Entête

1. Introduction

Table des matières

In the brain the interneuronal communications mediate behavior, perception, and thought. The properties of a neuronal population depend on the properties of individual neurons and the feed-back mechanisms that can affect individual neurons (Johnston and Brown 1986). These feedback mechanisms include synaptic connections, ephaptic interactions, and changes in extracellular space and ion concentrations. Many of the interactions are nonlinear and can lead to an output that is unstable and characterized by oscillations (Rutecki 1992). The important point to be made is that the behavior of individual neurons is defined by intrinsic ion conductances. Synaptic input will be amplified or dumped by these conductances and an output will be generated. The intrinsic currents generated by these conductances are affected by synaptic activities or changes in extracellular ion concentrations. The synaptic interactions between local neuronal constellations will influence other neurons and can result in coordinated activity of many neurons contributing to a functional output. Thus, change in either individual neuronal properties or feedback mechanisms of a neural network result in a change in the network output. These underlying changes may be rather small, but when processed in neural network, a small change may result in more dramatic network changes.

The Ca2+ as an endogenous ion plays a major role in modulation of both intrinsic neuronal excitability and synaptic activity (Crochet et al. 2005; Hille 2001; Katz 1969). However, the action of Ca2+ in these two phenomena is opposite. Extracellular calcium concentration ([Ca2+]o) is not fixed and fluctuate during both slow-wave sleep (Crochet et al. 2005; Massimini and Amzica 2001) and paroxysmal sleep oscillations (Amzica et al. 2002; Heinemann et al. 1977). As the consequence of this modulation the synaptic excitability decreases when [Ca2+]o decreases (Crochet et al. 2005; Katz 1969), but in the same time the intrinsic excitability increases (see below).

The following sections of this introduction describe the electrophysiological properties of cortical neurons; the modulation of these properties by network activity during both slow-wave sleep and paroxysmal activity.

Cortical neurons generate complicated patterns of activity during various network states, which depend on the interaction of very large numbers of interconnected neurons, their intrinsic properties and the state levels of neuromodulator activities. They display a heterogeneous distribution of electrophysiological properties that correlate with their morphology, anatomical connections and perhaps even neurotransmitter content (Connors and Gutnick 1990; McCormick et al. 1985).

The intrinsic properties of a neuron depend on unique sets of ionic channels, specific to a given neuron and on their distribution in different compartments of the neuron. The diversity of channels in neurons is large and results in a variety of patterns of action potential generation induced by a constant input. In the neocortex, four basic electrophysiological firing patterns of neuron have been identified: (a) Regular-spiking (RS) neurons constitute the majority of cortical neurons. They display trains of single spikes that adapt quickly or slowly to maintained stimulation. (b) Fast-rhythmic-bursting (FRB) neurons give rise to high-frequency (300-600 Hz) spike-bursts recurring at fast rates (30-50 Hz). (c) Intrinsically bursting (IB) neurons generate clusters of action potentials, with spike inactivation, followed by hyperpolarization and neuronal silence. (d) Fast-spiking (FS) neurons fire thin action potential and sustain tonically very high firing rates (up to 800 Hz) without frequency adaptation (Gray and McCormick 1996; McCormick et al. 1985; Steriade et al. 1998b). The duration of intracellularly recorded action potentials at half amplitude, measured during the state of natural waking in chronically implanted cats, shows modes between 0.6 and 1 ms in RS neurons. Slightly longer spikes are fired by IB neurons. In contrast, both FRB and FS neurons demonstrate much shorter action potentials, with modes at about 0.3 ms (fig. 1.1) (Steriade et al. 2001). Generally, RS, FRB and IB neurons are pyramidal-shaped neurons, while FS firing patterns are conventionally regarded as defining local GABAergic cells (GABA is γ-aminobutyric acid). However, in addition to pyramidal-shaped FRB neurons, other neurons, with the same FRB firing patterns, are local-circuit, sparsely spiny or aspiny interneurons (Steriade et al. 1998b). Some local inhibitory interneurons discharge like RS or bursting cells (Thomson et al. 1996). Thus, each of the aforementioned four firing patterns does not necessarily apply to a single of morphological class of neurons. Furthermore, the electrophysiological characteristics of different cortical neurons are flexible and a same neuron can change its firing pattern as a function of membrane potential, network activity and state of vigilance (Steriade et al. 1998b; Steriade et al. 2001; Timofeev et al. 2000a) and modulation of [Ca2+]o (see chapter 2). Intrinsic firing patterns enable a neuron to modify the input signal to a structured output pattern. Only fast-spiking neurons possess linear input-output characteristic. Below (see chapter 2) we will demonstrate that changes in [Ca2+]o affect intrinsic neuronal discharges.

