Heritable Characteristics Affecting Learning

For far too long, the intricate cellular and molecular machinations underpinning learning and memory have captivated the focus of neuroscience. One might even say it was an obvious direction. However, only in the more recent, enlightened epoch has scientific curiosity — or perhaps, desperation — finally pivoted to the truly fascinating, if somewhat convoluted, epigenetic mechanisms. These, as it turns out, are the silent orchestrators behind the dynamic, ever-shifting landscape of gene transcription that is fundamentally responsible for both the initial spark of memory formation and its stubborn, sometimes inconvenient, persistence.

At its core, epigenetic gene regulation operates with a kind of elegant, if somewhat brutal, efficiency. It often involves the physical 'marking' – a rather poetic term for chemical modification – of the very DNA itself, or the associated proteins that cradle it. These modifications are not mere fleeting whims; they are designed to induce, or conversely, to permit, profound and long-lasting alterations in gene activity. Among the most prominent of these epigenetic mechanisms, which have been painstakingly demonstrated to play a pivotal role in the complex theatre of learning and the fragile art of memory, are DNA methylation and the various sophisticated modifications of histones, including methylation, acetylation, and deacetylation. It's almost as if the cell has its own internal annotation system, deciding which chapters of its genetic library are open for reading and which are best left undisturbed, for now. [1]

DNA Methylation

DNA methylation is, to put it mildly, a rather precise affair. It entails the deliberate attachment of a methyl group to a 5' cytosine residue. This chemical tag doesn't just appear anywhere; it predominantly targets cytosines that are conveniently situated as part of a cytosine-guanine dinucleotide, often referred to as CpG sites. The consequences of this methylation are far from trivial, as it possesses the power to either ignite or extinguish gene transcription. This delicate balance is meticulously overseen by the enzymatic prowess of DNA methyltransferases (DNMTs). Specifically, DNMT3A and DNMT3B are the architects of de novo methylation, establishing fresh patterns at previously unadorned CpG sites. Meanwhile, DNMT1 serves as the vigilant guardian, ensuring that established methylation patterns are faithfully maintained across cellular divisions, a testament to the system's commitment to continuity. [2] The entire process, of course, wouldn't be possible without S-adenosyl methionine, which graciously volunteers its services as the essential methyl donor. [3]

One of the more compelling current hypotheses attempting to unravel the mysterious contribution of DNA methylation to the very storage of our memories suggests a temporal ballet of dynamic DNA methylation changes. These shifts, it is posited, unfold with a precise timing, activating the transcription of genes that encode for proteins specifically tasked with the stabilization and endurance of memory. Another, perhaps more unsettling, hypothesis posits that alterations in DNA methylation, even those occurring in the earliest stages of life, possess a remarkable staying power, persisting relentlessly through adulthood. Such early imprints would, by extension, profoundly influence how genes are subsequently capable of being activated in response to the myriad environmental cues encountered throughout an organism's existence.

The foundational revelation concerning the role of epigenetics in the intricate dance of learning and memory emerged from the seminal work of Szyf and Meaney (PMID 15220929). Their landmark study elegantly demonstrated that the seemingly simple act of licking and grooming by mother rats – a form of maternal care – actively prevented the methylation of the glucocorticoid receptor gene. The long-term implications were striking: these pups, when they matured into adults, exhibited a markedly superior ability to cope with stressors compared to their counterparts. The latter, having been deprived of such maternal attention as pups, instead accumulated methylation at the very same glucocorticoid receptor gene, a subtle yet profound difference with lasting behavioral consequences. This isn't just a quaint observation; it’s a stark reminder that even seemingly small interactions can leave indelible biological marks.

DNMTs and memory

In a study that illuminated the direct involvement of these molecular architects, Miller and Sweatt meticulously observed that rats subjected to a contextual fear conditioning paradigm displayed significantly elevated levels of mRNA for both DNMT3a and DNMT3b within their hippocampus. [4] For the uninitiated, fear conditioning is a rather straightforward, if somewhat unsettling, associative memory task. An animal is exposed to a neutral context, let's say a specific room, which is then unpleasantly paired with an aversive stimulus, such as a mild foot shock. Animals that have successfully formed this association will predictably exhibit higher levels of freezing behavior when subsequently exposed to that same context, even in the blissful absence of the aversive stimulation. A clear, if rather dramatic, demonstration of memory.

However, the plot thickened when these researchers intervened. When rats were treated with DNMT inhibitors – specifically zebularine or 5-aza-2′-deoxycytidine – immediately following their fear-conditioning session, they demonstrated a discernible reduction in their learning capacity, manifesting as decreased freezing behavior. Interestingly, when these treated rats were re-trained a full 24 hours later, they performed just as competently as their untreated counterparts. Furthermore, a crucial temporal detail emerged: if these DNMT inhibitors were administered a mere 6 hours after the initial training, and the rats were subsequently tested 24 hours later, they displayed perfectly normal fear memory. This pointed to a rather specific conclusion: DNMTs are not involved in the initial acquisition of memory, but rather, are critical players specifically in the intricate process of memory consolidation. [4] These findings, if nothing else, starkly underscore the profound importance of dynamic shifts in methylation status for the robust formation of memory. It’s not enough to learn; you have to cement it.

