Point Mutation in Syntaxin-1A Causes Abnormal Vesicle Recycling, Behaviors, and Short Term Plasticity*

Background: Roles of the syntaxin-1A·CaMKII interaction are not physiologically understood in vivo. Results: A point mutation in syntaxin-1A caused abnormal plasticity, recycling, and behaviors in mice. Conclusion: The CaMKII/syntaxin-1A interaction is essential for maintenance of neuronal plasticity. Significance: Syntaxin-1A is involved in regulatory pathways in higher brain functions. Syntaxin-1A is a t-SNARE that is involved in vesicle docking and vesicle fusion; it is important in presynaptic exocytosis in neurons because it interacts with many regulatory proteins. Previously, we found the following: 1) that autophosphorylated Ca2+/calmodulin-dependent protein kinase II (CaMKII), an important modulator of neural plasticity, interacts with syntaxin-1A to regulate exocytosis, and 2) that a syntaxin missense mutation (R151G) attenuated this interaction. To determine more precisely the physiological importance of this interaction between CaMKII and syntaxin, we generated mice with a knock-in (KI) syntaxin-1A (R151G) mutation. Complexin is a molecular clamp involved in exocytosis, and in the KI mice, recruitment of complexin to the SNARE complex was reduced because of an abnormal CaMKII/syntaxin interaction. Nevertheless, SNARE complex formation was not inhibited, and consequently, basal neurotransmission was normal. However, the KI mice did exhibit more enhanced presynaptic plasticity than wild-type littermates; this enhanced plasticity could be associated with synaptic response than did wild-type littermates; this pronounced response included several behavioral abnormalities. Notably, the R151G phenotypes were generally similar to previously reported CaMKII mutant phenotypes. Additionally, synaptic recycling in these KI mice was delayed, and the density of synaptic vesicles was reduced. Taken together, our results indicated that this single point mutation in syntaxin-1A causes abnormal regulation of neuronal plasticity and vesicle recycling and that the affected syntaxin-1A/CaMKII interaction is essential for normal brain and synaptic functions in vivo.

ties. Notably, the R151G phenotypes were generally similar to previously reported CaMKII mutant phenotypes. Additionally, synaptic recycling in these KI mice was delayed, and the density of synaptic vesicles was reduced. Taken together, our results indicated that this single point mutation in syntaxin-1A causes abnormal regulation of neuronal plasticity and vesicle recycling and that the affected syntaxin-1A/CaMKII interaction is essential for normal brain and synaptic functions in vivo.
Syntaxin-1A is a t-SNARE, and in neurons, it is a main component of the presynaptic SNARE complex that executes exocytosis (1). Neuronal SNARE complexes include syntaxin-1A, SNAP-25 (another t-SNARE), and VAMP-2 (a v-SNARE) and are essential for both vesicle docking and fusion during the presynaptic exocytosis that leads to transmitter release. The neuronal isoform of syntaxin-1A, designated syntaxin-1B, can also be incorporated into SNARE complexes; however, the specific function of syntaxin-1B is not clear. Syntaxin-1A interacts with many kinds of proteins, and most of these interactions require the syntaxin-1A H3 domain, which is also crucial for SNARE complex formation. However, the interaction between syntaxin-1A and autophosphorylated Ca 2ϩ /calmodulin-dependent protein kinase II (CaMKII) 2 requires a linker domain in syntaxin-1A that connects its N-terminal helices (Ha-Hc) and its H3 domain but not the H3 domain itself (2,3). The interaction between syntaxin-1A and autophosphorylated CaMKII is Ca 2ϩ -dependent (2,3). At the cellular level, an intermediate complex is necessary for exocytosis because CaMKII binds to syntaxin-1A; importantly, a missense mutation in syntaxin-1A (R151G) can inhibit this binding and thereby prevent syntaxin-1A from regulating exocytosis (2,4).
CaMKII is an oligomeric protein kinase and is particularly abundant in neurons that have highly plastic properties, such as neurons in the hippocampus, cerebral cortex, or cerebellum (5,6). CaMKII plays a central role in expression of postsynaptic long term plasticity. Whereas the role of CaMKII in long term plasticity is well documented, its role in presynaptic short term plasticity (STP) is not well understood. Notably, CaMKII is a major component of the presynaptic termini and the surface of synaptic vesicles (7,8), and it is also a major component of postsynaptic densities.
