FIGURE 3 A schematic representation of the thalamocortical interactions. NRT, nucleus reticularis of the thalamus. (Courtesy of Pierre-Olivier Polack.)
FIGURE 3 A schematic representation of the thalamocortical interactions. NRT, nucleus reticularis of the thalamus. (Courtesy of Pierre-Olivier Polack.)
1987, 1990) or in the WAG/Rij model (Inoue et al., 1993; Kandel et al., 1996).
The thalamus is the major source of subcortical inputs to the cerebral cortex. Most of the thalamocortical axons terminate in cortical layers IV and III (Alloway et al., 1993; Jones, 1985; Lu and Lin, 1993). Reciprocally, pyramidal cells from the cortical layers V and VI provide an important innervation to the thalamus (Boussara et al., 1995; Burkhaler and Charles, 1990; Chmielowska et al., 1989; Jones, 1985), and thalamocortical and corticothalamic connections are mainly glutamatergic (Deschênes and Hu, 1990; Fonnum et al., 1981; Kanedo and Mizumo, 1988; Kharazia and Weinberg, 1994; Ottersen et al., 1983; Zilles et al., 1990). In addition, all thalamic relay nuclei receive a massive GABAergic inhibitory projection from the reticular thalamic nucleus, the primary source of GABA in the rat thalamus (Cox et al.,
1996; De Biasi et al., 1988; Jones, 1985; Pinault et al., 1995). Reticular thalamic neurons do not project directly to the cerebral cortex; rather, they receive excitatory collaterals from thalamocortical and corticothalamic neurons (Bourassa et al., 1995; Contreras et al., 1993; de Curtis et al., 1989; Frassoni et al., 1984; Harris, 1987; Jones, 1985; Luebke, 1993; Spreafico et al., 1991; Warren et al., 1994) (Figure 3) (see also Chapter 7).
Electrophysiologic in vitro and in vivo studies have shown that thalamocortical neurons spontaneously exhibit two different modes of intrinsic activity during various states of the wakefulness-sleep cycle (Llinas, 1988; McCormick, 1992; McCormick and Bal, 1997; Steriade, 1993; Steriade and Deschênes, 1984; Steriade et al., 1993). Periods of wakefulness and attentiveness are associated with a tonic mode of discharge of fast, sodium- and potassium-mediated action potentials, allowing transmission of information to the cortex (Llinas, 1988; McCormick, 1992). On the contrary, during stages of drowsiness and slow-wave sleep, thalamocortical neurons generate bursts of spikes triggered by rhythmic low-threshold calcium spikes (Leresche et al., 1991; Llinas, 1988; McCormick, 1992; McCormick and Pape, 1990; Steriade, 1993; Steriade and Deschênes, 1984; Steriade et al., 1993). For more information about the organization and functions of the reticular nucleus, see the review by Pinault (2004).
Neurons of the thalamocortical and reticular nuclei have the capability of generating spindles, a typical oscillatory activity (Crunelli and Leresche, 1991; Gloor et al., 1990; McCormick, 1992; McCormick and Bal, 1997; Steriade, 1990; Steriade et al., 1993). Spindles consist of rhythmic oscillations of 7 to 14c/s that occur during the initial phase of slow-wave sleep (Steriade, 1993; Steriade and Deschênes, 1984). In vivo and in vitro recordings have demonstrated that these oscillations are generated by GABAergic inhibitory neurons of the reticular thalamic nucleus, imposing synchronous inhibitory postsynaptic potentials (ipsps) in a rhythmic manner to thalamocortical neurons (Bal and McCormick, 1993; Destexhe et al., 1996; Steriade, 1993; Steriade and Deschênes, 1984; Steriade et al., 1993; Von Krosigk et al., 1993). Recent in vivo studies indicate that corticothalamic glutamatergic projections to the reticular and relay thalamic neurons strongly reinforce the synchronization of spindles throughout the thalamocortical circuit (Contreras and Steriade, 1996; Contreras et al., 1996).
