h’euroscience Vol. 14, No. 2, pp. 375GtO3. 1985 030&4522/85 S3.00 + 0.

00
Printed in Great Britain Pergamon Pres Ltd
IBRO

COMMENTARY

LIMBIC SEIZURE AND BRAIN DAMAGE PRODUCED

BY KAINIC ACID: MECHANISMS AND RELEVANCE
TO HUMAN TEMPORAL LOBE EPILEPSY

Y. BEN-ARI
Laboratoire de Physiologic Nerveuse, Departement de Neurophysiologie appliquee, 91190 Gif-sur-Yvette,
France

“Perhaps the word kainic comes from Cam”

U.S. Von Euler at a meeting

CONTENTS
Introduction
Specific binding sites for kainate
Potent excitatory effects of kainate
Preferential action of kainate on limbic structures
Parenteral administration of kainate: a limbic status epilepticus
Intra-amygdaloid injection of kainate: dissociation between local and “distant” seizure-related damage
Kainate-like endotoxins
Mechanisms of local damage
Mechanisms of “distant” damage
Relevance to human epilepsy
Clinical considerations
Relevance of animal models for human temporal lobe epilepsy
Conclusions

Kainic acid (KA-in Japanese literally the ghost of
the sea) was isolated three decades ago from the
seaweed Digenea simplex’“’ which has been exten-
sively used in post-war Japan to eradicate ascariasis.
The active principle of other biological extracts used
for the same purpose has also been identified (notably
domoic and quisqualic acids, ihid.). The anti-
helminthic action is shared by other analogues of
glutamate in particular those having carboxyl (and
carboxymethyl) substituents in the 2 and 3 positions.
The next important step was made by Shinozaki and
Konishi’“3.‘bl who showed that when microionto-
phoretically ejected upon cortical neurons these
agents have a potent excitatory action, consisting of
prolonged spike discharge. Kainate was the most
potent analogue.
It had been known for some time that parenterally
administered monosodium glutamate destroys retinal
ganglion cells in developing mice.” Olney’-“,‘”
showed that glutamate also destroys neurons of
various brain regions lying outside the blood-brain
barrier, notably the arcuate nucleus. The neuro-
toxicity of KA was first shown by Olney er u!.‘~*
using the same experimental paradigm. Parenteral
KA in immature mice caused rapid necrotic changes
Ahhreviarionr: 2DCi, 2deoxyglucose; GABA. y-amino-
butyric acid; KA, kainic acid.
in the somata and dendrites of arcuate neurons; this
was associated with hyperplasia of glial elements and
followed by phagocytosis which could be manifest as
early as 3-4 h after administration. Afferent fibers
and axons in the necrotic area appeared normal (also
see Ref. 134).
In subsequent reports, the relationship between the
structure of a series of glutamate analogues, their
excitatory potency (as suggested from micro-
iontophoretic experiments’%) and their toxicity to-
ward the arcuate was studied.39.‘34.‘5*Since every
analogue of glutamate which produced neuronal
excitation also caused dendrosomatotoxic activity,
the term “excitotoxin” was proposed by Olney and
coworkers’36.13nwhich implies that neuronal cell death
is due to excessive depolarization produced by the
excitatory substances. It was proposed that the sensi-
tivity of a given neuron to KA is determined by the
percentage of its dendritic surface occupied by glu-
tamatergic receptors. The resistance of axons was
interpreted as resulting from the lack of amino acid
receptors on fiber tracts.
Intracerebral iontophoretic ejection of glutamate
produces a lesion in the rat cerebral cortex.‘90 The
demonstration that direct intracerebral injection of
KA also produces an “axon-sparing lesion”‘*
prompted the use of KA as a tool in neurobiology.“*
The rational for this use stems primarily from the

375
316 Y. Ben-Ari

possibility to destroy selectively the neurons of a the epileptogenic properties of KA is of more than
given brain region and thus to circumvent the prob- theoretical interest.
lem of the contribution offibres depassuge, which are The preferential effect of KA on limbic structures
also destroyed by conventional (electrolytic) lesions. is the main subject of the present commentary. The
Such studies have provided several experimental reader is referred to other monographs for more
models of human diseases such a hemiballism74 or general reviews of the action of KA and other
Huntington’s chorea (e.g. Coyle et al.,” McGeer et excitotoxins,36".37,6*,10*

~1.‘~’ and Refs therein).

Other more recent studies indicate, however, that SPECIFIC BINDING SITES FOR KAINATE

both the use of KA as an axon-sparing device and the

High and low afinity sites
postulated mechanisms to explain KA neurotoxicity
must be reassessed. Following the initial observations of Simon et
(a) The concept that KA acts through glutamate al I66the presence in the brain of saturable, specific
receptors has been eroded by extensive biochemical biiding sites for KA has been confirmed and ex-
and pharmacological studies;6~37~y3~‘2y~‘30~‘56 KA does tended by several groups. 6,93~‘56.‘92 Whereas a single
not act on most glutamatergic receptors but it may site was described in the earlier report, two sites were
involve a special class of glutamatergic receptors with found in later studies. In the comprehensive work of
unique pharmacological profile and regional distribu- London and CoyleT’ the high and low affinity sites
tion in the brain.‘15 (&s of 4- 16 and 27-66 nM respectively) were dis-
(b) There is a clear-cut gradient of vulnerability of tinguished on the basis of their different association
neurons to kainate; thus the pyramidal neurons of the and dissociation constants. Kainate dissociates
CA3 region of the hippocampus are the most vulner- within seconds from the low affinity site, in contrast
able neurons in the brain KA’*’ whereas neurons of the dissociation from high affinity sites occurs in 1 h
the diagonal band of Broca or the mesencephalic or more (M. L. Berger, in preparation). The affinity
trigeminal nucleus are resistant to its toxic effects.34,88 of various analogues for the two sites differs as well
Even within the same region striking differences are as the regional distribution in the brain.93 The stri-
noted, viz. the CA3 region and the granular layer of atum, hippocampus and forebrain regions are partic-
the hippocampal formation.‘*‘.‘*’ Several obser- ularly enriched with high affinity sites whereas more
vations, notably those made by Nadler and than 90% of the total specific binding in the cere-
coworkers”“~‘2Z suggest that this gradient is not due to bellum or pons-medulla are to low affinity sites. The
a differential distribution or density of glutamatergic distribution of high affinity binding sites shows some
receptors (also see Refs 89 and 157). Other potent correlation with the regional vulnerability to KA (see
excitatory analogues do not reproduce this gradient below).
but destroy the neurons as a function of their prox-
imity to the site of injection.2.88.96.‘2’ Regional distribution
(c) Other observations suggest a dissociation be- To gain a better understanding of the possible
tween the excitatory and toxic properties of KA, for physiological significance of KA binding sites, it is
instance lesion of the corticostriatal pathway confers essential to map carefully their distribution in the
a protection of striatal neurons from the toxic action brain. In spite of the low specific activity of currently
of KAl8.102 without altering the excitatory effect of available radioactive KA, the distribution of KA
KA on striatal neurons.“’ ’ binding sites has been described recently with auto-
(d) The axon-sparing characteristic of KA injec- radiographic techniques. ‘5,“4~‘88The highest level of
tions has been disputed9*.“‘.‘*’ and, at least in some specific binding sites in the brain is located in the
brain regions, there is little doubt that fibre tracts are stratum lucidum of CA3 in the hippocampus (see Fig.
destroyed rapidly by local injection of KA.“’ 1C) i.e. the region innervated by mossy fibers origi-
(e) Intracerebral injections of KA also produces nating from the granular layer of the fascia dentata.
recurrent epileptiform discharge,“.6’ motor This region is also the most vulnerable to the ex-
convulsions”,6’.‘6’ and subsequently damage at the citatory and toxic properties of KA (see below).
site of injection and damage in structures distant Quantitative densitometry shows that the density of
from the injected site. 11.61.135.161.198 The distant damage
silver grains (representing bound kainate) in the
which is not caused by other major excitatory ana- mossy fiber region is ten times higher than in adjacent
logues (notably ibotenic acidsa,‘*‘) was initially attrib- layers.15 This labelling is due to binding to the high
uted to diffusion of the toxin combined with a greater affinity (slowly displaceable) sites (ibid.), as confirmed
vulnerability of the distant neurons to KA;‘35,‘98 we by a more recent study (M. L. Berger, G. Charton
have however discovered the role of the limbic status and Y. Ben-Ari, in preparation) in which the CA3
epilepticus and paroxysmal discharge generated by and CA1 plus fascia dentata were microdissected
KA as a cause of remote brain damage.” Since the separately; in contrast to CA3, the CA1 and fascia
regional distribution of this damage is reminiscent of dentata contained lower affinity sites. The auto-
that associated with human temporal lobe radiographs also revealed that other forebrain re-
epilepsy 33.54,64.65,9'.151,155.172,175
a better understanding of
gions are enriched in KA binding notably the amyg-
 Limbic seizure and brain damage produced by kainic acid
dala, deep layers of piriform cortex, striatum,
reticularis nucleus of the thalamus, whereas most
thalamic nuclei, the septum or the pons-medulla are
poor in silver grains.‘5.“4,‘88
Another crucial question, namely the pre- or post-
synaptic localization of KA binding sites is still
controversial. Thus, injections of kainate in the stri-
atum at doses which destroys most neurons produce
a 50-60% drop in specific binding sites, whereas
destruction of major inputs to the striatum produce
a modest or negligible reduction in specific KA
binding sites,6,'8.37.78.156
A
more recent study (M. L.
Berger, in preparation) suggests that decortication
has no effects on the number of KA binding sites
but leads to a slight reduction in their affinity for
the ligand. All of these observations suggest, in
keeping with eiectrophysioiogical studies7’ ‘05*‘06a
postsynaptic localization of both binding sites and
neuroexcitatory actions of KA.
Other observations suggest a presynaptic inter-
action between KA and the endogenous transmitter
causes the neuronal damage. Thus decortication pro-
tects vulnerable neurons from being kiIled,‘s~‘o’with-
out altering the excitatory effects produced by micro-
iontophoretic applications of KA.“’ Interestingly, as
this protection appears only after several days, this
might be due to a slow degeneration of boutons
possessing some of the KA binding sites (also see
below).
In the hippocampus, KA binding sites are readily
and rapidly reduced following a variety of pro-
cedures. Thus, both colchicine treatment” and intra-
cerebroventricular kainate12’ at doses which destroy
respectively the granular Layer of the fascia dentata
(and their mossy fibers) and the pyramidal layer of
CA3 eliminate the labelling as demonstrated by auto-
radiographs.“4 We have recently reexamined this by
using injections of KA in the amygdala to avoid
direct administration of toxins to the hippocampus;
we also used 9-12 day survival periods to avoid
problems related to the reorganization of synaptic
circuitry following KA.“’ A 60-707; reduction in
high affinity KA sites was found together with a
destruction of CA3 pyramidal layer but with limited
change in the mossy fibers as assessed with Timm’s
stain (M. L. Berger, G. Charton and Y. Ben-Ari, in
preparation). These findings, in keeping with mat-
uration studies (see below), suggest that the KA sites
are primarily located on the postsynaptic side in this
region.
Kainate-glutamate interactions
Electrophysiological studies made primarily in the
spinal cord’94 suggest that there are various classes of
excitatory amino acids including: (a) a N-methyl-
D-aspartate-preferring site also activated by ibo-
tenate, p-glutamate and antagonized by low concen-
trations of divalent metal cations and a series of
377
D-amino acids; (b) a second class comprising quis-
qualate and kainate subpopulations (also see Ref.
73). However, since these studies rely on extracellular
comparison of spike-inducing activity of various
compounds, it is difficult to meet strict pharma-
cological criteria for receptor classification (e.g. re-
views and discussion in Refs 129 and 130).
Using autoradiographic techniques, Monaghan et
a1.“5 have recently shown that KA sites in stratum
lucidum of CA3 also bind [3H]glutamate although
with ionic dependence and antagonist sensitivity
which differs from the major glutamate site observed
in biochemical studies. These authors suggest the
presence in the hippocampus of four pharma-
cologically and anatomically distinct glutamate bind-
ing sites. Because of the particular vulnerability of the
CA3 region to KA and the fact that this region is
particularly enriched in KA high affinity binding
sites, the pharmacological profile of these sites has
been extensively studied on membrane preparations.
Transmitters such as acetylcholine or centrally active
agents (diazepam, cardiazol or morphine,‘56 but also
flurazepam or [Leulenkephalin) do not displace KA
(M. L. Berger and Y. Ben-Ari, unpublished). The
mossy fibers are also particularly enriched in
dynorphiniu3 and ZnZ+ ‘v’ (e.g. Fredericksanb” and
Refs therein) but neither the former nor the latter (in
millimolar concentrations) influence the binding of
KA (M. L. Berger and Y. Ben-Ari, unpublished); we
have aiso found no displacement of KA by nerve
growth factor even if there are suggestions that this
substance may be released in this region.“’ Putative
endogenous kainate-like substances, notably folic
acid or quinolinic acid (see chapter below on
Kainate-like endotoxins) also produced negligible
displacement of KA from the CA3 sites (ibid.). These
observations suggest that the specific high affinity KA
sites in CA3 have rather restricted conformational
requirements. It bears stressing that the transmitter
released by the mossy fibers has not beeu fully
characterized; the evidence in favor of glutamate is
for instance less compelhng than it is for the Schaffer
collaterals or the perforant path.ls”.18~
POTENT EXCITATORY EFFECTS OF KAINATE
The particularly powerful excitatory action of
KA has been described in several brain
areaS 32.5R.72.105.1U6,l48.194
This effect has a slow onset and
prolonged duration; in several preparations recovery
is not obtained. The principal actions of KA on
invertebrate preparations is to potentiate glutamate-
induced depolarization.35,42.‘b3 In vertebrate spinal
cord, KA directly produces a large and often irre-
versible depolarization’7,3~,5~ which is associated with
a reduction in input resistance4’ and depends on the
presence of extraceliular sodium.‘o0 As stressed by
Nistri,“’ it is however possible that the activity of KA
is due to an increase in Ca2+ permeability. This would
readily explain the strong conductance increase and
378 Y. Ben-Ari

Fig. 1. Preferential metabolic alteration of the CA3 region. Following parenteral administration of KA,
a rise in 2-deoxyglucose autoradiographic staining is noticeable in the CA3 region in immature (A-12
days of age) or adult animals (arrow, D) (B). In immature animals, this is the only brain region in which
we have detected a rise in 2-DC. In adult, at doses which produce hmbic motor seizures other structures
of the hmbic system are labelled (D). Furthermore in both immature (not shown) and adult (C) specific
high affinity KA+ binding sites are concentrated in the CA3 region. The 2-deoxyglucose pictures in (A)
and (B) were superimposed in the stained sections from which the autoradiography had been obtained
(e.g. Ben-Ari et al.,14 Berger et al., I6 Tremblay et al.‘86). Abbreviations: cing, cingulate gyrus; Fi, fimbria;
g, granular layer; h, hilus; mol, molecular layer; p, pyramidal layer; SR, stratum radiatum.

Fig. 2. Selective and non-selective seizure-related damage produced by kainate. In (A) and (C), damage
restricted to parts of CA3 by intra-amygdaloid injection of KA. (A) Neuronal cell loss with Nissl stain
after survival period of 5 days; (B) extensive necrosis caused by intracerebroventricular (or also parenteral)
KA. (C) Argyrophylia seen also after intra-amygdaloid injection of KA with short survival periods with
a Fink-Heimer stain. Nissl stain.

