learning and memory - university of texas at...

Learning and memory

Memory• Nondeclarative (implicit) memory - skills, habits, conditioning

• non-Associative

• habituation (repeat stimulus and one stops responding) and sensitization (enhanced response to a benign tone after a noxious stimulus).

• Associative pairing of two things the US + CS --> UR

• Classical associative conditioning - All of the various Pavolovian variants

• Operant conditioning - trial and error learning, press one key not another to get a reward.

• Declarative (explicit) memory - things that you remember often subdivided into episodic (events) and semantic (facts) - hippocampus

• Long-term memory - requires protein synthesis, >= 24hrs

• Short-term memory - protein synthesis independent

Associative learning Terms

• Operant conditioning - a behavior is reinforced or punished, learning produces modified behavior

• Classical conditioning - two stimuli are delivered at the same time or so that one predicts the occurrence of the other.

• US - unconditioned stimulus (foot shock)

• CS - conditioned stimulus (a tone)

Associative learning Terms - classical conditioning

• Pavlovian classical conditioning - pairing of US and CS. Most people are familiar with this.

• Context-dependent fear conditioning - the cage (room) with certain dimensions and features. The place is the US and a shock is the CS.

Associative learning

• Sound tone produces a no response

• Foot shock produces a strong response

• Pair them repeatedly and get a strong response

• Sound tone = conditional stimulus

• Foot shock = unconditional stimulus

Behavioral assays

Contextual Fear Conditioning

hippocampus and amygdala are important

animal learns to associate novel context with aversive stimulus

Video courtesy of Farah Lubin

3-shock training


Keep in mind that when placed in a chamber a rat normally does a thorough inspection.

Test day #1 24 hrs later

No new shock is given.

The animal is just placed in the chamber


Test day #1 24 hrs later

Obviously different context?

No freezing

Morris water maze• http://www.youtube.com/watch?

v=_90ch04yk5Y&feature=related

• http://www.youtube.com/watch?v=LrCzSIbvSN4

• Hippocampal learning

Electrophysiological Assays


• Sound tone produces a no response

• Foot shock produces a strong response

• Pair them repeatedly and get a strong response

• Sound tone = conditional stimulus

• Foot shock = unconditional stimulus


CS (tone)

neuron B

Weak Response

neuron C

neuron A


US (shock)

Strong Response

neuron B

neuron C

neuron A


CS (tone)

US (shock)

neuron B

neuron C

neuron A

Strong Response


CS (tone)

neuron B

neuron C

neuron A

Strong Response

Hebb’s Postulate: “When an axon of cell A . . . excites cell B and repeatedly or persistently takes part in firing it, some . . . metabolic change takes place in one or both cells so that A’s efficiency as one of the cells firing B is increased”. Hebb DO. 1949. The Organization of Behavior. New York: Wiley

CS (tone)

US (shock)

neuron B

neuron C

neuron A

Strong Response

neuron B Response

neuron A

neuron B Response

neuron A

neuron B Response

neuron A

Post-tetanic potentiation PTP

tetanic stimulation

LTPLong-term potentiation

http://www.unmc.edu/physiology/Mann/mann19.html

The requirement for LTP is simultaneous depolarization of the post-synaptic cell by more than

one synaptic input.

http://users.rcn.com/jkimball.ma.ultranet/BiologyPages/L/LTP.html

EPSP

Derived from Kandel, ER, JH Schwartz and TM Jessell (2000) Principles of Neural Science. New York: McGraw-Hill.

The requirement for LTP is simultaneous depolarization of the post-synaptic cell by more than

one synaptic input (multiple from same neuron)

Hippocampus and Learning and memory

1957 patient HMbilateral removal of hippocampus to relieve extreme epilepsyLost his ability to form long term memories

Miller CA, Sweatt JD (2007) Covalent modification of DNA regulates memory formation.

Neuron 53: 857-869.

DNMT

• Fear conditioned animals show increase in DNMT3A and 3B expression in hippocampus CA1 region.

in vivo (Levenson et al., 2004a). Thus, a model is emergingwhereby hippocampus-dependent memory formation isinitiated by activation of NMDA receptors, which leadsto an influx of calcium, activation of signaling pathways,and altered gene transcription mediated in part bychanges in chromatin structure.

