Membrane Targeting of L-type Calcium Channels

In this study, we report that palmitoylation was a critical determinant of the subcellular localization of the rat β2a subunit of voltage-dependent calcium channels. Immunohistochemical staining of transfected cells revealed that a palmitoylation-deficient β2a subunit exhibited a diffuse intracellular staining pattern, in contrast to the plasma membrane distribution seen with the wild-type β2asubunit. Unexpectedly, mutations in regions distal to the palmitoylation sites at Cys3 and Cys4 affected palmitoylation of the β2a protein. Mutations in ansrc homology 3 motif of the β2a subunit affected both palmitoylation and subcellular localization of the β2a protein. A mutation in the β interaction domain, which disrupted interactions between the expressed α1 and β subunits, also resulted in a decreased palmitoylation and diffuse intracellular localization of the β2a protein. Studies of chimeric proteins revealed that the 16-amino acid N terminus of the β2a subunit was sufficient to confer palmitoylation to the nonpalmitoylated β1b and β3 isoforms. However, palmitoylation of chimeric β subunits was by itself insufficient to restore the plasma membrane localization observed with the wild-type β2a protein. Treatment of transfected cells with brefeldin A increased the amount of palmitic acid incorporated in the β2a protein, suggesting that palmitoylation of β2a occurs during or shortly after protein synthesis. Two other β2 variants, the rabbit β2a and β2b, which lack the palmitoylation sties at Cys3 and Cys4, exhibited a diffuse intracellular staining pattern and were not palmitoylated.

In this study, we report that palmitoylation was a critical determinant of the subcellular localization of the rat ␤ 2a subunit of voltage-dependent calcium channels. Immunohistochemical staining of transfected cells revealed that a palmitoylation-deficient ␤ 2a subunit exhibited a diffuse intracellular staining pattern, in contrast to the plasma membrane distribution seen with the wild-type ␤ 2a subunit. Unexpectedly, mutations in regions distal to the palmitoylation sites at Cys 3  Studies of chimeric proteins revealed that the 16-amino acid N terminus of the ␤ 2a subunit was sufficient to confer palmitoylation to the nonpalmitoylated ␤ 1b and ␤ 3 isoforms. However, palmitoylation of chimeric ␤ subunits was by itself insufficient to restore the plasma membrane localization observed with the wild-type ␤ 2a protein. Treatment of transfected cells with brefeldin A increased the amount of palmitic acid incorporated in the ␤ 2a protein, suggesting that palmitoylation of ␤ 2a occurs during or shortly after protein synthesis. Two other ␤ 2 variants, the rabbit ␤ 2a and ␤ 2b , which lack the palmitoylation sties at Cys 3 and Cys 4 , exhibited a diffuse intracellular staining pattern and were not palmitoylated.
Voltage-dependent calcium channels are heteromultimeric proteins composed of a pore-forming ␣ 1 subunit, which determines many of the biophysical and pharmacological properties of the channel, and at least two other modulatory subunits, termed ␣ 2 ␦ and ␤ (1,2). Although these channels have been extensively studied electrophysiologically, less is known about the biochemical properties of these proteins due to their rarity in native tissues. The ␣ 2 ␦ and ␤ subunits contain no homology to any known proteins and are involved in the modulation of channel properties. To date, four separate ␤ isoforms have been identified, each of which contains a central conserved core flanked by unique N-and C-terminal regions specific to each isoform (1). Although the ␤ subunits are highly hydrophilic proteins with no predicted membrane-spanning domains, we recently demonstrated that the cardiac ␤ 2a isoform was localized to the plasma membrane even in the absence of an ␣ 1 subunit (3).
Co-expression of an accessory ␤ subunit with an ␣ 1 subunit in heterologous mammalian cell systems results in an increase in the number of drug/toxin binding sites (3)(4)(5)(6)(7), an increase in peak current amplitude (3, 8 -11), and an increase in the number of channels at the cell surface (3,8,10,11). The increase in channels at the plasma membrane has been demonstrated both biochemically (3) and electrophysiologically (8,10,11), and probably accounts for the increased drug/toxin binding sites and the increase in peak current amplitude observed upon ␤ subunit co-expression. In addition, it has been demonstrated recently that calcium currents, charge movements, and the number of dihydropyridine receptors were largely reduced in skeletal muscle myotubes of ␤ 1 subunit null mice (12) but could be restored by transfection of the ␤ 1 subunit, suggesting that the ␤ subunit played important roles in maintaining the expression of the ␣ 1 subunits. The major identified ␣ 1 -␤ interaction site involved regions conserved among all known ␣ 1 and ␤ subunit isoforms (13,14), and further characterization of this domain revealed that interactions between different ␤ subunits and a specific ␣ 1 subunit were fairly similar in affinity (15). Likewise, it has been shown that all four known ␤ subunit isoforms were capable of modulating currents from channels containing the cardiac ␣ 1C subunit (6,(15)(16)(17)(18)(19).