Fig. 3. Long-term sequelae of parenteral KA in adult animals. Two months after administration of KA.
the animal displayed spontaneous seizures (A) with paroxysmal discharge in both the hippocampus (H)
and amygdala (A). These seizures were elicited often by handling. (B) and (C) depict in the same animal
the extent of neuronal cell loss in the hilar region (C) and dark degenerating cells in (B); clear-cut evidence
of phagocytosis was also often observed suggesting a continuous process of neuronal degeneration (e.g.
Tremblay and Ben-Ari’84). (A) Scale in microvolts; LMS, limbic motor seizures.
Fig. 5. Parallelism between the distribution of neurons vulnerable to KA in the hilar region and
GABAergic neurons revealed by glutamate decarboxylase (GAD) immunocytochemistry. With the
Fink-Heimer technique following parenteral KA and short survival periods argyrophylic cells are
particularly frequent in stratum oriens and pyramidale of CA1 and in the hilar zone (medial part of CA3
C and polymorph zone B). In (C)-(F), GAD immunocytochemistry was performed in control animals
following injections of colchicine in the cortex (L. Nitecka, M. Tappaz, E. Tremblay and Y. Ben-Ari,
unpublished observations); (e.g. also Riback et al.‘46 and Somogyi et al.“‘). The boxed area in (D) is shown
at a higher magnification in (F). Arrows in (E) and (F) point to GABAergic neurons. Abbreviations: a,
alveus; f, fimbria; g, granular layer; m, molecular layer; o, stratum oriens; p, stratum pyramidale; r.
stratum radiatum.

Fig. 7. Repetitive electrical stimulation of the perforant path produces a reduction of the Timm stain in
the mossy fibers. (A), (C), (E) Control; (B), (D), (F) after electrical stimulation. from Sloviter’6x.

Fig. 9. Bleaching of Zn’+ stain produced a K + pulse. Push-pull cannulae were stereotaxically introduced
in the mossy fiber zone of each hemisphere. A K+ pulse (IOmin, 30mM) was applied to the right
hemisphere and the animal perfused immediately at the end of the pulse. Note the reduction of the
intensity of Zn*+ stain as compared to the contralateral (control) side. Timm stain (e.g. Ref. 168)
counterstained with cresyl violet. Abbreviations: g, granular layer; h, hilus; IPSI, ipsilateral; p, pyramidal
layer.
12 days Adult

cing

Bst

Fig. 1

379
380
 0

b
24 ,

I a

j
.
.
.
.
.
382
Fig. 7.

383
Fig. 9

3x4
 Limbic seizure and brain damage produced by kainic acid
the powerful depolarization, since Ca*+ carries more
current than Na+ and has a very positive reversal
potential because of its very low intracellular free
concentration (e.g. Puil’“).
Because of the particularly potent seizure-inducing
effect of KA in the hippocampus, a number of studies
have examined its electrophysiological effects in this
region. Microiontophoretic applications of KA to the
hippocampus in situ produce a prolonged increase in
firing rate of pyramidal cells. The effect is particularly
prominent in CA3 where removal of the ionto-
phoretic retaining current is often sufficient to pro-
duce an intense activation of neurons.” CA1 and
CA3 pyramidal cells have a similar sensibility to
glutamate or ibotenic acid, thereby further stressing
the unique features of the actions of KA in CA3.
De Montigny et ~1.~ have also suggested that the
GABAergic inhibition is not involved in this neu-
ronal activation since (a) GABA does not exert any
selective effect on the action of KA vs that of
glutamate; (b) benzodiazepines selectively block the
actions of KA in CAl-but not in CA3; this
antagonism is not associated with enhancement of
the inhibition produced by microiontophoretic appii-
cations of GABA (ibid.). Other studies have however
suggested that the action of kainate is at least partly
due to a reduction of the GABAergic inhibitory
drive. Thus, following both parenteral administration
of KA in the intact animal16’ or bath applications of
KA into slice preparation,‘06 there is a reduction of
the recurrent inhibition measured with the paired
pulse inhibition technique. A recent intracellular
study” has indeed shown that the principal action of
KA on in vitro CA3 and CA1 pyramidal neurons
appears to be a synaptic depression, which first
influence the inhibitory postsynaptic potentials and
(with higher doses) the excitatory postsynaptic poten-
tials. This depression of inhibition is associated with
minimal effects on membrane properties, action po-
tentials, fiber conduction, etc. It is not associated with
alterations in the response produced by pulses of
GABA, and it is best explained by a reduction of
presynaptic GABA release although a postsynaptic
effect is not completely ruled out. The excitatory
phenomena which occur following limited applica-
tions of KA were thus interpreted as being secondary
to decreased inhibition, in keeping with the hypothe-
sis that the removal of GABAergic inhibition plays
an important role in epileptogenesis in this region.’
The potent excitatory action of other agents-
notably acetylcholine’0~‘89-also involves a removal
of the inhibition. Interestingly, another study’48 has
shown that there is a large difference between the
concentration of KA required to induce a reversible
acceleration of spontaneous discharge (10 nM) and
the one required to produce depolarization with
apparently irreversible spike blockade (10 PM). This
difference must be kept in mind when the mechanisms
underlying the local and distant actions of KA are
examined (see below).
385
PREFERENTIAL ACTIONS OF KAINATE
ON LIMBIC STRUCTURES
Parenteral administration of kainate: a limbic status
epilepticus
In the adult rat parenteral administration of the
toxin produces a limbic seizure and brain damage
syndrome which has not been reproduced by other
excitatory amino acids. Since this illustrates, in a
remarkable manner, the central role of limbic cir-
cuitry in the actions of KA, it will be described with
some detail.
Clinical signs. The clinical signs produced by i.v.94
or i.p. I4 injections can be divided in several distinct
phases. During the first 20-30 min, the animals have
“staring” spells. This is followed by head nodding
and numerous wet-dog shakes for approximately
30min. The next phase which starts 1 h after KA is
characterized by the occurrence of individual recur-
rent limbic motor seizures. These are identical to the
seizures produced by repeated daily electrical stimu-
lation of the amygdala or other limbic structures69
and include masticatory and facial movements, fore-
paws tremor, rearing and loss of postural control.
The seizures then become progressively more com-
plex and prolonged, with a reduction in the interictal
pause. In the following phase (l-2 h), the animal
displays a full status epilepticus i.e. continuous con-
vulsions and abnormal electro encephalogram. We
have referred to this syndrome as a limbic motor
syndrome;14 tonico-clonic generalized convulsions
(i.e. of the bicuculline or picrotoxin type) are not
produced by kainate in adult animals, except with
massive doses about four or five times the lethal dose
(E. Tremblay and Y. Ben-Ari, unpublished). There-
fore the latter terminology which has often been used
in the literature for KA-induced seizures is inap-
propriate and should be reserved to describe pro-
cedures causing tonico-clonic convulsions and elec-
trographic seizures generalized to the entire cortex
from their onset.
Electric activity. Electrographic records from corti-
cal and subcortical structures have confirmed that the
three main phases of behavioral abnormalities (star-
ing, individual limbic seizures, status epilepticus)
correspond respectively to (a) localized paroxysmal
discharges in the hippocampus, (b) involvement of
other limbic structures-notably the amygdala-and
then (c) generalization to a number of other (non-
limbic) structures. The hippocampus has a particu-
larly low threshold to parenteral KA.‘4,94 This is in
keeping with electrophysiological studies showing
that in the sllice preparation, seizure activity is pro-
duced by particularly low concentrations of KA.‘48
The 2-deoxyglucose autoradiographic method. This
method also illustrates the central role played by the
hippocampus in the initial stage which follows par-
enteral administration of KA (Fig. IB). Adminis-
tration of lower doses of KA,94 or repeated injections
of diazepam at doses which block the seizures except
386 Y. Ben-Ari

at the most susceptible sites (Y. Ben-Ari and E. argyrophylic stains, darkly impregnated neurons are
Tremblay, unpublished) suggest that the temporal restricted to limbic and associated structures.‘”
pole of the hippocampal formation is preferentially With longer survival (1 day to 1 week) the histolog-
involved. A developmental study has also shown that ical changes, notably the neuronai cell loss, are
before the end of the third week of age (in rats), the largely restricted to the regions in which there had
Ammon’s horn is the only brain region in which a rise been a conspicuous rise in metabolism as reflected
in 2-deoxyglucose (2DG) labelling is observed follow- from the 2DG studies, i.e. hippocampus, amygdala,
ing parenteral KA (Fig. 1A and Tremblay et ~l.‘*~). piriform lobe, septum, medial thalamus, etc. (Fig. 2).
The lateral septum-which receives a direct As far as the late brain damage is concerned subtotal
projection from the Ammon’s horn-is also involved or total tissue necrosis is found in the temporobasal
at this early stage. During the second stage when portion of the brain where the damage extends
overt limbic motor seizures are produced, the 2DG through all cell layers from the piriform lobe to the
maps reveal an intense activation of the entire limbic occipital cortex up to the rhinal fissure dorsally (Fig.
system (Fig. 1D). With almost no exception, the 3B and Refs 12, 14, I61 and 174). Conversely there
structures in which there is a metabolic rise have are also regions of restricted damage such as the
massive direct axonal connections with one or more claustrum and adjacent deep cortical layers (clau-
of the following areas: the hippocampal formation, strum insulare) or the vulnerable part Ammon’s horn
the amygdaloid complex and the mediodorsal tha- (CA3) in which the neuronal cell loss is not associated
lamic complex (e.g. Refs 14 and 94, and Refs therein). with a full destruction of the neuropit. In the regions
Thus a rise in glucose consumption is noted in the of total or subtotal necrosis, in addition to the
target structures of the fornix pre- and post- neuronal cell loss, there is a loss of oligodendrocytes,
commissural efferent projections of the hippocampal demyelination, formation of astroglial scars and peri-
formation (i.e. the accumbens, ventral pallidum and venous hemorrhages as wet1 as vascular sprouting.‘7’
claustrum, anterior thalamic nuclei and infra-lilnbic ~(~~~-r~~~~ .~e4uelfft,. The long-term sequelae (I -3
cortex). The amygdaloid complex-which is also months) of parenteral KA further depict the in-
connected with the entorhinal cortex and other com- volvement of limbic structures. Firstly, in keeping
ponents of the hippocampal formation-is also with observations made following intracerebral injec-
highly labelled as well as structures to which the tions of KA,” and animals display “spontaneous
amygdala projects (notably the stria terminals and its seizures” notably when handled.‘XJ These seizures are
bed nucleus. Fig. ID). Finally, the labelhng in the of a fimbic type and are accompanied with par-
cortical mantle is restricted to the “limbic” cortex i.e. oxysmal seizure discharges in the amygdala and
the infra- and pre-limbic cortices as well as the hippocampus (Fig. 3A). Furthermore, examination
agranular insular, retrosplenial and perirhinal corti- of the brain (Fig. 3B) reveals a satellitosis in the
ces, several of these areas are projected upon by the Ammon’s horn and phagocytosis of neurons in strata
mediodorsal nucleus and the amygdaloid complex. oriens, radiatum and pyramidale several weeks after
Although the 2DG method does not allow conclu- the parenteral injection of KA i.e. at a time when the
sions to be reached concerning the labelling at the toxin has been removed from the brain.“’ This
cellular level. there is little doubt that the rise in reflects the on-going nature of the seizure and brain
metabolism occurs in relation to the propagation of damage.“’ In the pyramidal layer this late damage is
paroxysmal discharges to neuronal populations more conspicuous in CA3 than in CA 1.
which are axonally interconnected (e.g. Refs 14 and BIood,fhs and oxygen cwmmption. In order to test
94). The amygdala, hippocampal formation and me- directly the possible contribution to nerve tissue
diodorsal thalamus occupy a central position in the damage by a mismatch between increased local met-
postulated wiring diagram (ibid.). Another interesting abolic consumption and oxygen supply. we have
feature of the 2DG maps is the metabolic decrease measured blood flow, as well as partial pressure of
noted in several non-limbic structures such as most of oxygen and carbon dioxide in the vulnerable CA3
the neocortex, striatum, ventrobasal thalamus, etc. region of unanaesthetized freely breathing rats fol-
~~st~put~~~~l~~~~. The early histopathological lowing parenteral administration of KA.‘4”.““’ It was
changes (I -2 h after parenteral KA) are characterized found that seizure (as noticed by means of hippo-
by shrinkage and pycnosis of neuronal perikarya and campal records as well as the motor signs) are
swelling of dendrites. 136~‘6’.‘74 Pronounced oedema associated with a rise in the local blood flow which
and swelling of astrocytes are also apparent shortly more than compensates the increased oxygen con-
after KA application, at least with relatively high sumption (Fig. 4). Thus the PO, and PCO, remains
doses. These changes are largely distributed to the within normal limits during the observation period
hemisphere with minor alterations in cerebellum, (2 h), at a time when damage is caused. Severe
pons and medulla oblongata, and prominent changes seizures are actually associated with hyperoxia (and
in hippocampal formation and piriform cortex. These not hypoxia). This provides compelling evidence that
acute effects are associated with a dilatation of or- at least in this region the damage caused by par-
ganelles; the golgi apparatus, endoplasmic reticulum enteral kainate is not due to anoxic ischemic epi-
and adjacent astrocytes are swollen (ihid.). With sodes.‘““’
 Limbic seizure and brain damage produced by kainic acid 387

E A “,pp _‘.IcL . . ..Y_. .., =r;w-- w- ] 0.05rn”

346 356
363

I’(X), HlPV 60
~. k....i_._ _\
_ __-__L__ ___.--"Y__ .myx_ "l", Hr
40
3 20

12min. 3Omin. 14min

after Kainic Acid iniection (i.p, 9mg/kg /

Fig. 4. The damage in CA3 produced by kainate is not caused by a mismatch between local blood flow
and metabolism. The local partial pressure in CO, (PCOJ, oxygen (PO?), the blood flow (C.B.F.) and
hippocampal electroencephalogram (EA Hipp) were continuously recorded in a chronically implanted,
unanaesthetized freely breathing rat after parenteral administration of kainate. The PO, and PCO, were
measured by gas spectrometry, the gases are drawn via a special cannula into the analysis chamber by
an ultra high vacuum; the local blood flow was measured by the thermal clearance method. The cannulae
and electrodes (for the hippocampal electroencephalogram) were chronically implanted in the CA3 region
2 weeks before the experiment. At the day of the experiment the animal was placed in a hammock, left
to habituate for I h and the variables measured continuously and quantitatively before and (up to 3 h)
after the administration of KA. The rat was anaesthetized and perfused Immediately after the recording
sessions. Note that the seizures are associated with a rise in a local blood flow which more than
compensates the increased metabolism. Damage in CA3 was found even though there were no hypoxia
and hypercapnia in the CA3 region during the entire recording session. See e.g. Pinard et ~1.‘~’ and Refs
therein for techniques used.

