The calcium/calmodulin-dependent kinase type II (CaMKII) holoenzyme of the forebrain predominantly consists of heteromeric complexes of the αCaMKII and βCaMKII isoforms. Yet, in contrast to αCaMKII, the role of βCaMKII in hippocampal synaptic plasticity and learning has not been investigated. Here, we compare two targeted Camk2b mouse mutants to study the role of βCaMKII in hippocampal function. Using a Camk2b−/− mutant, in which βCaMKII is absent, we show that both hippocampal-dependent learning and Schaffer collateral–CA1 long-term potentiation (LTP) are highly dependent upon the presence of βCaMKII. We further show that βCaMKII is required for proper targeting of αCaMKII to the synapse, indicating that βCaMKII regulates the distribution of αCaMKII between the synaptic pool and the adjacent dendritic shaft. In contrast, localization of αCaMKII, hippocampal synaptic plasticity and learning were unaffected in the Camk2bA303R mutant, in which the calcium/calmodulin-dependent activation of βCaMKII is prevented, while the F-actin binding and bundling property is preserved. This indicates that the calcium/calmodulin-dependent kinase activity of βCaMKII is fully dispensable for hippocampal learning, LTP, and targeting of αCaMKII, but implies a critical role for the F-actin binding and bundling properties of βCaMKII in synaptic function. Together, our data provide compelling support for a model of CaMKII function in which αCaMKII and βCaMKII act in concert, but with distinct functions, to regulate hippocampal synaptic plasticity and learning.
Calcium/calmodulin-dependent kinase type II (CaMKII) is one of the most abundant proteins of the hippocampus, and its role in hippocampal plasticity and learning has been thoroughly investigated by pharmacological and genetic approaches. However, in the hippocampus there are two major isoforms of CaMKII, α and β, which cannot be distinguished using pharmacological approaches. In addition, almost all genetic approaches have focused on the α isoform (Silva et al., 1992; Mayford et al., 1996; Giese et al., 1998; Elgersma et al., 2002; Miller et al., 2002; Wang et al., 2008; Yamagata et al., 2009) (for an overview of CaMKII mutants, see Elgersma et al., 2004). Collectively, these studies have demonstrated that activation of αCaMKII is necessary for normal synaptic plasticity and learning, and, not surprisingly, its deregulation can lead to severe cognitive impairments (van Woerden et al., 2007).
In the hippocampus, αCaMKII and βCaMKII form a holoenzyme consisting of ∼12 subunits in a 2:1 ratio (Brocke et al., 1999). αCaMKII and βCaMKII are highly homologous, but they are encoded by two distinct genes (Camk2a and Camk2b, respectively) (Hudmon and Schulman, 2002a). The most noticeable difference between these isoforms is that βCaMKII is able to bind to F-actin in an activity-controlled manner, through its extra domain in the variable region (Shen et al., 1998; Shen and Meyer, 1999). Two βCaMKII subunits per holoenzyme are already sufficient to change the localization of the entire holoenzyme (Shen et al., 1998). Mainly due to these different actin-binding properties, αCaMKII and βCaMKII were shown to have opposing effects on synaptic strength in cultured neurons (Thiagarajan et al., 2002). Interestingly, βCaMKII not only binds to actin, but is also capable of bundling actin in a kinase-independent manner (O'Leary et al., 2006; Okamoto et al., 2007; Sanabria et al., 2009). This nonenzymatic bundling feature is likely achieved by single CaMKII oligomers binding to multiple actin filaments.
Studies addressing the role of βCaMKII in synaptic plasticity and learning have only recently been initiated. Inducible overexpression of βCaMKII in the dentate gyrus did not affect acquisition of hippocampal learning, but did affect the long-term consolidation of memories (Cho et al., 2007). Additionally, it was shown that the absence of βCaMKII reverses the polarity of plasticity at cerebellar parallel fiber–Purkinje cell synapses and causes significant cerebellar learning deficits (van Woerden et al., 2009). The reversal of plasticity is caused in part by a nonenzymatic property of βCaMKII, which prevents precocious activation of αCaMKII under low-calcium conditions.
Here we examined the role of βCaMKII in hippocampal synaptic plasticity and learning using two different βCaMKII mutants: (1) the Camk2b−/− mouse, which does not express βCaMKII; and (2) the Camk2bA303R mouse, where a point mutation blocks calcium/calmodulin binding, selectively preventing its enzymatic activation, while preserving its ability to bind to actin (Shen and Meyer, 1999; Fink et al., 2003; Lin and Redmond, 2008). We found that the absence of βCaMKII causes mislocalization of αCaMKII, impaired hippocampal synaptic plasticity, and impaired hippocampus-dependent learning. In contrast, these phenotypes were not present in the Camk2bA303R mutants, arguing that the actin binding and bundling function of βCaMKII governs a major aspect of its synaptic function. These results strongly suggest an essential, but nonenzymatic role for βCaMKII in hippocampal plasticity.