Further elucidating the roles of these enzymes, Feng et al. engineered double conditional knock out (DKO) mice, specifically targeting the genes for DNMT3a and DNMT1. These genetically modified mice exhibited a significantly diminished capacity for long-term potentiation (LTP) – the strengthening of synaptic connections crucial for learning – and, conversely, displayed a much more facile induction of long-term depression (LTD) in the hippocampus. When these DKO mice were put through the paces of the Morris water navigation task, a standard assay used to probe hippocampus-dependent spatial memory, they took considerably longer to locate the submerged platform compared to control mice. Intriguingly, single knock-out mice (SKO) for either DNMT3a or DNMT1 alone performed entirely normally, suggesting a redundancy in function. [5] The DKO mice also proved incapable of effectively consolidating memory after fear-conditioning, further solidifying the critical role of these enzymes. Given that SKO mice did not manifest the same learning and memory deficits as their DKO counterparts, the inescapable conclusion was that DNMT3a and DNMT1, rather than operating independently, play complementary and somewhat redundant roles in the intricate regulation of learning and memory. A rather inefficient design, one might argue, but biology rarely prioritizes elegance over robustness.

Further complicating the picture, when DNMTs are pharmacologically inhibited within the prefrontal cortex, the recall of existing memories is notably impaired. However, the formation of new memories remains curiously unaffected. [6] This intriguing observation suggests that DNA methylation, in its regulatory capacity for memory formation and maintenance, may operate with a surprising degree of circuit-specificity. It's not a global switch; it's more like a series of individually controlled dimmers, each affecting a particular neural pathway.

DNA methylation targets

Among the various genes that fall under the influence of DNA methylation during memory processes, the memory suppressor gene, protein phosphatase 1 (PP1), has been particularly scrutinized. Following contextual-fear conditioning, this gene was observed to undergo increased CpG island methylation. This increase in methylation directly correlated with a corresponding decrease in the levels of PP1 mRNA within the hippocampus of the trained rats. When the DNMTs responsible for this methylation were subsequently inhibited, the increased methylation at the PP1 gene was no longer observed, confirming the direct link. [4] These data, in a rather elegant fashion, suggest that during the critical phase of memory consolidation in associative learning tasks, CpG methylation serves a crucial purpose: to actively inhibit the expression of PP1, a gene whose very function is to negatively regulate memory formation. It's a subtle but powerful act of molecular suppression, ensuring that what needs to be remembered isn't immediately erased.

Demethylation and Memory

If DNA methylation is the subtle, often silencing, hand that inhibits genes implicated in memory suppression, then DNA demethylation is its energetic counterpart, vital for activating genes whose expression is unequivocally and positively correlated with robust memory formation. Sweatt and Miller, continuing their foundational work, also demonstrated that the gene reelin, a key player in the induction of long-term potentiation, exhibited a reduced methylation profile and a concomitant increase in reelin mRNA in fear-conditioned rats compared to control animals. [7] Similarly, Brain-derived neurotrophic factor (BDNF), another critically important gene renowned for its role in neural plasticity, has consistently shown reduced methylation and increased transcription in animals that have undergone various learning experiences. [7]

While these pioneering studies predominantly focused on the hippocampus, an area universally acknowledged as central to memory, more recent evidence has expanded the geographical scope. It has been shown that increased demethylation of both reelin and BDNF also occurs in the medial prefrontal cortex (mPFC), a region of the brain intimately involved in higher-order cognitive functions and the nuanced processing of emotion. [8] This expansion suggests a broader, more distributed epigenetic regulation across brain regions crucial for complex memory.

For a considerable time, the precise mechanisms underlying this experience-dependent demethylation response remained shrouded in a certain degree of mystery. Early evidence hinted that DNMTs themselves might surprisingly be involved in the demethylation process, adding a layer of paradoxical complexity. [7] There was also the suggestion that members of the DNA damage repair GADD45 family could contribute to this intricate demethylation cascade. [2] [3] However, our understanding has since sharpened. More recently, the pathways so elegantly illustrated in the figure below, provocatively titled "Demethylation of 5-Methylcytosine (5mC) in neuron DNA," particularly the TET-dependent pathway, have been definitively confirmed as primary routes for DNA demethylation. [9] Furthermore, the role of GADD45 has indeed been substantiated; it has been shown that GADD45 physically interacts with thymine-DNA glycosylase (TDG), and this interaction appears to promote the activity of TDG in its crucial role(s) during the conversion of 5mC back to cytosine. [9] It's a complex, multi-step process, proving that even removing a tag requires a surprising amount of effort.

Methyl-binding domain proteins (MBDs)

Beyond the direct modifications of DNA, proteins that specifically recognize and bind to methylated DNA also play a critical role. Mice engineered with genetic disruptions for CpG binding protein 2 (MeCP2) have consistently demonstrated significant impairments in hippocampus#Role in memory-dependent memory tasks. Furthermore, these mice exhibit compromised hippocampal LTP, underscoring the importance of MeCP2 in mediating the effects of DNA methylation on synaptic plasticity and memory. [2] It seems that simply having the methylation isn't enough; you need the right readers to interpret the message.

