Tải bản đầy đủ - 0trang
Section 5. Transcription Factors, Genes and Proteins
formation were conducted in Aplysia and Drosophila. Subsequent studies in vertebrate species
(including, in particular, mice and rats) using a variety of molecular-genetic tools suggest that
CREB has a highly conserved role in LTM formation.
CREB is a member of a family (CREB/ATF) of structurally similar, activity-regulated transcription factors. In mammals, at least three genes encode the CREB-like proteins, CREB,
CREM (cAMP Response Element Modulator) and ATF-1 (Activating Transcription Factor).38,55,94 The mammalian CREB gene comprises 11 exons,26,54,115 and alternative splicing
generates the three major activator isoforms of CREB: α, ∆, and β.11,47,118 Each of these is
highly expressed in all tissues. In addition to these transcriptional activators, the CREB family
also includes repressors of transcription. For example, the CREM gene codes at least four
isoforms that repress CRE-dependent transcription: the CREM α, β and γ proteins as well as
the inducible cyclic AMP early repressor (ICER).37,82
CREB regulates gene expression in response to a wide array of extracellular signals. In its
inactive state, CREB is prebound as a dimer to CRE sites in the promoter regions of target
genes. Neuronal stimulation may lead to the activation of CREB (via activation of various
CREB kinases). In its activated form CREB binds CREB-binding protein (CBP); the recruitment of CBP links CREB directly and indirectly to other components of the basal transcription machinery, thus promoting transcription.24
A large number of signaling pathways converge on CREB, indicating that the transcriptional activity of CREB is regulated by a wide variety of extracellular signals.31,77,100 Each of
these pathways activate CREB via CREB kinases that phosphorylate CREB at serine 133
(Ser133). This is the critical residue for the transcriptional activity46 since mutation of this
residue to a nonphosphorylatable alanine (Ala) residue abolishes the transcriptional response
to elevated cAMP levels.46,83 Although CREB was initially identified as a transcription factor
that responds to elevated levels of cAMP, it is now clear that CREB may be activated by three
major signaling pathways (Fig. 1): 1) cAMP, 2) Ca2+, and 3) growth factors.
1) cAMP: The activation of G-protein linked receptors (e.g., D1 receptors) leads to the
increases in the second messenger cAMP via activation of adenylate cyclase.44 Rises in levels of
cAMP lead to the activation of protein kinase A (PKA) by dissociating the regulatory (R) from
the catalytic (C) subunits. The C subunits of PKA passively translocate to the nucleus where
they may phosphorylate CREB at Ser133.5,29,50
2) Ca2+: Calcium is a pleiotropic second messenger that is activated via several different
mechanisms following changes in membrane potential. Extracellular Ca2+ may enter the cytoplasm via ligand-gated ion channels of NMDA and AMPA receptors, or via voltage-gated
calcium channels. In addition, Ca2+ may be released from intracellular stores.100 Calcium signals are then transduced via a large number of different CREB kinases which include: CamKII,
CaMKIV, RSK1-3 (via Ras-ERK), PKC and PKA.10,29,33,75,101,107 The different kinetics of
each of these pathways provides a mechanism for sustained CREB activation and CRE-mediated transcription. For example, activation of CaMKIV produces a wave of CREB phosphorylation with rapid on- and offset (lasting only minutes), whereas activation of the Ras-ERK-RSK2
pathway promotes a slower phase of CREB phosphorylation.116 Furthermore, the distinct kinetic properties of these upstream regulatory pathways may allow CREB to compute information regarding the exact nature of the stimuli, perhaps allowing for specific stimuli (or patterns
of stimulation) to be translated into specific patterns of gene expression. For example, recent
data indicate that Ca2+ influx into neurons causes the phosphorylation of CREB at Ser142 and
Ser143 (in addition to Ser133), and that CREB-induced transcription induced by this triple
phosphorylation may not require the participation of CBP.67 Therefore, Ca2+ influx promotes
CREB-mediated transcription via a set of mechanisms that are distinct from those produced by
other extracellular activation.
From Messengers to Molecules: Memories Are Made of These
Figure 1. Activation of CREB by a multiple signaling pathways. In the first pathway, a neurotransmitter may
bind to a receptor (R) that is linked to a G-protein (G), which leads to the increases in the second messenger
cAMP via activation of adenylate cyclase (AC). Rises in levels of cAMP leads to the activation of protein
kinase A (PKA) by dissociating the regulatory (R) from the catalytic (C) subunits. The C subunits of PKA
passively translocate to the nucleus where they may phosphorylate CREB at Ser133. In the second pathway
growth factors (such as NGF or BDNF) bind to and activate a Trk receptor. This, in turn, activates Ras and
the downstream kinases Raf, MEK and ERK. Activated ERKs stimulate the activity of MSKs and RSKs
which may then phosphorylate CREB at Ser133. In the third pathway, intracellular increases in Ca2+ which
binds to calmodulin (CaM) which activates CaM kinases (CaMKII, CaMKIV, CaMKK) which may also
phosphorylate CREB at Ser133.
