Search

Summary: Long-term sensitization of the gill withdrawal reflex in Aplysia requires heterosynaptic, modulatory input that is mediated in part by the growth of new synaptic connections between sensory neurons and their follower cells (intrinsic mediating circuit). Whether modulatory interneurons (the extrinsic modulatory circuit) also display learning-related structural synaptic plasticity remains unknown. To test this idea, we added a bona fide serotonergic modulatory neuron, the metacerebral cell (MCC), to sensory-motor neuron co-cultures and examined the modulating presynaptic varicosities of MCCs before and after repeated pulses of serotonin (5-HT) that induced long-term facilitation (LTF). We observed robust growth of new serotonergic varicosities that were positive for serotonin and capable of synaptic recycling. Our findings demonstrate that, in addition to structural changes in the intrinsic mediating circuit, there are also significant learning-related structural changes in the extrinsic modulating circuit, and these changes might provide a cellular mechanism for savings and for spread of memory. : Upreti et al. show that, in Aplysia californica, modulatory neurons can undergo robust structural and functional changes after long-term facilitation, suggesting that learning-induced changes must include remodeling and structural alterations of the extrinsic modulating circuit. Keywords: structural plasticity, neuromodulation, modulatory neurotransmitter, serotonin, synaptic plasticity, Aplysia californica

The gill- and siphon-withdrawal reflex of Aplysia is a simple defensive behavior that is mediated in part by a monosynaptic pathway. Despite this remarkable simplicity, the reflex can be modified by several forms of learning, including habituation, dishabituation, sensitization, and classical and operant conditioning. The forms of learning that have been explored in research studies exhibit many of the behavioral properties of learning in mammals, suggesting that they may involve similar neuronal mechanisms. In 1984, we proposed cellular mechanisms for several higher-order features of conditioning and incorporated those ideas in a quantitative model that simulated a broad range of behavioral properties. In this chapter, we summarize the current body of knowledge about the behavior, circuitry, and cellular and molecular mechanisms of learning and memory of the reflex. We then review our original model and suggest how recent advances may explain some of the behavioral properties that this model could not.

SummaryGABAergic signaling in the amygdala controls learned fear, and its dysfunction potentially contributes to posttraumatic stress disorder (PTSD). We find that sub-threshold fear conditioning leads to dopamine receptor D4-dependent long-term depression (LTD) of glutamatergic excitatory synapses by increasing inhibitory inputs onto neurons of the dorsal intercalated cell mass (ITC) in the amygdala. Pharmacological, genetic, and optogenetic manipulations of the amygdala regions centered on the dorsal ITC reveal that this LTD limits less salient experiences from forming persistent memories. In further support of the idea that LTD has preventive and discriminative roles, we find that LTD at the dorsal ITC is impaired in mice exhibiting PTSD-like behaviors. These findings reveal a novel role of inhibitory circuits in the amygdala, which serves to dampen and restrict the level of fear expression. This mechanism is interfered with by stimuli that give rise to PTSD and may also be recruited for fear-related psychiatric diseases.

Cellular studies of memory suggest that experience-dependent synaptic plasticity is a fundamental mechanism by which memories are encoded, processed, and stored within the brain. In this chapter, we focus on recent advances in our understanding of the molecular mechanisms of a simple form of learning-related plasticity, presynaptic facilitation, that contributes to short-term, intermediate-term, and long-term forms of memory in the marine invertebrate Aplysia californica . We also briefly review mechanisms of presynaptic plasticity in other experimental systems, which share many similarities with those in Aplysia and may play important roles in memory as well.

Long-term facilitation inAplysiais accompanied by the growth of new synaptic connections between the sensory and motor neurons of the gill-withdrawal reflex. One of the initial steps leading to the growth of these synapses is the internalization, induced by 5-HT, of the transmembrane isoform ofAplysiacell-adhesion molecule (TM–apCAM) from the plasma membrane of sensory neurons (Bailey et al., 1992). However, the mechanisms that govern the internalization of TM–apCAM and how this internalization is coupled to the molecular events that initiate the structural changes are not fully understood. Here, we report that the synthesis of membrane phosphatidylinositol 4,5-bisphosphate [PI(4,5)P2], which is known to be mediated by a signaling cascade throughAplysiaSec7 protein (ApSec7) and phosphatidylinositol-4-phosphate 5-kinase type I α (PIP5KIα) is required for both the internalization of TM–apCAM and the initiation of synaptic growth during 5-HT-induced long-term facilitation. Pharmacological blockade of PI(4,5)P2synthesis by the application of the inhibitor phenylarsine oxide blocked the internalization of apCAM. Furthermore, perturbation of the endogenous activation of ApSec7 and its downstream target PIP5KIα also blocked 5-HT-mediated internalization of TM–apCAM and synaptic growth. Finally, long-term facilitation was specifically impaired by blocking the ApSec7 signaling pathway at sensory-to-motor neuron synapses. These data indicate that the ApSec7/PIP5KIα signaling pathway is actively recruited during learning-related 5-HT signaling and acts as a key regulator of apCAM internalization associated with the formation of new synaptic connections during long-term facilitation.

The memory reconsolidation hypothesis suggests that a memory trace becomes labile after retrieval and needs to be reconsolidated before it can be stabilized. However, it is unclear from earlier studies whether the same synapses involved in encoding the memory trace are those that are destabilized and restabilized after the synaptic reactivation that accompanies memory retrieval, or whether new and different synapses are recruited. To address this issue, we studied a simple nonassociative form of memory, long-term sensitization of the gill- and siphon-withdrawal reflex in Aplysia , and its cellular analog, long-term facilitation at the sensory-to-motor neuron synapse. We found that after memory retrieval, behavioral long-term sensitization in Aplysia becomes labile via ubiquitin/proteasome-dependent protein degradation and is reconsolidated by means of de novo protein synthesis. In parallel, we found that on the cellular level, long-term facilitation at the sensory-to-motor neuron synapse that mediates long-term sensitization is also destabilized by protein degradation and is restabilized by protein synthesis after synaptic reactivation, a procedure that parallels memory retrieval or retraining evident on the behavioral level. These results provide direct evidence that the same synapses that store the long-term memory trace encoded by changes in the strength of synaptic connections critical for sensitization are disrupted and reconstructed after signal retrieval.

Despite considerable evidence for a critical role of neuroligin-1 in the specification of excitatory synapses, the cellular mechanisms and physiological roles of neuroligin-1 in mature neural circuits are poorly understood. In mutant mice deficient in neuroligin-1, or adult rats in which neuroligin-1 was depleted, we have found that neuroligin-1 stabilizes the NMDA receptors residing in the postsynaptic membrane of amygdala principal neurons, which allows for a normal range of NMDA receptor-mediated synaptic transmission. We observed marked decreases in NMDA receptor-mediated synaptic currents at afferent inputs to the amygdala of neuroligin-1 knockout mice. However, the knockout mice exhibited a significant impairment in spike-timing-dependent long-term potentiation (STD-LTP) at the thalamic but not the cortical inputs to the amygdala. Subsequent electrophysiological analyses indicated that STD-LTP in the cortical pathway is largely independent of activation of postsynaptic NMDA receptors. These findings suggest that neuroligin-1 can modulate, in a pathway-specific manner, synaptic plasticity in the amygdala circuits of adult animals, likely by regulating the abundance of postsynaptic NMDA receptors.

