Elsevier

Neuroscience

Volume 300, 6 August 2015, Pages 39-52
Neuroscience

Neuroscience Forefront Review
Building up and knocking down: An emerging role for epigenetics and proteasomal degradation in systems consolidation

https://doi.org/10.1016/j.neuroscience.2015.05.005Get rights and content

Highlights

  • Recently acquired memories depend on the hippocampus; maintenance of remote memories depends on the cortex.

  • Transcription is required for memory.

  • Epigenetic modifications regulate short-term transcription in the hippocampus.

  • Persistent cortical DNA methylation regulates transcription and memory maintenance.

  • Active protein degradation paves way for transcription to support memory formation.

Abstract

Memory formation is a protracted process in which recently acquired events are consolidated to produce stable and specific associations. Initially, newly acquired information undergoes cellular consolidation in the hippocampus, which transiently supports the storage of recently acquired memories. In contrast, remote, or “old” memories are maintained in the cortex and show almost complete independence from the hippocampus. Memories are transferred from the hippocampus to the cortex through a process termed systems consolidation. Emerging evidence suggests that recurrent activation, or “training” of the cortex by the hippocampus is vital to systems consolidation. This process involves prolonged waves of memory-related gene activity in the hippocampus and cortex long after the learning event has terminated. Indeed, molecular events occurring within hours and days of fear conditioning are essential for stabilizing and eventually transitioning the memory to the cortex. It is increasingly evident that molecular mechanisms that exhibit a capacity for prolonged activation may underlie systems consolidation. Processes that have the capacity to control protein abundance over long time scales, such as epigenetic modifications, are prime candidates for the molecular mechanism of systems consolidation. Indeed, recent work has established two types of epigenetic modifications as integral for systems consolidation. First, localized nucleosomal histone variant exchange and histone modifications are integral for early stages of systems consolidation, whereas DNA methylation appears to be utilized to form stable marks that support memory maintenance. Since systems consolidation also requires discrete and time-sensitive changes in protein abundance, additional mechanisms, such as protein degradation, need also be considered, although their role in systems consolidation has yet to be investigated. Here, we discuss the role of molecular mechanisms in systems consolidation and their implications for understanding how memories persist over time.

Introduction

Memory formation is a complex process that requires several rounds of molecular and cellular modifications within neurons that form the memory trace (Zovkic and Sweatt, 2013). These modifications occur at multiple and interacting levels that shift neurons from a pre-learning state, characterized by high levels of memory suppressor genes (Abel et al., 1998) to a consolidation state, characterized by high levels of memory-promoting genes (Zovkic and Sweatt, 2013). Such shifts require coordinated activity of machinery involved in both protein degradation and gene transcription, and indeed, interference with either of these processes is detrimental to memory (Frick, 2013, Zovkic et al., 2013, Jarome and Helmstetter, 2014). However, these processes tend to be studied independently and very little is known about the ways in which they interact. Moreover, the relative function and contribution of each level of regulation may change over the protracted time course of memory formation, which is characterized by modifications occurring in different brain regions at different points in time (Frankland and Bontempi, 2005, Miller et al., 2010, Lesburgueres et al., 2011, Zovkic et al., 2014). Here, we review the literature pertaining to the molecular mechanisms of systems consolidation, focusing particularly on epigenetic modifications that regulate the induction of gene activity and in protein degradation that is involved in resetting cellular states. Additionally we speculate how these mechanisms may be generalized to inter-regional communication, specifically systems consolidation.

Human case studies initially identified the hippocampus as a critical site for memory acquisition (Milner, 2005), but evidence for sparing of older, remote memories in patients with hippocampal lesions suggested that the hippocampus may have a time-limited role in memory (e.g., Rosenbaum et al., 2000, Squire and Bayley, 2007, Huijgen and Samson, 2015). Indeed, a growing number of studies in animal models indicate that many forms of memory are transiently dependent on the hippocampus and are subsequently transferred to the cortex for long-term storage and maintenance (Frankland and Bontempi, 2005, Preston and Eichenbaum, 2013). This process is referred to as systems consolidation to reflect the temporally graded and multi-regional characteristics of memory consolidation (Frankland and Bontempi, 2005, Preston and Eichenbaum, 2013). Recent studies have tremendously expanded our understanding of the molecular basis of hippocampal memory consolidation, but very little is known about the molecular mechanisms involved in network reorganization and long-term memory storage. This review will focus primarily on newly emerging studies of the molecular basis of systems consolidation and the open questions needed to understand the prolonged, multi-regional process of memory stabilization and storage.

Various lines of evidence demonstrate that memories undergo systems consolidation in rodents, including behavioral outcomes of localized lesions, time- and region-dependent changes in molecular markers, and morphological changes time-locked to different stages of memory formation. Some of the best characterized data come from lesion studies in the hippocampus and the cortex (reviewed in Frankland and Bontempi, 2005). In general, transient inactivation or permanent lesions of the hippocampus shortly after training (immediately or after 24 h) impair both recent (24 h–2 days) and remote (⩾7 days after training) memory, whereas hippocampal lesions at later time points have no impact on remote memory (e.g., Kim and Fanselow, 1992, Maviel et al., 2004), suggesting that remote memories do not require the hippocampus for recall once they are established. In contrast, inactivation of the prefrontal cortex (PFC) or the anterior cingulate cortex (ACC) shortly before testing selectively impair remote memory without impacting recent memory (Frankland et al., 2004, Maviel et al., 2004), supporting the hypothesis that memory is initially dependent on the hippocampus and is subsequently downloaded to the cortex for maintenance.

Consistent with evidence from lesion studies, molecular and imaging experiments have found shifts in activity from the hippocampus to the cortex during the recall of recent and remote memory, respectively, providing naturalistic support for data from lesion studies. For example, functional imaging studies in mice showed that the hippocampus is highly active when recalling recent radial arm maze memory, whereas high levels of ACC activity come online during remote (25 days) memory recall (Bontempi et al., 1999). Using a molecular approach, Maviel et al. (2004) showed that the recall of recent memory for a baited arm in a five-arm maze increased the expression of immediate early genes (IEGs) Egr1 and cFos in the hippocampus and not in the cortex, whereas the recall of remote memory selectively activated these IEGs in the prefrontal, ACC, and retrosplenial cortices. Similar results were obtained for water maze training and contextual fear conditioning, wherein Egr1 and cFos expression increased in the ACC, infralimbic, prelimbic and temporal cortices after remote (36 days), but not recent (1 day) memory recall (Frankland et al., 2004, Teixeira et al., 2006). Interestingly, memory recall at remote time points has been associated with reduced IEG expression in the hippocampus compared to the recent time point, suggesting that memory reorganization may involve active inhibition of the hippocampus during remote memory recall (Maviel et al., 2004).

