Identification of palmitoylation sites within the L-type calcium channel beta2a subunit and effects on channel function.

The hydrophilic beta2a subunit of the L-type calcium channel was recently shown to be a membrane-localized, post-translationally modified protein (Chien, A. J., Zhao, X. L., Shirokov, R. E., Puri, T. S., Chang, C. F., Sun, D. D., Rios, E., and Hosey, M. M. (1995) J. Biol. Chem. 270, 30036-30044). In this study, we demonstrate that the rat beta2a subunit was palmitoylated through a hydroxylamine-sensitive thioester linkage. Palmitoylation required a pair of cysteines in the N terminus, Cys3 and Cys4; mutation of these residues to serines resulted in mutant beta2a subunits that were unable to incorporate palmitic acid. Interestingly, a palmitoylation-deficient beta2a mutant still localized to membrane particulate fractions and was still able to target functional channel complexes to the plasma membrane similar to wild-type beta2a. However, channels formed with a palmitoylation-deficient beta2a subunit exhibited a dramatic decrease in ionic current per channel, indicating that although mutations eliminating palmitoylation did not affect channel targeting by the beta2a subunit, they were important determinants of channel modulation by the beta2a subunit. Three other known beta subunits that were analyzed were not palmitoylated, suggesting that palmitoylation could provide a basis for the regulation of L-type channels through modification of a specific beta isoform.

In this study, we demonstrate that the rat ␤ 2a subunit was palmitoylated through a hydroxylamine-sensitive thioester linkage. Palmitoylation required a pair of cysteines in the N terminus, Cys 3 and Cys 4 ; mutation of these residues to serines resulted in mutant ␤ 2a subunits that were unable to incorporate palmitic acid. Interestingly, a palmitoylation-deficient ␤ 2a mutant still localized to membrane particulate fractions and was still able to target functional channel complexes to the plasma membrane similar to wild-type ␤ 2a . However, channels formed with a palmitoylationdeficient ␤ 2a subunit exhibited a dramatic decrease in ionic current per channel, indicating that although mutations eliminating palmitoylation did not affect channel targeting by the ␤ 2a subunit, they were important determinants of channel modulation by the ␤ 2a subunit. Three other known ␤ subunits that were analyzed were not palmitoylated, suggesting that palmitoylation could provide a basis for the regulation of L-type channels through modification of a specific ␤ isoform.
The L-type calcium channel ␤ 2a subunit is a highly hydrophilic protein with no predicted membrane-spanning regions (1). Nevertheless, we previously demonstrated that this protein was membrane-localized when expressed in human embryonic kidney tsA201 cells (2). In addition, pulse-chase analysis sug-gested that the ␤ 2a subunit was post-translationally modified, resulting in an increase in apparent molecular mass from 68 kDa for the nascent protein to proteins of 70 -72 kDa. Here we demonstrate that one modification of the ␤ 2a subunit involves palmitoylation, a post-translational modification that has been shown to facilitate the membrane localization of other hydrophilic proteins. Palmitoylation involves the addition of palmitic acid to cysteine residues through a thioester linkage (3,4). This modification is thought to be dynamic due to the reversible nature of the thioester bond (3,4). However, despite the increasing number of palmitoylation sites identified, there appears to be no consensus motif for predicting candidate cysteine residues, which may be acylated. The residues involved in palmitoylation of the ␤ 2a subunit were identified, and the functional roles of these amino acids were investigated in biochemical and electrophysiological studies using mutant proteins. Three other known ␤ subunits were also analyzed and found not to undergo palmitoylation.
Generation and Characterization of the Generic ␤ Antibody-A BamHI-XhoI fragment encoding residues 17-354 of ␤ 2a was generated by PCR and subcloned into the BamHI-XhoI sites of pGEX-4T3 (Pharmacia Biotech Inc.), resulting in an in-frame fusion of ␤ 2a residues 17-354 to glutathione S-transferase. Purified fusion protein was obtained by standard procedures and injected into a goat at Bethyl Laboratories (Montgomery, TX).
Metabolic Labeling of Mammalian tsA201 and Sf9 Insect Cells-Transfection of tsA201 cells was performed in 100-mm tissue culture plates as described previously (2). Sf9 insect cells were cultured using standard techniques and infected with recombinant baculovirus containing the ␤ 2a cDNA (5). At 40 -45 h post-transfection, medium was removed, and cells were incubated with 0.5 mCi/ml of [ 3 H]palmitic acid (American Radiolabeled Chemicals, St. Louis, MO) in Dulbecco's modified Eagle's medium-Ham's F-12 medium (Sigma) or Sf900 (Life Technologies, Inc.)(for Sf9 cells) for 1-2 h at room temperature with gentle rocking. An aliquot of the medium was analyzed for 3 H content at the start and end of the metabolic labeling. Incorporation of radioactivity into the cells was estimated between 70 and 98%. Palmitoylation experiments were performed a minimum of three times with identical results. HPLC analysis to determine the identity of the acyl group on ␤ 2a was performed as described by Linder et al. (4). 3 H-Radiolabeled standards for myristate, palmitate, and stearate were purchased from * This work was supported in part by National Institutes of Health Grants HL23306 (to M. M. H.) and AR43113 (to E. R.) and by grants from the American Heart Association of Metropolitan Chicago (to R. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Whole-cell Patch-clamp Analysis-Cells were analyzed by whole-cell patch-clamp using data acquisition and pulse generation protocols similar to those previously described (2). Voltage pulse duration was 300 ms. The pipette solution contained (in mM): 110 cesium aspartate, 20 CsCl, 10 EGTA, 10 HEPES, 5 Mg-ATP. The extracellular solution contained (in mM): 150 tetraethylammonium chloride, 10 CaCl 2 , 10 HEPES. To record intramembrane charge movement, ionic currents were blocked by the addition of 10 M GdCl 3 in the bath (6). Holding potential was Ϫ90 mV. Asymmetric currents were obtained by subtracting control transients obtained by pulsing between Ϫ150 and Ϫ90 mV. Maximal amplitude of the ionic current and maximal intramembrane charge movement were related to cell capacitance in order to compare their densities.