Neurochemistry. Neurochemical changes after par- Ontogenesis of the ef/cts of parenteral kainate. In
enteral KA include a highly significant reduction in 3 week old animals, parenteral KA induces tonic or
the activity of glutamate decarboxylase, a marker for tonico-clonic convulsions, but not limbic motor
GABAergic synapses. This reduction is found in the signs.29,‘86 Limbic motor seizures are in fact first
hippocampus and amygdala but not in striatum or observed starting from the end of the third week; the
frontal cortex; the effect starts after 24 h and reaches shift from one type of syndrome to the other is
a maximum 2 days later. “J’~ This is in keeping with sudden with almost no intermediary stage. This
the electrophysiological experiments described ear- prompted us to study its anatomical, metabolic and
lier, and further reflects the preferential involvement biochemical correlates.‘6.‘3’.‘86 In young animals the
of these structures in the KA syndrome. Since both rise in 2DG labelling induced by KA is exclusively
in the hippocampus’ and amygdala” GABAergic restricted to the hippocampus (more precisely the
neurons are intrinsic, these observations reflect a CA3 region and hilus) (e.g. Fig. 1A) and to the lateral
direct damaging effect of the procedure on these septum, thus confirming the unique susceptibility of
structures. Neurons that are GABAergic are also these regions to KA. The rise in 2DG in this region
particularly vulnerable to KA in other structures is already observed in 3 day-old animals which have
(ibid. and Ref. 191) and in epileptogenic foci.‘45 immature mossy fibers when studied anatomically
Electrophysiological studies on hippocampal slices and electrophysiologically.3~‘78 Furthermore, in spite
from animals which received KA 2 weeks earlier also of the severe tonico-clonic type convulsions, there is
indicate a consistent reduction in the inhibitory no brain damage before the end of the third week of
mechanisms.59.9’ In keeping with these observations, life. It is only after, that the toxin produces limbic
the distribution of hippocampal neurons vulnerable motor seizures associated with metabolic rise in hip-
to parenteral KA (as revealed with argyrophylic pocampus, amygdala, limbic cortex, medial thal-
stains) is similar to the distribution of GABAergic amus, and with typical pathological changes.13’
interneurons as revealed by glutamate decarboxylase Therefore the occurrence of limbic seizures and the
immunocytochemistry (Fig. 5, see also Ribak et ~1.‘~~ presence of a fully mature mossy fiber system is
and Somogyi et d.“‘). In contrast, the cholinergic necessary to cause damage in the Ammon’s horn.
septohippocampal system is n&t sensitive to par- In the light of these results, the developing limbic
enteral KA77.X8.‘74 (see a!so note added in the proof). system seemed an interesting subject for testing the
388 Y. Ben-An

ontogeny and role of KA binding sites. Both Scat-
chard analysis and autoradiographs revealed high
affinity sites in the CA3 region at an early age but only
low affinity sites in the amygdala.‘” The high affinity
sites appear in the amygdala only at the end of the
third week of life, i.e. at a time when the KA
convulsions have a limbic aetiology (also see Ref. 25).
Suggested sequence of events ,following parenteral
kainate. To evaluate fully the sequence of events
produced by parenteral KA, it is essential first of all
to determine the amount of toxin reaching the brain,
in particular the vulnerable regions of the limbic
system. Earlier authors have underestimated
problem of the blood-brain barrier to KA and have
interpreted the effects of KA as due to a direct action.
This barrier is, however, highly impermeable to KA
as much as to other acidic amino acids. Using the
method described by Oldendorf’32 we found a brain
uptake index for KA in the hippocampus and other
limbic regions of 2-4x, this is close to the values
found with the impermeant tracer saccharose (M. L.
Berger, le Fauconnier, E. Tremblay and Y. Ben-Ari,
in preparation). After systemic injection of [‘H]KA,
about 250 fmol KA/mg wet tissue remained in brain
tissue unremovable by perfusion (ibid.). It is esti-
mated that the actual brain concentration of KA is
sufficient to produce paroxysmal discharge in CA3’4X
but two orders of magnitude lower than the concen-
trations required to directly produce damage by in
situ injections.6’ These observations cannot be recon-
ciled with the suggestions that the damage in the
hippocampus following parental KA is due to a direct
interaction of KA with (glutamatergic) receptors.“’
The progressive sequence of activation of the limbic
circuitry and the lack of damage (even with lethal
doses) to structures such as the striatum (rich in KA
receptors” and readily destroyed by direct injection
of the toxin37.3X) argue against the hypothesis of a
direct “toxic” action of parenterally injected KA.
A more likely explanation is that seizures are hrst
elicited directly by KA in CA3. The disinhibition
produced by removal of the powerful GABAergic
inhibition will contribute to enhance the synchronous
paroxysmal discharge. If the discharge is sustained, it
will propagate to the entorhinal cortex and the
lateral septum. Subsequently, it will recruit a limbic
circuitry in which the amygdala has a key role. I
suggest that recurrent activation of the amygdalo-
hippocampal axis leads to the epileptogenic damage.
Other studies in which the toxin was directly injected
in vulnerable regions have also provided compelling
evidence for a seizure-related damage produced by
KA. These will be described prior to a discussion on
the mechanisms of the damage.
Intra-atnygdaloid injection of kainate: dissociation
between local and “distant” seizure-related damage
Intra-amygdaloid injection of KA produces ini-
tially limbic motor seizures.“.‘3 Later, the symptom-
atology in the rat is more complex and includes barrel
rotation, circling behavior etc., reflecting the second-
ary involvement of extra-limbic structures. This is a
made1 of limbic seizures with a focal origin. In the
cat’*’ and monkey “.“* the symptomatology is also of
an amygdaloid type (chewing, salivation, ipsilateral
cloniae of the face in the cat and oral automatism in
the monkey). There is a secondary generalization to
other structures particularly in the cat but not so in
the monkey (ibid.).
The epileptiform activity which is first apparent in
the injected amygdala rapidly propagates to the
the ipsilateral Ammon’s horn-notably the CA3
region-and the contralateral amygdala.’ ’
2-Deoxyglucose autoradiographs’x7 reveal an ip-
silateral regional rise in labelling reminiscent of that
obtained after i.p. injection of the toxin. At the site
of injection-in keeping with the local effect of KA
in a large number of structures”.‘7.38.“X--there is
a rapid neuronal cell loss associated with profuse
microglial proliferation clearly apparent 1-2 days
after the injection. Damage is also seen in several
“distant” structures. The earliest pathological alter-
ations appear after 30-40 min from the local injection
after the animal has displayed a few limbic motor
seizures and are exquisitely restricted to the CA3a,b
region of the ipsilateral Ammon’s horn”.‘X7 (also see
Fig. 2A,C). Patches of pyramidal neurons in CA3a
(or b) and their distant apical dendrites are densely
argyrophylic. The stratum lucidum where the mossy
fibers terminate is in contrast vacuolated. Other parts
of the ipsilateral Ammon’s horn are progressively
affected: the CA3 area and hilus, then the CA1 area
whereas CA2 and the granular layer of the fascia
dentata are spared. In addition to the hippocampus,
other “distant” areas show degeneration after intra-
amygdaloid KA, these are part of the limbic sys-
tem,ll.l”7
The simplest explanation for the distant lesions is
that KA diffuses (either directly or through the
ventricle) and that a sufficient concentration of the
toxin reaches vulnerable neurons (such as the CA3
pyramidal neurons) to produce directly the damage.
However we have shown that administration of di-
azepam in doses sufficient to antagonize the limbic
seizures blocked the distant damage in the CA3
region without affecting the damage at the site of
injection.” It was therefore suggested that the KA
brain damage following intra-amygdaloid injections
had a dual aetiology: (a) local damage due to the
direct toxic action of KA and (b) “distant” action
mediated by the paroxysmal discharge and its associ-
ate convulsions. Other lines of supporting evidence
for a major role of the epileptiform discharge in
mediating the “distant” damage in the hippocampus
include (e.g. Refs 8 and 12):
(a) Similar metabolic and hippocampal changes
can be produced following microiontophoretic appli-
cations of KA in the amygdala at doses which are not
compatible with a significant diffusion to the hippo-
 Limbic seizure and brain damage produced by kainic acid 389

campus. *J*’ Also 2-5 nmol KA in the amygdala are
enough to produce destruction of the vulnerable CA3
region: this concentration is similar to that found
neurotoxic when applied directly into the ventricle’*’
and only twice the one which must be applied directly
in the hippocampus to produce a similar damage*.6’
(b) With a lateral or an oblique approach to inject
the toxin in the amygdala (i.e. conditions in which
diffusion to the ventricle is unlikely) there is a similar
hippocampal damage whereas injections of KA in the
vicinity of the ventricle (i.e. in the septum-preoptic
area’* or in the nucleus of the diagonal band36) do not
readily produce this damage, in keeping with the
fact that the symptomatology of the seizures pro-
duced by injections in the septum is not of the limbic
type, at least with brief survival periods.‘*
(c) The distribution of [3H]kainate following direct
intracerebral injections in the striatum is not consis-
tent with the hypothesis that the necrotizing action of
KA in distant regions is due to diffusion.‘24,‘54
(d) The temporal pole of the hippocampus adjacent
to the injected amygdala is less readily lesioned than
the (distant) septal pole.”
(e) There is an excellent correlation between the
severity of epileptiform activity recorded in situ and
the damage in the CA3 region. In an earlier study,‘*
every second of the hippocampal electrographic
tivity was classified in five categories (normal, post-
ictal depression, paroxysmal discharge with or with-
out concomitant motor events and record with
regular or irregular spikes); the hippocampal damage
correlated significantly with the total duration
post-ictal depressions and paroxysmal discharges as-
sociated with motor events but not with other abnor-
malities.
(f) Damage in the medial thalamus is occasionally
more severe contralaterally to the injected site.“.‘*
(g) Transection of the perforant path (through
which the paroxysmal discharge probably propagates
to the Ammon’s horn, see below) reduces post-ictal
depressions and the distant but not the local dam-
age.‘* In contrast, this procedure does not protect
the hippocampal damage following intracerebro-
ventricular injection of KA.‘20.‘22
(h) The correlations between the electrographic
records, the metabolic maps and the regional distri-
bution of neuronal damage suggest that axonal con-
nections with the amygdala rather than proximity to
the injection site play an important role. Several
*The statement made by French et aL6’ that doses used in
the hippocampus in their study were as much as 1000
times lower than these previously employed in the work
of Ben-Ari et al. (e.g. p. 2525) is unconvincing, since
French et al. compare the lowest doses injected in the
hippocampus which produce seizures and no damage,
with the highest doses which produce damage when
injected in the amygdala.“~12~‘87 The minimal dose pro-
ducing damage in the hippocampus following local
injection was 1.25 nmol (in 0.2 ~1) in the study of French
et al.t’
examples of this important feature of KA action have
been provided in the literature following application
of the toxin at various sites.s~‘2~36~‘so
(i) There are major differences between the charac-
teristics of local and distant damage, the former and
not the latter is associated with a massive, early
microglial proliferation” perhaps of vascular ori-
pin.‘18 In contrast, glial proliferation in the hippo-
campus after injection of kainate into the amygdala
is manifested 4-6 days after the injection, i.e. sub-
sequent to the onset of neuronal damage. The direct
action produced by KA on glial cells is indeed one of
the most impressive and consistent effects of local
damage.
(j) Repetitive electrical stimulation of the perforant
path, which plays a central role in mediating the
damage caused in the hippocampus by intra-
amygdaloid KA,‘* reproduces a similar pattern of
damage. 16’
The possibility that the distant damage is due to
orthograde or retrograde transport of the toxin can
be also excluded since it is not compatible with the
time course of events especially in the CA3 region
(e.g. Ref. 8).
All of these observations therefore strongly suggest
that following intra-amygdaloid KA, the paroxysmal
ac- discharge is generated in the entorhinal cortex where
the amygdala heavily projects.‘43 This discharge will
exert a powerful excitatory action on CA3 pyramidal
neurons primarily via the granule cells and their
mossy fibers.
of
KAINATE-LIKEENDOTOXINS
The observations that KA does not act through the
predominant type of glutamic acid receptors raise the
possibility that these effects could be mediated by
another endogenous substance. The regional distribu-
tion of KA receptors as revealed by autoradiography
in the Ammon’s horn reinforces this hypothesis. At
present, the following candidates have been proposed
as endogenous toxins since they reproduce some of
the features of KA.
Folates
Ruck et al.‘49 reported that a folate derivative
bound strongly to [3H]KA binding sites on rat cere-
bellar membranes and suggested that folates could be
KA-like substances. This idea was reinforced by
studies by Olney and coworkers’37 who showed that
intra-amygdaloid injections of folates reproduce the
seizures and distant brain damage caused by KA but
not the damage at the site of injection. Other obser-
vations, however, cannot be reconciled with this
scheme. In fact, recent studies have found that folates
bind very weakly to [-‘H]KA binding sites in cere-
bellum, striatum, frontoparietal cortex and also
major limbic structures including the hippocampus
and amygdala (IC,, of about 2 mM to displace 20 nM
KA).j7.1g5 In addition, injections of folates into the
390 Y. Ben-Ari