Activity of Thalamocortical and Reticular Neurons During SWDs and Their Relationship with Spindles
In GAERS models, in vivo extracellular and intracellular recordings showed that thalamocortical neurons fire during the spike component of the SWD (Pinault et al., 1998). Similarly, extracellular recordings showed that thalamic neurons located in the ventrolateral and ventroposterior nuclei fire in phase with the spike component of the SWDs in WAG/Rij rats (Inoue et al., 1993). In these neurons, the phasic discharges are generated by a rhythmic excitatory postsynaptic potential/GABAA inhibitory postsynaptic potential sequence superimposed on a tonic hyperpolarization (Pinault et al., 1998). The GABAA-mediated inhibition appears to result from reticular inputs. Similarly, a large-amplitude hyperpo-larization was recorded in the reticular neurons at the start of a SWD and a short interruption of burst firing was shown to occur during the early phase of a SWD, which is mediated by a slowly decaying depolarization (Slaght et al., 2002). During the full development of SWD, thalamocorti-cal and reticular neurons progressively discharge in synchrony about 12 ms before the spike component of the SWD complex, suggesting that both types of neurons are driven by a common input (Pinault et al., 2001).
It has been suggested that thalamic circuits underlying SWDs are the same as those generating spindles (van
Luijtelaar, 1997), and the reticular nucleus has been proposed to be a pacemaker for both kinds of oscillations (Avanzini et al., 1992, 1993, 2000; Seidenbecher et al., 1998). However, this hypothesis has never been conclusively demonstrated. In particular, SWDs do not always occur during the same state of vigilance as spindles, and this hypothesis does not explain the bilateral and widespread generalization of SWDs. Furthermore, short-lasting (<3 seconds) bursts of 5- to 9-c/s oscillations of medium voltage were recorded interictally in the GAERS but also in nonepileptic rats and are concomitant with immobility (Pinault et al., 2001). These bursts occur during quiet wake-fulness associated with desynchronized EEG and are distinct from spindles, which are observed during sleep, with a higher frequency (10 to 16c/s) and for a shorter duration. Such 5- to 9-c/s bursts often precede SWDs in GAERS, and the extracellular patterns of discharges of thalamocortical and reticular neurons in both GAERS and nonepileptic rats are very similar to what is recorded during SWDs. These observations suggest that SWDs and medium-voltage 5- to 9-c/s oscillations share similar cellular mechanisms and that SWD may evolve from 5- to 9-c/s oscillations in GAERS, probably as a result of some genetically mediated dysfunctions in the thalamocortical network (Pinault et al., 2001).
The possibility that the thalamus, and more particularly the GABAergic projections coming from the reticular thalamic nucleus, play a critical role in the generation of SWD was raised by the key role of the GABAergic receptors in the capability of the thalamic neurons to generate rhythmic oscillations (Crunelli and Leresche, 1991; McCormick, 1992; Steriade et al., 1993). The aggravating effect observed following microinjections of GABAa and GABAb agonists into relay nuclei, as well as the suppression of SWD induced by injection of GABAB antagonists into the same sites (Liu et al., 1991; 1993; Marescaux et al., 1992c), suggested that a thalamic enhancement of GABAergic neurotransmission could underlie absence seizures by increasing thalamic synchronizing mechanisms. Indeed, an increase in the extracellular concentration of GABA in thalamic relay nuclei was shown in GAERS by microdialysis (Richards et al., 1995). This increase appears to result from a lower uptake of GABA that is probably mediated by the GABA transporter-1 in the thalamus of GAERS (Sutch et al., 1999). However, no differences in the number of GABAergic neurons that are immunoreactive to GABA and to its synthesis enzyme glu-tamic acid decarboxylase (GAD) (Spreafico et al., 1993) were observed in the thalamic reticular nucleus between adult GAERS and nonepileptic control rats. Likewise autoradiographic and immunocytochemical studies have reported that the number, affinity, and expression of GABAA receptor subtypes were identical in the reticular and relay thala-mic nuclei between adult GAERS and control nonepileptic rats (Knight and Bowery, 1992; Snead et al., 1992; Spreafico et al., 1993).
