Dravet syndrome (DS) is caused by dominant mutations of the SCN1A gene, encoding the NaV1.1 sodium channel α subunit. Gene targeted mouse models of DS mutations replicate patients' phenotype and show reduced γ-aminobutyric acid (GABA)ergic inhibition. However, little is known on the properties of network hyperexcitability and on properties of seizure generation in these models. In fact, seizures have been studied thus far with surface electroencephalography (EEG), which did not show if specific brain regions are particularly involved. We have investigated hyperexcitability and epileptiform activities generated in neuronal networks of a mouse model of DS.
We have studied heterozygous NaV1.1 knock-out mice performing field potential recordings in combined hippocampal/cortical slices in vitro and video/depth electrode intracerebral recordings in vivo during hyperthermia-induced seizures.
In slices, we have disclosed specific signs of hyperexcitability of hippocampal circuits in both the pre-epileptic and epileptic periods, and a specific epileptiform activity was generated in the hippocampus upon application of the convulsant 4-aminopyridine in the epileptic period. During in vivo hyperthermia-induced seizures, we have observed selective hippocampal activity in early preictal phases and pronounced hippocampal activity in the ictal phase.
We have identified specific epileptiform activities and signs of network hyperexcitability, and disclosed the important role of the hippocampus in seizure generation in this model. These activities may be potentially used as targets for screenings of antiepileptic approaches.
Dravet syndrome (DS), also known as severe myoclonic epilepsy of infancy (SMEI), is a devastating drug-resistant epileptic encephalopathy, characterized by febrile/hyperthermia seizures at onset and later development of afebrile seizures, cognitive impairment, elevated mortality, and ataxia (Dravet et al., 2005). About 80% of patients with DS carry heterozygous missense or truncating mutations in the SCN1A gene encoding the NaV1.1 sodium channel α subunit (Claes et al., 2001; Marini et al., 2011), which cause haploinsufficiency: a 50% reduction of functional channels without negative dominance (Bechi et al., 2012). Notably, NaV1.1 is the target of hundreds of epileptogenic mutations causing different types of epilepsy (Catterall et al., 2010; Mantegazza et al., 2010; Marini & Mantegazza, 2010).
Heterozygous NaV1.1 knockout mice (NaV1.1+/−), in which deletion of exon 26 in Scn1a leads to haploinsufficiency of NaV1.1, recapitulate the phenotype of DS: they exhibit hyperthermia-induced and spontaneous seizures, premature death, mild ataxia, and cognitive impairment (Yu et al., 2006; Kalume et al., 2007; Oakley et al., 2009; Han et al., 2012). Although NaV1.1 starts to be expressed in rodents at about postnatal day 10 (P10; Beckh et al., 1989; Ogiwara et al., 2007), NaV1.1+/− mice show hyperthermia-induced seizures after P19 and spontaneous seizures after around P30 (Oakley et al., 2009). A knock-in mouse model of the R1407X NaV1.1 truncating mutation shows a similar phenotype (Ogiwara et al., 2007; Ito et al., 2012). These features are consistent with the patients' phenotype (Dravet et al., 2005) and set the beginning of the epileptic period in NaV1.1+/− mice at about P19, identifying a preepileptic period in which the underlying cause of epilepsy (NaV1.1 mutation) is present but there are no seizures. Electrophysiologic studies in dissociated hippocampal neurons from DS mice have shown selective loss of sodium current and excitability in γ-aminobutyric acid (GABA)ergic interneurons (Yu et al., 2006), and some immunohistochemistry studies have shown selective expression of NaV1.1 in GABAergic neurons (Ogiwara et al., 2007). These results point to selective impairment of GABAergic interneurons in DS mice, a hypothesis that has been confirmed by recent studies (Cheah et al., 2012; Dutton et al., 2012). Recently, it has also been shown that GABAergic inhibition is actually reduced in brain slices from DS mice (Han et al., 2012).
