Molecular Determinants of the CaVβ-induced Plasma Membrane Targeting of the CaV1.2 Channel*

CaVβ subunits modulate cell surface expression and voltage-dependent gating of high voltage-activated (HVA) CaV1 and CaV2 α1 subunits. High affinity CaVβ binding onto the so-called α interaction domain of the I-II linker of the CaVα1 subunit is required for CaVβ modulation of HVA channel gating. It has been suggested, however, that CaVβ-mediated plasma membrane targeting could be uncoupled from CaVβ-mediated modulation of channel gating. In addition to CaVβ, CaVα2δ and calmodulin have been proposed to play important roles in HVA channel targeting. Indeed we show that co-expression of CaVα2δ caused a 5-fold stimulation of the whole cell currents measured with CaV1.2 and CaVβ3. To gauge the synergetic role of auxiliary subunits in the steady-state plasma membrane expression of CaV1.2, extracellularly tagged CaV1.2 proteins were quantified using fluorescence-activated cell sorting analysis. Co-expression of CaV1.2 with either CaVα2δ, calmodulin wild type, or apocalmodulin (alone or in combination) failed to promote the detection of fluorescently labeled CaV1.2 subunits. In contrast, co-expression with CaVβ3 stimulated plasma membrane expression of CaV1.2 by a 10-fold factor. Mutations within the α interaction domain of CaV1.2 or within the nucleotide kinase domain of CaVβ3 disrupted the CaVβ3-induced plasma membrane targeting of CaV1.2. Altogether, these data support a model where high affinity binding of CaVβ to the I-II linker of CaVα1 largely accounts for CaVβ-induced plasma membrane targeting of CaV1.2.

forming Ca V ␣1 subunit, Ca V 1 and Ca V 2 channels arise from the multimerization of three other proteins (7): a cytoplasmic Ca V ␤ subunit, a mostly extracellular Ca V ␣2␦ subunit, and calmodulin constitutively bound to the C terminus of Ca V ␣1 (8 -12).
A considerable body of work documents the interaction and modulation of the Ca V ␣1 subunit of Ca V 1 and Ca V 2 channels (13)(14)(15)(16)(17)(18) by the auxiliary Ca V ␤. The high affinity Ca V ␣1-Ca V ␤ interaction site on the pore-forming Ca V ␣1 subunit is a conserved 18-residue sequence in the I-II linker called the ␣ interaction domain (AID) (19,20) that has been structurally resolved by high resolution x-ray crystallography (21)(22)(23). Structural work showed that the AID forms a ␣-helix that binds to the ␣ binding pocket (ABP) in the Ca V ␤ nucleotide kinase (NK) domain. It has been proposed that the MMQKAL cluster of residues within the latter determines the high affinity nanomolar interaction between the two proteins (24 -29). Numerous mutational analyses of the AID residues have correlated the Ca V ␤-induced biophysical modulation with the high affinity binding of Ca V ␤ to the AID peptide in a variety of Ca V ␣1 isoforms for Ca V 1 and Ca V 2 channels (25, 29 -32).
The association of Ca V ␣1 and Ca V ␤ subunits is also critical for proper channel maturation and cell surface expression of Ca V 2.2 (17), Ca V 1.2 (33,34), and Ca V 2.3 (35). In Ca V 2.2, the I-II linker is presumed to play a role in this process (17,18), and mutations within the AID motif eliminated its cell surface expression and biophysical modulation by Ca V ␤1b and Ca V ␤3 (32). In addition, the Ca V ␤2-induced increase in Ca V 1.2 whole cell currents was abolished with the AID-defective YWI/AAA mutant (29), suggesting that high affinity binding of Ca V ␤ onto AID is required to traffic Ca V ␣1 to the plasma membrane. Nonetheless, the unique character of the high affinity AID-ABP interface in the membrane targeting of Ca V ␣1 has been questioned (27, 36 -40). In particular, it has been suggested that Ca V ␤-mediated plasma membrane targeting could be uncoupled from Ca V ␤-mediated modulation of channel gating (26,41) with important contributions from other intracellular regions (33,39,(42)(43)(44).