Materials and Methods
Generation of the Camk2bA303R mutants.
The Camk2bA303R targeting construct was generated as follows. The Camk2b genomic sequence (ENSMUSG00000057897) was obtained from a public database (Ensembl) and used to design the primers for the targeting constructs. PCR fragments encompassing exon 6–11 using 5′ primer: 5′-GGTACCTGAGGAAGGTGCCAGCTCTGTCCC-3′ and 3′ primer: 5′-GTCGACCAGGGTAGTCACGGTTGTCC-3′ (5.3 kb; exon denotation according to ENSMUST00000019133) and exon 11–12 using 5′ primer: 5′-GCGGCCGCCTGTTAAAGGAATGGTTCTC-3′ and 3′ primer: 5′-ATGCATCTAAAAGGCAGGCAGGATGATCTGC-3′ (6 kb) were amplified using High Fidelity Taq Polymerase (Roche) on ES cell genomic DNA and cloned on either site of a PGK (phosphoglycerate kinase)–Neomycin selection cassette. All exons were sequenced to verify that no mutations were introduced accidentally. Site-directed mutagenesis was used to introduce the point mutation Ala303Arg. For counter selection, a gene encoding diphtheria toxin chain A (DTA) was inserted at the 5′ of the targeting construct. The targeting construct was linearized and electroporated into embryonic day 14 (E14) ES cells (derived from 129P2 mice). Cells were cultured in BRL cell-conditioned medium in the presence of leukemia inhibitory factor. After selection with G418 (200 μg/ml), targeted clones were identified by PCR (long-range PCR from neomycin resistance gene to the region flanking the targeted sequence). A clone with normal karyotype was injected into blastocysts of C57BL/6 mice. Male chimeras were crossed with female C57BL/6 mice (Harlan). The resulting F1 heterozygous mice (in the 129P2-C57BL/6 background) were used to generate F2 homozygous mutants and wild-type littermate controls. These mice, of either sex, were used for all the behavioral and electrophysiological experiments. The experimenter was blind for the genotype, but homozygous mice were easily recognizable by the ataxic gait. Therefore, a second person blind to the genotype also analyzed the data. Mice were housed on a 12 h light/dark cycle with food and water available ad libitum, and were used between 2 and 6 months of age for all experiments described (including electrophysiology). All animal procedures were approved by a Dutch Ethical Committee for animal experiments.
Lysates were prepared by quick dissection of the brain and by homogenization of the brain tissue in lysis buffer (10 mm TRIS-HCl 6.8, 2.5% SDS, 2 mm EDTA, and protease and phosphatase inhibitor cocktails; Sigma). The concentration of the lysates was adjusted to 1 mg/ml. Ten micrograms was used for Western blot analysis. Western blots were probed with antibodies directed against αCaMKII (MAB3119, 1:10,000; Millipore Bioscience Research Reagents), βCaMKII (CB-β1, 1:10,000; Zymed), Ph-T286/T87 CaMKII antibody (1:5000; catalog #06–881, Millipore Cell Signaling Solutions), and actin (MAB1501R, 1:2000; Millipore Bioscience Research Reagents). Blots were stained using Enhanced ChemoLuminescence (catalog #32106, Pierce). Western blot quantification was performed using NIH-Image.
Immunocytochemistry was performed on free-floating 40-μm-thick frozen sections using a standard avidin-biotin-immunoperoxidase complex method (ABC, Vector Laboratories) with βCaMKII (CB-b1, 1:2000; Zymed) as the primary antibody and diaminobenzidine (0.05%) as the chromogen (Hansel, 2006). For gross brain morphology, sections were stained with thionin.
Golgi–Cox staining on unfixed hippocampi of three Camk2b−/− mutants and three wild-type mice was performed using the FD Rapid GolgiStain Kit (FD NeuroTechnologies), according to the manufacturers' instructions. Sagittal sections, 100 μm thick, were cut on a microtome with cryostat adaptations. Pyramidal cell counting and selection for further detailed analysis was done by two independent observers who were both blind for genotype. A calibration grid was used to count the number of spines per 10 μm, using a 40× objective.
Primary hippocampal cultures and immunohistochemistry.
