially excitotoxic,glutamatergic activity during the postischemic period.Journal of Cerebral Blood Flow & Metabolism (2006) 26, 112–124. doi:10.1038/sj.jcbfm.9600171; published online 15June 2005

Keywords: chloride; EGABA; gramicidin; GABAA receptor; hippocampus; IPSC; perforated patch

Introduction

Transient cerebral ischemia kills neurons in vulner-able regions of the brain, including the hippocampus,striatum, and somatosensory cortex (Pulsinelli et al,1982; Crain et al, 1988). Within the hippocampus ofboth humans and rodents, area CA1 is particularlysensitive to cerebral ischemia (Pulsinelli et al, 1982;Kirino, 1982; Petito et al, 1987). The pyramidal cellsdegenerate within a few days after cerebral ischemia,whereas most of the interneurons remain intact

(Johansen et al, 1983; Schlander et al, 1988; Nitschet al, 1989; Tortosa and Ferrer, 1993). However, someinterneurons do exhibit degenerative changes andmorphological abnormalities within weeks or monthsafter ischemia (Fukuda et al, 1993; Inglefield et al,1997; Arabadzisz and Freund, 1999). Various mechan-isms have been proposed to account for the relativeresistance of CA1 interneurons to transient ischemia,including (1) the presence of Ca2þ -buffering proteins(e.g., parvalbumin) within interneurons (Freund et al,1990), (2) the expression of inhibitory neuropeptidessuch as somatostatin (Bering et al, 1997) and (3)increased neurotrophin signaling within interneurons(Larsson et al, 2001). Although these factors might beinvolved, they do not explain interneuron resistancecompletely. For example, parvalbumin-containinginterneurons within the striatum degenerate aftertransient cerebral ischemia (Larsson et al, 2001).

Received 11 December 2004; revised 12 April 2005; accepted 13May 2005; published online 15 June 2005

Correspondence: Dr RD Schwartz-Bloom, Department of Pharma-cology and Cancer Biology, Duke University Medical Center, Box3813, Durham, NC 27710, USA. E-mail: [email protected]

This work was supported by NIH grant NS28791 (RDS).

Journal of Cerebral Blood Flow & Metabolism (2006) 26, 112–124& 2006 ISCBFM All rights reserved 0271-678X/06 $30.00

www.jcbfm.com

Changes in synaptic responses early after anischemic insult may explain, at least partially, thedifference in vulnerability between CA1 pyramidalcells and interneurons. Enhanced excitatory trans-mission in area CA1 has been observed within a fewhours after ischemia, along with impairment ofCA1b pyramidal cell excitability (Urban et al,1989; Shinno et al, 1997; Gao et al, 1998; Mitaniet al, 1998). Decreases in inhibitory transmissionhave been observed in neocortex 10 months aftercerebral ischemia (Luhmann et al, 1995), but, to ourknowledge, inhibitory synaptic responses have notbeen examined in hippocampal pyramidal cellsearly after ischemia nor have either the excitatoryor inhibitory synaptic responses of CA1 interneur-ons. Synaptic responses in area CA1 interneuronshave been measured during anoxia in vitro; bothexcitatory and inhibitory synaptic responsesare depressed, but recover within 10 mins (Khazipovet al, 1995).

Interneurons within hippocampal area CA1 pro-vide inhibitory GABA innervation to pyramidalcells and other interneurons. It has been hypothe-sized that GABA neurotransmission in hippocampusis reduced early after cerebral ischemia (Li et al,1993; for a review, see Schwartz-Bloom and Sah,2001). In support of this idea, we showed thatGABAA receptors are downregulated in hippocampalarea CA1 within 30 mins after cerebral ischemia(Alicke and Schwartz-Bloom, 1995). Consistentwith this finding, GABA-stimulated chloride influxis reduced in isolated rat forebrain synapto-neurosomes within hours after ischemia (Verheulet al, 1993). Similarly, GABA-stimulated chlorideinflux is reduced after oxygen-glucose deprivation ofhippocampal slices, an in vitro model of cerebralischemia (Inglefield and Schwartz-Bloom, 1998;Galeffi et al, 2004b). This impairment of GABAfunction in hippocampal pyramidal cells wasassociated with an increase in intracellular chloride.If intracellular chloride increases after cerebralischemia in vivo, one would expect to see a positiveshift in the reversal potential for GABA-evokedcurrents (EGABA).

The objective of this study was to determine ifcerebral ischemia alters GABA function in pyrami-dal neurons and/or interneurons of area CA1 withina few hours after transient cerebral ischemia. Wealso determined whether any impairment of GABAA

receptor-mediated responses could be explained bya reduced driving force for chloride. To preserve thechloride gradient across the plasma membrane,recordings were made with the gramicidin-basedperforated patch technique (Akaike, 1996).

Materials and methods

Forebrain Ischemia

All experiments were performed in accordance withthe National Institutes of Health Guide for the Care

and Use of Laboratory Animals and were approved bythe Duke University Institutional Animal Care andUse Committee.

Transient forebrain ischemia was produced using two-vessel occlusion, as described by Zhan et al (2001).Briefly, adult male Sprague–Dawley rats (Charles River,Raleigh, NC, USA) that weighed 160 to 200 g (42 to 47 daysold) were fasted for 7 to 10 h before surgery, but allowedfree access to water. Animals were anesthetized with 2.5%halothane delivered through a face mask and intubatedtracheally; the lungs were ventilated mechanically with30% O2/70% N2O. The inspired concentration of ha-lothane was adjusted to maintain a mean arterial pressurebetween 80 and 100 mm Hg. A digital thermistor probewas inserted into the rectum to record core temperatureand another one was placed under the left temporalismuscle to monitor pericranial temperature. Cortical sur-face electroencephalograph (EEG) was monitored continu-ously during the surgery with active subdermal electrodespositioned over the parietal cortex bilaterally and aground lead inserted into the tail. The left femoral arterywas cannulated for continuous monitoring of bloodpressure and for blood sampling. The right jugular veinwas cannulated with a soft catheter for drug infusionand blood withdrawal. Both common carotid arteries wereisolated from the carotid sheaths via a ventral midlineincision. At 15 min before occlusion of the commoncarotid arteries, a sample of arterial blood (200 mL) waswithdrawn for blood gas analysis and plasma glucoseassay. Mean blood pressure was lowered to 30 to 40 mmHg by injecting 0.4 mg of phentolamine into the arterialcatheter, followed by withdrawal of blood from the jugularvein with a heparinized syringe. Subsequently, bothcommon carotid arteries were occluded with smallvascular clips for 8 mins. During the period of occlusion,the core temperature was controlled at 37.21C70.21C,whereas the pericranial temperature was allowed todecrease gradually from 37.21C70.21C at the beginningof the occlusion to 36.81C70.21C at the end of theocclusion. The clips were then removed and the with-drawn blood was reinfused. Reperfusion in each arterywas verified visually. Metabolic acidosis caused byhypoperfusion was corrected by intravenous administra-tion of 8.4% (w/v) sodium bicarbonate (0.3 ml) immedi-ately after reperfusion. Before discontinuation ofanesthesia, the vascular catheters were removed and thewounds were infiltrated with 1% lidocaine and sutured.The endotracheal tube was removed after recovery ofspontaneous respiration and the righting reflex. Theanimal was then maintained in a warm (301C to 321C)humidified chamber for another 3 h before being returnedto its cage at room temperature. Rats accepted for furtherstudy met the following criteria: (1) blood gas parameterswere within the normal range (pH 7.35 to 7.45, pCO2 35 to45 mm Hg, pO2 100 to 180 mm Hg); (2) plasma glucoseconcentration ranged between 90 and 150 mg/dL; and(3) the EEG flattened immediately after occlusion andremained flattened throughout the entire occlusion peri-od. Rats were killed 4 h later for electrophysiologicalstudies or 7 days later for cell counting and immuno-fluorescence studies.

