Keywords

Introduction

In this review, we will discuss the role of the neural cell adhesion molecule [1] in emotion and learning. Classically, the behavioral role attributed to NCAM has been in learning, memory, and neural plasticity [2, 3]. However, increasing evidence presented over the past few years is also unraveling a role of NCAM on emotional behavior. After a brief introduction about the NCAM molecule, we will start reviewing its role in emotion, particularly in unlearned emotional responses, such as anxiety and aggression. In the second part of this review, we will address the role of NCAM in learning and memory processes to finally propose a role for this molecule at the interface between emotion and learning.

General Features of NCAM in the Central Nervous System: Molecular Structure and Function

NCAM is a member of the immunoglobulin superfamily of cell adhesion molecules. It is characterized by the presence of immunoglobulin homology domains (Ig-domains) in its extracellular part. NCAM is encoded by a single gene on chromosome 9 in mice [4] and 11 in humans [5] and undergoes differential splicing of the messenger RNA [6, 7]. The three main splice variants of NCAM are named according to their approximate molecular weights: NCAM180, NCAM140, and NCAM120. Within the central nervous system, NCAM180 appears to be the isoform enriched at postsynaptic sites, while NCAM140 is expressed both in neurons (pre and postsynaptically) and glia, and NCAM120 predominantly in glia [8, 9].

Post-translational attachment of long chains of the polysaccharide polysialic acid (PSA) to NCAM (PSA-NCAM) allows NCAM an additional mechanism to control synaptic functioning. In the adult brain, PSA-NCAM is mainly present in regions capable of undergoing some kind of structural plasticity [10], such as the hypothalamo-neurohypophyseal system [11], the olfactory bulb [12], the piriform and entorhinal cortices [13], the amygdala [14] hippocampus [13], and prefrontal cortex [14]. PSA is proposed to decrease NCAM-mediated membrane-membrane adhesion in vitro [15, 16], presumably due to its very large hydrated volume or negative charge or both [9, 17, 18].

NCAM is highly expressed at synaptic junctions. Neuronal activity regulates the functioning of synapses with a potential to either enhance or depress synaptic strength. During development, selective expression of cell adhesion molecules is proposed to regulate embryogenesis by dictating patterns of cell differentiation followed either by stabilization or selective elimination of synapses, as a mechanism of finetuning cellular connections [19]. Moreover, during development, neurite outgrowth is associated with the cell being in a de-adhesive state. When the neurite reaches and innervates the correct area, adhesiveness is increased so that the cell becomes locked in position. This modification in adhesion is controlled by homophilic cell-cell binding via cell adhesion molecules such as NCAM [20]. In addition to development, learning, and memory, LTP, aging, stress, and neuro-regeneration are all events that can also stimulate synaptic reorganization. Cumulative evidence indicates a key role for NCAM in the neural remodeling accompanying all these events.

NCAM and Emotion

Altered emotional behaviors have mainly been studied in adult male mice expressing a null mutation in the NCAM gene. Initial work demonstrated that NCAM-KO mice show enhanced anxiety in an emergence test, as indicated by their increased latencies to leave from a small box and explore an open environment [21]. An increased anxiety-like behavior of NCAM-deficient mice was also found in the light/dark avoidance test [22]. This effect was then shown to be gender and genetic background-independent and could be influenced by application of a low dose of the 5-HT1A serotonin receptor agonists, either buspirone or 8-hydroxy-2-(di-n-propylamino)-tertraline (8-OH-DPAT), which at the same dose showed no anxiolytic effect in wild-type controls [22]. This suggested a functional alteration of the serotonergic system in these mice likely to be involved in their altered anxiety-like responses. Although the authors found little evidence for the existence of regional differences in serotonin receptor expression, they did find a slight reduction in the level of the 5-HT1A receptor expression in the hippocampus and the amygdala of NCAM-KO mice.

In support of an altered serotonergic system is also the finding that NCAM-KO mice show enhanced aggression toward an unfamiliar intruder male, which was accompanied by a greater post-intruder hormonal stress response [23]. Analysis of c-fos mRNA levels to monitor neuronal activation after the intruder stress revealed greater neuronal activity when compared with wild type controls in limbic areas, including the amygdala, a brain region known to be involved in modulating emotional responses.