βCaMKII or βCaMKII-A303R heterozygous mice were crossed, and wild-type, heterozygous, and knock-out hippocampal neuron cultures were prepared from brains of single E18 embryos out of mixed genotype litters. Mouse hippocampal neurons were isolated and prepared as described previously (Goslin and Banker, 1991). In short, the two hippocampi were removed from the embryonic brain, collected in 1 ml of DMEM on ice, washed two times with 1 ml of DMEM, and incubated in trypsin/EDTA solution (Invitrogen ) at 37°C for 15 min. After washing with 1 ml of DMEM, the cells were resuspended in neurobasal medium (NB) supplemented with 2% B27, 1% penicillin/streptomycin, and 1% glutamax (Invitrogen ), and dissociated using a gently flamed Pasteur pipette. Neurons were plated in a small drop on poly-l-lysine- (100 μg/ml, Sigma) and laminin-coated (50 μg/ml, Sigma) 15 mm glass coverslips at a density of 75,000 per coverslip in 12 well plates. After 2 h, 1 ml of NB supplemented with 2% B27, 1% penicillin/streptomycin, and 1% glutamax was added to the coverslips. Neuronal activity was suppressed by treating the neuronal cultures with TTX for 24 h (2 μm, Sigma).
For immunohistochemistry, two [guanylate kinase domain-associated protein (GKAP)] or three (bassoon) independent cultures of 14 d in vitro (DIV14) neurons were fixed for 10 min with 4% formaldehyde/4% sucrose in PBS at room temperature (Jaworski et al., 2009). After fixation, the cells were washed two times in PBS for 30 min at room temperature, and were incubated with primary antibodies in GDB buffer (0.2% BSA, 0.8 m NaCl, 0.5% Triton X-100, 30 mm phosphate buffer, pH 7.4) overnight at 4°C. The following primary antibodies were used: rabbit anti-bassoon (Synaptic Systems), rabbit anti-GKAP (Santa Cruz Biotechnology), and mouse anti-aCaMKII (Sigma). Phalloidin (Invitrogen) was used to stain filamentous actin (F-actin). Neurons were then washed three times in PBS for 30 min at room temperature and incubated with Alexa488- and Alexa568-conjugated secondary antibodies (Invitrogen) in GDB for 2 h at room temperature, and washed three times in PBS for 30 min. Slides were mounted using Vectashield mounting medium (Vector Laboratories). Confocal images were acquired using a LSM510 confocal microscope (Zeiss) with a 40× or 63× oil objective.
For quantification, confocal images of four to five neurons obtained from two to three independent experiments were obtained, with sequential acquisition settings at the maximal resolution of the microscope (1024 × 1024 pixels). Each image was a z-series of six to eight images each averaged two times was chosen to cover the entire region of interest from top to bottom. The resulting z-stack was “flattened” into a single image using maximum projection. Images were not further processed and were of similar high quality to the original single planes. The confocal settings were kept the same for all scans when fluorescence intensity was compared. Morphometric analysis, quantification, and colocalization were performed using MetaMorph software (Universal Imaging Corporation). The ratio of αCaMKII in the synapse over αCaMKII in the shaft was calculated by measuring the average intensity of the fluorescent αCaMKII signal in the synapses and the shaft.
Fear conditioning was performed in a conditioning chamber (Medical Associates) equipped with a grid floor via which the footshock could be administered. Each mouse was placed inside the conditioning chamber for 180 s. A footshock (2 s, 0.4 mA) was delivered 148 s after placement in the chamber. Twenty-four hours later, context-dependent freezing was measured during 3 min.
After the animals had been killed, sagittal slices (400 μm) were obtained and submerged in ice-cold artificial CSF (ACSF) using a vibratome, and hippocampi were dissected out. These sagittal hippocampal slices were maintained at room temperature for at least 1.5 h to recover before experiments were initiated. Then they were placed in a submerged recording chamber and perfused continuously at a rate of 2 ml/min with ACSF equilibrated with 95% O2, 5% CO2 at 31°C. ACSF contained the following (in mm): 120 NaCl, 3.5 KCl, 2.5 CaCl2, 1.3 MgSO4, 1.25 NaH2PO4, 26 NaHCO3, and 10 d-glucose. Extracellular recordings of field EPSPs (fEPSPs) were made in CA1 stratum radiatum with platinum (Pt)/iridium (Ir) electrodes (Frederick Haer Company). A bipolar Pt/Ir was used to stimulate Schaffer collateral/commissural afferents with a stimulus duration of 100 μs. Stimulus–response curves were obtained at the beginning of each experiment, 20 min after placing the electrodes. Long-term potentiation (LTP) was evoked using the following two different tetani: (1) 100 Hz (1 train of 1 s at 100 Hz); and (2) 200 Hz (4 trains of 0.5 s, spaced by 5 s). Both protocols were performed at one-third of the maximum fEPSP. fEPSP measurements were done once per minute. Potentiation was measured as the normalized increase of the mean fEPSP slope for the duration of the baseline. Only stable recordings were included, and this judgment was made blind to genotype. Average LTP was defined as the mean last 10 min of the normalized fEPSP slope.
