Identification and Subcellular Localization of the Subunits of L-type Calcium Channels and Adenylyl Cyclase in Cardiac Myocytes*

The properties of cardiac L-type channels have been well characterized electrophysiologically, and many such studies have demonstrated that the channels are regulated by a cAMP-dependent pathway. However, the subunit composition of native cardiac L-type calcium channels has not been completely defined. Furthermore, a very important question exists regarding the status of the C-terminal domain of the pore-forming α1subunit, as this domain has the potential to be the target of protein kinases but may be truncated as a result of post-translational processing. In the present studies, the α1C and β2 subunits were identified by subunit-specific antibodies after partial purification from heart membranes, or immunoprecipitation from cardiac myocytes. Both the β2and the full-length α1C subunits were found to be expressed and co-localized in intact cardiac myocytes along T-tubule membranes. Using a quantitative antibody binding analysis, we demonstrated that the majority of the α1C subunits in intact cardiac myocytes appear to be full-length. In addition, we observed that adenylyl cyclase is localized in a pattern similar to the channel subunits in cardiac myocytes. Taken together, our results provide new insights into the structural basis for understanding the regulation of L-type calcium channels by a cAMP-mediated signaling pathway.

The regulation of ion channels by protein phosphorylation and dephosphorylation is a common theme in neurobiology. One of the most extensively studied examples involves the cAMP-mediated regulation of the voltage-activated L-type calcium channel in heart. The activation of ␤-adrenergic receptors by norepinephrine facilitates the opening of cardiac L-type channels and regulates cardiac contractility through a protein kinase A (PKA) 1 -mediated phosphorylation of the channels or related proteins (1,2). Although many electrophysiological studies have centered on this important regulatory pathway, very little is known about the biochemical properties of the rare membrane proteins that comprise the channels and the molecular events that underlie their regulation.
Important issues need to be resolved to fully understand the regulatory processes that occur in this prototypical system. An essential first step is to understand the subunit composition of cardiac calcium channels. Most voltage-activated calcium channels are multisubunit complexes composed minimally of pore-forming ␣ 1 subunits along with accessory ␤ and ␣ 2 ␦ subunits. Earlier studies have demonstrated that purified L-type calcium channels contain an ␣ 1 subunit and the universal ␣ 2 ␦ subunit (3,4). Subsequently, cDNA cloning predicted the ␣ 1C isoform to be part of the cardiac calcium channels (5). While the ␣ 1C cDNA predicts a protein with a molecular mass of 242 kDa (5), L-type channel proteins purified from avian and mammalian heart contained ␣ 1 subunits of 190 -200 kDa (3,6). Recent studies have confirmed the suspicion that the purified proteins were ␣ 1C subunits that were truncated at the C terminus (7). Similar C-terminal truncations have been identified in ␣ 1 subunits from skeletal and neuronal L-type channels (8 -10) and the N-type calcium channel (11). It is important to ascertain if these truncations arise as a result of post-translational modifications or as artifacts of isolation and purification of the channel, since the truncations could have serious consequences for channel function and regulation. For example, the truncated cardiac ␣ 1C subunit is not a substrate of PKA in vitro (3,12), while the full-length protein is a substrate (7,12). In addition, a truncated ␣ 1C subunit has been shown to conduct much larger currents than the full-length protein (13). Thus, an important question is: are the ␣ 1C subunits full-length or truncated proteins in the cardiac myocyte? Another important unanswered question is which ␤-subunit(s) complex with the ␣ 1C to form cardiac L-type channels? This question also directly relates to the mechanism of regulation of the channels by PKA, as several ␤-isoforms are predicted to be potential substrates of PKA, while others are not. Northern analysis and polymerase chain reaction cloning have indicated that mRNAs for ␤ 1b , ␤ 2 , ␤ 3 , and ␤ 4 subunits are present in cardiac muscle (14). However, no studies have been performed to demonstrate the expression and the distribution of the ␤ subunits at the protein level in myocytes. In addition, the regulation of the L-type channel by PKA appears to be self-limiting as the cAMP-generating adenylyl cyclase that is present in cardiac myocytes appears to be negatively regulated by Ca 2ϩ influx through L-type channels (15). However, it is not known if the channel subunits and adenylyl cyclase are spatially localized in cardiac myocytes in such a way that the increased Ca 2ϩ resulting from * This work was supported in part by National Institutes of Health Grant HL23306 (to M. M. H.). 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.
§ calcium channel activation and/or Ca 2ϩ -induced Ca 2ϩ release could cause direct inhibition of the adenylyl cyclase. To address these issues, we used multiple channel-specific antibodies to determine whether the ␣ 1C subunit is a full-length or truncated protein in the intact myocyte and to identify its partner ␤ subunit. In addition, we demonstrated the expression and the localization of the calcium channel subunits and adenylyl cyclase in cardiac myocytes.

