A phase code for memory could arise from circuit mechanisms in entorhinal cortex

Abstract

Neurophysiological data reveals intrinsic cellular properties that suggest how entorhinal cortical neurons could code memory by the phase of their firing. Potential cellular mechanisms for this phase coding in models of entorhinal function are reviewed. This mechanism for phase coding provides a substrate for modeling the responses of entorhinal grid cells, as well as the replay of neural spiking activity during waking and sleep. Efforts to implement these abstract models in more detailed biophysical compartmental simulations raise specific issues that could be addressed in larger scale population models incorporating mechanisms of inhibition.

Figure 1

A. Summary of the circuitry of medial entorhinal cortex. Input from other cortical areas (Cortex) and subiculum (sub) enters in layer II and III. Layer II contains both stellate and pyramidal cells, and these cells send recurrent connections to layer II and afferent connections to dentate gyrus and CA3. Layer III has recurrent connections to layer II and III and afferent connections to CA1 and subiculum. Region CA1 and subiculum send return connections to layer V which projects to other cortical regions. B. Whole cell patch recording in slice preparations shows that layer II entorhinal stellate cells generate subthreshold membrane potential oscillations in between the generation of action potentials (Giocomo & Hasselmo, 2008b). Blowup focuses on subthreshold oscillations. C. Whole cell patch recording in the presence of cholinergic or mGlulr agonists shows that layer III and V pyramidal cells exhibit persistent spiking that is maintained after the termination of a square pulse current injection (Yoshida et al., 2008).

Figure 2

Schematic representation of oscillations with different frequencies that could be regulated by neuronal input. A. Higher frequency oscillation (f=6 Hz). B. Lower frequency oscillation (f=4 Hz). C. Experimental data from different populations of stellate cells recorded at different membrane potentials shows a difference in mean oscillation frequencies.

Figure 3

Plot of the phase of membrane potential oscillations in a single cell cos(φ_m(t)) interacting with the network theta rhythm oscillation cos(φ(t)). A. Input h(t) with magnitude 1.0 and duration 1.25, causes a shift in the frequency and phase of cos(φ_m(t)) relative to cos(φ(t)) that is proportional to the magnitude and duration of input. Thus, the input associated with a prior item input can alter the phase representation, providing memory for the item in the form of a shift in phase that is maintained over time. B. Example of the shift in frequency and phase caused by input h(t) with magnitude 0.2 and duration 2.0.

Figure 4

The phase code of memory can be read out by spiking activity due to interference between oscillations. The top row shows two oscillations that start out in antiphase with each other. The depolarizing input h(t) to cos(φ_m(t)) causes the frequency of the oscillation to increase and the phase to shift relative to the reference oscillation cos(φ(t)). The sum of the oscillations then shifts from showing destructive interference at the start to showing constructive interference. This constructive interference brings the summed oscillation over threshold, generating spiking activity.

Figure 5

Mechanism for interaction of persistent firing cells to cause grid cell firing. A. Spiking activity over time of three different groups of persistent firing neurons. Here, each group consists of three persistent spiking cells firing with a baseline frequency of 3 Hz with different phases. Cells receive input from head direction (HD) cells with 0 degree preferred angle for Group 1, 120 degree angle for Group 2, and 240 degree angle for Group 3. Grid cell firing arises from the convergent spiking of the three groups of persistent firing neurons. When all three persistent firing groups fire in synchrony, the grid cell will fire (dots). B. Grid cell spiking (dots) occurs only when all of the persistent firing neurons fire at the same phase, resulting in a typical grid cell firing pattern. Gray line indicates rat trajectory from experimental data (Hafting et al., 2005).

Figure 6

Model of encoding and replay of trajectories. A. During encoding, behavior drives the activity of head direction cells drives h(t) that drive the activity of grid cells in entorhinal cortex layers II and III. The grid cells drive place cell firing p(t) in the hippocampus. Links between state (place) and action (speed and head direction) are made by strengthening synapses between place cells and head direction cells W_PH. B. During retrieval, the activity of place cells activates head direction cells coding the velocity from that state which then activates the next encoded location. C-D. The model simulates temporally structured replay of spiking activity of place cells during REM sleep. The speed of replay depends on the strength of connections W_PH. Column C shows place cell activity during waking, with the same speed of movement. Column D shows spiking during simulated REM replay. In each row of Column D, the connections W_PH are multiplied by a different value (0.5, 1.0 or 1.5) to change overall strength during REM replay.