SummaryTranssynaptic interactions between neurons are essential during both developmental and learning-related synaptic growth. We have used Aplysia neuronal cultures to examine the contribution of transsynaptic signals in both types of synapse formation. We find that during de novo synaptogenesis, specific presynaptic innervation is required for the clustering of postsynaptic AMPA-like but not NMDA-like receptors. We further find that the cell adhesion molecule Dscam is involved in these transsynaptic interactions. Inhibition of Dscam either pre- or postsynaptically abolishes the emergence of synaptic transmission and the clustering of AMPA-like receptors. Remodeling of both AMPA-like and NMDA-like receptors also occurs during learning-related synapse formation and again requires the reactivation of Dscam-mediated transsynaptic interactions. Taken together, these findings suggest that learning-induced synapse formation recapitulates, at least in part, aspects of the mechanisms that govern de novo synaptogenesis.

SummaryRepeated pulses of serotonin (5-HT) induce long-term facilitation (LTF) of the synapses between sensory and motor neurons of the gill-withdrawal reflex in Aplysia. To explore how apCAM downregulation at the plasma membrane and CREB-mediated transcription in the nucleus, both of which are required for the formation of LTF, might relate to each other, we cloned an apCAM-associated protein (CAMAP) by yeast two-hybrid screening. We found that 5-HT signaling at the synapse activates PKA which in turn phosphorylates CAMAP to induce the dissociation of CAMAP from apCAM and the subsequent translocation of CAMAP into the nucleus of sensory neurons. In the nucleus, CAMAP acts as a transcriptional coactivator for CREB1 and is essential for the activation of ApC/EBP required for the initiation of LTF. Combined, our data suggest that CAMAP is a retrograde signaling component that translocates from activated synapses to the nucleus during synapse-specific LTF.

Consolidation of implicit memory in the invertebrate Aplysia and explicit memory in the mammalian hippocampus are associated with remodeling and growth of preexisting synapses and the formation of new synapses. Here, we compare and contrast structural components of the synaptic plasticity that underlies these two distinct forms of memory. In both cases, the structural changes involve time-dependent processes. Thus, some modifications are transient and may contribute to early formative stages of long-term memory, whereas others are more stable, longer lasting, and likely to confer persistence to memory storage. In addition, we explore the possibility that trans-synaptic signaling mechanisms governing de novo synapse formation during development can be reused in the adult for the purposes of structural synaptic plasticity and memory storage. Finally, we discuss how these mechanisms set in motion structural rearrangements that prepare a synapse to strengthen the same memory and, perhaps, to allow it to take part in other memories as a basis for understanding how their anatomical representation results in the enhanced expression and storage of memories in the brain.

Cellular studies of implicit and explicit memory suggest that experience-dependent modulation of synaptic strength and structure is a fundamental mechanism by which these memories are encoded, processed, and stored within the brain. In this review, we focus on recent advances in our understanding of the molecular mechanisms that underlie short-term, intermediate-term, and long-term forms of implicit memory in the marine invertebrate Aplysia cali- fornica, and consider how the conservation of common elements in each form may contribute to the different tem- poral phases of memory storage.

Summary Application of Clostridium difficile toxin B, an inhibitor of the Rho family of GTPases, at the Aplysia sensory to motor neuron synapse blocks long-term facilitation and the associated growth of new sensory neuron varicosities induced by repeated pulses of serotonin (5-HT). We have isolated cDNAs encoding Aplysia Rho , Rac , and Cdc42 and found that Rho and Rac had no effect but that overexpression in sensory neurons of a dominant-negative mutant of ApCdc42 or the CRIB domains of its downstream effectors PAK and N-WASP selectively reduces the long-term changes in synaptic strength and structure. FRET analysis indicates that 5-HT activates ApCdc42 in a subset of varicosities contacting the postsynaptic motor neuron and that this activation is dependent on the PI3K and PLC signaling pathways. The 5-HT-induced activation of ApCdc42 initiates reorganization of the presynaptic actin network leading to the outgrowth of filopodia, some of which are morphological precursors for the learning-related formation of new sensory neuron varicosities.

Recent cellular and molecular studies of both implicit and explicit memory storage suggest that experience-dependent modulation of synaptic strength and structure is a fundamental mechanism by which these diverse forms of memory are encoded and stored. For both forms of memory storage, some type of synaptic growth is thought to represent the stable cellular change that maintains the long-term process. In this review, we discuss recent findings on the molecular events that underlie learning-related synaptic growth in Aplysia and discuss the possibility that an active, prion-based mechanism is important for the maintenance of the structural change and for the persistence of long-term memory.

The time course and functional significance of the structural changes associated with long-term facilitation of Aplysia sensory to motor neuron synaptic connections in culture were examined by time-lapse confocal imaging of individual sensory neuron varicosities labeled with three different fluorescent markers: the whole-cell marker Alexa-594 and two presynaptic marker proteins—synaptophysin-eGFP to monitor changes in synaptic vesicle distribution and synapto-PHluorin to monitor active transmitter release sites. Repeated pulses of serotonin induce two temporally, morphologically, and molecularly distinct presynaptic changes: (1) a rapid activation of silent presynaptic terminals by filling of preexisting empty varicosities with synaptic vesicles, which parallels intermediate-term facilitation, is completed within 3–6 hr and requires translation but not transcription and (2) a slower generation of new functional varicosities which occurs between 12–18 hr and requires transcription and translation. Enrichment of empty varicosities with synaptophysin accounts for 32% of the newly activated synapses at 24 hr, whereas newly formed varicosities account for 68%.

Long-term memory for sensitization of the gill-withdrawal reflex inAplysia is associated with the growth of new synaptic connections between sensory and motor neurons. The duration of this structural change parallels the behavioral retention of the memory. Such changes can be reconstituted in dissociated cell culture by repeated presentations of the modulatory neurotransmitter serotonin (5HT) and are associated with an activity-dependent downregulation of NCAM-related cell adhesion molecules thought to contribute to cell recognition and axonal outgrowth during development. Thus, aspects of the mechanisms utilized for learning-related synaptic growth initiated by experience in the adult may eventually be understood in the context of the molecular logic that shapes synaptic circuitry during the later stages of neuronal development.Key words: learning, synapse, invertebrate, habituation, sensitization.

The requirement for transcription during long-lasting synaptic plasticity has raised the question of whether the cellular unit of synaptic plasticity is the soma and its nucleus or the synapse. To address this question, we cultured a single bifurcated Apylsia sensory neuron making synapses with two spatially separated motor neurons. By perfusing serotonin onto the synapses made onto one motor neuron, we found that a single axonal branch can undergo long-term branch-specific facilitation. This branch-specific facilitation depends on CREB-mediated transcription and involves the growth of new synaptic connections exclusively at the treated branch. Branch-specific long-term facilitation requires local protein synthesis in the presynaptic but not the postsynaptic cell. In fact, presynaptic sensory neuron axons deprived of their cell bodies are capable of protein synthesis, and this protein synthesis is stimulated 3-fold by exposure to serotonin.

Synapses of the hippocampal mossy fiber pathway exhibit several characteristic features, including a unique form of long-term potentiation that does not require activation of the N-methyl-D-aspartate receptor by glutamate, a complex postsynaptic architecture, and sprouting in response to seizures. However, these connections have proven difficult to study in hippocampal slices because of their relative paucity (

The storage of long-term memory in Aplysia is associated with the growth of synaptic connections, new protein synthesis and altered gene expression. These changes can now be induced in dissociated cell culture by repeated presentation of the modulatory neurotransmitter serotonin (5HT) where they are associated with a transient internalization of NCAM-related cell adhesion molecules thought to contribute to axonal outgrowth during development. Thus, learning-related synaptic growth initiated by experience in the adult may share mechanisms in common with the activity-dependent refinement of synaptic connections that shapes the developing nervous system.

