Proteolytic cleavage of the hydrophobic domain in the CaVα2δ1 subunit improves assembly and activity of cardiac CaV1.2 channels

Voltage-gated L-type CaV1.2 channels in cardiomyocytes exist as heteromeric complexes with the pore-forming CaVα1, CaVβ, and CaVα2δ1 subunits. The full complement of subunits is required to reconstitute the native-like properties of L-type Ca2+ currents, but the molecular determinants responsible for the formation of the heteromeric complex are still being studied. Enzymatic treatment with phosphatidylinositol-specific phospholipase C, a phospholipase C specific for the cleavage of glycosylphosphatidylinositol (GPI)-anchored proteins, disrupted plasma membrane localization of the cardiac CaVα2δ1 prompting us to investigate deletions of its hydrophobic transmembrane domain. Patch-clamp experiments indicated that the C-terminally cleaved CaVα2δ1 proteins up-regulate CaV1.2 channels. In contrast, deleting the residues before the single hydrophobic segment (CaVα2δ1 Δ1059–1063) impaired current up-regulation. CaVα2δ1 mutants G1060I and G1061I nearly eliminated the cell-surface fluorescence of CaVα2δ1, indicated by two-color flow cytometry assays and confocal imaging, and prevented CaVα2δ1-mediated increase in peak current density and modulation of the voltage-dependent gating of CaV1.2. These impacts were specific to substitutions with isoleucine residues because functional modulation was partially preserved in CaVα2δ1 G1060A and G1061A proteins. Moreover, C-terminal fragments exhibited significantly altered mobility in denatured immunoblots of CaVα2δ1 G1060I and CaVα2δ1 G1061I, suggesting that these mutant proteins were impaired in proteolytic processing. Finally, CaVα2δ1 Δ1059–1063, but not CaVα2δ1 G1060A, failed to co-immunoprecipitate with CaV1.2. Altogether, our data support a model in which small neutral hydrophobic residues facilitate the post-translational cleavage of the CaVα2δ1 subunit at the predicted membrane interface and further suggest that preventing GPI anchoring of CaVα2δ1 averts its cell-surface expression, its interaction with CaVα1, and modulation of CaV1.2 currents.

In cardiac cells, Ca 2ϩ signals control the force necessary for the myocardium to meet the physiological needs of the body (1). During diastole, the intracellular free ionized Ca 2ϩ is maintained in the nanomolar range by the concerted action of mechanisms that prevent Ca 2ϩ entry, promote its extrusion (mostly via the Na ϩ /Ca 2ϩ exchanger), and ensure its storage in the sarcoplasmic reticulum (2). Upon depolarization, Ca 2ϩ enters the cell through the cardiac high-voltage-activated L-type Ca V 1.2 channel and initiates the myocardium contraction via Ca 2ϩinduced Ca 2ϩ release from the sarcoplasmic reticulum. Regulation of the L-type Ca 2ϩ current has profound physiological significance. Alterations in density or the activation/inactivation gating of L-type Ca V 1.2 channels have been implicated in a variety of cardiovascular diseases such as hypertension (3), atrial fibrillation (4 -7), heart failure (8,9), and congenital arrhythmias (10 -12).
L-type Ca V 1.2 channels are heteromultimeric protein complexes composed of the main pore-forming Ca V ␣1 subunit non-covalently bound to the cytoplasmic Ca V ␤ auxiliary subunit (Ca V ␤1-␤4), the EF-hand protein calmodulin (constitutively bound to the C terminus of Ca V ␣1), and Ca V ␣2␦1 subunit (13)(14)(15)(16)(17). The full complement of auxiliary subunits is required to produce Ca V 1.2 channels with the typical biophysical and biochemical properties of the native cardiac channels (18). Ca V ␤ promotes the cell-surface trafficking of Ca V 1.2 channels through a high-affinity interaction (19) in part by preventing its degradation by the ubiquitin/proteasome system (20). Ca 2ϩ -dependent inactivation of Ca V 1.2 channels ensues following the interaction of Ca 2ϩ with intracellular calmodulin (21). Co-expression of Ca V ␣2␦1 with Ca V ␤-bound Ca V ␣1 promotes the activation of Ca V 1.2 at more physiological voltages (22)(23)(24)(25)(26)(27) by stabilizing the channel voltage sensors (28).
Over the last 15 years, structural studies have revealed the high-affinity interaction between Ca V ␤ and Ca V ␣1 as well as the Ca 2ϩ -calmodulin/Ca V ␣1 association by X-ray crystallography (29). By contrast, there was until recently little structural information on Ca V ␣2␦1. The reason can be found in the extreme complexity of the Ca V ␣2␦1 protein topology that results from multiple co-and post-translational modifications. Ca V ␣2␦1 arises from a single gene and is post-translationally cleaved into the large extracellular Ca V ␣2 and the single-pass transmembrane Ca V ␦ proteins bound by disulfide bridges (30 -33). Ca V ␣2␦1 includes 20 cysteine residues, and it has been argued that intra-molecular disulfide bonds are required to stabilize its higher order structure. In addition, Ca V ␣2␦1 is glycosylated at 16 asparagine sites, which are required for the protein folding and stability (27,34). These features represent significant hurdles for carrying the conventional protein expression and purification in bacterial systems. The three-dimensional (3D) structure of the skeletal muscle Ca V 1.1 channel, recently solved by single particle cryo-electron microscopy (cryo-EM), provides so far the best high-resolution look at the native Ca V ␣2␦1 protein purified from a rabbit skeletal muscle T-tubule preparation in complex with the pore-forming Ca V ␣1 from Ca V 1.1 (Fig. 1A) (17,35). The extracellular portion of the Ca V ␣2␦1 protein is constructed around five structural domains as follows: Cache1, von Willebrand factor A, Cache2, Cache3, and Cache4. There was, however, insufficient electron density to support amino acid assignment in a few specific regions, most notably in the "913-972" region and at the C-terminal domain between residues 1074 and 1106 (rabbit numbering) (17). These gaps in the 3D structure might be a consequence of the unintended action of proteases during the purification process, intrinsically disordered segments, or from an actual posttranslational deletion. Although more than 11 different proteases can theoretically target 250 different sites in Ca V ␣2␦1 (36), the two deletions identified in the structure correspond to regions previously identified. From the first report of its purification, Ca V ␣2␦1 is known to be cleaved in the "913-972" region into Ca V ␣2 and Ca V ␦ proteins (30,32,37). The cleavage of the C-terminal region is also compatible with the proposed attachment of Ca V ␣2␦1 at the cell membrane through a glycosylphosphatidylinositol (GPI) 4 anchor (32,38,39) at/or around residue Cys-1074 (rabbit isoform). To date, more than 150 different human proteins are known to be GPI-anchored (40). This process, also referred to as glypiation, occurs in the endoplasmic reticulum where the C terminus of type 1 transmembrane proteins is cleaved to be replaced by a GPI anchor (41)(42)(43)(44)(45)(46)(47)(48)(49). GPI modification promotes localization with membrane microdomains (40,50) and/or confers biological activity by espousing the optimal conformation for protein-protein interaction (51).
Herein, we show that the native Ca V ␣2␦1 protein from rat cardiomyocytes is a substrate for prokaryotic phosphatidylinositol-phospholipase C (PI-PLC). Deletion of the last 24 C-terminal residues, including the hydrophobic domain of the rat Ca V ␣2␦1, had little impact on Ca V 1.2 currents demonstrating that the C-terminally cleaved Ca V ␣2␦1 proteins achieve the conformation congruent with the modulation of Ca V 1.2 channels. In contrast, deletion of four residues surrounding Cys-1059 (rat isoform), which was the last amino acid identified in the 3D structure, impaired up-regulation of Ca V 1.2 currents.
The migration profile of the C-terminal fragments was also significantly altered by single mutations in the "1059 -1061" region, suggesting that the proteolytic cleavage was influenced by the chemical nature of the side chain in the site. More importantly, mutations of the predicted GPI-anchor sites markedly reduced the plasma membrane localization of Ca V ␣2␦1 proteins and prevented its co-immunoprecipitation with Ca V 1.2. Altogether, our data are compatible with a model where GPIanchored Ca V ␣2␦1 proteins are preferentially assembled and trafficked to the cell surface with the Ca V 1.2 channel complex.