We reasoned that interactions between syntaxin-1A and CaMKII may play an important role in the regulation of STP; therefore, we generated syntaxin-1A (R151G) knock-in (KI) mice that carry a single point mutation in syntaxin-1A. Based on the biochemical findings described here, we concluded that interactions between syntaxin-1A and CaMKII and the consequent recruitment of complexin to the SNARE complex were attenuated in the R151G KI mice. Based on our electrophysiological data, we concluded that the paired-pulse facilitation (PPF), an STP form, and post-tetanic potentiation (PTP), another STP form, were enhanced in the KI mice. In contrast, SNARE complex formation itself and basal transmission were normal in these KI mice. These phenotypes, together with abnormal behaviors indicative of excessive brain activity, may have resulted from insufficient CaMKII activity in presynaptic termini.
Additionally, the recycling of synaptic vesicles (SVs) was delayed in these KI mice, and the SV density was reduced. Therefore, knock-in of this single point mutation (syntaxin-1A (R151G)) attenuated the interaction between syntaxin-1A and CaMKII and resulted in abnormal regulation of exocytosis, STP, SV recycling, and higher brain functions that regulate behaviors. We conclude that the action of CaMKII in the presynaptic terminal is closely related to the regulation of STP because of an interaction between CaMKII and syntaxin-1A and the consequent action of complexin.

Animals
This research was approved by the respective Animal Care and Experimentation Committee of each institution to which the collaborators belong, and all experiments were performed according to guidelines of these committees.
The point mutation was confirmed by direct sequencing of the resultant PCR product, which included exon 6. The resulting chimeric mice were mated to C57BL/6N mice to yield heterozygous mice (9). Homozygous mutant mice and control mice were obtained by crossing heterozygous pairs. Genotypes were confirmed by a Southern blot analysis that is described below and, routinely, determined by PCR using the following primers: 5-GAGTGGGAACCCTGCCATC-3Ј, 5Ј-ATAGAC-AACTTGCAGTTCTCTC-3Ј, and 5Ј-CCAGACTGCCTTGG-GAAAAG-3Ј. A 539-bp product resulted from the wild-type (WT) locus, and a 695-bp product resulted from the KI locus.
The engineered R151G mutation generated a new restriction site for BstNI. The point mutation was confirmed by restriction fragment length polymorphism-PCR. Exon 6 was amplified with the oligonucleotide primers 5Ј-AGTCTCTGGGTTTTC-CGTAG-3Ј and 5Ј-CTCAAACCACAACCCCAGGC-3Ј. The 234-bp PCR product was digested with BstNI to monitor to the new restriction site generated by the mutation. The resulting products were WT 218 bp and KI 163 and 55 bp.
Pairs of synaptosome samples included one KI sample and one WT sample (1-20 g of protein) that were prepared at the same time and were then applied to the same SDS-polyacrylamide gel in triplicate. The obtained values were normalized to the values of valosin-containing protein or of Rho-guanine nucleotide dissociation inhibitor protein detected on the same blot. The normalized values were further standardized relative to the average of those of WT. The ECL prime chemiluminescence kit (GE Healthcare) was used to detect protein bands on immunoblots, and images were captured with a cooled CCD camera system (Light Capture-II, Atto Corp., Tokyo, Japan).

Immunoprecipitation Studies
For co-immunoprecipitation assays and analysis of syntaxin-1A⅐CaMKII complexes, synaptosomes dissolved in Triton X-100 and CHAPS (1.5 mg/assay) were preincubated with protein G-Sepharose 4Fast Flow (GE Healthcare) at 4°C for 2 h. The supernatant was collected and divided into identical samples that were then incubated overnight either with anti-syntaxin-1A mAb and normal mouse IgG or with anti-CaMKII polyclonal antibody and normal rabbit IgG at 4°C. For co-immunoprecipitation assays with complexin and syntaxin-1A, synaptosomes dissolved in Targitol, CHAPS, and n-dodecyl-␤-D-maltoside (600 g) were used. Next, anti-complexin polyclonal antibody was added to each sample, and the mixtures were incubated at 4°C for 2 h and then incubated with protein G-Sepharose beads for another 1 h. The one-sample t test was used for statistical comparisons of densitometric values.

Assay of SDS-resistant Complexes
To study SNARE complex formation, synaptosomes from KI or from WT mice homogenized in 20 mM HEPES (pH 7.2), 150 mM NaCl were directly lysed in 1% SDS. The protein concentration of each lysates was determined, and each lysate was then mixed with SDS sample buffer, and the sample from each genotype was then divided in two tubes as follows: for each genotype, one tube was boiled, and the other was kept at room temperature. The same blot was sequentially probed with antisyntaxin-1A, syntaxin-1B, syntaxin-1A/B, and ␣-tubulin antibodies. The densitometric value of the monomeric form of each SNARE protein was analyzed.
