Sorcin Associates with the Pore-forming Subunit of Voltage-dependent L-type Ca2+ Channels*

Intracellular Ca2+release in muscle is governed by functional communication between the voltage-dependent L-type Ca2+ channel and the intracellular Ca2+ release channel by processes that are incompletely understood. We previously showed that sorcin binds to cardiac Ca2+ release channel/ryanodine receptors and decreases channel open probability in planar lipid bilayers. In addition, we showed that sorcin antibody immunoprecipitates ryanodine receptors from metabolically labeled cardiac myocytes along with a second protein having a molecular weight similar to that of the α1 subunit of cardiac L-type Ca2+ channels. We now demonstrate that sorcin biochemically associates with cardiac and skeletal muscle L-type Ca2+ channels specifically within the cytoplasmically oriented C-terminal region of the α1 subunits, providing evidence that the second protein recovered by sorcin antibody from cardiac myocytes was the 240-kDa L-type Ca2+ channel α1 subunit. Anti-sorcin antibody immunoprecipitated full-length α1 subunits from cardiac myocytes, C2C12 myotubes, and transfected non-muscle cells expressing α1 subunits. In contrast, the anti-sorcin antibody did not immunoprecipitate C-terminal truncated forms of α1 subunits that were detected in myotubes. Recombinant sorcin bound to cardiac and skeletal HIS6-tagged α1 C termini immobilized on Ni2+ resin. Additionally, anti-sorcin antibody immunoprecipitated C-terminal fragments of the cardiac α1 subunit exogenously expressed in mammalian cells. The results identified a putative sorcin binding domain within the C terminus of the α1 subunit. These observations, along with the demonstration that sorcin accumulated substantially during physiological maturation of the excitation-contraction coupling apparatus in developing postnatal rat heart and differentiating C2C12 muscle cells, suggest that sorcin may mediate interchannel communication during excitation-contraction coupling in heart and skeletal muscle.

The release of Ca 2ϩ from muscle sarcoplasmic reticulum (SR) 1 is the principal link between electrical excitation of the sarcolemma and mechanical activation of the myofilaments, a process known as excitation-contraction (E-C) coupling. Ca 2ϩ release from stores in the cardiac SR occurs via Ca 2ϩ release channels that are referred to as ryanodine receptors (RyRs). In the heart, Ca 2ϩ release from the SR is largely triggered by Ca 2ϩ influx at the plasma membrane via voltage-dependent L-type Ca 2ϩ channels that are also dihydropyridine receptors (DHPRs) (1)(2)(3)(4)(5)(6)(7). In skeletal muscle, DHPRs serve as voltage sensors to detect depolarization of the sarcolemmal transverse tubules (T-tubules) and provide the physical impetus for opening of the SR Ca 2ϩ release channels (1, 8 -11). In both tissues, the geometry and close spatial relationship between sarcolemmal L-type channels and SR RyRs is critical in determining the time course of Ca 2ϩ release and E-C coupling; however, the molecular mechanisms that mediate interchannel communication are incompletely understood (1,(12)(13)(14). The possibility that additional proteins or factors might be interposed to facilitate cross-talk has often been suggested (1,8,12).
Sorcin, a 22-kDa Ca 2ϩ -binding protein first identified in multidrug-resistant cells, is widely distributed among mammalian tissues, including heart and skeletal muscle (15)(16)(17)(18)(19). At the subcellular level, sorcin localizes to T-tubule junctions of cardiac SR (19) and co-localizes with brain RyR in rat brain caudate-putamen nucleus (20,21). We previously demonstrated that introduction of sorcin into nonmuscle cells confers the property of caffeine-activated intracellular Ca 2ϩ release, suggesting a role for sorcin in modulating RyR function (19). This hypothesis was further strengthened by the demonstration that sorcin completely inhibits ryanodine binding to cardiac RyRs and substantially decreases the open probability of the Ca 2ϩ release channels reconstituted in lipid bilayers (22). Sorcin is, therefore, one of a group of modulators of RyR gating that includes calmodulin and FK506-binding protein, ligands that bind to a RyR domain that connects directly to a cytoplasmic extension of the transmembrane assembly of the receptor (23)(24)(25)(26)(27).