Residues at the membrane interface in Ca V ␣2␦1 are essential for the functional modulation of Ca V 1.2 currents
The high-resolution 3D cryo-EM structure of the purified rabbit Ca V ␣2␦1 subunit did not resolve the last 31 C-terminal residues, deduced from the nucleotide sequence suggesting that these residues are cleaved in the mature protein and replaced by a GPI anchor in the skeletal L-type Ca 2ϩ channel ( Fig. 1A) (17). To examine the presence of the GPI-anchored Ca V ␣2␦1 subunit in cardiomyocytes, rat ventricular myocytes were treated with prokaryotic phosphatidylinositol-phospholipase C (PI-PLC), which cleaves GPI anchors and releases lipid-anchored proteins (52). As seen, the cell-surface fraction of the PLC-treated cardiac Ca V ␣2␦1 proteins was significantly decreased (41 Ϯ 5%, n ϭ 3) but not completely eradicated (Fig.  2), suggesting that transmembrane and GPI-anchored forms of Ca V ␣2␦1 may co-exist in native tissues as shown for other proteins (53). To evaluate the functional impact of the C-terminal residues, deletion mutants of the rat mCherry-Ca V ␣2␦1-HA construct (Fig. 1B) were produced, and whole-cell currents were measured after recombinant expression with Ca V 1.2 and Ca V ␤3 (Fig. 3A). The mCherry-Ca V ␣2␦1-HA construct, described earlier (25,27), sports a constitutive fluorescent mCherry signal at the C terminus allowing the robust detection of C-terminal fragments on Western blots as well as confirming the translation of the protein. A 9-residue hemagglutinin (HA) tag from the human influenza virus, inserted in the extracellular domain, enables the identification of Ca V ␣2 at the cell surface using a FITC-conjugated anti-HA antibody.
As reported before (25,27), co-expression of the rat Ca V ␣2␦1 WT construct with Ca V 1.2 WT enhanced whole-cell peak current densities from Ϫ2.5 Ϯ 0.3 pA/pF (n ϭ 34) (mock vector) to Ϫ15 Ϯ 1 pA/pF (n ϭ 109) in the presence of 2 mM Ca 2ϩ as the charge carrier (Fig. 3B). The increase in peak current density was associated with a ϷϪ20-mV leftward shift in the activation potential of Ca V 1.2 suggesting that Ca V ␣2␦1 WT might increase channel function in part by improving the channel open state (Fig. 3C). Deletion of 24 residues within the C terminus (mCherry-Ca V ␣2␦1-HA ⌬1061-1085, Fig. 1B) produced proteins that expressed at the cell surface (Fig. 3D) and up-regulated Ca V 1.2 currents in a typical fashion with a Ϸ5-fold increase in peak current density and a ϷϪ16-mV shift in the activation potential (Table 1 for details). Smaller deletions of ⌬1062-1068, ⌬1062-1085, and ⌬1069 -1085 were also consistently found to modulate Ca V 1.2 currents. In contrast, the deletion of four residues in the "1059 -1063" region produced volt- 4 The abbreviations used are: GPI, glycosylphosphatidylinositol; ⌬Gact, free energy of activation; HEKT, human embryonic kidney 293 cells stably expressing an SV40 temperature-sensitive T antigen; ⌬MedFI, relative median fluorescent intensity; PI-PLC, phosphatidylinositol-specific phospholipase C; WGA, wheat germ agglutinin; pF, picofarad; LDLR, LDL receptor.
age-activated currents with biophysical properties indistinguishable from currents measured in mock-transfected cells, although Ca V 1.2/Ca V ␤3 activated at voltages significantly more negative in the presence of the ⌬1059 -1063 construct than in the complete absence of Ca V ␣2␦1. All constructs, even the nonfunctional Ca V ␣2␦1 ⌬1059 -1063, produced proteins with the expected molecular masses (Western blotting data, not shown) demonstrating that proteins were appropriately translated. In addition, all constructs were expressed at the cell surface as attested by the surface fluorescence signals measured using a flow cytometry assay (25,27,54). Nonetheless, the cell-surface fluorescence signals measured for these deleted constructs were reduced by 80 -90% as compared with the mCherry-Ca V ␣2␦1-HA WT (Table 2). It should be noted that the ⌬1059 -1063 construct emitted fluorescence in the upper limit of this range with 29 Ϯ 3% (n ϭ 6) suggesting that a defect in its cell-surface trafficking could not fully account for the decreased functional regulation of Ca V 1.2/Ca V ␤3 currents. These data confirm that the hydrophobic domain (1069 -1085) in Ca V ␣2␦1 is not required to produce functional Ca V 1.2 channels. In contrast, some or all residues included within the 1059 -1063 proximal region are essential for modulating the activity of the channel.