Synaptic responses were recorded using an Axopatch 1D amplifier (Molecular Devices, Sunnyvale, CA); signals were filtered at 1 kHz and digitized at 10 kHz with Digidata 1322A (Molecular Devices). Extracellular field-potential recordings were made in the stratum radiatum of the CA1 region using a glass electrode filled with 3 M NaCl. Whole-cell patch clamp recordings were performed using a patch pipette filled with the internal solution containing the following (mM): cesium gluconate 122.5, CsCl 17.5, NaCl 8, HEPES 10, EGTA 0.2, Mg-ATP 2, and Na 3 -GTP 0.3 (pH 7.2; 290 -310 mosM). To evoke synaptic responses, Schaffer collateral/commissural fibers were stimulated at 0.1 Hz. For extracellular recordings, stimulus strength was adjusted to obtain ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptor-mediated excitatory postsynaptic potentials (EPSPs) with a slope value between 0.10 and 0.15 mV/ms. For whole-cell patch clamp recordings, stimulus strength was adjusted to evoke excitatory postsynaptic currents (EPSCs) with an average amplitude of about 150 pA.
To examine input-output relationships, a low concentration of 6-cyano-7-nitroquinoxaline-2,3-dione (1 M), a non-Nmethyl-D-aspartate (NMDA) receptor antagonist, was included to partially block AMPA receptor-mediated synaptic responses; this blocker reduced the nonlinear summation of EPSPs and enabled more accurate measurements of EPSP slopes. These experiments were performed in the presence of 25 M D-2amino-5-phosphonovaleric acid (D-APV) to block NMDA receptor-mediated synaptic responses. PPF was induced by two consecutive stimuli delivered at inter-pulse intervals of 30, 50, 100, 200, or 300 ms in the presence of 25 M D-APV. The paired-pulse ratio was calculated by dividing the slope value of the second EPSP by that of the first EPSP. PTP was elicited by tetanic stimulation (100 Hz, 1 s) of afferent fibers in the presence of 50 M D-APV.
For estimating the size of the readily releasable pool (RRP) of SVs and the rate of SV supply from the reserve pool, EPSCs were evoked at 20 Hz, and the cumulative EPSC amplitude (normalized to the base-line EPSC amplitude) was plotted against the stimulus number. The regression line was then drawn by fitting a line to the last 10 points (from the 91st to the 100th pulse), and the slopes of these regression lines and their intercepts with the y axis (cumulative EPSC) were determined (11).
All values are expressed as the mean Ϯ S.E. The Student's t test was used to determine whether there was a statistically significant difference (p Ͻ 0.05) between the means of groups (*, p Ͻ 0.05; **, p Ͻ 0.01).
Scanning fluorescence images were acquired using a laser scanning microscope (LSM 510 META, Carl Zeiss Inc.) through a Plan-Apochromat ϫ100/1.4 objective at a sampling of 106 nm/pixel (912 ϫ 912 pixels). ImageJ software (National Institutes of Health) was used to perform all measurements. Individual puncta of 9 ϫ 9 pixels were manually selected to mark individual synapses. Larger puncta were rejected to avoid clusters of smaller synapses (11). For each experiment, more than 150 puncta were counted from five images per coverslip. For each set of samples in an experiment, each fluorescent value was normalized to the median of the fluorescent values from wild-type samples.

Electron Microscopy
While under deep pentobarbital anesthesia (100 mg/kg of body weight), each of three wild-type and each of three mutant mice were perfused transcardially with 2% paraformaldehyde, 2% glutaraldehyde in 0.1 M sodium cacodylate buffer (pH 7.4). Hippocampal slices were further post-fixed with 1% osmium tetroxide in 0.1 M sodium cacodylate buffer (pH 7.4) for 2 h, stained in block with 2% uranyl acetate for 1 h, dehydrated in graded alcohols, and embedded in Epon 812 resin. Ultrathin sections were cut with Ultracut TM (Leica). For quantitative analyses of morphological features at axo-spinous asymmetrical synapses in the distal striatum radiatum of the hippocampal CA1 region, electron micrographs were taken with an H7100 electron microscope (Hitachi, Tokyo, Japan). The length of each active zone and the densities of synapses, of SVs, and of docked SVs were measured using IPLab software (Scanalytics, Rockville, MD). Docked synaptic vesicles were defined as vesicles that were separated by less than 10 nm from the presynaptic membrane.

