Ca2+-independent Activation of Ca2+/Calmodulin-dependent Protein Kinase II Bound to the C-terminal Domain of CaV2.1 Calcium Channels*

Background: Calmodulin regulation of Ca2+ channels mediates short term synaptic plasticity. Results: Binding of CaMKII to CaV2.1 channels induces Ca2+-independent kinase activity. Conclusion: A complex of CaMKII and CaV2.1 channels is required for short term synaptic plasticity. Significance: CaMKII bound to CaV2.1 may regulate synaptic plasticity. Ca2+/calmodulin-dependent protein kinase II (CaMKII) forms a major component of the postsynaptic density where its functions in synaptic plasticity are well established, but its presynaptic actions are poorly defined. Here we show that CaMKII binds directly to the C-terminal domain of CaV2.1 channels. Binding is enhanced by autophosphorylation, and the kinase-channel signaling complex persists after dephosphorylation and removal of the Ca2+/CaM stimulus. Autophosphorylated CaMKII can bind the CaV2.1 channel and synapsin-1 simultaneously. CaMKII binding to CaV2.1 channels induces Ca2+-independent activity of the kinase, which phosphorylates the enzyme itself as well as the neuronal substrate synapsin-1. Facilitation and inactivation of CaV2.1 channels by binding of Ca2+/CaM mediates short term synaptic plasticity in transfected superior cervical ganglion neurons, and these regulatory effects are prevented by a competing peptide and the endogenous brain inhibitor CaMKIIN, which blocks binding of CaMKII to CaV2.1 channels. These results define the functional properties of a signaling complex of CaMKII and CaV2.1 channels in which both binding partners are persistently activated by their association, and they further suggest that this complex is important in presynaptic terminals in regulating protein phosphorylation and short term synaptic plasticity.

nal domain (33). Interaction with CaMKII increases Ca V 2.1 channel activity and enhances Ca V 2.1 channel facilitation by slowing inactivation and shifting the voltage dependence of inactivation to more positive membrane potentials (33). These effects of CaMKII did not require the catalytic activity of the kinase, suggesting that binding per se was sufficient for channel regulation (33). Here we report that CaMKII binds directly to a site in the C-terminal domain of Ca V 2.1 channels and that autophosphorylation of CaMKII stimulates binding to this site. Autophosphorylated CaMKII can bind to the Ca V 2.1 channel and synapsin-1 simultaneously. Binding of Ca V 2.1 to CaMKII induces Ca 2ϩ -independent kinase activity, which mediates both autophosphorylation and phosphorylation of synapsin-1 at Ser-603. Block of this interaction with competing peptides or the endogenous brain-specific CaMKII inhibitor, CaMKIIN, prevents short term synaptic facilitation and depression. We present a molecular model in which the pore-forming ␣ 1 subunit of the Ca v 2.1 channel serves as a platform for association of CaMKII at the site of influx of Ca 2ϩ , where it is persistently activated and poised to phosphorylate synapsin-1 and other nearby substrates to regulate synaptic vesicle dynamics and synaptic plasticity.
Culture and Transfection of tsA-201 Cells-TsA-201 cells were plated and grown at the density of 9 ϫ 10 5 cells/100 mm dish in Hyclone DMEM/F medium and afterward used in experiments (33). The cells were transfected using transit-LT1 (Mirus) and 8 g of total DNA.
Protein Extraction and Immunoblotting-At least 24 h post transfection, tsA-201 cells were processed to extract recombinant protein. The Petri dishes were placed on ice, and cells were harvested with a rubber scraper and sedimented at 3500 rpm at 4°C for 20 min. Cells were washed once with 20 ml of ice-cold PBS to remove serum proteins. The cell pellet was resuspended in 500 l of lysis buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1 mM EGTA, 1 mM EDTA, 50 mM NaF, 20 mM ␤-glycerophosphate, 320 mM sucrose, complete protease inhibitor mixture (Roche Diagnostics), Halt TM phosphatase inhibitor mixture (Thermo Scientific), and pipetted up and down 10 times. The preparation was sedimented at 2500 rpm at 4°C for 5 min to remove the nuclear fraction. The supernatant was collected and sedimented at 14,000 rpm at 4°C for 30 min. The pellet was washed with 500 l of lysis buffer and sedimented again at 14,000 rpm at 4°C for 30 min to obtain pure pellet devoid of contaminants from the previous supernatant fraction. The pellet was resuspended in 20 l of 2ϫ sample buffer and heated at 50°C for 5 min. Proteins were resolved on 4 -20% SDS-PAGE gels, and immunoblotting was performed with the indicated antibodies.
Co-immunoprecipitation-TsA-201 cells were cultured and transfected as described above. For co-immunoprecipitation studies, the cells were processed as described previously (33).
