Calretinin Regulates Ca2+-dependent Inactivation and Facilitation of Cav2.1 Ca2+ Channels through a Direct Interaction with the α12.1 Subunit*

Background: Ca2+-dependent inactivation and facilitation of Cav2.1 Ca2+ channels are major determinants of neuronal excitability and synaptic plasticity. Results: The Ca2+-binding protein calretinin interacts with Cav2.1 and inhibits Ca2+-dependent inactivation and enhances facilitation of Cav2.1. Conclusion: In addition to its role as a diffusible Ca2+ buffer, calretinin can interact with targets such as Cav2.1 and modulate their function. Significance: Calretinin-Cav2.1 interactions may shape Ca2+ signaling dynamics in neurons. Voltage-gated Cav2.1 Ca2+ channels undergo dual modulation by Ca2+, Ca2+-dependent inactivation (CDI), and Ca2+-dependent facilitation (CDF), which can influence synaptic plasticity in the nervous system. Although the molecular determinants controlling CDI and CDF have been the focus of intense research, little is known about the factors regulating these processes in neurons. Here, we show that calretinin (CR), a Ca2+-binding protein highly expressed in subpopulations of neurons in the brain, inhibits CDI and enhances CDF by binding directly to α12.1. Screening of a phage display library with CR as bait revealed a highly basic CR-binding domain (CRB) present in multiple copies in the cytoplasmic linker between domains II and III of α12.1. In pulldown assays, CR binding to fusion proteins containing these CRBs was largely Ca2+-dependent. α12.1 coimmunoprecipitated with CR antibodies from transfected cells and mouse cerebellum, which confirmed the existence of CR-Cav2.1 complexes in vitro and in vivo. In HEK293T cells, CR significantly decreased Cav2.1 CDI and increased CDF. CR binding to α12.1 was required for these effects, because they were not observed upon substitution of the II-III linker of α12.1 with that from the Cav1.2 α1 subunit (α11.2), which lacks the CRBs. In addition, coexpression of a protein containing the CRBs blocked the modulatory action of CR, most likely by competing with CR for interactions with α12.1. Our findings highlight an unexpected role for CR in directly modulating effectors such as Cav2.1, which may have major consequences for Ca2+ signaling and neuronal excitability.

increases CDF and inhibits CDI (17). CaBP1 and VILIP-2 bind to the same sites as CaM in the C-terminal domain of ␣ 1 2.1, but structural differences from CaM underlie their distinct modulation of Ca v 2.1 (17,18).
Within the EF-hand superfamily of Ca 2ϩ -binding proteins, CaM, CaBP1, and VILIP-2 are considered Ca 2ϩ sensors, so were defined by their abilities to undergo Ca 2ϩ -dependent conformational change and interact with effectors. In contrast, Ca 2ϩ -binding proteins such as parvalbumin and calbindin D-28k generally act as diffusible Ca 2ϩ buffers that modulate cytoplasmic Ca 2ϩ signals (19). Like Ca 2ϩ chelation by EGTA (7), parvalbumin and calbindin D-28k can significantly alter CDI of Ca v 2.1 channels in transfected HEK293T cells (20). However, because parvalbumin and calbindin D-28k were not reported to associate directly with ␣ 1 2.1, their actions on Ca v 2.1 CDI are likely to be indirect and are primarily due to their actions as diffusible Ca 2ϩ buffers (20).
Calretinin (CR) is a EF-hand Ca 2ϩ -binding protein that is highly expressed in cerebellar granule neurons where Ca v 2.1 channels are also expressed (19,21,22). Like parvalbumin and calbindin D-28k, CR is considered mainly as a Ca 2ϩ signal modulator in neurons. Based on its Ca 2ϩ chelating properties, CR can significantly shape the spatiotemporal properties of Ca 2ϩ signals generated by plasma membrane and intracellular Ca 2ϩ channels (23), yet CR undergoes large conformational changes upon binding to Ca 2ϩ (24,25), which may mediate Ca 2ϩ -dependent interactions with integral membrane proteins (26). While investigating the potential targets of CR, we discovered a consensus CR-binding motif in ␣ 1 2.1. Ca 2ϩ -dependent binding of CR to this region nullifies CDI and enhances CDF in transfected HEK293T cells, which suggests that local tethering of CR to ␣ 1 2.1 is required for channel modulation. Our findings provide the first evidence that CR can directly alter Ca 2ϩ signaling through interactions with effectors, thus raising new possibilities for how CR may modulate neuronal and network excitability (27,28).

