JLE

Epileptic Disorders

MENU

Relationship between oscillations in the basal ganglia and synchronization of cortical activity Volume 4, supplément 3, Supplement 3, December 2002

The basal ganglia constitute a network of interconnected subcortical structures, which provides a major integrative system of the forebrain (see Slaght et al.; Lehéricy and Gerardin in this issue). It has become clear that basal ganglia neurons can discharge in a rhythmic, oscillatory-burst mode and that this bursting activity plays a role in facilitating changes in the functional state of the brain and, more specifically, the cortex. Oscillatory discharge in subcortical regions, conveyed to the cortex via the thalamocortical network may have powerful effects on cortical processing [1]. Intracerebral recordings from parkinsonian patients treated by deep brain stimulation, have demonstrated the role of basal ganglia oscillations in the generation of tremors.

The functions of oscillations within the basal ganglia are poorly understood [2]. As the basal ganglia loops are involved in motor control, some oscillations may have something to do with movement. It is tempting to think that these oscillations may play a role in a particular oscillatory involuntary movement such as tremor [3-5]. However, since oscillations occur in basal ganglia even in the absence of tremor, they may participate in the physiology of normal, voluntary movement.

In the present paper, we will discuss the possible physiological or pathological roles of oscillatory activities within basal ganglia and their relationship with cortical oscillations. In the first part, we present recent data from animal studies that led to hypotheses about the generation of neuronal oscillations in the basal ganglia and their effect on cortical activity. In the second part, the different lines of evidence favouring the involvement of basal ganglia oscillations in involuntary as well as voluntary movements will be reviewed. Finally, the influence of a basal ganglia dysfunction on cortical oscillatory activity will be discussed with reference to a study that investigated the effect of high-frequency stimulation (HFS) of the internal globus pallidus (GPI) and the subthalamic nucleus (STN) in parkinsonian patients, on the reactivity of cortical rhythms related to motor programming.

Animal studies

Several animal studies, mainly performed in rodents have described neuronal oscillations in the basal ganglia. Oscillatory discharges have been directly recorded in cortico- striatal neurons [6], striatal interneurons [7], the substantia nigra pars compacta (SNc) [8], external globus pallidus (GPE) [9], GPI and STN [10].

Basal ganglia oscillatory pacemaker

It has been shown, that the excitatory influences of the STN and the inhibitory effects of GPE form a feedback system that generates synchronized bursting [11]. In this study, recordings from organotypic cortex-striatum-STNGPE cultures were performed and the results demonstrated that neurons in STN and GPE spontaneously produce synchronized oscillating bursts, which are abolished after pallidal lesions. Conversely, a cortical lesion favoured bursting activity at 0.8Hz. Plenz and Kital proposed that the STN and GPE constitute a central pacemaker modulated by striatal inhibition of GPE neurons.

Relationship between cortical and basal ganglia oscillations

Another study demonstrated that, in healthy animals, oscillatory activity of striatal neurons are synchronized by cortical activity [12]. In this study, the relationship between cortical activity and the activity of cortico-striatal neurons (C-S) and striatal output neurons (SONs) was assessed in rats by recording in vivo, both the electroencephalographic (EEG) activity and the intracellular activity of C-S or SONs. Using three different types of anesthesia (barbiturate, ketamine and fentanyl) that induced different patterns of cortical activity, the results provided experimental evidence that a certain degree of synchrony in C-S neurons is required for significant depolarization in SONs. Furthermore, the authors demonstrated that neuronal oscillation inside the striatum was dependent on the level of cortical synchronization (see also Slaght et al. in this issue). In another study [13], simultaneous recordings of the frontal electrocorticogram and membrane potential of SONs confirmed that fluctuations of membrane potential in striatal neurons are correlated with slow oscillatory activity in the frontal cortex under barbiturate anesthesia. Conversely, neurons of the globus pallidus (GP) and substantia nigra reticulata (SNR) continue to display regular tonic firing without any correlation with the synchronization of cortical activity.