There is some evidence to suggest that increased IEG expression in the cortex at remote time points is blocked in α-CaMKII+/− mutant mice, which have a specific deficit in remote memory (Frankland et al., 2001). Indeed, α-CaMKII+/− mice exhibit a dissociation between recent and remote memory, wherein memory for conditioned fear and the water maze is normal at recent and impaired at remote time points (Frankland et al., 2001). These mice have normal long-term potentiation (LTP) in the CA1 subregion of the hippocampus and impaired LTP in the cortex (visual and temporal) (Frankland et al., 2001), providing further support for the preferential role for the cortex in remote memory.

Consistent with the time course of molecular changes in the hippocampus and the cortex, spine density in the hippocampus increases rapidly and transiently after learning, whereas increased spine density in the ACC emerges more slowly (within 8 days of training) and persists for at least 48 days (Restivo et al., 2009, Vetere et al., 2011). Inhibiting the early growth of spines in the ACC impairs training-induced increase in spine density and memory recall 48 days after training (Vetere et al., 2011), indicating that changes in spine density support the maintenance of remote memory. Interestingly, MEF2, a negative regulator of spine density, was effective at inhibiting cortical spine density and remote memory when administered within 1 week of training, but not when administered 7 weeks after training (Vetere et al., 2011). These data indicate that early changes in spine density may reflect an ongoing reactivation process that is less pronounced at remote stages (Vetere et al., 2011). Remote memory has also been associated with increased cortical expression of GAP-43 (Growth associated protein 43), a marker of new synapses (Frankland et al., 2004, Maviel et al., 2004), providing further support for persistent structural reorganization of cortical regions during remote memory.

The phenomenon of temporally graded reorganization of stored information may be a ubiquitous feature of memory storage, as suggested by recent evidence for an analogous process in drosophila. In drosophila, olfactory memories appear to shift between different classes of mushroom body neurons (MBNs) at early (15 min–3 h) and late (24 h) time points after training (Cervantes-Sandoval et al., 2013). Parallel to inter-regional memory transfer in mammals, α’/β’ MBNs are activated shortly after learning and their inactivation selectively impairs early memory, whereas α/β MBNs are activated at later time points and their inactivation impairs older memory (Cervantes-Sandoval et al., 2013).

These data indicate that systems consolidation may be a conserved process that at least partly divides the tasks of memory acquisition and memory maintenance. Although the purpose of such a division is not clear, computational models suggest that systems consolidation allows for the acquisition of new memories without simultaneously erasing old ones. In this system, highly plastic cells mediate the formation of strong memories that degrade quickly, whereas cells with lower levels of plasticity support stable, long-lasting memories (Roxin and Fusi, 2013). Such differences in functional end points of memory acquisition and memory stability may thus dictate a need for distinct neural and cellular systems for each process (Abraham and Robins, 2005, Roxin and Fusi, 2013). This proposition is consistent with evidence that hippocampal neurogenesis may actively promote forgetting of older memories (Feng et al., 2001, Frankland et al., 2013, Akers et al., 2014), and that hippocampal activity is inhibited when recalling remote memories (Bontempi et al., 1999). In both cases, the data are consistent with a system in which maintenance of old memories is balanced against interference from newly acquired information through reliance on separate systems. The evidence from lesion, morphological and molecular studies suggest that each memory stage is indeed supported by distinct modifications in distinct brain regions, supporting the notion of separate but related systems underlying memory acquisition and maintenance.

The main premise of systems consolidation theory in general and this review in particular is that memory formation is a multistage process that involves some degree of information transfer from one brain region to another. For our purposes, we will focus primarily on the transfer of memories from the hippocampus to the cortex because this network is best characterized at the present time. We will also work under a certain set of assumptions to organize our arguments. First, we assume that initial stages of memory consolidation occur in the hippocampus and that concurrent reactivation of the hippocampus and the cortex underlies systems consolidation. PFC is rapidly activated in response to learning (Lesburgueres et al., 2011, Zovkic et al., 2014), indicating that the hippocampus and the cortex process memories in parallel. However, impairments of both recent and remote memory produced by hippocampal lesions shortly after training (Frankland and Bontempi, 2005) suggest that the cortical memory trace is initially insufficient to support memory and must first undergo hippocampus-dependent ‘training’ (Born and Wilhelm, 2012). In agreement with this interpretation, communication between the hippocampus and the cortex is ongoing during sleep, with activity in the hippocampus preceding activity in the cortex (Born and Wilhelm, 2012). Overall, these data suggest that memory transfer and reorganization is gradual and occurs through repeated interactions between the hippocampus and the cortex.

For discussion purposes, we assume that the formation of remote memory goes through three main stages, beginning with cellular or synaptic consolidation in the hippocampus. This stage is initiated rapidly after learning (0–3 h) and is characterized by local molecular alterations in gene expression, protein synthesis, and protein degradation, which support synaptic restructuring and are necessary for recent memory formation (see Fig. 1) (Rudy and Sutherland, 2008, Jarome and Helmstetter, 2014). As mentioned previously, disrupting cellular consolidation shortly after training also impairs remote memory, indicating that early hippocampal activity is necessary for systems consolidation.

The systems consolidation stage may be initiated in parallel with cellular consolidation, but is much slower and longer lasting, persisting from the time of learning until approximately 7–30 days later in rodents (see Fig. 1). This stage involves ongoing communication between the hippocampus and the cortex and ultimately results in reduced reliance on the hippocampus and increased reliance on the cortex as memories enter the final maintenance stage (Frankland and Bontempi, 2005). It is important to note that maintenance is not static and that remote memories may still be subject to reconsolidation, which can result in memories being modified after recall, or updated with the addition of new information (Abraham and Robins, 2005, Graff et al., 2014). Moreover, categorically similar memories may be incorporated into higher order schemas, which rapidly integrate new memories into existing systems of knowledge (Tse et al., 2011, Wang et al., 2012, van Buuren et al., 2014, van Kesteren et al., 2014). Thus, the ‘final’ stage of memory maintenance does not preclude ongoing plasticity and can be considered as an effective tool for characterizing and understanding various stages of memory formation.

Cellular consolidation in the hippocampus occurs within hours of learning (see Fig. 1) and has been the primary subject of research into the molecular basis of memory. This consolidation stage has been extensively reviewed elsewhere (Peixoto and Abel, 2013, Zovkic et al., 2013, Kandel et al., 2014), so we will mention it here only briefly. Studies of cellular consolidation have identified key molecular mechanisms that are conserved across different stages of memory formation, most prominently alterations in protein synthesis and gene expression (Zovkic et al., 2013). Although less well established, the regulated removal of pre-existing gene products by the proteasome system is also emerging as a crucial factor in cellular consolidation (see Fig. 1) (Jarome and Helmstetter, 2014) and the implications of this process for memory will be discussed in detail in later sections.