Reduced GABA responses after ischemiaR-Z Zhan et al

113

Journal of Cerebral Blood Flow & Metabolism (2006) 26, 112–124

Cell Counting and Immunofluorescence

To determine whether hippocampal CA1 interneuronswere more resistant to ischemic damage than pyramidalcells in the relatively young rats used in this study,hippocampal sections were double-stained with antibo-dies against microtubule-associated protein-2 (MAP-2)and parvalbumin, a Ca2þ -binding protein that is expressedin a large population of interneurons in stratum pyrami-dale of area CA1 (Aika et al, 1994; Shetty and Turner,1998). Adjacent sections were stained with cresyl violetfor counting intact neuronal cell bodies.

At 7 days after reperfusion, rats (4 sham-operatedcontrols and 4 rats subjected to transient cerebral ischemia)were anesthetized with sodium pentobarbital (60 mg/kg,intraperitoneally) and perfused through the left ventricle,first with heparinized 0.1 mol/L phosphate-buffered saline(PBS) and then with a fixative that contained 2% (w/v)paraformaldehyde, 0.075 mol/L lysine, 0.01 mol/L sodiumperoxidate and 0.0375 mol/L sodium phosphate, pH 6.8.Brains were removed and tissue blocks that contained theneocortex and hippocampus were cut. Tissue blocks werepostfixed with the same fixative for 4 h. After the blockswere washed in gradually increasing concentrations ofphosphate-buffered sucrose, they were frozen rapidly inethanol cooled with solid CO2. A cryostat was used to cutcoronal sections of 14mm (for immunofluorescence) or30mm (for cresyl violet staining) thickness from the rostralhippocampus as far as 4.16 mm posterior to bregma. Theprotocol for double immunofluorescence staining was asfollows. After sections were exposed to 100% acetone at41C for 60 secs, they were washed with 0.1 mol/L PBS 3times for 5 mins each. Sections were incubated with2% (w/v) IgG-free bovine serum albumin in 0.1 mol/LPBS for 90 mins, followed by an incubation with primaryantibodies at 41C overnight. The concentrations of rabbitpolyclonal anti-MAP-2 (Chemicon, Temecula, CA, USA)and mouse monoclonal antiparvalbumin (Sigma, St Louis,MO, USA) used were 12.5 and 5mg/mL, respectively.Sections were washed 3 times with 0.1 mol/L PBS (10 minseach) and then incubated with a mixture of Alexa Fluor594 goat anti-rabbit IgG (1:400) and Alexa Fluor 488 goatanti-mouse IgG (1:400) for 150 mins at 41C. The twosecondary antibodies were obtained from MolecularProbes (Eugene, OR, USA). After washing, sections weremounted with 75% glycerol in 0.1 mol/L PBS andvisualized under a fluorescence microscope. Images werecaptured with a digital camera controlled by AdobePhotoshop (Adobe, San Jose, CA, USA).

Cell counting was performed in cresyl violet-stainedsections adjacent to those used for immunofluorescence.Intact neuronal cell bodies in stratum pyramidale of thewhole CA1 area were counted in both hemispheres from atleast three sections per rat. The values, expressed as thenumber of cells per millimeter length of the stratumpyramidale, were averaged for each animal.

Preparation of Acute Brain Slices (‘Ex Vivo’)

At 4 h after reperfusion, the animal was reanesthetizedwith halothane and decapitated. The whole head was

immersed immediately in cold (51C to 61C) oxygenatedartificial cerebrospinal fluid (ACSF) and the forebrain wasremoved. Coronal hippocampal slices (400-mm thickness)were prepared with a Vibratome in oxygenated (95% O2/5%CO2) ACSF at 51C to 61C. Slices were cut from the rostralhippocampus as far as 4.16 mm posterior to bregma(Paxinos and Watson, 1986). The ACSF consisted of125 mmol/L NaCl, 2.5 mmol/L KCl, 2 mmol/L CaCl2,1.0 mmol/L MgCl2, 1.25 mmol/L NaH2PO4, 26 mmol/L NaH-CO3, 20 mmol/L D-glucose, and 1 mmol/L ascorbic acid, pH7.4 (saturated with 95% O2/5% CO2). The osmolality wasadjusted to 315 mosm. After a 30-min incubation at 331C,slices were kept in ACSF at room temperature.

Electrophysiological Recording

A slice was transferred to a plexiglas recording chamberand held in place with platinum wires. The slice wassuperfused continuously (2.5 mL/min) with ACSF at roomtemperature (221C to 241C). Pyramidal cells and inter-neurons located within the stratum pyramidale of areaCA1b were identified visually based on their size, shape,location, and dendritic morphology under a Nikon EclipseE600FN microscope equipped with infrared-differentialinterference contrast (IR-DIC) optics, a � 40 water immer-sion objective, and epifluorescence illumination (NikonInc., Melville, NY, USA). Images were captured with aninfrared charge-coupled device (CCD) video camera (IR-1000; DAGE MTI, Michigan City, IN, USA) and displayedon a monitor.

Patch electrodes were pulled from borosilicate glass (OD:1.5mm; ID: 1.1mm; Sutter Instruments, Novato, CA, USA)with a Flaming/Brown electrode puller (Sutter Instruments,Novato, CA, USA). Pipettes with resistances of 4 to 6 MOwere filled with an internal solution that contained136mmol/L CsCl, 1mmol/L MgCl2, 0.1mmol/L CaCl2,1mmol/L Ethylene glycol-bis(2-aminoethylether)-N,N,N 0,N 0

-tetraacetic acid (EGTA), 10mmol/L N-(2-Hydroxyethyl)piperazine-N0-(2-ethanesulfonic acid) (HEPES), 2mmol/Ladenosine 50-triphosphate (ATP) tris salt, and 0.4mmol/Lguanosine 50-triphosphate (GTP) tris salt, pH 7.30. Theosmolality ranged between 294 and 297mosm. To visualizethe recorded cells, Alexa Fluor 488 hydrazide (MolecularProbes, Eugene, OR, USA) was added to the internal solutionat a concentration of 0.003% (w/v). For perforated-patchrecordings, patch electrodes were immersed in the gramici-din-free internal solution for 1min and then back-filled withthe same solution that contained 40mg/mL of gramicidin.Gramicidin was dissolved in methanol (5mg/mL), mixed for1min, and sonicated for 2mins before being diluted into theinternal solution. The gramicidin-containing solutions wereused within 90mins after preparation.

Signals were recorded with an Axopatch 200B amplifier(Axon Instruments, Union City, CA, USA). After a gigasealwas formed, the progress of perforation was monitoreduntil the access resistance dropped below 100 MO. Seriesresistance (20 to 30 MO) was compensated to 90%. Seriesresistance was monitored throughout the experiment toensure its constancy (715%). The resting membranepotential was recorded when the access resistance was


114


B100 MO and was corrected for a 4 mV liquid junctionpotential. Recordings were made only from neurons with aresting membrane potential that exceeded �50 mV. Min-iature inhibitory postsynaptic currents (mIPSCs) andresponses to applied GABA were recorded at a holdingpotential of 0 mV in the presence of 0.5mmol/L tetrodotox-in (TTX), 100mmol/L DL-2-amino-5-phosphonopentanoicacid (DL-AP5), 10 mmol/L 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX), and 1 mmol/L CGP 55845 to block voltage-gated sodium channels, ionotropic glutamate receptors,and GABAB receptors, respectively. After recordingmIPSCs during a 2.5-mins period, GABA (100 mmol/L)was added to the superfusion medium for 90 secs and thechange in membrane current was monitored.