Behavioral Analysis
All behavioral tests were carried out with male mice of C57BL/N strain, including the KI mice that were at least 10 weeks old at the start of testing. Raw data from the behavioral tests, the date on which each experiment was performed, and the age of each mouse at the time of each respective experiment are shown in the mouse phenotype database. Mice were housed in groups (3-4 mice per cage) in a room with a 12-h light/dark cycle (lights on at 7:00 a.m.) with ad libitum access to food and water. All behavioral testing procedures were approved by the Animal Research Committee, Graduate School of Medicine, Kyoto University, and the National Institute for Physiological Sciences (12,13). The software applications used for the behav-ioral studies (Image HC, Image FZ, Image TS, and Image RM) were based on the public domain ImageJ program (rsb.info.nih. gov), which was modified for each test by authors (available through O'Hara & Co., Tokyo, Japan). Statistical analysis was conducted with StatView (SAS Institute, Cary, NC). Data were analyzed by two-way analysis of variance or two-way analysis of variance with repeated measures, unless noted otherwise. Values in graphs are expressed as the mean Ϯ S.E.
Open Field Test-Locomotor activity was measured via an open field test as described previously (13). In each trial, data were collected for 120 min.
Social Interaction Test in Home Cage-Social interaction monitoring in the home cage was conducted as described previously (14). Social interaction was measured by counting the number of particles in each frame as follows: two particles indicated the mice were not in contact with each other, and one particle indicated contact between the two mice.
Contextual and Cued Fear Conditioning-The contextual and cued fear conditioning test was performed as described previously (12,13). White noise (60 db), which served as the conditioned stimulus, was presented for 30 s, followed by a mild (2 s, 0.35 mA) foot shock, which served as the unconditioned stimulus. For each mouse, context testing was conducted for 180 s 1 day after conditioning in the same chamber. Data acquisition, control of stimuli (i.e. tones and shocks), and data anal- Representative images of (A) and graphs (B) depicting the quantifications of immunoblots (IB) resulting from the co-immunoprecipitation of Stx-1A (left) and CaMKII (right) from brain synaptosomes isolated from WT and Stx-1A (R151G) KI mice. IP, immunoprecipitation. The amounts of co-immunoprecipitated proteins were normalized to immunoprecipitated Stx-1A or CaMKII␣, and a percentage calculated relative to the amount in WT mice is shown. The amount of Stx-1A that co-immunoprecipitated with CaMKII was significantly reduced in KI mice compared with that in WT mice (mean Ϯ S.D., 27.00 Ϯ 4.583%, n ϭ 3; *, p Ͻ 0.05 in one sample t test). C and D, interaction between Stx-1A and complexin (Cplx) was attenuated in the samples from KI mice. Representative images (C) and graphs depicting quantification (D) of immunoblots containing the Stx-1A that co-immunoprecipitated with Cplx in synaptosomes isolated from WT and KI mice are shown. The amounts of co-immunoprecipitated Stx-1A, normalized to the value of immunoprecipitated Cplx, were significantly lower in KI mice than in WT mice  . SNARE complex formation in the synaptosomes from KI mice was similar to those from WT mice. A, representative immunoblot of boiled (ϩ) or non-boiled synaptosome from KI or WT mice. B, quantification of the analysis depicted in A showed fold change of monomeric syntaxins after boiling. No significant differences were observed between KI and WT samples (paired t test). The result of immunoprecipitation and GST pulldown assays indicated that binding of complexin to the SNARE complex was reduced; therefore, we examined whether the formation of the SNARE complex was affected by the R151G mutation introduced into the KI mice. The assembled SNARE complexes were resistant to SDS sample buffer at RT (20) but dissociated into monomeric SNARE proteins at 100°C. To evaluate the amount of assembled SNARE complex, we compared WT and KI samples with regard to the increase in monomeric syntaxins induced by the boiling of SDS-resistant SNARE complex. Statistically significant analyses were performed with data from two duplicate experiments using three synaptosome pairs from different preparations. No significant difference was observed between WT and KI samples in paired t test. C and D, ternary SNARE complex was assembled in the mixture of recombinant SNARE proteins. Recombinant SNAP-25 and VAMP-2 were mixed with GST-syntaxin-1A (WT) or GST-syntaxin-1A (R151G) (R). The SDS-containing sample buffer was added to the mixture but was kept at room temperature. The same PVDF membrane was sequentially immunoblotted using anti-VAMP-2, syntaxin-1, and SNAP-25 antibodies. Images obtained using the CCD camera (C) were pseudo-colored and merged (D) using ImageJ software. The arrows indicated the band recognized by all those three antibodies. The white asterisks in C indicated the monomers recognized by each of three antibodies, respectively. NOVEMBER 29, 2013 • VOLUME 288 • NUMBER 48 ysis were performed automatically with a program based on ImageJ. When this stimulated area was below a certain threshold (i.e. 20 pixels), the behavior was categorized as "freezing."

Mutation in Syntaxin Causes Abnormal Short Term Plasticity
Porsolt Forced Swim Test-Depression-related behavior was assessed via the Porsolt forced swim test as described previously (13). Mice were placed in the cylinders, and immobility and distance traveled were recorded over a 10-min test period. When the area equaled or exceeded the threshold, the mouse was considered to be "moving." Immobility lasting less than 2 s was not included in the analysis.
