How to build a grid cell

Abstract

Neurons in the medial entorhinal cortex fire action potentials at regular spatial intervals, creating a striking grid-like pattern of spike rates spanning the whole environment of a navigating animal. This remarkable spatial code may represent a neural map for path integration. Recent advances using patch-clamp recordings from entorhinal cortex neurons in vitro and in vivo have revealed how the microcircuitry in the medial entorhinal cortex may contribute to grid cell firing patterns, and how grid cells may transform synaptic inputs into spike output during firing field crossings. These new findings provide key insights into the ingredients necessary to build a grid cell.

Keywords: entorhinal cortex; grid cell; neural circuit; patch clamp; path integration; spatial navigation.

Figure 1.

Recurrent connectivity in MEC II is provided by interneurons. (a) Quadruple whole-cell recording in vitro from a fast-spiking interneuron (1) and three stellate cells (2, 3 and 4). Only the interneuron responds with excitatory postsynaptic potentials to the stimulation of a stellate cell (middle). By contrast, all stellate cells display inhibitory responses when the interneuron is stimulated (right). (b) Connectivity rates for excitation (black) and recurrent inhibition (red) as a function of postnatal age. Note near-absence of excitatory connections in adult animals. (c) Connectivity rates for inhibitory connections from fast-spiking to stellate cells, excitatory connections from stellate to fast-spiking cells, and the reciprocity of these connections. ((a–c) Adapted with permission from Couey et al. [24].)

Figure 2.

Membrane potential dynamics of fast-spiking neurons in the MEC of navigating mice. (a) Electrophysiological characterization of a fast-spiking neuron from MEC of an awake mouse. The inset shows a representative action potential. (b) Animal speed (top) and membrane potential (bottom) of the fast-spiking cell shown in (a). (c) Animal speed (top), membrane potential (middle) and LFP (bottom) during the first movement period in (b) plotted at higher magnification. (d) Firing rate (top), subthreshold membrane potential (middle) and theta MPO amplitude were plotted against the position of the animal along the long axis of the track. High basal firing rates and spatial modulation of firing are consistent with the predictions for inhibitory conductances by CAN models. ((a–c) Adapted from Schmidt-Hieber & Häusser [26].)

Figure 3.

Membrane potential of stellate cells shows theta periodicity during running. Whole-cell recordings from MEC II of mice navigating on a spherical treadmill. (a) Mouse speed (blue) and membrane potential of a stellate cell (black) during a running period. Note that onset of oscillatory activity (black arrow) precedes onset of running (blue arrow). (b) Spectrograms of membrane potential (middle) and LFP (bottom) were aligned to the onset of movement (top) before computing the average for stellate cells (left) and putative pyramidal cells (right). (c) Average power spectra for stellate cells (top) and putative pyramidal cells (bottom) at rest (black) and while running (red). Note the distinct theta peak in the power spectrum of stellate cells during running. ((a–c) Adapted from Schmidt-Hieber & Häusser [26].) (d) Membrane potential traces of an MEC II neuron with large theta oscillations (top), an MEC II neuron with small theta oscillations (middle) and an MEC III neuron (bottom). Theta oscillation amplitudes range from 2 to 12 mV. (Adapted with permission from Domnisoru et al. [4].)

Figure 4.

Grid cell firing is driven by sustained depolarizations. Whole-cell recordings from MEC of mice navigating on a spherical treadmill. (a) Membrane potential of an MEC neuron (black) during a run along a linear track. Two firing fields were crossed, as indicated at the bottom. Membrane potential was decomposed into a ramp (red) and a theta oscillation (grey). Note the sustained increase in membrane potential during field crossings. (b) The ramp voltage increased more than theta oscillation amplitude during firing field crossings, indicating that sustained depolarizations drive grid cell firing. ((a,b) Adapted with permission from Domnisoru et al. [4].) (c) Average firing rate (top), subthreshold membrane potential (middle) and theta MPO amplitude (bottom) were plotted against normalized position in a firing field of a stellate cell. Theta MPOs contributed only little to the depolarization in the field centre. (d) Normalized firing rates of stellate cells were plotted against deviations of theta MPO amplitudes (left) and subtheta membrane potential (right) from the mean. By contrast to subtheta membrane potential changes, changes in theta MPO amplitudes did not significantly correlate with firing rates, indicating that firing was primarily driven by slow depolarization. ((c,d) Adapted from Schmidt-Hieber & Häusser [26].)

Figure 5.

Action potentials are in phase with MPOs during grid field crossings. (a) The phases of action potentials (APs) with respect to LFP theta (left), theta MPOs with respect to LFP theta (middle) and action potentials with respect to MPOs (right) were plotted as a function of normalized position within firing fields of stellate cells. (Adapted from Schmidt-Hieber & Häusser [26].) (b) Mean phase of APs with respect to LFP theta (left) and with respect to theta MPOs (right) in the first and last eighth of each field. Action potentials showed no significant phase precession with respect to theta MPOs. (Adapted with permission from Domnisoru et al. [4].)

Figure 6.

Comparing models of phase precession to experimental data. (a–d) Schematic drawings of phase precession models. Phase precession (bottom) is plotted as in figure 5a. (a) An oscillatory interference model with two VCO inputs correctly predicts the experimentally observed phase precession. (b) A depolarizing ramp model with theta input phase-locked to LFP theta produces phase precession of APs with respect to LFP theta. Shunting recurrent inhibition prevents repetitive firing within theta periods, and thereby sharpens phase precession. However, APs show phase precession with respect to LFP theta, contrary to the experimental data. (c) Strongly hyperpolarizing recurrent inhibition abbreviates theta MPOs so that the experimental phase precession is reproduced. (d) Both dendritic excitatory inputs and somatic inhibitory inputs are phase-locked to LFP theta, but phase-shifted by 60° against each other. During the depolarizing ramp, excitatory drive decreases and inhibitory drive increases because of opposing changes in driving force. This shift in excitation–inhibition balance can produce phase precession of action potentials with respect to LFP theta [83].