EXPERIMENTAL PROCEDURES
Materials-Adult male rabbits were purchased from commercial suppliers. The human embryonic kidney (HEK) 293 cells were a gift from Dr. Ron Kopito (Stanford University). A stable HEK293 cell line expressing type V AC (HEK/AC V ) was a gift from Dr. Jack Krupinski, Weis Center for Research, Danville, PA. FITC-and TRITC-coupled secondary antibodies were obtained from Kirkegaard & Perry Laboratories (Gaithersburg, MD). Monoclonal anti-␣-actinin (sarcomeric) (EA-53) antibodies were obtained from Sigma. All other reagents were from standard sources.
Antibody Preparation and Purification-Calcium channel subunitspecific antibodies including Card C, Card I, ␤ 2 , and ␤ gen were described previously (16,17). The Card C antibodies were purified by protein G chromatography following standard procedures. The Card I antibody was affinity purified against ␣ 1C proteins heterologously expressed in Sf9 insect cells. 2 Crude Sf9 cell membranes containing the ␣ 1C proteins were separated by SDS-PAGE and transferred to nitrocellulose. The region on the blots containing the ␣ 1C proteins were cut out and incubated with Card I overnight at 4°C. After extensive washing, the bound antibody was eluted with 0.2 M glycine, pH 2.5, and rapidly neutralized with 1 M K 2 HPO 4 . The specificity of purified Card I was tested using expressed proteins on immunoblots (data not shown).
Partial cDNA clones for the type V and VI adenylyl cyclase (AC) were obtained and sequenced after screening a chicken heart library (Stratagene). A generic AC antibody, AC com , was generated against a chick type VI AC sequence in the C-terminal loop corresponding to amino acids 993-1112 in the dog type VI AC (18), which was identified by sequence alignment using LaserGene to be highly conserved in all ACs. An antibody targeted to type V and VI isozymes, AC V/VI , was generated against a sequence from the 6 -7 loop region of the chick type V AC corresponding to amino acids 547-661 in the dog type V AC (19). This sequence is highly conserved between type V and VI isozymes but has low homology with other AC isozymes (identified by LaserGene alignment). To produce fusion proteins for antibody production, the cDNAs encoding the chick type V and type VI sequences noted above were generated by polymerase chain reaction of the partial chick heart clones and subcloned into pGEX-4T3 (Pharmacia Biotech Inc.), resulting in an in-frame fusion of the AC residues to glutathione S-transferase. While both fusion proteins were insoluble, they were purified by SDS-PAGE and electroelution. Briefly, isopropyl-1-thio-␤-D-galactopyranoside-induced bacteria were resuspended and sonicated in PBS containing 1% Triton X-100. The insoluble protein pellet was washed three times in the same buffer to remove soluble proteins, and the final protein pellet was solubilized in SDS sample buffer and subjected to SDS-PAGE directly. The insoluble fusion protein was the major protein band on the gel revealed by Coomassie staining (data not shown). The fusion proteins were purified by excision from the gel followed by electroelution into SDS electrophoresis buffer (20). Antibodies to the purified fusion proteins were prepared in goats and rabbits at Bethyl Laboratories (Montgomery, TX). Four antibodies were obtained: rabbit AC com , goat AC com , rabbit AC V/VI , and goat AC V/VI (see "Results").
Biotinylation of Antibodies-To avoid the background problem caused by secondary antibodies in studies of low abundance proteins when the same antibody is used for immunoblotting after immunoprecipitation, Card I, ␤ gen , and ␤ 2 antibodies were biotinylated using an ImmunoPure NHS-LC-Biotinylation kit (Pierce) following the standard procedures suggested by the manufacturers.
Myocyte Isolation-Adult rabbit cardiac myocytes were isolated using previously described procedures (21). After isolation, myocytes were transferred to M199 media (Life Technologies, Inc.) and kept at 37°C until used.
Membrane Preparation and Partial Purification of Calcium Channels from Rabbit Hearts-Crude membranes from rabbit heart were prepared as described previously (22) except for the inclusion of prote-ase inhibitors used in Ref. 16. For immunoprecipitation, membranes were solubilized in solubilization buffer (50 mM Tris, pH 7.4, 5 mM EDTA, 5 mM EGTA, 0.1%SDS, 1% Triton X-100 and protease inhibitors (16)) and incubated with the ␤ gen antibody coupled to protein G. Immunoprecipitates were subjected to SDS-PAGE and channel subunits were identified by immunoblotting. For channel purification, membranes were prelabeled with (ϩ)-[ 3 H]PN 200-110 (Amersham) after which channel proteins were solubilized and partially purified with wheat germ agglutinin (WGA)-Sepharose chromatography (Sigma) as described previously (3). Fractions containing the peak of dihydropyridine binding were concentrated, and the protein components of the channel were analyzed by SDS-PAGE followed by immunoblotting.
