Main

Epilepsy arises from imbalances between excitatory and inhibitory synaptic transmission in key brain areas such as hippocampus and temporal cortex, in which fast synaptic excitatory neurotransmission is mediated via activation of ionotropic glutamate receptors. Ionotropic glutamate receptors are divided, on pharmacological and molecular biological criteria, into N-methyl-D-aspartate (NMDA), α-amino-3-hydroxy-5-methyl-4-isoxazolepropionate (AMPA) and kainate subtypes1,2. Kainic acid is a powerful convulsant and animal models of epilepsy have been built around this observation3,4,5.

Kainate receptors have diverse roles in both physiological and pathological events in the brain6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30. In particular, developments with novel pharmacological agents10,12,19,22,31 have enabled insights into the involvement of the GLUK5 (IUPHAR nomenclature32, formerly known as GluR5 or iGlu5; Table 1) subunit of kainate receptors in synaptic transmission8,13,14,29 and plasticity19,25,26,30. For example, within the hippocampus, GLUK5-containing kainate receptors are involved in frequency facilitation30 and induction of long-term potentiation (LTP)19 in mossy fiber pathways, and in excitatory drives of inhibitory CA1 interneurons14. In addition, pharmacological activation of GLUK5-containing kainate receptors inhibits both excitatory16 and inhibitory10 synaptic transmission. Given these diverse functions, GLUK5 antagonism could potentially increase or decrease excitability, with a net effect that could be either pro- or anti-convulsant. In addition, it still remains conjectural to what extent, if any, these GLUK5 receptors are activated by endogenous neurotransmitter during epileptic states.

Table 1 Selectivity profile of GLUK5 receptor antagonists.
Full size table

Here we used subtype-selective GLUK5 receptor antagonists LY377770 (refs. 12 and 22; an active enantiomer of LY294486, ref. 10) and LY382884 (refs. 12 and 19) to investigate whether GLUK5-containing kainate receptors are involved in pilocarpine-induced epileptiform activity in the CA3 region of hippocampal slices and pilocarpine-induced limbic seizures in conscious rats. In both models, the GLUK5 antagonists prevented both the induction and maintenance of such activity. Additionally, GLUK5 antagonism prevented stimulus train–induced bursting in hippocampal slices and seizure activity induced by 6 Hz corneal stimulation in vivo. We propose, therefore, that the GLUK5 subtype of the kainate receptor can be used as a target to develop future anticonvulsant drugs.

Results

Subtype-selectivity of GluK5 antagonists

The decahydroisoquinoline used initially for this investigation was LY377770, the selectivity of which is presented in Table 1. This antagonist inhibited human recombinant homomeric GLUK5 receptors transfected in HEK293 cells with an inhibitory constant (Ki) value of 2.3 μM. Ki values for other kainate receptor subunits tested exceeded 100 μM. Homomeric AMPA receptors were also inhibited by LY377770 but with Ki values in excess of 20 μM. LY377770 inhibited kainate-induced currents in DRG cells with a median inhibitory concentration (IC50; versus 30 μM kainate) of approximately 0.5 μM. In contrast, it was less potent at inhibiting AMPA (30 μM)- and NMDA (10 μM)-induced currents in hippocampal neurons, with approximate IC50 values of 13 and 50 μM, respectively (Table 1). The equivalent data for the other decahydroisoquinoline used, LY382884, has been published previously19. This compound is less potent than LY377770 as an antagonist at homomeric GLUK5 receptors but has an even greater selectivity versus AMPA and NMDA receptors. Both LY377770 and LY382884 were also effective against heteromeric assemblies of GLUK5 and GLUK6 (GluR6) subunits (Table 1).

Because pilocarpine is a muscarinic agonist, we also examined the possibility that the GLUK5 antagonists act directly on muscarinic cholinergic receptors. Both LY377770 and LY382884 were inactive at 10 μM in both radioligand binding and functional assays on cloned M1-M5 muscarinic receptors expressed in HEK293 cells (C. Felder, unpub. observ.).

