Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
J Neurosci. Author manuscript; available in PMC 2010 May 4.
Published in final edited form as:
PMCID: PMC2818627

Inositol 1,4,5-trisphosphate 3-kinase A Functions as a Scaffold for Synaptic Rac Signaling


Activity-dependent alterations of synaptic contacts are crucial for synaptic plasticity. The formation of new dendritic spines and synapses is known to require actin cytoskeletal reorganization specifically during neural activation phases. Yet the site-specific and time-dependent mechanisms modulating actin dynamics in mature neurons are not well understood. In this study, we show that actin dynamics in spines is regulated by a Rac anchoring and targeting function of inositol 1,4,5-trisphosphate 3-kinase A (IP3K-A), independent of its kinase activity. Upon neural activation, IP3K-A bound directly to activated-Rac1 and recruited it to the actin cytoskeleton in the post-synaptic area. This focal targeting of activated Rac1 induced spine formation through actin dynamics downstream of Rac signaling. Consistent with the scaffolding role of IP3K-A, IP3K-A knockout mice exhibited defects in accumulation of PAK1 by LTP-inducing stimulation. This deficiency resulted in a reduction in the reorganization of actin cytoskeletal structures in the synaptic area of dentate gyrus. Moreover, IP3K-A knockout mice showed deficits of synaptic plasticity in perforant path and in hippocampal-dependent memory performances. These data support a novel model in which IP3K-A is critical for the spatial and temporal regulation of spine actin remodeling, synaptic plasticity, and learning and memory via an activity-dependent Rac scaffolding mechanism.

Keywords: IP3K-A, Rac, synaptic plasticity, F-actin, dendritic spine, Actin


Memory formation and storage in the brain is believed to be mediated by changes in synaptic efficacy and strength (Bliss et al., 2003; Segal, 2005). Dendritic spines, the postsynaptic part of synapses, display a highly dynamic morphology in the developing (Dailey and Smith, 1996) and the mature brain (Yuste and Bonhoeffer, 2001). Because the number and morphology of dendritic spines are linked to synaptic efficacy and neuronal plasticity (Engert and Bonhoeffer, 1999; Maletic-Savatic et al., 1999; Kim et al., 2002; Leuner et al., 2003), mechanisms that modulate the formation and differentiation of spines are thought to be essential for learning and memory (Kennedy et al., 2005). Additionally, abnormal spine morphology has been observed in brain tissue samples of patients with mental retardation (Purpura, 1974).

Filamentous actin (F-actin) is highly enriched in dendritic spines (Matus et al., 1982) and the regulation of actin dynamics plays an essential role in the molecular processes underlying spine morphogenesis (Star et al., 2002; Bonhoeffer and Yuste, 2002) and plasticity (Penzes et al., 2003; Fukazawa et al., 2003; Lin et al., 2005). The key regulators of actin remodeling in postsynaptic spines are the small GTPase proteins such as Rho-A, Cdc42, and Rac1 (Luo, 2002). In particular, Rac1 has been shown to regulate the morphogenesis of dendritic spines by affecting actin dynamics (Nakayama et al., 2000; Tashiro et al., 2000). The activation of Rac1 requires guanine-nucleotide exchange factors (GEFs) such as Tiam1 and kalirin (Penzes et al., 2001, 2003; Tolias et al., 2005). However, whether there are mechanisms that are responsible for activity-dependent positioning of activated Rac1 to the dendritic actin cytoskeleton, remains unclear.

IP3K-A is a brain- and neuron-specific molecule that is enriched in dendritic spines (Mailleux et al., 1991). Previously, the only known activity of IP3K-A was its ability to convert inositol 1,4,5-trisphosphate (IP3) to inositol 1,3,4,5-tetrakisphosphate (IP4), thereby reducing the influx of calcium released from endoplasmic reticulum via IP3 receptor signaling (Irvine et al., 1986; Choi et al., 1990). In addition to its catalytic domain, IP3K-A also has an F-actin-binding domain of unknown significance in its N-terminus (Schell et al., 2001). Because IP3K-A is enriched in dendritic spines, it has been speculated that IP3K-A is involved in spine remodeling through its kinase function (Schell and Irvine, 2006). Furthermore, we recently reported that IP3K-A mRNA and protein are upregulated in rat brain following spatial learning tasks (Kim et al., 2004). Therefore, IP3K-A appears to be a strong candidate to be involved in the synaptic plasticity underlying learning and memory. Previous analysis of IP3K-A knockout mice demonstrated the loss of IP3K-A does not affect intracellular calcium levels in neurons (Jun et al., 1998). Thus whether or not IP3K-A plays a functional role via its catalytic activity to modulate synaptic signaling has been an open question. It is possible IP3K-A may have other, as of yet unexplored neuro-physiologic properties that are independent of IP4 levels.

In the present study, we demonstrate that IP3K-A promotes structural remodeling of dendritic spines by affecting actin dynamics. Surprisingly, these effects were not induced by the kinase activity of IP3K-A, but rather by IP3K-A-mediated targeting of activated Rac1 to the actin cytoskeleton of dendrites. Most importantly, this process occurred during neural activation. We observed that IP3K-A directly interacts with activated Rac1 and recruits it to the actin cytoskeleton in the spine-rich region following long-term potentiation (LTP) induction. This IP3K-A scaffolding function supports the local induction of Rac downstream events such as p21-activated kinase (PAK) phosphorylation and dendritic spine formation. Furthermore, IP3K-A knockout mice exhibited defects in synaptic plasticity in the dentate gyrus and hippocampal-dependent memory formation. Taken together, these data suggest a novel kinase independent role for IP3K-A as an activity dependent Rac scaffolding protein in dendritic spine remodeling and memory formation.

Materials and Methods


Male Sprague Dawley rats (Orient, Kyeonggido, Korea) and mice were housed in a temperature (22°C), humidity (50%) and light-controlled vivarium (light on 07:00 p.m./light off 07:00 a.m.) with ad libitum access to food and water. The genetic background of mice was 129Sv × C57BL/6. They were backcrossed more than 10 times with C57BL/6. We used only littermate mice from heterozygous parents. Animals were treated in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals, and formal approval to conduct this experiment has been obtained from the animal subjects review board of Korea University.