Eight-arm Radial Maze-Working memory was measured in fully automated eight-arm radial maze apparatuses (O'Hara & Co., Tokyo, Japan) as described previously (15). Data acquisition and data analysis were performed using Image RM software. The software was based on the public domain ImageJ program (http://rsb.info.nih.gov/ij/), which was modified for the test by authors (available through O'Hara & Co., Tokyo, Japan).

Generation of Syntaxin-1A (R151G) KI Mice-Syntaxin-1A
is a t-SNARE that is a main component of the presynaptic SNARE complex. The docking-fusion activity of syntaxin results from a structural conversion from the closed (inactive) form to the open (active) form (16), and the linker domain in syntaxin is believed to play an important role in this conversion. Currently, the only known protein that binds to the linker domain is the autophosphorylated form of CaMKII (2). To examine the functional role of the structural conversion of syntaxin-1A, we used the C57BL/6N strain of mice to generate KI mice that carry a syntaxin-1A (R151G) mutation (Fig. 1A). This missense mutation within the syntaxin-1A linker domain severely reduced the CaMKII/syntaxin-1A interaction (4), and in KI mice, the syntaxin-1A (R151G) mutation was confirmed via both Southern blot analysis and Western blot analysis (Fig. 1, B and C). The KI mice developed normally and were fertile; moreover, they did not show any marked macroscopic abnormalities or any abnormal phenotypes when assayed under a dissecting microscope.
Syntaxin-1A and an isoform, syntaxin-1B, are among many varied synapse-associated proteins for which the amount of protein in these (R151G) KI mice did not differ relative to that of the respective protein in wild-type (WT) mice (Fig. 1, D and E).

Reduction of the CaMKII/Syntaxin Interaction in KI Mice Inhibits Recruitment of Complexin to SNARE Complexes-
First, we analyzed the interaction between CaMKII and syntaxin-1A in KI and control mice. Based on immunoprecipitation data from synaptosomal fractions, we confirmed that samples from KI mice had much less signal from CaMKII⅐syntaxin-1A complexes than did samples from WT mice (Fig. 2, A and B). Next, we focused on the interaction between complexin I/II (also called synaphin) and syntaxin-1A in SNARE complexes. Complexin (complexin I/II in the brain), a small presynaptic cytosolic protein, binds to the SNARE complex (1,(17)(18)(19); as a result, complexin can inhibit or stimulate exocytosis. In KI mice, the binding of syntaxin-1A to complexin I/II in the synaptosomal fraction was reduced to 40% of that in WT mice (Fig. 2, C and D). To confirm that the R151G mutation inhibited the interaction between syntaxin-1A and complexin, we performed an in vitro reconstitution study. In the presence of the autophosphorylated CaMKII, the amount of complexin bound to R151G syntaxin-1A was reduced to less than 50% of the WT syntaxin-1A-bound complexin (Fig. 2, E  and F; for complexin, p ϭ 0.0411). The amount of SNAP-25 bound to mutated syntaxin-1A did not significantly differ from that bound to the WT. These results indicated that the R151G mutation resulted not only in an attenuated CaMKII/syntaxin interaction but also in attenuated recruitment of complexin to the SNARE complex.
We examined whether the SNARE complex formation was perturbed by the R151G mutation. At room temperature, for-mation of the assembled SNARE complex, i.e. the SDS-resistant complex (20), was not changed (Fig. 3A), and there were no statistically significant differences between WT and mutant samples (Fig. 3B). These results indicate that this point mutation does affect formation of the SNARE complex, which is involved in docking and fusion directly. With samples from either WT or mutant mice, the amount of syntaxin-1B in the SNARE complex did not change (Fig. 3, B-D).
KI Mice Exhibited Abnormal STP-Next, we used electrophysiological techniques and acutely dissected hippocampal slices to examine basal synaptic transmission and presynaptic STP at Schaffer collateral-CA1 synapses from KI mice and their WT littermates. The input-output relationships of EPSPs that are mediated by AMPA receptors in KI mice were not significantly different from those in WT mice (Fig. 4A), indicating that basal synaptic transmission was normal in mutants and that the initial SV release probability was similar between the KI and WT mice. In contrast, at inter-pulse intervals of 100 ms ; p ϭ 0.0199), PPF that depends upon the initial release probability (21) were significantly larger in mutants than in WTs; however, there were no differences between KI and WT mice in PPF at inter-pulse intervals of 30 ms (WT, 1.52 Ϯ 0.02; KI, 1.55 Ϯ 0.02; p ϭ 0.2527), 50 ms (WT, 1.54 Ϯ 0.02; KI, 1.59 Ϯ 0.02; p ϭ 0.0675), or 300 ms (WT, 1.18 Ϯ 0.01; KI, 1.20 Ϯ 0.01; p ϭ 0.1304) (Fig. 4B). In addition, PTP, another form of STP induced by high frequency (100 Hz, 1 s) stimulation, was also significantly higher in KI mice than in WT mice (Fig. 4C). These results revealed that short term plasticity was impaired in KI mice; however, this impairment did not result simply from a change in initial release probability but instead from time-and use-dependent changes in release mechanisms, presumably through the regulation of SV fusion to the terminal membrane and/or vesicular trafficking in the presynaptic terminal (22).