In addition to Ca V ␤, the ancillary subunit Ca V ␣2␦ and the ubiquitous calmodulin (CaM) protein have also been proposed to modulate HVA channel maturation and targeting (9). For instance, co-expression of Ca V ␣2␦ promoted the trafficking of the Ca V ␣1 subunit of Ca V 2.2 in COS-7 cells (45), suggesting that Ca V ␣2␦ could promote targeting of all HVA Ca V ␣1 sub-units. CaM is a soluble, 17-kDa Ca 2ϩ -binding protein that serves as a critical Ca 2ϩ sensor for Ca 2ϩ -dependent inactivation and facilitation upon Ca 2ϩ binding in many Ca V 1 and Ca V 2 channels (46), of which Ca V 1.2 and Ca V 2.1 have been best characterized (8,47). Constitutive apocalmodulin binding was reported on multiple sites in the Ca V ␣1 subunit of Ca V 1.2 (48) of which the C-terminal pre-IQ and IQ domains are best characterized (49). Mutations (TLF/AAA and I/E) in the pre-IQ and the IQ CaM-binding domains of the C terminus decreased the whole cell current density of Ca V 1.2, suggesting that Ca 2ϩ / CaM could modulate channel trafficking through its interaction with the C terminus (50) as it has been shown for small activated potassium channels (51).
To gauge the synergetic role of intracellular domains and auxiliary subunits in the steady-state plasma membrane expression of Ca V 1.2, we used a flow cytometry assay with an extracellularly HA-tagged Ca V 1.2 protein. Co-expression with Ca V ␤3 produced a robust enhancement in the plasma membrane targeting of the Ca V ␣1 subunit of Ca V 1.2. The WI residues in the AID helix of the I-II linker of Ca V 1.2 were critical for Ca V ␤-stimulated plasma membrane targeting of Ca V 1.2. No other combination with or without the auxiliary calmodulin and/or the Ca V ␣2b␦ subunit produced any significant increase in the plasma membrane targeting of Ca V 1.2. Hence, Ca V ␤ appears to be the most potent determinant in the plasma membrane targeting of Ca V 1.2. Altogether, our data support a model where high affinity binding of the ABP of Ca V ␤ to the AID helix of Ca V ␣1 largely accounts for Ca V ␤-induced plasma membrane targeting of Ca V 1.2.

EXPERIMENTAL PROCEDURES
Recombinant DNA Techniques-The rabbit Ca V 1.2 (GenBank TM accession number X15539), the rat Ca V ␤3 (GenBank TM accession number M88751) (52), the rat brain Ca V ␣2b␦ (GenBank TM accession number NM_000722) (53), and the human CaM (GenBank TM accession number M27319) were used. All of the subunits were subcloned in commercial vectors under the control of the cytomegalovirus promoter (see supplemental text for details).
For the Ca V ␤3 deletion mutants, flanking NotI sites were inserted around the region(s) to be deleted. Following restriction digest of the NotI fragment and religation of the cohesive ends, the resulting NotI site was mutated back to the wild type amino acids. The Ca V ␤3 fragments (numbered from their deduced amino acid sequence) were subcloned into the NotI sites of the pCMV-Tag5a vector (see supplemental text for details) that is a C-terminal c-Myc tagging vector. A Kozak sequence and an ATG initiation codon were inserted at the 5Ј-end of the nucleotide sequence.
Insertion of the HA Tag in the Ca V ␣1 Subunit-The hemagglutinin (HA) epitope tag (YPYDVPDYA) was inserted in the first extracytoplasmic predicted loop in Domain I at position 574 (nucleotide) for Ca V 1.2. The biophysical properties of the HA-tagged Ca V ␣1 subunit of Ca V 1.2 expressed in HEKT cells with the auxiliary Ca V ␤3 subunit were found not to be significantly different from the wild type Ca V 1.2 channel expressed under the same conditions (see Fig. 1). In addition, cDNA injection of Ca V 1.2-HA constructions in concert with Ca V ␣2b␦ and Ca V ␤3 subunits in Xenopus oocytes yielded a biophysical profile not significantly different from that reported previously for the Ca V 1.2 (56) channels expressed under the same conditions. Hence, the HA-tagged version of the Ca V ␣1 subunit of Ca V 1.2 will be referred to as Ca V 1.2 wt throughout the text.
Cell Culture and Transfections-tsA-201 (HEK293T or HEKT), a subclone of the human embryonic kidney cell line HEK-293 that expresses the simian virus 40 T-antigen, and COS1 cells were grown in Dulbecco's high glucose minimum essential medium supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin at 37°C under 5% CO 2 atmosphere. COS1, HEKT, stable Ca V ␤3, and Ca V ␣2b␦ cells lines were transiently transfected with HA-tagged Ca V 1.2 cDNA using Lipofectamine 2000 (Invitrogen) as per the manufacturer's instructions. Protein expression of the auxiliary subunits in the stable and transient cell lines were confirmed routinely by Western blotting (see Fig. 2B). Transfection rate of the control pEGFP plasmid was estimated to be 66 Ϯ 2% (n ϭ 4) as assessed by flow cytometry from the fluorescence of the green fluorescent protein. Preliminary tests showed that Ca V 1.2 protein expression peaked 24 -36 h after transfection.
