To further substantiate this finding, we also analyzed action potential timing during blockade of inhibition at the single-cell level. To do so, we applied DNDS, which blocks GABAAR-mediated AZD5363 inhibition from the intracellular side without

changing action potential firing (Dudek and Friedlander, 1996) (Figure 8). Using this approach we found that action potentials were locked to ripples (spike-time histograms in Figures 8F and 8H; log10 p values in the range of −54.3 and −1.6; n = 1,119 spikes associated with 1,564 SWRs; 7 cells). Together, these experiments demonstrate that the ripple-locked excitatory inputs remaining after block of inhibition can effectively regulate the spike timing of target principal neurons.

In a final set of experiments, we studied the Alpelisib timing of ripple-associated inhibition relative to phasic excitation during SWRs. At the excitatory reversal potential (∼−6 mV; Figure S8), we observed complex outward currents reflecting the superposed inhibitory inputs present during ripples (Figures 9A and 9B). These currents were also significantly locked to ripples, as demonstrated by onset phase analysis, onset-triggered LFP averaging, and peeling reconstruction analysis (Figures 9C–9E; n = 849 events in 6 cells). We compared the timing of phasic excitation and inhibition, as derived from cEPSC and cIPSC slope onsets and peeling reconstruction. Figure 9E juxtaposes the dynamics of ripple-locked excitatory and inhibitory currents for two cells, and Figure 9F summarizes the averaged fitted onset histograms for 8 and 6 cells, respectively. During the initial course of ripples, excitation is slightly phase-advanced, leading inhibition by ∼1.5 ms. In later periods, the phases of the two components converge (phase difference plot in Figure 9F, black line). This finding is confirmed by comparing

the lags of correlation peaks determined for 48 excitatory-inhibitory cell pairs early versus late in the ripple (Figure 9G). Cross-correlation peaks computed on the earlier period, between −16 and 0 ms relative to the SWR peak, clustered around −2 ms (median: −2.0 ms; blue histogram), whereas those computed between Florfenicol 0 and +16 ms clustered around 0 ms (median: 0 ms; green histogram). Together, these analyses reveal high precision of ripple-associated inputs and a progressive synchronization of excitation and inhibition during the course of ripples. Here, we combined an in vivo approach (Crochet and Petersen, 2006, Margrie et al., 2002 and Poulet and Petersen, 2008) and an in vitro model (Maier et al., 2009) to study synaptic input onto CA1 pyramidal cells during hippocampal ripples. We found that PSCs are phase-locked to ripples and coherent among principal neurons. These currents contained strong excitatory components: First, they could be observed at a membrane voltage with low driving force for Cl−, i.e.