Ablation of Ca2+ Channel β3 Subunit Leads to Enhanced N-Methyl-d-aspartate Receptor-dependent Long Term Potentiation and Improved Long Term Memory*

The β subunits of voltage-dependent Ca2+ channels (VDCCs) have marked effects on the properties of the pore-forming α1 subunits of VDCCs, including surface expression of channel complexes and modification of voltage-dependent kinetics. Among the four different β subunits, the β3 subunit (Cavβ3) is abundantly expressed in the hippocampus. However, the role of Cavβ3 in hippocampal physiology and function in vivo has never been examined. Here, we investigated Cavβ3-deficient mice for hippocampus-dependent learning and memory and synaptic plasticity at hippocampal CA3-CA1 synapses. Interestingly, the mutant mice exhibited enhanced performance in several hippocampus-dependent learning and memory tasks. However, electrophysiological studies revealed no alteration in the Ca2+ current density, the frequency and amplitude of miniature excitatory postsynaptic currents, and the basal synaptic transmission in the mutant hippocampus. On the other hand, however, N-methyl-d-aspartate receptor (NMDAR)-mediated synaptic currents and NMDAR-dependent long term potentiation were significantly increased in the mutant. Protein blot analysis showed a slight increase in the level of NMDAR-2B in the mutant hippocampus. Our results suggest a possibility that, unrelated to VDCCs regulation, Cavβ3 negatively regulates the NMDAR activity in the hippocampus and thus activity-dependent synaptic plasticity and cognitive behaviors in the mouse.

with Ca v ␤3 in the hippocampus (18 -20). In addition, N-and L-type VDCCs have been strongly implicated in activity-dependent long lasting synaptic changes, such as LTP, as well as in learning and memory (21,22). Therefore, we examined the Ca v ␤3-deficient mice (23) for hippocampus-dependent learning and memory and synaptic plasticity. Interestingly, long term memory and NMDAR-dependent LTP were increased in the Ca v ␤3-deficient mice, whereas there was no significant change in Ca 2ϩ currents. Furthermore, the mutant mice showed increased NMDAR-mediated synaptic responses and an increased NR2B level in the hippocampus. These results reveal Ca 2ϩ channel-independent functions of Ca v ␤3 in the hippocampus.

EXPERIMENTAL PROCEDURES
Animals-The generation of mice lacking Ca v ␤3 was described in our previous study (23). Ca v ␤3 heterozygous (Ca v ␤3 ϩ/Ϫ ) mice were backcrossed into two inbred backgrounds, C57BL/6J and 129S4/SvJae, each over 18 generations. Ca v ␤3 wild-type (Ca v ␤3 ϩ/ϩ ) and Ca v ␤3-deficient (Ca v ␤3 Ϫ/Ϫ ) mice used for analysis were obtained from interbreeding Ca v ␤3 ϩ/Ϫ mice of the two backgrounds. Animal care and handling were carried out according to the institutional guidelines. The mice were maintained with free access to food and water under a 12:12-h light/dark cycle. Behavioral experiments were performed on 8 -12-week-old mice. All experiments were performed in a blind manner with respect to the genotype.
Contextual and Cued Fear Conditioning-The fear conditioning was carried out as described in our previous study (24). A fear-conditioning shock chamber (19 ϫ 20 ϫ 33 cm) containing a stainless steel rod floor (5 mm diameter, spaced 1 cm apart) and a monitor was used (WinLinc behavioral experimental control software, Coulbourn Instruments). For conditioning, mice were placed in the fear-conditioning apparatus chamber for 2 min, and then a 28-s acoustic conditioned stimulus (CS) was delivered. Following the CS, a 0.5-mA shock of unconditioned stimulus was immediately applied to the floor grid for 2 s. This protocol was performed twice at 60-s interval. To assess contextual learning, the animals were placed back into the training context 24 h after training, and then freezing behavior was observed for 4 min. To assess cued learning, the animals were placed in a different context (a novel chamber, odor, floor, and visual cues) 24 h after training, and their behaviors were monitored for 5 min. During the last 3 min of this test, animals were exposed to the tone. Fear response was quantified by measuring the length of the time when the animal showed freezing behaviors, which was defined as lack of movements with a crouching position, except for respiratory movements (25). Foot-shock intensity was evaluated by placing naive animals in the conditioning chamber used for fear conditioning. Animals were subjected to a 1-s series of gradually increasing mild foot-shock amperage at 20-s intervals as follows: 0.1, 0.15, 0.2, 0.25, 0.3, 0.4, 0.5, and 0.6 mA. The shock intensity that evoked initial sensation responses (flinching and running), vocalization, and jumping was recorded for each mouse.
