Featured Review Cellscience Reviews Vol 2 No.2 ISSN 1742-8130

Toward a Molecular Biology of Learning-Related Synaptic Growth In Aplysia

Craig H. Bailey² , Eric R. Kandel^1,2, Kausik Si³ & Yun-Beom Choi²

Introduction

Recent studies in a variety of memory systems, ranging in complexity from simple forms of implicit memory in higher invertebrates to more complex forms of explicit memory in mammals, suggest that the storage of long-term memory is associated with altered gene expression, the synthesis of new proteins and the growth of new synaptic connections (Kandel, 2001). For both forms of memory storage, the synaptic growth is thought to represent a final cellular change that stabilizes the long-term process (Bailey and Kandel, 1993; Bailey, Bartsch and Kandel, 1996; Bailey, Kandel and Si, 2004).

Despite this association of synaptic growth with various forms of long-term memory, surprisingly little is known about the cellular and molecular mechanisms that give rise to the learning-related structural changes. This in turn raises three questions central to the molecular study of memory storage: (1) How does synaptic activity convert alterations in the strength of pre-existing connections during short-term memory into the growth of new synapses? Does synaptic growth require changes in gene expression? If so, what are the signaling pathways that trigger gene induction, and what are the cellular and molecular requirements for initiation of the growth process? (2) Is the stability of long-term memory, which would seem to require some mechanism that can survive molecular turnover, represented by the relative stability of synaptic structure? If so, what are the molecules that stabilize learning-induced synaptic growth and how does stabilization of these newly formed synapses lead to the persistence of memory storage? (3) Do the molecular mechanisms that regulate the initiation and maintenance of learning-related synapse formation in the adult brain resemble those that govern synaptogenesis and the fine-tuning of synaptic connections during development?

In this review, we address these questions by focusing on molecular and structural studies of long-term memory storage in Aplysia. We begin by examining the remodeling and growth of identified sensory neuron synapses that accompany long-term sensitization - an elementary form of implicit memory. We then turn to recent in vitro studies of the sensory-to-motor neuron synapse reconstituted in dissociated cell culture and consider the molecular mechanisms that underlie these learning-related structural changes and the functional contribution of these changes to the different temporal phases of memory storage. Finally, we outline some of the molecular insights that have been provided by these studies of synaptic growth associated with long-term memory in Aplysia and discuss how they may relate to the mechanisms that regulate de novo synapse formation during development.

Learning-Induced Growth of New Sensory Neuron Synapses Parallels Retention of Long-Term Sensitization

The nervous system of the marine snail Aplysia californica has proven useful as a model system for studying the cellular and molecular bases of learning and memory. It contains approximately 20,000 large, identifiable nerve cells clustered into 10 major ganglia. The ability to identify individual neurons and record their activity has made it possible to define the major components of the neuronal circuits of specific behaviors and to delineate the critical synaptic sites and underlying mechanisms used to store memory-related representations.

The molecular mechanisms contributing to implicit memory storage have been most extensively studied for the gill-withdrawal reflex of Aplysia (Kandel, 2001). As is the case with other types of defensive reflexes, several different forms of implicit learning can modify the gill-withdrawal reflex. We will focus here on sensitization, an elementary form of nonassociative learning by which an animal learns about the properties of a single noxious stimulus. The animal learns to strengthen its defensive reflexes and to respond vigorously to a variety of previously neutral stimuli after it has been exposed to a potentially threatening stimulus. In Aplysia, sensitization of the gill-withdrawal reflex can be induced by a strong stimulus applied to the tail. This activates facilitatory interneurons that synapse on identified sensory neurons to strengthen the synaptic connection between the sensory neurons and their target motor neurons. The behavioral memory for sensitization of the gill-withdrawal reflex is graded and retention is proportional to the number of training trials. A single stimulus to the tail gives rise to short-term sensitization lasting minutes to hours. Repetition of this stimulus produces long-term sensitization that can last for days or weeks (Frost et al., 1985).

Short- and long-term sensitization lead to enhanced synaptic transmission at the monosynaptic connection between identified mechanoreceptor sensory neurons and motor neurons. Although this component accounts for only part of the behavioral modification measured in the intact animal, its simplicity has facilitated the cellular and molecular analysis of both the short- and long-term forms of synaptic plasticity. The monosynaptic pathway can be reconstituted in dissociated cell culture in which serotonin (5-hydroxytryptamine; 5-HT), a modulatory neurotransmitter normally released by sensitizing stimuli, can substitute for the tail shock used during behavioral training in the intact animal (Montarolo et al., 1986). In parallel to behavioral sensitization, a single application of 5-HT produces short-term changes in synaptic effectiveness, whereas five spaced applications given over a period of 1.5 h produce long-term changes lasting 1 or more days.

Biophysical studies of this monosynaptic connection suggest that both the similarities and the differences in memory reflect, at least in part, intrinsic cellular mechanisms of the nerve cells participating in memory storage. Thus, studies of the connections between sensory and motor neurons in both the intact animal and in cells in culture indicate that the long-term changes are surprisingly similar to the short-term changes. A component of the increase in synaptic strength observed during both the short- and long-term changes is due, in each case, to enhanced release of transmitter by the sensory neuron, accompanied by an increase in the excitability of the sensory neuron, attributable to the depression of specific sets of potassium channels (Frost et al., 1985; Montarolo et al., 1986; Dale et al., 1987; Klein and Kandel, 1980; Hochner et al., 1986; Scholz and Byrne, 1987).

Despite this phenotypic similarity, the short-term changes differ fundamentally from the long-term changes in two important ways. First, the short-term change involves only covalent modification of preexisting proteins and an alteration of preexisting connections. Both short-term behavioral sensitization in the animal and short-term facilitation in dissociated cell culture do not require ongoing macromolecular synthesis: the short-term change is not blocked by inhibitors of transcription or translation (Schwartz et al., 1971; Montarolo et al., 1986). By contrast, these inhibitors selectively block the induction of the long-term changes both in the semi-intact animal (Castellucci et al., 1989) and in primary cell culture (Montarolo et al., 1986). Most striking is the finding that the induction of long-term facilitation at this single synapse in Aplysia exhibits a critical time window in its requirement for protein and RNA synthesis characteristic of that necessary for other forms of learning in both vertebrates and invertebrates. From a molecular perspective, these studies indicate that both the long-term behavioral and the long-term cellular changes require the expression of new genes unlike the short term changes.

Second, as described below, the long-term but not the short-term process involves a structural change. Bailey and Chen (1983; 1988a,b; 1989) first demonstrated that long-term sensitization training is associated with the growth of new synaptic connections by the sensory neurons onto their follower cells. This synaptic growth can be induced in the intact ganglion by the intracellular injection of cAMP, a second messenger activated by 5HT (Nazif et al., 1991) and can be reconstituted in sensory-motor neuron co-cultures by repeated presentations of 5HT (Glanzman et al., 1990; Bailey et al., 1992).

To examine the morphological basis of the synaptic plasticity that may underlie the transition from short-term to long-term memory, Bailey and Chen combined selective intracellular labeling techniques with the analysis of serial thin sections and used transmission electron microscopy to study complete reconstructions of unequivocally identified sensory neuron synapses from both control and behaviorally modified animals. They found the storage of long-term memory for sensitization (lasting several weeks) was accompanied by a family of distinct structural changes at identified sensory neuron synapses. These changes reflected a learning-induced remodeling of the functional architecture of presynaptic sensory neuron varicosities at two different levels of synaptic organization: (a) alterations in focal regions of membrane specialization of the synapse that mediate transmitter release—the number, size, and vesicle complement of sensory neuron active zones were larger in sensitized animals than controls (Bailey and Chen 1983; 1988b) and (b) a parallel but more pronounced and widespread effect involving modulation of the total number of presynaptic varicosities per sensory neuron (Bailey and Chen, 1988a). Sensory neurons from long-term sensitized animals exhibited a two-fold increase in the total number of synaptic varicosities, as well as an enlargement in the size of each neuron’s axonal arbor (Figure 1).

Figure 1. Serial reconstruction of sensory neurons from long-term sensitized and control animals. Total extent of the neuropil arbors of sensory neurons from one control and two long-term-sensitized animals are shown. In each case, the rostral (row 3) to caudal (row 1) extent of the arbor is divided roughly into thirds. Each panel was produced by the super-imposition of camera lucida tracings of all horseradish peroxidase-labeled processes present in 17 consecutive slab-thick sections and represents a linear segment through the ganglion of roughly 340 µm. For each composite, ventral is up, dorsal is down, lateral is to the left, and medial is to the right. By examining images across each row (rows 1, 2, and 3), the viewer is comparing similar regions of each sensory neuron. In all cases, the axonal arbor of long-term-sensitized cells is markedly increased compared to control and parallels the concomitant increase in the number of sensory neuron presynaptic varicosities. (From Bailey and Chen, 1988a)

To determine which class of structural changes at sensory neuron synapses might contribute to the retention of long-term sensitization, Bailey and Chen (1989) compared the time course for each morphological change with the behavioral duration of the memory. They found that not all of the structural changes persisted as long as the memory. The increase in the size and synaptic vesicle complement of sensory neuron active zones which are present 24 hr following the completion of behavioral training was back to control levels when tested 1 week later. These data indicated that, insofar as modulation of active zone size and associated synaptic vesicles is one of the structural mechanisms underlying long-term sensitization, it is associated with the expression and initial maintenance of the long-term process but not with its persistence. By contrast, the duration of changes in varicosity and active zone number, which persisted unchanged for at least one week and were only partially reversed at the end of three weeks, paralleled the behavioral time course of memory, indicating that only the changes in the number of sensory neuron synapses contribute to the long-term retention of sensitization.

These findings demonstrated that clear structural changes could accompany long-term behavioral modifications in Aplysia and indicated, for the first time, that these changes could be detected at the level of identified synaptic connections known to be critically involved in the behavior. Results from these initial studies of structural synaptic plasticity also provided evidence for an intriguing notion – that active zones are plastic rather than immutable components of the synapse, and that even elementary forms of learning can alter the organization and number of transmitter release sites to modulate the functional expression of synaptic connections. In addition, these studies indicated the growth of new sensory neuron synapses may represent the final and perhaps most stable phase of long-term memory storage and suggested that the stability of the long-term process may be achieved, at least in part, because of the relative stability of synaptic structure.

This long-lasting growth of new synaptic connections between sensory neurons and their follower cells (both interneurons and motor neurons) during long-term sensitization can be reconstituted in dissociated sensory-motor neuron co-cultures by repeated applications of 5-HT (Glanzman et al., 1990; Bailey et al., 1992b). In culture, the structural change can be correlated with the long-term (24 hr) enhancement in synaptic effectiveness and depends upon the presence of an appropriate target cell similar to the synapse formation that occurs during development. The nature of this interaction between the presynaptic cell and its target is not known. The signal from the postsynaptic neuron may be cell-associated, such as a constituent of the motor cell’s membrane, or diffusible, perhaps being released locally from the motor cell’s processes.