In another set of experiments, EGABA was determined byapplying GABA to the recorded cells at holding potentialsfrom �100 to 0 mV in 10 mV increments. Each holdingpotential was maintained for 4 secs, during which GABA(100 mmol/L) was delivered with a Picospritzer (GeneralValve, Fairfield, NJ, USA) for 10 ms beginning 250 ms afterthe change of holding potential. Pressure ejection pipetteshad a tip resistance of 3 to 5 MO and were placed 20 to30 mm from the somata of recorded cells.

At the end of each recording, the membrane was rupturedto allow Alexa Fluors 488 access to the cytoplasm, and therecorded cell was visualized at an emission wavelength of535 nm to confirm its identity 15 to 20 mins later. Cells werediscarded if the patch broke during the recording, asindicated by the appearance of intracellular fluorescenceor a sudden drop in EGABA. Additionally, data were notincluded from cells that could not be identified morpholo-gically as either a pyramidal cell or an interneuron.

Data Acquisition

Data were acquired using an IBM-compatible computerequipped with a Digidata 1322 interface and pClamp 8.1software (Axon Instruments, Union City, CA, USA).Whole-cell currents were filtered at 1 to 2 kHz anddigitized at 20 kHz for off-line analysis with pClamp 8.1and/or MiniAnalysis 5.2.8 (Synaptosoft Inc., Decatur, GA,USA). Miniature inhibitory postsynaptic currents werefirst screened automatically with a set of prespecifiedparameters and then accepted or rejected manually withan event detection amplitude threshold at 5 pA (Molnarand Nadler, 2001). Current amplitudes were measured attheir absolute maximum after subtraction of baselinenoise. The frequency of mIPSCs in a given cell wascalculated from the number of events over a 2.5-minsperiod. To construct cumulative probability plots, mIPSCrecordings within groups were pooled together and thenreanalyzed with MiniAnalysis. EGABA was determined inindividual cells from the I/V curve by a least-squaresregression. The intracellular chloride concentration wascalculated from the Nernst equation.

Statistical Analyses

Data are presented as means7s.e.m. unless otherwisespecified. Data were analyzed with two-way and one-way

analyses of variance (ANOVAs) with repeated measures,followed by Fisher’s protected-least-significant-difference(PLSD) test when interactions of main effects weresignificant. When only two groups of means werecompared, a Student’s t-test was used. The Kolmogorov–Smirnov test was used to compare the cumulativeprobability curves for amplitude, charge transfer, decaytime constant, 10% to 90% rise time, and intereventinterval. Values were considered statistically significant atPo0.05.

Materials

g-Aminobutyric acid, IgG-free bovine serum albumin,gramicidin, DL-lysine, paraformaldehyde, and sodium per-oxidate were obtained from Sigma Chemical Co. (St Louis,MO, USA). DL-2-amino-5-phosphonopentanoic acid, CGP55845, CNQX, and TTX were purchased from TocrisCookson Inc. (Ellisville, MO, USA).

Results

Relative Sensitivity to Ischemic Damage of PyramidalCells and Interneurons in Area CA1

In the present study, rats younger than thosenormally used in studies of cerebral ischemia (42to 47 days postnatal) were required so that gramici-din-based perforated patch recordings could bemade. Similar to the extent of neuronal damagecaused by 2-vessel occlusion for 8 mins in older rats(Zhan et al, 2002), cresyl violet staining revealedthat approximately 90% of neurons in the stratumpyramidale of area CA1 had degenerated 7 days afterreperfusion. The number of intact neurons per mmlength of stratum pyramidale was reduced signifi-cantly from 26174 in 4 sham-operated animals to 2572 in rats subjected to transient cerebral ischemia(Po0.001, unpaired Student’s t-test). To confirmthat interneurons were more resistant to ischemicdamage than pyramidal cells in area CA1, we used adouble-labeling approach with MAP-2 and parval-bumin antibodies to distinguish the loss of pyra-midal neurons from the loss of interneurons,respectively. As shown in Figure 1, both pyramidalcells and interneurons were stained with an anti-MAP-2 antibody. At 7 days after reperfusion, therewere only few MAP-2-positive cells remaining inarea CA1 stratum pyramidale. Most of the MAP-2immunolabeled cells were also parvalbumin-posi-tive, indicating that interneurons are more resistantto ischemia than the neighboring pyramidal cells.

GABAergic Responses in CA1 Pyramidal Cellsand Interneurons 4 h after Reperfusion

With the use of IR-DIC visualization and fluorescentimaging, most cells recorded in hippocampal slicesprepared 4 h after ischemia could be identified


115


easily as pyramidal cells or interneurons. Individualpyramidal cells had a triangular soma and a thickapical dendrite that extended into the stratumradiatum (Figure 2). In general, pyramidal cellsrecorded in slices prepared from rats 4 h afterischemia or sham operation exhibited similar cel-lular morphology. Area CA1 interneurons werediverse in their somatic shape and size and in theappearance of major dendrites. The morphology ofinterneurons obtained from animals subjected totransient ischemia was indistinguishable from thatof interneurons in sham-operated rats.

To record GABAergic currents without disruptingthe chloride gradient across the plasma membrane,we made gramicidin perforated patch recordingsfrom the somata of pyramidal cells and interneuronsin area CA1 stratum pyramidale. When the mem-brane potential (Vm) was changed from �60 to 0 mV,a potential considerably more positive than ECl, theoutwardly directed mIPSCs became substantiallylarger (Figure 3). Transient cerebral ischemia alteredmIPSCs in several respects (Figures 3–5, Table 1).First, there was a significant reduction in the meanfrequency (B60%) and mean peak amplitude(B50%) of mIPSCs in pyramidal cells, but not ininterneurons (Table 1). The mean frequency andpeak amplitude differences in pyramidal neuronsare supported by the corresponding cumulativeprobability curves for the interevent interval(a measure of frequency) and the mIPSC amplitude(Figures 4 and 5; Kolmogorov–Smirnov test,Po0.01). The reduced number of mIPSCs in pyr-

amidal cells might be explained either by reducedprobability of GABA release or by the accompanyingreduction in amplitude of mIPSCs. A reduction inamplitude would likely cause some of the smallermIPSCs to become lost in the background noise. Todiscriminate these possibilities, we recalculatedmIPSC frequency in rats subjected to ischemia onthe assumption that ischemia reduced all mIPSCamplitudes by B50%. By increasing the detectionthreshold to 10 pA (twice the standard threshold),the frequency of mIPSCs fell by only B30%. Thisreduction was substantially less than the B60%expected, if reduced amplitude accounted comple-tely for the lower mIPSC frequency observed afterischemia. These considerations led us to concludethat GABA release, in addition to postsynapticGABAA receptor function, was probably impairedby transient cerebral ischemia.