Exploratory behavior is a basic adaptive behavioral response in rodents. When an animal is presented with a new environment, it is normally motivated to explore. Perturbations in this response are indicative of alterations in emotionality. NCAM-KO mice were found to show enhanced exploratory activity in response to challenging and novel environments, such as the light/dark avoidance test [22] and the elevated plus maze [24], which has been proposed to be related to their enhanced amygdala activation and/or the greater stress response (see above). On its turn, this enhanced exploratory behavior of NCAM-deficient mice could also explain the contradictory anxiolytic behavior, which was reported in the elevated plus maze [24].

In order to further investigate the role of specific NCAM isoforms in emotional behaviors, the effect of manipulating the levels of the NCAM180 isoform was analyzed in the presence or absence of endogenous NCAM in mice. While transgenic overexpression of NCAM180 was without apparent behavioral and morphological effects, its expression in NCAM-deficient mice rescued many of the effects induced by NCAM ablation; i.e., it counteracted the following effects observed in NCAM-KO: (1) the enhanced intermale aggression displayed in the intruder test [25]; (2) the increased anxiety-like behavior in the light/dark avoidance test; and (3) partially, the hyperactivity displayed in the elevated plus maze [24]. Transgenic induction of NCAM180 in NCAM-KO mice also prevented the hypersensitivity of NCAM-KO mice to the anxiolytic effects of buspirone, which suggests the involvement of NCAM180 on the development and/or maintenance of the functionality of the serotonergic system, which might represent an important link between NCAM and the regulation of emotional behaviors. Together, these studies indicate that deletion of the NCAM gene is associated with abnormal emotional responding and that the NCAM180 isoform plays a pivotal role in emotional behavior.

Research to elucidate the role of NCAM in learning and memory has been greatly aided by the development of mimetic peptides with the ability to bind and modulate the activity of NCAM. In addition to the above described findings in genetically mutant mice, there is upcoming evidence from pharmacological studies that further supports a role for NCAM in emotional behaviors. Peptides which interact either with NCAM homophilic binding (P2 peptide; a 12-amino-acid peptide derived from the second immunoglobulin-like (Ig) module of NCAM and able to interact with cis-homophilic) have been shown to alter emotional behaviors. For example, administration of the P2 peptide intracerebroventricularly was shown to reduce anxiety, associated with performance in a learning task (i.e. the T-maze), and exploratory activity in an open field test [26]. Another peptide, C3d, binds to the NCAM IgI module and is able to trigger intracellular signaling cascades in vitro, which are similar to those activated by homophilic NCAM binding [27]. Intracerebroventricular injection of this peptide had no acute effect on exploratory behavior in an open field test, but dramatically reduced exploratory activity when C3d injected rats were re-exposed to the same arena 3-h and 24-h post-injection. This effect was not a consequence of sensorimotor impairments as peptide-treated rats performed without any significant differences on a rotarod and when explorative motility was accessed in an activity cage [28].

Emotional behavior is to a large extent associated with the development of psychiatric disorders. A few recent studies have shown some evidence to implicate NCAM in mood disorders. Thus, an increase in the CSF levels of the soluble 100-120kDa NCAM fragments has been found in patients with bipolar mood disorder type I and recurrent unipolar major depression [29, 30]. This increased presence of 120-kDa NCAM fragment in the CSF was postulated to derive form enhanced proteolytic cleavage of most of the extracellular region of transmembrane NCAM isoforms [31]. The complex nature of the extracellular region of the NCAM molecule has been also analyzed by creating transgenic mice that overexpress a soluble extracellular fragment of NCAM from the neuron-specific enolase promoter which leads to expression of this transgene in late neuronal differentiation (maximally in the neocortex and hippocampus) in adult mice [32]. Although these mice have normal sensory and motor functions, as compared to wild-type controls they display enhanced basal activity in the open field and in response to amphetamine and MK-801. However, and despite presenting a decreased number of synaptic terminals of subpopulations of GABAergic interneurons in the amygdala, these mice showed no alterations in anxiety-levels when tested in the zero maze [32].