Recently, we identified sites of palmitoylation in the rat ␤ 2a subunit of voltage-dependent calcium channels (8). Palmitoylation is a post-translational modification involving the reversible addition of a 16-carbon palmitic acid group to the cysteine residues of proteins through a labile thioester linkage (20). The mechanisms involved in the addition and removal of palmitic acid are still unclear, although palmitoyl transferases that exhibit catalytic selectivity for palmitic acid have been recently purified (21,22). Palmitoylation can be dynamically regulated due to the labile nature of the thioester bond, and studies have demonstrated the receptor-regulated depalmitoylation of certain proteins (23,24). Three other known ␤ subunit isoforms (␤ 1 , ␤ 3 , and ␤ 4 ) were found not to be palmitoylated (8), making palmitoylation the first identified biochemical modification unique to a specific ␤ subunit isoform. Consequently, palmitoylation could provide a mechanism for the selective regulation of voltage-dependent calcium channels containing a ␤ 2a subunit.
Palmitoylation of ␤ 2a required both the Cys 3 and Cys 4 residues in the N terminus, since site-directed substitution of either of these residues resulted in the loss of palmitate incorporation (8). The mutation of these residues in the palmitoylation-deficient ␤ 2a (C3S/ C4S) protein resulted in dramatic changes in whole-cell ionic con-ductance without affecting the number of functional channels at the cell surface (8). The studies presented herein describe the effects of palmitoylation on the subcellular localization of the ␤ 2a protein and further define the determinants of palmitoylation in the ␤ 2a sequence.

EXPERIMENTAL PROCEDURES
Materials-The large T-antigen-transformed human embryonic kidney cells (tsA201) were the generous gift of Dr. Richard Horn (Thomas Jefferson University). The Card C antiserum that detects the ␣ 1C subunit, the ␤ 2a antiserum that is specific for the ␤ 2a isoform, and the "␤-general" (␤ GEN ) antiserum that detects all ␤ subunit isoforms were previously described (3,8). Protein G-Ultralink resin and the Supersignal enhanced chemiluminescence detection kit were purchased from Pierce. [ 3 H]Palmitic acid was purchased from American Radiolabeled Chemicals (St. Louis, MO). Fluorescein isothiocyanate-and tetramethylrhodamine isothiocyanate-coupled secondary antibodies for immunohistochemical studies were purchased from Kirkegaard & Perry Laboratories (Gaithersburg, MD). All other reagents were from standard sources.
Cell Culture and Transfection Protocols-Cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% fetal bovine serum (Life Technologies) and 1% penicillin/streptomycin at 37°C in 5% CO 2 . Transfections were performed using the HEPES-buffered calcium phosphate method as described previously (8). A total of 30 -40 g of plasmid DNA/plate were used for each transfection. Briefly, plasmid DNA precipitates were layered on cells for 4 -6 h, after which cells and plasmid precipitate were incubated in 10% Me 2 SO 4 /Dulbecco's modified Eagle's medium for 6 min. Following this Me 2 SO shock, cells were refed with fresh medium and analyzed 40 -50 h later. For metabolic labeling experiments, two plates of transfected cells were used for each data point.
Metabolic Labeling with [ 3 H]Palmitic Acid-Transfected cells were metabolically labeled with 0.5 mCi/ml [ 3 H]palmitic acid in DME/F-12 (Sigma) as described previously (8). Palmitoylation experiments were performed a minimum of two times with each mutant, with virtually identical results. Membrane particulate fractions from metabolically labeled cells were solubilized with 0.4 M NaCl and 1% digitonin and then immunoprecipitated with the ␤ GEN antiserum. Gels for fluorography were treated with ENLIGHTNING (NEN Life Science Products) and then exposed to film for 2-4 weeks at Ϫ80°C.