In vivo LTP

In vivo LTP induction using rats was performed essentially as described (Fukazawa et al., 2003) with some modifications. Rats were anesthetized with sodium pentobarbital (60 mg/kg) injected intraperitoneally. A concentric electrode was positioned in the region of the perforant path (7.5 mm posterior and 4.0 mm lateral to the bregma). A recording electrode made of tungsten (125 µm in diameter, impedance 1-6M; A-M Systems) was positioned ipsilaterally in the DG granule cellular layer (3.8 mm posterior and 2.0 mm lateral to the bregma). For the baseline field potential recording, 50% of the maximum slope was used. To induce LTP, the biphasic square wave form of strong tetanic stimulation (four trains with 15 min intertrain intervals, with each train consisting of 20 bursts of 30 pulses at 400 Hz, delivered at 5 sec interburst intervals) was applied to the perforant path. These evoked potentials were amplified (× 1,000), filtered at 0.1 Hz-10k Hz bandwidth, and digitized at 10 kHz. Ten min after the last tetanus, potentiation was monitored for 45 min.

Cell culture

The rat hippocampus (embryonic, 18.5 days) was dissected and digested with 0.25% trypsin-EDTA (Gibco BRL) and trituration. The dissociated cells were plated on poly-D-lysine (Sigma) (50 µg/ml) precoated coverslips at a density of 7 × 104 cells/24-well plate. The complete growth medium consisted of neurobasal medium (Gibco BRL) containing B27 supplement (Gibco BRL), 0.5 mM L-glutamine (Gibco BRL), 25 mM glutamic acid (SIGMA), and antibiotics. The neurons were treated with 1 µM AraC (Sigma) to reduce the growth of contaminating nonneuronal cells. HeLa cells were seeded at a density of 2 × 104 cells/24-well plate and maintained in DMEM (Gibco BRL) plus 10% fetal bovine serum (Gibco BRL) and antibiotics.

Preparation of constructs

To construct pEGFP-tagged wild-type and deletion and point mutants of IP3K-A (Genbank access #NM-031045), inserts were created by PCR amplification of rat brain cDNA with the appropriate primers (see Supplemental table), digested with BamHI and HindIII (Promega, Madison, WI), then ligated into pEGFP-C1 vector (Clontech) with T4 DNA ligase (Promega). The GFP-IP3K-A-K262A was generated by site-directed mutagenesis using the QuickChange II XL mutagenesis kit (Stratagene) with the appropriate primers (Togashi et al., 1997; see Supplemental table). To construct DsRed-tagged Rac1 (Genbank access #NM-134366) and CRIB (Genbank access #NM-002576), inserts were created by PCR amplification of rat brain cDNA and AGS cell cDNA, respectively, with appropriate primers (see Supplemental table). The inserts were digested with EcoRI and BamHI (Promega) and ligated into pDsRed2-C1 vector (Clontech). Next, each construct was transformed into E. coli DH5α. For GST-tagged constructs, inserts were created by PCR amplification of rat brain cDNA with the appropriate primers (see Supplemental table), then amplified inserts were digested with BamHI and EcoRI (Promega) and ligated into pGEX-5X-1 vector (pGEX-2TK vector for GST-CRIB) (Amersham Biosciences), followed by transformation into E. coli BL-21 (CodonPlus®; Stratagene).

Adenoviral infection and transfection

Adenoviruses (synthesized at Neurogenex, Seoul, Korea) at a concentration of 200 MOI (multiplicity of infection) were used. For both neurons and HeLa cells, a half volume of medium was removed and replaced by virus-containing medium, and the cells returned to the incubator. Transfections of plasmids or siRNA into primary hippocampal neurons were performed in 24-well plates using the calcium phosphate method. Transfections of plasmids or siRNA into HeLa cells are performed using LIPOFECTAMINE™ 2000 (Invitrogen) according to the manufacturer’s instructions. After expression, neurons or HeLa cells were fixed with 4% paraformaldehyde, followed by further experiments.

Statistical analyses

For analysis of a difference between two groups, Student’s t-test or paired t-test (SPSS 12.0K) was used. When comparing among more than two groups, one way ANOVA (SPSS 12.0K) was used. ANOVA with repeated measures was employed to analyze data from the object recognition and radial arm maze tests (SAS). Scheffe, Tukey test and Bonfferoni correction were used for post hoc test (SPSS 12.0K). All values were expressed as mean ± SEM, and results were considered statistically significant if p<.05.

The experimental procedures of In vivo LTP for mice, Chemical LTP, GST pulldown assay, Immunoprecipitation, Western blotting, Immunohistochemistry, Immunocytochemistry, SiRNA, Electroporation, Protein expression, Subcellular fractionation, Purification of Rac1 protein, Rac1 activity assay, Golgi staining, Object recognition test, and Radial arm maze test can be found as Supplemental Experimental Procedure.


IP3K-A accumulates in dendritic spines upon LTP induction

Previously, we have reported that learning enhances IP3K-A protein expression in the hippocampal formation (Kim et al., 2004). The molecular mechanisms underlying this effect were further analyzed in vivo and in vitro. After induction of LTP in the dentate gyrus (DG) perforant path in vivo, we observed a marked accumulation of IP3K-A immunoreactivity (IR) in the middle and outer molecular layers of DG, which is the synaptic contact region with the medial and lateral perforant path from the entorhinal area (Figures 1A). Densitometric analysis for IP3K-A-IR in DG showed significant increase of density ratio (middle molecular layer / inner molecular layer) in the LTP-induced hemisphere compared to the non-treated contralateral side (t=11.09, df=6, p<.0001).

Figure 1
IP3K-A accumulates in dendritic spines upon LTP induction

For a more detailed examination of individual neurons, LTP was chemically induced in DIV 22 hippocampal primary neurons, and IP3K-A IR was quantified. Thirty minutes after chemical LTP (c-LTP) induction, the number of spines was increased (Figure 1B and Movie S1) and high levels of GluR1 signals were observed on the spines (Figure S1), which is consistent with previous reports (Xie et al., 2005). Interestingly, neurons with a greater number of spines displayed a stronger IP3K-A-IR signal in those spines (Figures 1B). A strong correlation (r=0.726, p<.0001) existed between spine number and mean value of the density ratio of IP3K-A (spine/shaft) (Figure 1C). These data suggest dynamic targeting of IP3K-A to the spine area by neural activation and a possible involvement of IP3K-A in activity-dependent spine formation.