Further evidence for the importance of NCAM and its polysialylated form PSA-NCAM in depression comes also from animal studies, which show that prolonged exposure of rodents to aversive stimuli (i.e. chronic stress protocols are widely used to model depression-like symptoms in rodents) lead to a reduction of NCAM mRNA and protein levels mainly in the hippocampus and the prefrontal cortex [33-43]. In contrast, PSA-NCAM expression was upregulated in the hippocampus, but downregulated in the amygdala of rats, exposed to 3 weeks of chronic stress [44]. Interestingly, recent studies in NCAM-KO (either constitutive or conditional, in which the cre-recombinase is regulated under the control of the αCaMKII promoter) mice have shown that these mice show enhanced depressive-like symptoms such as higher immobility time in the tail suspension test (Bisaz et al. unpublished observations) and a lower preference for sucrose solution [45]. In addition, chronic antidepressant treatments of rats and mice, with either the selective serotonin reuptake inhibitor fluoxetine or the tricyclic antidepressant imipramine, have shown to increase PSA-NCAM expression in the hippocampus and the prefrontal cortex [14, 46, 47]. These findings strongly support the hypothesis of a critical link between stress-related mood disorders and altered NCAM and PSA-NCAM expression [45].

In conclusion, the revised work indicates an important association between NCAM and emotion, particularly in domains such as anxiety, intermale aggression and depression. (See Table 1 for a comprehensive summary.) We will now review the literature that indicates a link between NCAM and learning and memory processes.

Table 1 Role of NCAM and PSA-NCAM in emotion
Full size table

NCAM in Learning: Functional Studies

Learning can be defined as a process by which new information is acquired, whereas memory is the process by which this information is retained. The process of transferring learned information into memory is known as consolidation and is intimately linked to the functioning of the hippocampus [48, 49]. Cognitive tests for rodents have been developed to capitalize on normal behavioral responses often using fear, hunger or innate curiosity to motivate and strengthen learning. For example, in the watermaze, animals are forced to learn a novel spatial map in order to escape from water, whereas in avoidance and fear conditioning paradigms electrical shocks are used to initiate associative memory formation. In particular, a key role and requirement of NCAM function during learning has been particularly demonstrated using hippocampus-dependent tasks, including avoidance conditioning [50] and spatial learning [51]. In addition, the role of NCAM in “emotional learning” is more typically studied with fear conditioning paradigms. Classical contextual and auditory fear conditioning are often utilized to this end. These tasks involve the induction of learned fear (generally manifested as a freezing response) to an initially neutral stimulus (respective, either a new context or a tone) that is associated with a naturally aversive stimulus (normally a footshock). Recall of the fear response can be tested by re-exposing the animal to either the context or tone and evaluating its conditioned fear (freezing) responses. Acquisition and consolidation of either fear conditioning modality relies on the basolateral complex of the amygdala [52-58], while the hippocampus is also required in the case of contextual fear conditioning [59].

The role of NCAM in learning has been investigated by a variety of approaches, including the examination of particular behaviors in genetically mutant (transgenic or knockout) mice, after application of “blocking” NCAM-related antibodies or peptides that affect (either mimicking or impairing) NCAM functioning. Constitutive NCAM-KO are perturbed in the Morris water maze when compared to wild-type controls [21]. As mentioned in the introduction, NCAM expression is critical during post-natal development and, therefore, one of the problems with the constitutive NCAM-KO mice is that their altered learning and memory might be related to the elimination of NCAM during such period (and its consequent developmental effects), but unrelated to the absence of NCAM in the adult brain. To overcome this problem, a conditional knockout mouse was generated where expression of NCAM was controlled by a Cre-recombinase system using a forebrain specific promoter. In this conditional knockout mouse, NCAM expression is reduced starting at around P22 onwards and, therefore, its deficit occurs after the major neurodevelopmental events have already taken place [60]. This conditional KO of NCAM also leads to a spatial memory deficit in adulthood, demonstrating that post-natal NCAM expression is required for learning and memory. Evidence that NCAM mediated processes contribute to fear memories also stem from work performed in constitutive NCAM-KO mice. In these mice, both auditory and contextual fear memories are impaired [24], suggesting that NCAM-mediated consolidation processes might also be implicated in brain regions other than the hippocampus, because auditory fear conditioning relies on the amygdala, but does not implicate the hippocampus.

