Absence seizures are characterized by bilaterally synchronous spike-and-wave discharges (SWDs) in the electroencephalogram, which reflect abnormal oscillations in corticothalamic networks. Although it was suggested that basal ganglia could modulate, via their feedback circuits to the cerebral cortex, the occurrence of SWDs, the cellular and network mechanisms underlying such a subcortical control of absence seizures remain unknown. The GABAergic projections from substantia nigra pars reticulata (SNR) to thalamocortical neurons of the ventral medial (VM) thalamic nucleus provide a potent network for the control of absence seizures by basal ganglia. The present in vivo study provides the first description of the activity of VM thalamic neurons during seizures in the genetic absence epilepsy rats from Strasbourg, a well established model of absence epilepsy. Cortical paroxysms were accompanied in VM thalamic neurons by rhythmic bursts of action potentials. Pharmacological blockade of excitatory inputs of nigrothalamic neurons led to a transient interruption of SWDs, correlated with a change in the activity of thalamic cells, which was increased in frequency and converted into a sustained arrhythmic firing pattern. Simultaneously, cortical neurons exhibited a decrease in their firing rate that was associated with an increase in membrane polarization and a decrease in input resistance. These new findings demonstrate that an inhibition of SNR neurons changes the activity of their thalamic targets, which in turn could affect cortical neurons excitability and, consequently, the generation of cortical epileptic discharges. Thus, the nigro-thalamo-cortical pathway may provide an on-line system control of absence seizures.
- ventral medial thalamic nucleus
- absence epilepsy
- basal ganglia
- in vivo
- substantia nigra pars reticulate
Absence epilepsy is a generalized, nonconvulsive epilepsy of multifactorial genetic origin (Panayiotopoulos, 1997; Crunelli and Leresche, 2002). Absence seizures consist of a brief and sudden impairment of consciousness concomitant with bilateral synchronized spike-and-wave discharges (SWDs) in the electroencephalogram (EEG) over wide cortical areas (Panayiotopoulos, 1997). Electrophysiological recordings in patients (Williams, 1953) and various animal models of absence epilepsy (for review, see Danober et al., 1998; Timofeev and Steriade, 2004) revealed that SWDs result from abnormal oscillations in thalamocortical loops. Studies in genetic absence epilepsy rats from Strasbourg (GAERS), a well established model of absence epilepsy (Marescaux et al., 1992; Danober et al., 1998), suggest that basal ganglia provide a subcortical system controlling absence seizures via a modulation of activity of the substantia nigra pars reticulata (SNR), the main basal ganglia output structure (for review, see Danober et al., 1998; Deransart et al., 1998; Deransart and Depaulis, 2002). For instance, pharmacological activation of GABAergic striatonigral neurons, as well as application of GABAA agonists within the SNR, suppresses absence seizures, whereas an aggravation of cortical paroxysms is obtained after application of GABAA antagonists in the SNR (Danober et al., 1998; Deransart et al., 1998). Moreover, recent in vivo electrophysiological studies in GAERS demonstrated that SWDs propagate in basal ganglia circuits with a severe functional imbalance between the cortico-striato-nigral and cortico-subthalamo-nigral pathways (for review, see Paz et al., 2005b) (Fig. 1). Specifically, propagation of epileptic discharges in corticostriatal axons produces a powerful feedforward inhibition of GABAergic striatal output neurons (Slaght et al., 2004; Paz et al., 2005b), whereas paroxysmal discharges in the corticosubthalamic pathway generate synchronized bursting activity in glutamatergic subthalamo-nigral neurons (Paz et al., 2005a,b) (Fig. 1). Such an increase in subthalamo-nigral activity, together with the silencing of striatonigral neurons, are consistent with the rhythmic bursting observed in SNR cells during ictal activity in freely moving GAERS (Deransart et al., 2003). Together, these findings suggest that changes in the firing of nigral neurons could modulate “on-line” the occurrence of absence seizures. However, the cellular and network mechanisms by which the SNR could affect the cortical excitability and regulate the emergence of SWDs remain unknown.
The GABAergic nigrothalamic projections to the ventral medial nucleus (VM) of the thalamus (Di Chiara et al., 1979; Yoshida and Omata, 1979; Kilpatrick et al., 1980; Chevalier and Deniau, 1982; Ueki, 1983) represent a major efferent system of basal ganglia through which electrical signals propagated in basal ganglia networks are conveyed to the neocortex (Herkenham, 1979, 1980; Glenn et al., 1982; Moran et al., 1982) (Fig. 1). Thus, the nigro-thalamo-cortical pathway provides a potent feedback circuit that could control absence seizures. In the present study, we first characterized, in vivo in the GAERS, the firing pattern and intracellular events of VM thalamic neurons during SWDs. We further examined the mechanisms of control of cortical seizures by the nigrothalamic pathway by determining the impact of a transient blockade of subthalamo-nigral glutamatergic transmission on the firing of VM thalamocortical cells and its repercussion on cortical neurons excitability.
Materials and Methods
All experiments were performed in accordance with local ethical committee and European Union guidelines (Directive 86/609/EEC) and every precaution was taken to minimize stress and the number of animals used in each series of experiments.