Alanine-substituted Ca V ␣2␦1 mutants are compatible with up-regulation of Ca V 1.2 currents
GPI anchoring usually includes a precise sequence of residues formed by a linker region localized 11 residues before the cleavage site "," and a spacer region of 5-7 amino acids before the transmembrane hydrophobic domain (Fig. 1B) (49,53). The "" cleavage site is usually formed by small amino acids (glycine, alanine, cysteine, valine, or serine) such as the ones found in the 1059 -1063 region, which includes Cys-1059, Gly-1060, Gly-1061, Val-1062, and Ser-1063. Algorithm searches conducted with PredGPI (55) and Big-PI (49) suggest that Gly-1061 or Val-1062 could form a functional cleavage site in Ca V ␣2␦1. Given that the post-translational modification by the GPItransamidase requires small amino acids, residues 1059 -1062 were substituted either with the larger amino acid isoleucine (124 Å 3 ) or with the smaller alanine residue (67 Å 3 ) (56). In addition, the C1059S mutant was also tested because it is predicted to form a disulfide bridge with the neighboring Cys-404 residue in the 3D cryo-EM structure of Ca V 1.1 (17). As seen, C1059A produced typical voltage-activated inward Ca 2ϩ currents, whereas substitutions with serine or glycine impaired up-regulation of Ca V 1.2 currents despite the observation that all these mutants were detected at the cell surface with similar intensities (Fig. 4 and Tables 1 and 2).
Ca V ␣2␦1 G1060A, G1061A, and V1062A within the putative cleavage site for GPI-anchoring produced mutant proteins that functionally modulated Ca V 1.2 currents ( Fig. 5 and Table  1). These mutants were largely detected at the cell surface with the relative fluorescence signals being 60 -80% of the signals measured for mCherry-Ca V ␣2␦1-HA WT ( Table 2). Although channel function is compatible with single substitution with alanine residues at either position 1060 or 1061, the double alanine-substituted mCherry-Ca V ␣2␦1-HA construct G1060A/G1061A failed to increase Ca V 1.2/Ca V ␤3 currents ( Fig. 5 and Table 1). The cell-surface fluorescence was, how-  (25). The color code in the primary sequence of the transmembrane ␦ protein identifies the putative regions required for recognition by the glycosylphosphatidylinositol transamidase to carry the post-translational modification. These regions were determined by combining results obtained with multiple algorithms (Big-PI, PredGPI, and GPI-SOM) (49,55). The transmembrane domain was determined from a GlobPlot analysis (73) (please note that the JBC is not responsible for the long-term archiving and maintenance of this site or any other third party hosted site.). Single-letter codes were used to represent the amino acids. ever, equivalent for C1059A, C1059S, and G1060A/G1061A thus ruling out a change in the cell-surface expression as the unique mechanism responsible for the decreased function in these mutants.
The same amino acids were then substituted one at a time with the bulkier isoleucine residue, predicted to prevent GPI cleavage. The functional impact of the isoleucine mutants was strongly position-specific. Isoleucine substitution at position Cys-1059 (Fig. 4) and Gly-1060 ( Fig. 6) yielded voltage-activated currents with properties that were nearly indistinguishable from the ones produced with the mock vector ( Table 1). As seen with the ⌬1059 -1063 construct, G1061I produced Ca V 1.2/Ca V ␤3 currents with reduced peak densities but still activated on average at more negative voltages. Finally, wholecell currents measured with V1062I were similar to currents measured with mCherry-Ca V ␣2␦1-HA WT.
The decrease of whole-cell currents measured in the presence of Ca V ␣2␦1 C1059I, G1060I, and G1061I mutants was associated with a near-elimination of their cell-surface fluorescence, with relative fluorescence intensities (1-16% of the fluorescence intensities measured for the WT construct) (Figs. 4D and 6D and Table 2). This was especially true for G1060I, which was not significantly detected at the cell surface in intact cells (no right shift in the x axis), although the protein produced a robust mCherry fluorescent signal (upper shift in the y axis). However, G1060I was detected by the FITC-conjugated anti-HA antibody in permeabilized cells confirming that the protein was translated and that the HA epitope remained accessible. Although V1062I also experienced a significant decrease in its cell-surface expression, with a relative fluorescence intensity that was 40 -50% lower than measured for the WT construct (Table 2 and Figs. 6 and 8), it did not prevent up-regulation of whole-cell currents. These results show that the correlation between the cell-surface fluorescence of the Ca V ␣2␦1 mutants and the up-regulation of Ca V 1.2 currents is not linear, in agreement with previous reports (25,27).
Altogether, these results show that volume matters more than hydrophobicity at positions 1059, 1060, or 1061 for the cell-surface localization of Ca V ␣2␦1 and the functional modulation of Ca V 1.2 currents. In particular, channel function requires an alanine or a cysteine residue at position 1059 and is not prevented by the introduction of an alanine residue at positions 1060 or 1061 in Ca V ␣2␦1, as long as there is at least one glycine residue at either position. More importantly, it can be concluded that the decrease in the cell-surface expression observed when amino acid substitutions were made near the putative GPI-cleavage site was not the only factor responsible for the reduction in channel function.

Alanine-substituted Ca V ␣2␦1 mutants are cleaved by PLC
GPI-linked proteins are tethered to the outer leaflet of the cell plasma membrane resulting in a protein that can be released in the medium by the action of the PI-PLC enzyme (46,57,58). The surface distribution of mCherry-Ca V ␣2␦1-HA, GFP-GPI, and a non-GPI-anchored GFP-tagged protein (LDLR-GFP) was analyzed by imaging live cells. LDLR-GFP is a fusion protein of the low-density lipoprotein receptor, which does not form a GPI anchor but rather uses the full transmembrane domain, in tandem with a green fluorescent protein (GFP) tag at its C terminus (59 -61). The GFP-GPI construct was used as a positive control (60,62). As seen, the LDLR-GFP, GFP-GPI, and mCherry-Ca V ␣2␦1-HA proteins all co-localized with wheat germ agglutinin (WGA) at the cell surface under control conditions (Fig. 7A) as illustrated by the bright white border produced by merging the fluorescence intensities for the test construct and the cell-surface marker. Compared with typical transmembrane proteins, GPI-anchored proteins reside in higher proportion on the cell surface (53). Incubation with 3 units/ml PI-PLC significantly impaired the surface localization of mCherry-Ca V ␣2␦1-HA WT proteins (Fig. 7, A and B, panels ii) with a global reduction of 47 Ϯ 7% (Ϸ2500 cells) in the HA-FITC fluorescence relative to the fluorescence signal for WGA. This reduction was similar to the impact of the PI-PLC treatment on the cell-surface fluorescence of the GPI-anchored GFP protein. In contrast, the cell-surface fluorescence of LDLR-GFP was relatively constant (90 -120% of the fluorescence signal measured in control dishes) (Fig. 7, A and B, panels iii). Cell-surface expression of the alanine-substituted Ca V ␣2␦1 G1060A and G1061A proteins was also significantly reduced after incubation with PI-PLC (Fig. 7, A and B, panels iv and v) suggesting that these mutations do not prevent the formation of the GPI anchor.
The release of plasma membrane-bound Ca V ␣2␦1 by PI-PLC was further confirmed quantitatively using large-volume high-  The current density scale is either 5 or 10 pA/pF, as indicated. B, averaged current-voltage relationships for the deleted constructs. The absolute peak currents densities measured with mCherry-Ca V ␣2␦1-HA WT varied from Ϫ5 to Ϫ44 pA/pF over the 8-month recording period with a mean of Ϫ15 Ϯ 1 pA/pF (n ϭ 109). Averaged peak current densities obtained with the mock mCherry vector are shown. Co-expression with Ca V ␣2␦1 left-shifted the voltage dependence of activation of Ca V Except for ⌬1059 -1063, the peak current densities of the deleted constructs were similar to values measured in the presence of mCherry-Ca V ␣2␦1-HA WT. Statistical analyses were performed with a one-way ANOVA test: *, p Ͻ 0.01 and **, p Ͻ 0.001 against the mock vector. See Table 1 for details. C, distribution of the free energies of activation. The free energies of activation (⌬Gact) measured in the presence of the mock vector and in the presence of mCherry-Ca V ␣2␦1-HA WT are centered at 0.5 Ϯ 0.1 and Ϫ0.69 Ϯ 0.03 kcal mol Ϫ1 , respectively. The distribution of the ⌬Gact values for the deleted constructs but for ⌬1059 -1063, overlapped with the ⌬Gact values measured in the presence of mCherry-Ca V ␣2␦1-HA WT. D, representative two-dimensional plots of mCherry versus FITC fluorescence. The cell-surface expression of the mCherry-Ca V ␣2␦1-HA construct was measured using a flow cytometry assay with 10,000 intact cells. The construct allows for detection of N-and C-terminal fragments using, respectively, a FITC-conjugated anti-HA (x axis of left, middle, and right panels) and an anti-mCherry (y axis of left panels). The robust mCherry signal (y axis of left panels) confirms that the proteins were translated up to the end of the coding sequence. The total protein fluorescence (right panels) was estimated from the relative HA-FITC signal estimated after cell permeabilization from the same pool of cells. The distribution of the fluorescence intensity measured for the fluorescence-positive cells are shown in gray, whereas the distribution of fluorescence intensity for the non-transfected cells is displayed as a single transparent peak. For the mutants shown herein, the cell-surface fluorescence for FITC was significantly higher than the background fluorescence measured with the mock vector (data not shown) but significantly lower than measured for the WT construct. The single straight line drawn at the maximum of the fluorescence signal for the WT construct was reported in all bar graphs to better visualize the changes in the signal. Numerical values are shown in Table 2. Table 1 Biophysical properties of Ca V 1.2 channels Ca V 1.2 WT was co-expressed with Ca V ␤3 and either pmCherry-mock vector, pmCherry-Ca V ␣2␦1-HA WT, or pmCherry-Ca V ␣2␦1-HA mutants using a 4:4:4 g ratio. Biophysical parameters were measured in the presence of 2 mM Ca 2ϩ as described elsewhere (25,27). Activation properties (E 0.5, act and ⌬Gact) were estimated from the mean I-V relationships and fitted to a Boltzmann equation. Only the cells with voltage-activated currents were kept for further analysis. Null-current cells outnumbered the positive-current cells for the following mutants: G1060I (24 null cells); G1061A (8 null cells); and G1060A/G1061A (15 null cells). The data are shown as the mean Ϯ S.E. of the number of cells (one cell per experiment), and the total number of experiments carried over several months is shown in parentheses. Statistical analysis was carried out against the values obtained in the presence of the mock pmCherry-N1 vector. *, p Ͻ 0.01; **, p Ͻ 0.001.