Membrane Isolation, Immunoprecipitation, and Immunoblotting-Membrane particulate fractions were isolated from transfected cells as described previously (3). For co-immunoprecipitation of ␣ 1 and ␤ subunits, membrane particulate fractions were solubilized with a combination of 0.4 M NaCl, 1% Triton X-100, and 0.1% SDS. Cells were solubilized on ice for 30 min and then centrifuged in a Beckman Ty65 rotor at 45,000 rpm for 1-2 h. Soluble fractions were incubated with 40 l of antiserum and 40 l of protein G-Ultralink resin (50% slurry) overnight with agitation at 4°C. Immunoprecipitates were washed with homogenization buffer 3-5 times and eluted with SDS sample buffer. Immunoblotting procedures were performed as described previously (3). Detection of immunoreactive bands was performed using either enhanced chemiluminescence (Pierce) or colorimetric enhanced diaminobenzidine substrate reaction (Pierce).
Immunofluorescence and Confocal Microscopy-For immunohistochemical studies, transfected cells were transferred to coverslips coated with 20 mM poly-L-lysine immediately following Me 2 SO shock. At 40 -50 h after Me 2 SO shock, cells were fixed and permeabilized using methanol/acetone (1:1, v/v) at 4°C for 10 min. Cells were then incubated for 6 -12 h with the primary antiserum in the presence of 0.1% bovine serum albumin (w/v). Secondary fluorescein isothiocyanate-and tetramethylrhodamine isothiocyanate-coupled antibodies were subsequently added and incubated for 2-4 h in the presence of 0.1% bovine serum albumin (w/v). Confocal immunofluorescent microscopy was performed in the Northwestern University Cell Imaging Facility using a Zeiss LSM-10 laser-scanning microscope.

RESULTS AND DISCUSSION
The Loss of Palmitoylation Affected the Subcellular Localization of the ␤ 2a Protein-Previously, we had performed biochemical studies demonstrating that the expressed rat ␤ 2a subunit fractionated to crude particulate fractions (3). In parallel, results of immunohistochemical studies demonstrated the localization of the ␤ 2a subunit to the plasma membrane even in the absence of a co-expressed ␣ 1 subunit (3). Mutation of Cys 3 and Cys 4 in the ␤ 2a (C3S/C4S) mutant eliminated palmitoylation of the ␤ 2a subunit (8); however, while this mutant protein still fractionated almost exclusively to crude particulate fractions (8), we did not examine the subcellular distribution of the mutant protein by confocal microscopy. Here we report the results of immunohistochemical studies that were performed to address further the subcellular distribution of the ␤ 2a (C3S/ C4S) protein in transiently transfected tsA201 cells. Transfected cells expressing either the wild-type rat ␤ 2a subunit or the palmitoylation-deficient ␤ 2a (C3S/C4S) subunit were fixed with methanol/acetone as described under "Experimental Procedures," immunohistochemically stained with the ␤ 2a antibody, and visualized using confocal immunofluorescence microscopy. Cells expressing the wild-type ␤ 2a subunit exhibited continuous staining along the plasma membrane and/or immediately adjacent to the plasma membrane ( Fig. 1A), consistent with our previously reported results (3). In marked contrast, the palmitoylation-deficient ␤ 2a (C3S/C4S) subunit displayed a diffuse intracellular staining pattern (Fig. 1B). The staining pattern did not noticeably change following treatment of these cells with the protein synthesis inhibitor cycloheximide (Fig.  1B, right), which was previously shown to enhance visualization of plasma membrane staining of the ␤ 2a subunit in transfected cells. These results suggested that palmitoylation of the ␤ 2a subunit was critical for the plasma membrane localization of this subunit when it was expressed in the absence of an ␣ 1 subunit.
