Coordinated Emergence of Hippocampal Replay and Theta Sequences during Post-natal Development

Abstract

Hippocampal place cells encode an animal's current position in space during exploration [1]. During sleep, hippocampal network activity recapitulates patterns observed during recent experience: place cells with overlapping spatial fields show a greater tendency to co-fire ("reactivation") [2], and temporally ordered and compressed sequences of place cell firing observed during wakefulness are reinstated ("replay") [3-5]. Reactivation and replay may underlie memory consolidation [6-10]. Compressed sequences of place cell firing also occur during exploration: during each cycle of the theta oscillation, the set of active place cells shifts from those signaling positions behind to those signaling positions ahead of an animal's current location [11, 12]. These "theta sequences" have been linked to spatial planning [13]. Here, we demonstrate that, before weaning (post-natal day [P]21), offline place cell activity associated with sharp-wave ripples (SWRs) reflects predominantly stationary locations in recently visited environments. By contrast, sequential place cell firing, describing extended trajectories through space during exploration (theta sequences) and subsequent rest (replay), emerge gradually after weaning in a coordinated fashion, possibly due to a progressive decrease in the threshold for experience-driven plasticity. Hippocampus-dependent learning and memory emerge late in altricial mammals [14-17], appearing around weaning in rats and slowly maturing thereafter [14,15]. In contrast, spatially localized firing is observed 1 week earlier (with reduced spatial tuning and stability) [18-21]. By examining the development of hippocampal reactivation, replay, and theta sequences, we show that the coordinated maturation of offline consolidation and online sequence generation parallels the late emergence of hippocampal memory in the rat.

Keywords: consolidation; development; hippocampus; memory; place cell; reactivation; replay; sleep; theta; theta sequence.

Figure 1

Reactivation of Cell Pair Co-firing Patterns during POST-Experience Rest Is Already Present at P17, but the Amount of RUN Co-firing Required to Induce Plasticity Is Greater in Young Rats (A) Schematic of experimental paradigm. Rats explored a square open field during RUN and rested in a separate holding box in the same room before (PRE-sleep) and after (POST-sleep) open field exploration. (B) Example of cell pair co-activity correlation between RUN and temporally adjacent rest sessions in one simultaneously recorded ensemble. Data were recorded at P17 and contain all cell pairs of the ensemble. x axes show place field similarity (PFS) in RUN (Pearson’s r correlation of rate map bin values), y axes correlation of co-firing during SWR events (correlation of cell pair activity across all SWRs in rest) in PRE (left panel) or POST (right panel) rest sessions. Points are colored according to magnitude of SWR spiking correlation in PRE and scaled in size according to their PFS in RUN. Regression statistics are in top right corner. Cell pairs with high PFS show an increase in SWR co-firing during POST-sleep. (C) Bar chart showing Pearson’s r values (±SE of correlation) of place field similarity (RUN) and SWR spiking correlation for PRE (pale colors) and POST (bold colors) rest sessions for all recorded cell pairs across development. ^∗∗ indicates differences at p < 0.001, ^∗ differences significant at p < 0.05. (D) Bar chart showing Pearson’s r values (±SE of correlation) of theta cycle co-firing (RUN) and SWR spiking correlation for PRE (pale colors) and POST (bold colors) rest sessions across development. Asterisks indicate differences at p ≤ 0.001. (E and F) Cell pair plasticity (change in cell pair SWR spiking correlation from PRE- to POST-sleep) as a function of cell pair co-firing in RUN. (E) Mean cell pair plasticity (±SEM) as a function of the number of theta cycles in which both cells fire during RUN. (F) Mean cell pair plasticity (±SEM) as a function of the number of spikes fired in theta cycles in which both cells fire. For (E) and (F), colored asterisks mark the smallest x axis bin in which cell pair plasticity is significantly different from zero (t test of mean against 0; p < 0.05) at each age.

Figure 2

Complex Spike Cell Firing on Square Track Environment in Developing Rats (A) Schematic of experimental paradigm. Rats explored a square track during RUN and rested in a separate holding box in the same room before (PRE-sleep) and after (POST-sleep) track exploration. (B) Example place field maps for RUN sessions on square track at different ages. For each age, each row represents the spatial firing of one cell along the length of the square track, filtered for one running direction. False colors show firing rate, scaled to the peak firing rate for each cell. Cells are ordered according to position of peak spatial firing on the track. Dashed white lines indicate the corners of the square track. (C and D) CS cell spatial firing evenly covers the extent of the square track. (C) Mean spatial distributions of normalized firing rate of all CS cells recorded within each age group. (D) Histograms showing the proportion of CS cell peak spatial firing locations at different positions on the square track within each age group.

Figure 3

Gradual Emergence of Replay between P17 and P32 (A) Significant linear trajectory events in POST-sleep at different ages (four examples per age). For each event, top panel shows time-by-position probability posterior derived from Bayesian decoding of position, based on event spiking. False colors show decoded probability, and white lines indicate the band of the best linear fit. Summed probability within fit lines (p) and speed of event (speed) are indicated above the posteriors. Bottom panel shows spike raster of complex spike cell activity during replay events. Cells are ordered by the position of their spatial peak firing on the track. Linear trajectories are predominantly stationary at younger ages, with replay emerging gradually in older animals. (B) Percentages (±95% confidence interval) of events with a significant linear trajectory during PRE- and POST-sleep sessions across development. Dotted line represents 95% confidence threshold. In all age groups, significantly more events than expected by chance showed a significant linear trajectory in POST sessions (binomial test; p < 0.001 for all groups). (C and D) Mean characteristics of significant linear trajectory events in each POST-sleep session. For all plots, each data point represents mean (±SEM) of all significant linear events in one experimental session (one rat/day). Adult data represent overall mean across all sessions. For each measure, r² and p values of linear regression over age are indicated above plots (adult data are always excluded from regression analysis). Distance covered (C) and speed of decoded trajectories (D) are shown. (E and F) Cumulative distributions of the distance covered (E) and the speed (F) of all significant linear trajectory events in the age groups P17–P20, P21–P24, P25–P32, and in adult animals. See also Figures S1, S2, and S3.

Figure 4

The Gradual Maturation of Theta Sequences between P17 and P32 Is Correlated with the Emergence of Replay (A) Examples of theta sequence emergence across development. Each plot shows a probability posterior derived from a single RUN session, where the x axis shows the proportion of time elapsed during the theta cycle and the y axis shows position on the track relative to the current location of the rat. The horizontal white line shows current rat location, and the vertical white lines demarcate one theta cycle. Hot colors show high decode probabilities. Numbers above the plots show theta sequence score, defined as the circular-linear weighted correlation of the probability posterior. Theta sequences are indicated by a shift in the decoded position from behind to ahead of the rat within the theta cycle: this emerges gradually between P17 and P32. (B) Mean (±SEM) theta sequence scores in each age group. ^∗∗ indicates differences significant at p < 0.001 (1-way ANOVA comparison of age groups). (C and D) Theta sequence scores are correlated with the distance covered (C) and speed (D) of replay trajectories during development. Each data point represents mean (±SEM) of all significant linear events in one experimental session for all developing rats. For each measure, r² and p values of linear regression over age are indicated above plots. Correlations reported in (C) and (D) remain significant even after controlling for age; see main text. See also Figure S4.