Ca²⁺/Calmodulin Regulates Trafficking of Ca_V1.2 Ca²⁺ Channels in Cultured Hippocampal Neurons

Abstract

As the Ca²⁺-sensor for Ca²⁺-dependent inactivation, calmodulin (CaM) has been proposed, but never definitively demonstrated, to be a constitutive Ca_V1.2 Ca²⁺ channel subunit. Here we show that CaM is associated with the Ca_V1.2 pore-forming α_1C subunit in brain in a Ca²⁺-independent manner. Within its CaM binding pocket, α_1C has been proposed to contain a membrane targeting domain. Because ion channel subunits assemble early during channel biosynthesis, we postulated that this association with CaM could afford the opportunity for Ca²⁺-dependent regulation of membrane targeting. We showed that the isolated domain functioned as a Ca²⁺/CaM regulated trafficking determinant for CD8 (a model transmembrane protein) using fluorescent-activated cell sorting analysis and, using green fluorescent protein-tagged α_1C subunits expressed in cultured hippocampal neurons, that Ca²⁺/CaM interaction with this domain accelerated trafficking of Ca_V1.2 channels to distal regions of the dendritic arbor. Furthermore, this Ca²⁺/CaM-accelerated trafficking was activity dependent. Thus, CaM imparts Ca²⁺-dependent regulation not only to mature Ca_V1.2 channels at the cell surface but also to steps during channel biosynthesis.

Introduction

Neuronal responses to electrical activity, such as neurotransmitter release or long-term potentiation, result from changes in intracellular Ca²⁺, which modulates a multitude of Ca²⁺ binding proteins (CaBPs) with distinct but overlapping effectors. Because ion channels are among these effectors, Ca²⁺ can provide feedback regulation to the very proteins that regulate membrane excitability. Most extensively studied is Ca²⁺-dependent inactivation (CDI) of L-type Ca_V1.2 Ca²⁺ channels (Budde et al., 2002).

The CDI sensor is thought to be calmodulin (CaM), the prototypical CaBP. CaM controls CDI through interaction with a ∼110 aa binding pocket in the C terminus (CT) of α_1C in which one critical determinant is a consensus CaM binding “IQ” motif (Peterson et al., 1999; Zühlke et al., 1999). Other sites in the binding pocket are also essential, and mutations throughout the binding pocket that disrupt CaM interaction occlude the expression of CDI (Kim et al., 2004). Based on biochemical, imaging, and functional analyses, CaM has been predicted (but never definitively demonstrated) to be an obligate Ca_V1.2 subunit, bound to α_1C in basal levels of [Ca²⁺]_i and poised to trigger CDI when [Ca²⁺]_i increases (Pitt et al., 2001).

Because ion channel subunits associate early during channel biosynthesis (Deutsch, 2003), we hypothesized that CaM could also affect diverse processes during channel biosynthesis in addition to regulating CDI. Thus, it was particularly intriguing that the α_1C CT contains a domain within the CaM binding pocket that is necessary for proper plasma membrane targeting of Ca_V1.2 channels (Gao et al., 2000). Because Ca_V1.2 channels are central to cAMP response element-binding protein (CREB)-dependent gene expression paradigms in the hippocampus that underlie learning and memory and specific targeting of Ca_V1.2 channels is necessary for CREB-dependent gene expression (Deisseroth et al., 2003; Weick et al., 2003), we postulated that Ca²⁺/CaM may regulate Ca_V1.2 trafficking in hippocampal neurons, thereby contributing to activity-dependent gene expression paradigms.

Materials and Methods

Molecular biology.

The green fluorescent protein (GFP)–α_1C construct was constructed as described previously (Grabner et al., 1998). The GFP–α_1C–CaM and GFP–α_1C–CaM₁₂₃₄ concatemers with a 12× glycine linker between the truncated α_1C (at amino acid 1669) and CaM or CaM₁₂₃₄ were generated following the general strategy as outlined previously (Mori et al., 2004). The myc–CD8–Ca²⁺ in activation region peptide (CIRP) constructs were generated by standard molecular biology techniques. The construct consists of an extracellular myc tag linked to the extracellular and transmembrane domains of CD8α (amino acids 1–206) linked to the CIRP region of α_1C (amino acids 1507–1669).

Fluorescent-activated cell sorting.