The role of NCAM in non-associative learning was also tested by examining habituation (a decrease in behavioral responding to a stimulus) and sensitization (increased responding to an aversive stimulus) processes in NCAM-KO mice. While acoustic and tactile responses were altered in NCAM-KO mice (as evaluated by their startle responses), their ability to habituate to these stimuli was the same as wild type mice, demonstrating habituation learning is intact. In contrast, NCAM-KO mice exhibited impaired footshock sensitisation learning when compared with the wild type controls [61]. Footshock sensitization is a form of contextual conditioning, during which the context becomes a fear inducing stimulus leading to an increase in startle response to a tone [62].

In addition to the work reviewed above in genetically mutant mice, the critical role played by cell adhesion molecules in learning-related synaptic plasticity has been further demonstrated using blocking antibodies and peptides that bind to NCAM. Through these pharmacological approaches, the temporal implication of NCAM molecules during memory consolidation could also be explored. For example, studies where antibodies against NCAM were infused by intracerebroventricular injection in rats trained in the passive avoidance task found NCAM antibodies to induce memory impairment only when they where administered between 6 and 8 h post-training, but they were ineffective if given at the time of training or at any other time point up to 10 h following training [63]. In agreement with these observations, NCAM-specific antibodies were also found to impair passive avoidance learning in chicks when administered to a brain region critical for that learning (the intermediate medial hyperstriatum ventrale) at around 6-8 h post-training time [1, 64, 65]. The requirement of NCAM in long-term memory formation was similarly demonstrated through the use of oligonucleotides directed to NCAM [66].

In vitro evidence demonstrated that the C3d peptide could disrupt NCAM mediated cell adhesion and modulate neuritogenesis and synaptogenesis. When given to rats following training in the passive avoidance task, C3d prevented memory consolidation but only when the peptide was administered within a restricted time window either 20 min before training or at 6-8 h post-training [67]. Moreover, the C3d peptide also impaired both acquisition and recall of the Morris water maze [68] and the consolidation of contextual fear memories when administered 5.5 post-training [69].

After homophilic binding, NCAM promotes neurite outgrowth through mechanisms involving its interaction with the fibroblast growth factor receptor (FGFR1) [25, 70, 71]. The region of NCAM that binds FGFR1 is found in the second FnIII module of NCAM. A 15-aa peptide mimicking this region, termed the FGL peptide, has been shown to bind to and activate FGFR1 and to stimulate neurite outgrowth [25]. In vitro, the FGL peptide promotes synaptogenesis, synaptic outgrowth and pre-synaptic functioning [25, 72], as opposed to the C3d peptide that it similarly induces synaptogenesis and synaptic outgrowth, but impairs presynaptic functioning [27]. In vivo, FGL effects also contrast to those induced by C3d peptide (for details, see above). FGL strongly enhances spatial memory as shown in experiments, in which it was administered in rats immediately after the first and second day of spatial training in the water maze [72]. This peptide was ineffective if given 2 days prior to training indicating that it does not perturb learning of new information. Moreover, post-training FGL treatment also improved performance in a subsequently given reversal learning challenge, suggesting that FGL is beneficial in promoting behavioral flexibility [72].

The FGL peptide has also significantly enhanced our understanding of the role of NCAM in the prototypic emotional learning task, fear conditioning. Intracereboventricular infusion of FGL just after fear conditioning improved contextual memory performance when tested 24 h, and 7 and 28 days later. However, auditory fear memories were only enhanced when tested 28 days later, but not earlier, suggesting different consolidation mechanisms for conditioned fear to tones, which might become apparent only after longer time periods. FGL did not affect emotional responses per se, having no affect on open field behavior when administered for 2 days prior to testing. Therefore, similarly to its effect in spatial learning, FGL was shown to also enhance emotional learning.