The majority of molecular changes that occur during cellular consolidation and underlie the formation of recent memory in the hippocampus are transient, meaning that molecular changes are induced rapidly after learning (within 15 min) and return to pre-training levels within 2–3 h (Fig. 1) (Huff et al., 2006, Zovkic et al., 2013). Such rapid and stimulus-mediated induction of molecular events is consistent with the proposition that high levels of plasticity are required for memory acquisition, whereas slower responding systems are needed for prolonged memory storage (Roxin and Fusi, 2013). Also consistent with this proposition is the idea that the memory itself is preferentially susceptible to pharmacological manipulations of protein degradation, synthesis, transcription and epigenetic modifications during this time early time period (0–3 h), indicating that molecular modifications occurring during the cellular consolidation window underlie rapid stimulus-induced plasticity needed for memory acquisition (Miller and Sweatt, 2007, Rossato et al., 2009, Jarome and Helmstetter, 2014). The relevance of this early phase is further supported by evidence that many of the manipulations that disrupt memory shortly after learning do not impact recent memory when they occur outside of this initial consolidation window [e.g., 6 h after training (Miller and Sweatt, 2007, Maddox and Schafe, 2011)], indicating that this early wave of molecular events is sufficient for memory stabilization.

The tremendous relevance of this rapid consolidation window for memory formation has resulted in nearly exclusive research focus on modifications occurring shortly after learning. However, some recent studies identified additional and delayed waves of hippocampal gene expression that occur beginning at 8 h after training and selectively impact remote memory (Taubenfeld et al., 2001, Merhav et al., 2006, Bekinschtein et al., 2007, Katche et al., 2010, Katche et al., 2012, Katche et al., 2013), suggesting that temporally distinct events in the hippocampus regulate different stages of memory. For example, a second wave of cFos expression re-appears 24 h after learning, but only for memories that are strong enough to persist at remote time points (Katche et al., 2010). Transcriptional inhibitors and inhibition of cFos expression 24 h after learning selectively interfere with remote memory (7 days), leaving recent memory (48 h) intact (Katche et al., 2010). Similar data were found with Egr1 and Bdnf with a slightly different time course, with a second wave of expression occurring between 8 h and 12 h after training and selectively regulating remote memory. The selective disruption of remote memory obtained with inhibition of these individual target genes was replicated with general inhibitors of transcription or protein-synthesis at these delayed time points, indicating that similar molecular events underlie cellular and systems consolidation, but with a different time course. That is, whereas initial stimulus-induced molecular events can stabilize recent memories, an additional and delayed wave of gene activity is required to ensure the longevity of those memories at the remote stage.

Many questions remain regarding the mechanisms that mediate the onset and termination of the distinct waves of gene activity. One possibility is that changes in gene expression occur in stages, with genes activated in an earlier phase stimulating the expression of genes in the later phase. Indeed, IEGs have a well-established role in regulating the expression of delayed response genes and are themselves subject to regulation by other activity-regulated genes (Alder et al., 2003, Glorioso et al., 2006, Loebrich and Nedivi, 2009). Consistent with this notion, blocking training-induced Bdnf induction 12 h after learning blocked delayed cFos expression at 24 h (Bekinschtein et al., 2007), indicating that training-induced gene expression is sequentially propagated in the hippocampus to support ongoing changes in gene activity needed for systems consolidation. In turn, delayed waves of Bdnf expression may be mediated by delayed waves of dopamine activity (Bernabeu et al., 1997, Rossato et al., 2009), suggesting that temporally specific neural activity is essential for coordinating the transcription patterns needed to stabilize memories.

Given the multiregional interaction in systems consolidation, molecular events in the hippocampus likely induce changes in the cortex and are in turn modified by feedback from cortical regions (Preston and Eichenbaum, 2013). Although this notion is relatively well-accepted, there is a notable absence of molecular evidence for this interaction. However, Lesburgueres and colleagues (2011) recently reported that inactivating the hippocampus at early time points after training blocked memory-related increases in cortical spine density at remote time points, suggesting that cortical changes in spine morphology are driven by hippocampal activity. Similarly, cortical inactivation resulted in prolonged IEG expression in the hippocampus at delayed time points (Lesburgueres et al., 2011), indicating a bidirectional influence between these regions. Interestingly, inactivating the PFC at early time points impaired remote memory even though recent memory was intact, indicating that early cortical activation is required to establish connections that support remote memory (Lesburgueres et al., 2011). The authors interpreted this early cortical activity as evidence of a cortical tagging process, which stabilizes appropriate neuronal communication between the hippocampus and the cortex during the protracted process of systems consolidation.

The final stage of memory maintenance has received comparatively less attention than the cellular consolidation stage. However, available evidence suggests that memory storage is an active and ongoing process that can also be disrupted by molecular manipulations (see Fig. 1). For example, inducible and transient knockout of the NR1 subunit of the NMDA receptor in the forebrain for 4 weeks, starting 6 months after training, disrupted recall of contextual fear memory 9 months after training, whereas 7 days of disruption was not sufficient to produce a memory deficit (Cui et al., 2004). It is not clear which of the affected forebrain regions this effect can be attributed to, but LTP was selectively impaired in the cortex (Cui et al., 2004) and CA1-specific NMDA deletion 4 weeks after training had no effect on remote memory (Shimizu et al., 2000). The authors hypothesized that the NMDA receptor may be a synaptic re-entry reinforcement (SRR) mechanism, allowing neurons that reactivate together to remain connected and the synaptic machinery to be accurately placed during receptor turnover, thereby maintaining memory specificity (Cui et al., 2004).

Epigenetic modifications initially entered memory research because of their potential to persist for the duration of the memory, based on their established role in maintaining cellular identity (Barrero et al., 2010). In addition to their potential for stability, epigenetic mechanisms are attractive candidates for memory because of their role as transcriptional regulators. In contrast to specific transcription factors, epigenetic mechanisms have the capacity to orchestrate gene expression programs by altering the likelihood that a gene will turn on or off in response to a particular stimulus (McQuown and Wood, 2011, Baker-Andresen et al., 2013). Epigenetic modifications of histones operate upstream of gene expression by regulating access of transcriptional machinery to underlying DNA. Specifically, stretches of ∼147 bp of DNA form a nucleosome by wrapping around a histone octamer, composed of two each of histones H2A, H2B, H3 and H4. Modifications of existing histones, such as acetylation, phosphorylation and methylation, modify histone–DNA interactions and mediate the recruitment of additional DNA-binding factors (Peixoto and Abel, 2013), thereby mediating gene expression. Numerous studies have now established a connection between histone modifications in the hippocampus and the formation of recent memory (Levenson et al., 2004, Lubin et al., 2008, Miller et al., 2008, Peixoto and Abel, 2013). In addition, we recently showed that histone variant exchange, in which canonical histones are replaced by their variant counterparts, is a novel epigenetic regulator of memory and gene expression in the brain (Zovkic et al., 2014), adding additional complexity to the interplay between epigenetics and memory formation.

