N-Cadherin Regulates Cytoskeletally Associated IQGAP1/ERK Signaling and Memory Formation

N-CADHERIN REGULATES CYTOSKELETALLY-ASSOCIATEDIQGAP1/ERK SIGNALING AND MEMORY FORMATION

Christina Schrick1, Andre Fischer2, Deepak P. Srivastava3, Natalie C. Tronson1, PeterPenzes3, and Jelena Radulovic1,*

1Department of Psychiatry and Behavioral Sciences, The Asher Center for Depressive Disorders,Northwestern University, Feinberg School of Medicine, Chicago, USA

2Neuropathology Group, European Neuroscience Institute, Goettingen, Germany

3Department of Physiology, Northwestern University, Feinberg School of Medicine, Chicago, USA

SummaryCadherin-mediated interactions are integral to synapse formation and potentiation. Here we showthat N-cadherin is required for memory formation and regulation of a subset of underlyingbiochemical processes. N-cadherin antagonistic peptide containing the His-Ala-Val motif (HAV-N)transiently disrupted hippocampal N-cadherin dimerization and impaired the formation of long-termcontextual fear memory while sparing short-term memory, retrieval and extinction. HAV-N impairedthe learning-induced phosphorylation of a distinctive, cytoskeletally-associated fraction ofhippocampal Erk-1/2 and altered the distribution of IQGAP1, a scaffold protein linking cadherin-mediated cell adhesion to the cytoskeleton. This effect was accompanied by diminished of N-cadherin/IQGAP1/Erk-2 interactions. Similarly, in primary neuronal cultures, HAV-N preventedNMDA-induced dendritic Erk-1/2 phosphorylation and caused relocation of IQGAP1 from dendriticspines into the shafts. The data suggest that the newly identified role of hippocampal N-cadherin inmemory consolidation may be mediated, at least in part, by cytoskeletal IQGAP1/Erk signaling.

KeywordsN-cadherin; HAV peptide; fear conditioning; Erk; IQGAP1; cytoskeleton

IntroductionClassic type I cadherins, including neuronal (N)-cadherin, play a critical role in thedevelopmental organization of the brain (Redies, 2000) and synapse formation in the adultCNS (Junghans et al., 2005). These processes are regulated by the extracellular domainsmediating cell-cell adhesion in a calcium-dependent, predominantly homophilic manner(Shapiro and Colman, 1998). Intracellularly, cadherins are anchored to the actin cytoskeletonby multiprotein complexes including beta- and alpha-catenin (Angst et al., 2001) and associatedwith docking proteins to intracellular signaling pathways (Husi et al., 2000) enabling thetranslation of cell adhesion signals into long-term cellular responses (Marambaud et al.,2003;Perron and Bixby, 1999;Widelitz, 2005).

*Corresponding author: Jelena Radulovic, Department of Psychiatry and Behavioral Sciences, The Asher Center for DepressiveDisorders, Northwestern University, Feinberg School of Medicine, 303 East Chicago Avenue, Ward 9-188, Chicago, IL 60611, Email:[email protected], Phone: 312 503 4627, Fax: 312 503 0466Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customerswe are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resultingproof before it is published in its final citable form. Please note that during the production process errors may be discovered which couldaffect the content, and all legal disclaimers that apply to the journal pertain.

NIH Public AccessAuthor ManuscriptNeuron. Author manuscript; available in PMC 2008 September 6.

Published in final edited form as:Neuron. 2007 September 6; 55(5): 786–798.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Earlier studies reported a role of N-cadherin in synaptic potentiation. By applying blockingantibodies or antagonistic peptides containing the His-Ala-Val (HAV) motif for classic type Icadherin dimerization, Tang et al. (1998) observed a significant impairment of hippocampallong-term potentiation (LTP). Bozdagi et al. (2000) subsequently reported similar findingswith N-cadherin blocking antibodies in a model of late-LTP, and additionally demonstratedthat synaptic potentiation increases the synthesis and recruitment of N-cadherin to newlyformed synapses. These findings suggested that N-cadherin–mediated cell adhesion maycontribute to learning and memory (Hagler and Goda, 1998;Murase and Schuman, 1999).

A role of cadherins in memory has not yet been documented. Contrary to expectation, micelacking the classical type II cadherin-11 displayed enhanced LTP and normal spatial learning(Manabe et al., 2000). On this basis, it was suggested that cadherins might be functionallyredundant thereby masking the effects of individual members (Togashi et al., 2002). The roleof classical type I cadherins, such as N-cadherin, could not be studied with constitutive geneknockouts due to embryonic lethality (Radice et al., 1997). By using an alternative approach,conditional expression of dominant negative protein in the adult brain, (Edsbagge et al.,2004) attempted to nonselectively impair the function of type I cadherins. Surprisingly, thesemice did not show any alterations of synaptic potentiation or spatial learning. It should benoted, however, that N-cadherin was produced normally in the brain of transgenic mice, andthat the expected functional consequences of the transgene could not be demonstrated due toits expression later in life.

To overcome possible compensatory or developmental effects of genetic manipulations, weemployed acute injection of antagonistic peptides containing the His-Ala-Val motif (HAV-N)to examine the role of N-cadherin in learning and memory. HAV-N containing aspartic (D)amino acid flanking the HAV sequence was shown to specifically bind to N-cadherin whencompared to other classic cadherins (Williams et al., 2000). On this basis as well as theirthorough characterization in electrophysiological experiments (Tang et al., 1998), we selectedHAV-N and a scrambled peptide (HAV-S) to perform these studies. The hippocampally-dependent contextual fear conditioning paradigm served to establish the effect of HAV-N onmemory formation, retrieval and extinction. Furthermore, we investigated the interference ofHAV-N with postadhesion signaling associated with N-cadherin complexes, includingextracellular signal-regulated kinases 1 and 2 (Erk-1/2) and IQGAP1 (Derycke and Bracke,2004).

We demonstrated a role of hippocampal N-cadherin in early mechanisms leading toconsolidation of long-term contextual fear memory. Biochemical findings suggested that thesemechanisms include the activation of a discrete fraction of cytoskeletal Erk-1/2 and itsinteraction with IQGAP1. N-cadherin-mediated cytoskeletal IQGAP1/Erk-1/2 signaling maybe important for synaptic remodelling associated with memory.

ResultsHAV-N impairs memory formation without affecting retrieval or extinction

Mice were implanted with microcannula into the dorsal hippocampus and subjected to contextand tone-dependent fear conditioning. Dorsohippocampal injection of HAV-N immediatelyafter training dose-dependently impaired contextual freezing (F (4,45) = 4.57, p < 0.05) duringthe memory test performed 24 hr later (Figure 1A). Tone-dependent fear conditioning (Figure1A) was not affected by any of the applied treatments (F (4,45) = 1.43, p = 1.22). The specificityof the most effective dose (750 ng/mouse) was compared to a group receiving the same amountof a scrambled peptide (HAV-S) or vehicle. Only HAV-N impaired long-term memory ofcontextual fear, as indicated by a significant reduction of freezing (F (2,33) = 6.16, p < 0.01)and increase of mean activity (F (2,33) = 3.251, p < 0.05) during the contextual memory test

Schrick et al. Page 2

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

performed 24 hr later (Figure 1B). HAV-N injected in separate groups of mice immediatelyposttraining did not affect freezing determined 1 hr later (F(2,30) = 0.57, p = 0.75), indicatingthat HAV-N did not affect short-term memory (Figure 1C). Additional time-course studiesshowed that HAV-N impaired long-term memory formation when injected 15 min beforetraining (F (2,28) = 4.31, p < 0.01), however injections performed 60 min posttraining did notsignificantly affect consolidation, as indexed by freezing behavior (F (2,29) = 1.98, p = 0.93;Figure 2A). Pretraining injections of HAV-N did not reveal any effects of the peptide on meanactivity (F (2,28) = 0.62, p = 0.87) or activity burst to the shock (F(2,28) = 0.34, p = 0.43). Thegroup values (in cm/s) for activity were: vehicle 11.5 ± 1.3; HAV-S 11.8 ± 0.8; HAV-N 11.1± 0.9; and for shock response: vehicle 41.9 ± 5.6, HAV-S 43.3 ± 4.7, HAV-N 39.7 ± 4.9.

To test the effect of N-cadherin signaling on memory retrieval and extinction, HAV-N wasinjected 15 min before the test (Figure 2B) or immediately after nonreinforced trials (Figure2C), respectively. Neither treatment affected freezing behavior when compared to the controlinjections (pre-test: F (2,27) = 1.79, p = 0.98; extinction: F (8,135) = 0.56, p = 1.85). These resultsindicated that the HAV-N antagonistic peptide significantly impaired memory consolidationwithout affecting retrieval or extinction of contextual fear.

