| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vol. 7, No. 2, pp. 73-84, March/April 2000
Neuromodulation and Cognitive Processes, Institut des Neurosciences, Centre National de la Recherche Scientifique (CNRS) Unité Mixte de Recherche (UMR) 7624, 75005 Paris, France
A permanently existing "idea" which makes its appearance
before the footlights of consciousness at periodical intervals is as
mythological an entity as the Jack of Spades.
William James (1890)
Memory lends itself to study through its retrieval whether it is
evaluated by the behavior of a mouse in a swimming
pool, a verbal report from a human subject, or inferred from an
electrophysiological event. As William James so aptly pointed out,
"the only proof of there being retention is that recall actually
takes place." (1892). Such a view of memory as remembering is well
elaborated in the theoretical reflections of Bergson (1896) Although some memory retrieval is likely to occur spontaneously as a
result of random fluctuations of patterns of neuronal activity,
retrieval is usually brought about as a result of integration of
incoming environmental information with the "memory network" driven
by that information (Tulving and Thomson 1973 The theoretical emphasis on memory reactivation and reconsolidation
made here raises the issue, as yet ill addressed by neurobiological experiments, of factors that control or modulate these processes. We
first review the literature dealing with the neurobiological factors
that are involved in the actual retrieval process, a literature that
was, until recently, relatively sparse. The advent of noninvasive imaging technology applicable to human studies has, however, awakened interest in this topic and has resulted in a proliferation of studies
directly dealing with this aspect of memory function. These will be briefly
discussed. The second part of this review deals with recent neurobiological
evidence for reconsolidation of memories after their reactivation.
The Consolidation Hypothesis and its Origins
Nineteenth century clinical studies of retrograde amnesia after
cerebral trauma led Ribot (1882) Retrieval Facilitation After Experimental Amnesia
A caveat for interpretation of results of experiments within this
paradigm came from a series of experiments showing that after amnestic
treatments, retrieval of memory could be achieved by exposing rats,
right before the retention test, to cues associated with the original
training. This was demonstrated in many studies showing recovery from
ECS, hypoxia, or other experimentally induced retrograde amnesias.
Effective "reminders" included a weak foot shock, exposure to the
training context, or a combination of the training context and the foot
shock (Lewis et al. 1968 Memory Retrieval Facilitation After Forgetting: Contextual Cue Reminders
Memory can be viewed as a "multidimensional conglomerate of
attributes" (Spear 1974
INTRODUCTION
TOP
INTRODUCTION
RETRIEVAL
REACTIVATION AND...
CONCLUSION AND PERSPECTIVES
REFERENCES
, in the
seminal studies of Bartlett (1932) and later in those of Tulving
(Tulving and Thomson 1973) and Craik (1983), who argue, after Bergson, that remembering is an activity similar to perceiving, in the sense
that it involves the apprehension and comprehension of contemporary stimuli in the light of past experience.
). It follows from this
that retrieval will lead to the formation of new memories made on the
background of a retrieved prior experience. Therefore, it is
inconceivable that new memory can be acquired independently of
retrieval of past experience, in that it is memory of the past, that
organizes and provides meaning to the present perceptual experience.
Borrowing Tulving's terminology, new episodic memory, to be remembered
in a meaningful way, must be consolidated within a preexisting semantic
memory. This analysis does not draw a clear demarcation between
consolidation and retrieval processes and in this view, it can be
assumed that every retrieval operation should trigger a reconsolidation
process (Spear and Mueller 1984). Moreover, decoding or retrieval will
change the information content of the "trace" such that memory can
be viewed from a neurobiological point of view as an emergent, dynamic,
adaptive property of the nervous system.
RETRIEVAL
TOP
INTRODUCTION
RETRIEVAL
REACTIVATION AND...
CONCLUSION AND PERSPECTIVES
REFERENCES
to formulate the "Loi de Regression," which held that those events experienced immediately before the trauma were the most likely to be forgotten. These and later
clinical observations can be considered as the foundation of the
consolidation hypothesis, which holds that memories are made after the
initial experience, existing initially in a fragile form and
strengthened over time, becoming less and less vulnerable to
interference. From this hypothesis emerged the prevailing paradigm for
the study of brain mechanisms of memory for many years-experimentally induced retrograde amnesia in rodents. The protocol usually used a one
trial avoidance task followed by a post-training amnestic treatment at
various intervals after the training trial. Treatments effective in
inducing amnesia include electroconvulsive shock (ECS), hypoxia,
hypothermia, inhibitors of protein synthesis, and various other drugs.