In intact brain, the neuron receives several thousand synaptic contacts located throughout the dendrite and soma. The postsynaptic conductance change produced by neurotransmitters will act upon the many different voltage-gated channels in the dendrite and soma, and results in either postsynaptic subthreshold potential, or to different patterns of action potential generation (Pare and Lang 1998). The synaptic connectivity in the neocortex is very dense. Each pyramidal cell receives 5000 to 60000 synapses (Cragg 1967; DeFelipe and Farinas 1992). Local-circuit synapses have been estimated to account for as many as 70 % of the synapses present in some areas of the cortex (Gruner et al. 1974; Szentagothai 1965). Thus, this robust input influx should create tremendous synaptic bombardment onto postsynaptic neurons, affecting reliability of unitary responses (Crochet et al. 2005) or modulating theirs intrinsic properties, the rate of intrinsic bursting firing is much higher in isolated cortical territory compared to isolated large gyrus or intact cortex (Timofeev et al. 2000a; Timofeev et al. 2000b). Network activity during various functional states modifies the firing patterns generated by intrinsic neuronal properties (Steriade et al. 2001; Timofeev et al. 2000a). The long-lasting changes in activity influence the intrinsic excitability of cortical neurons (Topolnik et al. 2003a), this regulation is consistent with a role in stabilizing firing rates. Recent experimental work proves that preventing cortical neurons from firing for two days dramatically increased their intrinsic excitability (Cudmore and Turrigiano 2004; Desai et al. 1999). In response to injected current, activity-deprived neurons fired much more rapidly and did so in response to smaller current injections. This increase in excitability was mediated by selective regulation of the magnitudes of persistent sodium and potassium currents; the former increased, whereas the latter decreased. These findings demonstrate that the history of activity of a cortical neuron helps to determine its intrinsic excitability. This may allow a neuron to adjust the way it modifies synaptic input efficiency to maintain its responsiveness during periods of intense change in synapse number and strength (Desai et al. 1999)

The slow wave sleep is dominant by slow oscillations of EEG less than 1 Hz (generally 0.5-1 Hz). The slow oscillations are observed during the natural sleep in cat (Amzica and Steriade 1998a; Steriade et al. 1996; Steriade et al. 2001) and in human (Achermann and Borbely 1997; Amzica and Steriade 1997; Simon et al. 2000), just like under anesthesia with urethane, ketamine-xylazine, or nitrogen oxide (Contreras and Steriade 1995; Steriade et al. 1994a; Steriade et al. 1993a, b). The slow oscillation groups other sleep rhythms such as spindles and delta waves (Amzica and Steriade 1995b, 1998b; Contreras and Steriade 1995; Steriade and Amzica 1998). The cortical origin of slow oscillations was demonstrated by (a) its survival in the cerebral cortex after thalamectomy (Steriade et al. 1993a); (b) its absence in the thalamus of decorticated animals (Timofeev and Steriade 1996); (c) the disruption of its long-range synchronization after disconnection of intracortical synaptic linkages (Amzica and Steriade 1995a) and (d) its presence in isolated cortical preparations (Sanchez-Vives and McCormick 2000; Timofeev et al. 2000a). The sleeping cortex appears as an oscillator master imposing the slow rhythm to the thalamus and modulating activity originating in the thalamus (Steriade et al. 1994b). During slow oscillations, the membrane potential of cortical neuron alternates between depolarizing and hyperpolarizing phases (Steriade et al. 1993b). The neuronal depolarization is associated with a depth-negative EEG deflection, and the neuronal hyperpolarization is associated with a depth-positive EEG deflection (Contreras and Steriade 1995). The depolarizing phase corresponds to a general excitation of cortical neurons associated with excitatory postsynaptic potentials (EPSPs), inhibitory postsynaptic potentials (IPSPs), and action potentials (Steriade et al. 1993a, b). The long-lasting hyperpolarizing phase of the slow oscillation is associated with absence of firing of all cortical neurons, inducing a generalized disfacilitation in the cortical network (Contreras et al. 1996; Timofeev et al. 2001). The slow oscillation is accompanied: (1) by a progressive decrease of [Ca2+]o during the depolarizing phase with run down of synaptic efficacy (Crochet et al. 2005; Massimini and Amzica 2001), and (2) by the glial regulation of [K+]o, which likely modulates neuronal excitability (Amzica and Massimini 2002).