Antagonism of kainate-induced disinhibition

Based on the observation that GLUK5-selective kainate receptor antagonism inhibits kainate-induced depression of evoked GABAergic synaptic inhibition in the CA1 region of the hippocampus10, we predicted that LY377770 might possess anticonvulsant activity10. On the other hand, the ability of this antagonist to inhibit kainate receptor-mediated presynaptic inhibition of excitatory transmission16 would tend to suggest a proconvulsant profile. To determine which effect is predominant, we initially extended our observations in the CA1 region of the hippocampus. We used extracellular field potential recording to investigate the effects of 3 μM kainate on evoked population spikes in hippocampal slices. This treatment reliably led to the appearance of a secondary population spike or spikes, reflecting the underlying disinhibition33. This was fully reversible upon washout of kainate. Perfusion with LY377770 (1.5 μM) before application of kainate greatly reduced or eliminated, in a fully reversible manner, the appearance of the secondary population spike(s) (n = 8; Fig. 1). These data demonstrate the involvement of GLUK5-containing receptors in kainate-induced disinhibition.

Figure 1: Antagonism of GLUK5-containing kainate receptors suppresses kainate-induced disinhibition in the hippocampus.
figure 1

Field potential recordings from stratum pyramidale of area CA1 to illustrate the population effect of kainate-induced disinhibition and its antagonism by pre-treatment with LY377770. The graphs plot the mean ± s.e.m. amplitude of the primary (black circle) and secondary (white circle) population spikes versus stimulus intensity, for eight experiments. The traces are averages of four successive responses, evoked by a stimulus intensity of 15 V, from one representative experiment.

Full size image

Reduction of ongoing epileptiform activity

The finding that either a GLUK5 kainate receptor antagonist or genetic elimination of the GLUK6 kainate receptor subunit17 prevents kainate-induced epileptiform activity shows that these subunits can contribute to the mediation of excitability changes induced by exogenous activation of kainate receptors. What is not known is whether kainate receptors activated by endogenous glutamate are actually involved in the epileptic discharges. A more direct test of this possibility is to generate epileptiform activity via a standard model that is used for the evaluation of anticonvulsant drugs and that does not involve direct stimulation of kainate receptors. For this, we used the muscarinic agonist pilocarpine in both in vitro and in vivo models34,35,36. We made recordings from the CA3 cell body region of hippocampal slices in response to stimulation within the cell body layer of the dentate gyrus (Fig. 2). The response, in control medium, comprised an initial negative deflection followed by a slower positive wave (field excitatory postsynaptic potential; fEPSP). Perfusion with pilocarpine (15 μM, 20 min) caused the appearance of multiple population spikes in response to electrical stimulation and the appearance of spontaneous inter-ictal bursts (Fig. 2a ii). Such discharges appeared between 11 and 17 min after starting pilocarpine infusion and lasted for the duration of the experiment (up to 5 h; data not shown). This epileptiform activity was dramatically reduced (Fig. 2a iii) in a reversible manner (Fig. 2a iv), by LY377770 (1.5 μM) in all six slices tested. These data suggest that this in vitro epileptiform activity involves an action of endogenous L-glutamate on GLUK5-containing kainate receptors.

Figure 2: GLUK5 antagonists and pilocarpine-induced epileptiform activity.
figure 2

(a) Reduction of epileptiform activity in vitro by LY377770. Electrically evoked synaptic responses (top) and DC recordings (middle) in control conditions (i), following the addition of 15 μM pilocarpine (ii), following the addition of 1.5 μM LY377770 to the pilocarpine-containing medium (iii) and following washout of LY377770 and pilocarpine (iv). The synaptic responses are averages of four successive responses that were obtained at the times indicated on the graph (i–iv). The DC recordings are 10-min excerpts to illustrate the maximum effects of the different treatments. The regular deflections in the excerpts of the chart records are stimulus artifacts, upon which are superimposed the pilocarpine-induced spontaneous inter-ictal bursts. Bottom, electrode placements for recording spontaneous and electrically evoked responses in area CA3 of the rat hippocampal slice. (b) Prevention of the initiation of epileptiform activity in vitro by LY377770. These are equivalent electrically evoked synaptic responses, but 1.5 μM LY377770 was added before 15 μM pilocarpine. LY377770 prevented the development of epileptiform activity, as determined 2 h after washout of LY377770 (v), but at this time, a second application of pilocarpine readily induced epileptiform activity (vi).