The ␤ 1b and ␤ 3 Subunits Exhibited a Diffuse Intracellular Localization-We previously reported that two other ␤ subunit isoforms, ␤ 1b and ␤ 3 , were not palmitoylated (8). In order to assess their cellular localization, transiently transfected tsA201 cells expressing either the ␤ 1b or ␤ 3 subunit were frac-tionated into crude membrane particulate fractions and cytosolic supernatants as described previously (3) and analyzed by immunoblotting with the ␤ GEN antiserum ( Fig. 2A). Following high speed centrifugation, both the ␤ 1b and ␤ 3 subunits were found primarily in the crude membrane particulate fraction ( Fig. 2A). To characterize further the subcellular localization of these proteins, transfected cells were also immunohistochemically stained with the ␤ GEN antibody and viewed by confocal immunofluorescent microscopy (Fig. 2B). Unlike the plasma membrane staining observed with the wild-type ␤ 2a subunit (Fig. 1), both the ␤ 1b and ␤ 3 subunits exhibited a diffuse intracellular staining pattern (Fig. 2B). Pretreatment of cells expressing the ␤ 1b and ␤ 3 subunits with cycloheximide did not enhance any plasma membrane staining (data not shown), suggesting that the intracellular distribution was not due to delayed trafficking of overexpressed subunits. The subcellular distribution observed for the ␤ 1b and ␤ 3 subunits resembled that seen with the palmitoylation-deficient ␤ 2a (C3S/C4S) mutant ( Fig. 1), consistent with the hypothesis that palmitoylation is required for localization of ␤ subunits to the plasma membrane in the absence of a co-expressed ␣ 1 subunit. Clearly, palmitoylation confers unique properties that distinguish ␤ 2a from the other nonpalmitoylated ␤ subunits. The localization of nonpalmitoylated ␤ subunits to crude particulate fractions suggested that they may be associated with intracellular membrane systems. The subcellular distribution of the ␤ 1b subunit reported here differs from that reported in another study, which asserted a membrane localization for this protein in transfected COS-7 cells (25). Since these studies were done in a different cell line, it is possible that interaction with cell typespecific proteins could also provide a mechanism for the targeting of ␤ subunits.

Mutations in a src Homology 3 (SH3) Motif
Affected Subcellular Localization of the ␤ 2a Subunit-All four calcium channel ␤ subunits contain a region resembling SH3 domains, which have been implicated in mediating protein-protein interactions through the binding of proline-rich sequences (26). The SH3like motif is located in the first of two domains that are highly conserved among the four ␤ subunit isoforms identified to date (Fig. 3A). Fig. 3A shows a Lipmann-Pearson alignment of the SH3 domain from ␤ 2a and the original SH3 domain from src (26). Also shown for comparison is an alignment of the src SH3 domain with the SH3 domain of spectrin (27). To assess whether the SH3 domain of ␤ 2a played a role in membrane localization, site-directed mutagenesis was used to target residues in the C-terminal region of the ␤ 2a SH3 domain corresponding to residues shown to be important for the binding of the src SH3 domain to proline-rich ligands (28). The ␤ 2a (P119A) and ␤ 2a (I115A/F117A/P119L) subunits were transiently transfected into tsA201 cells and analyzed by immunohistochemical staining. As seen in Fig. 3B, the ␤ 2a (P119A) mutant exhibited considerable staining at the plasma membrane as well as some diffuse intracellular distribution. The ␤ 2a (I115A/F117A/P119L) mutant appeared to be localized intracellularly, with less clear staining visible at the plasma membrane (Fig. 3C). These results initially suggested that mutations disrupting the ␤ 2a SH3 domain affected targeting of FIG. 2. The ␤ 1b and ␤ 3 subunits fractionate with crude membranes and exhibit a diffuse intracellular localization. A, transiently transfected cells expressing the ␤ 1b or ␤ 3 subunits were fractionated by high speed centrifugation into crude membrane particulate fractions (P) or cytosolic supernatants (S) and immunoblotted with the ␤ GEN antiserum following SDS-polyacrylamide gel electrophoresis on an 8% polyacrylamide gel. Equivalent proportions of crude membranes and supernatants were loaded for each sample, and the locations of the molecular mass markers are indicated on the right. Both the ␤ 1b and ␤ 3 subunits localized primarily to particulate fractions. B, transfected ␤ 1b and ␤ 3 cells were immunohistochemically stained with the ␤ GEN antiserum and viewed by confocal immunofluorescence microscopy. Both the ␤ 1b and ␤ 3 subunits exhibited a diffuse pattern of intracellular staining, with no specific plasma membrane staining discernible. The staining patterns were unchanged following pretreatment of the cells with 100 g/ml of cycloheximide (data not shown). the ␤ 2a protein to the plasma membrane, potentially through disruption of an SH3-mediated interaction. However, the possibility that these mutations may have disrupted the global structure of the ␤ 2a protein cannot be discounted, although the ability of the mutants to be recognized by the ␤ 2 and ␤ GEN antibodies suggests that the mutations do not drastically alter the structure of the protein.