Immunoprecipitation and Immunoblotting-Freshly isolated rabbit myocytes were homogenized in buffer A (250 mM sucrose, 0.25 M KCl, 10 mM imidazole, pH 7.4, 5 mM MgCl 2 , 10 mM EDTA, and protease inhibitors (16)) using a Tri-R Dounce homogenizer. The homogenates were centrifuged at 5,000 ϫ g for 10 min. Pellets were rapidly washed three times with buffer A containing 0.6 M KCl to extract myosin. The pellets were then washed once with buffer B (50 mM Tris-HCl, pH 7.4, 2 mM EDTA, 2 mM EGTA, and protease inhibitors), and resuspended in buffer B containing 1% SDS. Solubilized proteins were diluted 5-fold in buffer C (buffer B containing 0.8% digitonin, 0.25% cholate, 0.2 M NaCl) and immunoprecipitated using Card I and/or the ␤ gen antibodies coupled to immobilized protein G (Pierce). Pellets were washed 5 times in 20 volumes of buffer C containing 0.5 M NaCl. Immunoprecipitates were analyzed by SDS-PAGE followed by immunoblotting. Channel subunits were detected with enhanced chemiluminescence (Pierce) and horseradish peroxidase as described (16). When biotinylated antibodies were used for immunoblotting, proteins were visualized using neutroavidin and biotin-conjugated horseradish peroxidase (Pierce).
To test the specificity of the AC antibodies, HEK/AC V cells were sonicated in homogenization buffer (20 mM HEPES, 1 mM EDTA, pH 7.4, and protease inhibitors (16)) and crude membrane fractions were obtained by centrifugation. For immunoprecipitation, membranes were solubilized in immunoprecipitation buffer (40 mM sodium phosphate, pH 7.4, 50 mM sodium fluoride, 5 mM EDTA, 5 mM EGTA, 0.5 M NaCl, 1% Nonidet P-40, 0.5% deoxylcholate, 0.1% SDS, and protease inhibitors) and solubilized proteins were incubated with the indicated AC antibodies immobilized to protein G (Pierce). Immunoprecipitates were washed with immunoprecipitation buffer, subjected to SDS-PAGE and immunoblotting.
Immunofluorescence Staining-Freshly isolated rabbit myocytes were seeded onto poly-L-lysine-coated coverslips and incubated at 37°C for 1-2 h in M199 media (Life Technologies, Inc.). Prior to staining, myocytes were washed twice with relaxation buffer (0.1 M KCl, 5 mM EGTA, 5 mM MgCl 2 , 0.25 mM dithiothreitol in phosphate-buffered saline (PBS), pH 6.8), followed by fixing in pre-cooled (Ϫ20°C) methanol/ acetone (1:1) for 5-10 min at 4°C. Fixed myocytes were incubated overnight in labeling buffer (1% bovine serum albumin, 2% normal goat serum in PBS) to block nonspecific binding. Different primary antibodies were diluted in labeling buffer and incubated with myocytes overnight at 4°C. In the cases of double labeling, two primary antibodies were used simultaneously. Different mixtures of secondary antibodies were used subsequently, including: FITC-conjugated rabbit anti-goat IgG, FITC-conjugated goat anti-rabbit IgG, and TRITC-conjugated goat anti-mouse IgG. Coverslips were mounted onto slides and viewed on a scanning confocal microscope (Zeiss). For all the channel and AC antibodies, negative control experiments were performed using preimmune antibody, secondary antibody alone, and/or antigen-preabsorbed antibody; in each case we observed no specific staining (data not shown).
Cell Culture and Transfection-Transient expression of different calcium channel subunits in HEK293 cells has been described previously (16,17).
Quantifying Binding of Antibodies to Proteins in Intact Cells-As will be described below, it was of interest to determine the relative binding efficiencies of different antibodies to recognize the same proteins. Specifically, we designed experiments to allow comparison of recognition of the ␣ 1C subunit by Card I versus Card C. To do this we used heterologously expressed channel subunits as the standards. HEK293 cells were transiently transfected with plasmids encoding the ␣ 1C , ␤ 2a , and the ␣ 2 ␦ subunits (16). Cells were split into poly-L-lysine-coated 24-well plates and were fixed ϳ48 h post-transfection using the method described above for immunofluorescence studies. Cells were then incubated with 2% bovine serum albumin in PBS for 1-2 h at room temperature, followed by overnight incubation with the indicated primary antibodies at 4°C. After washing out unbound primary antibodies, 125 I-labeled protein G (NEN Life Sciences Products) was incubated with the cells for 2-4 h at 4°C. After extensive washing with PBS, bound radioactivity was eluted from the plates using 2% SDS and quantified using a ␥-counter. Nonspecific binding from primary antibodies was determined after preabsorbing antibodies with their immunizing antigens.