HAV-N blocks N-cadherin dimerizationHAV-N is expected to interfere with N-cadherin function by decreasing the dimerization ofN-cadherin molecules. By employing one way ANOVA with factor Group (naïve, vehicle,HAV-S, and HAV-N) for each dose and time point or two way ANOVA with Treatment andTime as factors, we analysed the dorsohippocampal levels of N-cadherin monomers and dimersafter HAV-N treatment (Figure 3A-3D). All mice except for those of the naïve group wereexposed to fear conditioning and injected immediately thereafter with vehicle, HAV-S orHAV-N. Immunoblot signals of total dorsohippocampal lysates showed a significant increaseof N-cadherin monomers after fear conditioning (Figure 3A, 3B) in all groups when comparedto the naïve (Group: F(18, 76) = 18.7, p < 0.01) group independently of treatment. Consideringthe very short lifetime (Klingelhofer et al., 2002) and low level (Bozdagi et al., 2000) ofcadherin dimers, the effect of HAV-N on dimerization was examined under nonreducingconditions ((-β-mercaptoethanol; -β-ME). This approach, selected to preserve disulfide bondspromoting and stabilizing cadherin dimerization (Makagiansar et al., 2002), enhanced dimerdetection in the control groups (Figure 3C). N-cadherin dimers in the membrane fractions werealso significantly elevated after fear conditioning (vehicle and HAV-S groups) when comparedto naïve mice. HAV-N significantly impaired N-cadherin dimer formation depending on thedose (F(3, 56) = 116.446, p < 0.001) and postinjection time (F(2,56) = 63.067, p < 0.001). Post-hoc analyses revealed significant effects of 500 ng/mouse (Figure 3D) and 750 ng/mouse(Figure 3C-3D) of HAV-N at the time points 15 and 30 min posttraining when compared tothe other groups. Analyses of membrane fractions of dorsohippocampal lysates obtained fromnaïve, vehicle-, or HAV-S-treated mice showed weak but clear bands at 230 kDa (Bozdagi etal., 2000) under nonreducing but not under reducing (+βME) conditions, indicating N-cadherindimers (Figure 3C). This band was not detectable in lysates obtained 15 and 30 min after HAV-N injection and reappeared, although with diminished intensity in the 500 (F(2,12) = 4.65, p <0.05) and 750 ng groups (F(2,12) = 4.12, p < 0.05), after 1 hr. The decrease of N-cadherindimerization by HAV-N (Figure S1C) was confirmed by a mouse monoclonal N-cadherinantibody (clone G-C4, Sigma). These results showed that the behaviorally effective doses ofHAV-N prevented N-cadherin dimerization in a restricted time window posttraining. Notably,the effect of HAV-N on N-cadherin dimerization was dissociable from N-cadherin production.Thus, N-cadherin dimerization early after fear conditioning and its later up-regulation mayindependently contribute to the early and late biochemical processes leading to long-termmemory formation. Attenuated N-cadherin cell-cell adhesion activates the small GTP-ase Rac1(Charrasse et al., 2002), whereas disruption of cell-cell adhesion may alter the subcellular

Schrick et al. Page 3

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

distribution of N-cadherin and beta-catenin (Rhee et al., 2002). To examine whether HAV-Naffected these parameters associated with N-cadherin function, we determined the levels of N-cadherin and beta-catenin as well as the localization and activity of Rac1 (see SupplementalProcedures) in several subcellular hippocampal fractions. Immunoblot analyses indicated thatHAV-N did not significantly affect beta-catenin-mediated signaling or N-cadherinredistribution under these conditions (Figure S1A, S1B). However, the activity of Rac1, aprotein recruited to the membrane when cell adhesion is weakened (Kaibuchi et al., 1999), waselevated in the membrane fractions (Figure S1D). Thus, some effects of decreased N-cadherindimerization, such as activation of Rac1, have been reproduced by HAV-N. Expectedly, thepeptide did not mimic the alterations accompanying severely disrupted cell-cell adhesiontypically involving N-cadherin and beta-catenin redistribution.

Hippocampal N-cadherin levels increase after contextual fear conditioningIncreased levels of N-cadherin may contribute to LTP (Bozdagi et al., 2000) and possiblymemory consolidation by strengthening cell-cell adhesion. On this basis, we examined thelevels of N-cadherin at different time points after training. In order to provide a maximal signal/noise ratio, we employed a dilution of anti-N-cadherin antibody that gave just detectablesignals. N-cadherin levels significantly increased after fear conditioning (F (7,40) = 5.97, p <0.01). An up-regulation was observed shortly after training, peaking 1 hr later and exhibitinga persistent elevation, particularly in the CA2 subfield, up to 5 days posttraining (Figure4A-4C). The specificity of N-cadherin production after fear conditioning was shown in twoexperiments. To control for the persistent N-cadherin elevation after conditioning, a group ofmice exposed to daily extinction trials leading to fear reduction was introduced (Fischer et al.,2004). Nonreinforced trials abolished the increase of N-cadherin levels observed 5 daysposttraining (Figure 4A,4C), suggesting that the up-regulation of N-cadherin levels werespecific for conditioning relative to extinction. Whether the observed down-regulation of N-cadherin contributes to fear extinction remains to be established.

In a separate experiment, a group of mice was exposed to an inverse presentation of trainingstimuli so that an immediate shock was followed by tone and context. This group exhibitedsignificantly lower freezing than the paired group (F (2,27) = 8.45, p < 0.01, Figure 4D) asobserved in our earlier work (Fischer et al., 2002;Sananbenesi et al., 2002), and did not showup-regulation of N-cadherin levels comparable to the paired context-tone-shock group(F (2,15) = 4.78, p < 0.01, Figure 4E).

Taken together, these findings demonstrated a long lasting increase of N-cadherin levels inhippocampal neurons specifically accompanying consolidation of associative learning.

HAV-N selectively disrupts the activation of Erk-1/2 associated with the cytoskeletonN-cadherins are associated with a number of signaling molecules within multiproteincomplexes linking the cell membrane to the actin cytoskeleton (Husi et al., 2000). On this basiswe hypothesized the memory-impairing actions of HAV-N may be due to interference withsignal transduction pathways involved in memory formation. We tested this possibility bymonitoring the activation of Erk-1/2, cAMP-dependent protein kinase binding protein (CREB),p38 mitogen-activated protein kinase (p38MAPK) and stress-activated protein kinase (SAPK)in hippocampal subcellular fractions prepared 1 hr after fear conditioning and immediateposttraining injection with HAV-N, HAV-S or vehicle. In all subcellular fractions pErk-1/2levels increased (vehicle vs naïve: membrane: t (7) = 5.17, p = 0.01; cytosol: t (7) = 3.95, p =0.05; cytoskeleton: t (7) = 4.82, p < 0.01) after training (Figure 5). Analyses of immunoblotsdemonstrated significant effects of HAV-N on Erk-1/2-mediated signaling (Figure 5A-5C).Whereas the levels of total Erk-1/2, but not the pErk/Erk ratio (F (2,10) = .929, p = 0.426), wereincreased in the membrane fraction (total Erk: F (2,10) = 7.42, p < 0.01; Figure 5A, lower panel),

Schrick et al. Page 4

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

the phosphorylation of Erk-1/2 within the cytoskeletal protein pool (Figure 5C, upper panel)was significantly reduced (F (2,10) = 6.975, p < 0.01). In contrast to pErk-1/2, the activation ofp38MAPK and pSAPK in the cytoskeleton, and CREB in the nucleus were not significantlyaffected (Figure S2A-S2C) at the corresponding time point.

Given that both pErk-1/2 (Sananbenesi et al., 2002, 2003) and N-cadherin levels peak 1 hr afterfear conditioning, we selected this time point to examine by immunofluorescence whether thesemolecules colocalize within the hippocampus. When compared to naïve mice (Figure 6A),mice exposed to fear conditioning exhibited a marked up-regulation of pErk-1/2 and N-cadherin in the pyramidal cell layer and str. radiatum (Figure 6B). Analyses of highmagnification captures revealed that 65% ± 8 of pyramidal neuron soma within thehippocampal CA1 subfield were positive for both N-cadherin and pErk-1/2 (Figure C). N-cadherin signals typically representing synaptic puncta (Bozdagi et al., 2000) were alsoobserved in 75% ± 11 of pErk-1/2-positive apical dendrites (Figure 6 lower pannel). It shouldbe noted that pErk-1/2 showed 53% ± 7 and 57% ± 5 overlap in the soma and dendrites,respectively (data not shown). Thus, most N-cadherin signals colocalized with pErk-1/2whereas a subset of pErk signals colocalized with N-cadherin. This finding was expected giventhe coupling of Erk-1/2 to a vast number of receptor molecules (Fukunaga and Miyamoto,1998).