Human embryonic kidney 293T (HEK293T) cells were maintained in 5% CO₂ at 37°C in DMEM supplemented with 10% FBS, l-glutamine, penicillin, and streptomycin. Cells were transfected using calcium phosphate (Chemicon, Temecula, CA). In brief, 11–21 μg of DNA in toto were used to transfect cells grown to a ∼60% density on a 6 cm dish. The DNA ratio of GFP to myc–CD8–CIRP was 1:10. When CaM or CaM₁₂₃₄ was cotransfected with myc–CD8–CIRP, the DNA ratio of CaM to myc–CD8–CIRP was 1:1. Cells were harvested ∼40 h after transfection and surface stained with an anti-myc antibody conjugated to the fluorophore PBXL-3 (Martek, Columbia, MD). Staining was performed following a standard fluorescent-activated cell sorting (FACS) staining protocol with at least 45 min of incubation with a 1:100 dilution of the antibody. The FACS wash used during spins and staining contained 1% BSA and 0.1% sodium azide in PBS. Cells were fixed with 4% paraformaldehyde in PBS. Samples were analyzed on a BD LSRII at the Columbia University Flow Cytometry Facility. A total of 30,000 cells were collected for each sample for each of three trials. The LSRII uses BD FACS DiVa software. Data analysis was performed using Flow-Jo (version 8.1.0), and statistical comparisons were done using Student's t test.

HEK293 cell culture and transfection for electrophysiology and confocal imaging.

HEK293 cells were maintained in 5% CO₂ at 37°C in DMEM supplemented with 10% FBS, l-glutamine, penicillin, and streptomycin. A stable cell line expressing the calcium channel auxiliary subunits β₁ and α₂δ was maintained as above with the addition of G418 (500 μg/ml). Cells were transfected using CaCl₂. In brief, 5–10 μg of DNA in toto were used to transfect cells grown to a ∼25% density on a 10 cm dish. When GFP was used to visualize transfected cells, the DNA ratio of GFP to α_1C was 1:10. When CaM or CaM₁₂₃₄ was cotransfected with α_1C, the DNA ratio of CaM or CaM₁₂₃₄ to α_1C was 10:1, increasing the likelihood that cells transfected with α_1C also contained transfected CaM or CaM₁₂₃₄.

Electrophysiological recordings were performed 2 d after transfection. Whole-cell patch-clamp currents recorded from HEK293 cells were obtained with an Axopatch 200B amplifier. The internal pipette solution contained the following (in mm):190 N-methyl-d-glucamine (NMG), 5 EGTA, 4 MgCl₂, 40 HEPES, pH 7.25, and 2 ATP, supplemented with 12 μm phosphocreatine. The external solution contained the following (in mm): 130 NMG-aspartate, 1 MgCl₂, 10 glucose, 10 4-aminopyridine, 10 HEPES, pH 7.4, and 10 BaCl₂. Series resistance was compensated by 70%. Currents were sampled at 10 kHz and filtered at 2 kHz. Gating currents were obtained by depolarizations to the reversal potential. Data are presented as mean ± SEM.

For confocal imaging, HEK cells grown on coverslips were rinsed with PBS, fixed with ice-cold methanol, and mounted in Citifluor. Images were obtained on a Zeiss (Oberkochen, Germany) LSM 510 confocal microscope at the Columbia University Health Sciences Optical Microcopy Facility.

Neuronal cultures and imaging.

Hippocampi from 1–2 d newborn rats (Sprague Dawley strain) were dissociated through enzymatic treatment with 0.25% trypsin and subsequent trituration. The cells were plated on glass coverslips previously coated with poly-d-lysine. Hippocampal cells were grown in neurobasal A medium (Invitrogen, Carlsbad, CA) supplemented with 2% B27, 2 mm glutamine, and 10% heat-inactivated bovine calf serum. After 24 h, this medium was replaced by one containing 0.5 mm glutamine, 1% heat-inactivated bovine calf serum, 0.5 mm kynurenic acid, 70 μm uridine, and 25 μm 5-fluorodeoxyuridine. Hippocampal neurons grown for 6–7 d in culture were transfected with the indicated plasmids using calcium phosphate precipitation, as described previously (Deisseroth et al., 1998). Some neurons were cotransfected with red fluorescent protein (RFP) (pDsRed2; Clontech, Mountain View, CA). Fluorescent images in live cells were obtained at 1, 3, 5, and 14 d after transfection. When testing the activity dependence of trafficking, cells were made quiescent 6 h after transfection by incubation for 60 min in Tyrode's solution containing (in mm) 119 NaCl, 3 KCl, 20 HEPES, 2 CaCl₂, 2 MgCl₂, 30 glucose, and 1 μm TTX, pH 7.3, and then stimulated with hyperkalemic solution containing the following (in mm): 77 NaCl, 45 KCl, 20 HEPES, 2 CaCl₂, 2 MgCl₂, and 30 glucose, pH 7.3 for 3 min. All images were acquired using a Bio-Rad (Hercules, CA) MRC 1000 confocal laser scanning system coupled to a Zeiss Axiovert 100 inverted microscope. For each image acquisition, a Kalman average of five frames was used. Trafficking was quantified by measuring the longest fluorescent dendritic branch with NIH ImageJ 1.35p, and data were analyzed with the Kolmogorov–Smirnov test.