Increase of basal ganglia oscillation under dopamine depletion

Several hypotheses about the physiological significance of oscillatory discharges in the basal ganglia remain unclear. However, the role of oscillatory discharges is better defined under pathological conditions such as dopaminergic depletion, as in animal models of Parkinson's disease. In the case of chronic nigrostriatal lesions, SNR neurons display rhythmic bursting activity [14, 15] that is strongly modulated by intrastriatal administration of dopamine receptor agonists [16]. This suggests that after chronic nigrostriatal lesions, an increased excitability of striatal neurons allows transmission of slow cortical rhythms to striatal target nuclei (figure 1). Another study demonstrated that oscillatory activity in the STN-GP network in anesthetized rats is phase-locked to rhythmic cortical activity [17]. The activity of STN neurons is correlated with concomitant cortical slow wave activity in healthy animals and 6-hydroxydopamine-(6-OHDA) lesioned animals. However, the removal of the ipsilateral cortex abolished low-frequency oscillation in STN neurons. Interestingly, the intensity of oscillatory activity was greater following dopamine depletion [18]. Moreover, the low frequency oscillation was observed in GP only in 6-OHDA-lesioned animals. These results indicate that following dopamine depletion, rhythmic cortical activity is propagated to the GP, suggesting that dopamine regulates the impact of the cerebral cortex on the STN-GP network (figure 2).

Altogether, animal studies suggest that neuronal oscillations are spontaneously generated within the basal ganglia system, especially from the GPE and STN, but are mainly synchronized by cortical activity via the striatal inputs. Dopamine depletion results in a global increase of oscillations within the whole basal ganglia system, particularly in the GP-NST network.

Basal ganglia oscillations and movement

In this chapter, the term ÒmovementÓ encompasses physiological, voluntary movements as well as abnormal, involuntary movements. Among the latter we will just refer to tremor because of its oscillatory nature. Regarding the physiology of movement, the relationship with basal ganglia oscillations are generally assessed by coherence analyses. Both aspects will be considered.

Basal ganglia oscillations and tremor

Thalamus

Studies dealing with the activity of basal ganglia neurons have been closely related to the surgical treatment of tremor, whether parkinsonian or debilitating essential tremor [19]. The first deep structure to be studied was the thalamus. Although this structure is usually not considered to be part of the basal ganglia, it is interesting to note that the destruction of the ventralis intermedius nucleus (Vim), which was the first surgical technique to alleviate tremor (not only of parkinsonian origin), allowed tremor-related thalamic neuronal activity to be recorded. About 33% of thalamic Vim neurons discharge in an oscillatory way, coherent with the tremor frequency [20, 21]. Some of these Òtremor-cellsÓ were sensitive to movement and sensory inputs, so they may detect the mechanical oscillation in the periphery. Others were not sensory cells, and have been supposed to contribute to the genesis of tremors.

Globus pallidus internus

When pallidotomy or GPI stimulation again became of interest in the treatment of Parkinson's disease, human data about GPI neurons became available [22, 23]. About 12-25% of GPI neurons discharge at tremor frequency [24]. However, their discharge frequency is almost never coherent with tremor frequency, so the primary role of the thalamus in tremor genesis was reinforced. Studies in nonhuman primates have the advantage that neurons can be studied before and after 1-methyl-4-phenyl-1,2,3,6-tetrahydro- pyridine (MPTP), a drug used to induce a parkinsonian syndrome, to highlight the effects of dopaminergic depletion. Administration of MPTP in monkeys resulted in (i) the appearance of correlated discharges in GPI neurons, never seen in normal monkeys, (ii) an increase of burst discharges and (iii) an increase in oscillatory discharges (figure 3) [25, 26]. About 30% of GPI neurons oscillated at 4-8Hz, and 12% oscillated at 10 (8-20) Hz, whereas no such oscillatory neurons were recorded before MPTP.

Subthalamic nucleus

During the last ten years, the STN has become a key structure in the physiology of the basal ganglia circuitry, and also a target in the surgical treatment of parkinsonian patients. In parkinsonian patients, as well as in monkeys rendered parkinsonian by MPTP, there was a dramatic increase in burst discharges and oscillatory discharges in STN neurons (figure 4) [27-31]. All these studies suggest that at least some oscillations within the basal ganglia may be a pathological behaviour. Indeed, the main effect of dopaminergic depletion is an increase in the oscillatory pattern of neuronal activities.