Taken together, these data showed that although learning activated Erk-1/2 in all subcellularfractions, HAV-N selectively impaired the fear conditioning-induced phosphorylation ofcytoskeletally-bound Erk-1/2. Notably, this delayed effect (1 hr posttraining) probably resultedfrom an early disruption of N-cadherin interactions, given that newly synthesized N-cadherindid not rescue the phosphorylation of Erk-1/2 at the examined time point. The colocalizationof pErk-1/2 and N-cadherin support the view that these proteins may interact, probablyindirectly, within individual pyramidal cells.

HAV-N triggers redistribution of IQGAP1Several docking proteins compartmentalize Erk-1/2 with upstream regulators thereby tightlycontrolling its activity. On the basis of its ability to bridge cadherin-mediated adhesion tocytoskeletal alterations and Erk-1/2 signaling (Brown and Sacks, 2006;Roy et al., 2004), wehypothesized that IQGAP1 may be involved in the effects of HAV-N on pErk-1/2. HAV-Ninjections differentially affected IQGAP1 levels in hippocampal subcellular fractions (Figure7A-7C). The significant decrease and increase of IQGAP-1 in the cytosolic (F (2,10) = 4.17, p< 0.05) and cytoskeletal (F (2,10) = 4.302, p < 0.05) fractions (Figure 7B,C), respectively,indicated redistribution of IQGAP1 from the cytosol to the cytoskeleton in response to HAV-N. However, the peptide did not affect (F (2,10) = 1.085, p = 0.374) membrane levels of IQGAP1(Figure 7A). In response to fear conditioning, the distribution of IQGAP1 between subcellularcompartments of vehicle controls was not altered (Figure 7D; membrane: t (7) = 0.177, p =0.65; cytosol: t (7) = 0.251, p = 0.73; cytoskeleton: t (7) = 0.213, p = 0.69), suggesting thatmaintaining the localization of IQGAP1 during learning may be an important role of N-cadherin. Such effects were not observed in identically treated but non-trained (Figure S3),confirming that HAV-N exerts maximal activity when applied concurrently with fearconditioning.

To study in more detail the relationship between cytoskeletal IQGAP1 and pErk-1/2, weperformed time course and dose-response studies with HAV-N. In the same cytoskeletalpreparations, HAV-N decreased and increased the levels of pErk-1/2 and IQGAP1,respectively, with maximal effects observed 1 hr after training/injection and only with thebehaviorally effective doses of 500 and 750 ng/mouse (Figure 7E). Two-way ANOVA revealedsignificant effects of dose (pErk-1/2: F (4,80) = 7.446, p < 0.001; IQGAP1: F (4,80) = 6.756, p< 0.001) and time (pErk-1/2: F (3,80) = 3.97, p < 0.05; IQGAP1: F (3,80) = 13.114, p < 0.001).

Schrick et al. Page 5

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

The behaviorally ineffective doses of 325 and 1500 ng/mouse did not produce significantbiochemical alterations. Given that increased concentrations of IQGAP1 negatively regulateErk-1/2 phosphorylation (Roy et al., 2004), these results suggest that HAV-N may disrupt theactivity of cytoskeletal Erk-1/2 by increasing the levels of IQGAP1 bound to the cytoskeleton.

HAV-N attenuates the interactions between N-cadherin, IQGAP1 and ErkWe further examined whether HAV-N affects the functional interactions between N-cadherin,IQGAP1 and Erk. As cytoskeletal preparations contain high levels of denaturating agents andlow protein amount, we performed coimmunoprecipitation studies using totaldorsohippocampal lysates. The samples were prepared from fear conditioned mice injectedposttraining with vehicle, HAV-S (750 ng) or HAV-N (750 ng) and euthanized 1 hr later. N-cadherin, IQGAP1, Erk-1/2 and pErk-1/2 were coimmunoprecipitated and detected by each ofthe employed antibodies, showing that these proteins interact in the dorsohippocampal tissue(Figure 8A-8D). We detected only Erk-2 in the immunoprecipitates with N-cadherin andIQGAP1 (Figure 8A, 8B), despite the presence of the Erk-1 isoform in the input samples (Figure8E). Analyses of N-cadherin bound complexes demonstrated a significant decrease of pErk-2(p < 0.01), Erk2 (p < 0.01) and IQGAP1 (p < 0.01) in response to HAV-N treatment. Similarly,significantly lower levels of N-cadherin (p < 0.01) were detected in immunoprecipitatesperformed with IQGAP1, pErk-1/2, and Erk-1/2 (Figure 8B-8D). IQGAP1 complexes withpErk-2 were significantly increased after fear conditioning in vehicle and HAV-S samples (p< 0.05), whereas HAV-N caused a reduction of pErk-2 bound to IQGAP1 (p < 0.01) despitean increased interaction with total Erk-2 (p < 0.01). These observations were confirmed withimmunoprecipitates employing pErk-1/2 and Erk-1/2 antibodies (Figure 8C, 8D). Consistentwith increased pErk levels in all cellular fractions, input pErk levels in the total lysates wasexpectedly higher. Supporting specific interaction with IQGAP1, pErk was not enhanced inN-cadherin (Figure 8A) or control IgG (Figure 8E) coimmunoprecipitates. Furthermore, thelevel of total Erk was not increased in the IQGAP1 complexes (Figure 8B).

These findings indicated that N-cadherin, IQGAP1 and Erk-2 form complexes in the dorsalhippocampus. Notably, although IQGAP1 levels were not affected by fear conditioning, itsinteraction with Erk-2 was significantly enhanced. The composition of N-cadherin/IQGAP1/Erk-2 complexes was disrupted by HAV-N, possibly as a result of an impaired N-cadherindimerization.

HAV-N inhibits Erk-1/2 phosphorylation and removes IQGAP1 from dendritic spine headsTo study in more detail the localization of N-cadherin, Erk-1/2 and IQGAP1 and theirrelationship to N-cadherin function we employed primary cortical cultures. We first establishedthat N-cadherin and Erk-1/2 colocalize with IQGAP1 in this cultures (Figure 9A, 9B). NMDA-induced activation of these cells resulted in a significant (p < 0.001) rise of pErk-1/2 levelsthat was completely blocked by pre-treatment with HAV-N but not HAV-S (Figure 9C, 9D).Under the same conditions, HAV-N did not affect total IQGAP1 levels, as indicated byfluorescence intensity, but resulted in decreased IQGAP1 in dendritic spines and anaccumulation in dendritic shafts (Figure 9E-9G), revealed by a decrease of spine/shaft ratio (p< 0.01) of IQGAP1 distribution. These findings replicated our key in vivo observationsindicating a role of N-cadherin in IQGAP1/Erk-1/2 signaling, and suggested that the decreaseof pErk-1/2 by HAV-N may be associated with the removal of IQGAP1 from dendritic spineheads and its increase in the dendritic shafts. Owing to the similarities in N-cadherin functionand its interactions with synaptic proteins in hippocampal and cortical neurons (Nuriya andHuganir, 2006), as well as consistencies between the in vitro and in vivo data, we hypothesizethat the identified role of N-cadherin in IQGAP1/Erk signaling encompasses both cortical andhippocampal plasticity.

Schrick et al. Page 6

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

DiscussionWe demonstrated an important role of N-cadherin in the consolidation of hippocampallydependent memory and associated neuronal signaling mechanisms. Biochemically, N-cadherinwas identified as a regulator of the distribution of IQGAP1 and activation of a distinctive,learning-induced, cytoskeletal fraction of pErk-1/2. In turn, learning mechanisms led to long-term up-regulation of hippocampal N-cadherin levels. Above all, this increase persisted in theCA2 hippocampal subfield known to transmit delayed excitatory signals to CA1 neurons(Sekino et al., 1997) and maintain long-range axonal connectivity (Zaidel, 1999), the link tothe basolateral amygdala being particularly significant for the integration of emotional andcognitive processing (Benes et al., 2004;Nakao et al., 2004). The N-cadherin antagonisticpeptide HAV-N disrupted fear conditioning-induced N-cadherin dimerization and cytoskeletalErk-1/2 signaling but not N-cadherin up-regulation. This observation is not surprising giventhat plasticity-induced increase of N-cadherin requires de novo protein synthesis mediated bynuclear effects of cAMP-dependent protein kinase signaling (Bozdagi et al., 2000), a pathwaynot affected by the employed HAV-N treatment.