hippocampus produce little or no motor abnormal- agent in order to characterize it in the context of
ities or pathological sequelae in contrast to KA.57 A other exogenous excitotoxins.
detailed comparison of the electrographic, clinical,
metabolic and histopathological sequelae of intra- MECHANISMS OF LOCAL DAMAGE
amygdaloid injections of folic acid and KA reveals
major differences between the two syndromes’*’ and Local damage includes that one produced at the
cannot support the observations of Olney et ~1.‘~’ site of injection or structures poorly protected by the
Thus, following KA injections, the 2DG maps reveal blood-brain barrier following intracerebral or par-
an enhanced metabolism in the injected amygdala enteral KA respectively. According to the excitotoxic
and adjacent piriform lobe, but the most conspicuous hypothesis of neuronal death, interaction of KA or
rise in labelling was noted in the entire frontoparietal other excitatory amino acids with appropriate recep-
cortex up to the cingulate region, and in the ventral tors (which in the initial version of the concept were
thalamic complex and globus pallidus, i.e. structures assumed to be glutamatergic and located post-
not labelled following KA treatment.‘85,‘87 In keeping synaptically) produces intense depolarization. It was
with this, we found a complete necrosis of the postulated that this depolarization will cause
piriform lobe and a massive anoxic-ischemic type of “energy-dependent homeostatic mechanisms to draw
damage in the frontoparietal cortex including status heavily on the ceil’s energy stores in an effort to
spongiosus, ischemic cell changes with or without restore ionic balance, with cell death ensuing when
incrustations in the superficial layers of the cortex, energy stores are exhausted or lethal alterations in the
etc. Interestingly, several features of the damage composition of the cells internal milieu have oc-
caused by KA-in particular in the boundary zones of curred”.’ In addition, glia at the site of injection are
the frontoparietal cortex-are reminiscent of the grossly oedematous and swollen, presumably because
pathological alterations induced in primates and sub- of the large increase in extracellular K + associated
primates species by anoxiceischemic episodes, no- with the seizure. This may impair glial uptake, pro-
tably atmospheric decompression or intracarotid air duce a reduction in extracellular space and contribute
embolism,2’~‘27 conditions which are associated with to neuronal damage.
convulsions. We suggest that folic acid produces In keeping with the suggested sequence of events,
those toxic actions of KA which are related to electrophysiological studies have indeed shown that
anoxic-ischemic changes caused by the convulsion in hippocampal slices. high concentrations of KA
but not the neuronal cell loss which is selectively and produce intense depolarization associated with an
directly produced by the paroxysmal discharge in irreversible loss of spike discharge.14’ This effect
vulnerable regions (such as CA3, see below). appears to depend neither on synaptic
transmission’“‘~‘“’ nor on the presence of presynaptic
Cholinomimetics terminals.‘“5 Extensive structure activity studies made
Olney and coworkers’35” also reported that choli- primarily in the striatum.“” retina”’ or the arcuate
nomimetics injected in the amygdala caused a seizure region”15 suggest a relation between neuroexcitation
and brain damage syndrome which is un- potency and neurotoxicity. although the former has
distinguishable from KA except that, like folates, been usually assessed from extracellular studies made
these agents do not cause damage at the site of in different brain sites. For instance the dimethyl ester
injection but only the distant damage. From the data derivative of kainate or its N-acetyl derivative, both
presently available it would seem that the distant of which are electrophysiologically inactive, are de-
damage is also of the non-selective seizure-related void of neurotoxic effects at doses 40P 100 fold higher
type (see below). than KA itself. Strong support for the excitotoxic
hypothesis comes from studies showing that KA also
Quinolinic acid
does not kill neurons insensitive to its excitatory
Schwartz and coworkers’59 have recently reported actions (e.g. neurons of the mesencephalic trigeminal
that nanomolar amounts of the tryptophan metabo- nucleus34,45) whereas. neurons which are readily ex-
lite quinolinic acid, injected intracerebrally, produce cited by KA such as CA3 pyramidal neurons are also
axon-sparing neuronal damage. Neurochemical, clin- very vulnerable to its toxic effects.‘”
ical and pathological observations show that quin- Other studies however cannot be reconciled with
olinic acid mimics the effects of excitotoxins. Thus, this scheme and suggest a more complex aetiology.
the agent shares with ibotenic acid (but not KA) the Thus blockade of the excitation produced by KA in
ability to produce circumscribed damage with few cerebellar cultures does not prevent the damage.“’
convulsions and no distant seizure-related damage. Other observations suggest that the damage caused
Other observations derived from tissue culture ex- locally by KA is not explicable by the seizure dis-
periments indicate that, like for the case of KA, the charge generated by the toxin. Thus. administration
actions of quinolinic acid depend on the presence of of anticonvulsants at doses sufficient to prevent the
normal synaptic patterns and appropriate syn- development of paroxysmal discharges, fails to pre-
aptogenesis in regions where it produces a toxic vent the local neurotoxicity of KA. In contrast,
effect.‘95 Further studies must be performed with this anaesthesia with y-butyrylacetone or chloralhydrate-
 Limbic seizure and brain damage produced by kainic acid
pentobarbital protects hippocampal neurons,
suggesting that the anaesthetic and not the
anticonvulsant action of drugs account for
local protectionsZW (also see Ref. 12). Furthermore,
a comparison of seizure and damage produced in the
hippocampus by a variety of excitatory amino acids
shows that the neurotoxicity poorly correlates with
the neuroexcitatory effects;2~‘2’~‘y9
for instance N-
methyl-D-aspartate produces continuous electro-
encephalogram seizures for 2 h with little subsequent
damage (ibid., also see Ref. 61).
Lesion experiments suggest a dissociation between
the excitatory (postsynaptic) and toxic actions (in-
volving the presynaptic elements) of KA,‘8.‘0’.‘05and
raise the possibility of different receptors for both
effects.‘06 The toxicity of intrahippocampal KA de-
pends on the presence of the septohippocampal (cho-
linergic) and perforant (presumably glutamater~c)
pathways. ‘20.122 Interestingly, transection of the per-
forant path protects the granules of the fascia dentata
from kainate but not from ibotenate lesions.89,‘J7
Other observations also emphasize the unique fea-
tures of KA neurotoxicity which are not shared by
other potent excitatory analogues. The actions of KA
but not that of other excitants depends on the
presence of mature synaptic patterns; thus KA but
not ibotenate fails to produce damage before the end
of the third week of age in the rat.25.‘3’.‘47Also,
neurons are differentially susceptible to KA but uni-
formly vulnerable to other excitants, the damage
produced by the latter depending simply on their
proximity to the injection site.z.88.96.‘2’ Several neu-
ronal populations resistant to local KA-notably the
cholinergic neurons of the medial septum and the
diagonal band-are readily destroyed by local ibo-
tenate.** This suggests that the two agents exert their
toxic effects by different mechanisms.
To account for these observations, a number of
alternative changes to the initial excitotoxic concept
have been proposed. These include (a) accumulation
of toxic concentrations of glutamate through in-
hibition of glutamate uptake by KA’O’ (e.g. also
Johnston ef aLXS),(b) cooperative interaction between
KA and endogenous glutamate released from pre-
synaptic terminalsI and (c) presynaptic localization
of KA receptors which would release toxic doses of
glutamate and/or aspartate.56 There is however no
compelling evidence in support of these mechanisms
which have been challenged by other expe~menta~
observations. For instance, (a) dihydrokainate which
is a more potent inhibitor of glutamate uptake is
devoid of neurotoxic action,36” (b) the toxicity of KA
is unaltered by coadministration of glutamate in the
cerebellum63 and (c) the concentrations of KA re-
quired to release glutamate from cerebellar parailel
fibers are 100 fold higher than the concentrations
sufficient to kill the cells innervated by these fibers.63
Also the hypothesis that excitotoxicity is due to a
presynaptic release of toxic concentrations of glu-
tamate (or aspartatej is also handicapped by the poor
391
efficacy of these amino acids in causing damage when
directly administered intracerebrally. Thus, to cause
necrotic changes in the striatum, glutamate must be
chronically infused for a week or more at particularly
high concentrations. s-99 These high doses are as-
sumed to be necessary in order to overcome the
combined capacity of high and low affinity uptake
systems. It is however not clear in this perspective
how the damage would be so rapidly caused (l-2 h)
following electrical stimulation or kainate adminis-
tration in vulnerable regions (uide supra).
Summing up, the mechanisms of the “local” dam-
age remains to be clarified. In the opinion of the
present reviewer, the local toxic action of KA is
caused by a multiple and complex series of effects
which are not necessarily due to a single mechanism
and cannot be categorized in a single framework. The
mechanism of the toxicity would differ according to
the locus of injection and epileptogenicity of the
injected structure, the experimental conditions (ana-
esthesia, concentrations, the doses used etc.). A pos-
sible direct effect of KA on the vasculature of the
injected locus also deserves investigation.
MECHANISMS OF “DISTANT’* DAMAGE
This includes the damage produced in structures
distant from the injection site or in structures protec-
ted by the blood-brain barrier following intra-
cerebral or parenteral KA respectively. Two types of
distant damage should be discussed separately (e.g.
Fig. 2).
Selective seizure-related damage
This is causally related to the propagation of
seizure discharge through a wel~~haracteriz~ syn-
aptic connection and without the involvement of
deleterious effects associated with the convulsions.
The evidence in favor of this type of damage is
compelling for the CA3 region. Indeed, as already
indicated above, the damage produced by parenteral
KA in this region is not explicable by hypoxia or a
mismatch between oxygen demand and blood sup-
pl~.‘~* However a number of other structures may
share with CA3 this feature, notably the claustrum,
deepest layers of insular cortex (presumably clau-
strum insulare), lateral part of the amygdaloid com-
plex etc. (see below).
The evidence that damage to CA3 produced by
parenteral or intra-amygdaloid KA is caused by the
seizure has been discussed above and does not need
further elaboration here. There is also evidence sug-
gesting that seizure activity contributes to the hippo-
campal damage by intracerebroventricular KA.‘20.‘2)
Nadler and coworkers have stressed the crucial role
of the mossy fibers for the damage produced in CA3;
in fact destruction of the mossy fibres provides
immediate protection of the CA3 region against
intracerebroventricular KA but not against direct
intrahippocampal KA.‘20.‘22
392 Y. Ben-Ari

Fig. 6. Delayed changes in extracellular and intracellular free-Ca2+ concentration [Ca’+] in CA2-3 region
following repetitive stimulation of the fimbria. Upper [Ca?+] trace: extracellular recording with
Ca’+-sensitive microelectrode. Lower [Ca2+l trace: intracellular recording with same electrode: note large
fall in [Cal when electrode penetrated cell. Oscilloscope traces (at bottom) were recorded at times indicated
by broken arrows. Initially &he hmbrial stimulation evoked an antidromic spike and a partly reversed
inhibitory postsynaptic potential. During the 10 Hz tetanus (between arrows), there were only small
fluctuations in [Ca’+] (reference barrel contained 3 M KCI), but there followed a period of post-tetanic
depression, characterized by an absence of responses (including inhibitory postsynaptic potentials) to the
same fimbrial stimulation. About 30s after the end of the tetanus, paroxysmal responses suddenly
appeared, in the form of bursts of population spikes. This was accompanied by a huge increase in [Cal
(to a peak of about 700 JIM) that lasted about 1 min. The return of [Cal to the intracellular base line level
(close to 1 PM) was followed by gradual return to the inhibitory postsynaptic potential (now fully
reversed).
The upper [Cal trace shows a comparably delayed large fall in extracellular [Cal evoked in a second
run by &hesame stimulation shortly after the intracellular recording. Data were obtained in situ, from a
rat under urethane anaesthesia (K. Krnjevic, M. E. Morris and N. Ropert, unpublished observations).

Two types of mechanisms can be proposed to
underline the selective seizure-related damage. The
first one implies large increase in jntracelluIar Ca2+
associated with the seizures; this will exceed the cell
capacity to buffer the Ca?+ thus leading to an increase
in intracellular free Ca*+ which, by means of pro-
teolysis or blockade of the Ca2+ proton exchange
across the mitochondrial membrane may cause neu-
ronal damage.‘52 In support of this view, histo-
chemical data indicate that the CA3 pyramidal cells
have a limited Ca*.+ binding capacity whereas the
granules of the fascia dentata and their mossy fibers
are rich in Ca2+ binding proteinss3 (also see Ben-Ari’).
Direct evidence in favor of this hypothesis has been
recently obtained by Krnjevic and coworkers~,“~
who have determined the changes in extracellular and
intracellular Ca2+ in the hippocampus during seizures
with ion-sensitive electrodes. It was found that a
particularly important decrease in extracellular Ca2 +
occurs in the pyramidal layer of CA3 during seizures,
and that this is associated with an increase in intra-
cellular Ca*+ (Fig. 6). Other experiments suggesting
the involvement of Ca*+ in excitotoxic damage have
been reported.“‘.‘“’
Other observations also indicate that factor(s) per-
haps exclusively localized to the mossy fibers, may be
released and produce damage during the seizures. A
number of substances found in the mossy fibers
deserve some discussion. For instance, the mossy
fibers are rich in dynorphinio3 which conceivably
could play a role in this event. Perhaps more likely is
the possibility of an endogenous kainate-like com-
pound since this region is amongst the richest in the
brain in high affinity KA receptors.‘5~“4~‘88Inter-
estingly, KA binding in the stratum lucidum (in the
mossy fiber zone) and not in other hippocampal
regions is decreased by Ca’+;“’ this is consistent with
the observation of Beaumont et ~1.~that Ca’+ is a
potent inhibitor of KA binding. Since a significant
reduction in extracellular Ca*+ is readily observed
during seizures, 9o this will further contribute to the
toxicity by increasing the effectiveness of KA on its
receptor sites.
Nevertheless, it is possible that the toxic factor is
not the transmitter itself but another agent released
with the transmitter during severe seizure discharge.
Zn2’ is a Iikely agent. Hence, (a) Zn2+ is particuIarly
concentrated in the mossy fibers76 where it is found
in synaptic vesicles75 (the actual concentration of
Zn2+ in this region is amongst the highest in the
brain3’,@). (b) Zn2+ is accumulated by the mossy
fibers following parenteral administration of the
 Limbic seizure and brain damage produced by kainic acid 393

0 Artificlol CSF

Spontaneous release
rmm

t K+ (30 n-M) 2 min

Time (min)

Fig. 8. In situ release of endogeneous zinc in the Amman’s horn. A push-pull cannula was stereotaxically
implanted in the CA3 zone in urethane anaesthetized animals. Zinc was measured by atomic adsorption.
A spontaneous release of zinc was observed approximately twice the concentration of zinc was found in
the artificial cerebrospinal fluid (CSF). A pulse of K+ (arrow, 30 mM, 2 min) caused a five-fold increase
in the amount of zinc released followed by a prolonged decrease of zinc with levels close to the blank;
IS-20 min elapsed before the recovery (Ben-Ari et al. *’, G. Charton, C. Rovira, Y. Ben-Ari and V. Leviel,
submitted for publication).

radioactive metal.‘93 (c) There is some evidence that
the population spike produced in CA3 by the activa-
tion of the mossy fibers is reduced by chelating agents
or diet conditions in which zinc levels are re-
duced.79,‘93(d) During repetitive electrical stimulation
of the fascia dentata (when CA3 damage is produced)
there is a loss of zinc stain in the mossy fibers (Fig.
7, and Sloviter’68); this might be due to release of
Zn2+ during the paroxysmal discharge. In a recent
study, we have measured by atomic adsorption the
spontaneous and evoked release of endogenous Zn2+
in the Ammon’s horn of unanaesthetized rats using
push-pull cannulae stereotaxically implanted.%” It
was found that when the cannula is located in the
vicinity of the mossy fibers there is a significant
spontaneous release of Zn2+ (two to four times the
blank levels). A brief pulse of K+ (I min) produces an
evoked release of Zn2+ to (maximal) levels which are
two orders of magnitude above the blank. When the
cannula was located in the fimbria or other parts of
the Ammon’s horn there was neither spontaneous nor
evoked release (e.g. Fig. 8, and G. Charton, C.
Rovira, Y. Ben-Ari and V. Leviel, submitted for
publication). Also, a K+ pulse applied unilaterally
just prior to perfusion-fixation produced a bleaching
of histologically demonstrable Zn2+ (Fig. 9 and Ref.
8a). Release of Zn ‘+ has also been found in slice
preparations3”). (e) Earlier studies have also shown
that intracerebral application of Zn2+ also produce
seizures.30,‘40 The possible mechanisms through
which Zn2+ might contribute to the damage are not
yet clear in view of the large spectrum of cellular
activities in which Zn2+ is involved.“3 It is however
of particular interest to note that zinc blocks very
efficiently sodium-potassium adenosine triphos-
phatase46 and glutamate decarboxylase.‘97
Summing up, several features of the mossy
fibers-CA3 synapse might cause the selective seizure-
related damage of the CA3 pyramidal neurons. Per-
haps a likely sequence of events includes orthodromic
generation of paroxysmal discharge in the granular
layer of the fascia dentata by intra-amygdaloid KA;
this causes a removal of the GABAergic inhibition in
the hilar zone and thus a disinhibition of the mossy
fiber input to CA3. Since the granular cells are
resistant to epileptogenic procedures, they will dis-
charge for prolonged periods. The release of an
endokain will excite the CA3 pyramidal neurons by
means of the postsynaptic KA receptors which might
be directly involved in the control of Ca2+ channels.
This is associated with a decrease in the extracellular
Ca*+-and presumably an increased effectiveness of
the endokain-and an increase in intracellular Ca2+
directly producing the damage. In addition, large
shifts in the concentrations of zinc will contribute to
kill the CA3 pyramidal neurons by unclarified mech-
anisms. It is probably the convergence of all these
factors which underlines the particular vulnerability
of the CA3 pyramidal neurons which can be de-
stroyed by rather brief seizure episodes without an
involvement of vascular or other deleterious condi-
tions.
Non -selective seizure -related damage
This is also related to the seizures and, as such,
blocked by administration of anticonvulsants. How-
394 Y. Ben-Ari

