BRAINA JOURNAL OF NEUROLOGY

The epileptic human hippocampal cornuammonis 2 region generates spontaneousinterictal-like activity in vitroLucia Wittner,1,2,3,4 Gilles Huberfeld,1,5 Stephane Clemenceau,1,5,6 Lorand Ero00 ss,3

Edouard Dezamis,1,6 Laszlo Entz,2,3 Istvan Ulbert,2,3,7 Michel Baulac,1,5 Tamas F. Freund,4

Zsofia Magloczky4 and Richard Miles1

1 INSERM U739, Faculte de Medecine Pitie-Salpetriere, Paris, France

2 Institute for Psychology, Hungarian Academy of Sciences, Budapest, Hungary

3 National Institute of Neurosurgery, Budapest, Hungary

4 Institute of Experimental Medicine, Hungarian Academy of Sciences, Budapest, Hungary

5 Epilepsy Unit, Hopital Pitie-Salpetriere, Paris, France

6 Neurosurgery Unit, Hopital Pitie-Salpetriere, Paris, France

7 Department of Information Technology, Peter Pazmany Catholic University, Budapest, Hungary

Correspondance to: Lucia Wittner,

Institute for Psychology,

Hungarian Academy of Sciences,

1068 Budapest,

Szondi u. 83-85,

Hungary

E-mail: wittner@cogpsyphy.hu

The dentate gyrus, the cornu ammonis 2 region and the subiculum of the human hippocampal formation are resistant to the cell

loss associated with temporal lobe epilepsy. The subiculum, but not the dentate gyrus, generates interictal-like activity in tissue

slices from epileptic patients. In this study, we asked whether a similar population activity is generated in the cornu ammonis 2

region and examined the electrophysiological and neuroanatomical characteristics of human epileptic cornu ammonis 2 neurons

that may be involved. Hippocampal slices were prepared from postoperative temporal lobe tissue derived from epileptic patients.

Field potentials and multi-unit activity were recorded in vitro using multiple extracellular microelectrodes. Pyramidal cells were

characterized in intra-cellular records and were filled with biocytin for subsequent anatomy. Fluorescent immunostaining was

made on fixed tissue against the chloride–cation cotransporters sodium-potasium-chloride cotransporter-1 and potassium-

chloride cotransporter-2. Light and electron microscopy were used to examine the parvalbumin-positive perisomatic inhibitory

network. In 15 of 20 slices, the hippocampal cornu ammonis 2 region generated a spontaneous interictal-like activity,

independently of population events in the subiculum. Most cornu ammonis 2 pyramidal cells fired spontaneously. All cells

fired single action potentials and burst firing was evoked in three cells. Spontaneous excitatory postsynaptic potentials were

recorded in all cells, but hyperpolarizing inhibitory postsynaptic potentials were detected in only 27% of the cells. Two-thirds of

cornu ammonis 2 neurons showed depolarizing responses during interictal-like events, while the others were inhibited, accord-

ing to the current sink in the cell body layer. Two biocytin-filled cells both showed a pyramidal-like morphology with axons

projecting to the cornu ammonis 2 and cornu ammonis 3 regions. Expression of sodium-potasium-chloride cotransporter-1 and

potassium-chloride cotransporter-2 was reduced in some cells of the epileptic cornu ammonis 2 region, but not to an extent

corresponding to the proportion of cells in which hyperpolarizing postsynaptic potentials were absent. Numbers of parvalbumin-

positive inhibitory cells and axons were shown to be decreased in the epileptic tissue. Electron microscopy showed the

doi:10.1093/brain/awp238 Brain 2009: 132; 3032–3046 | 3032

Received March 9, 2009. Revised July 31, 2009. Accepted July 31, 2009. Advance Access publication September 18, 2009

� The Author (2009). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved.

For Permissions, please email: journals.permissions@oxfordjournals.org

preservation of somatic inhibitory input of cornu ammonis 2 cells, and confirmed the loss of parvalbumin from the interneurons

rather than their death. An extra excitatory input (partly coming from sprouted mossy fibres) was demonstrated to innervate

cornu ammonis 2 cell bodies. Our results show that the cornu ammonis 2 region of the sclerotic human hippocampus can

generate an independent epileptiform activity. Inhibitory and excitatory signalling were functional but modified in epileptic

cornu ammonis 2 pyramidal cells. Overexcitation and the altered functional properties of perisomatic inhibitory network, rather

than a modified chloride homeostasis, may account for the perturbed �-aminobutyric acid-ergic signalling and the generation of

interictal-like activity in the human epileptic cornu ammonis 2 region.

Keywords: temporal lobe epilepsy; chloride homeostasis; perisomatic inhibition; synaptic re-organization

Abbreviations: CA = cornu ammonis; CSD = current source density; EPSPs = excitatory postsynaptic potentials; IPSPs = inhibitorypostsynaptic potentials; LFPg = local field potential gradient; MUA = multi-unit activity; PV = parvalbumin; PCL = pyramidal cell layer;TLE = temporal lobe epilepsy

IntroductionTemporal lobe epilepsy (TLE) is often associated with hippocampal

sclerosis. Hippocampal sclerosis consists of a stereotyped and

severe neuronal loss in the cornu ammonis (CA)1 and -3 areas

of the hippocampus with a slight/moderate cell loss in the dentate

gyrus and the CA2 region (Loup et al., 2000; Proper et al., 2000;

Pirker et al., 2001; Andrioli et al., 2007). There seems to be little

or no neuronal death in the subiculum of patients with classical

hippocampal sclerosis (Andrioli et al., 2007). The pattern of cell

loss in epilepsies of the temporal lobe is highly heterogeneous and

several grading systems have been used by different groups to

describe hippocampal sclerosis (Wyler, 1992; Proper et al., 2000;

de Lanerolle et al., 2003; Wittner et al., 2005).

Work on excised human epileptic tissue has demonstrated

considerable anatomical changes in preserved regions. In the

dentate gyrus, which is the largest and best studied of these

regions, there is an intense re-organization of both excitatory

and inhibitory neuronal networks. Dentate granule cells are

dispersed (Houser, 1990), certain inhibitory interneuron classes

are selectively lost (de Lanerolle et al., 1989; Magloczky et al.,

2000; Wittner et al., 2001), and axonal sprouting occurs for

surviving excitatory (Houser et al., 1990; Babb et al., 1991) and

inhibitory cells (Magloczky et al., 2000; Wittner et al., 2001).

In the CA2 region of epileptic patients, while there is only

limited pyramidal cell loss, there is evidence for a major

re-organization of perisomatic inhibition. The density of parvalbu-

min (PV)-containing perisomatic inhibitory cells in the CA2 region

is considerably decreased (Andrioli et al., 2007). An inhomo-

geneous distribution of the axon terminals of axo-axonic cells

has been described with a lack of staining in some regions and

an increase in the complexity of many surviving terminals

(Arellano et al., 2004). Furthermore, the identity of GABAA recep-

tor subunits, contributing to perisomatic inhibition in the CA2

region was altered (Loup et al., 2000), suggesting perturbations

at the subcellular level with possible functional consequences for

GABAergic signalling.

This re-organization of inhibitory and excitatory circuits may

contribute to the generation of epileptiform activity by surviving

regions of hippocampal formation. In vitro studies of human tissue

have shown epileptiform population activities in both the dentate

gyrus and the subiculum. Interictal-like activity is generated spon-

taneously in the subiculum (Cohen et al., 2002; Wozny et al.,

2005; Huberfeld et al., 2007). In the dentate gyrus, seizure-like

events have been induced by electrical stimuli or by elevating

extracellular potassium levels (Gabriel et al., 2004). Single cell

records show profound changes in synaptic responses and cellular

excitability in surviving regions. In the dentate gyrus, an impaired

inhibition is correlated with the degree of mossy fibre sprouting

(Franck et al., 1995). It probably contributes to hyperexcitable

population responses to anti-dromic and ortho-dromic stimuli

(Masukawa et al., 1989, 1992, 1996, 1997, 1999; Uruno et al.,

1994; Isokawa and Fried, 1996). Individual granule cells also

discharge more intensely in response to current injection

(Dietrich et al., 1999). In the CA2 region, hyperpolarizing inhibi-

tory synaptic responses are weak or absent (Williamson and

Spencer, 1994). In the subiculum, depolarizing as well as hyper-

polarizing inhibitory synaptic potentials in pyramidal cells may

contribute to spontaneous interictal-like activity (Cohen et al.,

2002; Wozny et al., 2005; Huberfeld et al., 2007).