RESULTS
The ␤ 2a Subunit Is Expressed and Complexed with the ␣ 1C Subunit in Rabbit Cardiac Myocytes-To identify which components comprise the cardiac calcium channels, we partially purified the L-type calcium channels from rabbit heart using WGA-Sepharose chromatography. Cardiac membranes were prelabeled with the dihydropyridine ligand (ϩ)-[ 3 H]PN 200-110 (3) and purification was followed by the specific binding (Fig. 1A). Fractions enriched in binding were analyzed with the ␣ 1C subunit specific antibody Card I (Fig. 1B). The major component recognized by Card I, which is directed against an internal domain of the ␣ 1C subunit (16), was a diffuse protein migrating at ϳ200 kDa (Fig. 1B). This is likely to be the truncated form of the ␣ 1C subunit of the L-type calcium channel (3,7). The fractions enriched in PN 200-110 binding were pooled and analyzed for reactivity with Card C, which is directed against the C-terminal region of the ␣ 1C subunit (16). Card C will only detect full-length ␣ 1C and not react with a C-terminal truncated protein. The Card C antibody recognized a protein migrating at 240 kDa (Fig. 1C, lane 2), that migrated similarly to Sf9 cell expressed ␣ 1C subunit (Fig. 1C, lane 1) which was included as a positive control, since previous studies showed that this heterologously expressed ␣ 1C subunit is a full-length protein with intact N and C termini (16). 2 Card I, which will recognize the full-length and C-terminal truncated protein, recognized essentially only the smaller 190 -200-kDa protein (Fig. 1C, lane 3). The lack of significant staining at the 240-kDa position with this antibody suggested that most of the ␣ 1C subunit in the partially purified preparation is truncated at its C terminus. The stronger staining of the full-length ␣ 1C subunit by Card C versus Card I reflected the greater sensitivity of the Card C antibody to detect denatured ␣ 1C proteins on Western blots (this was also observed using the Sf9 full-length ␣ 1C as a standard).
To detect which ␤ subunit co-purified with the ␣ 1C subunit, we stained the pooled WGA-Sepharose eluates with ␤ subunitspecific antibodies. In the same fractions that contained ␣ 1C , a diffuse protein of 68 -70 kDa was specifically immunostained by the ␤ 2 -specific antibody (Fig. 1C, lane 6), and exhibited mobility similar to the heterologously expressed ␤ 2a subunit from Sf9 cells that was run as a control (Fig. 1C, lane 1). Taken together, the results from the WGA-Sepharose purification suggest that ␣ 1C and ␤ 2 subunits are likely components of cardiac L-type calcium channels.
To begin to address whether the truncated ␣ 1C subunit is an artifact of isolation or a physiological component of the channels, we sought to minimize possible proteolysis during the isolation procedure by using cardiac myocytes. In addition, since heart is a heterogeneous tissue, studies with the cardiac myocytes were performed to ensure that the ␣ 1C and ␤ 2 subunits were indeed expressed in cardiac myocytes versus other cell types (e.g. smooth muscle) that are present in whole heart preparations. Crude membranes were quickly prepared from freshly isolated myocytes, solubilized in 1% SDS, and directly subjected to immunoprecipitation with a mixture of the Card I and ␤ gen (a generic ␤ subunit antibody (17)) antibodies. The immunoprecipitates were subjected to SDS-PAGE and analyzed for ␣ 1C content by immunoblotting with either Card C or biotinylated-Card I (Bio-I). A 240-kDa protein was detected by Card C in the immunoprecipitates, which corresponded to the full-length ␣ 1C subunit with an intact C terminus (Fig. 2A, lane  1). On the other hand, a major component of ϳ200 kDa and a trace of the 240-kDa component corresponding to the truncated and the full-length forms of the ␣ 1C subunit, respectively, were identified by Bio-I in the immunoprecipitates. Despite the fact that multiple protease inhibitors were used, the majority of the ␣ 1C subunit was truncated at the C terminus. The immunostained proteins migrated over a broad range, suggesting the presence of multiple proteolytic products that might arise as a result of proteolytic breakdown that occurs during isolation of the protein.
The immunoprecipitates from myocytes were also analyzed for the ␤ 2 subunit. The same 68 -70-kDa proteins discussed earlier were immunoprecipitated by the ␤ gen antibody from myocytes and were identified by subsequent immunoblotting with the ␤ 2 antibody (Fig. 2A, lane 1). In addition, blotting of the immunoprecipitates with biotinylated-␤ gen (Bio-␤ gen ) resulted in staining of the same proteins ( Fig. 2A, lane 4). Thus the ␤ 2 subunit is present in cardiac myocytes.
We asked whether any other ␤ subunits could be identified from heart by the ␤ gen antibody (17). As a positive control, we ran heterologously expressed ␤ 1b , ␤ 2a , ␤ 3 , and ␤ 4 subunits, which were readily detected by Bio-␤ gen (Fig. 2B, lanes 3-6). However, the only immunoreactive protein detected in the immunoprecipitates from rabbit heart membranes by Bio-␤ gen was the same as that detected by the ␤ 2 antibody (Fig. 2B, lanes  1 and 2). These results suggested that ␤ 2 is the major cardiac isoform of the ␤ subunit as detected by our antibodies.