Second, consistent with the reduced amplitude ofthe mIPSCs, the charge transfer per event wassignificantly smaller in pyramidal cells recordedfrom animals subjected to ischemia compared withsham-operated rats (Table 1, Figure 4; Kolmogorov–Smirnov test, Po0.01). Third, 10% to 90% rise timesof mIPSCs in both pyramidal cells and interneuronswere shorter after ischemia (Table 1), but themagnitude of the effect was substantially greater inpyramidal cells (Figures 4 and 5; Kolmogorov–Smirnov test, Po0.01). Fourth, the average decaytime constant was insensitive to ischemia in bothpyramidal cells and interneurons, although thecumulative probability distributions for the decay

Figure 1 CA1 interneurons are more resistant to ischemic damage than surrounding CA1 pyramidal cells. Sections were prepared 7days after reperfusion. Anti-MAP-2 and antiparvalbumin antibodies were used to label all neurons and interneurons, respectively.Note that a large percentage of interneurons survived transient cerebral ischemia, whereas most pyramidal cells were lost. Scalebar¼50 mm.


116


time constant were slightly different in sham-oper-ated rats versus rats subjected to ischemia (Figures 4and 5; Kolmogorov–Smirnov test, Po0.01).

The reduced mean mIPSC amplitude in pyramidalcells suggested impairment of the postsynapticresponse to GABA (Cossart et al, 2001). To test thispossibility, we measured GABA-evoked currents at aholding potential of 0 mV. GABA (100mmol/L) wasapplied to pyramidal cells by addition to thesuperfusion medium for 90 secs. The peak currentdecreased significantly from 138723 pA in sham-operated animals to 5775 pA in animals subjectedto transient cerebral ischemia (Figure 6).

Determination of EGABA

To determine if the reduced GABAergic responsescould be due to a reduced driving force for chloride,we measured EGABA, which reflects the chloridegradient across the plasma membrane. In sham-operated animals, EGABA did not differ betweenpyramidal cells and interneurons (Figure 7). Simi-larly, there was no significant difference in EGABA

between pyramidal cells and interneurons from

animals subjected to transient cerebral ischemia.There were no significant differences in the intra-cellular chloride concentration estimated from theNernst equation (Figure 8).

Discussion

Here we show that transient forebrain ischemiareduces GABA-evoked currents and the meanfrequency, amplitude, and rise time of mIPSCs inCA1 hippocampal pyramidal cells, but not in CA1interneurons, 4 h after reperfusion. The selectivereduction in GABAergic responses parallels thedifferential vulnerability of these neurons to ische-mia-induced cell death. With use of the gramicidinperforated patch technique to measure EGABA, wedetermined that our findings could not be explainedby a reduction in the driving force for chlorideinflux, at least at the level of the soma and at asurvival time of 4 h. We reported previously thatoxygen-glucose deprivation in hippocampal slicesreduces GABAA receptor-mediated responses of CA1pyramidal cells to applied GABA, coincident withan increase in somatic chloride concentration (In-glefield and Schwartz-Bloom, 1998; Galeffi et al,2004b). The difference between our ex vivo and invitro findings may relate to temporal issues. Oxygen-glucose deprivation of hippocampal slices kills CA1pyramidal cells within hours, whereas death ofthese cells is delayed by several days after transientcerebral ischemia. The delayed onset of degenera-tive cellular events in vivo allows time for chloridetransporters to normalize intracellular chlorideconcentration. In addition, elements of the physio-logical milieu present in the intact brain, but notin vitro, may promote restoration of the chloridegradient.

Many investigators have used acutely preparedbrain slices (ex vivo) to examine neuronal electro-physiological properties and intracellular ionicchanges caused by a previous episode of cerebralischemia (Urban et al, 1989, 1990; Kirino et al, 1992;Tsubokawa et al, 1992; Shinno et al, 1997; Taga et al,2000). Compared with in vivo recordings, the ex vivoslice offers experimental advantages, especially theability to use patch-clamp techniques. However, theex vivo approach does have some limitations. First,the process of slice preparation may render theslices temporarily ischemic, and this could affectneurons from the sham-control and ischemia groupsdifferently. Second, some neuronal connections aredisrupted by slicing. Third, superfusion with ACSFalters the environment of both damaged and intactneurons compared with the natural environment ofneurons in vivo. Despite these limitations, severalstudies showed that electrophysiological changesproduced by cerebral ischemia observed in theex vivo slice were comparable to those deter-mined from in vivo recordings. For example, Urbanet al (1989, 1990) first showed that transient

Figure 2 Morphological identification of recorded cells with IR-DIC visualization and Alexa Fluor 488 fluorescent imaging.Infrared-differential interference contrast images (the left ineach panel) were taken within a few minutes after sealformation and fluorescent images were obtained 15 to 20 minsafter membrane rupture. Upper panels: pyramidal cells wereidentified by their triangular soma and thick apical dendrite thatpenetrated the stratum radiatum. Lower panels: interneuronswere identified by their multiple major dendrites and the variedshape and size of their somata. Scale bar¼10 mm.


117


ischemia-enhanced excitatory transmission in CA1pyramidal cells of ex vivo slices is an N-methyl-D-aspartic acid (NMDA) receptor-dependent process.Subsequently, several investigators confirmed thatCA1 excitatory neurotransmission was also en-hanced after cerebral ischemia in the intact brain(Miyazaki et al, 1993; Gao and Xu, 1996; Gao et al,1999). This correspondence predicts that the pyr-amidal cell-selective reduction of GABA transmis-sion observed in the present study also occurs invivo. This hypothesis awaits confirmation.

The reduced mean mIPSC amplitude observed inthe ex vivo slice might be explained either by asmaller postsynaptic response to released GABA orby a reduced vesicular concentration of GABA.However, reduction of the membrane current evokedby application of GABA to CA1 pyramidal cellsimplies the involvement of a postsynaptic mechan-ism, reflecting a reduced GABAA receptor conduc-tance and/or activation of fewer GABAA receptors.These results are consistent with previous reports.GABAA receptors are downregulated (possibly byinternalization) in area CA1 within 30 mins afterglobal cerebral ischemia in the gerbil (Alicke and

Schwartz-Bloom, 1995) and in cultured corticalneurons within hours after oxygen-glucose depriva-tion (Mielke and Wang, 2005). Furthermore, specificGABAA receptor subunits, including a1, a2, a5and g, decline in vulnerable brain regions afterfocal cerebral ischemia in the rat (Redecker et al,2002). The shorter mean 10% to 90% rise timeindicates that the onset kinetics of pyramidal cellmIPSCs are accelerated. This result might beexplained by a change in the relative expression ofGABAA receptor subunits or declustering of GABAA

receptors. With respect to the latter possibility,destabilization of microtubules in cultured hippo-campal neurons resulted in the declusteringof GABAA receptors associated with a reduced risetime (Petrini et al, 2003). Cerebral ischemia causesrapid degradation of microtubule-associated pro-teins (Kitakawa et al, 1989; Matesic and Lin, 1994),which are important in stabilizing microtubules(for a review, see Hirokawa, 1994). Thus, tran-sient cerebral ischemia may lead to declustering ofpostsynaptic GABAA receptors, reducing the num-ber of receptors in close proximity to presynapticactive zones.

Figure 3 Transient cerebral ischemia alters the properties of mIPSCs in CA1 hippocampal pyramidal cells, but not in CA1interneurons, 4 h after reperfusion. Gramicidin perforated patch recordings were made in the presence of 0.5 mmol/L TTX, 20mmol/LCNQX, 100 mmol/L DL-AP5, and 1 mmol/L CGP 55845 at a holding potential of 0 mV. Larger events of mIPSCs were frequently seenin pyramidal cells from sham-operated control rats, but rarely seen in pyramidal cells subjected to transient cerebral ischemia.Ischemia had no such effect in CA1 interneurons. Spontaneous events were recorded from each cell for a 2.5-mins period. Resultsshown were obtained from representative experiments.