IP3K-A promotes dendritic spine formation

To investigate the function of IP3K-A, we introduced small interference RNA (siRNA) specific for IP3K-A into neurons (the specificity of the IP3K-A-siRNA is shown in Figure S2). Following transfection of IP3K-A-siRNA in combination with GFP-expressing vectors into DIV 22 hippocampal neurons, GFP+ neurons exhibited markedly reduced IP3K-A expression (Figure 2A). Quantification revealed that the density of dendritic spines was also greatly reduced in IP3K-A-siRNA-transfected/GFP+ neurons when compared with control-siRNA-transfected/GFP+ neurons (t=14.17, df=52, p<.0001). To avoid off-target effects of siRNA, we also did rescue experiments and found that the effect of IP3K-A siRNA on spine formation could be rescued by expression of an siRNA-resistant N-terminal fragment of GFP-IP3K-A (Figure S2). These data indicate that IP3K-A is required for the development or maintenance of dendritic spines.

Figure 2
IP3K-A promotes dendritic spine formation

To further address the role of IP3K-A in spontaneous spine formation, IP3K-A was overexpressed in DIV 22 hippocampal neurons by infection with adenovirus containing full-length IP3K-A. Eighteen hours after infection, the neurons exhibited a markedly increased number of dendritic structures, and overexpressed GFP-IP3K-A signals were highly localized in the heads of dendritic protrusions (Figures 2B and 2C). Quantification revealed that the number of protrusions in IP3K-A-expressing neurons was significantly increased compared with that of the control GFP-infected group (F(3, 201)=90.03; Tukey test, p<.05) (Figure 2G). Thus, IP3K-A appears to be able to induce production of new dendritic protrusions.

IP3K-A catalytic activity regulates calcium levels by modulating the metabolism of IP3. Yet previous work suggests IP3K-A may not function to regulate calcium flux in neurons (Jun et al., 1998). Thus, it is important to evaluate whether the function of IP3K-A in protrusion formation is mediated by its kinase activity. To address this issue, we overexpressed a mutant IP3K-A containing a point mutation (IP3K-A-K262A) that lacks kinase activity (Togashi et al., 1997). Remarkably, GFP-IP3K-A-K262A overexpression produced results similar to those for wild-type IP3K-A overexpression (Tukey test, p<.05) (Figures 2E and 2G). This finding suggested that some function other than IP3K-A’s kinase activity is essential for the spinogenic activity observed. Consistent with this, deletion of the actin-binding domain (66 amino acids of the N-terminal end; GFP-IP3K-A-Δact) completely abolished the spinogenic activity of IP3K-A. Additionally, GFP-IP3K-A-Δact showed a dominant negative effect on spine density when compared to the GFP control (Tukey test, p<.05) (Figures 2D and 2G), suggesting that the F-actin binding ability of IP3K-A is essential for spine forming activity. GFP-IP3K-A-WT and GFP-IP3K-A-Δact signals were perfectly merged with F-actin (Figure 3A) and DsRed (Figure S8) signals respectively, indicating that both GFP signals can be used as markers for dendritic protrusion morphology.

Figure 3
IP3K-A influences actin dynamics and binds directly to only activated Rac1

Collectively, these data suggest that IP3K-A is involved in the formation or maintenance of dendritic spines via its ability to interface with the actin cytoskeleton.

IP3K-A influences actin dynamics

The above results indicated a functional interplay between IP3K-A and F-actin in the process of dendritic protrusion formation. Indeed, GFP-IP3K-A co-localizes with post-synaptic F-actin, further suggesting that spine remodeling in neurons by IP3K-A might be related to actin dynamics (Figure 3A). To examine whether actin reorganization is involved in the IP3K-A-induced increase in protrusion density, HeLa cells were infected with GFP-IP3K-A-expressing adenovirus. This heterologous system has two major advantages: 1) F-actin fibers of HeLa cells are clearly resolved, allowing close observation of actin dynamics; and 2) HeLa cells do not express IP3K-A, eliminating any possible influence of endogenous IP3K-A. Following infection, GFP-IP3K-A signals completely overlapped with F-actin fiber signals. GFP-IP3K-A-infected cells also substantially reorganized their F-actin structure, resulting in filopodia-like fiber formation (Figure 3B). The results of these HeLa cell-based assays indicate that a major effect of IP3K-A is to alter actin dynamics.

IP3K-A acts as a scaffold for activated Rac1

Because small GTPase proteins are known to be key factors for modulating dendritic spine formation through actin polymerization (Nakayama et al., 2000; Tashiro et al., 2000), we asked whether IP3K-A interacts with small GTPase proteins. In pulldown assays with rat hippocampal lysate (P2) using GST-IP3K-A, we found that IP3K-A interacted with Rac1, but not with RhoA and cdc42 (Figure 3C). Because Rac1 is active when GTP binds to it, but inactive when GDP binds to it (Van Aelst and D'Souza-Schorey, 1997), we asked whether the binding affinity of IP3K-A for Rac1 is dependent on the activity status of Rac1. To address this issue, GDP- or GTPγs-bound purified Rac1 was used in GST pulldown assays with GST-IP3K-A. This experiment also allowed us to examine whether IP3K-A binds directly to Rac1 or not. Interestingly, only GTPγs-bound/activated Rac1 exhibited efficient binding affinity to IP3K-A (Figure 3D). Furthermore, an immunoprecipitation assay with hippocampal lysate demonstrated that endogenous IP3K-A was able to bind endogenous Rac1 (Figure 3E). Increased binding between IP3K-A and Rac1 was observed when GTPγs was added to the lysate, whereas no Rac1 binding was detected in the control group that was precipitated by anti-GST antibody. Interaction between overexpressed IP3K-A and Rac1 also occurred in a HeLa cell model (Figure S3).

Because the catalytic activity of IP3K-A is not required for efficient spine remodeling, we suspected it may not function as a downstream effector of Rac. Instead our data indicated that IP3K-A might serve as a novel scaffold for active Rac1. One hallmark feature of a scaffold is that the binding interaction does not occlude other effectors from binding to Rac1. This predicts that while bound to IP3K-A, Rac1 should be able to still bind to downstream effectors such as PAK. To address this question we examined whether a trimeric complex consisting of IP3K-A, Rac1, and PAK can exist (Figure 3F). Hippocampal lysates were loaded with either GDP or GTPys and the Rac binding domain of PAK (CRIB domain) was used to isolate activated Rac1. As expected, western blot analysis showed that Rac1 was enriched in GTPγs-treated lysate. Immunoblot analysis of these samples showed IP3K-A was also co-precipitated by the CRIB pulldown assay, but only in GTPγs treated lysate that also contained Rac1. These results indicate that complexes containing IP3K-A and active Rac1 are able to simultaneously bind to Rac effectors such as PAK1.