However, little is known on the properties of network hyperexcitability and on the mechanism of seizure generation in these models. In fact, properties of local circuits have not been investigated and seizures have been studied thus far with surface electroencephalography (EEG) (Yu et al., 2006; Ogiwara et al., 2007; Oakley et al., 2009), which did not show if specific brain regions are particularly involved. Notably, some studies have revealed, in comparison with the hippocampus, lesser impairments of NaV1.1 haploinsufficiency on GABAergic neuron functions in the cerebellum, reticular nucleus of the thalamus, and cerebral cortex (Yu et al., 2006; Kalume et al., 2007; Abe et al., 2010; Kalume et al., 2007), although there are limited studies done in brain slices. Shedding light on these issues is important for identifying specific network activities and brain regions that may be particularly implicated, which may be used as targets in drug screenings for the development of therapeutic approaches.
We have performed electrophysiologic recordings in brain slices from NaV1.1+/− mice in the period preceding the appearance of behavioral seizures (preepileptic period: P14–P18) and in the epileptic period (P30–P36), and intracerebral recordings with depth electrodes, identifying specific epileptic activities and network hyperexcitability generated in the hippocampus.
Experimental protocols were approved by local ethic committees. Mice were genotyped as in Yu et al. (2006); NaV1.1+/− mice and their wild-type (NaV1.1WT) littermates were selected for the experiments and compared. Brain slices were prepared as in Mantegazza et al. (1998) and Aracri et al. (2006). Depth electrodes were implanted in 3-month-old mice under isoflurane anesthesia; video EEG was recorded wideband (0.1–2.0 kHz) and sampled at 2 kHz/channel. See supporting information for details.
Paired-pulse stimulation discloses network hyperexcitability in NaV1.1+/− hippocampal slices
Because, as highlighted above, some studies have revealed greater impairments of NaV1.1 haploinsufficiency in GABAergic neurons of the hippocampus in comparison with the cerebral cortex and other brain areas (Yu et al., 2006; Kalume et al., 2007; Abe et al., 2010; Kalume et al., 2007), we have initially investigated the alterations of hippocampal network hyperexcitability in brain slices. Since nonstimulated slices did not show spontaneous network activity, we evaluated by field potential recordings the modifications in the response of the overall neuronal circuit in the CA1 area of the hippocampus stimulated using a paired pulse (PP) paradigm. Pairs of stimuli were applied to the Schaffer collaterals in the stratum radiatum, and population spikes were recorded from the stratum pyramidale in the CA1 area. We applied PP stimulations with various interpulse intervals ranging from 5 to 500 msec and evaluated the depression/facilitation observed in the amplitude of the second population spike in the pair (PS2) in comparison with that of the first one (PS1), calculating the paired pulse ratio (PPR = PS2/PS1), which depends both on local GABAergic and glutamatergic synaptic transmission and on the intrinsic excitability of the neurons forming the circuit (Waldbaum & Dudek, 2009).
In slices from P14–P18 NaV1.1WT mice (n = 16), we observed a clear PP depression with interpulse intervals between 5 and 6.4 msec, obtaining PPR ranging from 0.33 ± 0.06 to 0.92 ± 0.13 ms respectively (Fig. 1Aa,B). Notably, in slices from NaV1.1+/− mice (n = 13), we observed a lower depression of the conditioned response at interpulse intervals of 5 and 5.2 msec, and facilitation at interpulse intervals between 5.8 and 6.4 msec (p < 0.05 for all the interpulse intervals, Mann-Whitney U test; Fig. 1Ab,B). These modifications are consistent with hyperexcitability of the hippocampal network in NaV1.1+/− mice in the pre-epileptic period.