Mobility of the C-terminal fragments is altered in isoleucinesubstituted glycine mutants
Our data suggest that preventing the proteolytic cleavage at the level of Gly-1060 and Gly-1061 impairs cell-surface localization of Ca V ␣2␦1 and functional modulation of Ca V 1.2 currents. To confirm that these glycine residues play a role in the proteolytic cleavage of Ca V ␣2␦1, the mobility of the C-terminal fragments was examined in denaturing immunoblots of the mCherry-Ca V ␣2␦1-HA fusion protein (Fig. 9). As seen in Fig. 9A, the N-terminal fragment of the highly glycosylated mCherry-Ca V ␣2␦1-HA WT construct was detected by the Ca V ␣2 antibody at 170 kDa. Given that N-linked glycosylation alters the protein mobility by Ϸ50 kDa (27)  Recordings were made in the presence of 2 mM Ca 2ϩ from a holding potential of Ϫ100 mV. Time scale is 100 ms throughout. The current density scale is either 5 or 10 pA/pF. B, averaged current-voltage relationships. Averaged peak current densities obtained with the mock mCherry vector are shown. The peak current densities of mCherry-Ca V ␣2␦-HA WT, C1059S, and C1059A were not significantly different from one another, whereas mCherry-Ca V ␣2␦-HA C1059I behaved like the mock vector. Statistical analyses were performed with a one-way ANOVA test: *, p Ͻ 0.01, and **, p Ͻ 0.001 against the mock vector. See Table 1 for details. C, distribution of the free energies of activation. The free energy of activation (⌬Gact) for C1059S and C1059A overlapped with the values measured in the presence of mCherry-Ca V ␣2␦1-HA WT, whereas the C1059I mutant behaved like the mock vector. D, representative two-dimensional plots of mCherry versus FITC fluorescence are shown for the mutants shown in A. Cell-surface expression decreased significantly for mCherry-Ca V ␣2␦-HA C1059I. Analysis was carried out as described in Fig. 3. The single straight line drawn at the maximum of the fluorescence signal for the WT construct was reported in all bar graphs to better visualize the changes in the signal. Numerical values are shown in Table 2. Table 2 Cell surface expression of Ca V ␣2␦1 constructs Ca V 1.2 WT was co-expressed in stable Ca V ␤3 HEKT cells with pmCherry-Ca V ␣2␦1-HA WT or pmCherry-Ca V ␣2␦1-HA GPI mutant using a 1:1 weight DNA ratio (4:4 g). Flow cytometry experiments were conducted to determine cell-surface expression of tagged proteins. Fluorescence intensity was estimated with the FlowJo software. Relative expression of Ca V ␣2␦1 was calculated based on ⌬MedFI estimated for fluorophore FITC as compared with the value obtained for the WT construct measured the same day under the same conditions. The relative cellsurface expression was expressed as the ratio of ⌬MedFI for FITC in intact cells over the value of ⌬MedFI for FITC in permeabilized cells. The total number of experiments is provided in parentheses, with each experiment being the result of a different transfection in a separate cell dish (Ͼ10,000 cells). Statistical analysis was carried out against the ⌬MedFI for FITC measured with pmCherry-Ca V ␣2␦1-HA WT. *, p Ͻ 0.01; **, p Ͻ 0.001.
kDa protein is the main species detected in the plasma membrane fractions (Fig. 9B). Two C-terminal fragments of 46 and 33 kDa were detected by the anti-mCherry in the total cell lysates (Fig. 9, A and B). The more intense 33-kDa band is compatible with the production of the hydrophobic domain (3.7 kDa) attached to mCherry (28.8 kDa) ϭ 32.5 kDa. The fainter 46-kDa band is compatible with the protein being cleaved in the "898 -956" region (rat numbering) (17), thus producing a band at 45.8 kDa (17-kDa Ca V ␦ ϩ 28.8-kDa mCherry). Ca V ␣2␦1 hence appears to be simultaneously cleaved at the level of the GPI anchor and at the level of the Ca V ␦ protein. The C-terminal fragments of the alanine-substituted G1060A and G1061A mutants behaved essentially like the WT construct with the more intense band showing up at 33 kDa and a weaker band at 46 kDa. In contrast, the 33-kDa band was absent or much weaker in the isoleucinesubstituted G1060I and G1061I mutants, although the larger C-terminal 46-kDa fragment was clearly present. In addition, both proteins were detected by the anti-Ca V ␣2␦1 in total cell lysates at the expected molecular mass of 205 kDa but not in the plasma membrane fraction from the G1061I-transfected cells (n ϭ 3 different transfections and cell fractionation assays) (Fig.  9B). These data suggest that the presence of an isoleucine residue at positions 1060 or 1061 (G1060I and G1061I) results into a larger fraction of Ca V ␣2␦1 proteins not being cleaved at the level of the GPI-anchor.