IP3K-A regulates dendritic spines through Rac1 activity

The ability of IP3K-A to scaffold active Rac1 suggests Rac1 may be required for IP3K-A-induced effects in neurons. Thus, we next examined whether IP3K-A overexpression effects are dependent on Rac1 activity. When HeLa cells were treated with a Rac1-specific inhibitor (NSC23766) filopodia formation induced by GFP-IP3K-A was blocked, suggesting that Rac1 activity is essential for IP3K-A’s effects on cellular actin (Figure 4A). To obtain direct evidence of Rac1-mediated IP3K-A effects on the formation of dendritic protrusions, we treated DIV 22 hippocampal neurons that were infected by GFP-IP3K-A-expressing adenovirus with the NSC23766. After 18 hrs of treatment, IP3K-A-induced protrusion formation was partially blocked by the Rac1 inhibitor (F(2, 194)=65.2; Scheffe test, p<.05) (Figures 4B and 4C). These results indicate that IP3K-A-induced protrusion formation is dependent on Rac1 activity.

Figure 4
Rac1 activity is essential for IP3K-A to exert its effects on dendritic spines

Based on the above results, a question arises as to whether IP3K-A is involved in regulating the activity status of Rac1 (GDP- or GTP-bound). GST-CRIB pulldown assays using lysates of HeLa cells that had been transfected with GFP-IP3K-A revealed that overexpressed GFP-IP3K-A did not affect Rac1 activity, indicating that IP3K-A is not involved in the exchange of GDP-Rac1 to GTP-Rac1 (Figure S4).

LTP induction promotes IP3K-A/Rac complex formation and synaptic Rac signaling

At the cellular level, LTP induces cytoskeletal remodeling in dendritic spines (Nagerl et al., 2004; Chen et al., 2007). Accordingly, Fukazawa et al. (2003) reported increases in F-actin in the synaptic field by in vivo LTP induction. Because we observed enhanced IP3K-A-IR in the synaptic field by LTP induction (Figure 1), we asked whether Rac activity is enhanced in the synaptic area where IP3K-A has accumulated. Because immunostaining for activated Rac is not currently available, we instead tracked the phosphorylated PAK (p-PAK), a product of activated Rac. As was observed for IP3K-A (Figure 1A), p-PAK and F-actin were observed to be more accumulated in the synaptic field of the LTP-induced hemisphere than in the contralateral side (Figures 5A) [(F-actin: t=8.95, df=6, p<.0001), (p-PAK: t=7.05, df=6, p<.0001)]. In fact, total PAK1 was also accumulated in the synaptic field by LTP induction (Figure S5), suggesting that this Rac signaling pathway may be augmented. Furthermore, GST-CRIB pulldown assays with the synaptosomal (P2) fraction from c-LTP-induced hippocampal neurons demonstrated that the activated Rac1 pulled down with GST-CRIB was increased in c-LTP-induced neurons compared with that in the nontreated control group. Importantly, IP3K-A was pulled down together with activated Rac1, showing this signaling complex is organized downstream of synaptic activity (Figure 5C). Immunocytochemical assay also showed that p-PAK-IR overlapped with spine-targeted IP3K-A-IR (Figure 5B). Densitometry analysis revealed that more p-PAK-IRs were accumulated in dendritic spines by c-LTP induction (t=4.03, df=24, p<.0001), corresponding to the report of Chen et al. (2007) describing the p-PAK accumulation in the CA1 dendritic spines by LTP induction using rat brain slices. Because isolated dentate granule neurons cannot be cultured we examined the localization of IP3K-A in CA neurons as the best available proxy. These results likely hold true in both neuronal types since IP3K-A is expressed equally in the CA1 and DG. Taken together, these results suggest IP3K-A acts as an activity dependent scaffold for activated Rac in dendritic spines, where Rac downstream events are stimulated upon neural activation.

Figure 5
LTP induction activates Rac1 downstream events in dendritic spines

IP3K-A recruits activated Rac onto F-actin fibers

Based on the above observations, we hypothesized that IP3K-A recruits activated Rac1 onto the F-actin in the dendrites and spines following neuronal activation. To test this hypothesis, first we infected HeLa cells with adenoviruses containing full-length IP3K-A (GFP-IP3K-A-WT) or GFP-IP3K-A-Δact or GFP vector, then examined the localization of Rac1. Under normal conditions (GFP vector expression), Rac1-IR was diffusely localized in the cytoplasm and was not associated with F-actin (Figure 6A-a). However, following GFP-IP3K-A-WT overexpression, a majority of Rac1-IR and GFP-IP3K-A-WT signals were colocalized with augmented F-actin fibers (Figure 6A-c), whereas Rac1-IR was diffusely localized in cytoplasm when GFP-IP3K-A-Δact was overexpressed (Figure 6A-b). Z-axis image analysis indicated the colocalization of the three signals (F-actin, Rac1, and IP3K-A-WT) in IP3K-A-WT overexpressed cells. Because both GFP-IP3K-A-WT and GFP-IP3K-A-Δact can bind to Rac1 (Figure S6), this difference is not attributable to differential binding ability. Thus, we reasoned that F-actin binding ability of IP3K-A is required for the targeting of Rac1 onto the actin cytoskeleton. Moreover, cotransfection of GFP-IP3K-A and DsRed-CRIB into HeLa cells revealed that IP3K-A indeed recruits activated Rac onto the actin fibers (Figure 6B). Furthermore, p-PAK-IRs were dramatically merged with overexpressed GFP-IP3K-A signals, which are localized on the F-actin in the head of dendritic protrusions (Figure 3A and Figure 6C), suggesting that IP3K-A-dependent targeting of activated Rac onto F-actin cytoskeleton also occurs in neurons.