In the course of studying proteins involved in long-term facilitation in Aplysia, we found that 5-HT and cAMP, a second messenger activated by 5-HT, lead to the removal of a set of N-CAM-related cell adhesion molecules (apCAMs) from the surface membrane of sensory neurons by means of receptor-mediated endocytosis. Here we describe that, as part of this coordinated program for endocytosis, 5-HT and cAMP also induce in the sensory neurons an increase in the density of coated pits and coated vesicles and an increase in the expression of the light chain of Aplysia clathrin (apClathrin). The clathrin-related endocytosis seems designed to internalize and redistribute apCAMs and other surface membrane proteins in the sensory neurons, and thus it appears to constitute one of the initial steps in the growth of new synaptic connections that accompanies long-term facilitation.

Synaptic connections between the sensory and motor neurons of Aplysia in culture undergo long-term facilitation in response to serotonin (5-HT) and long-term depression in response to FMRFamide. These long-term functional changes are dependent on the synthesis of macromolecules during the period in which the transmitter is applied and are accompanied by structural changes. There is an increase and a decrease, respectively, in the number of sensory neuron varicosities in response to 5-HT and FMRFamide. To determine whether macromolecular synthesis is also required for the structural changes, we examined in parallel the effects of inhibitors of protein (anisomycin) or RNA (actinomycin D) synthesis on the structural and functional changes. We have found that anisomycin and actinomycin D block both the enduring alterations in varicosity number and the long-lasting changes in synaptic potential. These results indicate that macromolecular synthesis is required for expression of the long-lasting structural changes in the sensory cells and that this synthesis is correlated with the long-term functional modulation of sensorimotor synapses.

The long-term facilitation of synaptic efficacy that is induced by serotonin in dissociated cell cultures of sensory and motor neurons of Aplysia is accompanied by the growth of new synaptic connections. This growth is associated with a down-regulation in the sensory neuron of Aplysia cell adhesion molecules (apCAMs). To examine the mechanisms of this down-regulation, thin-section electron microscopy was combined with immunolabeling by gold-conjugated monoclonal antibodies specific to apCAM. Within 1 hour, serotonin led to a 50% decrease in the density of gold-labeled complexes at the surface membrane of the sensory neuron. This down-regulation was achieved by a heterologous, protein synthesis-dependent activation of the endosomal pathway, which leads to internalization and apparent degradation of apCAM. The internalization is particularly prominent at sites where the processes of the sensory neurons contact one another and may act there to destabilize process-to-process contacts that normally inhibit growth. In turn, the endocytic activation may lead to a redistribution of membrane components to sites where new synapses form.

1 Memory's Two Major Forms 2 Long-Term Synaptic Plasticity 3 Learning-Induced Growth of New Sensory Neuron Synapses 4 Functional Contribution of Presynaptic Structural Changes to Long-Term Facilitation 5 5-HT-Induced Regulation of the Presynaptic Actin Network 6 Chromatin Remodeling and Epigenetic Changes During Long-Term Memory Storage 7 Consolidation of Long-Term Memory 8 Stabilization of Learning-Related Synaptic Growth 9 Summary Keywords: aplysia; cell adhesion molecules; explicit memory; gene expression; implicit memory; learning and memory; local protein synthesis; long-term facilitation; LTF; long-term potentiation; LTP; long-term sensitization; long-term memory; memory storage; nascent synapse; silent synapse; structural changes; synapse; synapse formation; synaptic growth; synaptic plasticity; synaptic remodeling; transcription; translation

Structural remodeling of preexisting synapses and the growth of new synapses accompany various forms of long-term memory. Long-term sensitization of the gill-withdrawal reflex in Aplysia has been extensively studied in this regard and is associated with the presynaptic remodeling and growth of new synaptic connections made by identified sensory neurons onto their postsynaptic target neurons. Recent studies of living sensory–motor neuron synapses in culture have begun to define the cellular and molecular events that underlie these learning-related structural changes. This in turn has suggested the molecules and mechanisms the governing de novo synapse formation during development to be reutilized in the adult for the purposes of synaptic plasticity and memory storage.

The morphological basis of long-term sensitization of the gill-and-siphon withdrawal reflex in Aplysia was explored by examining the structure of identified sensory neuron synapses in control and behaviorally modified animals. Following long-term training, sensitized animals displayed an increase in the number of sensory neuron synapses compared to control animals. The relative permanence of these structural changes and their similarity in time course to the behavioral duration of sensitization suggest a role for synapse number changes during long-term memory.

Neuroligin-1 is a potent trigger for the de novo formation of synaptic connections, and it has recently been suggested that it is required for the maturation of functionally competent excitatory synapses. Despite evidence for the role of neuroligin-1 in specifying excitatory synapses, the underlying molecular mechanisms and physiological consequences that neuroligin-1 may have at mature synapses of normal adult animals remain unknown. By silencing endogenous neuroligin-1 acutely in the amygdala of live behaving animals, we have found that neuroligin-1 is required for the storage of associative fear memory. Subsequent cellular physiological studies showed that suppression of neuroligin-1 reduces NMDA receptor-mediated currents and prevents the expression of long-term potentiation without affecting basal synaptic connectivity at the thalamo-amygdala pathway. These results indicate that persistent expression of neuroligin-1 is required for the maintenance of NMDAR-mediated synaptic transmission, which enables normal development of synaptic plasticity and long-term memory in the amygdala of adult animals.