To examine the involvement of SV dynamics in STP, we analyzed synaptic responses to prolonged low frequency (5 Hz) stimulation; these responses are dependent upon the release probability, the size of SV pools, and the rate of SV supply to the pools. Once a stable base line had been maintained for more than 5 min (Fig. 4D), 5-Hz stimulation was applied for 3 min, and recordings were taken (Fig. 4E). Just after the start of 5-Hz stimulation, EPSPs exhibited transient facilitation in WT mice; this facilitation was followed by synaptic depression that persisted throughout the remaining stimulation (Fig. 4E). KI mice also showed similar changes, but throughout the stimulations, the synaptic responses were consistently larger in KI mice than in WT mice (Fig. 4, D and E). Furthermore, the synaptic depression that became evident soon after stimulation was significantly smaller in KI mice than in WT mice (WT, 69.6 Ϯ 1.8%; KI, 76.7 Ϯ 2.1%; p ϭ 0.0127). In contrast, the peaks of the depression (WT, 61.3 Ϯ 1.4%; KI, 62.1 Ϯ 2.4%; p ϭ 0.7917) and the recovery time constants after the 5-Hz stimulation (WT, 2.98 Ϯ 0.36 min; KI, 2.86 Ϯ 0.32 min; p ϭ 0.8043) were indistinguishable between the genotypes (Fig. 4D). These results indicated that the rate of SV supply was faster and/or the release probability during repetitive stimulation was higher in KI mice than in WT mice and that the initial release probability did not differ between KI and WT mice; these findings suggested that only plastic changes in synaptic responses were affected, but basal synaptic transmission was intact in KI mice.
Estimation of the Size of SV Pools and of Release Probability during Repetitive Synaptic Activation-Measurements of cumulative synaptic responses during high levels of synaptic activation are often used to accurately estimate the size of an average RRP of SVs and the rate of SV supply to an average RRP (23). Therefore, we used the whole-cell patch clamp recording technique to analyze synaptic responses at the Schaffer collateral-CA1 synapse; we used 20-Hz stimulation because it is sufficient to deplete SVs in the RRP (23). We found that the responses that were induced by high frequency stimulation (e.g. 20-Hz) in KI mice had larger amplitudes than those in WT mice (Fig. 5A). EPSCs were transiently facilitated during the early phase 20-Hz stimulation and then depressed during the later phase 20-Hz stimulation (Fig. 5B). As observed in field-potential recordings at 5 Hz (Fig. 4D), the late phase depression was significantly smaller in KI mice than in WT mice (Fig. 5B). Furthermore, transient potentiation after 20-Hz stimulation was also larger in KI mice than in WT mice (Fig. 5F), as is evident from the field potential recordings at 100 Hz (Fig. 4C).