The common feature of these diverse insults to the brain lies in their
temporal gradient of efficacy in inducing amnesia. For the most part,
the efficacy of the treatment depends on temporal contiguity to the
training episode; the shorter the interval between the training and the
treatment, the greater the amnestic effect. Time-dependent impunity to
these amnestic treatments was taken as evidence that the memory was now in a
fixed, consolidated, stable state (McGaugh 1966; Squire and Alvarez 1995).
; Quartermain et al. 1972; Miller and Springer
1972; Sara 1973; DeVietti and Hopfer 1974; Sara et al. 1975; Gordon and
Mowrer 1980). Small doses of analeptic drugs, such as strychnine,
amphetamine, or piracetam, when administered before the retention test,
also were shown to reverse ECS or hypoxia-induced amnesia, presumably
by acting directly on retrieval processes (Sara and David-Remacle 1974;
Sara and Remacle 1977). A series of important papers by Warrington and
Weiskrantz, around the same time, showed that human amnestics could
express normal memory performance if they were cued before the
retention test. They suggested that at least some forms of amnesia are
due to retrieval dysfunction rather than failure to consolidate
memories (Weiskrantz 1966; Warrington and Weiskrantz 1970). These
observations, reinforced by a strong conceptual framework provided by
Spear (1973), encouraged further studies of retrieval in animals.
, p. 56) among which are included both external and endogenous context. The simple passage of time may weaken
memories in the sense that they become less readily accessible, less
likely to be expressed at retention test. It is well known, however,
that the behavioral expression of forgetting or retention may be
altered by manipulation of the context before or during the retention
test (for review; see Spear 1974). We developed an animal model of
"spontaneous forgetting" in which rats trained to run in a six-unit
place discrimination maze for food reward showed a reliable retention
deficit when tested 3 weeks after training. The rapid acquisition (five
single daily trials) is probably the key to this memory deficit over
time. Forgetting could be alleviated by pretest "reminders", with
timing being a crucial factor in determining the efficacy of the
reminder treatment. Rats exposed for 1-2 min to the context in which
the learning had taken place made fewer errors than rats placed in a
neutral environment before the retention test. The context reminder had to be given immediately before the test; rats reminded 1 hr before the
test did not show this facilitation (Deweer et al. 1980; Deweer and
Sara 1984).
View larger version (8K):
[in a new window]
Figure 1:
Facilitation of retrieval by a contextual cue reminder before the
retention test. (Left) Number of errors at each daily trial
during acquisition; (right) retention performance of rats
tested 3 weeks after the last training trial. ( 
) Control rats
habituated to the contextual reminder by exposure to it every day
during the 3-week retention interval; (
) reminded rats, presented
with the contextual reminder once, just before the retention test.
Nonreminded control rats make significantly more errors than at the
last training trial and significantly more errors than rats that are
exposed to the context for 90 sec before the test. The contextual cue
reminder alleviates forgetting and the performance of reminded rats is
not different from that of the last training trial (adapted from Deweer
et al. 1980).
Pretest exposure to the experimental context in which discriminative
avoidance training had taken place also alleviates forgetting. These
experiments compared the effectiveness of the contextual cue and the
conditioned stimulus (CS) as reminders and found that pretest priming
with the CS facilitated performance at short training to test
intervals, whereas the contextual cue was only effective after a long
retention interval, when control animals showed considerable forgetting
(Gisquet-Verrier and Alexinsky 1986; Gisquet-Verrier et al. 1989).
State-Dependent Retrieval
Closely related to facilitation of memory retrieval by contextual
cue reminders is the phenomenon of "state dependent" retrieval. Here the endogenous context, or physiological state of the organism, including neurohumoral and hormonal state, is supposedly incorporated into the conglomerate of memory attributes and exerts control over
retrieval of the memory (Spear 1974; Izquierdo 1984). The phenomenon is
easily demonstrated using pharmacological manipulation with amphetamine
or barbiturates (Overton 1974) or opioids (Bruins Slot and Colpaert
1999). Animals trained with the drug and tested without show poor
retention, while those receiving drug treatment before both training
and test, show good retention.
Pretest treatment with hormones released during stress, such as
adrenocorticotrophic hormone (ACTH) (Mactutus et al. 1980), opioids
(Izquierdo 1984), epinephrine (Izquierdo and McGaugh 1987), and
vasopressin (Sara et al. 1982; Almedia and Izquierdo 1984) has been
shown to effectively reinstate memory for aversive events, both after
experimental amnesia and in normal forgetting. A widely accepted
interpretation of these results is that the hormone treatment reinstates the internal context of training that then facilitates access to the target memory, in much the same way that an exogenous contextual cue does (see Spear 1974; Riccio and Concannon 1981; McGaugh
1983; Izquierdo 1984).