Recombinant Protein Production-Purified recombinant CaMKII was a kind gift of Dr. Thomas Soderling (Oregon Health and Science University, Portland, OR (35)). GST-tagged Ca V 2.1(1848 -1964), Ca V 2.1(1959 -2035), or GST alone were expressed in Escherichia coli (BL21) in baffled shaker flasks. Isolated single colonies were inoculated and grown overnight in 50 ml of LB medium containing 100 l of ampicillin (100 mg/ml) at 37°C and 220 rpm to obtain precultures. Large-scale cultures were prepared using 400 ml of LB containing 800 l of ampicillin with the addition of 10 ml of overnight precultures. The cells were grown at 37°C and 220 rpm until the absorbance increased to 1.0 at 600 nm. Protein expression was induced by the addition of 400 l of 100 mM isopropyl 1-thio-␤-D-galactopyranoside for 14 h at 11°C. Cells were harvested by centrifugation at 3500 rpm, 4°C for 30 min, and washed 3 times by resuspension in 50 ml of cold PBS. The 15-ml cell suspension (in PBS) was incubated with 500 l of lysozyme (10 mg/ml) for 1 h at 4°C with shaking and subsequently sonicated four times at 20-s intervals. Cell-free supernatant was obtained by 2 rounds of centrifugation at 13,000 rpm at 4°C for 1 h. Glutathione-Sepharose beads (100 l) were washed 3 times in 50 ml of cold PBS and sedimented at 1500 rpm at 4°C for 1 min before use. Cell-free supernatant from the earlier step was incubated overnight with glutathione-Sepharose beads. Nonspecific binding was reduced by washing the bound protein six times with cold PBS. The washes were carried out for 30 min at 4°C on a shaker, and beads/resin were sedimented at 1500 rpm for 1 min at 4°C. The yield of fusion protein was estimated by Coomassie Blue staining after SDS-PAGE using a calibration curve with bovine serum albumin. These proteins are pure by SDS-PAGE analysis and are native with respect to binding of CaM as expected (Fig. 1).
For some experiments, CaMKII was expressed with a maltose-binding protein (MBP) epitope tag on its N terminus (36). MBP-CaMKII or MBP alone was expressed in E. coli (BL21) in baffled shaker flasks. Isolated single colonies were inoculated, grown overnight in 50 ml of LB medium containing 100 l ampicillin (100 mg/ml) at 37°C and 220 rpm to obtain precultures. Large-scale cultures were prepared using 400 ml of LB containing 800 l of ampicillin with the addition of 10 ml overnight precultures. The cells were grown at 37°C and 220 rpm until the absorbance increased to 1.0 at 600 nm. Protein expression was induced by the addition of 400 l of 100 mM isopropyl 1-thio-␤-D-galactopyranoside for 14 h at 11°C. Cells were harvested by centrifugation at 3500 rpm and 4°C for 30 min and washed 3 times by resuspension in 50 ml of cold PBS. The 15-ml cell suspension (in PBS) was incubated with 500 l of lysozyme (10 mg/ml) for 1 h at 4°C with shaking and subsequently sonicated 4 times at 20-s intervals. Cell-free supernatant was obtained by 2 rounds of centrifugation at 13,000 rpm and 4°C for 1 h. Amylose resin (100 l) was washed 3 times in 50 ml of cold PBS and sedimented at 1500 rpm and 4°C for 1 min before use. Cell-free supernatant from the earlier step was incubated overnight with amylose resin beads. Nonspecific binding was reduced by washing the bound protein six times with cold PBS. The washes were carried out for 30 min at 4°C on a shaker, and beads/resin were sedimented at 1500 rpm for 1 min at 4°C. The yield of fusion protein was estimated by Coomassie Blue staining after SDS-PAGE using a calibration curve with bovine serum albumin.
Binding Experiments-Binding of Ca V 2.1-GST fusion proteins and CaMKII was analyzed by GST pulldown assays using Tris binding buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 0.1% Tween 20, and 0.1% BSA). Autophosphorylation of Thr-286 in purified CaMKII was carried out by incubation on ice for 5 min in Tris binding buffer containing 0.025 mM CaCl 2 ,0.125 M CaM, 0.025 mM ATP, and 0.125 mM MgCl 2 (37). This procedure resulted in essentially complete phosphorylation of CaMKII as judged by its change in mobility in SDS-PAGE (Fig.  1C). The binding reaction was carried out for 1 h at 4°C. After incubation, washing was performed in binding or washing buffer with the addition of 5 mM EGTA where indicated. The proteins were washed 3 times for 5 min at 4°C by sedimentation at 500 rpm. The beads were boiled at 95°C in 10 l of 4ϫ sample buffer. Proteins were resolved on 10% SDS-PAGE, and immunoblotting was performed. Binding of CaMKII was detected using anti-CaMKII (monoclonal, BD Biosciences). CaMKII phosphorylation at Thr-286 was detected using antiphospho-CaMKII(Thr-286) (polyclonal, PhosphoSolutions). For re-probing of immunoblots, nitrocellulose membranes were stripped as described (38). Equal bait loading was confirmed by blotting with anti-GST (Sigma). Immunoblots (ECL detection) were documented with Chemidoc XRS (Bio-Rad).
Binding was quantified using densitometric measurement of band intensity using NIH ImageJ software.
Expression and Electrophysiological Recording in Cultured Neurons-Superior cervical ganglion (SCG) neurons were cultured as described to allow synapse formation (39). cDNAs encoding ␣ 1 2.1 subunit, eGFP, and (where indicated) CaMKIIN was microinjected into the nuclei of SCG neurons through glass micropipettes with 5% fast-green dye (Sigma). Entry of the injected reagents into the cell nucleus was monitored by the intensity of green dye in the nucleus. The cells were maintained at 37°C in a 95% air, 5% CO 2 -humidified incubator for 2-3 days.