Novel Object Recognition Memory Task-The task was performed as described (24,26,27). The mice were individually habituated to an open-field box (40 ϫ 40 ϫ 40 cm) for 3 days.
During the training trial, two objects were placed in the box, and animals were allowed to explore them for 5 min. A mouse was considered to be exploring the object when its head was facing the object within 1-inch distance. Following retention intervals (1 or 24 h), animals were placed back into the box with two objects in the same locations, but one of the familiar objects was replaced by a novel object, and mice were then allowed to explore the two objects for 5 min. The preference percentage, percentage of the time spent exploring the novel object over the total time spent exploring both objects, was used to quantitate the recognition memory.
Social Transmission of Food Preference Task-This task was performed as described previously (21,28,29), with slight modifications. "Demonstrator" mice were given a distinctively scented food (cinnamon or cocoa) for 2 h and then immediately allowed to interact with "observer" mice for 30 min. Either 1 or 24 h later, observers were given a choice between two scented foods: either the same scented food that the demonstrators had eaten (cued) or another distinctively scented food (non-cued). Half of the observers in each genotype was subjected to interaction with the demonstrators that had eaten cinnamon as cued food and the other half with those that had eaten cocoa as cued food to control for the possibility of food preference bias.
Whole-cell Patch Clamp Recording on Acutely Isolated CA1 Pyramidal Neurons and on Hippocampal Slices-All experiments were performed on 2-3-week-old mice. Preparation of and recording from hippocampal slices (400 m thick) were as described in our previous study (21,30). Hippocampal slices were prepared in oxygenated, cold ACSF (124 mM NaCl, 3.5 mM KCl, 1.25 mM NaH 2 PO 4 , 2.5 mM CaCl 2 , 1.3 mM MgSO 4 , 26 mM NaHCO 3 , and 10 mM glucose, pH 7.4). For the measurement of Ca 2ϩ currents, acutely isolated CA1 pyramidal neurons were prepared from hippocampal slices, as described in our previous study (30). The recorded CA1 neurons were voltage-clamped at Ϫ60 mV using glass pipette electrodes (3-5 M⍀ series resistance Ͻ20 M⍀) and the I-V curve was generated in a stepwise fashion: ϩ10-mV increments from Ϫ60 to ϩ40 mV. Internal pipette solution contained the following, 130 mM CsCl, 10 mM EGTA, 10 mM HEPES, 4 mM MgCl 2 , 4 mM MgATP, 0.3 mM Tris-GTP, 5 mM tetraethylammonium chloride, and was brought to pH 7.4 with NaOH. Extracellular solution contained the following, 25 mM tetraethylammonium chloride, 5 mM 4-aminopyridine, 20 mM HEPES, 3 mM KCl, 5 mM CaCl 2 , 2 mM MgCl 2 , 100 mM NaCl, 0.001 mM tetrodotoxin, and was brought to pH 7.4 with NaOH. For the measurement of after hyperpolarization (AHP) currents, visually guided CA1 pyramidal neurons in hippocampal slice were held at Ϫ55 mV, and currents were evoked by depolarizing voltage commands to 20 mV for 200 ms followed by a return to Ϫ55 mV for 10 s. During recording, the slices were superfused with ACSF at room temperature. Glass pipettes (3-5 M⍀) were filled with solution containing 140 mM KMeSO 4 , 8 mM NaCl, 1 mM MgCl 2 , 10 mM HEPES, 2 mM Mg-ATP, 0.4 mM Na 2 -GTP, and 0.02 mM EGTA (pH 7.3, 290 mosM). In addition, action potentials (APs) were triggered under current clamp mode by depolarizing current injection (from ϩ30 to ϩ90 pA), and the number of AP (from threshold to the peak) and AP durations (width at half-height) were measured. The internal solution for mEPSC (miniature excitatory postsynaptic currents) recording was filled with the following buffer, 135 mM potassium gluconate, 5 mM KCl, 2 mM MgCl 2 , 5 mM EGTA, 10 mM HEPES, 0.5 mM CaCl 2 , 5 mM Mg-ATP and 0.3 mM Na-GTP, and was brought to pH 7.4 with KOH. The experiment was performed in the presence of tetrodotoxin (1 M) and bicuculline (10 M, a GABA type a receptor antagonist). The recorded CA1 pyramidal neurons were voltageclamped at Ϫ70 mV. The