ever, in contrast to the selective seizure-related dam- experimental animals. and i\ often a life time process
age, this is not associated with a specific synaptic with a complex interaction occurring between an
connection, releasing a toxic factor but rather with initial event (occurring at birth) creating a scar$ the
more general deleterious conditions caused by the latter favors the occurrence of seizures which may
convulsions in some brain structures. This is best also contribute to further brain damage. Therefore
illustrated by the necrosis seen in the piriform region even if species difl’erences are put aside, an animal
(e.g. Fig. 2) after parentera]“,‘” I” “’ or model cannot be expected to reproduce all the fea-
intracerebroventricular”‘.“5,“’ KA, but also intra- tures of such a complex disorder. The researcher is
amygdaloid folic acid.“’ The necrosis involves the faced with lifetime neurological disorders such as
entire frontobasal and temporobasal portions of the temporal lobe epilepsy v,ith the obvious necessity of
brain separated from normal cortex by the fissura adopting a very pragmatic attitude and accepting the
rhinalis dorsally, reaching rostrally the olfactory bulb fact that the tnodel will at best reproduce circum-
and occipital cortex. With survival times of l-3 days, scribed features of the disease. Hence. I shall briefly
the damage includes in addition to neuronal cell loss. present the aspects of human temporal lobe epilepsy
demyelination, astroglial scaring perivenous hemor- which deserve particular attention as they can be
rhages and extensive vascular sprouting.‘7J As partly interpreted in the framework of experimental
stressed recently by Sperk et al.,“” this incomplete data.
parenchymal necrosis is a lesion typical ol The hippocur?rpu/ formution und am~~gdulu occup~~ N

anoxic-ischemic damage, although the topography 01 cm trul positior2 in tcwiporirl lohc of’ c>pikpsJ. Initial11
the lesions differ from that produced by generalized described by Jackson,“’ temporal lobe epilepsy is
hypoxia or ischemia. The authors therefore suggested associated with a symptomatology which includes: (I)
that the overactivity caused by KA in this region sensory symptoms (visual. olfactory. gustatory, sen-
leads to release of water and metabolites and sub- sations. etc. .): (2) mental symptoms (clouded con-
sequently oedema; this by compressing drainage ves- sciousness, “forced thinking”, &jti VU,hallucinations,
sels against the skull in the front0 basal region result etc.); (3) visceral symptoms (epigastric sensations.
in local disturbance of blood flow and chewing with salivation. oral automatisms, etc.); (4)
anoxic&schemic changes. somatomotor symptom5 (tonico-clonic move-
This non-selective seizure-related damage is prob- ments.h’.h’.hu
ably also produced in a number of other brain regions There is a general consensus that the temporal lobe
including the lateral septum, the CA1 field of the occupies a central position in this syndrome. hence
Ammon’s horn and perhaps also medial thalamic the name. Thus, although initially controversial, elec-
nuclei. A severe parenchymal necrosis is also pro- troencephalographic studies using superficial and
duced in these regions’74 notably in CA1 which is deep electrodes have shown that the clinical symp-
more vulnerable to anoxic-ischemic episodes than tomatology is associated with the presence of par-
CA3.*” Interestingly, these regions have little or uone oxysmal discharge in the temporal lobe region.”
of the features associated with the selective seizure- More recently. the involvement or these regions is
related damage in CA3; for instance the lateral also suggested by positron emission tomography.“”
septum and medial thalamus are virtually devoid of Although the paroxysmal discharge is often present
high affinity specific KA receptors’5.‘h.“4.‘XX and zinc.7h in the hippocampal formation~~~in keeping with its
particularly tow threshold to epileptogenic
RELEVANCE TO HUMAN EPILEPSY procedures” the amygdaloid complex plays a more
direct role in the occurrence of several symptoms
Clinical considerations
notably the oro-alimentary signs.‘~hX.“7 Thus, in the
Psychomotor or temporal lobe epilepsy or complex case report made by Wieser,“‘h the episode began with
partial seizures as defined in the international recurring visceral symptoms. These were associated
classification”4 is seen mainly in adults.‘.“.hh Although with hippocampal right discharges usually starting in
seizure disorders frequently have their onset early in the amygdala: bilateral hippocampal discharges were
life, the seizures are more frequently of the gener- associated with an impairment of consciousness.
alized type in children younger than 15 years whereas Direct electrical stimulation of the amygdala- but
temporal lobe epilepsy is much less frequent (21”,, as not of the hippocampus reproduces several of these
compared to 56:/, in adults@j). Furthermore since signs which closely resemble the symptoms displayed
temporal lobe epilepsy is amongst the most difficult by patients during “spontaneous” seizures. Further-
type to treat medically, surgery has become a useful more, Buser et cd.” have shown that whereas all
form of therapy. 5’.55The symptomatology of tempo- patients display hippocampal-evoked responses to
ral lobe epilepsy, its electrographic, metabolic and stimulation of the amygdala. an amygdaloid response
histopathological sequelae have been reviewed in to stimulation of the hippocampus is only observed
earlier monographs to which the reader is re- in temporal lobe epilepsy, raising the possibility that
fcrrc ~,24.54.64.hfi.97.151.1S5.l72.l7S
This disorder is complex, the facilitation of hippocampoamygdaloid con-
has a large number of different aetiologies, includes nections is instrumental for the aetiology of this
a symptomatology which is not readily defined in disorder. This IS partIcularI\ important in view of the
 Limbic seizure and brain damage produced by kainic acid 395

anatomical evidence for the involvement of the amyg- the hypothesis and interpreted the lesion as ictal
dala (and not the hippocampus) in several clinical image or more precisely as resulting from
signs associated with epilepsy of the temporal lobe. In hypoxic-ischemic episodes associated with the seiz-
fact, in both rats and primates, the central amygdala ures. However, since epilepsy is associated with vas-
directly projects to motor structures coordinating cular dilatation and not constriction e.g. Ref. 141,
facial and masticatory movements; the amygdah is other suggestions have been made including excessive
also in a position to directly control vegetative and increase in metabolism (i.e. a mismatch between
autonomic function through its massive projections metabolic demand and blood supply), increased per-
to the hypothalamus and brain stem centerss0s’43 (e.g. meability of the blood-brain barrier, or also in-
Ben-Ari’ and Refs therein). Projections from the creased brain volume (cerebral oedema which would
amygdala to the striatum8’ are probably involved in lead to compression of vessels with herniation at the
some of the stereotypies associated with temporal supratentorial region and compression of the poste-
lobe epilepsy. These anatomical observations are rior cerebral arteries). Other theories include early
particularly relevant since they constitute, in the damage to the hippocampus due to birth or early
opinion of the present reviewer, a common ground infancy head trauma or febriles convulsions.33,h4,‘4’,‘55
for discussion between the clinician and the re- Since, there is no clinical consensus for aetio-
searcher. pathology of the damage, animal models are useful to
The temporal lobe region harbors the main patho- clarify this issue.
logical substrate of psychomotor epilepsy. Five de-
cades after the observations of Bouchet and Relevance of animal models for human temporal lobe
Cazauviehl” made in 18 15, Sommer made a detailed epilepsy
description of the “Ammon’s horn sclerosis” which To bear some relevance to human temporal lobe
consisted in pyramidal cell loss with a typical gradient epilepsy, the animal model must fulfill at least the
of vulnerability. “* The observation that a sclerosis of following criteria: (a) the hippocampus, amygdala
the hippocampus constitutes the most frequent dam- and other limbic structures must play a central role
age in post-mortem examination of chronic epileptics in the symptomatology; (b) it must reproduce a
has been repeatedly confirmed (it is on average found pattern of brain damage which is reminiscent of
in 50-7O’A of the cases). In the work of Margerison Ammon’s horn sclerosis; (c) it must reproduce the
and Corsellis9’ and Sano and Malamud”’ the patients spontaneous and repetitive seizures of the temporal
were clinically and electroencephalographically stud- lobe type that are the hallmarks of epilepsy; (d) it
ied and a significant correlation between the symp- must be relatively resistant to anticonvulsivants in
toms and the subsequent brain damage was noted. parallel to human temporal lobe epilepsy’4’” in order
The brain damage involves in addition to the to constitute a useful test model for developing new
hippocampus-where the most frequent degeneration drug treatment. Several animal models have been
affects the Sommer sector, the endfohum and in used to experimentally reproduce temporal lobe epi-
severe cases the granules of the fascia dentata-the lepsy.
amygdala, adjacent cortical regions, cerebellum, and Animal models which produce generalized
midline thalamic structures.33*4’,54.64 Several of these tonico-clonic generalized convulsions. Parenteral ad-
regions are vulnerable to a variety of deleterious ministration of a number of potent convulsants such
conditions such as febrile convulsions but also anoxia as bicuculline, allylglycine or pentetrazole, produce
and other conditions which are not associated with generalized convulsions (e.g. Refs 19, 108, 1 I I and
convulsions. 176, and Refs therein). These animal models however
Temporal lobe seizures are particulary resistant to do not fulfil the above mentioned criteria.
drug treatment and therefore surgery has been used (a) Electrographic and clinical data indicate that
for treatment of intractable epilepsy. With the advent the hippocampus and other limbic structures are not
of these procedures, the nature and characteristics preferentially involved in these convulsions as the
of the lesions have been examined in more detail. seizures are generalized from the onset to the entire
Studies such as those of the Scheibels’53 (also see cortical mantle (ibid.). In immobilized artificially
Brown”) have revealed the progressive time course of ventilated rats, Siesjii and Abdul-Rahman’hS have
pathological changes. Several reports have shown shown by means of 2DG autoradiography a meta-
that when the pathological abnormalities are re- bolic rise in the hippocampus; however in freely
moved, the patients benefit substantially from the moving animals there is no rise in metabolism in
procedure, i.e. there is general consensus that the hippocampus. I4 A recent study, in which a quan-
sclerotic area contributes to the temporal lobe sei- titative determination of 2DG in several brain regions
zures was made, has confirmed and extended this obser-
In contrast, there is less consensus amongst the vatiom4’ the overall picture with bicuculline in freely
clinicians on the aetiopathology of the brain dam- moving animals is that of a marked regional meta-
age. 33,54.64Initially Sommer”* suggested that the bolic depression, including in the hippocampus. This
sclerosis was the cause of epilepsy; the studies of also reflects the difficulties in extrapolating from
German pathologists notably Spielmeyer’ reversed immobilized preparations-on which most of the

N.S.C‘4,Z-B
396
studies using bicuculline have been made-to the
freely moving conditions.
(b) In a number of studies made in immobilized
artificially ventilated animals, bicuculline and similar
convulsants produce h~ppocampal damage; however
the lesions are quite widespread and include several
non-limbic structures (superficial layers of neocortex,
basal ganglia, cerebellum, etc.).9~‘08~‘W~“’Other studies
have failed to demonstrate significant hippocampal
damage,14.23.52.160
In one study made in freely moving
animals, damage to the basket cells of the cerebellum
was the most consistent observation.‘4 Soderfeldt and
coworkers have furthermore shown that after brief
periods of recovery the oedema caused by the con-
vulsants has vanished and most neurons look nor-
mal 4.170.171
(c) To the best of the present reviewer’s knowledge,
“spontaneous’~ seizures are not observed after such
procedures (e.g. Meldrum’08). Unless such seizures
are produced-and particularly seizures of the limbic
type-the relevance of these procedures to study
temporal lobe epilepsy can be questioned.
Using the bicuculline model, Meldrum and coworkers
initially concluded that damage to the hipp~ampus
primarily due to anoxic-Gschemic episodes.‘~~“” The vacu-
Jization noted in several brain regions was interpreted as
mitochrondrial swelling due to the anoxic events in keeping
with earlier studies in which similar alterations were re-
produced by anoxic episodes.2’ This view however has not
been confirmed by Siiderfeidt and coworkers who showed
that the vacuolization seen by light microscopy reflected
dilatation of the endoplasmic re&dum and cytoplasmic
vacuoles, and not mitochondrial swelling.“~“““~ In sub-
sequent studies, Meldrum et ~1.“’ noted that the cerebellar
damage was more reduced than the hippocampal one by
paralyzing the animal, suggesting that the former may be
causally related to the motor convulsions. Since in a later
study” hypoxia surprisingly reduced-rather than en-
hanced as expected-the hippo~mpal damage, the authors
suggested that the epileptic seizures per se could play an
important role in the aetiology of the lesion. Subsequent
studies in which the blood flow was measured provided
further evidence that the damage was not solely due to an
imbalance between metabolic demand and s~pply.‘~* These
experiments have not been confirmed however by Soderfeldt
et ul. using a similar preparation, since it was shown (a) that
hypoxia augments rather than reduces the severity of the
damagei70,‘7’and (b) the measurement of the ratio between
the local cerebral blood flow and glucose consumption
indicates that a mismatch in the hippocampus probably
plays an important role in the damage seen with the
bicuculline model.”
To summarize, procedures which produce
tonico-clonic generalized convulsions have provided
controversial results concerning the nature and re-
gional distribution of the damage and the aetiology
of this damage. It is the present reviewer’s opinion
that the pathological sequelae noted after severe
tonico-clonic generalized convulsion are explicable
by the general deleterious conditions associated with
the convulsions i.e. these procedures do not produce
the selective seizure-related damage. To obtain com-
pelling evidence for a causal role of the paroxysmal
discharge per se it is essential: (a) to directly measure
the Iocal PO,, PC02 and blood flow in vulnerable
Y. Ben-Ari
regions in order to exclude local hypoxia as an
important factor in the pathology; and (b) to define
in detail the anatomical substrate of the propagation
seizure activity and the possible preferential par-
ticipation of a given excitatory input to the hippo-
campus or other vulnerable regions.
Animal models which produce focal seizures with a
limbic symptomatology. To produce focal-secondarily
generalized seizures with a limbic symptomatology,
several experimental approaches have been used,
including local injection into the amygdala of alumi-
num cream6’ or cobalt t ‘1y~‘26~128Zn2+ into the lateral
ventricle30,46or of ouabain into the lateral ventricle or
septum. ‘39Another method is based on continuous
electrical stimulation of the amygdala in previously
kindled rat.““ Some of these procedures fulfil a
number of the aforementioned criteria including hip-
pocampal damage and ‘~spontaneous” seizures.
The kainate model-which has been more exten-
sively studied than the aforementioned procedures-
is particularly useful for the study of the evolution,
propagation and pathological consequences of epi-
leptic discharge in the limbic system. It clearly fultils
is the criteria mentioned above with regard to human
temporal lobe epilepsy since: (a) the hippocampus.
amygdala and other limbic structures play a central
role in its symptomatology’4.94”2.‘6’~‘R2(e.g. Ben-Ari’
and Refs therein); (b) the pattern of brain damage is
clearly reminiscent of Ammon’s horn sclerosis with a
similar gradient of vulnerability. A number of
differences have been noted, however, including the
neuronal cell loss in the granular layer of the fascia
dentata in humans4’ whereas this is resistant in the
animal models. However, cell counts were not made
in these studies and in a more recent work, patches
of necrosis were noted in this region after survival
periods of I-2 months (L. Nitecka, E. Tremblay and
Y. Ben-Ari, in preparation); (c) spontaneous seizures
with a limbic symptomatology are consistently noted
following parenterallg4 (Fig. 3) or intracerebral
KA.27~28In keeping with this, electrophysiological
studies in hippo~ampal slice suggest that the kainate-
lesioned hippo~ampi become epileptogeni~;s9,9’ (d)
available anticonvulsants are weak against the
seizures generated by KA.3’.‘79
The kainate model can shed some light on tempo-
ral lobe epilepsy in humans. For instance, the obser-
vation that selective seizure-related damage in CA3
can be produced without hypoxia or h~er~apn~a’42
suggests that restricted paroxysmal discharge may
produce a selective seizure-related localized damage
in the human hippocampus. Conversely, the damage
noted in the human temporal cortex might be due to
general disturbances associated with the seizures. If
the Amman’s horn region to which the mossy fibers
project is enriched in high affinity kainate receptors
(present in human brain@), this may indicate that the
release of an endokain from mossy fibers plays a role
in the aetiopathology of the sclerosis. Auto-
radiographic techniques and zinc stains on human
 Limbic seizure and brain damage produced by kainic acid 397

