-
R E V I E W
580 TINS Vol. 23, No. 11, 2000 0166-2236/00/$ see front matter
2000 Elsevier Science Ltd. All rights reserved. PII:
S0166-2236(00)01659-3
ANIMAL MODELS of epilepsy do not provide all thecomplex
etiologies and variety of syndromes thathave been identified in
humans. Nevertheless, becauseof similar basic features,
experimental models haveallowed the determination of the basic
molecular andcellular mechanisms of epileptogenesis and its
relationto brain damage. This is particularly exemplified in
thefield of temporal lobe epilepsy (TLE). Indeed, studiesusing the
kainate model of epilepsy have considerablyimproved our
understanding of important issues inthe field of the epilepsies:
are seizures the cause or theconsequence of brain damage? Does
hyperactivity perse lead to cell loss? And, if so, how? Where are
seizuresgenerated and why? Since the publication of an
earliercomprehensive review1, the combined use of molecu-lar
biology, patch clamp and imaging techniques aswell as novel in
vitro preparations have significantlyimproved our understanding of
the mechanisms ofaction of kainate. Here we compare recent and
earliermodels proposed to explain how kainate generatesseizures in
the hippocampus. In particular, this reviewconcentrates on the
effects of kainate on the multiplefacets of GABA-mediated
inhibition and their contri-bution to epileptogenesis. We propose a
model inwhich kainate excites all its targets in the hippo-campus,
i.e. pyramidal neurons and interneurons viadifferent kainate
receptor subtypes. The excitation ofinterneurons leads to a massive
increase of tonicGABA-mediated inhibition in principal cells.
This,however, fails to prevent epileptiform activitiesbecause of
the strong excitation of CA3 pyramidalneurons, first by the
selective activation of kainatereceptors at mossy fiber synapses
and second by gluta-matergic recurrent collateral synapses that
have a lowthreshold for the generation of synchronized activ-ities.
Thus, the generation of seizures by kainate is notcaused by a
collapse of inhibition, a conclusion thathas been reinforced by
observations made in chronicmodels of epilepsy in which there is no
general failureof action-potential driven inhibition.
The kainate experimental animal model of TLE
Studies performed two decades ago have shown thatsystemic or
intracerebral injections of kainate cause
epileptiform seizures in the CA3 region of the hippo-campus.
These seizures propagate to other limbic struc-tures and are
followed by a pattern of cell loss that issimilar to that seen in
patients suffering from TLE1,2.
CA3 pyramidal neurons are indeed amongst themost responsive
neurons to kainate in the brain,because they readily degenerate
following local or dis-tal injections of kainate. However, studies
using thekainate model have also shown that CA3 pyramidalneurons
are highly vulnerable to network hyperactiv-ity per se and readily
degenerate following recurrentseizures probably because of a
sustained release of glu-tamate leading to an activation of kainate
receptors.Thus, injections of kainate in structures that are
distalfrom the hippocampus, at concentrations that do notdiffuse to
the hippocampus, are sufficient to generatea seizure and brain
damage syndrome that includesCA3 damage1. In addition, the neuronal
damage inCA3 following distal injections is prevented by block-ade
of seizures using diazepam injections suggestingthat this damage is
caused by the repetitive activationof afferent pathways during
seizures. This is furthersupported by the direct relationship
between epilep-tiform activities and the extent of damage in CA3
andby the fact that lesion of hippocampal afferent path-ways
abolishes most of the damage in the hippo-campus. The conclusion
that damage in that region isselectively caused by seizure per se
is confirmed by thedirect measure of local blood flow and oxygen
con-sumption in situ, which shows that there is no imbal-ance
between oxygen supply and neuronal activity(reviewed in Ref. 1).
Furthermore, repetitive high-fre-quency stimulation of CA3
pyramidal neuronsinduces a selective cell loss in this region also
suggest-ing that excessive activation of synaptic inputs to
CA3pyramidal neurons is toxic3. Therefore, CA3 pyramidalneurons are
susceptible to repetitive synchronizedactivities that leads to cell
loss probably as a result ofthe sustained release of glutamate.
The CA3 region is also the hippocampal pacemakerfor the
generation of synchronized activities. This islargely the
consequence of the dense network of recur-rent collateral
glutamatergic axons (associated to AMPA receptor-mediated synapses)
that interconnect
Kainate, a double agent that generatesseizures: two decades of
progressYehezkel Ben-Ari and Rosa Cossart
Studies using kainate, an excitatory amino acid extracted from a
seaweed, have provided majorcontributions to the understanding of
epileptogenesis.Here we review pioneering and more recentstudies
aimed at determining how kainate generates seizures and, in
particular, how inhibition isaltered during seizures.We focus on
target and subunit-specific effects of kainate on
hippocampalpyramidal neurons and interneurons that lead to an
excitation of both types of neurons and thusto the parallel
increase of glutamatergic and GABAergic spontaneous currents.We
propose thatkainate excites all its targets,the net consequence
depending on the level of activity of the network.Trends Neurosci.
(2000) 23, 580587
Yehezkel Ben-Ariand Rosa Cossart
are at the INMED,INSERM U29, Parc
scientifique deLuminy, BP 13,
13273 Marseille,France.
-
TINS Vol. 23, No. 11, 2000 581
pyramidal neurons. More recently, extensive physiologi-cal and
modeling studies have shown that activation ofeven a small
percentage of recurrent excitatory collateralsynapses that
interconnect pyramidal neurons is suffi-cient to generate
synchronized activities4. Because ofthis feature, which is unique
in the hippocampus, awide range of convulsive agents with different
modes ofaction, such as bicuculline, kainate, carbachol or
4-AP,generate seizures in CA3 but not in the isolated
CA1.Therefore, the CA3 region is the pacemaker for the gen-eration
of synchronized activities that subsequentlypropagate to CA1 and
other brain regions.
Several observations suggest that the epileptogeniceffects of
kainate in CA3 are largely caused by the acti-vation of
high-affinity kainate receptors that are pref-erentially expressed
in the mossy fiber synaptic region.This was first suggested by the
earlier observation thatin both humans and various animal species,
the mossyfiber synaptic region (stratum lucidum) is enrichedwith
high-affinity kainate receptors (Kd within therange of 520 nM)5,6
that can be activated even by thesmall concentrations of kainate
crossing thebloodbrain barrier during seizures induced by sys-temic
injections of the toxin7. Furthermore, selectivelesion by neonatal
irradiation of the granule cells andtheir mossy fibers eliminates
the epileptogenic effectsof kainate but not that of high K1
concentrations8
(Fig. 1b). Parallel developmental studies show that theneuronal
damage induced by kainate is only observedonce granule cells and
mossy fiber synapses are opera-tional, at approximately the third
postnatal week1. Inaddition, in vivo9 and slice recordings1012
indicate thatlow concentrations of kainate (50250 nM) that
selec-tively activate kainate and not AMPA receptors13 gen-erate
seizures in CA3 pyramidal neurons that propa-gate to CA1 and to
other limbic structures10,12. Evenhigh concentrations of kainate
(greater than micro-molar) do not generate seizures in the
disconnectedCA1 area suggesting that the epileptiform
activitiesobserved in CA1 following local infusions of
kainate14
are in fact generated in CA3 following diffusion of thetoxin.