Excitatory postsynaptic potentials (EPSPs) were recorded from SCG neurons cultured for 6 weeks as described (39). Injected neurons were identified with an inverted microscope equipped with an epifluorescence unit. Conventional intracellular membrane potential recordings were made from two neighboring neurons using microelectrodes filled with 1 M KAc (70 -90 m⍀). Paired action potentials were generated in the injected, presynaptic neuron expressing the ␣ 1 2.1 channels and eGFP by passing 1-2 nA of current for 5 ms through the intracellular recording electrode. EPSPs were recorded from a neighboring non-injected neuron. CaMKIIN was co-expressed with Ca V 2.1 channels or Ca V 2.1(1897-1912) was injected 30 min before recordings as indicated. Endogenous synaptic transmission was blocked by bath application of 3 M -conotoxin GVIA in a modified Krebs solution consisting of 136 mM NaCl, 5.9 mM KCl, 1 mM CaCl 2 , 1.2 mM MgCl 2 , 11 mM glucose, and 3 mM Na-HEPES, pH 7.4. For recording sub-threshold EPSPs, the membrane potential of the postsynaptic cell was held at Ϫ70 or Ϫ80 mV by passing current (0.2-0.4 nA) through the recording electrode. Each 30-s recording protocol was repeated 4 times for each inter-stimulus interval, and the ratio of the second EPSP to the first EPSP was averaged for each synapse. Data values with associated error shown in the text and figures represent the mean Ϯ S.E.

RESULTS
CaMKII Binds Directly to the C-terminal Domain of Ca v 2.1 Channels-As shown in previous studies, we confirmed that CaMKII binds to the C-terminal domain of Ca V 2.1 channels in transfected cells, as assessed by co-immunoprecipitation ( Fig.  2A) (33). However, it was not known whether this binding interaction was direct or required other intermediary proteins. To measure direct interactions between CaMKII and Ca V 2.1, we investigated binding of CaMKII in vitro to adjacent segments of the C-terminal domain that contain the two components of the calcium sensor protein interaction site: the IQ-like motif (Ca V 2.1(1848 -1964)) and the CaM binding domain (Ca V 2.1(1959 -2035)) (40,41). We prepared autophosphorylated CaMKII in which Thr-286 was essentially completely phosphorylated (Fig. 1C). We incubated purified preparations of nonactivated, Ca 2ϩ /CaM-treated, and autophosphorylated CaMKII with purified GST-tagged Ca V 2.1(1848 -1964), Ca V 2.1(1959 -2035), or GST alone in vitro and detected the complex by GST pulldown assay. Binding of autophosphorylated CaMKII to GST-Ca V 2.1(1848 -1964) (Fig. 2B, lane a3) was substantially greater than binding of nonactivated CaMKII Activation of CaMKII by Ca V 2.1 Channels FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 or CaMKII incubated with Ca 2ϩ /CaM but without ATP (Fig.  2B, lanes a1 and a2). In contrast, GST-tagged Ca V 2.1(1959 -2035) did not bind any of the three forms of CaMKII (Fig. 2B, lanes a4 -a6), and GST itself was also unable to bind CaMKII ( Fig. 2B, lanes a7-a9). Quantification of the results revealed that binding of the autophosphorylated kinase to GST-Ca V 2.1(1848 -1964) was ϳ8-fold greater than binding to the non-   1-3), syntaxin 1A (lanes 4 -6), or GST (lanes 7-9) were incubated with CaMKII (20 nM) and Mg 2ϩ (5 mM) in the presence/absence of Ca 2ϩ /CaM (5 M) and ATP (1 mM) as indicated. a, binding of CaMKII probed using anti-CaMKII. b, the same blot after stripping and re-probing using anti-GST to show loading of Ca V 2.1(1848 -1964), syntaxin 1A, or GST alone. The section of each blot containing GST-labeled proteins or GST itself was aligned for presentation even though the molecular weights are different. activated or Ca 2ϩ /CaM-treated kinase (Fig. 2C). These results demonstrate direct and specific binding of CaMKII to Ca V 2.1(1848 -1964) containing the IQ-like motif.
Location of the CaMKII Binding Site-Ca V 2.1(1848 -1964) contains an IQ-like motif as part of a bipartite regulatory site that is important for CaM-induced facilitation and inactivation of Ca V 2.1 channels (9). IQ-like motifs bind CaM, which could potentially serve as a docking site for CaMKII (43). On the other hand, CaM-independent docking of CaMKII to cardiac Ca V 1.2 channels is well described, suggesting that CaM is not required for CaMKII binding to Ca 2ϩ channels (37 (Fig. 3A, lane a4), but its binding to Ca V 2.1(1848 -1964)/EEDAAA was significantly reduced to 0.42 Ϯ 0.02 of WT when quantified using phospho-CaMKII-(Thr-286)-specific antibody (Fig. 3A, lane a5; Fig. 3B, p Ͻ 0.01). Autophosphorylated CaMKII did not bind to the GST control (Fig. 3A, lane a1), and non-activated CaMKII did not bind to either Ca V 2.1(1848 -1964)/TGVKIY or Ca V 2.1(1848 -1964)/  Samples were washed in the absence or presence of 5 mM EGTA as indicated. a, the blot was probed with anti-CaMKII antibody. b, the same blot after stripping and re-probing using anti-phospho-CaMKII(Thr-286). c, the same blot after stripping and reprobing using anti-CaM. d, the same blot after stripping and re-probing using anti-GST to show equal loading of Ca V 2.1 (1848 -1964). B, quantitation of relative CaMKII binding using anti-CaMKII under the indicated conditions (mean Ϯ S.E.; **, p Ͻ 0.01; ***, p Ͻ 0.001 by Student's t test; n ϭ 4).  (Fig. 3A, lanes a2 and a3), confirming the specificity of binding to phospho-CaMKII(Thr-286). These results indicate that the TGYKIY motif forms an important part of the binding site for CaMKII, but other nearby sequence elements must also contribute substantially to kinase binding when this sequence is mutated.