CaM coimmunoprecipitation and immunoblots.

Adult rat brain was washed twice with ice-cold PBS and homogenized with glass homogenizer 20 times in 4 vol of ice-cold PBS with protease inhibitor cocktail (Roche, Indianapolis, IN). The homogenate was centrifuged at 16,000 × g at 4°C for 20 min. The pellet was resuspended in 150 mm NaCl, 50 mm Tris, pH 7.5, 10 μm CaCl₂ (buffer A) plus protease inhibitors and 1% Triton X-100 and incubated for 1 h at 4°C and then recentrifuged. Protein (1.8 mg) was incubated in buffer A plus 0.25% Triton X-100 and free Ca²⁺ as indicated (Ca²⁺ buffering was performed with 5 mm BAPTA using calculations derived from WEBMAXC STANDARD, http://www.stanford.edu/∼cpatton/maxc.html). Anti-α_1C (Alomone Labs, Jerusalem, Israel) was added to the samples, which were incubated for 1 h at 4°C, and then overnight at 4°C with protein A plus G beads. Samples were washed with buffer A plus 0.25% Triton X-100 supplemented with free Ca²⁺ as indicated and protein (10 μg) was loaded on SDS-PAGE gels. For anti-α_1C immunoblot, protein was transferred to nitrocellulose. For anti-CaM immunoblot, protein was transferred to polyvinylidene difluoride at 4°C for 16 h in 50 mm potassium phosphate, pH 7.0, and fixed to the membrane by incubation with freshly prepared glutaraldehyde (0.2% in the phosphate buffer) for 20 min. EGTA (1 mm) was added to all samples before SDS-PAGE. Immunoblots were performed with anti-CaM antibody (Upstate Biotechnology, Lake Placid, NY).

For GFP immunoblots, HEK293 cells were washed with ice-cold TBS (150 mm NaCl and 50 mm Tris, pH 7.5), harvested, and resuspended in ice-cold TBS with 1% Triton X-100 and protease inhibitor 2 d after the cells had been transfected with the respective plasmids. Cells were lysed by pipetting up and down and then centrifuged at 16,000 × g for 20 min. Protein was then separated by SDS-PAGE, transferred to nitrocellulose, and immunoblotted with anti-GFP antibody (Covance, Princeton, NJ).

Results

CaM is constitutively associated with Ca_V1.2 channels in brain

To test whether CaM is constitutively associated with Ca_V1.2 in tissue, we checked whether CaM coimmunoprecipitated with α_1C from rat brain lysates. As shown in Figure 1A, CaM was detected in α_1C immunoprecipitates when elevated Ca²⁺ (10 μm) or nominal basal Ca²⁺ (100 nm) was added to the buffer. These results suggested that CaM functions as a true channel subunit and remains associated with Ca_V1.2 channels at basal levels of Ca²⁺.

Figure 1.

Constitutive association of CaM with α_1C in brain promotes activity-dependent Ca_V1.2 trafficking. A, Immunoblot for α_1C (top) and CaM (bottom) after immunoprecipitation of α_1C in the presence of 10 μm Ca²⁺ or 100 nm Ca²⁺. IgG controls were performed and did not show α_1C immunoprecipitation (data not shown). B, Immunoblot for GFP of lysates of untransfected HEK cells or cells transfected with GFP, GFP–α_1C, or GFP–α_1C (TLF). Single arrowhead, GFP–α_1C; double arrowhead, GFP. C, Fluorescent confocal images of hippocampal neurons transfected with RFP and cotransfected with GFP–α_1C, GFP–α_1C (I/E), or GFP–α_1C (TLF). Scale bar, 20 μm. D, Normalized frequency versus distance from soma of GFP in the longest dendrite for each neuron for GFP–α_1C (n = 193), GFP–α_1C (TLF) (n = 108), and GFP–α_1C (I/E) (n = 55) in cultures imaged 3 d after transfection. E, Normalized frequency versus distance from soma of RFP in the longest dendrite for each neuron for GFP–α_1C (n = 63), GFP–α_1C (TLF) (n = 64), and GFP–α_1C (I/E) (n = 55). F, Normalized frequency versus distance from soma of GFP in the longest dendrite for each neuron stimulated with 45 mm K⁺ for 3 min at 6 h after transfection (n = 69) versus GFP without stimulation (control, n = 70) and imaged 24 h after stimulation. G, Same as in F, but for GFP–α_1C (n = 115, stimulated; n = 55, control). H, Same as in F, but for GFP–α_1C (I/E) (n = 53, stimulated; n = 52, control). I, Same as in F, but for GFP–α_1C (TLF) (n = 41, stimulated; n = 41, control).