Nevertheless GAERS were reported to be more sensitive to the convulsive effects of antagonists of the GABAa receptor as well as to benzodiazepine inverse agonists, whereas no differences were found for convulsions induced by glutamate analogues (Vergnes et al., 2000). In particular, the ß-carbolines (FG 7142 and DMCM) and the imidazobenzodiazepines (RO 19-4603) were several times more convulsant in GAERS than in nonepileptic rats (Vergnes et al., 2001). These convulsions mainly involve the cortex and the hippocampus and suggest that, in these structures, the number of GABAA receptors and their intrinsic binding properties are different in GAERS. However, no differences were found in the number and sensitivity of the GABAA receptors in the two strains (Vergnes et al., 2001); these findings are in agreement with previous studies. Furthermore the inhibitory response to clonazepam was not significantly modified in the nucleus reticularis in GAERS (Badiu, 2004). It is presently difficult to explain, at the cellular level, the increased sensitivity of rats with absence epilepsy to drugs reducing the GABAA mediated neurotransmission.
GABAB receptor activation mediates a late and long-lasting inhibitory postsynaptic potential, which produces the hyperpolarization necessary to elicit rhythmic low-threshold calcium currents (IT). Such de-inactivation of these calcium channels results in repeated bursts of action potentials in thalamocortical neurons and may be involved in the generation of SWDs (Crunelli and Leresche, 1991). In particular, intracellular recordings of nRT neurons suggested that the large-amplitude hyperpolarization of these neurons during SWD is mediated by GABAb (Slaght et al., 2002). Using an autoradiographic method, no significant difference in [3H]GABA binding to GABAB receptor in the thalamus was found between adult GAERS and control nonepileptic rats (Knight and Bowery, 1992). Similarly, the density and affinity of GABAB receptors, measured with GABAB agonist and antagonist radioligands [3H]-CGP 27492 and [3H]-CGP 54626, respectively, on thalamic membranes (Mathivet et al., 1994; 1996) appeared similar in adult GAERS and in control rats. However, a recent study using in situ hybridization and immunohistochemistry showed an increase of mRNA coding for the GABAB1 subunit in the somatosen-sory cortex in GAERS, whereas a decrease was observed in the ventrobasal nuclei (Princivalle et al., 2003). On the contrary, an increase in both B1 and B2 subunits was observed in all regions, suggesting the existence of an up-regulation of the GABAB receptor in GAERS. These data were obtained in adult rats and may result from reiteration of SWD. It is thus difficult to determine whether an overexpression of GABAB receptor is a key factor in SWD ontogeny. In GAERS, the amplitude of the IT calcium current was found to be increased in the reticular neurons, which could strengthen the synchronizing mechanisms in thalamocortical circuits (Tsakiridou et al., 1995). Although this finding remains controversial (Guyon et al., 1993), a significant increase in mRNA levels of alpha-1G in ventrobasal nuclei and of alpha 1H in the nRT was found using in situ hybridization (Talley et al., 2000). The increase in alpha-1H was also found in juvenile GAERS before they express SWD. This suggests that this modest but significant increase could facilitate IT current and thus participate in the occurrence of SWDs.
Other calcium-channel changes were found in WAG/Rij rats. Quantification of channel expression indicated that the development of SWDs in WAG/Rij rats is concomitant with an increased expression of the P/Q type in the nRT. These channels are mainly presynaptic, as revealed by double immunofluorescence involving the presynapse marker syntaxin (van de Bovenkamp-Janssen et al., 2004a). Dysfunction of the P/Q type of calcium channels (i.e., reshuffling of the beta subunit in this type of channel) was proposed to underlie the pathological phenotype of lethargic mice (Burgess and Noebels, 2000).