3) Growth Factors: CREB mediates gene expression in response to a wide variety of growth
factors, including nerve growth factor (NGF), fibroblast growth factor (FGF), epidermal growth
factor (EGF) and brain-derived neurotrophic factor (BDNF) (see Brandner, this book). Signaling is then mediated by a large number of growth-factor-induced kinases. For example, NGF
stimulation activates NGF receptors (tyrosine kinase receptor, Trk receptors) that stimulates
guanine-nucleotide release factors (GRFs) that activate Ras, a small G protein. Activated Ras,
in turn, stimulates the serine/threonine kinase, Raf, that triggers activation of MEK, and its
targets, the ERK 1 /2 members of the MAPK family.12 One downstream substrate of the Ras/
ERK pathway is a 90 kDa ribosomal S-6 kinase-2 (RSK-2). Upon activation, both ERKs and
RSKs translocate to the nucleus where they may phosphorylate CREB at Ser133.23,36,117
Just as phosphorylation of Ser133 seems to be critical for activation of CREB, dephosphorylation of this residue is important for inactivation of CREB. As with all other phosphoproteins, therefore, the level of CREB phosphorylation at Ser133 reflects a balance between the
oppositional actions of kinases and phosphatases, such as protein phosphatase 1 and 2 (PP-1
and PP-2).49 For example, dephosphorylation of CREB at Ser 133 may be initiated by the
activation of calcineurin (PP-2B) by the Ca2+-CaM pathway.10 The transcriptional activity of
phosphorylated CREB may also be actively suppressed by transcriptional repressors, such as
CREM α, β and γ isoforms or ICER.37,70,82
The complexity of the pathways upstream from CREB may permit tight, fine-tuned regulation of CRE-mediated transcription, allowing it to produce distinct patterns of gene expression in response to different patterns of stimulation. For example, CREB activation may be
moderated by phosphorylation events at sites other than Ser133 (e.g., Ser142 and/or Ser143),
and also indirectly by phosphorylation or dephosphorylation of other components of the transcription machinery that CREB interacts with (e.g., CBP, POL II etc).
CREB and Electrophysiological Studies of Long-Term Plasticity
The withdrawal of the gill—an external respiratory organ— in the marine mollusk Aplysia
can be produced by mechanical stimulation of either the siphon or mantle shelf. The reflex
serves a defensive purpose: the retraction of the gill protects it from potential damage. This
reflex exhibits a number of forms of plasticity. In particular, the sensitization of the withdrawal
reflex—that is its enhancement following noncontingent shock applied to the tail of the animal—has been instrumental in the identification of many of the cellular and molecular mechanisms mediating synaptic and behavioral plasticity. The persistence of the reflex sensitization is
related to the number of shocks applied to the tail: one shock produces a transient sensitization, lasting minutes, whereas 5 or more shocks produce a LTM lasting days or
longer.6,18,25,39,45,62,91 Long-term sensitization at the synaptic level can be studied in reduced
preparations containing the sensory-motor synapse: short- and long-term facilitation of this
synapse mediates the behavioral sensitization of the reflex.
The role of CREB in memory and plasticity has been studied in cocultured Aplysia sensory
and motor neurons.28 Injection of oligonucleotides with CRE sequences into cultured sensory
neurons blocks long-term facilitation (LTF).28 Presumably, these CRE-oligonucleotides act as
competitive antagonists, trapping the CREB proteins needed for the transcriptional activation
of genes that ultimately mediate LTF.4,61 Moreover, a similar injection of a reporter gene driven
by a CRE-containing promoter shows that stimuli that produce LTF also trigger CREB activation, while stimuli that do not produce LTF similarly do not trigger CREB activation.61
There are several CREB-like proteins in Aplysia. The CREB1 gene encodes three proteins
(ApCREB1a, ApCREB1b and ApCREB1c) by alternative splicing.7 The ApCREB1a shares
structural and functional homology with CREB transactivators in mammals, while ApCREB1b
resembles mammalian ICER, a repressor of CREB transcription. Injection of antibodies or
antisense against CREB1a blocks LTF (but not short-term facilitation) while injection of phosphorylated ApCREB1a protein alone induces LTF.7 Application of ApCREB1b blocks LTF
while decreasing ApCREB1b function lowers the threshold for producing LTF.7 ApCREB1c is
a cytoplasmic protein that lacks a nuclear localization signal. Injection of unphosphorylated
CREB1c followed by a single pulse of serotonin enhances STF and induces LTF. Therefore,
this cytoplasmic form of CREB may play an important role not only in the modulation of
CREB-mediated transcription necessary for LTF but also in STF.7 Aplysia CREB2 is structurally unrelated to Aplysia CREB1 but shares some homology with mouse ATF4.51 Decreasing
ApCREB2 function decreases the threshold for producing LTF.7 However, the precise mechanism underlying the effects that ApCREB2 exerts on LTF is unclear.