All data are presented as means ± SEM and were tested for normality of distribution using the D'Agostino–Pearson test. If normality of distribution was violated or sample size was too small to determine normality, we used the Mann–Whitney U test. In all other cases, an appropriate t test was used to analyze differences between genotypes for spine density, freezing time, and LTP induction (based on the average of the last 10 min). A repeated-measures ANOVA was used to analyze differences between genotypes for fiber volley and fEPSP slope, even when in some cases the distribution was not normal because there is no nonparametric alternative for the repeated-measures ANOVA.
Camk2b−/− mice show normal hippocampal morphology and no change in αCaMKII protein levels and autophosphorylation
Generation of the Camk2b−/− mouse has been described previously (van Woerden et al., 2009). Using immunohistochemistry and Western blot, we confirmed the absence of βCaMKII in the hippocampus of the Camk2b−/− mouse (Fig. 1a,b). Since in vitro experiments showed that upregulation of βCaMKII causes downregulation of αCaMKII (Thiagarajan et al., 2002), we tested whether the absence of βCaMKII caused upregulation of αCaMKII in vivo. However, we did not observe a change in αCaMKII protein levels (wild-type mice: 100 ± 5.4, n = 6; Camk2b−/− mice: 91.5 ± 6.3, n = 7; Mann–Whitney U test, U = 13.00, p = 0.29) (Fig. 1b), nor was there a significant change in basal levels of αCaMKII Thr286 phosphorylation (wild-type mice:100 ± 17.1 n = 6; Camk2b−/− mice: 92.7 ± 13.9 n = 7; Mann–Whitney U test, U = 20.00, p = 0.95) (Fig. 1b). These data show that in Camk2b−/− mice βCaMKII is absent and that protein expression and basal levels of autophosphorylation of αCaMKII are unaltered.
Previous studies showed that upregulation or downregulation of βCaMKII in vitro caused respectively an increased or decreased dendritic arborization, suggesting that βCaMKII might be critical for normal dendritic development in vivo (Fink et al., 2003). Therefore, we performed a detailed examination of the hippocampus using thionin staining. However, we found no evidence of significant changes in hippocampal structure at the light microscopy level (Fig. 1c). Furthermore, we investigated the morphology of the CA1 pyramidal cells using Golgi–Cox staining. We found no significant change in the density of spines (wild-type mice: 6.35 ± 0.36 n = 15 cells from 3 mice; Camk2b−/− mice: 7.07 ± 0.61 n = 15 from 3 mice; unpaired two-tailed t test, t(28) = 1.8, p = 0.09) (Fig. 1d), which is consistent with our previous findings for the cerebellum (van Woerden et al., 2009). Together, these data show that gross neuronal development is preserved in Camk2b−/− mice.