This has also been directly implemented in theintact hippocampus
preparation in vitro15.
These studies suggest that local or systemic injec-tions of
kainate first activate CA3 pyramidal neuronsvia high-affinity
receptors present in mossy fibersynapses. The activation of CA3
recurrent collateralsynapses generates synchronized network-driven
glutamatergic currents that propagate to other hippo-campal
regions. It is important to stress that the cru-cial role of the
mossy fiber synapses is also confirmedby the observations that
episodes of status epilepticus,such as those generated by kainate,
also lead to theformation of novel aberrant mossy fiber synapses
onCA3 pyramidal cells and on granule cell neurons1618,to an
increased density of kainate receptors and to areduction of seizure
threshold (Fig. 1a). Therefore,seizures beget seizures because
although seizuresinduce damage through neuronal hyperactivity,
theneuronal hyperactivity will facilitate seizure genera-tion via
the formation of novel mossy fiber synapses.
However, to unravel the precise mechanisms of theaction of
kainate, it is essential to analyze its action onionotropic
receptor-mediated currents and on voltage-gated currents. Studies
performed primarily in CA1pyramidal neurons have reported a
plethora of effectsof kainate including: (1) a reduction of the
amplitude
of evoked GABAergic IPSCs and of the frequency ofminiature IPSCs
(but see below) effects that arethought to facilitate seizure
generation; (2) an increaseof tonic inhibition that might have
opposite conse-quences; (3) a blockade of two currents that are
impor-tant in regulating cell excitability: IAHP and IH (Ref.
19)and (4) a reduction of voltage-gated Ca21 currents20 andof
glutamate release21 (but see below). However, the rel-evance of
these effects to the epileptogenic action ofkainate is not clear as
a wide range of concentrations ofkainate were used (50 nM to 27 mM)
and the effects arenot consistently directly associated to kainate
receptor-mediated synaptic currents. The following discussionfirst
concentrates on the effects induced by low con-centrations
(submicromolar) of kainate that generate
Y. Ben-Ari and R. Cossart Kainate and seizures RE V I E W
(a)
(b)
(c)
trends in Neurosciences
200 m V120 s
200 m V120 s
KA (300 nM) KA (500 nM)
Fig. 1. Seizures are generated in CA3 by activation of kainate
receptors located at mossyfiber synapses. Seizures induce
collateral sprouting of mossy fibers in the CA3 region45,(a) shows
photomicrographs depicting the sprouting of mossy fibers
(Timm-stained) in the CA3region of a control (left) versus an
epileptic rat (right). Note the aberrant infrapyramidal bandof
mossy fibres (arrows). (b) Photomicrographs of Timm-stained mossy
fibers from control (left)and irradiated (right) hippocampi of an
adult rat (P40). The traces below show the field-poten-tial
recordings. In the non-irradiated CA3 region (left trace), bath
application of kainate (KA)(300 nM) induced spontaneous and evoked
epileptiform discharges. In the right trace from theirradiated CA3
region, KA (500 nM) decreased the amplitude of the field potentials
and did notinduce bursts. Neonatal irradiation reduces the density
of Timm-stained mossy fibers and pre-vents the epileptic action of
kainate8. (c) Photomicrographs from receptor autoradiographyusing
3[H]kainate on coronal sections from wild-type (left) and GluR6
mutant mice (right).GluR6 deficient mice are less susceptible to
systemic administration of kainate (20 mg kg1)than control mice as
revealed by the number of mice showing seizures25, 6 out of 6 for
controland 1 out of 17 for GluR6 knockout mice. (a) Adapted, with
permission, from Ref. 45, (b)adapted, with permission, from Ref. 8
and (c) adapted, with permission, from Ref. 25.
-
582 TINS Vol. 23, No. 11, 2000
paroxysmal discharges in the hippocampus. Theseeffects are
associated to synaptic currents and are, atleast in part, target
and subunit selective.
Target and subunit-selective effects of kainateassociated to
synaptic currents
Activation of GluR6-containing kainate receptors at mossy
fibersynapses located on CA3 pyramidal neurons
Repetitive electrical stimulation of the mossy fiberpathway
generates slow EPSCs in CA3 pyramidal neur-ons that are mediated by
kainate receptors and notAMPA receptors because they are resistant
to the selec-tive AMPA receptor antagonist GYKI53655 (Refs
22,23).
The mossy fiber synapses are located close to the somaof
pyramidal neurons and should therefore generateEPSCs that will
efficiently propagate to the cell bodyand its intracellular
machinery. These EPSCs are notgenerated in CA1 pyramidal neurons,
confirming thespecific involvement of high-affinity kainate
receptorsfor synaptic transmission in CA3 pyramidal cells24.
Cloning of several kainate receptor subunits andgeneration of
knockout mice have made it possible toidentify the subunit involved
in kainate receptor-medi-ated synaptic currents. Thus, granule
cells and CA3pyramidal cells are enriched with
GluR6-containingkainate receptors and the synaptic currents
generatedby the stimulation of mossy fibers, after blockade ofAMPA
receptors, are eliminated in GluR6 knockouts25
(Fig. 1c). In GluR6 knockouts higher concentrations ofkainate
are also required to generate seizures. Theseobservations provide
direct evidence that GluR6 sub-units mediate the epileptogenic
actions of kainate inCA3. Collingridge and colleagues suggested
that mossyfiber synapses also include pre- and postsynaptic
GluR5containing kainate receptors26. However, the selectiveGluR5
ag-onist ATPA does not generate a postsynapticcurrent in CA3
pyramidal cells26 and the mRNAsencoding GluR5 are weakly expressed
in granule cellsor CA3 pyramidal cells27. In addition, these data
arebased on the inhibition by a GluR5 antagonist of EPSCsthat are
considered to be mediated by mossy fibersynapses. However,
stimulating the fascia dentata notonly activates the mossy fibers
but also activates theextensively arborized glutamatergic recurrent
collater-als of the CA3 pyramidal cell axons. Because mossyfiber
EPSCs are larger than recurrent collateral ones28, itwill be
interesting to determine the effects of GluR5antagonists on
identified mossy fiber-mediated EPSCs.