118


Other factors that depress GABAA receptor func-tion include elevated intracellular calcium, arachi-donic acid, and oxygen-free radicals (for a review,see Schwartz-Bloom and Sah, 2001; Sah et al, 2002).Transient cerebral ischemia increases all of these.However, it is unclear whether these metabolicevents persist in the ex vivo preparation or reverseonce the hippocampus is removed from the post-ischemic milieu and glutamate receptors are

blocked. Further studies are required to assess thesepossibilities.

EGABA was unchanged 4 h after ischemia, suggest-ing that the transmembrane chloride gradient waspreserved in both CA1 pyramidal cells and inter-neurons at or near the soma. Thus, responses toGABA would be expected to remain hyperpolariz-ing. In contrast, it has been suggested that thechloride gradient might be dissipated or reversed.As stated above, the somatic chloride concentrationincreases after oxygen-glucose deprivation in vitro(Inglefield and Schwartz-Bloom, 1998; Galeffi et al,

Figure 4 Cumulative probability plots of amplitude, chargetransfer area, 10% to 90% rise time, and decay time constantreveal that ischemia had profound effects on miniature IPSCs inpyramidal cells. Recordings in pyramidal cells from sham-operated rats (solid line, n¼6) and from ischemic rats (dottedlines, n¼6) were pooled together and analyzed with MiniAna-lysis. Note that ischemia produced a considerable left shift ofthe amplitude, charge transfer area, 10% to 90% rise time, andinterevent interval probability curves (Kolmogorov–Smirnovtest, Po0.01).

Figure 5 Cumulative probability plots show that cerebralischemia had negligible effects on all electrophysiologicparameters measured in interneurons. Recordings from inter-neurons within stratum pyramidale (area CA1) from sham-operated rats (solid line, n¼6) and from ischemic rats (dottedlines, n¼6) were pooled together and analyzed with MiniAna-lysis.

Table 1 Effects of transient cerebral ischemia on the properties of mIPSCs in CA1 hippocampal pyramidal cells and CA1interneurons

Pyramidal cell Interneuron

Sham-operated Ischemia Sham-operated Ischemia

Frequency (Hz) 1.070.1 0.470.1* 1.470.3 1.070.2Amplitude (pA) 13.671.7 7.370.8* 10.070.8 11.871.2Charge transfer (fC) 201.9722.5 105.9722.9* 207.3744.0 235.6729.310–90% rise time (ms)w 2.970.3 2.370.3 2.770.1 2.470.2Decay time constant (ms)ww 17.470.7 16.471.5 19.572.4 22.471.0

Values are means7s.e.m. for six cells in each group. Data obtained from all mIPSCs were averaged to express a single value for each neuron. *Po0.01compared with pyramidal cells recorded from sham-operated rats, multiple ANOVAs, followed by Fisher’s PLSD post hoc test; wPo0.05, ischemia versussham-operated, 2-way ANOVA; wwPo0.05, pyramidal neurons versus interneurons, 2-way ANOVA.


119


2004b). Increases in extracellular Kþ concentrationor glutamate receptor activation, conditions thatoccur after ischemia, stimulate Naþ/Kþ /2Cl� co-transporter (NKCC-1) activity (Sun and Murali,1998; Su et al, 2000) or reverse the direction of the Kþ /Cl� cotransporter (KCC2) to increase intracellularchloride (Payne et al, 1996; Jarolimek et al, 1999;Kakazu et al, 1999; DeFazio et al, 2000). NKCC-1protein expression and NKCC-1 activity increase ininfarcted brain regions from rats subjected to focalcerebral ischemia (Yan et al, 2001, 2003). Finally,downregulation of KCC2 is associated with in-creased intracellular chloride concentration afteroxygen-glucose deprivation (Galeffi et al, 2004b)and other pathological conditions (Coull et al, 2003;Nabekura et al, 2002; Toyota et al, 2003). We suggestthat an initial postischemic rise in intracellularchloride concentration reverses within 4 h of reper-fusion. At least two factors may contribute to thereversal of high intracellular chloride. Immediatelyafter ischemia, extracellular GABA levels are highenough to saturate GABAA receptors (Globus et al,

1988; Schwartz et al, 1995). However, within 1 h ofreperfusion, extracellular GABA levels normalizedue to restoration of GABA transport. Thus, 4 h afterthe onset of reperfusion, GABAA receptors areprobably unsaturated with agonist, allowing restora-tion of the normal chloride gradient. It is nottechnically feasible to evaluate this hypothesis bymeasuring EGABA earlier than 4 h after ischemiabecause CA1 neurons remain leaky and depolarizedfor some time. It is extremely difficult to form andmaintain a gigaohm seal under these conditions.Another possibility is that the expression/activity ofchloride transporters changes in a direction thatrestores the previously disrupted chloride gradient.In this regard, Kang et al (2002) reported that NKCC-1 immunoreactivity in CA1 pyramidal cells isdownregulated within 12 h after cerebral ischemiain the gerbil.

The lack of change we observed in EGABA does notpreclude the possibility that dendritic chlorideconcentration remained elevated at the 4 h survivaltime, because the membrane potential of distaldendrites is not well controlled by somatic currentinjection. Interestingly, there is a high density ofKCC2 expressed in the vicinity of excitatorysynapses on the distal dendrites in area CA1 (Gulyaset al, 2001). This extrusion mechanism is proposedto counteract the rise in intracellular chlorideassociated with excitation-induced dendritic swel-ling. Recently, we showed that KCC2 is down-regulated in the hippocampus after oxygen-glucosedeprivation in vitro and after cerebral ischemia(Galeffi et al, 2004a, b), possibly limiting its abilityto restore the dendritic chloride gradient. Thus, anattenuated chloride gradient in distal dendritescould explain the reduced amplitude of somemIPSCs 4 h into reperfusion. Theoretically, thispossibility could be investigated by recordingdirectly from the dendrites. However, it is techni-cally very difficult to make gramicidin perforatedpatch recordings from dendrites in acutely preparedhippocampal slices. An alternative approach isrequired.

One of the most interesting findings in the presentstudy was that CA1 interneurons retained normalGABA responses at a time when CA1 pyramidalcells responded poorly to GABA. This differencemay contribute to the differing vulnerability of theseneuronal populations to ischemic damage. Bothglutamate neurotransmission and calcium influxmay play a role. Although some interneuronsexpress more calcium-permeable AMPA receptorsthan pyramidal cells (He et al, 1998; Catania et al,1998), the increase in intracellular calcium inducedby AMPA is smaller in hippocampal interneuronsthan in pyramidal cells (Segal and Greenberger,1992). In addition, pyramidal cells express moreNMDA receptors (Nyıri et al, 2003), and both NMDAcurrents and calcium influx through the NMDAchannel are larger in pyramidal cells than ininterneurons (Avignone et al, 2003). Furthermore,

Figure 6 Transient cerebral ischemia reduces the size ofcurrents evoked by the application of GABA to CA1 pyramidalcells 4 h after reperfusion. Recording conditions were the sameas described in Figure 2. GABA (100 mmol/L) was applied forthe 90-sec period indicated by the horizontal bar above eachtrace. (A): pyramidal cell from a sham-operated rat. (B):pyramidal cell recorded from a rat subjected to transientcerebral ischemia. (C): GABA evoked a smaller current inpyramidal cells from rats subjected to transient cerebralischemia. Values are means7s.e.m. for six cells in each group.*o0.01 (Student’s t-test).