Full size image

Reduction of induction of epileptiform activity

Next we wanted to determine whether LY377770 was also able to prevent initiation of epileptiform activity by pilocarpine. Previous treatment with LY377770 (1.5 μM) had no obvious effects on basal synaptic excitatory transmission in all four slices tested (Fig. 2b ii). Then we applied pilocarpine (15 μM) in the presence of LY377770. In all slices, the effects of pilocarpine were almost completely prevented by LY377770 (Fig. 2b iii), and no evidence for multiple spikes was seen during 2 hours of washout (Fig. 2b iv and v). To ensure that this prolonged inhibition of epileptiform activity was not exerted by residual LY377770, pilocarpine was re-applied 2 hours after commencing washout of LY377770. This second application of pilocarpine readily induced epileptiform activity (Fig. 2b vi), which in turn was sensitive to a short perfusion with LY377770 (Fig. 2b vii). Rapid return of the epileptiform activity was observed within 30 min of LY377770 withdrawal (Fig. 2b viii), illustrating that this compound has a relatively short duration of action in these slice experiments. This experiment also showed that pilocarpine effectively washed out of the recording system rapidly; otherwise, epileptiform activity would have been initiated upon washout of the first application of LY377770.

AMPA receptors are not needed for induction

AMPA receptor antagonists attenuate epileptiform activity in several epilepsy models37,38. Whereas LY377770 is selective for GLUK5 kainate receptors over AMPA receptors, a subliminal involvement of AMPA receptors could not be fully excluded. We therefore compared the effects of a selective AMPA receptor antagonist, GYKI53655 (refs. 2,39,40), with those of LY382884, which compared to LY377770 is more selective, though less potent, as a GLUK5 kainate receptor antagonist (Fig. 3).

Figure 3: The induction of epileptiform activity requires activation of kainate but not AMPA receptors.
figure 3

Data are presented as in Fig. 2b. (a) The selective GLUK5 receptor antagonist, LY382884 (10 μM), did not affect low-frequency synaptic responses (ii), but prevented the development of pilocarpine-induced epileptiform activity (iii–v). A second application of pilocarpine readily induced epileptiform activity 2.5 h after washout of LY382884 (vi). (b) The AMPA receptor antagonist, GYKI53655 (30 μM), substantially antagonized the low-frequency synaptic response before, during and immediately after (ii) co-perfusion of 15 μM pilocarpine. After washout of GYKI53655, epileptiform activity appeared, although pilocarpine was not re-applied (iii–iv). This epileptiform activity was reversibly blocked by LY382884 (10 μM; v and vi). The graph above, on the same time scale, shows the time course of washout of 30 μM GYKI53655 monitored using recordings of mossy fiber–evoked fEPSPs. Note that full recovery takes approximately 2 h.

Full size image

Application of LY382884 alone (10 μM) did not affect basal synaptic excitatory transmission (Fig. 3a ii, n = 5). Again, co-application of LY382884 with pilocarpine (15 μM) prevented the development of multiple spikes in all five slices tested (Fig. 3a iii–v). A second application of pilocarpine after 2 hours of washout readily induced epileptiform activity (Fig. 3a vi). Perfusion with 30 μM GYKI53655 substantially antagonized the fEPSP, but not the initial negative waveform. Co-perfusion with pilocarpine (15 μM) did not obviously affect this depressed response (Fig. 3b ii). After washout of GYKI53655, there was the appearance of multiple population spikes within 2 hours in all five slices tested (Fig. 3b iii–iv). The delayed development of epileptiform activity can be attributed to the slow washout of GYKI53655 (Fig 3b inset; see also ref. 19). Addition of LY382884 (10 μM) dramatically reduced this epileptiform activity, in a rapidly reversible manner (Fig. 3b v and vi).

These data show that selective antagonism of GLUK5-containing kainate receptors inhibited both the induction and expression of epileptiform activity. In contrast, selective antagonism of AMPA receptors suppressed the expression of epileptiform activity but did not prevent the induction of epileptiform activity, which was manifest upon washout of the antagonist.

