Part 1:Epigenetic Regulation Of The Circadian Gene Per1 Contributes To Age-related Changes in Hippocampal Memory
Mar 19, 2022
Contact: Audrey Hu audrey.hu@wecistanche.com
Janine L. Kwapis1,2,3, Yasaman Alaghband1,2, Enikö A. Kramár1,2, Alberto J. López1,2, Annie Vogel Ciernia4, André O. White5, Guanhua Shu1,2, Diane Rhee1,2,
Christina M. Michael1,2, Emilie Montellier6, Yu Liu7,8, Christophe N. Magnan7,8, Siwei Chen7,8, Paolo Sassone-Corsi6, Pierre Baldi7,8, Dina P. Matheos1,2 & Marcelo A. Wood1,2,3
Aging is accompanied by impairments in both circadian rhythmicity and long-term memory. Although it is clear that memory performance is affected by circadian cycling, it is unknown whether age-related disruption of the circadian clock causes impaired hippocampal memory. Here, we show that the repressive histone deacetylase HDAC3 restricts long-term memory, synaptic plasticity, and experience-induced expression of the circadian gene Perlin in the aging hippocampus without affecting rhythmic circadian activity patterns. We also demonstrate that hippocampal Per] is critical for long-term memory formation. Together, our data challenge the traditional idea that alterations in the core circadian clock drive circadian-related changes in memory formation and instead argue for a more autonomous role for circadian clock gene function in hippocampal cells to gate the likelihood of long-term memory formation.
Animals have an internal circadian clock that drives the rhythmic cycling of biological processes every ~24 h. Circadian rhythms drive numerous physiological events, including the sleep-wake cycle, feeding behavior, body temperature, and metabolism. In the master circadian clock, the suprachiasmatic nucleus (SCN), a group of core clock genes oscillate in a negative feedback loop that cycles every ~24 h1,2. In addition to regulating basic biological processes, the circadian clock also has a strong influence on memory. Long-term memory, which is transcription-dependent, shows a strong time-of-day effect, with peak memory performance during the day (inactive phase) in mice3,4. Notably, both long-term memory and circadian rhythmicity are impaired with age5, suggesting that common molecular mechanisms might underlie both processes. One idea is that clock genes located in memory-relevant structures, like the dorsal hippocampus, might gate an animal’s ability to form long-term memory based on the time of day6. Consistent with this, dis- ruption of several individual clock genes throughout the brain can impair hippocampal long-term memory in young animals. As no study to date has selectively disrupted circadian gene function within the dorsal hippocampus, it is unclear whether clock genes act within hippocampal cells to affect long-term memory formation or whether these memory deficits result from off-target effects in other brain regions, such as impaired circadian rhythms, sleep deficits, or even developmental abnormalities.
Gene expression is decreased in the aging brain, which may be the consequence of a more repressive chromatin structure. Transcription is controlled in part through changes in chromatin structure, which can dynamically promote or restrict access to neuronal DNA following a learning event. One hypothesis put forth by Barnes and Sweatt posits that the epigenome is altered in aging neurons, resulting in a repressive chromatin structure that prevents normal gene expression required for long-term memory formation7. Several studies support this idea showing altered histone modification mechanisms in the aging brain8–10. How- ever, whether chromatin modification mechanisms abnormally regulate circadian gene expression in a key learning and memory brain region is unknown. Here, we examined this possibility by focusing on the role of histone deacetylase 3 (HDAC3)-dependent regulation of age-related memory and gene expression. We found that deletion or disruption of HDAC3 in the dorsal hippocampus ameliorates age-related impairments in long-term memory and synaptic plasticity in 18-month-old mice, an effect that appears to be mediated in part by the core circadian clock gene Period1 (Per1). Broadly, this work suggests that Per1 may be a mechanism contributing to age-related impairments in both long-term memory and circadian rhythmicity, depending on the structure.
Results
HDAC3 contributes to age-related memory impairments. We first tested whether the repressive histone deacetylase HDAC3 plays a role in age-related memory decline. HDAC3 is a potent negative regulator of memory formation and disruption of HDAC3 in young animals can transform a subthreshold learning event into one generating persistent long-term memory for multiple tasks11–14. We used two methods of disrupting HDAC3 in the dorsal hippocampus of aging (18-month-old) mice. First, we created focal homozygous deletions of HDAC3 by infusing AAV2.1-CaMKII-Cre recombinase (1 μL per side) into the dorsal hippocampi of HDAC3flox/flox mice (Supplementary Fig. 1a). Second, to selectively block the enzymatic activity of HDAC3, we used a dominant-negative point mutant virus, AAV2.1-CMV- HDAC3(Y298H)-v5 that specifically blocks HDAC3 deacetylase activity without affecting protein-protein interactions (see ref. 12,15–17 Supplementary Fig. 1b). Viruses were infused two weeks before training, allowing for tight spatial and temporal control over our HDAC3 manipulations to avoid potential side effects that might occur from prolonged HDAC3 disruption during development18,19. Two weeks after AAV-CaMKII-Cre infusion (Supplementary Fig. 1c), we observed that Hdac3 mRNA expression was not affected by training in object location memory (OLM), but genetic deletion of hippocampal Hdac3 disrupted expression of Hdac3 mRNA (Supplementary Fig. 1d) in addition to HDAC3 protein (Supplementary Fig. 1a).