In addition to histone dynamics, methylation of cytosine bases is a classical epigenetic mechanism that has been associated with gene repression, although recent evidence suggests that this effect may vary based on specific properties, such as the site of DNA methylation (Suzuki and Bird, 2008, Medvedeva et al., 2014). In general, DNA methylation in the promoter tends to be associated with gene repression, whereas methylation within gene bodies may be permissive for transcription (Suzuki and Bird, 2008). Although DNA methylation was long thought to be a stable and irreversible epigenetic mark, it is now known that the conversion of methylated to hydroxymethylated cytosines by ten-eleven-translocation (TET) enzymes is a pathway to active DNA demethylation (Bhutani et al., 2011, Gong and Zhu, 2011), indicating that rapid cycling between methylation and de-methylation may be crucial for memory. Indeed, recent studies showed that TET1 regulates hippocampus-dependent fear memory and the expression of memory-related genes (Kaas et al., 2013, Rudenko et al., 2013), suggesting that conversion of methylated to hydroxymethylated cytosines may be essential for memory formation. In addition to acting as an intermediate in DNA demethylation, hydroxymethylation may be a relatively stable epigenetic mark in its own right (Hahn et al., 2014, Irwin et al., 2014), indicating a potentially unique role in memory. Although the function of hydroxymethylation in remote memory and systems consolidation has never been studied directly, efforts to tease apart distinct types of methylation will be essential for clarifying their role in memory.

Initially, the capacity of DNA methylation to mediate long-term changes in gene expression, as evidenced by its role in maintaining cellular identity (Barrero et al., 2010), led to the hypothesis that DNA methylation is a self-perpetuating mark with a unique capacity to persist for the duration of memory (Crick, 1984, Miller and Sweatt, 2007). Indeed, changes in DNA methylation occur in association with memory formation and are required for consolidating long-term fear memory (Miller and Sweatt, 2007, Lubin et al., 2008, Miller et al., 2008, Miller et al., 2010). However, in contrast to the hypothesized stability of DNA methylation, this modification is rapidly induced (within 1 h) and reversed (within 24 h) to baseline levels in the hippocampus during fear conditioning (Miller and Sweatt, 2007), likely through a demethylation process involving TET-mediated conversion of methylated to hydroxymethylated cytosines (Kaas et al., 2013, Rudenko et al., 2013). Several studies have replicated these findings in the amygdala using cued fear conditioning (Maddox and Schafe, 2011, Monsey et al., 2011, Maddox et al., 2014), indicating that dynamic DNA methylation has a temporally limited role in the hippocampus during memory formation. Indeed, the half-life of DNA methylation varies with the model used and the brain region being studied. In some models, DNA methylation has a half-life as short as 10 min, while in brain regions like the cortex, DNA methylation can persist for at least 30 days (Miller and Sweatt, 2007, Miller et al., 2010, Meagher, 2014).

Similarly, rapid turnover is also found for histone modifications during the process of memory formation, with modifications emerging within 30 min and returning to baseline levels within hours (Peixoto and Abel, 2013). According to reports in various model systems, histone modifications can have half-lives as short as 2 min (Meagher, 2014), suggesting that PTMs have the capacity to respond very rapidly to environmental stimuli. Most recently, we identified histone variant exchange as another chromatin modification that occurs within 30 min of fear conditioning and returns to baseline levels within 2 h in the hippocampus (Zovkic et al., 2014). Specifically, we showed that that histone H2A.Z, a variant of the canonical histone H2A, is removed from the first nucleosome downstream of the transcription start site (TSS) of memory-related genes and that this removal is associated with gene induction. Overall, these data demonstrate that multiple levels of epigenetic modifications in the hippocampus occur rapidly in response to learning and are terminated shortly thereafter to regulate the initial stage of memory consolidation by modifying gene expression. Indeed, such rapid turnover of histones and their modifications may help the hippocampus become more responsive to external stimuli.

Given the time-limited role of the hippocampus in memory consolidation, Miller and colleagues posited that transient changes in DNA methylation mimic the transient role of the hippocampus in memory, hypothesizing that persistent changes in DNA methylation would instead occur in the cortex, where remote memories are maintained over time (Miller et al., 2010). Consistent with this hypothesis, fear conditioning resulted in reduced methylation of the Egr1 promoter that persisted for 30 days and was associated with increased Egr1 expression 1 h after remote memory recall at 30 days (Miller et al., 2010). Similarly, the memory suppressor calcineurin exhibited a delayed increase in promoter methylation 1 day after fear conditioning that persisted for at least 30 days after training and was associated with reduced calcineurin expression at 30 days (Miller et al., 2010). Blocking cortical methylation at the remote time point impaired memory recall, whereas blocking methylation 1 day after training had no effect on remote memory, indicating that stable changes in cortical DNA methylation support memory maintenance. Taken together, these data suggest that the hippocampus, which supports recent memory, requires transient changes in the epigenome, whereas stable epigenetic modifications in the cortex are maintained over the duration of memory maintenance.

Memory maintenance was also investigated in relation to histone modifications using the novel object recognition paradigm, which confirmed the transient role of epigenetic modifications in the hippocampus and their prolonged role in the PFC (Graff et al., 2012). In contrast to previous studies, which measured epigenetic changes in the absence of recall, Graff et al. (2012) found that changes in histone acetylation were only evident after recalling recent and remote memories. In general, histone modifications in the hippocampus peaked after recent memory recall, although the exact temporal pattern varied for distinct modifications. For example, phosphorylation of H3 at serine 10 was increased during short-term (10 min) and recent (1 day) memory recall, whereas acetylation of H3 at lysines 14 and 5 was only increased after recalling a recent memory. In contrast, all of these modifications were increased after both recent and remote (7 day) recall in the PFC, consistent with a persistent role of epigenetic modifications in the cortex. These data support a role for histone modifications in memory maintenance, but it is unclear whether the changes observed at 7 days would persist at later time points that are typically associated with memory maintenance (i.e., 30 days after learning). In addition, the recall-specific occurrence of these effects implicates histone modifications as potential regulators of memory updating during recall-induced reconsolidation rather than as regulators of memory maintenance.