Long-Term Facilitation is Associated with Activation of Silent Presynaptic Varicosities and Growth of New Functional Synaptic Varicosities

All of the studies of sensory to motor neuron connections mentioned above, during both long-term sensitization in the behaving animal (Bailey and Chen 1983; 88a,b; 89) as well as long-term facilitation in culture (Glanzman et al., 1990; Bailey et al., 1992b), were done on populations of sensory neuron varicosities. Thus, they could not follow structural changes at the same specific synaptic varicosities continuously over time, and did not examine the functional contribution of these presynaptic structural changes to the different time-dependent phases of 1ong-term facilitation. As a result, these earlier studies could not determine whether the increase in synaptic strength resulted from the conversion of pre-existing but non-functional (silent) synapses or from the addition of newly formed synapses or both.

To examine these alternative possibilities, and as a first step in exploring the molecular mechanisms that contribute to learning-induced structural changes, Kim et al. (2003) combined time-lapse confocal imaging of individual presynaptic varicosities of sensory neurons labeled with three different fluorescent markers: the whole cell marker Alexa-594, and two presynaptic marker proteins: synaptophysin-eGFP which monitors changes in the distribution of synaptic vesicles within individual varicosities and synapto-PHluorin, a monitor of active transmitter release sites (Miesenbock et al., 1998). They found that repeated pulses of 5-HT induce two temporally, morphologically, and molecularly distinct classes of presynaptic changes: 1) a rapid activation of silent presynaptic terminals through the filling of pre-existing empty varicosities with synaptic vesicles, which requires translation but not transcription and 2) a generation of new synaptic varicosities which occurs more slowly and requires both transcription and translation. The enrichment of pre-existing but empty varicosities with synaptophysin is completed within 3 to 6 hr, parallels intermediate-term facilitation and accounts for approximately 32% of the newly activated synapses evident at 24 hr. By contrast, the new sensory neuron varicosities, which account for 68% of the newly activated synapses at 24 hr, do not form until 12-18 hr after exposure to 5 pulses of 5-HT (Figure 2). The rapid activation of silent presynaptic terminals suggests that in addition to its role in long-term facilitation, this modification of pre-existing synapses may also contribute to the intermediate phase of synaptic plasticity and memory storage (Ghirardi et al., 1995; Mauelshagen et al., 1996; Sutton et al., 2001).

Figure 2. Time course of 5-HT-induced filling of pre-existing empty sensory neuron varicosities and the growth of new varicosities. (A) Clustering of synaptophysin-GFP into pre-existing empty varicosities begins almost immediately (0.5hr) after 5 x 10 µM 5-HT. Accumulation of synaptophysin-GFP at these varicosities is completed within 3-6 hr and remains relatively stable for 24 hr. The time course of the mean enrichment of empty varicosities after 5-HT treatment and representative time-lapse images of the filling of a pre-existing empty varicosity are shown at each time point. (B) Time course of the mean enrichment and the change in the number of new varicosities are shown at the same time points as in (A) after normalization with the maximum number of new varicosities at 24 hr. Most of the new varicosities are formed between 12 and 18 hr after 5-HT treatment and this process is preceded by enrichment of some varicosities with synaptophysin. Time-lapse images indicate that synaptophysin-GFP accumulates in a pre-existing axonal swelling and is followed by either the splitting and subsequent division of this swelling or the budding off of synaptic vesicles and associated active zone material from the original varicosity resulting in the appearance of two new synaptophysin-enriched varicosities between 12 and 18 hr after 5-HT treatment. These two varicosities (both the original and the newly formed ) are still enriched and present 24 hr after 5-HT treatment. (Modified from Kim et al., 2003)

In this study, Kim and colleagues employed a reduced 5-HT protocol to selectively induce facilitation in the intermediate-term time domain without inducing long-term facilitation (Ghirardi et al., 1995). They found that isolated intermediate-term facilitation was also accompanied by the redistribution and clustering of synaptic vesicle proteins into empty sensory neuron varicosities at 0.5 hr and 3 hr similar to that which occurred when intermediate- and long-term facilitation were recruited together. However, the presynaptic structural changes induced by the reduced 5-HT protocol differed from those induced by long-term training in at least two ways. First, there was no growth of new sensory neuron varicosities in the isolated intermediate phase. Second, unlike the filling of pre-existing empty varicosities during the intermediate-term phase induced by the long-term protocol, the newly-filled varicosities did not persist for 24 hr and were unaffected by inhibitors of protein synthesis suggesting that the structural remodeling induced by the reduced 5-HT protocol involved only a simple rearrangement of pre-existing synaptic components (Figure 3). This may reflect a fundamental difference in the molecular mechanisms recruited by the two 5-HT protocols. Although both protocols induce intermediate-term facilitation, the long-term protocol may activate additional molecular events (including the machinery for translational activation) required to set up the long-term phase, perhaps by stabilizing the intermediate phase. At present it is not known how the covalent modifications that lead to the rearrangement of pre-existing synaptic proteins at empty varicosities is converted by the long-term protocol to a more stable, protein-synthesis dependent process.

Figure 3. Time course of facilitation and synaptophysin-GFP enrichment following the selective induction of facilitation in the intermediate-term time domain. Facilitation and clustering of synaptophysin-GFP into pre-existing empty varicosities are indicated following selective induction of intermediate-term facilitation isolated from long-term facilitation by a reduced 4x10nM 5-HT training protocol. Only intermediate-term facilitation is induced without the onset of long-term faciliation. The mean percentage of pre-existing empty varicosities filled at the indicated time points are shown. Representative time-lapse images of two pre-existing empty varicosities are presented before the start and after the completion of four pulses of 10nM 5-HT. Merged images (yellow) of Alexa-594(red) and synaptophysin-GFP (green) are presented. Although the reduced 5-HT protocol clearly induces synaptophysin enrichment during the intermediate-term time domain (0.5 and 3 hr), this clustering of synaptic vesicles does not persist and is not evident at 24 hr. (Modified from Kim et al., 2003)

In most model systems used to study long-lasting forms of synaptic plasticity, the functional contribution of the structural changes that accompany memory storage remains largely unknown. In particular one would like to know if changes in the number or structure of synaptic connections induced by learning are functionally effective and capable of contributing to the storage of long-term memory. The synPH data of Kim et al. (2003) represents an important step in addressing this issue (Figure 4). Previous studies have shown that specific 5-HT protocols (Casadio et al., 1999) or experimental manipulation in Aplysia (Hatada et al., 2000; Udo et al., 2005) can induce long-term facilitation at 24 hr without the formation of new varicosities. How might such an increase in synaptic strength persist for 24 hr in the absence of synaptic growth? The results of Kim and associates suggest that additional modifications of pre-existing connections including the activation of previously silent synapses, may play a role in the initial phases of synaptic maintenance and highlight the fact that there are likely to be multiple types of structural mechanisms that can contribute to long-term facilitation at 24 hr (Schacher et al., 1997; Sutton and Carew, 2000; Bailey et al., 2000).

Figure 4. 5-HT-induced activation of pre-existing silent varicosities and the growth of new functional synaptic varicosities. The functional state of individual sensory neuron varicosities as determined before and 24 hr after 5 x 10 µM 5-HT. (A) The merged images (red, Alexa-594; green, synPH) reveal that a pre-existing empty varicosity lacking synPH (red) at -3 hr becomes enriched (yellow) at 24 hr after 5-HT treatment. The pseudocolor images before (rest) and after (stim) depolarization of the sensory neuron indicate that there is no significant change in fluorescence intensity at -3 hr (presynaptically silent and not competent for evoked transmitter release) but illustrate a significant increase in fluorescence intensity (presynaptically active and competent for evoked transmitter release) 24 hr after 5-HT treatment. (B) Only sensory neuron axonal processes are present at -3 hr, but a new varicosity is formed and enriched in synPH (yellow) at 24 hr after 5-HT treatment. The pseudocolor images show an increase in fluorescence intensity, indicating that the newly formed presynaptic varicosity is functional. (C) A pre-existing and synPH-enriched varicosity is competent both before and after 5-HT treatment. There is no substantial change in varicosity structure or synPH distribution. The pseudocolor images also indicate that the varicosity is functional at both –3 hr and 24 hr following 5-HT treatment. The pseudocolor scale shows fluorescence intensity of synPH (in arbitrary fluorescence units) for rest/stim panels of (A)-(C). (Modified from Kim et al., 2003)

Of the two classes of presynaptic structural plasticity induced by 5-HT in culture, synaptic growth appears to contribute more to the synaptic enhancement present at 24 hr than does the activation of pre-existing silent synapses. It will be of interest to see if the functional contribution by newly formed synapses increases with time when the growth process is more fully developed and memory storage is likely to be more stable. This would be consistent with the earlier studies in the intact animal outlined above, which have shown that only the increases in the number of sensory neuron varicosities and active zones persist for several weeks in parallel with the behavioral duration of the memory, as well as more recent work in culture which has demonstrated that synaptic growth plays a more prominent role in the expression of the later phases of long-term facilitation (Martin et al., 1997b; Casadio et al., 1999).

5-HT-Induced Regulation of the Presynaptic Actin Network: a Nodal Point for Learning-Related Synapse Formation

The 5-HT-induced enrichment of synaptic vesicle proteins and concomitant recruitment of active zone components in both pre-existing and newly formed sensory neuron synapses during long-term facilitation is likely to involve an activity-dependent rearrangement of the actin cytoskeleton. For example, the remodeling of synapses in response to physiological activity requires reorganization of the actin network (Colicos et al., 2001; Huntley, Benson, and Colman, 2002) and the inhibition of actin function blocks synapse formation and interferes with long-term synaptic plasticity (Hatada et. al., 2000; Krucker, Siggins, and Halpain, 2000; Zhang and Benson, 2001). Furthermore, several synaptic proteins such as synapsin can bind to the actin cytoskeleton and modulate the trafficking of synaptic vesicles (Humeau et al., 2001).

How do repeated applications of 5-HT lead to reorganization of the actin cytoskeleton? The balance between actin polymerization and depolymerization is tightly regulated by extracellular signaling molecules, many of which act through the Rho family of GTPases. These small GTPases are thought to participate at different stages during the development of the central nervous system, for example, in the establishment of polarity, axon guidance, dendritic growth and maintenance of dendritic spines (Yuan et al., 2003; Sin et al., 2002; Nakayama et al., 2000; Bradke et al., 1999; Threadgill et al., 1997).