Methylation and Learning and Memory Disorders

It's hardly a surprise that disruptions in these finely tuned epigenetic mechanisms would correlate with neurological disorders. Changes in the expression of genes associated with post-traumatic stress disorder (PTSD) – a condition notoriously characterized by an impaired extinction of traumatic memories, a failure to unlearn fear – may well be mediated by aberrant DNA methylation patterns. [10] The persistent shadow of trauma, it seems, can be etched into the very epigenome.

In individuals grappling with schizophrenia, a distinct epigenetic signature has been observed: the gene reelin is consistently found to be down-regulated. This suppression is attributed to increased DNA methylation at its promoter regions, specifically within GABAergic interneurons. Concurrently, DNMT1 has been shown to be upregulated in these same cells, suggesting a coordinated epigenetic dysregulation contributing to the complex pathology of the disorder. [10] It’s a stark reminder that the delicate balance of epigenetic marks is essential for normal brain function.

Histone Methylation

The methylation of histones is another layer of epigenetic complexity, a nuanced system where the outcome — whether gene transcription is boosted or suppressed — hinges on several precise variables. It depends entirely on which histone is modified, the specific amino acid residue that receives the modification, and even the exact number of methyl groups appended. [11] It’s not a blunt instrument; it’s a surgical tool.

In the case of lysine methylation, the system boasts three distinct states of modification: monomethylated, dimethylated, or trimethylated lysines. These subtle quantitative differences have profound qualitative effects. For instance, the di- or trimethylation of histone H3 at lysine 9 (H3K9) is consistently associated with regions of the genome that are transcriptionally silent, effectively locking down gene expression. Conversely, the di- or trimethylation of histone H3 at lysine 4 (H3K4) signals the opposite: it is strongly associated with genes that are actively being transcribed, serving as a beacon for gene expression. [12]

Histone 3 lysine 4 trimethylation and memory formation

The hippocampus, as we've established, is a central hub in the intricate process of memory formation. H3K4 trimethylation, as noted, is a hallmark of active transcription. In compelling contextual fear conditioning experiments conducted on rats, researchers observed a significant surge in H3K4 trimethylation levels within the hippocampus following the fear conditioning protocol. [13] In these specific experiments, Gupta et al. were able to draw a direct and critical link between these observed changes in histone methylation and the concomitant active gene expression that occurs during the vital consolidation of associative memories. [13]

This same study also revealed a crucial dynamic: these histone methylations were not permanent fixtures; they were, in fact, reversible. The levels of trimethylation of H3K4 returned to their baseline within a mere 24 hours, strongly indicating that active demethylation processes were at play following memory consolidation. To further dissect the precise role of methyltransferases in the formation of long-term memory, the researchers subjected rats deficient in Mll – a methyltransferase specifically targeting H3K4 – to the same fear conditioning tests. Rats carrying a heterozygous mutant Mll+/- gene demonstrated a significant and undeniable reduction in their capacity to form long-term memories when compared to their normal counterparts with an intact Mll gene. The conclusion was clear: H3K4 methyltransferases, such as Mll, are not merely accessory players but hold an essential, indispensable role in the formation of long-term memory within the hippocampus. [13]

It is not merely a genome-wide increase in histone methylation that drives memory formation, but rather precise, localized changes in the methylation state of histones at the promoters of specific genes. [13] Genes like Zif268 and BDNF are universally recognized as absolutely critical for the process of memory consolidation. [14] Following contextual fear conditioning, a marked increase in H3K4 trimethylation is observed specifically around the promoters of both the Zif268 and BDNF genes, precisely when these genes become transcriptionally active. This elegant synchronicity powerfully demonstrates that, at the very moment of memory consolidation, the transcription of these crucial memory formation genes, such as Zif268 and BDNF, is meticulously regulated by histone methylation. [13]

Histone 3 lysine 9 dimethylation and memory formation

In stark contrast to H3K4 trimethylation, histone H3 lysine 9 dimethylation is consistently associated with transcriptional silencing. [12] The G9a/G9a-like protein (GLP) complex is the specific methyltransferase responsible for orchestrating this particular modification. [15] One compelling study delved into the role of G9a/GLP-mediated transcriptional silencing within the hippocampus and entorhinal cortex (EC) during the critical phase of memory consolidation. The findings were quite revealing: inhibiting the G9a/GLP complex specifically in the EC, but notably not in the hippocampus, resulted in a significant enhancement of long-term memory formation. [16]

Furthermore, the inhibition of G9a/GLP in the entorhinal cortex also induced discernible alterations in histone H3 lysine 9 dimethylation within the Cornu Ammonis area 1 of the hippocampus, strongly implying the importance of this complex in mediating the intricate functional connectivity between these two crucial brain regions. Therefore, it is clear that the G9a/GLP complex is not a passive observer but an active and important player in histone methylation, and consequently, in long-term memory formation across both the hippocampus and the EC. [16] It's a delicate interplay, where silencing in one area can surprisingly enhance function in another.