In the current study, we explored the potential role ofanother epigenetic mechanism, cytosine-50 methylation,in memory formation. Many developmentally importantprocesses utilize this ‘‘prima donna’’ of epigenetics (Scar-ano et al., 2005; Santos et al., 2005), including gene im-printing, cell differentiation, X chromosome inactivation,and long-term transcriptional regulation (Bestor et al.,1988; Okano et al., 1998). This covalent modification ofDNA is catalyzed by DNA (cytosine-50) methyltransferases(DNMTs) and involves the transfer of a methyl group to the50 position of cytosine residues, canonically at CG dinucle-otides. Expression and activity of DNMTs is generallyrestricted to dividing cells and is very high during earlydevelopment (Szyf et al., 1985, 1991; Monk et al., 1987;Singer-Sam et al., 1990; Goto et al., 1994). DNA methyla-tion can induce long-term transcriptional silencingthrough direct interference with transcription factor bind-ing. In addition, methylated DNA can counter the tran-scriptional effects of histone acetylation by recruitingchromatin remodeling enzymes, including histone deace-tylases (HDACs), via the action of methyl-CpG bindingdomain proteins (MBDs) like MeCP2 (Becker et al.,1987; Nan et al., 1997, 1998; Jones et al., 1998; Crosset al., 1997).

DNAmethylation has been studied extensively in devel-opment and has long been considered a static processfollowing cell differentiation, because typically DNMT ex-pression greatly diminishes once terminal differentiationhas occurred (Bestor et al., 1988; Szyf et al., 1985, 1991;Monk et al., 1987; Singer-Sam et al., 1990; Goto et al.,1994; Deng and Szyf, 1999). Because the mammalianbrain primarily consists of postmitotic neurons and glialcells that possess relatively low proliferative potential, re-ports that the adult mammalian CNS possesses highlevels of DNMT mRNA and enzymatic activity were unex-pected (Monk et al., 1987; Goto et al., 1994; Brooks et al.,1996). Early studies into the function of DNMT in the brainsuggested that this enzyme might be involved in DNA re-pair and neurodegeneration (Brooks et al., 1996; Endreset al., 2000; Fan et al., 2001; Endres et al., 2001). However,recent studies have also implicated misregulation of DNAmethylation and DNMTs in such cognitive disorders asschizophrenia, Rett syndrome, and Fragile X mental retar-dation (Veldic et al., 2004; Amir et al., 1999; Sutcliffe et al.,1992).

To begin investigating a potential role for DNA methyla-tion in the adult CNS, we examined a provocative possibil-ity contrary to the prevailing model of an exclusive role forDNA methylation in development. Thus, we investigatedwhether DNA methylation regulates memory consolida-tion in the adult CNS via gene-specific control of trans-cription within the hippocampus.

RESULTS

DNA Methyltransferase Activity Is Necessaryfor Memory FormationWe recently characterized the effects of DNMT inhibitionon hippocampal synaptic plasticity (Levenson et al.,2006). We found that DNA of the gene reelin, which isinvolved in the induction of synaptic plasticity, exhibitsrapid decreases in cytosine methylation when DNMT ac-tivity is blocked in acute hippocampal slices. We alsofound that DNMT inhibition prevents the induction ofLTP. These findings suggested that DNA methylationmight be dynamically regulated in the adult nervous sys-tem and serve as an additional epigenetic mechanismgoverning memory formation. To pursue this idea, wefirst investigated whether or not DNMT mRNA levels inthe hippocampus are altered by contextual fear condi-tioning, a hippocampus-dependent associative memoryparadigm. Using real-time quantitative PCR, we exam-ined DNMT mRNA levels in the adult rat hippocampus30 min after training for contextual fear conditioning.We assayed levels of three DNMT subtypes, DNMT 1,3A, and 3B, as well as the immediate-early gene c-fos,which is rapidly induced in the hippocampus by fearconditioning (Melia et al., 1996; Maciejak et al., 2003;Huff et al., 2006). Though there is some overlap in func-tion, DNMT1 has preferential activity for hemimethylatedDNA and is traditionally considered a maintenancemethyltransferase in DNA replication, while 3A and 3Bare responsible for de novo methylation (Siedlecki andZielenkiewicz, 2006). Animals exposed to the associativecontext-plus-shock training displayed an increase inDNMT3A and 3B mRNA in area CA1 relative to animalsexposed only to the novel context of the fear conditioningchamber (DNMT3A: t7 = 2.88; DNMT3B: t7 = 4.56; c-fos:t7 = 2.81; p values < 0.05; Figure 1). These initial findingssuggested the intriguing possibility that DNMT activitymight be dynamically regulated in the adult CNS in vivoin response to environmental sensory stimulation.