Western Blots-Protein expression of all constructs was confirmed by Western blotting in total cell lysates. HAtagged Ca V 1.2 constructs were detected with anti-HA. The procedures are detailed in the supplemental text. Briefly, the membranes were incubated with anti-HA (1:500) (Covance Biotechnology, Québec, Canada) and revealed with an antimouse horseradish peroxidase secondary antibody (1:10000; Jackson Immunoresearch).
Fluorescence-activated Cell Sorting (FACS) Experiments-Cell surface expression of the Ca V 1.2 subunits was determined by flow cytometry using a FACScalibur flow cytometer (Becton Dickinson) at the flow cytometry facility of the Department of Microbiology of the Université de Montréal. The cells expressing the extracellular HA tag were detected using an anti-HAconjugated FITC fluorophore with a FITC filter (530 nm). The relative intensity of staining provided a metric to quantify cell surface expression of the HA-tagged Ca V 1.2 proteins (see supplemental Figs. S1 and S2 and supplemental text for details). The HA-tagged Ca V 1.2 construct was systematically tested as a control with the mutant channels.
Immunofluorescence-For fluorescence microscopy, Ca V ␤3 stable cells were grown on sterile poly-D-lysine-coated coverslips. The cells were fixed 24 h after transfection in 4% paraformaldehyde, permeabilized with 0.075% saponin for 10 min at room temperature, washed in phosphate-buffered saline, and blocked in IgG-free 2% bovine serum albumin in phosphate-buffered saline for 20 min. The cells were incubated with FITC-conjugated anti-HA antibody (1:100) for 1 h at room temperature prior to the cells being mounted (Prolong antifade kit; Invitrogen) on glass microscope slides. HA-tagged Ca V 1.2 channels (wild type and mutant) were visualized (ϫ60) using an Olympus microscope IX-81 microscope along with Image-Pro Plus 5.0 software.
Statistical Analysis-Statistical analyses were performed using the built-in one-way analysis of variance fitting routine for two independent populations of Origin 7.0. The data were considered statistically significant at p Ͻ 0.01.

RESULTS
Ca V ␣2␦ Increases Whole Cell Currents of Ca V 1.2-Co-expression of Ca V 1.2 and Ca V 2.1 with the auxiliary Ca V ␣2␦ subunit was shown to stimulate whole cell currents (45) in COS-7 cells, suggesting that Ca V ␣2␦ could promote plasma membrane targeting of HVA Ca V ␣1 subunits. In Ca V 1.2, the gating charge appears to be unaffected by co-expression with Ca V ␣2␦, suggesting that Ca V ␣2␦ stimulates channel facilitation by setting Ca V 1.2 channels in a conformational state very close to the open state without increasing protein density (57). To evaluate the functional role of Ca V ␣2␦, Ca V 1.2 wt and HA-tagged Ca V 1.2 ␣1 subunits were transiently transfected in the Ca V ␤3 stable HEKT cell line in the absence and in the presence of Ca V ␣2b␦. As shown in Fig. 1A, whole cell currents, recorded in the presence of a physiological solution containing 2 mM Ca 2ϩ (see the supplemental text), were significantly larger when measured in the presence of the Ca V a2␦, confirming that Ca V ␣2␦ stimulates whole cell currents of Ca V 1.2 (9). As shown in Fig. 1B, average whole cell current density increased from Ϫ7 Ϯ 2 pA/pF (n ϭ 6) for the wild type Ca V 1.2 channel in the stable Ca V ␤3 stable cell line as compared with a current density of Ϫ41 Ϯ 9 pA/pF (n ϭ 7) for the wild type Ca V 1.2 channel measured in the same cell line after transient transfection with Ca V ␣2b␦ subunit. Similar results were obtained for the HA-tagged Ca V 1.2 channels (Fig. 1B).