The deficit in GABAergic signalling in the human epileptic sub-

iculum has been associated with a perturbed chloride homeostasis

(Palma et al., 2006; Huberfeld et al., 2007). The internal chloride

concentration is largely controlled by two cation–chloride cotrans-

porters (Payne et al., 2003; Sipila et al., 2006). The Na–K–2Cl

cotransporter NKCC1 and the K–Cl cotransporter KCC2 have

functionally opposing effects. NKCC1 increases, while KCC2

decreases intra-neuronal chloride level. Thus, NKCC1 transport

favours depolarizing and KCC2 favours hyperpolarizing responses

to GABAA receptor activation. Epileptiform activity has been linked

to changes in the function or expression of NKCC1 and KCC2

(Payne et al., 2003; Rivera et al., 2004; Palma et al., 2006;

Huberfeld et al., 2007). In the human epileptic subiculum,

NKCC1 was found to be upregulated (Palma et al., 2006) and

KCC2 was downregulated (Palma et al., 2006; Huberfeld et al.,

2007; Munoz et al., 2007), resulting in elevated levels of internal

chloride and depolarizing GABAergic events (Palma et al., 2006;

Huberfeld et al., 2007).

Little is known about the physiological and anatomical proper-

ties of the human epileptic CA2 region. Here, we show that an

interictal-like activity is generated in the CA2 region in hippocam-

pal slice preparations derived from TLE patients. We used

Epileptiform activity in the CA2 region Brain 2009: 132; 3032–3046 | 3033

electrophysiological and anatomical techniques to clarify the

properties and possible mechanisms of this epileptiform activity.

Simultaneous multiple channel extracellular and intra-cellular

records were made to describe excitatory and inhibitory signalling

in the CA2 region, and to relate cellular responses to the syn-

chronous population events. We explored the role of neuronal

chloride homeostasis, which might be related to depolarizing

GABAergic responses, by double immunofluorescence against the

Cl-cotransporters KCC2 and NKCC1. Changes in the perisomatic

inhibitory input were verified by light and electron microscopic

analysis of PV-positive interneurons.

Methods

Epileptic patients—control subjectsHippocampal tissue was obtained from operations on patients (age

range, 28–52 years; seizures for 5–39 years) with pharmacoresistant

epilepsies of the temporal lobe. All patients gave a written consent,

and our protocol was approved either by the Comite Consultatif

National d’Ethique (France) or by the Hungarian Ministry of Health

(Hungary). Control hippocampi (n = 2, age 53 and 56 years) were

kindly provided by the Lenhossek Human Brain Program,

Semmelweis University, Budapest, Hungary. Control subjects died

from causes unrelated to any brain disease and were processed for

autopsies in the Department of Forensic Medicine of the Semmelweis

University. The autopsy confirmed the absence of any neurological

disorders. Brains were removed 2–4 h after death, and fixed by perfu-

sion through the internal carotids and vertebral arteries, as described

(Wittner et al., 2001, 2002, 2005). Both control brains (C10 and C11)

have been used in previous studies (Wittner et al., 2001, 2002, 2005).

The study was approved by the Regional and Institutional Committee

of Science and Research Ethics of Scientific Council of Health (TUKEB

5-1/1996, further extended in 2005) and performed in accordance

with the Declaration of Helsinki.

Tissue preparationTissue was transported from the operating room to the laboratory in

a cold, oxygenated solution containing (in milliMolar) 248 D-sucrose,

26 NaHCO3, 1 KCl, 1 CaCl2, 10 MgCl2 and 10 D-glucose, equilibrated

with 5% CO2 in 95% O2. Hippocampal–subicular slices of 400mm

thickness were cut with a vibratome. They were maintained at

35–37�C and were put in an interface chamber perfused with a

solution containing (in miliMolar) 124 NaCl, 26 NaHCO3, 4 KCl,

2 MgCl2, 2 CaCl2 and 10 D-glucose and equilibrated with 5% CO2

in 95% O2.

RecordingsIntra-cellular records were made with microelectrodes that contained

2M KAc bevelled to a resistance of 50–100 M�. The data were

obtained within 10–20 min of penetration. Signals were amplified

with either an Axoclamp 2B (Axon Instruments, Union City, CA,

USA) or with a BA-1S (NPI Electronic GmbH, Tamm, Germany) ampli-

fier operated in current-clamp mode. In acceptable records, membrane

potential was more negative than –50 mV, input resistance was higher

than 20 M� and action potentials were overshooting. Extracellular

records were made with tungsten electrodes of �10 mm tip diameter

(Cohen and Miles, 2000), using a differential amplifier (1700; AM

Systems, Everett, WA, USA). They were digitized with a 12-bit,

16-channel analogue-to-digital converter (Digidata 1200A; Molecular

Devices, Union City, CA, USA) at 20 kHz sampling rate and saved on a

personal computer. Data analysis were made with pClamp 8 program

(Axon Instruments)

The local field potential gradient (LFPg) was recorded with laminar

multiple channel microelectrodes (24 channels, distance between con-

tacts: 150mm) (Ulbert et al., 2001, 2004a, b; Fabo et al., 2008) using

a custom made 48-channel voltage gradient amplifier of pass-band

0.01 Hz to 10 kHz. Signals were digitized with two 32-channel,

16-bit resolution analogue-to-digital converter (National Instruments,

Austin TX, USA) at 20 kHz sampling rate, recorded with a home

written routine in LabView7 (National Instruments, Austin TX, USA).

Current source density (CSD), an estimate of population transmem-

brane currents, and multi-unit activity (MUA) were calculated from the

LFPg using standard techniques (Ulbert et al., 2001, 2004b). For the

CSD and MUA colour maps �100–350 events were averaged. Data

were analysed with the Neuroscan Edit4.3 program (Compumedics

Neuroscan, Charlotte, NC, USA) and a home written program for

MATLAB (The MathWorks, Natick, MA, USA).

Cell morphologyBiocytin was injected from electrodes containing 1.6% in 2 M KAc

(1.5 nA depolarizations of 200 ms duration repeated at 1–3 Hz for at

least 10 min). After injection, slices were maintained for at least 1 h in

the recording chamber. They were then fixed overnight in 4% para-

formaldehyde with 15% picric acid in 0.1 M phosphate buffer (PB) at

4�C. They were resectioned at 70 mm and freeze-thawed above liquid

N2 in PB containing 30% sucrose. Cells containing biocytin were

revealed with the ABC reaction (Vector, 1.5 h, 1:250) using diamino-

benzidine (DAB, Sigma, St Louis, MO, USA) as the chromogen.

Sections were osmicated (20 min, 0.5% OsO4), dehydrated in ethanol

and mounted in Durcupan (ACM; Fluka, Buchs, Switzerland). Cells

were digitally reconstructed in three dimensions using the

NeuroLucida system (MicroBrightField Inc. Williston, VT, USA).

ImmunohistochemistrySixty-micrometre thick sections were cut from fixed tissue (see above),

washed with 0.1M PB and freezed–thawed over liquid N2 in PB con-

taining 30% sucrose. Endogenous peroxidase activity was blocked

with 1% H2O2. Non-specific staining was suppressed with 5% skim

milk powder and 2% BSA in PB. KCC2 immunostaining used a

polyclonal rabbit antibody (1:2000 dilution) (Payne et al., 1996),

NKCC1 was stained with a monoclonal mouse antibody (T4,

Developmental Studies Hybridoma Bank, Iowa City IA, USA, 1:2000

dilution) for 24 h at 4�C. The specificity of the KCC2 antibody has

been detailed previously (Payne et al., 1996; Williams et al., 1999).

The T4 NKCC1 antibody showed no staining in NKCC1 knockout mice

(Chen et al., 2005). Alexa594-bound donkey anti-rabbit and

Alexa488-bound horse anti-mouse fluorescent secondary antibody

(3 h, 1:250; Invitrogen, San Diego, CA, USA) were used as secondary

antibodies. Images of fluorescent immunostaining were made with a

confocal microscope (SP2; Leica, Nussloch, Germany). The specificity

of immunostaining was verified according to the following criteria:

(i) incubating sections without primary antibody gave no specific stain-

ing; (ii) somatic or perisomatic dendritic membrane was clearly stained;

and (iii) membrane staining could be distinguished from non-specific

lipofuscin autofluorescence.

3034 | Brain 2009: 132; 3032–3046 L. Wittner et al.

A monoclonal mouse antibody was used against PV (1:5000, Sigma,

St Louis, MO, USA) for 24 h at 4�C. The specificity of the antibody

was tested by the manufacturer. For visualization of immunopositive

elements, biotinylated anti-mouse immunoglobulin G (1:300, Vector)

was applied as secondary antiserum followed by avidin–biotinylated

horseradish peroxidase complex (ABC; 1:300, Vector). The immuno-

peroxidase reaction was developed by DAB dissolved in Tris buffer

(TB, pH 7.6) as a chromogen. Sections were dehydrated as described

above. Dark, specific staining due to the antibody could be clearly

distinguished from the light non-specific osmium staining of lipofuscin.

Electron microscopyAfter light microscopic examination, areas of interest were

re-embedded and sectioned for electron microscopy. Ultrathin serial

sections were collected on Formvar-coated single slot grids, stained

with lead citrate and examined with a Hitachi 7100 electron micro-

scope. For the analysis of synaptic input to CA2 pyramidal cells, blocks

were re-embedded from one control (C11) and three sclerotic epileptic

hippocampi, containing the entire width of the pyramidal layer

(one block from each patient). About 15–25 cells were digitized

from each subject or patient (each from one 55 nm thick section). In

one specimen (Patient P960830) numerous degenerating pyramidal

cells were found. These cells were not included in the analysis. The

perimeter of pyramidal cell somata and the contact length of all affer-

ent synaptic boutons made with the soma were measured using the

ImageJ program (Wayne Rasband, National Institutes of Health, USA).