Figure 1. Fear Conditioning Is Associated with an Upregula-tion of DNMT mRNADNMT3A, DNMT3B, and c-fos mRNA in area CA1 are upregulated

within 30 min of fear conditioning in context-plus-shock animals, rela-

tive to context-only controls. *p < 0.05. Error bars represent SEM.

858 Neuron 53, 857–869, March 15, 2007 ª2007 Elsevier Inc.

Neuron

DNA Methylation and Memory Formation

DNMT inhibitor infused into CA1 blocks learning

Because fear conditioning led to an upregulation of hip-pocampal DNMT mRNA, we next tested the necessity ofDNMT activity for memory formation. For this experiment,we infused one of two distinct DNMT inhibitors, 5-aza-deoxycytidine (5-AZA) or zebularine (zeb), directly intoarea CA1 of the hippocampus immediately after contex-tual fear conditioning. Infusions were administered aftertraining for this hippocampus-dependent task to avoidstate-dependent effects of the drug. There were no differ-ences in time spent freezing between the animals infusedwith 5-AZA and those infusedwith zeb during the retentiontests (p > 0.05), nor were there any differences betweentheir respective vehicle groups (0.8% acetate and 10%DMSO; p > 0.05). Therefore, the 5-AZA and zeb datawere collapsed into one DNMT inhibitor group, as werethe two vehicle groups. When memory was assessed24 hr later (test day 1), animals infused with a DNMT inhib-itor (5-AZA or zeb) displayed significantly less freezingthan their vehicle-treated (VEH) counterparts (F(1,22) =103.9; p < 0.001; Figure 2A), indicating that hippocampalDNMT activity is necessary for memory consolidation.

DNA methylation is not generally considered to bea plastic process; in development, alterations in DNAmethylation are essentially permanent. This considerationraised the question of whether or not the effects of DNMTinhibition on the capacity for memory formation are per-manent. To address this, we assessed the ability of thesesame DNMT inhibitor-treated animals to form the fearmemory later on, in the absence of the drug. For this ex-periment, animals treated with DNMT inhibitor or vehicle24 hr earlier were retrained for contextual fear conditioningimmediately after testing on test day 1. Twenty-four hourslater, fear memory was again assessed (test day 2). Thefreezing behavior in vehicle-treated animals on test day2 was significantly greater than their freezing during test1 (p < 0.005) and lasted for nearly the entire test period(Figure 2A). This result was expected, as this test followeda second training trial for a task in which a single trial is suf-ficient to form a strong, long-lasting memory. Animalstreated with DNMT inhibitor after the first training trialand subsequently retrained on test day 1 showed signifi-cantly greater freezing on test day 2 as compared to theirperformance on test day 1 (p < 0.005). This result estab-lishes that the drug infusion did not damage the hippo-campus. More importantly, it demonstrates that the ef-fects of DNMT inhibition are not immuteable. Rather, thechanges are plastic, allowing the DNA methylation statesnecessary for memory consolidation to be re-establishedafter transient DNMT inhibition and DNA demethylation.

Interestingly, test day 2 freezing in previously DNMTinhibitor-treated animals was equivalent to the freezingdisplayed by vehicle-treated animals on test day 1 and

Figure 2. DNMT Inhibition Blocks Memory Consolidation ina Plastic Manner(A) Intra-CA1 infusion of DNMT inhibitor immediately after contextual

fear conditioning training blocked consolidation, as demonstrated by

an absence of freezing behavior at the 24 hr test (test day 1). However,

memory is formed normally (test day 2) if these same animals are re-

trained immediately after test 1 and allowed to consolidate thememory

in the absence of drug. A third round of training establishes that DNMT

inhibitor animals are capable of forming memories equal in strength to

vehicle-treated animals (test day 3) (F(5,54) = 73.08). * Denotes signifi-

cantly greater than test day 1 DNMT inhibitor, p < 0.005. # Denotes sig-

nificantly different from all others, p < 0.005.