CaM controls activity-dependent trafficking of Ca_v1.2 channels in hippocampal neurons

The previous identification of a plasma membrane targeting sequence (amino acids 1623–1666) within the CaM binding pocket of α_1C (Gao et al., 2000), and the constitutive nature of the interaction between CaM and α_1C led us to investigate whether CaM regulated α_1C trafficking in hippocampal neurons. In this role, CaM could allow Ca²⁺-dependent trafficking of α_1C that was responsive to activity. To assess trafficking, we expressed a GFP-tagged version of α_1C in cultured hippocampal neurons. The addition of GFP to the α_1C N terminus does not alter any functional properties of Ca_V1.2 channels (Grabner et al., 1998) and was thus a convenient tool for visualizing the subcellular distribution of channels. GFP–α_1C showed intense signal in the soma and prominent fluorescence throughout the dendritic arbor (Fig. 1C), consistent with the pattern observed previously in hippocampal neurons after transfection of GFP–α_1C or immunocytochemistry of endogenous α_1C (Obermair et al., 2004). Immunoblots of GFP–α_1C expressed in HEK cells confirmed that the fluorescent signal represented GFP–α_1C and not cleaved GFP (Fig. 1B). To assess trafficking of Ca_V1.2, we first devised a metric in which we measured the distance from the soma to the farthest GFP fluorescent point in the longest dendrite in each neuron and plotted these data as a normalized (to the number of neurons in each dataset) frequency distribution. We compared the normalized frequency distributions for a wild-type (WT) α_1C, an α_1C with ¹⁵⁹¹TLF¹⁵⁹³ mutated to AAA (TLF), and an α_1C with I1654E (Fig. 1C). We had shown previously that both mutations, I/E in the IQ motif (I/E) and TLF in a separate region of the CaM binding pocket, disrupt CaM interaction with the α_1C C terminus (Kim et al., 2004). As shown in Figure 1D, the CaM disrupting mutations significantly reduced α_1C trafficking assessed 3 d after transfection. This was not attributable to a mutant-induced decrease in dendritic length because cotransfection of RFP in a subpopulation of these cells showed that dendritic length was unaltered (Fig. 1E), nor was it attributable to a decrease in mutant protein expression because the TLF mutant was expressed at similar levels to WT GFP–α_1C, as shown in Figure 1B.

The potential role for CaM in trafficking of α_1C prompted us to test whether this process was sensitive to neuronal activity and the consequent Ca²⁺ influx that could activate CaM. We therefore compared trafficking in neurons depolarized with 45 mm K⁺ for 3 min with neurons held quiescent. There was no activity-dependent change in distribution of GFP transfected as a control (Fig. 1F). However, when we compared GFP–α_1C trafficking, we found that activity facilitated trafficking (Fig. 1G). The TLF and I/E mutations, in contrast, prevented activity-dependent facilitation of trafficking (Fig. 1H,I). Together, these data suggest that Ca²⁺/CaM may provide dynamic, activity-dependent trafficking to Ca_v1.2 channels.

The CaM binding pocket contains a Ca²⁺/CaM-sensitive membrane targeting domain

To determine whether the previously identified membrane targeting sequence in α_1C (amino acids 1623–1666) contributed to Ca²⁺/CaM trafficking of Ca_v1.2 channels in hippocampal neurons, we first tested whether this isolated sequence supported Ca²⁺/CaM-dependent targeting of the unrelated, type I transmembrane protein CD8α, which is known to be constitutively trafficked to the plasma membrane (Iodice et al., 2001). We fused the CaM binding pocket of α_1C (CIRP, amino acids 1507–1669) to the C terminus of a myc-tagged CD8α (myc–CD8–CIRP) and assessed whether Ca²⁺/CaM promoted its surface expression in HEK293 cells by comparing the effects of CaM versus CaM₁₂₃₄, a mutated CaM unable to bind Ca²⁺. The myc–CD8–CIRP was coexpressed with CaM or CaM₁₂₃₄, and surface expression was analyzed in nonpermeabilized cells by FACS. In these experiments, GFP was cotransfected to assess the success of transfection. As can be seen in Figure 2, no GFP signal or myc–CD8 surface labeling was detected in untransfected cells. In the absence of CaM or CaM₁₂₃₄ expression, ∼5% of the cells demonstrated surface expression of the myc-tagged construct; similar levels were obtained with CaM coexpression, suggesting that endogenous CaM was not limiting. In comparison, coexpression of CaM₁₂₃₄ reduced myc–CD8–CIRP surface labeling by 56 ± 6% (n = 3 transfections; p = 0.02), revealing that coexpression of CaM promotes significantly higher membrane targeting. These results indicate that CIRP may contain a retention motif that is masked by Ca²⁺/CaM. The presence of myc–CD8–CIRP surface labeling in cells cotransfected with CaM₁₂₃₄ could reflect either that Ca²⁺-free CaM is able to promote membrane targeting, albeit less effectively than Ca²⁺/CaM, and/or that the coexpressed CaM₁₂₃₄ did not completely preclude the actions of endogenous CaM.