The synaptic coverage of each soma was measured as the ratio of

the total length of afferent terminals divided by the somatic surface

perimeter (Wittner et al., 2001, 2005).

Results

Hippocampal anatomyIn this study, we examined population activities of the human

epileptic CA2 region and cellular properties of CA2 neurons in

slice preparations from 13 TLE patients (11 with hippocampal

sclerosis and 2 without hippocampal sclerosis). Fixed tissue was

used for anatomical analysis: the expression of the chloride-

cotransporters NKCC1 and KCC2 by neurons of the CA2 region

was studied in seven patients (six hippocampal sclerosis, one non-

hippocampal sclerosis), and the perisomatic input of CA2 pyrami-

dal cells in PV-stained sections was examined in four patients

(three hippocampal sclerosis, one non-hippocampal sclerosis).

We determined the boundaries of the CA2 region of the human

hippocampal formation from previous descriptions (Lorente de No,

1934; Rosene and Van Hoesen, 1987). TLE is often associated

with hippocampal sclerosis consisting of a severe atrophy of the

CA1 and CA3 regions, whereas the subiculum, the dentate gyrus

and the CA2 regions are relatively spared—although the degree of

cell loss may vary in all regions (Sommer, 1880; Proper et al.,

2000; Pirker et al., 2001; Andrioli et al., 2007). Our sclerotic

samples showed classical signs of hippocampal sclerosis with a

relatively homogeneous pattern of principal cell loss. The CA1

region was very atrophic, medium cell loss was observed in the

CA3a/b subregions, and many neurons were lost in the CA3c/hilar

region. The CA2 region was demarcated by the absence of CA1

pyramidal cells and a reduced thickness of the CA3 pyramidal cell

layer (PCL) (see Supplementary Fig. 1). Although the CA2 region

was clearly much better preserved than the CA1 and CA3 regions,

it remains possible that a quantitative analysis would reveal a

slight/moderate cell loss. In the patients with no hippocampal

sclerosis, the border between the CA2 and CA1 region was deter-

mined by the thickness of the cell layer: CA2 pyramids form a

densely packed, relatively thin layer, whereas the PCL in the

CA1 and CA3 regions is thicker and cell bodies are more

dispersed. Thickness of the cell body layer and borders of the

CA2 region were visible in the unstained slices from which record-

ings were made. Electrophysiological recordings were made with

both extra- and intra-cellular electrodes placed in the middle of

the CA2 region to avoid either CA1 or CA3 cells. In fixed and

immunostained tissue, we used additional characteristics to deter-

mine the boundaries of the CA2 region. Somata of CA2 pyramidal

cells are larger than those of either CA3 or CA1 pyramidal cells,

are strongly stained with calbindin (Seress et al., 1992), and they

usually contain more lipofuscin. Both immunostaining procedures

used (fluorescence and peroxidase reaction combined with

osmication) showed a non-specific lipofuscin staining clearly

distinguishable from the specific stainings (see Methods section).

Based on a combination of these features, we could distinguish

CA2 region from the two neighbouring hippocampal subfields.

Quantitative anatomical experiments, including cell counts

(KCC2 and NKCC1) and electron microscopy of CA2 pyramidal

cells were also made from the middle of the CA2 region avoiding

the region boundaries.

Spontaneous interictal-like activity isgenerated in the CA2 region in vitroInterictal spikes detected on scalp electroencephalographic (EEG)

recordings are a hallmark of focal epilepsies. Such events are

absent in healthy subjects and represent one of the most impor-

tant diagnostic signs of focal epilepsies. Interictal discharges in

both human and experimental focal epilepsies are characterized

by high amplitude (450 mV), fast EEG transients or spikes, usually

followed by a slow wave lasting several hundreds of milliseconds

(for review see de Curtis and Avanzini, 2001). In human TLE, the

use of intra-cranial electrodes reaching the parahippocampal gyrus

confirmed the presence of interictal spikes in the deep structures

and showed that they were more active than the neocortex

(Clemens et al., 2003). In vivo intra-hippocampal (Ulbert et al.,

2004b; Fabo et al., 2008) recordings with microelectrodes and

high frequencies of data acquisition showed that an increased

neuronal firing accompanied the field potential deflection of

interictal spikes in the subiculum of patients with TLE.

Spontaneous synchronous discharges were generated by slices

prepared from the subiculum of patients with TLE (Cohen et al.,

2002; Wozny et al., 2005; Huberfeld et al., 2007). In vitro

synchronous population events show similarities to interictal

spikes recorded in vivo, consisting of rhythmically recurring fast

field potential deflections associated with an increased neuronal

firing. Caution is needed in comparing these population events

with interictal spikes generated in vivo. Since neither in vivo

Epileptiform activity in the CA2 region Brain 2009: 132; 3032–3046 | 3035

intra-hippocampal recordings nor hippocampal tissue from healthy

humans are available, we know very little about single cell activity

or population oscillations that a healthy hippocampus generates.

Based on the resemblances that we have noted, we will describe

the synchronous population bursts observed in vitro as interictal-

like activity.

The dentate gyrus, the CA2 region and the subiculum of the

human hippocampal formation resist the severe cell loss associated

with TLE. The subiculum, but not the dentate gyrus, generates an

interictal-like activity in tissue slices from epileptic patients (Cohen

et al., 2002; Wozny et al., 2005; Huberfeld et al., 2007).

We asked whether a similar activity is generated by the CA2

region of epileptic human hippocampus.

Spontaneous interictal-like activity was observed in the CA2

region in vitro, in physiological bathing solution in 15 of 20

slices, derived from 10 of 13 TLE patients. The frequency of the

synchronous events varied between 0.1 and 1.5 Hz, with a mean

of 0.50� 0.42 Hz (mean� SD, n = 12 slices). In the majority of the

slices (n = 13), interictal-like activity region was generated repeti-

tively with a similar interval between events (Fig. 1A1), but in two

slices, from two different patients, interictal-like events tended to

occur in groups separated by long silences (Fig. 1A2 and 1A3). In 10

slices, field potentials were recorded with two or three electrodes

located separately in the pyramidal layer. These records showed

that interictal activities were synchronized within the region.

We examined the activity of the dentate gyrus in extracellular

recordings made from the granule cell layer in 6 of 10 patients

where an interictal-like activity was detected in CA2. In all cases,

we detected a spontaneous MUA in the granule cell layer.

Synchronous events, characterized by an acceleration of MUA,

together with a field potential, were not detected in the dentate

region in any of these slices (n = 6, Supplementary Fig. 2).

Intra-cellular records from dentate granule cells revealed a

spontaneous discharge of single action potentials and an absence

of events indicative of population synchrony (n = 3, Supplementary

Fig. 2).

We asked whether there was a temporal relation between

synchronous activity generated in the CA2 region and that

recorded in the same slice from the subiculum. We performed

simultaneous recordings from CA2 and the subiculum (n = 6

slices), in slices where both regions generated an interictal-like

activity. The occurrence of synchronous events in the CA2

region was independent of the subicular activity (Fig. 1B).

Furthermore, interictal-like events were generated independently

in the CA2 region in slices containing the hippocampus proper and

lacking the subiculum (n = 4 slices). When interictal-like events

were observed both in the CA2 region and the subiculum,

the frequency of synchronous events in the CA2 region was

usually lower (0.51� 0.50 Hz) than that in the subiculum

(1.31� 0.79 Hz).

Figure 1 Spontaneous interictal-like activity emerged in vitro in the hippocampal CA2 region of TLE patients, showing either

continuous rhythmicity (A1) or inhomogeneous incidence (A2 and A3). This interictal-like activity was independent of the subicular

activity (B). Upper traces are wide band recordings, lower traces show field potential (band pass filtered, 1–50 Hz). Asterisks indicate

interictal-like events. Single events, magnified on the right side, consisted of field potential deflections with elevated neuronal firing.

From epileptic patients P250105 (A1), P180105 (A2 and A3), P050607 (B).

3036 | Brain 2009: 132; 3032–3046 L. Wittner et al.

The shape of the interictal-like events varied between slices, but

a field potential transient of �30–100 mV amplitude, recorded in

the PCL, associated with an increased frequency of multi-unit

firing was observed in every case (Fig. 1). The field potential

transient was either biphasic with a positive and a negative

component, or showed only a positive deflection (Fig. 1, insets).

Multi-unit firing usually increased during the rising phase of the

field potential transient, and often continued during later phases

(Fig. 1).