αCaMKII is mislocalized in Camk2b−/− neurons
Since βCaMKII is able to bind F-actin in an activity-controlled manner (Shen et al., 1998; Shen and Meyer, 1999) and pharmacologically induced changes in actin bundling have a large effect on CaMKII delivery in spines (Allison et al., 2000; Okamoto et al., 2004), it is possible that βCaMKII can change actin dynamics and the localization of the CaMKII holoenzyme. Therefore, we hypothesized that absence of βCaMKII might result in abnormalities in synaptic localization of αCaMKII. However, using brain sections for quantitative analysis of changes in localization of αCaMKII is challenging, because of its ubiquitous distribution, and its ability to self-aggregate and redistribute depending on multiple factors including fixation conditions (Tao-Cheng et al., 2002). Hence, we investigated the localization of endogenous αCaMKII by immunostaining of neuronal cultures obtained from E18 hippocampal neurons from Camk2b−/− and wild-type mice. Neurons were fixed at DIV14 and stained with an antibody against bassoon (a marker for the presynaptic active zone), GKAP (a marker for the postsynaptic density), phalloidin (which stains F-actin), and an antibody against αCaMKII (Fig. 2). Whereas αCaMKII showed a striking synaptic labeling in wild-type neurons, this was clearly less pronounced in the Camk2b−/−-derived neurons. To quantify this, we measured the ratio of αCaMKII in the synapse (colocalizing with bassoon or GKAP) to αCaMKII in the dendritic shaft, and found that this ratio was reduced by >40% in the Camk2b−/− neurons when measured over bassoon (αCaMKIIsynapse/αCaMKIIshaft ratio: wild-type neurons, 3.3 ± 0.2, n = 15; Camk2b−/− neurons, 2.1 ± 0.2, n = 14; Mann–Whitney U test, U = 17.00, p = 0.0002) (Fig. 2a), and by 60% when measured over GKAP (αCaMKIIspine/αCaMKIIshaft ratio: wild-type neurons, 3.42 ± 0.16, n = 10; Camk2b−/− neurons, 1.50 ± 0.14 n = 10; Mann–Whitney U test, U = 8.00, p = 0.001) (Fig. 2b). This distinct difference was still observed after decreasing or increasing neuronal activity of Camk2b−/− neuronal cultures with tetrodotoxin or bicuculline for 24 h, as neither treatment changed the αCaMKII localization significantly (αCaMKIIsynapse/αCaMKIIshaft ratio: control medium, 1.66 ± 0.19, n = 5; tetrodotoxin, 1.74 ± 0.18, n = 5; bicuculline, 1.80 ± 0.13, n = 6; ANOVA F(2,13) = 0.05; p = 0.96). This indicates that the changes were not due to differences in spontaneous activity between the cultures. In addition, even though the cultures were derived from E18 neurons, the ratio of αCaMKII/βCaMKII in wild-type DIV14 cultures was comparable to the ratio observed in the adult mouse hippocampus (Fig. 2d). Together, these findings highlight the importance of βCaMKII in regulating the distribution of endogenous αCaMKII between the synaptic pool and the adjacent dendritic shaft.
Camk2b−/− mice show impaired LTP
Given that overall neuronal morphology was unaffected in Camk2b−/− mutants with an impaired localization of αCaMKII into dendritic spines, we examined the functional implications by investigating synaptic plasticity at the hippocampal CA1 synapse. Using extracellular recordings in acute hippocampal slices, we focused on the Schaffer–collateral pathway given the large literature implicating this synapse in many forms of hippocampus-dependent learning and memory. No significant impairment in basal synaptic transmission was observed in the Camk2b−/− mice (Fig. 3a), with significant changes in neither fiber volley (repeated-measures ANOVA, F(1,135) = 0.22; p = 0.64) nor fEPSP slope (repeated-measures ANOVA, F(1,135) = 1.4; p = 0.23). However, we found a significant deficit in LTP in Camk2b−/− mutants compared with wild-type littermates (wild-type mice, 152.6 ± 7.7, n = 27; Camk2b−/− mice, 119.3 ± 4.4, n = 21; Mann–Whitney U test, U = 130.00, p = 0.0015) (Fig. 3b), indicating that βCaMKII plays an essential role in synaptic plasticity at the hippocampal Schaffer–collateral pathway. Notably, the LTP deficit in the Camk2b−/− mice induced by a 100 Hz/1 s tetanus, is as severe as after the loss of the far more abundant αCaMKII (Elgersma et al., 2002). Moreover, it should be noted that hippocampal LTP is unaffected in the heterozygous Camk2a mutant (Frankland et al., 2001; Elgersma et al., 2002), which shows a larger decrease in total CaMKII level compared with the homozygous Camk2b mutant. This indicates that the LTP deficit of the Camk2b−/− mice cannot only be explained in terms of loss of CaMKII activity.
It has previously been shown that changes in the F-actin/G-actin equilibrium affect αCaMKII localization (Allison et al., 2000; Okamoto et al., 2004) and that stimulation of NMDA receptors affects CaMKII localization as well as the F-actin/G-actin equilibrium (Shen and Meyer, 1999; Okamoto et al., 2004). Together, these findings support a model in which CA1 LTP is highly dependent upon proper targeting of αCaMKII into spines, for which βCaMKII is required. If the LTP deficit is due to mislocalization of αCaMKII, it is conceivable that the LTP deficit can be rescued by using a stronger LTP-inducing protocol (4 trains of 200 Hz for 0.5 s, spaced 5 s apart) (Grover and Teyler, 1990). Indeed, using this stimulation protocol, LTP is normal in Camk2b−/− mice, confirming the functional implications of the altered CaMKII localization (wild-type mice, 131.6 ± 4.4, n = 14; Camk2b−/− mice, 129.7 ± 5.7, n = 12; unpaired two-tailed t test, t(24) = 0.36, p = 0.72) (Fig. 3c). To confirm that this 200 Hz LTP protocol requires αCaMKII, we also performed this experiment in Camk2a−/− mice. Indeed, Camk2a−/− mice show a significant impairment (wild-type mice, 179.6 ± 13.3, n = 12; Camk2a−/−, 133.3 ± 4.8, n = 18; Mann–Whitney U test, U = 36.00, p = 0.0025) (Fig. 3d), indicating that this LTP protocol is indeed dependent on CaMKII activity. Together, these results show that the LTP deficit in the Camk2b−/− mice can be overcome by a strong LTP-inducing protocol. These findings are consistent with, but not a proof of, the hypothesis that the deficit in hippocampal synaptic plasticity is a result of a failure to properly target αCaMKII into dendritic spines of the Camk2b−/− mouse.
