How good vibes turn bad: rise of ictal events from oscillatory network activity. Focus on "transient depression of excitatory synapses on interneurons contributes to epileptogenesis during gamma oscillations in the mouse hippocampal slice".

Synchronous network oscillations are a common feature of normal cerebral activities associated with a variety of brain functions from sleep to exploratory behaviors and learning and memory. Synchronous network events, however, can take place also under certain pathological conditions, for example, in epilepsy. A major, long-standing question regarding the nature of the relationship between the normal and pathological synchronous activities concerns the mechanisms that underlie the transition from physiological background activity to pathological hypersynchronicity. In this issue of the Journal of Neurophysiology (p. 1225–1235), Traub et al. make a significant contribution toward a resolution of this complex issue by combining experimental data obtained from hippocampal slices with computational network modeling to demonstrate that short-term depression of excitatory inputs to GABAergic interneurons plays a major role in phase transitions during a bistable network activity comprising of gamma oscillations and epileptiform bursts (Traub et al. 2005). Gamma oscillations (30–70 Hz) are present in vivo in the hippocampus and neocortex, and persistent gamma oscillations can also be elicited in vitro by a variety of manipulations, including the application of kainate or activators of metabotropic acetylcholine or glutamate receptors. Interneurons had been identified previously as key players in the generation of hippocampal network gamma oscillations both in vivo and in vitro, where they may exert their effects through multiple mechanisms, for example, through feedback inhibition that limits action potential generation to narrow time windows during population oscillations. Synchronous neuronal activity, such as paroxysmal depolarization shifts (PDS), can become particularly conspicuous during the initiation and maintenance of epileptic seizures. However, the mechanisms that underlie the triggering of PDS and epileptiform bursts are not yet fully understood. Curiously, prior in vivo recordings from rodents and humans during the onset of population bursts suggested an involvement of apparently “normal” physiological rhythms in epileptic burst generation. Specifically, both very fast oscillations ( 70 Hz) and gamma oscillations have been detected immediately preceding the onset of population bursts (Fisher et al. 1992; Medvedev et al. 2000), suggesting that the ability to hypersynchronize pyramidal neuronal output and generate PDS might be intrinsic to physiological rhythms, and, by extension, that even relatively small alterations in network properties might destabilize the network. In the first, experimental, part of the study, Traub et al. used high concentrations (10–100 M) of the metabotropic glutamate receptor agonist (S)-3,5-dihydroxyphenylglycine (DHPG) to evoke transitions from physiological to pathological oscillatory activity in the CA3 region of the hippocampus. Application of DHPG is known to be epileptogenic in vivo, and, under the in vitro conditions used in this study, DHPG resulted in oscillatory field potentials in the gamma range interspersed with epileptiform burst activity recordable from the CA3 pyramidal layer. In addition, very fast oscillations became apparent immediately before and during the bursts. Thus the DHPG-induced in vitro activity mimicked some of the defining features observed during the onset of ictal events in vivo. Subsequently, intracellular recordings showed that the transitions from gamma oscillations to burst activity were accompanied by an increase in excitatory input and a decrease in GABAergic input to CA3 pyramidal neurons. Conversely, the excitatory synaptic activity onto interneurons of the pyramidal layer sharply declined prior to the occurrence of a population burst, transiently weakening the interneuronal control of the pyramidal cells. Considering the complexity of the neuronal circuits that are involved in seizures, straightforward interpretation of data obtained from experimental studies of epilepsy is often difficult. The major challenge is posed by the fact that, in most acute and chronic experimental models of seizure generation and epilepsy, numerous synaptic and cellular parameters undergo alterations simultaneously, frequently superimposed on network reorganizations that take place due to cell loss and reactive axon sprouting. Given the multiple factors that get modified during the development of seizures and epilepsy, it appears to be a particularly daunting task to isolate the variables that are key to hyperexcitability and lowered seizure thresholds. Anatomically and biophysically realistic, largescale computational network modeling, however, has been proving to be uniquely useful in determining the relative contributions of the various parameters that have been found to be modified during experimental investigations (e.g., Santhakumar et al. 2005). In this spirit, in the second part of their study, Traub et al. implemented the experimentally observed transient changes in synaptic activity in a large-scale computational model. Similar to the experimental results, the computational model generated a bistable network activity switching between gamma oscillations and epileptiform population discharges, strongly supporting the conclusion that time-deAddress for reprint requests and other correspondence: I. Soltesz, Dept. of Anatomy and Neurobiology, University of California, Irvine, CA 92697 (E-mail: [email protected]). J Neurophysiol 94: 905–906, 2005. doi:10.1152/jn.00210.2005.