Experiments were performed in vivo from 28 GAERS (20 females and 8 males; 3–10 months of age; 170–340 g) and 7 nonepileptic Wistar rats (5 females and 2 males; 3–5 months of age; 200–280 g). Animals were initially anesthetized with sodium pentobarbital (40 mg/kg, i.p.; Sanofi, Libourne, France) and ketamine (100 mg/kg, i.m.; Imalgène, Rhone Mérieux, France). A cannula was inserted into the trachea and the animal was placed in a stereotaxic frame. Wounds and pressure points were repeatedly (every 2 h) infiltrated with lignocaine (2%). Once the surgical procedures had been completed (see below), ear bars were removed and the head was held via a metallic rod cemented to the skull. Rats were subsequently maintained in a narcotized and sedated state by injections of fentanyl (4 μg/kg, i.p.; Janssen-Cilag, Issy-Les-Moulineaux, France) repeated every 20–30 min (Simons and Carvell, 1989; Pinault et al., 1998; Charpier et al., 1999; Slaght et al., 2002a, 2004; Bruno et al., 2003; Paz et al., 2005a). To obtain long-lasting stable intracellular recordings, rats were immobilized with gallamine triethiodide (40 mg, i.m., every 2 h; Specia, Paris, France) and artificially ventilated. The degree of anesthesia was assessed by continuously monitoring the EEG and heart rate, and additional doses of fentanyl were administered at the slightest change toward a waking pattern (i.e., an increase in the frequency and reduction in amplitude of EEG waves and/or an increase in the heart rate). Body temperature was maintained (36.5–37.5°C) with a homeothermic blanket. At the end of the experiments, animals received an overdose of sodium pentobarbital (200 mg/kg, i.p.).
EEG recordings were obtained with a low-impedance (≈60 KΩ) silver electrode placed on the dura above the orofacial motor cortex (12.5 mm anterior to the interaural line, 3.3–4 mm lateral to the midline) (Hall and Lindholm, 1974; Neafsey et al., 1986; Paxinos and Watson, 1986) and the somatosensory cortex (6.7 mm anterior to the interaural line, 4 mm lateral to the midline) (Paxinos and Watson, 1986). The reference electrode was placed in the muscle on the opposite side of the head.
Intracellular recordings were performed using glass micropipettes filled with 2 m potassium acetate (40–70 MΩ). Measurements of apparent membrane input resistance and time constant were based on the linear electrical cable theory applied to an idealized isopotential neuron (Rall, 1969). The voltage–current (V–I) relationship was measured from variations of the membrane potential in response to intracellular injections of hyperpolarizing current pulses (−0.2 to −1.2 nA; 200 ms duration; every 1.55 s; n ≥ 6) applied through the recording electrode. Apparent input resistance was measured from the mean (n ≥ 6) membrane potential change at the end of hyperpolarizing current pulses of low intensity (−0.4 nA; 200 ms duration; every 1.55 s) applied through the recording electrode or by measurement of the slope of the linear portion of the V–I curve (see Fig. 4 A2,B2). The values of membrane potential were corrected according to the potential recorded extracellularly immediately after termination of the intracellular recording. The membrane time constant, calculated from −0.4 nA current pulses, was the time taken for the membrane potential to reach 63% of its final value. The mean membrane potential was calculated during interictal activity from recording periods of 25 s.
For single-unit extracellular recordings and juxtacellular labeling (see below), glass electrodes were filled with 0.5 m NaCl and 1.7% neurobiotin (10–20 MΩ; Vector Laboratories, Burlingame, CA).
Cortical cells, located in the orofacial motor cortex, were recorded within 300 μm of the EEG electrode at the following coordinates: 12.5 mm anterior to the interaural line, 3–4 mm lateral to the midline, and 0.93–2.72 mm below the cortical surface, indicating a somatic localization in the layers V and VI (Zilles, 1985).
Intracellular, single-unit and multiunit extracellular recordings in the VM thalamic nucleus were obtained from the region projecting to the orofacial motor cortex (Herkenham, 1979, 1980; Glenn et al., 1982; Deniau and Chevalier, 1985) and receiving inputs from the orofacial region of SNR (Deniau and Chevalier, 1992). The corresponding stereotaxic coordinates were 5.9–6.7 mm anterior to the interaural line, 1.3–1.8 mm lateral to the midline, and 6.40–7.14 mm below the cortical surface. Intracellular recordings from ventral lateral (VL) thalamic neurons were obtained at the corresponding coordinates, 5.9–6.7 mm anterior to the interaural line, 1.4–1.8 mm lateral to the midline, and 5.50 – 6.30 mm below the cortical surface (Paxinos and Watson, 1986).
Extracellular recordings from SNR neurons were obtained at the following coordinates: 3.4–3.7 mm anterior to the interaural line, 2.4–3.2 mm lateral to the midline, and 7.05–8.10 mm below the cortical surface.
In all experiments, intracellular or single-unit and multiunit extracellular recordings were simultaneously performed with the corresponding ipsilateral EEG of orofacial motor cortex.
Pharmacological blockade of glutamatergic transmission in the SNR was obtained after local injection of kynurenate (KYNU) (100 nl; 75 mm; Sigma, Fallavier, France), a nonselective NMDA and AMPA receptors antagonist (Stone, 1993), via a glass micropipette (tip diameter, 40–50 μm) fitted to a 1 μl Hamilton syringe (Chevalier et al., 1985). Kynurenate was dissolved in NaCl (0.9%) and all solutions were adjusted to pH 7.4. Kynurenate was injected within the SNR at the following stereotaxic coordinates: 3.2 mm anterior to the interaural line, 2.2–2.4 mm lateral to the midline, and 7.7–8 mm below the cortical surface. When SNR neurons were simultaneously recorded (n = 5 experiments), the injection micropipette was positioned obliquely within the SNR and its tip was sufficiently distant (∼0.20 mm) from the vertical recording electrode to avoid any mechanical interference between the injection and the electrophysiological recording. Kynurenate injection, at a volume of 100 nl that diffuses for distances <0.2–0.3 mm (Myers, 1971; Robledo and Feger, 1990), was thus restricted to the SNR without diffusion in neighboring structures. For each animal, the exact location of injection sites in SNR was determined on histological sections using the atlas of Paxinos and Watson (1986).