Camk2b−/− mice show impaired hippocampus-dependent learning
To test whether the mislocalization of CaMKII in the Camk2b−/− mouse also affected hippocampal learning, we made use of contextual fear conditioning. In this task, mice are conditioned to associate a certain context with a mild, aversive footshock. Learning is assessed by measuring freezing behavior (i.e., the cessation of all movement except respiration), which is a natural expression of fear in mice. Camk2b−/− mutants did not differ from their wild-type littermates in preshock baseline freezing behavior (wild-type mice, 0.9 ± 0.2%, n = 11; Camk2b−/− mice, 0.7 ± 0.2%, n = 9; two-tailed t test, t(18) = 0.72, p = 0.48) (Fig. 3e), but showed significantly less freezing in the 24 h long-term memory test, demonstrating an impairment of hippocampus-dependent memory (wild-type mice, 44.8 ± 4.6%, n = 11; Camk2b−/− mice, 27.8 ± 2.7%, n = 9; two-tailed t test, t(18) = 3.0, p = 0.0079) (Fig. 3e).
Generation and characterization of the Camk2bA303R mouse
The results above suggest that βCaMKII strongly influences hippocampal plasticity by regulating αCaMKII localization into dendritic spines. However, the affinity of βCaMKII for calcium/calmodulin is nearly 10-fold higher than αCaMKII, and the sensitivity range of the heteromeric holoenzyme is dependent on the ratio of α to β subunits (De Koninck and Schulman, 1998; Brocke et al., 1999). Hence, we cannot rule out that the deficits in CA1 LTP and hippocampus-dependent learning are caused by the loss of the enzymatic activity of βCaMKII, rather than the abnormal localization of αCaMKII. Therefore, we sought to distinguish between these possibilities using a well described mutation of βCaMKII (A303R), which prevents kinase activation (by interfering with calcium/calmodulin binding) while preserving F-actin binding and bundling (Shen and Meyer, 1999; Fink et al., 2003; O'Leary et al., 2006). In addition, the βCaMKII-A303R protein does not show a dominant-negative effect on dendritic arborization in vitro, in contrast to cells expressing the catalytically dead βCaMKII-K42R protein, which cannot bind ATP (Fink et al., 2003). Accordingly, this mutation elegantly permits us to dissect the requirement of βCaMKII kinase activity and F-actin bundling on LTP and learning.
We created a knock-in mutant of the Camk2b gene, substituting alanine303 for arginine (A303R) (Fig. 4a,b). Immunostaining and Western blot analysis of brains of homozygous point mutants (designated as Camk2bA303R mice) revealed no change in expression of βCaMKII and αCaMKII in the Camk2bA303R mouse (Fig. 4c,d). Since binding of calcium/calmodulin is a prerequisite for autophosphorylation of CaMKII at Thr286/287 (for review, see Hudmon and Schulman, 2002b; Lisman et al., 2002; Colbran, 2004), we used the phospho-Thr286/287 antibody to confirm that this mutation renders the βCaMKII-A303R protein insensitive to calcium/calmodulin activation. Indeed, Western blot analysis showed that βCaMKII Thr287 phosphorylation was entirely absent in the Camk2bA303R mouse (Fig. 4d), confirming in vitro studies that the A303R mutation blocks activation of βCaMKII. We also observed a significant reduction of αCaMKII Thr286 autophosphorylation (wild-type mice, 100 ± 7.2, n = 7; Camk2bA303R mice, 69.9 ± 8.3, n = 5; Mann–Whitney U test, U = 4.0, p = 0.03) (Fig. 4d), which is not unexpected given that most of the αCaMKII subunits are associated with βCaMKII subunits, and that it takes two adjacent activated CaMKII subunits to get inter-subunit autophosphorylation at Thr286/Thr287 (Hudmon and Schulman, 2002b; Lisman et al., 2002; Colbran, 2004). Thionin staining did not reveal any gross morphological changes in the brain (Fig. 4e), indicating that development of the brain is normal despite the presence of an inactive form of βCaMKII.