frequency and amplitude of mEPSCs were analyzed with MiniAnalysis (Synaptosoft) (21). For the measurement of AMPAR-and NMDAR-mediated synaptic currents in visually guided CA1 pyramidal neurons, pipettes (3-5 M⍀) were filled with the internal solution (130 mM cesium gluconate, 5 mM KCl, 0.1 mM CaCl 2 , 2.0 mM MgCl 2 , 5 mM EGTA, 10 mM HEPES, 10 mM QX-314, 4 mM Na-ATP, and 0.4 mM Na-GTP, brought to pH 7.3 with CsOH). The currents were measured in the presence of bicuculline (10 M) and CGP 55845 (5 M, a GABA type B receptor antagonist). The synaptic currents were evoked by a bipolar tungsten electrode that was placed in the stratum radiatum. NMDAR-and AMPAR-mediated responses were discriminated based on their distinct kinetics and voltage dependence; the NMDAR-mediated currents were measured at ϩ40 mV, 100 ms after the response onset, whereas the AMPAR-mediated currents were taken as the peak amplitude response recorded at Ϫ70 mV (31). D-AP5 (50 M) blocked the late component of the currents recorded at ϩ40 mV, whereas CNQX (10 M), an AMPA receptor blocker, eliminated the currents recorded at Ϫ70 mV. Whole-cell patch clamp currents were recorded and digitized with a MultiClamp 700A amplifier and a Digidata 1320 or 1322A (Axon Instruments, CA), and acquired data were analyzed with the pCLAMP version 9.2 (Axon Instruments) and the Mini Analysis Program (Synaptosoft).
Extracellular Recording on Hippocampal Slices-Preparation of hippocampal slices and the method of field excitatory postsynaptic potentials (fEPSPs) recording have been described previously (21,24). Hippocampal slices (400 m) were prepared from 7-8-week-old mice, as described above. Slices were then placed in a warm, humidified (32°C, 95% O 2 , 5% CO 2 ) recording chamber containing oxygenated ACSF and maintained for 1.5 h prior to experiments. A bipolar stimulating electrode was placed in the stratum radiatum in the CA1 region, and extracellular field potentials were also recorded in the stratum radiatum using a glass microelectrode (borosilicate glass, 3-5 M⍀, filled with 3 M NaCl). Test responses were elicited at 0.033 Hz. Base-line stimulation was delivered at an intensity that evoked a response that was ϳ40% of the maximum evoked response. LTP was induced electrically by one of the following protocols: 1) 100-Hz LTP was induced for 100 ms, 300 ms, or 1 s; 2) 200-Hz LTP was induced by 10 trains of 200-ms stimulation at 200 Hz delivered every 5 s. LTD was elicited by paired-pulses low frequency stimulation (PP-LFS) (50-ms pulse interval at 1 Hz for 15 min). Drugs were added to the perfusion medium at least 30 min before recording.
Immunohistology and Western Blot-Immunostaining was performed as described previously (32,33). Animals were anesthetized and perfused through the heart with 50 ml of cold saline and 50 ml of 4% paraformaldehyde in 0.1 M phosphate buffer. Brains were then removed and were post-fixed over-night. Coronal sections containing hippocampus were stained with the following primary antibodies: anti-␤ 3 subunit (anti-Ca v ␤3, Alomone Labs), anti-SMI-32, and anti-GAD. A biotinylated secondary antibody and the avidin/biotin system were used for each antibody followed by a 3,3Ј-diaminobenzidene reaction. Some of the DAB reactions incorporated a nickel intensification procedure. For gross morphology of the hippocampus, Nissl staining was used. For Western blot analysis, total hippocampal proteins were prepared as described previously (34). 25 g of protein were loaded per lane and analyzed by SDS-PAGE followed by Western blotting. The following antibodies have been described previously: NMDAR 2A/2B (35) and GluR1/2 (34). The following fusion protein was used for the generation of the following polyclonal antibody: H 6 -rat NMDAR1 (amino acids 340 -561; 1740 guinea pig). Antibody for ␣-tubulin was purchased from Sigma.
Statistical Analysis-All data are given as mean Ϯ S.E. Twoway repeated ANOVA, one-way ANOVA, and Student's t test were used for statistical analyses. p Ͻ 0.05 was considered statistically significant.