One neuron may participate in the storage of multiple memories. Therefore, activity-dependent changes must be synapse-specific so that the same neuron can encode multiple patterns of stimulation. Experiments using a single sensory neuron composed of two branches
that contact two spatially separated motor neurons show that local application of serotonin
onto a single synapse induces LTF that is specific to that branch.22,73 This branch-specific LTF
requires local protein synthesis (presumably at the synapse to be modified) as well as CREB
activation in the nucleus of the presynaptic neuron. Repeated application of serotonin onto the
cell body of the sensory neuron (rather than the branch) induces a transient, cell-wide LTF that
does not persist beyond 48 hours. This transient LTF is CREB-dependent, but is not accompanied by synaptic growth. A similar pattern of transient LTF and no synaptic growth is pro-
From Messengers to Molecules: Memories Are Made of These
duced by injection of phospho-CREB1 into the sensory neuron. In order for this transient LTF
to become stable and for synaptic growth to appear, a single pulse of serotonin at either synapse
is required. Thus, CREB-mediated transcription cooperatively induces synaptic changes in
concert with local stimulation by serotonin, representing a mechanism by which individual
synapses may be strengthened.
CREB and Memory in Drosophila
The molecular mechanisms underlying LTM have been successfully studied in Drosophila
(or fruit flies). Learning in flies has been studied using an associative olfactory conditioning
paradigm. Flies will learn to avoid a previously neutral odor that was paired with shock in favor
of another odor that was not paired with shock in a T-maze.112 Both forward and reverse
genetic approaches have been used to study the involvement of CREB in memory in Drosophila.112 Using a forward genetic approach, the progeny of flies that were treated with a mutagen were screened for learning and memory impairments. Two mutants identified by this
screen were subsequently determined to have disruptions in Ca2+/CaM-stimulated adenylate
cyclase (rutabaga) and in cAMP-specific phosphodiesterase (dunce), both key enzymes in the
regulation of intracellular levels of cAMP.16,71,112
Just as in other species, LTM (produced by multiple training trials) is dependent on protein
synthesis.113 Using a reverse genetics approach, Yin and colleagues120 showed that disrupting
CREB function in Drosophila blocks LTM produced by multiple training trials, suggesting
that protein synthesis required for LTM is mediated, at least in part, by CREB. They found
that transgenic over-expression of a CREB transcriptional repressor (dCREB2b) impairs LTM,
but not STM, in this task. The finding that STM is intact indicates that the over-expression of
this CREB repressor does not disrupt acquisition, and furthermore suggests that basic perceptual, motor, and motivational processes required for the task are intact in these flies.120
In species ranging from Aplysia to human, spaced training (training trials presented with
intervening rest intervals) is more effective than massed training (the same number of training
trials presented shorter intervening rest intervals) in producing LTM. The same is true in flies:
multiple spaced training produces maximal LTM, whereas the same number of trials presented
in a massed fashion produces strong STM but weak or no LTM. However, massed training
alone is sufficient to produce maximal LTM if a CREB activator (dCREB2a) is over-expressed
in transgenic flies prior to training. The over-expression of this CREB activator produces robust LTM following even just one training trial,121 perhaps the fly equivalent of ‘photographic’
or ‘flashbulb’ memory.119 Transgenic flies over-expressing a mutant activator, where Ser231
(similar to Ser133 of the mammalian CREB gene) was replaced by an Ala, do not show LTM
after one training trial, indicating that phosphorylation of CREB at this residue is required for
the enhancement of LTM.121 Together, these results show the importance of CREB in LTM
formation in Drosophila and, furthermore, suggest that CREB may be a rate-limiting component of this process.
CREB and LTM in Mammals
Targeted Disruption of CREB Function in a Mouse
The study of the role of CREB in mammalian memory was first made possible by the
generation of a mouse in which the CREB gene was disrupted. A neomycin resistance (neo)
gene was inserted into exon 2 of the CREB gene, which was believed to contain the translation
initiation site for all CREB isoforms.56 This neo insertion resulted in the loss of two main
isoforms of CREB (α and ∆). However, the translation of a previously unknown CREB isoform
(CREBβ) starts from exon 4. Therefore insertion of the neo gene into exon 2 did not disrupt
the translation of this isoform; rather, in these CREBα∆ mice, expression of the CREBβ isoform
is elevated.11 The expression levels of CREM activator (τ) and repressor (α and β) isoforms
were also increased in these mice. However, importantly, CREB-dependent transcription is
decreased in these CREBα∆ mutant mice.11 The homozygous deletion of all major CREB
isoforms (α, β and ∆; CREBnull) is lethal.96
Since the CREBα∆ mice were generated, they have been exhaustively characterized at the
behavioral level. Consistent with the effects of protein synthesis inhibition,2,13,99 CREBα∆ mice
exhibit normal STM but impaired LTM in several fear conditioning paradigms. For example,
CREBα∆ mice show normal conditioned freezing to both tone and context when tested shortly
(<1 hour) after training. However, both contextual and tone fear conditioning are impaired if
these mice are tested 24 hours after training.14,66 A similar pattern of results has been observed
using a different assay of conditioned fear—fear-potentiated startle.34
A parallel set of findings has been observed in studies examining two forms of social learning in CREBα∆ mice. Rodents develop a preference for foods recently smelled on the breath of
other rodents.15,41,42 Memory for this socially transmitted food preference is normal in CREBα∆
mice when tested immediately following training. However, just as in fear conditioning, CREBα∆
mice are impaired when tested 24 hours following training.42,66 The ability of rodents to remember conspecifics can be assessed in a social recognition task. Recognition is inferred from
a decrease in the amount of time spent investigating a familiar (vs. unfamiliar) conspecific.