Intriguingly, the early emergence of epigenetic modifications in the cortex within 24 h of learning (Miller et al., 2010, Graff et al., 2012, Zovkic et al., 2014) indicates that cortical epigenetic changes may be relevant for remote memory long before the memory becomes reliant on the cortex for maintenance (7 and 30 days) (Miller et al., 2010). Indeed, this early stage of cortical activation likely reflects its communication with the hippocampus and is consistent with evidence for cortical histone acetylation “tagging” cells to which the memory will ultimately be transferred (Lesburgueres et al., 2011). It is probable that the time-limited nature of histone modifications in the cortex can be attributed to the termination of ongoing hippocampal input as the memory becomes increasingly dependent on the cortex. Indeed, histone modifications appear to be particularly responsive to stimuli, which is consistent with their rapid turnover (Meagher, 2014) and with evidence that histone modifications at delayed time points occur only in response to recall (Graff et al., 2012). In contrast, selected changes in cortical DNA methylation emerge only after 24 h and persist for at least 30 days (Miller et al., 2010), suggesting that DNA methylation may be better suited for stable regulation of memory maintenance. The general sequence of events, in which histone modifications are rapidly induced compared to DNA methylation, has been observed in other systems, which find that post-translational modifications tend to precede changes in DNA methylation (Strunnikova et al., 2005, Meldi et al., 2012). Together, these data suggest that histone modifications and DNA methylation may have distinct roles in memory acquisition and maintenance.

Of note, however, changes in histone H2A.Z incorporation persisted in the cortex 7 days after learning even in the absence of recall (Zovkic et al., 2014), suggesting that histone variants may have greater capacity for longevity compared to histone acetylation. However, even H2A.Z modifications returned to basal levels by 30 days, leading us to hypothesize that H2A.Z may be preferentially involved in systems consolidation rather than memory maintenance (Zovkic and Walters, 2015). We discuss this concept in greater detail below.

Evidence is beginning to emerge in support of epigenetic regulation of the early stages of systems consolidation, beginning immediately after learning and lasting for approximately 7 days. Using the social transmission of food preference paradigm, in which rats learn about food safety by interacting with a rat who recently consumed a novel food, Lesburgueres and colleagues (2011) showed that changes in histone acetylation are rapidly established in the orbitofrontal cortex (OFC) 1 h after training. Blocking histone acetylation in the OFC with an MSK1 inhibitor 1 h before training resulted in impaired remote memory at 30 days, whereas increasing histone acetylation with sodium butyrate selectively enhanced remote memory. Importantly, increasing histone acetylation was only effective at improving remote memory when administered for 6 days immediately after training and not when administered between 15 and 21 days after training, indicating that cortical histone acetylation is especially relevant during the early period of systems consolidation. The authors proposed that cortical acetylation functions as a tagging system that ensures accurate reactivation of appropriate cortical neurons during the protracted process of systems consolidation. Such rapid time course of histone effects on memory is similar to that observed for early interference with cortical spine density discussed earlier (Vetere et al., 2011), indicating that this early period is especially relevant for laying down remote memories. Overall, these data suggest that molecular modifications critical for systems consolidation occur rapidly after training and involve concurrent activation between the hippocampus and the cortex.

In a recent study, we showed that histone variant exchange in the cortex also regulates systems consolidation, particularly during the initial 7 days of fear conditioning. Specifically, we showed that histone H2A.Z, a variant of the canonical histone H2A, exhibits a distinct binding pattern to memory-related genes during early stages of systems consolidation that are no longer evident at 30 days (Zovkic et al., 2014). Cortical H2A.Z was rapidly removed from nucleosomes positioned immediately after the TSS (+1 nucleosomes) within 2 h of fear conditioning and in general, H2A.Z binding at this nucleosome returned to basal levels by 7 days. In contrast, H2A.Z binding exhibited a delayed increase at the first nucleosome before the TSS (−1 nucleosome) 7 days after fear conditioning, indicating that H2A.Z has distinct functions at different stages of systems consolidation. Moreover, the switch in H2A.Z binding between +1 and −1 nucleosomes over time suggests that the position of H2A.Z binding provides an additional level of regulation that is differentially utilized with changing demands required by distinct stages of memory stabilization (for a detailed review of this hypothesis, see (Zovkic and Walters, 2015). In addition to implicating histone variant exchange as a novel mechanism of systems consolidation, these data suggest that systems consolidation is a dynamic process, such that epigenetic modifications need not be static during this process. Moreover, group differences in H2A.Z binding were no longer evident at 30 days, indicating that distinct processes may be involved in systems consolidation and memory maintenance.

Consistent with the early removal of H2A.Z from memory-related genes (within 2 h of training), we showed that virally mediated depletion of cortical H2A.Z before training improved recall 7 days and 30 days later, but not 24 h after training, indicating that epigenetic modifications within the first 7 days are critical for regulating systems consolidation (Zovkic et al., 2014). However, H2A.Z was depleted throughout the training and testing periods, thus precluding precise conclusions about the temporal parameters underlying H2A.Z function in systems consolidation and memory maintenance. We are actively pursuing this question at the present time.

Our focus thus far has been on gene induction, but studies of gene regulation indicate that mRNA synthesis and protein degradation are intertwined processes that work in concert to regulate responses to stimuli (Schwanhausser et al., 2013a, Schwanhausser et al., 2013b). Indeed, the final stage of a protein is its eventual degradation, but degradation machinery has received limited attention in memory literature because it was originally viewed as a static housekeeping process, with little active role in dynamic memory-regulating processes. However, there is growing recognition that the ubiquitin proteasome system (UPS) is capable of dynamically regulating vital and time-dependent cellular processes, best exemplified by its role as a checkpoint in mitosis (Glotzer et al., 1991). The UPS accomplishes precise regulatory outcomes through the coordinated action of E1, E2, and E3 ligases, which act in coordination to specifically add ubiquitin to target proteins. This process can be repeated to form a poly-ubiquitin chain, which signals most efficiently for degradation of the protein at the proteasome. Ubiquitination is not a one-way reaction, as many de-ubiquitinating enzymes exist that can antagonize ubiquitination and impede degradation. This gives rise to a highly regulated process that can temporally and spatially modulate specific protein abundance.

With the speed and precision available to the UPS, it is little surprise that recent studies have identified an integral role for the UPS in synaptic activity. In the post-synapse, proteasomes are recruited to spines based on synaptic activity (Bingol and Schuman, 2006), regulate activity-dependent spine growth (Hamilton et al., 2012), and show increased rates of degradation after synaptic stimulation (Djakovic et al., 2009). Consistent with these observations, many proteins in the post-synaptic compartment are actively regulated by the UPS in response to synaptic activity, including such key players as the NMDA and AMPA receptors, as well as scaffolding proteins such as PSD95 (Tsai, 2014). The UPS is also indispensable in signaling pathways. For example, the UPS degrades IkB, the inhibitory component of the NF-kB pathway, allowing NF-kB to translocate to the nucleus (Chen, 2005), a key action for learning and memory (Kaltschmidt et al., 2006). The role of the UPS is not limited to post-synaptic proteins and has been demonstrated to play an important, but poorly understood role in regulating pre-synaptic processes. The UPS has been implicated in regulating vesicle numbers, both in vitro (Willeumier et al., 2006) and in vivo (Walters et al., 2014). Similarly, the UPS modulates the release probability of the synapse (Speese et al., 2003, Willeumier et al., 2006, Yao et al., 2007), as well as degradation of pre-synaptic proteins (Yao et al., 2007), implicating the UPS in memory-relevant processes.