Normal Gross Morphology of the Hippocampus in the
Ca v ␤3 Ϫ/Ϫ Mice-We first examined the cytoarchitectonic divisions in the brain of the Ca v ␤3 Ϫ/Ϫ mice, especially in the hippocampus. The Ca v ␤3 Ϫ/Ϫ mice exhibited normal hippocampal divisions, including CA1, CA2, CA3, and dentate gyrus. No expression of Ca v ␤3 was observed in the Ca v ␤3 Ϫ/Ϫ hippocampus ( Fig. 1A), whereas Ca v ␤3 was abundant in the wild-type hippocampus as was shown previously (17). The immunoreactivities and the expression patterns of SMI-32 (a neurofilament protein) (Fig. 1B) and GAD (GABA-synthesizing enzyme) ( Fig.  1C) were normally observed in the hippocampus of the Ca v ␤3 Ϫ/Ϫ mice as in the Ca v ␤3 ϩ/ϩ mice. In addition, Nissl staining of the coronal brain sections revealed no gross abnormalities in the hippocampus of the Ca v ␤3 Ϫ/Ϫ mice (Fig. 1D).
Enhanced Contextual Fear Conditioning in the Ca v ␤3 Ϫ/Ϫ Mice-Because Ca v ␤3 is highly expressed in the hippocampus and is known to be associated with N-or L-type VDCCs, which play important roles in hippocampus-dependent learning and memory in animals (17)(18)(19), we examined whether the deletion of Ca v ␤3 affected the animal's capacity for hippocampus-dependent learning and memory. First, we subjected the mice to the fear conditioning assay that is known to require the function of the hippocampus (36). The Ca v ␤3 ϩ/ϩ (n ϭ 14) and Ca v ␤3 Ϫ/Ϫ (n ϭ 14) mice showed similar levels of freezing response during the training ( Fig. 2A). In the contextual fear memory assay performed 24 h after the training, the Ca v ␤3 Ϫ/Ϫ mice displayed more freezing behavior than the Ca v ␤3 ϩ/ϩ (F (1, 26) ϭ 8.36, p Ͻ 0.01, two-way repeated ANOVA), indicating an enhanced long term memory of the Ca v ␤3 Ϫ/Ϫ mice for contextual fear conditioning. A post hoc test (Scheffe's test) also revealed significant differences between the two genotypes during the 2nd (p Ͻ 0.05), the 3rd (p Ͻ 0.05), and the 4th min (p Ͻ 0.05) (Fig. 2B). On the other hand, no difference was observed between the two genotypes in the cued fear conditioning assay (Fig. 2C), indicating that the enhanced memory in the Ca v ␤3 Ϫ/Ϫ mice is limited to the hippocampus-dependent fear conditioning. There was no sig-nificant difference in response to variable electric intensities between Ca v ␤3 Ϫ/Ϫ (n ϭ 7) and Ca v ␤3 ϩ/ϩ (n ϭ 9) mice, indicating comparable reactivity or sensitivity to electric foot-shock of the two genotypes (Fig. 2D).
Enhanced Novel Object Recognition Memory in the Ca v ␤3 Ϫ/Ϫ Mice-We next subjected the mice to the novel object recognition task that is based on the animal's ability to discriminate a novel object from a familiar one, which requires the hippocampus (37). We first assessed the amount of time spent by the animals exploring the two objects during the training trial, and we found that both of the genotypes, Ca v ␤3 ϩ/ϩ (n ϭ 17) and Ca v ␤3 Ϫ/Ϫ mice (n ϭ 14), explored the two objects for equal time (Fig. 2E), which indicated no preference of the animals for either object. At a 1-h retention interval, when one of the familiar objects was replaced by a novel one, both Ca v ␤3 ϩ/ϩ (n ϭ 8) and Ca v ␤3 Ϫ/Ϫ mice (n ϭ 7) exhibited increased preference for the novel object to the familiar one (F (1, 13) ϭ 22.86, p Ͻ 0.001, two-way repeated ANOVA). No difference, however, was found between the two genotypes (F (1,13) 2F). At the 24-h retention test, however, Ca v ␤3 Ϫ/Ϫ mice (n ϭ 7) showed increased preference for the novel object compared with Ca v ␤3 ϩ/ϩ (n ϭ 9) (F (1,14) ϭ 36.14, p Ͻ 0.001, two-way repeated ANOVA, Scheffe's post hoc test, p Ͻ 0.01) (ϩ/ϩ, 62.68 Ϯ 6.26%; Ϫ/Ϫ, 88.90 Ϯ 3.23%) (Fig. 2F), indicating that the Ca v ␤3 Ϫ/Ϫ mice have an enhanced performance in the object recognition memory task.