In two slices with interictal-like activity in the CA2 region,

a laminar micro-electrode (see Methods section) was used

to measure the LFPg (Fig. 2). The 24-channel electrode

(distance between contacts: 150mm) was placed perpendicular

to the PCL, permitting recording from the entire width of the

CA2 region, along the somatic–dendritic axis. Deflections in the

LFPg signal were detected in both the pyramidal layer and prox-

imal dendritic areas. MUA was visible in the cell body layer

(Fig. 2A and B). CSD analysis shows an estimate of the transmem-

brane currents of the local neuronal population. Inward currents

(sink) could reflect either an active, synaptically driven excitation

or a passive return current. Outward currents (source) might

represent an active inhibition, or the passive return current of a

sink. During interictal-like activity, CSD analysis showed a sink in

the stratum pyramidale and sources in dendritic regions of stratum

radiatum and stratum oriens (Fig. 2C). This confirms that synchro-

nous events are locally generated in the CA2 region, and suggests

the presence of an excitation in the cell body layer. MUA analysis

confirmed an increase in neuronal firing in the pyramidal layer

during interictal-like activity (Fig. 2D).

Cellular properties of human CA2pyramidal cellsIntra-cellular records were made from 19 human CA2 pyramidal

cells (Fig. 3, Table 1). Most of them fired in the absence of

injected current (n = 13/19). The mean resting membrane potential

of six CA2 pyramidal cells that did not fire was –71.9� 3.9 mV.

Spontaneous firing consisted of single action potentials in all cells.

Burst firing could be elicited by current injection in three cells

(Table 1). The mean input resistance of CA2 pyramidal cells

(n = 17) was 25.6� 6.1 M� and their time constant measured

from responses to 200 ms duration, 0.5 mV amplitude negative

current injections was 15.5� 7.1 ms. In the only previous study

on CA2 pyramidal cells from human hippocampus with

Ammon’s horn sclerosis, Williamson and Spencer (1994) noted

that hyperpolarizing postsynaptic potentials (IPSPs) were weak or

Figure 2 Wide band LFPg recording of interictal-like events (asterisk) in the epileptic CA2 region (A). Channels 10–12 are in the

striatum radiatum, 13–18 in the striatum pyramidale and 19–20 in the striatum oriens. One event is shown in detail in (B). The colour

map of the CSD analysis shows a sink (red) in the pyramidal layer and two sources (blue) in proximal dendritic regions (C). MUA (red)

increased in the PCL during interictal-like events (D). For both colour maps, average of 353 sweeps, channel numbers are indicated on

the left side (ch), arbitrary units (AU) on the right side. A schematic pyramidal cell shows the orientation of the region. Layers are

indicated on the maps. R = striatum radiatum; P = striatum pyramidale; O = striatum oriens. From epileptic patient P081126.

Epileptiform activity in the CA2 region Brain 2009: 132; 3032–3046 | 3037

Figure 3 The majority of the recorded human CA2 cells fired action potentials spontaneously (A). Spontaneous EPSPs (asterisks) were

detected in all cells, and IPSPs (arrows) were observed in about one-third of the cells. Most of the CA2 pyramidal cells (n = 11/15)

received depolarizing synaptic potentials simultaneously with interictal-like events (B), while the remaining cells were hyperpolarized

during synchronous events (C). B1, C1 show single sweeps, B2, C2 show superimposed sweeps. Cells P041012_C11 (A and C),

P050125_C4 (B).

Table 1 Cellular properties of human CA2 pyramidal cells

Cell Spontaneous firing Burst firing Spontaneous EPSPs Spontaneous IPSPs Response to synchronous events

P040803_C3 + � + + Hyperpolarizing

P040803_C4 � + + + Hyperpolarizing

P041012_C11 + � + + Hyperpolarizing

P041129_C1 + � + � N/A

P050118_C1 � � + � Depolarizing

P050125_C1 � � + � Depolarizing

P050125_C2 � � + � Depolarizing

P050125_C3 � � + + Hyperpolarizing

P050125_C4 + � + � Depolarizing

P050125_C5 + � + � Depolarizing

P050215_C1 + � + + N/A

P050215_C2 � � + � N/A

P050531_C2 + + + � Depolarizing

P070406_C1 + � + � Depolarizing

P070406_C1 + � + � Depolarizing

P070406_C6 + � + � Depolarizing

P070502_C1 + + + + N/A

P090512_C1 + � + � Depolarizing

P090512_C2 + � + � Depolarizing

3038 | Brain 2009: 132; 3032–3046 L. Wittner et al.

rare at resting potential. Our data reinforce this observation.

Spontaneous excitatory postsynaptic potentials (EPSPs) could be

easily detected in all recorded cells (n = 19) but hyperpolarizing

potentials were evident in only six CA2 cells (Fig. 3A, Table 1).

Behaviour of human CA2 cells duringinterictal-like activityDuring interictal-like events in the subiculum, the majority of

pyramidal cells (�80%) hyperpolarize, while a minority of

�20% depolarize and sometimes discharge. We asked whether

CA2 pyramidal cells also display different behaviours during syn-

chronous events (n = 15 cells). We found both hyperpolarizing and

depolarizing responses of CA2 cells, in a rather different propor-

tion to that recorded from the subiculum (Table 1). At resting

membrane potential, large depolarizing synaptic potentials

occurred simultaneously with interictal-like events in 73% of the

cells (n = 11/15), whereas a smaller group of cells (27%, n = 4/15)

showed a larger hyperpolarizing responses (Fig. 3B and C). At rest,

the responses of a given cell to consecutive synchronous events

were either depolarizing or hyperpolarizing. Mixed responses, or

variable responses were not detected. The depolarizing or hyper-

polarizing nature of the response in a given cell did not change as

neuronal membrane potential varied between –50 and –70 mV.

Spontaneous IPSPs were detected exclusively in cells, where

interictal-like events were associated with hyperpolarizing poten-

tials (n = 4). In cells with depolarizing responses, only spontaneous

synaptic depolarizations, but no spontaneous hyperpolarizations,

were detected (n = 11).

Morphology of human CA2 cellsThree CA2 cells were fully recovered after biocytin injection.

All were pyramidal cells, with a pyramidal shaped cell body, one

thick apical dendrite traversing striatum radiatum and lacunosum-

moleculare and several thinner basal dendrites in striatum oriens.

Axon collaterals were well labelled for two of these cells and

ramified in the CA2 region as in the rat hippocampus (Mercer

et al., 2007). One of the cells projected to the CA3 region as

well (Fig. 4). In both cases, axons arborized in both the striatum

oriens and radiatum. Since our tissue samples were from epileptic

patients, it is possible that this axonal distribution results in part

from a reactive axonal re-organization.

Cation–chloride cotransporters in thehuman epileptic CA2 regionHuberfeld et al. (2007) suggested that different cellular behaviours

during interictal-like events in the subiculum reflect differences in

intra-cellular chloride homeostasis. We first used immunostaining

techniques to search for possible differences in the expression of

molecules that regulate Cl-homeostasis in CA2 neurons. Double

immunofluorescent staining was used to examine the presence of

the intra-cellular chloride accumulating NKCC1 and the chloride

extruding KCC2 molecules in specimens from seven epileptic

patients (Fig. 5, Table 2). As a control, we used hippocampal

tissue derived from two healthy monkeys perfused in another

study. While, we note that monkey data cannot be directly com-

pared with human epileptic tissue, both species are primates and

many anatomical features of the hippocampus are comparable

(e.g. Chan-Palay et al., 1986; Nitsch and Leranth, 1991; Seress

et al., 1991, 1993; Sloviter et al., 1991; Lavenex et al., 2009).

A cautious comparison might then be realistic. In the monkey

hippocampus, 99.4% of examined CA2 pyramidal cells were posi-

tive for NKCC1 (n = 788/793) and 98.6% of the cells were posi-

tive for KCC2 (n = 782/793). These results suggest that both

chloride cotransporter molecules are present in nearly all pyramidal

cells of the healthy primate CA2 region. Our electrophysiological

results, where about two-thirds of CA2 cells showed depolarizing

response to synchronous bursts, led us to expect a significant

decrease in KCC2-positive cells in the CA2 region. However, we

found that the proportion of the NKCC1-positive cells was lower,

87.5� 5.4% (n = 1681/1892 cells from seven patients) than in the

monkey CA2 region, and the proportion of KCC2-positive cells

was slightly lower, at 93.8� 2.2% (n = 1785/1892 cells). There

was some variability in tissue from the different patients, but con-

siderable numbers of NKCC1-negative pyramidal cells were always

evident. Interestingly, the proportion of NKCC1+/KCC2� cells

was similar in control monkey and epileptic human tissue

Figure 4 CA2 pyramidal cells project to the CA2 (n = 2) and to the CA3 (n = 1) regions. Axon collaterals were found in the strata

oriens, pyramidale and radiatum. In this NeuroLucida drawing, cell body and dendritic tree are shown in red, axons in blue. Inset shows

a photomicrograph from the same cell. P050125_C5.

Epileptiform activity in the CA2 region Brain 2009: 132; 3032–3046 | 3039

(1.01 and 1.05%, respectively). The ratio of double negative cells

was elevated in the epileptic tissue (5.10% compared with 0.38%

in control monkey CA2), suggesting that, in this region, reactive

changes of the two chloride cotransporters in an epileptic brain

could be linked.