Ca V ␤ Promotes Membrane Targeting of Ca V 1.2-To determine whether Ca V ␣2b␦ stimulates plasma membrane targeting of Ca V 1.2 channels, protein density of the extracellularly HA-tagged Ca V 1.2 channel was quantified with an anti-HA-conjugated FITC fluorophore. Fig. 2A shows the histogram of the fluorescent signal measured after transient expression of the Ca V ␣1 and the auxiliary subunit (either Ca V ␤3 or Ca V ␣2b␦) in nonpermeabilized cells. Protein expression was confirmed by Western blotting (Fig. 2B). As seen, less than 0.5% of the cell population produced autofluorescence, whereas only 1% of the cells were fluorescent after the addition of the FITC antibody to control nontransfected cells (see raw data in supplemental Figs. S1 and S2). Transient co-expression of the HA-tagged Ca V 1.2 subunit in the stable Ca V ␤3 cell line increased the number of proteins detected at the membrane from a value of 4.5 Ϯ 0.5% (n ϭ 25) in the nontransfected cell line to 23 Ϯ 1% (n ϭ 29) with Ca V ␤3 (p Ͻ 0.001) ( Table 1).
The results obtained with Ca V ␤3 contrast with the effect observed when co-expressing HA-tagged Ca V 1.2 with Ca V ␣2b␦ (p Ͼ 0.1). No further increase in the fluorescent signal was observed in the combined presence of the two auxiliary subunits. Similar results were obtained when Ca V 1.2 was transiently expressed, either in a background of stably transfected Ca V ␤3 or in a background of stably transfected Ca V ␣2b␦ cells (see Table 1 for numerical values). The maximum fluorescence obtained with Ca V ␤3 confirms that Ca V ␣2b␦ has little effect by itself on the Ca V 1.2 protein density at the plasma membrane. Among Ca V ␤ subunits, transient co-expression of Ca V 1.2-HA with Ca V ␤4 caused a similar boost in plasma membrane expression, whereas Ca V ␤2a was found to be slightly less potent for a Ca V ␤3 Ϸ Ca V ␤4 Ͼ Ca V ␤2a ranking (Table 1). Altogether, these results validated the fluorescence sorting analysis of HA-tagged Ca V ␣1 proteins to evaluate steady-state protein level in intact cells independently of channel gating.
␣ Interaction Domain: the Role of the WI Pair-Crystallographic analyses have shown that the AID-Ca V ␤ interaction is show a typical voltage-dependent activation with a mean current density of Ϫ7 Ϯ 2 pA/pF (n ϭ 6) for the wild type Ca V 1.2 channel in the stable Ca V ␤3 stable cell line as compared with a current density of Ϫ41 Ϯ 9 pA/pF (n ϭ 7) for the wild type Ca V 1.2 channel measured in the same cell line after transient transfection with Ca V ␣2b␦ subunit. The activation potential of 3 Ϯ 3 mV for the Ca V show a typical voltage-dependent activation with a mean current density of Ϫ9 Ϯ 2 pA/pF (n ϭ 5) for the Ca V 1.2-HA channel in the stable Ca V ␤3 stable cell line as compared with a current density of Ϫ36 Ϯ 8 pA/pF (n ϭ 7) for the Ca V 1.2-HA channel measured in the same cell line after transient transfection with Ca V ␣2b␦ subunit. The activation potential of 2 Ϯ 3 mV for the Ca V 1.2-HA/Ca V ␤ was shifted to Ϫ12 Ϯ 2 mV in the presence of Ca V ␣2b␦. Patch clamp experiments were carried out in the whole cell configuration in the presence of a 2 mM Ca 2ϩ saline solution.
anchored through a set of six residues, Asp, Leu, Gly, Tyr, Trp, and Ile, distributed among three ␣-helical turns of the I-II linker of Ca V 1.2 (21)(22)(23), with the WI pair of residues being most critical for the AID-Ca V ␤ protein interaction (29). To evaluate whether the AID-Ca V ␤ interaction controls the plasma membrane targeting of Ca V 1.2, the HA-tagged Ca V 1.2 subunit was transiently co-expressed in HEKT cells or in Ca V ␤3 stable cells. Ca V ␤3 stimulated the plasma membrane targeting of N-terminal mutants: L464A, G466A, G466F, Y467G, Y467A, Y467S, and Y467F (Fig. 3A, supplemental Fig. S4, and supplemental Table SII). When compared with the control situation (ϮCa V ␤3), there was a 4 -8-fold stimulation in the plasma membrane targeting mutations in the order G466A Ϸ G466F Ͼ Y467F Ͼ Y467G Ͼ L464A Ϸ Y467A Ͼ Y467S. Hence, no single point mutation in the N-terminal region of the AID completely abolished the Ca V ␤3-induced stimulation in the plasma membrane targeting of Ca V 1.2. In contrast, the Ca V ␤3 stimulation effect was completely eradicated in the double mutant G466A/ Y467F, even though each individual mutation behaved like the wild type channel, suggesting that each residue contributes to the high affinity interaction with Ca V ␤3.