Extracellularly recorded neurons were labeled by juxtacellular injection of Neurobiotin (Pinault, 1996; Mailly et al., 2003). Briefly, positive current pulses (1–8 nA; 200 ms) were applied at a frequency of 2.5 Hz through the bridge circuit of the amplifier. The current was slowly increased while the electrode was advanced toward the neuron in 1 μm steps (LSS-1000 Inchworm Motor Positioning System; Burleigh Instruments, Fishers, NY) until the cell discharge was driven by the injected current. Current pulses were applied for a 10–30 min period to obtain a reliable labeling of neuronal processes. For intracellular recordings, depolarizing current pulses (0.2–1 nA; 100–200 ms) were applied at a frequency of 2.5 Hz at the end of the recording period. The histochemical methods used to reveal the morphology of neurobiotin-filled neurons have been described in detail previously (Slaght et al., 2002a). The position of labeled neurons within the structures was confirmed using the atlas of Paxinos and Watson (1986).
Data acquisition and analysis.
Intracellular recordings were obtained under current-clamp conditions using the active bridge mode of an Axoclamp 2B amplifier (Molecular Devices, Union City, CA). Data were stored on-line on a DRA 800 digital tape recorder (Biologic, Claix, France) and then digitized with a sampling rate of 20 kHz (intracellular signals), 10 kHz (extracellular signals), or 300 Hz (EEG) for off-line analysis. To perform spectral analysis of EEG potentials, fast Fourier transforms were applied using Spike 2 (CED Software; Cambridge Electronic Design, Cambridge, UK). Cross-correlograms of the firing between two simultaneously recorded units were obtained by first encoding the position of the peak of the action potentials into separate channels using the memory buffer function of Spike 2, and the event correlation function was then used to produce the cross-correlogram. The amplitude of action potentials was calculated as the potential difference between their voltage threshold, measured as the membrane potential at which the dV/dt exceeded 10 V · s−1 (Mahon et al., 2003), and their peak. Numerical values are given as means ± SEM unless stated otherwise. Statistical significance was assessed by performing appropriate statistical tests, Student's paired t test, one-way ANOVA, or Mann–Whitney test. In some measurements, the normality of distributions was tested using the Kolmogorov–Smirnov test and a Gaussian–Laplace fit was performed. Statistical analysis and curve fitting were performed with Origin 7.0 (Microcal Software, Northampton, MA) and SigmaStat 3.0 (SPSS, Chicago, IL).
To quantify the degree of statistical association and the time delay between EEG waves and intracellular activity of VM thalamic neurons, the nonlinear correlation coefficient h 2 was calculated between these signals as a function of a time shift (τ) (Lopes da Silva et al., 1989; Slaght et al., 2004). In contrast with classical methods, such as cross-correlation or coherence, which are only sensitive to linear relationships, this nonlinear index can reveal signal interdependencies under more general conditions (Meeren et al., 2002).
To quantify the dependence of a signal Y on a signal X, an estimator of the nonlinear correlation index (h 2) was computed as follows: , where g(xi ) denotes a piecewise linear regression curve, used to predict the signal Y at any value of X, and 〈y〉 denotes the average of the time series yi over N samples. Statistically, this nonlinear correlation index quantifies the reduction of variance that can be obtained by predicting the Y values on the basis of the regression curve. The nonlinear index ranges between 0, when both signals are independent, and 1, for a perfect dependence. Theoretical and practical aspects of this method have been described in detail previously (Lopes da Silva et al., 1989).
In the present study, the degree of association between EEG and intracellular signals and the corresponding time delays during seizures were obtained by computing h 2 as a function of a time shift (τ) between the signals y and x. The shift for which the maximum of h 2(τ) was reached provided an estimate of the time delay between the activities (Lopes da Silva et al., 1989; Meeren et al., 2002). Because epileptic activities are not stationary in time, correlation indices were computed for sliding short epochs of 2 s with 98% (≈1.96 s) of temporal overlap.
Properties of cortical SWDs
The temporal properties of SWDs recorded by the surface EEG electrode, as well as the shape of individual spike-wave complexes, were similar to those previously described under analogous experimental conditions (Pinault et al., 1998; Charpier et al., 1999; Slaght et al., 2002a,b, 2004; Pinault, 2003; Paz et al., 2005a) and in freely moving GAERS (Marescaux et al., 1992; Deransart et al., 2003). Specifically, SWDs (n = 1777 from 28 GAERS) recorded in the present study had a mean duration of 4.7 ± 0.4 s (from 0.9 to 140 s) and recurred once every 52.5 ± 10.6 s. The intra-SWD frequency, revealed by spectral analysis of the EEG signal, ranged from 7 to 9.5 Hz.
Extracellular activity of VM thalamic neurons
We first examined the spontaneous activity of VM thalamic neurons in GAERS in extracellular recordings of single units (n = 30 cells from six rats). Recorded cells, which were morphologically identified by juxtacellular injection of Neurobiotin (see Materials and Methods), were located in the region of the VM thalamic nucleus (Figs. 2 A1, 3 D, inset) related to the orofacial motor cortex (Herkenham, 1979, 1980; Glenn et al., 1982; Deniau and Chevalier, 1985). They exhibited the morphological characteristics of VM thalamic cells (Figs. 2 A1, 3 D), either a fusiform or polygonal soma, with a diameter of ∼ 25 μm, and widespread dendritic ramifications (Yamamoto et al., 1985). The axonal projections, which could be followed 1.5 mm from the soma, pursued rostrolateral course toward the cerebral cortex.