NCAM in Learning: Correlative Studies

Functional studies reviewed above suggest a role for NCAM in memory formation. However, they do not allow one to address whether the functional consequences derived from interference with NCAM (expression or function) are due to a primary effect on already existing molecules or, instead, on learning-induced NCAM regulation. This is a critical issue as the former possibility might imply a non-specific effect of treatments on normal circuit functioning, whereas the latter would highlight NCAM as a key player in memory-associated circuit remodeling [68]. We will review here those studies that have addressed a potential regulation of NCAM expression by learning experiences.

The 6-8 h post-training period was highlighted by interventive (antibody injection) studies as critical for the involvement of NCAM in memory consolidation, Interestingly, in chicks, NCAM was found to be enriched in synaptic active zones in a memory-relevant region (the lobus parolfactorius) 5-6 h after a one-trial passive avoidance learning experience [73]. In rats, spatial training in the water maze was found to induce an increase in synaptosomal, but not total, expression level of hippocampal NCAM140 24 h post-training [68]. In the zebrafish, using an active avoidance paradigm (fish learn to cross a hurdle to avoid mild electric shocks when presented with a conditioned light signal) NCAM mRNA levels were increased in the optic tectum (a region important for avoidance learning) 3 h following avoidance conditioning, indicating that some learning-dependent changes in NCAM expression are transcriptionally mediated [74].

In addition to this temporal increase in NCAM expression observed several hours after training, there is also evidence that NCAM expression is decreased for a restricted period on the first phases of memory consolidation. Work in simple invertebrate organisms such as the sea snail Aplysia has allowed the study of neurobiological mechanisms linked to different types of learning, notably habituation and sensitisation [75]. Sensitisation in Aplysia is induced by presenting a noxious stimulus to the tail of the snail whereupon it withdraws its siphon and gill. If this is followed by stimulation of the siphon, it will withdraw both the siphon and gill in a more sensitized manner [76]. The neuronal circuit involved in long-term sensitisation has been identified [77]. Elements of this circuit (a siphon sensory neuron synapsing onto a gill motor neuron) can be isolated and cultured in vitro and synaptic facilitation can be induced artificially by presenting a puff of 5-TH directly to the cultured sensory neuron [78]. Interestingly, Aplysia expresses a homologue of mammalian NCAM known as apCAM. Induction of long-term sensitisation in cultured sensory neurons of Aplysia is accompanied by the growth of new synaptic connections and requires downregulation of apCAM [79].

Remarkably, in rodents NCAM is also selectively degraded 2-6 h post-training and this is necessary for passive avoidance memory consolidation [67]. Moreover, in this study the authors demonstrated that the C3d peptide, which impairs memory formation, seemed to prevent the temporal reduction in NCAM that occurs 2-6 h following learning in the passive avoidance paradigm indicating, as had been demonstrated in Aplysia, that a temporal reduction in NCAM expression is required for effective learning and memory.

PSA-NCAM in Learning

Polysialylation of NCAM is a potent modulator of NCAM functioning that significantly impacts on the role of NCAM in learning. Substantial evidence indicates that this posttranslational modification mechanism plays a key role in activity-dependent synaptic plasticity [2, 80] and memory formation. Most of the original work on this topic focused in the hippocampus. The requirement of PSA-NCAM for spatial learning has been indicated by different approaches. Removal of PSA from NCAM by endo-neuraminidase NE (endo-N), an enzyme which specifically cleaves α-2, 8-linked PSA, impedes the acquisition and retention of spatial memory [81, 82]. Mice expressing a null mutation in the polysialyltransferase (PST) gene, an enzyme critical for the postnatal polysialylation of NCAM, are impaired in spatial learning [83]. Conversely, a synthetic PSA-mimetic peptide administered in the mouse hippocampal CA3 region 5 h after massed training in the water maze was shown to significantly improve recall up to 4 weeks after training [84].

The role of PSA-NCAM in fear conditioning has received much attention in the past few years. Mice deficient in the PST gene were shown to display a very mild deficit in contextual fear conditioning. By contrast, auditory fear conditioning was normal in these PST-KO mice [83]. Strikingly, contextual fear conditioning was also impaired by application of PSA or PSA-NCAM 6 h, but not 2 h, following training [85]. Adult mice lacking the prenatally important St8SiaII/STX polysialyltransferase exhibit impaired memories (but not acquisition) in fear learning-related paradigms, auditory and contextual fear conditioning [86], suggesting an involvement of PSA-NCAM in memory processes related to fear conditioning. However, since PSA-NCAM expression levels in the amygdala of adult St8SiaII/STX knockout mice are normal, the possibility exist that their alterations in emotional learning are due to developmentally caused alterations in amygdala function.