Again, LTM, but not STM, for social recognition is disrupted in CREBα∆ mice.65 Together
with the fear conditioning data, these findings show that the CREBα∆ mutation specifically
affects LTM, and not STM, in a variety of behavioral paradigms with widely varying performance demands. The extent of these impairments is influenced by gene dosage.42 Further
disruption of CREB function can be achieved by combining the CREBα∆ and CREBnull mutations to produce mice carrying only a single allele for the CREBβ isoform (CREBcomb). Memory
impairments are more severe in these CREBcomb mice compared to the CREBα∆ mice carrying
two alleles for the CREBβ isoform.
Drawing an intriguing parallel with the fly experiments, Kogan et al66 showed that the
LTM deficits in the CREBα∆ mice were rescued by increasing the spacing between training
trials. This was true in three different forms of LTM: spatial (Morris water maze), contextual
(fear conditioning) and social (socially transmitted food preference). These parallels suggest
that the levels of activated CREB are rate-limiting for memory formation: The over-expression
of the CREB activator (dCREB2a) in the transgenic flies removes the requirement for spaced
training trials for LTM formation; Conversely, the reduced levels of CREB in the CREBα∆
mice necessitates multiple, spaced training, rather than fewer massed trials, to produced stable
One difficulty in the analysis of knockout mice in learning and memory studies is distinguishing between the effects of a given mutation on mnemonic vs. nonmnemonic processes.
This problem is largely circumvented in the CREBα∆ and related mice since these mice show
normal learning and STM. Therefore, compromising CREB function alone does not seem to
have nonspecific effects on sensory, motor and motivational processes required for the acquisition and expression of learning. Rather, compromising CREB function appears to specifically
affect the formation of LTM.
Gaining Temporal and Spatial Control of CREB Function
One of the problems with traditional knockout approaches is that the target protein is
deleted throughout development and in all tissues. For example, compensatory upregulation of
the CREBβ and CREM isoforms complicates the analysis and interpretation of the CREBα∆
mice. Therefore, achieving both spatial and temporal control over gene expression has been
one of the major goals, and three approaches have attempted to meet this challenge.
First, two studies have examined the effects of CREB antisense oligonucleotides on learning
and memory in rats. Guzowski and McGaugh48 examined acquisition in the hidden version of
the water maze following injections of antisense against CREB mRNA directly into the dorsal
hippocampus of rats. These injections disrupted acquisition in the hidden version of the water
From Messengers to Molecules: Memories Are Made of These
maze, a form of learning known to be dependent upon the hippocampus. Similar injections 2
days post-training had no effect on subsequent performance in the water maze, indicating that
decreasing CREB function does not affect the expression of a previously consolidated memory.
Acutely disrupting CREB function in the amygdala has also been shown to disrupt the development of a conditioned taste aversion.69 Injections of antisense directed against CREB blocked
long-term (3-5 days), but not short-term (2 hour), CTA memory. Sense control infusions, as
well as infusions of antisense into brain regions (basal ganglia) not critical for plasticity underlying CTA, had no effect.
Second, a transgenic line of mice has been developed that inducibly expresses a CREB
repressor (αCREBS133A).63 The induciblity of the system is produced by fusing the CREB
repressor to a ligand-binding domain (LBD) of a human estrogen receptor with a G521R
mutation (LBDG521R). The activity of this mutated LBD is regulated not by estrogen but by
the synthetic ligand, tamoxifen.27,35,72 In the absence of tamoxifen, the LBDG521R-CREBS133A
fusion protein is bound to heat shock proteins and is therefore inactive.35 However, administration of tamoxifen activates this inducible CREB repressor (CREBIR) fusion protein, allowing it to compete with endogenous CREB and disrupt CRE-mediated transcription. This mouse
has been used to dissect the role of CREB in potentially dissociable memory processes. By
administering tamoxifen to activate the repressor in CREBIR transgenic mice at key time points
in a fear conditioning protocol, the effects of acutely disrupting CREB function on 1) encoding or STM, 2) consolidation into LTM, 3) storage or maintenance, 4) retrieval were assessed.
CREB is crucial for the consolidation of long-term conditioned fear memories, but not for
encoding, storage or retrieval of these memories. While acute over-expression of a CREB repressor disrupts LTM, chronic over-expression of the same transgene throughout development
has much milder effects.93 The weaker effects associated with chronic over-expression of a
CREB repressor (compared to conditional over-expression of this transgene in the CREBIR
mice) may be due to upregulation or compensation through development. Alternatively, the
weaker phenotype might be due to a milder disruption of CREB function in these mice: for
example, transgene expression levels may not be sufficiently high to compete effectively with
A third approach has used viral vector-mediated gene transfer technology to manipulate
CREB levels.19-21 Josselyn et al60 used herpes simplex viral vector-mediated gene transfer technology to specifically increase CREB expression in the amygdala of rats. This method exploits
the natural ability of the herpes simplex virus to insert DNA into specific neuronal populations.103 These rats were fear-conditioned using massed training that normally only produces
STM but no or weak LTM for a light-shock pairing (Fig. 2). However, the over-expression of
CREB in the amygdala neurons now results in normal LTM. These data are consistent with
results in Drosophila showing that increasing CREB levels reduces the number of training trials
required to produce LTM, or overcomes the requirement for trial spacing to produce LTM.121
Detecting CREB Activation During Learning
Complementary to approaches demonstrating that disruption of CREB function blocks
the formation of LTM are those showing that CREB is activated following learning. These
studies are invaluable since they provide a powerful synergy between systems and molecular
approaches. They not only show that CREB-mediated transcription is critical for the formation of long-term memories, but they identify where and when these processes occur.