These observations and the dramatic effects of gran-ule cell and
mossy fiber lesion by neonatal irradiation(see above) suggest that
the epileptogenic actions ofkainate are mediated at least in part
by GluR6-contain-ing kainate receptors present on mossy fiber
synapses.The secondary activation of the CA1 region, the
majoroutput gate from the hippocampus, leads to the propa-gation of
seizures to other limbic structures, notably tothe entorhinal
cortex and other cortical structures andthus to the generation of a
limbic partial type ofseizure. Therefore, postsynaptic kainate
receptors con-taining GluR6 subunits and located on CA3 mossyfiber
synapses are key players in the generation ofseizures by kainate.
In spite of this, evidence also existsfor the presence of
presynaptic kainate receptors inmossy fiber synapses, but their
role in the epilepto-genic effects of kainate is presently unclear
(see below).Activation of GluR5 containing receptors located
oninterneurons
Two recent studies have shown that applications oflow
concentrations (submicromolar) of kainate in thepresence of
selective NMDA and AMPA receptorsantagonists produce a massive
long-lasting depolar-ization of CA1 interneurons and a powerful and
sus-tained barrage of action potentials13,24. As expected,the
consequence of this strong excitation of interneu-rons is an
increase of the spontaneous inhibitionrecorded in CA1 pyramidal
neurons (Fig. 2); indeed,an eightfold increase of the frequency of
tonic IPSCswas attained using 250 nM kainate. This effect is
selec-tive for interneurons because similar or higher
con-centrations of kainate do not significantly depolarize
Y. Ben-Ari and R. Cossart Kainate and seizuresRE V I E W
trends in Neurosciences
ATPA 1 m M
ATPA 1 m M
KA 250 nM KA 250 nM
100 s50 pA100 ms5 mV
1 min100 ms10 mV
10 mV100 ms
GluR5GABA-R
Interneuron Pyramidal cell
(a)
(b)
(d)
(c)
Fig. 2. Activation of GluR5-containing kainate receptors located
on CA1 interneuronsincreases tonic GABA-mediated inhibition on
pyramidal cells13. (a) Diagram showing CA1interneuron as the blue
cell and the CA1 pyramidal neuron as the red cell. (b) and (c)
Lefttraces shown kainate (KA) depolarization in CA1. Current clamp
recordings (I holding: 0 pA)were measured in the presence of
GYKI53655 (30 mM) and D-APV ( 50 mM). KA (250 nM) orATPA (1 mM),
the selective agonist for GluR5-containing kainate receptors,
caused a reversibledepolarization of the membrane potential and
repetitive action potential firing. Right tracesshow that kainate
increases tonic inhibition in CA1 pyramidal cells. Voltage-clamp
recordingsof spontaneous IPSCs at the reversal potential for
glutamatergic currents (Vhold: 110 mV). Inthe presence of GYKI53655
(30 mM) and D-APV (50 mM), KA (250 nM) or ATPA (1 mM)reversibly
increases the spontaneous IPSCs frequency. (d) Left trace shows
that the kainate-receptor-mediated EPSP evoked after stimulation
(shown by arrow) in stratum radiatum in thepresence of GYKI53655
(30 mM), D-APV (50 mM) and bicuculline (20 mM) triggers a burst
ofaction-potentials. Right trace shows that the EPSP evoked in CA1
pyramidal cells by stimula-tion (arrow) in s. radiatum is
completely blocked by GYKI53655 (30 mM) and D-APV (50 mM).Adapted,
with permission, from Ref. 13.
-
TINS Vol. 23, No. 11, 2000 583
CA1 pyramidal neurons. Morphological identificationof the
recorded interneurons indicated that a widerange of GABAergic
neurons in strata oriens, radiatumor lacunosum moleculare that
project to the cell bodyor to the apical dendrites of pyramidal
neurons areactivated by kainate (Fig. 3). This suggests that this
isa widespread property of various interneurons types.Preliminary
observations suggest that CA3 interneur-ons are also depolarized
massively by kainate. Mostimportantly, kainate receptor-mediated
synaptic cur-rents could be generated by electrical stimulation
ofstratum radiatum in CA1 inter-neurons but not inpyramidal
neurons. Kainate receptor-mediated EPSCshave slower kinetics than
AMPA receptor-mediatedEPSCs in the same neurons (rise time, 6 ms
versus 2ms; decay time, 30 ms versus 13 ms, respectively),which
appears to be a general property of kainatereceptor-mediated
synaptic currents. The mechanismsunderlying this difference are
unknown, but it is pos-sible that it results from a distal
distribution of kainatereceptors at the edge of the postsynaptic
density.Interestingly, at low concentrations, kainate did notalter
other parameters of inhibition in pyramidalneurons, including
miniature and evoked inhibition(see below). Therefore, there is a
network of postsy-naptic kainate receptors present in interneurons
butnot in pyramidal cells. Activation of this network pro-duces a
paradoxical overinhibition of the target neur-ons that might reduce
seizure generation.
Cossart et al.13 also reported that the effects ofkainate are
mediated in part by GluR5-containingreceptors because they could be
mimicked by theselective GluR5 subunit agonist ATPA and blocked
bythe relatively selective antagonist, LY293558. Mostinterneurons
in stratum oriens responded to ATPA(Fig. 3). By contrast, only 20%
of stratum radiatuminterneurons were affected by the GluR5
agonist,although most of them were depolarized using kainate(250
nM). Therefore, kainate excites most interneurontypes, some via
GluR5 subunit-containing receptors,others via different receptor
subtypes. Indeed, a recentstudy suggested that kainate was still
able to depolar-ize stratum radiatum interneurons in mice that
lackedthe GluR5 subunit29. Therefore, it is probable that
inaddition to the GluR5-mediated network of interneu-rons, other
interneurons overinhibit principal neur-ons via activation of
different conformations ofkainate receptors. Nevertheless,
activation of GluR5subunit-containing receptors appears to be
restrictedto interneurons in CA1 (and also in other hippo-campal
regions) and thus might act to reduce thepropagation of seizures
and the generation of syn-chronized activities. The heterogeneity
of interneurontypes results in a wide-range of selective
modulationby kainate of GABA-mediated inhibition and of net-work
excitability. Future studies are required to deter-mine whether the
selective activation of the inhibitorynetwork is sufficient to
prevent epileptogenesis.