αCaMKII shows normal subcellular distribution in Camk2bA303R neurons
To directly examine the influence of βCaMKII on the localization of αCaMKII, we tested whether synaptic targeting of αCaMKII was altered in neurons of the Camk2bA303R mice. Remarkably, and in strong contrast to the findings in Camk2b−/− neurons, Camk2bA303R neurons showed normal αCaMKII synaptic localization (αCaMKIIsynapse/αCaMKIIshaft: wild-type neurons, 4.2 ± 0.7, n = 6; Camk2bA303R neurons, 3.9 ± 0.6, n = 7; Mann–Whitney U test, U = 14.00, p = 0.37) (Fig. 5). These results demonstrate that βCaMKII protein but not its calcium/calmodulin-dependent activation is required for targeting αCaMKII to synapses.
The Camk2bA303R mice show normal LTP
Camk2bA303R mice retain normal synaptic localization of αCaMKII, despite a complete abrogation of calcium/calmodulin-dependent kinase activity of βCaMKII. This provides a unique opportunity to dissect the mechanism by which βCaMKII influences synaptic plasticity and learning.
As observed in Camk2b−/− mice (Fig. 3a), extracellular recordings in acute hippocampal slices of Camk2bA303R mice showed a slight, but not significant, reduction in basal synaptic transmission (basal synaptic transmission: repeated-measures ANOVA fiber volley, F(1,46) = 0.44, p = 0.51; fEPSP slope, F(1,46) = 0.93; p = 0.34; n = 25 and 23, respectively, for wild-type and Camk2bA303R mice) (Fig. 6a). However, in contrast to the Camk2b−/− mice, which have a severe deficit in 100 Hz LTP, Camk2bA303R mice showed normal 100 Hz LTP (wild-type mice, 149.7 ± 9.2, n = 15; Camk2bA303R mice, 133.4 ± 7.5, n = 10; two-tailed t test, t(23) = 1.40, p = 0.17) (Fig. 6b). These results indicate that the kinase activity of βCaMKII is dispensable for hippocampal synaptic plasticity and suggest that βCaMKII functions principally to regulate the targeting of αCaMKII into dendritic spines.
Contextual fear learning of the Camk2bA303R mice
Given that Camk2b−/− mice showed a significant deficit in contextual fear conditioning (Fig. 3e), the results from Camk2bA303R mice provide an opportunity to determine the mechanism of βCaMKII function in hippocampus-dependent learning. Camk2bA303R mice showed indistinguishable freezing behavior from their wild-type littermates (wild-type mice, 62.6 ± 4.1, n = 13; Camk2bA303R mice, 53.5 ± 5.0 n = 11; two-tailed t test, t(22) = 1.42, p = 0.17) (Fig. 6c) without any change in baseline freezing behavior (wild-type mice, 3.5 ± 0.8, n = 13; Camk2bA303R mice, 4.6 ± 1.6, n = 11; Mann–Whitney U test, U = 65.00, p = 0.73) (Fig. 6c). Hence, together with the results obtained using Camk2b−/− mice, these results suggest that βCaMKII-dependent localization of αCaMKII is required for normal hippocampus-dependent learning.
Through a genetic dissection in mice, we have identified that βCaMKII functions at the Schaffer collateral–CA1 synapse principally to target αCaMKII into dendritic spines. We found that the complete loss of βCaMKII protein leads to mislocalization of αCaMKII in cultured neurons. Furthermore, we found that loss of βCaMKII severely impairs hippocampal learning and synaptic plasticity. In contrast, the βCaMKII-A303R mutant provided an elegant dissection of βCaMKII function, since it fully retains F-actin binding despite a loss of calcium/calmodulin-dependent kinase activity. Indeed, Camk2bA303R mice showed proper αCaMKII localization, robust LTP, and normal learning. Since we found that the subcellular localization of αCaMKII is changed only in the absence of βCaMKII, and that the LTP deficit in the absence of βCaMKII can be overcome by using a strong LTP-inducing protocol, our data support a role of βCaMKII, in which it regulates the targeting of αCaMKII into spines. Given the extensive literature showing that βCaMKII has F-actin binding and bundling properties, which are fully conserved in the βCaMKII-A303R mutant, we have concluded that the synaptic plasticity and learning deficits observed in the Camk2b−/− mutants result from the impaired βCaMKII-dependent targeting of αCaMKII into spines.