Characterization of the in vivo focal pilocarpine model

Acute pilocarpine administration, focally in the hippocampus or systemically, leads to limbic seizures in rats with characteristics of human complex partial seizures34,35. To study the anticonvulsant potential of GLUK5 receptor antagonists in vivo, we used simultaneous microdialysis, electrocorticography and behavioral observations35 of seizures induced by pilocarpine (10 mM), perfused via a microdialysis probe in the dorsal hippocampus for 40 min. Behaviorally, a pre-convulsive state, typified by scratching, wet-dog shakes and tremor, invariably progressed into excess salivation, masticatory movements, rearing and finally into limbic motor convulsions. The appearance of the seizure activity was associated with concomitant, long-lasting and significant increases in extracellular hippocampal glutamate and GABA concentrations, as measured by microdialysis (Fig. 4a). This provided a neurochemical measure of the increase in network synaptic activity. The simultaneously recorded electrocorticogram showed a pattern typical of tonic–clonic seizures (Fig. 4a). Frequent periods of intermittent seizures, which did not evolve into status epilepticus, lasted for the duration of the experiment (up to 8 h) in all control animals (n = 8).

Figure 4: LY377770 is able to block established pilocarpine-induced limbic seizures.
figure 4

(a) Intrahippocampal pilocarpine (10 mM; 40 min) perfusion invariably induces tonic–clonic seizures in conscious rats. Electrographic recordings taken from a single experiment and dialysate concentrations of glutamate and GABA (percent of baseline levels) for 8 similar experiments during baseline (i), after sham injection with physiological saline (ii), 20–40 min (iii) and 80–100 min after ending pilocarpine perfusion (iv). (b) Data of six similar experiments during baseline (i) and when pilocarpine-induced seizures were fully expressed (ii). (iii–iv) 75 mg/kg of LY377770 can abolish the tonic–clonic pilocarpine-induced seizures. *P < 0.05; **P < 0.01.

Full size image

Blockade of ongoing seizures by ly377770

To determine whether LY377770 was able to block ongoing limbic seizures, we administered 30 and 75 mg/kg intraperitoneally (i.p.) after pilocarpine-induced seizures had been established. The lower dose of the GLUK5 antagonist abolished the ongoing clonic–tonic seizures in two of four rats tested. The administration of 75 mg/kg blocked seizure activity within 20 min in all animals (n = 6). The rats displayed no further seizures during the subsequent 2 h of the experiment. Concomitantly, the glutamate and GABA levels returned to baseline values and the electrographic patterns characteristic of the initial tonic–clonic seizure activity decayed via rhythmical peak/wave discharges (2 cycles/s) to a post-ictal pattern (Fig. 4b). These data indicate that GLUK5-containing receptors are involved in the maintenance of seizures, although, in these in vivo experiments, specificity for such receptors cannot be guaranteed (see Discussion).

GLUK5 antagonists prevent seizure initiation

To investigate the ability of GLUK5 antagonists to prevent the onset of seizures in freely moving animals, LY377770 (10–75 mg/kg) was injected intraperitoneally 30 min before the pilocarpine perfusion. There were no obvious behavioral, electrocorticographic or biochemical changes associated with the administration of the GLUK5 receptor antagonist alone, but seizure activity was clearly blocked in a dose-dependent manner. Thus, 10 mg/kg of LY377770 (n = 2) was without effect and 15 mg/kg protected one of two animals, whereas 30 and 75 mg/kg each protected all six animals tested. The latter doses completely abolished all behavioral, neurochemical and electroencephalographic signs of seizure activity (Fig. 5a).

Figure 5: GLUK5 receptor antagonists prevent focally evoked pilocarpine-induced limbic seizures and accompanying rises in extracellular (EC) glutamate and GABA concentrations in conscious rats.
figure 5

(a) Data are presented as in Fig. 4 for 6 similar experiments where the rats were pretreated with LY377770 (30 mg/kg, i.p.). Electrocorticographic recordings and dialysate concentrations of glutamate and GABA (in percent of baseline levels) during baseline (i), after i.p. injection with LY377770 (ii), 20–40 min (iii) and 80–100 min after ending pilocarpine perfusion (iv). (b) Equivalent data for six animals pretreated with LY382884 (150 mg/kg).