Activation of CaMKII by
Binding of Autophosphorylated CaMKII Persists after Ca 2ϩ / CaM Dissociation-Binding of CaMKII to Ca V 1.2 and Ca V 2.1 channels modulates their function (33,37). The local Ca 2ϩ concentration at the active zone is tightly regulated to create a Ca 2ϩ nanodomain (9). Because of its slowly reversible autophosphorylation, activation of CaMKII can integrate repetitive Ca 2ϩ signals and serve as a molecular memory of synaptic activity (15). Therefore, it is important to test the persistence of the kinasechannel interaction as the level of Ca 2ϩ declines. To address this question, we chelated Ca 2ϩ using 5 mM EGTA in the washing buffers. CaMKII was isolated by binding to Ca V 2.1(1848 -1964) immobilized to glutathione-Sepharose beads, and the binding of CaMKII was analyzed in the absence or presence of 5 mM EGTA in the washing buffer (Fig. 4A). CaMKII is highly phosphorylated in the presence of Ca 2ϩ /CaM and ATP. Autophosphorylated CaMKII bound more effectively to Ca V 2.1(1848 -1964) (Fig. 4A, lane a3) than nonactivated or Ca 2ϩ /CaM-treated CaMKII (Fig. 4A, lanes a1 and a2). Both Ca 2ϩ /CaM and ATP are essential components in triggering the autophosphorylation reaction, and omission of ATP followed by chelation of Ca 2ϩ using EGTA (5 mM) in the washing reac-tion reduced binding significantly (Fig. 4A, lane a4). However, binding of the autophosphorylated kinase persisted even after Ca 2ϩ /CaM was removed from the kinase-channel complex by washes with EGTA-containing washing buffer (Fig. 4A, lane  a5), suggesting that Ca 2ϩ /CaM removal does not rapidly reverse binding of autophosphorylated CaMKII. The efficacy of chelation of Ca 2ϩ /CaM by EGTA treatment was probed using anti-CaM. No bound CaM was detected after EGTA treatment (Fig. 4A, lanes c4 -c5). We quantified the relative CaMKII binding in each of these conditions and found significant retention of bound CaMKII after EGTA treatment (Fig. 4B). Collectively, these results show that binding of CaMKII to Ca V 2.1(1848 -1964) is greatly enhanced by Ca 2ϩ /CaM-dependent autophosphorylation of CaMKII, but once formed, the complex persists even after removal of Ca 2ϩ /CaM.
Dephosphorylation of CaMKII Does Not Reverse Binding-To test whether autophosphorylation is required to maintain binding of CaMKII to Ca V 2.1(1848 -1964), we dephosphorylated the channel-bound CaMKII and tested the fate of kinase-channel interaction. Protein phosphatases PP1, PP2A, PP2B, and PP2C represent the majority of serine/threonine phosphatase activity in brain and other tissues (44). CaMKII in isolated postsynaptic densities was mostly dephosphorylated by PP1 (44). Autophosphorylated CaMKII preferentially bound to Ca V 2.1(1848 -1964) (Fig. 5A, lane a1). Dephosphorylation of autophosphorylated CaMKII by PP1 treatment before incuba- Washes were carried out in the presence or absence of 5 mM EGTA as indicated. A, binding of CaMKII to Ca V 2.1(1848 -1964) GST was detected by GST pulldown assay. a, the blot was probed with anti-CaMKII. b, the same blot is shown after stripping and re-probing using anti-phospho-CaMKII(Thr-286). c, the same blot after stripping and re-probing using anti-GST to show equal loading of Ca V 2.1 (1848 -1964). B, binding of CaMKII to GST alone (control) was detected by GST pulldown assay. a, the blot was probed with anti-CaMKII. b, the same blot after stripping and re-probing using anti-phospho-CaMKII-(Thr-286). c, the same blot after stripping and re-probing using anti-GST to show equal loading of GST. tion of CaMKII with the channel peptide or both before and after incubation greatly reduced its binding (Fig. 5A, lane a2  and a3), confirming that autophosphorylation is prerequisite for CaMKII binding to the channel. The remaining faint signal in the absence of autophosphorylation is in accordance with our earlier finding of a low level of binding of nonactivated or Ca 2ϩ /CaM-treated CaMKII to the channel (Fig. 2B). Surprisingly, dephosphorylation of autophosphorylated CaMKII by post-PP1 treatment did not destabilize its binding (Fig. 5A, lane  a4).Evidently,oncekinaseisboundtothechannel,dephosphorylation does not reverse its binding, even when dephosphorylation is essentially complete (Fig. 5A, lanes b2-b5). Combining post-PP1 treatment and washing the kinase-channel complex with EGTA-containing buffer also did not reverse CaMKII binding (Fig. 5A, lane a5). Together these results show that once the autophosphorylated CaMKII is bound to Ca V 2.1 (1848 -1964), the interaction cannot be rapidly reversed by dephosphorylation (Fig. 5A, lane a4), Ca 2ϩ /CaM removal (Fig. 5A, lane a6), or both processes carried out simultaneously in vitro (Fig. 5A, lane a5). GST alone does not have any effect (Fig. 5B), further strengthening evidence for the specificity of these interactions.