Disulfide bonds (Makagiansar et al., 2002) and other interactions (Troyanovski et al., 2006)promote and stabilize cadherin dimerization. HAV peptides are thought to prevent theseinteractions, thereby disrupting N-cadherin dimerization and postadhesion signaling (Harrisonet al., 2005;Renaud-Young and Gallin, 2002;Shapiro et al., 1995). Accordingly, it wasdemonstrated that the formation of cadherin dimers is required for cadherin-mediated signaling(Kim et al., 2005). HAV-N may also attenuate N-cadherin dimerization, at least in part, bycausing redistribution of the N-cadherin/Erk-1/2 docking protein IQGAP1 known to stabilizesurface N-cadherin molecules (Noritake et al., 2005). We confirmed those observations invivo, by showing that selected doses of HAV-N transiently impaired N-cadherin dimerization,pErk-1/2-mediated signal transduction and IQGAP1 distribution in hippocampal neurons.Importantly, only the doses of HAV-N that disrupted fear conditioning caused thesebiochemical effects, suggesting a link between the observed molecular and behavioralalterations.

HAV-N significantly impaired memory formation of contextual fear at selected doses relativeto training. U-shaped dose-response curves are typically observed with in vivo pharmacologicalstudies employing biologically active peptides (Calabrese and Baldwin, 2003). Although themechanisms of such effects are not known in detail, it may be speculated that at higher dosesthe HAV-N molecules might interact with one another, with other related cadherins, or withunrelated proteins containing N-cadherin motifs, such as fibroblast growth factor receptors(Williams et al., 2001).

The involvement of N-cadherin in LTP induction but not LTP maintenance (Tang et al.,1998), when N-cadherin production and synaptic recruitment significantly increase (Bozdagiet al., 2000), is consistent with our in vivo experiments. Despite this early involvement, N-cadherin mediated the development of late but not early LTP, as revealed by the use of blockingantibodies (Bozdagi et al., 2000). Similarly, in our study HAV-N impaired long- but not short-term memory. These results indicated a role of N-cadherin in early biochemical events initiatingmemory consolidation (Miyamoto, 2006). Lack of effects of HAV peptides at later time pointsafter training, during retrieval or extinction does not exclude a role of learning-induced N-cadherin in these processes as posttraining actions of HAV-N may be limited by several factors.First, in response to synaptic activity, N-cadherin levels increase in the form of molecularlymodified, stable, protease-resistant dimers (Tanaka et al., 2000). Such modifications couldlimit the actions of HAV-N on dimerization of newly synthesized N-cadherin and behavior.Second, the effects of HAV-N critically and inversely depend on the concentration ofextracellular Ca2+ (Tang et al., 1998). The transition of calcium ions from extracellular spaces

Schrick et al. Page 7

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

into the cell during early phases of memory formationmay enhance the efficiency of HAVpeptides during these phases, as suggested earlier (Tang et al., 1998). Thus, elevated N-cadherinlevels, stabilization of N-cadherin molecules, or weaker calcium ion fluctuations may result indiminished efficiency of HAV-N. Consistent with the latter possibility, extinction mechanisms,although shared to some extent with those involved in conditioning, appear to recruit a morelimited set of biochemical pathways (Berman and Dudai, 2001;Lin et al., 2003). Furthermore,processes other than dimerization, including cleavage or decreased N-cadherin production,could contribute to extinction. Such alterations may be required to weaken the synaptic contactsestablished during fear conditioning, so that acquired fear would not persist without aversivereinforcement. The potential roles of N-cadherin in the later phases of memory formation,retrieval and extinction, remaining to be established by more potent interfering agents ortransient genetic manipulations, are likely to reveal additional mechanisms associated with thefunction of newly synthesized N-cadherin.

The significant involvement of N-cadherin dimerization in the activation of Erk-1/2, a keykinase involved in memory consolidation (Atkins et al., 1998), was not surprising. However,the selective role of N-cadherin dimerization in the phosphorylation of a segregated,cytoskeletal fraction of Erk-1/2 (but not pCREB, p38MAPK, pSAPK or beta-catenin signalingat the same time-points) was unexpected, given that signal transduction by PKA (Dell’acquaet al., 2006) as well as beta-catenin (Yu and Malenka, 2003) also depends on their interactionswith N-cadherin. Possibly, the functions of the latter pathways are predominantly regulated byN-cadherin cleavage (Reiss et al., 2005), or alternative cascades including the Wnt pathway(De Ferrari et al., 2003).

Contrary to the significant decrease of Erk-1/2 activation in cytoskeletal preparations of HAV-N injected hippocampi, pErk-1/2 levels in the membrane extracts increased proportionally tothe levels of total Erk-1/2 protein in this fraction. Weakening of cell-cell interactions by theantagonistic peptide may trigger the activation of Rac1 (Charrasse et al., 2002), as confirmedhere, or redistribution of other signaling molecules towards the cell membrane (Retta et al.,2006). Although membrane interactions and docking of Erk-1/2 significantly contribute tosynaptic plasticity and learning (Morozov et al., 2003;Shalin et al., 2006), it appears that theycould not compensate for the memory impairing effect of HAV-N. Selective activation ofcytoskeletal Erk-1/2 by N-cadherin may thus be required for memory formation in anonredundant manner to membrane-docked Erk-1/2. Because the threshold for intracellularErk-1/2 phosphorylation is markedly higher than for membrane-bound Erk-1/2 (Harding et al.,2005), N-cadherin interactions may selectively lower the activation threshold for Erk-1/2bound to the cytoskeleton. Such effects may be mediated by IQGAP1, a scaffold for cadherinsand cytoskeletal elements, that has recently emerged as a main regulator of cell adhesion-induced cytoskeletal rearrangement and signal transduction (Fukata et al., 2001;Noritake etal., 2005). Fear conditioning did not affect IQGAP1 levels, however its interaction with pErk-2,the main isoform activated by fear conditioning (Atkins et al., 1998), was significantlyenhanced. HAV-N triggered redistribution of IQGAP1 from the cytosol to the cytoskeletonand diminished N-cadherin binding to IQGAP1 and pErk, suggesting that normal N-cadherinfunction was required for optimal interaction of these molecules. Relatively small increases ordecreases of IQGAP1 and other scaffold proteins are known to decrease Erk-1/2 activity byaltering the stoichiometry of the mitogen-activated protein kinases serving as upstream Erkactivators (Levchenko et al., 2000;Roy et al., 2004). In agreement with the model proposed byRoy et al. (2005), the enhanced binding of IQGAP1 to Erk-2 was accompanied by significantlydecreased Erk-2 phosphorylation in response to HAV-N treatment. Thus, IQGAP1 may be astrong candidate for linking N-cadherin and Erk-1/2 signaling during contextual fear memoryformation.

Schrick et al. Page 8

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

The cytoskeletal preparations employed here contained both actin and microtubules that arepredominantly localized in dendritic spines and shafts, respectively (Kaech et al.,1997;Widelitz, 2005;Wilson and Keith, 1998). To more closely determine the redistributionof IQGAP1 and its relationship to Erk-1/2 phosphorylation we employed neuronal culturesactivated by NMDA with or without pre-treatment with HAV-S and HAV-N. Consistent withthe in vivo generated data, HAV-N abolished the phosphorylation of pErk-1/2. At the sametime, IQGAP1 signals increased in the dendritic shafts while decreasing in the spines,suggesting that N-cadherin may be required for maintaining IQGAP1/Erk-1/2 interactions inthe dendritic spines. In view of the well established role of IQGAP1 in cellular polarization indeveloping cells (Noritake et al., 2005), similar interactions between N-cadherin and IQGAP1in adult neurons may lead to localized changes in potentiated synapses underlying associativeneuronal plasticity (Frey and Morris, 1997).

In conclusion, we demonstrated that N-cadherin is required for memory formation ofcontextual fear. Data generated both in vivo and in vitro suggest that these behavioral effectsmay be based on N-cadherin-mediated interactions of IQGAP1 and Erk-1/2 within thecytoskeleton of dendritic spines. The N-cadherin-regulated cytoskeletal pool of Erk-1/2 maybe an important mediator of activity-induced synaptic remodelling (Okamura et al., 2004) andmemory formation mechanisms involving actin rearrangement (Fischer et al., 2004).

Experimental ProceduresAnimals

Nine-week old C57BL/6J mice were obtained from Jackson Laboratories. The mice wereindividually housed in a satellite facility provided with a separate ventilation system (15 airexchanges/hr), a 12/12 dark light cycle (7 am-7 pm), 40-50% humidity, and 20 ± 2°Ctemperature adjacent to the behavioral room. All studies were approved by NorthwesternUniversity the Animal Care and Use Committee in compliance with National Institutes ofHealth standards. The number of mice per group was 10-12 for behavioral, 4-5 for immunoblotand 6 for immunohistochemical experiments.