Introduction: Memory can be divided into declarative and non-declarative processes. Declarative or explicit memory is the conscious recall of knowledge about facts and events and is particularly well developed in the vertebrate brain. Non-declarative or implicit memory is the non-conscious recall of skilled behavior and other tasks and includes simple associative forms such as classical conditioning and non-associative forms such as sensitization 12. During sensitization an animal learns about the properties of a single noxious stimulus enabling the formation of a learned fear response. The cellular and molecular mechanisms that underlie the storage of implicit memory have been most extensively analyzed in the gill- and siphon-withdrawal reflex of marine mollusk Aplysia. This organism offers several unique advantages for the study of learning and memory, such as a relatively simple and tractable central nervous system, large identified neurons, and well-characterized neural circuits related to specific behaviors that can be modified by learning 3. In Aplysia, sensitization of the gill- and siphon-withdrawal reflex is induced by a strong stimulus to its tail 4. Repetitive stimuli produce long-term sensitization that lasts days to several weeks whereas a single stimulus induces short-term sensitization lasting only minutes to a few hours 56. These two forms of memory can be reconstituted in dissociated sensory-motor neuron cultures by the modulatory neurotransmitter serotonin (5-HT) 7. A single pulse of 5-HT induces short-term facilitation (STF), whereas five applications of 5-HT induce long-term facilitation (LTF) 8. In contrast to the short-term synaptic changes, which involve covalent modifications of preexisting molecules leading to an alteration of preexisting synaptic connections, long-term synaptic changes require the synthesis of new macromolecules including mRNAs and proteins 891011. In addition, long-term changes are accompanied by structural modifications including the growth of new synaptic connections between the sensory neurons and their target motor neurons 1213. These features of long-term, learning-related synaptic plasticity, are highly conserved throughout evolution of the nervous system. A variety of experimental systems, ranging in complexity from Aplysia to rodents, have been used to examine the molecular mechanism underlying long-term synaptic changes 121011. In this review, we focus on the role of nuclear transcription factors in the presynaptic sensory neurons of Aplysia during LTF. A model system for examining the molecular biology of long-term memory – the Aplysia sensory to motor neuron synapse: In sensory-motor neuron cultures, STF and LTF can be induced by applying 5-HT. One pulse of 5-HT activates PKA and PKC, probably via the activation of different types of G proteins. Activated PKA phosphorylates a potassium channel (S channel), resulting in the elevated influx of calcium leading to an increase in membrane excitability and spike broadening. PKC facilitates the mobilization of synaptic vesicles to the presynaptic active zone. Together, these kinases enhance transmitter release by modifying preexisting molecules 14151617. LTF induced by repeated pulses of 5-HT requires the synthesis of both new proteins and RNAs. Inhibitors of protein or RNA synthesis selectively block LTF, but not STF when applied within a critical time window that encompasses the training protocol 8. Analogous to STF, cAMP-dependent protein phosphorylation is also involved in LTF and cAMP analogs induce LTF 18. Bacskai and colleagues first demonstrated that in response to repeated pulses of 5-HT the catalytic subunit of PKA translocates into the nucleus of the presynaptic sensory neuron to activate CREB-1 19. Repeated pulses of 5-HT also induce phosphorylation of MAPK and the activated MAPK also translocates into the nucleus of the sensory neuron where it removes the repressive influence of ApCREB-2 20. Dash et al. provided the first evidence of the involvement of cAMP-inducible genes expressed during LTF in Aplysia. Microinjection of CRE (cAMP-responsive element) oligonucleotides into the nucleus of sensory neurons selectively blocked 5-HT-induced LTF without affecting short-term changes 21. These data first suggested that LTF requires the activation of cAMP-inducible genes, and that CRE oligonucleotides prevent interactions between CRE-binding protein (CREB)-related transcription factors and these genes. Using a newly developed gene transfer technique, Kaang and colleagues next showed directly that four or more pulses of 5-HT stimulate CRE-mediated gene expression. Moreover, transcription induced by 5-HT requires the phosphorylation of CREB on Ser119 by PKA 22. Indeed Bartsch et al. went on to show that injection of phosphorylated CREB into the sensory neuron can by itself initiate the long-term process 23. Collectively these results suggested that a signaling axis composed of cAMP-PKA-CREB participates in the molecular cascade leading to the expression of LTF. Subsequent studies confirmed the involvement of a number of related downstream molecules in the induction of LTF, including CAMAP, ApC/EBP, and ApAF. ApCREBs – Central modulators of LTF: In Aplysia, ApCREB1a and ApCREB2 have been characterized as an activator and repressor of LTF, respectively. ApCREB2 is a homolog of mammalian CREB2/ATF4 which is also identified as a transcriptional repressor 2425. Inhibition of ApCREB2 by injecting antiserum or double-strand RNA (dsRNA) into sensory neurons allowed a single pulse of 5-HT to produce translation-dependent LTF and the growth of new synapses 2526. Conversely, injection of an anti-CREB1 antibody into sensory neurons selectively blocked LTF. Moreover, introduction of the phosphorylated transcriptional activator, ApCREB1a, was sufficient to induce LTF. 23. PKA-mediated activation of ApCREB1a stimulated the downstream transcription factor, ApC/EBP, via recruiting CBP and subsequently facilitating histone acetylation 27. For activation of downstream gene expression by ApCREB1a, the inhibitory constraint of ApCREB2 must be relieved. This de-repression is mediated by the phosphorylation of ApCREB2 by nuclear translocated MAPK 2028. Together these results suggest that LTF requires not only the activation of memory-enhancer genes but also the inactivation of memory-suppressor genes. Moreover, CREBs appear to play a critical role in maintaining the dynamic balance between these positive and negative factors. CREB, a basic leucine zipper transcription factor, is also reported to be involved in long-term plasticity in the nervous systems of other organisms, including fly and mouse. In transgenic Drosophila, expression of dCREB2b led to the blockage of long-term memory, whereas dCREB2a facilitated memory 2930. Similarly, CREB-deficient mice displayed impaired long-term potentiation (LTP) and long-term memory 31. The threshold for late phase LTP was lowered in the hippocampus from mice expressing the constitutively active form of CREB 32. ApC/EBP – A key downstream gene for LTF: In view of the critical roles of CREB in memory consolidation, characterization of its downstream effectors has been a major focus of research in both invertebrate and vertebrate learning models 333435. In Aplysia the CCAAT enhancer-binding protein (ApC/EBP), an immediate early gene during the consolidation phase of LTF, was found by Alberini et al. to be a downstream target of ApCREB1 36. The C/EBP family of transcription factors contains a basic leucine-zipper domain. Specifically, expression of ApC/EBP is rapidly induced in response to 5-HT treatment and this occurs to an immediate early gene in a translation-independent manner 36. Inhibition of ApC/EBP by injection of ERE oligonucleotides, ApC/EBP antiserum or dsRNA into the sensory neurons in sensory-to-motor neuron cultures during a critical time window blocked 5-HT-induced LTF 3637. Moreover, a single pulse of 5-HT which normally induces only STF, produces LTF when ApC/EBP is overexpressed in the sensory neuron 37. These findings support the idea that ApC/EBP is both necessary and sufficient to consolidate short-term memory into long-term memory. However, since overexpression of ApC/EBP alone does not induce LTF, an additional component must be required for converting STF to LTF. Lee and colleagues used the RNA interference technique designed to block the function of ApC/EBP and similarly found that this blocked LTF in Aplysia 2637. Moreover, in addition to these studies in Aplysia, C/EBPβ and -δ are induced in the rodent hippocampus after inhibitory avoidance learning, suggesting that C/EBPs are highly conserved molecular components of the CREB-dependent signal pathway involved in memory consolidation 38. ApC/EBP is also known to be induced by the neural activity. The depolarization-induced ApC/EBP induction appears to be mediated by transient induction of the nucleolar protein, ApLLP, which was recently characterized as a novel transcription factor induced by neural activity in Aplysia sensory neuron 39. Kim et al. also showed that a single pulse of 5-HT can produce LTF when the synapse is pretreated with high potassium solution. LTF induced by this protocol was completely blocked by the injection of anti-ApLLP or anti-ApC/EBP antibody 39. One important feature of immediate early genes such as ApC/EBP, is that their expression is tightly regulated within a specific and narrow time window. ApC/EBP mRNA displays a peak expression at 2 h after induction that rapidly returns to the basal levels 36. Recent studies have found that the ApC/EBP 3' UTR contains putative AU-rich element (ARE) sequences which are cis-acting regulatory elements commonly found in labile mRNAs 40. Different sets of ARE binding proteins may interact with ApC/EBP mRNA to regulate its stability and/or translatability 414243. Yim et al. found that an ARE binding protein, ApELAV binds to and stabilizes ApC/EBP mRNA, suggesting that post-transcriptional regulation of ApC/EBP may also play an important role during LTF 40. ApAF – A binding partner for both ApCREB2 and ApC/EBP: Both ApC/EBP and ApCREBs are transcription factors containing the basic leucine zipper (bZIP) domain in the C-terminus. This domain is involved in both DNA binding and multimerization 232544. Using the bZIP domain of ApC/EBP as bait, Bartsch and colleagues screened a cDNA library to identify additional transcription factors acting downstream of ApCREB1. In this fashion, they cloned a novel transcription factor ApAF, whose activity is regulated by PKA 45. In contrast to ApC/EBP, which is induced in response to 5-HT, ApAF is a constitutively expressed gene. Interestingly, an in vitro binding assay revealed that ApAF interacts with both ApCREB2 and ApC/EBP, but not ApCREB1. Inhibition of ApAF by injection of a specific antibody blocked LTF induced by repeated pulses of 5-HT, suggesting that ApAF is necessary for LTF. Previously, it had been found that injection of phosphorylated ApCREB1a by itself or anti-ApCREB2 antibody combined with a single pulse of 5-HT induced LTF that phenocopied 5 × 5-HT-induced LTF 2325. ApAF is involved in both forms of LTF: an anti-ApAF antibody blocked LTF induced by phosphorylated CREB1a as well as that by ApCREB2 antibody injection paired with one pulse of 5-HT. Moreover, overexpression of ApAF enhanced both LTF induced by 5 × 5-HT and ApCREB2 antibody with 1 × 5-HT 45. Thus, ApAF may be a potential memory enhancer gene downstream of ApCREB1 and ApCREB2. To determine whether ApC/EBP is the critical partner of ApAF, Lee and colleagues investigated the effects of silencing the ApAF gene on LTF induced by ApC/EBP overexpression paired with one pulse of 5-HT 46. ApAF inhibition by dsRNA completely blocked LTF induced by both ApC/EBP overexpression and 5 × 5-HT. In combination with a single pulse of 5-HT, the ApAF-ApC/EBP heterodimer produced LTF, even in the absence of CRE- and CREB-mediated gene expression. These results provide direct evidence that the ApAF-ApC/EBP heterodimer is a key downstream effecter of ApCREB. Furthermore, ApAF enhances ERE-mediated gene expression by cooperating with ApC/EBP and phosphorylation at Ser-266 of ApAF by PKA is required for activation of the ApAF-ApC/EBP heterodimer during 5-HT-induced LTF 46. These data explain why ApC/EBP overexpression in the absence of 5-HT could not convert STF to LTF. The single pulse of 5-HT in ApC/EBP-induced LTF possibly functions in triggering the phosphorylation of ApAF by activated PKA. Lee et al. also examined the role of Ser-266 phosphorylation of ApAF in the relief of ApCREB2 repression 46. Overexpression of a dominant negative mutant of ApAF which cannot be phosphorylated at Ser-266 completely blocked both LTF induced by 5 × 5-HT and that by the ApC/EBP overexpression combined with 1 × 5-HT. Moreover, this mutant restored ApCREB1-mediated gene expression and 5-HT-induced LTF repressed by ApCREB2 as efficiently as its wild type counterpart, suggesting that Ser-266 phosphorylation of ApAF is not required to relieve ApCREB2 repression 46. However, the precise signaling pathway that regulates the interactions between ApAF and ApCREB2 remains to be characterized. Taken together, these studies of ApAF indicate that transcriptional regulation of memory consolidation is quite diverse and can recruit both direct and indirect interactions between transcription factors. CAMAP – A retrograde signal from the membrane to the CRE promoter: Neurons display a distinct highly differentiated form which consists of three basic compartments: a cell body or soma which contains the nucleus harboring genomic information and two types of processes axons and dendrites. Dendrites are input elements of the neuron. Together with the cell body they receive synaptic contacts from other neurons. Axons are the output elements of the neuron. The branches of each axon form numerous synaptic connections with other neurons. It is well known that long-term synaptic plasticity requires the synthesis of both new RNAs and proteins. This raises three fundamental questions: 1) Does synaptic plasticity always occur in a cell-wide manner? 