EXPERIMENTAL PROCEDURES
cDNAs and Molecular Biology-The following Ca v 2.1 subunit cDNAs were used: ␣ 1 2.1 (rbA isoform (29)), ␤ 2A (30), and ␣ 2 ␦ (31). GST-␣ 1 2.1 fusion proteins containing one or more CR-binding (CRB) motifs were generated by PCR and subcloning of the corresponding sequence in EcoRI/XhoI sites of pGEX-4T1 (GE Healthcare). For the chimeric Ca 2ϩ channel ␣ 1 2.1-1.2 subunit, amino acids 864 -983 from rat brain ␣ 1 2.1 were replaced by amino acids 807-923 of rat brain ␣ 1 1.2 (rbCII (32)). The corresponding DNA sequence of ␣ 1 1.2 was amplified by PCR and cloned into AscI/SgrAI sites of the ␣ 1 2.1-containing expression plasmid. GFP-CR was generated by cloning a PCR fragment containing the cDNA coding for human CR (836 bp) in the plasmid pEGFP-C1 (Invitrogen) using CR-specific primers comprising EcoRI and BamHI sites; the sequence of the insert was verified prior to use in experiments. For the expression construct containing the first three CRB motifs of ␣ 1 2.1 (mcherry-CRB1-3), the corresponding sequence (amino acids 899 -953) and Kozak sequence was amplified from rat brain ␣ 1 2.1 by PCR and subcloned into XhoI and HindIII restriction sites of pEGFPN1-mcherry (a kind gift from S. England). All of the cDNA constructs were subject to sequencing prior to use in experiments.
Phage Display and Binding Assays-Purified human recombinant CR-coated 7.5-cm 2 plastic plates and the Ph.D. TM -7 phage display peptide library kit (New England Biolabs Inc., Biocencept, Allschwil, Switzerland) were used according to the manufacturer's protocol with three rounds of panning. The cDNA coding for the surface-exposed heptapeptides (part of the pIII coat protein) was extracted from Escherichia coli strain ER2738, amplified by PCR, and subject to DNA sequencing (Microsynth, GmbH, Balgach, Switzerland).
For experiments with GFP-CR expressed in HEK293T cells, transfected cells were harvested and homogenized in 1 ml of ice-cold cell lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, 0.25% (w/v) sodium deoxycholate, 1 mM EDTA, pH 7.4, and protease inhibitors) containing either 1.6 mM CaCl 2 or 1 mM EGTA. The homogenate was rotated at 4°C for 1 h to solubilize membrane proteins, and the insoluble material was separated by centrifugation at 16,100 ϫ g (30 min). The supernatant (300 l) was incubated with 40 l of a 50% slurry of immobilized GST-CRB1-3 brought to 1 ml with the lysis buffer at 4°C overnight. The beads were washed three times with 1 ml of ice-cold lysis buffer, and the bound proteins were eluted with SDS-containing sample buffer, subjected to SDS-PAGE, and transferred to nitrocellulose. Mouse monoclonal antibodies anti-GFP (1:3000; Santa Cruz Biotechnology, Santa Cruz, CA) were used to detect bound GFP-CR by Western blot.
Coimmunoprecipitation-For coimmunoprecipitation from transfected HEK293T cells, transfected cells were harvested 48 h after transfection. The cell lysates were prepared as described in binding assays and incubated with 5 g of CR antibodies (Swant) and 40 l of protein A-Sepharose (50% slurry) overnight, rotating at 4°C. After three washes with 1 ml of cell lysis buffer, the proteins were eluted and analyzed by SDS-PAGE. Coimmunoprecipitated proteins were detected by Western blotting with anti-␣ 1 2.1 antibodies (1:300; Alomone Labs, Jerusalem, Israel) and anti-GFP antibodies (1:3000; Santa Cruz Biotechnology).
