A Synthetic Peptide Ligand of Neural Cell Adhesion Molecule (NCAM) IgI Domain Prevents NCAM Internalization and Disrupts Passive Avoidance Learning


  • Andrew G. Foley,

    1. Department of Pharmacology, Conway Institute, University College Dublin, Dublin, Ireland*Protein Laboratory, Panum Institute, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Barbara P. Hartz,

    1. Department of Pharmacology, Conway Institute, University College Dublin, Dublin, Ireland*Protein Laboratory, Panum Institute, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • * Helen C. Gallagher,

    1. Department of Pharmacology, Conway Institute, University College Dublin, Dublin, Ireland*Protein Laboratory, Panum Institute, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Lars C. B. Rønn,

    1. Department of Pharmacology, Conway Institute, University College Dublin, Dublin, Ireland*Protein Laboratory, Panum Institute, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • Vladimir Berezin,

    1. Department of Pharmacology, Conway Institute, University College Dublin, Dublin, Ireland*Protein Laboratory, Panum Institute, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • * Elisabeth Bock,

    1. Department of Pharmacology, Conway Institute, University College Dublin, Dublin, Ireland*Protein Laboratory, Panum Institute, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author
  • and * Ciaran M. Regan

    1. Department of Pharmacology, Conway Institute, University College Dublin, Dublin, Ireland*Protein Laboratory, Panum Institute, University of Copenhagen, Copenhagen, Denmark
    Search for more papers by this author

  • Lippincott Williams & Wilkins, Inc., Philadelphia

  • Abbreviations used: C3d, dendrimer of C3 peptides; NCAM, neural cell adhesion molecule; PAGE, polyacrylamide gel electrophoresis; SDS, sodium dodecyl sulphate; TBS-T, 10 mM Tris-HCl, 150 mM NaCl, and 0.5% (vol/vol) Tween-20 (pH 7.4).

Address correspondence and reprint requests to Prof. C. M. Regan at Department of Pharmacology, University College Dublin, Belfield, Dublin 4, Ireland. E-mail: Ciaran.Regan@ucd.ie


Abstract: The neural cell adhesion molecule (NCAM) mediates cell adhesion and signal transduction through trans-homophilic- and/or cis-heterophilic-binding mechanisms. Intraventricular infusions of anti-NCAM have revealed a functional requirement of NCAM for the consolidation of memory in rats and chicks in a specific interval 6-8 h after training. We have now extended these studies to a synthetic peptide ligand of NCAM (C3) with an affinity for the IgI domain and the capability of inhibiting NCAM-mediated neurite outgrowth in vitro. Intraventricular administration of a single 5 μg bolus of C3 strongly inhibited recall of a passive avoidance response in adult rats, when given during training or in the 6-8-h posttraining period. The effect of C3 on memory consolidation was similar to that obtained with anti-NCAM as the amnesia was not observed until the 48-h recall time. The unique amnesic action of C3 during training could be related to disrupted NCAM internalization following training. In the 3-4-h posttraining period NCAM 180, the synapse-associated isoform, was down-regulated in the hippocampal dentate gyrus. This effect was mediated by ubiquitination and was prevented by C3 administration during training. These findings indicate NCAM to be involved in both the acquisition and consolidation of a passive avoidance response in the rat. Moreover, the study provides the first in vivo evidence for NCAM internalization in learning and identifies a synthetic NCAM ligand capable of modulating memory processes in vivo.

Change in synapse number and/or structure is a consistent correlate of learning across species and paradigms (Geinisman et al., 1991; Doubell and Stewart, 1993; Hunter and Stewart, 1993; O’Malley et al., 1998; Toni et al., 1999). Considerable evidence suggests these changes to be reliant on molecular functions mediated by immunoglobulin superfamily cell adhesion molecules (Rose, 1996). Antibody interventive studies have demonstrated the neural cell adhesion molecule (NCAM) and L1 to be required for the induction and maintenance of long-term potentiation and in the stable retention of memory for spatial, appetitive, or aversive tasks (Doyle et al., 1992a; Lüthl et al., 1994; Rønn et al., 1995; Arami et al., 1996; Alexinsky et al., 1997). In behaving animals, NCAM function is restricted to a 6-8-h posttraining period as administration of antibodies before this time, or immediately following it, is without effect (Doyle et al., 1992a; Scholey et al., 1993; Alexinsky et al., 1997; Roullet et al., 1997). Similarly, L1 function is necessary during task acquisition and in the 6-8-h and 15-18-h posttraining periods of consolidation (Lühti et al., 1994; Scholey et al., 1995; Tiunova et al., 1998).