Anti-sorcin antibody immunoprecipitates two proteins from metabolically labeled cardiac myocytes (19), one of which we previously identified as the 565-kDa cardiac RyR (19), and the other (ϳ220 kDa) displayed an electrophoretic mobility similar to that of the major, pore-forming ␣ 1 subunit of the cardiac L-type Ca 2ϩ channel (␣ 1C ) (28,29). Here we directly test the possibility that the unidentified protein is ␣ 1C by analyzing the interaction of sorcin and the ␣ 1 subunits of cardiac and skeletal muscle Ca 2ϩ channels. The ability of sorcin to interact with sarcolemmal L-type channels, in concert with its modulatory effect on RyR gating, positions sorcin as a candidate regulator of interchannel cross-talk.

MATERIALS AND METHODS
Cell Lines and Cultured Cardiac Myocytes-COS-1 cells were obtained from ATCC (Rockville, MD) and grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Culture conditions for human embryonic kidney (HEK) 293 and Sf9 insect cells have been described (29 -31). Mouse C2C12 myoblasts (32)(33)(34) were maintained in Dulbecco's modified Eagle's medium containing 15% fetal bovine serum and were transferred to Dulbecco's modified Eagle's medium containing 2% horse serum to initiate differentiation, a process that serves as a model of normal myogenesis. Formation of multinucleated myotubes was observed within 48 h after medium transfer. Preparation of rat cardiac myocytes was carried out according to published procedures (19,35).
Membrane Preparations-Sf9 and HEK 293 cell membranes were prepared as described previously (29,30). Fractionation of rat heart tissue for soluble and crude membrane components was carried out by differential centrifugation according to our published procedures (29). In brief, rat heart tissue was homogenized in buffer containing 0.25 M sucrose, 0.25 M KCl, 10 mM imidazole (pH 7.4), 5 mM MgCl 2 , 10 mM EDTA, and protease inhibitors. After centrifugation at 5000 ϫ g for 10 min, supernatants were removed and frozen, and pellets were quickly washed with lysis buffer containing 0.6 M KCl to extract myosin, which obscures ␣ 1c on gels. Washed pellets were solubilized and analyzed for ␣ 1c content, and supernatants were analyzed for sorcin by Western blot as described below. In some experiments, heart tissue was lysed by sonication in 50 mM Tris (pH 7.4) containing 1% Nonidet P-40, 150 mM NaCl, 20 mM NaF, 10 g/ml aprotinin, and 2 mM phenylmethylsulfonyl fluoride (buffer A).
Antibodies and Expression Vectors-Preparation and characterization of sorcin antibodies (19); of SKN, SKC, Card I, and Card C antibodies; and of ␣ 1C and ␣ 1S expression vectors has been described previously (30,36). Briefly, the SKN and SKC antibodies recognize N-terminal and C-terminal domains on the ␣ 1S subunit (36), while the Card I and Card C antibodies recognize internal and C-terminal domains of the ␣ 1C subunit (30). The RyR antibody was a generous gift of Dr. Gerhard Meissner (University of North Carolina) (37). The antimyosin heavy chain (MHC) antibody F59 has been described (34). Anti-polyhistidine was purchased from Sigma, and horseradish peroxidase-conjugated secondary antibodies were obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, CA) and used according to the manufacturer's directions.