(B) DNMT inhibitor infusions fail to block memory formation if adminis-

tered 6 hr after training.

(C) Location of needle tips for all intra-CA1 infusions. Diagram repre-

sents histology from animals whose behavioral data are depicted in

(A) and (B) and Figure 5. Because of the extensive overlap between

the infusion needle tips of these animals, not all tip locations are resolv-

able on this diagram.

Error bars represent SEM.

Neuron 53, 857–869, March 15, 2007 ª2007 Elsevier Inc. 859

Neuron


Inhibitor is given immediately after the first training sesson.

Because fear conditioning led to an upregulation of hip-pocampal DNMT mRNA, we next tested the necessity ofDNMT activity for memory formation. For this experiment,we infused one of two distinct DNMT inhibitors, 5-aza-deoxycytidine (5-AZA) or zebularine (zeb), directly intoarea CA1 of the hippocampus immediately after contex-tual fear conditioning. Infusions were administered aftertraining for this hippocampus-dependent task to avoidstate-dependent effects of the drug. There were no differ-ences in time spent freezing between the animals infusedwith 5-AZA and those infusedwith zeb during the retentiontests (p > 0.05), nor were there any differences betweentheir respective vehicle groups (0.8% acetate and 10%DMSO; p > 0.05). Therefore, the 5-AZA and zeb datawere collapsed into one DNMT inhibitor group, as werethe two vehicle groups. When memory was assessed24 hr later (test day 1), animals infused with a DNMT inhib-itor (5-AZA or zeb) displayed significantly less freezingthan their vehicle-treated (VEH) counterparts (F(1,22) =103.9; p < 0.001; Figure 2A), indicating that hippocampalDNMT activity is necessary for memory consolidation.

DNA methylation is not generally considered to bea plastic process; in development, alterations in DNAmethylation are essentially permanent. This considerationraised the question of whether or not the effects of DNMTinhibition on the capacity for memory formation are per-manent. To address this, we assessed the ability of thesesame DNMT inhibitor-treated animals to form the fearmemory later on, in the absence of the drug. For this ex-periment, animals treated with DNMT inhibitor or vehicle24 hr earlier were retrained for contextual fear conditioningimmediately after testing on test day 1. Twenty-four hourslater, fear memory was again assessed (test day 2). Thefreezing behavior in vehicle-treated animals on test day2 was significantly greater than their freezing during test1 (p < 0.005) and lasted for nearly the entire test period(Figure 2A). This result was expected, as this test followeda second training trial for a task in which a single trial is suf-ficient to form a strong, long-lasting memory. Animalstreated with DNMT inhibitor after the first training trialand subsequently retrained on test day 1 showed signifi-cantly greater freezing on test day 2 as compared to theirperformance on test day 1 (p < 0.005). This result estab-lishes that the drug infusion did not damage the hippo-campus. More importantly, it demonstrates that the ef-fects of DNMT inhibition are not immuteable. Rather, thechanges are plastic, allowing the DNA methylation statesnecessary for memory consolidation to be re-establishedafter transient DNMT inhibition and DNA demethylation.

Interestingly, test day 2 freezing in previously DNMTinhibitor-treated animals was equivalent to the freezingdisplayed by vehicle-treated animals on test day 1 and

Figure 2. DNMT Inhibition Blocks Memory Consolidation ina Plastic Manner(A) Intra-CA1 infusion of DNMT inhibitor immediately after contextual

fear conditioning training blocked consolidation, as demonstrated by

an absence of freezing behavior at the 24 hr test (test day 1). However,

memory is formed normally (test day 2) if these same animals are re-

trained immediately after test 1 and allowed to consolidate thememory

in the absence of drug. A third round of training establishes that DNMT

inhibitor animals are capable of forming memories equal in strength to

vehicle-treated animals (test day 3) (F(5,54) = 73.08). * Denotes signifi-

cantly greater than test day 1 DNMT inhibitor, p < 0.005. # Denotes sig-

nificantly different from all others, p < 0.005.

(B) DNMT inhibitor infusions fail to block memory formation if adminis-

tered 6 hr after training.