Somatic inhibitory input to humanepileptic CA2 pyramidal cells

Light microscopy

The hypothesis that the strength of synaptic inhibition is reduced

in an epileptic brain has been debated for many years. A decrease

in the number of PV-positive interneurons has been described in

the epileptic CA2 region, suggesting that inhibitory circuits are

altered (Andrioli et al., 2007). An absence of spontaneous hyper-

polarizing synaptic potentials in some CA2 pyramidal cells suggests

that inhibitory systems in this region may be changed but that

interneurons remain functional. The inhomogeneous distribution

of perisomatic inhibitory axons reported in the CA2 region

(Arellano et al., 2004) further suggests that structural changes in

inhibitory systems might result in the preservation of inhibitory

inputs to some, but not all cells. We examined this point

by comparing PV-immunostaining of the CA2 region in tissue

from two control subjects and five epileptic patients (four hippo-

campal sclerosis, one non-hippocampal sclerosis). In the control

CA2 region, numerous PV-positive interneurons were present

(see also Braak et al., 1991; Seress et al., 1993). The somata of

Figure 5 Double immunofluorescent data show that the chloride accumulating cotransporter NKCC1 (green) and the chloride

extruding KCC2 (red) are expressed by nearly all cells in the healthy monkey CA2 region (A). Arrowheads show the NKCC1+/KCC2+

cells. In the human epileptic CA2 region (B) both cotransporters are downregulated, although to different extents: 13% of CA2

pyramidal cells were NKCC1-negative (arrows), while only 6% were immunonegative for KCC2. Numerous double negative cells were

observed in the epileptic CA2 region (asterisks). From Patient P050118.

Table 2 Proportions of NKCC1- and KCC2-positive pyramidal cells in control monkey and epileptic human CA2 region

NKCC1+KCC2+ (%)

NKCC1+KCC2� (%)

NKCC1�KCC2+ (%)

NKCC1�KCC2� (%)

NKCC1+(%)

KCC2+(%)

Monkey (n = 793) 98.3 1.0 0.3 0.4 99.4 98.6

Epileptic human (n = 1 892 cells from 7 patients) 86.4�5.6 1.1� 0.4 7.4� 5.2 5.1� 2.3 87.5� 5.4 93.8� 2.2

3040 | Brain 2009: 132; 3032–3046 L. Wittner et al.

the majority of the multi-polar inhibitory cells were located in the

striatum pyramidale, with long dendrites traversing the entire

width of the CA2 region, while other horizontal PV-positive cells

were observed in the striatum oriens (Fig. 6A1). The axonal net-

work in the pyramidal layer was homogeneous (Fig. 6A2). As in

the previous studies (Arellano et al., 2004; Andrioli et al., 2007),

we noted a strong decrease in PV-positive cells and axonal

branches in the human epileptic CA2 region. Very few PV-positive

cells were observed in epileptic tissue (Fig. 6B1). Axonal staining

was inhomogeneous, with basket formations visible around the

somata of some cells, but apparently absent from others

(Fig. 6B2).

Electron microscopyThese data suggest that perisomatic inhibition of CA2 pyramidal

cells may be reduced. However, PV has been shown to disappear

from cell bodies and dendrites in rat (Sloviter, 1991; Magloczky

and Freund, 1995) and non-human primate (Scotti et al., 1997a, b)

models of epilepsy, as well as in the human epileptic hippo-

campus (Wittner et al., 2001, 2005). Therefore, we used electron

microscopy to examine the profiles of synaptic terminals contact-

ing the soma of CA2 cells from one control and three epileptic

patients with hippocampal sclerosis. This procedure provides

another view of inhibitory innervation that does not depend on

the PV content of presynaptic axon terminals. We measured the

soma perimeter of CA2 cells and the active zone lengths of all

synapses they received. The synaptic coverage (micrometres of

synaptic length/100 mm soma perimeter) (Wittner et al., 2001,

2005) was determined for each examined cell.

In control tissue, CA2 pyramidal cell bodies were innervated

exclusively by symmetrical, presumably inhibitory synapses

(n = 28 cells, Fig. 7A, Table 3). The percentage of PV-positive term-

inals was 17.5%, the proportion of perforated synapses (more than

one active zone/axon terminal) was 10.7% (n = 103 boutons exam-

ined). The synaptic coverage of the cells varied between 0.14 and

1.49 mm synaptic length/100 mm soma perimeter, with a mean of

0.80� 0.08 (mean� SEM, Fig. 7C), and showed a normal

distribution (Kolmogorov–Smirnov normality test). The average

length of active zones was 0.175� 0.006 mm (mean� SEM).

In the dentate gyrus (Wittner et al., 2001) and CA1 region of

human epileptic tissue (Wittner et al., 2005), numerous symme-

trical, presumably inhibitory axon terminals innervate principal cell

somata. Asymmetrical, presumed excitatory synapses terminated

on cell bodies only occasionally. In the CA2 region of all three

epileptic samples, many symmetrical synapses (Fig. 7B) terminated

Figure 6 PV-stained interneurons in the CA2 region of the human control (A) and epileptic (B) hippocampus. Numerous multi-polar

cells are located in the PCL in the control tissue (arrows on A1), whereas their number dramatically decreases in epileptic tissue.

A homogeneous axonal network is stained in the control CA2 region (arrowheads on A2). In epileptic tissue, PV-positive basket

formations (arrowheads on B2) surrounded some cells (triangle), but were absent from others (asterisk). From control C11 and

from epileptic patient P960930.

Epileptiform activity in the CA2 region Brain 2009: 132; 3032–3046 | 3041

on pyramidal cell bodies. Unexpectedly, we also observed frequent

contacts made by asymmetrical synapses (Table 3, Fig. 7C). Most

presumed excitatory boutons showed similar characteristics to

those of mossy terminals (Amaral and Dent, 1981). They were

of large size, typically with a diameter 42 mm, densely packed

with many round vesicles, some dense core vesicles and numerous

mitochondria (Fig. 7B3). We therefore quantified symmetrical

(inhibitory) and asymmetrical (excitatory) synaptic coverage

separately for CA2 pyramidal cells (Table 3, Fig. 7C). In all three

samples, we examined the somatic input of CA2 cells in a single

plane (from one 55 nm thick section), to avoid biased sampling.

With this method, asymmetrical synapses were detected on the

soma of 50, 70 and 62% of CA2 pyramidal cells examined, while

symmetrical synapses were present on all examined cells.

The mean symmetrical synaptic coverage of the cells in the

epileptic CA2 region was similar to that of cells from control

subjects (Table 3, Fig. 7C). However, the variability between

cells was higher in epileptic tissue (from 0.11 to 2.75 mm synaptic

length/100 mm soma perimeter). In one patient with numerous

degenerating pyramidal cells in the CA2 region (P960830), the

mean symmetrical synaptic coverage of the non-degenerating

pyramidal cells was lower than in the others, but the difference

was not significant (Kruskall–Wallis one-way ANOVA). The distri-

bution of the symmetrical synaptic coverage of CA2 cells in epi-

leptic tissue was comparable with that of control tissue, although it

did not pass the normality test. The proportion of PV-stained axon

terminals facing CA2 cell bodies was dramatically decreased:

6.25% of terminals were PV-positive in one patient (n = 80 axon

Figure 7 Electron micrographs show the somatic input of CA2 pyramidal cells in the control (A) and the epileptic (B) CA2 region.

Pyramidal cell somata are shown on A1 and B1. PV-positive (A2 and B2) and PV-negative (A3 and B3) axon terminals formed

symmetrical synapses on the cell bodies. Asymmetrical synapses including large mossy terminals (B4, MT) innervated CA2 cell somata

(PC) only in the epileptic tissue. Symmetrical synaptic coverage (micrometre synaptic length/100 mm soma perimeter) was found to be

similar in the control and epileptic tissue (C, mean� SEM). From control C11; and epileptic Patients P960930 (B1–3), P020117 (B4),

P960830.

3042 | Brain 2009: 132; 3032–3046 L. Wittner et al.

terminals examined), and no PV-positive terminals were found in

two patients (n = 54 and 46 axon terminals examined). The synap-

tic coverage data did not reveal two distinct groups of cells, one

displaying and the other lacking somatic inhibition. The percentage

of perforated synapses was increased in the epileptic samples:

26.7% on average (Table 3). The average length of the synaptic

active zones of symmetrical synapses was similar to the control

value in two patients, and was significantly lower in the patient

with the degenerating pyramidal cells (P50.01, Mann–Whitney

rank sum test).

In summary, electron microscopic analysis showed that somatic

inhibitory input CA2 pyramidal cell somata is preserved in epileptic

tissue. PV-immunopositivity is largely absent from inhibitory

terminals and an extra excitatory input, which seems likely to

derive from sprouted mossy fibres, innervates pyramidal cell bodies.