Pharmacological Facilitation of Retrieval of "Forgotten" Memories
There are relatively few pharmacological studies of direct effects
on memory retrieval. Drugs facilitating retrieval when injected before
the retention test include strychnine (Gordon and Spear 1973; Sara and
Remacle 1977), cocaine (Rodriguez et al. 1993), nootropic drugs (Sara
and David-Remacle 1974; Sara et al. 1979; Sara, 1980) nicotine (Faiman
et al. 1992; Zarrindast et al. 1996), and glucose (Manning et al. 1998). There
are reports of vasopressin facilitation of retrieval and the effects appear to
be mediated through nicotinic receptors (Faiman et al. 1992).
Amphetamine, a drug with multiple central and peripheral actions, among
which include enhancement of release of both dopamine and noradrenaline
(NA), facilitated retrieval of the forgotten maze task, when the
injection was made before the retention test, 3 weeks after training.
The effect was specific to retrieval after a forgetting interval; there
was no effect on when it was given before or after acquisition trials
(Sara and Deweer 1982). Retrieval of a forgotten conditioned emotional
response was facilitated by pretest treatment with amphetamine as well,
when the animals were treated just before the retention test (Sara
1984; Quartermain et al. 1988). It is important to note that in none of
these experiments was there any evidence of state dependency, although this
had been reported in several earlier studies (for review, see Overton 1974)
It is noteworthy that those drugs reported to directly facilitate
retrieval after experimentally induced amnesia or spontaneous forgetting share the common action of increasing arousal or vigilance, even if by different mechanisms. Further evidence for the importance of
arousal in memory retrieval processes comes from experiments in which
low level electrical stimulation of the mesencephalic reticular
formation (MRF) just before the test alleviated forgetting in the same
maze task (Sara et al. 1980). Subsequent experiments indicated that the
MRF stimulation-induced increase in arousal alone did not facilitate
retrieval; the memory had first to be "primed" or reactivated by
exposure to the context in which the training had taken place.
Moreover, the effectiveness of the contextual cue reminder was
potentiated by concurrent stimulation of the MRF (Dekeyne et al. 1987).
Memory Retrieval and the Noradrenergic System
Later studies, using the same maze forgetting paradigm, implicated
the noradrenergic system, by showing facilitation of retrieval with
pretest injection of the 
2 receptor antagonist yohimbine (Sara
1985) or the more specific antagonist idazoxan (Sara and Devauges
1989b). Both of these drugs increase firing of the noradrenergic neurons in the locus coeruleus (LC) and increase release of NA from
terminals in the forebrain target regions by antagonistic action on
inhibitory autoreceptors. It should be noted, however, that these
systemic injections of idazoxan cause a peripherally mediated increase
in blood pressure (V. Devauges and S.J. Sara, unpubl.), which would
result in an increase in cerebral blood flow and could account for
cognitive facilitation, independent of effects on the central
noradrenergic system. Subsequent studies, however, lent further support
for the suggestion that the noradrenergic system mediates contextual
cue reminder induced retrieval facilitation. In chronically implanted
rats, electrical stimulation of the noradrenergic nucleus locus
coeruleus just before the retention test facilitated retrieval in the
maze-forgetting paradigm, as illustrated in Figure 2
(Sara and Devauges 1989a). The facilitation was
blocked by prior systemic injection of the 
-adrenergic receptor
antagonist propranolol (Devauges and Sara 1991).
|
Neuroanatomical Loci of Retrieval Animal Studies
Based on a review of the early literature on hippocampal lesions in
rats, Hirsh (1974) attempted to provide a unifying theory of
hippocampal function, calling the hippocampus the gateway to memory as
part of a system mediating contextual retrieval. More than 20 years
later a direct test of this proposition was unable to provide
supporting evidence. Discrete lesions of the hippocampus did not
prevent the facilitation of memory retrieval by pretest presentation of
either the CS or the experimental context (Gisquet-Verrier and Schenk
1994). On the other hand, recent studies using reversible functional
inactivation of the dorsal hippocampus have implicated this structure
in memory retrieval processes. Local injection of tetrodotoxin impaired
retrieval of a passive avoidance response, when the injections were
made 1 hr before the retention test (Ambrogi Lorenzini et al. 1996). In
another series of studies (Moser and Moser 1998) it was shown that
partial inactivation of the hippocampus by local infusion of the GABA
agonist muscimol temporarily impaired retrieval of spatial memory, but
did not affect new learning. These studies showed that although spatial
information can be acquired and retained with only a small number of
local ensembles of neurons in the hippocampus, retrieval of spatial
information depends on a widely distributed network and requires the
integrity of at least 70% of the dorsal hippocampus. Furthermore, it
appears that the integrity of AMPA/kainate receptors
within the hippocampus is necessary for retrieval (Riedel et al. 1999),
whereas the NMDA receptors are involved only in the encoding, but not
retrieval of spatial information (Steele and Morris 1999).