Given the differential disruption of auditory fear conditioning in PST as compared to SXT-KO mice it initially seemed, at least in adulthood, PSA-NCAM may be more important in regulating hippocampal learning when compared with amygdala-dependent learning. To resolve this issue, our group examined the role of PSA-NCAM in the amygdala during fear learning and found that auditory fear conditioning under conditions that employed a high intensity shock (1 mA) enhances the amygdaloid expression of PSA-NCAM 12 h post-training [87]. However, fear learning appeared not to require the induction of PSA-NCAM since endo-N cleavage failed to prevent either fear learning or its consolidation. However, removal of PSA-NCAM in the amygdala enhanced memory extinction, suggesting that PSA-NCAM modulations during emotional learning may be important in determining the intensity of the memory trace.

On the other hand, there is extensive evidence showing that PSA-NCAM levels in the hippocampus are modified by learning experiences. Notably, selective enhancement of PSA-NCAM-positive cells in the rat hippocampal dentate gyrus (in particular in a population of cells located at the dentate infragranular zone) has been found following initiation of learning in numerous behavioral tasks. For example, a temporal modification of PSA-NCAM levels is known to occur 10-12 h after passive avoidance [88-92], water maze [93, 94], olfactory learning [95, 96] and contextual fear conditioning [97, 98]. This up-regulation in PSA-NCAM can be sustained, being in some cases also evident at 24 h post-training [68, 94]. Moreover, repetitive training in the water maze induces PSA-NCAM upregulation following each training session while animals are still improving their performance levels [93] suggesting that enhanced expression of PSA-NCAM in the hippocampus is a molecular signature of plasticity-related to hippocampal learning. Despite PSA-NCAM being a marker for immature neurons, the spatial learning-dependent increase in PSA-NCAM does not result from increased neurogenesis or progenitor cells survival [95], indicating some other function for this selective modification in PSA-NCAM. Together, these studies suggest that selected enhancement of hippocampal PSA-NCAM can facilitate memory formation, whereas the role of amygdaloid PSA-NCAM in memory function deserves further studies.

NCAM and PSA-NCAM: Sensitive Indices of “Emotional Learning”

In this review about the role of NCAM and PSA-NCAM in learning, we have detailed many instances were both NCAM and PSA-NCAM are important for “emotional learning” particularly in the context of auditory and contextual fear conditioning. We should also note that in our view the concept of “emotional” learning goes beyond fear learning tasks, since in our view virtually all animal learning models involve an emotional motivation too (for example, escaping from water stress in the water maze task). Work from our lab has found that NCAM [99] and PSA-NCAM [98] expression in the hippocampus are regulated by emotional learning depending on the intensity of the emotional experience. Contextual fear conditioning was found to induce time- and shock-intensity dependent alterations in the expression of hippocampal NCAM and PSA-NCAM. The intensity of the training experience can be modulated by altering the shock received by the animal, applying 0.2, 0.4 or 1 mA, which corresponds to low, medium and high intensity shocks, respectively. Previously, we showed that the intensity of the shock has a positive correlation with both the extent and duration of conditioned fear and post-training corticosterone levels [100]. Training rats at a moderate intensity (0.4 mA) led to a significant enhancement of hippocampal PSA-NCAM 12 h post-training [98], similarly to changes found after passive avoidance conditioning and spatial learning [68, 88]. By contrast, 24 h hours post-training only animals trained at 1 mA showed a significant enhancement of NCAM expression and, interestingly, this group also exhibited the greatest retention of the task and highest post-training corticosterone induction.