Activation of CREB leads to the transcription of genes with CRE sites in their promoter
regions. Transgenic mice with a β-galactosidase reporter construct under the regulation of a
CRE-containing promoter (CRE-LacZ) have been used to identify where in the brain learning-related CREB-mediated transcription occurs.58,59 Following fear conditioning significant
increases in CRE-dependent gene expression are observed in both the hippocampus and the
amygdala, consistent with the idea that plasticity in these structures is critical for learning
context-US and tone-US associations. In a clever control study Impey and colleagues showed
that CRE-dependent gene expression related to tone-US associations was limited to the amygdala
Figure 2. Effects of CREB over-expression in the amygdala on fear-potentiated startle. A) Massed training
produces weak LTM, as assessed by mean fear-potentiated startle difference scores (difference between mean
startle amplitude on light-tone (LT) trials from tone-alone (TA) trials). B) The same number of trials
presented in a spaced fashion produces robust LTM. C) Infusion of HSV-LacZ herpes simplex viral vectors
encoding LacZ (HSV-LacZ) into the basolateral amygdala produces high expression of β-galactosidase that
is restricted to the basolateral amygdala. D) A high-power image of the amygdala following infusion of
HSV-CREB showing over-expression of CREB that is restricted to the lateral nucleus of the amygdala. E)
Infusion of HSV-LacZ into the amygdala does not change the weak LTM normally induced by massed
training. F) Infusion of HSV-CREB into the amygdala prior to massed training enhances LTM.
using a latent inhibition protocol. To minimize the likelihood of the context becoming associated with shock, mice were pre-exposed to the training context for 12 hours prior to training.
These procedures produced significant increases in CRE-dependent gene expression in the
amygdala, but not the hippocampus. Consistent with this, when these mice were tested they
only showed conditioned freezing when re-exposed to the tone, but not the context.
From Messengers to Molecules: Memories Are Made of These
A second approach has been to use immunocytochemical procedures to detect
learning-induced changes in levels of phosphorylated CREB (pCREB). For example, levels of
pCREB are elevated in the olfactory bulbs following olfactory conditioning in neonate rat
pups.79 Consistent with the effects intra-amygdala infusions of CREB antisense oligonucleotides on the development of a conditioned taste aversion,69 increases in pCREB levels are
observed in the lateral nucleus of the amygdala following pairing of saccharin (CS) and
LiCl-induced illness (US). Similar increases are not observed if the rats are exposed to the CS
(saccharin) or US (LiCl) alone, indicating that activation of CREB is related to associative
Several studies have examined pCREB levels in fear-motivated learning paradigms. Inhibitory avoidance training, for example, induces phosphorylation of CREB in the CA1 and Dentate Gyrus regions of the hippocampus.9,17,88,108-110,114 These immunocytochemical data confirm similar findings using the CRE-reporter mouse.59 Contextual fear conditioning increases
pCREB levels in both the hippocampus and amygdala,105 again consistent with the observations of Impey.59
The contribution of these studies is that they show that CREB activation is restricted to the
brain regions that have been shown to critically mediate learning in each of these tasks. Furthermore, they allow us to characterize the time course of CREB activation following a learning event. Indeed, both contextual fear conditioning and inhibitory avoidance training appear
to produce two waves of CREB activation:8,105 pCREB levels are initially increased 0-30 minutes following training, and later 3-6 hours following training. These observations are consistent with the idea that LTM formation may involve multiple rounds of protein synthesis. For
example, protein synthesis inhibition immediately following, or 4 hours following training,
disrupts long-term contextual fear memories.13 It is speculated that these later waves may be
mediated by sustained PKA activity: In Aplysia CREB activation leads to the induction of a
number of immediate response genes, including carboxy-terminal ubiquitin hydrolase. This
hydrolase removes the regulatory subunit of PKA, allowing the kinase to become persistently
CREB and Reconsolidation
Two studies showed that either CRE-dependent gene expression59 or CREB activation105
are detected in the amygdala following fear conditioning training. A third study has shown that
pCREB levels are elevated in the amygdala following testing.52 Therefore retrieval, as well as
encoding, of fear memories initiates signaling cascades that culminate in CREB activation and
presumably gene expression. These findings support the idea that memories are dynamic and
modifiable entities.81,85-87,98 That is, memory retrieval may induce a state of plasticity in which
memories become labile before becoming stable again. The process of re-stabilization of the
trace, or reconsolidation, following retrieval has been shown to be protein-synthesis dependent.86 Consistent with the role for CREB in regulating gene expression required for initial
consolidation of memories, recent data supports the role for CREB in regulating gene expression required for reconsolidation, implied by the Hall52 study. Using the inducible CREB
repressor transgenic mice, Kida et al63 showed that acutely repressing CREB function following memory reactivation impairs the stability of memory. Although the exact molecular mechanisms mediating consolidation and reconsolidation may differ,108 CREB appears to be necessary for both.