In Aplysia, the application of toxin B, a general inhibitor of the Rho family, blocks 5-HT-induced long-term facilitation, as well as growth of new synapses in sensory-motor neuron co-cultures. Moreover, repeated pulses of 5-HT selectively induce the spatial and temporal regulation of the activity of one of the small GTPases, Cdc42 leading to a rearrangement of the presynaptic actin network and the assembly, insertion and functional maturation of active transmitter release sites at sensory neuron varicosities. The 5-HT activation of ApCdc42 is dependent on signaling through the P13K and PLC pathways and in turn ApCdc42 activates the downstream effectors PAK (p21-Cdc42/Rac-activated kinase) and N-WASP (neuronal Wiskott-Aldrich syndrome protein) (Fig. 5, Udo et al., 2005).

The activation of ApCdc42 in sensory neurons leads to the outgrowth of filopodia from presynaptic varicosities. Interestingly, 5-HT stimulation by itself naturally induces filopodia, which is dependent on the activation of ApCdc42. Filopodia have been proposed to be a morphological precursor of dendritic spines in the mammalian nervous system, and this process may be regulated by neuronal activity (Jontes and Smith, 2000). The 5-HT-induced activation of CdC42 in Aplysia triggers not only the formation of filopodia but also the molecular maturation of new transmitter release sites. A major synaptic vesicle protein, synaptophysin, accumulates at the tips of 5-HT-induced filopodia, some of which then give rise to new varicosities (Figure 5D). These observations support the following ideas: (1) filopodia represent one of the morphological precursors for the growth of new presynaptic varicosities during learning-related synaptic plasticity, and (2) the formation of filopodia and initial assembly of the presynaptic compartment can be induced by the activation of Cdc42. Thus, 5-HT-induced regulation of the Cdc42 signaling pathways and the consequent reorganization of the presynaptic actin network appear to be a part of the initial molecular cascade required for the growth of new sensory neuron varicosities associated with the storage of long-term memory.

Figure 5. Temporal and spatial activation of ApCdc42 at individual sensory neuron varicosities. A. FRETc images of a sensory neuron expressing both CFP-ApCdc42 and YFP-N-WASP-CRIB. Images are pseudo-colored to indicate the relative activity of ApCdc42. Sensory neuron varicosities on the initial segment (red arrows, such as a and b) or on the minor processes (green arrows, such as c) of the motor neuron were monitored simultaneously. 5-HT stimulation selectively induced the activation of ApCdc42 at varicosities on the initial segment of the motor neuron. B. Temporal- and spatial-specific activation of ApCdc42. Relative changes of FRETc signals of sensory neuron varicosities at both on-target or off-target sites. Individual varicosities were analyzed at different time points. C. Activation of ApCdc42 at filopodia and at new varicosities. The increased activity of ApCdc42 was often found at the tips of filopodia and newly formed varicosities. White, green and red arrowheads indicate filopodia, tips of filopodia and newly formed varicosities, respectively. D. Time course of 5-HT-induced formation of sensory neuron filopodia and new presynaptic varicosities. The graph illustrates a cumulative plot for the formation of filopodia and varicosities in which the values at -3 hr and 24 hr were normalized to 0 and 100, respectively. The 5-HT induction of filopodia precedes the formation of new sensory varicosities, consistent with their role as a morphological precursor for the generation of some new presynaptic sites. Filopodia formed between 3 hr and 12 hr after 5-HT stimulation, whereas new varicosities appeared between 9 hr and 18 hr. This time course for the formation of filopodia is similar to that of ApCdc42 activation suggesting that ApCdc42 is activated at the initial stages of new varicosity formation. The time-lapse images illustrate that 5-HT treatment induces two distinct types of new sensory neuron varicosity formation: one (top) is mediated by the interstitial outgrowth of filopodia from pre-existing varicosities and the other (bottom) is dependent on an apparent division of or a budding off process from a pre-existing transmitter release site. In both cases, the formation of new varicosities occurred 9-18 hr after 5-HT stimulation, which is approximately 6 hr after the formation of filopodia. Sensory neurons expressing synaptophysin-GFP (green) were injected with Alexa-594 (red). A pre-existing varicosity (red arrowhead) developed a filopodium (white arrowhead) after 5-HT stimulation. Accumulation of synaptophysin was observed at the tip of the filopodium (blue arrowhead). Numbers indicate hours. E. Spatial-specific activation of sensory neuron varicosities. SynaptopHluorin (synPH) was used to monitor the activity of sensory neuron varicosities formed at target or off-target sites of the motor neuron L7. A merged image of phase contrast and synPH fluorescence (left) illustrates the spatial distribution of varicosities at target (A) and off-target sites (B), and pseudo-colored subtracted images (right) indicate their respective synPH activities before and after 0.2 M KCl stimulation for 1 min. The percentage change of synPH signals is shown in the graph. F. Inhibition of PLC or PI3K blocks the activation of ApCdc42 at varicosities. Cultures were incubated with KT5720 (10 mM), Wortmannin (200 nM) and U73122 (1 mM) and the percentage changes of FRETc at 6 hr were compared. Similar results were obtained by the Cdc42 pull-down assay, where Wortmannin and U73122 blocked the activation of ApCdc42. G. Inhibition of LTF and the growth of new sensory neuron varicosities by PLC and PI3K and PKA inhibitors. (Modified from Udo et al., 2005)

Initiation of Long-Term Facilitation Requires Activation of cAMP-Responsive Genes and Coordinated Recruitment of CREB-Related Transcription Factors

Repeated presentations of 5-HT to sensory-motor co-cultures produce a long-term enhancement in transmitter release. 5-HT binds to cell surface receptors on the sensory neurons that activate the enzyme adenylyl cyclase, which converts ATP to the diffusible second-messenger cAMP, thereby activating the cAMP-dependent protein kinase (PKA). How PKA participates in long-term facilitation was illustrated by Bacskai and associates (1993) who monitored the subcellular localization of the free PKA catalytic subunit and found that with one pulse of 5-HT, which produces short-term facilitation, the catalytic subunit was restricted to the cytoplasm. With repeated pulses of 5-HT, which induce long-term facilitation, the catalytic subunit translocated to the nucleus, where it phosphorylates transcription factors and thereby regulates gene expression. Both cAMP and PKA are essential components of the signal-transduction pathway for consolidating memories, not only in Aplysia but also for certain types of memory in Drosophila and mammals. Several olfactory learning mutants in Drosophila map to the cAMP pathway (Davis, 1996; Davis et al., 1995; Drain, Folkers, and Quinn, 1991), indicating that blocking PKA function blocks memory formation in flies. In parallel, the late but not the early phase of LTP of the CA3-to-CAl synapse in the hippocampus is impaired by pharmacological or genetic interference with PKA (Abel et al., 1997; Frey, Huang, and Kandel, 1993; Huang, Lin, and Kandel, 1994).

These studies suggest that the long-term regulation of transmitter release requires PKA-related gene activation. PKA activates gene expression by the phosphorylation of transcription factors that bind to the cAMP-responsive element (CRE). One of the major transcription factors that recognize the CRE is a protein called CRE-binding protein (CREB1), which functions as a transcriptional activator only after it is phosphorylated by PKA or another second messenger kinase. Microinjection of CRE containing oligonucleotides into sensory neurons inhibits the function of CREB1 and blocks long-term facilitation but has no effect on the short-term process (Dash et al., 1990). Not only is CREB1 activation necessary for long-term facilitation, it is also sufficient to induce long-term facilitation, albeit in reduced form and in a form that is not maintained beyond 24 hrs. Thus, sensory cell injection of recombinant CREB1a phophorylated in vivo by PKA led to an increase in EPSP amplitude at 24 hours in the absence of any 5-HT stimulation (Bartsch et al., 2000).

Bartsch and associates (1995) have found that the genetic switch that converts short- to long-term facilitation is not only composed of the CREB1 regulatory unit but also another member of the CREB gene family, ApCREB2, a CRE-binding transcription factor constitutively expressed in sensory neurons. ApCREB2 resembles human CREB2 and mouse ATF4 (Hai et al., 1989; Karpinski et al., 1992), and functions as a repressor of long-term facilitation. Thus, injection of anti-ApCREB2 antibodies into Aplysia sensory neurons causes a single pulse of 5-HT, which normally induces only short-term facilitation lasting minutes, to evoke facilitation that lasts more than 1 day. This response requires both transcription and translation and is accompanied by the growth of new synaptic connections.

These studies reveal that long-term synaptic changes are governed by both positive and negative regulators, and suggest the transition from short-term facilitation to long-term facilitation requires the simultaneous removal of transcriptional repressors and activation of transcriptional activators. These transcriptional repressors and activators can interact with each other both physically and functionally. It is likely that the transition is a complex process involving temporally distinct phases of gene activation, repression, and regulation of signal transduction. The balance between CREB activator and repressor isoforms is also critically important in long-term behavioral memory, as first shown in Drosophila. Expression of an inhibitory form of CREB (dCREB-2b) blocks long-term olfactory memory but does not alter short-term memory (Yin et al., 1994). Overexpression of an activator form of CREB (dCREB-2a) increases the efficacy of massed training in long-term memory formation (Yin et al., 1995). The CREB regulatory unit can be modulated by a number of kinases and therefore serve to integrate signals from various signal transduction pathways. This ability to integrate signaling, as well as to mediate activation or repression, may explain why CREB is so central to memory storage (Martin and Kandel, 1996).

To examine directly the role of CREB-mediated responses in the integration of long-term synaptic activity, Guan et al. (2002) focused on the interactions of two opposing modulatory transmitters, 5-HT and FMRFamide, important for behavioral modification in Aplysia These opposing inputs are integrated in the neuron’s nucleus and are evident in the repression of the CAAT-box-enhancer-binding protein (C/EBP), a transcription regulator critical for long-term facilitation. Whereas 5-HT induces C/EBP by activating CREB1 and recruiting CREB-binding protein to acetylate histones, FMRFamide displaces CREB1 with CREB2 and recruits histone deacetylase to deacetylate histones. When 5-HT and FMRFamide are given together, FMRFamide overrides 5-HT by recruiting CREB2 and the deacetylase to displace CREB1 and CBP, thereby inducing histone deacetylation and repression of C/EBP. Thus, the facilitatory and inhibitory modulatory transmitters that are important for long-term memory in Aplysia activate signal transduction pathways that can alter nucleosome structure bidirectionally through acetylation and deacetylation.

As outlined above, the conversion of short- to long-term memory requires not only the activation of CREB1 but also the removal of the repressive action of CREB2, which lacks consensus sites for PKA phosphorylation (Bartsch et al., 1995). CREB2 does, however, have two conserved sites for MAP kinase (MAPK) phosphorylation. MAPK is activated by 5-HT in Aplysia neurons and, like PKA, MAPK translocates to the nucleus with prolonged 5-HT treatment (Martin et al., 1997a; Michael et al., 1998). The involvement of MAPK in long-term plasticity may be quite general: Martin and associates (1997a) found that cAMP also activated MAPK in mouse hippocampal neurons, suggesting that MAPK may play a role in hippocampal long-term potentiation. The requirement for MAPK during hippocampal LTP has been shown by English and Sweatt, (1996, 1997), who demonstrated that ERK1 is activated in CAl pyramidal cells during LTP and that bath application of MAPK kinase inhibitors blocks LTP.