Increased Autophosphorylation of CaMKII Bound to Ca V 2.1(1848 -1964)-Binding of CaMKII to its target sites on NMDA-type glutamate receptors and K V 11 channels causes Ca 2ϩ -independent activation of the kinase (45,46). Although there is no detectable amino acid sequence similarity between Ca V 2.1 channels, K V 11 channels, and glutamate receptors, expression of the C-terminal tail of Ca V 2.1 channels, Ca V 2.1(1766 -2212), without the pore domain leads to enhancement of autophosphorylation of Thr-286 on endogenous CaMKII in human embryonic kidney tsA-201 cells (Fig.  6A, lane a2). These results suggest that binding of CaMKII to Ca V 2.1 channels per se may be sufficient to activate the kinase and increase its autophosphorylation substantially. The addition of ionomycin, which increases entry of Ca 2ϩ into the cytosol, resulted in a comparable increase in CaMKII autophosphorylation on Thr-286 (Fig. 6A, lane a3). Increased cytosolic Ca 2ϩ activates the Ca 2ϩ -dependent phosphoprotein phosphatase calcineurin (47), which might limit the level of autophosphorylation induced by ionomycin. However, in the presence of ionomycin (5 M) and the calcineurin inhibitor cyclosporin A (5 M), there was little increase in autophosphorylation compared with ionomycin alone (Fig. 6A, lane a4), suggesting that calcineurin does not play a major role in control of CaMKII phosphorylation in tsA-201 cells. Quantification of these results revealed an ϳ6-fold increase in autophosphorylation upon transfection of Ca V 2.1(1766 -2212) or treatment with ionomycin plus cyclosporin A (Fig. 6B).
To extend these findings of increased autophosphorylation of CaMKII bound to Ca v 2.1 channels in transfected tsA-201  1(1848 -1964). a, the blot was probed with anti-phospho-CaMKII(Thr-286). b, the same blot after stripping and re-probing using anti-CaMKII.  2(1848 -1964). a, the blot was probed with anti-phospho-CaMKII(Thr-286). b, the same blot after stripping and re-probing using anti-CaMKII. D, the quantification of relative CaMKII autophosphorylation using anti-phospho-CaMKII(Thr-286) under the indicated conditions (mean Ϯ S.E.; not significant by Student's t test; n ϭ 3). FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7 cells to pure proteins, we reconstituted the kinase-channel dimeric complex in vitro and measured autophosphorylation. We activated CaMKII in the presence of Ca 2ϩ /CaM, ATP, and Mg 2ϩ and studied the autophosphorylation levels in the absence or presence of the Ca V 2.1(1848 -1964) peptide (Fig.  7A). Autophosphorylation of CaMKII increased in the presence of increasing concentrations of the Ca v 2.1 channel peptide (Fig.  7Aa, lanes 1-4) and then decreased at higher concentrations of the peptide (Fig. 7Aa, lanes 4 -6). Quantification of the levels of CaMKII autophosphorylation using anti-phospho-CaMKII-(Thr-286) (Fig. 7B) showed that direct binding of Ca V 2. 1(1848 -1964) in vitro mimics the activation of kinase autophosphorylation observed in transfected tsA-201 cells. These results confirm that direct binding of Ca V 2.1(1848 -1964) activates CaMKII autophosphorylation. As negative controls, we used a GST-tagged protein of the same size from the C-terminal of Na v 1.2 channels (Fig. 7, C and D) or GST itself (data not shown), and we found no effect on CaMKII activity, further supporting the specificity of this interaction.

Activation of CaMKII by Ca V 2.1 Channels
Phosphorylation of Synapsin-1 by CaMKII Bound to Ca V 2.1(1766 -2212)-Synapsin-1 is a major presynaptic phosphoprotein that is a prominent substrate for CaMKII, and phosphorylation by CaMKII regulates the effects of synapsin-1 on synaptic vesicle trafficking (23). Phosphorylation of synapsin-1 by CaMKII substantially increases synaptic transmission at the squid giant synapse (28,29). Expression of Ca V 2.1(1766 -2212) with synapsin-1 in tsA-201 cells led to a substantial increase in synapsin-1 phosphorylation at Ser-603 (Fig. 8A, lane a3) compared with untransfected tsA-201 cells (Fig. 8A, lane  a1) or cells expressing synapsin-1 alone (Fig. 8A, lane a2). Increasing the cytosolic Ca 2ϩ concentration with ionomycin, which triggers CaMKII autophosphorylation, also led to a significant increase in synapsin-1 phosphorylation at Ser-603 (Fig.  8, lane a4), and these levels are comparable to those observed when synapsin-1 is coexpressed with Ca V 2.1(1766 -2212). Ionomycin treatment of tsA-201 cells co-expressing Ca V 2.1(1766 -2212), and synapsin-1 shows further enhancement in synapsin-1 phosphorylation (Fig. 8, lane a5). These results indicate that the C-terminal domain of Ca V 2.1 channels stimulates activation and autophosphorylation of CaMKII as effectively as Ca 2ϩ /CaM, and this leads to phosphorylation of synapsin-1 at Ser-603 and potentially to phosphorylation of other presynaptic substrates.