Spontaneous transitions between interictal and ictal periods were characterized in VM thalamic cells by a switch from arrhythmic, single-spike activity to burst firing mode (Fig. 2 A2). Across cells, the mean firing frequency during interictal periods was 2.40 ± 0.24 Hz (from 0.12 to 4.84 Hz; n = 30 cells) (Fig. 2 B1). During SWDs, the firing rate of thalamic neurons reached a mean value of 7.86 ± 0.72 Hz (range, 2.20–17.28 Hz; n = 659 SWDs from 30 cells) (Fig. 2 B1) corresponding to a probability of discharge associated with individual EEG spikes of 0.42 ± 0.03 (from 0.07 to 0.80; n = 30 cells). The end of cortical paroxysms could be associated in VM thalamic cells (n = 20 cells) with a transient period of excitation manifested by repetitive discharges of action potentials (Fig. 2 A2). The percentage increase in firing rate of VM thalamic neurons during SWDs declined exponentially as a function of the mean interictal firing rate (Fig. 2 B2) and had a mean value of 359.3 ± 73.1% (n = 30 cells) (Fig. 2 B1).
Cortical epileptic discharges also resulted in a drastic change in the firing pattern of VM thalamic neurons, which was characterized by high-frequency bursts of action potentials shortly before the spike component of the EEG (Figs. 2 A2,C1; 3 A,B2). Bursts contained from two to six action potentials (2.69 ± 0.01; n = 4548 bursts from 670 SWDs; n = 30 neurons) (Figs. 2 C1, 3 B2) and had a duration of 4.81 ± 0.03 ms (range, 2.10–30.81 ms; n = 4548), with a mean intraburst frequency of 355.01 ± 0.94 Hz (range, 74.96–543.48 Hz; n = 4548 bursts). The delay of the first action potential, using the peak of the corresponding EEG spike component as the zero-time reference, was −15.84 ± 0.12 ms (n = 6261 action potentials from 670 SWDs; n = 30 cells) (Fig. 2 C2). A similar analysis using all action potentials in a burst indicated a mean delay of −14.05 ± 0.09 ms (n = 14,070 action potentials from 4548 bursts in 670 SWDs; n = 30 cells) (Fig. 2 C3).
We could perform simultaneous extracellular recordings of two units (n = 3 pairs from two GAERS) in the VM thalamus (Fig. 3 A). These multiunit recordings highlighted how typically random firing patterns of neighboring thalamic neurons in the absence of paroxysms (Fig. 3 B1,B3) became highly correlated and time-locked to the EEG spike during SWDs (Fig. 3 B2,B4). They also indicated that the first action potential in a burst from one VM unit could either precede or follow the first action potential in the concomitant burst of the neighboring cell, with a mean delay between these two action potentials of 5.33 ± 0.58 ms (range, 0.4–37.9 ms; n = 368 delays from 75 SWDs; n = 3 double-unit recordings) (Fig. 3 C). Consistent with the single-unit recordings (Fig. 2 A2), the cessation of SWDs was associated in the two simultaneously recorded thalamic neurons with a rebound of excitation evident as either single action potentials (Fig. 3 A, gray unit) or long clusters of 4–14 action potentials (Fig. 3 A, black unit). During multiunit recording experiments, both extracellularly recorded cells could be labeled by the juxtacellular injection of Neurobiotin (Fig. 3 D).
Intracellular recordings of VM thalamic neurons: electrical membrane properties
The hyperactivity of GAERS VM thalamic neurons during SWDs might result from an alteration of their intrinsic excitability. To test this hypothesis, we used intracellular recordings to compare passive and active electrical membrane properties of VM thalamic cells in GAERS (n = 14 cells from 10 GAERS) and nonepileptic rats (n = 9 cells from 7 Wistar rats). We first examined the membrane V–I relationship, which was determined during interictal periods in GAERS, by measuring membrane potential changes in response to a series of intracellular square current pulses (Fig. 4 A1,A2,B1,B2). The apparent input resistance, measured from the linear portion of the V–I plot (Fig. 4 A2,B2), was similar in both rat strains (Table 1). In both cell populations, a marked inward rectification was consistently observed in response to negative current pulses of high intensity (Fig. 4 A1,A2,B1,B2). This resulted from a depolarizing “sag” of membrane potential (Fig. 4 A1,B1, arrow), likely caused by the hyperpolarization-activated inward cationic current (I h) (McCormick and Pape, 1990). Large-amplitude current-induced hyperpolarizations were systematically followed by a rebound of depolarization (Fig. 4 A1,B1, arrowhead) that might result from a mixture of the low-threshold activated calcium current (I T) (Llinas and Jahnsen, 1982; Steriade et al., 1990) and I h. A characteristic electrical feature of relay thalamic neurons (Llinas and Jahnsen, 1982; Steriade et al., 1990) was that depolarizing current pulses applied from the resting potential generated a tonic firing pattern in VM thalamic cells of both GAERS and control animals (Fig. 4 A1;A3, left;B1;B3, left). During DC hyperpolarization, which deinactivates I T (Steriade et al., 1990), the same current pulse induced in both cell populations a low-threshold Ca2+ potential (LTCP) crowned by a burst of Na+ action potentials (Fig. 4 A3;B3, right panels). Moreover, basic electrical membrane properties of thalamic cells, including membrane potential, membrane time constant, and action potentials properties, were similar in both GAERS and normal rats (Table 1).
Because passive and active intrinsic membrane properties of VM thalamic neurons recorded from GAERS and normal Wistar rats did not differ significantly, it is unlikely that the sustained bursting activity of thalamic cells during SWDs was caused by an alteration in their intrinsic excitability.