Full size image

We also tested the weaker but more selective GLUK5 receptor antagonist LY382884, again injected intraperitoneally 30 min before pilocarpine perfusion. As with LY377770, there were no obvious behavioral, electrocorticographic or biochemical changes associated with its administration. At a dose of 50 mg/kg, LY382884 did not prevent pilocarpine-induced convulsions (n = 2), whereas 100 mg/kg prevented the full limbic seizures leaving only some pre-convulsive signs, such as wet-dog shakes (n = 2). Complete protection against all behavioral, electrographic and biochemical signs was obtained with 150 mg/kg (Fig. 5b, n = 6).

As GLUK5 receptor antagonists inhibited ongoing seizures, this apparent suppression of seizure induction could have been due to a prolonged action of the antagonists in vivo. We therefore extended the period of observation following the subsequent pilocarpine administration to more than 5 hours. No behavioral, biochemical or electrographic signs of seizures were detected at any time throughout this period (Fig. 6a, n = 6). However, in parallel experiments, a second administration of pilocarpine approximately 4 hours after the administration of LY377770 resulted in full-blown seizures in all six rats (Fig. 6b). Therefore, because at 4 hours or less there was insufficient LY377770 to prevent this seizure activity, the result in Fig. 6a can be attributed to its ability to block the induction of seizures.

Figure 6: LY377770 (30 mg/kg) prevents the initiation of pilocarpine-induced seizures in conscious rats.
figure 6

Data presented as in Fig. 4. (a) Seizures and changes in amino acids are absent for more than 5 h after pilocarpine perfusion following systemic LY377770 injection (n = 6). (b) A second pilocarpine perfusion approximately 4 h after treatment with LY377770 readily induced seizure activity (n = 6). *P < 0.05.

Full size image

Action of GLUK5 in other models

To determine how generalized the effectiveness of GLUK5 receptor antagonism was in models of epileptiform activity, we investigated the action of LY382884 in two additional in vitro models. In the first model, known as stimulus train–induced bursting (STIB)41, we delivered multiple tetani to the mossy fiber pathway. This treatment readily induced synaptically evoked, inter-ictal activity that persisted beyond the stimulus trains. LY382884, present during the tetani, dramatically reduced the development of this activity (Fig. 7a). In the second model, we applied the non-competitive GABAA receptor antagonist, picrotoxin42. This treatment resulted in both evoked and spontaneous inter-ictal–like activity, which was not prevented by LY382884 (Fig. 7b).

Figure 7: GLUK5 receptor antagonists inhibit the generation of electrically induced but not picrotoxin-induced epileptiform activity.
figure 7

(a) Ten periods of tetanic stimulation were delivered and the duration of the first synaptic response following each tetanus is plotted for controls (n = 9) and slices in which LY382884 (10 μM) was present during each tetanus (n = 4). The traces are single responses from a representative control (a–d) and LY382884-treated slice (e–h) at the time points indicated. In controls, the STIB protocol led to a rapid induction of epileptiform activity (b) that persisted for at least 1 h following the last tetanus (d). These changes were greatly reduced by LY382884. (b) LY382884 (10 μM) did not prevent epileptiform activity induced by picrotoxin (50 μM). The traces are single responses from a representative experiment using LY382884. The histograms plot pooled data of burst duration at the corresponding time points for four experiments and four interleaved control experiments. (c) Number of animals displaying seizures (gray bars) and the number per group (white bars) for the 6 Hz corneal stimulation in mice and picrotoxin treatment in rats.

Full size image

We next investigated two in vivo models. In an electrically induced seizure model, which comprised 6 Hz corneal stimulation in mice43, LY377770 inhibited the generation of seizure activity in a dose-dependent manner (median effective dose (ED50), 28 ± 1 mg/kg i.p.; n = 8 per group; Fig. 7c). These mice showed loss of righting reflex and some deaths at 300 mg/kg but not at 100 mg/kg LY377770. In contrast, picrotoxin-induced seizures42 in male Wistar rats were not affected by LY377770 (75 mg/kg i.p. injected 30 min before 3 mg/kg i.p. picrotoxin; n = 6; Fig. 7c).