Subcellular Localization of Cardiac Calcium Channels-Ltype calcium channels have been localized to the T-tubular system in skeletal muscle (23,24), but only very limited information is available concerning their distribution in mammalian cardiac muscle. We investigated the subcellular localization of the ␤ 2 and ␣ 1C subunits in cardiac myocytes using immunofluorescence staining. Previous studies have demonstrated that the T-tubule network is closely apposed to the Z-lines in ventricular myocytes (25,26). An anti-␣-actinin antibody was used to label the Z-lines in intact myocytes (27), to obtain an indication of T-tubule localization. To reveal the distribution of the ␤ 2 subunit, freshly isolated rabbit ventricular myocytes were co-stained with the ␤ 2 and the anti-␣-actinin antibodies, and visualized using FITC-and TRITC-conjugated secondary antibodies, respectively (Fig. 3, A-C). The staining pattern for both the ␤ 2 subunits and ␣-actinin were observed as regularly-spaced and evenly-distributed transverse bands (Fig.  3, A and B). To further determine whether the ␤ 2 subunit is expressed on T-tubules, two confocal images obtained from the same section of the myocyte (Fig. 3, A and B) were overlaid using Adobe software (Fig. 3C). The predominant yellow color in the merged image indicated that the ␤ 2 subunit and ␣-actinin are closely associated. The most reasonable interpretation of these data is that the ␤ 2 subunit is localized along T-tubule membranes in cardiac myocytes. To corroborate these findings, we used the ␤ gen and the anti-␣-actinin antibodies to co-label the myocytes; similar striated staining associated with the T-tubules was also observed (data not shown). These results provided the first demonstration that the ␤ 2 subunits are expressed in cardiac myocytes and localized on the T-tubules.
We next investigated the presence and the subcellular localization of the ␣ 1C subunits using immunofluorescence staining in intact cardiac myocytes. We were particularly interested in the possibility of detecting the full-length ␣ 1C subunit in the intact cells where the possibility of proteolysis should be minimal. Either the purified Card I (pCard I) or the purified Card C (pCard C) antibodies were applied separately or co-applied with the anti-␣-actinin antibody to stain the fixed myocytes. Images obtained from pCard I (Fig. 3D) showed the same distribution as obtained with the ␤ 2 antibody. The images of pCardI and the anti-␣-actinin antibody (data not shown) were merged, and the resulting overlaid image is shown in Fig. 3E. The staining pattern again formed regularly-spaced transverse bands in yellow color indicating that the ␣ 1C subunit is also associated with T-tubule membranes as is the ␤ 2 subunit. Similar results were obtained with pCard C staining (Fig. 3, F and  G). Importantly, the images obtained with pCard C (Fig. 3F) resulted in a staining pattern that was indistinguishable from that of pCard I (compare Fig. 3D with 3F, and 3E with 3G), and that of the ␤ 2 subunit (Fig. 3C). These results strongly suggested that the full-length ␣ 1C subunits are present in intact myocytes and likely associated with the ␤ 2 subunit (as well as the ␣ 2 ␦ subunits (26)) along the T-tubules. Experiments to co-stain the myocytes with combinations of ␣ 1C subunit specific and ␤ subunit-specific antibodies (e.g. Card I with anti-␤ 2 or Card C with ␤ gen ) resulted in no specific staining (despite the fact that strong signals were observed when used separately). These results were likely due to steric hindrance between the two antibodies trying to access two closely associated antigens (the ␣ 1C and the ␤ 2 subunits).
Are All the ␣ 1C Subunits Full-length in Intact Cardiac Myocytes?-Cardiac L-type calcium channels can be modulated by cAMP-dependent phosphorylation. The ␣ 1C subunit is a potential target for PKA-mediated phosphorylation, however, an intact C terminus of the ␣ 1C subunit is required (7,12), since the truncated ␣ 1C subunit is not a PKA substrate (3). The data discussed so far have demonstrated that the full-length ␣ 1C subunits were present and localized on T-tubules in intact myocytes (Fig. 3F), however, the majority of the immunoprecipitated or purified ␣ 1C proteins were truncated (Figs. 1 and  2). It was therefore of particular interest to attempt to assess the amount of the full-length ␣ 1C subunits in the intact myocytes. To do so we developed a quantitative assay using Card C and Card I to analyze the ratio of full-length versus truncated ␣ 1C proteins in situ. Methanol fixation of intact cells was used to avoid possible proteolysis during cell lysis, and 125 I-labeled protein G was used to determine the amount of primary antibody bound to the antigens. Fluorescence-conjugated secondary antibodies were not used because it is very difficult to quantify and normalize the fluorescent signals produced by different secondary antibodies.