Figure 6
IP3K-A recruits activated Rac1 onto F-actin fibers

In vivo impairment of activity-dependent accumulation of PAK1/F-actin and synaptic plasticity in IP3K-A knockout mice

The combined results so far indicate a model whereby IP3K-A plays an important role in synaptic targeting of active Rac. Our in vitro data predicts that in the absence of IP3K-A there may be alterations to the developmental course leading to synapse formation. Golgi staining followed by camera lucida analyses with young KO mice (4-week-old) revealed a significantly decreased density of dendritic spines in DG neurons (F(3, 14)=125.2; Scheffe test, p<.05). However, the spine density recovered to the WT value in 8-week-old KO mice (Figure S7). These observations suggest a retardation of the spine development in KO mice, which is rescued by an unknown compensatory mechanism at later time points. Nonetheless these data are in agreement with our in vitro data implicating IP3K-A function during active spinogenesis.

Our model also predicts that in the absence of IP3K-A, synaptic activation of Rac dependent pathways and actin remodeling may be blunted. To test this, we examined whether adult IP3K-A KO mice have defects in neural activation-induced accumulation of F-actin and PAK1 in the synaptic area of dentate gyrus (DG). Following tetanic stimulation in the perforant path of 3-month-old mice, we observed less PAK1 and F-actin in MML and OML of the DG in KO mice when compared with littermate controls (t=43.81, df=4, p<.0001 for PAK1; t=8.50, df=4, p<.001 for F-actin) (Figures 7A and 7B), suggesting that IP3K-A plays an essential role in activated Rac targeting as well as in activity-dependent actin dynamics in DG neurons in vivo.

Figure 7
IP3K-A knockout mice show defects in LTP and actin dynamics in dentate gyrus

To elucidate the physiological significance of IP3K-A in adult brains, we next investigated the synaptic plasticity in IP3K-A KO mice. Previously, Jun et al. (1998) reported that the hippocampal slices of IP3K-A KO mice showed enhanced LTP in CA1 region. Additionally, there was a slight, but distinct disruption of DG-LTP in KO mice during first 30 min even though the magnitude of potentiation was under 0.15mV induced by 100Hz stimulations. To examine plasticity in the DG in a more physiologic context, we used an in vivo LTP induction and measurement method. For this, we induced LTP in the anesthetized mice (10-week-old) using stereotactically placed probes. LTP was induced by high frequency stimulation (HFS) in perforant path-DG synapses. Notably, IP3K-A KO mice exhibited disrupted LTP compared with WT littermate mice. The magnitude of potentiation following HFS was enhanced 363.4 ± 8.6% or 102.6 ± 5.9% over the initial baseline average in the WT or KO mice, respectively (Figure 7C). The difference of the enhancement between the two genotypes was statistically significant (paired t-test; t=29.5, df=49, p<.0001). Although KO mice also showed inductions of population spikes by HFS (72% of WT mice), the induced potentiations were not maintained over 20 minutes, suggesting the possibility that rapid remodeling of actin by IP3K-A is important to the maintenance of synaptic plasticity in vivo. Previous report showed that there was no difference in paired pulse facilitation in perforant path between WT and KO (Jun et al, 1998). Basically, in addition, IP3K-A does not express in axon including pre-synapse (data not shown), suggesting that IP3K-A may not be essential for axonal input.

IP3K-A knockout mice show defects in memory performances

Previously it has been reported that there was no significant difference in escape latency between IP3K-A KO and WT mice in the Morris water maze test (Jun et al., 1998); we have also confirmed this result (data not shown). However, we observed that IP3K-A KO mice have augmented fear levels compared with WT mice (manuscript in preparation). Because Morris water maze may evoke fear-related responses, a high level of fear in KO mice should enhance the driving force to search for the hidden platform, complicating the interpretation of these results. Thus, we tested the capability for memory formation in IP3K-A KO mice using alternative hippocampal-dependent memory tasks that are less influenced by fear motivations. We first performed an object recognition test. In this test, mice are instinctively interested in a novel object rather than a familiar one. If a mouse remembers the object that has been explored before, it will normally prefer to explore and contact a new object more frequently. As expected, WT mice recognized and spent more time exploring the new object in a 24 hrs choice test. In contrast, IP3K-A KO mice did not recognize the novel object and spent similar amounts of time exploring each object (Scheffe test, p<.05) (Figure 8A)(See Supplemental Information for additional statistical analysis). Thus, KO mice have a reduced ability to process novel object recognition memory. A 72 hour choice was subsequently performed, using the familiar versus yet another novel object. In this trial, where both genotypes have been exposed to the familiar object on three occasions, both WT and KO mice recognized the novel object. These data indicate that with additional trial the IP3K-A KO mice can distinguish the familiar versus novel object and confirm these mice retain novelty-seeking behavior. Importantly, neophobic responses to the objects did not confound the behavior in either cohort, as the total number of touches on objects was not different between either genotype (data not shown).

Figure 8
IP3K-A knockout mice show defects in synaptic plasticity and memory performances

Memory deficits were also noted in an independent assay to evaluate spatial memory using the radial arm maze test. In this task, mice were required to remember previously visited maze arms extending from a central starting platform in order to get rewards. Spatial memory errors were measured as the number of visits to the same arm more than once, where the reward had already been obtained. Interestingly, both WT and KO mice improved their error rate in the first three trials. After this point WT mice continued to improve, while KO mice did not. Statistical analyses with ANOVA (repeated measures) followed by Bonfferoni correction revealed that the KO mice were impaired in spatial memory on sessions 6 and 7 (p<.05) (Figure 8B)(See Supplemental Information for additional statistical analysis). These results suggest that the spatial memory of IP3K-A KO mice was also impaired in this test.


In the present study, we identified a novel function of IP3K-A that regulates activity-dependent remodeling of dendritic spine actin by scaffolding active Rac. We found that IP3K-A is translocated to dendritic spines by neural activation, where it recruits and anchors activated Rac to the actin cytoskeleton. Our in vivo and in vitro data may provide new insights into the site-specific and time-dependent signaling mechanisms of synaptic plasticity, which appears to be important for the processes of memory formation.