The intercept of the regression line of the cumulative plot of EPSCs with the axis of the cumulative EPSC amplitude typically represents the RRP size, and the slope of the regression line represents the rate of SV supply when the release probability is constant (24,25). The slope of the regression line in KI mice (51.5 Ϯ 4.3 (n ϭ 23)) was significantly larger (p ϭ 0.0374) than that in WT mice (39.8 Ϯ 3.4 (n ϭ 24)) ( Fig. 5, C and D), but the intercepts did not differ between KI and WT mice (Fig. 4, C and  E), indicating that the pool size was similar between the genotypes (WT, 4,794.1 Ϯ 377.0; KI, 4645.2 Ϯ 291.3; p ϭ 0.7574). Analyses were performed automatically with Image HC software. *, p Ͻ 0.05. C, contextual memory in KI mice. Panel a, KI mice showed less freezing following the conditioning than did WT mice (F 1,37 ϭ 11.802, p ϭ 0.0015). Panel b, reduced freezing was observed in KI mice during the contextual test (left) (F 1,37 ϭ 10.879, p ϭ 0.0022) and during the cued testing in the altered context (right) (pretone period, F 1,37 ϭ 12.279, p ϭ 0.0012; tone period, F 1,37 ϭ 3.406, p ϭ 0.0730), which was conducted 24 h after conditioning. Panel c, KI mice also showed decreased freezing during the contextual test (left) (F 1,37 ϭ 9.633, p ϭ 0.0037), and during the cued testing in the altered context (right) (pretone period, F 1,37 ϭ 14.062, p ϭ 0.0006; tone period, F 1,37 ϭ 10.515, p ϭ 0.0025), which was conducted 12 days after conditioning. **, p Ͻ 0.01. D, altered performance of KI mice in the Porsolt forced swim test. For the tests conducted on the 1st day, there were no significant differences in immobility between KI and WT mice (F 1,38 ϭ 0.669, p ϭ 0.4185), but the KI mice traveled longer distances than did WT mice (F 1,38 ϭ 6.630, p ϭ 0.0140; *, p Ͻ 0.05). On the following day, KI mice traveled longer distances than did WT mice (p ϭ 0.0091; **, p Ͻ 0.01). Notably, the KI mice displayed decreased immobility (F 1,38 ϭ 8.278, p ϭ 0.0065; † †, p Ͻ 0.01) and increased distance traveled (F 1,38 ϭ 10.962, p ϭ 0.0020; † †p Ͻ 0.01) relative to the WT mice during first 1 min of the test. E, KI mice showed a performance deficit in the initial phase of training of the eight-arm radial maze test (1-18 session, F 1,36 ϭ 7.512, p ϭ 0.0095), which assesses working memory. After sufficient training sessions, the performance of the KI mice was comparable with that of WT (19 -30 session, F 1,36 ϭ 0.133, p ϭ 0.7171). **, p Ͻ 0.01.
However, the SV supply rate and/or the release probability during higher frequency stimulation were higher in KI mice than in WT mice, suggesting that SV turnover after frequent stimulation is specifically overactivated in KI mice relative to that in WT mice.
Delayed Recycling of SVs in Syntaxin (R151G) KI Mice-Next, we examined whether endocytosis, like exocytosis, was affected in these KI mice. Using the cultured hippocampal neurons, we confirmed that KI cells showed normal synaptogenesis in vitro (Fig. 6A). The distribution of synaptic proteins in KI cells was quantitatively similar to that in WT (Fig. 6B). FM4-64 is a hydrophilic fluorescent dye that only becomes incorporated into the cell membrane during endocytosis. Using this dye and high K ϩ stimulation, we examined SV recycling and observed a significant delay of SV recycling (Fig. 6, C and D) in KI cells.
If SV recycling is continuously disturbed for long periods, KI mice may exhibit an abnormal distribution of SVs. Thus, we performed a quantitative electron microscopic analysis of hippocampal neurons. Although the number or the size of the synapses was not changed in KI mice, the SV density was reduced (Fig. 7A) to approximately two-thirds of that in WT mice (Fig.  7B). These results indicated that the (R151G) mutation affected normal SV recycling, and as a result, mature KI mice had a lower SV density than mature WT mice.
Behavioral Analyses Indicate that KI Mice Exhibit Hyperactive Locomotion-Generalized characteristics did not differ significantly between KI and WT mice. Under these conditions, behavioral tests are very useful for quantitatively evaluating the higher brain functions of transgenic mice (12). The KI mice exhibited locomotor hyperactivity in the open field and fewer anxiety-like behaviors in the open field (Fig. 8A). The KI mice were also hyperactive in their home cages (Fig. 8B). Contextual memory tests revealed that the KI mice showed less freezing following conditioning (Fig. 8C). In the Porsolt forced swim test, the KI mice traveled longer distances than WT mice on the 1st day as well as the following day. In particular, the KI mice exhibited significantly longer distances traveled and significantly less immobility at the first 1 min of the 2nd days' test (Fig.  8D, see † †). The KI mice showed a performance deficit in the initial phase of training of the eight-arm radial maze test, which is a test that assesses working memory (Fig. 8E).
Based on all these findings, the KI mice, which have only a single point mutation (syntaxin-1A R151G), showed disorders in several higher brain functions. In particular, like CaMKII␣ϩ/Ϫ mice, the KI mice showed hyperactivity, decreased anxiety-like behaviors, decreased depression-like behavior, and impairment of working memory (Fig. 8, A, B, D, and E) (13,15). These findings indicate that the R151G mutation causes insufficient CaMKII action in the presynaptic terminal.