Consolidation of Long-Term Facilitation Requires the Induction of Immediate-Early Genes

The activation of adenylyl cyclase, the increase in cAMP concentration with the resultant dissociation of the catalytic subunit of PKA and its translocation to the nucleus, as well as the phosphorylation of CREB are all unaffected by inhibitors of protein synthesis. Where then does the protein synthesis-dependent step that characterizes the consolidation phase appear? Clearly, it must require an additional step - the activation of genes by CREB. To follow further the sequence of steps whereby CREB leads to the stable, self-perpetuation of the long-term process, Alberini and colleagues (1994) characterized the intermediary, immediate-early genes induced by cAMP and CREB. In a search for possible cAMP-dependent regulatory genes that might be interposed between constitutively expressed transcription factors and stable effector genes, Alberini and colleagues (1994) focused on the CCAAT-box-enhancer-binding protein (C/EBP) transcription factors. They cloned an Aplysia C/EBP homologue (ApC/EBP) and found that its expression was induced by exposure to 5-HT. Induction of ApC/EBP mRNA is rapid and transient, and occurs in the presence of protein synthesis inhibitors, characteristic for immediate-early genes. Inhibition of ApC/EBP activity blocked long-term facilitation but had no effect on short-term facilitation. Thus, the induction of ApC/EBP seems to serve as an intermediate component of a molecular switch activated during the consolidation period.

The existence of C/EBP, a cAMP-regulated immediate-early gene that is itself a transcription factor and regulates other genes, leads to a model of sequential gene activation. CREB1a, CREB1b, CREB1c, and CREB2 represent the first level of control, since all are constitutively expressed. Stimuli that lead to long-term facilitation perturb the balance between CREB1-mediated activation and CREB2-mediated repression, through the action of PKA, MAPK, and possibly other kinases. This leads to the up-regulation of a family of immediate-early genes. Some of these immediate-early genes are transcription factors such as C/EBP; others are effectors, such as ubiquitin hydrolase, that contribute to consolidation by either extending the inducing signal or initiating the changes at the synapse that cause long-term facilitation.

5-HT-Induced Internalization of ApCAM: a Preliminary and Permissive Step in Learning-Related Synaptic Growth

How does this sequential gene induction lead to the growth of new sensory neuron synapses that accompany long-term facilitation in Aplysia? To address this issue, Mayford et al. (1992) focused on four proteins, D1-D4, which decrease their expression in a transcriptionally-dependent manner following the application of 5HT or cAMP and found that they encoded different isoforms of an immunoglobulin-related cell adhesion molecule, designated ApCAM, which is homologous to NCAM in vertebrates and Fasciclin II in Drosophila. Imaging of fluorescently labeled MAbs to ApCAM revealed that not only is there a decrease in the level of expression but that even preexisting protein is lost from the surface membrane of the sensory neurons within one hour after the addition of 5HT (Mayford et al., 1992).

To examine the subcellular mechanisms by which 5HT modulates ApCAM and the possible significance this modulation might have for the synaptic growth that is induced by 5HT, Bailey et a1. (1992a) combined thin-section electron microscopy with immunolabeling using a gold-conjugated monoclonal antibody specific to ApCAM. Down-regulation of ApCAM was particularly prominent at adherent axonal processes of the sensory neurons and was achieved by a protein-synthesis-dependent activation of the clathrin-mediated endosomal pathway, leading to internalization and apparent degradation of ApCAM. As is the case for the down-regulation at the level of expression, the 5HT-induced internalization of ApCAM can be simulated by cAMP. Concomitant with the internalization of ApCAM, Hu et a1. (1993) found that, as part of this coordinated program for endocytosis, 5HT and cAMP also induce an increase in the number of coated pits and coated vesicles in the sensory neurons and an increase in the expression of the light chain of clathrin (ApClathrin). Since the ApClathrin light chain contains the important functional domains of both LCa and LCb of mammalian clathrin thought to be essential for the coated pit assembly and disassembly, the increase in clathrin is likely to be an important component in the learning-induced activation of the endocytic cycle required for the internalization of ApCAM. The 5-HT-induced, clathrin-mediated decrease in ApCAM is thought to have at least two major structural consequences: 1) disassembly of homophilically associated fascicles of the sensory neurons (defasciculation), a process that may destabilize adhesive contacts normally inhibiting growth, and 2) endocytic activation that may lead to a redistribution of membrane components to sites where new synapses form.

Aplysia neurons express two isoforms of ApCAM - a membrane form (TM) and a phosphoinositol-linked form (GPI). Only the transmembrane isoform is internalized following exposure to 5-HT (Bailey et al., 1997). This internalization can be blocked by overexpression of transmembrane ApCAM with a point mutation in the two MAPK phosphorylation consensus sites, as well as by injection of a specific MAPK antagonist into sensory neurons. These data suggest that activation of the MAPK pathway is important for the internalization of ApCAMs and may represent one of the initial and perhaps permissive stages of learning-related synaptic growth in Aplysia. Furthermore, the combined actions of MAPK both at the membrane and in the nucleus suggest that MAPK plays multiple roles in long-lasting synaptic plasticity and appears to regulate each of the two distinctive processes that characterize the long-term process: activation of transcription and growth of new synaptic connections.

Is the down-regulation of the transmembrane isoform of ApCAM required for the synaptic growth that accompanies 5-HT-induced long-term facilitation? To address this question, Han et al. (2004) overexpressed various HA-epitope tagged recombinant ApCAM molecules in Aplysia sensory neurons and investigated their effects on long-term facilitation. Overexpression of the TM isoform, but not the GPI-linked isoform of ApCAM , blocked both the long-term facilitation, as well as 5-HT-induced synaptic growth. This inhibition of long-term facilitation was restored by interrupting the adhesive function of ApCAM with the anti-HA antibody. In addition, long-term facilitation was completely blocked by the overexpression of the cytoplasmic tail portion of ApCAM fused with GFP, designed to bind proteins such as MAP kinase p42 that might normally bind to the cytoplasm of the sensory neuron. Overall, these studies provide additional evidence that the extracellular domain of transmembrane ApCAM has an inhibitory function that must be neutralized by internalization to induce long-term facilitation and suggest that the cytoplasmic tail may provide an interactive platform for both signal transduction and the internalization machinery.

Stabilization of Learning-Related Synaptic Growth for Persistence of Long-Term Facilitation: Local Activation of Protein Synthesis and Translational Regulators

The stability of long-term facilitation in Aplysia requires both the activation of a nuclear program and the persistence of learning-related synaptic growth. These findings raise an important question: Are all long-term changes associated with memory storage necessarily cell-wide or can they be restricted to some synapses and not others?

To address this question, Martin et al., (1997b) developed a culture system in Aplysia, where a single bifurcated sensory neuron of the gill withdrawal reflex was plated in contact with two spatially separated gill motor neurons and focused on the role of local protein synthesis in the maintenance of synapse-specific long- term plasticity. In this culture system, application of a single pulse of 5-HT to one of the two sets of synapses, results in a synapse-specific short-term facilitation of preexisting connections that lasts for minutes. Five pulses of 5-HT, designed to simulate the spaced training that leads to long-term memory, elicit a synapse-specific long-term facilitation that lasts for three or more days. Whereas the short-term form does not require new protein synthesis, the long-term form requires both CREB-dependent transcription and local protein synthesis at activated synapses, and is accompanied by the growth of new sensory neuron synapses. This long-term facilitation as well as the synaptic growth can be captured by a single pulse of 5-HT applied at the opposite sensory to motor neuron synapse. By contrast, to the synapse-specific forms, cell-wide long-term facilitation generated by repeated pulses of 5-HT at the cell body is not associated with growth and does not persist beyond 48 hrs. However, this also can be captured and growth can be induced in a synapse-specific manner by a single pulse of 5-HT applied to one set of the peripheral synapses (Casadio et al., 1999). Casadio and colleagues further found that when an inhibitor of local protein synthesis was applied to the synapse that received a single pulse of 5-HT, there was facilitation and synaptic growth at the marked synapse at 24 hr, but these did not persist to 72 hr. These results first suggested that local protein synthesis may not be required for the initiation of the 5-HT-induced changes in function and structure at 24 hr but is required for these 5-HT-induced changes to persist to 72hr.

Thus, CREB-mediated transcription is required for the establishment of all four forms of long-term facilitation in Aplysia and for the initial maintenance of the synaptic plasticity at 24 hr, but it is not sufficient for the self-maintained stabilization of the plastic changes. To obtain persistent facilitation and specifically to obtain the growth of new synaptic connections, one needs, in addition to CREB-mediated transcription, a marking signal produced by a single pulse of 5-HT applied to the synapse. This single pulse of 5-HT has at least two marking functions. First, it produces a PKA-mediated covalent modification that marks the captured synapse for growth. Second, it stimulates a rapamycin-sensitive component of local protein synthesis, which is required for the long-term maintenance of the plasticity and stabilization of the growth beyond 24 hrs.

The finding of two distinct components for the marking signal suggested that there is a mechanistic distinction between the initiation of long-term facilitation and synaptic growth, which does not require local protein synthesis and the stable maintenance of the long-term functional and structural changes which are dependent on local protein synthesis. What might constitute the mark necessary for stabilizing long-term facilitation? Since mRNAs are made in the cell body, the need for the local translation of some mRNAs suggests that these mRNAs may be dormant before they reach the marked synapse. If that is true, then the synaptic mark for stabilization might be a regulator of translation capable of activating mRNAs that are translationally dormant.

Si et al. (2003a) have recently identified the Aplysia homologue of CPEB (cytoplasmic polyadenylation element binding protein), a protein capable of activating dormant mRNAs. This novel, neuron-specific isoform of CPEB is present in the processes of sensory neurons and can be induced in the process by a single pulse of 5-HT. The induction of ApCPEB is independent of transcription but requires new protein synthesis and is sensitive to rapamycin and to inhibitors of P13 kinase. Moreover, the induction of CPEB coincides with the polyadenylation of neuronal actin, and blocking ApCPEB locally at the activated synapse blocks the long-term maintenance of synaptic facilitation but not its early expression at 24 hrs. Thus, ApCPEB has all the properties required of the local protein synthesis-dependent component of marking and supports the idea that there are separate molecular mechanisms for initiation of the long-term process and its stabilization.