If binding of Ca V 2.1 to CaMKII can lead to phosphorylation of synapsin-1, a stable ternary complex of Ca V 2.1 and synapsin-1 bound to CaMKII may be formed. To test this possibility we expressed CaMKII in bacteria with a MBP epitope tag and purified the resulting fusion protein. MBP-CaMKII attached to amylose resin was able to bind both Ca V 2.1(1848 -1964) and synapsin-1 simultaneously (Fig. 9), whereas control experiments with MBP showed no binding (Fig. 9B). Formation of this ternary complex in presynaptic terminals would allow local phosphorylation of synapsin-1 by CaMKII bound to Ca V 2.1 channels to modulate the dynamics of synaptic vesicle function in active zones containing these proteins. As CaMKII is a dodecamer, this ternary complex may be formed within a single subunit or may reflect binding of Ca V 2.1 and synapsin-1 to different subunits in the CaMKII complex.
Functional Role of Interaction of CaMKII with Ca V 2.1 Channels in Synaptic Plasticity-Binding of CaMKII to Ca V 2.1 channels enhances their functional activity by inhibiting their inactivation (33) and enhances the activity of CaMKII by increasing its autophosphorylation and its phosphorylation of other substrates as shown above. To critically test the potential effects of this specific interaction on synaptic transmission, it is necessary to manipulate the activity of CaMKII bound specifically to Ca V 2.1 channels in the presynaptic terminal without altering the functional activity of CaMKII in the postsynaptic compartment or CaMKII in other locations in the presynaptic terminal. Accordingly, we expressed Ca V 2.1 channels in SCG neurons in cell culture by microinjection of cDNA into the nucleus of a single cell using methods that were well defined in previous work (42,48,49). After 2-3 days, we impaled the injected cell and a nearby uninjected postsynaptic partner, and we measured synaptic transmission driven exclusively by the transfected Ca V 2.1 channels by blocking endogenous N-type Ca 2ϩ currents with -conotoxin GVIA. We studied paired-pulse facilitation as a measure of short term synaptic plasticity. In this experimental paradigm, the synaptic response to the second stimulus in the pair is enhanced by residual Ca 2ϩ remaining in the nerve terminal from the first stimulus (9). Paired-pulse facilitation of synaptic transmission in this transfected SCG neuron preparation is primarily caused by facilitation of Ca V 2.1 channel activity by Ca 2ϩ /CaM binding to the Ca 2ϩ sensor protein binding site in the C-terminal domain (48). As illustrated in Fig. 10A, Ca V 2.1 channels expressed alone generate synaptic transmission in which the paired-pulse ratio is highly dependent on the inter-stimulus interval (ISI) between the paired pulses. At short ISI, synaptic depression is dominant, and paired-pulse ratio values are less than 1.0. At longer ISI, synaptic facilitation becomes dominant, peaks at ϳ1.75 for an ISI of 80 ms, and declines to 1.0 at long ISI (Fig. 10A). Perfusion of a competing peptide that blocks the interaction of CaMKII with Ca V 2.1 channels (Ca V 2.1(1897-1912)) prevented both paired-pulse facilitation and paired-pulse depression at this model synapse (Fig. 10A), suggesting that binding of CaMKII to Ca V 2.1 channels is required for expression of this regulatory effect. Similarly, expression of the brain-specific CaMKII inhibitor CaMKIIN (50), which prevents CaMKII binding to Ca V 2.1 channels (33), also prevented paired-pulse facilitation and depression (Fig.  10A). This is consistent with previous results showing that facilitation of Ca V 2.1 channels expressed in tsA-201 cells also requires binding of CaMKII (33). It is unlikely that the basal release probability is affected by competing peptide injection or CaMKIIN expression because the mean amplitudes of the first EPSPs are unchanged (Fig. 10B). Because the competing peptide Ca V 2.1(1897-1912) applied acutely through the recording pipette and CaMKIIN expressed from cDNA both reduce paired-pulse facilitation of synaptic transmission, as expected from their inhibition of paired-pulse facilitation of Ca V 2.1 current, these results support the conclusion that these are specific effects. To further support the specificity of action of these agents, we applied Ca V 2.1(1897-1912) through the recording pipette and CaMKIIN by expression of injected cDNA and analyzed their effects on synaptic transmission initiated by endogenous Ca V 2.2 channels. We found no effect of either of these agents (Fig. 10, C and D), further supporting the specificity of their effects on facilitation of Ca V 2.1 currents and neurotransmission initiated by Ca V 2.1 channels. Evidently, binding of  1(1848 -1964), CaMKII, and synapsin-1. CaMKII was expressed as a MBP fusion protein and immobilized on amylose resin. A, incubations were carried out as depicted in the figure. a, binding of synapsin-1 was detected using anti-synapsin-1 antibody (PhosphoSolutions, polyclonal, 1;5000 TBST). b, shown is the same blot as a but after stripping and reprobing with anti-GST to show the binding of Ca V 2. 