Using PP stimulations with interpulse intervals between 50 and 500 msec, we observed PP facilitation in both NaV1.1WT and NaV1.1+/− slices. In fact, PPR varied between 1.44 ± 0.05 (with interpulse interval of 50 msec) and 1.18 ± 0.03 (with 500 msec) in NaV1.1WT slices, and between 1.50 ± 0.04 and 1.22 ± 0.04 in NaV1.1+/− slices; facilitation was not significantly different at any of the interpulse intervals used (Fig. 1C). To find out if in the epileptic period the hyperexcitability of hippocampal circuits is more prominent, we used slices obtained from P30–P36 mice: at this age, NaV1.1+/− mice show spontaneous seizures and high mortality rate (Yu et al., 2006). As in P14–P18 slices, we observed PP facilitation with interpulse intervals between 50 and 500 msec (Fig. 1D): PPR varied between 1.45 ± 0.05 (with interpulse interval of 50 msec) and 1.19 ± 0.03 (with 500 msec) in NaV1.1WT slices, and between 1.36 ± 0.04 and 1.17 ± 0.02 in NaV1.1+/− slices. However, as in P14–P18 slices, facilitation was not significantly different at any of the interpulse intervals. Of interest, with interpulse intervals that induced PP facilitation, the second stimulus in the pair frequently elicited multiple (2–3) population spikes in NaV1.1+/− slices in both the age periods, whereas they were observed much more rarely in NaV1.1WT slices (Fig. 1E). In particular, using an interpulse interval of 150 msec in P14–P18 slices, we observed multiple spikes in 58% (7 of 11) of NaV1.1+/− slices, but only in 26% (4 of 19) of NaV1.1WT slices (p < 0.05, Fisher's exact test). Comparing the mean number of spikes elicited by the second stimulus in the pair, we found a trend toward an increase but no significant difference (1.32 ± 0.13 spikes in NaV1.1WT; 1.67 ± 0.19 in NaV1.1+/−; p = 0.09 Mann-Whitney U test; Fig. 1F). These experiments showed a further sign of hyperexcitability in NaV1.1+/− slices in the preepileptic period. Furthermore, multiple spikes were observed in 23% (3 of 11) of NaV1.1WT slices and in most of NaV1.1+/− slices (75%, 9/12; p < 0.05; Fisher's exact test) at P30–P36 and, differently than in slices from P14–P18 NaV1.1+/− mice; also the mean number of spikes was significantly increased (1.27 ± 0.14 in NaV1.1WT, 2.00 ± 0.23 in NaV1.1+/−; p < 0.05, Mann-Whitney U test; Fig. 1F), suggesting that the hyperexcitability of the circuit is more pronounced in the epileptic period and that hippocampal slices reproduce the phenotype of the mice.
Posttetanic afterdischarges are potentiated in NaV1.1+/− slices, particularly in the epileptic period
We better characterized the response to electrical stimuli of the NaV1.1+/− hippocampal CA1 network using tetanic stimulations applied to the Schaffer collaterals and recording field potentials in the pyramidal layer (Fig. S1). Tetanic stimulations can induce afterdischarges in the γ frequency range, which can evolve in epileptiform activities (Whittington et al., 1997); we used 10 stimuli at 100 Hz with amplitude that induced the maximal population spike amplitude. In the preepileptic period (P14–P18), we observed afterdischarges in NaV1.1WT slices (n = 17) formed by 18.2 ± 1.7 spikes with an average frequency of 35.8 ± 3.7 Hz (Fig. S1A left, B). In NaV1.1+/− slices (n = 18); the afterdischarges were formed on average by more spikes than in NaV1.1WT slices (28.0 ± 3.3; p < 0.05, Mann-Whitney U test) with frequency that was not different (40.8 ± 3.6 Hz; Fig. S1A right, B). In the epileptic period (P30–36), afterdischarges observed in both NaV1.1WT (n = 12) and NaV1.1+/− (n = 23) slices were formed by fewer spikes than at P14–18 (5.2 ± 1.6 spikes for NaV1.1WT, 14.3 ± 2.3 for NaV1.1+/−, p < 0.05, Mann-Whitney U test), but the number of spikes was again significantly higher in NaV1.1+/−, slices (p < 0.05, Mann-Whitney U test; Fig. S1B), and, notably, it showed a larger increase in the epileptic period (2.7-fold) than in the preepileptic period (1.5-fold; Fig. S1C). These results confirm that NaV1.1+/− slices are hyperexcitable, and are consistent with an increase in hyperexcitability during the transition from the preepileptic to the epileptic period.