Calcium influx into cells or release of calcium from intracellular stores has a variety of consequences including mediator release from synaptic vesicles, activation of second messenger systems, gene transcription, and opening of calcium dependent ion channels. Important currents that control firing properties of neurons are calcium dependent potassium currents ( IC and IAHP ). They are largely activated following Ca2+ influx via voltage gated Ca2+channels during the action potentials and generating the afterhyperpolarization that follows action potentials.

In neurons, action potentials are followed by an afterhyperpolarization that may have three components. These have been called the fast afterhyperpolarization (fAHP), the medium AHP (mAHP), and the slow AHP (sAHP) (Sah and Faber 2002). The fAHP is activated immediately after the action potential and lasts several tens of milliseconds. The mAHP is also activated rapidly following the action potential (<5 ms) but decays with a time course of several hundred milliseconds. Finally, the third component of the AHP is the slow AHP, which rises to a peak over several hundred milliseconds; and can last up to 5 s following an action potential (Sah and Faber 2002). While in some neurons slow AHPs have been described following a single action potential (Hirst et al. 1985; Sah and McLachlan 1991), it is more commonly seen following a train (4–10) of spikes (Faber et al. 2001; Lancaster and Nicoll 1987; Schwindt et al. 1988). All three types of AHP are known to be mediated by calcium-activated potassium channels, which are activated in response to calcium influx via voltage dependent calcium channels that open during the action potential (Lancaster and Nicoll 1987; Storm 1987, 1990). The current underlying the fAHP has been named I c. This current is voltage dependent (Adams et al. 1982) and is blocked by low concentrations of TEA, iberiotoxin, and paxilline indicating that the underlying channels are BK-type channels (Adams et al. 1982; Lancaster and Nicoll 1987; Shao et al. 1999). The exact identity of these channels, however, has not been determined. In contrast, the mAHP is unaffected by BK channel blockers but is blocked by apamin indicating that it is due to the activation of SK-type channels (Pennefather et al. 1985; Sah and McLachlan 1991, 1992; Schwindt et al. 1988). The current that underlies the mAHP has been called I AHP (Adams et al. 1982). It peaks rapidly following calcium influx (<5 ms) and decays with a time constant of 50 to several hundred milliseconds (Pennefather 1988; Sah 1992). It is notable that while activation of BK channels generates the fAHP, these channels also contribute to action potential repolarisation. In contrast, activation of SK channels does not contribute to action potential repolarisation (Lancaster and Nicoll 1987; Sah 1996; Storm 1987, 1990). The current that underlies the slow AHP was first described in neurons in the myenteric plexus (Hirst et al. 1985). Following calcium influx, this current has a time to peak on the order of hundreds of milliseconds, and decays to baseline with a time constant of 1–2 s at 30 °C. To distinguish it from I AHP, this current has been designated I sAHP (Sah 1996). As with I AHP, this current requires a rise in cytosolic calcium for activation and is voltage insensitive. I sAHP is not blocked by apamin or TEA. However, I sAHP is modulated by a range of neurotransmitters including noradrenaline, serotonin, glutamate, and acetylcholine all of which block the current (Nicoll 1988).