Intracellular recordings of VM thalamic neurons: cellular events associated with absence seizures
In vivo intracellular recordings from GAERS VM thalamic neurons (n = 14) confirmed and extended the observations made with extracellular recordings. Intracellularly and extracellularly recorded cells were located in the same thalamic region and exhibited identical morphological features (data not shown). When an SWD appeared in the EEG, the firing of VM thalamic cells switched from single spike activity to burst firing (Fig. 5 A1), accompanied by membrane potential oscillations temporally correlated with spike-wave complexes (Fig. 5 A1,A2). Repetitive membrane depolarizations were superimposed on a “croissant”-shaped hyperpolarizing envelope of amplitude 12.2 ± 1.1 mV (range, 6.6–20.1 mV; n = 14 cells) that lasted for the entire SWD (Fig. 5 A1). Measurements of hyperpolarizing envelope amplitude as a function of the membrane potential (data not shown), which was maintained at different levels of polarization by DC injections, indicated a virtual reversal potential of approximately −100 mV, consistent with the equilibrium potential of K+ in thalamic neurons (Destexhe and Sejnowski, 2001).
The suprathreshold oscillations in thalamic neurons during SWDs were composed of temporally summating high-frequency synaptic potentials (Fig. 5 A2, inset) and a LTCP-like depolarization (Fig. 5 A2, arrow) giving rise to multiple Na+ action potentials, whereas subthreshold events (Fig. 5 A2, crossed arrow) were characterized by a smooth depolarizing rising phase (∼40 ms duration) followed by a slower (∼100 ms duration) decaying component.
The end of the SWD in the cortical EEG coincided with the cessation of intracellular thalamic oscillations and of the hyperpolarization (Fig. 5 A1). The membrane repolarization was often (n = 40 SWDs from nine GAERS) followed by repetitive firing (Fig. 5 A1) similar to the postictal discharge observed during extracellular recordings (Figs. 2 A2, 3A).
Intracellular recordings of VL thalamic neurons
In the course of thalamic recordings, we obtained intracellular records (n = 5) (Fig. 6) from the GAERS VL thalamic nucleus, which is located just above the VM thalamic nucleus (see Materials and Methods). Ventral lateral thalamic cells had a mean interictal membrane potential (−58.0 ± 1.7 mV; n = 5 cells), apparent input resistance (25.6 ± 1.8 MΩ; n = 5 cells), and membrane time constant (13.5 ± 1.9 s; n = 5 cells) similar (p > 0.3 for each parameters) to that calculated in VM thalamic neurons (Table 1). During interictal periods, thalamic VL cells exhibited a mean firing frequency (2.21 ± 0.55 Hz; n = 5 cells) (Fig. 6 A1) comparable (p > 0.9) to that of VM thalamic units (Fig. 2 B1). As observed in intracellularly recorded VM thalamic cells (Fig. 5 A1), the occurrence of cortical paroxysms were accompanied in VL thalamic cells with membrane oscillations (Fig. 6 A1,A2) superimposed on a croissant-shaped hyperpolarization (Fig. 6 A1). However, in contrast with VM thalamic cells that showed a significant increase in their mean firing rate during SWDs (Fig. 2 B1,B2), the rate of discharge in VL thalamic neurons was not significantly modified by the seizures (1.31 ± 0.21 Hz; n = 47 SWDs from five cells; p > 0.5) (Fig. 6 A1) and was markedly lower (p < 0.001) to that measured in VM thalamic cells (Fig. 2 B1). This relative weak activity of VL thalamic cells during SWDs was not attributable to a specific alteration in their intrinsic excitability because they exhibited, as VM thalamic neurons (Fig. 4 A1), the typical active membrane properties of thalamocortical neurons, including a depolarizing sag of membrane potential during current-induced membrane hyperpolarization, a postinhibitory rebound of depolarization and a tonic firing pattern in response to depolarizing current pulses applied from the resting potential (Fig. 6 A3).
Association strength and temporal relationship between VM intracellular oscillations and cortical SWDs
To assess the functional links and the directionality of information flow between the cerebral cortex and the VM thalamic nucleus, we measured the strength of association (h 2) (see Materials and Methods) and the temporal delays between EEG and thalamic intracellular activities. As illustrated in Figure 5 B, the seizure onset was characterized by a gradual increase in the degree of association between the signals and, after reaching a maximum in the middle of the crisis, h 2 progressively recovered its preictal value. From four paired recordings of surface cortical and intracellular thalamic activities (n = 23 seizures), the average degree of association between both signals during the first second of the SWD (h 2 = 0.38 ± 0.13; n = 23 SWDs) was significantly higher than before the ictal epoch (h 2 = 0.19 ± 0.12; p < 0.001). Furthermore, values for the time delay derived from nonlinear correlation of EEG waves and the intracellular activity of VM thalamic cells were profoundly modified by the seizure (Fig. 5 B2). Before SWDs, low values for h 2 were associated with highly variable time delays. In contrast, as the seizure began, a clear and stable time lag emerged between thalamic intracellular and cortical EEG rhythmic activities and was maintained throughout the SWD (Fig. 5 B2). For the 23 analyzed seizures, oscillatory thalamic signals preceded the cortical oscillations by 12.2 ± 8.5 ms.