2) If not, how does the nucleus identify the correct synapses for delivery of gene products to achieve synapse-specific plasticity? 3) Is the nucleus the only site responsible for transcription and translation? To address the question of synapse specificity, Martin and her colleagues developed a new culture system in Aplysia consisting of a bifurcated Aplysia sensory neuron which makes synapses with two spatially separated motor neurons in culture 47. They found that 5-HT can induce branch-specific LTF that was dependent on CREB activation in the nucleus of presynaptic sensory neuron. Branch-specific LTF was also accompanied by synapse-specific growth of new sensory neuron synapses. These studies highlight the importance of a retrograde signal propagating from the stimulated synaptic site to the nucleus. The CREB-downstream molecules produced in the cell body can be captured by other synapses which have been tagged or marked. This synaptic mark is PKA-dependent; however, rapamycin-dependent local protein synthesis is required for LTF to persist for more than 72 h 4748. Further molecular studies have dissected the characteristics of the retrograde signal and the synaptic mark. An early step in the growth of new synaptic connections is the internalization of an NCAM immunoglobulin-related cell adhesion molecule – apCAM. In recent studies Lee et al. addressed the question: How does this internalization of apCAM relate to the activation of transcription? Lee found that CAMAP is an interacting partner of apCAM in the sensory neuron. CAMAP serves as a transcriptional co-activator that is also a crucial retrograde signaling molecule involved in the initiation phase of LTF 49. ApCAM is down-regulated upon application of 5-HT 50. Serotonin leads to clathrin-mediated endocytosis of the transmembrane isoform of apCAM (TM-ApCAM) from the surface membrane of sensory neurons and this internalization depends on phosphorylation of its cytoplasmic tail by MAPK 5152. When TM-apCAM was overexpressed in sensory neurons, five pulses of 5-HT failed to produce synaptic facilitation or enhancement of synaptic growth, suggesting that down-regulation of apCAM is required for both LTF and the presynaptic structural changes induced by 5-HT 53. Whereas MAPK is known to be involved in apCAM downregulation, the subcellular mechanisms responsible for apCAM internalization remain to be characterized. CAMAP (apCAM-Associated Protein) was cloned by yeast two-hybrid screening using the cytoplasmic tail of TM-ApCAM as bait 49. CAMAP is colocalized with apCAM at the surface of the plasma membrane in the basal state and is translocated to the nucleus of sensory neuron after treatment with 5-HT. Nuclear translocation and dissociation from apCAM are modulated by PKA-mediated phosphorylation of CAMAP. A phosphorylation-mimicking mutant of CAMAP can dissociate from apCAM and translocate to the nucleus in the absence of 5-HT stimulation 49. How does CAMAP translocate to the nucleus? Importins, which transport cargo molecules from the cytosol to nucleus, are critically involved in synaptic plasticity in both Aplysia and rodent brain 54. This raises the possibility that importin may transport CAMAP into the nucleus of the sensory neuron. In view of nuclear translocation of CAMAP from the plasma membrane, Lee et al. suggested that CAMAP acts as a retrograde signaling molecule in the induction of LTF. To act as a retrograde signal, a molecule must meet several criteria, specifically, 1) it should translocate into the nucleus from the synaptic site in response to stimuli that induce synaptic plasticity, such as LTF and LTP, 2) it should play a specific function in the nucleus, and 3) blockage of its translocation and function should inhibit the long-term synaptic change. CAMAP appears to fulfill all of the above requirements. First, CAMAP translocates to the nucleus following both cell-wide and synapse-specific applications of 5-HT. After 5-HT treatment, the mobility of CAMAP is significantly increased at the distal neurites of sensory neurons. Second, CAMAP binds to ApCREB1 in the nucleus, where it acts as a transcriptional co-activator that can induce ApC/EBP expression. Interestingly, the N-terminal region, but not full-length CAMAP displayed transcriptional activity. The C-terminus of CAMAP appears to act as an autoinhibitory domain. This inhibitory clamp is relieved by PKA phosphorylation of CAMAP. Details of the molecular mechanisms that underlie the co-activator function of CAMAP and the PKA phosphorylation-dependent restoration of transcriptional activity of the N-terminus of CAMAP are currently not known. Finally, blockage of CAMAP expression by dsRNA completely suppresses 5-HT-induced ApC/EBP upregulation and LTF. Moreover, CAMAP dsRNA blocks synapse-specific LTF induced by local application of 5-HT at the synapse. Taken together, these results indicate that CAMAP is a critical retrograde messenger in the initiation of LTF. The precise role of CAMAP in the internalization of apCAM remains unclear. We have recently found that overexpression of mutant CAMAP that cannot be phosphorylated by PKA impaired the 5-HT-induced internalization of apCAM 49. This finding suggests that CAMAP tethers apCAM to the plasma membrane in the basal state, and that the phosphorylation and nuclear translocation of CAMAP are necessary for the subsequent down-regulation of apCAM. Chromatin alteration and epigenetic changes in gene expression with memory storage: Although epigenetic mechanisms were widely known to be involved in the formation and long-term storage of cellular information in response to transient environmental signals, the discovery of their putative relavance in adult brain function is relatively recent 2755. The epigenetic marking of chromatin, such as histone modification, chromatin remodeling and the activity of retrotransposons, may have long-term consequences in the transcriptional regulation of specific loci involved for long-term synaptic changes 56. The contribution of histone tail acetylation, a modification that favors transcription and is associated with active loci, to LTF formation was first revealed by the study of Guan et al. in Aplysia neurons 27. This study found that both facilitatory and inhibitory stimuli bidirectionally alter the acetylation stage and structure of promoters driven by the expression of genes involved in the maintenance of LTF, such as ApC/EBP. It also demonstrated that enhancing histone acetylation with histone deacetylase (HDAC) inhibitors facilitates the induction of LTF (Guan et al. 2002). HDAC inhibitors have recently been shown to have similar effects during L-LTP in the Schaffer collateral pathway of mammals and to enhance memory formation in hippocampus-dependent tasks 5758596061. HDAC inhibitor, sodium butyrate, has been shown to induce the growth of new dendrites and synapses, which might be an underlying mechanism of its memory enhancing effects 61. Moreover, mice with reduced histone acetyltransferase activity, such as different mouse models for Rubinstein-Taybi mental retardation syndrome, have deficits in both long-lasting forms of memory and LTP 57586263. These results indicate that critical chromatin changes occur during the formation of long-term memory and that these changes are required for the stable maintenance of these memories. Perspectives During the last decade molecular studies have increased our understanding of the signaling and regulatory mechanisms that underlie LTF (Figure 1). Several research groups have generated a variety of cellular models of long-term memory by investigating 5-HT-induced LTF and other forms of long-term synaptic changes induced by different stimulation paradigms in Aplysia. Figure 1 Schematic model of signaling pathways underlying long-term facilitation in Aplysia sensory neuron Schematic model of signaling pathways underlying long-term facilitation in Aplysia sensory neuron. The repeated treatments with neurotransmitter 5-HT activate a G-protein coupled receptor that stimulates adenylyl cyclase, which in turn activates PKA. MAPK are also activated and translocates into the nucleus. At the synaptic site, PKA stimulates the nuclear translocation of the retrograde signal molecule CAMAP via phosphorylating its Ser148. This phosphorylation results in both the dissociation from TM-apCAM and the restoration of its transcriptional activity from autoinhibition. In the nucleus, MAPK phosphorylates CREB2 which represses CREB1 and ApAF in the absence of 5-HT. Once freed from CREB2 and stimulated by PKA, CREB1 forms a homodimer to activate the downstream target gene, ApC/EBP. Translocated CAMAP acts as a co-activator of CREB1. ApC/EBP interacts with ApAF that is activated by PKA to form a core downstream effector of CREB1. ApC/EBP-ApAF heterodimer induces the late genes which are critical for the consolidation and maintenance of LTF. Robust neural activity induces and activates the transcription factor, ApLLP in the nucleus in a calcium-dependent manner. ApLLP induces ApC/EBP expression and lowers the threshold for LTF induction. Elucidating the downstream molecule of ApC/EBP remains to be challenged. SNS, strong noxious stimulus. Recently, operant conditioning was demonstrated in the gill-withdrawal and feeding behavior in Aplysia and an electrophysiological study revealed that operant and classical conditioning of feeding behavior differentially modify the intrinsic excitability of an identified neuron 646566. PolyADP-ribose-polymerase 1 (PARP1) facilitates the transcription of long-term memory related genes by decondensing chromatin structure in neurons that mediate operant conditioning 67. However, the transcription factors involved in this behavioral modification are yet to be identified. Interestingly, ApC/EBP is the common downstream molecule of the novel transcription factors, ApLLP and CAMAP. ApC/EBP, which pairs with ApAF, could activate the transcription of effector genes critically involved in the consolidation and maintenance of long-term memory 36. Genes encoding structural proteins, such as clathrin light chain and the chaperon BiP were identified as late effector genes 6869. The elongation factor 1 alpha was also suggested to be essential for maintaining newly formed synapses 70. However, the number of late genes identified thus far represents only a beginning. Since the DNA-binding motifs of ApC/EBP and ApAF homodimer or heterodimer have been analyzed, completion of the ongoing Aplysia genome sequencing project should facilitate the identification of other novel late effector molecules 45. Recently, advances have also been made in clarifying the molecular mechanisms that contribute to learning-related synaptic plasticity in the mammalian brain, particularly those that underlie the induction and expression of LTP and LTD 7172. When it comes to transcriptional regulation, however, our current understanding is far from complete. Aplysia was the first organism in which cAMP was shown to play a critical role in learning-related synaptic plasticity 2873. Since signaling cascades that underlie the expression of long-term memory are surprisingly well conserved throughout the species, insights from the molecular studies of Aplysia should provide an important foundation for future studies into the transcriptional regulation of memory formation in the more complex mammalian brain. Authors' contributions: All authors participated in developing the ideas, the writing, discussion and integration of the information. All authors read and approved the final manuscript.