Histone methylation and other epigenetic modifications

The intricate tapestry of histone methylation marks is rarely isolated; it often correlates with other significant epigenetic modifications, such as histone deacetylation and DNA methylation, all converging in the context of learning and memory. For instance, a reduction in histone deacetylation is consistently correlated with an increase in H3K9 dimethylation, that very modification we've just discussed as being associated with transcriptional silencing. [13] This implies a potential therapeutic avenue: histone deacetylase inhibitors could theoretically be employed to increase histone acetylation, thereby suppressing H3K9 dimethylation and, in turn, increasing gene transcription. A rather elegant workaround, if it proves consistently effective.

In the case of DNA methylation, a fascinating observation emerged: increases in H3K4 trimethylation were found to correlate with altered DNA methylation of CpG sites at the promoter of Zif268, a gene universally acknowledged for its involvement in memory formation, particularly after fear conditioning. Gupta et al. demonstrated that DNA methylation at the Zif268 promoter actually increased after fear conditioning, and this increase remarkably correlated with an increase in Zif268 gene expression. [13] This finding was, to say the least, surprising, as the conventional wisdom had long held that DNA methylation invariably resulted in transcriptional silencing. [13] It seems biology, once again, delights in defying simplistic expectations.

Histone Acetylation

Acetylation is a fundamental chemical modification that, in its most basic form, involves the replacement of a hydrogen atom with an acetyl group. In the complex biological arena, acetylation is most famously, and perhaps most critically, associated with the modification of proteins, particularly the histones around which our genetic material is so tightly wound. This acetylation reaction is, more often than not, expertly catalyzed by a class of enzymes endowed with histone acetyltransferase (HAT) activity.

Histone acetyltransferases (HATs)

HATs are the enzymatic maestros responsible for orchestrating the acetylation of specific amino acid residues. Their modus operandi involves a clever biochemical maneuver: they acetylate the lysine side group of target amino acids by deftly transferring an acetyl group from an acetyl CoA molecule, thereby creating acetyl lysine. While HAT enzymes are most prominently associated with histone proteins, where they diligently work to regulate the intricate interaction between histones and the DNA they embrace, their catalytic reach extends far beyond. They are not merely restricted to the acetylation of histones; they are capable of acetylating a multitude of other proteins intimately implicated in the nuanced manipulation of gene expression, including crucial transcription factors and various receptor proteins. Their influence, it seems, is far-reaching.

Chromatin remodeling

Acetylation stands as one of the principal mechanisms intricately implicated in the dynamic process of chromatin remodeling. This remodeling fundamentally influences the regulation of gene expression by artfully altering the relationship between nucleosomes and the DNA wrapped around them. The acetylation of histones, in a rather elegant biochemical move, serves to neutralize their positive charge. This reduction in positive charge, in turn, diminishes the electrostatic attraction between the formerly positively charged histone proteins and the negatively charged phosphate groups that constitute the DNA backbone. This crucial alteration in charges induces a relaxation of the DNA from the nucleosome, making that particular segment of DNA more accessible. These relaxed, acetylated regions are consistently observed to exhibit significantly higher levels of gene expression compared to their non-acetylated counterparts. It’s like loosening a tightly wound spring to allow its energy to be released.

Acetylation as an epigenetic marker

The dynamic patterns of histone acetylation have proven to be an invaluable source of epigenetic information. Their utility stems from their remarkable ability to faithfully reflect changes in transcription rates and, crucially, to underpin the stable maintenance of gene expression patterns. This intricate "acetylation code" can then be meticulously deciphered, yielding a wealth of information for the study of the inheritance patterns of epigenetic changes, particularly those relevant to the complex phenomena of learning, memory, and various disease states. It's a biological shorthand, a subtle signature of cellular history.

Acetylation as a mechanism for learning and memory

The profound role of epigenetic mechanisms and the subsequent chromatin remodeling has been robustly implicated in both synaptic plasticity – the very malleability of our neural connections – and the precise regulation of neuronal gene expression. Studies employing inhibitors of the histone deacetylase complex, such as SAHA, toluene, garcinol, trichostatin A, and sodium butyrate, have unequivocally demonstrated that acetylation is not merely a bystander but an essential player in the synaptic plasticity of the brain. By inhibiting these deacetylase complexes, researchers observed a global increase in total acetylation rates within the brain, which, in turn, led to elevated rates of transcription and a remarkably enhanced memory consolidation. [17] [18]

Through the strategic use of various learning assays, including the classic Morris water maze test and fear conditioning paradigms, in conjunction with drugs designed to manipulate acetylation, it has been convincingly shown that acetylation patterns within the hippocampus are absolutely integral to the formation of memory associations and the manifestation of learning behavior. [19] Further studies employing diverse HDAC inhibitors during periods of neural development have consistently reported increased learning and memory capabilities, a direct consequence of an elevated acetylation state. Conversely, studies conducted with HAT inhibitors – which reduce acetylation – yielded precisely the opposite effect: a marked impairment of memory consolidation and an overall decrease in learning capacity. [20] The evidence, it seems, is rather compelling.