The plasticity-dependent localization of IP3K-A in dendritic spines

We demonstrated here the targeting of IP3K-A in dendritic spines follows LTP induction. Previously, it has been reported that treatment of cultured neurons with glutamate or NMDA diminishes IP3K-A levels in dendritic spine heads (Schell and Irvine, 2006). In agreement with this, we also observe a similar loss of IP3K-A from spine heads following 100 µM glutamate treatment; yet with this treatment, the spines eventually collapsed (data not shown). Accumulating evidence suggests that direct treatment of brain slices with NMDA produces long-term depression-like phenomena (Lee et al., 1998), which promote the shrinkage of spines (Zhou et al., 2004). Furthermore, we have also previously shown that aberrant neuronal activations by chemical reagents (kainic acid) or electric stimulation (electroconvulsive shock) reduce the expression of IP3K-A (Kim et al., 1994; Sun et al., 2006). Because these conditions reduce synaptic efficacy and induce amnesia (Squire and Spanis, 1984; Holmes, 1991), these results collectively suggested that the expression and localization of IP3K-A might be dynamically regulated by neuronal plasticity. Our current study evaluated the dynamics of IP3K-A localization following LTP induction. Following in vivo and in vitro c-LTP induction, IP3K-A dramatically accumulated in the synaptic area, demonstrating that the localization of IP3K-A is regulated by neuronal activity. Therefore, the localization of IP3K-A in the dendrites appears to be intimately paired with mechanisms of synaptic plasticity in vivo.

IP3K-A induced remodeling of dendritic spines utilizes a novel kinase independent mechanism

Overexpression of IP3K-A in neurons markedly increased the number of dendritic protrusions, whereas knockdown of IP3K-A expression reduced the number of these protrusions. Together these results suggest that IP3K-A can modulate structural reorganization of dendritic spines. Until now, the only known role of IP3K-A was its kinase activity, which phosphorylates IP3 to IP4 (Irvine et al., 1986). IP4 is known to regulate calcium influx from the ER in combination with IP3 (Hermosura et al., 2000). Because intra-spine calcium is essential for the control of actin dynamics and spine morphology (Oertner et al, 2005), it was natural to assume that the mechanism of IP3K-A to modify actin dynamics is related to its catalytic activity. Surprisingly, the present results exclude this idea. Overexpression of IP3K-A-K262A, a point mutant lacking catalytic activity, recapitulated the spine-forming effect of normal IP3K-A, suggesting that the catalytic activity of IP3K-A is not essential for spine formation. In line with our result, cytoskeletal reorganization induced by IP3K-A overexpression was observed in H1299 cells, independent of its catalytic activity (Windhorst et al., 2008). Furthermore, Jun et al. (1998) reported that the responses of intracellular calcium levels to glutamate treatment in the neurons of IP3K-A KO mice were similar to those of neurons in WT mice, adding to the likelihood that the kinase activity of IP3K-A is not relevant to its function in spines.

Although IP3K-A lacking catalytic activity could still induce spine formation, deletion of IP3K-A’s actin-binding domain completely eliminated this effect, indicating that the spine-forming activity of IP3K-A is related to its ability to interact with F-actin. Because IP3K-A was found to be a direct binding partner of the active form of Rac1, and inhibition of Rac1 partially blocked IP3K-A’s spine-forming activity, we hypothesized that IP3K-A may play an anchoring role between activated Rac and actin cytoskeleton in dendritic spines.

Rac signaling remodels F-actin structures, producing a specific morphology in many types of cells, including neurons (Hering and Sheng, 2001; Etienne-Manneville and Hall, 2002). Constitutively active Rac disrupts normal spine morphology of neurons, causing a net increase of dendritic protrusions. Conversely, dominant negative Rac causes a progressive reduction of the number of spines (Nakayama et al., 2000; Tashiro et al., 2000), indicating that Rac activity is important for the maintenance of spine density. The regulation of Rac activity status (GTP-GDP exchange) by GEFs and GTPase-activating proteins (GAPs) is crucial for activity-dependent synaptic remodeling (Van Aelst and D'Souza-Schorey, 1997). For instance, Rac-GEFs betaPIX, Tiam1 and kalirin-7 are localized within the spine head and their activity is regulated by synaptic activity (Penzes et al., 2001; Tolias et al., 2005; Saneyoshi et al., 2008). In addition to Rac activation, the anchoring and targeting of activated Rac to ‘hot-spots’ of actin remodeling may be required to promote actin dynamics-mediated spine formation. Additionally, if the positioning of activated Rac is restricted to pre-existing spines, it would be difficult to explain the phenomena of the formation of new spines from the dendritic shaft by the Rac pathway. We propose in this study that IP3K-A recruits activated Rac to F-actin-rich domains where Rac signaling is required for mechanisms of synaptic plasticity. This is supported by our findings that following LTP induction, p-PAK accumulated with an IP3K-A/F-actin complex in the synaptic area in adult rat DG. Importantly, IP3K-A KO mice showed a clear reduction of PAK1 and F-actin accumulation in the synaptic field. These data highlight the possibility that IP3K-A plays an essential scaffolding role in support of Rac signaling events by localizing activated Rac within the actin-rich spines of neurons.

Impairment of synaptic plasticity and memory formation in IP3K-A knockout mice

Consistent with the reduced PAK1 and F-actin accumulations we observed in IP3K-A KO mice, in vivo LTP was also greatly impaired, suggesting that the IP3K-A/Rac/F-actin complex we have identified is essential for synaptic plasticity in vivo.

These defects were also manifested by behavioral impairments in the IP3K-A KO mice. Object recognition and radial arm maze tests consistently demonstrated the impairment of hippocampal-dependent memory formation in IP3K-A KO mice. Recognition memory is associated with the function of the hippocampal formation regions in humans, monkeys, and rodents (Myhrer, 1988; Reed and Squire, 1997; Zola et al., 2000). For example, rats seem to be unable to recognize novelty when hippocampal perforant path projection is disrupted (Myhrer, 1988). In present study, IP3K-A KO mice showed a deficit in remembering the object seen the day before the testing, indicating reduced ability of recognition memory of KO mice.

We also used the radial arm maze test, in which the ability to master the task being tested is also dependent on the hippocampal formation, including the DG (McNaughton et al., 1989). Although the radial arm maze test can evaluate the capacity of spatial working memory, this task also requires reference memory (Satoh et al., 2007). Interestingly, IP3K-A KO mice showed no difference in radial arm maze test error rate as compared with WT animals in the first 3 sessions, suggesting that there was no specific working memory deficit during that early period. Yet, while WT mice continued to improve their error rate after session 3, KO mice did not. Therefore, it is possible that the spatial memory deficit of KO mice in late phases may be caused by the retardation of working memory improvement, or by differences in reference memory. In either scenario, our analysis using both the novel object recognition and radial arm tests are consistent with deficits in learning and memory as a consequence of the loss of IP3K-A dependent pathways.