NCAM involvement in memory most likely involves change in cell adhesion and signal transduction events through modulation of trans-homophilic and/or cis-heterophilic interactions. These changes may regulate synaptic growth as evidenced by the doubling of NCAM-immunopositive synaptic zones observed in the chick lobus parolfactorius at 5.5 h posttraining in an avoidance learning paradigm (Skibo et al., 1998). In the rodent, the functional requirement of NCAM during the consolidation of a passive avoidance response coincides with the 6-8-h posttraining period in which a transient twofold increase in axospinous synapses is observed in the hippocampal dentate gyrus (O’Malley et al., 1998). The loosening of cell—cell interactions that would be necessary to permit the synaptic growth associated with memory consolidation may also be dependent on modulation of NCAM function. ApCAM, the Aplysia NCAM homologue, has been demonstrated to down-regulate rapidly and to become internalized in an in vitro model of long-term sensitization of the gill and siphon-withdrawal reflex (Bailey et al., 1992; Mayford et al., 1992).

The NCAM trans-homophilic binding mechanism has been proposed to involve a double reciprocal interaction between the first and second Ig domains of opposing NCAM molecules (Kiselyov et al., 1997; Thomsen et al., 1998; Atkins et al., 1999; Jensen et al., 1999). Recently, an undecapeptide (C3) that binds to NCAM IgI with a dissociation constant similar to that for NCAM IgII has been identified (Kiselyov et al., 1997; Rønn et al., 1999). In vitro, this peptide has been shown to stimulate neurite outgrowth and to interfere with NCAM homophilic binding mediated by the IgI and IgII domains. In vivo, we now demonstrate that intraventricular infusion of the C3 peptide during training and in the 6-8-h posttraining period results in amnesia of the passive avoidance response in the adult rat and prevents NCAM internalization in the 3-4-h period following task acquisition.



The anesthetic agents ketamine hydrochloride (Vetalar) and xylocaine (Rompun) were purchased from Parke-Davis, U.K. and Bayer, U.K., respectively. Antibodies to ubiquitin C-terminal hydrolase and ubiquitin protein conjugates were purchased from Affiniti Research Products Ltd., U.K. Secondary antibodies and routine laboratory chemicals were from Sigma, U.K.

Animal maintenance

Postnatal day 80 male Wistar rats (weighing 300-350 g) were obtained from the Biomedical Facility, University College, Dublin, Ireland, and housed individually in a 12-h light/dark cycle with food and water available ad libitum. Animals were introduced, maintained, and handled in the test environment for 3 days before the commencement of studies. Moreover, each day the animals were placed in an open field apparatus (80 × 80 × 15 cm, divided into 64 squares), where their locomotor activity was assessed by the number of lines crossed in a 5-min period. Analysis of behavioural anomalies used the nonparametric Kruskal—Wallis test.

Passive avoidance paradigm

The procedure was identical to the one that we have described previously (Fox et al., 1995). In brief, animals were trained in a one-trial, step-through, light—dark passive avoidance paradigm. The smaller, illuminated compartment was separated from a larger, dark compartment by a shutter, which contained a small entrance. The floor of the training apparatus consisted of a grid of stainless steel bars that could deliver a remotely controlled, scrambled shock (0.75 mA every 0.5 ms) of 5 s in duration when the animal entered the dark chamber. The animals were tested for recall of this inhibitory stimulus at 24 and 48 h posttraining by placing them in the light compartment and noting their latency to enter the dark compartment. A criterion period of 600 s was used. Values significantly different from the control were determined using the Mann—Whitney U test for nonparametric data, and p values of <0.05 were considered to be significant. Immediately before training and recall, animals were placed in an open field apparatus, and their spontaneous locomotion was monitored.