Mutations of the ␤ Interaction Domain (BID) Affected Subcellular Localization of the ␤ 2a Subunit-Site-directed mutagenesis was used to create a mutation at Pro 234 in the previously identified BID, which has been suggested to provide the major site of interaction of ␤ subunits with ␣ 1 subunits (14). Mutation of the corresponding Pro 237 residue in the ␤ 1b subunit to Arg appeared to eliminate completely ␣ 1 -␤ interaction (14). Transfected cells expressing the ␤ 2a (P234R) mutant were immunohistochemically stained and analyzed by confocal immunofluorescence microscopy to assess subcellular distribution (Fig. 3D). Unexpectedly, staining with the ␤ 2a antibody revealed that the ␤ 2a (P234R) subunit exhibited a diffuse intracellular staining pattern (Fig. 3D) similar to that seen with the ␤ 2a (C3S/C4S) mutant and the nonpalmitoylated ␤ 1b and ␤ 3 subunits (see Figs. 1 and 2). Since the ␤ 2a (P234R) mutation is expected to impair the BID, this result suggested the possibility that the plasma membrane staining normally seen with the wild-type ␤ 2a subunit might be due to interaction with an endogenous ␣ 1 subunit. However, this possibility seemed unlikely given that the high levels of heterologous ␤ 2a expression were most certainly higher than levels of endogenous channels. Additionally, electrophysiological measurements did not reveal the presence of measurable calcium currents or charge movement in cells transfected with only the ␤ 2a subunit (data not shown). An alternative conclusion is that mutation of Pro 234 may have affected subcellular localization through global disruption of the ␤ 2a protein structure, although the mutant protein retained sufficient native structure to be identified by the anti-␤ antibodies.
Mutations Distal to the N Terminus of ␤ 2a Affected Palmitoylation-Mutations in the BID and SH3 domain resulted in a diffuse intracellular localization of the ␤ 2a protein that resembled the subcellular localization observed with the palmitoylation-deficient ␤ 2a (C3S/C4S) mutant (Fig. 1B) and the nonpalmitoylated ␤ 1b and ␤ 3 subunits (Fig. 2B). Metabolic labeling of transfected cells with [ 3 H]palmitic acid was performed to address whether the ␤ 2a (P234R), ␤ 2a (P119A), and ␤ 2a (I115A/ F117A/P119L) mutants were palmitoylated yet unable to target to the plasma membrane due to disruption of a protein-protein interaction. Following the metabolic labeling of transfected cells, proteins were immunoprecipitated from solubilized membrane fractions using the ␤ GEN antiserum as described under "Experimental Procedures." Immunoprecipitated fractions were analyzed both by quantitative immunoblotting and by fluorography. Incorporation of [ 3 H]palmitic acid in the immunoreactive bands was quantified using liquid scintillation counting and normalized by protein amounts to values obtained for the wildtype ␤ 2a subunit. A representative experiment is summarized in Fig. 4. Unexpectedly, the BID and SH3 mutations all resulted in decreased amounts of palmitate incorporation compared with wild-type ␤ 2a , although the reductions in palmitoylation of the ␤ 2a (P234R) and ␤ 2a (I115A/F117A/P119L) mutants were much more extensive than the very modest reduction observed with the ␤ 2a (P119A) mutant. Since metabolic labeling measures steady-state levels of [ 3 H]palmitate incorporation, it was unclear whether the decreased palmitoylation observed in the different mutants resulted from changes in the kinetics of either palmitate addition or removal.