As known, Ca2+ is a key factor in synaptic activity; the release of neurotransmitter from vesicles is dependent on Ca2+ entry induced by arriving of an action potential to the presynaptic terminal (Katz 1969). Ca2+ enters through clusters of channels near docked synaptic vesicles in active zones. This Ca2+ acts at extremely short distance (tens of nanometers) in short time (200 µs) and at very high local concentration (~100 µM), in calcium microdomains, by binding cooperatively to a low-affinity receptor with fast kinetics to trigger exocytosis. When Ca2+ channels close, these microdomains of high Ca2+ return to near resting concentrations quickly and the evoked response is terminated. Speed, efficiency, and flexibility are the hallmarks of this process (Macleod et al. 2004)

As mentioned above, the [Ca2+]o fluctuates as a function of neuronal activity, it decreases by approximately 20% during active states of the cortical network (Massimini and Amzica 2001). In the study Modulation of synaptic transmission in neocortex by network activities (Crochet et al. 2005) we investigated the effects of fluctuations of [Ca2+]o on synaptic transmission. We observed an increase in number of failures and a decrease in amplitude and duration of postsynaptic responses during active network states. To study the effects of fluctuations of [Ca2+]o on synaptic transmission, we performed two types of experiments. First , in ketamine-xylazine anesthetized cats we measured the modulation of response amplitude elicited by microstimulation and the parallel changes of [Ca2+]o (Fig. 1.2). The amplitude and the duration of responses during active network states were reduced as compared to the silent network states (Fig. 1.2, a). Measurements of individual responses demonstrated (a) a progressive increase in the amplitude of responses starting from the onset of silent state as estimated form the onset of depth-positive field potential and from the onset of neuronal hyperpolarization, (b) an abrupt decrease in the EPSPs’ amplitude at the onset of active state and (c) a recovery of the response amplitude as network goes back to silent states. The progressive increase in the EPSPs’ amplitude occurred in parallel with an increase in the [Ca2+]o (Fig. 1.2, b-c). The decrease in the EPSPs’ amplitude at the onset of active state was likely due to the decrease in the input resistance of neuron and due to the decrease in the [Ca2+]o. Since the time constant of Ca2+ electrodes was long (hundreds of milliseconds) the exact changes of the [Ca2+]o measured in these experiments were likely underestimated and the decrease in the [Ca2+]o could occur immediately at the onset of active state. Second , in order to provide further evidences of Ca2+ nature of high failure rates during active network states, we combined microstimulation with microdialysis of artificial cerebrospinal fluid (ACSF) containing different [Ca2+]o, and direct [Ca2+]o measurements with Ca2+ sensitive electrodes in barbiturate anesthetized cats (Fig.1.3a). Microdialysis of 1.0 mM of Ca2+ (control solution) yielded free [Ca2+]o of 1.1-1.2 mM; using the high Ca2+ solution (5 mM) raised [Ca2+]o to 2.5-3.0 mM and employment of Ca2+ free solution lowered [Ca2+]o to 0.7-0.8 mM. Microstimulation evoked responses (n=7) were compared for the three conditions when [Ca2+]o reached steady levels. The total averaged response was increased in high calcium condition and decreased in low calcium condition (Fig. 1.3b). As shown by histograms of response amplitude (Fig. 1.3c), both the amplitude of successful responses and the failure rate were affected by changes in [Ca2+]o. In control condition the mean amplitude of successful EPSPs was 0.83±0.14 mV and the overall failure rate was 54±12 %. Raising [Ca2+]o increased the amplitude of EPSPs to 1.23±0.18 mV, and decreased failure rates to 29±11 %. Lowering [Ca2+]o reduced the amplitude of EPSPs to 0.49±0.06 mV and increased failure rate to 76±6 %. Paired t-test revealed that these differences were significant at p<0.05. In addition, on some occasions (n=7) we observed bimodal distribution of histograms of successful responses (see control in Fig. 1.3c). Invariantly, the second peak of histogram increased in high [Ca2+]o conditions and it was abolished in low [Ca2+]o conditions. These data suggests that presynaptic stimuli in control conditions activated one or several release sites. Increase in [Ca2+]o increased probability of simultaneous activation of two release sites and decrease in [Ca2+]o allowed activation of maximum one release site per stimulus (Crochet et al. 2005).