Since the method relies on quantifying the signal from Card I versus Card C, it was very important to determine the relative sensitivities of Card I and Card C toward the ␣ 1C subunits in native cells, as we previously recognized differences in the ability of the antibodies to detect the proteins after SDS-PAGE FIG. 2. Immunoprecipitation of cardiac L-type calcium channel subunits. A, freshly isolated rabbit myocytes were homogenized and solubilized with 1% SDS followed by immunoprecipitation with a mixture of Card I and the ␤ gen antibody. The immunoprecipitates were analyzed by immunoblotting with specific antibodies as indicated at the bottom for the ␤ or the ␣ 1C subunits. Immunoprecipitated proteins were blotted with either a mixture of Card C (C) and the ␤ 2 antibody (lanes 1 and 2) or a mixture of Bio-I and Bio-␤ gen (lanes 3 and 4). Lanes 1 and 4 were stained with immune sera, and lanes 2 and 3 were stained with preimmune sera. A 240-kDa protein (␣ 1 ) was recognized by Card C (lane 1), while two proteins of 240 (␣ 1 ) and 200 kDa (Ç␣ 1 ) were detected by Bio-I (lane 4). A diffuse protein band at ϳ68 -70 kDa was recognized by both the ␤ 2 antibody (lane 1) and Bio-␤ gen (lane 4). Preimmune sera revealed no specific staining (lanes 2 and 3). B, crude membranes were prepared from frozen rabbit heart and solubilized proteins were immunoprecipitated with the ␤ gen antibody and blotted with either Bio-␤ 2 (lane 1) or Bio-␤ gen (lane 2). A diffuse protein band at ϳ68 -70 kDa was detected from immunoprecipitates by both Bio-␤ 2 and Bio-␤ gen (lanes 1  and 2). HEK293 cells were transfected with cDNAs encoding ␤ 1b , ␤ 2a , ␤ 3 , or ␤ 4 , and membrane fractions were prepared from transfected cells and used here as standards to demonstrate the expected immunoreactivity of Bio-␤ gen against other ␤ isoforms. Shown in lanes 3-6 are heterologously expressed ␤ 1 , ␤ 2a , ␤ 3 , and ␤ 4 proteins, respectively, detected by Bio-␤ gen . and immunoblotting. However, we did not use the differences detected by Western blotting to quantify antibody recognition of ␣ 1C , since it is well recognized that antibodies differ in their ability to detect antigens in various types of immunodetection assays. To quantify and compare Card I and Card C reactivity, we used heterologously expressed full-length ␣ 1C protein as a standard. In previous studies we demonstrated expression of functional calcium channels in HEK293 cells, in which all the ␣ 1C subunits were expressed as the full-length form (16). HEK cells were transfected, fixed, and incubated with Card I and Card C followed by 125 I-protein G as described under "Experimental Procedures." As shown in Table I, the total binding from both antibodies was first determined (value T). The Card I and Card C antibodies were preabsorbed with the antigens, and nonspecific bindings were obtained (Table I, value NS). The relative sensitivity (value S) of Card I or Card C antibodies to detect a similar amount of expressed ␣ 1C subunit was determined by subtracting the nonspecific counts from the total bound 125 I-labeled protein G counts (Table I, S ϭ T Ϫ NS). By averaging the results from four independent experiments, the relative binding sensitivity of the Card I versus the Card C antibody (S I /S C ) was assessed to be 2.95 Ϯ 0.4 (n ϭ 4). This ratio implied that the signals produced by the specific binding of the Card I antibody to an equivalent amount of the ␣ 1C subunit under these conditions was ϳ3 times stronger compared with that of the Card C antibody.
We next asked how much full-length ␣ 1C subunit was present in intact cardiac myocytes. The rationale was that the ratio of Card I/Card C binding would reflect the proportion of fulllength versus truncated ␣ 1C subunits in the intact myocytes. If all the ␣ 1C was full-length, the ratio should be as in the transfected cells. However, if a significant portion of the ␣ 1C was truncated in the intact myocyte, then the binding of Card C (which only binds the full-length protein) would decrease while the binding of Card I would be unchanged (as Card I recognizes both the truncated and full-length forms of ␣ 1C ). Accordingly, the Card I/Card C binding ratio would become larger. To compare the results with those obtained from transfected HEK cells, the myocytes were stained with Card I and Card C followed by 125 I-protein G incubation under identical conditions FIG. 3. Subcellular localization of calcium channel ␣ 1C and ␤ 2 subunits in rabbit ventricular myocytes. Images were produced using a laser scanning confocal microscope (Zeiss) to examine rabbit myocytes after labeling with anti-␣-actinin antibody and antibodies specific for either the ␤ 2 or the ␣ 1C subunit. Visualization was done using FITC-and TRITC-conjugated secondary antibodies. Images shown are: A, immunolocalization of the ␤ 2 subunit detected by the ␤ 2 specific antibody; and B, expression pattern of ␣-actinin in ventricular myocytes. C, merged image produced by superimposing A and B. The predominant yellow color suggests that the ␤ 2 subunit is distributed along the T-tubules at the level of the Z-lines in cardiac ventricular myocytes. Images shown in D and E were obtained from myocytes stained with Card I and anti-␣-actinin antibody. D, the staining pattern of ␣ 1C subunits revealed by Card I. E, a merged image generated by superimposing images from double staining experiments using Card I and anti-␣-actinin antibody. Images shown in F and G were obtained from myocytes stained with Card C and anti-␣-actinin antibody. F, the staining pattern of ␣ 1C subunits revealed by Card C. G, a merged image generated by superimposing images from double staining experiments using Card C and anti-␣-actinin antibody. Note the identical staining patterns identified by Card I and Card C.