RESULTS AND DISCUSSION
Palmitoylation of Different ␤ Subunit Isoforms-An epitope common to the four known ␤ subunit isoforms (1, 7-10) was used to generate a generic ␤ antiserum (␤ GEN ). The ␤ GEN antiserum was able to specifically recognize ␤ 1b (ϳ72-75 kDa), ␤ 2a (68 -72 kDa), ␤ 3 (ϳ58 kDa), and ␤ 4 (ϳ58 kDa) in tsA201 cells transiently transfected with cDNAs for each isoform (Fig.  1A). In order to investigate the possibility that ␤ subunits may be palmitoylated, transiently transfected tsA201 cells expressing different ␤ isoforms were metabolically labeled with [ 3 H]palmitic acid, immunoprecipitated with the ␤ GEN antiserum, and subsequently analyzed for acylation of ␤ proteins. The resulting fluorogram (Fig. 1B) indicated incorporation of 3 H-radiolabel in a ϳ70-kDa protein in ␤ 2a cells (Fig. 1A, second  lane), suggesting the addition of an acyl group. However, no specific radiolabeling of the ␤ 1b , ␤ 3 , and ␤ 4 subunits was observed. The presence of the different ␤ subunits in the immunoprecipitates was confirmed by immunoblotting (data not shown). Experiments were performed three times with identical results. Acylation of the ␤ 2a protein also occurred in Sf9 insect cells, indicating that this phenomenon was not cellspecific (data not shown).
Characterization of the Acyl Moiety on ␤ 2a -Acylation with palmitic acid occurs most frequently through a thioester linkage, which is sensitive to base hydrolysis (3,4). To assess the nature of the acylation linkage on ␤ 2a , immunoprecipitated ␤ 2a from metabolically labeled cells was split into two equal fractions and treated with either 1 M hydroxylamine (pH 7.5) or 1 M Tris-HCl (pH 7.5) and analyzed by subsequent SDS-polyacrylamide gel electrophoresis and fluorography (Fig. 1C). Immunoblotting was used to confirm that equal amounts of ␤ 2a protein were loaded in each lane (Fig. 1C, upper panel). Treatment with hydroxylamine led to a clear loss of 3 H signal on the fluorogram (Fig. 1C, lower panel), indicating that the acyl moiety on the ␤ 2a protein was attached via a base-sensitive Because palmitic acid can be naturally metabolized to myristic acid, reverse phase HPLC was used to identify the basesensitive radioactive label. After labeling Sf9 cells expressing the ␤ 2a protein with [ 3 H]palmitic acid, base hydrolysis and reverse phase HPLC were performed as described in Linder et al. (4). Fractions eluted off the column were assayed using scintillation counting. The base-hydrolyzed fraction eluted as a single peak, corresponding to the same elution profile as the palmitate standard. A presample run prior to loading of the base-hydrolyzed fraction onto the column gave counts undistinguishable from background. These results demonstrated that the ␤ 2a protein was acylated with palmitic acid through a base-sensitive linkage.
Identification of Sites Important for Palmitoylation of ␤ 2a -The addition of palmitic acid usually occurs through a thioester linkage on cysteine residues, although no consensus sequence exists for predicting the exact location of the modification. The N-terminal region of ␤ 2a contains a Met-Xaa-Cys motif found in several palmitoylated proteins ( Fig. 2A; Ref. 3). To assess whether this region of the ␤ 2a protein was involved in palmitoylation, an N-terminal mutant was constructed that contained a 15-amino acid deletion (residues 2-16) of the unique region of the ␤ 2a N terminus (see "Experimental Procedures"). This mutant was not able to incorporate palmitic acid (data not shown), suggesting that palmitoylation of ␤ 2a required the Nterminal region of ␤ 2a . To further investigate the site or sites of palmitoylation, site-directed mutants were constructed in which cysteine residues 3 and 4 were mutated to serine residues. These mutants, ␤ 2a (C3S), ␤ 2a (C4S), and ␤ 2a (C3S/C4S), were expressed in tsA201 cells and analyzed for incorporation of palmitic acid. The results indicated that only the wild-type ␤ 2a was palmitoylated, whereas all three of the mutants did not incorporate palmitic acid (Fig. 2B). A corresponding Western blot from the same experiment demonstrated the amount of protein in each immunoprecipitation (Fig. 2B). The immunoreactive bands were excised from the nitrocellulose and analyzed with liquid scintillation counting, which confirmed the results of the fluorogram; values obtained in this experiment are shown at the bottom of Fig. 2B. These results indicated that both Cys 3 and Cys 4 were necessary for palmitoylation of ␤ 2a .