Retrieval in Humans: Functional Imaging
Functional imaging technology is beginning to reveal brain areas
specifically engaged during retrieval, and this will undoubtedly prove
to be a powerful tool in the future to address this question. Already
it has provided some clues to the engagement of particular anatomical
regions of the brain in different aspects of the retrieval process. For
example, Tulving's studies suggest that during effortful retrieval,
the right frontal cortex be activated, whereas the hippocampus is
engaged when the retrieved memory is recognized as such (Calabrese et
al. 1996). In fact, there is a growing consensus from both positron
emission tomography (PET) and magnetic resonance imaging (MRI) studies
that the right frontal cortex is selectively engaged during retrieval
attempt (Nyberg et al. 1996a; Fletcher et al. 1998; Wagner et al.
1998). Although some attempts are being made to analyze networks using
these techniques (Nyberg et al. 1996b), these approaches thus far can
only suggest the gross anatomy of regions that show changes in
metabolic activity during memory retrieval (for review, see Cabeza and
Nyberg 1997). Information concerning dynamics of implicated networks
can, for the time being, be best provided by invasive techniques of
recording neuronal activity from multiple electrode sites.
Understanding at a cellular level will require other in vivo and ex
vivo techniques based on well-validated animal models.
Contextual Cue Reminders, Retrieval, and the Truncated Conditioned Reflex
Although the anatomical studies from both rats and humans implicate
specific structures in memory retrieval operations, namely hippocampus
and frontal cortex, the prevailing view is that memories are widely
distributed in the brain, and that specific information is actually
stored in sensory cortices. Retrieval must somehow involve initial
activation of relevant intrinsic networks, selection of relevant
extrinsic stimuli, and integration of these different sources of
information into a meaningful trace. From subjective experience we know
that memory retrieval takes time
it may be a matter of milliseconds,
but can extend to minutes or more. Retrieval can occur spontaneously,
but it can be the fruit of great effort as well. The role of subtle,
but significant, environmental stimuli in triggering these processes is
intuitively obvious, and has been investigated systematically in
animals and humans. Nevertheless, virtually nothing is known about the
physiological processes underlying the act of remembering. The initial
process must involve some orientation of attention to a particular
stimulus or ensemble of stimuli. How those particular stimuli are
recognized as "meaningful" or how they can activate the specific
distributed network presumed to be the neuronal substrate of the memory
still remains unknown.
Formal experiments in the rat, some of which have been described above,
have consistently demonstrated the efficacy of exposure to the
experiment context in improving memory performance after a long
training to test interval, where nonreminded rats show forgetting.
Speculatively, this could be the equivalent to the déjà vu
phenomenon that we have all experienced. We walk into a room and have
an immediate sensation of familiarity, without being able to evoke a
particular episodic memory associated with the context. We experience
an increase in arousal and attention, and initiate a search for cues or
relevant stimuli within the context to facilitate retrieval of the
target memory. That operation can take several seconds or minutes or
even more before "ecphory" (Tulving and Markowitsch 1997).
The contextual cue reminder may act to facilitate memory in rats in a
similar way. The context elicits a conditioned arousal response, which
then facilitates brain mechanisms underlying retrieval. Pavlov believed
that cortical activation was regulated through conditioning and his
pupil Kupalov formalized this idea, under the name of truncated
conditioned reflex (Kupalov 1961). In this analysis, the experimental
context, because of its regular association with the reinforcement,
comes to elicit a nonspecific conditioned responsean increase in
cortical tonus (Sara 1985, 1991). Konorski (1967) later developed this
idea, referring to it as the preparatory response and assigning it a
major role in the conditioning process. For the present analysis such a
mechanism could account for the action of contextual cue reminders in
facilitating retrieval. The context, which had been associated with the
reinforcement during learning, acts as a CS to elicit an arousal
response. Such a response could involve activation of multiple
peripheral and central mechanismsthe adrenal-pituitary axis, neurons
of the MRF, locus coeruleus, and other brainstem neuromodulating systems.