120


interneurons buffer intracellular calcium moreeffectively than do pyramidal cells, due to theirhigh expression of calcium-binding proteins (for areview, see Baimbridge et al, 1992). An increasein intracellular calcium might be important inmediating the reduced GABA response. First, anischemia-induced increase in intracellular calciummay reduce GABAA responses via a G protein-dependent mechanism (Chen and Wong, 1990).Second, calcium influx through NMDA receptorchannels is important for the loss of microtubule-associated proteins (Buddle et al, 2003), resulting indeclustering of GABAA receptors in postischemicbrain. Notably, application of MK-801, a high-affinity blocker of the NMDA channel, can attenuatethe downregulation of GABAA receptor subunitsafter focal cerebral ischemia (Redecker et al, 2002)and the suppression of GABA currents in vitro(Allen et al, 2004).

Depressed GABAA receptor-mediated synaptictransmission may, in turn, account for part of thepostischemic enhancement of the NMDA receptorcontribution to excitatory synaptic responses in CA1pyramidal cells. The latter phenomenon has been

attributed to increased phosphorylation of theNMDA receptor by cyclin-dependent kinase 5(Wang et al, 2003). However, synaptically drivenNMDA receptor-mediated responses are limited inboth amplitude and duration by the hyperpolarizingfeedforward IPSP (Wu et al, 2004). Reduction offeedforward IPSPs is expected to enlarge and pro-long the NMDA receptor contribution to EPSPs inCA1 pyramidal cells and to promote calcium influxthrough NMDA channels. Thus, depressed GABAinhibition may promote excitotoxicity in two ways:by enhancing the overall excitability of area CA1,which would increase calcium influx by severalroutes, and specifically by enhancing calcium influxthrough NMDA channels.

In addition to the phasic inhibition (IPSCs)mediated by synaptically released GABA, there isanother type of GABAergic inhibition called tonicinhibition. The average charge carried by thetonically active GABA receptors is several timeslarger than the charge carried by all of the IPSCs,even when IPSCs occur at high frequencies (Nusserand Mody, 2002). Tonic inhibition is presentin hippocampal interneurons under physiological

Figure 7 Transient cerebral ischemia does not change EGABA in either CA1 pyramidal cells or CA1 interneurons 4 h after reperfusion.Traces show the response to somatic GABA application at different holding potentials in the same cell. GABA was applied for 10 msat each holding potential beginning at the arrow. The corresponding current–voltage (I–V) relationship is shown beneath each trace.Results shown were obtained from representative experiments.


121


conditions and in pyramidal cells upon disturbanceof GABA uptake (Semyanov et al, 2003). It isconsidered to be important in regulating neuronalgain and signal-to-noise ratio (Mitchell andSilver, 2003; Chadderton et al, 2004; for a review,see Semyanov et al, 2004). Given that the concen-tration of extracellular GABA is increased duringand within the first hour after ischemia, furtherstudies are required to determine if tonic inhibi-tion is altered by cerebral ischemia and if theeffect differs between CA1 pyramidal cells andinterneurons.

We conclude that GABA responses are reduced inCA1 hippocampal pyramidal cells, but not ininterneurons, within a few hours after reperfusion.This deficit in synaptic inhibition may contribute tothe selective vulnerability of pyramidal cells and tothe enhancement of NMDA receptor-mediated sy-naptic responses observed at the same survival time.It may also explain the substantial neuroprotectiveefficacy of benzodiazepines and other GABA-enhan-cing drugs observed in animal models of globalcerebral ischemia (for a review, see Schwartz-Bloomand Sah, 2001). Thus, normalizing GABA synapticfunction might be a useful approach to counteractexcessive glutamatergic excitation during thepostischemic period, thereby attenuating pyramidalcell death.

References

Aika Y, Ren JQ, Kosaka K, Kosaka T (1994) Quantitativeanalysis of GABA like-immunoreactive and parvalbu-min-containing neurons in the CA1 region of the rathippocampus using a stereological method, the dis-sector. Exp Brain Res 99:267–76

Akaike N (1996) Gramicidin perforated patch recordingand intracellular chloride activity in excitable cells.Prog Biophys Biol 65:251–64

Alicke B, Schwartz-Bloom RD (1995) Rapid down-regula-tion of GABAA receptors in gerbil hippocampusfollowingcerebral ischemia. J Neurochem 65:2808–11

Allen N, Rossi D, Attwel D (2004) Sequential release ofGABA by exocytosis and reversed uptake leads toneuronal swelling in stimulated ischemia of hippo-campal slices. J Neurosci 24:3837–49

Arabadzisz D, Freund TF (1999) Changes in excitatory andinhibitory circuits of the rat hippocampus 12 to 14months after complete forebrain ischemia. Neuro-science 92:27–45

Avignone E, Frenguelli BG, Irving AJ (2003) CA1 hippo-campal interneurons and pyramidal cells show differ-ent responses to NMDA. Program #467.6 Washington,DC: Society for Neuroscience

Baimbridge KG, Celio MR, Rogers JH (1992) Calcium-binding proteins in the nervous system. Trends Neu-rosci 15:303–8

Bering R, Draguhn A, Diemer NH, Johansen FF (1997)Ischemia changes the coexpression of somatostatin andneuropeptide Y in hippocampal interneurons. ExpBrain Res 115:423–9

Buddle M, Eberhardt E, Ciminello LH, Levin T, Wing R,DiPasquale K, Raley-Susman KM (2003) Microtubule-associated protein 2 (MAP2) associates with the NMDAreceptor and is spatially redistributed within rathippocampal neurons after oxygen-glucose depriva-tion. Brain Res 978:38–50

Catania MV, Bellomo M, Giuffrida R, Stella AM, AlbaneseV (1998) AMPA receptor subunits are differentiallyexpressed in parvalbumin- and calretinin-positiveneurons of the rat hippocampus. Eur J Neurosci 10:3479–90

Chadderton P, Margrie TW, Hausser M (2004) Integrationof quanta in cerebellar granule cells during sensoryprocessing. Nature 428:856–60

Chen Q-X, Wong RKS (1990) Suppression of GABAA

receptor responses by NMDA application in hippocam-pal neurons acutely isolated from the adult guinea-pig.J Physiol 420:207–21

Cossart R, Tyzio R, Dinocourt C, Esclapez M, Hirsch JC,Ben-Ari Y, Bernard C (2001) Presynaptic kainatereceptors thatenhance the release of GABA on CA1hippocampal interneurons. Neuron 29:497–508

Coull JA, Boudreau D, Bachand K, Prescott SA, Nault F,Sik A, De Koninck P, De Koninck Y (2003) Trans-synaptic shift in anion gradient in spinal lamina Ineurons asa mechanism of neuropathic pain. Nature424:938–42

Crain BJ, Westerkam WD, Harrison AH, Nadler JV (1988)Selective neuronal death after transient forebrainischemia in the Mongolian gerbil: a silver impregnationstudy. Neuroscience 27:387–402

DeFazio RA, Keros S, Quick MW, Hablitz JJ (2000)Potassium-coupled chloride cotransport controls intra-cellular chloride in rat neocortical pyramidal neurons.J Neurosci 20:8069–76

Figure 8 Neither EGABA nor the calculated intracellular chlorideconcentration ([Cl�]i) was altered by transient cerebral ischemiain either cell type 4 h after reperfusion. (A) EGABA wasdetermined by least-squares regression of the I–V curve in 5cells from each group. (B) [Cl�]i was calculated from the Nernstequation, assuming that EGABA¼ ECl

�; ¼ sham-operated;¼ ischemia.