1(1848 -1964). c is the same blot as a and b but after stripping and reprobing with anti-CaMKII (Invitrogen, monoclonal, 1:1000 in TBST) to show the CaMKII bait. B, incubations were carried out as depicted in figure. a, binding of synapsin-1 was detected using anti-synapsin-1 antibody (Phospho-Solutions, polyclonal, 1;5000 TBST). b, shown is the same as a but after stripping and reprobing with anti-GST to show the binding of Ca V 2. 1(1848 -1964), and c is same blot as a and b but after stripping and reprobing with anti-CaMKII (Invitrogen, monoclonal, 1:1000 in TBST) to show the CaMKII bait. FIGURE 10. Binding of CaMKII to Ca V 2.1 channels is required for short term synaptic plasticity. A, SCG neurons were cultured and injected with cDNA encoding Ca V 2.1 channels as described under "Experimental Procedures." Where indicated, CaMKIIN was co-expressed with Ca V 2.1 channels, or Ca V 2.1(1897-1912) was injected through a whole-cell patch electrode 30 min before recording as described (39). Sharp microelectrode impalements were made in the previously transfected, presynaptic neuron and a neighboring, synaptically connected neuron. Action potentials were generated in the presynaptic cell, and EPSPs were recorded in the postsynaptic cell and analyzed as described under "Experimental Procedures." The paired-pulse ratio (PPR) was plotted against inter-stimulus interval (mean Ϯ S.E.; *, Ͻ 0.05, Student's t test, n ϭ 6 -9). WT, green; CaMKIIN, blue; Ca V 2.1(1897-1912), orange. B, amplitudes of the first EPSP in paired pulse experiments are shown. C, a similar experiment to that described in panel A was carried out with Ca V 2.1(1897-1912) injected through a whole-cell patch electrode in untransfected SCG neurons, and paired-pulse facilitation of neurotransmission initiated by the endogenous Ca V 2.2 channels was measured in the absence of -conotoxin GVIA. D, a similar experiment to that described in panel A was carried out with expression of CaMKIIN in untransfected SCG neurons, and paired-pulse facilitation of neurotransmission initiated by the endogenous Ca V 2.2 channels was measured in the absence of -conotoxin GVIA. Normalized paired-pulse ratios of control and CaMKIIN-expressing neurons are plotted. FEBRUARY 15, 2013 • VOLUME 288 • NUMBER 7

JOURNAL OF BIOLOGICAL CHEMISTRY 4645
CaMKII by Ca V 2.1 channels is required for both up-regulation of channel activity in paired pulses and for Ca 2ϩ -independent activation of CaMKII by Ca V 2.1, and one or both of these effects is necessary for normal short term synaptic plasticity.

DISCUSSION
CaMKII Binds Directly to Ca V 2.1 Channels-Previous studies showed that CaMKII can be co-immunoprecipitated with Ca V 2.1 channels from transfected cells and that binding of CaMKII per se was sufficient for up-regulation of Ca V 2.1 channel activity in transfected cells and neurons (33), but no evidence was provided for direct interaction of these two key Ca 2ϩ -signaling proteins. Our results show that CaMKII does indeed bind directly to the C-terminal domain of Ca V 2.1 channels at an interaction site located in Ca V 2.1(1848 -1964). Autophosphorylation enhances binding of CaMKII, and autophosphorylated CaMKII remains bound to the C-terminal domain of Ca V 2.1 channels even after dephosphorylation and removal of the original Ca 2ϩ /CaM stimulus for binding. Thus, Ca V 2.1 channels with bound CaMKII are likely to serve as a signaling complex in the presynaptic active zone in neurons.
Binding to Ca V 2.1 Channels Induces Sustained Ca 2ϩ -independent Activity of CaMKII-CaMKII activation normally requires binding of Ca 2ϩ /CaM, but previous studies have demonstrated Ca 2ϩ /CaM-independent activation of CaMKII by interaction with NMDA-type glutamate receptors and K V 11 channels (45,46). Our results show that binding of CaMKII to Ca V 2.1 channels persistently activates its catalytic activity, as measured by autophosphorylation. CaMKII remains activated even after removal of the Ca 2ϩ /CaM stimulus. This is a provocative result, as it implies that CaMKII bound to Ca V 2.1 channels is poised to phosphorylate nearby substrates and thereby regulate synaptic transmission locally.
It is interesting to compare the regulatory effects of CaM and CaMKII on Ca V 2.1 channels and Ca V 1.2 channels. These two types of Ca 2ϩ channels are less than 50% identical in amino acid sequence and serve different physiological roles: Ca V 2.1 in initiation of synaptic transmission versus Ca V 1.2 in initiation of excitation-contraction coupling in muscle and in postsynaptic regulation in neurons (8). Although both proteins bind CaM and CaMKII to nearby sites in their C-terminal domains, the regulatory consequences are quite different. Binding of Ca 2ϩ / CaM to Ca V 2.1 channels causes facilitation followed by inactivation, whereas binding to Ca V 1.2 channels causes only Ca 2ϩdependent inactivation (9,51). Binding of CaMKII to Ca V 2.1 channels enhances their activity and their facilitation, whereas phosphorylation of Ca V 1.2 channels by CaMKII bound to the C-terminal domain and/or the Ca V ␤2 subunit is required for facilitation of their activity (9,37,52,53). The structural and mechanistic basis for this differential regulation of Ca V 1.2 and Ca V 2.1 channels by CaM and CaMKII bound to nearby sites in their C-terminal domains is an interesting area for further research.