Synaptic remodeling and synaptic growth accompany various forms of long-term memory. Storage of the long-term memory for sensitization of the gill-withdrawal reflex in Aplysia has been extensively studied in this respect and is associated with the growth of new synapses by the sensory neurons onto their postsynaptic target neurons. Recent time-lapse imaging studies of living sensory-to-motor neuron synapses in culture have monitored both functional and structural changes simultaneously so as to follow remodeling and growth at the same specific synaptic connections continuously over time and to examine the functional contribution of these learning-related structural changes to the different time-dependent phases of memory storage. Insights provided by these studies suggest the synaptic differentiation and growth induced by learning in the mature nervous system are highly dynamic and often rapid processes that can recruit both molecules and mechanisms used for de novo synapse formation during development.

SummaryThe time course of the requirement for local protein synthesis in the stabilization of learning-related synaptic growth and the persistence of long-term memory was examined using Aplysia bifurcated sensory neuron-motor neuron cultures. We find that, following repeated pulses of serotonin (5-HT), the local perfusion of emetine, an inhibitor of protein synthesis, or a TAT-AS oligonucleotide directed against ApCPEB blocks long-term facilitation (LTF) at either 24 or 48 hr and leads to a selective retraction of newly formed sensory neuron varicosities induced by 5-HT. By contrast, later inhibition of local protein synthesis, at 72 hr after 5-HT, has no effect on either synaptic growth or LTF. These results define a specific stabilization phase for the storage of long-term memory during which newly formed varicosities are labile and require sustained CPEB-dependent local protein synthesis to acquire the more stable properties of mature varicosities required for the persistence of LTF.

Recent studies of implicit and explicit memory suggest that modulation of synaptic strength and structure is a fundamental mechanism by which these memories are encoded, processed, and stored within the brain. Two model systems have been extensively studied as examples of these two forms of memory: sensitization in the marine snail Aplysia californica as an example of implicit memory, and spatial memory formation in rodents as an example of explicit memory. In this review, we discuss and compare critical synaptic sites and the underlying cellular and molecular mechanisms of short-term, intermediate-term, and long-term memory and consider how the conservation of common elements in each form may contribute to the different temporal phases of both implicit and explicit memory storage.