For coimmunoprecipitation from mouse brain, the cerebellum was homogenized in 1 ml of 250STMDPS buffer (250 mM sucrose, 50 mM Tris-HCl, 5 mM MgCl 2 , 1 mM DTT, pH 7.4, and protease inhibitors). The nuclear fraction was removed by centrifugation at 800 ϫ g for 15 min. The membrane fraction was separated from the cytosolic fraction by ultracentrifugation at 100,000 ϫ g for 1 h. The membrane pellet was solubilized in 1 ml of solubilization buffer (Tris-buffered saline (50 mM Tris-HCl, 150 mM NaCl, pH 7.5), 1% Triton X-100, and protease inhibitors) at 4°C for 30 min, and insoluble material was removed by ultracentrifugation at 100,000 ϫ g for 1 h. Either 5 g of rabbit IgG (Invitrogen) or anti-CR antibodies (Swant, Marly, Switzerland) were added to the solubilized membrane proteins along with 50 l of protein A-Sepharose (50% slurry; Sigma-Aldrich). The reactions were continued overnight with end over end rotation at 4°C. The resin was collected by centrifugation and rinsed three times with 1 ml of solubilization buffer. Bound proteins were eluted and resolved by 4 -12% SDS-polyacrylamide gel and Western blotting with anti-CR antibodies (1:5000; Swant) and anti-␣ 1 2.1 antibodies as described above.
Electrophysiology of Transfected HEK293T Cells-Whole cell patch clamp recordings were acquired 36 -60 h of post-transfection with a HEKA (Lambrecht/Pfalz, Rheinland-Pfalz, Germany) EPC-9 patch clamp amplifier. External recording solution contained 150 mM Tris, 1 mM MgCl 2 , and 10 mM CaCl 2 or BaCl 2 . Internal solution contained 140 mM N-methyl-D-glucamine, 10 mM HEPES, 2 mM MgCl 2 , 2 mM Mg-ATP, and 0.5 mM EGTA. The pH of both solutions was adjusted to 7.3 with methanesulfonic acid. Electrode resistances were 1-2 M⍀ in the bath solution, and series resistance was ϳ2-4 M⍀, compensated 60 -80%. The membrane potential was held at Ϫ60 mV prior to action potential waveform protocols or to Ϫ80 mV prior to all other experiments. The acquired data were analyzed using Igor Pro software (Wavemetrics), and statistics were performed using SigmaPlot (Systat Software). All of the averaged data are presented as the means Ϯ S.E. In the figures, numbers of cells (n) are indicated in parentheses.

RESULTS
CR Interacts with Ca v 2.1-While screening a heptapeptide phage display library for CR-interacting proteins, we identified a consensus CR-binding sequence, H(R/K)HRRR(E/D), consisting of five or six basic residues (His/Arg) flanked by an acidic (Glu/Asp) residue (Fig. 1A). Bioinformatic analysis revealed that this CRB sequence was repeated five times in the cytoplasmic loop between domains 2 and 3 of ␣ 1 2.1 (Fig. 1B). The CRBs are C-terminal to the synaptic protein interaction site ("synprint") (33) and are highly conserved in ␣ 1 2.1 between species. The CRB-containing region is also present in the Ca v 2.2 ␣ 1 subunit (␣ 1 2.2; supplemental Fig. S1) but absent in Ca v 1 ␣ 1 subunits. To test whether the CRB(s) directly interact with CR, pulldown assays were performed with GST fusion proteins containing the first three CRBs in ␣ 1 2.1 (GST-CRB1-3). CR bound to GST-CRB1-3, but not to control GST, in the presence of Ca 2ϩ (1.6 mM). Binding of CR was Ca 2ϩ -dependent because it was prevented when free Ca 2ϩ was chelated with EGTA (1 mM; Fig. 1C). To estimate the affinity of CR binding to ␣ 1 2.1, GST-CRB1-3 was incubated with varying amounts of fluorescently tagged CR. Binding was relatively low affinity (EC 50 ϭ ϳ10 M) and was stronger with than without Ca 2ϩ (Fig. 1D). Compared with CR binding to a GST fusion protein containing only the most N-terminal CRB (CRB1), CR binding in the linear range of the binding curve was increased by 19.6 Ϯ 7.9% and 32.6 Ϯ 13.6% for GST proteins containing two (CRB1,2) and three (CRB1-3) motifs, respectively (data not shown). These results confirm that CR binds in a Ca 2ϩ -dependent manner to the CRBs in ␣ 1 2.1.