How might ApCPEB stablize the late phase of long-term facilitation? As outlined above, the stability of long-term facilitation seems to result from the persistence of the structural changes in the synapses between sensory and motor neurons, the decay of which parallels the decay of the behavioral memory. These structural changes include the remodeling of pre-existing facilitated synapses, as well as the growth and establishment of new synaptic connections. Reorganization and growth of new synapses has two broad requirements: (1) structural (changes in shape, size, and number) and (2) regulatory (where and when to grow). The genes involved in both of these aspects of synaptic growth might be potential targets of ApCPEB. The structural aspect of the synapses are dynamically controlled by reorganization of the cytoskeleton, which can be achieved either by redistribution of pre-existing cytoskeletal components or their local synthesis. The observation that N-actin and T?1 tubulin (Martin et al., unpublished observation) are present in the peripheral population of mRNAs at the synapse and can be polyadenylated in response to 5-HT suggests that at least some of the structural components for synaptic growth can be controlled through ApCPEB-mediated local synthesis (Kim et al., 1999). In addition, CPEB has been found to be involved in the regulation of local synthesis of EphA2 (Brittis et al., 2002), a member of the family of receptor tyrosine kinases, which have been implicated in axonal pathfinding and the formation of excitatory synapses in the mammalian brain. Thus, CPEB might contribute to the stabilization of learning-related synaptic growth by controlling the synthesis of both the structural molecules such as tubulin and N-actin and the regulatory molecules such as CAMK11 and members of the Ephrin family.

If there is a continuous need for the local protein synthesis to maintain the learning-related synaptic changes over long periods of time, how can these enduring changes be achieved by a translational regulator such as CPEB in the face of a continuous turnover of the protein? One possible answer to this question comes from the subsequent finding by Si et al. (2003b) that the neuronal isoform of CPEB shares properties with prion-like proteins. Prions are proteins that can assume at least two stable conformational states, one of which is usually active while the other is inactive. Furthermore, one of the conformational states, the prion state, is self-perpetuating, promoting the conformational conversion of other proteins of the same type. Work on yeast suggests the Aplysia neuronal CPEB exists in two stable, physical states that are functionally distinct. As with other prions, one of these states has the ability to self-perpetuate in a dominant epigenetic fashion. However, unlike the known prion proteins where the dominant state is the inactive form of the protein, surprisingly, in the case of Aplysia CPEB, the dominant form is the active form of the protein capable of activating translationally dormant mRNAs. Because the activated CPEB can be self-perpetuating, it could contribute to a self-sustaining synapse-specific long-term molecular change and provide a mechanism for the stabilization of learning-related synaptic growth and the persistence of memory storage.

An Overall View

The storage of long-term memory is represented at the cellular level by activity-dependent modulation of both the function and the structure of specific synaptic connections (Kandel, 2001). Although a number of molecular components that underlie the functional changes (i.e. changes in synaptic strength) associated with memory storage have been characterized, little is known about how these are regulated by and coupled to the signaling pathways that give rise to the synaptic structural changes. Recent studies in the mammalian central nervous system have provided some evidence that activity-dependent remodeling of pre-existing synapses and the growth of new synaptic connections might play an important role in complex neuronal networks by modulating and perhaps reconfiguring the activity of the neural network in which this occurs (Buchs and Muller 1996; Durand et al., 1996; Engert and Bonhoeffer, 1999; Maletic-Savatic et al., 1999; Toni et al, 1999). However, in the mammalian brain, these structural changes are difficult to study because the effects are often modest and the contribution of individual synapses to the learning process is not yet well defined (Malenka and Nicoll, 1997; Yuste and Bonhoeffer, 2001; Bliss et al., 2003; Lamprecht and LeDoux, 2004).

A useful model system for delineating the molecular and structural mechanisms that underlie memory storage is the gill-withdrawal reflex of Aplysia where the persistence of learning-induced growth of new sensory neuron synaptic connections parallels both the behavioral duration of the memory and the enhancement of synaptic strength. This synaptic growth can be reconstituted in sensory-motor neuron co-cultures by repeated presentations of 5-HT (Montarolo et al., 1986). Moreover, the Aplysia sensory-motor neuron culture preparation has five primary advantages for the molecular and functional analysis of learning-related synaptic growth: 1) Individual neurons can be cultured readily. 2) The synaptic connections formed in culture are precise. Autapses do not form nor do neurons form connections with inappropriate targets. 3) Learning-induced structural changes are robust, highly reproducible and easy to study. 4) The presynaptic glutamatergic sensory neuron varicosities that form connections with motor neurons are large and individual varicosities can be followed continuously for several days. 5) The cell body of both the sensory and motor neuron is large and it is easy to express constructs that alter gene expression.

Recent time-lapse imaging studies in the Aplysia culture preparation have found that long-term facilitation of the sensory-motor neuron synapse is accompanied by two temporally and morphologically distinct classes of presynaptic structural change: the rapid activation of silent pre-existing varicosities by filling with synaptic vesicles and the slower growth of new functional varicosities (Kim et al., 2003). These findings, the first to be made on individually identified presynaptic varicosities, suggest that the duration of the changes in synaptic effectiveness that accompany memory storage may be reflected by the differential regulation of two fundamentally disparate forms of presynaptic compartment: (1) nascent (empty) silent varicosities that can be rapidly and reversibly remodeled into active transmitter release sites and (2) mature, more stable and functionally competent varicosities that following long-term training may undergo a process of fission to form new stable synaptic contacts. What are the cellular and molecular mechanisms responsible for these two distinct classes of learning-related structural change?

Some potential insights to this question are suggested by recent live imaging studies of developing synapses in the mammalian central nervous system (for review see, Ziv and Garner, 2004). In Aplysia, the 5-HT-induced enrichment and subsequent activation of pre-existing silent varicosities occurs very rapidly, initial changes in the recruitment of synaptic vesicle proteins to empty presynaptic varicosities can be detected at 30 min after 5-HT training. This short delay is similar to what has been reported for the establishment of functional transmitter release sites in cultured hippocampal neurons and suggests that organization of the presynaptic compartment could be achieved by a rapid recruitment of pre-assembled active zone components to the sites of cell contact (Ahmari et al., 2000; Friedman et al., 2000).

One attractive hypothesis that has been proposed to account for this rapid differentiation of the presynaptic compartment is based on a unitary model of active zone assembly (Zhai et al., 2001; Shapira et al., 2003). Here proteins that comprise the cytoskeletal matrix of the active zone (CAZ) are packaged into transport vesicles for delivery and fusion with the plasma membrane at nascent synaptic contacts. These precursor vesicles contain multiple active zone components including Piccolo and Bassoon - hence the name Piccolo/Bassoon transport vesicles (PTVs) - as well as other CAZ scaffolding molecules implicated in synaptic vesicle exocytosis such as Rim, Munc13s, Munc18-1, syntaxin, Snap25, N-type calcium channels, α-liprin and ERC/CAST but typically not synaptic vesicle proteins. The assembly of each new active zone appears to be preceded by the recruitment and fusion of integer multiples of Piccolo/Bassoon transport vesicles - typically 2-5.

This model in developing mammalian synapses provides a molecular mechanism of active zone assembly that is consistent with the rapid remodeling and presynaptic activation of pre-existing empty sensory neuron varicosities induced by 5-HT in Aplysia culture. Moreover, the apparent heterogeneity in the content of these mobile pre-assembled packets – some of which contain synaptic vesicle proteins and the others which contain components required for assembly of the active zone (Friedman et al., 2000; Zhai et al., 2001) could explain why more than a half of the sensory neuron varicosities enriched in synaptophysin following 5-HT treatment are not functional. It is likely the maturation of transmitter release sites is not yet complete in these varicosities and that varicosities which are enriched only in synaptic vesicle proteins but lack a fully differentiated active zone would not be functionally competent.

How are these modular transport packets induced by learning for presynaptic assembly targeted to empty sensory neuron varicosities? In Aplysia culture, only a specific subset of presynaptic varicosities become enriched in synaptic vesicles and activated whereas others do not? Since repeated 5-HT treatment leads to a differential enrichment only in specific sensory neuron varicosities, proteins produced locally at each varicosity might contribute to this structural alteration. Alternatively, these protein components might not be synthesized within or nearby each varicosity, but might be transported and captured at specific varicosities that have been “tagged” following 5-HT stimulation (Frey and Morris, 1997; Martin et al., 1997b). Thus, the 5-HT-induced recruitment of synaptic vesicle and active zone proteins at a specific subset of sensory neuron varicosities and the subsequent functional activation of these previously silent synapses may be one of the local structural consequences of long-term synapse-specific plasticity. It should be noted that although considerable turnover of synapses and concomitant renewal of active zone proteins characterize the mature nervous system, Piccolo/Bassoon transport vesicles are exceedingly rare when compared to developing synapses (Zhai et al., 2001). It will be of interest to see if this developmental mechanism for presynaptic differentiation can be induced by learning and memory in the adult brain and if the learning-related recruitment of active zone proteins and subsequent assembly of transmitter release sites at empty sensory neuron varicosities is associated with an increase in the frequency of these modular precursor transport vesicles.

The second general class of learning-related presynaptic change associated with long-term facilitation in Aplysia is the 5-HT-induced formation of new sensory neuron varicosities. In culture, this increase in synapse number appears to be accomplished, at least in part, by the division or splitting of preexisting varicosities (Hatada et al., 2000; Kim et al., 2003; Udo et al., 2005). Based on time-lapse confocal imaging, the apparent sequence of structural plasticity that gives rise to new varicosities involves the 5-HT-induced recruitment of synaptic vesicle proteins to a pre-existing varicosity leading to both an enrichment of synaptic vesicles as well as an overall increase in the size of that varicosity. This remodeling and growth of pre-existing varicosities ultimately results in the generation of new varicosities that are also enriched in synaptic vesicle proteins.

Precisely how these new varicosities come into being is still not clear. The close spatial association of pre-existing and newly formed sensory neuron varicosities in time-lapse images, including the occasional confluence of both presynaptic compartments, suggests a process involving either the division of the original varicosity or a budding off of some components of the transmitter release site of that varicosity. Aspects of this reorganization of the presynaptic compartment that precedes learning-related synaptic growth in Aplysia have been reported at developing synapses in mammals. For example, recent imaging studies of the early stages of synapse formation have shown that presynaptic sites formed immediately after initial contact of axonal and dendritic processes are highly unstable. Moreover, even apparent mature presynaptic sites are relatively unstable as occasionally “orphan release sites” break off from fully formed boutons and either migrate to adjacent presynaptic sites or participate in the formation of completely new ones (Krueger et al., 2003; Friedman et al.; 2000; Hopf et al., 2002).

Combined these observations indicate that differentiation of the presynaptic compartment, either induced by learning in the mature nervous system or as a mechanistic step during development, is a highly dynamic and rapid process that can recruit both preexisting proteins as well as preassembled synaptic components and suggest that there may be an upper limit to the size of an active zone and associated synaptic vesicle cluster that a neuron can construct. Any increase beyond this limit may cause the entire transmitter release apparatus to become unstable, perhaps because of some inherent metabolic or thermodynamic constraints. The final result is that smaller units of active zone material and cognate synaptic vesicle clusters can bud from the established site and translocate along the axon where, under appropriate conditions, they may participate in the assembly of new presynaptic sites.