Intracerebroventricular administration of the dendrimer of C3 peptides (C3d)

Animals were anesthetised by intraperitoneal administration of a solution (1 ml/kg) containing 4 parts ketamine hydrochloride (100 mg/ml) and 3 parts xylocaine (20 mg/ml). Depth of anaesthesia was assessed by monitoring respiration rate and palpebral and pedal withdrawal reflexes. Once the animal was anaesthetised, the skull was exposed and trepanned (1 mm in diameter), and a guide cannula was stereotactically implanted (1 mm deep) at a position 1 mm posterior and 1.6 mm lateral to bregma. Two flat-bottom stainless steel screws were placed laterally to the cannula, and the entire complex was secured with dental acrylic (Simplex Rapid). When the acrylic had polymerised, a dummy cannula, cut to the same dimensions as the guide cannula, was screwed into place to maintain the patency of the guide cannula and to minimise possible infection. Thereafter, the incision was closed with interrupted silk sutures, and the animal was placed in a heated cage (35°C) and monitored carefully until recovery was complete. No further procedures were allowed for a minimum of 1 week following surgery. During this period body weight was monitored, and the lack of anomalies was confirmed by ANOVA.

The C3 peptide (ASKKPKRNIKA) was administered in dendrimer form (C3d), which consisted of four peptide monomers coupled to a lysine backbone. Animals were restrained but conscious, and the peptide was administered 20 min before training or at the described posttraining time points. The dummy cannula was removed, and the patency of the guide cannula was confirmed using a spare internal cannula. The C3d peptide was delivered by inserting an internal cannula (28 gauge, 16 mm long) through the guide cannula and into the right lateral ventricle. The C3d peptide was prepared in sterile 0.9% NaCl, and a 5-μl aliquot was delivered at a rate of 1 μl/min. An alanine-substituted peptide dendrimer (ASAAPAANIKA), which was inactive in vitro, served as a control. This is one of a series of control peptides previously determined to lack the neuritogenic properties of C3d (Rønn et al., 1999). One minute following administration, the internal cannula was slowly withdrawn and replaced with the dummy cannula, and the animal was returned to its home cage.

All experimental procedures were approved by the Review Committee of the Biomedical Facility of University College, Dublin, and were carried out by individuals who held the appropriate licence issued by the Department of Health.

Preparation of hippocampal dentate gyrus homogenates

Animals were killed at various posttraining times by cervical dislocation. The dentate gyrus was bilaterally dissected from the hippocampus and hand-homogenised in ice-cold 0.32 M sucrose containing 1 mM 3-isobutyl-1-methylxanthine. Light microscopic analysis of sections from the dissected region indicated it to comprise the hilus and dentate gyrus proper and in some cases a small amount of the CA4 region of the hippocampus. Protein concentration was determined according to the methods of Lowry et al. (1951). Samples containing equal protein were prepared for electrophoresis in sodium dodecyl sulphate (SDS)—polyacrylamide gel electrophoresis (PAGE) sample buffer containing 62.5 mM Tris-HCl (pH 6.8), 2% SDS, 10% glycerol, 0.01% bromophenol blue, and 42 mM dithiothreitol and boiled for 10 min.

NCAM immunoprecipitation

Samples containing 500 μg of protein in a 250-μl volume were diluted 1:2 with 2× immunoprecipitation buffer [1× = 1% (vol/vol) Triton X-100, 150 mM NaCl, 10 mM Tris (pH 7.4), 1 mM EDTA, 1 mM EGTA, 0.2 mM phenylmethylsulphonyl fluoride, and 0.5% (vol/vol) Nonidet P-40] and immunoprecipitated with rabbit anti-rat NCAM IgG (1:500) overnight at 4°C. Protein A-Sepharose 4-B beads were washed and preswollen for 2 h in immunoprecipitation buffer at 4°C. A 100-μl aliquot of a 50% protein A suspension was added to each tube, containing a total volume of 500 μl, and rocked for 2 h at 4°C. The immunoprecipitate was pelleted by centrifugation at 10,000 rpm for 10 min at 4°C and washed three times in 1× immunoprecipitation buffer. The final pellet was resuspended in 50 μl of SDS-PAGE sample buffer and boiled for 10 min to release the immunoprecipitable fraction from the bead. The sample was centrifuged at 14,000 rpm for 2 min to pellet the Sepharose beads, and the supernatant was loaded onto SDS-PAGE gels for immunoblot analysis as described below.