In line with its role as a synaptic regulator, the UPS has a well-documented role in both short (Wilson et al., 2002, Walters et al., 2008, Walters et al., 2014) and long-term synaptic plasticity (Dong et al., 2008, Dong et al., 2014), and has been implicated in regulating memory formation (Jarome and Helmstetter, 2013). Infusing proteasome inhibitors into the amygdala and the PFC each produced memory deficits (Jarome et al., 2011, Reis et al., 2013, Fukushima et al., 2014), whereas infusion into the hippocampus produced mixed results. Specifically, infusing proteasome inhibitors into the hippocampus had no effect on contextual fear conditioning (Lee et al., 2008), whereas others did report deficits in inhibitory avoidance (Lopez-Salon et al., 2001) and the Morris water maze (Artinian et al., 2008). Thus, the proteasome may have region-specific effects on memory and play a more nuanced role in the hippocampus. The role of the proteasome has not been directly investigated in systems consolidation, but evidence that this system preferentially affects established memories (Lee et al., 2008, Fukushima et al., 2014) and has different roles in distinct memory-related regions implicates the UPS as a candidate for regulating systems consolidation. Here, we discuss some of the early evidence for the proteasome in memory formation and speculate that this system will be essential for understanding systems consolidation and memory maintenance.

The most intriguing role of the UPS may be to destabilize previously acquired fear memories, indicating that the UPS has a unique role in regulating memories once they are established. This role was demonstrated with studies of memory reconsolidation, which is induced in response to memory recall and can be disrupted by translational inhibitors (Nader et al., 2000). Specifically, recent studies found that effects of translational inhibitors on inhibitory avoidance and contextual fear conditioning are blocked by simultaneous treatment with a proteasome inhibitor, such that memories are spared from the normal amnesic effects of these drugs (Lee et al., 2008, Fukushima et al., 2014). Given that reconsolidation provides an opportunity for a memory to be modified or updated by new information, these data were interpreted to mean that the proteasome is needed to remove the prior memory for updating to occur.

To test this hypothesis, two separate research groups utilized novel modifications on inhibitory avoidance (Fukushima et al., 2014) and contextual fear conditioning (Lee et al., 2008) paradigms to examine the role of protein degradation on previously established memories. To this end, Lee and colleagues (2008) exposed mice to contextual fear conditioning followed by two recall sessions, with the first serving to place the already formed memory in a labile state and the second to test memory recall. Consistent with previous studies, translational inhibitors administered after the first recall session on day 2 impaired memory recall on day 3. However, when translational inhibitors were co-administered with a proteasome inhibitor, the memory was spared, such that the level of freezing on day 3 was nearly identical to the level of freezing on day 2 (Lee et al., 2008). Lee and colleagues further validated these findings using an extinction paradigm, which consists of repeated exposures to the context in the absence of foot shocks. Blocking proteasome activity prevented the extinction of contextual fear memory, supporting the hypothesis that the proteasome has a particularly essential role in removing existing memories (Lee et al., 2008).

In separate experiments, Fukushima and colleagues (2014) sought to differentiate recall from extinction to determine whether recall itself affects memory. Using the inhibitory avoidance task, mice were trained to avoid the naturally preferred dark chamber by pairing it with a mild foot shock and the memory is expressed as increased latency to enter the dark chamber on subsequent trials (Fukushima et al., 2014). In a varied version of the paradigm, mice were re-exposed to the training apparatus and immediately removed when they crossed the threshold to the dark chamber. This additional exposure resulted in even higher latency to enter the dark chamber, indicating that recall followed by reconsolidation produced a stronger memory compared to original training alone. Similar to the previous paradigm, administering translational inhibitors immediately after recall erased all memories. Intriguingly, co-administering translational inhibitors with a proteasome inhibitor prevented the increase in cross-over latency normally conferred by the recall session but left the original memory intact. Specifically, the mouse displayed the learning typical of a single training session (Fukushima et al., 2014) and lost all potentiation caused by the recall event. These data suggest that memory updating involves the clearance of proteins associated with original learning, indicating that the proteasome may destabilize the original memory to free up space for memory updating. This also implies that when memories are reconsolidated, the protein trace within neurons associated with the original memory is ablated and is replaced by newly synthesized proteins that establish the updated memory. Overall, these data suggest that the proteasome plays a unique role in synapses that maintain existing memories by removing proteins that underlie the initial memory during reconsolidation.

If we extend these observations to a hypothetical role for the UPS in systems consolidation, we may explain some of the issues that arise in studies of remote memories. We often operate under the assumption that remote memories are exact replicas of recent memories, but data from humans and rodents suggest that memories change over time, as evidenced by the gradual loss of contextual detail, or by the fear incubation phenomenon, in which fear memories strengthen over time (Rosenbaum et al., 2001, Pickens et al., 2009). Such normative changes in memory quality may be an outcome of memory updating assisted by the UPS. Systems consolidation involves passive reactivation of memory during sleep (O’Neill et al., 2008) and delayed reactivation of memory-related genes in the hippocampus (Katche et al., 2010), such that repeated instances of reactivation may provide extensive opportunities for memory reconsolidation or updating. In this scenario, each new wave of gene activity would be accompanied by a corresponding wave of UPS activity that creates space in which the newly activated genes can exert their effects. Given the role of the UPS in synaptic clearance (Tsai, 2014), epigenetic modifications that are relatively protected from protein degradation may tag the appropriate neurons for reactivation (Lesburgueres et al., 2011), resulting in a novel composition and abundance of proteins within the cell to reflect the updated memory. This new protein composition would allow for ongoing memory updating within the specific cell populations dedicated to the particular memory. The degree of degradation achieved by the UPS under normal circumstances is not clear, but existing data lead us to speculate that each instance of hippocampus-induced reactivation may result in some degree of resetting of prior synaptic information. As cortical connections are strengthened, there is less need for hippocampal input into the trace and the hippocampus disengages from this memory trace. The hippocampal synapses that underlie the disengagement need to be reset and prepared for different synaptic weights in the future, a clear role for the UPS. Overall, the UPS may be specifically involved in refining connections between the hippocampus and the cortex, allowing for circuit flexibility that is necessary to maintain memory updating.