Blockade of cortical SWDs after intranigral injection of kynurenate
GABAergic projections of the SNR to VM thalamocortical neurons provide a potent feedback to the cerebral cortex that could underlie the control of absence seizures by the basal ganglia (see Introduction). To test this hypothesis, we first examined the impact on the EEG of intranigral injection of a glutamate receptor antagonist (kynurenate), expected to block the intranigral glutamatergic transmission, mainly originating from the subthalamus (Smith and Parent, 1988; Robledo and Feger, 1990), and decrease the activity of SNR cells (see below). A unilateral injection of 100 nl of kynurenate in SNR abolished ipsilateral SWDs in both motor and somatosensory cortices (Fig. 7) for 20 ± 3.7 min (n = 13 injections with recovery). After recovery, which occurred simultaneously in both cortical regions, cortical paroxysms had a duration (8.29 ± 0.74 s; n = 368 SWDs) and a period of recurrence (55.58 ± 5.40 s; n = 352 inter-SWD intervals) similar (p > 0.05 for both parameters) to that calculated during preinjection periods (control duration, 6.78 ± 0.40 s; n = 513 SWDs; control inter-SWD intervals, 51.76 ± 4.96 s; n = 489).
The specificity of kynurenate injection in the blockade of generalized seizures was attested by the lack of EEG changes after intranigral vehicle injections (0.9% NaCl; 100–200 nl; n = 5 injections) (Fig. 8 A). Duration of cortical paroxysms (before vehicle, 7.51 ± 0.58 s; n = 194 SWDs; after vehicle, 7.76 ± 0.54 s; n = 130 SWDs) as well as the inter-SWD interval (before vehicle, 22.39 ± 1.99 s; n = 189; after vehicle, 18.42 ± 2.21 s; n = 125) were not significantly modified by the vehicle injection (p > 0.1 for both parameters).
Effect of intranigral injections of kynurenate on the activity of SNR and VM thalamic neurons
To determine the mechanisms of control of absence seizures by the nigrothalamic pathway, we first correlated the antiepileptic effects of intranigral injection of kynurenate with changes in activity of SNR and VM thalamic neurons.
In control conditions, extracellularly recorded SNR neurons (n = 11 from five GAERS) showed an elevated interictal firing rate of 19.30 ± 1.01 Hz (n = 11 cells). In five recordings, of duration sufficient to test the effect of kynurenate injection, the blockade of SWDs was associated with a significant decrease in the firing rate of nigral neurons (before kynurenate, 19.72 ± 0.94 Hz; after kynurenate, 4.17 ± 2.40 Hz; n = 5 cells; p < 0.001) (Fig. 8, compare B1, B2; D1). The recovery of seizures was concomitant with a restoration of the interictal firing activity of SNR neurons (22.42 ± 2.95 Hz; n = 4 cells; p > 0.4, control vs recovery) (Fig. 8, compare B3, B1; D1).
In this set of experiments, the five extracellularly recorded VM thalamic neurons had, during control condition, a mean interictal firing rate of 0.93 ± 0.08 Hz (Fig. 8 C1,C2,D2). As expected from the inhibitory influence of SNR on VM thalamic nucleus (Deniau and Chevalier, 1985), the pharmacological blockade of glutamatergic transmission in the SNR, which also interrupted SWDs, was reflected in thalamic cells by a marked increase in firing rate (4.46 ± 2.26 Hz; n = 4 cells; p < 0.01, control vs kynurenate) (Fig. 8 C1,C3). The restoration of SWDs, that could be obtained during three extracellular thalamic records, was correlated with a recovery in the interictal firing rate of VM thalamic neurons (1.59 ± 0.31 Hz; n = 3 cells; p > 0.2, control vs recovery) (Fig. 8 C1,C4,D2).
Demonstrating the specific action of intranigral injection of kynurenate, vehicle injections (0.9% NaCl; 100–200 nl; n = 5) into the SNR had no effect on either firing frequency of VM neurons (before vehicle, 1.43 ± 0.15 Hz, n = 10 cells; after vehicle, 1.36 ± 0.17 Hz, n = 5 cells; p = 0.6) or cortical EEG pattern (Fig. 8 A).
Intranigral injection of kynurenate decreases cortical neurons excitability
To determine the cortical mechanisms underlying the control of absence seizures by the SNR, we examined the effect of intranigral injections of kynurenate on simultaneously recorded EEG and cortical intracellular activities. In the 10 intracellularly recorded cortical cells, the activity between seizures, before kynurenate injection, was characterized by irregular, depolarizing and hyperpolarizing, membrane potential fluctuations that generated a random firing pattern at a mean frequency of 3.05 ± 0.24 Hz (Fig. 9 A1). SWDs were accompanied in cortical neurons with repetitive membrane depolarizations superimposed on a tonic hyperpolarization that lasted for the entire surface paroxysm. These interictal and ictal intracellular activities are similar to those previously described in the same orofacial cortical region of GAERS (Charpier et al., 1999; Slaght et al., 2004; Paz et al., 2005a,b). After kynurenate application, the suppression of SWDs was correlated with a significant decrease in cortical neuron firing frequency (0.44 ± 0.25 Hz; n = 10 cells; p < 0.001) (Fig. 9 A2). When intracellular recordings were kept until recovery of SWDs (n = 7 cells), the interictal firing rate of cortical cells was restored close its baseline value (2.19 ± 0.28 Hz; n = 7 neurons) (Fig. 9 A3). Moreover, the cellular correlates of surface paroxysms (Fig. 9 A3) were similar to those observed during control periods.