post-mortem material can be used to study the bio- thought to produce axon-sparing lesions. On the
chemical organization of this region in normal and other hand, studies made with kainate in the last few
epileptic brains. It will be of particular interest to see years reflect the considerable usefulness of KA and
whether there is also an abnormal sprouting of mossy other excitotoxins to reproduce experimentally a
fibers in the affected Ammon’s horn. In fact, Nadler number of models of interest to human pathology. As
and coworkers’*’ have recently shown that following the emphasis of the present reviewer’s own research
destruction of the CA3 pyramidal layer, the mossy suggests, it is the field of temporal lobe epilepsy which
fibers deprived of their postsynaptic targets, establish will particularly benefit from studies of the mech-
new connections, with the granules of the fascia anisms of action of kainate. The use of KA un-
dentata. This new circuit increases the excitability of doubtedly will provide us with a better understanding
dentate granule cells which now become epi- of the normal and pathological functioning of the
leptogenic. Further animal studies should examine hippocampus and other limbic regions.
whether surgical removal of the necrotic region elim-
inates spontaneous seizures in temporal lobe epilepsy;
there is some indication in the literature that this may Acknowledgements-1 am indebted to Drs A. Nistri, M.
be indeed the case.26 Berger and R. Naquet for their criticism and remarks. I also
acknowledge the help of Mrs D. Giraud for the bibli-
ography. Our studies on the kainate model were supported
CONCLUSIONS by the Fondation pour la Recherce Medicale and INSERM
(CRE 1953). I am also indebted to Dr. M. C. Demantay
Kainate is perhaps not as useful a tool as initially for her support.

REFERENCES

1. Aicardi J. and Chevrie J. J. (1970) Convulsive status epilepticus in infants and children. A study of 239 cases. Epilepsia
11, 167-197.
2. Aldinio C., French E. D. and Schwartz R. (1983) The effects of intrahippocampal ibotenic acid and its blockade by
2-amino-7-phosphonoheptanoic acid: morphological and electroencephalographic analysis. Expl Bruin Res. 51,3644.
3. Amaral D. G. and Dent J. A. (1981) Development of the mossy fibers of the dentate gyrus: I. A light and electron
microscopic study of the mossy fibers and their expansions. J. camp. Neurol. 195, 51-86.
3a. Assaf S. Y. and Chung S. (1984) Release of endogenous Zn 2+ from brain tissue during activitv. Nature 308. 734-736.
4. Atillo A., Soderfelt B., Kalimo H., Ollson Y. and Siesjo B. K. (1985) Pathogenesis of brain lesions caused by
experimental epilepsy. Light and electron microscopic changes in the rat hippocampus following bicuculline-induced
status epilepticus. Acta neuropath., Berl. In press.
5. Bancaud J., Talairach J., Morel P. et Bresson M. (1966) La come d’Ammon et le noyau amygdalien: effets cliniques
et electriques de leur stimulation chez I’homme. Revue Neural. 115, 329-352.
6. Beaumont K., Maurin Y., Reisine T. D., Fields J. Z., Spokes E., Bird E. D. and Yamamura H. I. (1979) Huntington’s
disease and its animal model: alterations in kainic acid binding. Life Sci. 24, 809-816.
7. Ben-Ari Y. (1981) (ed.) (1981) The Amygdaloid Complex. Elsevier/North-Holland, Amsterdam. p. 516.
8. Ben-An Y. (1981) Epilepsy: changes in local glucose consumption and brain pathology produced by kainic acid. Adv.
Biochem. Psychopharmac. 27, 385-394.
8a. Ben-Ari Y., Charton G., Leviel V. and Rovira C. (1985) Snontaneous and evoked release of Zn*+ in the Ammon’s
horn of the anaesthetized rat. J. Physiol., Lond. In press. .
9. Ben-Ari Y., KmieviC K. and Reinhardt W. (1979) Hinnocamnal __ _ seizures and failure of inhibition. Can. J. Phvsiol.
Pharmac. 57, 1462-1466.
10. Ben-Ari Y., KmjeviC K., Reinhardt W. and Ropert N. (1981) Intracellular observations on disinhibitory action of
acetylcholine in the hippocampus. Neuroscience 6, 2475-2484.
Il. Ben-Ari Y., Tremblay E. and Ottersen 0. P. (1980) Injections of kainic acid into the amvgdaloid complex of the rat:
an electrographic, clinical and histological study in relation to the pathology of epilepsy: Neuroscience 5, 515-528.
12. Ben-Ari Y., Tremblav E., Ottersen 0. P. and Meldrum B. S. (1980) The role of enilentic activitv , in hiDDOCamDa1 1r --r and ~~~
“remote” cerebral lesions induced by kainic acid. Brain Res.- 191,’ 79-97. * *
13. Ben-Ari Y., Tremblay E., Ottersen 0. P. and Naquet R. (1979) Evidence suggesting secondary epileptogenic lesions
after kainic acid: pre-treatment with diazepam reduces distant but local damage. Bruin Res. 165, 362-365.
14. Ben-An Y., Tremblay E., Riche D., Ghilini G. and Naquet R. (1981) Electrographic, clinical and pathological
alterations following systemic administration of kainic acid, bicuculline and pentetrazole: metabolic mapping using
the deoxyglucose method with special reference to the pathology of epilepsy. Neuroscience 6, 1361-1391.
15. Berger M. and Ben-Ari Y. (1983) Autoradiographic visualization of ‘H kainic acid receptor subtypes in the rat
hiDDOCamDUS. Neurosci. Leti. 39. 237-242.
16. Bdr’ger M.1 Tremblay E., Nitecka’L.and Ben-Ari Y. (1984) Maturation of kainic acid seizure-brain damage syndrome
in the rat. III. Postnatal development of kainic acid binding sites in the limbic system. Neuroscience 13, 000-000.
17. Biscoe T. J., Evans R. H., Headley P. M., Martin M. R. and Watkins J. C. (1976) Structure activity relations of
^
excitatory ammo acids on frog and rat spmal neurones. Br. J. Pharmac. 58, 373-383.
18. Biziere K. and Coyle J. T. (1979) Effects of cortical ablation on the neurotoxicity and receptors binding of kainic acid
in the striatum. J. Neurosci. Res. 4, 383-398.
19. Blennow G., Brierley J. B., Meldrum B. S., Siesjo B. K. (1978) Epileptic brain damage; the role of systemic factors
that modify cerebral energy metabolism. Bruin 101, 687-700.
20. Bouchet C. et Cazauviehl B. (1825) De l’epilepsie consideree dans ses rapports avec I’aliination mentale. Archs ghn.
Acad. 9, 510-542.
398 Y. Ben-Ari

21. Brierley J. B. (1976) Cerebral hypoxia. In Greer?fi’eld’s Neuroparhology (eds Blackwood W. and Corsellis J. A. N.),
pp. 43-86. Edward Arnold Press, London.
22. Brown W. J. (1973) Structural substrates of seizure foci in the human temporal lobe. In Epilepsy: its Phenomena in
Man (ed. Brazier M. A. B.), pp. 339-374. Academic Press. New York.
23. Brown W. J., Mitchell A. G. Jr., Babb T. L. and Crandall P. H. (1980) Structural and physiological studies in
experimentally induced epilepsy. Expl Neural. 69, 543-562.
24. Buser P., Bancaud J. and Talairach J. (1973) Depth recording in man in temporal lobe epilepsy. In Epi1ep.s.y: its
Phenomena in Man (ed. Brazier M. A. B.), pp. 67-97. Academic Press, New York.
25. Campochiaro P. and Coyle J. T. (1978) Ontogenetic development of kainate neurotoxicity: correlates with
glutamatergic innervation. Proc. n&n. Acad. Sci. U.S.A. 75, 2025-2029.
26. Cavalheiro E. A., Calderazzo Filho L., Riche D.. Feldblum S. and Le Gal La Salle G. (1983) Amygdaloid lesion
increases the toxicity of intrahippocampal kainic acid injection and reduces the late occurrence of spontaneous
recurrent seizures in rats. Brain Res. 262, 201-207.
21. Cavalheiro E., Riche D. and Le Gal La Salle G. (1983) Long-term effects of intrahippocampal kainic acid in rats:
a method for inducing spontaneous recurrent seizures. Electroenceph. clin. Neurophysiol. 53, 581-582.
28. Cepeda C., Tenaka T., Riche D. and Naquet R. (1982) Limbic status epilepticus: behavior and sleep alterations after
intra-amygdaloid kainic acid microinjections in Pupio pupio baboons. Elecfroencephul. clin. Neurophvsiol. 54,
603-613.
29. Cherubini E., de Feo M., Mecardlli 0. and Ricci G. F. (1983) Behavioral and electrographic patterns induced by
systemic administration of kainic acid in developing rats. Deal Bruin Res. 9, 69-77.
30. Chung S. H. and Johnson M. S. (1983) Divalent transition-metal ions (Ca” and Zn’+ ) in the brains of epileptogenic
and normal mice. Bruin Res. 280, 323-334.
31. Clifford D. B., Lothman E. W., Dodson W. E. and Ferendelli J. A. (1982) IXects of anticonvulsant drugs on kaimc
acid-induced epileptiform activity. ISpI Neural. 76, 156&167.
32. Collingridge G. L., Kehl R.. Cook and McLennan H. (1983) Effects of kainic acid and other amino acids on synaptic
excitation in rat hippocampal slices. 1. Extracellular analysis. Expl Brain Res. 52, 170-17X.
33. Colloque sur 1es problemes d’anatomie normale et pathologique posCs par les dtcharges kpileptiques (1965). Acru
neural. psychiat. Beg. 56.
34. Colonnier M., StCriade M. and Landry P. (1979) Selective resistance of sensory cells of the mesencephalic trigeminal
nucleus to kainic acid-induced lesions. Bruin Res. 172, 552-556.
35. Constanti A. and Nistri A. (1976) A comparative study of the effects of glutamate and kainate on the lobster muscle
fibre and frog spinal cord. Br. J. Pharmac. 57, 259-268.
36. Contestabile A. and Villani L. (1983) The use of kainic acid for tracing neuroanatomical connections in the
septohabenulo interpeduncular system of the rat. J. camp. Neurol. 214, 459-469.
36a. Coyle J. T. (1983) Neurotoxic action of kainic acid. J. Neurochem. 41, 1-l 1.
37. Coyle J. T., Bird S. J., Evans R. H.. Gulley R. L., Nicklas W. J. and Olney J. W. (1981) Excitatory amino-acid
neurotoxins: selectivity, specificity and mechanisms of action. Neurosci. Res. Prog. Bull. 19, l-427.
38. Coyle J. T., Molliver M. R. and Kuhar M. J. (1978) In situ injection of kainic acid: a new method for selectively
lesioning neuronal cell bodies while sparing axons of passage. J. camp. Neural. 180, 301~318.
39. Coyle J. T., Zaczek R., Slevin J. and Collins J. (1981) Neuronal receptor sites for kainic acid: correlations with
neurotoxicity. In Glutumutr us u Neurolrunsmirrer (eds Di Chiari G. and Gessd G. L.). pp. 337-346. Raven Press.
New York.
40. Cudennec A., Duverger D., Lloyd K. G., McCulloch J. and McKenzie E. T. (1985) In Epilep.psl.und GABA Recep!or
Aeonisrs: Ba.~ic und Therapeutic Reseurch (eds Bartholini G.. Bessi L.. Lloyd K. G. and Morselli P.). Raven Press,
Niw York. In press. ’
41. Dam A. M. (1980) Epilepsy and neuron loss in the hippocampus. Epilepsirr 21, 617-~629.
42. Daoud A. and Usherwood P. N. R. (1975) Action of kainic acid on a glutamatergic synapse. Comp. Biochem. Physiol.
52c, 51-53.
43. Della Bernardina B., Colamaria V., Capovilla G. and Bondavalli S. (1983) Nosological classification of epilepsies in
the first three years of life. In Epilepsy: un Update on Research and Therupy (eds Nistico G.. Di Perri R. and Meinard
H.), pp. 165-183. Alan Liss, New York.
44. De Montigny C., De Bonnel G., and Tardiff D. (1983) Electrophysiological studies on kainic acid induced neuronal
activation. In Excitotoxins (eds Fuxe K., Roberts P. and Schwartz R.), pp. 157-169. McMillan Press, London.
45. De Montigny C. and Lund J. P. (1980) A microiontophoretic study of the action of kainic acid and putative
neurotransmitters in the rat mesencephalic trigeminal nucleus. Neuroscience 5, 1621~ 1628.
46. Donaldson J., St Pierre T.. Minnich J. and Barbeau A. (1971) Seizures in rats associated with divalent cation inhibition
of Na+-K’ ATPase. Can. J. Biochem. 48, 1217-1224.
47. Duce T. R., Donaldson P. L. and Usherwood P. N. R. (1983) Investigation into the mechanisms of excitant amino-acid
cytotoxicity using a well-characterized glutamatergic system. Brain Res. 263, 77-87.
48. Earle K., Baldwin M. and Penfield W. (1953) Incisural sclerosis and temporal lobe seizures produced by hippocampal
herniation at birth. Archs Neural. Psychiut.. Chicago 69, 27-38.
49. Engberg I., Flatman J. A. and Lambert J. D. C. (1978) The action of N-methyl-D-aspartic and kainic acid on
motone&ones with special emphasis on conductance changes. Br. J. Pharmac. 64, 384.
50. Eneel J. Jr.. Brown W. J., Kuhl D. E., Phelps M. E., Mazziotta J. C. and Crandall P. H. (1982) Pathological findings
underlying focal temporal lobe hypometadolism in partial epilepsy. Ann. Neurol. 12, 51&51?. -
51. Engel J. Jr., Driver M. V. and Falconer M. A. (1978) Electrophysiological correlates of pathology and surgical results
in temporal lobe epilepsy. Bruin Res. 98, 129-156.
52. Escobar A. and Nieto D. (1980) Absence of lesions in Ammon’s horn neurons following metrazol induced seizures
in the hyperthermic rat. In Advances in Epileptology (eds Wada J. A. and Penry J. K.). pp. 494-501. Raven Press,
New York.
53. Evans R. H. (1980) Evidence supporting the indirect depolarization of primary afferent terminals in the frog by
excitatory amino acids. J. Physiol., Lund. 298, 25-35.
54. Falconer M. A. (1970) The pathological substrate of tempora! lobe epilepsy. Gu>,‘.s Hosp. Rep. 119, 47-60.
 Limbic seizure and brain damage produced by kainic acid 399