We have thus far treated proteasomal degradation and epigenetics as separate mechanisms, but there is plenty of interplay between them. Histones themselves are degraded by the proteasome (Haas et al., 1990, Qian et al., 2013), and even histone variants such as H2A.Z may be regulated by proteasomal degradation (Baptista et al., 2013). Although we did not test a possible role for the proteasome in our work with H2A.Z, we did demonstrate that levels of H2A.Z in the hippocampus are reduced in response to fear conditioning, suggesting a potential role for the proteasome (Zovkic et al., 2014). Additionally, the proteasome is integral for various processes that regulate gene expression, including nucleosome eviction during transcriptional initiation (Chaves et al., 2010). Moreover, many histone modifying enzymes are degraded by the proteasome (Zou and Mallampalli, 2014), including histone de-methylases (Tan et al., 2011) and histone deacetylases (Kramer et al., 2003), which are known regulators of memory formation (Peixoto and Abel, 2013).

In addition, LTP is dependent on both gene transcription (Nguyen et al., 1994) and the proteasome (Dong et al., 2008) for induction, suggesting a temporal link between these processes. Recent evidence demonstrates that the proteasome may be involved in regulating gene activity during LTP, as proteasomal inhibition prevents transcription-promoting histone marks from occurring (Bach et al., 2015). Finally, some evidence suggests that DNA methylation is also regulated by the proteasome through regulated degradation of DNA methyltransferase enzymes (Esteve et al., 2009), thus positioning the proteasome as a regulator of crucial epigenetic processes. Although the direct links between these processes in memory formation remain to be studied, the importance of both processes for memory formation suggests that the interplay between the proteasome and epigenetics is crucial for regulating memory formation, particularly in relation to the protracted waves of gene activity and clearance.

In broad terms, systems consolidation rests on the assumption that the hippocampus has a time-limited role in memory. Although there is ample evidence to support a time-limited role for the hippocampus, it is important to note that such observations may at least partly be a product of the available laboratory methods (for an extended review of this issue, please see (Wiltgen and Tanaka, 2013). For example, a study using optogenetics, which provides much better temporal resolution than pharmacological approaches, showed that hippocampal inactivation can indeed produce deficits in remote memory, but only if the hippocampus is inactivated immediately before recall (Goshen et al., 2011). Consistent with results from lesion studies, these authors did not find deficits in remote memory when the hippocampus was inactivated with longer acting pharmacological tools (Goshen et al., 2011), indicating that the hippocampus may be involved in remote memory recall under normal circumstances, but that compensatory mechanisms are engaged when the hippocampus is inactivated for longer periods of time.

One potential explanation for mixed results is that even though remote memories can be efficiently recalled without the hippocampus, there may nevertheless be a loss of detail and precision compared to memories recalled with an intact hippocampus (Rosenbaum et al., 2001, Clark et al., 2005a, Clark et al., 2005b, Teixeira et al., 2006). In addition, hippocampal involvement in remote memory may vary with the type of memory. For example, inactivating the dorsal hippocampus with lidocaine impaired both recent and remote memory in the water maze (Teixeira et al., 2006), indicating that complex spatial tasks may rely on the hippocampus throughout the life of the memory. In contrast to studies with fear conditioning, water maze testing induced cFos expression in CA1 and CA3 subregions of the hippocampus at both the recent and remote time points, suggesting a prolonged involvement of these regions in spatial memory. However, this effect was also found in control mice, for which spatial cues did not predict platform location, indicating that the hippocampus is activated whenever spatial navigation is required (Teixeira et al., 2006). Thus, task demands may at least partly mediate hippocampal involvement at recent and remote time points.

Another potential confound is the lack of prefrontal involvement in memory at recent stages of memory formation. A number of studies suggest that memory formation immediately involves the PFC and that relative independence of recent memories of the cortex is a byproduct of relatively simple paradigms used to study systems consolidation (Preston and Eichenbaum, 2013). In more complex scenarios, new memories are synthesized with existing knowledge into contextually rich schemas (Tse et al., 2011, Wang et al., 2012, Preston and Eichenbaum, 2013), such that the PFC is immediately engaged in updating existing schemas. In the lab, many single-trial learning tasks, such as contextual fear conditioning, are well-suited for investigating the stimulus-locked molecular events associated with learning and memory, but the relative lack of prior experiences in which to integrate new learning may alter cortical involvement during the recent memory phase (Preston and Eichenbaum, 2013). Work from Richard Morris’s lab has developed behavioral tasks that assess the development of schemas in rodents and indeed, those studies have shown much more rapid reliance of memories on the cortex compared to memories without schemas (Tse et al., 2011, Wang et al., 2012). Utilizing such models in molecular studies of memory will be essential for developing a complete understanding of memory formation over time.

Section snippets

Future directions

Much remains to be discovered regarding the molecular changes that drive the communication between hippocampal and cortical regions in memory stabilization. However, as we learn more about the molecular basis of this process, it will become increasingly important to investigate how these molecular changes drive memory formation at all stages. Features of memory, such as the strength or the duration of the stimulus, tend to be associated with different magnitudes of molecular outcomes, with

References (123)

  • I. Goshen et al.

    Dynamics of retrieval strategies for remote memories

    Cell

    (2011)
  • J. Graff et al.

    Epigenetic priming of memory updating during reconsolidation to attenuate remote fear memories

    Cell

    (2014)
  • A. Haas et al.

    Ubiquitin-mediated degradation of histone H3 does not require the substrate-binding ubiquitin protein ligase, E3, or attachment of polyubiquitin chains

    J Biol Chem

    (1990)
  • M.A. Hahn et al.

    5-Hydroxymethylcytosine: a stable or transient DNA modification?

    Genomics

    (2014)
  • A.M. Hamilton et al.

    Activity-dependent growth of new dendritic spines is regulated by the proteasome

    Neuron

    (2012)
  • J. Huijgen et al.

    The hippocampus: A central node in a large-scale brain network for memory

    Rev Neurol (Paris)

    (2015)
  • R.E. Irwin et al.

    5-Hydroxymethylation marks a class of neuronal gene regulated by intragenic methylcytosine levels

    Genomics

    (2014)
  • T.J. Jarome et al.

    The ubiquitin-proteasome system as a critical regulator of synaptic plasticity and long-term memory formation

    Neurobiol Learn Mem

    (2013)
  • G.A. Kaas et al.

    TET1 controls CNS 5-methylcytosine hydroxylation, active DNA demethylation, gene transcription, and memory formation

    Neuron

    (2013)
  • E.R. Kandel et al.

    The molecular and systems biology of memory

    Cell

    (2014)
  • C. Katche et al.

    Maintenance of long-term memory storage is dependent on late posttraining Egr-1 expression

    Neurobiol Learn Mem

    (2012)
  • C. Katche et al.