Other effects of kainate
Presynaptic effects of kainate on the release of GABAThe
observation that kainate enhances spontaneous
GABAA-mediated inhibition was unexpected becauseintuitively
epileptogenesis should be associated with areduction of
GABA-mediated inhibition. In fact, a col-lapse of inhibition has
been repeatedly suggested tounderlie the epileptogenic effects of
kainate11,14,30. Two
parameters have been examined in detail: evokedGABA-mediated
IPSCs and miniature TTX-insensitiveIPSCs. However, in contrast to
the clear cut effects ofkainate on tonic inhibition, these effects
are contro-versial and require high concentrations of agonist(Box
1) even if they can be mimicked by glutamatereleased during
repetitive high-frequency stimula-tion31. The decrease of evoked
GABA-mediated inhibi-tion has been proposed to be mediated by a
presynap-tic subtype of kainate receptor involving ametabotropic
function32. This interpretation hasrecently been challenged in a
study showing that theeffects of kainate on evoked IPSCs could be
explainedby indirect mechanisms resulting from the highamount of
GABA released by interneurons duringkainate application33 [(Box 1)
but see also Ref. 32].
Another important point to stress is that there areseveral
parameters to consider in order to evaluatethe strength of
inhibition and the IPSC evoked byelectrical stimulation is not
necessarily the most rel-evant. Also, the hypothesis that
epileptogenesis isassociated to a general fall of the
GABA-mediatedinhibitory drive has not been confirmed in acuteand
chronic models of epilepsy (Box 2). Indeed, thenotion of
kainate-induced decrease of inhibitiondepends upon the parameter
used to assess the levelof inhibition. For example, in the most
recent stud-ies examining the fate of GABA-mediated inhibitionin
the kainate model of TLE, the modifications ofinhibition are
specific for each inhibitory pathwayand locus-dependent within each
specific pathway.Thus, measuring only one parameter might pointto a
deficit of inhibition even though, globally, inhibition is
enhanced. This stresses the necessity tocheck all the parameters
characterizing inhibitionand not one in particular (Box
2).Presynaptic effects of kainate on the release of glutamate
Early morphological and biochemical studies sug-gest that
kainate receptors are located presynapticallyon mossy fiber
terminals8,17,34. Thus, the selective
Y. Ben-Ari and R. Cossart Kainate and seizures RE V I E W
trends in Neurosciences
SO
SP
Bistratified interneuron
SO
SPSR
SR
SLM
SOSP
SLM
SR
Perforant path-associatedinterneuron
(a) O-LM interneuron (b) Basket cell
Fig. 3. Different types of interneurons tested for their
sensitivity to kainate . Reconstructionsusing the Neurolucida
system of four types of interneurons depolarized by kainate (cell
bodyand dendrites are shown in red and axons are shown in black).
(a) 100% of stratum oriensinterneurons are depolarized by kainate
(250 mM) or by ATPA (the GluR5 agonist). (b) 85%of stratum radiatum
interneurons are depolarized by kainate (250 nM) and 20% of
stratumradiatum interneurons are depolarized by APTA (Ref. 13).
Abbreviations: SO, stratum oriens;SP, stratum pyramidale; SR,
stratum radiatum; SLM, stratum lacunesum moleculare.
-
584 TINS Vol. 23, No. 11, 2000
lesion of dentate gyrus neurons and of mossy fibersmarkedly
reduced the high-affinity binding sites instratum lucidum (Fig. 1)
and immunolabeling ofGluR6/7 subunits was observed in
unmyelinatedaxons of the CA3 region34. Furthermore, it wasrecently
shown that low concentrations of kainateaugment the presynaptic
afferent volley recorded inCA3 following stimulation of mossy
fibers, an effectsuggested to result from depolarization of mossy
fiberaxons35. By contrast, kainate application does notaffect the
frequency of miniature EPSCs recorded in
CA3 pyramidal neurons (see below and Ref. 24) evenif a selective
modulation by kainate of mossy fiberminiature EPSCs, which are
large amplitude eventsoccurring at a low frequency28, cannot be
excluded.
Several observations suggest that different concen-trations of
kainate can have opposite effects. Thus, inthe elegant study of
Kamiya and Ozawa35, low concen-trations of kainate (200 nM) that
generate seizures inCA3 (Ref. 36), increase the mossy fiber
afferent volleybut reduce the mossy fiber EPSP. Higher
concentra-tions of kainate (3 mM) reduce both the afferent
volley
Y. Ben-Ari and R. Cossart Kainate and seizuresRE V I E W
Kainate has repeatedly been shown to depress GABA-mediated
inhi-bition in CA1 pyramidal cells, but both the mechanisms and
thephysiological relevance of this effect are unclear.
(1) The disinhibitory action of kainate hypothesis relies
mainlyon the decrease of the evoked IPSC amplitude by kainate.
However,evoked responses cannot conclusively distinguish either a
mecha-nism (pre- or postsynaptic, or direct or indirect) or a
general level ofinhibition (see Box 2). First, a decrease of the
amplitude of theevoked IPSCs can occur as a result of a pure
postsynaptic phenom-enon [i.e. a change in postsynaptic membrane
resistance (seeFig. I)], or to an indirect presynaptic phenomena
such as exhaustionof the terminal, a change in the probability of
transmission failurein the axon or activation of presynaptic GABAB
receptors [see points(3) and (4) and Fig. I]. Thus, the strongest
evidence for a presynap-tic population of kainate receptors on
GABAergic terminals shouldbe based on the study of miniature IPSCs
or on the study of theeffects of kainate on evoked IPSCs obtained
with paired recordingsfrom connected neurons. The latter experiment
has not been per-formed and the former has generated contradictory
resultsac.Furthermore, the decrease in the evoked IPSC amplitude
observedonly with high concentrations of kainate does not imply
that thereis a reduction of GABA quantal release. Marty et al.d
have shown
that a reduction of evoked IPSC can be associated with an
increaseof the miniature IPSCs frequency.
(2) Reduction of evoked IPSCs by kainate requires
micromolarconcentrations (Fig. I). Thus, in the study of
Rodriguez-Moreno etal.c, the depressant effect of kainate on evoked
IPSCs was shown tofollow a bell-shaped curve, with optimal
concentrations of1030 mM. This bell-shaped curve effect was
suggested to occur as aresult of the fact that low and high
concentrations of kainate hin-der steady-state receptor activity:
low concentrations are unable toactivate receptors and high
concentrations rapidly desensitizes thereceptors. However, in the
same study, kainate was bath applied for30 min, which is long
enough for the receptors to be largely desen-sitized. In the study
of Cossart et al.a, the reduction of evoked inhi-bition by kainate
is also observed at high (greater than micromolar)but not at low
(submicromolar) concentrations (Fig. I). By contrast,the target and
subunit-specific effects of kainate have been reportedat low
concentrations.