Basal ganglia oscillations and the physiology of movement

The role played by the basal ganglia in the physiology of movement is not yet fully understood [32]. The analysis of basal ganglia neuronal activity requires microelectrode recordings, which can, in humans, only be performed during the surgical treatment of Parkinson's disease. Recordings through implanted chronic macroelectrodes only detect local field potentials and consequently merely allow the analysis of oscillations within the target structure. The physiology of movement has been most often studied through oscillations within the cortex (using EEG or MEG), and within the muscles since motor units usually discharge rhythmically. Therefore, we will first review the main results of studies dealing with cortico-muscular relationships, and then more specific studies in which oscillations within the basal ganglia were recorded. All these studies used coherence analysis to demonstrate these relationships, which is a normalised measure of the linear correlation between two signals in a given frequency. A value of 0 means that there is no correlation, a value of 1 means that both signals are linearly correlated at this frequency. When this is the case, phases may be studied in order to determine which signal leads or lags the other and by what delay. One must be aware that there is no direct link between the frequency of an oscillation and the rate of discharge of the neurons that contributes to this oscillation: such an oscillation is the result of a population effect. Action potentials of different neurons occur at specific times and not at random. As an example, neurons discharging at 20 Hz may generate a 10 Hz oscillation and, conversely, a 20 Hz oscillation may originate from neurons spiking at 10 Hz.

Corticomuscular coherence and the influence of basal ganglia

Piper rhythm

The Piper rhythm is a muscle oscillation in the gamma band, around 40 Hz, caused by motor unit synchronization mainly during powerful contractions. This rhythm, first described about 100 years ago by Piper [33], originates in the primary motor cortex. Indeed, a significant coherence at 35-60 Hz was found between the Piper rhythm and MEG oscillations within the contralateral primary motor cortex with a phase delay compatible with a propagation from cortex to muscle via fast corticospinal pathways (figure 5) [34, 35]. The Piper rhythm is typically present during maximal and submaximal, tonic and also phasic contractions. Though generated by the motor cortex, it is sensitive to basal ganglia dysfunction since it is lost in untreated parkinsonian patients [36]. Indeed, the power of the Piper rhythm was shown to be markedly decreased or even lost in five untreated parkinsonian patients who also exhibited a decrease in the exerted force. Interestingly, both abnormalities were reversed by levodopa.

Beta rhythms

Muscles and cortex may also be coherent within the beta 15-30 Hz band as demonstrated in animal as well as in human studies. In humans, cortico-muscular coherence was found using either MEG [37, 38] or EEG [39] signals during weak, tonic contractions, but never during movement. Non-human primates exhibited such a coherence during the maintenance of a precision grip task [40]. Consequently, coherence within the beta band appears to be a way for the brain to maintain a stable motor output at least expense. The most often cited theory is the Òbinding theoryÓ, according to which coherence would serve to bind together two distant elements involved in the same (namely motor in this particular case) function [41]. Therefore, it is possible to find EEG-EEG coherence between two distant cortical areas involved in the same motor task [42-45]. In the same way, the finding of an EMG-EMG coherence reveals the existence of a link between two muscles as well as the common influence of the motor cortex on them [46-48].

Influence of the basal ganglia

What role do the basal ganglia play in these different types of cortico-muscular coherence? To address this question, it is interesting to know how cortico-muscular synchronization behaves in Parkinson's disease? In PD patients, the coherence in the beta band is decreased, as well as in the gamma band as mentioned above. These fast oscillations are replaced by 10 Hz oscillations, presumably resulting in the appearance of an action tremor, and are associated with a decrease in the generated force due to unfused motor unit contractions and bradykinesia. Interestingly, all these abnormalities are reversed by medical (levodopa) as well as surgical (high frequency STN stimulation) therapies (figure 6).

Coherence between cortex,muscles and basal ganglia

Although the previous studies only analysed coherence between cortex and muscles, the influence of the basal ganglia could be inferred from results obtained in parkinsonian patients. Nevertheless, a few studies examined the relationships between cortex, muscles and some deep structures within the basal ganglia.