ERK/MAPK cascade

Further upstream in this intricate signaling network, studies have unequivocally demonstrated that the ERK/MAPK cascade plays a pivotal role in the precise regulation of lysine acetylation within the insular cortex of the brain. This particular region, it's worth noting, is intimately implicated in the formation of taste memories – a rather specific and visceral form of learning. The activation of the ERK/MAPK cascade was conspicuously observed in mice immediately following the introduction of a novel taste, and this cascade was subsequently proven to be an absolute necessity for the memory of that taste to be successfully formed.

The proposed mechanism for how this cascade operates is that MAPK effectively regulates histone acetylation and the subsequent chromatin remodeling through a series of downstream effectors. A prominent example of such an effector is the CREB binding protein (CBP), which, conveniently enough, possesses intrinsic HAT activity. [21] [22] [23] By meticulously observing the rates of acetylation within the insular cortex, researchers were able to precisely discern which specific patterns of acetylation were attributable to the activity of deacetylases or acetylases, and which were the direct result of lysine acetyltransferase activity. [22] It’s a delicate balance of molecular forces, all conspiring to help you remember that particularly awful flavor you once encountered.

Long term potentiation

Long term potentiation (LTP) is, in essence, the sustained enhancement of signal strength between neurons. It is the very bedrock of synaptic plasticity and, as such, plays an absolutely pivotal role in the formation of memory. LTP is fundamentally dependent on the activity of NMDA receptors within the brain, and it has been consistently demonstrated that NMDA activity directly influences acetylation states.

When NMDA receptors are activated, they trigger a crucial influx of calcium into the cell. This calcium surge, in turn, activates various downstream signaling pathways that ultimately converge on and activate the ERK pathway. The ERK pathway then modulates key transcription factors such as CREB. CREB, in a rather collaborative fashion, then recruits a HAT to assist in the creation and stabilization of long-term memory formation, often through a self-perpetuating cycle of acetylated histones. Studies specifically investigating the acetylation of histone H3 in the CA1 region of the hippocampus have consistently shown that the activation of NMDA receptors leads to an increase in H3 acetylation. Conversely, the inhibition of the ERK pathway in the CA1 region results in a discernible decrease in H3 acetylation. [23]

In summary, the intricate ballet of molecular events appears to follow a clear sequence:

NMDA-R activation directly increases the phosphorylation of ERK and, consequently, the acetylation of Histone H3.
Robust memory formation demonstrably requires proper NMDA-R function.
Memory conditioning itself increases both the phosphorylation of ERK and the acetylation of Histone H3.
ERK activity is meticulously regulated by phosphorylation.
The acetylation of Histone H3 is, in turn, regulated by ERK.
Interestingly, Histone H4 does not appear to be regulated by ERK, suggesting a specificity in the pathway.
HDAC inhibitors consistently enhance LTP, and this enhancement is demonstrably dependent on the rate of transcription.
Crucially, HDAC inhibitors do not directly affect NMDA-R function, indicating their action is downstream.

It's a complex, but ultimately logical, chain of command.

Histone Deacetylation

HDACs' role in CREBP-dependent transcriptional activation

Figure 1: HDAC inhibition enhances memory and synaptic plasticity through CREB:CBP. Figure adapted from Vecsey et al. (2007) [24]

Histone deacetylases (HDACs) are the molecular counterparts to HATs, fulfilling the crucial role of removing acetyl groups (-COCH3) from histones. This enzymatic action leads to profound alterations in chromatin structures, effectively decreasing the accessibility of transcriptional factors to the underlying DNA, and thereby reducing the transcription of genes. It’s a mechanism for tightening the genetic reins. HDACs have been convincingly shown to play a significant role in both learning and memory, primarily through their precise regulatory influence on the CREB-CBP pathway.

Studies have consistently concluded that HDAC inhibitors, such as trichostatin A (TSA), lead to an increase in histone acetylation and, consequently, a measurable improvement in both synaptic plasticity and long-term memory (Fig 1A). CREB, a cAMP response element-binding protein and a potent transcriptional activator, forms a complex with CBP, known as the CREBP complex. This complex is instrumental in activating genes that are critically involved in synaptic formation and the enduring quality of long-term memory (Fig 1B).

When TSA treatments were applied specifically to the hippocampal CA1 region of mice, researchers observed elevated acetylation levels and a marked enhancement of long-term potentiation (LTP), a mechanism, as previously discussed, that is fundamentally involved in learning and memory (Fig 1B). However, a critical distinction emerged: TSA treatments administered to CBP mutants, specifically those lacking KIX domains, failed to elicit any effect on LTP in mice (Fig 1D). The KIX domain, it should be noted, is absolutely essential for the physical interaction between CREB and CBP; thus, its absence effectively disrupts the formation of the crucial CREBP complex. Similarly, genetic knockouts of CREB itself produced results strikingly similar to those observed in the mutant CBP mice (Fig 1C). The implication is clear: both HDAC inhibition and the functional association of CREBP are indispensable for the robust development of memory.