Synaptic remodeling and synaptic growth accompany various forms of long-term memory. Storage of the long-term memory for sensitization of the gill-withdrawal reflex in Aplysia has been extensively studied in this respect and is associated with the growth of new synapses by the sensory neurons onto their postsynaptic target neurons. Recent time-lapse imaging studies of living sensory-to-motor neuron synapses in culture have monitored both functional and structural changes simultaneously so as to follow remodeling and growth at the same specific synaptic connections continuously over time and to examine the functional contribution of these learning-related structural changes to the different time-dependent phases of memory storage. Insights provided by these studies suggest the synaptic differentiation and growth induced by learning in the mature nervous system are highly dynamic and often rapid processes that can recruit both molecules and mechanisms used for de novo synapse formation during development.

Cellular and molecular studies of both implicit and explicit memory suggest that experience-dependent modulation of synaptic strength and structure is a fundamental mechanism by which these memories are encoded and stored within the brain. In this review, we focus on recent advances in our understanding of two types of memory storage: (i) sensitization in Aplysia, a simple form of implicit memory, and (ii) formation of explicit spatial memories in the mouse hippocampus. These two processes share common molecular mechanisms that have been highly conserved through evolution., Research in this review was supported in part by the Howard Hughes Medical Institute (to ERK), National Institutes of Health grant MH37134 (to CHB) and the Kavli Institute for Brain Sciences. AB is supported by the Marie Curie Excellence grant MEXT-CT-2003-509550.

Studies of sensitization and classical conditioning of the gill-withdrawal reflex in Aplysia have shown that the synaptic connections between identified glutamatergic sensory neurons and motor neurons can be enhanced in one of two ways: by a heterosynaptic (modulatory input-dependent) mechanism that gives rise with repetition to long-term facilitation and by a homosynaptic (activity-dependent) mechanism that gives rise with repetition to a facilitation that is partially blocked by 2-amino-5-phosphonovaleric acid and by injection of 1,2-bis(2-aminophenoxy)ethane- N,N,N′,N′ -tetraacetate (BAPTA) into the postsynaptic cell and is similar to long-term potentiation in the hippocampus. We here have examined how these two forms of facilitation interact at the level of an individual synaptic connection by using a culture preparation consisting of a single bifurcated sensory neuron that forms independent synaptic contacts with each of two spatially separated motor neurons. We find that the homosynaptic facilitation produced by a train of action potentials is cell wide and is evident at all of the terminals of the sensory neuron. By contrast, the heterosynaptic facilitation mediated by the modulatory transmitter serotonin (5-HT) can operate at the level of a single synapse. Homosynaptic activation gives rise to only a transient facilitation lasting a few hours, even when repeated in a spaced manner. The heterosynaptic facilitation produced by a single pulse of 5-HT, applied to one terminal of the sensory neuron, also lasts only minutes. However, when one or more homosynaptic trains of spike activity are paired with even a single pulse of 5-HT applied to one of the two branches of the sensory neuron, the combined actions lead to a selective enhancement in synaptic strength only at the 5-HT-treated branch that now lasts more than a day, and thus amplifies, by more than 20-fold, the duration of the individually produced homo- and heterosynaptic facilitation. This form of synapse-specific facilitation has unusual long-term properties. It does not require protein synthesis, nor is it accompanied by synaptic growth.

In 1894, Ramón y Cajal first proposed that memory is stored as an anatomical change in the strength of neuronal connections. For the following 60 years, little evidence was recruited in support of this idea. This situation changed in the middle of the twentieth century with the development of cellular techniques for the study of synaptic connections and the emergence of new formulations of synaptic plasticity that redefined Ramón y Cajal's idea, making it more suitable for testing. These formulations defined two categories of plasticity, referred to as homosynaptic or Hebbian activity-dependent, and heterosynaptic or modulatory input-dependent. Here we suggest that Hebbian mechanisms are used primarily for learning and for short-term memory but often cannot, by themselves, recruit the events required to maintain a long-term memory. In contrast, heterosynaptic plasticity commonly recruits long-term memory mechanisms that lead to transcription and to synpatic growth. When jointly recruited, homosynaptic mechanisms assure that learning is effectively established and heterosynaptic mechanisms ensure that memory is maintained.

The expression of tissue plasminogen activator (tPA) is increased during activity-dependent forms of synaptic plasticity. We have found that inhibitors of tPA inhibit the late phase of long-term potentiation (L-LTP) induced by either forskolin or tetanic stimulation in the hippocampal mossy fiber and Schaffer collateral pathways. Moreover, application of tPA enhances L-LTP induced by a single tetanus. Exposure of granule cells in culture to forskolin results in secretion of tPA, elongation of mossy fiber axons, and formation of new, active presynaptic varicosities contiguous to dendritic clusters of the glutamate receptor R1. These structural changes are blocked by tPA inhibitors and induced by application of tPA. Thus, tPA may be critically involved in the production of L-LTP and specifically in synaptic growth.

The synaptic growth that accompanies 5-HT-induced long-term facilitation of the sensory to motor neuron connection in Aplysia is associated with the internalization of apCAM at the surface membrane of the sensory neuron. We have now used epitope tags to examine the fate of each of the two apCAM isoforms (membrane bound and GPI-linked) and find that only the transmembrane form is internalized. This internalization can be blocked by overexpression of transmembrane constructs with a single point mutation in the two MAPK consensus sites, as well as by injection of a specific MAPK antagonist into sensory neurons. These data suggest MAPK phosphorylation at the membrane is important for the internalization of apCAMs and, thus, may represent an early regulatory step in the growth of new synaptic connections that accompanies long-term facilitation.

The storage of long-term memory is associated with a cellular program of gene expression, altered protein synthesis, and the growth of new synaptic connections. Recent studies of a variety of memory processes, ranging in complexity from those produced by simple forms of implicit learning in invertebrates to those produced by more complex forms of explicit learning in mammals, suggest that part of the molecular switch required for consolidation of long-term memory is the activation of a cAMP-inducible cascade of genes and the recruitment of cAMP response element binding protein-related transcription factors. This conservation of steps in the mechanisms for learning-related synaptic plasticity suggests the possibility of a molecular biology of cognition.

Perhaps the most striking finding in the biology of memory is that long-term memory involves structural changes. A variety of memory processes ranging in complexity from those produced by elementary forms of nonassociative learning in invertebrates to those produced by higher order associative tasks in mammals are accompanied by alterations in the structure of synaptic connections. This finding raises three interesting conceptual issues. First, how does short-term memory, which involves only covalent modification of pre-existing proteins and an alteration of pre-existing connections, become transformed into a structural change? Second, is the stability of long-term memory achieved because of the stability of synaptic structures? If so, is loss of memory with time reflected in loss of synaptic connections? Finally, how do the structural changes that accompany long-term memory compare with the de novo synapse formation and synaptic pruning that occur during development? In this review we consider the recent evidence for structural changes in learning by focusing on these three issues. We begin by examining

The morphology of growth cones from identified neurons of Aplysia californica was analysed both with video-enhanced contrast differential-interference contrast (VEC-DIC) microscopy, and through serial electron microscopic reconstructions of the same growth cones. The largest structures seen in the living growth cones, the large irregular refractile bodies (LIRBs), were shown in electron micrographs to be unique structures, composed predominantly of dense-core vesicles but including mitochondria and smooth membrane profiles. The LIRBs were stratified in the growth cones, occurring predominantly in sections distant from the substrate and relatively devoid of microtubules. VEC-DIC observations showed that LIRBs formed in the peripheral regions of the organelle-rich central growth cone, and grew in size through fusion with other LIRBs, accumulating into a large central mass in more proximal regions. The distribution of microtubules and LIRBs and the movements of LIRB suggest that there is an overall circulatory pattern in the growth cones, with the delivery of new vesicles occurring at distal areas close to the substrate, and the accumulation and retrograde processing of organelles occurring in more proximal areas away from adhesive contacts.