Much effort, using a wide variety of tools, has been focused on identifying the molecular
mechanisms underlying learning and memory. Establishing that a particular molecule participates critically in these processes relies, it might be argued, on presentation of at least two types
of evidence.74,95 First, disruption of normal molecular function should interfere with memory
formation. Second, activation of the molecule, in a predictable, region-specific manner, should
be observed following learning. Reliance on evidence from one line of inquiry increases the
potential for false-negative and false-positive results.43,64 For example, targeted deletion of a
particular molecule may cause learning impairments not because that molecule directly participates in processes critical for plasticity; rather, the loss of that molecule may produce more
global disruption of cellular processes that indirectly impair the neuron’s ability to respond
appropriately to extracellular signals.
The case for the critical involvement of CREB in LTM is compelling since both types of
evidence have been brought to bear on the problem. That is, disrupting CREB function, be it
via the generation and testing of genetically-engineered mice or via the infusion of oligonucleotides, specifically disrupts LTM, but not learning. Secondly, studies using reporter mice or
immunocytochemical approaches, have shown that CREB is activated following learning in a
temporally- and region-restricted manner. In rodents, this conclusion is strengthened since
these observations are drawn from a wide variety of tasks, each with widely varying stimulus
properties and performance demands.
Similar manipulations of CREB function produce qualitatively similar effects in a wide
variety of species including Aplysia, Drosophila, Chasmagnathus crab, honey bees, and song
birds.1,3,57,68,84,92,97,102,119 In humans, it is particularly noteworthy that the cognitive disabilities in several disorders appear to be directly related to disruption of CREB-mediated transcription. Mutations in RSK2, a protein kinase that activates CREB by phosphorylation at
Ser133, are associated with Coffin-Lowry syndrome,111 as well as nonspecific mental retardation.80 For example, in tissue from Coffin-Lowry patients, reductions in RSK2-mediated CREB
phosphorylation (following EGF stimulation) are linearly related to severity of cognitive deficits.53 In addition, Rubinstein-Taybi syndrome, which is caused by a mutation of CBP—the
cofactor that is essential for transcriptional activation of CREB—is associated with mental
retardation.90 Consistent with this, mice that are heterozygous for CBP mutation exhibit learning
impairments.89 These studies, along with those from sea slugs and flies, mice and rats, suggest
an evolutionarily conserved role for CREB-transcription in role in long-term memory formation.
We thank Rui M. Costa for comments and discussion.
1. Abel T, Martin KC, Bartsch D et al. Memory suppressor genes: Inhibitory constraints on the
storage of long- term memory. Science 1998; 279:338-341.
2. Abel T, Nguyen PV, Barad M et al. Genetic demonstration of a role for PKA in the late phase of
LTP and in hippocampus-based long-term memory. Cell 1997; 885:615-26.
3. Alberini CM. Genes to remember. J Exp Biol 1999; 21:2887-2891.
4. Alberini CM, Ghirardi M, Metz R et al. C/EBP is an immediate-early gene required for the consolidation of long-term facilitation in Aplysia. Cell 1994; 76:1099-1114.
5. Bacskai BJ, Hochner B, Mahaut-Smith M et al. Spatially resolved dynamics of cAMP and protein
kinase A subunits in Aplysia sensory neurons. Science 1993; 105:222-226.
6. Bailey CH, Montarolo P, Chen M et al. Inhibitors of protein and RNA synthesis block structural
changes that accompany long-term heterosynaptic plasticity in Aplysia. Neuron 1992; 94:749-58.
7. Bartsch D, Ghirardi M, Skehel PA et al. Aplysia CREB2 represses long-term facilitation: Relief of
repression converts transient facilitation into long-term functional and structural change. Cell 1995;
8. Bernabeu R, Cammarota M, Izquierdo I et al. Involvement of hippocampal AMPA glutamate receptor changes and the cAMP/protein kinase A/CREB-P signalling pathway in memory consolidation of an avoidance task in rats. Braz J Med Biol Res 1997; 308:961-965.
9. Bevilaqua LR, Cammarota M, Paratcha G et al. Experience-dependent increase in cAMP-responsive element binding protein in synaptic and nonsynaptic mitochondria of the rat hippocampus.
Eur J Neurosci 1999; 11:3753-3756.
10. Bito H, Deisseroth K, Tsien RW. CREB phosphorylation and dephosphorylation: A Ca(2+)- and
stimulus duration-dependent switch for hippocampal gene expression. Cell 1996; 877:1203-1214.
From Messengers to Molecules: Memories Are Made of These
11. Blendy JA, Kaestner KH, Schmid W et al. Targeting of the CREB gene leads to up-regulation of
a novel CREB mRNA isoform. Embo J 1996; 155:1098-1106.
12. Blenis J, Chung J, Erikson E et al. Distinct mechanisms for the activation of the RSK kinases/
MAP2 kinase/pp90rsk and pp70-S6 kinase signaling systems are indicated by inhibition of protein
synthesis. Cell Growth Differ 1991; 26:279-285.