Further analysis of TSA treatments revealed an increased expression of Nr4a1 and Nra2 genes, while other CREB-regulated genes remained curiously unaffected. This suggests that HDAC inhibitors do not simply trigger a global transcriptional surge but rather improve memory through the activation of specific genes that are precisely regulated by the CREBP complex. [24] It's a targeted strike, not a broad assault.

HDAC2

While the individual roles of specific HDACs in the nuanced processes of learning and memory are still being meticulously unraveled, HDAC2 has emerged as a particularly interesting, if somewhat problematic, player. It has been shown to act as a negative regulator of both memory formation and synaptic plasticity. [19]

When researchers induced overexpression (OE) of both HDAC1 and HDAC2 in mice, the predictable outcome was a decrease in the levels of acetylated lysines. However, the behavioral consequences were strikingly different. After subjecting these mice to context and tone-dependent fear conditioning experiments, HDAC1 OE mice showed no discernible change in their freezing behavior. In stark contrast, HDAC2 OE mice exhibited a significant decrease in freezing behavior, strongly suggesting an impairment in memory formation. Conversely, mice with HDAC2 knockouts (KO) displayed increased freezing levels compared to wild-type (WT) mice, while HDAC1 KOs exhibited freezing behaviors similar to WTs. In summary, Guan et al. [19] meticulously demonstrated that:

It is HDAC2, and not HDAC1, that meticulously regulates synaptogenesis and synaptic plasticity. The overexpression of HDAC2 leads to a decrease in spine density in CA1 pyramidal neurons and dentate gyrus granule cells, whereas HDAC2 KOs show a distinct increase in spine density.
Long-term potentiation in CA1 neurons was conspicuously absent in HDAC2 OE mice but was remarkably easy to induce in HDAC2 KO mice. LTP, however, remained unaltered between HDAC1 KO and OE mice, further highlighting the specificity of HDAC2.
HDAC2 actively suppresses neuronal gene expression. Intriguingly, HDAC2 was found to interact more robustly than HDAC1 with the promoters of specific memory-forming genes such as Bdnf, Egr1, Fos, and GLUR1, suggesting a direct mechanism of repression.
CoREST, a known co-repressor, preferentially associates with HDAC2, not HDAC1, indicating a specific complex formation.
SAHA, a broad-spectrum HDAC inhibitor, notably increased the freezing behavior of HDAC2 OE mice in both contextual fear and tone-dependent experiments, but had no effect on HDAC2 KO mice. This strongly suggests that HDAC2 is a major, if not the primary, target of SAHA's memory-enhancing effects.

HDAC3

HDAC3 also plays a rather inconvenient role as a negative regulator of long-term potentiation formation, another gatekeeper in the memory process. McQuown et al. [25] have painstakingly detailed that:

Genetic knockouts of HDAC3 specifically within the dorsal hippocampus resulted in a measurable enhancement of memory during object location tests (OLM), indicating its suppressive role.
RGFP136, a selective HDAC3 inhibitor, demonstrably enhances LTP for both object recognition and location tasks, further solidifying HDAC3's negative influence.
This enhancement of LTP by RGFP136 operates through a CBP-dependent mechanism, linking it back to the critical CREBP pathway.
HDAC3 deletions were associated with increased expression of Nr4a2 and c-Fos genes, suggesting that HDAC3 normally represses these genes.
HDAC3 interacts with NCoR which? and HDAC4 to execute its role in memory formation, highlighting a complex of interacting proteins.

HDACs' role in CNS disorders

It should come as no great surprise that such fundamental regulators like HDACs and HATs play a crucial, if often problematic, role in various central nervous system (CNS) disorders, including the devastating Rett syndrome. [26]

Consider Rubinstein-Tabyi syndrome, a condition characterized by intellectual disability, often linked to possible mutations in CREB-binding protein and p300. Encouragingly, strategies aimed at enhancing the expression of CREB-dependent genes or, alternatively, inhibiting HDAC activity, have shown promise in partially restoring LTP loss and ameliorating deficits in late LTP. This suggests that HDAC inhibitors, such as TSA, might offer a plausible therapeutic avenue for Rubinstein-Tabyi syndrome, a glimmer of hope in a challenging field.

Beyond these, a growing list of other memory-deficit disorders are also being investigated for their potential responsiveness to HDAC inhibitors as a therapeutic strategy:

The sheer breadth of these potential applications underscores the profound, and often overlooked, importance of these epigenetic regulators.

Role of DNA topoisomerase II beta in learning and memory

During the crucible of a new learning experience, a specific, highly orchestrated set of genes is rapidly expressed within the brain. This induced gene expression is not merely incidental; it is considered absolutely essential for the efficient processing of the information being learned. Such genes are, rather aptly, referred to as immediate early genes (IEGs). It turns out that DNA topoisomerase II beta (TOP2B) activity is not just helpful, but essential for the expression of these IEGs during a particular type of learning experience in mice, termed associative fear memory. [27]

Such a learning experience appears to rapidly trigger TOP2B to induce deliberate double-strand breaks in the promoter DNA of IEG genes that are crucial for neuroplasticity. This isn't damage; it's a controlled demolition. The subsequent repair of these induced breaks is intriguingly associated with DNA demethylation of the IEG gene promoters, a process that then allows for the immediate and robust expression of these IEG genes. [27] It's a rather dramatic way to get a gene expressed, but apparently, it works.

