Moreover, in response to a pharmacological increase in the excitation/inhibition balance onto MCs, long-range γ synchronization is enhanced to preserve the mean firing activity of MCs and the amplitude of recurrent click here and lateral inhibition that they receive. Such excitation/inhibition manipulation impairs odor mixture discrimination and slows the time required to discriminate between related odors. In brain circuits, γ rhythms rely on different modes of

network interactions (reviewed in Wang, 2010, Whittington et al., 2011 and Buzsáki and Wang, 2012). Following the excitatory-inhibitory network model (E-I model), reciprocally connected networks of excitatory and inhibitory cells interact so that excitatory neurons drive inhibitory neurons,

which in turn gate and synchronize excitatory neurons (Brunel and Wang, 2003 and Wang, 2010). Experimental and computational studies of OB γ rhythms have generally supported this model (Bathellier et al., 2006 and Lagier et al., 2007). Here, in the awake mouse, we show the crucial role of the reciprocal coupling between MCs and GCs in generating and tuning the frequency of γ oscillations. The decrease in γ frequency after PTX was associated with a lengthening of the apparent kinetics of MC spiking inhibition, with light-evoked recurrent and lateral inhibition occurring ∼1–2 ms later and decaying slower. These kinetic properties increased the time for MC spiking to recover from evoked Diminazene inhibition Proteasome inhibitor and may prolong each γ cycle of synaptic inhibition, resulting in the observed decrease in γ frequency. Our study also highlights the importance of the MC population in generating γ rhythms and reveals a band-pass effect characteristic of a resonance property. In addition, this OB circuit γ resonance is tuned by the excitatory/inhibitory synaptic properties of the dendrodendritic circuit. Interestingly, this contrasts with results obtained in the cortex where optogenetic driving of fast-spiking basket interneurons, but not excitatory pyramidal

cells, amplifies γ oscillations (Cardin et al., 2009). According to a second model, the synaptic interactions within a network of inhibitory interneurons (I-I model) represents a mechanism by which γ rhythms can be generated (Whittington et al., 2011 and Wang, 2010). Using a selective knockin strategy, we show in the awake mouse that γ oscillations rely exclusively on the dendrodendritic inhibition received by MCs and not from the synaptic inhibition received by GCs. A third model for network synchronization includes excitatory coupling between principal neurons, but several elements clearly discard this possibility in the OB. First, the pharmacological blocking of gap junction does not affect γ oscillations.