TABLE I
Quantitative binding of calcium channel ␣ IC subunit specific antibodies in transfected HEK293 cells and rabbit myocytes Results shown are radioactive counts per minute (cpm) of 125 I-labeled protein G bound to the transfected HEK cells and rabbit myocytes under the conditions described. The results shown are the raw data from a single experiment (except for the means given on the last line) and are representative of four separate experiments in which antibody binding was performed under identical conditions to paired groups of transfected HEK cells and cardiac myocytes. T, total binding: NS, nonspecific bindings; as described under "Experimental Procedures." The ratio of specific binding of Card I/Card C (S I /S C ) is given as the mean Ϯ S.E. from four independent experiments. 125  as described above. We quantified the binding of these antibodies in the myocytes in at least four separate experiments, and the results indicated that the ratio of radioactivity due to the specific binding of the two antibodies to the ␣ 1C subunits in the myocytes was 2.8 Ϯ 0.3 (n ϭ 4) ( Table I). This ratio is similar to that obtained in the transfected cells (Table I), indicating that the total amount of the ␣ 1C subunits detected by Card C was similar to that detected by Card I in the intact myocytes. These results provided the first experimental evidence that all the ␣ 1C subunits expressed in myocytes contain an intact C terminus, and the truncation observed in the isolated protein likely occurs artificially during channel isolation.
Are the Calcium Channel Subunits Co-localized with Adenylyl Cyclase at the Subcellular Level?-Previous mRNA analysis predicted that heart preparations contain the type V and VI isoforms of AC (28), however, no data are available to demonstrate the presence of those isozymes at the protein level. Expressed type V and VI AC have been demonstrated to be inhibited by micromolar Ca 2ϩ (28), and indeed, adenylyl cyclase in cardiac myocytes is known to be inhibited by Ca 2ϩ (15). This Ca 2ϩ -dependent inhibition can be relieved by blockers of L-type calcium channels (15). To test the hypothesis that the inhibition of AC by Ca 2ϩ influx through the L-type channel was achieved by a close spatial proximity of the channel and AC protein, we developed two new antibodies (described under "Experimental Procedures") that should recognize the cardiac AC isoforms. The specificity of the antibodies was tested using an HEK293 cell line stably expressing rat type V AC isozyme (HEK/AC V ). Fig. 4, panel A, shows a Western blot in which a crude membrane preparation of HEK/AC V cells was solubilized, immunoprecipitated with polyclonal goat sera (G-AC V/VI or G-AC com ), and probed with rabbit AC V/VI (R-AC V/VI ) or preimmune sera. A major protein band of ϳ200 kDa was detected by the R-AC V/VI antibodies from the crude solubilized membrane preparation by immunoblotting (Fig. 4A, lane 1) and in preparations that were immunoprecipitated by either the G-AC com or G-AC V/VI antibodies and subsequently immunobloted with the R-AC V/VI antibody (Fig. 4A, lanes 2 and 4). The size of this protein is the same as that expected for glycosylated rat type V adenylyl cyclase. 3 This protein band was not detected in the immunoblots by the preimmune rabbit serum (Fig. 4A,  lanes 3 and 5), when preimmune goat sera were used in the immunoprecipitation step (data not shown), or when untrans-fected HEK293 cells were probed with R-AC V/VI antibody (data not shown). It is thus clear that the R-AC V/VI antibody specifically recognizes the type V AC protein in immunoblotting experiments and that both the G-AC V/VI and G-AC com antibodies specifically immunoprecipitate the expressed type V AC protein.
Rabbit cardiac myocytes were analyzed in immunofluorescence studies with G-AC V/VI to assess the subcellular distribution of the cardiac enzyme. Images obtained from immunofluorescence staining indicated that the antibody recognized the cardiac AC and that it had a distribution pattern identical to the channel subunits in cardiac myocytes (Fig. 4B). Co-staining of the cells with G-AC V/VI and anti-␣-actinin antibodies demonstrated close localization of these two proteins, and, as was the case for the channel subunits, strongly suggested a Ttubule localization of the AC protein (Fig. 4C). Identical results were obtained with G-AC com (data not shown). Experiments that attempted to co-stain the myocytes with the AC antibodies and channel-specific antibodies together revealed no specific staining, as was the case in similar studies that attempted to use two channel antibodies together. These results provide the first direct visualization at the protein level of cardiac AC. Since the enzyme could be detected with an antibody targeted to type V or VI AC, the results strongly suggested that cardiac myocytes express a type V or VI AC. In addition, they are the first to demonstrate the subcellular localization of the AC and its close proximity to the L-type calcium channels. DISCUSSION The results of the present study have demonstrated the expression and localization of L-type calcium channel subunits and adenylyl cyclase in cardiac myocytes using biochemical and immunocytochemical approaches. The novel findings were: (i) cardiac myocytes express the ␤ 2 subunit of L-type calcium channels and it is co-localized on T-tubule membranes with the ␣ 1C subunit; (ii) the ␣ 1C subunit is a full-length protein in cardiac myocytes, and the C-terminal truncated version of the protein likely arises during isolation procedures; (iii) cardiac myocytes express a type V and/or VI isoform of AC and it co-localizes with the L-type calcium channels on the T-tubule membranes. Further discussion of these points follows.
Recent studies have revealed that the ␤ subunits of calcium channels play multiple roles in modulating channel formation and function in heterologous systems. The direct interaction between the ␤ subunits and the ␣ 1 subunits has been shown in 3 J. Krupinski, personal communication.

FIG. 4. Expression and subcellular localization of AC in rabbit ventricular myocytes.