PeptidesThe HAV-N peptide, Ala-Arg-Phe-His-Leu-Arg-Ala-His-Ala-Val-Asp-Ile-Asn-Gly-Asn-Gln-Val, and a scrambled peptide (HAV-S), Val-Ala-Val-Leu-Tyr-Glu-Lys-Ser-Gly-Ile-Ala-Tyr-His-Asn-Ser-Ala-Ser, were synthesized and kindly provided by Lars van Werven, MaxPlanck Institute for Experimental Medicine, Goettingen. The peptides were dissolved inartificial cerebrospinal fluid immediately before injections at the indicated concentrations.

Cannulation and injectionsDouble cannulae were placed into the dorsal hippocampus (anteroposterior - 1.5 mm,mediolateral 1 mm, dorsoventral 2 mm) as described earlier (Radulovic et al., 1999). Injectionswere delivered bilaterally (0.25 μl/side) over a 30 s period at the indicated doses and timesrelative to the behavioral task. The cannula position was determined for each mouse bymethylene-blue injection after the end of experiments and only data obtained from mice withcorrectly inserted cannula were analyzed. For all biochemical experiments, the peptides wereinjected immediately after fear conditioning.

Fear conditioningContextual and tone-dependent fear conditioning was performed in an automated system (TSEInc.) and consisted of a single exposure to context (3 min) followed by a 30-s tone (10 kHz,75 dB SPL) and a footshock (2s, 0.7 mA, constant current) as described previously (Radulovic

Schrick et al. Page 9

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

et al., 1998). Context-dependent freezing was measured 24 hr later every 10th second over 180s by two observers unaware of the experimental conditions and expressed as percentage of totalnumber of observations. Freezing to the tone was similarly scored in a novel context during a3-min exposure. Extinction trials were performed at 24 hr intervals and consisted ofnonreinforced 3-min exposures to the context (Fischer et al., 2004).

ImmunohistochemistryMice were anesthetized with an intraperitoneal injection of 240 mg/kg of Avertin at indicatedtimes after training, and transcardially perfused with ice-cold 4% paraformaldehyde inphosphate buffer (pH 7.4, 150 ml/mouse). Brains were processed for immunihistochemistrywith specific antibodes as described in detail in the Supplemental Procedures. Quantificationof immunostaining signals was performed as described previously (Winder et al.,1999;Sananbenesi et al., 2002). Digital images were captured with a cooled color CCD camera(RTKE Diagnostic Instruments) and SPOT software for Macintosh. Adobe Photoshop 5.0 forMacintosh was used for image processing. The background corresponding to areas withoutimmunopositive cells was subtracted. Images were binarized and digitally contrast stretchedfor maximum resolution of N-cadherin-positive cells. The optical density within the CA1 andCA2 subfield and stratum radiatum were determined for each image with the ImageJ software(NIH). The average pixel intensity reflecting the mean value of these three areas, was thendivided by the total area to determine integrated pixel density.

For coimmunolabeling studies the TSA Fluorescence System (NEN Life Science Products)was employed, using fluorescein as substrate for pErk-1/2 and rhodamine for N-cadherin.Multicolor immunofluorescence was captured using a Zeiss LSM5 Pascal confocal microscopeand Z-stacks of images were taken using the 10x and 63x objectives and analyzed with aMetamorph software (Macintosh Universal Imaging). The number of coimmunolabeled somasand dendrites was determined as percentage of a total of 100 cells obtained from 3 capturedimages/mouse.

Protein extraction and immunoblotMice were injected with vehicle, HAV-N or HAV-S immediately after training, and euthanizedby cervical dislocation 1 hr later. Individual dorsal hippocampi (the rostral 2.5 mm septal pole)were dissected, frozen in liquid nitrogen and kept at -80°C. Cytoplasmic, membrane,cytoskeletal and nuclear fractions were prepared by using the ProteoExtract kit for subcellularproteome extraction (EMD Biosciences) according to the instructions. Aliquots of individualsamples were stored in loading buffers with or without β-mercaptoethanol (β-ME) for analysesunder reducing or nonreducing conditions. The purity of the nuclear, cytoplasmic andcytoskeletal samples was confirmed by immunoblot with membrane (Na/K ATP-ase), nuclear(CREB), cytoskeletal (β-catenin) and cytoplasmic (LDH) markers (Figure S2D). Afterdetermining the protein concentration (Bio-Rad), the lysates (5-20 μg/well) were subjected to10% SDS polyacrylamide gel electrophoresis and subsequently blotted to PVDF membranes(Millipore) as described previously (Sananbenesi et al., 2002). The membranes were saturatedwith I-block (Tropix) and then incubated with the primary (Supplemental Procedures) andcorresponding secondary antibodies, enhancer (Nitro-block II, Tropix) and chemiluminescentsubstrate (CDP Star, Tropix). Blots were exposed to X-ray films and developed in the rangeof maximal chemiluminescence emission (10 min). Molecular weight and densitometriccalculations of pErk-1/2 were performed with the computer software ImageJ (NIH). Inter-assayvariability between blots was determined by a standard control sample for each individualfraction obtained from pooled hippocampi of five naïve mice. The levels of individual proteinswere normalized to β-actin. The signals of pErk-1/2 were additionally normalized to those oftotal Erk-1/2 in corresponding individual samples.

Schrick et al. Page 10

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

ImmunoprecipitationNaive mice or mice exposed to fear conditioning were injected i.h. immediately afterwardswith vehicle, HAV-S or HAV-N (n = 9/group). Individual dorsal hippocampi were collected1 hour later, when HAV-N typically causes a significant decrease of pErk. The hippocampiwere lysed in a modified radioimmunoprecipitation buffer, incubated 15 min on ice andcentrifuged for 15 min, 15000 × g, 4°C as described (Fischer et al., 2007). After determiningthe protein concentration for each lysate (Bio-Rad protein assay), 2 mg of total protein/lysatewas prepared by pooling 3 samples of each group. Five hundred μg/sample/group wasincubated for 1h at 4°C with 4 μg anti-N-cadherin, IQGAP1, Erk or pErk-1/2 antibodies.Immunoprecipitation was performed by the Catch and Release kit (Upstate) as described inthe user’s manual. Subsequent immunoblot and analyses were performed as described above.

Neuronal culture and treatmentsMedium density cortical neuron cultures were prepared from rat E18 embryos as describedpreviously (Liao et al., 1999; see Supplemental Procedures). Chemical activation of NMDAreceptors was preformed at DIV 28 as follows. Neuron cultures were maintained in presenceof 200 μM D,L-amino-phosphonovalerate (D,L-APV). For treatments, neurons were pre-incubated in aCSF (in mM: 125 NaCl, 2.5 KCl, 26.2 NaHCO3, 1 NaH2PO4, 11 glucose and2.5 CaCl2 ± 200 μM APV and 1.25 MgCl2) for 30 min at 37°C. Coverslips were also pre-treated or not, with the peptides HAV-S or HAV-N at a concentration of 200 μM for 2 hours.Following a wash in treatment medium (aCSF plus 10 μM glycine, 100 μM picrotoxin and 1μM strychnine), coverslips were transferred to the treatment chambers in aCSF ± peptides.Treatments were allowed to proceed for the indicated times after which cells were fixed for 10minutes with methanol chilled to -20°C. Coverslips were then processed for immunostainingand analyses as previously described (Penzes et al., 2003; See supplemental Procedures)

Data analysisStatistically significant differences were determined by Student’s t-test, or one (factorTreatment) and two way ANOVA (factors Dose × Time) followed by Scheffe’s test for post-hoc comparisons. The results are presented as mean ± S.E.M.

Supplementary MaterialRefer to Web version on PubMed Central for supplementary material.

Acknowledgements

We thank Dr. Deanna Benson (Mount Sinai School of Medicine) for helpful discussion. This work was supported bythe NIMH grant MH073669 to J.R.