SDS-PAGE and immunoblotting

Proteins were separated on 7.5% or 15% polyacrylamide minigels at 100 V constant voltage and transferred to nitrocellulose membranes in a wet transfer system (200 V constant voltage) using standard techniques. NCAM immunoblotting was performed with a polyclonal antibody that recognises all isoforms, under conditions described by Andersson et al. (1993). Polyclonal antibodies to ubiquitin C-terminal hydrolase and ubiquitin protein conjugates were used. Following transfer, the membrane was blocked in 2% bovine serum albumin diluted in 10 mM Tris-HCl, 150 mM NaCl, and 0.05% (vol/vol) Tween-20 (pH 7.4; TBS-T) for 1 h. Primary antibody was diluted 1:1,000 in TBS-T and incubated overnight at 4°C. The membrane was washed extensively and incubated with a peroxidase-conjugated goat anti-rabbit IgG secondary antibody (1:20,000 in TBS-T) for 2 h at room temperature. After further washing, the blot was developed with an enhanced chemiluminescent detection system (Pierce, U.K.) and exposed to x-ray film for appropriate times. Semiquantitative scanning densitometry was performed using the Imagequant software package (Molecular Dynamics). Bands for analysis were outlined, and their areas and/or volumes were integrated to obtain an indication of band intensity.


Initially, the influence of the C3d peptide on memory consolidation was determined at the 6-h posttraining period at which administration of NCAM antibodies is known to produce amnesia (Doyle et al., 1992a; Alexinsky et al., 1997). Following a single intraventricular bolus of 5 μg of C3d, the animals exhibited a significant decrease in their 48-h recall latency with respect to animals treated with saline and control peptide (Fig. 1). Thus, the slowly emerging amnesia induced by C3d, and also observed with NCAM antibodies, is not attributable to decay of task recall over time as it was not observed in animals receiving saline or the inactive peptide. At the 24-h recall time, the escape latency times in C3d-treated animals tended to become reduced; however, this was only significant with respect to the peptide control. The effect was time- and dose-dependent as the peptide was without effect at the 24- or 48-h recall times when administered at the 10-h posttraining period or at the lower dose of 0.7 μg. Unexpectedly, C3d also induced a slowly emerging amnesia when administered just before training. This effect was distinct from the amnesic action observed at the 6-h posttraining time, as administration of an equivalent dose of C3d in the intervening period (3 h posttraining) was without effect on recall at either the 24- or 48-h recall time. None of the amnesic actions of C3d appeared to be attributable to gross behavioural anomalies, as no significant difference in locomotor activity was observed by open-field analysis (Table 1). Neither were they due to extinction as a result of the first (24-h) recall because none of the animals reentered the dark chamber during this period and control animals, which did not display amnesia, were also subjected to both recalls.

Figure 1.

Time-dependent influence of C3d on passive avoidance learning. The agents were administered intracerebroventricularly (ICV) 20 min before training (0 h) or at the other indicated posttraining times. The escape latencies at the 24-h (open columns) and 48-h (solid columns) recall times are expressed as mean ± SEM (bars) values, and n values are indicated in the columns. *p < 0.05, significantly different from saline control; †p < 0.05, significantly different from the control (cont) peptide.

Table 1. Effect of intracerebroventricular administration of C3d on locomotion in an open field apparatus
Treatment, administration time (h)Time of assessment (h)
  1. Compounds were administered at the indicated posttraining time points with administration at 0 h occurring 20 min before training. Locomotion was assessed in the open field apparatus 72, 48, and 24 h pretraining and 0, 24, and 48 h posttraining. Data are mean ± SEM values (4 ≤ n ≤ 6).