Regulatory sequence in a promoter at a transcription start site with a paused RNA polymerase and a TOP2B-induced double-strand break

Brain regions involved in memory formation including medial prefrontal cortex (mPFC)

The double-strand breaks that are so precisely induced during a learning experience are not immediately patched up; they linger, for a short but crucial period. Approximately 600 regulatory sequences located in promoters and roughly 800 regulatory sequences found in enhancers appear to be critically dependent on these double-strand breaks, specifically those initiated by topoisomerase 2-beta (TOP2B), for their activation. [28] [29] The induction of these particular double-strand breaks is remarkably specific, tailored to their inducing signal. For instance, when neurons are activated in vitro, a mere 22 TOP2B-induced double-strand breaks occur across their entire genomes, highlighting the precision of this mechanism. [30]

These TOP2B-induced double-strand breaks are not solitary events; they are accompanied by a retinue of at least four enzymes belonging to the non-homologous end joining (NHEJ) DNA repair pathway. This includes DNA-PKcs, KU70, KU80, and DNA LIGASE IV (as depicted in the Figure). These dedicated enzymes efficiently repair the double-strand breaks within a relatively short window, typically ranging from about 15 minutes to two hours. [30] [31] Crucially, the double-strand breaks within the promoter are thus intimately associated with TOP2B and at least these four repair enzymes. These proteins are found simultaneously present on a single promoter nucleosome – that compact unit of DNA wrapped around histones, comprising about 147 nucleotides – strategically located near the transcription start site of their target gene. [31]

The double-strand break introduced by TOP2B appears to serve a rather ingenious purpose: it physically frees the part of the promoter at an RNA polymerase-bound transcription start site. This liberation allows this promoter segment to physically move and interact directly with its associated enhancer (see regulatory sequence). This crucial interaction then permits the enhancer, along with its bound transcription factors and mediator proteins, to directly engage with the RNA polymerase, which is typically paused at the transcription start site, thereby initiating transcription. [30] [32] It's a remarkable example of molecular choreography, where a deliberate break facilitates a critical connection.

Contextual fear conditioning in mice, a powerful learning paradigm, results in the mouse developing a long-term memory and fear of the specific location where the aversive event occurred. This type of conditioning triggers hundreds of DSBs within the neurons of the mouse brain's medial prefrontal cortex (mPFC) and hippocampus (as illustrated in the Figure: Brain regions involved in memory formation). These DSBs, rather than being random acts of destruction, predominately activate genes intimately involved in synaptic processes, which are, of course, absolutely vital for learning and memory. [33]

Roles of ROS and OGG1 in memory and learning

Initiation of DNA demethylation at a CpG site. In adult somatic cells DNA methylation typically occurs in the context of CpG dinucleotides (CpG sites), forming 5-methylcytosine-pG, or 5mCpG. Reactive oxygen species (ROS) may attack guanine at the dinucleotide site, forming 8-hydroxy-2'-deoxyguanosine (8-OHdG), and resulting in a 5mCp-8-OHdG dinucleotide site. The base excision repair enzyme OGG1 targets 8-OHdG and binds to the lesion without immediate excision. OGG1, present at a 5mCp-8-OHdG site recruits TET1 and TET1 oxidizes the 5mC adjacent to the 8-OHdG. This initiates demethylation of 5mC. [34]

Demethylation of 5-Methylcytosine (5mC) in neuron DNA. As reviewed in 2018, [9] in brain neurons, 5mC is oxidized by the ten-eleven translocation (TET) family of dioxygenases (TET1, TET2, TET3) to generate 5-hydroxymethylcytosine (5hmC). In successive steps TET enzymes further hydroxylate 5hmC to generate 5-formylcytosine (5fC) and 5-carboxylcytosine (5caC). Thymine-DNA glycosylase (TDG) recognizes the intermediate bases 5fC and 5caC and excises the glycosidic bond resulting in an apyrimidinic site (AP site). In an alternative oxidative deamination pathway, 5hmC can be oxidatively deaminated by activity-induced cytidine deaminase/apolipoprotein B mRNA editing complex (AID/APOBEC) deaminases to form 5-hydroxymethyluracil (5hmU) or 5mC can be converted to thymine (Thy). 5hmU can be cleaved by TDG, single-strand-selective monofunctional uracil-DNA glycosylase 1 (SMUG1), Nei-Like DNA Glycosylase 1 (NEIL1), or methyl-CpG binding protein 4 (MBD4). AP sites and T:G mismatches are then repaired by base excision repair (BER) enzymes to yield cytosine (Cyt).