The adage "sleep and epilepsy are common bedfellow" is supported by much clinical and experimental evidence showing that epileptic seizures of different types preferentially occur during slow-wave sleep (SWS or non-REM). It is considered the possibility that many spontaneous electrographic seizures in "normal" subjects may not be recognized, and that those sleeping individuals pass in and out of seizures during slow sleep oscillations. Needless to say, these are electrical seizures and it is refrain from using the term epilepsies (Timofeev and Steriade 2004)

The electrographic pattern of seizures accompanying Lennox-Gastault syndrome as well as their occurrence during slow-wave sleep described below in animal model, resemble the clinical Lennox-Gastaut syndrome of humans (Halasz 1991; Kotagal 1995; Niedermeyer 1999a, b). These seizures are generated intracortically and spontaneously without discontinuity from the sleep-like slow oscillations, they are characterized by SW or spike-wave/polyspike-wave (SW/PSW) complexes of 1.5-3 Hz, intermingled with episodes of fast runs at ~7-16 Hz (Fig.1.4) (Neckelmann et al. 1998; Steriade et al. 1998a; Steriade and Contreras 1998; Timofeev et al. 1998). The intracortical origin suggested because the majority of thalamocortical neurons do fire spikes during these seizures (Steriade and Contreras 1995; Timofeev et al. 1998) and similar seizures are recorded in cats with large with matter undercuts (Topolnik et al. 2003b) or could be elicited in isolated neocortical slabs (Timofeev et al. 1998). During the initial part of the seizure, SW/PSW complexes characterize the EEG recording, while, at neuronal level, the slope as well as the amplitude of the shift in the membrane potential from hyperpolarized to depolarized levels increased compared to the pre-seizure epoch. The action potentials were partially inactivated and AHP disappeared during paroxysmal depolarizing shift (PDS). The following period, with fast runs, was characterized by EEG "spikes", at the neuronal level, a tonically depolarized membrane potential, with smaller depolarization superimposed. The synchrony between cortical neuronal pools during fast runs is impaired that suggests their focal generation; (see chapter III). The end of the seizure was associated with a short period of hyperpolarization, followed by resumption of the sleep-like slow oscillation (Steriade et al. 1998a; Timofeev and Steriade 2004).

The modulation of neuronal excitability by many factors is responsible of shift from normal to paroxysmal activity. Various neuronal ionic currents contribute to the increase in the susceptibility to seizures. The discovery of voltage-dependent Na+ and Ca2+ channels in dendrites changed the model of dendrites with only passive properties and demonstrated that different intrinsic currents can amplify synaptic signals (Schwindt and Crill 1995), which may eventually lead to abnormal cellular excitation and paroxysmal discharges (Ward and Schmidt 1961). During cortically driven seizures (Timofeev and Steriade 2004), the persistent Na+ current [ I Na(p)] and probably the high-threshold Ca2+ current contribute to the generation of paroxysmal depolarizing shifts (PDSs). Ca2+-activated K+ current [ I K(Ca)] take also part in the control of the amplitude and duration of PDSs. The hyperpolarizing components of seizures largely depended on Cs+-sensitive K+ currents. I K(Ca) plays a significant, while not exclusive, role in the mediation of hyperpolarizing potentials related to EEG "waves" during spike-wave seizures (Timofeev et al. 2004). Recent intracellular studies of neocortical neurons in vivo (Timofeev et al. 2002b) as well as work on human and rat slices (Cohen et al. 2002; Fujiwara-Tsukamoto et al. 2003) demonstrate that PDSs contain an important inhibitory component. The hyperplarization activated depolarizing current plays a critical role in the generation of subsequential PDSs (Timofeev et al. 2002a). The intrinsic propensity of some neocortical neurons to bursting is also a factor that predisposes to seizures. The FRB neurons fire high-frequency (approximately 300–400 Hz) spike-bursts at fast rates (approximately 30–40 Hz), they have the highest propensity to drive seizures. This is coupled with the crucial role played by FRB neurons in the generation of ultra-fast oscillations, called ripples (80–200 Hz), which initiate seizures in both animal experiments (Grenier et al. 2003) and humans (Fisher et al. 1992). The firing frequency of IB (8 Hz) and number of theirs spikes fired during fast runs suggest their role in generation of fast runs (Fig. 3.6) (see chapter III). It is widely accepted that the development epileptiform activity results from a shift in the balance between excitation and inhibition toward excitation (Dichter and Ayala 1987; Galarreta and Hestrin 1998; Nelson and Turrigiano 1998; Tasker and Dudek 1991). Other two factors (increased inhibition and decreased excitation) also may lead to functional unbalance in cortical circuits; however, their role in seizure initiation was poorly explored. By increased "inhibition" we mean either increased number of IPSPs or disfacilitation, i.e. temporal absence of network activity. By decreased excitation we mean significantly decreased activity in afferent structures or decreased effectiveness of EPSPs. If one of the above factors occurs, it would create conditions that are favorable to seizure generation. One condition for seizure generation is the functional heterogeneity of cortical networks, such as the presence of two or more different cortical regions with relatively high and low levels of synaptic activity. Transitory or persistent reduction in synaptic activity within some cortical foci would increase the sensitivity of cortical neurons in those foci and in surrounding areas (Abbott et al. 1997; Desai et al. 1999) and the synaptic inputs from cortical regions exhibiting moderate or high levels of activity would lead to an increased responsiveness in those cortical areas.