CaMKII Bound to Ca V 2.1 Channels Phosphorylates Synapsin-1-Synapsin-1 is abundant in presynaptic terminals, where it tethers synaptic vesicles to the actin cytoskeleton and is required for normal replenishment of synaptic vesicles during periods of high synaptic activity (54). Actin surrounds clusters of synaptic vesicles in presynaptic terminals and concentrates synapsin-1 at vesicle clusters (55). It also plays an important role in the dynamics of synaptic vesicle transfer to the readily releasable pool that is poised for rapid exocytosis, and both synapsin-1 and CaM are involved in those processes (55)(56)(57). Synapsin-1 is phosphorylated at Ser-603, which regulates trafficking of synaptic vesicles in vivo (54). Our results show that CaMKII bound to Ca V 2.1 channels is effective in phosphorylating Ser-603 in the absence of stimulation by Ca 2ϩ /CaM. In the nerve terminal, phosphorylation of Ser-603 detaches synapsin-1 from synaptic vesicles and renders the vesicles mobile (58,59). Thus, CaMKII bound to Ca V 2.1 channels may phosphorylate synapsin-1 nearby and regulate synaptic vesicle dynamics in and near the active zones in the presynaptic terminal.
Binding of CaMKII to Ca V 2.1 Channels Is Required for Short Term Synaptic Plasticity-Recent results from studies of neurotransmission at the Calyx of Held, a large synapse in the auditory system, and in cultured SCG neurons show that Ca 2ϩ -dependent facilitation and inactivation of Ca V 2.1 channel activity contribute substantially to short term synaptic facilitation and depression (9). Our results show that binding of CaMKII to Ca V 2.1 channels is required for short term synaptic plasticity in SCG neurons. Block of CaMKII binding with a competing peptide from its Ca V 2.1 binding site inhibits short term facilitation and depression of synaptic transmission, as does binding of the brain-specific CaMKII inhibitor CaMKIIN. Evidently, CaMKII binding to Ca V 2.1 channels is a necessary prerequisite for short term synaptic plasticity mediated by Ca V 2.1 channels, as observed previously in studies of synaptic plasticity in genetically modified mouse strains (31). Binding of CaMKII to Ca V 2.1 channels may play a permissive role by enhancing the activation of Ca V 2.1 channels in response to trains of depolarizing stimuli and the resulting influx of Ca 2ϩ , because binding of the kinase does not activate or facilitate channel activity by itself (33). In addition to the role of CaMKII binding to Ca V 2.1 channels in short term plasticity demonstrated here, it is possible that phosphorylation by CaMKII bound to Ca V 2.1 channels may also be essential in the longer-term effects of synapsin-1 in regulating synaptic vesicle dynamics and synaptic transmission in the local environment of Ca V 2.1 channels at active zones.
Functional Roles of the Presynaptic Ca V 2.1 Signaling Complex-Previous studies show that Ca V 2.1 channels are regulated by binding of SNARE proteins, G proteins, CaM and CaM-like calcium sensor proteins, and CaMKII (9). Proteomic analysis revealed a complex of ϳ100 proteins associated with Ca V 2.1 channels in isolated nerve terminals from the mouse brain (10). This large protein complex serves to bring the essential machinery for neurotransmitter release close to presynaptic Ca V 2.1 channels, which provide the trigger of Ca 2ϩ entry to initiate rapid exocytosis. It also serves to regulate the activity of Ca V 2.1 channels in response to Ca 2ϩ and other regulatory messengers. Prior binding of CaMKII to Ca V 2.1 channels is required for facilitation and inactivation of the Ca V 2.1 channel during trains of repetitive depolarizations or action potentials. In this way, CaMKII binding to Ca V 2.1 serves as a molecular switch to turn on or off the millisecond timescale modulation of channel activity by Ca 2ϩ -dependent facilitation and inactivation.
The Effector Checkpoint Model for Calcium Channel Regulation-Voltage-gated Ca 2ϩ channels are regulated by their effectors such that the channels are more active when the effectors of their Ca 2ϩ signal are bound. Examples include regulation of the skeletal muscle Ca 2ϩ channel by the ryanodinesensitive Ca 2ϩ release channel (60), its effector in excitationcontraction coupling, and regulation of presynaptic Ca 2ϩ channels by SNARE proteins, which are the effectors for Ca 2ϩdependent exocytosis (9). Regulation of Ca v 2.1 channels by CaMKII also fits this regulatory theme (33). Binding of CaMKII to Ca v 2.1 increases the activity of both binding partners, and their interaction is required for facilitation of synaptic transmission and perhaps for other aspects of presynaptic function. Enhancement of the activity of Ca 2ϩ channels whose effectors are bound would focus Ca 2ϩ entry and Ca 2ϩ -dependent protein phosphorylation in locations where it can effectively generate a cellular response via local Ca 2ϩ signaling. This mechanism would enhance local signal transduction and reduce ineffective Ca 2ϩ entry and protein phosphorylation at other sites.