0120.0 ± 10.9108.5 ± 10.1107.7 ± 20.2127.0 ± 13.2129.3 ± 19.2130.0 ± 21.0
3199.6 ± 6.2191.0 ± 5.5197.2 ± 12.0171.2 ± 12.3150.8 ± 21.7161.2 ± 20.2
6114.0 ± 14.7135.3 ± 17.1122.0 ± 10.7161.2 ± 9.8129.3 ± 25.295.7 ± 22.9
10165.3 ± 7.1165.7 ± 4.2165.3 ± 12.2169.3 ± 15.4164.8 ± 6.6172.0 ± 5.5
C3d control      
6113.5 ± 14.6167.5 ± 58.9159.5 ± 78.4164.0 ± 33.7107.5 ± 52.0141.0 ± 19.7
C3d (0.7 μg)       
6143.2 ± 9.3145.3 ± 14.2162.4 ± 12.5172.3 ± 21.5169.5 ± 17.8168.4 ± 16.0
C3d (5.0 μg)       
0133.0 ± 17.0123.2 ± 14.0132.3 ± 15.3189.7 ± 17.7147.0 ± 18.2163.3 ± 18.2
3170.7 ± 11.4159.0 ± 9.2163.3 ± 14.5176.3 ± 18.1149.0 ± 19.5149.0 ± 22.6
6135.4 ± 20.0157.0 ± 20.2140.0 ± 9.5123.7 ± 11.6143.3 ± 23.2159.7 ± 17.3
10157.6 ± 21.2155.6 ± 17.7146.0 ± 18.0142.0 ± 18.3142.0 ± 26.7155.6 ± 17.7

FIG. 1.


Given the identified functional requirement of NCAM during memory acquisition, we determined if this was followed by periods of NCAM internalization within the temporal frame identified by the C3d peptide ligand. NCAM expression was analyzed in homogenates of hippocampal dentate gyrus obtained from animals killed at increasing posttraining times. All animals analysed had acquired the task, with their escape latencies exceeding the criterion time used. In comparison to naive animals (data not shown) or those killed immediately following training, only those in the 3-6-h posttraining period had altered NCAM expression. These had significantly reduced levels of the 180-kDa NCAM isoform (Figs. 2A and 3A). At 12 h posttraining varying amounts of NCAM180 were observed compared with controls, although the difference was nonsignificant (Fig. 3A). Reduced expression of NCAM180 at 3-6 h posttraining was correlated with an increased expression of ubiquitin C-terminal hydrolase during the same period (Figs. 2B and 3D), suggesting that during memory consolidation NCAM may undergo regulated proteolysis by the ubiquitin-dependent pathway.

Figure 2.

NCAM (A) and ubiquitin C-terminal hydrolase (B) expression in the rat dentate gyrus following passive avoidance training. Animals were killed at the indicated times posttraining, and homogenates prepared from dissection of the dentate gyrus were subjected to SDS-PAGE and immunoblot analysis with polyclonal antibodies. Animals killed immediately following training served as a control.

Figure 3.

Densitometric quantification of NCAM and ubiquitin C-terminal hydrolase expression following avoidance training: 180-, 140-, and 120-kDa NCAM isoforms (A—C, respectively) and ubiquitin C-terminal hydrolase expression (D). The integrated areas or volumes of the relevant bands yielded a value in arbitrary units that was expressed as a percentage of the value for a control animal killed immediately following training, with this being normalised to 100% in each case. Data are mean ± SEM (bars) values of three separate and independent experiments. *p < 0.05, significantly different from control.

FIG. 2.

FIG. 3.

To confirm this assumption we sought to demonstrate directly ubiquitination of NCAM. Immunoprecipitation of NCAM from the dentate fraction was followed by immunoblot analysis with a polyclonal antibody recognising ubiquitinated proteins. This indicated all NCAM isoforms to be ubiquitinated in the 3-4-h posttraining period, consistent with the view that increased expression of the hydrolase reflects targeted degradation of NCAM (Fig. 4). It is notable that there was no evidence of NCAM ubiquitination at 6 h posttraining, indicating that active proteolysis following learning must be additionally directed towards other proteins because ubiquitin C-terminal hydrolase expression was increased at this time point.