The first mapping data in GAERS showed that SWDs of maximal amplitude were recorded primarily recorded in the sensorimotor cortex (Vergnes et al., 1990). In addition, they suggested that SWDs occur simultaneously in the thalamus and cortex but that thalamus was sometimes leading, whereas the reverse never seems to occur. Using multisite electrode recordings in WAG/Rij and nonlinear association analysis of the signals, a consistent cortical focus was found within the perioral subregion of the somatosensory cortex (Meeren et al., 2002). Coupling between this region and connected thalamic sites in the ventrobasal nuclei was found to change direction during the course of the seizure. During the first 500 msec of a SWD, the cortical focus appears to lead the thalamus. This is further supported by the observation that cortical SWDs can sometimes occur without concomitant thalamic discharge, whereas the reverse was never observed (Meeren et al., 2002). Furthermore, it is consistent with the fact that rhythmic unit firing was reported to start a few cycles earlier in the cortex than in the thalamus in GAERS (Seidenbecher et al., 1998). This study also suggested a fast intracortical spread of the SWDs, with the intrahemispheric spread being slower than the interhemi-spheric one. Although the existence of the cortical perioral focus remains to be confirmed in GAERS, the intrahemi-spheric versus the interhemispheric spread could explain the discrepancies with the previous studies suggesting the leading role of the thalamic nuclei. Indeed, in the first mapping studies, the cortical electrode was never located in the perioral area and was generally placed in a more posterior region. In this region, cortical activity occurs a few cycles later according to the study of Meeren et al. (2002). These findings led this group of investigators to suggest that absence seizures have a focal origin. The primary driving source for the rhythmic activity could be the cortex (or at least a specific region of the cortex), not the thalamus. However, once the seizure is initiated, the cortex and the thalamus form an oscillatory network in which both structures drive each other (Meeren et al., 2002). The generalized and synchronous aspects of these seizures could be due to the very fast cortical spread of seizure activity.
The focal cortical origin of SWDs is in agreement with the recent finding that application of ethosuximide at the somatosensory cortex suppresses absence seizures in GAERS, whereas no or weak effects were observed when this compound was injected in the thalamus. Moreover, local injections at other cortical areas were not effective (Manning et al., 2004). This finding is also in agreement with the fact that focal unilateral injection of lidocaine into the perioral region of the cortex in WAG/Rij caused a general decrease in the number of SWDs recorded in other cortical areas in both hemispheres (Sitnikova and van Luijtelaar, 2004). This finding is consistent with clinical data suggesting that generalized SWD tend to begin in the cortex (see Sitnikova and van Luijtelaar, 2004). In addition, the somatosensory region of the cortex is known to produce physiologic oscillations in the frequency domain of 7 to 12c/s (Ritz et al., 1997). This "somatosensory rhythm" is supposed to be due to the synchronous functioning of the cortical pyramidal cells and inhibitory interneurons (Silva et al., 1991). Because SWD share several similarities with somatosensory rhythms, it has been suggested that SWD could derive from these cortical physiologic oscillations (Sitnikova and van Luijtelaar, 2004).
Several data are also in agreement with the existence of a cortical and eventually "focal" generator of SWD. In the cerebral cortex, the main source of inhibition comes from local nonpyramidal GABAergic interneurons (basket cells, chandelier cells, double bouquet cells, and bipolar cells), which strongly control the activity of cortical pyramidal neurons (Beaulieu, 1993; Defelipe, 1993; Meinecke and Peters, 1987; Somogyi and Cowey, 1981). In WAG/Rij rats, extracellularly and intracellularly recorded synaptic responses revealed an intracortical hyperexcitability accompanied by a reduction of the efficiency of GABA-ergic inhibition in slices of the frontal cortex (Luhmann et al., 1995). In addition, immunocytochemical studies on the WAG/RiJ cortex showed that some areas lacked the calcium-binding protein parvalbumin, which is co-localized with GABA (van de Bovenkamp-Janssen et al., 2004b). Finally, it was shown that voltage-gated sodium channels, as measured by quantitative polymerase chain reaction (PCR) and immuno-cytochemistry, are upregulated selectively at the facial somatosensory cortex in WAG/Rij rats (Klein et al., 2004). Further experiments are required to determine the cellular mechanisms underlying SWD in the focal cortical generator and whether such a focus also exists in the GAERS and can be assessed in the human patients.
Both genetic models, GAERS and WAG/Rij rats, offer a very high predictability for an antiabsence effect of an AED under development (see preceding discussion). In addition, both models have been useful in predicting adverse effects of some AEDs, such as vigabatrin and tiagabine, which should not be prescribed in cases of childhood absence epilepsy or in other cases of generalized absence seizures. In general, GAERS and WAG/Rij are very useful models to estimate the potential antiepileptic or adverse effect of a new drug candidate for the central nervous system (van Luijtelaar et al., 2002). Designing new drugs dedicated to suppressing oscillatory rhythms in the cortex or the thalamus may lead to important cognitive or vigilance side effects.
Was this article helpful?