The notion of an anatomical basis for learning has its roots in ancient Greece and has been restated and refined in modern terminology by numerous contributors 1–4 . A fundamental tenet of these early structural theories was that the basic wiring diagram of the brain could be modified throughout life and that learning was capable of producing continuous structural changes, typically including the growth of novel connections. Prior to the last two decades, evidence for this type of neuronal growth was restricted to embryonic development and regeneration of peripheral tissues. Later studies demonstrated that injury, environmental pressures, and internal body state could also produce structural changes in the adult CNS 5 . More recently it has become clear that the arrangement of synaptic connections in the mature nervous system can undergo striking changes even during normal functioning 6 . As the diversity of species and plastic processes subjected to morphological scrutiny has increased, convergence upon a set of structurally detectable phenomena has begun to emerge. Although several aspects of synaptic structure appear to change with experience, the most consistent potential substrate for memory storage during behavioral modification is an alteration in the number and/or pattern of synaptic connections.

We have used the gill and siphon withdrawal reflex of Aplysia to study the morphological basis of the persistent synaptic plasticity that underlies long-term sensitization. One critical locus for storage of the memory for sensitization is the set of monosynaptic connections between identified siphon sensory neurons and gill and siphon motor neurons. To complement previous morphological studies of the presynaptic terminals of identified sensory neurons, we examined the effects of long-term sensitization on the structure of an identified postsynaptic target--the gill motor neuron L7. We found an increase in the frequency, size, and vesicle complement of presynaptic contacts onto L7 processes in sensitized compared to control animals. Combined, these data indicate a striking increase in the percentage of the surface area of L7 that is occupied by synaptic contacts after long-term training. These results are consistent with our observations that sensitization produces an increase in the synapses that the sensory neurons make on their target cells and provide additional support for the hypothesis that changes in synapse number may represent a mechanism underlying long-term memory.

The morphological consequences of long-term habituation and sensitization of the gill withdrawal reflex in Aplysia california were explored by examining the total number of presynaptic varicosities of single identified sensory neurons (a critical site of plasticity for the biochemical and biophysical changes that underlie both types of learning) in control and behaviorally trained animals. Sensory neurons from habituated animals had 35% fewer synaptic varicosities than did sensory neurons from control animals. In contrast, sensory neurons from sensitized animals had twice as many varicosities per sensory neuron compared to controls, as well as enlarged neuropil arbors. These changes suggest that modulation of synapse number may play a role in the maintenance of long-term memory.

We have used the gill- and siphon-withdrawal reflex of Aplysia californica to explore the morphological basis of the synaptic plasticity that underlies long-term sensitization. In earlier studies (Bailey and Chen, 1983, 1988a), we described 2 classes of structural changes at identified sensory neuron synapses that occur following long- term sensitization: (1) increases in the number, size, and vesicle complement of active zones and (2) an overall increase in the total number of synaptic varicosities per sensory neuron. In the present study, we have begun to examine which of these anatomical changes might be necessary for the maintenance of long-term sensitization by exploring the time course over which they occur and, in particular, their duration relative to the persistence of the memory assessed behaviorally. Toward this end we have quantitated changes in both the total number of varicosities and their active zone morphology in single HRP-labeled sensory neurons taken from long-term sensitized and control animals at different intervals (1–2 d, 1 week, and 3 weeks) following training. We have found that long-term sensitized animals examined within 48 hr after the completion of training demonstrate an increase in the total number of varicosities per sensory neuron as well as an increase in the incidence, size, and vesicle complement of their synaptic active zones compared with control animals. The increase in the number of varicosities and active zones persists unchanged for at least 1 week, and the increase in active zone number is only partially reversed at the end of 3 weeks.(ABSTRACT TRUNCATED AT 250 WORDS)

In a culture system where a bifurcated Aplysia sensory neuron makes synapses with two motor neurons, repeated application of serotonin (5-HT) to one synapse produces a CREB-mediated, synapse-specific, long-term facilitation, which can be captured at the opposite synapse by a single pulse of 5-HT. Repeated pulses of 5-HT applied to the cell body of the sensory neuron produce a CREB-dependent, cell-wide facilitation, which, unlike synapse-specific facilitation, is not associated with growth and does not persist beyond 48 hr. Persistent facilitation and synapse-specific growth can be induced by a single pulse of 5-HT applied to a peripheral synapse. Thus, the short-term process initiated by a single pulse of 5-HT serves not only to produce transient facilitation, but also to mark and stabilize any synapse of the neuron for long-term facilitation by means of a covalent mark and rapamycin-sensitive local protein synthesis.

Summary Neurotrophins control the development and adult plasticity of the vertebrate nervous system. Failure to identify invertebrate neurotrophin orthologs, however, has precluded studies in invertebrate models, limiting our understanding of fundamental aspects of neurotrophin biology and function. We identified a neurotrophin (ApNT) and Trk receptor (ApTrk) in the mollusk Aplysia and found that they play a central role in learning-related synaptic plasticity. Blocking ApTrk signaling impairs long-term facilitation, whereas augmenting ApNT expression enhances it and induces the growth of new synaptic varicosities at the monosynaptic connection between sensory and motor neurons of the gill-withdrawal reflex. Unlike vertebrate neurotrophins, ApNT has multiple coding exons and exerts distinct synaptic effects through differentially processed and secreted splice isoforms. Our findings demonstrate the existence of bona fide neurotrophin signaling in invertebrates and reveal a posttranscriptional mechanism that regulates neurotrophin processing and the release of proneurotrophins and mature neurotrophins that differentially modulate synaptic plasticity. Video Abstract

SummaryNeurexin and neuroligin, which undergo heterophilic interactions with each other at the synapse, are mutated in some patients with autism spectrum disorder, a set of disorders characterized by deficits in social and emotional learning. We have explored the role of neurexin and neuroligin at sensory-to-motor neuron synapses of the gill-withdrawal reflex in Aplysia, which undergoes sensitization, a simple form of learned fear. We find that depleting neurexin in the presynaptic sensory neuron or neuroligin in the postsynaptic motor neuron abolishes both long-term facilitation and the associated presynaptic growth induced by repeated pulses of serotonin. Moreover, introduction into the motor neuron of the R451C mutation of neuroligin-3 linked to autism spectrum disorder blocks both intermediate-term and long-term facilitation. Our results suggest that activity-dependent regulation of the neurexin-neuroligin interaction may govern transsynaptic signaling required for the storage of long-term memory, including emotional memory that may be impaired in autism spectrum disorder.

The switch from short- to long-term facilitation induced by behavioral sensitization in Aplysia involves CREB-like proteins, as well as the immediate-early gene ApC/EBP. Using the bZIP domain of ApC/EBP in a two-hybrid system, we have cloned ApCREB2, a transcription factor constitutively expressed in sensory neurons that resembles human CREB2 and mouse ATF4. ApCREB2 represses ApCREB1-mediated transcription in F9 cells. Injection of anti-ApCREB2 antibodies into Aplysia sensory neurons causes a single pulse of serotonin (5-HT), which induces only short-term facilitation lasting minutes, to evoke facilitation lasting more than 1 day. This facilitation has the properties of long-term facilitation: it requires transcription and translation, induces the growth of new synaptic connections, and occludes further facilitation by five pulses of 5-HT.

The ultrastructure of histaminergic synaptic terminals was studied by the means of intrasomatic injection of horseradish peroxidase into the identified histaminergic neuron C2 of Aplysia . The axonal tree of C2 was found to consist, in part, of varicosities that display putative release sites similar in morphology to those described in other neurons in Aplysia . The varicosities contain at least two populations of vesicles: a conspicuous class of large vesicles with an electron-dense core that almost fills the entire vesicle and a heterogeneous class of large and small electron-lucent vesicles. The small lucent vesicles preferentially cluster near active zones.