If CR is an interacting partner of Ca v 2.1, it should also associate with the intact channel. To test this prediction, we coexpressed CR and Ca v 2.1 in HEK293T cells and used CR antibodies in coimmunoprecipitation experiments. CR antibodies brought down ␣ 1 2.1 in cells cotransfected with Ca v 2.1ϩ CR. Some ␣ 1 2.1 was brought down nonspecifically by CR antibodies because a small amount of ␣ 1 2.1 was detected in immunoprecipitations from cells transfected with Ca v 2.1 alone (Fig. 1E). However, a larger signal corresponding to ␣ 1 2.1 was consistently detected when Ca v 2.1 was coexpressed with CR, indicating some specific association of CR with ␣ 1 2.1. To verify these results in the native context, we performed coimmunoprecipitation experiments with extracts from mouse cerebellum given that CR and Ca v 2.1 are both expressed highly in cerebellar granule neurons (19,21,22). In these experiments, CR antibodies but not control IgG coimmunoprecipitated ␣ 1 2.1 (Fig. 1F) Fig. 2).
Ca 2ϩ -dependent binding of CR to the channel could dynamically alter Ca 2ϩ -dependent modulation of Ca v 2.1. Because of a reliance on global elevations in Ca 2ϩ , CDI can be suppressed by Ca 2ϩ buffers such as EGTA (5-10 mM). In contrast, CDF, which depends on rapid, local Ca 2ϩ elevations, is not blunted by high concentrations of EGTA (6,7). If CR acted as a Ca 2ϩ buffer when bound to the channel, it could suppress CDI of Ca v 2.1. To test this, we performed experiments dissecting voltage-from Ca 2ϩ -dependent inactivation (VDI and CDI, respectively). Voltage protocols consisted of two test pulses (P1 and P2) separated by a conditioning prepulse (Fig. 3A, pre) to varying voltages. With both Ca 2ϩ and Ba 2ϩ as the charge carrier, the second test current is smaller than the first because of inactivation induced by the prepulse. The ratio of the P2:P1 test current amplitudes can therefore be used as a metric for inactivation (Fig. 3B). For Ca v 2.1 I Ca , inactivation is significantly stronger than for I Ba (P2/P1 ϭ 0.31 Ϯ 0.07 for I Ca versus 0.68 Ϯ 0.04 for I Ba , p Ͻ 0.001). In addition, the P2:P1 ratio exhibits U-shaped dependence on prepulse voltage with a minimum at the prepulse voltage evoking the maximal inward I Ca (Fig. 3B). In contrast, I Ba shows a more modest and monotic increase in inactivation with prepulse voltage caused by VDI (Fig. 3, A and B).
Because Ca v 2.1 current density positively affects CDI (14,20), we restricted analysis to cells exhibiting maximal current densities between 13 and 19 pA/pF.
Compared with the strong CDI in cells transfected with Ca v 2.1 alone, CDI was virtually undetectable across the full range of prepulse voltages in cells cotransfected with CR (Fig. 3,  A and B). At prepulse voltages between Ϫ20 and ϩ10 mV, I Ca actually underwent facilitation (P2/P1 Ͼ 1). Facilitation was Ca 2ϩ -dependent in that there was no effect of CR on inactivation of I Ba (Fig. 3C). Consistent with protocols that induce CDF (6), a prepulse to 0 mV induced a significant acceleration in activation kinetics of the P2 compared with the P1 current for both Ca v 2.1 alone ( P1 ϭ 6 Ϯ 1.3 ms versus P2 ϭ 2 Ϯ 0.3 ms, n ϭ 6, p ϭ 0.02) and Ca v 2.1ϩCR ( P1 ϭ 12.4 Ϯ 1 ms, versus P2 ϭ 1.9 Ϯ 0.2 ms, n ϭ 9, p Ͻ 0.001). However, the prepulse-induced speeding of the P2 activation kinetics was significantly greater for cells cotransfected with CR compared with Ca v 2.1 alone (ϳ125%, p ϭ 0.001). To quantitate the Ca 2ϩ -dependent mod-  NOVEMBER 16, 2012 • VOLUME 287 • NUMBER 47

JOURNAL OF BIOLOGICAL CHEMISTRY 39769
ulation seen with a 0-mV prepulse, the P2/P1 ratio of I Ba was subtracted from that for I Ca . This analysis clearly indicated that unlike Ca v 2.1 alone, which shows significant CDI, CR caused overt CDF (Fig. 3C).
