Our data reveal that the effectiveness of recurrent inhibition depends on the dendritic excitatory input pattern and the intrinsic excitability of dendritic branches. On dendritic branches exhibiting weak excitability, local inputs evoked EPSPs and weak dendritic spikes. These were reliably suppressed by recurrent inhibition. In contrast, strong dendritic

spikes evoked on branches with high intrinsic excitability resisted recurrent inhibition and therefore provided persistent input-output coupling. Furthermore, we found that plasticity of branch excitability enabled weakly excitable branches to increase their resistance to inhibition. Previous studies on excitatory signal integration have shown that dendritic spikes amplify

spatially and temporally PF-02341066 clinical trial correlated inputs from presynaptic ensembles and consequently facilitate the conversion of these inputs to an action potential output (Gasparini et al., 2004; Losonczy and Magee, 2006; Remy et al., 2009; Stuart et al., 1997). Our experiments now show that the activation of recurrent inhibition significantly reduces the set of dendritic branches that are able to generate dendritic spike-triggered action potential output. We show that inhibition virtually excluded dendritic branches on which weak spikes and EPSPs were generated from direct triggering of action potential output. In contrast, strong dendritic spikes converted correlated branch input to highly precise, spikelet-triggered action potential output despite the presence of recurrent inhibition. PD-1/PD-L1 inhibitor cancer This resistance was not only present when

recurrent synapses were selectively activated, but also when local branch inhibition was evoked using GABA microiontophoresis, which is not selective for either recurrent or feedforward circuits. Resistance to inhibition was also not restricted to a specific timing of excitation and inhibition, an observation suggesting that strong dendritic spikes may also withstand feedforward activation of dendritic inhibitory synapses. Indeed, some dendrite targeting interneuron subtypes participating in recurrent inhibition have been shown to also be recruited by CA3-Schaffer collateral input in a feedforward manner (Somogyi and during Klausberger, 2005). The interaction of inhibitory synapses with dendritic excitation and spike generation provided by these subtypes was not a direct focus of this study, but in our experiments feedforward inhibitory synapses were coactivated with recurrent synapses when we performed GABA microiontophoresis on a branch. By exhibiting resistance to recurrent inhibition strong dendritic spikes may ensure effective input to output coupling for correlated inputs on highly excitable dendritic branches, whereas weakly spiking dendrites become much less effective. Thus, inhibition segregates branches, and their presynaptic afferent assemblies, into two distinct populations based on their output efficacy.