The question now is to what extent the context actually does elicit
firing of these neurons during retrieval. Experiments from our
laboratory have shown that LC cells do respond vigorously to
information about the context in which are embedded the conditioned stimuli (CS+ and CS
), during discriminative conditioning (Sara and
Segal 1991; see Fig. 3 for details). It still remains
to be demonstrated that LC neurons fire when the animal is exposed to the context in which learning took place, after a long retention interval.
|
A schematic diagram summarizing the relationship between context and
retrieval is provided in Figure 4. In accordance with Kupalov, the experimental context, because of its association with the
reinforcement, comes to elicit a conditioned response that includes
firing of LC neurons. This would result in release of NA in the forebrain.
|
There is an extensive literature on postsynaptic effects of NA in
sensory pathways, with several investigators suggesting that NA
increases signal/noise by inhibiting background neuronal firing while sparing evoked activity (Foote et al. 1975; Waterhouse and
Woodward 1980; Hasselmo et al. 1997) or tuning sensory responses by
narrowing the receptive field of sensory neurons (Waterhouse et al.
1990; Manunta and Edeline 1999). Such actions would serve to enhance
perceptual acuity. If memory retrieval is intimately related to
perceptual processes as we suggested in the introduction, then this
action of NA or other neuromodulators could facilitate retrieval by
orienting attention, and gating and tuning responses to sensory stimuli
(Sara 1985, 1991).
| |
REACTIVATION AND RECONSOLIDATION |
|---|
|
TOP
INTRODUCTION
RETRIEVAL
REACTIVATION AND...
CONCLUSION AND PERSPECTIVES
REFERENCES
|
|---|
Reactivation and Experimental Amnesia
Embedded in the extensive literature on experimental amnesia and
memory consolidation were several iconoclastic papers representing the
work of just three or four laboratories, showing that a temporally graded retrograde amnesia could be obtained for a memory that was
reactivated or retrieved just before the amnestic treatment. If the
rat, well trained to a specific task and thus having a well-consolidated memory, is exposed to part of the learning
environmentusually the reinforcement or a contextual cueand this is
followed by the amnestic treatment, then the animal shows amnesia for
the task on a subsequent retention test. This was first demonstrated after reactivation of passive avoidance training followed by ECS (Misanin et al. 1968). This same group later showed that the phenomenon not to be limited to memories forfoot shock after single trial avoidance training. After extensive training in a complex maze task,
memory was reactivated by exposure to the start box and the click of
the opening of its door (the start box alone was not a sufficient cue).
When this was followed by the amnestic agent, ECS, amnesia was obtained
(Lewis et al. 1972; Lewis and Bregman 1973). Numerous control
procedures assured that the specific cues associated with the original
learning were essential to the effect and not merely a reinstatement of
an emotional or motivational state, as had been suggested by other
investigators (Misanin et al. 1968; Robbins and Meyer 1970).
Hypothermia-induced amnesia for a well-trained task was obtained after
exposure of the rat to either the unconditioned stimulus (UCS) (foot
shock) or the context and UCS, immediately before the amnestic
treatment (Mactutus et al. 1979; Richardson et al. 1982). These
researchers even found that reactivated memories were more susceptible
to the hypothermic treatment than newly acquired memories, in that less
cooling was required to obtain amnesia after reactivation. Finally,
inhibition of protein synthesis produces amnesia for a well
consolidated memory in mice, provided that the memory is reactivated by
presentation of the CS before the drug treatment (Judge and Quartermain 1982).
Reactivation and Memory Facilitation
Reactivated memory is not only vulnerable to amnestic agents, but it
can be facilitated by treatments that enhance memory consolidation.
Experiments by DeVietti et al. (1977) demonstrated that electrical
stimulation of the MRF, which improves memory consolidation when
administered within a short time after acquisition, improved memory for
a single trial-conditioned fear response in rats when it was applied
after memory reactivation and the rat was tested 24 hr later. The
shorter the interval between the reactivation and the stimulation, the
better the memory enhancement, the temporal gradient of efficacy being
quite similar to the postacquisition gradient.
Although of great theoretical and clinical importance, these
reactivation studies did not receive the attention that they merited at
the time, perhaps because the experimental amnesia paradigm was more or
less abandoned with the discovery of long-term potentiation (LTP),
which became the prevailing paradigm for memory research in the 1980s
and 1990s. Nevertheless, the results clearly indicate that it is not
necessarily the newness of the memory which determines its lability,
but whether it is active or inactive at the time of treatment. A recent
series of studies has lent support to this view, showing that both
glucose and fructose enhance memory, not only for newly acquired
memories, but also for reactivated passive avoidance training, in a
dose- and time-dependent manner. For the treatment to be effective, it
must be given within 30 min of the reactivation experience, a reminder
foot shock (Horne et al. 1997; Rodriguez et al. 1999).