These results suggested that the altered subcellular localiza-tion seen with the BID and SH3 domain mutants could be due solely to changes in the palmitoylation of ␤ 2a rather than through the disruption of protein-targeting interactions mediated by these domains. In support of this hypothesis, the ␤ 2a (P119A) mutant, which should disrupt SH3 ligand binding (28), exhibited only a small decrease in palmitoylation and was associated with the plasma membrane with only a small amount of intracellular localization. Potentially, mutation of the BID and SH3 domains affected protein interactions that were critical for palmitoylation of the ␤ 2a protein. In the absence of a known tertiary structure for the ␤ 2a protein, it is uncertain whether these domains may be in close topological proximity on the surface of the protein. Conceivably, the changes in subcellular localization of the SH3 and BID mutants, as well as any potential effects on channel function, could result primarily from a disruption of palmitoylation of the ␤ 2a protein. These results caution that structure-function analysis of calcium channel ␤ subunits through site-directed mutagenesis, in the absence of biochemical studies, could result in misleading interpretations. Elucidation of the three-dimensional structure of ␤ subunits should facilitate the characterization of these proteins and reveal the mechanisms by which different mutations could affect protein structure and/or function. The present findings do not completely rule out a role for the BID and/or SH3 domain in ␤ subunit localization. Further studies may further characterize protein interactions involving these domains and their potential role in channel regulation. replaced with the 16-amino acid N terminus from either the wild-type ␤ 2a or the palmitoylation-deficient ␤ 2a (C3S/C4S) subunits. The chimeric ␤ 2a/1b and ␤ 2a/3 subunits were expressed in transiently transfected tsA201 cells. An immunoblot of membrane particulate fractions from these cells stained with the ␤ GEN antiserum (Fig. 5A) demonstrated the expression of the different ␤ subunit proteins. Indicated below the immunoblot are the predicted molecular masses from the cDNA sequences and the apparent molecular masses observed upon SDS-polyacrylamide gel electrophoresis. The ␤ 1b and ␤ 3 subunits each exhibited a molecular mass higher than that predicted by the cDNA sequences (Fig. 5A), in agreement with previously published reports (8). Surprisingly, replacement of the 15-amino acid N terminus of ␤ 3 with the larger 16-amino acid N terminus of ␤ 2a resulted in a decreased apparent molecular mass of the ␤ 2a/3 chimera, suggesting that the increased electophoretic mobility observed with the ␤ 3 protein probably involved structural determinants and/or post-translational modifications in the ␤ 3 N terminus. The ␤ 2a/1b chimeras still exhibited molecular masses higher than those predicted by the cDNA sequences, suggesting that structural elements responsible for the aberrant electrophoretic mobility of this protein were located in the isoform-specific C-terminal region rather than the N terminus. Post-translational modifications unique to the ␤ 1b and ␤ 3 isoforms could be the cause of the higher molecular masses observed with these two proteins.
Transiently transfected cells expressing the different chimeric ␤ subunits were metabolically labeled with [ 3 H]palmitic acid. Subsequently, the chimeric ␤ proteins were isolated by immunoprecipitation and analyzed both by immunoblotting and fluorography (Fig. 5B). For comparison, cells expressing the wild-type ␤ 2a subunit were also metabolically labeled in parallel. The immunoblot stained with the ␤ GEN antiserum (Fig. 5B, top) indicated the expression and immunoprecipitation of each ␤ subunit. The fluorogram (Fig. 5B, bottom) indicated the incorporation of [ 3 H]palmitic acid only in ␤ subunits containing the wild-type ␤ 2a N terminus. By contrast, chimeric ␤ subunits containing the ␤ 2a (C3S/C4S) mutation were not palmitoylated. These results demonstrated that the 16-amino acid N-terminal region of ␤ 2a contained all of the structural determinants necessary to confer palmitoylation to nonpalmitoylated ␤ subunits.
Palmitoylation by Itself Was Not Sufficient to Confer Plasma Membrane Localization-Transiently transfected cells expressing the different chimeric ␤ subunits were analyzed by immunohistochemical staining and confocal immunofluorescence microscopy (Fig. 6). Although the ␤ 2a/1b and ␤ 2a/3 subunits were palmitoylated (Fig. 5B), the distribution of these proteins was intracellular (Fig. 6, left panels), resembling the distribution of FIG. 5. Chimeric ␤ subunits resulted in changes in relative electrophoretic mobility and demonstrated that the ␤ 2a N terminus was sufficient to confer palmitoylation. A, an immunoblot shows the relative electrophoretic mobility of different ␤ subunits expressed in transiently transfected tsA201 cells. Membrane particulate fractions were electrophoresed on an 8% polyacrylamide gel, transferred to nitrocellulose, and visualized by immunoblotting with the ␤ GEN antiserum. The predicted molecular mass and the apparent molecular mass of each ␤ subunit are indicated below. The location of molecular mass markers is indicated on the right. B, transfected tsA201 cells expressing the different ␤ subunit chimeras were metabolically labeled with [ 3 H]palmitic acid. The ␤ subunits were immunoprecipitated with the ␤ GEN antiserum, and the presence of each protein in the immunoprecipitates was confirmed by immunoblotting with the ␤ GEN antiserum (top). The same immunoprecipitate was analyzed by fluorography for the incorporation of the nonpalmitoylated ␤ 1b and ␤ 3 subunits (Fig. 2) and the palmitoylation-deficient ␤ 2a (C3S/C4S) mutant (Fig. 1). The staining pattern of the palmitoylated chimeras (Fig. 6, left panels) was difficult to distinguish from that of the nonpalmitoylated ␤ 2a/1b (C3S/C4S) and ␤ 2a/3 (C3S/C4S) subunits (Fig. 6,  right panels). These results demonstrated that palmitoylation by itself was insufficient to confer the plasma membrane localization typically seen with the ␤ 2a subunit. Potentially, other elements of the ␤ 2a sequence, most likely in the isoform-specific C-terminal region, are also involved in determining subcellular localization. However, the results with the different ␤ 2a subunits described above, which localize intracellularly despite the presence of intact C-terminal regions, suggest that membrane targeting of the ␤ 2a protein requires both palmitoylation and other structural elements in the ␤ 2a sequence.