DiscussionHippocampal tissue obtained during surgery from epileptic patients

provides an excellent opportunity to explore cellular and network

properties of the human hippocampus. This study focused on the

CA2 region, which is resistant to the severe cell loss related to TLE

(Sloviter et al., 1991). The CA2 region forms quite a large subfield

in human hippocampus. However, it has been relatively neglected

in work on epilepsy, in favour of studies on the robust synaptic

re-organization in the dentate gyrus and the degenerative changes

in the CA1 region (for reviews see Cossart et al., 2005; Magloczky

and Freund, 2005). One study has described electrophysiological

properties of CA2 cells in slices of tissue obtained from epileptic

patients (Williamson and Spencer, 1994), but network activity in

the human CA2 region has not yet been investigated. Here, we

report that the human epileptic CA2 region generates an indepen-

dent epileptiform activity, similar in some respects to the interictal-

like activity described in the subiculum.

Excitatory signalling was functional in epileptic CA2 pyramidal

cells but hyperpolarizing inhibitory signalling was clearly present in

only a minority of cells (cf. Williamson and Spencer, 1994).

Immunostaining for Cl cotransporters suggested that chloride

homeostasis of CA2 cells was considerably perturbed, although

the reduced expression of NKCC1, and to a lesser extent KCC2,

would be expected to strengthen hyperpolarizing inhibition.

Immunostaining for the inhibitory cell marker PV, suggested that

perisomatic inhibition of CA2 cells might be reduced. However,

electron microscopic analysis instead revealed that somatic inhibi-

tory inputs were preserved and showed that PV was absent

from surviving interneurons. It also demonstrated aberrant somatic

excitatory inputs.

Role of excitatory signallingIn the human epileptic subiculum in vitro, spontaneous interictal-

like activity seems to depend on both glutamatergic and

GABAergic signalling. Depolarizing GABAergic responses in a

minority of pyramidal cells, with an altered chloride homeostasis,

have been suggested to contribute to the generation of this

interictal-like activity (Cohen et al., 2002; Huberfeld et al.,

2007). The present study demonstrates that similar synchronous

bursts, consisting of a field potential deflection with elevated

neuronal firing, are generated in the CA2 region in vitro. As in

the subiculum, pyramidal cells were either depolarized or hyper-

polarized during interictal-like events. However, the proportion of

the two responses differed. In the CA2 region about two-thirds of

the pyramidal cells showed depolarizing responses, whereas in the

subiculum only �20% of the cells depolarize during interictal-like

events. This result might be biased by the relatively small number

of cells (n = 15) recorded in our study. However, the current sink

in the cell body layer demonstrated by CSD analysis also indicates

the importance of excitatory signalling in the generation of

the synchronous events. Differences between the cellular compo-

sition, lamination and afferent connections in the CA2 region and

the subiculum (Lorente de No, 1934; Sloviter et al., 1991; Seress

et al., 1993) might also contribute to this dissimilarity. The com-

bination of a functional excitation in all cells with the reduced

inhibition suggests that the balance of excitation and inhibition

is turned towards excitation, possibly favouring the emergence

Table 3 Symmetrical (inhibitory) and asymmetrical (excitatory) synaptic coverage of CA2 cell somata in control andepileptic tissue

Control 11(n = 28 cells)

P960930(n = 21 cells)

P020117(n = 18 cells)

P960830(n = 10 cells)

Symmetrical synaptic coverage (micrometre synaptic length/100 mmsoma perimeter, mean� SEM)

0.80� 0.08 0.81� 0.12 0.83� 0.15 0.61�0.12

Asymmetrical synaptic coverage (mm synaptic length/100 mm somaperimeter, mean� SEM)

– 0.46� 0.15 0.26� 0.11 0.48�0.19

Cells receiving asymmetrical synapses (%) – 62 50 70

Number of axon terminals giving synapses to somata n = 103 n = 121 n = 69 n = 64

Ratio of boutons giving symmetrical synapses (%) 100 66.1 78.3 71.9

Ratio of PV+ terminals (%) 17.5 6.3 0 0

Ratio of boutons giving perforated synapses (%) 10.7 25.0 25.9 30.4

Average length of synaptic active zones (mm) 0.175 0.168 0.177 0.137�

Ratio of boutons giving asymmetrical synapses (%) – 33.9 21.7 28.1

Ratio of boutons giving perforated synapses (%) – 39.0 26.7 38.9

Average length of synaptic active zones (mm) � 0.175 0.216 0.219

*P50.01.

Epileptiform activity in the CA2 region Brain 2009: 132; 3032–3046 | 3043

of synchronous population bursts. Our electron microscopic find-

ings support previous data, that in the human epileptic hippocam-

pus, mossy fibres sprout into the CA2 region (Houser et al., 1990)

and form a novel somatic excitatory input to pyramidal cells.

Synaptic excitation mediated by these aberrant inputs may con-

tribute to the somatic current sink and the generation of interictal-

like bursts in this region. This point could be tested further by

asking whether the depolarizing cellular responses that accompany

interictal-like events in most CA2 cells possess a GABAergic

component, or whether they depend exclusively on glutamatergic

signalling.

Although interictal-like events recorded in the CA2 region and

in the subiculum differ in some respects, we should note the

similarities between synchronous events recorded from the CA2

region in vitro and in vivo records of interictal spikes from the

subiculum of epileptic patients. In vivo, two types of subicular

interictal spikes display different CSD patterns (Fabo et al.,

2008). Type 1 interictal spikes, like the synchronous events in

the CA2 region in vitro, showed a current sink in the cell body

layer, suggesting the presence of an active, perisomatic, excitatory

mechanism (Fabo et al., 2008).

Role of GABAergic signallingWe detected spontaneous EPSPs and IPSPs in CA2 pyramidal cells

suggesting that both excitatory and inhibitory synaptic circuits are

functional in this region. Williamson and Spencer (1994) reported

that inhibitory synaptic potentials were absent in most CA2

pyramidal cells from epileptic patients with hippocampal sclerosis.

Inhibitory potentials were recorded in only 3 of 9 recorded cells

and some events possessed slow kinetics, suggesting that they

could have resulted from the activation of GABAB rather than

GABAA receptors. In this work, spontaneous IPSPs with fast

rather than slow kinetics were detected in 6 out of 19 recorded

CA2 cells.

There was a perfect correlation between cells in which hyper-

polarizing spontaneous IPSPs were detected and those in which

hyperpolarizing responses accompanied synchronous interictal-like

bursts (n = 4). In contrast, spontaneous synaptic events were

exclusively depolarizing in cells which exhibited depolarizing

responses during interictal-like bursts (n = 11). These distinct cellu-

lar responses, together with the inhomogeneous distribution of

perisomatic axon terminals revealed by PV immunostaining,

could be consistent with a structural impairment of GABAergic

inhibition in a large subset of CA2 cells. However, this hypothesis

was not supported by the electron microscopic data which

revealed no or little difference between the numbers of presumed

inhibitory symmetric boutons contacting the soma of epileptic

and control CA2 pyramidal cells. Similar to the findings in

the CA1 region (Wittner et al., 2005), the symmetrical synaptic

coverage was decreased only in samples with evidence for

pyramidal cell degeneration (P960830). Changes in dendritic

inhibition or alterations in the expression of GABAA receptor

subunits (Loup et al., 2000), neither of which we examined,

might also contribute to the absence of spontaneous IPSPs in

some pyramidal cells.

The decrease in PV-positive elements coupled with the preser-

vation of somatic inhibitory contacts indicates a disappearance of

PV from perisomatic interneurons, a reactive response also

observed in the dentate gyrus and the CA1 region (Wittner

et al., 2001, 2005). Similar results have been found in a non-hu-

man primate model of epilepsy, with a significant loss of

perykaryal PV in the CA2 region of the Mongolian gerbil (Scotti

et al., 1997a). In the human epileptic CA2 region, PV immuno-

staining is much decreased in perisomatic terminals as well as the

cell body. This differs from data obtained from the dentate gyrus

and the CA1 region (Wittner et al., 2001, 2005), where the ratio

of PV-positive boutons terminating on axon initial segments and/

or somata was unchanged or increased in the epileptic samples,

compared to the control. The lack of PV modifies Ca2+-buffering

and so changes the functional properties of these interneurons (for

review see Schwaller, 2009). Investigations in PV�/� knockout

mice showed that PV deficiency facilitated the depolarizing

switch in the polarity of GABAergic responses induced by high-

frequency stimulation. In the PV–KO mice, these depolarizing

responses to GABA resulted in hypersynchronous firing and an

increased susceptibility to seizures induced by pentylenetetrazol,

a GABAA receptor blocker (Schwaller et al., 2004). Similar

mechanisms might operate in the human epileptic CA2 region,

resulting in an absence of hyperpolarizing potentials in some

cells even while somatic inhibitory inputs were preserved. In this

way, the absence of PV might contribute to the generation of

interictal-like activity in the CA2 region.