Thus, memories exist in an active state where they are labile and
susceptible to disruption by amnestic agents or enhancement by memory
modulators, and in an inactive or dormant state during which they are
resistant to amnestic brain insults or memory enhancing treatments. In
fact, recent results suggest that even the lowly terrestrial slug
Limax flavus demonstrates dynamic memory reorganization after
reactivation. Hypothermia can induce amnesia for an odor aversion in
this species, if it is applied within 1 min after acquisition or 1 min
after the well-established memory is reactivated by a brief exposure to
the carrot conditioned stimulus (Yamada et al. 1992; Sekiguchi et al. 1997).
Memory Reactivation and Consolidation During Sleep
That a dream is the subjective experience of the brain reprocessing
information acquired during the waking state is a compelling idea and
has led to much speculation over the years concerning the relationship
between sleep and memory (see review chapters in Fishbein 1981).
Convincing experimental evidence supporting the hypothesis that memory
is further processed during sleep episodes subsequent to learning has
accumulated, with early studies showing a spontaneous increase in the
rapid eye movement (REM) phase of sleep subsequent to learning (Leconte
and Hennevin 1971; Hennevin and Leconte 1977) and amnesia when animals
or human subjects are deprived of REM sleep after learning (Fishbein
1970; Fishbein and Gutwein 1977; for review, see Hennevin et al. 1995;
also Smith 1985). Spear and Gordon (1981) proposed a conceptual
framework for information processing during sleep, suggesting that
memories are reactivated, particularly during the REM phase, and it is this active state of memory that allows further processing or reinforcing of the underlying neural circuits in much the same way that
reactivated memories can be impaired or improved in wakefulness (see
preceding section). Hennevin and her colleagues have provided strong
experimental evidence for reactivation during sleep. They showed that
very low level electrical stimulation of the MRF that facilitates
memory when applied immediately after acquisition or immediately after
reactivation also facilitates memory when applied during REM sleep
(Hennevin et al. 1989). Using the rationale that, if memory
consolidation does occur during sleep that follows learning, then it
could be reinforced by reactivating specific circuits related to the
learning, a previously learned CS was presented as a reminder during
REM sleep. Rats undergoing this reactivation treatment showed better
memory when tested in subsequent wakefulness (Hars et al. 1985;
Hennevin and Hars 1985). A physiological substrate for the
interpretation of this reactivation experiment has been provided by
more recent experiments showing that brain structures comprising a
circuit involved in the initial learning maintain and express
plasticity during REM sleep. Conditioned responses to tone paired with
shock during the awake state can be expressed in both the hippocampus
and the medial geniculate nucleus during sleep (Hennevin et al. 1993).
Conditioned neuronal responses in the amygdala, which probably mediates
the affective component of the memory, are likewise elicited by the CS
during REM sleep (Hennevin et al. 1998).
There is some evidence that postacquisition information processing
occurs in the hippocampus during slow wave sleep (SWS). Pavlides and
Winson (1989) reported that particular hippocampal place cells that
fired during exploratory behavior were selectively more active during
the SWS episode following this behavior. Recording from multiple
hippocampal place cells, McNaughton and colleagues have shown that
neurons activated together by a behavioral experience during the awake
state tend to fire together during subsequent SWS episodes, as revealed
by cross correlogram analysis (Wilson and McNaughton 1994; Skaggs and
McNaughton 1996; Kudrimoti et al. 1999). They suggest that this
correlated activity is the result of a reactivation during sleep, which
serves to reinforce a neuronal ensemble representing the memory of the
behavioral experience. Buzsaki (1989, 1998) has proposed a similar
hypothesis concerning memory consolidation occurring during SWS based
on observations that postlearning sleep episodes are characterized by
an increase in sharp waves or bursting activity in the CA3 region of
the hippocampus, which could be reinforcing synapses activated during
learning. It is not clear from these analyses what factors determine
which ensembles are reactivated during sleep and consequently, how
representations are selected for further consolidation.
In the case of Hennevin's experiments, which have included complex
maze learning, discriminative avoidance learning and associative conditioning, REM sleep seems to be the important phase for memory processing. In the hippocampal recording studies, where focus has been
on neurons associated with spatial information processing, reactivation
of ensemble firing appears to occur only during SWS. The discrepancy
between the two data sets raises the question of whether different
types of information are processed during different brain states. The
manipulations during REM sleepMRF stimulation, reactivation
cuesdemonstrably facilitate subsequent memory performance in the
awake animal, but do not give any indication of the circuits involved.