Heteromeric protein assembly is impaired in the Ca V ␣2␦1 ⌬1059 -1063 construct
Given that the decrease in channel function observed with the GPI-impaired mutants could not solely result from a diminution in the cell-surface expression, we last examined whether GPI-anchored Ca V ␣2␦1 proteins were preferentially interacting with Ca V 1.2/Ca V ␤3 channels. The uncoupling between cell-surface expression and channel modulation suggests indeed that Ca V ␣2␦1 might be expressed at the surface without interacting with Ca V 1.2 as suggested by recent single-molecule imaging studies (64). and stained for imaging as described under "Experimental procedures." The GFP-GPI protein confirmed the activity of PI-PLC, whereas the pmCherry mock vector monitored the cell transfection efficiency (panels ii and iii). All mCherry-Ca V ␣2␦1-HA proteins (panels i, iv, and v) displayed a strong mCherry signal thus confirming that the protein was translated up to the C terminus. Close to 2500 cells expressing the mCherry-Ca V ␣2␦1-HA WT protein were examined for each condition in six paired experiments (12-cell dishes) that were conducted over an 8-month period. For mCherry-Ca V ␣2␦1-HA WT, PI-PLC caused a reduction of 47 Ϯ 7% (22-66% of the control signal) in the HA-FITC fluorescence density at the cell surface relative to the fluorescence signal for WGA in control dishes. For comparison, PI-PLC caused a 40 Ϯ 10% reduction (30 -80% of the control signal) in the cell-surface fluorescence for GFP-GPI, and no significant change for the cell-surface fluorescence of LDLR-GFP (90 -127% of the control signal). Furthermore, PI-PLC impaired the cell-surface expression of pmCherry-Ca V ␣2␦1-HA G1060A with a 50 Ϯ 8% reduction in the relative intensity for the HA-FITC surface labeling (n ϭ 507 cells without PI-PLC and n ϭ 436 cells with PI-PLC) and a smaller 32 Ϯ 10% reduction for G1061A (n ϭ 733 cells without PI-PLC and n ϭ 382 cells with PI-PLC). Altogether, these data suggest that mCherry-Ca V ␣2␦1-HA G1060A and G1061A proteins are GPI-anchored. Scale bar, 10 m.
Co-immunoprecipitation assays were carried out by pulling down Ca V ␣2␦1 WT, G1060A, and ⌬1059 -1063 from the antic-Myc-coated beads in the presence of Ca V 1.2/Ca V ␤3-c-Myc (Fig 10). As seen, Ca V ␣2␦1 WT (Fig. 10A) and G1060A (Fig.  10B) were well expressed (input lanes) and yielded robust signals after column elution. In contrast, Ca V ␣2␦1 ⌬1059 -1063 (Fig. 10C) did not seem to interact strongly with Ca V 1.2/Ca V ␤3, although the input signal was quite strong. In fact, the pulldown assays were performed after loading onto the beads twice the amount of starting material for Ca V ␣2␦1 ⌬1059 -1063 (10 g in Fig. 10C compared with 5 g of protein for A and B). Even under these conditions, the signal for Ca V ␣2␦1 WT was stronger than for Ca V ␣2␦1 ⌬1059 -1063. The weaker interaction is further emphasized when comparing the different exposure times (Fig. 10C, panels i-iii). The relative intensity of the signal in the bound fraction was about Ϸ60 times stronger for the WT than for the ⌬1059 -1063 construct despite similar intensities in the input lanes. Control Western blotting of proteins found in the unbound fraction, oth-erwise known as the flow-through fraction, confirmed that ⌬1059 -1063 was present in the protein mixture as were Ca V ␣2␦1 WT and G1060A (Fig. 10C, panel iv). These data suggest that the cleavage and subsequent GPI modification may confer in part the biological activity of Ca V ␣2␦1 by promoting the optimal conformation for the interaction between Ca V 1.2 and Ca V ␣2␦1.

Residues Cys-1059 -Gly-1062 in Ca V ␣2␦1 are essential for the functional modulation of L-type Ca V 1.2 channels
The Ca V ␣2␦1 protein is known to increase whole-cell currents of high voltage-activated Ca 2ϩ channels. In particular, Ca V ␣2␦1 promotes the channel activation gating of the L-type Ca V 1.2 channel (22-27) by stabilizing the channel voltage sensors in repeats I-III (28,35). In contrast, Ca V ␣2␦1 augments whole-cell currents of Ca V 2.2 and Ca V 2.3 channels with little changes in their activation potential (24, 37, 63, 67) supporting . Two-dimensional dot plots are shown in the leftmost graphs. The single straight line drawn at the maximum of the fluorescence signal for the WT construct was reported in all bar graphs to better visualize the changes in the signal. Analysis was carried out as described in Fig. 3. The distribution of the cell-surface FITC fluorescence intensity is shown in the middle histogram, and the distribution of the total mCherry fluorescence intensity is shown in the right histogram. The cell-surface fluorescence for FITC was significantly lower for C1059S relative to the fluorescence of the WT construct but was similar between V1062I and the WT construct. Of note, the relative fluorescence for mCherry was similar for the WT construct, C1059S, and V1062I. C, total fluorescence intensity for the control condition was normalized to 100% for each fluorophore. PI-PLC caused a 35 Ϯ 3% reduction (32-41%) in cell-surface FITC fluorescence of mCherry-Ca V ␣2␦1-HA (n ϭ 6 transfections). D and E, PI-PLC released 45 Ϯ 4% (n ϭ 3) and 33 Ϯ 2% (n ϭ 3) of the C1059S and V1062I surface-bound proteins supporting the view that C1059S and V1062I do not prevent the formation of a GPI anchor. The single straight line drawn at the maximum of the fluorescence signal for the WT construct was reported in all bar graphs to better visualize the changes in the signal.
the view that association of Ca V ␣2␦1 within the heteromeric complex might depend upon the molecular makeup of the Ca V ␣1 subunit (64).
Herein we show that deleting five residues between Cys-1059 and Ser-1063 (⌬1059 -1063) impaired the negative shift in the activation potential as well as the up-regulation of Ca V 1.2 cur-   2). Immunoblotting was carried out on total WT (20 g), G1060A (20 g), and ⌬1059 -1063 (40 g) proteins before the immunoprecipitation assay (input lane). The signal for the housekeeping protein GAPDH is shown below each blot. Co-immunoprecipitation assays were carried out with mCherry-Ca V ␣2␦1-HA WT (5 g), G1060A (5 g), and ⌬1059 -1063 (10 g) proteins. Images were captured after short (1 s, panels i) or longer exposure times (30 s and 200 s, panels ii and iii). Ca V ␤3 and Ca V 1.2 proteins migrated, respectively, at 60 and 250 kDa. All Ca V ␣2␦1 proteins migrated at Ϸ175 kDa, which is consistent with cleavage at or around Cys-1059. Ca V ␣2␦1-HA WT and G1060A consistently yielded a stronger co-immunoprecipitation signal than Ca V ␣2␦1-HA ⌬1059 -1063 at all exposure times despite a larger amount of starting material. In average, the normalized signal for Ca V ␣2␦1-HA ⌬1059 -1063 bound to the Ca V 1.2-Ca V ␤3 complex was 60 Ϯ 10 times smaller than the luminescent signal for Ca V ␣2␦1-HA WT. The quantification was carried out by reporting the ratio of the intensity measured under non-saturating conditions (30-s exposure) normalized by the signal measured in the input lane over the GAPDH signal. All immunoblots were carried out in parallel under the same transfection and extraction conditions. The mCherry-Ca V ␣2␦1-HA ⌬1059 -1063 proteins that did not bind to the antibody-bead complex (referred to as the unbound fraction) were collected, diluted in a Laemmli buffer, and fractionated by SDS-PAGE using an 8% gel and revealed with the anti-Ca V ␣2␦1 (panel iv). As seen, mCherry-Ca V ␣2␦1-HA ⌬1059 -1063 migrated at 175 kDa confirming that the proteins were appropriately translated and were present in the preparation in detectable quantities throughout. All experiments, carried out four times with both mutants and eight times for the WT construct over a period of 2 months, yielded similar results. rents. In contrast, deleting larger segments within the C terminus of Ca V ␣2␦1 (⌬1061-1085, ⌬1062-1068, and ⌬1069 -1085 constructs) did not prevent the up-regulation of L-type Ca V 1.2 currents. This is in agreement with a previous study showing that deleting the hydrophobic domain of Ca V ␣2␦1 after Cys-1059 (to form the equivalent ⌬1060 -1091) was not incompatible with modulation of Ca V 2.1 currents (66). The Ca V ␣2␦1 deletion constructs in our study as well as the ␣ 2 ␦-1⌬Cterm-HA construct (66) displayed a significant decrease in their cell-surface expression, yet only Ca V ␣2␦1 ⌬1059 -1063 failed to up-regulate Ca V 1.2 currents. This impaired modulation occurred despite the cell-surface expression of Ca V ␣2␦1 ⌬1059 -1063. It is hence tempting to conclude that the proteolytic cleavage of the hydrophobic domain is required to achieve the optimal interaction/orientation between the extracellular Ca V ␣2 and the voltage sensors (28,35) of the Ca V ␣1 subunit in Ca V 1.2. This interpretation is supported by the co-immunoprecipitation data showing that the interaction between the two proteins was significantly decreased when the hydrophobic "transmembrane" domain was present in the Ca V ␣2␦1 ⌬1059 -1063 construct. Hence, the molecular nature of the residues within that region in Ca V ␣2␦1 controls indirectly the activity of Ca V 1.2 channels.
Up-regulation of Ca V 1.2 currents was compatible with the presence of the small hydrophobic alanine residues at positions Cys-1059, Gly-1060, and Gly-1061. Although the behavior of the single alanine-substituted mutants was reasonably similar to the WT protein, the double mutant Ca V ␣2␦1 G1060A/ G1061A disrupted the cell-surface expression and failed to promote Ca V 1.2 currents indicating that a glycine residue at either position 1060 or 1061 is required for Ca V ␣2␦1 to function normally. In contrast, the isoleucine-substituted Ca V ␣2␦1 C1059I, G1060I, and G1061I mutants significantly 1) reduced the steady-state cell-surface localization of Ca V ␣2␦1 as characterized by confocal imaging, Western blotting of membrane protein fractions, and two-color flow cytometry assays; and 2) prevented the Ca V ␣2␦1-mediated increase in the peak current density and voltage-dependent gating of Ca V 1.2. Of note, the functional impact of these single mutations was greater than many single mutations preventing N-linked glycosylation (27). Functional modification was limited to these three positions because V1062I caused only a minor decrease in the cell-surface expression of Ca V ␣2␦1 and did not prevent the up-regulation of Ca V 1.2 currents. Only Gly-1060 is conserved in isoforms Ca V ␣2␦-2 and -3. This observation, coupled with the patch-clamp data showing that Ca V ␣2␦1 ⌬1061-1085 upregulated Ca V 1.2 currents, suggests that a small side chain at position 1060 is crucial for carrying surface anchoring of Ca V ␣2␦1.