55. Falconer M. A. and Serafetinides E. A. (1963) A follow up study of surgery in temporal lobe epilepsy. J. Neural.
Neurosurg. Psychiat. 26, 1544165.
56. Ferkany, J. W. and Coyle J. T. (1983) Kainic acid selectively stimulates the release of endogenous excitatory acidic
amino acids, J. Pharmac. exp. Ther. 223, 399-406.
57. Ferkany J. W., Slevin J. T., Zaczek R. and Coyle J. T. (1982) Failure of folic acid derivatives to mimic the actions
of kainic acid in brain in L&O and in vitro. Neurobeh~. Toxicol. Terutol. 4, 573-580.
58. Fisher R. S. and Alger B. E. (1985) Electrophysiolo~cal action of kainic acid (KA) induced epileptiform activity in
the rat hippocampal slice. J. Neurosci. In press.
59. Franck J. E. and Schwartzkroin P. A. (1983) Kainate lesioned hippocampi become epileptogenic. Sot. Neurosci. Abslr.
9, 908.
60. Frederickson C. J. (ed.) (1985) Neurobiology of Zinc. A. R. Liss, New York. In press.
61. French E. D., Aldinio C. and Schwartz R. (1982) Intrahippocampal kainic acid, seizures and local neuronal
degeneration: relationships assessed in unanesthetized rats. Neuroscienre, 7, 2525-2536.
62. Fuxe K., Roberts P. J. and Schwartz R. (eds) (1984) The Exeifofox~~s. McMillan Press, London.
63. Garthwaite J. and Wilkins G. P. (1982) Kainic acid receptors and neurotoxicity in adult and immature rat cerebellar
slices. Neuroscience, 7, 2499-2514.
64. Gastaut H. (1980) L’ipilepsie temporale. Une nouvelle etude critique apres un quart de siecle. Concours med., Suppl.
15, l-48.
65. Gastaut H. and Fischer-Williams M. (1957) The physiopathology of epileptic seizures. In Handbook of
Physiology -Neurophysiology, pp. 329-363.
66. Gaustaut H., Gaustaut J. L., Goncalvez Silva G. E. and Fernandez Sanchez G. P. (1975) Relative frequency of
different types of epilepsy: a study employing the classification of the inte~ational ligue against epilepsy. Epjlepsiu
16,457-461.
67. Gastaut H., Vigouroux P. et Naquet R. (1952) Lesions ipileptogenes amygdalo-hippocampiques provoquees chez le
chat par injection de “crime d’alumine”. Revue Neurol. 87, 607-609.
68. Gloor P., Olivier A. and Quesnay L. F. (1981) The role of the amygdala in the expression of psychic phenomena in
temporal lobe seizures. In The Amygdaloid Complex (ed. Ben-Ari Y.), pp. 485498. Elsevier/North-Holland,
Amsterdam.
69. Goddard G. V., McIntyre D. C. and Leech C. K. (1969) A permanent change in brain function resulting from daily
electrical stimulation. Expr Neural. 25, 295-330.
70. Goldschmidt R. B. and Stewart 0. (I 982) Neurotoxic effects of colchicine: differential susceptibility of CNS neuronal
populations. Neuroscience 7, 695-714.
71. Green J. D. (1964) The hippocampus. Physiol. Rev. 44, 561-608.
72. Hall J. G., Hicks T. P. and McLennan H. (1978) Kainic acid and the glutamate receptor. Neurosci. Lett. 8,
171-175.
73. Hall J. G., Hicks T. P., McLennan H., Richardson T. L. and Wheal H. B. (lY7Y) ‘i’he excitatton ot mammaIjan central
neurons by amino acids. J. Physiol., Lond. 286, 29-39.
74. Hammond C., Feger J., Bioulac B. and Souteyrand J. P. (1979) Experimental hemiballism in the monkey produced
by unilateral kainic acid lesion in corpus luysii. Brain Res. 171, 577-580.
75. Haug F. M. S. (1967) Electron microscopilcal localization of the zinc in hippocampal mossy fibre synapses by a
modified sulphide silver procedure. Hisiochemie 8, 3SS-368.
76. Haug F. M. S. (1973) Heavy metals in the brain. A light microscopic study of the rat with Timm’s sulphide silver
method. Methodological considerations and cytological and regional staining patterns. Adv. Anat. Embryol. CeN Biol.
47, l-71.
77. Heggli D. E. and Malthe-S0renss~ D. (1982) Systemic injection of kainic acid: effect on neurotransmitter markers
in piriform cortex, amygdaloid complex and hippocampus and protection by cortical lesioning and anticonvulsants.
Neuroscience 7, 125771264.
78. Henke H. and Cuenod M. (1979) L-Glutamate specific 3H kainic acid binding in the rat neostriatum after degeneration
of the cortico-striatal pathway. Neurosci. Left. 979, 341-345.
79. Hesse G. W. (1979) Chronic zinc deficiency alters neuronal function of hippocampal mossy fibers. Science, N. Y. 205,
1005-1008.
80. Hopkins D. A. and Holstege G. (1978) Amygdaloid projections to the m~encephalon, pons and medulla oblongata
in the cat. Bxp/ Brain Res. 32, 529-547.
8 1. Ingvar M. (1982)Regional cerebral metabolism and blood flow during experimentally induced seizures. Ph.D. Thesis,
Lund.
82. Jackson J. H. (1888) On a particular variety of epilepsy (“intellectual aura”): one case with symptoms of organic
disease. Brain 11, 179.
83. Jande S. S., Maler L. and Lawson E. M. (1981) Immunohistochemical mapping of vitamin D dependent calcium
binding protein in brain. Nature 294, 765-767.
84. Jasper H. and Kershmann J. (1941) Electroencephalographic classification of the epilepsies. Archs Neurof. Psychiut.,
Chicago 45, 907-943.
85. Johnston G. A. R., Kennedy S. M. E. and Twitchin B. (1979) Action of the neurotoxin kainic acid on high affinity
uptake of L-glutamatergic acid in rat brain slices. J. Neurochem. 32, 121-127.
86. Karnushina I., Susuki R., Padgett W. and Daly J. W. (1983) Degeneration of CA1 neurons in hippocampus after
ischemia in mongolian gerbils: cyclic AMP systems. Brain Res. 268, 87-94.
87. Kelley A. E., Domesick V. B. and Nauta W. J. H. (1982) The amygdalostriatal projection in the rat-an anatomical
study by anterograde and retrograde tracing methods. Neuroscience 7, 615630.
88. KBhler C. and Schwartz R. (1983) Comparison of ibotenate and kainate neurotoxicity in rat brain: a histological
study. Neuroscience 8, 819-835.
89. Kohler C., Schwartz R. and Fuxe K. (1979) Hippocampal lesions indicate differences between excitotoxic properties
of acidic amino acids. Neurosci Left. 175, 366-371.
90. Krnjevic K., Morris M. E. and Reiffenstein R. J. (1980) Changes in extracellular Ca*+ and K+ activity accompanying
hippocampal discharges. Can. J. Physiol. Pharmac. 58, 5799583.
400 Y. Ben-Ari

91. Lancaster B., Wheal H. V. and Ashwood T. J. (1983) Hippocampal electrophysiology after kainic acid treatment: a
chronic model of focal epilepsy. Sec. Neurosci. Absfr. 9, 908.
92. Le Gal La Salle G., Paxinos G., Emson P. and Ben-Ari Y. (1978) Neurochemical mapping of GABAergic systems
in the amygdaloid complex and bed nucleus of the stria terminalis. Brain Res. 155, 397403.
93. London E. D. and Coyle J. T. (1979). Specific binding of ‘H kainic acid to receptor sites in rat brain. Molec. Pharmac.
15, 492-505.
94. Lothman E. W. and Collins R. C. (1981) Kainic acid-induced limbic motor seizures: metabolic, behavioral,
electroencephalographic and neuropathological correlates. Bruin Res. 218, 299-3 18
95. Lucas D. R. and Newhouse J. P. (1976) The toxic effects of sodium L-glutamate on the inner layers of the retina.
Archs Opthal., N.Y. 58, 193-204.
96. Mangano R. M. and Schwartz R. (1983) Chronic infusion of endogenous excitatory amino acids into rat striatum
and hippocampus. Brain Res. Bull. 10, 47-51.
97. Margerison J. H. and Corsellis Y. A. N. (1966) Epilepsy and the temporal lobes. A clinical electroencephalographic
and neuropathological study of the brain in epilepsy, with particular reference to the temporal lobes. Brain 89,
499-530.
98. Mason S. T. and Fibiger H. C. (1979) On the specificity of kainic acid. Science, N. Y. 204, 1339-1341.
99. McBean G. J. and Roberts P. J. (1984) Chronic infusion of L-glutamate causes neurotoxicity in rat striatum. Bruin
Res. 294, 372-375.
100. McDonald J. F. and Porietis A. V. (1982) Two conductances activated by excitatory amino acids. Neurosci. Absir.
12, 796.
101. McGeer E. G., McGeer P. L. and Singh K. (1978) Kainate-induced degeneration of neostriatal neurons: dependency
upon corticostriatal tract. Bruin Res. 139, 381-383.
102. McGeer E. G., Olney J. W. and McGeer P. L. (eds) (1978) Kainic Acid us u Tool in Neurobiology. Raven Press, New
York.
103. McGinty J., Henriksen S. J., Goldstein A., Terenius L. and Bloom F. E. (1983) Dynorphin is contained within
hippocampal mossy fibers, immunochemical alterations after kainic acid administration and colchicine induced
neurotoxicity. Proc. natn. Acad. Sci. U.S.A. 80, 5899593.
104. McIntyre D. C., Natharison D. and Edson N. (1982) A new model of partial status epilepticus based on kindling.
Brain Res. 250, 53-63.
105. McLennan H. (1980) The effect of decortication on excitatory amino acid sensitivity of striatal neurons. Neurosci.
Left. 18, 3133316.
106. McLennan H., Collinridge G. L. and Kehl S. J. (1984) Electrophysiological actions of kainate and other excitatory
amino acids and the structure of their receptors. In Excitotoxins (eds Fuxe K., Roberts P. and Schwartz R.), pp. 19-32.
McMillan Press, London.
107. Meibach R. C., Brown L. and Brooks F. H. (1978) Histofluorescence of kainic acid-induced striatal lesions. Brain
Res. 148, 219-223.
108. Meldrum B. S. (1983) Metabolic factors during prolonged seizures and their relation to nerve cell death. In Status
Epilepticus Mechanisms of’ Brain Damage and Treatment (eds Delgado-Escueta A. V., Wansterlain C. G.. Treiman
D. M. and Porter R. J.). Advances in Neurology, Vol. 34, pp. 261-275. Raven Press, New York.
109. Meldrum B. S. and Brierley J. B. (1973) Prolonged epileptic seizures in primates: ischemic cell changes and its relation
to ictal physiological events. Archs Neural. Psychiat., Chicago 28, 10-47.
110. Meldrum B. S. and Horton R. W. (1973) Physiology of status epilepticus in primates. Archs Neural. Psychiat., Chicago
28, 1-9.
II I. Meldrum B. S., Vigouroux R. A. and Brierley J. B. (1973) Systemic factors and epileptic brain damage. Prolonged
seizures in paralyzed artificially ventilated baboons. Archs Neural. Psychiat., Chicago 29, 82-87.
112. Minini Ch., Meldrum B. S., Riche D., Silva-Comte C. and Stutzmann J. M. (1980) Sustained limbic seizures induced
by intra-amygdaloid kainic acid in the baboon: symptomatology and neuropathological consequences. Ann. Neural.
8, 501-509.
113. Milvan A. S. (1970) Metals in enzyme catalysis. In The Enzymes (3rd Edn). Vol. II, pp. 445-536.
114. Monaghan D. T. and Cotman C. W. (1982) The distribution of ‘H kainic acid binding sites in rat CNS as determined
by a&radiography. Bruin Res. 252, 91-100.
115. Monaahan D. T.. Holets V. R.. Tov D. W. and Cotman C. W. (1983) Anatomical distribution of four
pharm\cologically ‘distinct ‘H L-glutamate binding sites. Nature 306, 176- 179.
116. Morris M. E., Krnjevic K. and Ropert N. (1983) Changes in free Ca++ recorded inside hippocampal pyramidal
neurons in response to fimbrial stimulation. Sot. Neurosci. Abstr. 9, 395.
117. Munari C.. Bancaud J., Bonis A., Buser P., Talairach J., Szikla G. et Philippe A. (1979) Role du noyau amygdalien
dans la survenue de manifestations oro-alimentaires au tours des crises tpileptiques chez l’homme. Rev. Electroenceph.
Neurophysiol. 9, 236-240.
118. Murabe Y., Ibata Y. and Sano Y. (1980) Morphological studies on neuroglia. III. Macrophage response and “neuro
gliocytosis” in kainic acid-induced lesions. Cell Tissue Res. 218, 75-86.
119. Mutaris R. (1967) Cobalt experimental hippocampal epilepsy. Epilepsia 8, 228-240.
120. Nadler J. V. and Cuthbertson G. J. (1980) Kainic acid neurotoxicity toward hippocampal formation: dependence on
specific excitatory pathways, Brain Res. 195, 47-56.
121. Nadler J. V., Evenson D. A. and Cuthbertson G. J. (1981) Comparative toxicity of kainic acid and other acidic amino
acids toward rat hippocampal neurons. Neuroscience 6, 2505-2517.
122. Nadler J. V., Evenson D. A. and Smith E. M. (1981) Evidence from lesion studies for epileptogenic and
non-epileptogenic neurotoxic interactions between kainic acid and excitatory innervation. Brain Res. 205, 405-
410.
123. Nadler J. V., Perry B. W., Gentry C. and Cotman C. W. (1980) Degeneration of hippocampal CA3 pyramidal cells
induced by intraventricular kainic acid. J. camp. Neural. 192, 333-359.
124. Nadler J. V., Shelton D. L., Perry B. W. and Cotman C. W. (1980) Regional distribution of ‘H kainic acid after
intraventricular injection. L(fe Sri. 26, 1333138.
125. Nadler J. V., Tanck D. L.. Everson D. E. and Davis J. N. (1983) Synaptic rearrangements in the kainic acid model
 Limbic seizure and brain damage produced by kainic acid 401