We have recently examined the regional specificity of contextual fear conditioning on hippocampal PSA-NCAM expression. We found [97] differential expression of hippocampal PSA-NCAM in the ventral and dorsal hippocampus that corresponds to a different functional involvement of these discrete regions in learning tasks [51]. Context exposure alone led to a significant increase in PSA-NCAM in the ventral and dorsal hippocampal dentate gyrus 24 h post-training [97]. However, following training in the contextual fear conditioning task (i.e., when context was paired with a shock), PSA-NCAM expression was only enhanced in the dorsal hippocampus. Moreover, infusion on Endo-N to the dorsal, but not the ventral, hippocampus impaired retention of the contextual memory [97]. More recently, we have demonstrated that prevention of a very rapid reduction in PSA-NCAM in the ventral hippocampus of rats exposed to the radial arm water maze is linked to a facilitation of memory retrieval (Conboy et al. unpublished observations).

Exposing rats during 30 min to a traumatic experience (i.e., predator stress, more specifically cat exposure) immediately following massed training in the radial arm water maze can impair recall of the platform location and induce a correlative reduction in hippocampal NCAM180 expression [101]. In contrast, the novel anti-depressant treatment agomelatine can prevent this stress-induced deficit in memory retrieval. Our recent findings show that in parallel to this behavioral effects, agomelatine facilitates the synaptic insertion of NCAM within 30 min of cessation of the learning task (Conboy et al. unpublished observations).

For a more comprehensive list of the role of NCAM and PSA-NCAM in learning and memory see Table 2.

Table 2 NCAM and PSA-NCAM in learning and memory
Full size table

Mechanisms Related to NCAM Actions on Learning

Humans retain the ability to form new memories, in the absence of dementia, throughout their whole life, which indicates that the implicated brain structures must retain the potential to continuously restructure their synapses. Adult learning and memory has been proposed to imply, to a certain extent, a replay of neurodevelopmental events and as such utilize the same plasticity-related molecules, including NCAM [102, 103]. Numerous studies have shown modifications of hippocampal synaptic morphology as a result of learning and memory [104-107]. As the selective expression of cell adhesion molecules during neurodevelopment is important in determining synaptic structuring [19], analogous cell adhesion molecule modulations presumably regulate synaptic restructuring during memory formation.

At the conceptual level, the reviewed evidence led us to propose the following role for NCAM during memory consolidation. During the early consolidation period, 2-6 h following training NCAM expression seems to be downregulated [67]. A transient increase in spine number has been found in the hippocampus in the early hours after a training experience [104-106]. In Aplysia, it has been shown that the growth of new synaptic connections requires endocytosis and degradation of the NCAM homologue apCAM [79]. It is therefore conceivable that the temporal reduction in NCAM found in rodents may similarly enable synaptic loosening to facilitate synaptic growth. During the later periods of consolidation, robust learning and memory has been associated with an enhancement of synaptosomal NCAM expression [68, 98]. In chicks, avoidance training induces the localisation of NCAM in the synaptic active zone of the lobus parolfactorius 5-6 h after a one-trial passive avoidance learning experience [73], indicating that NCAM may localize to newly formed synapses. Moreover, contextual fear conditioning leading to a strong memory [100] correlates with an enhancement of synaptically localized NCAM [98]. Correlative work in vitro has suggested that increased concentrations of NCAM can selectively increase synapse formation. For example, transfection of NCAM deficient neurons with any of the three NCAM molecules leads to the formation of synapses preferentially between NCAM-NCAM containing neurons [108].

Learning and memory also require temporal modulations in PSA-NCAM. Learning induced synaptic modifications occurring in the hippocampus are transient [106] indicating that a period of synaptic pruning or selection also contributes to memory consolidation. As PSA attachment to NCAM reduces NCAM mediated cell adhesion, activity dependent upregulation in PSA-NCAM during the later memory consolidation period (12-24 h) [68, 88] may enable synaptic loosening which facilitates selection and pruning of hippocampal connections.

In conclusion, we have presented evidence that NCAM and PSA-NCAM regulate both emotion and learning and memory processes, and we have presented a model suggesting that the functioning of these molecules might be related to the modulation of learning induced by emotional aspects. This evidence supports the use of recently developed NCAM-related compounds, such as the FGL peptide, for the treatment of devastating neurological disorders of cognitive dysfunction like Alzheimer’s disease.