Figure 2.

The CIRP region of α_1C contains a forward-trafficking domain regulated by Ca²⁺/CaM. FACS analysis of HEK293 cells cotransfected with GFP, myc–CD8–CIRP, and either CaM or CaM₁₂₃₄ and stained with an anti-myc antibody. Surface staining for control untransfected cells (left), myc–CD8–CIRP plus CaM transfected cells (middle), and myc–CD8–CIRP plus CaM₁₂₃₄ transfected cells (right). Double-positive cells present in top right quadrant are shown magnified below.

Ca²⁺/CaM regulates membrane targeting of Ca_v1.2 channels

We next tested whether membrane targeting of α_1C in HEK cells was affected by mutations that disrupted CaM binding. We measured the resultant currents of two different Ca_V1.2 channels containing CaM binding mutations in α_1C (I/E and TLF as in Fig. 1) and currents from channels containing WT α_1C. As shown in Figure 3, A and B, current amplitudes from both the I/E and the TLF mutants were decreased compared with WT α_1C. To correlate whether this effect could be attributed to effects on trafficking/targeting, we used the GFP-labeled α_1C and analyzed the fluorescent distribution for the WT and mutant channels. The GFP fluorescence pattern in cells expressing TLF or I/E was different from the pattern in cells expressing WT α_1C. Rather than the plasma membrane-like pattern observed in cells expressing WT α_1C, the pattern in cells expressing the mutant channels was more clustered toward the center of the cell in an endoplasmic reticulum-like pattern (Fig. 3C). We also measured gating charge (Q_ON) during voltage steps to the reversal potential (E_rev) to obtain a quantitative correlate of the number of functional channels at the cell surface. The decrease in Q_ON for channels containing the CaM binding mutations suggested that there were significantly less functional channels at the plasma membrane than for the WT α_1C (Fig. 3D).

Figure 3.

Mutations in the CaM binding domain of α_1C affect channel targeting. A, Mutation of amino acids ¹⁵⁹¹TLF¹⁵⁹³ to alanines (TLF) or mutation of I1654E (I/E) reduced current amplitude in HEK293 cells as shown by normalized I–V relationships of I_Ba (n = 5–7). B, Exemplar whole-cell I_Ba currents. Calibration: 200 pA, 200 ms. C, Representative confocal images of GFP-tagged α_1C WT, TLF, and I/E and accompanying line scans showing altered cellular localization of the CaM binding mutants. D, Summary of ON gating current amplitude (n = 10–14) elicited by stepping to the reversal potential (E_rev of approximately +55 mV). Exemplar current from a cell expressing the WT channel evoked by a 20 ms test pulse to +55 mV from V_h of −90 mV (Calibration: 2 nA, 20 ms), and a 250% magnification of the gating current is shown below.

To confirm that these effects were attributable to CaM interaction with CIRP and to test the Ca²⁺ dependence of CaM-mediated trafficking, we compared the effects of CaM versus CaM₁₂₃₄. In an attempt to achieve effective competition between CaM₁₂₃₄ and endogenous CaM, we truncated α_1C after the IQ motif at amino acid 1669 and fused CaM₁₂₃₄ (or CaM as a control) via a flexible 12 glycine linker, thereby providing a high local concentration of the mutant CaM. This general strategy, originally developed to assess the contribution of CaM to CDI, does not affect the resultant Ca²⁺ currents (Mori et al., 2004). An added advantage is that this approach restricts the actions of CaM₁₂₃₄ to α_1C, limiting possible nonspecific effects on the myriad CaM-related pathways that could contribute to protein trafficking. In our α_1C–CaM concatemers, we included GFP at the N terminus as shown in Figure 4A to correlate a quantitative assessment of channels at the plasma membrane (using electrophysiological recordings) with images showing the subcellular localization of the concatemers. Confocal images in Figure 4B of cells expressing the concatemers with CaM (GFP–α_1C–CaM) showed a fluorescence pattern consistent with trafficking of the channels to the membrane and similar to GFP–α_1C (Fig. 4B). In contrast, the predominant fluorescence in cells expressing concatemers with CaM₁₂₃₄ (GFP–α_1C–CaM₁₂₃₄) was excluded from the cell periphery and was instead in a perinuclear distribution (Fig. 4B), suggesting that less channels trafficked to the plasma membrane. Peak currents recorded from these cells were markedly reduced compared with cells expressing GFP–α_1C–CaM (Fig. 4C). Q_ON during voltage steps to the E_rev, normalized for cell capacitance because the cells expressing GFP–α_1C–CaM₁₂₃₄ tended to be significantly larger than the cells expressing GFP–α_1C–CaM, suggested that the linked CaM concatemers yielded significantly more functional channels at the plasma membrane than the linked CaM₁₂₃₄ concatemers (4.4 ± 0.7 vs 0.4 ± 0.3 mC/F; n = 5 each; p = 0.001). Thus, these data suggested that Ca²⁺/CaM is a potent stimulator for channel trafficking through its interaction with the targeting domain in α_1C.

