RECONSTRUCTING SHARP WAVE-RIPPLES FROM SUBTHRESHOLD ACTIVITY OF HILAR MOSSY CELLS

Information is processed as it propagates through successive layers of neural circuits. However, how information is efficiently encoded when transmitted from layers with many neurons to those with fewer remains poorly understood. Here, we investigated how hippocampal sharp wave-ripples (SWRs), synchronized population events implicated in memory consolidation, are represented in the subthreshold membrane dynamics of hilar mossy cells (MCs), a numerically limited neuronal population. In this study, we performed in vitro whole-cell recordings simultaneously from up to five MCs in hippocampal slices, as well as in vivo whole-cell recordings from single MC in anesthetized mice, while concurrently recording SWRs. To assess how SWR is encoded in dispersed subthreshold responses, we applied machine learning algorithms to predict CA3 SWR waveforms from MC subthreshold membrane potentials (Vms). We found that Vms from five MCs could reconstruct approximately 30 % of the total SWR waveform. Although the SWRs predicted from individual MCs partially overlapped, each MC contributed distinct information, indicating a distributed and efficient representation of SWR-related information despite the sparse MC population. These findings suggest that hippocampal circuits can compress population-level activity while minimizing information loss, providing new insights into how neural circuits transmit and transform information across layers with bottlenecks. Such distributed subthreshold representations may support memory-related information flow in the hippocampus.