How much inhibition in an epileptiform burst?

GABAA receptor-mediated signalling and localisation-related epilepsy have a long and tortuous relationship. Since GABAA receptors mediate fast inhibition in the cortex the simple view is that they prevent seizures. In keeping with this, loss-of-function mutations of GABAA receptor subunits are associated with rare familial forms of epilepsy. Furthermore, experimental epilepsy in rodents has been shown to be associated with loss of interneurons that release GABA in brain regions involved in focal seizure generation such as the hippocampal formation. Although neuronal loss is not restricted to interneurons, some studies have reported a decrease in the number and strength of GABAergic synapses on surviving principal (excitatory) neurons. And indeed, potentiating GABAergic neurotransmission with barbiturates, benzodiazepines or GABA uptake blockers such as tiagabine, is a highly effective anti-epileptic strategy. But these drugs do not always work, and some forms of epilepsy arising from cortical foci can be stubbornly resistant to pharmacotherapy. Why is this? Under normal conditions GABAergic neurotransmission not only inhibits principal neurons but, helped by the fast kinetics of some interneurons and GABAergic synapses, also provides a mechanism to synchronise their firing. This feature of GABAergic signalling depends in particular on ‘perisomatic’ synapses made on the cell bodies and axon initial segments of principal cells. Recent studies on tissue from patients with intractable epilepsy (Wittner et al. 2005) and rodents with experimental epilepsy (Cossart et al. 2001) show that, unlike ‘dendritic’ inhibition, perisomatic GABAergic inputs can be preserved or even enhanced relative to control tissue. This phenomenon might be compensatory, counteracting an increased level of activity in the pathological network. However, it may also shift the balance between the two effects of GABAergic transmission: away from dendritic inhibition (which is generally thought to counteract integration of excitatory inputs) towards perisomatic inhibition, which could lead to excessive synchronization. This, together with an increased propensity of pyramidal neurons to fire in bursts, may contribute to abnormal network dynamics. A further twist is that GABA could even treacherously swap sides because of a change in intracellular chloride ion concentration, and turn into an excitatory neurotransmitter (Cohen et al. 2002). Against this background, a qualitative description of how GABAergic signalling changes in epilepsy is insufficient: what we need is a quantitative understanding of how many interneurons contribute to generate perisomatic and dendritic currents, and of the relative amplitudes and kinetics of the different types of GABAA receptor-mediated signals. In a recent issue of The Journal of Physiology Marchionni & Maccaferri (2009) take the first step towards such a quantitative description. They take advantage of the fact that epileptiform network activity does not require the physical loss of GABAergic synapses per se: with appropriate ionic or pharmacological manipulations both interictal and ictal discharges can be induced in brain tissue from non-epileptic animals. The authors compare the amplitudes of monosynaptic GABAergic signals elicited in pyramidal neurons by action potentials in individual interneurons before and after switching a hippocampal slice to a magnesium-free perfusion solution with increased potassium concentration in order to induce spontaneous bursts of activity. Using the results from previous morphological studies, they estimate the number of GABAergic neurons recruited by such bursts. It turns out that most if not all available perisomatic-targeting interneurons fire. They extend this to some technically challenging simultaneous recordings from the cell bodies and dendrites of individual pyramidal neurons, in order to ask whether the almost universal involvement of interneurons is restricted to those that are specialised to project to the perisomatic region. Although what happens in distal dendrites remains to be determined, it turns out that bursts of inhibition in proximal dendrites behave very similarly to those detected at the soma. What are the implications of these findings for epilepsy? Although near-universal recruitment of perisomatic-projecting interneurons may predispose the network to hypersynchronous behaviour, another possibility is that such massive activation acts to retard the spread of epileptiform activity. Such a role for feed-forward inhibition as a final barrier to the propagation of excessive activity has been demonstrated in layer 5 of the neocortex in a similar in vitro model (Trevelyan et al. 2007). Indeed, the intermittent burst discharges studied by Marchionni and Maccaferri may be less a model of epileptic seizures than of interictal discharges as occur in patients with epilepsy, and which have been proposed to have an anti-ictogenic role (de Curtis & Avanzini, 2001). Clearly, much work remains to be done to understand the roles of different forms of fast GABAergic signalling in ‘real’ epilepsy, not least by examining other in vitro manipulations to trigger spontaneous activity, by studying tissue from rodents at distinct developmental stages where chloride homeostasis may be quite different, and eventually by comparing to tissue from animals with established epilepsy. The quantitative approach taken by Marchionni and Maccaferri shows that a synthesis of anatomical, electrophysiological and pharmacological tools needs to be brought to bear on this difficult problem.