Enhanced Long Term Memory in the Social Transmission of Food Preference Task in the Ca v ␤3 Ϫ/Ϫ Mice-Finally, we carried out the social transmission of food preference assay, another hippocampusdependent memory task. This task exploits the tendency of mice to prefer food that they have recently smelled on the breath of other mice (demonstrator mice), and subsequently, this tests their ability to learn and remember the information transmitted by olfactory cues during social interactions. 1 h after social interactions with demonstrator mice, both Ca v ␤3 ϩ/ϩ (n ϭ 7) and Ca v ␤3 Ϫ/Ϫ (n ϭ 6) mice preferred the "cued" food to the "non-cued" food, and there was no significant difference between the two genotypes (ϩ/ϩ, 83.70 Ϯ 3.63%; Ϫ/Ϫ, 75.87 Ϯ 7.34%, F (1, 11) ϭ 0.72, p ϭ 0.41, one-way ANOVA) (Fig. 2G). The amount of total food eaten was not different between geno- types during this task (Fig. 2H). These results indicate that the mice were not deficient in olfaction or social interactions.
Normal Intrinsic Firing Properties and AHP Currents in the Ca v ␤3 Ϫ/Ϫ -As a close coupling was reported by co-immunoprecipitation between Ca v ␤3 and N-or L-type VDCCs in hippocampal neurons (18 -20), we measured N-or L-type VDCCsmediated cellular properties in CA1 neurons. Ca 2ϩ influx Ca v ␤3 Ϫ/Ϫ (n ϭ 14). Solid line indicates the duration of CS (tone, 28 s), and the triangles indicate unconditioned stimulus (foot shock, 2 s). B, contextual fear conditioning 24 h after training. Ca v ␤3 Ϫ/Ϫ mice displayed more freezing behavior than the Ca v ␤3 ϩ/ϩ for contextual fear conditioning. *, p Ͻ 0.05, Scheffe's post hoc test. C, cued fear conditioning 24 h after training. CS (tone) presentation is indicated by the solid line. D, responses to variable foot shock intensities. F, flinching; R, running; V, vocalization; J, jumping. E and F, novel object recognition task. E, mean exploratory preference during training in Ca v ␤3 ϩ/ϩ (n ϭ 17) and Ca v ␤3 Ϫ/Ϫ mice (n ϭ 14). F, exploration to a novel object after each retention time. At 24 h retention, Ca v ␤3 Ϫ/Ϫ mice show increased preference for the novel object compared with Ca v ␤3 ϩ/ϩ . Dotted line indicates equal exploration of all objects. **, p Ͻ 0.01, Scheffe's post hoc test. G and H, social transmission of food preference task. G, at 1-h retention, both genotypes show a preference for the cued food, and there is no difference between genotypes (ϩ/ϩ, n ϭ 7; Ϫ/Ϫ, n ϭ 6). After 24 h, however, Ca v ␤3 Ϫ/Ϫ mice (n ϭ 10) exhibit more preference to cued food than Ca v ␤3 ϩ/ϩ mice (n ϭ 10). *, p Ͻ 0.05, one-way ANOVA. H, total food eaten during each test time. There are no differences in total current density between genotypes. The current density was estimated by dividing the peak amplitude by the cell capacitance (pA/pF). C, Ca 2ϩ current divided by maximum values of Ca 2ϩ current (I/I max ). D, value was obtained by fitting current traces evoked at 0 mV to a single exponential curve.
through N-or L-type VDCCs is known to be linked to the functions of Ca 2ϩ -activated K ϩ channels that are involved in shaping of APs, including the duration of AP and after hyperpolarization (AHP), and thus can modulate firing properties (40). First we produced AP discharges by a depolarizing current injection under the current clamp mode (Fig. 4A). The CA1 pyramidal neurons of Ca v ␤3 ϩ/ϩ (n ϭ 8) and Ca v ␤3 Ϫ/Ϫ (n ϭ 13) displayed very similar firing patterns. No significant difference was observed in the interspike intervals (Fig. 4C), the number (Fig. 4B) and duration (Fig. 4D) of APs. To directly assess the functions of Ca 2ϩ -activated K ϩ channels, we recorded AHP currents. Again, there was no difference in the AHP current between the Ca v ␤3 ϩ/ϩ (n ϭ 8) and the Ca v ␤3 Ϫ/Ϫ (n ϭ 9) (ϩ/ϩ, 131.60 Ϯ 15.86 pA; Ϫ/Ϫ, 112.16 Ϯ 15.42 pA, p ϭ 0.41, Student's t test) (Fig. 4E). These results show that the Ca v ␤3 Ϫ/Ϫ mutation did not affect intrinsic firing behaviors of hippocampal CA1 neurons.