13. Bourtchouladze R, Abel T, Berman N et al. Different training procedures recruit either one or two
critical periods for contextual memory consolidation, each of which requires protein synthesis and
PKA. Learning Memory 1998; 5:365-374.
14. Bourtchouladze R, Frenguelli B, Blendy J et al. Deficient long-term memory in mice with a targeted mutation of the cAMP-responsive element-binding protein. Cell 1994; 791:59-68.
15. Bunsey M, Eichenbaum H. Selective damage to the hippocampal region blocks long-term retention of a natural and nonspatial stimulus-stimulus association. Hippocampus 1995; 56:546-556.
16. Byers D, Davis RL, Kiger Jr JA. Defect in cyclic AMP phosphodiesterase due to the dunce mutation of learning in Drosophila melanogaster. Nature 1981; 289:79-81.
17. Cammarota M, Bevilaqua LR, Ardenghi P et al. Learning-associated activation of nuclear MAPK,
CREB and Elk-1, along with Fos production, in the rat hippocampus after a one-trial avoidance
learning: Abolition by NMDA receptor blockade. Mol Brain Res 2000; 761:36-46.
18. Carew TJ, Castellucci VF, Kandel ER. An analysis of dishabituation and sensitization of the
gill-withdrawal reflex in Aplysia. Int J Neurosci 1971; 2:79-98.
19. Carlezon Jr WA, Boundy VA, Haile CN et al. Sensitization to morphine induced by viral-mediated gene transfer. Science 1997; 277:812-814.
20. Carlezon Jr WA, Nestler EJ, Neve RL. Herpes simplex virus-mediated gene transfer as a tool for
neuropsychiatric research. Crit Rev Neurobiol 2000; 141:47-67.
21. Carlezon Jr WA, Thome J, Olson VG et al. Regulation of cocaine reward by CREB. Science 1998;
22. Casadio A, Martin KC, Giustetto M et al. A transient, neuron-wide form of CREB-mediated
long-term facilitation can be stabilized at specific synapses by local protein synthesis. Cell 1999;
23. Chen RH, Sarnecki C, Blenis J. Nuclear localization and regulation of erk- and rsk-encoded protein kinases. Mol Cell Biol 1992; 123:915-927.
24. Chrivia JC, Kwok RP, Lamb N et al. Phosphorylated CREB binds specifically to the nuclear protein CBP. Nature 1993; 365:855-859.
25. Cleary LJ, Lee WL, Byrne JH. Cellular correlates of long-term sensitization in Aplysia. J Neurosci
26. Cole TJ, Copeland NG, Gilbert DJ et al. The mouse CREB (cAMP responsive element binding
protein) gene: Structure, promoter analysis, and chromosomal localization. Genomics 1992;
27. Danielian PS, White R, Hoare SA et al. Identification of residues in the estrogen receptor that
confer differential sensitivity to estrogen and hydroxytamoxifen. Mol Endocrinol 1993; 72:232-240.
28. Dash PK, Hochner B, Kandel ER. Injection of the cAMP-responsive element into the nucleus of
Aplysia sensory neurons blocks long-term facilitation. Nature 1990; 345:718-721.
29. Dash PK, Karl KA, Colicos MA et al. cAMP response element-binding protein is activated by
Ca2+/calmodulin- as well as cAMP-dependent protein kinase. Proc Natl Acad Sci USA 1991;
30. Davis HP, Squire LR. Protein synthesis and memory. Psychol Bull 1984; 6:518-559.
31. Deisseroth K, Tsien RW. Dynamic multiphosphorylation passwords for activity-dependent gene
expression. Neuron 2002; 34:179-182.
32. Dudai Y. Consolidation: Fragility on the road to the engram. Neuron 1996; 173:367-370.
33. Enslen H, Sun P, Brickey D et al. Characterization of Ca2+/calmodulin-dependent protein kinase
IV. Role in transcriptional regulation. J Biol Chem 1994; 269:15520-15537.
34. Falls WA, Kogan JH, Silva AJ et al. Fear-potentiated startle, but not prepulse inhibition of startle,
is impaired in CREBalphadelta-/- mutant mice. Behav Neurosci 2000; 1145:998-1004.
35. Feil R, Brocard J, Mascrez B et al. Ligand-activated site specific recombination in mice. Proc Natl
Acad Sci USA 1996; 93:10887-10890.
36. Finkbeiner S, Tavazoie SF, Maloratsky A et al. CREB: A major mediator of neuronal neurotrophin
responses. Neuron 1997; 195:1031-1047.
37. Foulkes NS, Borrelli E, Sassone-Corsi P. CREM gene: Use of alternative DNA-binding domains
generates multiple antagonists of cAMP-induced transcription. Cell 1991; 644:739-749.
38. Foulkes NS, Mellstrom B, Benusiglio E et al. Developmental switch of CREM function during
spermatogenesis: From antagonist to activator. Nature 1992; 355:80-84.
39. Frost WN, Castelluci VG, Hawkins RD et al. The monosynaptic connections made by the sensory
neurons of the gill– and siphon–withdrawal reflex participate in the storage of long–term memory
for sensitization. Proc Nat Acad Sci USA 1985; 82:8266.