Figure 4.

Trafficking of Ca_V1.2 channels to the plasma membrane is Ca²⁺/CaM dependent. HEK293 cells stably expressing β1 and α₂δ were transfected with GFP-tagged α_1C linked to either CaM or CaM₁₂₃₄ (A–C) or with GFP-α_1C and additionally with CaM or CaM₁₂₃₄ separately (D). A, Schematic of the CaM-linked GFP-tagged α_1C construct. B, Representative confocal fluorescent images and accompanying line scans of cells expressing GFP–α_1C–CaM or GFP–α_1C–CaM₁₂₃₄. C, I–V relationships of I_Ba currents for cells expressing GFP–α_1C–CaM and GFP–α_1C–CaM₁₂₃₄ (n = 7–11). D, I–V relationships of I_Ba currents for cells expressing GFP–α_1C and CaM or CaM₁₂₃₄ (n = 7–11).

As a control for the linked CaM strategy, we also assessed the trafficking/membrane targeting of α_1C when CaM or CaM₁₂₃₄ was coexpressed separately. As seen in Figure 4D, coexpression of CaM generated significantly higher current density than coexpression of CaM₁₂₃₄. Gating charge measurements were also higher (5.6 ± 0.4 mC/F for coexpressed CaM vs 2.5 ± 0.8 mC/F for CaM₁₂₃₄; n = 5 each; P = 0.007) than coexpression of CaM₁₂₃₄. Thus, the CaM-mediated trafficking of α_1C was Ca²⁺ dependent. Compared with the CaM₁₂₃₄-linked constructs (Fig. 4C), the effects with unlinked CaM₁₂₃₄ were diminished. This incomplete inhibition was similar to what we observed for the CD8–CIRP experiments (Fig. 2), suggesting that endogenous CaM was able to compete with coexpressed unlinked CaM₁₂₃₄ and at least partially reverse the effects.

Ca²⁺/CaM control Ca_v1.2 channel trafficking in hippocampal neurons

Because Ca²⁺/CaM interaction with the CIRP in α_1C appeared to mediate trafficking of α_1C in HEK cells, we next tested whether Ca²⁺/CaM was responsible for the trafficking of α_1C in hippocampal neurons. Our general strategy was to compare the distribution of GFP–α_1C–CaM and GFP–α_1C–CaM₁₂₃₄ after transfection into cultured neurons using a similar method of analysis as in Figure 1D. Before proceeding with those experiments, we first analyzed whether either the addition of CaM to the α_1C C terminus or the truncation of α_1C after the IQ motif affected the distribution of α_1C in hippocampal neurons. This concern was motivated by previous reports showing that the distal C terminus of α_1C is important for localizing Ca_V1.2 with the machinery necessary to support CREB-dependent gene expression (Dolmetsch et al., 2001; Weick et al., 2003). When we compared the distribution of GFP–α_1C–CaM with GFP–α_1C, we found that their distributions were similar, both showing intense signal in the soma and prominent fluorescence throughout the dendritic arbor (compare Fig. 5A with 1C). Moreover, measurement of the distance from the soma to the farthest GFP fluorescent point in the longest dendrite for each neuron revealed no difference between GFP–α_1C–CaM and GFP–α_1C (Fig. 5B, data for 3 d after transfection). Thus, the distal end of α_1C is not critical for α_1C trafficking to distal regions of dendrites.

Figure 5.

Trafficking of α_1C in hippocampal neurons is accelerated by Ca²⁺/CaM. A, Fluorescent confocal image of a hippocampal neuron transfected with GFP–α_1C–CaM. Scale bar, 20 μm. B, Normalized frequency versus distance from soma of GFP in the longest dendrite for each neuron for GFP–α_1C–CaM (n = 153) or GFP–α_1C (n = 193) in cultures 3 d after transfection. C, Normalized frequency versus distance from soma of GFP in the longest dendrite for each neuron for GFP–α_1C–CaM versus GFP–α_1C–CaM₁₂₃₄ in cultures imaged at day 1 (CaM, n = 38; CaM₁₂₃₄, n = 44), day 3 (CaM, n = 153; CaM₁₂₃₄, n = 188), day 5 (CaM, n = 26; CaM₁₂₃₄, n = 24), and day 14 (CaM, n = 22; CaM₁₂₃₄, n = 25) after transfection. D, Same format as in C. GFP–α_1C (WT) versus GFP–α_1C (TLF) for day 1 (WT, n = 73; TLF, n = 111), day 3 (WT, n = 193; TLF, n = 130), day 5 (WT, n = 127; TLF, n = 180), and day 14 (WT, n = 32; TLF, n = 38).