On the other hand, the relation between the increased neuronal ensemble
firing of specified neurons and subsequent spatial memory performance
expressed at a behavioral level remains to be demonstrated. Buzsaki
(1998) has recently suggested that both REM and SWS are critical for
memory formation, the function of the former being to update the
information input from neocortex to CA3, which is then reinforced
during SWS by bursting activity critical for synaptic plasticity and
long-term consolidation.
Pharmacological Blockade of Reconsolidation
A serendipitous finding in our laboratory opened the door to
pharmacological investigation of postreactivation reconsolidation. In
experiments aimed at assessing the effect of NMDA receptor blockade on
the performance of a spatial task, rats were well trained in a radial
maze to choose three of eight baited arms, always the same three
relative to the spatial configuration of the room cues. Treatment with
low doses of the NMDA noncompetitive receptor antagonist MK-801 had no
effect on performance of the task when injected beforehand, but the
following day rats expressed an unexpected memory deficit when tested
in absence of the drug. Interpreting these results in the light of the
nearly forgotten retrograde amnesia-reactivation studies, we
hypothesized that the daily trial in the maze reactivated memory, which
was then susceptible to disruption. A series of experiments was
designed to test specifically the hypothesis that spatial reference
memories undergo an NMDA receptor-dependent reconsolidation process
after reactivation. Rats were trained to criteria in the radial arm maze with a fixed 3/8-arm pathway. A single trial, which
must be errorless, served as the reactivation procedure and
pharmacological treatments were administered at varying intervals after
this reactivation, to determine the temporal gradient of efficacy of
the drug. Using this procedure, we found that blockade of NMDA
receptors by MK-801 produces a memory impairment when injections are
made within 1 hr of the reactivation and memory is tested 24 hr or 48 hr later (Przybyslawski and Sara 1997). A group of rats receiving the
drug treatment outside of the experimental context did not show
amnesia. Further evidence for a role of NMDA receptors in a
postreactivation reconsolidation process in another species has been
provided from studies of passive avoidance memory in the day-old chick.
Chicks treated with the receptor antagonist AP5
intracerebroventricularly immediately after being presented with a
visual reminder of the training presented transient memory impairment,
whereas chicks receiving the drug without the reminder showed no
deficit (Summers et al. 1997).
-adrenergic antagonists have proved to be effective amnestic
agents in the spatial memory paradigm described above, but with a
longer temporal gradient than that of NMDA receptor antagonists. Systemic injections of propranolol are effective in inducing memory impairment when given up to 2 hr after the reactivation
treatment (Przybyslawski et al. 1999). Trained animals receiving the
drug treatment without the reactivation trial did not show amnesia when
tested a day or two later. Such amnestic effects are not limited to
this spatial reference memory paradigm; we observed similar effects of
propranolol after reactivation of a conditioned fear memory. Rats were
trained in a passive avoidance task and received the drug treatment
right after training or after reactivation by a simple retention test,
48 hr after training. Rats showing perfect retention at the test and
then treated with propranolol demonstrated significant amnesia when
retested 24 hr later (Przybyslawski et al. 1999). Interestingly, in
this experiment the reactivated memory was found to be more vulnerable
to the amnestic effects of propranolol that the newly acquired memory,
a result reminiscent of that reported by Riccio's group for
hypothermia-induced amnesia (see above). The results of the present
experiments suggest that memories, be they appetitively motivated
spatial memories or based on conditioned fear, require intact
-receptors to reconsolidate the memory trace after use.
Using intracerebroventricular rather than systemic injections to more
accurately control the time and central site of action, we defined a
specific time window, 1-2 hr after training, in which receptors
play a role in consolidation of newly acquired odor-reward associations. If the receptor antagonist is given immediately after or
5 hr after training, there is no memory impairment (Sara et al. 1999).
Recent studies using the radial maze-based spatial memory paradigm
described above revealed an extended temporal gradient during which the
-antagonist is effective, suggesting that this late
-receptor-dependent phase exists during reconsolidation processes
as well (Roullet and Sara 1998).
These pharmacological experiments reinforce the earlier literature
reviewed above, showing that reactivated memories are susceptible to
interference by a variety of amnestic agents. The results of those
early experiments, although they did not extend our knowledge of the
neurobiological processes underlying these reconsolidation processes,
did encourage the view that memory is dynamic and that new memories are
formed on the foundation of reactivated old memories. Our experiments
show that postreactivation amnesia can be induced by NMDA receptor
blockade for a short period after reactivation (Pryzbyslawski and Sara
1997) and by -receptor blockade within a rather precise time
window, 1-2 hr after reactivation (Roullet and Sara 1998).