βCaMKII is not required for neuronal development
Considering the developmentally earlier expression of βCaMKII compared with αCaMKII (Sahyoun et al., 1985; Bayer et al., 1999) and the observation that knockdown or overexpression of βCaMKII affects synapse number and neurite outgrowth in cultured hippocampal neurons or neuroblastoma cells (Nomura et al., 1997; Fink et al., 2003), a large effect on hippocampal development could be expected. However, we found that neither the absence nor the inactivation of βCaMKII affects brain development at the light microscopic level. Hence, we conclude that βCaMKII is not required for normal brain development and for normal spine number of CA1 pyramidal neurons. This is similar to previous findings in the cerebellum (van Woerden et al., 2009). However, although our data show that βCaMKII is dispensable for normal brain development, we cannot rule out that germ line mutations in βCaMKII result in homeostatic compensatory mechanisms to prevent the changes seen in neuronal cultures shortly after the expression of βCaMKII was changed.
βCaMKII-dependent hippocampal synaptic plasticity
We showed that βCaMKII is required for normal hippocampal NMDA receptor-dependent plasticity as well as learning. Notably, despite the fact that βCaMKII has a 10-fold higher affinity for calcium/calmodulin compared with αCaMKII, the impairment in NMDA receptor-dependent LTP in Camk2b−/− mice cannot be explained by a reduction in calcium/calmodulin sensitivity, since LTP deficits were not observed in the Camk2bA303R mice, in which βCaMKII has a negligible affinity for calcium/calmodulin. Rather, the deficit in NMDA-dependent LTP is most likely caused by the reduced synaptic localization of αCaMKII, as our data show that the subcellular localization of αCaMKII is strongly influenced by βCaMKII. Specifically, there is approximately a 50% reduction in the αCaMKII spine/shaft ratio in Camk2b−/− neurons, whereas the localization of αCaMKII in Camk2bA303R neurons is indistinguishable from wild-type mice. Hence, βCaMKII functions independently of its calcium-dependent kinase activity to regulate the synaptic localization of αCaMKII.
We realize that our study has some limitations. First, to be able to quantify the subcellular localization of αCaMKII we had to revert to (ex vivo) dissociated neuronal cultures obtained from the mutant mice, which is the commonly used technique to study CaMKII localization. Whether the observed changes also hold true in vivo remains to be shown. In addition, our localization studies cannot distinguish between presynaptic and postsynaptic αCaMKII. However, given the relative abundance of postsynaptic αCaMKII, and the important role of postsynaptic αCaMKII in LTP, we believe that our findings strongly suggest that βCaMKII is particularly important for targeting postsynaptic CaMKII. Last, although the observed reduction of synaptic αCaMKII is likely underlying the NMDA-receptor-dependent LTP deficit, we cannot rule out that the loss of βCaMKII also affects other synaptic processes associated with the βCaMKII-dependent actin binding and bundling properties. But regardless of the precise nature of the deficits, it is interesting to note that the plasticity deficits can be overcome by a very strong LTP-inducing protocol consisting of four trains of 200 Hz (Grover and Teyler, 1990). Possibly, this protocol is able to activate the pool of αCaMKII, which is present in the shaft. In this respect, it is interesting to note that a recent study showed that voltage-gated calcium channels (VGCCs) are able to activate a different pool of αCaMKII than NMDA receptors, with a stronger activation of αCaMKII in dendritic shafts compared with αCaMKII in spines (Lee et al., 2009). However, whether activation of VGCCs is responsible for the LTP rescue remains to be investigated.
Together, our data show that βCaMKII is essential for hippocampus-dependent learning and for normal plasticity at the Schaffer collateral–CA1 synapse. Despite the fact that βCaMKII has a higher affinity for calcium/calmodulin compared with αCaMKII, we found that the calcium/calmodulin-dependent activation of βCaMKII was fully dispensable for hippocampal LTP and learning. Rather, our data show that βCaMKII in hippocampal pyramidal neurons plays a structural role, which serves to target αCaMKII to synapses.
This work was supported by grants from NWO-ZonMW and Neuro-Bsik to Y.E. We thank members of the Elgersma laboratory, Steven Kushner, and Roger Colbran for stimulating discussions and/or critically reading the manuscript. We thank Dmitri Andreev, Mehrnoush Agha Davoud Jolfaei, John Kongasan, Minetta Elgersma, and Hans van der Burg for technical support.
- Correspondence should be addressed to Ype Elgersma, Department of Neuroscience, Erasmus University Medical Centre, Drive Molewaterplein 50, 3015 GE, Rotterdam, The Netherlands.