122


Freund TF, Buzsaki G, Leon A, Baimbridge KG, Somogyi P(1990) Relationship of neuronal vulnerability andcalcium binding protein immunoreactivity in ischemia.Exp Brain Res 83:55–66

Fukuda T, Nakano S, Yoshiya I, Hashimoto PH (1993)Persistent degenerative state of non-pyramidal neuronsin the CA1 region of the gerbil hippocampus following-transient forebrain ischemia. Neuroscience 53:23–38

Galeffi F, Pond BB, Zhan RZ, Mickuca AE, Schwartz-Bloom RD (2004a) Changes in expression of the K+/Cl�

cotransporter, KCC2, in the rat hippocampus aftertransient global ischemia. Soc Neurosci Abstr 229.13

Galeffi F, Sah R, Pond BB, George A, Schwartz-Bloom RD(2004b) Changes in intracellular chloride after oxygen-glucose deprivation of the adult hippocampal slice:effect of diazepam. J Neurosci 24:4478–88

Gao TM, Pulsinelli WA, Xu ZC (1998) Prolonged enhance-ment and depression of synaptic transmission in CA1pyramidal neurons induced by transient forebrainischemia in vivo. Neuroscience 87:371–83

Gao TM, Pulsinelli WA, Xu Z (1999) Changes inmembrane properties of CA1 pyramidal neurons aftertransient forebrain ischemia in vivo. Neuroscience90:771–80

Gao TM, Xu Z (1996) In vivo intracellular demonstrationof an ischemia-induced postsynaptic potential fromCA1 pyramidal neurons in rat hippocampus. Neu-roscience 75:655–69

Globus MY, Busto R, Dietrich WD, Martinez E, Valdes I,Ginsberg MD (1988) Effect of ischemia on the in vivorelease of striatal dopamine, glutamate, and gamma-aminobutyric acid studied by intracerebral microdia-lysis. J Neurochem 51:1455–64

Gulyas AI, Sik A, Payne JA, Kaila K, Freund TF (2001) TheKCl cotransporter, KCC2, is highly expressed in thevicinity of excitatory synapses in the rat hippocampus.Eur J Neurosci 13:2205–17

He Y, Janssen WGM, Vissayajjhala P, Morrison JH (1998)Synaptic distribution of GluR2 in hippocampal GA-BAergic interneurons and pyramidal cells: a double-label immunogold analysis. Exp Neurol 150:1–13

Hirokawa N (1994) Microtubule organization and dy-namics dependent on microtubule-associated proteins.Curr Opin Cell Biol 6:161–72

Inglefield JR, Wilson CA, Schwartz-Bloom RD (1997)Effect of transient cerebral ischemia on gamma-amino-butyric acidA receptor a1-subunit immunoreactiveinterneurons in the gerbil hippocampus. Hippocampus7:511–23

Inglefield JR, Schwartz-Bloom RD (1998) Optical imagingof hippocampal neurons with a chloride-sensitivedye: early effects of in vitro ischemia. J Neurochem70:2500–9

Jarolimek W, Lewen A, Misgeld U (1999) A furosemide-sensitive K+-Cl�cotransporter counteracts intracellularCl� accumulation and depletion in cultured rat mid-brain neurons. J Neurosci 19:4695–704

Johansen FF, Jorgensen BM, Diemer NH (1983) Resistanceof hippocampal CA1 interneurons to 20 mins oftransient cerebral ischemia in the rat. Acta Neuropathol61:135–40

Kakazu Y, Akaike N, Komiyama S, Nabekura J (1999)Regulation of intracellular chloride by co-transportersin developing lateral superior olive neurons. J Neurosci19:2843–51

Kang T-C, An S-J, Park S-K, Hwang IK, Yoon D-K, Shin H-S, Won MH (2002) Changes in Na+–K+–Cl� cotransport

immunoreactivity in the gerbil hippocampus followingtransient ischemia. Neurosci Res 44:249–54

Khazipov R, Congar P, Ben-Ari Y (1995) Hippocampal CA1lacunosum-moleculare interneurons: comparison ofeffects of anoxia on excitatory and inhibitory post-synaptic currents. J Neurophysiol 74:2138–49

Kirino T (1982) Delayed neuronal death in the gerbilhippocampus following ischemia. Brain Res 239:57–69

Kirino T, Robinson HPC, Niwa A, Tamura A, Kawai N(1992) Disturbance of membrane function precedingischemic delayed neuronal death in the gerbil hippo-campus. J Cereb Blood Flow Metab 12:408–17

Kitakawa K, Matsumoto M, Niinobe M, Mikoshiba K, HataR, Ueda H, Handa N, Fukunaga R, Isaka Y, Kimura K,Kamada T (1989) Microtubule-associated protein 2 as asensitive marker for cerebral ischemic damage-immu-nohistochemical investigation of dendritic damage.Neuroscience 31:401–11

Larsson E, Lindvall O, Kokaia Z (2001) Stereologicalassessment of vulnerability of immunocytochemicallyidentified striatal and hippocampal neurons afterglobal cerebral ischemia in rats. Brain Res 913:117–32

Li H, Siegel RE, Schwartz RD (1993) Rapid decline ofGABAA receptor subunit mRNA expression in hippo-campus following transient cerebral ischemia in thegerbil. Hippocampus 3:527–37

Luhmann HJ, Mudrick-Donnon LA, Mittmann T, Heine-mann U (1995) Ischaemia-induced long-term hyperex-citability in rat neocortex. Eur J Neurosci 7:180–91

Matesic DF, Lin RC (1994) Microtubule-associatedprotein 2 asan early indicator of ischemia-inducedneurodegeneration in the gerbil forebrain. J Neurochem63:1012–20

Mielke JG, Wang YT (2005) Insulin exerts neuroprotec-tion by counteracting the decrease in cell-surfaceGABAA receptors following oxygen-glucose depriva-tion in cultured cortical neurons. J Neurochem 92:103–1113

Mitani A, Namba S, Ikemune K, Yanase H, Arai T, KataokaK (1998) Postischemic enhancements of N-methyl-D-aspartic acid (NMDA) and non-NMDA receptor-mediated responses in hippocampal CA1 pyramidalneurons. J Cereb Blood Flow Metab 18:1088–98

Mitchell SJ, Silver RA (2003) Shunting inhibition mod-ulates neuronal gain during synaptic excitation. Neu-ron 38:433–45

Miyazaki S, Katayama Y, Furuichi M, Kinoshita K, mKawamata T, Tsubokawa T (1993) Post-ischemic po-tentiation of Schaffer collateral/CA1 pyramidal cellresponses of the rat hippocampus in vivo: involvementof N-methyl-D-aspartate receptors. Brain Res 611:155–9

Molnar P, Nadler JV (2001) Lack of effect of mossyfiber-released zinc on granule cell GABAA receptorsin the pilocarpine model of epilepsy. J Neurophysiol85:1932–40

Nabekura J, Ueno T, Okabe A, Furuta A, Iwaki T, Shimizu-Okabe C, Fukuda A, Akaike N (2002) Reduction ofKCC2 expression and GABAA receptor-mediated ex-citation after in vivo axonal injury. J Neurosci 22:4412–4417