NaV1.1+/− slices in the epileptic period show specific 4-aminopyridine (4-AP)–induced recurrent multicomponent epileptiform discharges that can be independently generated by the hippocampus
The experiments presented above disclosed hippocampal network hyperexcitability, but epileptiform activities were not observed. Because NaV1.1+/− mice show hyperthermia-induced seizures, we attempted to induce epileptiform activities in slices from P30–P36 mice by increasing the temperature of the recording chamber to febrile levels. However, with our experimental system, the only effect that we observed was induction of spreading depression (SD; Fig. S2; Dreier, 2011). In fact, neither NaV1.1WT (n = 7) nor NaV1.1+/− (n = 10) slices showed spontaneous activity at up to 40°C. With temperature higher than 40°C, we observed SD in both NaV1.1WT and NaV1.1+/− slices (in three NaV1.1WT and seven NaV1.1+/− slices; proportion that is not significantly different, Fisher exact test). Figure S2 shows SD events recorded in hippocampal CA1 area (stratum radiatum), which showed similar features in NaV1.1WT and NaV1.1+/− slices. Notably, afterdischarges were more pronounced and showed higher frequency in NaV1.1+/− (22 ± 6 spikes with mean instantaneous frequency of 1.5 ± 0.4 Hz in NaV1.1WT; 35 ± 5 spikes and 2.6 ± 0.3 Hz in NaV1.1+/−).
However, because SD is not considered an epileptiform activity, we stimulated slices applying the convulsant 4-AP: a K+ channel blocker that increases both excitatory and inhibitory neurotransmitter release and that has been used extensively to induce experimental epileptiform discharges in vitro (Avoli & de Curtis, 2011). In particular, in combined entorhinal cortex (EC)–hippocampal horizontal slices, in which connections between ventral hippocampus and EC are preserved, 4-AP induces both short (about 100 msec in duration) interictal-like discharges and longer ictal-like activities (seizure-like discharges that last tens of seconds).
We simultaneously recorded in combined EC-hippocampal slices local field potentials in the CA1 area of the hippocampus (stratum radiatum or pyramidale) and in the EC (layer V) between 1 and 1.5 h after the beginning of 4-AP continuous perfusion. When slices from mice in the age range P30–P36 (epileptic period) were used, we observed in NaV1.1WT slices (n = 10) the typical 4-AP-induced epileptiform discharges that have been already described in WT rodents (Fig. 2A; Table S1). Of interest, in slices from NaV1.1+/− mice (n = 23) a further type of epileptiform discharge, which we named recurrent multicomponent discharges (RMDs), was evident in 45% of the slices recorded (Fig. 2B). These discharges had mean duration of 1.4 ± 0.1 s and interval of occurrence of 5.2 ± 0.5 s. In the EC (Fig. 2B, upper panel), they were formed by an initial negative deflection of the field potential with superimposed fast field potential transients. In the CA1 area of the hippocampus, they showed negative deflection in the stratum radiatum (Fig. 2B lower panel) and positive deflection in the stratum pyramidale (not shown), with superimposed fast field potential transients that were less evident in comparison with the EC. Notably, in slices in which RMDs were present, we never observed typical 4-AP–induced ictal-like activities generated in the EC. We also observed interictal activities that were similar to those recorded in NaV1.1WT slices, which often overlapped with RMDs (Fig. 2B). Examination of the traces at a larger time scale and quantification of the delay of onset of RMDs between CA1 and EC (Fig. 2B, right) showed that they were observed first in the hippocampus and then in the EC (delay of onset was 37 ± 2 msec).
Of interest, in most of the recorded slices we sometimes observed RMDs that were substantially more prolonged than the average, as those indicated by the leftmost asterisks in the traces of Fig. 2B (left). Moreover, in four NaV1.1+/− slices (from three different mice), we observed even longer RMDs that lasted on average 19.5 ± 3.5 s (Fig. 2C); thus, they could be interpreted as infrequent ictal-like activities specific of NaV1.1+/− slices. However, their features are different in comparison with the typical 4-AP ictal-like activities (Fig. 2D): they were shorter and characterized by a sharper negative voltage deflection with superimposed fast field potential transients of homogeneous frequency (14.3 ± 0.6 Hz), whereas classic 4-AP ictal-like activities show initial high frequency fast field potentials (tonic-like phase) followed by lower frequency discharges (clonic-like phase; Avoli & de Curtis, 2011). As for the shorter RMDs, their onset was recorded first in the hippocampus proper and then in the EC, although this does not demonstrate that they are generated in the hippocampus.
