The role of inhibition in shaping memory-encoding hippocampal sequences

Hippocampal pyramidal (PY) spiking sequences link temporally contiguous memories by encoding sensory cues as well as temporal intervals between them, constructing trajectories across ‘memory-space’. But what is the role of inhibition – particularly parvalbumin (PV) and somatostatin expressing (SST) interneurons (INs) - in shaping such sensory and temporal representations? We pioneered longitudinal, high-frequency voltage imaging in vivo on the CA1 of PV-Cre and SST-Cre mice while they performed an odor-cued delayed non-match-to-sample task (DNMS), shown to yield odor-specific PY sequences during the odors and ensuing delays. We recorded action potentials and membrane dynamics from PV and SST cells during untrained trial exposure as well as during trained performance, following the same cells across multiple days or before and after training. Roughly half of PV and SST cells exhibited a significant firing field during the odor-cue presentation (‘odor-cells’). Unlike PYs, these neurons responded to both odors similarly, and practically no cells encoded any delay timepoints, yielding no delay-sequences. The number of odor-cells was stable per day, with a fixed cell-turnover, whereas some INs retained stable odor-fields for days, even throughout DNMS training. Surprisingly, at the odor onset, cells exhibited a sharp delta-frequency hyperpolarization which was triggered by a short burst seen only in PV cells. It briefly reset the intracellular theta phase of INs and condensed the rebound odor-spiking into a single theta-cycle. Such pronounced and fine-timed inhibition should silence many PYs during odors. Indeed, through in vivo 2-photon calcium imaging in transgenic mice under the same setup, we found that roughly half PYs were inhibited during odors. We thus propose that the role of this very fine-timed but non-specific and behaviorally-independent inhibition, is to silence background activity during a cue and thus increase the signal-to-noise-ratio of the few PYs that encode the cue presentation and initiate the ensuing spiking sequence.