Figure 4.

Immunoblot detection of ubiquitinated NCAM following passive avoidance training. Dentate gyrus homogenates, prepared at various posttraining times, were subjected to immunoprecipitation with polyclonal anti-NCAM, and the fraction collected on protein A-Sepharose was immunoblotted with an antibody recognising ubiquitinated proteins. Equal protein loading and immunoprecipitation efficiency are indicated by the 55-kDa Ig heavy chain band, and the molecular masses of the major NCAM isoforms are indicated by the arrows.

FIG. 4.

Given that task acquisition was followed by NCAM internalization, we determined the influence of prior C3d administration on this process. Animals rendered amnesic by C3d administration just before task acquisition and killed 4 h posttraining exhibited reduced levels of ubiquitin-mediated NCAM degradation. This time point was selected as it occurs midway through the established transient period of increased proteolysis during memory consolidation. In comparison with saline-treated control animals, those exposed to C3d had a higher expression of the three major NCAM isoforms in the dentate gyrus (Fig. 5A). In support of this finding, ubiquitin C-terminal hydrolase expression was significantly reduced 4 h posttraining by C3d administration (Fig. 5B). This provides evidence to support both ubiquitin-mediated NCAM proteolysis as a learning-specific event and C3d as an NCAM-directed amnesic agent.

Figure 5.

Influence of C3d administration on learning-associated modulation of (A) NCAM isoform and (B) ubiquitin C-terminal hydrolase expression. Animals were treated with either saline or C3d by intracerebroventricular administration immediately before training, and dentate gyrus homogenates were prepared after the animal was killed at 4 h posttraining (n = 4). As all samples were electrophoresed on a single blot for each protein, normalisation of densitometric values was not required, and these are indicated by arbitrary units. Data are mean ± SEM (bars) values of the four samples for each treatment. *p < 0.05, significantly different from control.

FIG. 5.


These results demonstrate that the C3d synthetic peptide ligand of the NCAM IgI domain prevents the consolidation of the passive avoidance response in two distinct time periods. The amnesic action of the C3d peptide, when administered at the 6-h posttraining time, coincides with the period in which NCAM function has been previously demonstrated to play a critical role in information processing. However, the amnesia observed following administration of C3d during training was unexpected because administration of NCAM antibodies at this time is without effect (Doyle et al., 1992a; Scholey et al., 1993, 1995). Nevertheless, the action of the C3d peptide is most likely specific to NCAM as analyses of its binding properties indicate a specificity for the IgI-like NCAM domain and not the structurally similar NCAM IgII domain (Rønn et al., 1999), and, as such, it is unlikely to bind other nonhomologous Ig domains. Moreover, in both cases, the slow emergence of the amnesic effect, which only becomes apparent at the 48-h recall time, is similar to that observed following administration of anti-NCAM (Nolan et al., 1987; Doyle et al., 1992a; Alexinsky et al., 1997).

As the requirement of NCAM trans-homophilic binding during acquisition of a passive avoidance response is not known to be associated with de novo synapse growth, at least in the dentate gyrus of the hippocampal formation (O’Malley et al., 1998), we investigated the possibility that NCAM trans-homophilic binding was involved in regulating NCAM internalization as suggested originally by the in vitro studies of Kandel and colleagues (Bailey et al., 1992; Mayford et al., 1992). These studies provided evidence for a similar NCAM internalization mechanism in the early events of memory formation and that its degradation involved the ubiquitin pathway. Although NCAM degradation was primarily directed to the synapse-specific 180-kDa isoform during the 3-6-h period, which coincided with the increased expression of ubiquitin C-terminal hydrolase, the ubiquination of all three isoforms was unexpected. This would imply that degradation of all three NCAM isoforms occurs in the period immediately following training. The significant degradation of the 180-kDa isoform is consistent with in vitro observations indicating the 140-kDa isoform to have a much faster turnover rate (Park et al., 1997). Such differential turnover rates may also explain the apparent discrepancy in the degree of ubiquitination we observed between different isoforms because larger amounts of NCAM140 would thus be targeted for degradation. Moreover, the observation that C3d-induced amnesia increases the expression of all NCAM isoforms tends to support the premise that all isoforms are involved in the structural remodelling that accompanies the early phase of memory consolidation and that the action of this peptide is not limited to the NCAM180 isoform.