Effects of Palmitoylation on the Membrane Association and Solubility of ␤ 2a -To test for a potential role for palmitoylation in the membrane localization of the ␤ 2a subunit, transfected tsA201 cells expressing either the wild-type ␤ 2a or the palmitoylation-deficient mutants were fractionated by centrifugation into membrane particulate fractions and soluble "cytosolic" fractions as described previously (2). Both the palmitoylated ␤ 2a protein and the palmitoylation-deficient ␤ 2a mutants still localized to membrane particulate fractions (Fig. 3A). The wildtype ␤ 2a is expressed as a multiple series of bands of 68 -72 kDa that were previously shown to arise from unidentified posttranslational modifications (2). In contrast, the palmitoylationdeficient mutants did not appear to convert to the higher molecular mass isoforms, as evidenced by the decreased immunoreactivity at 70 -72 kDa (Fig. 3A). Rather, all the palmitoylation-deficient ␤ 2a subunits exhibited distinctly lower molecular masses (Fig. 3A). These result suggest that palmitoylation might be required for the post-translational modification to the larger size forms seen with the wild-type ␤ 2a protein. In addition, the palmitoylation-deficient mutants appeared to be more susceptible to proteolysis, because several smaller immunoreactive bands were seen for each mutant (Fig.  3A). Further studies were performed to assess contributions of palmitoylation to the interaction of the ␤ 2a subunit with the membrane. The ␤ 2a protein could be immunoprecipitated from  (Fig. 3B, top). However, only ␤ 2a immunoprecipitated from the detergentsolubilized fraction contained [ 3 H]palmitate-labeled ␤ 2a (Fig.  3B, bottom). This result demonstrates that the palmitoylation of ␤ 2a modifies the nature of its association with the membrane, confirming our previous hypothesis, which predicted distinct salt-and detergent-soluble populations of the ␤ 2a subunit (2).
Effects of Palmitoylation on Channel Targeting and Channel Function-To address potential roles of palmitoylation on channel function and/or protein multimerization, cells expressing both the cardiac ␣ 1C subunit (11) as well as either ␤ 2a or ␤ 2a (C3S/C4S) were analyzed. Both ␤ 2a and ␤ 2a (C3S/C4S) could be co-immunoprecipitated with ␣ 1C , indicating that the subunits could be co-expressed and directly interact (data not shown). However, electrophysiological analyses demonstrated striking differences in the biophysical properties of channels with the ␤ 2a (C3S/C4S) subunit (Fig. 4, A and B). Measurements of whole-cell calcium currents demonstrated that the expression of ␤ 2a increased currents relative to ␣ 1C alone; however, drastic reductions were seen in cells expressing ␣ 1C and ␤ 2a (C3S/C4S) (Fig. 4A, left panel). In contrast, charge movement was increased to similar values in either ␣ 1C /␤ 2a or ␣ 1C / ␤ 2a (C3S/C4S) cells relative to ␣ 1C cells, which exhibited charge movement that was at the lower limits of detection (Fig. 4A,  right panel). This latter result indicated that similar numbers of functional channels were present in the plasma membrane of cells expressing either the wild-type or mutant ␤ 2a subunits. Because previous studies have demonstrated a role for ␤ subunits to recruit functional channels to the membrane (2,12,13), the present results demonstrate that wild-type and palmitoylation-deficient ␤ 2a subunits are equivalent in this function. However, a plot of peak current amplitude versus charge movement demonstrated that the relationship of the amount of peak current to charge movement was dramatically decreased in channels with the ␤ 2a (C3S/C4S) mutants compared with channels with wild-type ␤ 2a (Fig. 4B), which indicated that less current was carried per functional channel in channels containing ␤ 2a (C3S/C4S). These results suggested that the lack of palmitoylation and/or the lack of further post-translational modification that could require palmitoylation may have resulted in a ␤ 2a subunit that was able to target channels to the membrane as described previously (2) but was altered in its ability to modify calcium channel currents.
Although it is difficult to attribute these functional changes directly to the loss of palmitic acid, it is clear that Cys 3 and Cys 4 were important determinants of palmitoylation and ␤ 2a modulation of channel function. The molecular mechanisms underlying dynamic palmitoylation and de-palmitoylation remain unclear, although it has been suggested that agonists of certain receptors can stimulate the de-palmitoylation of proteins (14,15). Intracellular enzymes involved in either the palmitoylation or de-palmitoylation of proteins have only recently been identified (16,17). Continuing investigations on the pathways and enzymes involved in the regulation of palmitoylation and depalmitoylation should facilitate studies on the role of this post-translational modification with respect to channel function.