A, the specificity of the AC antibodies was determined using HEK/AC V cells. Crude cell membranes were prepared, solubilized, and subjected to either SDS-PAGE directly (lane 1) or used for immunoprecipitation. Solubilized proteins were immunoprecipitated by either G-AC com (lanes 2 and 3) or G-AC V/VI (lanes 4 and 5). The immunoprecipitates were separated by SDS-PAGE and blotted with R-AC V/VI immune (lanes 1, 2, and 4, a dilution of 1:2500 of antisera was used) or preimmune sera (lanes 3 and 5). The immunoprecipitates were from five times the amount of the solubilized membrane material that was run in lane 1. Images shown in B and C were obtained from the myocytes stained with G-AC V/VI and anti-␣-actinin antibody. B, expression and subcellular localization of AC as revealed by G-AC V/VI . C, a merged image produced by superimposing images from double labeling the myocytes with G-AC V/VI and anti-␣-actinin antibody.
vitro (29). However, there are only a few reports demonstrating the expression of ␤ subunits and their association with the ␣ 1 subunits in native tissues (30). In the present study, the association of the ␤ 2 subunit with the ␣ 1C subunit in cardiac myocytes has been clearly established. First, we have shown that the ␤ 2 subunit co-purified through WGA-Sepharose chromatography with the ␣ 1C subunits from cardiac muscle. Second, the ␤ 2 subunit co-localized with the ␣ 1C subunit to the T-tubule network in cardiac myocytes.
Although the results of Northern analyses have suggested that there could be more than one ␤ isoform present in heart (14, 31), we did not detect the expression of other ␤ subunits besides the ␤ 2 isoform using the ␤ gen antibody. It could be argued that this might be due to the lack of sensitivity of the ␤ gen antibody to other ␤ isoforms, however, this antibody readily detects the other expressed ␤ subunits (17). Thus, a more likely explanation of the present results is that, if there are other ␤ subunits expressed in heart, the level of expression compared with that of the ␤ 2 subunit is low and beyond the detection of the ␤ gen antibody. The results here strongly suggest that the ␤ 2 subunit is the major cardiac isoform.
An important finding of the present study is that all the ␣ 1C subunit appears to be full-length in cardiac myocytes. The development of a quantitative assay to assess the amount of full-length ␣ 1C subunits provided the first demonstration that the C terminus of the ␣ 1C subunit is intact in myocytes. Since the C terminus of the ␣ 1C subunit may play critical roles in mediating phosphorylation and regulation of the calcium channels (2), this finding provides an important piece of information concerning the receptor-mediated regulation of the channels. If the C terminus was truncated in the intact cells, as it is after isolation of the channels by purification or immunoprecipitation, it would be difficult to assign a role to the ␣ 1 subunit in channel regulation by PKA, because the truncated ␣ 1C subunit is not a PKA substrate while the full-length protein is a substrate (3,7,12). The finding that the C terminus is intact in myocytes allows serious consideration of the ␣ 1C subunit as a major PKA target in vivo. Several possible scenarios may explain why the majority of the ␣ 1C subunit appears to be truncated when examined in purified preparations. The truncation may result from proteolysis that occurs by a protease whose activity is not inhibited by the protease inhibitors used by ourselves and others. Alternatively, the proteolysis may occur in vivo as a result of post-translational modification, but the C terminus may remain associated with the rest of the channel and is "lost" only upon cell disruption. Other scenarios are also possible and further testing will be required to resolve this issue.
Many electrophysiological studies have shown that calcium channel function is regulated by ␤-adrenergic receptor agonists through adenylyl cyclase and the cAMP-dependent signaling pathway (1). However, this pathway appears to have an internal turn-off mechanism, as cardiac AC activity is counterregulated by Ca 2ϩ influx through L-type calcium channels (15). The Ca 2ϩ that enters cardiac myocytes through L-type calcium channels is known to induce further increases in Ca 2ϩ from ryanodine-sensitive calcium-release channels. One mechanism to explain calcium-mediated inhibition of cardiac AC would be close association of cyclase with the calcium channels. In the present study, we demonstrated that either the type V or type VI Ca 2ϩ -sensitive AC is localized in rabbit myocytes and is present on the T-tubule membranes in a spatial distribution indistinguishable from the calcium channel complex. Other studies have suggested a close localization of the L-type cal-cium channels with the calcium-release channels (26). Taken together, these findings suggest that the regulation of the calcium channel by cAMP signaling and the regulation of the AC by Ca 2ϩ are likely facilitated by a close spatial association between the channel and the AC proteins.
In summary, we have shown that the ␤ 2 subunit of the L-type calcium channel is expressed and localized on T-tubule membranes in association with the full-length ␣ 1C subunit. Thus the minimal proven subunit composition of cardiac L-type channels is ␣ 1C , ␤ 2 , and ␣ 2 ␦. The C terminus of the ␣ 1C subunit is intact in cardiac myocytes and loss of the C terminus of the ␣ 1C subunit likely occurs during cell lysis. Moreover, we demonstrated a similar subcellular localization of the Ca 2ϩ -sensitive type V and/or VI adenylyl cyclase with the calcium channel in cardiac myocytes.