To find out if NaV1.1+/− RMDs can be independently generated by different regions of the hippocampus proper or whether they need preserved connections between ventral hippocampus and EC, we recorded 4-AP–induced epileptiform activities in coronal slices comprising the dorsal hippocampus and the somatosensory or visual neocortex, in which there are no connections between hippocampus and cortex. In experiments in which we simultaneously recorded field potentials from the CA1 area of the dorsal hippocampus and the layer V of the neocortex, we observed epileptiform activities that were similar to those that we recorded in combined EC-hippocampal slices. In fact, recordings from NaV1.1WT slices (n = 14) showed typical interictal-like activities in the hippocampus and ictal-like activities in the neocortex (Fig. 3A–C; Table S1). Notably, in 47% of NaV1.1+/− slices (n = 15), in addition to these activities, we observed RMDs that were similar to those recorded in horizontal slices (Fig. 3D–F; Table S1). Their interval of occurrence was longer than in horizontal slices, but the mean duration and frequency of superimposed fast field potential transients were similar to that observed in horizontal slices (Table S1). We never observed in these slices RMDs lasting tens of seconds, but discharges of 2–4 s were often present. Unlike the EC in horizontal slices, we never observed RMDs in the neocortex. Moreover, different from horizontal slices, neocortical classic 4-AP ictal-like activities were always present in slices showing hippocampal RMDs, and their properties were not modified comparing with NaV1.1WT slices or with NaV1.1+/− slices in which hippocampal RMDs were not observed (Table S1), consistent with the lack of hippocampal-cortical connections in coronal slices. Typical 4-AP–induced interictal activities were often observed in the hippocampus of NaV1.1+/− slices, but their frequency was lower when RMDs were present (Table S1). These results confirm that RMDs are specifically generated in the hippocampus proper and can be generated also by the dorsal hippocampus.
We then studied the effect of 4-AP in NaV1.1WT and NaV1.1+/− slices obtained from mice in the preepileptic period (P14–P18, not shown). Both in NaV1.1WT (n = 17) and NaV1.1+/− (n = 12) slices, application of 4-AP induced typical epileptiform activities (short interictal-like and longer ictal-like discharges), which had similar duration and frequency in NaV1.1WT and NaV1.1+/− slices (Table S1). Notably, in this age range we never observed RMDs, which evidently are generated only in the epileptic period.
Block of glutamate receptors inhibits the generation of multicomponent epileptiform discharges and discloses GABAergic discharges with reduced frequency in NaV1.1+/− slices
To better characterize the specific 4-AP–induced epileptiform activity observed in slices from NaV1.1+/− mice, we applied specific blockers of glutamate and GABA receptors. Figure S3A shows field potential recordings form the CA1 area in a coronal slice from a NaV1.1+/− mouse in the epileptic period after 1 h application of 4-AP, in which we observed interictal discharges and RMDs (Fig. S3A, right panel). Application of 3 mM kynurenic acid, an unspecific inhibitor of glutamate receptors, blocked the generation of RMDs (Fig. S2B). Therefore, glutamatergic transmission is necessary for their generation.
In the presence of kynurenic acid, we observed recurrent discharges that were blocked by subsequent application of the GABA-A receptor inhibitor bicuculline (5 μM), applied with kynurenic acid (Fig. S3C). Therefore, these discharges are GABA-A–mediated field potentials, which have been already observed and characterized in 4-AP–treated slices; notably, it has been proposed that they are involved in the generation of the standard ictal activities induced by 4-AP (Barbarosie et al., 2002). Of interest, the frequency of these GABA-A–mediated discharges was lower in slices from NaV1.1+/− mice (interevent interval was 14.3 ± 1.0 s in NaV1.1WT slices, n = 16, and 18.1 ± 1.1 s in NaV1.1+/− slices, n = 15; p < 0.05 Mann-Whitney U test), whereas the average duration showed just a trend toward reduction (491 ± 65 msec in NaV1.1WT slices, 346 ± 49 msec in NaV1.1+/− slices), consistently with reduced GABAergic activity (Fig. S3D).