Chloride homeostasisTwo recent anatomical studies have suggested that the chloride

accumulating NKCC1 cotransporter is upregulated and KCC2, the

chloride extruding cotransporter, is downregulated in some cells of

the epileptic human subiculum (Palma et al., 2006; Munoz et al.,

2007). An increased expression of NKCC1 coupled with a reduced

KCC2 should elevate internal chloride concentrations and favour

depolarizing actions of GABA. Combined physiological and anato-

mical work revealed no immunopositivity for KCC2 in a proportion

of cells in which depolarizing events accompanied interictal field

potentials (Huberfeld et al., 2007). With a greater proportion of

CA2 cells showing depolarizing interictal-like responses, we

expected a considerable downregulation of KCC2, with a possible

upregulation of the NKCC1 cotransporter. In contrast, we found

that NKCC1 was absent from �13% of CA2 pyramidal cells,

and KCC2 from only �6%. While we noted a rather elevated pro-

portion of cells in which both cotransporters were absent, this situa-

tion should tend to reinforce a hyperpolarizing GABAergic inhibition

in the CA2 pyramidal cell population. Thus the discrepancy between

our electrophysiological and immunofluorescent data indicates that

changes other than expression of these cotransporters may account

for the impaired GABAergic signalling in epileptic CA2 cells.

ConclusionsWe have shown that interictal-like events are spontaneously gen-

erated in the CA2 region of the epileptic human hippocampus

3044 | Brain 2009: 132; 3032–3046 L. Wittner et al.

in vitro. Their form was similar to those recorded from the sub-

iculum, but our data suggest that the underlying mechanisms may

differ. Instead of a modified chloride homeostasis, an aberrant

perisomatic innervation of CA2 pyramidal cells by excitatory affer-

ents may contribute to a perturbation of the balance between

excitation and inhibition in this region. The importance of

excitation is supported by the presence of a current sink in the

pyramidal layer, a high proportion of cells that depolarized during

interictal-like events and the existence of mossy fibre terminals on

pyramidal cell somata. Some evidence seemed to suggest that a

reduction in the efficacy of synaptic inhibition might also modify

the balance between excitatory and inhibitory processes. It

included both single cell electrophysiology and the loss of

PV-immunopositivity both in somato-dendritic elements and

axonal processes. However, electron microscopy revealed that

somatic inhibitory inputs were preserved suggesting that PV

expression in perisomatic interneurons was lost. Further, immuno-

staining for Cl cotransporters showed that NKCC1 was more

strongly downregulated than KCC2. This modification is inconsis-

tent with a depolarizing shift in basal Cl equilibrium potentials in

CA2 cells. Thus, our results suggest that an aberrant excitatory

mossy fibre input to pyramidal cell somata together with func-

tional changes in the PV-positive perisomatic inhibitory network

resulting in depolarizing GABAergic responses may underly the

generation of interictal-like activity in the epileptic CA2 region.

AcknowledgementsWe wish to thank Drs M. Palkovits, P. Sotonyi and Zs.

Borostyanko00 i-Baldauf (Semmelweis University, Budapest) for

providing control human tissue, and Ms K. Ivanyi, K. Lengyel, E.

Simon and Mr Gy. Goda for excellent technical assistance. Thanks

to R. Csercsa and A. Magony for the data analysis program

SpikeSolution.

FundingINSERM (Poste vert to L.W.); the Bolyai Janos Research Fellowship

(to L.W.); French and Hungarian governments (NKTH-Fr38/2006,

ANR; OTKA49122, ETT135/2006); and the European Union (FP7

EPICURE FP6 EC LSH-CT-2006-037315, NeuroProbes EU IP

IST-027017).

Supplementary materialSupplementary material is available at Brain online.

ReferencesAmaral DG, Dent JA. Development of the mossy fibers of the dentate

gyrus: I. A light and electron microscopic study of the mossy fibers and

their expansions. J Comp Neurol 1981; 195: 51–86.

Andrioli A, Alonso-Nanclares L, Arellano JI, DeFelipe J. Quantitative

analysis of parvalbumin-immunoreactive cells in the human epileptic

hippocampus. Neuroscience 2007; 149: 131–43.

Arellano JI, Munoz A, Ballesteros-Yanez I, Sola RG, DeFelipe J.

Histopathology and reorganization of chandelier cells in the human

epileptic sclerotic hippocampus. Brain 2004; 127: 45–64.

Babb TL, Kupfer WR, Pretorius JK, Crandall PH, Levesque MF. Synaptic

reorganization by mossy fibers in human epileptic fascia dentata.

Neuroscience 1991; 42: 351–63.

Braak E, Strotkamp B, Braak H. Parvalbumin-immunoreactive structures

in the hippocampus of the human adult. Cell Tissue Res 1991; 264:

33–48.Chan-Palay V, Kohler C, Haesler U, Lang W, Yasargil G. Distribution of

neurons and axons immunoreactive with antisera against neuropeptide

Y in the normal human hippocampus. J Comp Neurol 1986; 248:

360–75.

Chen H, Luo J, Kintner DB, Shull GE, Sun D. Na(+)-dependent chloride

transporter (NKCC1)-null mice exhibit less gray and white matter

damage after focal cerebral ischemia. J Cereb Blood Flow Metab

2005; 25: 54–66.

Clemens Z, Janszky J, Szu00 cs A, Bekesy M, Clemens B, Halasz P. Interictal

epileptic spiking during sleep and wakefulness in mesial temporal lobe

epilepsy: a comparative study of scalp and foramen ovale electrodes.

Epilepsia 2003; 44: 186–92.Cohen I, Miles R. Contributions of intrinsic and synaptic activities to the

generation of neuronal discharges in in vitro hippocampus. J Physiol

2000; 524 (Pt 2): 485–502.

Cohen I, Navarro V, Clemenceau S, Baulac M, Miles R. On the origin of

interictal activity in human temporal lobe epilepsy in vitro. Science

2002; 298: 1418–21.

Cossart R, Bernard C, Ben-Ari Y. Multiple facets of GABAergic neurons

and synapses: multiple fates of GABA signalling in epilepsies. Trends

Neurosci 2005; 28: 108–15.

de Curtis M, Avanzini G. Interictal spikes in focal epileptogenesis. Prog

Neurobiol 2001; 63: 541–67.

de Lanerolle NC, Kim JH, Robbins RJ, Spencer DD. Hippocampal

interneuron loss and plasticity in human temporal lobe epilepsy.

Brain Res 1989; 495: 387–95.

de Lanerolle NC, Kim JH, Williamson A, Spencer SS, Zaveri HP, Eid T,

et al. A retrospective analysis of hippocampal pathology in human

temporal lobe epilepsy: evidence for distinctive patient subcategories.

Epilepsia 2003; 44: 677–87.

Dietrich D, Clusmann H, Kral T, Steinhauser C, Blumcke I, Heinemann U,

et al. Two electrophysiologically distinct types of granule

cells in epileptic human hippocampus. Neuroscience 1999; 90:

1197–206.

Fabo D, Magloczky Z, Wittner L, Pek A, Ero00 ss L, Czirjak S, et al.

Properties of in vivo interictal spike generation in the human

subiculum. Brain 2008; 131: 485–99.

Franck JE, Pokornny J, Kunkel DD, Schwartzkroin PA. Physiologic and

morphologic characteristics of granule cell circuitry in human epileptic

hippocampus. Epilepsia 1995; 36: 543–58.Gabriel S, Njunting M, Pomper JK, Merschhemke M, Sanabria ER,

Eilers A, et al. Stimulus and potassium-induced epileptiform activity

in the human dentate gyrus from patients with and without

hippocampal sclerosis. J Neurosci 2004; 24: 10416–30.

Houser CR. Granule cell dispersion in the dentate gyrus of humans with

temporal lobe epilepsy. Brain Res 1990; 535: 195–204.

Houser CR, Miyashiro JE, Swartz BE, Walsh GO, Rich JR, Delgado-

Escueta AV. Altered patterns of dynorphin immunoreactivity suggest

mossy fiber reorganization in human hippocampal epilepsy. J Neurosci

1990; 10: 267–82.

Huberfeld G, Wittner L, Clemenceau S, Baulac M, Kaila K, Miles R, et al.

Perturbed chloride homeostasis and GABAergic signaling in human

temporal lobe epilepsy. J Neurosci 2007; 27: 9866–73.Isokawa M, Fried I. Extracellular slow negative transient in the dentate

gyrus of human epileptic hippocampus in vitro. Neuroscience 1996;

72: 31–7.Lavenex P, Lavenex PB, Bennett JL, Amaral DG. Postmortem changes in

the neuroanatomical characteristics of the primate brain: hippocampal

formation. J Comp Neurol 2009; 512: 27–51.

Epileptiform activity in the CA2 region Brain 2009: 132; 3032–3046 | 3045

Lorente de No R. Studies on the structure of the cerebral cortex - II.Continuation of the study of the ammonic system. J Psychol Neurol

1934; 46: 113–77.

Loup F, Wieser HG, Yonekawa Y, Aguzzi A, Fritschy JM. Selective altera-

tions in GABAA receptor subtypes in human temporal lobe epilepsy.J Neurosci 2000; 20: 5401–19.

Magloczky Z, Freund TF. Delayed cell death in the contralateral hippo-

campus following kainate injection into the CA3 subfield. Neuroscience

1995; 66: 847–60.Magloczky Z, Freund TF. Impaired and repaired inhibitory circuits in the

epileptic human hippocampus. Trends Neurosci 2005; 28: 334–40.