(3) Nicoll and co-workerse suggested that the reduction of
theevoked IPSCs by kainate could be explained by indirect
mechanisms:kainate increased firing in interneurons leading to an
enhanced releaseof GABA. The resulting massive activation of
postsynaptic GABAAreceptors augments passive shunting of the
postsynaptic membrane,and the increase of GABA release activates
presynaptic GABAB receptorsthat in turn depress GABA releasee. The
involvement of GABAB recep-tors in the depressant action of kainate
could account for themetabotropic action of kainate proposed by
Lerma and co-workersf,g.
(4) An additional problem is caused by the fact that in
mostexperiments the connections between CA3 and CA1 were not
sur-gically removed, thus enabling the propagation of seizures
fromCA3 to CA1. For example, in the pioneering study of Alger
andFisherh, the reduction of the evoked IPSP occurs concomitantly
withpropagated epileptiform activities.
(5) Finally, we have recently shown that activation of
presynap-tic kainate receptors, selectively located at inhibitory
synapses onCA1 interneurons, does not decrease but instead
increases GABAquantal release (R. Cossart et al., unpublished
observations).
Referencesa Cossart, R. et al. (1998) GluR5 kainate receptor
activation in interneurons
increases tonic inhibition of pyramidal cells. Nat. Neurosci. 1,
470478b Frerking, M. et al. (1998) Synaptic activation of kainate
receptors on
hippocampal interneurons. Nat. Neurosci. 1, 479486c
Rodriguez-Moreno, A. et al. (1997) Kainate receptors
presynaptically
downregulate GABAergic inhibition in the rat hippocampus. Neuron
19,893901
d Glitsch, M. and Marty, A. (1999) Presynaptic effects of NMDA
in cerebellarPurkinje cells and interneurons. J. Neurosci. 19,
511519
e Frerking, M. et al. (1999) Mechanisms underlying kainate
receptor-mediated disinhibition in the hippocampus. Proc. Natl.
Acad. Sci. U. S. A.96, 1291712922
f Rodriguez-Moreno, A. and Lerma, J. (1998) Kainate receptor
modulationof GABA release involves a metabotropic function. Neuron
20, 12111218
g Rodriguez-Moreno, A. et al. (2000) Two populations of kainate
receptorswith separate signaling mechanisms in hippocampal
interneurons. Proc.Natl. Acad. Sci. U. S. A. 97, 12931298
h Fisher, R.S. and Alger, B.E. (1984) Electrophysiological
mechanisms ofkainic acid-induced epileptiform activity in the rat
hippocampal slice.J. Neurosci. 4, 13121323
Box I. Does kainate presynaptically reduce GABA-mediated
inhibition?
trends in Neurosciences
00.20.40.60.8
0
1.0
Cum
ula
tive p
roba
bility
Inter-event interval (s)0 0.5 1 1.5 2
00.20.40.60.8
0
1.0
TTXKA 250 nMWASH
TTXKA 10 m MWASH
Cum
ula
tive p
roba
bility
Inter-event interval (s)0 0.5 1 1.5 2
KA 250 nM
GYKI53655D-APV
KA 10 m M
40 ms50 pA
GYKI53655D-APV
(a)
(b)
Fig. I. Does kainate presynaptically reduce GABA-mediated
inhibition?Kainate-induced disinhibition of CA1 pyramidal neurons
requires high con-centrations of agonist13. (a) Kainate (KA) (250
nM) has no effect on the ampli-tude of the evoked IPSC in CA1
pyramidal cells or on the frequency of miniatureIPSCs (shown in
cumulative probability plot, right) recorded in the presence
ofGYKI53655 (30 mM) and D-APV (50 mM) (Vhold: 110 mV). (b)
Increasing theconcentration of kainate to 10 mM depresses the
evoked IPSC amplitude andreduces the frequency of miniature IPSCs
from 60% of the CA1 pyramidal cellsrecorded. n= x cells/y.
-
TINS Vol. 23, No. 11, 2000 585
Y. Ben-Ari and R. Cossart Kainate and seizures RE V I E W
Box 2.The fate of inhibition in temporal lobe epilepsy
TLE Interneuron
Control
(b)(i) (ii)
(iii)
Evokedepileptiform discharge
SO
SPSR
5 0 5 10 15 20 25 30
controlepileptic
20 pA
10 pA1 s
10 pA20 ms
20 ms
Spontaneousepileptiform discharge
AP frequency (Hz)
(a)
30 mV
1 min
1 2
2 s30 mV
1 2
KA (250 nM)
trends in Neurosciences
Fig. I. The fate of inhibition in temporal lobe epilepsy. (a)
Interneurons are alsorecruited during kainate-induced epileptiform
discharges 17. Bath application of kainate(KA) (250 nM) in the
neonatal intact hippocampal formation in vitro induces ictal
activ-ity in a CA3 stratum oriens interneuron shown by current
clamp recording. Differentphases of the ictal episode are marked by
arrows (1,2) and shown on the traces belowon an expanded time
scale. 1-interictal phase, 2-tonic oscillations. (b) The fate
ofinterneurons in animal models of temporal lobe epilepsy (TLE).
(i) Neurolucida recon-struction of a biocytin-labeled interneuron
from a KA-treated rat. (ii) Voltage clamprecordings (cell attached
configuration) of evoked and spontaneous epileptiform dis-charges
in interneurons from slices of KA-treated ratsd showing that
interneurons arehyperexcitable in TLE. (iii) Distribution of
spontaneous firing frequencies from controland epileptic
interneurons recorded in the cell attached mode showing that
interneu-rons are hyperactive in TLE. Traces show typical cell
attached recordings from controland TLE interneurons (right). Part
(i), adapted, with permission from Ref. d.
GABAA receptor-mediated inhibition is a concept that
encompassesa constellation of variables, including the tonic
activity ofGABAergic interneurons, the properties of the
presynapticGABAergic terminals impinging upon their target and the
propertiesof postsynaptic receptors. The multiplicity of inhibitory
interneu-rons types, each defining morphologically, physiologically
and func-tionally distinct classesa,b adds to the difficulty in
measuring inhibi-tion. In the kainic acid model of temporal lobe
epilepsy (TLE),several parameters characterizing inhibition have
been investigatedin two morphologically and functionally different
inhibitory path-ways in the CA1 region of the hippocampus: the
pathway compris-ing the class of interneurons that specifically
project to the periso-matic region of CA1 pyramidal cells and which
tightly control theiroutput and the pathway comprising the class
that project to the den-drites and which control excitatory inputs
and dendritic firingc.According to the parameter being measured,
inhibition can appearunchangedd, decreasedeg or increasedgi in TLE.