In slices from P14–P18 mice (preepileptic period; not shown), upon application of kynurenic acid, we observed recurrent activity that was blocked by subsequent application of bicuculline, similar to results in P30–P36 slices. However, unlike in P30–P36 slices, the duration and the frequency of this activity was not different in comparison with NaV1.1WT slices, although there was a trend toward the reduction of both duration and frequency (duration: 643 ± 74 msec in NaV1.1WT, n = 12 slices, 497 ± 66 msec in NaV1.1+/−, n = 11 slices; inter-event interval: 18.0 ± 1.7 s in NaV1.1WT, 21.7 ± 1.4 s in NaV1.1+/−).
Overall, these experiments show that in brain slices the hippocampus proper is particularly important for the generation of RMDs, a specific epileptiform activity of NaV1.1+/− slices generated by 4-AP, and that a GABAergic activity considered important for the generation of the standard 4-AP ictal activity is reduced in this model in the epileptic period.
In vivo depth electrode recordings during hyperthermia-induced seizures
To find out if also in vivo the hippocampus is particularly implicated in the generation of epileptic activity in comparison with the neocortex, we carried out in nonanesthetized freely moving mice synchronized video-EEG recordings with electrodes implanted into the CA1 region of both dorsal hippocampi and bilaterally in the somatosensory cortex, and a reference electrode placed in the skull above the cerebellum. In preliminary video-only recordings (2 weeks of observation of five NaV1.1+/− and two NaV1.1WT nonimplanted mice at 3 months of age), we observed no behavioral seizures in NaV1.1WT mice and spontaneous behavioral tonic–clonic seizures in the NaV1.1+/− mice, but seizure frequency was low in our mice (average frequency of 1 per 168 ± 10 h). However, hyperthermia consistently induced seizures in NaV1.1+/− mice older than P19, as reported (Oakley et al., 2009). Therefore, we implanted electrodes in four NaV1.1+/− mice and three NaV1.1WT mice, and performed video-EEG recordings during 13 hyperthermia induced seizures. Notably, hyperthermia-induced seizures are an important clinical feature of DS: the first seizures in patients are induced by fever or hyperthermia, which remain triggering factors also at later stages (Dravet et al., 2005). Moreover, it has been reported that the features of spontaneous generalized seizures in NaV1.1+/− mice are similar to those of hyperthermia-induced ones (Oakley et al., 2009). Before inducing hyperthermia, we recorded baseline activity at normal core body temperature (37.3 ± 0.3°C n = 13) for at least 30 min, allowing the animal to relax and explore the environment. All the NaV1.1+/− mice showed fully developed tonic–clonic seizures, which were induced on average by a core body temperature of 41.1 ± 0.5°C (n = 13). Neither seizures nor epileptiform activity were observed in NaV1.1WT mice up to a core body temperature of 42.5°C (n = 6). Figure 4 displays representative recordings obtained during a hyperthermia-induced seizure in a 14-week-old NaV1.1+/− mouse at core body temperature of 41.2°C, showing that both the hippocampus and the neocortex are involved in the observed epileptic activity (video-EEG available on demand). The analysis of the traces with an expanded time scale (bottom panels) shows that early electrographic activity developed first in the hippocampus (Fig. 4b). This was observed consistently in all seizures and mice, with a mean delay of 80 ± 21 s (n = 13) between cortical and hippocampal activity. This early activity (it began on average 167 ± 50 s before the tonic–clonic seizure, range 61–360 s) was not associated with modifications in behavior and thus can be considered a preictal activity. Later activity (Fig. 4c) was characterized by coincident spikes in both the hippocampi and the neocortex (asterisks; cortical spikes were observed on average 19 ± 14 msec after hippocampal spikes) and sequences of sharp waves localized in the hippocampi (arrows). During this phase, mice showed episodes of tail erection and irregular brief jerks of the limbs, resembling a mild myoclonic seizure. Coincident fast activity and spiking was observed in the hippocampi and cortex, with larger amplitude in the hippocampi, at the onset of tonic–clonic seizures (Fig. 4d), which consisted initially in rearing with symmetrical forelimb clonus, jumps, and then a fully developed tonic–clonic seizure. Late seizure activity was characterized by diffuse burst discharges in both the structures (Fig. 4e), which behaviorally corresponded to long-lasting clonus of all limbs.