Thalamus

Among seven patients whose Vim was targeted for deep brain stimulation to alleviate tremor, coherence in the 8-27 Hz range was found between cerebellar thalamic LFPs and the ipsilateral EEG or the contralateral EMG of distal upper limbs muscles in the three patients with an isolated tremor and 2/4 patients with an associated cerebellar syndrome. Phase analysis revealed that oscillatory activity in the Òcerebellar thalamusÓ lagged behind that in both the cortex and the muscles. This can be explained by a more rapid and direct pathway between the cortex and muscle than between the cortex and thalamus, presumably via the nucleus pontis and the cerebellum [49]. No coherence was found above 30 Hz, e.g. in the gamma band. It should be kept in mind that the Vim has no direct relationship with the basal ganglia, and is rather involved in the cerebellar system and the cortico-ponto-cerebellothalamo- cortical loop.

Subthalamic nucleus

The same group has performed a similar study in a group of ten parkinsonian patients who underwent an STN stimulation [50]. They found a significant coherence between STN LFPs and the ipsilateral EEG as well as the contralateral EMG over a 7-45 Hz range (figure 7). Therefore, the sensorimotor cortex, muscles and STN are coherent not only in the beta band, just like the cerebellar thalamus, but also in the gamma band, which suggests that there might be some frequency selectivity within different structures involved in motor control. The analysis of phase relationships is more complex compared to the previous study with the Vim. The cortex led the STN with a delay that is rather compatible with an indirect corticosubthalamic route. The EMG lagged behind the STN with time differences tending to cluster around two different values, namely 6 and 47 ms. The latter may correspond to a peripheral feedback from muscle while the former presumably refers to some sort of corollary discharge (central feedback) from the cortex to STN.

Globus pallidus internus and subthalamic nucleus

A step further was to look at coherence between different structures within the basal ganglia. This was done by recording LFPs simultaneously from GPI and STN in four patients with Parkinson's disease with and without dopaminergic stimulation [51, 52]. In off-drug conditions, at rest, STN and GPI were coherent at 6 Hz and at around 20 Hz. Coherence still persisted, though lower, during tonic contractions. After levodopa, the frequencies of oscillations below 30 Hz were greatly reduced, while a new frequency peak emerged at 70 Hz. Similarly, coherence at 20 Hz was strikingly decreased, while coherence at 70 Hz was increased. This was the case at rest as well as during tonic muscular activation, with a stronger coherence at rest [51]. Consequently, the level of dopaminergic stimulation appears to be critical for the expression of oscillations within the basal ganglia. When dopamine is lacking, as in untreated Parkinson's disease, frequencies below 30 Hz seem to be favoured in subthalamo-pallidal networks. These frequencies may be related either to tremor, especially oscillations at 4-10 Hz, or to some aspects of motor impairment such as bradykinesia. With appropriate dopaminergic stimulation, this subthalamo-pallidal network seems to oscillate at a higher frequency of around 70 Hz [51, 52]. This rhythm appears to be important for, but not directly related to, the organization of voluntary movement since it also exists at rest. Therefore, it could be linked to movement-related executive attention, or could act as a Òcarrier frequencyÓ necessary for motor commands to be executed [51]. These findings may explain the paradoxical results of functional neurosurgery in Parkinson's disease. In particular, although GPI lesions are expected to impair movement by destroying the major output of the basal ganglia, they lead to motor improvement in parkinsonian patients. To explain such a result, the authors put forward the hypothesis of two systems within the basal ganglia: a Òlow-frequency systemÓ and a Òhigh-frequency systemÓ. The Òlow-frequency systemÓ would impair movement, and would be blocked either by dopaminergic stimulation or focal destruction of GPI or STN, thus explaining the good results of pallidotomy for example. The Òhigh-frequency systemÓ would promote movement, and would be artificially enhanced by high frequency stimulation of either nucleus. Consequently deep brain lesioning and deep brain stimulation would act through different mechanisms.

Most previous studies analysed coherence between cortex, muscles and basal ganglia during ongoing tonic or sometimes phasic contractions. However, they gave no information about the influence of movement on these oscillations. Basal ganglia oscillations within the beta band were recently demonstrated to be reduced by voluntary movement [53]. Some oscillations may even be reduced before movement onset, thus suggesting their involvement in movement preparation (figure 8) [54].