The following alter-ations have been reported in brain slices from
epileptic animals:
(1) A reduction of synaptic and extrasynaptic GABAA
receptor-mediated currents following a probable modification of the
com-position of GABAA receptors subunits
d,j.(2) A deficit of GABA quantal release at perisomatic
synapses and
a depletion of the reserve pool of GABA-containing vesicles
consis-tent with the suggestion of an impairment of vesicular
release,although the number of perisomatic GABAergic terminals on
pyra-midal neurons is not modified in TLE (Ref. f).
(3) A reduction of the frequency of spontaneous IPSCs in
dendriticbut not somatic recordings that is probably caused by the
loss of den-dritic projecting interneuronsk.
(4) An increase of the excitability of various populations
ofinterneurons, i.e. both the number and the firing frequency of
spon-taneously firing interneurons are increased by 50% in TLE
(Ref. l).
These observations clearly show a multiplicity of
modifications,which can go in opposite directions even within a
giveninhibitory pathway. However, it is possible to get an overall
viewof inhibition in TLE by looking at the spontaneous
GABAergiccurrents received by the soma and dendrites of pyramidal
cellsduring steady state. This measurement reveals that the net
flux of
Cl2 through GABAA receptors is increased by 50% in somata inTLE,
i.e. the hyperactivity of perisomatic projecting interneuronsmore
than compensates for the pre- and postsynaptic deficits.
Bycontrast, the inhibitory drive is decreased in the dendrites,
i.e. thehyperactivity of the surviving dendritic projecting
interneuronsdoes not compensate totally for the loss of other
populations ofdendritic projecting interneurons and for the
postsynaptic deficit.This example demonstrates that the assessment
of the fate of inhi-bition necessitates the evaluation of each of
the parameters thatdefine inhibition and for each subspecific
pathway.
Referencesa Freund, T.F. and Buzsaki, G. (1996) Interneurons of
the hippocampus.
Hippocampus. 6, 347470b Parra, P. et al. (1998) How many
subtypes of inhibitory cells in the
hippocampus? Neuron 20, 983993c Miles, R. et al. (1996)
Differences between somatic and dendritic inhibition
in the hippocampus. Neuron 16, 815823d Esclapez, M. et al.
(1997) Operative GABAergic inhibition in hippocampal
CA1 pyramidal neurons in experimental epilepsy. Proc. Natl.
Acad. Sci.U. S. A. 94, 1215112156
e Buhl, E.H. et al. (1996) Zinc-induced collapse of augmented
inhibition byGABA in a temporal lobe epilepsy model. Science 271,
369373
f Hirsch, J.C. et al. (1999) Deficit of quantal release of GABA
inexperimental models of temporal lobe epilepsy. Nat. Neurosci. 2,
499500
g Gibbs, J.W. et al. (1997) Differential epilepsy-associated
alterations inpostsynaptic GABA(A) receptor function in dentate
granule and CA1neurons. J. Neurophysiol. 77, 19241938
h Nusser, Z. et al. (1998) Increased number of synaptic GABA(A)
receptorsunderlies potentiation at hippocampal inhibitory synapses.
Nature395, 172177
i Brooks-Kayal, A.R. (1998) Selective changes in single cell
GABA(A)receptor subunit expression and function in temporal lobe
epilepsy. Nat.Med. 4, 11661172
j Gibb, J.W. et al. (1996) Characterization of GABAA receptor
function inhuman temporal cortical neurons. J. Neurophysiol. 75,
14581471
k Bernard, C. et al. (1997) Selective loss of GABAergic
inhibition in theapical dendrites of CA1 pyramidal neurons in
temporal lobe epilepsy. Soc.Neurosci. Abstr. 838.10
l Bernard, C. et al. (1999) Increased inhibitory GABAergic drive
in the somaof CA1 pyramidal cells in experimental epilepsy. Soc.
Neurosci. Abstr.340.10
-
586 TINS Vol. 23, No. 11, 2000
and the EPSP probably because of a conduction blockas a result
of axonal membrane depolarization. Thisstudy not only reflects the
importance of the dose ofkainate used but also the lack of direct
relationshipbetween epileptogenesis and reduction of the
evokedEPSP. In physiological conditions, i.e. no
glutamateantagonists and intact preparations15, low concentra-tions
of kainate are sufficient to generate seizures,whereas larger
concentrations (micromolar doses) usu-ally produce an irreversible
loss of synaptic activity,presumably as a result of cell swelling
and cell death.
Kainate has also been suggested to modulate gluta-mate release
in the CA1 region. Indeed, kainatereduces the release of
[3H]L-glutamate from synapto-somes and the amplitude of evoked NMDA
receptor-mediated EPSCs (Ref. 21). However, because this
effectrequires particularly large concentrations (1100 mM)and
prolonged applications (1030 min), its physio-logical relevance
remains to be established.
Therefore, a presynaptic action of kainate that canreduce
glutamate release from mossy fibers probablyplays an important role
in mediating some effects ofkainate, but it is presently unclear
how this partici-pates in the epileptogenic actions of the
toxin.Other non-direct effects of kainate
Kainate has additional effects that might enhanceor reduce the
excitability of pyramidal neurons.Although these early studies were
carried out beforethe availability of selective AMPA receptor
antagon-ists, the effects observed with low concentrations
ofkainate were mediated by kainate and not AMPAreceptors because
only kainate receptors are activatedwith submicromolar
concentrations13. The followingeffects deserve emphasis.
(1) Low concentrations of kainate (100200 nM)facilitate the
repetitive firing of CA1 pyramidal neur-ons37. This effect is
mediated by the attenuation of theCa21-dependent K1 current (IAHP),
which is responsiblefor the afterhyperpolarization following Na1
spikesand by the reduction of the inward rectifier K1 current(IQ)
(Ref. 19).
(2) In the elegant study of Nistri and Cherubini20,kainate
(50400 nM) depressed the L-type Ca21 cur-rent. This effect was
prevented by intracellular dialysiswith BAPTA, suggesting that it
is mediated by anincrease in the inactivation of Ca21 currents via
a risein free intracellular Ca21. It is possible that such a riseof
intracellular Ca21 mediates other effects of kainate.
(3) Kainate reduces postsynaptic GABAB receptor-activated K1
currents38, further stressing possible second-messenger
cascades-mediated effects.
Concluding remarks
Kainate acts as a double-agent controlling thehippocampal
network activity via the activation of anheterogeneous network of
kainate receptors differen-tially distributed among inhibitory
interneurons andexcitatory pyramidal cells. Hence, kainate in
thenanomolar range generates seizures in CA3 at least inpart
through the activation of GluR6-containing recep-tors localized
postsynaptically at mossy fiber synapseson pyramidal cells.
