PHASE-DEPENDENCE OF FUNCTIONAL CONNECTIVITY IN GRID CELL NETWORKS

Grid cells in the medial entorhinal cortex are hypothesized to participate in a continuous attractor network (CAN), in which recurrent connectivity constrains population activity to a low-dimensional toroidal manifold. Although growing evidence supports this framework, a key aspect of grid cell CAN models remains untested, namely the assumption that functional connectivity within a grid module depends on grid phase. Specifically, many CAN models propose a Mexican hat-like connectivity profile with excitation between functionally similar cells and inhibition of more dissimilar neurons.
To test this prediction, we developed an all-optical approach to probe functional connectivity between grid cells. Grid cells are first recorded in freely moving mice using a miniature two-photon (2P) microscope (MINI2P), then re-identified in a benchtop 2P holographic stimulation system. Grid cells expressing rsChRmine are selectively photostimulated, and stimulation-triggered responses are measured in non-stimulated grid cells and related to their grid phase.
Using this approach, we record over 100 grid cells simultaneously with MINI2P. On average 80% of these neurons can be re-identified and photostimulated in the benchtop system. Preliminary results suggest enhanced excitatory functional connectivity between grid cells with similar phases whereas connectivity between phase-dissimilar cells is dominated by inhibition. In contrast, similar stimulation experiments for direction-tuned neurons in the parasubiculum reveal global inhibition, independent of preferred firing direction. If further validated, these findings may confirm the existence of CAN connectivity for grid cells, and a potential lack thereof for direction-tuned neurons in the parasubiculum, whose ring-like topology may be derived from upstream systems.