Influence of basal ganglia on cortical oscillatory activity

Reactivity of cortical oscillations

The cortex generates spontaneous oscillatory activity that can be modified by changes in cortical activation. It has been shown that a decrease in amplitude of cortical oscillations corresponds to the activation of sub-cortical structures [55], while an increase in amplitude of cortical oscillations may be related to a resting of sub-cortical areas [56, 57] or to somesthetic inputs [58]. These changes in cortical oscillations can be quantified by the method of event-related desynchronization (ERD) or synchronization (ERS).

According to previous studies [59, 60], spatio-temporal recruitment of cortical areas related to a movement is better assessed using the analysis of ERD/ERS of cortical rhythms. Therefore, the reactivity of µ and ß EEG rhythms during preparation and execution of voluntary movement has been related to sensorimotor cortex activity (see also Lehéricy and Gerardin in this issue). In a healthy subject, a contralateral desynchronization of µ rhythm appears nearly 2 s before movement onset, during the preparation phase, over the central area covering the primary sensorimotor (PSM). During the execution of the movement, this desynchronization is recorded bilaterally over the central regions. At the end of the movement, an overall synchronization of cortical rhythms has been described [56, 61]. Since the basal ganglia circuits participate to the control of activation of the motor cortex during the planning of voluntary movement, we wanted to know how this could influence the reactivity of cortical oscillatory activity.

Influence of dysfunction of basal ganglia on the reactivity of cortical rhythms

The exploration of the reactivity of cortical rhythms in parkinsonian patients has revealed an abnormal spatiotemporal pattern of desynchronization of central cortical rhythms [62]. A delay in the appearance of premovement desynchronization over the primary sensorimotor cortex was observed in PD patients compared with controls, mainly when the movement was executed with the more akinetic hand [63]. Study of ERD of µ rhythm allows detection of the motor programming delay at the early stage of the disease in untreated hemiparkinsonians [64]. Furthermore, this delay of ERD is partially corrected both by chronic [64] and acute administration of L-dopa [65].

Influence of basal ganglia stimulation on cortical oscillations

Inhibition by high frequency electrical stimulation of the hyperactive nuclei (GPI and STN) has recently been proposed in the treatment of Parkinson's disease. This hyperactivity is sustained by enhanced glutamatergic inputs from the disinhibitory STN, leading in turn to hyperactivity of the output nuclei of a circuit which includes the GPI and the substantia nigra pars reticulata [66].

High frequency stimulation of the internal globus pallidus (HF-GPI) has previously been used in order to reduce the hyperactivity of the GPI-STN network in advanced Parkinson disease [67], with a good clinical outcome, by reducing dyskinesia by 90% [68].

Influence of GPI stimulation on cortical oscillations

To better understand the relationships between basal ganglia and cortical oscillations, we evaluated the influence of such intracerebral stimulation within basal ganglia, on the reactivity of cortical oscillations related to voluntary movements. We studied six patients with PD, treated by bilateral posteroventral GPI stimulation because of Ldopa- induced, severe dyskinesia.

Clinical evaluation and electroencephalogram recordings were performed in 4 sets of conditions for each subject, and in the following order:

1) without any stimulation or L-dopa for at least the previous 15 hours (the condition was designated Òoff stimÓ);

2) with stimulation and without L-dopa (designated Òon stimÓ);

3) without stimulation and after an acute administration of L-dopa (designated Òon drugÓ), where the L-dopa dose was the usual, first morning dose used by each patient prior to the date of surgery to relieve their symptoms;

4) with stimulation and after an acute administration of L-dopa (designated Òbest onÓ). The Òbest onÓ condition was defined as the best parkinsonian triad control without (or at most with mild) dyskinesia.

For each patient, we recorded an EEG during the performance of repeated voluntary movements under the four conditions (Òoff stimÓ, Òon stimÓ, Òon drugÓ and Òbest onÓ). We could then compute the ERD of the µ rhythm related to the movement. The results are illustrated in figure 9 (GPI-HFS) and figure 10 (STN-HFS).