(C) Location of needle tips for all intra-CA1 infusions. Diagram repre-

sents histology from animals whose behavioral data are depicted in

(A) and (B) and Figure 5. Because of the extensive overlap between

the infusion needle tips of these animals, not all tip locations are resolv-

able on this diagram.

Error bars represent SEM.


Neuron


If you wait 6 hrs to give the inhibitor then there is no effect

DNMT blocks consolidation

Fear conditioning alters memory gene

• Increased CpG island methylation on protein phosphatase 1 - reduced PP1 expression. This enzyme promotes forgetting.

• DNMT inhbitors prevent this change

• Decreased CpG island methylation on Reelin - increased Reelin expression. This receptor promotes learning.

date, there is strong evidence supporting a role for chro-matin modifications in memory consolidation (Swankand Sweatt, 2001; Guan et al., 2002; Huang et al., 2002;Levenson et al., 2004a, 2006; Korzus et al., 2004; Alarconet al., 2004;Wood et al., 2005; Kumar et al., 2005; Chwanget al., 2006). The findings presented here, however, areparticularly significant because they are in such stark con-trast to the current understanding of the role of cytosinemethylation in development. It is thought that DNA meth-ylation is crucial for normal development and that embry-onic methylation patterns are maintained in perpetuitypostnatally—only to be perturbed in cases like cancer inwhich abberant hypomethylation occurs and a subse-quent loss of transcriptional regulation of these genes(Santos et al., 2005). The results of our study, however, in-dicate that, at least within the hippocampus, DNA methyl-ation levels can be rapidly and dynamically altered byenvironmental stimuli that induce associative learning.This finding necessitates a shift in the way we think aboutcellular roles for DNA methylation.Our findings also complement recent discoveries con-

cerning epigenetic aberrations observed in cancer andcognitive disorders. Just as we observed during memoryconsolidation, cancer research has revealed bidirectionalDNA methylation-dependent regulation of genes. For ex-ample, tumorogenisis appears to be driven by global hy-pomethylation working in concert with hypermethylationof a specific subset of genes (Luczak and Jagodzinski,2006). In addition, recent studies have implicated misre-gulation of DNA methylation in a number of cognitive dis-orders, including several autism spectrum disorders andschizophrenia (reviewed in Weeber et al., 2002b; Graysonet al., 2006). Fragile X mental retardation results from ab-normal trinucleotide expansion, which leads to decreasedgene expression through aberrant DNA methylation andrestrictive chromatin structure (reviewed in Weeberet al., 2002b). Rett syndrome is associated with mutationsin MeCP2, one of the MBDs recruited by methylated DNAthat contributes to gene silencing (Amir et al., 1999; Collinset al., 2004). And a recent study identified an overlappingpathway of gene dysregulation within 15q11-13 in Rett,

Angelman syndrome, and autism and implicated MeCP2function in all three through studies of MeCP2-deficientmice and human Rett, Angelman syndrome, and autismbrains (Samaco et al., 2005).

In addition, hypermethylation of the reelin gene is a rap-idly emerging hypothesis as a potential basis for schizo-phrenia, a disorder marked by a variety of cognitive defi-cits. In the cortex of schizophrenic patients, there istypically a 50% reduction in reelin mRNA, an effect asso-ciated with aberrant methylation of the gene (Chen et al.,2002). The findings presented here may provide an impor-tant and relevant piece of data to the schizophrenia field,as they provide evidence that reelinmethylation is subjectto modulation in response to experience and environmen-tal stimuli. In addition, the current findings indicate that notall alterations in DNA methylation are aberrant; rather,some changes naturally occur during normal memoryformation.

In this study we report that DNMT inhibition preventsmemory consolidation. However, because both 5-AZAand zebularine are potent DNMT inhibitors, affecting allthree subtypes (DNMT1, 3A, and 3B; Weisenbergeret al., 2004; Marquez et al., 2005), we are unable to deter-mine which of the subtypes is important for hippocampalmemory formation. We also report an increase in themRNA of the two de novo methyltransferases (3A and3B) within a half hour of fear conditioning training. Perhapsboth DNMT3A and 3B work in concert to bring about denovo methylation of the necessary genes, such as PP1,and transcriptionally silence these genes, aiding in mem-ory formation.