Discussion

Using pilocarpine-induced and electrically evoked epileptiform activity in vitro and seizure activity in vivo, we showed that GLUK5-containing receptor antagonists possess clear anticonvulsant activity, at doses with no overt behavioral side effects. Therefore, GLUK5 antagonists are a class of compounds with potential utility for the future treatment of human temporal lobe epilepsy.

Selectivity of the GLUK5 receptor antagonists

LY377770 and LY382884, like some other decahydroisoquinolines, are potent GLUK5 receptor antagonists on both cloned and native receptors2,10,12,19,22,31. As shown here, they are similarly effective on GLUK5 homomers and GLUK5/K6-containing heteromers, as well as the native GLUK5 assemblies in rat DRG neurons. They showed weak activity at GLUK2 homomers and GLUK2/K6 heteromers, with Ki values above 100 μM, and were inactive on homomeric GLUK6 and GLUK7. Thus, based on available information, we conclude that the effects on kainate receptors described here were due to an action on GLUK5 homomers or GLUK5-containing heteromeric receptors. The relative potency of LY377770 and LY382884 is consistent with this conclusion.

Decahydroisoquinolines also display some AMPA receptor antagonism2. However, the concentration of LY377770 used in vitro (1.5 μM) was just below the Ki value for GLUK5 and at least 14-fold less than the Ki value for the most sensitive AMPA subunit, that is, GLUA4. Moreover, LY382884, which is considerably more selective for GLUK5-containing kainate receptors versus AMPA receptors19, mimicked all of the effects of LY377770. With both antagonists, the effects of inhibiting epileptiform activity occurred without any antagonism of the low-frequency synaptic response, which is mediated predominantly via the synaptic activation of AMPA receptors. Furthermore, GYKI53655, which has a high degree of selectivity for AMPA versus kainate receptors, substantially inhibited low-frequency excitatory synaptic transmission, but did not prevent pilocarpine-induced epileptiform activity, after washout of the antagonist.

NMDA receptor antagonists are also effective in antagonizing epileptiform activity. However, an effect on NMDA receptors can be excluded, as LY377770 is only a weak NMDA receptor antagonist and LY382884 is essentially inactive on this receptor (IC50 versus 10 μM NMDA on hippocampal neurons > 300 μM). In hippocampal slices, neither antagonist affects GABAA or GABAB receptor-mediated synaptic inhibition. Furthermore, they antagonize the effects of the selective GLUK5 receptor agonist ATPA10,16,19 but not the effects of agonists for any of the other receptor subtypes that have been tested (mGlu, adenosine, muscarinic and adrenergic receptors19,44).

Direct blockade by these GLUK5 antagonists of the action of pilocarpine on muscarinic receptors can also be excluded, as LY377770 and LY382884 were inactive on all muscarinic subtypes. Furthermore, both GLUK5 antagonists were effective in electrical stimulation models that do not involve muscarinic receptors. Therefore, all effects of LY377770 and LY382884 observed in the in vitro experiments can be attributed to antagonism of GLUK5-containing kainate receptors rather than antagonism of AMPA, NMDA or muscarinic receptors.

For the systemic administration of antagonists, the precise concentrations of the compounds at their sites of action in the brain are not, of course, accurately known. It is possible, therefore, that the compounds, in particular, LY377770, were also acting at AMPA or other receptors. Hence, experiments with these drugs on GLUK5 knockout animals will be required to prove definitively that the anticonvulsant effects are on this kainate receptor subtype. However, the potency difference between LY377770 and the more selective LY382884 observed in the present in vivo experiments paralleled their relative potencies as GLUK5 antagonists and their actions on epileptiform activity in vitro. The lack of effect of LY377770 on picrotoxin-induced seizures also suggests a lack of functional AMPA receptor antagonism. Although the site of action following systemic application cannot be ascertained, an action within the hippocampus seems likely, given the effectiveness of the antagonists in the hippocampal slice experiments. In addition, intrahippocampal perfusion of LY377770, via the same microdialysis probe used to apply pilocarpine, was able to prevent fully the development of seizures (I.S., unpub. observ.). Thus, the effects of the GLUK5 receptor antagonists on both the initiation and expression of the limbic seizures in conscious rats were likely to be predominantly via GLUK5-containing kainate receptors within the hippocampus.