In summary, we define a previously unknown mechanism to scaffold active Rac to spines in response to plasticity inducing stimuli. Our data indicate that this mechanism is essential for in vivo Rac signaling pathways, synaptic plasticity, and aspects of hippocampal-dependent learning and memory.

Supplementary Material





We are grateful to Drs. Menahem Segal, William C. Wetsel, and Michael D. Ehlers for critical reading of the manuscript, to Drs. Im Joo Rhyu, Byung-Il Choi, Eunjoon Kim, and Dong Ki Kim for advices and consultations throughout this study, to Dr. Ryuichi Shigemoto for providing a GluR1 antibody, to Mi-Ra Noh and Jee-Woong Kim for technical support. This work was supported by Korea Science and Engineering Foundation (M103KV010023-07K2201-02310 and M1050000004905J000004900 to H.K), a grant of the Korea Health 21 R&D project, Ministry of Health and Welfare, Republic of Korea (02-PJ1-PG1-CH06-0001 to H.K.) and by a National Institutes of Health Grant (R01-NS059957 to S.H.S.).


Supplemental Data

Supplemental Data include eight figures, one Movie, Supplemental Experimental Procedures, Supplemental Statistical Analyses, and a Supplemental Table can be found with this article on line.


  • Bliss TV, Collingridge GL, Morris RG. Introduction. Long-term potentiation and structure of the issue. Philos Trans R Soc Lond B Biol Sci. 2003;358:607–611. [PMC free article] [PubMed]
  • Bonhoeffer T, Yuste R. Spine motility. Phenomenology, mechanisms, and function. Neuron. 2002;35:1019–1027. [PubMed]
  • Chen LY, Rex CS, Casale MS, Gall CM, Lynch G. Changes in synaptic morphology accompany actin signaling during LTP. J Neurosci. 2007;27:5363–5372. [PubMed]
  • Choi KY, Kim HK, Lee SY, Moon KH, Sim SS, Kim JW, Chung HK, Rhee SG. Molecular cloning and expression of a complementary DNA for inositol 1,4,5-trisphosphate 3-kinase. Science. 1990;248:64–66. [PubMed]
  • Dailey ME, Smith SJ. The dynamics of dendritic structure in developing hippocampal slices. J Neurosci. 1996;16:2983–2994. [PubMed]
  • Engert F, Bonhoeffer T. Dendritic spine changes associated with hippocampal long-term synaptic plasticity. Nature. 1999;399:66–70. [PubMed]
  • Etienne-Manneville S, Hall A. Rho GTPases in cell biology. Nature. 2002;420:629–635. [PubMed]
  • Fukazawa Y, Saitoh Y, Ozawa F, Ohta Y, Mizuno K, Inokuchi K. Hippocampal LTP is accompanied by enhanced F-actin content within the dendritic spine that is essential for late LTP maintenance in vivo. Neuron. 2003;38:447–460. [PubMed]
  • Hering H, Sheng M. Dendritic spines: structure, dynamics and regulation. Nat Rev Neurosci. 2001;2:880–888. [PubMed]
  • Hermosura MC, Takeuchi H, Fleig A, Riley AM, Potter BV, Hirata M, Penner R. InsP4 facilitates store-operated calcium influx by inhibition of InsP3 5-phosphatase. Nature. 2000;408:735–740. [PubMed]
  • Holmes GL. The long-term effects of seizures on the developing brain: clinical and laboratory issues. Brain Dev. 1991;13:393–409. [PubMed]
  • Irvine RF, Letcher AJ, Heslop JP, Berridge MJ. The inositol tris/tetrakisphosphate pathway--demonstration of Ins(1,4,5)P3 3-kinase activity in animal tissues. Nature. 1986;320:631–634. [PubMed]
  • Jun K, Choi G, Yang SG, Choi KY, Kim H, Chan GC, Storm DR, Albert C, Mayr GW, Lee CJ, Shin HS. Enhanced hippocampal CA1 LTP but normal spatial learning in inositol 1,4,5-trisphosphate 3-kinase(A)-deficient mice. Learn Mem. 1998;5:317–330. [PubMed]
  • Kennedy MB, Beale HC, Carlisle HJ, Washburn LR. Integration of biochemical signalling in spines. Nat Rev Neurosci. 2005;6:423–434. [PubMed]
  • Kim H, Ko JP, Kang UG, Park JB, Kim HL, Lee YH, Kim YS. Electroconvulsive shock reduces inositol 1,4,5-trisphosphate 3-kinase mRNA expression in rat dentate gyrus. J Neurochem. 1994;63:1991–1994. [PubMed]
  • Kim HT, Kim IH, Lee KJ, Lee JR, Park SK, Chun YH, Kim H, Rhyu IJ. Specific plasticity of parallel fiber/Purkinje cell spine synapses by motor skill learning. Neuroreport. 2002;13:1607–1610. [PubMed]
  • Kim IH, Park SK, Sun W, Kang Y, Kim HT, Kim H. Spatial learning enhances the expression of inositol 1,4,5-trisphosphate 3-kinase A in the hippocampal formation of rat. Brain Res Mol Brain Res. 2004;124:12–19. [PubMed]
  • Lee HK, Kameyama K, Huganir RL, Bear MF. NMDA induces long-term synaptic depression and dephosphorylation of the GluR1 subunit of AMPA receptors in hippocampus. Neuron. 1998;21:1151–1162. [PubMed]
  • Leuner B, Falduto J, Shors TJ. Associative memory formation increases the observation of dendritic spines in the hippocampus. J Neurosci. 2003;23:659–665. [PMC free article] [PubMed]
  • Lin B, Kramar EA, Bi X, Brucher FA, Gall CM, Lynch G. Theta stimulation polymerizes actin in dendritic spines of hippocampus. J Neurosci. 2005;25:2062–2069. [PubMed]
  • Luo L. Actin cytoskeleton regulation in neuronal morphogenesis and structural plasticity. Annu Rev Cell Dev Biol. 2002;18:601–635. [PubMed]
  • Mailleux P, Takazawa K, Erneux C, Vanderhaeghen JJ. Inositol 1,4,5-trisphosphate 3-kinase mRNA: high levels in the rat hippocampal CA1 pyramidal and dentate gyrus granule cells and in cerebellar Purkinje cells. J Neurochem. 1991;56:345–347. [PubMed]