We therefore performed measurements with GFP–α_1C–CaM and GFP–α_1C–CaM₁₂₃₄ at several time points after transfection over a 14 d period (Fig. 5C). For the first 3 d after transfection, the GFP–α_1C–CaM trafficked farther from the soma than the CaM₁₂₃₄-linked construct; by 5 d after transfection, the normalized frequency distributions were indistinguishable. That the CaM₁₂₃₄-linked construct was not retained within the soma, as might be expected from the distribution of this construct when expressed in HEK cells, likely reflects the specific manifestation of Ca²⁺/CaM-mediated effects on channel trafficking/targeting in polarized neurons compared with nonpolarized HEK cells. Additionally, we postulated that endogenous CaM might be able to compete more effectively with CaM₁₂₃₄ in these neuronal experiments because of the longer time frame. To test this hypothesis, we analyzed the time dependence of trafficking of GFP–α_1C (TLF) compared with GFP-α_1C (WT) over 14 d, because the mutation should prevent endogenous CaM interaction. As seen in Figure 5D, the WT channel maintained its advantage over TLF throughout the 14 d period.

Discussion

Underlying the formation of memory are certain well characterized types of synaptic plasticity that require Ca²⁺ influx and the consequent activation of gene expression cascades. Several activity-dependent gene expression pathways from synapse to nucleus have been dissected. In particular, activation of CREB has served as the quintessential model for this process (Deisseroth et al., 1996, 1998, 2003; West et al., 2002). Extensive investigation of the pathways leading to CREB activation in hippocampal neurons has revealed a privileged role for Ca²⁺ influx through Ca_V1.2 channels and the related Ca_V1.3 channels compared with other high voltage-activated (HVA) Ca²⁺ channels (e.g., Ca_V2.1 or Ca_V2.2) (Striessnig et al., 2006). Because the distinguishing feature among different HVA Ca²⁺ channels is the identity of the pore-forming α₁ subunit, the advantage of Ca_V1.2 and Ca_V1.3 channels must be attributable to specific determinants within the α₁ subunit rather than in the accessory β or α₂δ subunits, which are common to all HVA channels (Ertel et al., 2000).

One key component affording Ca_V1.2 and Ca_V1.3 channels their privileged role in excitation–transcription coupling may therefore be the association between their respective α₁ subunits and CaM. Although CaM has also been predicted to interact with and confer Ca²⁺ sensitivity to gating of other HVA Ca²⁺ channels (Lee et al., 1999; DeMaria et al., 2001; Liang et al., 2003), the magnitude of the effect is markedly smaller, suggesting a different mode of interaction. An essential role for CaM interaction with α_1C (Ca_V1.2) in CREB-dependent gene expression was suggested by the disruption of excitation–transcription coupling when the CaM binding IQ motif in α_1C was mutated (Dolmetsch et al., 2001). Because CaM is but one CaBP that has been postulated to modulate Ca²⁺ channel function in neurons (Lee et al., 2002; Zhou et al., 2004; Lautermilch et al., 2005; Yang et al., 2006), our demonstration that CaM coimmunoprecipitates with α_1C from brain lysates provides the first evidence that CaM is a bona fide Ca_V1.2 subunit and the likely Ca²⁺ sensor for CDI and gene expression. Moreover, the recent demonstration that CaBP4 competes with CaM to modulate Ca_V1.3 channels in certain neuronal subpopulations (Yang et al., 2006) may explain the apparent differences in coupling of CREB activation to Ca_V1.2 versus Ca_V1.3 channels (Zhang et al., 2006).