The role of NCAM in the 6-8-h posttraining period has been attributed to its involvement in a time-dependent, selective stabilisation of synapses from a population transiently overproduced following learning (Doyle et al., 1992b). Recent evidence supports this assumption as direct morphological studies have revealed a transient, learning-associated period of synaptic growth to coincide with the posttraining period in which anti-NCAM is effective (O’Malley et al., 1998). Time-dependent increases in synapse number are also observed following the induction of long-term potentiation, and at later times the frequency of synapses enriched in the synapse-specific NCAM180 isoform almost doubles (Schuster et al., 1998; Toni et al., 1999). However, NCAM may also strengthen existing synapses as following avoidance learning in the chick the frequency of NCAM-labelled synapses increases at the 5.5-h posttraining time, and this is in advance of the period in which de novo synapse formation is observed in the same brain region (Hunter and Stewart, 1993; Skibo et al., 1998).

The ability of C3d to induce amnesia at separate times in the cascade of molecular events that follow learning does not necessarily imply that the distinct NCAM functions during task acquisition and consolidation are unrelated. Interfering with NCAM homophilic binding may have the dual consequence of reducing synapse stability immediately following training and preventing de novo synapse growth in the 6-8-h period of consolidation. This common mechanism may be mediated by altered signal transduction and/or cell adhesion events that are known to be dependent on NCAM—NCAM interactions (Williams et al., 1994; Beggs et al., 1997; Saffell et al., 1997; Schmid et al., 1999). It is also possible that the C3d ligand may indirectly disrupt the carbohydrate-dependent cis-binding mechanism between L1 and NCAM as L1 function is known to be required during task acquisition and in the 6-8-h posttraining period (Kadmon et al., 1990; Horstkorte et al., 1993; Scholey et al., 1995). However, this is considered unlikely as the cis interaction between these cell adhesion molecules occurs through the IgIV domain of NCAM, whereas the C3d ligand binds to the IgI module (Horstkorte et al., 1993; Rønn et al., 1999). It is more likely that the amnesia induced by the C3d ligand arises from perturbation of trans-homophilic binding involving NCAM Ig domains I and II.

The importance of NCAM—NCAM trans-homophilic binding during task acquisition and consolidation may serve the common function of linking transient patterns of neural activity to adhesive changes in synapses. Although such a role is speculative, it is clear that NCAM degradation during learning occurs in a temporally regulated and highly directed manner involving transient activation of the ubiquitin pathway. Similarly, the time-dependent activation of ubiquitin C-terminal hydrolase during long-term sensitization in Aplysia (Hedge et al., 1997) and the proteolytic events underlying NMDA receptor activation and the induction of long-term potentiation in vitro (Vanderklish et al., 1995; Hoffman et al., 1998; Lynch, 1998) support our in vivo observations on the requirement of regulated proteolysis in the early events of memory consolidation. Moreover, such synaptic destabilisation may serve as a clearing event before the subsequent synaptic growth that follows the induction of long-term potentiation or the consolidation of a passive avoidance response (O’Malley et al., 1998; Toni et al., 1999). In these later periods of synapse growth trans-homophilic NCAM interactions may provide positive growth signals through increases in cell adhesion molecule-mediated signal transduction, cell adhesion molecule synthesis, and cell adhesion molecule-mediated synaptic strengthening.


This work was supported by an International Collaboration Grant from Enterprise Ireland, the Health Research Board of Ireland, the Lundbeck Foundation, a Brorsons research grant, the Danish Medical Research Council, and the Danish Cancer Society.