Continuous 2-week video-EEG recordings were performed during the period of seizure induction with hyperthermia, but we did not observe clear spontaneous seizures (noninduced by hyperthermia) in the implanted mice, although we observed interictal activity with no clear behavioral correlate. More recordings in different age periods should be done for studying spontaneous seizures.
These data show that electrographic preictal activity develops earlier in the hippocampus than in the somatosensory cortex during hyperthermia-induced seizures and that the hippocampus is strongly implicated in the generation of both preictal and ictal activity.
We have identified signs of network hyperexcitability, epileptiform activities, and modifications of network properties that characterize neuronal circuits in a mouse model of DS, and shown that the hippocampus is particularly involved in seizure generation.
With our experimental system in vitro, we did not observe spontaneous epileptiform activities in control conditions and hyperthermia-induced spreading depression both in NaV1.1+/− and NaV1.1WT slices. Therefore, to generate epileptiform discharges and evaluate their properties in NaV1.1+/− slices, we used the convulsant 4-AP, a common inducer of epileptiform activity in vitro (Avoli & de Curtis, 2011). Application of 4AP was able to induce epileptiform activity both in NaV1.1WT and NaV1.1+/−. In NaV1.1WT, we observed the standard activity reported in rodent slices upon application of 4-AP, with ictal-like activity and interictal like-activities observed both in combined entorhinal cortex-hippocampus horizontal slices and in coronal slices comprising the somatosensory/visual cortex and the dorsal hippocampus. In slices from NaV1.1+/− mice, we observed a specific epileptiform activity that we named recurrent multicomponent discharges (RMDs) and that was present only during the epileptic period (P30–36). When RMDs were present, we did not observe standard 4-AP ictal activities generated by the entorhinal cortex in horizontal slices (in which connections between the hippocampus and the entorhinal cortex are preserved). Notably, in coronal slices (in which there are no connections between the hippocampus and the cortex) RMDs were generated in the hippocampus, whereas the somatosensory cortex generated standard 4-AP ictal-like activities, suggesting that the hippocampus proper is sufficient for their generation, that hippocampocortical connections are necessary for inhibiting the generation of standard 4-AP ictal-like activities, and that the 4-AP–induced activity in the somatosensory/visual cortex is not modified by NaV1.1 haploinsufficiency. We observed long-lasting (up to tens of seconds) RMDs only in horizontal slices, suggesting that entorhinal cortex and parahippocampal structures are important for sustaining these discharges or that the ventral hippocampus is more prone to the their generation.
RMDs are inhibited by the block of glutamatergic receptors, showing that excitatory neurotransmission is essential for their generation, similarly to preictal activities observed in human hippocampus resected from temporal lobe epilepsy patients (Huberfeld et al., 2011). In these conditions, we observed GABAergic field potentials that were similar to those that have been proposed to be important for inducing standard 4-AP ictal-like discharges in the entorhinal cortex (Barbarosie et al., 2002; Avoli & de Curtis, 2011). Notably, we found that the frequency of these GABAergic potentials is reduced in NaV1.1+/− slices from the epileptic period, whereas in the preepileptic period we found just a trend toward a reduction. This is further evidence that GABAergic activity is reduced in NaV1.1+/− neuronal networks (Yu et al., 2006; Ogiwara et al., 2007; Han et al., 2012), but with a more prominent reduction in the epileptic period, and that standard 4-AP ictal-like discharges may not be a specific ictal-like activity of NaV1.1+/− slices.