Brefeldin A Increased Palmitoylation of ␤ 2a -To assess whether palmitoylation of the ␤ 2a protein occurs prior to transport to the plasma membrane, transfected cells were treated with the fungal metabolite brefeldin A, which prevents the trafficking of proteins to the plasma membrane through the trans-Golgi network. Cells were metabolically labeled with [ 3 H]palmitic acid following treatment with either brefeldin A or vehicle as described under "Experimental Procedures." Solubilized proteins from labeled cells were immunoprecipitated with the ␤ GEN antiserum, and the presence of ␤ 2a protein in the immunoprecipitated pellets was confirmed by immunoblotting with the ␤ 2a antiserum. The ␤ 2a protein was immunoprecipitated from both control cells and brefeldin A-treated cells (Fig.  7, upper panel). A fluorogram of the same immunoprecipitate (Fig. 7, lower panel) indicated that ␤ 2a from both control cells and brefeldin A-treated cells was palmitoylated. The amount of [ 3 H]palmitic acid was quantified from scintillation counting of immunoreactive bands (Fig. 7, upper panel) and normalized to the amount of protein in the immunoprecipitated pellet. Treatment with brefeldin A increased palmitoylation of the ␤ 2a protein about 2-6-fold in two separate experiments. Brefeldin A has been previously shown to increase (29,30) as well as decrease (31) the palmitoylation of other proteins. The apparent increase in the palmitoylation of ␤ 2a upon brefeldin treatment probably results from the retention of the protein in the endoplasmic reticulum, thus preventing dynamic depalmitoylation of ␤ 2a by enzymes localized at the plasma membrane. These results suggest that palmitoylation of the ␤ 2a protein occurs shortly after synthesis and prior to transport through the trans-Golgi network.
Rabbit ␤ 2 Isoforms Lacked Palmitoylation and Exhibited Diffuse Intracellular Localization-The sequences of several ␤ subunit clones isolated from rabbit cardiac tissue were identified as ␤ 2 splice variants (32) by virtue of the similarity of their C-terminal sequences to that of the rat ␤ 2a isoform, which was the first ␤ 2 subunit identified (6). The first rabbit ␤ 2 clone identified was given the designation of ␤ 2a (32), despite the fact that it contained an N terminus that differed significantly from that of the rat ␤ 2a subunit. The rabbit ␤ 2b also contains a distinct N-terminal domain (32). Transiently transfected tsA201 cells expressing the rabbit ␤ 2a and ␤ 2b subunits were metabolically labeled with [ 3 H]palmitic acid to address FIG. 8. The subcellular localizations of nonpalmitoylated ␤ 2 splice variants are different from that seen with the wild-type rat ␤ 2a protein. A, transiently transfected cells expressing the rat ␤ 2a subunit, the rabbit "␤ 2a " subunit, or the rabbit "␤ 2b " subunit were metabolically labeled with [ 3 H]palmitic acid. The ␤ 2 proteins were isolated by immunoprecipitation with the ␤ GEN antibody and visualized by immunoblotting with the ␤ 2a antibody (top). The same immunoprecipitate was analyzed by fluorography to assess the incorporation of [ 3 H]palmitic acid in each protein (bottom). B, the subcellular localization of these different rabbit ␤ 2 subunits was examined by immunohistochemistry in transiently transfected cells. Staining is shown for cells expressing the rabbit ␤ 2a (left) and the rabbit ␤ 2b (right) proteins. Cells were fixed and immunostained with the ␤ 2a antibody and then analyzed by confocal immunofluorescent microscopy. For comparison, see also the staining of the wild-type rat ␤ 2a subunit in Fig. 1A. whether these ␤ 2 splice variants were palmitoylated. Proteins were immunoprecipitated with the ␤ GEN antiserum and analyzed by immunoblotting and fluorography (Fig. 8A). Immunoblotting with the ␤ 2a antibody was used to visualize the immunoprecipitated proteins (Fig. 8A, top). The rabbit ␤ 2a and ␤ 2b subunits were detected as immunoreactive bands at ϳ69 and 72 kDa, respectively, as predicted by their cDNA sequences (32). As expected from the absence of Cys 3 and Cys 4 in the rabbit ␤ 2a and ␤ 2b sequences (32), the fluorogram indicated that [ 3 H]palmitic acid was only incorporated into the rat ␤ 2a protein (Fig. 8A, bottom). The lack of palmitoylation of the rabbit ␤ 2a subunit confirmed the expectation from the primary sequences that this protein is biochemically distinct from the rat ␤ 2a isoform, despite having the same subscript designation.