The instability of nascent presynaptic compartments during the early stages of neuronal differentiation is characterized by the dispersion of mobile packets of synaptic vesicles, synaptic vesicle precursors and active zone precursors and a renewal of their migration once the transient pre- and postsynaptic contacts breakup. As the neurons mature, an increasing proportion of these initial contacts develop into more stable and functionally competent presynaptic terminals (Ziv and Garner, 2004). Results in Aplysia indicate that these mature synaptic contacts can be selectively destabilized in an activity-dependent fashion during learning and memory and suggest that learning-related synaptic growth in the adult brain may reutilize some of the same cellular and molecular mechanisms important for de novo synapse formation during development.

How does the long-term process stabilize the transformation of labile, nascent presynaptic compartments into more mature and functionally competent release sites? Conversely, how are mature and fully functional presynaptic compartments destabilized to give rise to the formation of new release sites? Both the stabilization of presynaptic assembly as well as the recruitment of destabilizing factors that may remove molecular constraints and permit fission are likely to require some regulatory interaction with the postsynaptic neuron. Several studies have now suggested a potential role of the postsynaptic neuron in modulating the 5-HT-induced structural changes observed at presynaptic terminals in Aplysia (Bailey and Chen, 1988a; Glanzman et al., 1990; Bao et al., 1998; Schacher et al., 1999). This interaction between postsynaptic and presynaptic neurons is critical for the formation and maturation of synapses during development, and may play a key role in both the pre- and postsynaptic expression of the structural plasticity associated with learning and memory storage. Moreover, since the two classes of 5-HT-induced changes in presynaptic structure that accompany long-term facilitation in Aplysia culture share their postsynaptic counterpart, there must also be trans-synaptic signals both anterograde and retrograde to coordinate and regulate the learning-induced structural remodeling in an ongoing manner.

Obvious candidates for the molecules involved in triggering and stabilizing the presynaptic differentiation associated with long-term memory are those involved in synaptogenesis during development. With the extensive investigation of the developing neuromuscular junction as well as developing synapses in the central nervous system of both vertebrates and invertebrates, the list of such molecules has increased considerably in recent years (for review see Goda and Davis, 2003; Sanes and Lichtman, 1999; Jin, 2002; Scheiffele, 2003). Once axonal outgrowth and pathfinding are completed and the incipient axon-target interactions develop, signaling molecules can engage in bidirectional communication to coordinate the differentiation of pre- and postsynaptic membrane specializations. Several cell surface molecules have been identified that might mediate this process through their adhesion and signaling capabilities (Washbourne et al., 2004). Some of these trans-synaptic signaling systems employ heterophilic interactions that, in principle, could introduce an element of directionality required for the coordinated functional differentiation of the pre- and postsynaptic compartment (Scheiffele, 2003). Among the trans-synaptic molecular candidates the ß–Neurexin-Neuroligin interaction is particularly intriguing.

Neurexins are presynaptic transmembrane proteins present in many variants (Tabuchi and Sudhof, 2002). Neurexins associate with synaptic vesicles by interaction with presynaptic scaffolding proteins such as CASK and Mints that are present in the cytomatrix of the active zone and direct binding to synaptotagmin (Hata et al., 1993; Hata et al., 1996; Biederer and Sudhof, 2000). Neuroligins are postsynaptic transmembrane proteins and bind to the PDZ domains of PSD-95 (Irie et al., 1997), a scaffolding protein in the postsynaptic compartment of excitatory synapses. Thus, the ß-Neurexin-Neuroligin interaction may act as a trans-synaptic bridge bringing synaptic vesicles into alignment with neurotransmitter receptor-ion channel complexes in the postsynaptic density. This hypothesis gained support when Neuroligin expressed in nonneuronal cells was shown to cluster synaptic vesicles in contacting glutamatergic axons (Scheiffele et al., 2000). Moreover, antibody-induced clustering of recombinant Neurexin directly induced the co-clustering of synaptic vesicles (Dean et al., 2003).

Although details of Neuroligin-induced presynaptic differentiation are not completely understood, aspects of the proposed consequences of this trans-synaptic interaction during development might provide additional molecular insights into the mechanisms that underlie the two learning-related presynaptic changes in Aplysia. Neuroligin activity depends on the lateral clustering of individual Neuroligin molecules that appear to induce the clustering of Neurexin in the presynaptic membrane. These lateral clusters of Neurexin may, in turn, activate signaling cascades that promote presynaptic assembly by recruiting scaffolding molecules such as CASK, MINT and Veli that interact directly with the PDZ-binding motifs in the cytoplasmic tail of Neurexin.

Since CASK also interacts with proteins that regulate the actin-spectrin cytoskeleton, additional Neurexin molecules and/or other synaptic components could be inserted into this presynaptic scaffold at the cytomatrix of the active zone perhaps facilitating the 5-HT-induced growth and stabilization of the newly formed release sites at empty sensory neuron varicosities. Conversely, activity-dependent stimulation that gives rise to learning and memory could lead to a selective and synapse-specific interruption of the adhesive and/or signaling capabilities of the Neuroloigin-Neurexin trans-synaptic interaction at a subset of mature sensory neuron varicosities. This learning-induced alteration might, in turn, destabilize the cytoskeleton matrix of the presynaptic compartment - a transient and permissive step that could allow units of active zone material and associated synaptic vesicles to break off and to participate in the establishment of new presynaptic sites.

It is becoming increasingly clear that many of the molecular mechanisms and signaling interactions utilized for learning-related synaptic growth in the adult brain may share features in common with those that govern synaptogenesis during development. Recent studies of the synaptic growth that accompanies long-term facilitation in Aplysia have begun to characterize the sequence of cellular and molecular events responsible for both the initiation and persistence of these structural changes. This in turn has revealed that specific molecules and mechanisms important for synapse formation during the development of the nervous system can be reutilized in the adult for the purposes of synaptic plasticity and memory storage (Figure 6).

Figure 6. The molecular biology of memory storage Aplysia. In short-term sensitization (lasting minutes to hours) a single tail shock causes a transient release of serotonin that leads to covalent modification of preexisting proteins. The serotonin acts on a transmembrane serotonin receptor to activate the enzyme adenylyl cyclase (AC), which converts ATP to the second messenger cyclic AMP. In turn, cAMP recruits the cAMP-dependent protein kinase A (PKA) by binding to the regulatory subunits (spindles), causing them to dissociate from and free the catalytic subunits (ovals). These subunits can then phosphorylate substrates (channels and exocytosis machinery) in the presynaptic terminals, leading to enhanced transmitter availability and release. In long-term sensitization, repeated stimulation causes the level of cAMP to rise and persist for several minutes. The catalytic subunits can then translocate to the nucleus, and recruit the mitogen-activated protein kinase (MAPK). In the nucleus, PKA and MAPK phosphorylate and activate the cAMP response element-binding (CREB) protein and remove the repressive action of CREB-2, an inhibitor of CREB-1. CREB-1 in turn activates several immediate-response genes, including a ubiquitin hydrolase necessary for regulated proteolysis of the regulatory subunit of PKA. Cleavage of the (inhibitory) regulatory subunit results in persistent activity of PKA, leading to persistent phosphorylation of the substrate proteins of PKA. A second immediate-response gene activated by CREB-1 is C/EBP, which acts both as a homodimer and as a heterodimer with activating factor (AF) to activate downstream genes [including elongation factor la (EF1α)] that lead to the growth of new synaptic connections. (From Kandel, 2001.)

For example, these studies indicate that the functional and structural changes associated with long-term memory involve the flow of information from receptors on the cell surface to the genome, as seen in other processes of cell differentiation and growth. Such changes may reflect recruitment by environmental stimuli of developmental processes that are latent or inhibited in the fully differentiated neuron. Indeed, an increasing body of evidence suggests that the molecular and structural changes accompanying long-term memory storage share several features in common with the cascade of events that underlie neuronal differentiation and development. In both cases, there is a requirement for new protein and mRNA synthesis. These alterations can be initiated in the long-term process by modulatory transmitters that, in this respect, appear to mimic the effects of growth factors and hormones during the cell cycle and differentiation. Thus, modulatory transmitters important for learning activate not only a cascade of cytoplasmic events required for the short-term process, but also induce a genomic cascade by which the transmitter can exert a long-term regulation over both the excitability and synaptic architecture of the neuron through changes in gene expression.

Studies in Aplysia have further demonstrated that the earliest stages of long-term memory formation are associated with modulation of an immunoglobulin-related cell adhesion molecule homologous to NCAM. These cell adhesion molecules can be down-regulated by 5-HT, a modulatory transmitter important for both sensitization and classical conditioning in Aplysia and by cAMP, a second-messenger activated by 5-HT. This down-regulation appears to serve as a preliminary and permissive step for the growth of new synaptic connections that accompanies the long-term process. Thus, a molecule used during development for cell adhesion and axon outgrowth is retained into adulthood, at which point it seems to restrain or inhibit growth until the molecule is rapidly and transiently decreased at the cell surface by a modulatory transmitter important for learning.

The finding that 5-HT leads to the rapid down-regulation of only one isoform of ApCAM (the transmembrane isoform) and not the others (the GPI-linked isoforms) raises the interesting possibility that learning-related synaptic growth in the adult may be initiated by an activity-dependent recruitment of specific isoforms of adhesion molecules, similar to the modulation of cell-surface receptors during the fine-tuning of synaptic connections in the later stages of the developing nervous system. One consequence of isoform recruitment is that it would allow neuronal activity to regulate the surface expression of each isoform, a process that might take on additional functional significance if these surface molecules were distributed differentially along the three-dimensional extent of the neuron. These results also suggest that processing and storage of information in the nervous system may rely on the same mechanisms utilized by other cells in the body to organize and regulate membrane trafficking important for growth and cytoplasmic expansion.

Finally, insights from the molecular studies of learning and memory in Aplysia suggest that the critical time window for macromolecular synthesis that is a ubiquitous feature of long-term memory storage may be explained by a cascade of gene activation whereby one or more immediate-early genes control the transcription of late effector genes. The similarity between this critical period essential for memory storage in the mature nervous system and those found during the different stages of neuronal differentiation suggests that aspects of the regulatory mechanisms underlying both the functional and structural changes that accompany long-term memory in the adult may eventually be understood in the context of the basic molecular logic used for the formation and refinement of synaptic connections during development.

Acknowledgements

Research in this review was supported in part by the Howard Hughes Medical Institute (to E.R.K.), National Institutes of Health grant MH37134 (to C.H.B.), and the Kavli Institute for Brain Sciences.

References

Abel T.P.V., Nguyen V., Barad M., Deuel T.A., Kandel E.R., and Bourtchouladze R. 1997. Genetic demonstration of a role for PKA in the late phase of LTP and in hippocampus-based long-term memory. Cell 88:615-26.

Ahmari, S.E., Buchanan, J., and Smith, S.J. 2000. Assembly of presynaptic active zones from cytoplasmic transport packets. Nat. Neurosci. 3: 445-451.