Enzymatic cleavage of Ca V ␣2␦1 by PLC
The proposed proteolytic cleavage of the hydrophobic domain in Ca V ␣2␦1 agrees with the 3D structure of the Ca V 1.1 channel complex (17) and supports the GPI anchoring of the protein. GPI anchoring is generally inferred from the predicted primary amino acid sequence, but enzymatic cleavage of the membrane-bound protein by PI-PLC further supports GPI anchoring (53). As shown in this work, the cell-surface expres-sion of recombinant Ca V ␣2␦1 in HEKT cells and native Ca V ␣2␦1 proteins from cardiomyocytes was significantly reduced by an enzymatic treatment with PI-PLC. As reported herein, only some of the surface-bound Ca V ␣2␦1 WT proteins were released by PLC. GPI-anchored proteins are known to exist in vivo as heterogeneous mixtures with considerable variation in their glycan core modifications and lipid moieties (51,53) and in the composition of the amide-linked fatty acid forming the GPI anchor (57), which in turn can influence the affinity of PLC for its substrate. It is thus impossible to conclude whether the remaining fraction of non-releasable proteins was localized in surface microdomains inaccessible to PI-PLC or whether the remaining pool of Ca V ␣2␦1 proteins located at the cell surface was simply resistant to PI-PLC and thus not GPI-anchored. It is important to note, however, that surface-bound Ca V ␣2␦1 WT, C1059S, and V1062I proteins, which modulated Ca V 1.2 currents, were cleaved by PI-PLC to a similar extent. This observation suggests that a similar fraction of the Ca V ␣2␦1 proteins at the cell membrane is PLC-cleavable despite individual alterations in their cell-surface trafficking.

Molecular determinants underlying the cleavage of Ca V ␣2␦1
The strongest biochemical evidence linking the mutations of the glycine residues to a modification in the proteolytic processing of Ca V ␣2␦1 protein came from investigating the mobility of the C-terminal fragments using the mCherry antibody. A short 33-kDa C-terminal fragment was reproducibly associated with Ca V ␣2␦1 WT and the alanine-substituted mutants (G1060A, G1061A, and V1062A), all forms of the protein that supported whole-cell current up-regulation. This fragment is compatible with the migration of the short hydrophobic domain coupled with the mCherry protein. In contrast, this fragment was absent in the isoleucine-substituted Ca V ␣2␦1 G1060I and G1061I proteins, suggesting that proteolytic cleavage of the C terminus was affected by the mutation. Of note, all proteins produced a C-terminal fragment of 46 kDa, compatible with an additional cleavage site at a location predicted to overlap with the N-terminal end of the Ca V ␦1 protein. Given that the isoleucine-substituted Ca V ␣2␦1 C1059I, G1060I, and G1061I but not V1062I failed to up-regulate Ca V 1.2 currents, these data suggest that the sequence Cys-1059 -Gly-1060 -Gly-1061 forms the proteolytic cleavage of Ca V ␣2␦1. Deleting this region significantly reduced the co-immunoprecipitation of Ca V ␣2␦1 with Ca V 1.2/Ca V ␤3 channels suggesting that preventing the cleavage of the C terminus and subsequent membrane anchoring of Ca V ␣2␦1 impaired the heteromeric assembly of the L-type Ca V 1.2 channel. Altogether, our data are compatible with a model where the proteolytic cleavage and GPI anchoring of Ca V ␣2␦1 promotes the heteromeric assembly and the activity of the L-type Ca V 1.2 channel.

Recombinant DNA techniques
The rabbit Ca V 1.2 (GenBank TM X15539) and the rat Ca V ␤3 (GenBank TM M88751) were subcloned in commercial vectors under the control of the CMV promoter as described elsewhere (25,67). The coding sequence (1091 residues) of the rat brain Ca V ␣2␦1 clone (GenBank TM NM_012919) (68) was subcloned in the pmCherry-N1 vector, and the hemagglutinin (HA) epitope (YPYDVPDYA) was inserted in the extracellular domain of Ca V ␣2 between Asp-676 and Arg-677, as described ( Fig. 1B) (27,54). Point mutations were produced with the Q5 site-directed mutagenesis kit (New England Biolabs Inc., Whitby, Ontario, Canada) in the pmCherry-Ca V ␣2␦1-HA construct according to the manufacturer's instructions as described elsewhere (25,27). Ca V ␣2␦1 deletion mutants were produced by PCR overlap extension using non-mutagenic forward and reverse primers flanking the region to be deleted. Following the PCR, a kinase/ligase/DpnI enzyme mix was added to the amplified DNA for circularization and template removal before transformation into high-efficiency DH5-␣competent Escherichia coli. Constructs were verified by automated double-stranded sequence analysis (Genomics Platform, IRIC, Université de Montréal, Quebec, Canada). Protein expression of all constructs was confirmed by Western blotting in total cell lysates as described below (19,25,27).