We further analyzed the effects of CR on Ca v 2.1 CDF using action potential (AP) waveforms. During a train of 200-Hz AP stimuli, CDF is evident as a progressive increase in the amplitude of I Ca ; I Ba undergoes less facilitation, which is voltage-dependent (Fig. 4, A and B). Consistent with our data obtained with conditioning prepulses (Fig. 3), CR caused a significantly larger increase in I Ca but not I Ba by the end of the train (Fig. 4C). We measured CDF as the difference in the amplitude of I Ca and I Ba at the end (0.5 s) of the train. By this metric, CDF for Ca v 2.1 was significantly greater with CR (0.22 Ϯ 0.01, n ϭ 10) than without (0.13 Ϯ 0.03, n ϭ 8; p Ͻ 0.01, t test). These results confirm that CR enhances CDF of Ca v 2.1.
These effects of CR on Ca v 2.1 CDI and CDF could be due to its actions as a freely diffusible Ca 2ϩ buffer. Alternatively, CR binding to the ␣ 1 2.1 II-III linker may directly suppress CDI and enhances CDF. To distinguish between these possibilities, we took advantage of the fact that the CRBs present in ␣ 1 2.1 are not conserved in the Ca v 1.2 ␣ 1 subunit (␣ 1 1.2). If CR binding to ␣ 1 2.1 is necessary for Ca v 2.1 modulation, then chimeric channels in which the 2-3 linker of ␣ 1 1.2 is substituted for that in ␣ 1 2.1 (Ca v 2.1-1.2) should prevent effects of CR on CDI and CDF. Consistent with this prediction, Ca v 2.1-1.2 channels underwent CDI that was not affected by CR (Fig. 5). In addition, there was no significant difference in CDF during AP trains in cells transfected with Ca v 2.1-1.2 alone and cells cotransfected with CR (p ϭ 0.57; Fig. 6). These results show that the CRBs are required for CR modulation of Ca v 2.1 CDI and CDF.
To verify the importance of CR binding to the CRBs for Ca v 2.1 modulation, we tested the impact of a peptide, which should competitively displace CR from binding sites in the ␣ 1 2.1 II-III linker. Because our biochemical experiments indicated that CR binds to a peptide sequence including the first three CRBs (Fig. 1), we generated a mcherry-tagged peptide containing this region (CRB1-3). Because CR had no effect on I Ba either in inactivation or facilitation protocols (Figs. 3 and 4), we restricted analysis to I Ca and assumed effects on I Ca inacti-  vation and facilitation reflected effects on CDI and CDF, respectively. When cotransfected with Ca v 2.1ϩCR, CRB1-3 significantly opposed the decrease in CDI caused by CR (for a 0-mV prepulse, P2/P1 ϭ 1.0 Ϯ 0.05 for Ca v 2.1ϩCR versus 0.58 Ϯ 0.11 for ϩCRϩCRB1-3; p Ͻ 0.01; Fig. 7A). Coexpression of CRB1-3 also significantly inhibited the effect of CR on CDF by the end of a 1-s AP train (p Ͻ 0.05; Fig. 7B). In contrast, there was no difference in CDI (p ϭ 0.43 for a 0-mV prepulse) or CDF (p ϭ 0.93 at the end of the 1-s train) in cells transfected with Ca v 2.1 alone and those cotransfected with Ca v 2.1ϩCRB1-3 (Fig. 7, C and D), which argued against nonspecific inhibitory effects of CRB1-3 on Ca v 2.1 that were independent of CR modulation. Taken together, our results support a mechanism in which CR binding to the ␣ 1 2.1 II-III linker directly inhibits Ca v 2.1 CDI and enables enhanced CDF of Ca v 2.1 during repetitive stimuli.

DISCUSSION
Our study provides key evidence for a new role for CR as an integral component of Ca v 2.1 complexes that modulates Ca v 2.1 function. First, CR binds in a Ca 2ϩ -dependent manner to basic motifs in the ␣ 1 2.1 II-III linker. Second, CR forms a complex with Ca v 2.1 channels in the brain. Third, the interaction of CR with the ␣ 1 2.1 II-III linker inhibits CDI and enhances CDF of Ca v 2.1 channels and enhances CDF in transfected HEK293T cells. Because of the cellular overlap in CR and Ca v 2.1 expres-sion in the brain (22,34), CR interactions with Ca v 2.1 should be considered in models of how this protein controls Ca 2ϩ signaling and excitability in neurons.