Point mutations in the C-terminal WI pair yielded a different picture. I471L was the only mutant that was detected at the membrane to the same extent as the wild type channel in the presence of Ca V ␤3. However, Ca V ␤3 stimulated significantly the plasma membrane targeting of I471A and I471F mutants. W470Y, W470F, W470A, W470G, I417G, and I471R were not significantly different in the presence or in the absence of Ca V ␤3 ( Fig. 3C and supplemental Table SII). Western blots carried out in total cell lysates with the anti-HA confirmed that all of the Ca V 1.2 mutants tested produced proteins with the expected molecular weight (Fig. 3, B and D). Immunofluorescence microscopy confirmed that W470A disrupted the plasma membrane targeting of Ca V 1.2 in the presence of Ca V ␤3 (supplemental Fig. S5). Membrane expression of Ca V 1.2 in cultured hippocampal neurons was also disrupted after mutation of the key tryptophan residue to alanine (58). Furthermore, double mutations in the same region completely eradicated the Ca V ␤3 stimulation of Ca V 1.2 plasma membrane targeting (supplemental Fig. S6 and Table SI). Partial (⌬458 -463) or complete (⌬458 -475) removal of the AID-binding site within the I-II loop yielded similar results, confirming that no other low affinity Ca V ␤ site within the Ca V ␣1 subunit could promote the plasma membrane targeting of Ca V 1.2 in the absence of the AID region. Isothermal titration calorimetry assays have substantiated that the affinity of Ca V ␤2a for the AID region of Ca V 1.2 decreased after single-point mutations of these residues. There was an Ͼ1000-fold increase in the K d with the W/A and I/A mutants, whereas alanine mutation of the Leu and Gly residues imparted a smaller 5-10-fold decrease in the Ca V ␤2a affinity (29). Sub-stitution of the tryptophan residue by either tyrosine or phenylalanine only partly compensated for the mutation, confirming the requirement of a residue containing a double aromatic ring at this position. For the neighboring isoleucine position, mutation with the conserved leucine residue was found to preserve the Ca V ␤3-induced membrane targeting of Ca V 1.2. For comparison, I387L in Ca V 2.3 was the only mutant tested in the WI pair that supported Ca V ␤3 binding and Ca V ␤3 modulation of gating (31). In contrast, none of the Ca V 1.2 mutations identified in the short QT syndrome, an inherited form of cardiac arrhythmia (59), was shown to affect the Ca V ␤3 plasma membrane targeting of Ca V 1.2 (supplemental Fig. S7 and Table SII). Altogether our results support a strong correlation between Ca V ␤3 binding affinity to the AID region as determined from fusion proteins and from isothermal titration calorimetry assays (29) and its role in promoting the targeting of Ca V 1.2 proteins at the plasma membrane. More importantly our results suggest that the molecular determinants that account for Ca V ␤3 binding to the AID region are also responsible for the Ca V ␤3-induced stimulation of the plasma membrane targeting of Ca V 1.2 proteins. The NK Domain of Ca V ␤ Is Essential for Plasma Membrane Targeting of Ca V 1.2-The observation that different Ca V ␤s, which all share a conserved core containing the SH3 and NK domains, cause different biophysical effects on Ca V ␣1 subunits suggests that other regions besides the conserved AID-ABP interaction, could influence channel conformational changes (13). The SH3 domain of Ca V ␤2a was found to bind to the I-II linker of Ca V 2.1 channels, suggesting that low affinity interactions outside of the AID-ABP interface could contribute to the full functional effects of the Ca V ␤ subunit (40). A few years later, however, the conserved AID-NK domain interaction was found to be necessary for Ca V ␤-stimulated Ca V 2.1 channel surface expression (60). To evaluate whether the AID-NK interaction controls the plasma membrane targeting of Ca V 1.2, the HA-tagged Ca V 1.2 subunit was transiently co-expressed in HEKT cells in the presence of Ca V ␤3 full-length or deleted constructs as well as with Ca V ␤3 fragments.
We found that the NK domain of Ca V ␤3 (180 -364) (Fig. 4A) was required for the plasma membrane targeting of Ca V 1.2 ( Fig. 4B and supplemental Table SIII). Targeted deletion of the SH3 domain between residues 57 and 123 preserved 80% of the Ca V 1.2 protein detected at the membrane (Fig. 4B). The deleted Ca V ␤3 ⌬57-123 construct preserved the typical Ca V ␤3 modulation of channel gating and inactivation current kinetics. Peak whole cell current density was not significantly affected (Fig. 5).