Immunoblotting of total cell lysates from HEKT cells
Twenty four hours after transfection, cells were washed twice with ice-cold PBS and lysed with a RIPA buffer (150 mM NaCl, 1% IGEPAL, 0.5% sodium deoxycholate, 0.1% SDS, 50 mM Tris (pH 8.0)) containing a protease inhibitor mixture, including 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, aprotinin, bestatin, E-64, leupeptin, and 1 mM EDTA (Sigma) for 30 min at 4°C. The cell lysates were sonicated and centrifuged at 13,000 rpm for 30 min at 4°C. Supernatant was collected, and proteins were quantified with the Pierce BCA protein assay kit (Thermo Fisher Scientific, Ottawa, Ontario, Canada). Immunoblotting was carried out with fresh lysates. Proteins were mixed with the Laemmli sample buffer in the presence of 0.4 mM 2-mercaptoethanol and electrophoresed on an 8% SDS-polyacrylamide gel alongside the Precision Plus Protein TM dual color standard (Bio-Rad). After electroblotting and blocking with 5% (w/v) skim milk for 30 min, the supported nitrocellulose membranes (Bio-Rad) were incubated with the anti-Ca V ␣2␦1 (Alomone, Jerusalem, Israel, 1:1000) or the anti-mCherry (Biovision, 1:20,000). The extracellular N-terminal fragments were identified using the anti-Ca V ␣2␦1, whereas the anti-mCherry was used to detect C-terminal fragments. Membranes were stripped and incubated with an anti-GAPDH as a loading control (Sigma, 1:25,000) unless stated otherwise. Signals were detected with the ECL substrate. Blots were visualized with the ChemiDoc Touch documentation system (Bio-Rad). Molecular weights were estimated using Image Lab TM Software version 5.2 (Bio-Rad) by linear regression of standard molecular weight markers. The glycosylated native Ca V ␣2␦1 protein migrated at 160 kDa (27). The calculated molecular mass of the recombinant glycosylate mCherry-Ca V ␣2␦1-HA construct is 205 kDa. GAPDH migrated as a monomer close to 37 kDa in accordance with its calculated mass.

Isolation of the plasma membrane fraction from recombinant HEKT cells
Four different protein fractions (total cell lysates, cytosolic, total membrane, and plasma membrane fraction) were prepared as explained before (27,69). Briefly, transfected HEKT cells cultured in 100-mm dishes were homogenized at 4°C in a Tris-based solution containing a mixture of protease inhibitors (Sigma) and 1 mM EDTA at pH 7.4. The cell homogenate was aliquoted into three tubes. After a 2-h incubation period at 4°C with 1% (v/v) Triton X-100, the first tube was centrifuged at 10,000 ϫ g for 10 min to remove cell debris, nuclei, and mitochondria. The supernatant was kept as the total protein fraction (whole-cell lysates). The second tube was centrifuged at 200,000 ϫ g and 4°C for 20 min. The supernatant is referred to as the cytosolic fraction. The pellet was resuspended in homogenizing buffer containing 1% (v/v) Triton X-100. After 30 min of incubation on ice, a second centrifugation was done at 200,000 ϫ g. The resulting supernatant is referred to as the total membrane protein fraction. The third tube was centrifuged at 10,000 ϫ g for 10 min. The supernatant obtained was centrifuged at 200,000 ϫ g and 4°C for 20 min. The pellet was resuspended in the homogenizing buffer containing 0.6 M KCl. Subsequent centrifugations were performed at 200,000 ϫ g and 4°C for 20 min to wash out the KCl. The final pellet was resuspended in the homogenizing buffer and is considered to be enriched in plasma membrane proteins. Proteins were electrophoresed on an 8% SDS-polyacrylamide gel and blotted with the anti-Ca V ␣2␦1 (Alomone, 1:1000), anti-mCherry (Biovision, 1:20,000), anti-GAPDH (Sigma, 1:25,000), and anti-pan-cadherin (Life Technologies, Inc., 1:5000) as a marker of the plasma membrane fraction. The extracellular N-terminal fragments were identified using the anti-Ca V ␣2␦1, whereas the anti-mCherry was used to detect C-terminal fragments.

Enzymatic cleavage and immunoblotting of adult rat cardiomyocytes
Experiments were approved by the Animal Protection Committee of the Montreal Heart Institute (protocol no. 2014-1775; 2014-10-01) and were performed in accordance with the guidelines of the Canadian Council for Animal Care and the Guide for the Care and Use of Laboratory Animals 8th Edition (2011). Male Sprague-Dawley adult rats (150 -200 g) (Charles River Laboratories, St. Constant, Canada) were anesthetized with a mix of pentobarbital/heparin. Hearts were quickly removed and placed on ice-cold Tyrode's solution containing 130 mM NaCl, 5.4 mM KCl, 1 mM MgCl, 0.33 mM Na 2 HPO 4 , 10 mM HEPES, 5.5 mM glucose, and 1 mM CaCl 2 (pH 7.4). Ventricular myocytes were then isolated as described elsewhere (70). Briefly myocytes were homogenized at 4°C in a Tris-based solution containing a mixture of protease inhibitors (Sigma), including 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, aprotinin, bestatin, E-64, leupeptin, and 1 mM EDTA at pH 7.4 (69). Cell lysates were incubated 2 h either with 5 units/ml phosphatidylinositol-specific phospholipase C enzyme (Life Technologies, Inc.) (52) or with the vehicle solution. Protein fractions (total cell lysates, cytosolic, total membrane, and plasma membrane fraction) were isolated and immunoblotted as explained above (27,69

Phosphatidylinositol-specific phospholipase C digestion of recombinant proteins
HEKT cells were transfected with Ca V 1.2, Ca V ␤3, and mCherry-Ca V ␣2␦1-HA constructs. Cells were dissociated 24 h after transfection and for confocal imaging and seeded for 6 h in Dulbecco's modified Eagle's medium high glucose ϩ 1% fetal bovine serum. Cells were incubated for 2-16 h with phosphatidylinositol-specific phospholipase C (Life Technologies, Inc.) at a concentration of 3 units/ml in a solution containing 20 mM Tris-HCl, 1 mM EDTA, 0.01% sodium azide, and 50% glycerol at pH 7.5 (52). Half the cells were incubated under the same conditions with the vehicle solution. PI-PLC-treated cells were used immediately for live-cell imaging or for flow cytometry assays.

Live-cell imaging
HEKT Ca V ␤3 stable cells were transiently transfected with pCMV-Ca V 1.2 WT and pmCherry-Ca V ␣2␦1-HA WT or mutants and stained 24 h after transfection. The plasma membrane was visualized with wheat germ agglutinin-Alexa 647 (WGA-647) (1:200, Life Technologies, Inc.) (25,27). WGA is a carbohydrate-binding protein that recognizes sialic acid and N-acetylglucosaminyl sugar residues. An anti-HA FITC-conjugated antibody (10 g/ml) (clone HA-7, Sigma) was used to detect the HA-tagged Ca V ␣2␦1 protein, and the nuclei were stained with DAPI (1:1000) (Life Technologies, Inc.) in 1ϫ PBS for 45 min at 4°C. Confocal fluorescent images were captured between 1 and 3 h after staining with a Zeiss LSM 710 confocal microscope system equipped with a ϫ63/1.4 oil objective. The 488-nm laser was used to detect either FITC or GFP fluorescence, and the mCherry fluorescence was measured at 594 nm. The fluorescence of the plasma membrane marker WGA-647 was measured at 633 nm. The images were analyzed using the FIJI software to delete background, subtract noise, and to produce co-localization pixel maps (shown in white) between the 488-and 633-nm channels using the co-localization finder plugin from ImageJ (71).
To determine the cell-surface expression level of the mCherry-Ca V ␣2␦1-HA proteins, cells were harvested 24 h after transfection, washed in 1ϫ PBS buffer, and stained with the FITCconjugated mouse monoclonal anti-HA epitope tag antibody at 5 g/ml (Sigma) or with the control IgG1-FITC murine isotype control (5 g/ml) at 4°C for 30 min. To determine the total quantity of both intracellular and extracellular expression of the tagged proteins, cells were fixed and permeabilized using BD Cytofix/Cytoperm TM fixation/permeabilization solution kit (BD Biosciences). Roughly 10,000 cells were counted using a FACSAria III SORP flow cytometer (BD Biosciences) at the flow cytometry facility hosted by the Department of Microbiologie, Infectiologie, and Immunologie at the Université de Montréal. The fluorescence intensity detected with the IgG1-FITC isotype control murine (5 g/ml) or with the anti-HA FITC-conjugated antibody (5 g/ml) in HEKT untransfected cells was not significantly different from the fluorescence measured in the complete absence of fluorophore (25). The following control conditions were carried out in triplicate with each series of experiments: (a) untransfected Ca V ␤3 cells without the anti-HA FITC-conjugated antibody; (b) untransfected Ca V ␤3 cells with the anti-HA FITC-conjugated antibody to assess the level of background staining; and (c) Ca V ␤3 cells transfected with pCMV-Ca V 1.2 and pmCherry-Ca V ␣2␦1-HA WT. Expressing mCherry-Ca V ␣2␦1-HA WT in HEKT cells produced a 3-log increase in the FITC (x axis) and mCherry fluorescence (y axis) on two-dimensional dot plots (leftmost plots in Figs. 3D, 4D, 5D, and 8, A and B), as shown previously (25,27,54).

