Coordination of entorhinal–hippocampal ensemble activity during associative learning

Abstract

2Neural activity was recorded from EC and CA1 of 17 rats trained to solve a simplified version of an odour-place association task thought to depend on interfacing of the hippocampus with inputs from the olfactory bulb and the piriform cortex via the EC 13-15 ( Fig. 1; Supplementary Video 1). On each trial, the animal sampled odours in a cue port for 1 s and then, depending on odour identity, ran to either of two cups for food reward (Fig. 1b). Following 3 weeks of training, the percentage of correct trials increased to asymptotic levels, reaching a criterion of two consecutive sessions of 85% correct performance after 16.8 ± 0.8 days (mean ± s.e.m; Fig. 1c; repeated measures ANOVA: F(20, 320) = 95.6, P < 0.001). Training-days number 2, 6, 10, 14 and 18 were defined as T1-T5, respectively. T5 was post-criterion for all animals.We first examined spectral activity in CA1 of 5 well-trained animals. After these animals had reached the 85% performance criterion, electrodes were implanted across the transverse axis of CA1 ( Fig. 1a; Extended Data Fig. 1a). Analyses focused on activity during the cue-sampling period, when recall of odour-place associations was expected to be initiated ( The observed 20-40 Hz coherence between LEC and CA1 could reflect disynaptic LEC-CA3 and CA3-CA1 coupling 12 . If this were the case, LEC-CA1 coherence would be expressed in both proximal CA1 (pCA1) and dCA1, considering that LEC input terminates indiscriminately along the CA3 transverse axis 13 . However, simultaneous recordings from LEC and pCA1 showed no increase in coherence during the cue interval (3 rats; Fig. 1h). At all frequencies from 13.7 to 33.2 Hz, pCA1-LEC coherence was weaker than dCA1-LEC coherence in the first group of rats (FDR corrected, q < 0.05; 12 tetrode pairs; Fig. 1i). The lack of coupling between LEC and pCA1, and between MEC and dCA1, suggests that the coherence is mediated by direct LEC-CA1 connections.We then asked whether and how oscillatory coupling between LEC and dCA1 evolves as the animals learn to use the odour cues to navigate. Tetrodes were targeted simultaneously to dCA1 and layer III of LEC in 5 rats. Once the electrodes were in place, the rats were trained to To determine whether coupling of 20-40 Hz oscillators in LEC and dCA1 is necessary for successful performance, we assessed activity at T5 on error trials ('T5e'; Supplementary Video 1).Results were compared to an identical number of correct trials from the same session (down-sampled correct trials, 'T5d'). The coherence of LEC and dCA1 oscillations in the 20-40Hz band decreased significantly on error trials ( Fig. 2a-c; paired t-test for all combinations of recording pairs, t(19) = 9.81, P < 0.001; for one pair per animal: t(4) = 5.0, P = 0.007). There was also a decrease in cross-frequency coupling in dCA1 ( Fig. 2f-g; t(9) = 6.30, P < 0.001). These findings suggest that 20-40 Hz coupling of dCA1 and LEC is necessary for successful odour-based navigation.K. M. Igarashi et al. 5The emergence of 20-40 Hz coupling was reflected in individual dCA...

Place Cells, Grid Cells, and Memory

“…Converging evidence has suggested that hippocampal neurons respond also to nonspatial features of the environment, such as odors (Eichenbaum et al 1987;Wood et al 1999;Igarashi et al 2014), tactile inputs (Young et al 1994), and timing (Hampson et al 1993). The same cells that respond to nonspatial stimuli fire like place cells when animals move around in space, suggesting that place cells express the location of the animal in combination with information about events that take place or took place there (Leutgeb et al 2005b;Moser et al 2008).…”

“…Recently, Igarashi et al (2014) recorded simultaneously from the lateral entorhinal cortex and CA1 region of the rat hippocampus as the animals learned an odor-place association. As the animals learned the association, the two structures showed an increasing degree of synchronous oscillatory activity in the 20-to 40-Hz range and a corresponding increase in spiking activity to the rewarded odors.…”

“…Whether entorhinal and hippocampal neurons influence each other depends strongly on the state of theta and gamma oscillations, which, during active awake behavior, predominates frequency spectra in both regions (Buzsáki et al 1983;Bragin et al 1995;Chrobak and Buzsáki 1998;Csicsvari et al 2003;Colgin et al 2009). Theta oscillations are generally coherent across most of the entorhinal -hippocampal network, but the coherence of beta and gamma oscillations is more local and fluctuates at subsecond timescales (Colgin et al 2009;Igarashi et al 2014). Such fluctuations may enable place cells to interact with different entorhinal subpopulations at different times.…”

“…Whether a place cell responds to inputs from grid or border cells may change with time, as may the influence of different modules of grid cells. Recordings from CA1 and lateral entorhinal cortex suggest that place cells also respond dynamically to nonspatial inputs, such as odors, with learned relationships to locations in the environment (Igarashi et al 2014). Beta and gamma oscillations may enable place cells to respond temporarily to information about the content of locations in the spatial environment.…”

Grid cells and cortical representation

“…Gamma oscillations provide a mechanism for dynamic coupling of selected cell assemblies 171,172 . Place cells in CA1 use fast gamma oscillations to couple to spatially modulated cell assemblies in MEC 173 whereas low-frequency betagamma oscillations enable coupling with odour-coding neurons in the lateral part of entorhinal cortex 174 . Because beta and gamma epochs are both short-lasting and regionally specific 173,174 , place cells may interact dynamically with a range of entorhinal cell assemblies each carrying a distinct type of information.…”