To determine whether deletion of HDAC3 improves memory in aging mice, we tested the effects of hippocampal HDAC3 deletion (HDAC3flox/flox) or activity disruption (HDAC3 (Y298H)) on long-term memory for OLM (Fig. 1a). Consistent with numerous reports of age-related hippocampal memory deficits, we found that aging, 18-m.o. wildtype (HDAC3+/+) mice displayed poor memory for OLM following 10-min training, showing no significant increase in DI between training and testing (Fig. 1b, gray bars). Importantly, a 10-min training session.
2 NATURE COMMUNICATIONS | (2018)9:3323 |DOI: 10.1038/s41467-018-05868-0 |www.nature.com/naturecommunications
Fig. 1 Deleting or disrupting HDAC3 ameliorates age-related impairments in hippocampal memory. an OLM procedure. AAV was infused 2 weeks before the training. b 18-m.o. HDAC3flox/flox mice showed significantly better memory for OLM compared to wild type (HDAC3+/+) littermates (Two- way ANOVA: Significant Genotype x Session interaction (F(1,29) = 15.96, p < 0.001), Sidak’s posthoc tests, ***p < 0.001, n = 14(5F), 17(6F)). c Total exploration was similar for both groups at test (t(29) = 1.67, ***p = 0.11). d Disrupting HDAC3 activity in the dorsal hippocampus with AAV-HDAC3 (Y298H)-V5 also ameliorated hippocampal memory impairments in 18-m. o. mice (Two-way ANOVA: main effect of session (F(1,16) = 15.96, p < 0.001), Sidak’s post hoc tests, ***p < 0.001, *p < 0.05, n = 9,10; all males). e Total exploration time was similar for both groups at test (t(16) = 0.28, p = 0.78). f ORM experimental procedure, 2 weeks after the completion of OLM. g Both 18-m.o. HDAC3flox/flox mice and HDAC3+/+ littermates showed little preference for the novel object (Two-way ANOVA, no main effects or interaction, n = 14(5 F), 17(6 F)). h Total exploration time was similar for both groups at test (t(29) = 0.59, p = 0.56). i Disrupting HDAC3 activity in the dorsal hippocampus with AAV-HDAC3(Y298H) also had no effect on ORM, with neither group showing a preference for the novel object (Two-way ANOVA, no main effects or interaction, n = 9,10; all males). j Groups showed similar total exploration time at test (t(16) = 0.28, p = 0.78). Data are presented as mean ± SEM; black circles, males; gray squares, females normally produce strong long-term memory in young mice20. In contrast, 18-m.o. HDAC3flox/flox littermates formed robust long-term memory (Fig. 1b, teal bars), with a significant increase in preference for the moving object at rest relative to training. HDAC3flox/flox mice showed a significantly higher DI at test than HDAC3+/+ mice despite similar levels of total exploration during the test session (Fig. 1c). We observed similar effects with the activity-specific AAV-HDAC3(Y298H) virus. Aging, 18-m.o. empty vector (EV) control mice showed poor memory for OLM whereas mice infused with AAV-HDAC3(Y298H) into the DH showed significantly higher preference for the moving object with no change in total exploration (Fig. 1d, e). In contrast to the poor long-term memory observed in 18-m.o. wildtype mice
(Fig. 1b, d), short-term memory for OLM (tested 60 m after
acquisition; Supplementary Fig. 2a) was intact for both HDAC3+/+ and HDAC3flox/flox mice (Supplementary Fig. 2b, c) and there was no significant difference in anxiety between these groups (Supplementary Fig. 2d). We also observed no significant difference in movement between the groups during habituation (Supplementary Figs. 2e, f, 8a, b). Together, these results demonstrate that age-related impairments in OLM are ameliorated by deletion or disruption of HDAC3 in the dorsal hippocampus.