Quantification of steady-state cell-surface expression by flow cytometry assays
Flow cytometry data were analyzed using the FlowJo software, version 10 (TreeStar, Ashland, OR) as described (25). Relative expression of Ca V ␣2␦1 was calculated based on ⌬median fluorescence intensity (⌬MedFI) for each fluorophore (mCherry or FITC) as explained elsewhere (54). Briefly, the gates for the positive cells (P2) and the negative cells (P3) were set manually on the two-dimensional dot plots. On the fluorescence histograms shown in the middle and right panels of Figs. 3D, 4D, 5D, and 8, A and B, the cell count (y axis) is shown as a function of the fluorescence intensity (x axis) within the region delineated by the P3 (negative cells) and the P2 (positive cells) gates. The ⌬MedFI for FITC was calculated by subtracting the FITC fluorescence density of the negative cells (P3) from the fluorescence density of the FITC-positive cells (P2). The same method was used to calculate the ⌬MedFI for mCherry (data not shown). ⌬MedFI for FITC measured in intact non-permeabilized cells was used as a relative index of the steady-state cellsurface expression of the HA-tagged Ca V ␣2␦1 (middle panels in Figs. 3D, 4D, 5D, 8A, and 8B), whereas the ⌬MedFI for mCherry attested that the protein was translated until the C terminus. The ⌬MedFI values for FITC were also measured after cell permeabilization (right panels in Figs. 3D, 4D, 5D, 8A, and 8D). This procedure was especially important for the mutants that failed to generate significant cell-surface fluorescence as a means to confirm the accessibility of the HA epitope.
⌬MedFI values were normalized to the maximum value measured the same day for mCherry-Ca V ␣2␦1-HA WT expressed under the same conditions to account for variations in the absolute fluorescence intensity of the anti-HA FITC-conjugated antibody. We are thus reporting changes in the relative cell-surface fluorescence as compared with the fluorescence intensity of the mCherry-Ca V ␣2␦1-HA WT. The normalized ⌬MedFI values for mCherry measured for each mutant in intact and permeabilized cells were not significantly different from one another (p Ͼ 0.1) (data not shown) suggesting that the cell permeabilization procedure did not distort significantly the relative fluorescence readout under most conditions.

Patch-clamp experiments in HEKT cells
Whole-cell patch-clamp experiments were carried out on isolated cells after transfection in HEKT cells in the presence of the peGFP vector coding for the GFP (0.2 g) as a control for transfection efficiency. Electrodes were filled with a solution containing (in mM) the following: 140 CsCl; 0.6 NaGTP; 3 MgATP; 10 EGTA; 10 HEPES; titrated to pH 7.3 with NaOH with a resistance varying between 2.8 and 3.2 megohms. Cells were bathed in a modified Earle's saline solution (in mM) as follows: 135 NaCl; 20 tetraethylammonium chloride; 2 CaCl 2 ; 1 MgCl 2 ; 10 HEPES, titrated to pH 7.3 with KOH. GFP-positive cells were selected for patching. On-line data acquisition was achieved with the Axopatch 200-B amplifier (Molecular Devices, Sunnyvale, CA) connected to the PClamp software Clampex 10.5 through the Digidata 1440A acquisition system (Molecular Devices) (25). A series of 450-ms voltage pulses were applied from a holding potential of Ϫ100 mV at a frequency of 0.2 Hz, from Ϫ60 to ϩ70 mV at 5-mV intervals. Series resistance was compensated to ϳ85% after on-line capacitive transient cancellation. Unless stated otherwise, whole-cell currents were sampled at 5 kHz and filtered at 1 kHz. PClamp software Clampfit10.5 was used for data analysis. Midpotentials of activation values (E 0.5, act ) were estimated from the peak I-V curves obtained for each channel composition and were reported as the mean of individual measurements Ϯ S.E. (25,72). The free energy of activation was calculated using the mid-activation potential shown in Equation 1, ⌬G act ϭ z ⅐ F ⅐ E 0.5, act (Eq. 1) where z is the effective charge displacement during activation, and F is the Faraday constant (65). The r100 ratio is defined as the ratio of peak whole-cell currents remaining after a depolarizing pulse of 100 ms (I 100 ms /I peak ) and was used as an indicator of the inactivation kinetics (reported in Table 2). To assess for internal consistency, the experiments carried out with novel mutants always included a control experiment performed with mCherry-Ca V ␣2␦1-HA WT (pCMV-Ca V 1.2 WT ϩ pCMV-Ca V ␤3 ϩ pmCherry-Ca V ␣2␦1-HA WT) thus explaining the larger sample size for mCherry-Ca V ␣2␦1-HA WT. Previous experiments confirmed that mCherry-Ca V ␣2␦1-HA WT sustains the functional modulation of Ca V 1.2 currents (25). Experiments performed under the same conditions yielded peak current densities that could vary by as much as Ϯ35% between each series of experiments. This variation appeared to be essentially linked to minor changes in the cell density at the time of transfection. Data from all experiments performed under the same conditions over a period of 10 months were pooled, and biophysical properties are reported in Table 2. Experiments were performed at room temperature (20°C).

Statistics
Results were expressed as mean Ϯ S.E. Tests of significance were carried out using the unpaired ANOVA with the Tukey test embedded in the Origin 7.0 analysis software (OriginLab Corp., Northampton, MA). Data were considered statistically significant at *, p Ͻ 0.01, and **, p Ͻ 0.001.
Author contributions-E. S. produced single and multiple mutants, performed and analyzed flow-cytometry experiments, conducted patch-clamp experiments, performed live-cell imaging, and carried out the immunoblotting of the cell fractions. B. B. conducted patchclamp experiments and performed the rat cardiomyocyte experiments. M. P. T. performed and analyzed flow-cytometry experiments and provided immunoblotting of the mutants under denaturing conditions. J. B. prepared Fig. 1 and carried out the coimmunoprecipitation assays. B. G. A. provided the rat cardiomyocytes. G. M. provided constructs and supervised the phospholipase C experiments. L. P. designed and coordinated the study, interpreted the data, and wrote the manuscript. All authors reviewed the results and approved the final version of this manuscript.