The kynurenate-induced decrease in cortical activity was associated with changes in the passive membrane properties of cortical cells. The membrane potential of recorded cells (before injection, −58.03 ± 0.42 mV; n = 10 cells) hyperpolarized significantly after intranigral application of kynurenate (−64.64 ± 1.33 mV; n = 10 cells; p < 0.001) (Fig. 9 A1,A2,B,D, top). This was correlated with a decrease in apparent input resistance (before kynurenate, 25.67 ± 1.67 MΩ, n = 10 cells; vs after kynurenate, 18.77 ± 1.31 MΩ, n = 9 cells; p < 0.001) (Fig. 9 C,D, middle) and time constant (before kynurenate, 12.13 ± 1.07 ms, n = 10 cells; vs after kynurenate, 7.67 ± 0.37 ms, n = 9 cells; p < 0.001) (Fig. 9 C,D, bottom). V–I relationships of cortical neurons, during both preinjection and postinjection periods, were linear (apparent input resistance was independent of the level of membrane polarization) (data not shown), indicating that the reduction in input resistance was not attributable to an inward rectification at membrane potentials reached after drug injection. Again, the recovery of SWDs led to a recovery of the control membrane input resistance (26.88 ± 3.91 MΩ; n = 6 cells; p > 0.7, control vs recovery) and time constant (9.77 ± 0.55 ms; n = 6 cells; p > 0.1, control vs recovery) (Fig. 9 D, middle and bottom). The diminution in cortical neurons firing rate did not result from changes in action potential properties. Their duration (before kynurenate, 1.68 ± 0.10 ms, n = 10 cells; vs after kynurenate, 1.60 ± 0.09 ms, n = 10 cells), amplitude (before kynurenate, 59.10 ± 1.90 ms, n = 10 cells; vs after kynurenate, 60.20 ± 2.9 ms, n = 10 cells) and voltage threshold (before kynurenate, −50.30 ± 0.60 ms, n = 10 cells; vs after kynurenate, −50.40 ± 0.70 ms, n = 10 cells) were not significantly modified (p > 0.7 for each parameter) after kynurenate injection.
The present study provides the first description of the activity of VM thalamic neurons during absence seizures and strongly suggests that the nigro-thalamo-cortical network participates in the control of cortical paroxysms. Our main findings are as follows: (1) during SWDs, VM thalamic neurons display synchronized repetitive high-frequency bursts of action potentials in-phase with EEG spikes; (2) intracellular recordings reveal that rhythmic bursting in thalamic cells is generated by LTCP-like depolarizations interacting with synaptic inputs; (3) the apparent hyperexcitability of VM thalamic neurons does not result from an alteration of membrane properties in epileptic animals; (4) pharmacological blockade of glutamatergic transmission in the SNR increases the rate of discharge in VM thalamic cells and leads to an irregular tonic firing pattern correlated with an interruption of cortical SWDs; and (5) this blockade of ictal activity is associated in cortical neurons with a membrane hyperpolarization and a decrease in input resistance.
Membrane oscillations and bursting of VM thalamic neurons during SWDs
The occurrence of SWDs in the EEG was accompanied in VM thalamic cells with oscillatory depolarizations superimposed on a croissant-shaped hyperpolarization (Fig. 5 A1). This intracellular waveform closely resembles that previously described from specific thalamocortical relay neurons in GAERS during spontaneous seizures (Pinault et al., 1998; Pinault, 2003) and in the cat during pharmacologically induced paroxysms (Steriade and Contreras, 1995; Timofeev et al., 1998; Timofeev and Steriade, 2004). The sustained hyperpolarization observed in VM thalamic cells during SWDs apparently reversed near −100 mV. This value is close to the equilibrium potential of K+, strongly suggesting the involvement of postsynaptic K+-dependent GABAB receptors (Destexhe and Sejnowski, 2001). The activation of these receptors, which generates a similar hyperpolarizing envelope in thalamocortical neurons in a pharmacological model of absence seizures in vitro (Bal et al., 1995), might result from repetitive discharges in converging GABAergic inputs arising from the reticular thalamic nucleus (Steriade and Contreras, 1995; Slaght et al., 2002a) and the SNR (Deransart et al., 2003) (our unpublished observations) during spike-and-wave activity.
The membrane oscillations in VM thalamic neurons during seizures likely result from complex interactions between synaptic and active intrinsic membrane properties. The early phase of thalamic rhythmic depolarizations during SWDs (Fig. 5 A2) could be attributable to the activation of I h (McCormick and Pape, 1990) by the sustained membrane hyperpolarization. The presence of I h in the recorded neurons was attested by the characteristic depolarizing sag during current-induced hyperpolarizations (Fig. 4 A1, arrow) (McCormick and Pape, 1990). Alternatively, the depolarizing phase of thalamic oscillations could originate from excitatory synaptic inputs and/or Cl−-dependent synaptic potentials that might be depolarizing from the membrane potential reached during the hyperpolarizing envelope (approximately −75 mV) (i.e., more negative that the Cl− equilibrium potential) (Destexhe and Sejnowski, 2001). These subthreshold rhythmic depolarizations could activate I T, from a deinactivated state caused by the membrane hyperpolarization, and generate an LTCP (Steriade et al., 1990). Such a Ca2+-dependent rebound of excitation, observed at the break of a hyperpolarizing current pulse (Fig. 4 A1, oblique arrowhead), could trigger during seizures a single or a burst of Na+ action potentials when coincident with a barrage of depolarizing synaptic potentials (Fig. 5 A2). These excitatory inputs probably arose from repetitive firing in corticothalamic neurons, a hypothesis consistent with the timing of action potentials discharge in corticofugal axons during absence seizures (Charpier et al., 1999; Pinault, 2003; Slaght et al., 2004; Paz et al., 2005a) and in agreement with the association strength of simultaneously recorded EEG and intracellular thalamic activities during the main body of the seizure (Fig. 5 B2).
SWDs in the EEG were associated with a drastic change in the firing pattern of VM thalamocortical neurons, characterized by a switch from interictal, irregular, single-spike activity to burst firing in-phase with EEG spikes. During SWDs, the mean firing rate of VM thalamic cells was tripled with a firing probability in association with individual EEG spikes as high as ∼0.5. This seizure-associated change in firing frequency and pattern contrasts with our observations in the present work from VL thalamic neurons (Fig. 6 A1) and with the majority of specific thalamocortical neurons previously recorded during spike-and-wave activity in GAERS (Pinault et al., 1998; Charpier et al., 1999) and cats (Steriade and Contreras, 1995; Timofeev et al., 1998; Timofeev and Steriade, 2004). Indeed, these specific thalamocortical neurons show no bursting activity and their firing decreases during SWDs, likely because of a sustained inhibition arising from the reticular thalamic nucleus (Steriade and Contreras, 1995; Slaght et al., 2002a).
