# Sparse Activity of Hippocampal Adult-Born Neurons during REM Sleep Is Necessary for Memory Consolidation

- **ID:** `kumar-2020-neuron`
- **Authors:** Kumar D, Koyanagi I, Carrier-Ruiz A, Vergara P, Srinivasan S, Sugaya Y, Kasuya M, Yu TS, Vogt KE, Muratani M, Ohnishi T, Singh S, Teixeira CM, Chérasse Y, Naoi T, Wang SH, Nondhalee P, Osman BAH, Kaneko N, Sawamoto K, Kernie SG, Sakurai T, McHugh TJ, Kano M, Yanagisawa M, Sakaguchi M
- **Published:** Neuron, vol. 107(3), 552-565.e10 — 2020-08
- **DOI:** 10.1016/j.neuron.2020.05.008
- **Link:** https://doi.org/10.1016/j.neuron.2020.05.008
- **PMID:** 32502462
- **Type:** Research article · peer reviewed
- **Role of M. Sakaguchi:** last corresponding
- **License:** Elsevier open archive user license (from 2021-08-05)
- **Themes:** adult_neurogenesis_memory, rem_sleep_memory, fear_memory_ptsd, imaging_tools

**In one sentence:** In mice, young adult-born dentate gyrus neurons active during fear learning are reactivated in later REM sleep, and silencing their sparse REM-sleep activity impairs contextual fear memory consolidation.

**Summary.** This study asked whether the small population of young adult-born neurons in the mouse dentate gyrus contributes to memory consolidation during sleep. Long-duration calcium imaging in freely moving mice showed that these neurons fire sparsely, even less in sleep, and that contextual fear learning further lowers their activity in REM sleep while the neurons active during learning tend to be reactivated in REM sleep. Optogenetically silencing young adult-born neurons only during REM sleep in the consolidation window left sleep architecture and dentate theta oscillations unchanged, but reduced activity-dependent gene expression, lengthened dendritic spine necks, and impaired contextual fear memory. Random optogenetic activation in REM sleep also impaired memory, which the authors interpret as a need for finely tuned activity.

**Question.** Does the activity of young adult-born dentate gyrus neurons during sleep, and REM sleep in particular, contribute to hippocampus-dependent memory consolidation?

**Approach.** Nestin-CreERT2 mice were used to express GCaMP3, halorhodopsin, or channelrhodopsin-2 in young adult-born neurons; activity was imaged across sleep-wake states with a head-mounted miniature microscope, and the neurons were silenced or activated in a sleep-stage-specific manner using real-time EEG/EMG staging after contextual fear conditioning, combined with RNA sequencing, spine morphology, and LFP analysis.

**Key findings** (each checked against the article text):

1. After contextual fear conditioning (delayed shock), the calcium event frequency of young adult-born neurons in REM sleep was further reduced, which did not occur in immediate-shock controls or in fully mature granule neurons. [Fig. 1G-H; Fig. S5A-C; mouse; n = 88, 47, 113, and 137 neurons in IS-BL, IS-AS, DS-BL, and DS-AS groups; p < 0.05, bootstrap]
2. About 37.9% of recorded young adult-born neurons were active at least once in post-learning REM sleep, and neurons active right after the shock were more likely to be active in subsequent REM sleep, whereas activity during the retrieval test did not predict REM activity. [Fig. 1I-K; mouse; n = 59 and 36 neurons in REM- and REM+ groups from 7 mice; p < 0.05, p < 0.01, bootstrap]
3. Silencing young adult-born neurons during REM sleep in the 6 h after conditioning did not change sleep architecture, EEG power spectra, or dentate gyrus theta autocorrelation (p = 0.72). [Fig. 3D-L; mouse; n = 11-13 mice/genotype; 434 and 468 LFP epochs; p = 0.72, Mann-Whitney test]
4. REM-sleep silencing of young adult-born neurons reduced dentate gyrus expression of immediate early genes such as Zif268, Npas4, and Fos, all decreased genes carried SRF binding motifs, and dendritic spine necks of these neurons became longer. [Fig. 4H-L; Fig. 5C-E; mouse; n = 4 mice each (RNA-seq); n = 30-40 dendrites/group (spines); SRF motif enrichment p < 0.0001]
5. Silencing young adult-born neurons only during REM sleep within the consolidation window reduced contextual freezing the next day, while silencing outside the window (p = 0.98), during NREM sleep or wakefulness, or of 2-week-old or 10-week-old neurons did not significantly impair memory. [Fig. 6A-Q; mouse; n = 12-13/genotype (within window); n = 9/genotype (outside window); p < 0.05, two-tailed unpaired t test]
6. Random 20-Hz optogenetic activation of young adult-born neurons during REM sleep also reduced contextual freezing, suggesting that finely tuned rather than simply more activity is needed for consolidation. [Fig. 6T-V; mouse; n = 10-12/genotype; p < 0.05, two-tailed paired t test]

**Methods:** in vivo calcium imaging with a head-mounted miniaturized microendoscope in freely moving mice; tamoxifen-inducible genetic labeling of adult-born neurons (pNestin-CreERT2); closed-loop optogenetic silencing (halorhodopsin) and activation (channelrhodopsin-2) during EEG/EMG-defined sleep stages; contextual and tone fear conditioning; dentate gyrus local field potential recording; slice patch clamp; laser microdissection and RNA sequencing; confocal 3D reconstruction of dendritic spines

**Models:** male F1 transgenic mice on a C57BL/6J background (pNestin-CreERT2 crossed with Rosa26 Cre-dependent GCaMP3, halorhodopsin, channelrhodopsin-2, or tdTomato reporter lines)

**Brain regions:** dentate gyrus; hippocampus

**Keywords:** adult neurogenesis; dentate gyrus; REM sleep; memory consolidation; contextual fear conditioning; calcium imaging; optogenetics; dendritic spines; immediate early genes

**Data and code availability:**
- data: https://doi.org/10.17632/gfbdv5kfrz.1 — “Data underlying the results described in this manuscript, except raw data for RNA analysis, are available at Mendeley Data”
- data: https://doi.org/10.17632/pfgkv43yn7.1 — “RNaseq raw data are available at Mendeley Data”

**Limitations stated by the authors:**
- Calcium indicators lack the temporal resolution needed to assess how adult-born neuron firing relates to theta phase.
- The sparseness of REM-active adult-born neuron activity prevented population-level ensemble analysis.
- Selectively controlling only the neurons active in REM sleep would require manipulation with temporal resolution matching single REM episodes of about a minute.
- The interval between conditioning and testing in the imaging experiments was shortened to 5.5 h to keep the same cells trackable, which lowered overall freezing.

---
Prepared by the Department of Neurophysiology, Kobe University Graduate School of Medicine. Quantitative statements were checked against the published articles. Index of all papers: https://www.med.kobe-u.ac.jp/neurophy/llms-full.txt