An additional implication of the current findings is thata DNA demethylase must exist and be regulated in theadult CNS. Identification of a demethylating enzyme isone of the more intriguing and controversial aspects ofthe current DNA methylation literature. At present there isno clearly identified DNA demethylase; however, at leasttwo candidates are identifiable based on current literature.Detich and colleagues have reported demethylase activityof an MBD in vitro (Detich et al., 2002)—overexpression ofMBD2 in cell culture induced demethylation at a number of

Figure 9. Schematic Representation ofthe Role DNA Methylation May Be Play-ing in the Transcriptional Regulation ofMemory Formation in the HippocampusNote: The receptors, kinases, and transcription

factors depicted in gray play established roles

in hippocampal memory consolidation. How-

ever, the present study does not address the

potential link between these proteins and the

DNA methylation we report here to be impor-

tant for memory formation.


Neuron


Lubin FD, Roth TL, Sweatt JD (2008) Epigenetic regulation of BDNF gene transcription in the consolidation of

fear memory. J Neurosci 28: 10576-10586.

Brain-derived neurotrophic factor (BNDF)

• Contributes to LTP and memory.

• A neurotrophic factor that promotes synapse strength and neuronal survival

moter and exonic regions of the bdnf gene (Fig. 2A) (see Materialsand Methods).

Data generated using MSP indicated that context exposurealone elicited a decrease in methylated DNA levels associatedwith exons I and VI relative to naive or shock-alone controls (Fig.2A). No changes in methylated DNA levels associated with exonsII and IV were observed with context exposure alone comparedwith naive or shock-alone controls. With contextual fear condi-tioning, we found significant decreases in methylated DNA levels

associated with exons I and IV in area CA1of hippocampus at 2 h compared with na-ive controls with no change at exon II (Fig.2A). We also observed prominent in-creases in methylated exon VI DNA at 2 hcompared with naive controls (Fig. 2A).As a control, we measured levels of un-methylated DNA within the bdnf gene af-ter context exposure or fear conditioningusing primer sequences designed to am-plify unmethylated CpG island sites (sup-plemental Fig. 1, available at www.jneurosci.org as supplemental material).Overall, these data support the idea thatDNA methylation controls exon-specificreadout of the bdnf gene, because there is acorrelation of decreased methylation atspecific exon-associated CpG islands (Fig.2A) with increased exon transcription(Fig. 1D). In addition, they indicate a sur-prising complexity to the control of DNAmethylation at the bdnf gene locus withspecific decreases and increases in methyl-ation at individual initiation start sites be-ing associated with specific bdnf tran-scripts regulated in response to a novelcontext versus associative fear conditioning.

As another independent assessment ofaltered methylation, we confirmed ourMSP data using direct bisulfite sequencingPCR (BSP) to examine site-specific meth-ylation within the exon IV region. We ex-amined exon IV specifically because it wasthe only bdnf transcript isoform we foundsignificantly increased in area CA1 withfear conditioning (Fig. 1D), and our MSPdata suggested that this increase in tran-scription was associated with decreasedmethylation at this locus (Fig. 2A). A sche-matic of the 12 CpG dinucleotides withinthe exon IV region screened by MSP isshown in Figure 2B. This exon IV regioncontains a cAMP response element (5!-TCACGTCA-3!, located between basepairs "38 and "31) site for the transcrip-tion factor cAMP response element-binding protein (Shieh et al., 1998; Tao etal., 1998), which encompasses CpG site 1.Sequencing data confirm active demethyl-ation of bdnf exon IV after fear condition-ing (Fig. 2B). Interestingly, although MSPfailed to detect bdnf exon IV methylationchanges with context exposure alone, BSPanalysis revealed active demethylation of

exon IV after context exposure. However, post hoc analyses re-vealed that demethylation of exon IV is greater at specific CpGsites with fear conditioning versus context exposure alone,which is consistent with the hypothesis of differential regula-tion of this site in response to different behavioral experiences.Thus, the bisulfite DNA sequencing data reiterate active reg-ulation of DNA methylation within the bdnf gene, and further-more demonstrate that bdnf methylation is regulated in re-sponse to fear conditioning or context exposure.