We next tested whether our focal HDAC3 manipulation affected object recognition memory (ORM), which does not require the dorsal hippocampus for retrieval20. In this task, one familiar object is replaced by a novel item (Fig. 1f). Deleting HDAC3 in the dorsal hippocampus did not rescue memory for ORM, with both HDAC3+/+ and HDAC3flox/flox mice showing no preference for the novel object at test compared to training (Fig. 1g). Again, groups did not differ in total exploration levels during the test session (Fig. 1h). Similarly, activity-specific disruption of HDAC3 in the hippocampus was unable to ameliorate age-related ORM impairments (Fig. 1i, j). Thus, age-related impairments in long-term ORM were not ameliorated by hippocampal deletion or disruption of HDAC3.
Together, our results indicate that deletion or disruption of HDAC3 in the dorsal hippocampus can ameliorate age-related long-term memory deficits for a hippocampus-dependent task (OLM; Fig. 1a–e) without affecting memory for a hippocampus- independent task (ORM; Fig. 1f–j). Importantly, all mice showed intact short-term memory for OLM (Supplementary Fig. 2), suggesting that these animals acquire memory normally but fail to consolidate this information into observable long-term memory. As short-term memory is transcription-independent (for review21), this finding is consistent with the idea that learning-induced gene expression is altered in aging mice, resulting in age-related impairments in long-term memory.
HDAC3 disruption reverses age-related impairments in LTP.
To test whether HDAC3 also contributes to age-related synaptic plasticity impairments, we examined long-term potentiation (LTP) in acute hippocampal slices following either deletion or disruption of HDAC3. LTP is also impaired with age, particularly when the stimulation protocol is close to the LTP induction threshold22. Two weeks after viral infusion, we prepared hippocampal slices and induced LTP with a single train of 5 theta bursts to Schaffer collateral inputs and recorded field EPSPs from apical dendrites of CA1b. This relatively mild form of stimulation produced a stable level of LTP in young HDAC3+/+ slices (Fig. 2a). Deleting HDAC3 in the hippocampus enhanced LTP, with HDAC3flox/flox mice showing significantly higher potentiation than wild-type controls. As predicted, aging-18-m.o. HDAC3+/+ mice showed impaired LTP and the HDAC3 deletion ameliorated this deficit, producing LTP comparable to that of young wild-type mice (Fig. 2b, c) with no effect on baseline synaptic transmission (Fig. 2g, h).
We observed similar results with the activity-specific disruption. Here, we used a within-subjects design in which young and old wild-type mice were infused with the control virus (AAV-EV) into one hippocampus and AAV-HDAC3(Y298H) into the contralateral hippocampus. As before, we found that disrupting HDAC3 activity enhanced LTP in slices from young mice (Fig. 2d) and ameliorated age-related LTP impairments in slices from aging mice (Fig. 2e, f) without interfering with baseline synaptic transmission (Fig. 2i, j). Either deletion or disruption of HDAC3 can therefore ameliorate age-related impairments in hippocampal LTP.
A subset of age-impaired genes is improved by HDAC3 deletion. Our data suggest that deleting or disrupting HDAC3 ameliorates age-related impairments in long-term memory and synaptic plasticity. We next asked whether age-related deficits in hippocampal gene expression could also be ameliorated by deleting HDAC3. We hypothesized that dysregulation of HDAC3 activity in the old brain leads to an unusually repressive chromatin structure that limits gene expression, which ultimately impairs long-term memory. To identify which specific genes are regulated by HDAC3 in the young and aging brain, we ran an RNA sequencing experiment in which we compared young (3-m.o.) wild-type mice, aging (18-m.o.) HDAC3+/+ mice, and aging (18-m.o.) HDAC3flox/flox littermates. To identify gene expression changes during memory consolidation, animals in each group were killed 60 m after 10-min OLM training and compared to homecage (HC) controls. After mapping and considering the haploid genome23,24, sequencing quality was assessed (Supplementary Fig. 3a, b) and significant differences in expression profiles were examined between all pairs of samples for p < 0.0520.
We expected that experience-induced gene expression would be altered in the old brain, as previously described, with a subset of genes failing to express after learning. We, therefore, focused on those genes expressed at significantly higher levels in the trained groups compared to homecage controls. While each group (Young WT, Old WT, Old HDAC3flox/flox) showed a substantial number of genes induced by OLM training, each group showed a unique gene expression profile (Fig. 3a, Supplementary Tables 1–3). Old brains upregulated in both the young wildtype and old HDAC3flox/flox groups that do not show experience-induced increases in the old wildtype hippocampus. These are the genes that fail to normally express in the old brain after OLM training but are rescued by HDAC3 deletion and may therefore function as a mechanism through which HDAC3 deletion ameliorates age-related memory impairments. Four genes were identified in this group: Nr4a1, Egr1, Tsc22d3, and Per1. All of these genes have been broadly implicated in memory formation6,26,27, although this is the first study to demonstrate that experience-induced expression of these genes is impaired with age and rescued with HDAC3 deletion.