Mode of action of GLUK5 antagonists

Here we identified the roles of GLUK5-containing kainate receptors in both the initiation and expression of epileptiform activity in vitro and seizure activity in vivo. Future studies will be required to establish the precise cellular mechanisms responsible for these effects. In this context, there are several potential mechanisms by which activation of GLUK5-containing kainate receptors could both initiate and sustain epileptiform or seizure activity. For example, first, the activation of GLUK5-containing kainate receptors results in inhibition of GABAergic synaptic transmission10. Conceivably, during the induction or expression of epileptiform or seizure activity, these receptors are activated and lead to a depression of synaptic inhibition. LY377770 and LY382884 would then work by preserving synaptic inhibition. Consistent with this possibility, GLUK5 antagonists were ineffective in the in vitro and in vivo picrotoxin models, where GABAA receptor-mediated synaptic inhibition was antagonized directly. Second, the synaptic activation of presynaptic GLUK5-containing kainate receptors results in frequency-dependent facilitation of mossy fiber synaptic transmission30. This produces a potentially unstable positive feedback whereby the release of L-glutamate leads to further augmentation of L-glutamate release. Thus, by inhibition of these receptors, LY377770 and LY382884 would greatly reduce high-frequency synaptic transmission without affecting low-frequency synaptic transmission in the mossy fiber pathway. Third, GLUK5-containing kainate receptors are involved in the induction of mossy fiber LTP19. Thus, it is possible that excessive activation of GLUK5 receptors induces epilepsy through a kindling process, analogous to NMDA receptor-dependent LTP and kindling at CA1 synapses and elsewhere. Consistent with this possibility, LY382884 dramatically inhibited the appearance of epileptiform activity evoked by stimulus train–induced bursting. Therefore, it is likely that multiple mechanisms could explain the involvement of GLUK5-containing kainate receptors in epileptiform or seizure activity.

The frequency-dependence of GLUK5-containing kainate receptor-mediated synaptic processes could explain why GLUK5 receptor antagonists were effective against epileptiform activity while not affecting low-frequency synaptic transmission8,19. This property, coupled to the limited expression of GLUK5 receptor subunits in the brain, may explain the lack of overt side effects observed in the present study, relative to other agents tested in this model for limbic epilepsy35. Side effects coupled with poor efficacy are issues associated with the currently available antiepileptic drugs45,46,47.

Role of GLUK5 receptors in other models

Here we showed that the sensitivity to GLUK5 antagonism includes seizure-like activity that is pilocarpine-induced and electrically evoked, but not picrotoxin-induced, both in vitro and in vivo. The former two models are thought to be predictive for complex partial seizures34,35,43, whereas picrotoxin is a model of more generalized seizures42. However, it has also been suggested that GLUK5 receptors are not important for the development and expression of amygdala kindling, which is also regarded as a model of complex partial seizures37. Thus, it is likely that the involvement of different classes of glutamate receptors depends critically on the experimental model and possibly the site of initiation of seizure activity.

Of potential relevance, allelic variants of the GLUK5 receptor gene are associated with susceptibility to juvenile absence epilepsy48, a common form of idiopathic generalized epilepsy that accounts for approximately 5% of all epilepsies.

Conclusion

Our results suggest that selective GLUK5 kainate receptor antagonists may provide a valuable approach for the future treatment of complex partial seizures, an intractable and debilitating form of epilepsy, which leaves many patients refractory to the current available medication. Both LY377770 and LY382884 lacked overt side effects at effective doses. Inhibiting GLUK5 receptors, which are recruited during high-frequency activity, is an effective way of targeting drugs to epileptic foci. There remains the potential to develop drugs with even more selectivity, as the post-transcriptional status and stochiometry of the relevant GLUK5 subunits is determined.

Methods

Ethical review.

Experiments were approved by local Ethics Committees on Animal Experimentation of Eli Lilly & Co., University of Bristol, University of Utah, and Vrije Universiteit Brussels and the appropriate national legislation.

Materials.