    Molecular signatures and mechanisms of long-lasting memory consolidation and storage

    Neurobiol Learn Mem

    (2013)
  • J.M. Levenson et al.

    Regulation of histone acetylation during memory formation in the hippocampus

    J Biol Chem

    (2004)
  • S.A. Maddox et al.

    DNA methyltransferase activity is required for memory-related neural plasticity in the lateral amygdala

    Neurobiol Learn Mem

    (2014)
  • S.C. McQuown et al.

    HDAC3 and the molecular brake pad hypothesis

    Neurobiol Learn Mem

    (2011)
  • C.A. Miller et al.

    Covalent modification of DNA regulates memory formation

    Neuron

    (2007)
  • C.A. Miller et al.

    DNA methylation and histone acetylation work in concert to regulate memory formation and synaptic plasticity

    Neurobiol Learn Mem

    (2008)
  • C.L. Pickens et al.

    Long-lasting incubation of conditioned fear in rats

    Biol Psychiatry

    (2009)
  • A.R. Preston et al.

    Interplay of hippocampus and prefrontal cortex in memory

    Curr Biol

    (2013)
  • M.X. Qian et al.

    Acetylation-mediated proteasomal degradation of core histones during DNA repair and spermatogenesis

    Cell

    (2013)
  • R.S. Rosenbaum et al.

    New views on old memories: re-evaluating the role of the hippocampal complex

    Behav Brain Res

    (2001)
  • A. Rudenko et al.

    Tet1 is critical for neuronal activity-regulated gene expression and memory extinction

    Neuron

    (2013)
  • J.W. Rudy et al.

    Is it systems or cellular consolidation? Time will tell. An alternative interpretation of the Morris group’s recent science paper

    Neurobiol Learn Mem

    (2008)
  • S.D. Speese et al.

    The ubiquitin proteasome system acutely regulates presynaptic protein turnover and synaptic efficacy

    Curr Biol

    (2003)
  • T. Abel et al.

    Memory suppressor genes: inhibitory constraints on the storage of long-term memory

    Science

    (1998)
  • K.G. Akers et al.

    Hippocampal neurogenesis regulates forgetting during adulthood and infancy

    Science

    (2014)
  • J. Alder et al.

    Brain-derived neurotrophic factor-induced gene expression reveals novel actions of VGF in hippocampal synaptic plasticity

    J Neurosci

    (2003)
  • J. Artinian et al.

    Protein degradation, as with protein synthesis, is required during not only long-term spatial memory consolidation but also reconsolidation

    Eur J Neurosci

    (2008)
  • T. Baptista et al.

    Regulation of histone H2A.Z expression is mediated by sirtuin 1 in prostate cancer

    Oncotarget

    (2013)
  • F. Bermudez-Rattoni

    Is memory consolidation a multiple-circuit system?

    Proc Natl Acad Sci U S A

    (2010)
  • R. Bernabeu et al.

    Involvement of hippocampal cAMP/cAMP-dependent protein kinase signaling pathways in a late memory consolidation phase of aversively motivated learning in rats

    Proc Natl Acad Sci U S A

    (1997)
  • B. Bingol et al.

    Activity-dependent dynamics and sequestration of proteasomes in dendritic spines

    Nature

    (2006)
  • B. Bontempi et al.

    Time-dependent reorganization of brain circuitry underlying long-term memory storage

    Nature

    (1999)
  • J. Born et al.

    System consolidation of memory during sleep

    Psychol Res

    (2012)
  • I. Cervantes-Sandoval et al.

    System-like consolidation of olfactory memories in Drosophila

    J Neurosci

    (2013)
  • S. Chaves et al.

    Cks1, Cdk1, and the 19S proteasome collaborate to regulate gene induction-dependent nucleosome eviction in yeast

    Mol Cell Biol

    (2010)
  • Z.J. Chen

    Ubiquitin signalling in the NF-kappaB pathway

    Nat Cell Biol

    (2005)
  • R.E. Clark et al.

    Hippocampus and remote spatial memory in rats

    Hippocampus

    (2005)
  • R.E. Clark et al.

    Impaired remote spatial memory after hippocampal lesions despite extensive training beginning early in life

    Hippocampus

    (2005)
  • F. Crick

    Memory and molecular turnover

    Nature

    (1984)
  • Cited by (14)

    • Epigenetics and memory: an expanded role for chromatin dynamics

      2021, Current Opinion in Neurobiology
      Citation Excerpt :

      Importantly, neuronal epigenetic marks also have a high level of stimulus sensitivity that promotes adaptation to environmental stimuli, as shown by rapid and transient induction of epigenetic changes after learning [4•,5–8]. As such, epigenetic mechanisms play a vital role in regulating learning-induced changes in transcription, a necessary step in establishing stable memories [9]. Epigenetic modifications can occur directly on DNA, as is the case for DNA methylation, or on histone proteins that package DNA into nucleosomes, the building blocks of chromatin.

    • Hormone-epigenome interactions in behavioural regulation

      2020, Hormones and Behavior
      Citation Excerpt :

      Indeed, hormone actions have long been characterized as organizational or activational, whereby hormonal surges during sensitive periods shape (i.e. organize) the developing brain, and circulating hormones activate established systems in the mature brain to regulate behaviour (Phoenix et al., 1959). Similarly, epigenetic factors can exert stable actions, particularly in shaping development and maintaining cell fate, but they also have an important role in regulating rapid stimulus-induced transcriptional modifications (Cheedipudi et al., 2014; Walters and Zovkic, 2015). As such, epigenetic and hormonal mechanisms play complementary roles in transcriptional regulation and their interactions have powerful effects during development and in activity-induced behavioural regulation.

    • Learning and Age-Related Changes in Genome-wide H2A.Z Binding in the Mouse Hippocampus

      2018, Cell Reports
      Citation Excerpt :

      Chromatin modifications are crucial for learning and memory and serve as key regulators of transcription (Walters and Zovkic, 2015).

    • Epigenetic mechanisms of learning and memory

      2018, Molecular-Genetic and Statistical Techniques for Behavioral and Neural Research
    • Memory-Associated Dynamic Regulation of the "Stable" Core of the Chromatin Particle

      2015, Neuron
      Citation Excerpt :

      In contrast to known deposition candidates, regulation of H3.3 removal from chromatin is poorly understood, and Maze et al. provided the first evidence implicating proteasome-mediated H3.3 eviction and degradation in neurons. This novel role for the proteasome is particularly interesting in light of recent evidence for proteasomal regulation of learning and memory, as well as its regulation of H2A variants outside the brain (Walters and Zovkic, 2015). Chromatin is subject to complex regulatory mechanisms that impact neural plasticity through effects on gene expression.

    View all citing articles on Scopus
    View full text