Deleting HDAC3 ameliorates age-related deficits in Per1. Of the genes identified through our RNA-seq, Period1 (Per1) was the most strongly induced by OLM training in the HDAC3flox/flox group (Fig. 3c). This target is particularly intriguing, as Per1 is typically studied in the context of circadian rhythms but has also been implicated in hippocampal memory formation6,28,29. As aging is known to be accompanied by impairments in circadian rhythms5 and memory is linked to time-of-day2,30, this target of HDAC3 may represent a critical and underexplored interface between aging, chromatin modification, and the circadian clock.
To further examine the expression of Per1 in old HDAC3+/+ and old HDAC3flox/flox animals, we used RT-qPCR and ChIP- qPCR. We found that OLM training failed to induce upregulation of Per1 in the dorsal hippocampus of 18-m.o. HDAC3+/+ mice, but in the absence of HDAC3 (HDAC3flox/flox), OLM training triggered a significant increase in Per1 mRNA (Fig. 3f). We also measured the expression of two additional genes that play a well-documented role in long-term memory formation: Arc and cFos31. Experience-induced expression of Arc was intact in the aging wildtype brain and was unaffected by HDAC3 deletion (Supplementary Fig. 3c). cFos expression, on the other hand, failed to be induced by OLM training in the old HDAC3+/+ hippocampus, but deleting HDAC3 was not sufficient to restore this failed induction (Supplementary Fig. 3d). Deleting HDAC3 therefore only restores expression of a subset of experience-induced genes in the aging brain, including Per1.
To determine whether deleting HDAC3 restores expression of Per1 by promoting histone acetylation along with its promoter, we next measured acetylation of histone 4, lysine 8 (H4K8Ac) at the Per1 CRE promoter site using chromatin immunoprecipitation (ChIP-qPCR). H4K8Ac is a marker of transcriptional activation32 and is thought to be a target of HDAC311,12. OLM training did not change H4K8Ac levels at the Per1 promoter in the old wildtype brain but in the absence of HDAC3, H4K8Ac levels at the Per1 promoter were significantly increased in response to OLM training (Fig. 3g). For Arc and cFos, we saw no change H4K8Ac occupancy across groups (Supplementary Fig. 3e, f). Together, these results suggest that deleting HDAC3 increases acetylation at the Per1 promoter and expression ofPer1 mRNA in response to learning.
One important question is whether the observed changes in Per1 are due to changes in the circadian rhythm of HDAC3flox/ flox mice. If deletion of HDAC3 in the dorsal hippocampus alters circadian rhythmicity, this could explain the observed changes in both Per1 expression and long-term memory formation. To rule this out, we assessed the circadian rhythmicity of young (3-m.o.) and aging (18-m.o.) HDAC3flox/flox mice and their HDAC3+/+ littermates following AAV-CaMKII-Cre infusion. After 2 weeks of entrainment to a 12 h light/dark cycle (LD), mice were put in constant darkness (DD) to measure endogenous circadian rhythms (Supplementary Fig. 4a). We observed no difference in circadian activity patterns between HDAC3+/+ and HDAC3flox/ flox mice at either age group (Supplementary Fig. 4b–f), suggesting that hippocampal HDAC3 has no effect on circadian rhythmicity. Thus, the observed changes in Per1 expression and long-term memory formation following HDAC3 deletion cannot be explained by changes in circadian rhythmicity.
Per1 is induced by OLM training and regulated by HDAC3.
We next wanted to determine whether hippocampal Per1 is required for long-term memory formation. First, we assessed whether hippocampus-dependent learning typically induces Per1 mRNA expression. We sacrificed young (3-m.o.) wildtype mice 60 m after the acquisition of either OLM or context fear conditioning (CFC)12. Per1 mRNA expression was significantly upregulated in animals trained with either OLM (Fig. 4a) or CFC (Fig. 4b) compared to homecage controls, indicating that Per1 mRNA expression is typically induced in the hippocampus during memory consolidation for multiple tasks.
To determine whether this experience-induced increase in Per1 might be mediated through HDAC3, we next used ChIP-qPCR to measure HDAC3 occupancy after OLM training at different sites along with the Per1 promoter in the young hippocampus (Fig. 4c, top). We found that HDAC3 occupancy at the Per1 promoter was reduced at all three tested sites following OLM training (Fig. 4c, bottom). Along with our previous finding that HDAC3 deletion restores both acetylations at Per1 and Per1 mRNA expression (Fig. 3), this strongly suggests that HDAC3 regulates Per1 expression in the dorsal hippocampus and dysregulation of HDAC3 may contribute to age-related impairments in experience-induced Per1 expression. PER1 may therefore be part of a key mechanism through which HDAC3 deletion ameliorates age-related impairments in hippocampal long-term memory formation.