Figure 1. Increased bdnf gene expression after context exposure and fear conditioning. A, Schematic representation of thecontextual fear-conditioning test protocol. Animals were exposed to the training chamber and either received a series of foot-shocks after being exposed to the context (Context # Shock), being exposed to the context alone (Context # No Shock), orreceiving only the footshock without being exposed to the context (No Context # Shock). B, Animals were reexposed to thetraining chamber 24 h later and tested for freezing behavior. n $ 8 –9/group; *p % 0.001 compared with shock-alone controls.C, bdnf mRNA in area CA1 of hippocampus is increased within 0.5 h of context exposure (t(8) $ 2.48, p $ 0.0381, n $ 5) andcontext # shock (t(8) $ 2.41, p $ 0.0425, n $ 5) compared with naive controls. At 2 h, bdnf mRNA expression peaks in area CA1of hippocampus in fear-conditioned animals (t(9) $ 3.15, p $ 0.0117, n $ 5– 6). At 24 h, bdnf mRNA levels returned to baselinein area CA1 of hippocampus from both context (t(6) $ 0.42, p $ 0.6887, n $ 4) and context # shock (t(7) $ 1.97, p $ 0.0894,n $ 4 –5) animals relative to naive controls. D, After context exposure alone, exon I and VI bdnf mRNA increased in area CA1 (exonI, t(3) $3.42, p$0.0418, n$4; exon VI, t(5) $2.66, p$0.0449, n$6) with a corresponding increase in total bdnf mRNA levels(exon IX, t(4) $ 3.38, p $ 0.0279, n $ 5) relative to naive control. After fear conditioning, exon IV bdnf mRNA increased (t(3) $5.88, p $ 0.0098, n $ 4) in area CA1 of hippocampus with an increase in total bdnf gene expression as assessed by exon IX mRNA(t(5) $ 3.49, p $ 0.0175, n $ 6). No significant changes in exon II bdnf mRNA were observed with context alone or context #shock relative to naive control. The solid line across the bars represents normalized naive control levels [one-sample t test, *p %0.05, **p % 0.01, compared with naive controls; Student’s t test, not significant (ns), #p % 0.05, ##p % 0.01, compared withcontext alone]. Error bars indicate SEM.

Lubin et al. • Epigenetic Regulation of bdnf Gene during Memory Consolidation J. Neurosci., October 15, 2008 • 28(42):10576 –10586 • 10579















Also see decreased CpG island methylation

served decrease in percentage methylation at cytosine residues atexon IV CpG sites with fear conditioning was significantly alteredwith DNMT inhibition (Fig. 6A). Consequently, we also foundthat DNMT inhibition significantly attenuated exon IV mRNAlevels with fear conditioning (Fig. 6B). These results are consis-tent with the finding that DNMT inhibition decreased histoneacetylation at bdnf promoter 4 during memory consolidation.Moreover, these findings provide a parsimonious explanation forthe mechanism by which DNMT inhibition blocks memory for-mation: through inhibition of transcription of a gene that pro-motes synaptic plasticity and memory, bdnf. It is important tonote that although DNMT inhibition reduced total bdnf mRNAlevels (exon IX) produced by fear conditioning, total bdnf expres-sion remained significantly elevated relative to naive controls(data not shown).

NMDA receptor activation is necessary for regulation of bdnfDNA methylation by contextual fear conditioningGiven the documented role of NMDA receptor activation inlong-term memory formation and hippocampal chromatin re-modeling (Levenson et al., 2004; Chwang et al., 2006), we deter-mined whether or not NMDA receptor activation was necessary

for bdnf DNA methylation and altered gene expression in contex-tual fear conditioning. Again, we focused on associative fear con-ditioning because the most robust increases in bdnf DNA meth-ylation in hippocampus were observed with associativecontextual fear learning.

We examined the effect of inhibiting NMDA receptor activa-tion on memory consolidation by injecting rats with the non-competitive antagonist MK801 1 h before training (Fig. 7A).Twenty-four hours later (test day 1), MK801-treated animals dis-played significantly less freezing than did vehicle-treated animals(t(14) ! 14.90, p ! 0.0001, n ! 8), as has been previously reported(Levenson et al., 2004; Chwang et al., 2006). We next isolatedmRNA from either vehicle- or MK801-treated fear-conditionedanimals at 2 h after training and assessed exon IX mRNA levels(all transcripts) in area CA1 of hippocampus. Vehicle-treatedanimals displayed a significant increase in total bdnf mRNA inarea CA1 of hippocampus relative to naive controls, whereasMK801 treatment blocked this increase (Fig. 7B). These findingsdemonstrate that NMDA receptor activation is necessary for in-creased bdnf gene transcription during consolidation of contex-tual fear memory. In control experiments we confirmed that theeffects of NMDA receptor blockade on bdnf mRNA expressionwere reversible after 24 h (data not shown).