Western Blot Analysis and Immunoprecipitation-Soluble fractions of cultured cells and tissues, as well as solubilized membrane fractions (described above) were prepared for analysis in buffer A. Samples containing 60 g of protein (38) were separated by gel electrophoresis on 6% (for analysis of ␣ 1 subunits, RyR, and MHC) or 11% (for sorcin and HIS 6 -tagged proteins) polyacrylamide gels under denaturing conditions (39). Proteins were transferred to nitrocellulose, and the membranes were processed for Western blotting with detection by chemiluminescence (Amersham Pharmacia Biotech) as described previously (19). Additional details are given in the figure legends. For immunoprecipitation assays, aliquots of protein samples in buffer A containing 100 or 200 g of protein in a total of 0.25 ml of buffer A were incubated with 1 g of antibody raised against peptides from either the N or C terminus of sorcin (19) for 1 h at 4°C. Antigen-antibody complexes were precipitated with protein G (Sigma), washed with buffer A, and solubilized in Laemmli (39) buffer at room temperature for 15 min. Immunoprecipitated proteins were analyzed by Western blot as described above and in the figure legends. Molecular weight markers were purchased from Amersham Pharmacia Biotech or Bio-Rad.
In Vitro Binding Assays-Fusion proteins of the L-type Ca 2ϩ channel ␣ 1C and ␣ 1S C-terminal domains were expressed in bacteria with vectors constructed by ligation of a genomic BglII-BamHI fragment of ␣ 1C (GenBank TM accession number X15539) or ␣ 1S (GenBank TM accession number X05921) into the BamHI site of pQE (QIAGEN, Valencia, CA). The proteins were expressed with an N-terminal HIS 6 tag fused to amino acids 1622-2171 (40) of the ␣ 1C C terminus or to amino acids 1497-1873 (40) of the ␣ 1S C terminus. For immobilization of fusion peptides on Probond Ni 2ϩ resin (Invitrogen Corp., San Diego, CA), bacterial lysates containing the fusion proteins were prepared according to the manufacturer's directions by sonication and freeze-thaw in phosphate-buffered saline (PBS) at pH 7.8 (binding buffer). Cleared supernatants were applied to washed resin beds by batch absorption, and treated resin samples were washed three times with binding buffer, twice with PBS at pH 6.0, and three times with PBS at pH 7.4. Aliquots of 10 g of recombinant sorcin (41) in PBS (pH 7.4) were then combined with the protein-containing resins for 30 min at 4°C. Resins were washed three times with PBS (pH 7.0), and bound proteins were eluted in 0.5 M imidazole. HIS 6 -tagged p27, a cyclin-dependent kinase inhibitor (42), was used as a negative control. Eluted proteins were analyzed by Western blot after electrophoresis on 11% acrylamide. Nitrocellulose membranes containing transferred proteins were sequentially probed with mouse monoclonal anti-polyhistidine antibody and then with rabbit polyclonal antibody to a peptide from the C terminus of sorcin (19).
Preparation and Transfection of HIS 6 -tagged C-terminal Fragments of ␣ 1C -A fragment (designated fragment A) of the C-terminal cytoplasmic domain of ␣ 1C (amino acids 1622-2171) (41) was removed from the ␣ 1C expression vector (30) and cloned in frame into the pHISA expression vector (Invitrogen), which added 30 amino acids, including the His 6 epitope, at the N terminus. Fragments B (amino acids 1622-1872), C (amino acids 1749 -1978), D (amino acids 1622-1772), and E (amino acids 1622-1748) from the ␣ 1C C terminus were prepared by polymerase chain reaction and cloned into pHISA. Polymerase chain reaction primers included restriction sites to facilitate subsequent cloning. The following primer pairs were used: fragment B, sense GGAAACCTGGAA-CAAGCCAAT (also used for fragments D and E) and antisense GCTCTAGATCCTCACGTCGTAGTTGTC; fragment C, sense GCG-GATCCTCCCCCAGACCTTCACTACG and antisense GCTCTAGAGC-GAGGTGGGGGAATGGCT; fragment D, antisense GCTCTAGACTAG-GAGTCCACCAGCTTCTCGTG; and fragment E, antisense GCTCT-AGATCCTCACGTCGTAGTTGTC. Plasmids encoding the fragments were transiently co-transfected with the full-length sorcin vector pFRC-MVSOR (19) in COS-1 cells or alone in HEK cells with the use of lipofectamine (Life Technologies, Inc.). Association of sorcin and fragments was studied by immunoprecipitation with sorcin antibody and Western blot analysis.

