Activation of Subthalamic Nucleus Outflow by High-Frequency Stimulation Is Consistent with the Nigral Control of Epilepsy Model

Zhang et al. [1] found that STN HFS at 130 Hz (the most popular frequency for HFS) for 1 h increased glutamate concentrations in the GP and SNr, which were maximal 1 h after stimulation, but still very elevated at 2 h. In contrast, HFS at 130 Hz (or 260 Hz) had absolutely no effect on GABA levels in the GP. However, both conditions led to substantial and sustained increases in GABA (with a similar time course to glutamate levels) in the SNr. In epileptic rats, baseline GABA levels were increased in the SNr (but not GP) and the GABA-inducing effects of STN HFS were greater than in normal rats. Finally, STN HFS at 260 Hz also markedly elevated SNr and GP glutamate levels (in normal and epileptic rats) with a similar time course but to a slightly lesser degree compared to 130 Hz, whereas the GABA-elevating effects in the SNr were slightly greater at 260 Hz than 130 Hz. How should these data be interpreted? First, it should be remembered that the GP in the rat is equivalent to the external segment of the GP (GPe) in primates, not the internal segment, which in rats is the entopeduncular nucleus [to emphasize this I will denote the GP of the rat as ‘GP(e)’]. The glutamate-elevating effects of STN HFS in the SNr and GP(e) are consistent with an activation of STN by HFS, with the SNr receiving the glutamatergic fibers of the STN, and the GP(e) receiving recurrent glutamatergic STN fibers ( fig. 1 c) [13, 14] . This stands in contrast to the notion behind the initial experiments on epileptic rats where it was thought a priori that STN HFS would inhibit its output. But does this glutamate elevation in the SNr drive an increase in its output? The finding that GABA is elevated as well in the SNr is critical and unexpected. This would result in overall inhibition of the SNr, despite an STN-mediated increase in glutamate release, since GABA receptors are located more proximally on cell soma thus negating excitation elicited by glutamate receptors located more distally on dendrites. As a result, nigral projections will likely be inhibited leading to increased antiepileptic outflow from the superior colliculus, consistent with the clinical results of STN HFS and the nigral control of epilepsy model. The observation of decreased metabolic activity [15] and firing rate [16] in the SNr with STN HFS is consistent with this prediction. By what mechanism might GABA levels be increased in the SNr by STN HFS? The SNr receives GABA afferents from the striatum (direct pathway), but also the GP(e) (indirect pathway) [17] . Thus, elevation of GABA in the SNr might result from the activation of pallidal-nigral efferents by recurrent STN-GP(e) glutamatergic fibers, consistent with the increased glutamate levels in the GP(e) found in this study. Other explanations may exist as well. However, antidromic activation of corticostriatal efferents (via the monosynaptic cortico-STN projection) is unlikely since this would be expected to lead to elevation of GABA in the GP(e) (via Zhang et al. [1] examined the effects of two levels of high-frequency stimulation of the subthalamic nucleus (STN) on the concentration of extracellular glutamate and -aminobutyric acid (GABA) in the globus pallidus (GP) and substantia nigra pars reticulata (SNr) of normal or kainate-induced spontaneously epileptic adult rats. They interpret their results in terms of the mechanism of antiepileptic properties of STN stimulation. This study is focused on an important question, namely the actual mechanisms of action at play during electrical stimulation within the so-called nigral control of epilepsy circuit first elucidated by Karen Gale and colleagues [2, 3] in the 1980s and subsequently by others including the Grenoble group of Deransart et al. [4] . The essential insight of Gale’s group was that bilaterally inhibiting the SNr (with the GABA agonist muscimol) blocked limbic seizures – induced by injection of the GABA antagonist bicuculline into the self-named area tempestas (‘tempest’ = ‘gale’) – as well as other types of seizures [5, 6] . It is thought that the mechanism of this nigral control of epilepsy circuit proceeds through decreased activity of the inhibitory projection of the SNr to the deep layers of the superior colliculus which upregulates a widespread projection to cortical regions ( fig. 1 a, b). The French group [7] extended these findings by using high-frequency stimulation (HFS) of the STN with the notion that this would inhibit the glutamatergic drive on the SNr and provide a more therapeutically practical way of decreasing SNr activity – the prevailing notion at the time being that STN HFS is inhibitory of its output. Electrical stimulation of the SNr was not considered because of its large size. Indeed, STN HFS was shown in various models including the absence rat to be effective, providing the rationale for two small pilot studies [8, 9] with mixed but in part promising results leading to an ongoing large clinical trial in France of STN deep brain stimulation for ring chromosome syndromic epilepsy. The problem is: the weight of the evidence [10] is that STN HFS actually drives its axonal projections (at a regular rate) rather than inhibiting them [11, 12] . So: what is STN HFS really doing that controls epilepsy in these animal and human studies? Published online: May 13, 2008