DISCUSSION
Syntaxin-1A is a key player in presynaptic exocytosis as a SNARE complex component. We generated and examined syntaxin-1A (R151G) KI mice, which have a single point mutation that attenuated the interaction between CaMKII and syntaxin-1A. A key biochemical abnormality caused by this KI mutation was reduced recruitment of complexin, which normally functions as a molecular clamp, to SNARE complexes (Fig. 2). The mutation affected neither normal SNARE complex formation (Fig. 3) nor the associated basal neurotransmitter release (Fig.  4). However, the KI mutation was associated with abnormal short term plasticity, excess SV release (Fig. 5), delayed SV recycling (Fig. 6), and reduced SV density in mature mice (Fig. 7); additionally, the mutation caused abnormal behaviors, including hyperactivity (Fig. 8). These results clearly indicated that syntaxin-1A is essential not only for exocytosis but also for the regulation of STP, SV recycling, and higher brain function through regulation of transmitter release. All of these were disturbed by a single R151G point mutation.
Role of Complexin in Regulating STP-Although the RRP size in syntaxin-1A-KI was similar to that in WT, the SV supply rate and/or the release probability during higher frequency stimulation was higher in KI than in WT (Fig. 4). Therefore, we asked the following two questions. 1) How does the CaMKII/syntaxin-1A interaction regulate exocytosis and play a role in STP? 2) How does impairment of this interaction lead to enhanced PPF? To answer these questions, we focused on complexin. Complexin is bound to the SNARE complex and thereby regu-  NOVEMBER 29, 2013 • VOLUME 288 • NUMBER 48 lates exocytosis after SV docking. However, the role of complexin remains unclear. In particular, the effect of complexin on exocytosis appears to be bidirectional. Initially, complexin adopts a conformation that inhibits vesicle fusion; it then undergoes a conformational change, thereby adopting a conformation that stimulates exocytosis (29 -33). Because CaMKII⅐syntaxin-1A was necessary for recruitment of complexin to the SNARE complex and because the mutant syntaxin-1A (R151G) inhibited this recruitment both in in vitro reconstitution (Fig. 2, A  and B) and in vivo immunoprecipitation studies (Fig. 2, A-D), the "clamping" action of complexin on the SNARE complex is probably insufficient and incomplete. Repetitive synaptic stimulation probably induces an accumulation and therefore a rise of [Ca 2ϩ ] i in the presynaptic terminal that includes the residual Ca 2ϩ that remained after the first pulse in the PPF experiments (34). These processes result in elevated autophosphorylation of CaMKII and enhance complexin recruitment to the SNARE complex. Data from many reconstitution studies and some electrophysiological studies have shown that complexin itself has an inhibitory effect on vesicular fusion and that it acts as a "clamp" (30,32,35); thus, it is very likely that, in the CNS including hippocampal neurons, recruitment of complexin via the CaMKII/syntaxin-1A interaction inhibits transmitter release during repetitive stimulation and that impairment of this recruitment leads to the increase of the second response in PPF. This hypothesis could also completely account for the enhanced PTP as well (Fig. 4B).

Mutation in Syntaxin Causes Abnormal Short Term Plasticity
Similarity of R151G Phenotypes to CaMKII Mutations-We demonstrated that syntaxin-1A (R151G) KI mice exhibited enhanced STP (Figs. 4 and 5) and abnormal behaviors (Fig. 8) and that both phenotypes were associated with and may be caused by an attenuated CaMKII/syntaxin-1A interaction and the consequent reduction in the recruitment of complexin to SNARE complexes (Fig. 2). However, the basal release probability in KI mice was not significantly different from that in WT mice (Fig. 4A), indicating that the CaMKII-dependent mechanisms acted only during the plastic stage, which occurs after stimulation. These results suggested that presynaptic CaMKII regulated STP and exocytosis via interactions with the proteins involved in the SNARE mechanism.
Our results seemed generally consistent with the results from previous experiments conducted by other groups that used CaMKII␣-KO homozygous or heterozygous mice (36,37) or presynaptic cell-specific CaMKII␣-KO mice ( Fig. 9) (38). In addition, the R151G KI mice exhibited abnormalities in higher brain functions that were very similar to those of CaMKII␣ ϩ/Ϫ mice (Fig. 8) (13,15,38). These findings appear also to be consistent with the hypothesis that presynaptic CaMKII activity was reduced because of the R151G mutation. Further studies are required to resolve these issues.
Abnormal SV Recycling and Syntaxin-1A-We also showed that the mutation caused a delay in SV recycling in the KI mice, and probably as a consequence of this delay, SV density was reduced in the brains of these mice. We do not currently know the exact reason for these abnormalities caused by a single point mutation of syntaxin-1A. One possible interpretation is that excess SV release caused by the stimulation also caused excess SV consumption, and endocytosis was not sufficiently rapid to prevent a delay in SV recycling. We could not compare the effects of the R151G mutation in syntaxin-1A directly to those of CaMKII mutations because the designs of the previous studies differed too greatly from our study design. Further investigation into whether the endocytic abnormality is dependent upon CaMKII or complexin should be a next step.