To ascertain whether NMDA receptor activation mediates al-tered DNA methylation at the bdnf locus during consolidation ofcontextual fear conditioning, we examined the effect of MK801on the methylation status of DNA associated with exon IV. Weobserved no change in exon IV methylation in area CA1 of hip-pocampus from MK801-treated animals, indicating that NMDAreceptor blockade during fear-conditioning training preventeddecreased bdnf methylation (Fig. 7C). Moreover, the increase inunmethylated exon IV levels normally seen with fear condition-ing was also blocked with MK801 treatment (supplemental Fig. 1,available at www.jneurosci.org as supplemental material; Fig.7C). Importantly, the lack of altered exon IV methylation afterMK801 treatment was paralleled by a block in exon IV mRNAlevels after fear conditioning when compared with vehicle treat-ment (Fig. 7D). BSP analysis confirmed that cytosine residues atexon IV CpG sites were demethylated with fear conditioning, andthat MK801 treatment significantly attenuated demethylation(Fig. 8). Interestingly, we found that MK801 treatment did notalter demethylation at CpG site 3 of exon IV normally observedwith fear conditioning, thus suggesting complex CpG site-specific regulation of DNA methylation within the bdnf gene. Inparallel experiments, we examined the effect of MK801 on themethylation status of another site within the bdnf gene, the bdnfcoding exon (exon IX) via two approaches, MSP and methylatedDNA immunoprecipitation. Using these two independent meth-ods, we confirmed that exon IX methylation also decreased withfear conditioning and that MK801 treatment blocked this effectat exon IX (supplemental Fig. 2, available at www.jneurosci.org assupplemental material). Together, these findings confirm thatNMDA receptor activation contributes to the regulation of bdnfgene methylation during contextual fear conditioning.

DiscussionIn this study we produced four main findings. First, we founddifferential regulation of bdnf exons in hippocampus after con-text exposure versus fear conditioning, and found that both typesof environmental stimuli triggered altered DNA methylation atthe bdnf gene locus. Second, we observed that DNA methylationis dynamic in the adult CNS and appears to regulate bdnf genereadout during memory consolidation in the adult hippocam-

Figure 5. Histone modifications at exon-specific bdnf promoters during memory consolida-tion. A, At the bdnf gene, PH3/AcH3 and AcH3 levels were increased at promoter 4 during fearmemory consolidation relative to naive controls (PH3/AcH3, t(2) ! 4.185, p ! 0.05; AcH3, t(2)

! 6.21, p ! 0.0250, n ! 3). AcH4 levels were significantly reduced at promoter 2 relative tonaive controls (t(2) ! 12.26, p ! 0.0066, n ! 3). The top shows the schematic location of thebdnf promoters preceding each exon (P1, P2, P4, and P6) as indicated by the star. In the graph,the areas with no bars indicate that the anti-histone antibody precipitated negligible levels ofthe bdnf promoter regions that were not detectable. B, At the bdnf promoter 4, AcH3 levels wereincreased during memory consolidation and significantly attenuated with DNMT inhibition[zebularine (ZEB)] relative to naive controls (t(8) ! 3.047, p ! 0.0318, n ! 4 – 6). C, Therewere no significant histone modifications (AcH3) around the !-actin promoter during memoryconsolidation (n ! 3; one-way ANOVA, *p " 0.05, compared with naive controls; Student’s ttest, #p " 0.05, compared with context # shock alone). Error bars indicate SEM.

10582 • J. Neurosci., October 15, 2008 • 28(42):10576 –10586 Lubin et al. • Epigenetic Regulation of bdnf Gene during Memory Consolidation

Infusion of a DNMT inhibitor

• Block methylation change at BDNF IV promoter, block change in BDNF expression, block learning.

NMDA receptor

• Ion channel whose modulation has been shown to underlie LTP.

• Blocking this channel during training blocks, BDNF expression change, change in CpG island methylation, and learning.

learning and memory - university of texas at...

Documents