Ca 2ϩ -dependent Binding of CR to Effectors-CR has long been considered a diffusible Ca 2ϩ buffer that can alter spatiotemporal aspects of Ca 2ϩ signaling (19,23). However, multiple lines of evidence suggest that CR may not be freely diffusible and may, in fact, interact with proteins in a manner analogous to "Ca 2ϩ sensors" such as CaM. First, CR undergoes large conformational changes upon binding to Ca 2ϩ (24,25). Similar Ca 2ϩ -dependent structural changes have been reported for the related EF-hand proteins calbindin D-28k and secretagogin (35,36). As in CaM, Ca 2ϩ binding to CR may expose hydrophobic regions of the protein, which allow Ca 2ϩ -dependent interactions with other proteins. Second, in addition to its cytosolic localization, CR is also abundant in the membrane fraction of cerebellar extracts and less so under conditions of low Ca 2ϩ  (26). The latter findings could be explained by Ca 2ϩ -dependent interactions of CR with integral membrane or membrane-associated proteins, much like the well established role of CaM in regulating pre-and post-synaptic effectors (37).
Although our in vitro experiments indicate a relatively low binding affinity (ϳ10 M) of CR for the ␣ 1 2.1 CRB sequences in GST-CRB1-3 (Fig. 1D), we believe this may underestimate the ability of CR to interact with ␣ 1 2.1 in vivo. Because we found greater CR binding to GST proteins containing three compared with two or one CRB sequences, it is likely that the five CRBs in the intact channel may increase the avidity of CR binding. In addition, CR is thought to be expressed at rather high concentrations in some neurons, ranging from ϳ40 to 80 M (27,38). These concentrations of CR would be sufficient to saturate the CRB sequence(s) in ␣ 1 2.1 to a large extent, particularly if the affinity between CR and CRBs is even higher within the intact channel complex than in vitro. Although our biochemical analyses cannot allow conclusions regarding the stoichiometry of CR binding to the ␣ 1 2.1 II-III linker, binding of CR to the GST-CRB1-3 protein was cooperative (Fig. 1D). Because GST-CRB1-3 contains three of the five CRB sequences in the ␣ 1 2.1 II-III linker, it is possible that cooperativity in the binding assay resulted from multiple CR molecules binding to GST-CRB1-3.
Alternatively, many EF-hand proteins including CR isoforms show reversible dimerization in vitro (39). The binding of CR dimers to the CRBs could therefore also lead to the apparent cooperativity in binding to GST-CRB1-3. Additional studies will be required to fully resolve the molecular thermodynamic properties of CR interactions with ␣ 1 2.1.
Although our findings that CR coimmunoprecipitates with Ca v 2.1 channels in the brain (Fig. 1F)    possible that CR binding to the CRBs in ␣ 1 2.2 could have important consequences for the modulation of Ca v 2.2 channels in neurons. The characterization of other CR-interacting proteins other than Ca v 2.1 and how they may be modulated by CR are crucial for fully understanding Ca 2ϩ signaling dynamics in the neuronal and non-neuronal cell-types in which CR is expressed (42,43).
Ca v 2.1 CDI/CDF Modulation by CR-Significant progress has been made in elucidating the mechanisms underlying Ca v 2.1 CDI and CDF (see Ref. 44 for review). CDI depends on the N-terminal lobe of CaM, which responds to global rather than local Ca 2ϩ signals. CDF depends on the C-lobe of CaM, which likely binds local Ca 2ϩ ions as they emerge from the channel pore (6,45). In support of this model, intracellular dialysis with EGTA (10 mM) prevents CDI but not CDF (7). Moreover, CDF but not CDI is observed at the single channel level (46). This latter result and the findings that CDI increases with Ca v 2.1 current density (14,20) illustrate that CDI depends on Ca 2ϩ influx through multiple open channels and so should be sensitive to Ca 2ϩ buffering. Based on the ability of CR to rapidly depress presynaptic Ca 2ϩ signals (47), it is assumed that CR will act as a Ca 2ϩ buffer when faced with a rise in intracellular Ca 2ϩ . Therefore, it is perhaps not surprising that coexpression of CR with Ca v 2.1 inhibited CDI. The remarkable result is that CR accomplishes this through its association with the ␣ 1 2.1 II-III linker (Figs. 6 and 7). Our biochemical and electrophysiological experiments show that disabling CR interactions with the ␣ 1 2.1 II-III linker prevents the channel modulation. The tethered CR may rapidly suppress global Ca 2ϩ elevations that support CDI, which subsequently enhances CDF during trains of AP waveforms (Figs. 3 and 4). Alternatively, CR binding to the II-III linker could allosterically modulate CDI in the manner of a Ca 2ϩ sensor. Despite the presence of molecular determinants for CDI and CDF in the C-terminal domain of Ca v ␣ 1 subunits, auxiliary Ca v ␤ subunit interactions with the I-II linker have been shown to modulate CDI and CDF of Ca v 1.2 channels (48). Experimental dissection of a Ca 2ϩ buffering versus Ca 2ϩ sensor mechanism is complicated by the possibility that disrupting the Ca 2ϩ buffering activity of CR might also negate its ability to interact with the channel but nevertheless is an important goal for future studies.