Alberini C.M., Ghirardi M., Metz R., and Kandel E.R. 1994. C/EBP is an immediate-early gene required for the consolidation of long-term facilitation in Aplysia. Cell 76: 1099-1114.

Antonova I., Arancio O., Trillat A-C., Wang H.G., Zablow L., Udo H., Kandel E.R., and Hawkins R.D. 2001. Rapid increase in clusters of presynaptic proteins at onset of long-lasting potentiation. Science 294: 1547-1550.

Bacskai B.J., Hochner B., Mahaut-Smith M., Adams S.R., Kaang B.K., Kandel E.R., and Tsien R.Y. 1993. Spatially resolved dynamics of cAMP and protein kinase A subunits in Aplysia sensory neurons. Science 260: 222-226.

Bailey C.H. and Chen M. 1983. Morphological basis of long-term habituation and sensitization in Aplysia. Science 220: 91-93.

Bailey C.H. and Chen M. 1988a. Long-term memory in Aplysia modulates the total number of varicosities of single identified sensory neurons. Proc Natl Acad Sci USA 85: 2373-2377.

Bailey C.H. and Chen M. 1988b. Long-term sensitization in Aplysia increases the number of presynaptic contacts onto the identified gill motor neuron L7. Proc Natl Acad Sci USA 85: 9356-9359.

Bailey C.H. and Chen M. 1989 Time course of structural changes at identified sensory neuron synapses during long-term in Aplysia. J. Neurosci. 9: 1774-1780.

Bailey C.H. and Kandel E.R. 1993. Structural changes accompanying memory storage. Annu. Rev. Physiol. 55: 397-426.

Bailey, C.H., Kandel, E.R. and Si, K. 2004. The persistence of long-term memory: a molecular approach to self-sustaining changes in learning-induced synaptic growth. Neuron 44:49-57.

Bailey C.H., Bartsch, D., and Kandel E.R. 1996. Toward a molecular definition of long-term memory storage. Proc Natl Acad Sci USA 93: 13445-13452.

Bailey C.H., Chen M., Keller F., and Kandel, E.R. 1992a. Serotonin-mediated endocytosis of apCAM: An early step of learning-related synaptic growth in Aplysia. Science 256: 645-649.

Bailey C.H., Montarolo P.G., Chen M., Kandel E.R., Schacher S. 1992b. Inhibitors of protein and RNA synthesis block the structural changes that accompany long-term heterosynaptic plasticity in the sensory neurons of Aplysia. Neuron 9:749-758.

Bailey, C.H., Giustetto, M., Zhu, H., Chen, M., and Kandel, E.R. 2000. A novel function for serotonin-mediated short-term facilitation in Aplysia: Conversion of a transient cell-wide homosynaptic Hebbian plasticity into a persistent, protein synthesis-independent synapse-specific enhancement. Proc. Natl. Acad. Sci. USA 97: 11581-11586.

Bailey C.H., Kanng B.K., Chen M., Marin C., Lim C.S., Casadio A., and Kandel E.R. 1997. Mutation in the phosphorylation sites of MAP kinase blocks learning-related internalization of apCAM in Aplysia sensory neurons. Neuron 18:913-924.

Bao, J.X., Kandel, E.R., and Hawkins, R.D. 1998. Involvement of presynaptic and postsynaptic mechanisms in a cellular analog of classical conditioning at Aplysia sensory-motor neuron synapses in isolated cell culture. J. Neurosci. 18: 458-466.

Bartsch D., Ghirardi M., Skehel P.A., Karl K.A., Herder S.P., Chen A., Bailey C.H., and Kandel E.R. 1995. Aplysia CREB2 represses long-term facilitation: Relief of repression converts transient facilitation into long-term functional and structural changes. Cell 83: 979-992.

Bartsch D., Ghirardi M., Casadio A., Giustetto M., Karl K.A., Zhu H., and Kandel E.R. 2000. Enhancement of memory-related long-term facilitation by ApAF, a novel transcription factor that acts downstream from both CREB1 and CREB2. Cell 103: 595-608.

Biederer, T., Sudhof, T.C. 2000. Mints as adaptors. Direct binding to Neurexins and recruitment of munc 18. J. Biol. Chem. 275:39, 803-806.

Bliss, T.V., Collingridge, G.L., and Morris, R.G. (2003) Introduction. Long-term potentiation and structure of the issue. Philos. Trans. R. Soc. Lond. B Biol. Sci. 358:607-611.

Bradke F. and Dotti C.G. 1999. The role of local actin instability in axon assembly. Science 283: 1931-1934.

Brittis P.A., Lu, Q., Flanagan J.G. 2002. Axonal protein synthesis provides a mechanism for localized regulation at an intermediate target. Cell 110: 223-235.

Buchs P.A. and Muller D. 1996. Induction of long-term potentiation is associated with major ultrastructural changes of activated synapses. Proc. Nail. Acad. Sci. USA 93: 8040-8045.

Casadio A., Martin K.C., Giustetto M., Zhu H., Chen M., Bartsch D., Bailey C.H., and Kandel E.R. 1999. A transient, neuron-wide form of CREB-mediated long-term facilitation can be stabilized at specific synapses by local protein synthesis. Cell 99: 221-237.

Castellucci V.F., Blumenfeld H., Goelet P., Kandel E.R. 1989. Inhibitor of protein synthesis blocks long-term behavioral sensitization in the isolated gill-withdrawal reflex of Aplysia. Science 220: 91-93.

Colicos M.A., Collins B.E., Sailor M.J., and Goda Y. 2001. Remodeling of synaptic actin induced by photoconductive stimulation. Cell 107: 605-616.

Dale N., Kandel E.R., and Schacher S. 1987. Serotonin produces long-term changes in the excitability of Aplysia sensory neurons in culture that depend on new protein synthesis. J. Neurosci. 7: 2232-2238.

Dash P.K., Hochner B., and Kandel E.R. 1990. Injection of the cAMP-responsive element into the nucleus of Aplysia sensory neurons blocks long-term facilitation. Nature 345: 718-721.

Davis R.L. 1996. Physiology and biochemistry of Drosophila learning mutants. Physiol. Rev. 76: 299-317.

Davis R.L, Cherry J., Dauwalder B., Han P.L., and Skoulakis E. 1995. The cyclic AMP system and Drosophila learning. Mol. Cell Biochem. 149-150: 271-278.

Dean, C., Scholl, F.G., Choih, J., DeMaria, S., Berg, J., Isacoff, E., and Scheiffele, P. 2003. Neurexin mediates the assembly of presynaptic terminals. Nat. Neurosci. 6, 708-716.

Drain P. and Folkers E. 1991. cAMP-dependent protein kinase and the disruption of learning in transgenic flies. Neuron 6:71-82.

Durand G.M., Kovalchuk Y., and Konnerth, A. 1996. Long-term potentiation and functional synapses induction in developing hippocampus. Nature 381:71-75.

Engert F. and Bonhoefier T. 1999. Dendritic spine changes associated with hippocampal long-term synaptic plasticity. Nature 399:66-70.

English J.D. and Sweatt J.D. 1996. Activation of p42 mitogen-activated protein kinase in hippocampal long-term potentiation. J. Biol. Chem. 271: 24329-24332.

English J.D. and Sweatt J.D. 1997. A requirement for the mitogen-activated protein kinase cascade in hippocampal long-term potentiation. J. Biol.Chem. 272: 19103-19106.

Friedman, H.V., Bresler, T., Garner, C.C., and Ziv, N.E. 2000. Assembly of new individual excitatory synapses: time course and temporal order of synaptic molecule recruitment. Neuron 27: 57-69.

Frey, U., and Morris, R.G. 1997. Synaptic tagging and long-term potentiation. Nature 385: 533-536.

Frey U., Huang Y.-Y., and Kandel E.R. 1993. Effects of cAMP simulate a late stage of LTP in hippocampal CA1 neurons. Science 260: 1661-1664.

Frost W.N., Castellucci V.F., Hawkins R.D., and Kandel E.R. 1985. Monosynaptic connections made by the sensory neurons of the gill- and siphon-withdrawal reflex in Aplysia participates in the storage of long-term memory for sensitization. Proc. Natl. Acad. Sci. USA 82: 8266-8269.

Ghirardi M., Montarolo P.G., and Kandel E.R. 1995. A novel intermediate stage in the transition between short- and long-term facilitation in the sensory to motor neuron synapse of Aplysia. Neuron 14: 413-420.

Glanzman D.L., Kandel E.R., and Schacher S. 1990. Target-dependent structural changes accompanying long-term synaptic facilitation in Aplysia neurons. Science 249: 779-802.

Goda, Y. and Davis, G.W. 2003. Mechanisms of synapse assembly and disassembly. Neuron 40, 243-264.

Guan Z., Giustetto M., Lomvardas S., Kim J.-H., Miniaci M.D., Schwartz J.H., Thanos D., and Kandel E.R. 2002. Integration of long-term memory related synaptic plasticity involves bidirectional regulation of gene expression and chromatin structure. Cell 111:483-493.

Hai T.W., Liu F., Coukos W.J., and Green M. K. 1989. Transcription factor ATF cDNA clones: An extensive family of leucine zipper proteins able to selectively form DNA-binding heterodimers [published erratum appears in Genes Dev. 1990 Apr: 4(4): 682]. Genes Dev. 3: 2083-2090.

Han, J.H., Lim, Y.S., Kandel, E.R., and Kaang, B.K. 2004. Role of Aplysia cell adhesion molecules during 5-HT-induced long-term functional and structural changes. Learn. Mem. 11, 421-435.

Hata, Y., Davletov, B., Petrenko, A.G., Jahn, R. and Sudhof, T.C. 1993. Interaction of synaptotagmin with the cytoplasmic domains of neurexins. Neuron 10, 307–315.

Hata, Y., Butz, S., and Sudhof, T.C. 1996. CASK: a novel dlg/PSD95 homolog with an N-terminal calmodulin-dependent protein kinase domain identified by interaction with neurexins. J. Neurosci. 16, 2488-94.

Hatada, Y., Wu, F., Sun, Z.Y., Schacher, S., and Goldberg, D.J. 2000. Presynaptic morphological changes associated with long-term synaptic facilitation are triggered by actin polymerization at preexisting varicosities. J. Neurosci. 20: RC82.

Hochner B., Klein M., Schacher S., and Kandel E.R. 1986. Additional components in the cellular mechanisms of presynaptic facilitation contributes to behavioral dishabituation in Aplysia. Proc. Natl. Acad. Sci. USA 83: 8794-8798.

Hopf F.W., Walters J., Mehta S. and Smith S.J. 2002. Stability and plasticity of developing synapses in hippocampal neuronal cultures. J. Neuroci. 22: 775-781.

Hu Y., Barzilai A., Chen M., Bailey C.H., and Kandel E.R. 1993. 5-HT and cAMP induce the formation of coated pits and vesicles and increase the expression of clathrin light chain in sensory neurons of Aplysia. Neuron 10: 921-929.