The Ca V ␣1-Ca V ␤ interaction appears to require the MMKQAL motif in the ␣3 helix of the NK domain and was identified in the crystal structure (21-23) as critical for the high affinity AID-ABP interaction. Indeed deletion of the 195-200 residue region of Ca V ␤3 completely abolished plasma membrane targeting of Ca V 1.2 (Fig. 4B), and the single point mutation M196A in Ca V ␤3, equivalent to M245 in Ca V ␤2a (29), significantly decreased plasma membrane targeting with only 13 Ϯ 1% (n ϭ 3) fluorescent cells (supplemental Table SIII). Nonetheless, the NK domain (181-362 fragment) alone was not sufficient for targeting Ca V 1.2 to the membrane (Fig. 4D). Only the larger fragment (58 -362) that includes part of the SH3 domain was found to stimulate significantly the plasma membrane targeting of Ca V 1.2. The integrity of the constructions was verified by Western blot (Fig. 4, C and E).
Calmodulin in the Plasma Membrane Targeting of Ca V 1.2-CaM interacts with multiple sites in the Ca V ␣1 subunit of Ca V 1.2 (48, 61), of which the C-terminal pre-IQ and IQ domains are best characterized. Constitutive CaM binding to the N terminus has also been reported (62). To determine whether low affinity binding of CaM to intracellular regions contributes to trafficking of Ca V 1.2 channels (50, 63), Ca V 1.2 was co-expressed with CaM wt or the dominant negative mutant of CaM (CaM 1,2,3,4 ) in HEKT control cells (supplemental Fig. S8) and in Ca V ␤3 stable cells. Overexpression of CaM wt or its negative dominant mutant in Ca V ␤3 stable cells did not significantly alter whole cell currents measured in the presence of 2 mM Ca 2ϩ with peak current densities of Ϫ8 Ϯ 2 pA/pF (n ϭ 9) (CaM wt) and of Ϫ9 Ϯ 3 pA/pF (n ϭ 9) (CaM 1,2,3,4 ), whereas co-expression with the latter significantly decreased calcium-dependent inactivation kinetics (supplemental Fig. S9). Cytometry flux assays also failed to show a change in the plasma membrane expression of Ca V 1.2 with or without Ca V ␤3 (Fig. 6 and supplemental Table SIV). These data contrast with previous reports that CaM 1,2,3,4 co-expression reduced peak Ca V 1.2 current amplitude in HEK cells compared with CaM co-expression (8). It suggests that Ca V ␤3 is the dominant subunit to promote plasma membrane targeting of Ca V 1.2 and that CaM does not act synergistically with Ca V ␤3 under these conditions.
Overexpression of CaM wt was reported to promote the plasma membrane targeting of Ca V 1.2 proteins in COS1 cells, provided there was a complete absence of Ca V ␤ (63). To test the hypothesis that CaM could chaperone Ca V 1.2 to the membrane in the presence of Ca V ␣2b␦ in our expression system, FACS experiments were carried out in the stable Ca V ␣2b␦ cell line. As seen in Fig. 6B, CaM was unable to increase the number of Ca V 1.2 proteins at the membrane in the absence of Ca V ␤3 under these conditions. Overexpression of CaM wt with the double mutant W470A/I471A (supplemental Table SIV) also failed to promote plasma membrane targeting of Ca V 1.2, thus ruling out a mechanism whereby low affinity binding of Ca V ␤ subunit to the AID region could mask the CaM effect.