It is noteworthy that the CRBs in the ␣ 1 2.1 II-III linker overlap with the bipartite synprint site in ␣ 1 2.1. Interactions between SNAREs and the synprint are thought to promote efficient coupling of Ca v 2 channels and exocytosis in presynaptic nerve terminals based on evidence that peptides containing the synprint site impair neurotransmission (49,50). Our findings suggest that such peptides may also influence CR regulation of Ca v 2.1 CDI and CDF and so may affect synaptic transmission via multiple mechanisms. In addition, splice variants lacking portions of the CRB/synprint region have been identified in neuroendocrine cells and various brain regions (51). The inability of CR to modulate CDI/CDF of such variants may further diversify Ca v 2.1 Ca 2ϩ signaling between neuronal subtypes.
Neurophysiological Significance of CR-Ca v 2.1-Immunohistochemical analyses indicate a number of neuronal cell groups in which CR and Ca v 2.1 colocalize. Ca v 2.1 channels are the major presynaptic Ca 2ϩ channels in the nerve terminals form-ing the Calyx of Held synapse in the auditory brainstem (9,52). CR is detected presynaptically at these synapses but only at significant levels (Ͼ18% in rats) after postnatal day 14 (53). Electrophysiological recordings at the Calyx of Held synapse, usually done in brainstem slices from juvenile rats (postnatal day 8 -10), indicate that the presynaptic Ca v 2.1 channels undergo CDI and CDF (8 -10, 54). Given our findings that CR inhibits CDI and enhances CDF, the developmental increase in CR would be expected to promote the activity-dependent Ca 2ϩ influx that may limit synaptic depression and/or increase reliability in the mature Calyx of Held synapse (55).
As our coimmunoprecipitation of Ca v 2.1 with CR from mouse cerebellum would indicate, CR-Ca v 2.1 complexes may play a role in cerebellar granule cells, the predominant cell types expressing CR in the cerebellum (56,57). Granule cells provide the major excitatory drive to Purkinje neurons in the form of parallel fibers. Genetic inactivation of CR in mice increases the intrinsic excitability of granule cells and Purkinje cell firing rate in vivo (27,28,58). With respect to Ca v 2.1 in granule cells, a loss of CR should inhibit Ca v 2.1 Ca 2ϩ influx by enhancing CDI (Fig.  3). Decreased I Ca may seem at odds with the hyperexcitable phenotype of CR Ϫ/Ϫ granule cells, because it would be expected to limit activation of Ca 2ϩ -activated BK channels and subsequently oppose repolarization following an action potential. However, we have observed compensatory changes in Ca v 2.1 subunit expression in cerebellar Purkinje neurons from mice lacking parvalbumin and calbindin D-28k (15), which could also explain the lack of correlation between our findings and that expected in CR Ϫ/Ϫ granule cells. In addition, Ca v 2.1/CR interactions may be more relevant presynaptically, where enhanced CDF may support residual Ca 2ϩ in parallel fiber terminals that causes short term synaptic plasticity at the parallel fiber-Purkinje cell synapse (59). In summary, our results implicate CR as a novel modulator of Ca v 2.1 channels, which may foreshadow yet additional roles for CR in actively regulating neuronal excitability and synaptic transmission through direct Ca 2ϩ -dependent interactions with other effectors.