As extensively reviewed by Massaad and Klann in 2011 [35] and further elaborated by Beckhauser et al. in 2016, [36] reactive oxygen species (ROS) – often perceived as agents of cellular damage – are, perhaps counterintuitively, absolutely required for the normal, healthy functions of learning and memory. It seems a little chaos is necessary for order.

One of the most frequently occurring DNA oxidation products resulting from ROS activity is 8-hydroxy-2'-deoxyguanosine (8-OHdG). While the removal of these oxidized bases from DNA typically occurs with remarkable speed, often within minutes – with 8-OHdG boasting a half-life of a mere 11 minutes [37] – the steady-state levels of endogenous DNA damages represent a dynamic equilibrium between their constant formation and their subsequent repair. 8-OHdGs are, in fact, among the most prevalent DNA damages found at steady-state, with an estimated average of 2,400 8-OHdG damaged nucleotides residing in the typical mammalian cell. [38] The steady-state 8-OHdG level observed in the brain is remarkably similar to that found in other tissues, suggesting a ubiquitous presence. [39]

The very presence of 8-OHdG in neurons appears to have a direct and crucial role in the intricate processes of memory and learning. The DNA glycosylase oxoguanine glycosylase (OGG1) is the primary enzymatic workhorse responsible for the precise excision of 8-OHdG during base excision repair. However, OGG1, while meticulously targeting and associating with 8-OHdG, also possesses a demonstrable role in adaptive behavior. This implies a physiologically relevant partnership between 8-OHdG and OGG1 in the complex domain of cognition within the adult brain. [40] [41] Specifically, heterozygous OGG1+/- mice, which possess approximately half the normal protein level of OGG1, consistently exhibit poorer learning performance in the Barnes maze compared to their wild-type counterparts. [42] It seems you need enough of the cleanup crew to keep the cognitive machinery running smoothly.

In adult somatic cells, such as neurons, DNA methylation, as we've established, typically occurs within the context of CpG dinucleotides (CpG sites), leading to the formation of 5-methylcytosine (5mC). [34] Thus, a CpG site can be methylated to form 5mCpG. The presence of 5mC at CpG sites within gene promoters is widely regarded as a fundamental epigenetic mark that acts to suppress transcription. [43] However, if the guanine at a 5mCpG site is subjected to attack by ROS, leading to the formation of 8-OHdG, OGG1 will then bind to this 8-OHdG lesion. Crucially, it does so without immediate excision of the 8-OHdG. While OGG1 is present at this 5mCp-8-OHdG site, it performs another critical function: it recruits TET1 to the 8-OHdG lesion. TET1 then oxidizes the 5mC directly adjacent to the 8-OHdG. This pivotal oxidation initiates the 5mC's entry into the DNA demethylation pathway (as clearly illustrated in the Figure titled "Initiation of DNA demethylation at a CpG site"). [34] This pathway begins with the formation of 5-hydroxymethylcytosine, which may either persist in the DNA or undergo further oxidative reactions, ultimately followed by base excision repair, to restore the nucleoside at that position back to cytosine (see Figure "Demethylation of 5-Methylcytosine (5mC) in neuron DNA"). It's a remarkably intricate way to remove a repressive mark, requiring damage, binding, and recruitment.

The human genome, in its vast complexity, contains approximately 28 million CpG sites, with an average frequency of about 1 per hundred base pairs. [44] When an intense learning situation, such as contextual fear conditioning, is applied to rats, [45] it can result in a life-long, indelible fearful memory after a single training event. [45] While the long-term memory of this event appears to be initially stored in the hippocampus, this storage is notably transient; it does not permanently reside there. [45] A significant portion of the long-term storage of contextual fear conditioning memory appears to migrate and take root in the anterior cingulate cortex. [46] (For visual reference, consult the Figure: Brain regions involved in memory formation, and this additional reference. [47])

When rats undergo contextual fear conditioning, a cascade of epigenetic changes ensues. More than 5,000 differentially methylated regions (DMRs), each spanning approximately 500 nucleotides, are observed within the rat hippocampus neural genome, both one hour and 24 hours after conditioning. [48] This profound epigenetic reorganization leads to the upregulation of about 500 genes (often due to hypomethylation of CpG sites) and the downregulation of approximately 1,000 genes (frequently due to newly formed 5mC at CpG sites in a promoter region). This specific pattern of induced and repressed genes within neurons appears to provide a molecular blueprint for forming this initial, transient memory of the training event in the rat brain's hippocampus. [48]

A similar picture emerges in mice. One hour after contextual fear conditioning, 675 genes were found to be demethylated and 613 genes hypermethylated in the hippocampal region of the mouse brain. [49] However, these changes, much like in rats, proved to be transient within hippocampal neurons, with almost none persisting after four weeks. Yet, in mice subjected to conditional fear conditioning, a stark contrast was observed after four weeks: there were over 1,000 differentially methylated genes and more than 1,000 differentially expressed genes in the anterior cingulate cortex. [49] This region, it bears repeating, is precisely where long-term memories are known to be stored in the mouse brain. [46] It seems memory, like everything else, is eventually outsourced to a more permanent, if equally complex, storage facility.

Epigenetics In Learning And Memory