Huang Y.-Y., Li X.C. and Kandel, E.R. 1994. cAMP contributes to mossy fiber LTP by initiating both a covalently mediated early phase and macromolecular synthesis-dependent late phase. Cell 79: 69-79.

Humeau Y., Doussau F., Vitiello F., Greengard P., Benfenati F., and Poulain B. 2001. Synapsin controls both reserve and releasable synaptic vesicle pools during neuronal activity and short-term plasticity in Aplysia. J. Neurosci. 21: 4195-4206.

Huntley, G.W., Benson, D.L., and Colman, D.R. 2002. Structural remodeling of the synapse in response to physiological activity. Cell 108: 1-4.

Irie, M., Hata, Y., Takeuchi, M., Ichtchenko, K., Toyoda, A., Hirao, K., Takai, Y., Rosahl, T.W., and Sudhof, T.C. 1997. Binding of neuroligins to PSD-95. Science 277, 1511–1515.

Jin, Y. 2002. Synaptogenesis: insights from worm and fly. Curr. Opin. Neurobiol. 12, 71-79.19

Jontes J.D., and Smith S.J. 2000. Filopodia, spines and generation of synaptic diversity. Neuron 27: 11-14.

Kandel E.R. 2001. The molecular biology of memory storage: a dialogue between genes and synapses. Science 294: 1030-1038.

Karpinski B.A., Morle, G.D., Huggenvik, J., Uhler, M.D. and Leiden, J.M. 1992. Molecular cloning of human CREB-2: An ATF/CREB transcription factor that can negatively regulate transcription from the cAMP response element. Proc. Natl. Acad. Sci. U.S.A. 89: 4820-4824.

Klein M. and Kandel E.R. 1980. Mechanism of calcium current modulation underlying presynaptic facilitation and behavioral sensitization in Aplysia. Proc. Natl. Acad. Sci. U.S.A. 77: 6912-6916.

Kim C.H. and Lisman J.E. 1999. A role of actin filaments in synaptic transmission and long-term potentiation. J. Neurosci. 19: 4314-4321.

Kim, J.-H., Udo, H., Li, H.-L., Youn, T.Y., Chen, M., Kandel, E.R., and Bailey, C.H. 2003. Presynaptic activation of silent synapses and growth of new synapses contribute to intermediate and long-term facilitation in Aplysia. Neuron 40: 151-165

Krucker T., Siggins G.R., and Halpain S. 2000. Dynamic actin filaments are required for stable long-term potentiation (LTP) in area CA1 of the hippocampus. Proc. Natl. Acad. Sci. USA 97: 6856-6861.

Krueger S.R., Kolar A. and Fitzsimonds R.M. 2003. The presynaptic release apparatus is functional in the absence of dendritic contact and highly mobile within isolated axons. Neuron 40: 945-957.

Kullman D.M. and Siegelbaum S.A. 1995. The site of expression of NMDA receptor-dependent long-term potentiation: new fuel for an old fire. Neuron 15: 997-1002.

Lamprecht, R. and LeDoux, J. 2004. Structural plasticity and memory. Nature Rev. Neurosci. 5: 45-54.

Malenka R.C., and Nicoll R.A. 1997. Silent synapses speak up. Neuron 19: 473-476.

Maletic-Savatic M., Maliriow R., and Svoboda K. 1999. Rapid dendritic morphogenesis in CAl hippocampal dendrites induced by synaptic activity. Science 283: 1923-27.

Mauelshagen J., Parker G.R., and Carew T.J. 1996. Dynamics of induction and expression of long-term synaptic facilitation in Aplysia. J. Neurosci. 16: 7099-7108.

Martin K.C. and Kandel E.R. 1996. Cell adhesion molecules, CREB and the formation of new synaptic connections during development and learning. Neuron 17: 567-570.

Martin K.C., Michael D., Rose J.C., Barad M., Casadio A., Zhu H., and Kandel E.R. l997a. MAP kinase translocates into the nucleus of the presynaptic cell and is required for long-term facilitation in Aplysia. Neuron 18:899-912..

Martin E.C., Casadio A.,Zhu H.,Yaping E., Rose J., Chen M., Bailey C.H., and Kandel E.R. 1997b. Synapse-specific long-term facilitation of Aplysia sensory somatic synapses: a function for local protein synthesis memory storage. Cell 91: 927-938.

Mayford M., Barzilai A., Keller F., Schacher S., and Kandel E.R. 1992. Modulation of an NCAM-related adhesion molecule with long-term synaptic plasticity in Aplysia. Science 256: 638-644.

Michael D. and Martin K.C. 1998. Repeated pulses of serotonin required for long-term facilitation activate mitogen-activated protein kinase in sensory neurons of Aplysia. Proc. Natl. Acad. Sci. U.S.A. 95: 1864-1869.

Miesenbock, G., De Angelis, D.A., and Rothman, J.E. 1998. Visualizing secretion and synaptic transmission with pH-sensitive green fluorescent proteins. Nature 394, 192-195.

Montarolo P.G., Goelet P., Castellucci V.F., Morgan J., Kandel E.R., and Schacher S. 1986. A critical period for macromolecular synthesis in long-term heterosynaptic facilitation in Aplysia. Science 234: 1249-1254.

Nakayama A.Y., Harms M.B., and Luo L. 2000. Small GTPases Rac and Rho in maintenance of dendritic spines and branches in hippocampal pyramidal neurons. J Neurosci. 15: 5329-5338.

Nazif F.A., Byrne J.H., Cleary L.J. 1991. cAMP induces long-term morphological changes in sensory neurons of Aplysia. Brain Res. 539: 324-327

Sanes, J.R., Lichtman J.W. 1999. Development of the vertebrate neuromuscular junction. Annu.Rev. Neurosci. 22, 389-442.

Schacher S., Castellucci V.F., and Kandel E.R. 1988. cAMP evokes long-term facilitation Aplysia sensory neurons that requires new protein synthesis. Science 240: 1667-1669.

Schacher, S., Wu., F., and Sun, Z.-Y. 1997. Pathway-specific synaptic plasticity: activity-dependent enhancement and suppression of long-term facilitation at converging inputs on a single target. J. Neurosci. 17: 597-606.

Schacher, S., Wu, F., Panyko, J.D., Sun, Z. Y., and Wang, D. 1999. Expression and branch-specific export of mRNA are regulated by synapse formation and interaction with specific postsynaptic targets. J. Neurosci. 19: 6338-6347.

Scheiffele, P. 2003. Cell-cell signaling during synapse formation in the CNS. Annu. Rev. Neurosci. 26, 485-508.

Scheiffele, P., Fan, J. Choih, J., Fetter, R., and Serafini, T. 2000. Neuroligin expressed in nonneuronal cells triggers presynaptic development in contacting axons. Cell 101, 657-669.

Schwartz H., Castellucci V.F., and Kandel E.R. 1971. Functions of identified neurons and synapses in abdominal ganglion of Aplysia in absence of protein synthesis. J. Neurophysiol. 34: 9639-9653.

Scholz K.P. and Byrne J.H. 1987. Long-term sensitization in Aplysia: biophysical correlates in tail sensory neurons. Science 235: 685-687.

Shapira M., Zhai R.G., Dresbach T., Bresler T., Torres V.I., Gundelfinger E.D., Ziv N.E., and Garner C.C. 2003. Unitary assembly of presynaptic active zones from piccolo-bassoon transport vesicles. Neuron 38: 237-252.

Si, K., Giustetto, M., Etkin, A., Hsu, R., Janisiewicz, A.M., Miniaci, M.C., Kim, J.-H., Zhu, H. and Kandel, E.R. 2003a. A neuronal isoform of CPEB regulates local protein synthesis and stabilizes synapse-specific long-term facilitation in Aplysia. Cell 115, 893-904.

Si, K., Lindquist, S. and Kandel, E.R. 2003b. A neuronal isoform of the Aplysia CPEB has prion-like properties. Cell 115, 879-891.

Sin W.C., Haas K., Ruthhazer E.S., and Cline H.T. 2002. Dendrite growth increased by visual activity requires NMDA receptor and Rho GTPases. Nature 419: 475-480.

Sutton M.A., Masters S.E., Bagnall M.W., and Carew T.J. 2001. Molecular mechanisms underlying a unique intermediate phase of memory in Aplysia. Neuron 31: 143-154.

Tabuchi, K. and Sudhof, T.C. 2002. Structure and evolution of neurexin genes: insight into the mechanism of alternative splicing. Genomics 79, 849–859.

Toni N., Buchs P.A., Nikonenko I., Bron C.R., Muller D. 1999. LTP promotes formation of multiple spine synapses between a single axon terminal and a dendrite. Nature 402: 421-425.

Threadgill R., Bobb K., and Ghosh A. 1997. Regulation of dendritic growth and remodeling by Rho, Rac and Cdc42. Neuron 19: 625-634.

Udo, H., Jin, I., Kim, J-H., Li, H-L., Youn, T., Hawkins, R.D., Kandel, E.R. and Bailey, C.H. 2005. Serotonin-induced regulation of the actin network for learning-related synaptic growth requires CdC42, N-WASP and PAK in Aplysia sensory neurons. Neuron 45:887- 901.15.

Washbourne, P., Dityatev, A., Scheiffele, P., Biederer, T., Weiner, J.A., Christopherson, K.S., and El-Husseini, A. 2004. Cell adhesion molecules in synapse formation. J. Neurosci. 24:9244-9249.

Yin J.C., Del Vecchio M., Zhou H., and Tully T. 1995. CREB as a memory modulator: Induced expression of a dCREB2 activator isoform enhances long-term memory in Drosophila. Cell 81: 107-115.

Yin J.C., Wallach J.S., Del Vecchio, M., Wilder E.L., Zhou H., Quinn W.G., and Tully T. 1994. Induction of a dominant negative CREB transgene specifically blocks long-term memory in Drosophila. Cell 79: 49-58.

Yuan X.B., Jin M., Xu X., Song Y.Q., Wu C.P., Poo M.M., and Duan S. 2003. Signaling and crosstalk of GTPases in mediating axon guidance. Nat. Cell Biol. 5:38-45.

Yuste R. and Bonhoeffer, T. 2001. Morphological changes in dendrititic spines associated with long-term synaptic plasticity. Annu. Rev. Neurosci. 24:1071-108.

Zhai R.G., Vardinon-Friedman H., Cases-Langhoff C., Becker B., Gundelfinger E.D., Ziv N.E., and Garner C.C. 2001. Assembling the presynaptic active zone: a characterization of an active one precursor vesicle. Neuron 29: 131-143.

Zhang W. and Benson D.L. 2001. Stages of synapse development defined by dependence on F-actin. J. Neurosci. 21: 5169-5181.

Ziv N.E. and Garner C.G. 2004. Cellular Mechanisms of Presynaptic Assembly. Nat.Rev. Neurosci. 5: 385-399.