Normal Basal Synaptic Transmission and Short Term Plasticity in the Ca v ␤3 Ϫ/Ϫ Mice-We then examined the basal synaptic function at hippocampal CA3-CA1 synapses in the Ca v ␤3 Ϫ/Ϫ mice. In mEPSCs (Fig. 5A), Ca v ␤3 Ϫ/Ϫ mice showed frequencies and amplitudes similar to those of Ca v ␤3 ϩ/ϩ mice (Fig. 5B). In addition, fEPSPs were recorded from the CA1 area of the hippocampus in response to stimulations of Schaffer collateral fibers. As illustrated in Fig. 5C, the input-output relation of synaptic transmission was not altered in the Ca v ␤3 Ϫ/Ϫ mice (ϩ/ϩ, n ϭ 10; Ϫ/Ϫ, n ϭ 12). We next studied the effect of the Ca v ␤3 mutation on paired-pulse facilitation (PPF), a presynaptic form of short term plasticity. PPF is a transient enhancement of neurotransmitter release induced by two closely spaced stimuli. This increase in release is usually attributed to intracellular Ca 2ϩ concentration in the presynaptic terminal following the first stimulus (41,42). There were no significant differences in all tested interpulse intervals between the Ca v ␤3 ϩ/ϩ (n ϭ 7) and the Ca v ␤3 Ϫ/Ϫ (n ϭ 9) (Fig. 5D). Taken together, these results indicate that the Ca v ␤3 mutation had no significant effect upon the basal synaptic function and the presynaptic short term plasticity in hippocampal CA3-CA1 synapses.
Increased NMDAR-mediated Synaptic Currents and NR2B Levels in the Ca v ␤3 Ϫ/Ϫ Mice-NMDAR is known to play a crucial role in LTP, as well as learning and memory (43)(44)(45)(46). Therefore, we examined the possibility that changes in the synaptic responses mediated by NMDAR might underlie the increased LTP in Ca v ␤3 Ϫ/Ϫ mice. To evaluate this possibility, we first measured the NMDAR-mediated fEPSPs by adding CNQX (10 M), an AMPA receptor blocker, to the buffer with reduced Mg 2ϩ concentration (0.1 mM). A significant difference was noted between the Ca v ␤3 ϩ/ϩ and the Ca v ␤3 Ϫ/Ϫ in these NMDAR-mediated field responses; the Ca v ␤3 Ϫ/Ϫ (n ϭ 13) exhibited higher NMDAR-mediated fEPSPs than the Ca v ␤3 ϩ/ϩ (n ϭ 12) (F (1, 23) ϭ 5.52, p Ͻ 0.05, two-way repeated ANOVA) (Fig. 7A). To assess this finding more directly, we measured the excitatory postsynaptic currents (EPSCs) evoked by stimulations at Schaffer collateral axons under the whole-cell voltage clamp conditions in CA1 neurons. It was found that there was no significant difference in the amplitude of AMPAR-mediated EPSCs at Ϫ70 mV between the two genotypes (Fig. 7B, left). However, a significant difference was noted in the NMDAR/ AMPAR amplitude ratio between Ca v ␤3 ϩ/ϩ (n ϭ 15, 0.28 Ϯ 0.04 at ϩ40 mV) and Ca v ␤3 Ϫ/Ϫ (n ϭ 13, 0.47 Ϯ 0.06 at ϩ40 mV) (p Ͻ 0.05, Student t test) (Fig. 7B, right). Together, these results indicate that NMDAR-mediated responses are increased in Ca v ␤3 Ϫ/Ϫ mice.
In an effort to obtain some clue for the mechanism underlying the increased NMDAR responses in the Ca v ␤3 Ϫ/Ϫ mice, we quantified the levels of NMDAR subunits by Western blot analysis. It was found that the protein level of NR2B subunit in the hippocampus of the Ca v ␤3 Ϫ/Ϫ mice (n ϭ 3, 1.14 Ϯ 0.05, normalized to Ca v ␤3 ϩ/ϩ values) was slightly increased relative to that of the Ca v ␤3 ϩ/ϩ mice (n ϭ 3) (p Ͻ 0.05, Student's t test) (Fig. 7C). There were no significant changes in the levels of other glutamate receptors.

