Despite several studies and models, much remains unclear about how the human basal ganglia operate. Deep brain stimulation (DBS) of the subthalamic nucleus (STN) is an effective treatment for complicated Parkinson's disease, but how DBS acts also remains unknown. The clinical benefit of DBS at frequencies >100 Hz suggests the possible importance of neural rhythms operating at frequencies higher than the range normally considered for basal ganglia processing (<100 Hz). The electrodes implanted for DBS also offer the opportunity to record neural activity from the human basal ganglia. This study aimed to assess whether oscillations at frequencies >100 Hz operate in the human STN. While recording local field potentials from the STN of nine patients with Parkinson's disease through DBS electrodes, we found a dopamine- and movement-dependent 300-Hz rhythm. At rest, and in the absence of dopaminergic medication, in most cases (eight out of 11 nuclei) the 100-1000 Hz band showed no consistent rhythm. Levodopa administration elicited (or markedly increased) a 300-Hz rhythm at rest [(mean +/- SD) central frequency: 319 +/- 33 Hz; bandwidth: 72 +/- 21 Hz; power increase (after medication - before medication)/before medication: 1.30 +/- 1.25; n = 11, P = 0.00098]. The 300-Hz rhythm was also increased by apomorphine, but not by orphenadrine. The 300-Hz rhythm was modulated by voluntary movement. Before levodopa administration, movement-related power increase in the 300-Hz rhythm was variably present in different subjects, whereas after levodopa it became a robust phenomenon [before 0.014 +/- 0.014 arbitrary units (AU), after 0.178 +/- 0.339 AU; n = 8, P = 0.0078]. The dopamine-dependent 300-Hz rhythm probably reflects a bistable compound nuclear activity and supports high-resolution information processing in the basal ganglia circuit. An absent 300-Hz subthalamic rhythm could be a pathophysiological clue in Parkinson's disease. The 300-Hz rhythm also provides the rationale for an excitatory--and not only inhibitory--interpretation of DBS mechanism of action in humans.

The basic information architecture in the basal ganglia circuit is under debate. Whereas anatomical studies quantify extensive convergence/divergence patterns in the circuit, suggesting an information sharing scheme, neurophysiological studies report an absence of linear correlation between single neurones in normal animals, suggesting a segregated parallel processing scheme. In 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-treated monkeys and in parkinsonian patients single neurones become linearly correlated, thus leading to a loss of segregation between neurones. Here we propose a possible integrative solution to this debate, by extending the concept of functional segregation from the cellular level to the network level. To this end, we recorded local field potentials (LFPs) from electrodes implanted for deep brain stimulation (DBS) in the subthalamic nucleus (STN) of parkinsonian patients. By applying bispectral analysis, we found that in the absence of dopamine stimulation STN LFP rhythms became non-linearly correlated, thus leading to a loss of segregation between rhythms. Non-linear correlation was particularly consistent between the low-beta rhythm (13-20 Hz) and the high-beta rhythm (20-35 Hz). Levodopa administration significantly decreased these non-linear correlations, therefore increasing segregation between rhythms. These results suggest that the extensive convergence/divergence in the basal ganglia circuit is physiologically necessary to sustain LFP rhythms distributed in large ensembles of neurones, but is not sufficient to induce correlated firing between neurone pairs. Conversely, loss of dopamine generates pathological linear correlation between neurone pairs, alters the patterns within LFP rhythms, and induces non-linear correlation between LFP rhythms operating at different frequencies. The pathophysiology of information processing in the human basal ganglia therefore involves not only activities of individual rhythms, but also interactions between rhythms.

Through electrodes implanted for deep brain stimulation in three patients (5 sides) with Parkinson's disease, we recorded the electrical activity from the human basal ganglia before, during and after voluntary contralateral finger movements, before and after L-DOPA. We analysed the movement-related spectral changes in the electroencephalographic signal from the subthalamic nucleus (STN) and from the internal globus pallidus (GPi). Before, during and after voluntary movements, signals arising from the human basal ganglia contained two main frequencies: a high beta (around 26 Hz), and a low beta (around 18 Hz). The high beta (around 26 Hz) power decreased in the STN and GPi, whereas the low beta (around 18 Hz) power decrease was consistently found only in the GPi. Both frequencies changed their power with a specific temporal modulation related to the different movement phases. L-DOPA specifically and selectively influenced the spectral power changes in these two signal bands.