Nitsch C, Scotti A, Sommacal A, Kalt G (1989) GABAergichippocampal neurons resistant to ischemia-inducedneuronal death contain the Ca2+-binding protein par-valbumin. Neurosci Lett 105:263–8

Nusser Z, Mody I (2002) Selective modulation of tonicand phasic inhibitions in dentate gyrus granule cells.J Neurophysiol 87:2624–8


123


Nyıri G, Stephenson FA, Freund TF, Somogyi P (2003)Large variability in synaptic N-methyl-D-aspartatereceptor density on interneurons and a comparisonwith pyramidal cell spines in the rat hippocampus.Neuroscience 119:347–63

Paxinos G, Watson C (1986) The rat brain in stereotaxiccoordinates. 2nd ed. San Diego: Academic Press

Payne JA, Stevenson TJ, Donaldson LF (1996) Molecularcharacterization of a putative K–Cl cotransporter in therat brain. J Biol Chem 271:16245–52

Petito CK, Feldmann E, Pulsinelli WA, Plum F (1987)Delayed hippocampal damage in humans followingcardiorespiratory arrest. Neurology 37:1281–6

Petrini EM, Zacchi P, Barberis A, Mozrzymas JW,Cherubini E (2003) Declusterization of GABAA recep-tors affects the kinetic properties of GABAergiccurrents in cultured hippocampal neurons. J Biol Chem278:16271–9

Pulsinelli WA, Brierley JB, Plum F (1982) Temporal profileof neuronal damage in a model of transient forebrainischemia. Ann Neurol 11:491–8

Redecker C, Wang W, Fritschy JM, Witte OW (2002)Widespread and long-lasting alterations in GABA(A)-receptor subtypes after focal cortical infarcts in rats:mediation by NMDA-dependent processes. J CerebBlood Flow Metab 22:1463–75

Sah R, Galeffi F, Ahrens R, Jordan G, Schwartz-Bloom RD(2002) Modulation of the GABAA-gated chloride chan-nel by reactive oxygen species. J Neurochem 80:383–91

Schlander M, Hoyer S, Frotscher M (1988) Glutamatedecarboxylase-immunoreactive neurons in the aging rathippocampus are more resistant to ischemia than CA1pyramidal cells. Neurosci Lett 91:241–6

Schwartz RD, Yu X, Katzman MR, Hayden-Hixson DM,Perry JM (1995) Diazepam, given postischemia, protectsselectively vulnerable neurons in the rat hippocampusand striatum. J Neurosci 15:529–39

Schwartz-Bloom RD, Sah R (2001) g-Aminobutyric acidA

neurotransmission and cerebral ischemia. J Neurochem77:353–71

Segal M, Greenberger V (1992) Acidic amino acids evoke asmaller [Ca2+]i rise in GABAergic than non-GABAergichippocampal neurons. Neurosci Lett 140:243–6

Semyanov A, Matthew C, Walker MC, Kullmann DM,Silver RA (2004) Tonically active GABAA receptors:modulating gain and maintaining the tone. TrendsNeurosci 27:262–9

Semyanov A, Walker MC, Kullmann DM (2003) GABAuptake regulates cortical excitability via cell type-specific tonic inhibition. Nat Neurosci 6:484–90

Shetty AK, Turner DA (1998) Hippocampal interneuronsexpressing glutamic acid decarboxylase and Ca2+-bind-ing proteins decrease with aging in Fischer 344 rats.J Comp Neurol 394:252–69

Shinno K, Zhang L, Eubanks JH, Carlen PL, Wallace MC(1997) Transient ischemia induces an early decrease ofsynaptic transmission in CA1 neurons of rat hippo-campus: electrophysiologic study in brain slices.J Cereb Blood Flow Metab 17:955–66

Su G, Haworth RA, Dempsey RJ, Sun D (2000) Regulationof Na+–K+–Cl� cotransporter in primary astrocytes bydibutyryl cAMP and high [K+]o. Am J Physiol 279:C1710–21

Sun D, Murali SD (1998) Stimulation of Na+–K+–2Cl�

cotransporter in neuronal cells by excitatory neuro-transmitter glutamate. Am J Physiol 275:C772–9

Taga K, Fujiwara N, Shimoji K (2000) Ischemic tolerancepreserves propagation of membrane depolarization.Neuroreport 11:2961–4

Tortosa A, Ferrer I (1993) Parvalbumin immunoreactivityin the hippocampus of the gerbil after transientforebrain ischemia: a qualitative and quantitativesequential study. Neuroscience 55:33–43

Toyota H, Ohno K, Yamada J, Ikeda M, Okabe A, Sato K,Hashimoto K, Fukuda A (2003) Induction of NMDAand GABAA receptor-mediated Ca2+ oscillations withKCC2 mRNA downregulation in injured facial moto-neurons. J Neurophysiol 89:1353–62

Tsubokawa H, Oguro K, Robinson HPC, Masuzawa T,Kirino T, Kawai N (1992) Abnormal Ca2+ homeostasisbefore cell death revealed by whole cell recording ofischemic CA1 hippocampal neurons. Neuroscience49:807–17

Urban L, Neill KH, Crain BJ, Nadler JV, Somjen GG (1989)Postischemic synaptic physiology in area CA1 of thegerbil hippocampus studied in vitro. J Neurosci 9:3966–75

Urban L, Neill KH, Crain BJ, Nadler JV, Somjen GG (1990)Postischemic synaptic excitation and N-methyl-D-as-partate receptor activation in gerbils. Stroke 21:III-23–7

Verheul HB, de Leeuw F-E, Scholten G, Tulleken CAF,Lopes da Silva FH, Ghijsen WEJM (1993) GABAA

receptor function in the early period after transientforebrain ischemia in the rat. Eur J Neurosci 5:955–60

Wang J, Liu S, Fu Y, Wang JH, Lu Y (2003) Cdk5 activationinduces hippocampal CA1 cell death by directlyphosphorylating NMDA receptors. Nat Neurosci6:1039–47

Wu SH, Ma CL, Kelly JB (2004) Contribution of AMPA,NMDA, and GABAA receptors to temporal pattern ofpostsynaptic responses in the inferior colliculus of therat. J Neurosci 24:4625–34

Yan Y, Dempsey RJ, Flemmer A, Forbush B, Sun D (2003)Inhibition of Na+–K+–Cl� cotransporter during focalcerebral ischemia decreases edema and neuronaldamage. Brain Res 961:22–31

Yan Y, Dempsey RJ, Sun D (2001) Na+–K+–Cl� cotranspor-ter in rat focal cerebral ischemia. J Cereb Blood FlowMetab 21:711–21

Zhan R-Z, Fujihara H, Baba H, Yamakura T, Shimoji K(2002) Ischemic preconditioning is capable of inducingmitochondrial tolerance in the rat brain. Anesthesiol-ogy 97:896–901

Zhan R-Z, Wu C, Fujiwara H, Taga K, Qi S, Naito M,Shimoji K (2001) Both caspase-dependent and caspase-independent pathways may be involved in hippocam-pal CA1 neuronal death because of loss of cytochrome cfrom mitochondria in a rat forebrain ischemia model.J Cereb Blood Flow Metab 21:529–40


124


Documents

Depressed responses to applied and synaptically-released GABA in CA1 pyramidal cells, but not in CA1 interneurons, after transient forebrain ischemia