LY377770 and LY382884 were synthesized in the laboratories of Eli Lilly & Co12. Other compounds were purchased from Tocris–Cookson (Bristol, UK) or Sigma Chemical (St. Louis, Missouri).

Ligand-binding studies.

These experiments were done using cell membranes from frozen human HEK293 cells expressing either human recombinant AMPA or kainate receptors using 3H-AMPA and 3H-kainate, respectively, as previously described10,19.

Electrophysiology in isolated neurons.

Antagonist activity and selectivity of GLUK5 ligands was tested on kainate receptor–mediated responses in acutely isolated rat dorsal root ganglion neurons and on AMPA and NMDA receptor–mediated responses of cultured hippocampal neurons, as previously described19.

Electrophysiology in slices.

Experiments were done in an interface recording chamber on transverse hippocampal slices (400 μm), obtained from Wistar rats 4–6 weeks old, superfused at a flow rate of 2 ml/min with Krebs medium, comprising 124 mM NaCl, 3 mM KCl, 26 mM NaHCO3, 1.25 mM NaH2PO4, 2 mM CaCl2, 1 mM MgSO4 and 10 mM D-glucose, bubbled with O2/CO2 (95/5%) at 30 ± 1°C. Extracellular (DC) field EPSPs and population spikes were recorded from cell body layers in CA3 or CA1, using glass microelectrodes (2–4 MΩ) containing 4 M NaCl. Synaptic responses were evoked by low-frequency (0.033-Hz) stimulation using bipolar electrodes, positioned in the cell body layer of the dentate gyrus or in stratum radiatum of area CA1, respectively. Stimulus train–induced bursting was generated by delivering 10 periods of tetanic stimulation, each comprising 120 shocks at 60-Hz test intensity, with 5 min between tetani, to the mossy fiber pathway. Test compounds were added to the perfusing medium.

Microdialysis.

Intracranial guides with cannula (CMA Microdialysis, Stockholm, Germany) were stereotaxically implanted in hippocampus of male Wistar rats (270–300 g) anesthetized with ketamine:diazepam (25:5 mg/kg). Coordinates relative to bregma were as follows: L, +4.6; A, −5.6 and V, +4.649. Rats received ketoprofen (4 mg/kg i.p.) to provide post-operative analgesia and were placed in microdialysis cages to recover overnight, with food and water ad libitum. The microdialysis probes (3 mm membrane; recovery, 10–15%) were perfused (2 μl/min) overnight with Ringer's solution, containing 147 mM NaCl, 2.3 mM CaCl2 and 4 mM KCl. The next day, dialysates were collected every 20 min from the freely moving rats. Drugs were dissolved in physiological saline when administered intraperitoneally or in Ringer's solution when applied via the microdialysis probe.

Electrocorticography.

Simultaneous microdialysis–electrocortico-graphy was done in at least two rats of each experimental group. A parasagittal groove above each hemisphere was drilled in the skull of the anesthetized rat. An array of six cortical electrodes, containing a prefrontal reference electrode, a ground electrode and four parietal monitoring electrodes, was then positioned on the dura mater, as described previously35. Monopolar generalized cortical seizure activity was recorded toward the reference electrode (time constant, 0.15 s; high cut-off filter, 70 Hz; sensitivity, 500 mV/cm) and was amplified polygraphically.

Microbore liquid chromatography.

Chromatographic conditions and precolumn derivatization procedures for analysis of GABA and glutamate in microdialysates have been described previously50. The mean basal levels (n = 38) of GABA and glutamate were 35 ± 3 nM and 379 ± 37 nM, respectively. Data were expressed as percent changes from basal.

Six-hertz seizure model.

Low-frequency (6 Hz) electrical stimulation (32 mA) was delivered via corneal electrodes to CF#1 mice for 3 s. This results in convulsive responses in 97% of unprotected animals. LY377770 was injected intraperitoneally 30 min earlier (n = 8 per group).

Statistical analysis.

Glutamate and GABA dialysate concentrations, uncorrected for probe recovery, were expressed as mean ± s.e.m. Significance between baseline and post-treatment amino acid levels within the same experimental group was determined by a two-tailed Wilcoxon test for paired measures (α = 0.05).