DISCUSSION
In this study, we analyzed the Ca v ␤3-deficient mice with respect to their capacity for learning/memory and synaptic plasticity. Although there was no change in VDCCs currents and basal synaptic transmission, we found that the deletion of Ca v ␤3 caused an increase of NR2B expression and NMDAR activities, including currents and LTP, in the hippocampus and an enhanced capacity for learning and memory. This study  There are no differences in AMPAR-mediated EPSCs at Ϫ70 mV between genotypes (left bar graph), but the ratio of NMDAR/AMPAR response in the Ca v ␤3 Ϫ/Ϫ is higher than that of the Ca v ␤3 ϩ/ϩ (right bar graph). NMDAR-mediated responses were taken from the amplitude of currents at ϩ40 mV, 100 ms after EPSCs onset, whereas the AMPAR-mediated responses were taken as the peak amplitude of EPSCs recorded at Ϫ70 mV. *, p Ͻ 0.05. C, Western blot analysis. The relative levels of glutamate receptors in hippocampal proteins of Ca v ␤3 Ϫ/Ϫ mice. The NR2B level of Ca v ␤3 Ϫ/Ϫ mice was relatively increased, whereas other glutamate receptors did not change. The equal amount of protein loading was confirmed by normalizing against the amount of tubulin. *, p Ͻ 0.05. demonstrates a previously unidentified outcome of the deletion of Ca v ␤3 in the adult brain.
Yet the Ca v ␤ subunits of VDCCs have been known to be associated with VDCCs and regulate Ca 2ϩ influx through VDCCs by modulating the properties of VDCCs ␣ 1 subunits, including trafficking of channel complexes to the plasma membrane, Ca 2ϩ current densities, and voltage-dependent activation or inactivation (4,5). Of the Ca v ␤ subtypes, the Ca v ␤3 is the predominant form in the brain (17), and its role in several neurons has been revealed by studies carried out using mice lacking the Ca v ␤3. In superior cervical ganglion neurons, the Ca v ␤3 Ϫ/Ϫ showed reduced N-and L-type Ca 2ϩ currents relative to the Ca v ␤3 ϩ/ϩ and shifting of voltage-dependent activation in P/Qtype Ca 2ϩ currents (23). In dorsal root ganglion neurons, the Ca v ␤3 Ϫ/Ϫ mice showed a reduced expression of N-type VDCCs and functional alterations of Ca 2ϩ currents, which was thought to be involved in the reduced pain responses of the Ca v ␤3 Ϫ/Ϫ mice (38). In olfactory sensory neurons, the Ca v ␤3 Ϫ/Ϫ mice also exhibited decreased protein expressions and Ca 2ϩ currents of L-type and N-type VDCCs, leading to increased olfactory neuronal activities (39). These reduced expressions of proteins or Ca 2ϩ currents of VDCCs might be considered to mostly result from deficiency in trafficking of channel complexes to the plasma membrane.
However, although the Ca v ␤3 is known to be highly expressed in the hippocampus (17) and has been shown to associate with 42% of the ␣ 1 subunits of L-type VDCCs in the hippocampus (18), we could not observe a change in nifedipinesensitive L-type Ca 2ϩ currents in hippocampal CA1 pyramidal neurons of the Ca v ␤3 Ϫ/Ϫ mice (supplemental Fig. 1). In addition, there were no clear differences in the patterns of the immunohistological labeling for the ␣1C (Ca v 1.2) and the ␣1D (Ca v 1.3) subunits of L-type VDCCs in the hippocampus, between the two genotypes (supplemental Fig. 2). Furthermore, although Ca v ␤3 in the brain was shown to associate with about 52% ␣1B subunit of N-type VDCCs that play a crucial role in neurotransmitter release at hippocampal CA3-CA1 synapses (19 -21, 47, 48), the basal synaptic transmission, including mEPSCs, was not altered at hippocampal CA3-CA1 synapses of the Ca v ␤3 Ϫ/Ϫ mice. Therefore, some compensation by other Ca v ␤ isotypes might have occurred for the deletion of Ca v ␤3 in the hippocampus of the Ca v ␤3 Ϫ/Ϫ mice, as was reported in olfactory sensory neurons of the Ca v ␤3 Ϫ/Ϫ mice (39).
Instead, however, we found an increased LTP at hippocampal CA3-CA1 synapses in the Ca v ␤3 Ϫ/Ϫ mice. The induction of LTP by a tetanic stimulation at 100 Hz is known to be dependent on NMDAR, and 200-Hz LTP requires both NMDAR and L-type VDCCs at hippocampal CA3-CA1 synapses (49). When NMDAR was blocked by D-AP5, the enhancement in 100-Hz and 200-Hz LTP of the Ca v ␤3 Ϫ/Ϫ mice was obliterated. This indicates that the increased potentiation in the Ca v ␤3 Ϫ/Ϫ is of the NMDAR-dependent component in LTP, rather than L-type VDCC-dependent. The increased LTP and long term memory in the Ca v ␤3 Ϫ/Ϫ mice could be analogous to other cases where an alteration of NMDAR-mediated synaptic responses resulting from the increased levels of NR2B was shown (45,46).
