Speed cells in the medial entorhinal cortex

EMILIO KROPFF; ERIC CARMICHAEL; EDVARD MOSER; MAY-BRITT MOSER

Washington

Conferencia; Society for Neuroscience Annual Meeting 2014; 2014

Society for Neuroscience

Grid cells in the medial entorhinal cortex (MEC) have spatial maps characterized by periodicity (Hafting et al, 2005). Their spatial code applies to several places within an environment as well as across environments (Fyhn et al, 2007), which suggests they could be the basis of a path-integration representation of space (McNaughton et al, 2006). A path integrator would receive a speed signal as an input and integrate it along a short window of time to obtain the displacement of the animal, using this result to update the spatial maps. Although weak correlations with theta frequency (Jeewajee et al, 2008) or grid cell firing (Sargolini et al, 2006; Wills et al, 2012) have been reported, the local availability and nature of a speed signal in MEC is still unclear. We recorded neural activity in the MEC and hippocampus of rats in a 1mx1m open field and in the Flintstone car, a task designed to control their speed. The car, which was ran by a computer along a 4m track, had no floor, constraining rats to run by their own means at an experimenter-determined speed in order to keep up and reach the end of the track, where a food reward was delivered. We report that running speed is coded in the firing rate of a large population of MEC cells, representing around 10% of the overall cell count in all layers. The firing of these speed cells is characterized by a positive-linear relationship with running speed and low spatial and head-directional information content. While other entorhinal populations (grid, head direction and border cells) present large overlaps with each other, their overlap with the population of speed cells is in all cases significantly lower than expected by chance. In agreement with the functional segregation of speed cells, when experimentally disentangling speed and position we were unable to find any modulation by speed in the grid cell population, and found only a weak modulation in hippocampal place cells. Finally, MEC speed cells exhibited prospective coding, anticipating the rats? movements around 60ms on average, although in some selected examples the anticipation reached 200-300ms. In turn, grid cells recorded in the same experiments behaved as if guided by a prospective rather than a regular path integrator, as evidenced by the anticipation of their firing position when acceleration was positive as opposed to negative, an effect that was absent in simultaneously recorded hippocampal place cells. This shift in firing position was stronger in a) layer II and b) the last portion of the theta cycle, not associated in grid cells with phase precession proper. Put together, our results point to the coordinated involvement of MEC grid cells and speed cells in path integration.