ReferencesAngst BD, Marcozzi C, Magee AI. The cadherin superfamily: diversity in form and function. J Cell Sci

2001;114:629–641. [PubMed: 11171368]Atkins CM, Selcher JC, Petraitis JJ, Trzaskos JM, Sweatt JD. The MAPK cascade is required for

mammalian associative learning. Nat Neurosci 1998;1:602–609. [PubMed: 10196568]Benes FM, Burke RE, Walsh J, Berretta S, Matzilevich D, Minns M, Konradi C. Acute amygdalar

activation induces an upregulation of multiple monoamine G protein coupled pathways in rathippocampus. Mol Psychiatry 2004;9:932–945. 895. [PubMed: 15170462]

Berman DE, Dudai Y. Memory extinction, learning anew, and learning the new: dissociations in themolecular machinery of learning in cortex. Science 2001;291:2417–2419. [PubMed: 11264539]

Schrick et al. Page 11

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Bozdagi O, Shan W, Tanaka H, Benson DL, Huntley GW. Increasing numbers of synaptic puncta duringlate-phase LTP: N-cadherin is synthesized, recruited to synaptic sites, and required for potentiation.Neuron 2000;28:245–259. [PubMed: 11086998]

Brown MD, Sacks DB. IQGAP1 in cellular signaling: bridging the GAP. Trends Cell Biol. 2006Calabrese EJ, Baldwin LA. Toxicology rethinks its central belief. Nature 2003;421:691–692. [PubMed:

12610596]Charrasse S, Meriane M, Comunale F, Blangy A, Gauthier-Rouviere C. N-cadherin-dependent cell-cell

contact regulates Rho GTPases and beta-catenin localization in mouse C2C12 myoblasts. J Cell Biol2002;158:953–965. [PubMed: 12213839]

De Ferrari GV, Chacon MA, Barria MI, Garrido JL, Godoy JA, Olivares G, Reyes AE, Alvarez A,Bronfman M, Inestrosa NC. Activation of Wnt signaling rescues neurodegeneration and behavioralimpairments induced by beta-amyloid fibrils. Mol Psychiatry 2003;8:195–208. [PubMed: 12610652]

Dell’acqua ML, Smith KE, Gorski JA, Horne EA, Gibson ES, Gomez LL. Regulation of neuronal PKAsignaling through AKAP targeting dynamics. Eur J Cell Biol. 2006

Derycke LD, Bracke ME. N-cadherin in the spotlight of cell-cell adhesion, differentiation,embryogenesis, invasion and signalling. Int J Dev Biol 2004;48:463–476. [PubMed: 15349821]

Edsbagge J, Zhu S, Xiao MY, Wigstrom H, Mohammed AH, Semb H. Expression of dominant negativecadherin in the adult mouse brain modifies rearing behavior. Mol Cell Neurosci 2004;25:524–535.[PubMed: 15033180]

Fischer A, Sananbenesi F, Schrick C, Spiess J, Radulovic J. Cyclin-dependent kinase 5 is required forassociative learning. J Neurosci 2002;22:3700–3707. [PubMed: 11978846]

Fischer A, Sananbenesi F, Schrick C, Spiess J, Radulovic J. Distinct roles of hippocampal de novo proteinsynthesis and actin rearrangement in extinction of contextual fear. J Neurosci 2004;24:1962–1966.[PubMed: 14985438]

Fukunaga K, Miyamoto E. Role of MAP kinase in neurons. Mol Neurobiol 1998;16:79–95. [PubMed:9554703]

Frey U, Morris RG. Synaptic tagging and long-term potentiation. Nature 1997;385:533–536. [PubMed:9020359]

Fukata M, Nakagawa M, Itoh N, Kawajiri A, Yamaga M, Kuroda S, Kaibuchi K. Involvement of IQGAP1,an effector of Rac1 and Cdc42 GTPases, in cell-cell dissociation during cell scattering. Mol Cell Biol2001;21:2165–2183. [PubMed: 11238950]

Hagler DJ Jr, Goda Y. Synaptic adhesion: the building blocks of memory? Neuron 1998;20:1059–1062.[PubMed: 9655493]

Harding A, Tian T, Westbury E, Frische E, Hancock JF. Subcellular localization determines MAP kinasesignal output. Curr Biol 2005;15:869–873. [PubMed: 15886107]

Harrison OJ, Corps EM, Kilshaw PJ. Cadherin adhesion depends on a salt bridge at the N-terminus. JCell Sci 2005;118:4123–4130. [PubMed: 16118243]

Husi H, Ward MA, Choudhary JS, Blackstock WP, Grant SG. Proteomic analysis of NMDA receptor-adhesion protein signaling complexes. Nat Neurosci 2000;3:661–669. [PubMed: 10862698]

Junghans D, Haas IG, Kemler R. Mammalian cadherins and protocadherins: about cell death, synapsesand processing. Curr Opin Cell Biol 2005;17:446–452. [PubMed: 16099637]

Kaech S, Fischer M, Doll T, Matus A. Isoform specificity in the relationship of actin to dendritic spines.J Neurosci 1997;17:9565–9572. [PubMed: 9391011]

Kaibuchi K, Kuroda S, Fukata M, Nakagawa M. Regulation of cadherin-mediated cell-cell adhesion bythe Rho family GTPases. Curr Opin Cell Biol 1999;11:591–596. [PubMed: 10508646]

Kim YJ, Johnson KR, Wheelock MJ. N-cadherin-mediated cell motility requires cis dimers. CellCommun Adhes 2005;12:23–39. [PubMed: 16371344]

Klingelhofer J, Laur OY, Troyanovsky RB, Troyanovsky SM. Dynamic interplay between adhesive andlateral E-cadherin dimers. Mol Cell Biol 2002;22:7449–7458. [PubMed: 12370292]

Levchenko A, Bruck J, Sternberg PW. Scaffold proteins may biphasically affect the levels of mitogen-activated protein kinase signaling and reduce its threshold properties. Proc Natl Acad Sci U S A2000;97:5818–5823. [PubMed: 10823939]

Schrick et al. Page 12

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Liao D, Zhang X, O’Brien R, Ehlers MD, Huganir RL. Regulation of morphological postsynaptic silentsynapses in developing hippocampal neurons. Nat Neurosci 1999;2:37–43. [PubMed: 10195178]

Lin CH, Yeh SH, Leu TH, Chang WC, Wang ST, Gean PW. Identification of calcineurin as a key signalin the extinction of fear memory. J Neurosci 2003;23:1574–1579. [PubMed: 12629159]

Makagiansar IT, Nguyen PD, Ikesue A, Kuczera K, Dentler W, Urbauer JL, Galeva N, Alterman M,Siahaan TJ. Disulfide bond formation promotes the cis- and trans-dimerization of the E-cadherin-derived first repeat. J Biol Chem 2002;277:16002–16010. [PubMed: 11856755]

Manabe T, Togashi H, Uchida N, Suzuki SC, Hayakawa Y, Yamamoto M, Yoda H, Miyakawa T, TakeichiM, Chisaka O. Loss of cadherin-11 adhesion receptor enhances plastic changes in hippocampalsynapses and modifies behavioral responses. Mol Cell Neurosci 2000;15:534–546. [PubMed:10860580]

Marambaud P, Wen PH, Dutt A, Shioi J, Takashima A, Siman R, Robakis NK. A CBP bindingtranscriptional repressor produced by the PS1/epsilon-cleavage of N-cadherin is inhibited by PS1FAD mutations. Cell 2003;114:635–645. [PubMed: 13678586]

Miyamoto E. Molecular mechanism of neuronal plasticity: induction and maintenance of long-termpotentiation in the hippocampus. J Pharmacol Sci 2006;100:433–442. [PubMed: 16799259]

Morozov A, Muzzio IA, Bourtchouladze R, Van-Strien N, Lapidus K, Yin D, Winder DG, Adams JP,Sweatt JD, Kandel ER. Rap1 couples cAMP signaling to a distinct pool of p42/44MAPK regulatingexcitability, synaptic plasticity, learning, and memory. Neuron 2003;39:309–325. [PubMed:12873387]

Murase S, Schuman EM. The role of cell adhesion molecules in synaptic plasticity and memory. CurrOpin Cell Biol 1999;11:549–553. [PubMed: 10508654]

Nakao K, Matsuyama K, Matsuki N, Ikegaya Y. Amygdala stimulation modulates hippocampal synapticplasticity. Proc Natl Acad Sci U S A 2004;101:14270–14275. [PubMed: 15381775]

Noritake J, Watanabe T, Sato K, Wang S, Kaibuchi K. IQGAP1: a key regulator of adhesion andmigration. J Cell Sci 2005;118:2085–2092. [PubMed: 15890984]

Nuriya M, Huganir RL. Regulation of AMPA receptor trafficking by N-cadherin. J Neurochem2006;97:652–661. [PubMed: 16515543]

Okamura K, Tanaka H, Yagita Y, Saeki Y, Taguchi A, Hiraoka Y, Zeng LH, Colman DR, Miki N.Cadherin activity is required for activity-induced spine remodeling. J Cell Biol 2004;167:961–972.[PubMed: 15569714]