The 1643-1666 fragment in the C terminus forms the high affinity (K d Ͼ 3 nM) IQ-binding domain that co-crystallized   Table 1 and supplemental Table SIV. B, HA-tagged Ca V 1.2 wt and the double W470A/I471A mutant were expressed transiently either in the stable Ca V ␤3 or in the stable Ca V ␣2b␦ cell line. Cell surface expression of the Ca V 1.2 protein was quantified as described for supplemental Fig. S2. The number of fluorescent cells was not significantly increased by overexpressing CaM wt in any cell line as compared with the control Ca V 1.2 cells (p Ͼ 0.05). Similar data were obtained with the W470A mutant that abrogated Ca V ␤3 binding and stimulation of Ca V 1.2 plasma membrane targeting. The numerical values can be found in Table 1 and supplemental Table SIV. with CaM (49). This high affinity binding site overlaps with the C-terminal "targeting domain" identified previously (42,64). To test the hypothesis that constitutive calmodulin binding to the IQ motif is required for plasma membrane targeting, FACS experiments were carried out after mutations of the aromatic residues responsible for the high affinity (K d Ϸ 3 nM) CaM binding (49). Complete deletion of the 1643-1666 fragment did not alter surface labeling, whereas the W470A mutation in the ⌬IQ channel eliminated plasma membrane targeting of Ca V 1.2 (supplemental Table SIV), suggesting that the IQ domain is not likely to act as a retention signal. Furthermore, point mutations Q1655A, Y1657A, and F1658A, as well as multiple mutations I1654A/F1658A and Y1657A/F1658A, and I1654A/Y1657A/ F1658A did not alter plasma membrane targeting of Ca V 1.2. Plasma membrane targeting was not affected by a triple mutation in the pre-IQ domain (Ca V 1.2 T1651A/F1652A/L1653A) and was modestly supported in the I1654A and I1654A/ Q1655A mutants (Fig. 7 and supplemental Table SIV). It should be remembered that the I/A mutation only moderately affected Ca 2ϩ /CaM binding to the C-terminal peptide of Ca V 1.2 as compared with the I/E mutant (65). Deleting the larger 1623-1666 region, identified as an important targeting domain (42), completely eradicated the plasma membrane expression of Ca V 1.2 both in the presence and in the absence of Ca V ␤3 (Fig. 7 and supplemental Table SIV). Furthermore, as shown by others before (50), the Ca V 1.2 protein could not be detected at the membrane in the presence of the triple T1591A/L1592A/ F1593A mutation. The three TLF residues are located in a pre-IQ apocalmodulin-binding site (peptide A) (48,66), but overexpression with CaM wt or CaM 1,2,3,4 did not rescue plasma membrane targeting (supplemental Table SIV). Altogether, these data highlight the role of the C terminus in the plasma membrane targeting of Ca V 1.2 and suggest that high affinity Ca 2ϩ /CaM binding is not critical for the plasma membrane targeting of Ca V 1.2.

DISCUSSION
To exhibit functional activity, ion channels must be targeted to the plasma membrane. Co-expression of Ca V 1.2 with either Ca V ␣2b␦ or CaM (alone or in combination) failed to promote significantly the detection of fluorescently labeled Ca V 1.2-HA channels in intact cells by flow cytometry. Co-expression of Ca V 1.2 in the presence of Ca V ␤3 with either Ca V ␣2␦ or CaM failed to further increase the number of Ca V 1.2 proteins detected at the plasma membrane. Furthermore, plasma membrane targeting of AID-disrupted Ca V 1.2 mutants (thus in the absence of high affinity Ca V ␤ binding) could not be recovered by overexpressing the calmodulin protein alone or in combination with the auxiliary Ca V ␣2b␦ subunit, suggesting that Ca V ␤ is the critical auxiliary subunit in the plasma membrane targeting of Ca V 1.2.
Plasma membrane targeting of Ca V 1.2 was decreased but not abolished in the double I1654A/Q1655A mutant in the presence of Ca V ␤3 and was not altered in the absence of Ca V ␤3, thus ruling out molecular models whereby the IQ motif, containing a polybasic motif, could serve as an endoplasmic reticulum retention signal (67). However, Ca V ␤-induced membrane expression was unaffected by the complete deletion of the high affinity Ca 2ϩ /CaM IQ-binding site (⌬C1643-1666), suggesting that high affinity Ca 2ϩ /CaM binding to this domain (1643-1666) is not required. In contrast, deletion of the larger 1623-1666 region in the C terminus and mutation of the TLF site, located in a pre-IQ apocalmodulin-binding site, abolished plasma membrane targeting of Ca V 1.2 in agreement with previous reports (43,44,50). Clearly the C terminus of Ca V 1.2 harbors key site targeting signals, but our current data do not support a model whereby high affinity Ca 2ϩ /CaM binding to the IQ domain (1643-1666) is required. Whether the C terminus is required for proper protein folding or for anchoring apocalmodulin onto low affinity CaM-binding sites on the C terminus (50,63) as in SK channels (51,68) will await further structural studies.