RESULTS
Sorcin Antibody Immunoprecipitates ␣ 1C -We previously showed that sorcin antibody recovered two proteins from rat cardiac myocytes metabolically labeled with [ 35 S]methionine, a 565-kDa protein shown to be the RyR and an unidentified ϳ220-kDa protein (19). Here we demonstrate that a single protein recognized by Card I, an ␣ 1C -specific antibody shown to recognize the 240-kDa ␣ 1C (29,30), was immunoprecipitated from rat heart tissue by sorcin antibody (Fig. 1A, lane 1). Card C, an antibody directed against the C terminus of ␣ 1C (30), also recognized this protein (see Fig. 4). Results of immunoprecipitations carried out in the absence of Ca 2ϩ (not shown) were indistinguishable from those shown in Fig. 1A. Rat heart contained the 22-kDa sorcin, which co-migrated with recombinant sorcin (41) (Fig. 1B, lane 1), and an 18-kDa species (Fig. 1B, lane 1). Both sorcin forms were detected in most rat (and mouse) heart samples; tissues from some animals contained only the 22-kDa form (19).
To confirm the identity of the cardiac protein recognized by Card I and Card C, we examined whether ␣ 1 subunits could be immunoprecipitated with sorcin antibody from Sf9 cells infected with a recombinant baculovirus directing expression of ␣ 1C or from HEK 293 cells transiently expressing full-length ␣ 1C . This ability would require the presence of endogenous sorcin species in those nonmuscle cells. We found that sorcin antibody recognized an 18-kDa protein in HEK 293 cells (Fig.  1B, lane 2), similar in size to the 18-kDa protein in heart, and an ϳ30-kDa protein in Sf9 cells (Fig. 1B, lane 3). The 18-, 22-, and ϳ30-kDa bands in Fig. 1B were not detected on Western blots with antibody preincubated with the antigenic sorcin peptide (data not shown). Card I recognized a protein immunoprecipitated by anti-sorcin antibody from ␣ 1C -expressing Sf9 (Fig. 1A, lane 4) or HEK 293 cells (Fig. 1A, lane 6). The immunoprecipitated bands co-migrated with bands detected by Card I by direct Western blot of proteins from Sf9 and HEK 293 cells heterologously expressing the ␣ 1C subunit (data not shown). No specific proteins identified by Card I were recovered by immunoprecipitation of ␣ 1C -expressing Sf9 cells with preimmune serum (Fig. 1A, lane 2) or by immunoprecipitation of uninfected Sf9 (Fig. 1A, lane 3) or untransfected HEK cells (Fig. 1A, lane 5) with anti-sorcin antibody. These results strongly suggested that a heretofore unidentified protein immunoprecipitated with sorcin antibody from metabolically labeled cardiac myocytes (19) was ␣ 1C .
The cardiac ␣ 1C subunit has been previously reported to exist in two forms in isolated cardiac membranes (29,43). A minor fraction of the cardiac ␣ 1C is the full-length protein of ϳ240 kDa that can be recognized by both the Card I and Card C antibodies (29), while the major fraction of ␣ 1C in cardiac membranes is a C-terminal truncated protein of ϳ190 kDa that reacts with Card I but not Card C (29). Interestingly, in the experiments described here, only the full-length form of the ␣ 1C was immunoprecipitated by the anti-sorcin antibody (Fig. 1A), suggesting a potential role of the C terminus for the ␣ 1C /sorcin interaction.