Penzes P, Beeser A, Chernoff J, Schiller MR, Eipper BA, Mains RE, Huganir RL. Rapid induction ofdendritic spine morphogenesis by trans-synaptic ephrinB-EphB receptor activation of the Rho-GEFkalirin. Neuron 2003;37:263–274. [PubMed: 12546821]

Perron JC, Bixby JL. Distinct neurite outgrowth signaling pathways converge on ERK activation. MolCell Neurosci 1999;13:362–378. [PubMed: 10356298]

Radice GL, Rayburn H, Matsunami H, Knudsen KA, Takeichi M, Hynes RO. Developmental defects inmouse embryos lacking N-cadherin. Dev Biol 1997;181:64–78. [PubMed: 9015265]

Radulovic J, Kammermeier J, Spiess J. Relationship between fos production and classical fearconditioning: effects of novelty, latent inhibition, and unconditioned stimulus preexposure. JNeurosci 1998;18:7452–7461. [PubMed: 9736664]

Radulovic J, Ruhmann A, Liepold T, Spiess J. Modulation of learning and anxiety by corticotropin-releasing factor (CRF) and stress: differential roles of CRF receptors 1 and 2. J Neurosci1999;19:5016–5025. [PubMed: 10366634]

Redies C. Cadherins in the central nervous system. Prog Neurobiol 2000;61:611–648. [PubMed:10775799]

Reiss K, Maretzky T, Ludwig A, Tousseyn T, de Strooper B, Hartmann D, Saftig P. ADAM10 cleavageof N-cadherin and regulation of cell-cell adhesion and beta-catenin nuclear signalling. Embo J2005;24:742–752. [PubMed: 15692570]

Renaud-Young M, Gallin WJ. In the first extracellular domain of E-cadherin, heterophilic interactions,but not the conserved His-Ala-Val motif, are required for adhesion. J Biol Chem 2002;277:39609–39616. [PubMed: 12154084]

Retta SF, Balzac F, Avolio M. Rap1: A turnabout for the crosstalk between cadherins and integrins. EurJ Cell Biol 2006;85:283–293. [PubMed: 16546572]

Schrick et al. Page 13

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Rhee J, Mahfooz NS, Arregui C, Lilien J, Balsamo J, VanBerkum MF. Activation of the repulsive receptorRoundabout inhibits N-cadherin-mediated cell adhesion. Nat Cell Biol 2002;4:798–805. [PubMed:12360290]

Roy M, Li Z, Sacks DB. IQGAP1 binds ERK2 and modulates its activity. J Biol Chem 2004;279:17329–17337. [PubMed: 14970219]

Roy M, Li Z, Sacks DB. IQGAP1 is a scaffold for mitogen-activated protein kinase signaling. Mol CellBiol 2005;25:7940–7952. [PubMed: 16135787]

Sananbenesi F, Fischer A, Schrick C, Spiess J, Radulovic J. Phosphorylation of hippocampal Erk-1/2,Elk-1, and p90-Rsk-1 during contextual fear conditioning: interactions between Erk-1/2 and Elk-1.Mol Cell Neurosci 2002;21:463–476. [PubMed: 12498787]

Sekino Y, Obata K, Tanifuji M, Mizuno M, Murayama J. Delayed signal propagation via CA2 in rathippocampal slices revealed by optical recording. J Neurophysiol 1997;78:1662–1668. [PubMed:9310451]

Shalin SC, Hernandez CM, Dougherty MK, Morrison DK, Sweatt JD. Kinase suppressor of Ras1compartmentalizes hippocampal signal transduction and subserves synaptic plasticity and memoryformation. Neuron 2006;50:765–779. [PubMed: 16731514]

Shapiro L, Colman DR. Structural biology of cadherins in the nervous system. Curr Opin Neurobiol1998;8:593–599. [PubMed: 9811622]

Shapiro L, Fannon AM, Kwong PD, Thompson A, Lehmann MS, Grubel G, Legrand JF, Als-Nielsen J,Colman DR, Hendrickson WA. Structural basis of cell-cell adhesion by cadherins. Nature1995;374:327–337. [PubMed: 7885471]

Tanaka H, Shan W, Phillips GR, Arndt K, Bozdagi O, Shapiro L, Huntley GW, Benson DL, Colman DR.Molecular modification of N-cadherin in response to synaptic activity. Neuron 2000;25:93–107.[PubMed: 10707975]

Tang L, Hung CP, Schuman EM. A role for the cadherin family of cell adhesion molecules in hippocampallong-term potentiation. Neuron 1998;20:1165–1175. [PubMed: 9655504]

Togashi H, Abe K, Mizoguchi A, Takaoka K, Chisaka O, Takeichi M. Cadherin regulates dendritic spinemorphogenesis. Neuron 2002;35:77–89. [PubMed: 12123610]

Troyanovsky RB, Sokolov EP, Troyanovsky SM. Endocytosis of cadherin from intracellular junctionsis the driving force for cadherin adhesive dimer disassembly. Mol Biol Cell 2006;17:3484–3493.[PubMed: 16760429]

Widelitz R. Wnt signaling through canonical and non-canonical pathways: recent progress. GrowthFactors 2005;23:111–116. [PubMed: 16019432]

Williams E, Williams G, Gour BJ, Blaschuk OW, Doherty P. A novel family of cyclic peptide antagonistssuggests that N-cadherin specificity is determined by amino acids that flank the HAV motif. J BiolChem 2000;275:4007–4012. [PubMed: 10660557]

Williams EJ, Williams G, Howell FV, Skaper SD, Walsh FS, Doherty P. Identification of an N-cadherinmotif that can interact with the fibroblast growth factor receptor and is required for axonal growth.J Biol Chem 2001;276:43879–43886. [PubMed: 11571292]

Wilson MT, Keith CH. Glutamate modulation of dendrite outgrowth: alterations in the distribution ofdendritic microtubules. J Neurosci Res 1998;52:599–611. [PubMed: 9632316]

Winder DG, Martin KC, Muzzio IA, Rohrer D, Chruscinski A, Kobilka B, Kandel ER. ERK plays aregulatory role in induction of LTP by theta frequency stimulation and its modulation by beta-adrenergic receptors. Neuron 1999;24:715–726. [PubMed: 10595521]

Yu X, Malenka RC. Beta-catenin is critical for dendritic morphogenesis. Nat Neurosci 2003;6:1169–1177. [PubMed: 14528308]

Zaidel DW. Quantitative morphology of human hippocampus early neuron development. Anat Rec1999;254:87–91. [PubMed: 9892421]

Schrick et al. Page 14

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 1.HAV-N Impairs Contextual Fear Conditioning(A) Dose-dependent impairment of contextual freezing by immediate posttraining injectionsof HAV-N (left panel). The same doses did not affect tone-dependent freezing (right panel).(B) A dose of 750 ng/mouse of HAV-N but not a scrambled peptide, HAV-S, resulted in asignificant decrease of freezing behavior (left) and increase of mean activity (right) during thetest of long-term contextual memory performed 24 hr after training.(C) The same dose did not affect freezing (left) or activity (right) during the test of short-termcontextual memory performed 1 hr after training.Statistically significant differences: *p < 0.05, **p < 0.01 vs vehicle; #p < 0.05, ##p < 0.01 vsHAV-S. The number of mice per group was 10 (A), 12 (B) and 11 (C).

Schrick et al. Page 15

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 2.Effect of HAV-N on Memory Formation, Retrieval and Extinction(A) Injections of HAV-N 15 min before or immediately after training impaired contextualfreezing during the memory test whereas injections 1 hr posttraining were ineffective.(B) HAV-N administered 15 min before the memory test did not affect freezing, indicatingnormal contextual memory retrieval.(C) HAV-N injected immediately after individual nonreinforced trials did not affect extinctionof contextual fear, as shown by a similar decrease of freezing behavior in all experimentalgroups. HAV-N and HAV-S were injected at a dose of 750 ng/mouse.Statistically significant differences: **p < 0.01 vs vehicle and HAV-S; #p < 0.05, ##p < 0.01vs test 1. The number of mice per group was 10-11.