of Ammon’s horn sclerosis. In Excitotoxins (eds Fuxe K., Roberts P. and Schwartz R.), pp. 256-270. McMillan Press,
London.
126. Naquet R. (1953) Sur les fonctions du rhinencephale d’aprts les resultats de sa stimulation chez I’homme. These de
Doctorat d’Etat, Marseille.
127. Naquet R., Arfel G., Choux M., Dubois D. and Riche D. (1966) Etude experimentale de l’ambolie gazeuse par voie
carotidienne chez le chat. Electroencepi. clin. Neurophysiol. 20, 181-196.
128. Naquet R., Tiberin P. et Vigouroux R. (1962) Rhinencephale et epilepsie experimentale, role respectifs du siege de
la lesion cicatricielle primitive et des lesions secondaires de l’hippocampe. In Physiologic de I’Hippocumpe (ed.
Passouant P.), pp. 120-128. CNRS, Paris.
129. Nistri A. (1985) Glutamate. In Neurotrunsmitter Action in the Vertebrate Nervous System (eds Barker J. L. and
Roganski M. A.). Plenum Press, New York. In press.
130. Nistri A. and Constanti A. (1979) Pharmacological characterization of different types of GABA and glutamate
receptors in vertebrates and invertebrates. Prog. Neurobiol. 13, 117-235.
131. Nitecka L., Tremblay E., Charton G., Bouillot J. P., Berger M. L. and Ben-Ari Y. (1984) Maturation of kainic acid
seizure-brain damage syndrome in the rat. II. Histopathological sequelae. Neuroscience 13, 1073-1094.
132. Oldendorf W. H. (1970) Measurement of brain uptake of radiolabelled substances using a tritiated water internal
substance. Brain Res. 24, 372-376.
133. Olney J. W. (1969) Brain lesion, obesity and other disturbances in mice treated with monosodium glutamate. Science,
N.Y. 164, 719-721.
134 Ghiey J. W. (1978) Neurotoxicity of excitatory amino acids. In Kuinic Acid us a Tool in Neurobiology (eds McGeer
E. G., Olney J. W. and McGeer P. L.), pp. 95-121. Raven Press, New York.
135 Olney J. W. and De Gubareff T. (1978) Extreme sensitivity of olfactory cortical neurons to kainic acid toxicity. In
Kainic Acid as a Tool in Neurobiology (eds McGeer E. G., Olney J. W. and McGeer P. L.), pp. 201-217. Raven Press,
New York.
135a. Olney J. W., De Gubareff T. and Labruyere J. (1982) Seizure-related brain damage produced by cholinergic agents.
Nature 301, 520-522.
136. Olney J. W., Fuller T. and De GubarefI T. (1979) Acute dendrotoxic changes in the hippocampus of kainate treated
rats. Brain Res. 176, 91-100.
137. Olney J. W., Fuller T. A. and De Gubareff T. (1981) Folates have kainate like neurotoxicity. Nufure 292, 165-167.
138. Olney J. W., Rhee V. and Ho 0. L. (1974) Kainic acid: a powerful neurotoxic analogue of glutamate. Brain Res. 77,
507-512.
139. Pealey T. A., Zuckerman E. C. and Glaser G. H. (1969) Epileptogenic effects of localized ventricular perfusion of
ouabain on dorsal hippocampus. Expl Neurol. 25, 210-219.
140. Pei Y., Zhao D., Huang J. and Lao L. (1983) Zinc-induced seizures: a new experimental model of epilepsy. Epilepsiu
24, 169-176.
141. Penfield W. (1956) Epileptogenic lesions. Acta neural. psychiut. belg. 56, 75-88.
141a. Penry J. K. (1975) Perspectives in complex partial seizures. In Advances in Neurology (eds Penry J. K. and Daly
0. D.). Vol. 11, pp. 1-14. Raven Press, New York.
142. Pinard E., Ben-Ari Y., Tremblay E. and Seylaz J. (1983) Is intratissular hypoxia responsible for lesions during seizures
induced by kainic acid? In Cerebral Blood Flow, Metabolism and Epilepsy (ed. Baldy-Moulinier M.), pp. 252-257. J.
Libbey, London.
142a. Pinard E., Tremblay E. Ben-Ari Y. and Seylaz J. (1984) Blood flow compensates oxygen demand in the vulnerable
CA3 region of the hippocampus during kainate-induced seizures. Neuroscience 13, 1039-1049.
143. Price J. L. (1981) The efferent connections of the amygdaloid complex in the rat, cat and monkey. In The Amygdaloid
Complex (ed. Ben-Ari Y.), pp. 125-132. Elsevier/North-Holland, Amsterdam.
144. Puil R. (1981) L-Glutamate: its interactions with spinal neurons. Brain Res. 3, 229-322.
145. Ribak C. E., Bradbume R. M. and Basil-Harris A. (1982) A preferential loss of GABAergic, symmetric synapses in
epileptic foci: a quantitative structural analyses of monkey neocortex. J. Neurosci. 2, 1725-1737.
146. Ribak C. E., Vaughn J. E. and Sato K. (1978) Immunocytochemical localization of glutamic acid decarboxylase in
neuronal somata following cholchicine inhibition of axonal transport. Brain Res. 146, 315-332.
147. Roberts P. J.. McBean G. J.. Steiner M. Y.. Kohler C. and Schwartz R. (1983) Lesionine effects of ibotenate in the
immature rat’ brain and protection by 2-amino-7-phosphoheptanoic acid.\ Sot.’ Neurosci.-Abstr. 9, 261.
148. Robinson J. H. and Deadwyler S. A. (1981) Kainic acid produces depolarisation of CA3 pyramidal cells in the in
vifro hippocampal slice. Bruin Res. 221, 117-127.
149. Ruck A., Kramer S., Metz J. and Brennan M. J. W. (1980) Methyltetrahydrofolate is a potent and selective agonist
for kainic acid receptors. Nature 287, 352-803.
150. Ruth R. E. (1981) Kainic acid lesions of the hippocampus produced iontophoretically: the problem of distant damage.
Expl Neurol. 76, 508-527.
151. Sano K. and Malamud N. (1953) Clinical significance of sclerosis of the cornu Ammonis. Archs Neurol. Psychiat.,
Chicago 78, 40-53.
152. Schanne F. A. X., Kane A. B., Young E. E. and Forber J. L. (1979) Calcium dependence of toxic cell death: a final
common pathway. Science, N.Y. 206, 700-702.
153. Scheibel M. E. and Scheibel A. B. (1973) Hippocampal pathology in temporal lobe epilepsy. A golgi survey. In
Epilepsy. Its Phenomena in Man (ed. Brazier M. A. B.), pp. 311-337. Academic Press, New York.
154. Scherer-Singler V. and McGeer E. G. (1979) Distribution and persistence of kainic acid in the brain. Life Sci. 24,
1015-1022.
155. Scholz W. (1959) The contribution of pathoanatomical research to the problem of epilepsy. Epilepsia 1, 36-55.
156. Schwartz R. and Fuxe K. (1979) 3H Kainic acid binding: relevance for evaluating the neurotoxicity of kainic acid.
Life Sci. 24, 1471-1480.
157. Schwartz R. and Klihler C. (1980) Evidence against an exclusive role of glutamate in kainic acid neurotoxicity.
Neurosci. Lett. 19, 243-249.
158. Schwartz R., Scholcz D. and Coyle J. T. (1978) Structure activity relations for the neurotoxicity of kainic acid
derivatives and glutamate analogues. Neuropharmacology 17, 1455 15 1.
402 Y. Ben-Ari

159. Schwartz R., Whetsell W. D. Jr. and Mangano R. M. (1983) Quinolinic acid: an endogenous metabolite that produces
axon-sparing lesion in rat brain. Science, N.Y. 219, 316-319.
160. Schwartz I. R., Broggi G. and Pappas G. D. (1970) Fine structure of cat hippocampus during sustained seizures. Brain
Rex. 18, 166-170.
161. Schwab J. E., Fuller T.. Price J. L. and Olney J. W. (1980) Widespread patterns of neuronal damage following systemic
or intracerebral injections of kainic acid: a histological study. Neuroscience 5, 991-1014.
162. Seil F. J., Blank N. K., Woodward W. R. and Lehman A. L. (1981) Kainate effects in cerebellar cultures. Ado. Biochem.
Psychopharmac. 27, 347-354.
163. Shinozaki H. (1981) Discovery of novel actions of kainic acid and related compounds. In Kainic Acid as a Tool in
Neurobiology (eds McGeer E., Olney J. W. and McGeer P.). pp. 17-36. Raven Press. New York.
164. Shinozaki H. and Konishi S. (1980) Actions of several antihelminthics and insecticides on rat cortical neurons. Bruin
Res. 24, 368-371.
165. Siesjij B. K. and Abdul-Rahman A. (1979) A metabolic basis for the selective status epilepticus. AC/Uphysiol. sand.
106, 377-378.
166. Simon J. R., Contrera J. F. and Kuhar M. J. (1976) Binding of )H kainic acid, an analogue of L-glutamate, to brain
membranes. J. Neurochem. 26, 141-147.
167. Sloviter R. S. (1983) Epileptic brain damage in rats induced by sustained electrical stimulation of the perforant path.
I. Acute electrophysiological and light microscopic studies. Bruin Res. Bull. 10, 675-698.
168. Sloviter R. S. (1984) A selective loss of Timm staining in the hippocampal mossy fiber pathway accompanies
electrically-induced hippocampal granule cell seizure activity in rats. In Neurobiology q/Zinc (ed. Frederickson C. J.).
A. R. Liss, New York. In press.
169. Sloviter R. S. and Damiano B. P. (1981) On the relationship between kainic acid-induced epileptiform activity and
hippocampal neuronal damage. Neuropharmacology 20, 1003-101 I,
170. Soderfeldt B. (1982) Epileptic brain damage. Histopathological studies on rats with experimental status epilepticus.
Ph.D. Thesis, Lund.
171. Siiderfelt B., Kalimo H., Olsson Y. and Siesjij B. (1981) Phatogenesis of brain lesions caused by experimental epilepsy:
light- and electron microscopic changes in the rat cerebral cortex following bicuculline induced status epilepticus. Acta
neuroputh., Berl. 54, 2 19-23 1
172. Sommer W. (1880) Erkrankung des Ammonshorn als aetiologisches Moment der Epilepsie. Arch. Psychiat.
NercKrunkh. 10, 631-678.
173. Somogyi P. A., Smith A. D., Nunzi G., Gorio A., Takagi H. and Wu J. Y. (1983) Glutamate decarboxylase
immunoreactivity in the hippocampus of the cat: distribution of immunoreactive synaptic terminal with special
reference to the axon initial segment of pyramidal neurons. J. Neurosci. 373, 1450-1468.
174. Sperk G., Lassmann H., Baran H., Kish S. J., Seitelberger F. and Hornykiewicz 0. (1983) Kainic acid induced seizures:
neurochemical and histopathological changes. Neuroscience 10, 1301- 13 15.
175. Spielmeyer W. (1927) Die pathogenesis des epileptischen krampfer. Z. Neurol. Psychiut. 109, 501-520.
176. Starzl T. E.. Niemer W. T. and Dell M. (1963) Cortical and subcortical electrical activity in experimental seizures
induced by metrazol. J. Neuroputh. e-up. Neurol. 12, 262-276.
177. Stewart G. R.. Frederickson C. J., Howell G. A. and Gage F. H. (1984) Cholinergic denervation-induced decrease
of chelatable zinc in mossy fiber region of the hippocampal formation. Brain Res. 290, 43-5 1.
178. Stirling R. V. and Bliss T. V. P. (1978) Hippocampal mossy fiber development at the ultrastructural level. Prog. Brain
Res. 48, 191-198.
179. Stone W. E. and Javid M. J. (1980) Effects of anticonvulsants and glutamate antagonists on the convulsive actions
of kainic acid. Archs inr. Phurmacodyn. Thhr. 243, 56-65.
180. Storm-Matthisen J. (1977) Localization of transmitter candidates in brain: the hippocampal formation as a model.
Prog. Neurobiol. 8, 119- 129.
181. Takemoto T. (1978) Isolation and structural identification of naturally occurring excitatory amino acids. In Kuinic
Acid as a Tool in Neurobiology (eds McGeer E. G., Olney J. W. and McGeer P.), pp. I-15. Raven Press, New
York.
182. Tanaka T., Kaijima M., Daita G., Oghani J., Yonemasu S. and Riche D. (1982) Electroclinical features of kainic acid
induced status epilepticus in freely moving cats. Microinjection to the dorsal hippocampus. Electroenceph. clin.
Neurophysiol. 54, 288-300.
183 Taxt T. and Storm-Mathisen J. (1984) Uptake of u-aspartate and L-glutamate in excitatory axon terminals in
hippocampus: autoradiographic and biochemical comparison with y-aminobutyrate and other amino acids in normal
rats and in rats with lesions. Neuroscience 11, 79-100.
184. Tremblay E. and Ben-Ari Y. (1985) Usefulness of parenteral kainic acid as a model of temporal lobe epilepsy. Ret,.
Eleclroenceph. Neurophysiol. In press.
185. Tremblav E.. Berzer M.. Nitecka L.. Cavalheiro E. and Ben-Ari Y. (1984) A multidisciplinary study of folic acid
neurotoxicity: in&actions with kainate binding sites and relevance to the aetiology of epilepsy. Neuroscience 12,
5699589.
186. Tremblay E., Nitecka L., Berger M. L. and Ben-Ari Y. (1984) Maturation of kainic acid seizure-brain damage
syndrome in the rat. I. Clinical, electrographic and metabolic observations. Neuroscience 13, 1051-1072.
187. Tremblay E., Ottersen 0. P., Rovira C. and Ben-Ari Y. (1983) Intra-amygdaloid injections of kainic acid: regional
metabolic changes and their relation to the pathological alterations, Neuroscience 8, 299-315.
188. Unnerstall J. R. and Wamsley J. K. (1983) Autoradiographic localisation of high-affinity ‘H kainic acid binding sites
in the rat forebrain. Eur. J. Pharmuc. 86, 361-371.
189. Valentino R. and Dingledine R. (198 1) Presynaptic, inhibitory effect of acetylcholine in the hippocampus. J. Neurosci.
1, 784792.
190. Van Harreveld A. and Fifkova E. (1971) Light and electron microscopic changes in central nervous tissue after
electrophoretic injection of glutamate. E.xpl Molec. Path. 15, 61-81.
191. Vilani L.. Migani P., Poli A., Niso R. and Contestabile A. (1982) Neurotoxic effect of kainic acid on ultrastructure
and GABAergic parameters in the goldfish cerebellum. Neuroscience 7, 25 15-2524.
192. Vincent S. R. and McGeer E. (1979) Kainic acid binding to membrane of striatal neurons. Life Sci. 24, 265-270.
 Limbic seizure and brain damage produced by kainic acid 403

193. Von Euler C. (1962) On the significance of the high zinc content in the hippocampal formation. In Physiologic de
I’Hippocampe (ed. Passouant P.). CNRS, Paris.
194. Watkins J. C. and Evans R. H. (1981) Excitatory amino acid transmitters. A. Reo. Pharmac. Toxicol. 21, 165-204.
195. Whetsell W. 0. W. Jr. and Schwartz R. (1983) Mechanisms of excitotoxins examined in organotypic cultures of rat
central nervous system. In Exeitotoxins (eds Fuxe K., Roberts P. and Schwartz R.), pp. 207-218. McMillan Press,
London.
196. Wieser H. G. (1980) Temporal lobe or psychomotor status epiiepticus. A case report. E~eetroe~ceph. cl&. ~earap~ys~a~.
48, 558-572.
197. Wu J. Y. and Roberts E. (1974) Properties of brain L-glutamate decarboxylase: inhibition studies. J. Neurochem. 23,
759-767.
198. Wuerthele S. M., Love11K. L., Jones M. Z. and Moore K. E. (1978) A histological study of kainic acid-induced lesions
in the rat brain. Brain Res. 149, 489-497.
199. Zaczek R. and Coyle J. T. (1982) Excitatory amino acid analogues: neurotoxicity and seizures. ~europharmuco~ogy
21, 15-26.
200. Zaczek R., Nelson M. F. and Coyle J. T. (1978) Effects of anaesthetics and anticonvulsants on the action of kainic
acid in the rat hippocampus. Eur. J. Pharmac. 52, 323-327.

(Accepted 19 Jury 1984)

Nare added in proof. In a recent study (L. Nitecka, D. Riche, G. Ghilini and Y. Ben-Ari, in preparation) using an antibody
directed against GABA, we have observed a rapid destruction of GABA-containing neurons in the instar zone after
parenteral administration of KA.