Magloczky Z, Wittner L, Borhegyi Z, Halasz P, Vajda J, Czirjak S, et al.Changes in the distribution and connectivity of interneurons in the

epileptic human dentate gyrus. Neuroscience 2000; 96: 7–25.

Masukawa LM, Burdette LJ, McGonigle P, Wang H, O’Connor W,

Sperling MR, et al. Physiological and anatomical correlates of thehuman dentate gyrus: consequences or causes of epilepsy. Adv

Neurol 1999; 79: 781–94.

Masukawa LM, Higashima M, Kim JH, Spencer DD. Epileptiform dis-

charges evoked in hippocampal brain slices from epileptic patients.Brain Res 1989; 493: 168–74.

Masukawa LM, O’Connor WM, Burdette LJ, McGonigle P, Sperling MR,

O’Connor MJ, et al. Mossy fiber reorganization and its possible

physiological consequences in the dentate gyrus of epileptic humans.Adv Neurol 1997; 72: 53–68.

Masukawa LM, Uruno K, Sperling M, Oconnor MJ, Burdette LJ. The

functional relationship between antidromically evoked field responsesof the dentate gyrus and mossy fiber reorganization in temporal lobe

epileptic patients. Brain Res 1992; 579: 119–27.

Masukawa LM, Wang H, O’Connor MJ, Uruno K. Prolonged field poten-

tials evoked by 1 Hz stimulation in the dentate gyrus of temporal lobeepileptic human brain slices. Brain Res 1996; 721: 132–9.

Mercer A, Trigg HL, Thomson AM. Characterization of neurons in the

CA2 subfield of the adult rat hippocampus. J Neurosci 2007; 27:

7329–38.Munoz A, Mendez P, DeFelipe J, Alvarez-Leefmans FJ. Cation-chloride

cotransporters and GABA-ergic innervation in the human epileptic

hippocampus. Epilepsia 2007; 48: 663–73.Nitsch R, Leranth C. Neuropeptide Y (NPY)-immunoreactive neurons in

the primate fascia dentata; occasional coexistence with calcium-

binding proteins: a light and electron microscopic study. J Comp

Neurol 1991; 309: 430–44.Palma E, Amici M, Sobrero F, Spinelli G, Di Angelantonio S, Ragozzino D,

et al. Anomalous levels of Cl- transporters in the hippocampal subicu-

lum from temporal lobe epilepsy patients make GABA excitatory. Proc

Natl Acad Sci USA 2006; 103: 8465–8.Payne JA, Rivera C, Voipio J, Kaila K. Cation-chloride co-transporters in

neuronal communication, development and trauma. Trends Neurosci

2003; 26: 199–206.Payne JA, Stevenson TJ, Donaldson LF. Molecular characterization of a

putative K-Cl cotransporter in rat brain. A neuronal-specific isoform.

J Biol Chem 1996; 271: 16245–52.

Pirker S, Czech T, Baumgartner C, Maier H, Novak K, Furtinger S, et al.Chromogranins as markers of altered hippocampal circuitry in temporal

lobe epilepsy. Ann Neurol 2001; 50: 216–26.

Proper EA, Oestreicher AB, Jansen GH, Veelen CW, van Rijen PC,

Gispen WH, et al. Immunohistochemical characterization of mossyfibre sprouting in the hippocampus of patients with pharmaco-resistant

temporal lobe epilepsy. Brain 2000; 123 (Pt 1): 19–30.

Rivera C, Voipio J, Thomas-Crusells J, Li H, Emri Z, Sipila S, et al.

Mechanism of activity-dependent downregulation of the neuron-specific K-Cl cotransporter KCC2. J Neurosci 2004; 24: 4683–91.

Rosene DL, Van Hoesen GW. The hippocampal formation of the primate

brain: a review of some comparative aspects of cytoarchitecture andconnections. In: Jones EG, Peters A, editors. Cerebral cortex. Vol. 6.

New York: Plenum Press; 1987. p. 345–456.

Schwaller B. The continuing disappearance of ‘‘pure’’ Ca2+ buffers. Cell

Mol Life Sci 2009; 66: 275–300.

Schwaller B, Tetko IV, Tandon P, Silveira DC, Vreugdenhil M, Henzi T,et al. Parvalbumin deficiency affects network properties resulting in

increased susceptibility to epileptic seizures. Mol Cell Neurosci 2004;

25: 650–63.

Scotti A, Bollag O, Kalt G, Nitsch C. Loss of perikaryal parvalbuminimmunoreactivity from surviving GABAergic neurons in the CA1 field

of epileptic gerbils. Hippocampus 1997a; 7: 524–35.

Scotti AL, Kalt G, Bollag O, Nitsch C. Parvalbumin disappears from

GABAergic CA1 neurons of the gerbil hippocampus with seizureonset while its presence persists in the perforant path. Brain Res

1997b; 760: 109–17.

Seress L, Gulyas AI, Ferrer I, Tunon T, Soriano E, Freund TF. Distribution,morphological features, and synaptic connections of parvalbumin- and

calbindin D28k-immunoreactive neurons in the human hippocampal

formation. J Comp Neurol 1993; 337: 208–30.

Seress L, Gulyas AI, Freund TF. Parvalbumin- and calbindin D28k-immunoreactive neurons in the hippocampal formation of the

macaque monkey. J Comp Neurol 1991; 313: 162–77.

Seress L, Gulyas AI, Freund TF. Pyramidal neurons are immunoreactive

for calbindin D28k in the CA1 subfield of the human hippocampus.Neurosci Lett 1992; 138: 257–60.

Sipila ST, Schuchmann S, Voipio J, Yamada J, Kaila K. The cation-chloride

cotransporter NKCC1 promotes sharp waves in the neonatal rat

hippocampus. J Physiol 2006; 573: 765–73.Sloviter RS. Permanently altered hippocampal structure, excitability, and

inhibition after experimental status epilepticus in the rat: the ‘‘dormant

basket cell’’ hypothesis and its possible relevance to temporal lobeepilepsy. Hippocampus 1991; 1: 41–66.

Sloviter RS, Sollas AL, Barbaro NM, Laxer KD. Calcium-binding protein

(calbindin-D28K) and parvalbumin immunocytochemistry in the

normal and epileptic human hippocampus. J Comp Neurol 1991;308: 381–96.

Sommer W. Erkrankung des Ammonshornes als Aetiologisches Moment

der Epilepsie. Arch Psychiat NervKrankh 1880; 10: 631–75.

Ulbert I, Halgren E, Heit G, Karmos G. Multiple microelectrode-recordingsystem for human intracortical applications. J Neurosci Methods 2001;

106: 69–79.

Ulbert I, Heit G, Madsen J, Karmos G, Halgren E. Laminar analysis ofhuman neocortical interictal spike generation and propagation: current

source density and multiunit analysis in vivo. Epilepsia 2004a; 45

(Suppl 4): 48–56.

Ulbert I, Magloczky Z, Ero00 ss L, Czirjak S, Vajda J, Bognar L, et al. In vivolaminar electrophysiology co-registered with histology in the hippo-

campus of patients with temporal lobe epilepsy. Exp Neurol 2004b;

187: 310–8.

Uruno K, O’Connor MJ, Masukawa LM. Alterations of inhibitory synapticresponses in the dentate gyrus of temporal lobe epileptic patients.

Hippocampus 1994; 4: 583–93.

Williams JR, Sharp JW, Kumari VG, Wilson M, Payne JA. The neuron-specific K-Cl cotransporter, KCC2. Antibody development and initial

characterization of the protein. J Biol Chem 1999; 274: 12656–64.

Williamson A, Spencer DD. Electrophysiological characterization of

CA2 pyramidal cells from epileptic humans. Hippocampus 1994; 4:226–37.

Wittner L, Ero00 ss L, Czirjak S, Halasz P, Freund TF, Magloczky Z.

Surviving CA1 pyramidal cells receive intact perisomatic inhibitory

input in the human epileptic hippocampus. Brain 2005; 128: 138–52.Wittner L, Ero00 ss L, Szabo Z, Toth S, Czirjak S, Halasz P, et al. Synaptic

reorganization of calbindin-positive neurons in the human hippo-

campal CA1 region in temporal lobe epilepsy. Neuroscience 2002;

115: 961–78.Wittner L, Magloczky Z, Borhegyi Z, Halasz P, Toth S, Ero00 ss L, et al.

Preservation of perisomatic inhibitory input of granule cells in the

epileptic human dentate gyrus. Neuroscience 2001; 108: 587–600.Wozny C, Knopp A, Lehmann TN, Heinemann U, Behr J. The subiculum:

a potential site of ictogenesis in human temporal lobe epilepsy.

Epilepsia 2005; 46 (Suppl 5): 17–21.

Wyler AA. Corpus callosotomy. Epilepsy Res Suppl 1992; 5: 205–8.

3046 | Brain 2009: 132; 3032–3046 L. Wittner et al.