The end of the SWD in the EEG coincided with the cessation of VM thalamic intracellular oscillations and removal of the tonic hyperpolarization. The postictal membrane repolarization was typically followed by a short (0.2–3.5 s) and intense (4–20 action potentials) period of firing. This postictal rebound of excitation, which was not associated with a prominent synaptic depolarization (Fig. 5 A1), could originate from a disinhibitory mechanism attributable to a decrease in the activity of nigrothalamic neurons at the end of cortical paroxysms (Deransart et al., 2003). However, we cannot exclude the contribution of a rebound response in corticothalamic neurons (Grenier et al., 1998; Timofeev et al., 2002) and/or of an intrinsic postinhibitory excitation attributable to I h and/or I T. The increase in firing of VM thalamic neurons associated with the termination of the seizure, could be responsible for the postictal rebound of depolarization in striatal output neurons (Arbuthnott et al., 1990; Slaght et al., 2004), which, in turn, might contribute to the transient interruption of firing in SNR neurons at the end of cortical paroxysms (Deransart et al., 2003).
Disinhibition of the thalamocortical pathway and decrease in cortical neurons excitability
We showed that unilateral injection of a nonselective antagonist of glutamatergic receptors in the SNR transiently suppresses cortical SWDs concomitantly with an elevated but arrhythmic firing in VM thalamocortical neurons. This finding is consistent with previous work showing that bilateral blockade of glutamatergic transmission in the SNR has antiepileptic effects (Deransart et al., 1996). Moreover, it has been already shown that the pharmacological blockade of GABAergic nigrothalamic activity in normal rats causes a time-locked increase in the activity of VM thalamocortical cells through a process of disinhibition (Deniau and Chevalier, 1985; Chevalier and Deniau, 1990). This transition of activity in thalamic neurons, associated with the desynchronized interictal EEG, is also consistent with the thalamic tonic firing recorded during brain-activated states of waking, whereas rhythmic bursting (as seen during SWDs) is mainly present during unconsciousness-associated cortical slow waves (Glenn and Steriade, 1982; Llinas and Steriade, 2006).
The disinhibition of thalamocortical neurons, which resulted in a cessation of seizures, was associated with a decrease in cortical neurons excitability, expressed by a decrease in firing rate and a decrease in apparent membrane input resistance and time constant. The increase in membrane conductance, together with alteration in membrane time constant, might produce a shunting effect on depolarizing synaptic potentials, leading to a dramatic decrease in their amplitude and making their temporal summation less effective (Rall, 1977). Such a decrease in membrane excitability could account for the lack of large-amplitude synaptic depolarizations in cortical neurons, which might explain, together with the hyperpolarization, the decrease in cortical firing rate (Fig. 9 A2).
The origin of the decreased excitability of cortical neurons after disinhibition of thalamocortical neurons remains unknown. However, it is plausible that the increased activity of nonspecific thalamocortical cells causes a widespread feedforward cortical inhibition by activating cortical GABAergic interneurons (Swadlow, 2003; Thomson and Bannister, 2003), resulting in an increase in the membrane conductance (Staley and Mody, 1992) of corticothalamic cells.
Basal ganglia as an on-line control system of absence seizures
The possibility that absence seizures are controlled by the SNR first emerged from pharmacological studies in GAERS showing that a bilateral inhibition of SNR suppresses cortical SWDs (Depaulis et al., 1988, 1989; Deransart et al., 1996, 1998, 2001). Because the blockade of the GABAergic transmission at the level of the superior colliculus, which likely originates from the SNR, results in an alteration of generalized nonconvulsive seizures (Depaulis et al., 1990), it has been first postulated the involvement of the nigrotectal pathway. The present findings, showing that changes in the electrophysiological activity of VM thalamic cells are correlated with a cessation of SWDs, reinforce the “nigrothalamic” hypothesis and suggest that basal ganglia could control on-line absence seizures. During SWDs, VM thalamic neurons exhibited repetitive bursting activity generated, at least in part, by the nigrothalamic inhibition responsible for a deinactivation of I T. Consistent with the temporal relationships and the association strength we measured between cortical and thalamic activities (Fig. 5 B) during the main body of SWDs, the rhythmic bursting in the thalamocortical loop should provide a resonant circuitry sustaining the cortical discharges and maintaining the coherence of seizure activity. The rebound of excitation of striatonigral neurons at seizure termination (Slaght et al., 2004) would be responsible for the decrease of firing in nigrothalamic neurons (Deransart et al., 2003). Thus, the subsequent tonic firing in VM thalamocortical neurons, through a disinhibitory process, might contribute to the cortical desynchronization (Glenn and Steriade, 1982) and, consequently, to the termination of the seizure and the recovery of conscious processes.
This work was supported by the Ministère Français de la Recherche and the Agence Nationale de la Recherche (ANR 2006). We are grateful to Drs. S. Mahon, R. Miles, and P. O. Polack for thoughtful discussion and critical reading of this manuscript. We also thank Dr. N. Maurice for assistance with the pharmacological experiments and A.-M. Godeheu for assistance with the histological processing.
- Correspondence should be addressed to Jeanne Tamar Paz, Institut National de la Santé et de la Recherche Médicale, Unité 667, Collège de France, F-75231 Paris, France.