Further studies were performed to assess the subcellular localization of the rabbit ␤ 2a and rabbit ␤ 2b subunits in transfected cells (Fig. 8B). Confocal microscopy of immunohistochemical staining with the ␤ 2a antibody revealed that both the rabbit ␤ 2a and ␤ 2b proteins were localized to intracellular structures (Fig. 8B). The subcellular staining pattern of the rabbit ␤ 2a and ␤ 2b subunits differed markedly from that of the rat ␤ 2a subunit (Fig. 1A), further demonstrating the biochemical distinction between the originally identified rat ␤ 2a subunit (6) and the subsequently identified rabbit ␤ 2a subunit (32). The biophysical properties of the rat versus the rabbit ␤ 2a subunits have not been systematically compared.
Palmitoylation Affects ␤ 2a Subcellular Localization and Channel Function-Mutations affecting palmitoylation, which include the ␤ 2a (C3S/C4S) mutant, the ␤ 2a (P234R) mutant, and the ␤ 2a (I115A/F117A/P119L) mutant, also resulted in the localization of the ␤ 2a protein to intracellular membrane systems. Likewise, nonpalmitoylated rabbit ␤ 2 isoforms localized intracellularly rather than to the plasma membrane. These results suggest that palmitoylation is pivotal for the plasma membrane localization observed with the wild-type rat ␤ 2a protein. However, studies on chimeric ␤ 1 and ␤ 3 subunits, which were still intracellularly localized despite being palmitoylated, also clearly indicated that palmitoylation was by itself insufficient to localize these ␤ subunits to the plasma membrane. Taken together, it appears that the unique plasma membrane localization of the rat ␤ 2a subunit requires a combination of both palmitoylation as well as other structural determinants that may be unique to the ␤ 2a sequence.
The mechanism behind the regulation of channel function by palmitoylation is still unclear. The ␤ 2a (C3S/C4S) subunit, when co-expressed with the ␣ 1C subunit in Xenopus oocytes, supported increases in current density that were similar to or greater than those observed with wild type ␤ 2a . 2 In contrast, a decrease in whole-cell calcium currents was seen with the ␤ 2a (C3S/C4S) subunit in mammalian HEK cells (8). An examination of electrophysiological data collected from a large population of ␣ 1C ␤ 2a cells revealed that for cells expressing a given number of channels, as measured by charge movement, there was a wide range in the corresponding amount of whole-cell calcium current (8). The diversity in the ratio of ionic current to charge movement from cell to cell may reflect variations in the population of palmitoylated channels between different cells and suggests that the palmitoylation of ␤ 2a may indeed have a dynamic component.
A recent study in Xenopus oocytes demonstrated that the electrophysiological properties unique to the rat ␤ 2a isoform could be attributed directly to palmitoylation of the ␤ 2a subunit (33), specifically the lack of prepulse facilitation seen upon co-expression with the ␣ 1C channel, the inhibition of voltageinduced inactivation of ␣ 1E channels, and the blockage of ␣ 1E channel inhibition by G protein-coupled receptors. Since palmitoylation is thought to be regulated through receptor-mediated processes (20), the dynamic palmitoylation and depalmitoylation of ␤ 2a could potentially allow humoral regulation of certain subsets of channels in different cells, resulting in modulation of channel activity and properties such as neuronal plasticity. Further studies involving the identification of biochemical pathways involved in the palmitoylation of ␤ 2a may lend insight into the mechanism by which this modification regulates channel function, as well as the role of channel palmitoylation in the regulation of calcium entry in different cells.