Kainate, at similar low concentra-tions, massively increases tonic
inhibition via the acti-vation of GluR5-containing receptors
localized at glu-tamatergic synapses on GABAergic inter-neurons.
Thisdifferential expression of kainate receptors betweenneuronal
subtypes is reminiscent of other pathways,
for example of glutamate acting on metabotropicreceptors39 or
ACh acting on nicotinic receptors40.
The multiple facets of inhibition (i.e. evoked, spon-taneous or
miniature) and the extremely diversifiednetwork of interneurons
provide a rich repertoire ofexcitability modulations that cannot be
classified sim-ply as an increase or decrease of inhibition.
Thus,there is now direct evidence for two distinct forms
ofinhibition on pyramidal cells41 originating from twobroad classes
of interneurons: those that innervate thedendrites, control the
input of the hippocampal net-work and the propagation to the soma
of large calciumcurrents, and those that innervate the soma,
controlthe generation of Na1 action potential and hence theoutput
of the hippocampal network. Furthermore, apopulation of
interneurons is specialized to innervateother interneurons42,
thereby enabling a fine controlof the excitability of these cells.
As the distribution ofkainate receptor subtypes appears to be age-,
subunit-and target-selective, the net consequence of the
acti-vation of kainate receptors will vary: an increase of
theinterneuronal activity by kainate might even result ina
paradoxical reduction of its epileptogenic effects.
However, because only evoked kainate responseshave been
observed, the physiological conditionsunder which kainate receptors
are activated are notpresently clear. Interestingly, it has been
recentlyreported43 that pure kainate receptor-mediated sponta-neous
PSCs (not associated to AMPA receptor-mediatedPSCs) are observed at
early stages of maturation. Thissuggests a differential activation
of AMPA and kainatereceptors that can be regulated by the neuronal
activ-ity level or conditioned by a concomitant activation ofother
transmitters or synaptic pathways. All these pos-sibilities point
to a wider repertoire of modulation ofionotropic glutamatergic
synapses than previouslyenvisaged. Interestingly, activation of
kainate receptor-mediated PSCs in CA3 pyramidal neurons
requiresbrief tetani, whereas in interneurons a single stimulusis
sufficient suggesting that under physiological condi-tions
postsynaptic kainate receptors on interneuronsmight be more
frequently recruited than those onpyramidal cells at mossy fiber
synapses. Furthermore,there are some examples where the efficacy of
EPSP-spike coupling is particularly strong and precise
ininterneurons resulting in an efficient feed-forwardrecruitment of
interneur-ons44. If this is the case,kainate receptor-mediated
EPSPs might play an impor-tant role in resetting endogenous
rhythmic activitiesthat are controlled by interneurons during the
genera-tion of seizures in addition to normal
physiologicalconditions. It remains to determine the
mechanismsresponsible for the shift of the
inhibitory/excitatorybalance in the somato-dendritic compartments
thatwill ultimately lead to epileptogenesis.
Selected references1 Ben-Ari, Y. (1985) Limbic seizure and brain
damage produced by
kainic acid: mechanisms and relevance to human temporal
lobeepilepsy. Neuroscience 14, 375403
2 Nadler, J.V. (1981) Minireview. Kainic acid as a tool for the
studyof temporal lobe epilepsy. Life Sci. 29, 20312042
3 Sloviter, R.S. (1996) Hippocampal pathology and
pathophysiologyin temporal lobe epilepsy. Neurologia 11, 2932
4 Miles, R. and Wong, R.K (1983) Single neurones can
initiatesynchronized population discharge in the hippocampus.
Nature306, 371373
5 Monaghan, D.T. and Cotman, C.W. (1982) The distribution
of[3H]kainic acid binding sites in rat CNS as determined
byautoradiography. Brain Res. 252, 91100
Y. Ben-Ari and R. Cossart Kainate and seizuresRE V I E W
-
TINS Vol. 23, No. 11, 2000 587
6 Tremblay, E. et al. (1985) Autoradiographic localization of
kainic acidbinding sites in the human hippocampus. Brain Res. 343,
378382
7 Berger, M.L. et al. (1986) Limbic seizures induced by
systemicallyapplied kainic acid: how much kainic acid reaches the
brain? Adv.Exp. Med. Biol. 203, 199209
8 Gaiarsa, J.L. et al. (1994) Neonatal irradiation prevents
theformation of hippocampal mossy fibers and the epileptic action
ofkainate on rat CA3 pyramidal neurons. J. Neurophysiol. 71,
204215
9 Debonnel, G. et al. (1990) Neurotoxic effect of domoic
acid:mediation by kainate receptor electrophysiological studies in
therat. Can. Dis. Wkly Rep. 16, 5968
10 Ben-Ari, Y. and Gho, M. (1988) Long-lasting modification of
thesynaptic properties of rat CA3 hippocampal neurones induced
bykainic acid. J. Physiol. 404, 365384
11 Fisher, R.S. and Alger, B.E. (1984)
Electrophysiologicalmechanisms of kainic acid-induced epileptiform
activity in therat hippocampal slice. J. Neurosci. 4, 13121323
12 Robinson, J.H. and Deadwyler, S.A. (1981) Kainic acid
producesdepolarization of CA3 pyramidal cells in the vitro
hippocampalslice. Brain Res. 221, 117127
13 Cossart, R. et al. (1998) GluR5 kainate receptor activation
ininterneurons increases tonic inhibition of pyramidal cells.
Nat.Neurosci. 1, 470478
14 Rodriguez-Moreno, A. et al. (1997) Kainate
receptorspresynaptically downregulate GABAergic inhibition in the
rathippocampus. Neuron 19, 893901
15 Khalilov, I. et al. (1999) Maturation of
kainate-inducedepileptiform activities in interconnected intact
neonatal limbicstructures in vitro. Eur. J. Neurosci. 11,
34683480
16 Cronin, J. and Dudek, F.E. (1988) Chronic seizures and
collateralsprouting of dentate mossy fibers after kainic acid
treatment inrats. Brain Res. 474, 181184
17 Represa, A. et al. (1987) Kainate binding sites in the
hippo-campal mossy fibers: localization and plasticity.