Schrick et al. Page 16

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 3.HAV-N transiently disrupts N-cadherin dimerization(A) The levels of N-cadherin increase after fear conditioning, as revealed by immunoblotanalyses performed under reducing conditions in total dorsohippocampal lysates. Under theseconditions, N-cadherin is detected as a monomer. The levels of N-cadherin were not affectedby HAV-N treatment.Statistically significant differences: *p < 0.001 vs naïve. The number of samples/group was 5.(B) Representative immunoblot showing up-regulation of N-cadherin monomers.(C) Representative immunoblots of N-cadherin in membrane dorsohippocampal fractionsresolved on 5% SDS gels show distinctive bands at 230 kDa under nonreducing conditionsindicating N-cadherin dimers. These bands were not observed in samples obtained from dorsalhippocampi of mice injected with 500 and 750 ng/mouse HAV-N and sacrificed 15 and 30 minlater. One hr after HAV-N injection, the band was detectable although weaker than in vehicle-and HAV-S-treated mice. The dimer band was not detectable under reducing conditions.(D) Fear conditioning triggers up-regulation of N-cadherin dimers in the vehicle, HAV-S,HAV-N (325 ng and 1500 ng) groups. One way ANOVA revealed significantly stronger signalsin these groups (p < 0.001) when compared to the naïve group. The behaviorally effective dosesof HAV-N (500 and 750 ng/mouse) significantly impaired N-cadherin dimerization 15 and 30min posttraining.

Schrick et al. Page 17

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Statistically significant differences: *p < 0.01 vs naïve; #p < 0.01 vs all other experimentalgroups except for naïve (this group was not included in two way ANOVA analysis because itrepresents a single time point). The number of samples per group was 4.

Schrick et al. Page 18

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 4.Fear Conditioning Up-regulates N-cadherin in the Hippocampus(A) N-cadherin levels significantly increased after contextual fear conditioning up to 120 hr(5 days) posttraining. Extinction (E) training abolished the increase of N-cadherin observed atthe 120 hr time point (120 + E, light green).(B) Representative micrographs of the CA1 subfield of naïve mice and 1 hr posttraining.(C) Representative micrographs of the CA2 subfield of naïve mice, 1 hr posttraining, 120 hrposttraining and 120 hr with extinction (120 hr + E). Statistically significant differences: *p <0.05, **p < 0.01, ***p < 0.01 vs time point 0 (naïve mice).(D) Freezing behavior was induced only by paired (CTS, context-tone-shock) but not reverse,unpaired (STC, shock-tone-context) presentation of the conditioned and unconditioned stimuli.Statistically significant differences: *p < 0.05 vs naïve, ***p < 0.01 vs naïve, ###p < 0.001 vsSTC. The number of mice per group was 7.(E) N-cadherin levels were significantly higher in mice of the CTS when compared to the STCand naïve groups (left). Representative micrographs (right).Statistically significant differences: *p < 0.05, ***p < 0.01 vs vehicle; ###p < 0.001 vs STC.The number of mice per group was 6.

Schrick et al. Page 19

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 5.HAV-N Affects Erk-1/2 Phosphorylation and Distribution(A) HAV-N did not affect pErk-1/2 (top) but increased total Erk-1/2 (bottom) levels inmembrane fractions of hippocampal extracts.(B) HAV-N did not exhibit significant alterations of pErk-1/2 and total Erk-1/2 in cytosolicfractions.(C) HAV-N disrupted the phosphorylation of Erk-1/2 without affecting total Erk-1/2 incytoskeletal fractions.HAV-N, HAV-S (750 ng/mouse each) and vehicle were used for treatment. Middle panels:representative immunoblots of hippocampal extracts obtained 1 hr posttraining. The levels oftotal Erk-1/2 were normalized to actin in individual extracts. The number of individual samplesper treatment was 4-5. Statistically significant differences: *p < 0.05, **p < 0.01 vs naive; #p< 0.05, ###p < 0.01 vs vehicle and HAV-S.

Schrick et al. Page 20

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 6.Up-regulation of pErk-1/2 in N-Cadherin-Positive Hippocampal Neurons(A) pErk-1/2 (green) and N-cadherin (red) in hippocampi of naïve mice.(B) Up-regulation of pErk-1/2 (green) and N-cadherin (red) in CA1 pyramidal cell soma anddendrites after fear conditioning.The top and middle panels represent 10x images of the CA1 region. Scale bar = 40 μm. Thebottom panel represents single z stacks of merged N-cadherin and pErk-1/2 high magnification(63x) images of the CA1 subfield and adjacent str. radiatum. showing colocalization withinthe soma (arrow) as well as apical dendrites (arrowhead). Scale bar = 5 μm.

Schrick et al. Page 21

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 7.HAV-N Alters the Distribution of IQGAP1(A) HAV-N did not affect membrane levels of IQGAP1.(B) HAV-N significantly decreased the level of IQGAP1 in the cytosol.(C) HAV-N caused an accumulation of IQGAP1 in the cytoskeleton.Statistically significant differences: *p < 0.05, **p < 0.01 vs vehicle; #p < 0.05, ##p < 0.01 vsHAV-S. The number of individual samples per group was 4-5.(D) Fear conditioning did not affect significantly the distribution of IQGAP1 in subcellularfractions of hippocampal extracts. The optical density for IQGAP1 was normalized to the levelof β-actin.(E) Dose-response and time course curves demonstrating the effects of HAV-N on IQGAP1(left) and pErk-1/2 (right) levels in cytoskeletal preparations. Data are expressed as percent ofmean O.D. in naïve mice. The number of individual samples per group was 4-5. Statisticallysignificant differences: *p < 0.001 vs vehicle, HAV-N 325 and 1500 ng/mouse; #p < 0.01 vsnaïve; ap < 0.001 vs time point 0 (immediately post-injection).

Schrick et al. Page 22

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 8.N-cadherin/IQGAP1/pErk-1/2 interactions are disrupted by HAV-N Dorsohippocampallysates were prepared from naïve mice or mice trained in the fear conditioning paradigm andinjected immediately afterwards with vehicle (white bars), HAV-S (yellow bars) or HAV-N(purple bars) (750 ng/mouse each). Dashed line represent values of the naïve group (100%).Data are expressed as percentage of naïve.(A) Decreased levels of pErk-2, Erk-2 and IQGAP1 in dorsohippocampal lysates of HAV-N-treated mice after immunoprecipitation with N-cadherin.(B) Decreased levels of pErk-2 and N-cadherin, and increased level of Erk-2 in the same HAV-N lysates after immunoprecipitation with IQGAP1.(C) Decreased levels of IQGAP1 and N-cadherin in the HAV-N lysates afterimmunoprecipitation with pErk-1/2.(D) Increased levels of IQGAP1 and decreased levels of N-cadherin in the HAV-N lysatesafter immunoprecipitation with Erk-1/2.(E) Levels of N-cadherin, IQGAP1, pErk-1/2 and Erk-1/2 in the input samples or afterimmunoprecipitation with a corresponding anti-rabbit or anti-mouse immunoglobulin (IgG)serving as a negative control.Statistically significant differences: *p < 0.01 vs naïve, #p < 0.01 vs vehicle and HAV-S. Thenumber of individual samples/group was 3. Labels: 1, naïve; 2, vehicle; 3, HAV-S; 4, HAV-N.

Schrick et al. Page 23

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

Figure 9.HAV-N Inhibits Erk-1/2 Phosphorylation and Removes IQGAP1 from Spine Heads.(A) IQGAP1 and N-cadherin colocalize at the plasma membrane and spine heads in corticalpyramidal neurons. Scale bars: 5 μm.(B) A similar colocalization pattern (arrow: plasma membrane; arrowhead: spine heads) isobserved for IQGAP1 and Erk.(C) Erk-1/2 phosphorylation in pyramidal neurons with or without chemical activation ofNMDA receptors for 30 min in the presence or absence of HAV-S or HAV-N. Images aredisplayed in pseudocolor to demonstrate changes in p-Erk-1/2 levels.(D) Quantification of pERK1/2 levels (n= 9 neurons, 3 experiments). Statistically significantdifferences: * p < 0.001 vs all other groups.(E) IQGAP1 immunofluorescence in pyramidal neurons with or without chemical activationof NMDA receptors for 30 min in the presence or absence of HAV-S or HAV-N: white arrowsshow immunofluorescence in spine heads of control neurons and after NMDA receptoractivation in presence or absence of HAV-S or HAV-N peptide, solid white arrow head showsIQGAP1 immunofluorescence in the dendritic shaft. Dashed boxes indicate spine head anddendritic shaft used in line scan quantification.

Schrick et al. Page 24

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

(F) HAV-N decreases the ratio of IQGAP1 signals in the dendritic spines vs dendritic shafts.Statistically significant differences: * p < 0.001 vs all other groups.(G) Line scan of IQGAP1 immunofluorescence intensity along a spine head and adjacentdendritic shaft. Blue/green line: 30 min activation + HAV-S; red/white line: 30 min activation+ HAV-N. Statistically significant differences: *p < 0.01 vs control, activated and HAV-S.

Schrick et al. Page 25

Neuron. Author manuscript; available in PMC 2008 September 6.

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript

NIH

-PA Author Manuscript