  • Maletic-Savatic M, Malinow R, Svoboda K. Rapid dendritic morphogenesis in CA1 hippocampal dendrites induced by synaptic activity. Science. 1999;283:1923–1927. [PubMed]
  • Matus A, Ackermann M, Pehling G, Byers HR, Fujiwara K. High actin concentrations in brain dendritic spines and postsynaptic densities. Proc Natl Acad Sci U S A. 1982;79:7590–7594. [PubMed]
  • McNaughton BL, Barnes CA, Meltzer J, Sutherland RJ. Hippocampal granule cells are necessary for normal spatial learning but not for spatially-selective pyramidal cell discharge. Exp Brain Res. 1989;76:485–496. [PubMed]
  • Myhrer T. Exploratory behavior and reaction to novelty in rats with hippocampal perforant path systems disrupted. Behav Neurosci. 1988;102:356–362. [PubMed]
  • Nagerl UV, Eberhorn N, Cambridge SB, Bonhoeffer T. Bidirectional activity-dependent morphological plasticity in hippocampal neurons. Neuron. 2004;44:759–767. [PubMed]
  • Nakayama AY, Harms MB, Luo L. Small GTPases Rac and Rho in the maintenance of dendritic spines and branches in hippocampal pyramidal neurons. J Neurosci. 2000;20:5329–5338. [PubMed]
  • Oertner TG, Matus A. Calcium regulation of actin dynamics in dendritic spines. Cell Calcium. 2005;37:477–482. [PubMed]
  • Penzes P, Johnson RC, Sattler R, Zhang X, Huganir RL, Kambampati V, Mains RE, Eipper BA. The neuronal Rho-GEF Kalirin-7 interacts with PDZ domain-containing proteins and regulates dendritic morphogenesis. Neuron. 2001;29:229–242. [PubMed]
  • Penzes P, Beeser A, Chernoff J, Schiller MR, Eipper BA, Mains RE, Huganir RL. Rapid induction of dendritic spine morphogenesis by trans-synaptic ephrinB-EphB receptor activation of the Rho-GEF kalirin. Neuron. 2003;37:263–274. [PubMed]
  • Purpura DP. Dendritic spine "dysgenesis" and mental retardation. Science. 1974;186:1126–1128. [PubMed]
  • Reed JM, Squire LR. Impaired recognition memory in patients with lesions limited to the hippocampal formation. Behav Neurosci. 1997;111:667–675. [PubMed]
  • Saneyoshi T, Wayman G, Fortin D, Davare M, Hoshi N, Nozaki N, Natsume T, Soderling TR. Activity-dependent synaptogenesis: regulation by a CaM-kinase kinase/CaM-kinase I/betaPIX signaling complex. Neuron. 2008;57:94–107. [PMC free article] [PubMed]
  • Satoh Y, Endo S, Ikeda T, Yamada K, Ito M, Kuroki M, Hiramoto T, Imamura O, Kobayashi Y, Watanabe Y, Itohara S, Takishima K. Extracellular signal-regulated kinase 2 (ERK2) knockdown mice show deficits in long-term memory; ERK2 has a specific function in learning and memory. J Neurosci. 2007;27:10765–10776. [PubMed]
  • Schell MJ, Erneux C, Irvine RF. Inositol 1,4,5-trisphosphate 3-kinase A associates with F-actin and dendritic spines via its N terminus. J Biol Chem. 2001;276:37537–37546. [PubMed]
  • Schell MJ, Irvine RF. Calcium-triggered exit of F-actin and IP(3) 3-kinase A from dendritic spines is rapid and reversible. Eur J Neurosci. 2006;24:2491–2503. [PubMed]
  • Segal M. Dendritic spines and long-term plasticity. Nat Rev Neurosci. 2005;6:277–284. [PubMed]
  • Squire LR, Spanis CW. Long gradient of retrograde amnesia in mice: continuity with the findings in humans. Behav Neurosci. 1984;98:345–348. [PubMed]
  • Star EN, Kwiatkowski DJ, Murthy VN. Rapid turnover of actin in dendritic spines and its regulation by activity. Nat Neurosci. 2002;5:239–246. [PubMed]
  • Sun W, Kang Y, Kim IH, Kim EH, Rhyu IJ, Kim HJ, Kim H. Inhibition of rat brain inositol 1,4,5-trisphosphate 3-kinase A expression by kainic acid. Neurosci Lett. 2006;392:181–186. [PubMed]
  • Tashiro A, Minden A, Yuste R. Regulation of dendritic spine morphology by the rho family of small GTPases: antagonistic roles of Rac and Rho. Cereb Cortex. 2000;10:927–938. [PubMed]
  • Togashi S, Takazawa K, Endo T, Erneux C, Onaya T. Structural identification of the myo-inositol 1,4,5-trisphosphate-binding domain in rat brain inositol 1,4,5-trisphosphate 3-kinase. Biochem J. 1997;326(Pt 1):221–225. [PubMed]
  • Tolias KF, Bikoff JB, Burette A, Paradis S, Harrar D, Tavazoie S, Weinberg RJ, Greenberg ME. The Rac1-GEF Tiam1 couples the NMDA receptor to the activity-dependent development of dendritic arbors and spines. Neuron. 2005;45:525–538. [PubMed]
  • Van Aelst L, D'Souza-Schorey C. Rho GTPases and signaling networks. Genes Dev. 1997;11:2295–2322. [PubMed]
  • Windhorst S, Blechner C, Lin HY, Elling C, Nalaskowski M, Kirchberger T, Guse AH, Mayr GW. Ins(1,4,5)P3 3-kinase-A overexpression induces cytoskeletal reorganization via a kinase-independent mechanism. Biochem J. 2008;414:407–417. [PubMed]
  • Xie Z, Huganir RL, Penzes P. Activity-dependent dendritic spine structural plasticity is regulated by small GTPase Rap1 and its target AF-6. Neuron. 2005;48:605–618. [PubMed]
  • Yuste R, Bonhoeffer T. Morphological changes in dendritic spines associated with long-term synaptic plasticity. Annu Rev Neurosci. 2001;24:1071–1089. [PubMed]
  • Zhou Q, Homma KJ, Poo MM. Shrinkage of dendritic spines associated with long-term depression of hippocampal synapses. Neuron. 2004;44:749–757. [PubMed]
  • Zola SM, Squire LR, Teng E, Stefanacci L, Buffalo EA, Clark RE. Impaired recognition memory in monkeys after damage limited to the hippocampal region. J Neurosci. 2000;20:451–463. [PubMed]