Sorcin Antibody Immunoprecipitates Full-length ␣ 1S -We next addressed the question of whether the ␣ 1C /sorcin association was ␣ 1 isoform-specific. SKN, an ␣ 1S -specific antibody generated against the N terminus of ␣ 1S and shown to recognize full-length (214-kDa) and C-terminal truncated (170 -190-kDa) forms of ␣ 1S (31, 36, 44), recognized three proteins in C2C12 myotubes (Fig. 1C, lane 2). None of these proteins were detected in undifferentiated C2C12 myoblasts (Fig. 1C, lane 1); up-regulation of the ␣ 1S subunit during skeletal muscle development has been established (45,46). The anti-sorcin antibody immunoprecipitated only the most slowly migrating form that was detected in myotubes, while lower M r forms that were reactive with SKN were not recovered (Fig. 1C, lane 4). The SKC antibody, directed against the ␣ 1S C terminus, recognizes the full-length 214-kDa ␣ 1S subunit but does not detect the C-terminal truncated forms of ␣ 1S that are commonly observed in skeletal muscle (44,47). In the present experiments, SKC recognized only the full-length subunit in C2C12 myotubes (Fig. 1C, lane 6), a band of similar size recovered by sorcin antibody (Fig. 1C, lane 4). The M r of full-length ␣ 1S has been established by Ferguson plot analysis to be 214 kDa, although under most conditions it migrates anomalously as an ϳ190-kDa protein (44). From these data, we concluded that the lower M r bands in C2C12 myotubes that were immunoreactive with SKN, but not with SKC, were C-terminal truncations of ␣ 1S and that the sorcin antibody only immunoprecipitated fulllength ␣ 1S . These results suggested that an intact C terminus in ␣ 1S was necessary for interaction with sorcin. As in the ␣ 1C studies, we examined whether sorcin antibody would recover ␣ 1S heterologously expressed in Sf9 and HEK 293 cells. SKN (and SKC, not shown) recognized a single protein immunoprecipitated by sorcin antibody from ␣ 1S -expressing Sf9 (Fig. 1C, lane 9) and HEK 293 (Fig. 1C, lane 11) cells. The bands comigrated with proteins recognized by SKN and SKC by direct Western blot (data not shown). No specific proteins identified by SKN were recovered by sorcin antibody immunoprecipitation of uninfected Sf9 (Fig. 1C, lane 7), Sf9 infected with ␣ 1C (Fig. 1C, lane 8), or untransfected HEK cells (Fig. 1C, lane 10).
The results are consistent with the interpretation that sorcin associates with the ␣ 1S and ␣ 1C subunits and that an intact C terminus in each protein is necessary for this interaction.
Association of Sorcin with ␣ 1 C Terminus Domains-In order to test the hypothesis that sorcin interacts with C-terminal domains of ␣ 1C and ␣ 1S , HIS 6 -tagged C terminus fragments of ␣ 1C and ␣ 1S were immobilized on Ni 2ϩ resin and incubated with recombinant sorcin protein. As shown in Fig. 2, recombinant sorcin (22 kDa) specifically bound to immobilized ␣ 1S (Fig. 2,  1, 3, and 5, and eluted proteins are shown in lane 2 (negative control protein HIS 6tagged p27), lane 4 (␣ 1S C terminus (C-term) and sorcin), and lane 6 (␣ 1C C terminus and sorcin). Eluted proteins were fractionated on 11% acrylamide gels and transferred to nitrocellulose membranes. Membranes were blocked, incubated with anti-polyhistidine, washed, incubated with horseradish peroxidase-conjugated goat anti-mouse antibody, and treated with materials for chemiluminescence detection. The membranes were then reblocked and probed with sorcin antibody followed by incubation with horseradish peroxidase-conjugated goat antirabbit antibody and detection by chemiluminescence as previously described (19). Sorcin signals are separated by a space that slightly exaggerates the appropriate distance between the 22-kDa recombinant sorcin signal and p27. lane 4) and ␣ 1C (Fig. 2, lane 6) C termini (57 and 70 kDa, respectively). In contrast, no specific association of sorcin with the irrelevant HIS 6 -tagged p27 protein was observed (Fig. 2,  lane 2).