The privileged role of Ca_V1.2 channels in excitation–transcription coupling may therefore derive from their specific manner of interaction with CaM. In its role as a channel subunit, CaM is positioned to contribute toward the Ca²⁺-dependent regulation of any of the steps in channel biosynthesis that occur after subunit assembly. We found that CaM provides a Ca²⁺-dependent acceleration of channel trafficking to distal regions of dendrites in hippocampal neurons. This adds to a growing list of examples in which CaM, as a constitutive subunit, modulates steps in channel biosynthesis such as trafficking of SK K⁺ channels (Joiner et al., 2001; Lee et al., 2003) or assembly of KCNQ1 K⁺ channels (Ghosh et al., 2006; Shamgar et al., 2006). For Ca²⁺/CaM acceleration of α_1C trafficking, two results suggest that the mechanism of action for CaM most likely involves an interaction with the CaM binding pocket. First, we demonstrated that the CaM binding pocket contains a Ca²⁺/CaM-dependent autonomous forward trafficking signal. This role for CaM and its binding pocket in α_1C trafficking/targeting fits well with the previous observation that a region in the α_1C CT, in which the CaM binding pocket resides, is important for membrane targeting (Gao et al., 2000). Second, mutations in the CaM binding pocket hindered α_1C trafficking. This latter result provides a possible explanation for the decreased gene expression seen with IQ motif mutations (Dolmetsch et al., 2001); these mutations possibly altered trafficking of α_1C to distal regions of the dendrites or reduced the amount of channels inserted into the membrane, thereby interfering with the gene expression cascade. In this context, it is interesting to note the previous report from Peterson et al. (1999), in which CaM₁₂₃₄ coexpression reduced peak Ca_V1.2 current amplitude in HEK cells compared with CaM coexpression. Because CaM₁₂₃₄ exerts its interference with CDI in a dominant-negative manner by precluding binding of CaM to the α_1C CaM binding pocket, the reduced current amplitude with CaM₁₂₃₄ coexpression recorded in that study, as in our results here, most likely arose from decreased trafficking/membrane insertion of α_1C subunits. Our quantification of gating charge and the visualization of the subcellular localization of GFP-tagged α_1C subunits here offer additional confirmation. As noted previously, it is important to consider that the specific manifestation of Ca²⁺/CaM-mediated effects on channel trafficking/targeting in nonpolarized HEK cells may differ from that in neurons; nevertheless, the delayed trafficking in neurons and the reduced surface expression in HEK cells both appear to result from Ca²⁺/CaM-mediated interaction with α_1C.

Because the mutations that disrupted CaM interaction markedly diminished both Ca_V1.2 current amplitude and affected trafficking, processes normally ascribed to Ca²⁺ channel β subunits (Dolphin, 2003), it is possible that CaM disruption impacts β subunit modulation. Consistent with this hypothesis, we demonstrated previously that the complex of α_1C CT and CaM could interact with the intracellular linker between the α_1C I–II domains (Kim et al., 2004), the primary sight for β subunit interaction. Moreover, the IQ motif contains the arginine- and lysine-rich sequence ¹⁶⁵⁹RKFKKRK¹⁶⁶⁵, a polybasic motif that could serve as an endoplasmic reticulum retention signal (Michelsen et al., 2005). Conserved among other HVA Ca²⁺ channels, the IQ motif may therefore contain the elusive α₁ ER retention signal, masked by CaM, that controls Ca²⁺ channel trafficking. The juxtaposition of CaM and the β subunit through the interaction with the I–II linker could explain the contribution of β subunits to the trafficking process. In the specific case of α_1C, the dynamic regulation provided by CaM could be important in synaptic plasticity that depends on Ca²⁺ influx through Ca_V1.2 channels and is consistent with the data showing that Ca²⁺/CaM interaction with the α_1C IQ motif contributes to activity-dependent gene expression (Dolmetsch et al., 2001).

In this regard, our observation that Ca_V1.2 channels were located in distal regions of dendrites, consistent with recent observations (Davare et al., 2001) that describe Ca_V1.2 channels in spines rather than an older report that suggested that Ca_V1.2 channels were restricted to the soma and proximal dendrites (Westenbroek et al., 1998), places Ca_V1.2 channels in the proper setting to initiate synapse to nucleus signaling. Ca²⁺/CaM regulation of dendritic trafficking might therefore be an important component to the emerging model in which Ca_V1.2 channels participate in local signaling microdomains to regulate plasticity, even when the contribution of Ca²⁺ influx through the channels is modest. For example, long-term depression of synaptic activity depends on Ca²⁺ signaling through Ca_V1.2 channels and downstream activation of CaM kinase II in dendritic spines, yet Ca²⁺ influx through Ca_V1.2 channels is too small to be separated accurately from the dominant signal mediated by Ca_v2.3 (R-type) channels (Yasuda et al., 2003). Likewise, it is possible that CaM regulation of Ca_V1.2 trafficking may increase the local number of Ca_V1.2 channels in response to activity and participate in the specific role of Ca_V1.2 channels in NMDA receptor-independent plasticity and spatial memory (Moosmang et al., 2005). Such signaling pathways, possibly regulated by Ca²⁺/CaM-mediated trafficking, may be vital for plasticity on the scale of individual synapses. Hence, determining the regulators of Ca_V1.2 localization in the dendritic arbor will prove to be important for our understanding of learning and memory.