Thus, a mosaic of intrinsic cell properties, changes in extracellular ion concentrations (K+, Ca2+) (Heinemann et al. 1977), as well as intracellular acidification (Xiong et al. 2000), may play a role in the induction and/or duration of seizures. Since seizure activities are regarded as hypersynchronous states, but the synaptic transmission during seizures is impaired due to low [Ca2+]o, we hypothesized that (a) the low [Ca2+]o enhance intrinsic neuronal responsiveness and (b) the local synchronization during seizures is also impaired. These issues were investigated experimentally and the results are presented in chapters 2 & 3.

Figure 1.1

Figure 1.1: Electrophysiological identification of different cell classes in neocortex. Chronically implanted, awake cat. Left column depicts responses of regular-spiking (RS), fast-rhythmic-bursting (FRB), fast-spiking (FS) and intrinsically bursting (IB) neurons from area 4 to depolarizing current pulses p(0.2s, 0.8 nA). At the right of each depolarizing current pulse, action potentials of each cell are shown; note thin spikes of FRB and FS neurons, compared to those of RS and IB neurons. Right column illustrates the width of action potentials (at half amplitude) in a sample of 117 neurons (48 RS, 37 FRB, 24 FS and 8 IB, corresponding to patterns depicted on the left). (Steriade et al. 2001).

Figure 1.2

Figure 1.2: Activity dependent modulation of responses elicited by microstimulation.

a. A period of spontaneous activity in neocortex in ketamine-xylazine unesthetized cat (upper panel) and averaged responses to microstimulation during active and silent network states (lower panel). Arrowheads indicate the time of stimulation. b. Wave-triggered average of EEG, intracellular activities and [Ca2+]o as well as amplitude of intracellular response to microstimuli applied during different phases of slow oscillation. The first maximum of EEG-depth negativity was taken as 0 time. c. Averaged amplitude of responses to microstimuli from 9 neurons during different phases of slow oscillation. The amplitude was averaged in following time windows: -800 to -600 ms, -600 to -400 ms, -400 to -200 ms, -200 to 0 ms, 0 to 200 ms, 200 to 400 ms, 400 to 600 ms and 600 to 800 ms. The time base in b and c is the same. (Crochet et al. 2005).

Figure 1.3

Figure 1.3: Modulation of microstimulation induced EPSPs by changes in [Ca2+]o.

a. Set up for intracellular recordings, microstimulation, microdialysis, direct Ca2+ and EEG measurements. b. Averaged responses in the conditions of different [Ca2+]o. Red, high [Ca2+]o; blue normal [Ca2+]o and green low [Ca2+]o. Microstimuli induced a single EPSP. High [Ca2+]o resulted in increased amplitude of the response, whereas low [Ca2+]o decreased the response. c. Histograms showing for each condition the distribution of the amplitude of the background synaptic noise (filled gray), the responses (failure excluded, open bars) and the failures (thick line). Color-code same as part b. Note the distribution of noise that become wider in high [Ca2+]o and narrower in low [Ca2+]o. d. Measured changes in the [Ca2+]o. (Crochet et al. 2005).

Figure 1.4

Figure 1.4: Fast-rhythmic-bursting (FRB) cortical neuron during SW and fast seizure developing spontaneously from slow sleep oscillation.

A: intracellular and depth-EEG recording from area 7. Seizure is indicated by arrows (below the EEG trace) and lasted for ~26 s. B: electrophysiological identification by depolarizing current pulse (0.2 s). See text. Epochs C and D in A are expanded below. C: slow oscillation before the seizure. D: both fast runs (~12 Hz) and PSW complexes (~2 Hz). (Steriade et al. 1998a)

© Soufiane Boucetta, 2005