Cell‐specific contribution to gamma oscillations

Abstract

Gamma oscillations are a particularly prominent form of rhythmic activity occurring during wakefulness and attentive behaviour that result from synchronous activity of cortical neurons at frequencies ranging between 20 and 80 Hz. Abnormalities in these oscillations might underlie neuropsychiatric disorders such as schizophrenia (Uhlhaas et al. 2006). Therefore understanding the precise mechanisms underlying gamma oscillations – in particular the respective role of synaptic and intrinsic currents – represents a major goal in fundamental and clinical neuroscience. Gamma oscillations are prevalent in the hippocampus in vivo and they can be induced in the isolated CA3 region of the hippocampus in vitro with bath application of carbachol (Fisahn et al. 1998). They require both excitatory and inhibitory synaptic transmission. In particular, in their pioneering study Fisahn and co-workers highlighted the role of GABAergic activity by showing that prolongation of the GABAA receptor-mediated inhibitory postsynaptic currents (IPSCs) with barbiturate reduced the frequency of gamma oscillations. Models of leaky integrate-and-fire neurons suggest that synaptic recurrent feedback loops between interneurons and pyramidal cells are necessary and sufficient to explain gamma frequency oscillations (Brunel & Wang, 2003). In a recent study, gamma oscillation fluctuations in amplitude and frequency were found to be correlated with cycle-to-cycle changes in excitation and inhibition (Atallah & Scanziani, 2009). In this issue of The Journal of Physiology, Oren, Hájos and Paulsen (Oren et al. (2010) examined the relative contribution of synaptic and voltage-dependent action currents in pyramidal neurons during gamma oscillations induced with carbachol in the CA3 region of the rat hippocampus in vitro. They combined recordings of the local field potential (LFP) to monitor the oscillatory activity of the principal neurons with cell-attached recordings from subclasses of GABAergic interneurons. Changes in the oscillation amplitude were quantified by the amplitude of the wavelet transform. Oren et al. confirm in this study that inhibitory currents contribute significantly to LFP amplitude fluctuations as a consequence of an increased probability of firing in subclasses of interneurons that are responsible for perisomatic inhibition. Indeed when performing cell-attached recordings from various cell types, the authors could observe that, for a given oscillation cycle, the mean cycle amplitude was significantly higher when perisomatic (or interneuron-selective) interneurons discharged. In contrast, there was no such direct relationship between the discharge of a pyramidal cell and LFP amplitude. Therefore cycle-to-cycle variation amplitude appears to be the direct consequence of interneuron firing probability. However, the synchronized firing of pyramidal cells, which can be detected as a population spike in some cycles during gamma waves, affected the LFP. Indeed, the authors provide evidence that slow action currents, resulting from the activation of repolarisation conductances contribute in a significant manner to the early component of the LFP. Local field potentials are probably one of the oldest, least invasive and simplest ways to perform high sampling rate recordings of not only neuronal assemblies, but also the discharge activity of single neurons with the help of combined on- and offline signal processing analysis. The possibility of performing high rate recordings with almost no recording time limitation can then become an advantage in comparison with imaging techniques. Oren, Hájos and Paulsen have proven that these techniques should not yet be considered out of date since their use, in combination with other electrophysiological techniques (cell-attached and whole-cell recordings), can be useful and very informative for studying network activity and the contribution of different cell subtypes to the genesis or modulation of cortical oscillations. In this study, the firing probability of dendritic-targeting interneurons was reported not to be correlated with LFP amplitude fluctuations. This in vitro observation is in agreement with in vivo experiments showing that the discharge of these interneurons is not coupled to gamma oscillations (Tukker et al. 2007). Knowing that these interneurons exert specific inhibitory functions on EPSP integration and dendritic electrogenesis but not on action potential firing (Miles et al. 1996) leaves them free to serve distinct functions than the ‘sole’ contribution to gamma oscillations.