Ca V ␣2␦ and Surface Targeting of Ca V 1.2-It is widely acknowledged that Ca V ␣2␦ subunits facilitate the voltage dependence of channel gating and potentiate whole cell currents in a number of recombinant HVA Ca V ␣1-Ca V ␤ subunit combinations (69 -71). For instance, electrophysiological measurements carried out with Ca V 2.3 in HEK cells showed a small but significant increase in peak current density in the presence of Ca V ␣2␦ that arose, at least in part, from an increase in the number of functional channels (72). Mutating the von Willebrand factor-A domain of Ca V ␣2␦-2 was shown to increase the intracellular localization of Ca V 1.2 or Ca V 2.2 throughout the cell (45), suggesting that Ca V ␣2␦ could play a role in the plasma membrane targeting of HVA Ca V ␣1 subunits. It can, however, be argued that the 30% decrease of Ca V 1.2 at the cell surface assessed from cell surface biotinylation assays cannot fully account for the 3-fold decrease in channel peak current density observed with the Ca V ␣2␦-2 mutant (45). Indeed, it has been shown elsewhere that the gating charge of Ca V 1.2 is unaffected by co-expression with Ca V ␣2␦, supporting a model where Ca V ␣2␦ stimulates peak currents by setting Ca V 1.2 channels in a conformational state very close to the open state without increasing protein density (57). Our current electrophysiological and flow cytometry measurements also support a role for Ca V ␣2␦ in the modulation of Ca V 1.2 channel gating rather than plasma membrane protein density. At this point, however, our data do not exclude a role for Ca V ␣2␦ at early steps of protein biosynthesis and/or recycling that could not be detected in our steady-state assay.
Ca V ␤ Stimulated Surface Labeling of Ca V 1.2-Co-expression with Ca V ␤ subunits produced a 3-5-fold enhancement in the plasma membrane targeting of the Ca V ␣1 subunit of Ca V 1.2, with Ca V ␤3 and Ca V ␤4 being the most potent effects. No other combination with or without the auxiliary CaM and/or the Ca V ␣2b␦ subunit produced any significant increase in the plasma membrane targeting of Ca V 1.2. Hence, Ca V ␤ appears to be the most potent determinant in the plasma membrane targeting of Ca V 1.2. Our detailed mutational analysis supports a model where high affinity binding of the ABP of Ca V ␤ to the AID helix of Ca V ␣1 largely accounts for Ca V ␤induced plasma membrane targeting of Ca V 1.2.
The mechanism whereby Ca V ␤ antagonizes ER retention of the Ca V ␣1 subunit remains debated (73). In the two-site model, Ca V ␤ stimulation of protein expression and modulation of gating are controlled by distinct sites through a two-to-one stoichiometry. This model opens up the possibility that secondary Ca V ␤-binding sites could contribute to plasma membrane targeting. Several observations could be suggestive of such a mechanism. Surface expression of Ca V 1.2-HA channels in Xenopus oocytes was not increased by injection of the Ca V ␤2a protein, and even decreased gating currents and surface expression of the Ca V 1.2-⌬AID-expressing oocytes (36,74). Covalently linking Ca V ␤2b to the C terminus of Ca V 1.2 stimulated whole cell currents but failed to modulate channel gating in HEK cells (37,38). Ca V ␤2b-induced modulation of trafficking and gating was also uncoupled in N-terminally truncated Ca V 1.2 (39). Deletion of a low affinity interaction site between the SH3 module of Ca V ␤ and the I-II linker of the Ca V 2.1 subunit (outside the AID-GK interaction) did not affect Ca V 2.1 protein trafficking (40). Small fragments of Ca V ␤2 arising from putative splice variants were also shown to bind to the C terminus of the Ca V 1.2 subunit where they promoted membrane targeting in the absence of the GK/SH3 module of Ca V ␤ subunits (75,76). It remains, however, difficult to assess the physiological relevance of these findings, given that they result from in vitro interaction studies between isolated peptides.
In the one-site model, Ca V ␤ interacts sequentially with the Ca V ␣1 subunit through a unique binding site in the I-II linker in a 1:1 stoichiometry to dislodge ER retention signals and modulate gating (25). In the intact channel, high affinity binding of Ca V ␤ onto the AID motif would account for both Ca V ␤-induced modulation of gating and the Ca V ␤-plasma membrane trafficking of Ca V ␣1 (32,77,78). For Ca V 1.2 expressed in a mammalian cell system, the Trp 470 and Ile 471 residues previously shown to account for the high affinity binding of Ca V ␤ (29) onto the Ca V 1.2 subunit were herein found to account for the Ca V ␤ stimulation of Ca V 1.2 plasma membrane targeting. As mentioned earlier, disruption of these residues alone or in combination had a dominant effect and abrogated cell surface labeling of Ca V 1.2. Our data hence support a model whereby high affinity binding of the MMQKAL motif of Ca V ␤ to the AID helix of the Ca V ␣1 subunit is required for chaperoning and modulating HVA Ca V channels.