GPI stimulation influenced the premotor cortex activation before the movement, and induced a selective and focal effect on contralateral PSM cortex activation during the movement execution (figure 9). Premovement desynchronization and desynchronization latency over central contralateral derivations were not improved under GPI stimulation alone, despite a clinical improvement. When GPI stimulation was associated with L-dopa, the contralateral PSM cortex activation was improved during the preparation and the execution phases of the movement. Central contralateral movement desynchronization was positively correlated with the UPDRS motor ratings and the bradykinesia subscore improvement. Consequently, it can be suggested that, in advanced PD patients, L-dopa associated with GPI stimulation may partially restore cortical activation during the preparation phase of the movement [69]. It was not possible to assess the L-dopa effect alone because of artefacts induced by dyskinesias. The lack of improvement of premovement desynchronization by GPI stimulation in the absence of L-dopa seems to indicate that GPI does not directly influence cortical oscillatory activity.

Influence of STN stimulation on cortical oscillations

STN-HFS is currently preferred to GPI-HFS because of its greater clinical efficiency, with UPDRS motor score decreases ranging from 41% to 74% [68, 70]. We have explored the reactivity of cortical rhythms during planning and execution of voluntary movement in ten PD patients treated with STN-HFS.

Clinical evaluation and electroencephalogram recordings were performed under the same four sets of conditions for each subject: Òoff stimÓ, Òon stimÓ, Òon drugÓ, Òbest onÓ.

In the Òoff stimÓ condition, there was an abnormal pattern of desynchronization with a delay over the contralateral sensorimotor region. STN stimulation alone restored a nearly normal pattern of cortical activation by i) increasing contralateral primary sensorimotor cortex activation during movement preparation and execution and ii) decreasing the pre-movement desynchronization diffusion over the bilateral premotor regions (figure 10). Moreover, sensorimotor cortex oscillatory activity changes were correlated with clinical improvement. STN-HFS seemed to have a dopamimetic-like effect on cortical desynchronization of the µ rhythm.

As compared to STN stimulation, GPI stimulation seemed to induce a more restricted temporal effect on the sensorimotor cortex during movement. The lack of an enhancing effect of GPI stimulation on the sensorimoror cortex during the preparation phase of a distal movement was also found in PET studies (see Lehéricy and Gerardin in this issue). The association of GPI stimulation and L-dopa led to a more limited temporal and spatial improvement of the sensorimotor cortex oscillatory activity than did the association of L-dopa and STN stimulation. We observed good concordance between sensorimotor cortex oscillatory activity and the clinical outcomes under STN stimulation and GPI stimulation.

Altogether, these studies of cortical reactivity related to planning of voluntary movement in parkinsonian patients provide evidence that it is possible to influence cortical reactivity through the basal ganglia system. High-frequency stimulation within basal ganglia circuits influence the reactivity of rhythms within the motor cortex. STN stimulation partly restores a normal pattern of desynchronization of cortical µ rhythm, while the influence of GPI stimulation seems to be less. Moreover, the effect of STN or GPI stimulation is best under L-dopa treatment, which suggests that dopaminergic inputs are involved in the control of cortical oscillatory activity by the basal ganglia system.

CONCLUSION

Taken together, animal studies and human neurophysiological investigations indicate that neuronal oscillations within the basal ganglia have a close relationship with cortical oscillatory activities. In animals, dopamine depletion induces an increase in neuronal bursting activity within the basal ganglia that is dependent on cortical oscillations. In humans, dopamine depletion related to Parkinson's disease induces an abnormal reactivity of cortical rhythms within the motor cortex that is improved by L-dopa and influenced by STN and GPI stimulation. Basal ganglia oscillations may have differential effects with respect to their frequency. Low frequency basal ganglia oscillations are related to tremor, bradykinesia or low force. High frequency basal ganglia oscillations, especially around 70 Hz, only exist when dopamine is present and are important for motor programs to be correctly executed. All these data demonstrate that basal ganglia and cortical oscillations are influenced by each other. Most importantly, changes in basal ganglia oscillations will result in modifications of oscillations in the cortical network.