40. Galef Jr BG, Wigmore SW. Transfer of information concerning distant foods: A laboratory investigation of the ‘information-centre’ hypothesis. Anim Behav 1983; 31:748-758.
41. Galef Jr BG, Mason JR, Preti G et al. Carbon disulfide: A semiochemical mediating socially-induced
diet choice in rats. Physiol Behav 1988; 42:119-124.
42. Gass P, Wolfer DP, Balschun D et al. Deficits in memory tasks of mice with CREB mutations
depend on gene dosage. Learning Memory 1998; 5:274-288.
43. Gerlai R. Gene targeting: Technical confounds and potential solutions in behavioral brain research.
Behav Brain Res 2001; 125:13-21.
44. Gilman AG. G proteins: Transducers of receptor-generated signals. Annu Rev Biochem 1987;
45. Goelet P, Castellucci VF, Schacher S et al. The long and short of long-term memory - a molecular
framework. Nature 1986; 322:419-422.
46. Gonzalez GA, Montminy MR. Cyclic AMP stimulates somatostatin gene transcription by phosphorylation of CREB at serine 133. Cell 1989; 594:675-680.
47. Gonzalez GA, Yamamoto KK, Fischer WH et al. A cluster of phosphorylation sites on the cyclic
AMP-regulated nuclear factor CREB predicted by its sequence. Nature 1989; 337:749-752.
48. Guzowski JF, McGaugh JL. Antisense oligodeoxynucleotide-mediated disruption of hippocampal
cAMP response element binding protein levels impairs consolidation of memory for water maze
training. Proc Natl Acad Sci USA 1997; 946:2693-2698.
49. Hagiwara M, Alberts A, Brindle P et al. Transcriptional attenuation following cAMP induction
requires PP-1-mediated dephosphorylation of CREB. Cell 1992; 701:105-113.
50. Hagiwara M, Brindle P, Harootunian A et al. Coupling of hormonal stimulation and transcription
via the cyclic AMP-responsive factor CREB is rate limited by nuclear entry of protein kinase A.
Mol Cell Biol 1993; 138:4852-4859.
51. Hai T, Liu F, Coukos W et al. Transcription factor ATF cDNA clones: An extensive family of
leucine zipper proteins able to selectively form DNA-binding heterodimers. Genes and Dev 1989;
52. Hall J, Thomas KL, Everitt BJ. Fear memory retrieval induces CREB phosphorylation and Fos
expression within the amygdala. Eur J Neurosci 2001; 137:1453-1458.
53. Harum KH, Alemi L, Johnston MV. Cognitive impairment in Coffin-Lowry syndrome correlates
with reduced RSK2 activation. Neurology 2001; 562:207-214.
54. Hoeffler JP, Meyer TE, Waeber G et al. Multiple adenosine 3',5'-cyclic monophosphate response
element DNA-binding proteins generated by gene diversification and alternative exon splicing. Mol
Endocrinol 1990; 46:920-930.
55. Hoeffler JP, Meyer TE, Yun Y et al. Cyclic AMP-responsive DNA-binding protein: Structure based
on a cloned placental cDNA. Science 1988; 242:1430-1433.
56. Hummler E, Cole TJ, Blendy JA et al. Targeted mutation of the cAMP response element binding
protein (CREB) gene: Compensation within the CREB/ATF family of transcription factors. Proc
Natl Acad Sci USA 1994; 91:5647-5651.
57. Impey S, Goodman RH. CREB signaling—timing is everything. Sci STKE 2001; 82:E1.
58. Impey S, Mark M, Villacres EC et al. Induction of CRE-mediated gene expression by stimuli that
generate long-lasting LTP in area CA1 of the hippocampus. Neuron 1996; 165:973-982.
59. Impey S, Smith DM, Obrietan K et al. Stimulation of cAMP response element (CRE)-mediated
transcription during contextual learning. Nature Neurosci 1998; 1:595–601.
60. Josselyn SA, Shi C, Carlezon Jr WA et al. Long-term memory is facilitated by cAMP response
element-binding protein overexpression in the amygdala. J Neurosci 2001; 21:2404-2412.
61. Kaang BK, Kandel ER, Grant SG. Activation of cAMP-responsive genes by stimuli that produce
long-term facilitation in Aplysia sensory neurons. Neuron 1993; 103:427-435.
62. Kandel ER, Klein M, Bailey CH et al. Serotonin, cyclic AMP, and the modulation of the calcium
current during behavioral arousal. In: Jacobs BL, Gelperin A, eds. Serotonin Neurotransmission
and Behavior. Cambridge, MA: MIT Press, 1981:211-254.
63. Kida S, Josselyn SA, de Ortiz SP et al. CREB required for the stability of new and reactivated fear
memories. Nature Neurosci 2002; 5:348-355.
64. Kim JJ, Baxter MG. Multiple brain-memory systems: The whole does not equal the sum of its
parts. Trends Neurosci 2001; 246:324-330.
65. Kogan JH, Frankland PW, Silva AJ. Long-term memory underlying hippocampus-dependent social
recognition in mice. Hippocampus 2000; 10:47-56.