Slow-wave sleep-imposed replay modulates both strength and precision of memory

Abstract

Odor perception is hypothesized to be an experience-dependent process involving the encoding of odor objects by distributed olfactory cortical ensembles. Olfactory cortical neurons coactivated by a specific pattern of odorant evoked input become linked through association fiber synaptic plasticity, creating a template of the familiar odor. In this way, experience and memory play an important role in odor perception and discrimination. In other systems, memory consolidation occurs partially via slow-wave sleep (SWS)-dependent replay of activity patterns originally evoked during waking. SWS is ideal for replay given hyporesponsive sensory systems, and thus reduced interference. Here, using artificial patterns of olfactory bulb stimulation in a fear conditioning procedure in the rat, we tested the effects of imposed post-training replay during SWS and waking on strength and precision of pattern memory. The results show that imposed replay during post-training SWS enhanced the subsequent strength of memory, whereas the identical replay during waking induced extinction. The magnitude of this enhancement was dependent on the timing of imposed replay relative to cortical sharp-waves. Imposed SWS replay of stimuli, which differed from the conditioned stimulus, did not affect conditioned stimulus memory strength but induced generalization of the fear memory to novel artificial patterns. Finally, post-training disruption of piriform cortex intracortical association fiber synapses, hypothesized to be critical for experience-dependent odor coding, also impaired subsequent memory precision but not strength. These results suggest that SWS replay in the olfactory cortex enhances memory consolidation, and that memory precision is dependent on the fidelity of that replay.

Keywords: memory; odor memory; odor object; olfaction; piriform cortex; sleep.

Figure 1.

A, The basic experimental paradigm for electrical OB stimulation and conditioning. Animals were implanted with three stimulating electrodes in the OB and a recording electrode in aPCX. During conditioning, stimulation of one of the OB electrodes was randomly chosen to be paired with a 0.5 mA foot shock (CS+) whereas stimulation of another was not paired (CS−). During the 4 h post-training period, the CS+ or a novel stimulus, the MisMatch electrode (MM), was given at least 20 times during natural bouts of slow-wave activityor during waking. 24 h later, the animal was placed in a new context and given three presentations of the CS+, the CS−, and MM to test for stimulus evoked freezing. B, A representative example of theLFP recorded in the aPCX and respiratory (Resp) response to a brief olfactomimetic OB stimulus. C, Top, A representative pseudocolor sonogram, LFP trace, and EMG trace from one animal during imposed replay (white line) while in SWS. Presenting an electrical OB stimulus during SWS did not waken animals. Bottom, A representative comparison of FFT power spectra for SWS during OB stimulation showing no difference in slow-wave power structure. D, The number of SWS bouts over the 4 h post-training period for animals that received OB stimulus presentation during SWS (n = 9) and animals that received no imposed replay (n = 8). There was not a significant difference between groups. E, The mean duration of SWS bouts post-training for animals that received OB stimulation was not significantly different from the mean duration of SWS bouts for animals that received no stimulation.

Figure 2.

A, Following differential electrical OB stimulation aversion conditioning, paired animals that had CS+ matching pattern reinforcement during post-training SWS (n = 6) showed enhanced freezing to the CS+ stimulus compared with animals in the paired group that received no post-training stimulation (n = 4). Animals in the paired group that received post-training CS+ stimulation during waking (n = 6) showed reduced freezing, consistent with extinction. All paired animals showed freezing selectively to the CS+ and not to the CS−. None of the unpaired conditioning groups showed any significant freezing response to either stimulus presented 24 h following pseudoconditioning. In this and subsequent figures, groups with the same alphabetic label are not significantly different from each other based on post hoc tests. Groups with different alphabetic labels are significantly different from each other based on post hoc tests (p < 0.05). B, To eliminate the floor effect between paired: no stimulation and paired: Awake groups, we increased the number of CS+/US pairings during conditioning to 7. With this additional training, paired awake CS+ animals (n = 3) showed significant extinction compared with the no stimulation group (n = 3). Data points from all individual animals are included to represent typical intersubject variation in this and all behavioral experiments. C, Following the presentation of a novel stimulus that did not match the CS+ during post-training SWS, paired animals that received mismatching pattern stimulation during the SWS (n = 4) showed generalized freezing on the test day. Paired animals that had CS+ matching pattern replay during SWS (n = 5), however, showed selective freezing to only the CS+. Unpaired animals (n = 4) showed no evoked freezing response on the test day. D, To examine the effects of CS+ imposed replay timing relative to piriform cortical sharp-waves, we presented the imposed CS+ either during the sharp-wave peak or 200 ms after the sharp-wave peak. Arrows mark onset and offset of stimulation. E, Animals receiving imposed CS+ replay 200 ms after sharp-waves (n = 3) showed enhanced freezing response on the test day compared with the no stimulation (n = 3) and the on-peak stimulation (n = 3) groups). The A-marked group represents significant selective freezing to the CS+ compared with CS−. The B-marked group signifies animals with delayed imposed replay froze significantly more than all other groups.

Figure 3.

A, Schematic diagram of aPCX circuitry including afferent (LOT) and association fibers. Baclofen selectively depresses the association fiber synapses (stars). B, Baclofen infused into the anterior piriform cortex selectively depressed intracortical association fiber synaptic responses while having no effect on afferent fiber evoked (LOT) synaptic responses. The bar represents a significant difference in LOT and association fiber evoked responses. Inset, Examples of LOT- and association fiber-evoked responses before and after baclofen infusion.

Figure 4.

A, Top, A representative example showing a transition from SWS into waking as recorded in the aPCX. SWS is characterized by high delta power activity and relatively low EMG activity. Below is an example of an aPCX sharp-wave and its amplitude measurement. B, Baclofen (n = 9; B-marked groups) reduced anterior piriform cortex sharp-wave amplitude compared with saline controls (n = 9; A-marked groups). This effect was greatest in animals trained in the paired baclofen condition (n = 5; Group C). C, Paired animals (n = 10; A-marked groups) spent significantly more time in SWS post-training than unpaired animals (n = 10; B-marked groups). Baclofen infusion into the anterior piriform cortex had no effect on the total time spent in SWS in either group (Fisher's PLSD p < 0.05). D, Bilateral baclofen infusions into the anterior piriform cortex following training significantly enhanced generalization of odor-evoked freezing, without impacting freezing to the CS+. Groups (n = 5/group) with the same alphabetic label are not significantly different from each other based on post hoc tests. Groups with different alphabetic labels are significantly different from each other based on post hoc tests, p < 0.05.