Neuroscience20, 739748
18 Nadler, J.V. et al. (1980) Loss and reacquisition of
hippocampalsynapses after selective destruction of CA3CA4 afferents
withkainic acid. Brain Res. 191, 387403
19 Gho, M. et al. (1986) Kainate reduces two
voltage-dependentpotassium conductances in rat hippocampal neurons
in vitro.Brain Res. 385, 411414
20 Nistri, A. and Cherubini, E. (1991) Depression of a
sustainedcalcium current by kainate in rat hippocampal neurones in
vitro.J. Physiol. 435, 465481
21 Chittajallu, R. et al. (1996) Regulation of glutamate release
bypresynaptic kainate receptors in the hippocampus. Nature379,
7881
22 Castillo, P.E. et al. (1997) Kainate receptors mediate a slow
post-synaptic current in hippocampal CA3 neurons. Nature 388,
182186
23 Vignes, M. and Collingridge, G.L. (1997) The synaptic
activationof kainate receptors. Nature 388, 179182
24 Frerking, M. et al. (1998) Synaptic activation of kainate
receptorson hippocampal interneurons. Nat. Neurosci. 1, 479486
25 Mulle, C. et al. (1998) Altered synaptic physiology and
reducedsusceptibility to kainate- induced seizures in
GluR6-deficient mice.Nature 392, 601605
26 Vignes, M. et al. (1998) The GluR5 subtype of kainate
receptorregulates excitatory synaptic transmission in areas CA1 and
CA3of the rat hippocampus. Neuropharmacology 37, 12691277
27 Bahn, S. et al. (1998) Kainate receptor gene expression in
thedeveloping rat brain. J. Neurosci. 14, 55255545
28 Henze, D.A. et al. (1997) Large amplitude miniature
excitatorypostsynaptic currents in hippocampal CA3 pyramidal
neurons areof mossy fiber origin. J. Neurophysiol. 77, 10751086
29 Bureau, I. and Mulle, C. (1998) Potentiation of
GABAergicsynaptic transmission by AMPA receptors in mouse
cerebellarstellate cells: changes during development. J.
Physiol.509, 817831
30 Clarke, V.R. et al. (1997) A hippocampal GluR5 kainate
receptor regulating inhibitory synaptic transmission. Nature389,
599603
31 Min, M.Y. et al. (1999) Synaptically released glutamate
reducesgamma-aminobutyric acid (GABA)ergic inhibition in
thehippocampus via kainate receptors. Proc. Natl. Acad. Sci. U. S.
A.96, 99329937
32 Rodriguez-Moreno, A. et al. (2000) Two populations of
kainatereceptors with separate signaling mechanisms in
hippocampalinterneurons. Proc. Natl. Acad. Sci. U. S. A. 97,
12931298
33 Frerking, M. et al. (1999) Mechanisms underlying kainate
receptor-mediated disinhibition in the hippocampus. Proc. Natl.
Acad. Sci.U. S. A. 96, 1291712922
34 Petralia, R.S. et al. (1994) Histological and
ultrastructurallocalization of the kainate receptor subunits, KA2
and GluR6/7,in the rat nervous system using selective antipeptide
antibodies.J. Comp. Neurol. 349, 85110
35 Kamiya, H. and Ozawa, S. (2000) Kainate
receptor-mediatedpresynaptic inhibition at the mouse hippocampal
mossy fibresynapse. J. Physiol. 523, 653665
36 Westbrook, G.L. and Lothman, E.W. (1983) Cellular and
synapticbasis of kainic acid-induced hippocampal epileptiform
activity.Brain Res. 273, 97109
37 Cherubini, E. et al. (1990) Effects of kainate on the
excitability ofrat hippocampal neurones. Epilepsy Res. 5, 1827
38 Rovira, C. et al. (1990) Block of GABAb-activated K1
conductanceby kainate and quisqualate in rat CA3 hippocampal
pyramidalneurones. Pflgers Arch. 415, 471478
39 McBain, C.J. et al. (1994) Activation of metabotropic
glutamatereceptors differentially affects two classes of
hippocampalinterneurons and potentiates excitatory synaptic
transmission.J. Neurosci. 14, 44334445
40 Frazier, C.J. et al. (1998) Acetylcholine activates an
alpha-bungarotoxin-sensitive nicotinic current in rat hippo-campal
interneurons, but not pyramidal cells. J. Neurosci. 18,
11871195
41 Miles, R. et al. (1996) Differences between somatic and
dendriticinhibition in the hippocampus. Neuron 16, 815823
42 Gulyas, A.I. et al. (1996) Interneurons containing calretinin
arespecialized to control other interneurons in the rat
hippocampus.J. Neurosci. 16, 33973411
43 Kidd, F.L. and Isaac, J.T. (1999) Developmental and
activity-dependent regulation of kainate receptors at
thalamocorticalsynapses. Nature 400, 569573
44 Csicsvari, J. et al. (1998) Reliability and state dependence
ofpyramidal cell-interneuron synapses in the hippocampus:
anensemble approach in the behaving rat. Neuron 21, 179189
45 Represa and Ben-Ari, Y. (1992) Kindling is associated with
theformation of novel mossy fibre synapses in the CA3 region.
Exp.Brain Res. 92, 6978
Y. Ben-Ari and R. Cossart Kainate and seizures RE V I E W
AcknowledgementsThe authors areindebted toC. Bernard,M. Esclapez
andJ.C. Hirsch for theirmajor contributionto most of theresults
reviewed inthis paper and fortheir helpful comments on
themanuscript.
B O O K R E V I E W S
Most theoretical articles and books onpain begin nowadays with
the official IASPdefinition of pain: an unpleasant sensoryand
emotional experience associated withactual or potential tissue
damage, ordescribed in terms of such damage1. Theythen go on to
agree with the definition andproceed with their discussion.
DonaldPrice in his book Psychological Mechanismsof Pain and
Analgesia follows a differenttack. He disagrees with the
definition,arguing that it does not emphasize the
experiential nature of pain. He proposes anew definition: pain
is a somatic perceptioncontaining (1) a bodily sensation with
qualitieslike those reported during tissue-damagingstimulation, (2)
an experienced threat associ-ated with this sensation, (3) a
feeling ofunpleasantness or other negative emotionbased on this
experienced threat.
Notice how much Price packs into hisdefinition of pain. Each
unit of pain con-tains the sensation of pain itself, the
feelingthat one is somehow being threatened and
a negative affective reaction to the sen-sation and feeling.
However, this is toomuch, for two reasons. First, we can
dis-sociate the negative affective reactionsfrom pain sensations,
either pharmacologi-cally using, for example, fentanyl, or
bio-logically using, for example, a frontal lobot-omy.
Understanding and treating theconcomitant emotional reactions to
painsensations are important and Price is cor-rect in stating that
scientists and cliniciansdo not pay enough attention to this
aspectof pain patients. However, acknowledgingthese facts does not
make reactions topain part of pain itself.
Psychological Mechanisms of Pain and Analgesiaby Donald D.
Price, IASP Press, 1999. $69.00 (xiii 1 248 pages) ISBN 0 931092 29
9