However, are RMDs specific epileptic activities of NaV1.1+/− slices? It has been reported that it is difficult to induce typical 4-AP ictal-like activities in slices of epileptic tissue obtained from human specimens or animal models of epilepsy, and this has been interpreted as a homeostatic modification of the network (Zahn et al., 2008, 2012). This property is shared by the NaV1.1+/− horizontal slices in which we have observed RMDs. However, RMDs in our experiments have been observed only in NaV1.1+/− slices, they are generated only in the epileptic period, they can be in some cases long enough to be considered ictal-like activities, and standard 4AP ictal-like activities can be generated in the neocortex in slices that generate hippocampal RMDs. Moreover, they are similar to the recurrent epileptiform activity observed in the entorhinal cortex during combined application of 4-AP and bicuculline, and to the late recurrent discharges observed after prolonged exposure to low Mg2+ artificial cerebrospinal fluid (ACSF) (Pfeiffer et al., 1996; Bruckner et al., 1999). Of interest, these activities are generated in conditions of reduced GABAergic transmission, similarly to RMDs in NaV1.1+/− slices, and are resistant to classic anticonvulsants. Conversely, typical 4-AP ictal-like discharges have been proposed to be generated by GABAergic activities (which are reduced in NaV1.1+/− slices) and can be blocked by several classic anticonvulsants (Bruckner et al., 1999; Barbarosie et al., 2002). Therefore, we propose that RMDs can be considered a specific epileptiform activity of NaV1.1+/− slices, which highlights the important role of the hippocampus in this mouse model of DS.
Depth electrode recordings during hyperthermia-induced seizures in vivo confirmed that the hippocampus is involved in the generation of ictal activity, and showed prominent hippocampal activity during early preictal phases in comparison with the somatosensory cortex. Although the accurate determination of brain areas particularly involved in seizure generation would require recordings with more electrodes, our results are consistent with the in vitro data.
Moderate electric stimulations of slices could not induce epileptiform activity. However, they disclosed specific signs of network hyperexcitability in NaV1.1+/− hippocampal slices. In particular, in the epileptic period we observed an increased number of population spikes in posttetanic afterdischarges and in response to the second stimulus of paired-pulse stimulations. Of interest, these signs of hyperexcitability were reduced in slices from the preepileptic period (P14–P18), when NaV1.1+/− mice do not show behavioral seizures (Yu et al., 2006; Ogiwara et al., 2007; Oakley et al., 2009). This evidence, together with the lack of RMDs and of modifications of 4-AP–induced GABAergic activities, suggests that reduction of GABAergic inhibition is larger in the epileptic period than in the preepileptic one, although the mutation of NaV1.1 (the underlying cause of epilepsy in DS) is present also in the preepileptic period. However, NaV1.1 begins to be expressed at around P10 and shows an increase in expression level between the second and third postnatal weeks in rodents (Beckh et al., 1989; Ogiwara et al., 2007). This can be the main reason for a progressive reduction of GABAergic inhibition in NaV1.1+/− mice during development, which can lead to neuronal network hyperexcitability in the preepileptic period (between P10 and P18) and to ictal activity in the epileptic period. However, other modifications in network functions, physiologic-developmental ones, or induced by the initial network hyperexcitability, may contribute in setting the transition from the two periods.
Notably, the specific activities that we have observed in NaV1.1+/− slices may be used as in vitro targets in preclinical screening of drugs.
We thank Alessandro Cattalini for skillful technical experience and Maia Chikhladze for help in setting up preliminary experiments. This study was supported by the LabEx ICST (MM), Mariani Foundation R-12-94 (MM and SF), Italian Ministry of Health GR-2007-657156 (MM), European Integrated Project EPICURE EFP6-037315 (MM and SF), the Fondation pour la Recherche Medicale (MM), ERANET NEURON “2P-imaging” (MdC), and Italian Telethon GGP10138A (MM and SF). CL has been the recipient of a “Fin de Thèse” fellowship from the Fondation pour la Recherche Medicale.
None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal's position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.