Although the Ca 2ϩ currents and mEPSCs were measured from 2-to 3-week-old mice, basal synaptic transmission and LTP were recorded in 7-to 8-week-old mice. Thus, no alteration in Ca 2ϩ currents of at least N-and L-type VDCCs could be expected in the adult Ca v ␤3 Ϫ/Ϫ mice, because they showed normal responses in basal synaptic transmission and NMDAindependent LTP, in which N-and L-type VDCCs have a crucial role, respectively (21,22,(47)(48)(49).
Our results suggest a possibility that Ca v ␤3 can be a multifunctional protein as was shown for other Ca v ␤ isotypes. The studies of crystal structures revealed that Ca v ␤ subunits belong to membrane-associated guanylate kinase family that has scaffolding functions, suggesting that the Ca v ␤ can play a role in scaffolding multiple signaling pathways by protein-protein interactions through SH3 and GK domains (6,8,9). Recently, it was suggested that the Ca v ␤ could directly interact with other proteins, and furthermore it could function without marked influences on the property of VDCCs (10,11,50). The physiological unbinding of the Ca v ␤ from the VDCCs complex has already been demonstrated from the inactivation heterogeneity of VDCCs and reversibility of the interaction with ␣ 1 subunits (51,52). It was reported that Ca v ␤ could directly bind to Gem and Rem, small G-proteins that have a GTPase activity, and this interaction inhibited the surface expression and the activity of VDCCs (12,13). In addition, it was also shown that Ca v ␤ could promote endocytosis of VDCCs by interaction with dynamin (14). A short splice variant of Ca v ␤4 could directly interact with CHCB2, a nuclear protein, and then translocate into the nucleus for the subsequent regulation of gene transcription in the cochlea (15). In this study, it was found that the Ca v ␤ could function independently from VDCCs without marked influences on the surface expression and voltage-dependent properties of VDCCs. Furthermore, inositol 1,4,5-trisphosphatemediated signaling was enhanced in Ca v ␤3-deficient pancreatic ␤ cells, whereas Ca 2ϩ currents of VDCCs were not affected (16). Similarly, Ca v ␤ were found to internalize Shaker K ϩ channels by association with dynamin (14). These activities of Ca v ␤ are considered to be completely independent of VDCCs regulation, and thus indicate that Ca v ␤ can function as a multifunctional protein by interactions with other proteins. In this light, it might be possible that the Ca v ␤3 can directly or indirectly associate with NR2B.
Although our results showed a modest increase of NR2B in the mutant, it is not clear whether this increase can totally explain how the NMDAR activities are enhanced. In the meantime, it was discovered that the C-terminal tail region of Ca v 1.3 L-type VDCC bound to the SH3 domain of Shank, a postsynaptic scaffolding protein (53)(54)(55). Shank is also known to associate with GKAP-PSD95-NR2B through postsynaptic density-95 (PSD-95)/Discs large/zona occludens-1 domain (56). One of the binding sites of Ca v ␤ is the C-terminal tail region of ␣ 1 subunits of VDCCs (6,8,9,57). In this light, the removal of Ca v ␤3 might have an influence on the interaction of VDCCs and their partners and then could lead to an alteration in the NMDAR activity. Alternatively, we cannot rule out the possibility that a compensatory increase of other Ca v ␤ isotypes or other developmental compensation, which may have occurred in the Ca v ␤3 Ϫ/Ϫ hippocampus, could also be linked to the alteration in the NMDAR activity. In addition, previously described behavioral alterations from the changes in dorsal root ganglion or olfactory neuronal activities in the Ca v ␤3 Ϫ/Ϫ mice (38,39) could contribute to the phenotypes shown in our results.
Initially, we started investigating the role of the Ca v ␤3 in synaptic transmission and hippocampus-dependent learning and memory because of its known relationship with N-or L-type VDCCs. Interestingly, we found that the ablation of Ca v ␤3 led to enhanced LTP and capacity for learning and memory in the animal. These phenotypes appear to be due to the increased NMDAR activity with increased NR2B levels in the Ca v ␤3 Ϫ/Ϫ mice. Even though the precise mechanism of the enhancement of the NMDAR activity in the Ca v ␤3 Ϫ/Ϫ mice is not yet completely understood, our experiments may reveal a potentially novel function of Ca v ␤3, unrelated to a role associated with VDCCs. Further studies of the relationship, including direct or indirect protein-protein interactions, between Ca v ␤3 and NMDAR will be needed to confirm this role of Ca v ␤3 in the adult brain.