Recently, we have obtained the same pattern of results when the
pharmacological treatments were made after initial learning, in a
rapidly acquired odor-reward association task. The NMDA receptor antagonist AP5, injected intracerebroventricularly, induced amnesia when injected immediately after training, but not 2 hr after (S. Tronel
and S.J. Sara, unpubl.). The -receptor antagonist timolol, injected under the same conditions, was effective in producing amnesia
at 2 hr, but not at 5 min, 1 hr or 5 hr after training (Sara et al.,
unpubl.). The respective intracellular biochemical pathways governed by
these receptors and their role in short-term and long-term synaptic
plasticity have been described using the LTP model system. Studying the
memory characteristics of pharmacologically or genetically modified
mice has provided further evidence at a behavioral level of the role of
NMDA receptors in the early stages of memory formation and the
importance of the cAMP cascade and phosphorylation of cAMP response
element binding (CREB) in long-term memory formation (Abeliovich et al,
1993; Bourtchouladze et al. 1994; for reviews, see Mayford et al. 1995;
Bailey et al., 1996). Beta-receptors, as a member of the family of
receptors positively linked to Gs protein would act by adenyl cyclase
to activate this pathway and thereby reinforce long-term memory
processes. A schematic diagram of these putative early NMDA receptor
dependent and late -adrenergic-dependent pathways, adapted from Mayford et al. (1995) is summarized in Figure 5.
|
Although the reactivation studies found in the experimental amnesia literature were performed in absence of any strong neurobiological hypotheses concerning mechanisms, the present results suggest that when a memory is reactivated by stimuli associated with the learning there is a reenactment of at least some of the cellular events that occur during the initial consolidation. To what extent the entire postacquisition cascade of intracellular events, shown in Figure 5, is recapitulated each time a memory is activated and reorganized is probably a function of the age and complexity of the memory and the amount of new information to be integrated into the circuit. It might be a function of the level of arousal, attention, or motivation of the animal at the time of retrieval, as well, since neuromodulatory influences would vary with those parameters.
| |
CONCLUSION AND PERSPECTIVES |
|---|
|
TOP
INTRODUCTION
RETRIEVAL
REACTIVATION AND...
CONCLUSION AND PERSPECTIVES
REFERENCES
|
|---|
The demonstration of the vulnerability of memory when it is in an
active state reinforces the idea that memories, reorganized as a
function of new experiences, undergo a reconsolidation process. Recent
pharmacological studies indicate that reconsolidation after reactivation recapitulates some of the cellular processes occurring after the initial memory acquisition (Przybyslawski and Sara 1997; Roullet and Sara 1998; S. Tronel and S.J. Sara, unpubl.; Sara et al.
1999). But these studies have not told us anything about the
neurobiology of the retrieval process itself. The rapidly developing
field of noninvasive functional imaging is providing tools to study
brain circuitry involved in retrieval processes at the network level.
At the same time genetic technology has advanced to the point where the
expression of transgenes can be induced rapidly and reversed in
selective regions of the mouse forebrain, providing tools to study
cellular mechanisms involved in retrieval and reorganization of memory.
Preliminary results confirm that memory storage, memory retrieval, and
its reconsolidation share some common processes (Mansuy et al. 1998).
We have argued that activation of brainstem neuromodulatory systems,
through a conditioned arousal response to the context, will play an
essential role in both retrieval and reconsolidation. Release of
neuromodulators, particularly NA, will facilitate attention and sensory
processing of incoming information during retrieval. The effects of NA
and other modulators in triggering intracellular processes upon which stable long-term memory is dependent would promote reconsolidation of
the newly reorganized memory. Thus, a high level of attention and
arousal at the time of retrieval will play a capital role in
reinforcing the memory, since neuromodulatory systems are activated during these behavioral states. This could account for the persistent, vivid memories associated with post traumatic stress disorder (Przybyslawski et al. 1999) and the persistent ability of the drug-taking context to induce craving.
| |
ACKNOWLEDGMENTS |
|---|
The author thanks Yves Moricard for assistance in preparing the manuscript.
| |
FOOTNOTES |
|---|
1 E-MAIL sjsara{at}ccr.jussieu.fr; FAX 331 44273251.
| |
REFERENCES |
|---|
|
TOP
INTRODUCTION
RETRIEVAL
REACTIVATION AND...
CONCLUSION AND PERSPECTIVES
REFERENCES
|
|---|