An association between sorcin and the ␣ 1 subunit C terminus was confirmed and extended in a mammalian expression system. Several different HIS 6 -tagged fragments of the ␣ 1C C terminus were generated and co-transfected with full-length sorcin into COS-1 cells or transfected alone into HEK cells for further delineation of the ␣ 1C sorcin-binding domain. Sorcin antibody consistently immunoprecipitated fragment A, extending from amino acid 1622 to the carboxyl end of ␣ 1C at amino acid 2171 (70 kDa) (Fig. 3, lane A). Two domains of fragment A, one extending from amino acid 1622 to 1978 and one from amino acid 1979 to 2171, were prepared and tested for sorcin association. The proximal fragment and not the distal C-terminal region was immunoprecipitated by anti-sorcin antibody (data not shown). The proximal region was then further examined by generating fragments B and C. Fragment B, extending from amino acid 1622 to 1872 (37 kDa), was readily recovered by sorcin antibody immunoprecipitation (Fig. 3, lane B). Fragment C, extending from amino acid 1748 to 1978 (39 kDa), was consistently unrecoverable by sorcin antibody (Fig. 3, lane C). Sorcin antibody immunoprecipitated fragments D (amino acids 1622-1772) (31 kDa) and E (amino acids 1622-1748) (26 kDa) (Fig. 3). Both direct Western (left lanes A-E) and immunoprecipitation/Western (right lanes A-E) analyses are shown in Fig.  3. We concluded that sorcin was associated with a domain of the ␣ 1C C terminus within amino acids 1622-1748 (Fig. 3). While we did not construct similar fusion proteins derived from the ␣ 1S C terminus, it is noteworthy that the cardiac and skeletal ␣ 1 isoforms are highly homologous (40) in the domain implicated in the ␣ 1 /sorcin interaction.
Increase in Sorcin Expression during Muscle Development-To further probe potential relationships between sorcin and L-type Ca 2ϩ channel ␣ 1 subunits, we examined the developmental expression of sorcin during physiological maturation of E-C coupling in developing muscle. It is well established that the expression of sarcolemmal and SR channel proteins is regulated in concordance with a program of muscle maturation (1, 45, 46, 48 -52). Sorcin abundance in postnatal rat heart substantially increased from day 1 to day 12 after birth (Fig.  4A). Both the 22-and 18-kDa species, depicted in Fig. 1B, increased in parallel. The ␣ 1C subunit was readily detectable as early as postnatal day 1 and also increased in abundance during this time period (Fig. 4A), in agreement with previous reports (45).
Sorcin was not present in neonatal rat cardiac myocytes 24 h after establishment in culture (Fig. 4B, lanes 1 and 2) but was detected on subsequent days as the cells commenced beating (Fig. 4B, lanes 3 and 5). In contrast to the intact heart (Fig. 1B,  lane 1, and Fig. 4A) and freshly isolated adult cardiac myocytes (data not shown), only the 22-kDa form was observed in neonatal myocytes. We next examined whether contractile arrest, produced by treating cells with verapamil (51), would affect sorcin expression. As shown in Fig. 4B, lanes 4 and 6, the level of sorcin was substantially reduced in verapamil-arrested cells.
The 22-kDa sorcin (the 18-kDa form was not detected) also accumulated in differentiating mouse skeletal muscle C2C12 cells (Fig. 5). RyR expression increased in parallel with sorcin, whereas myosin heavy chain was only detected after a 48-h period in mitogen-depleted medium (Fig. 5), consistent with previous reports (33,52). Full-length and C-terminal truncated forms of the ␣ 1S subunit were expressed in early stages of differentiation and were up-regulated in parallel during progression to fully differentiated myotubes (Fig. 5, lanes 3 and 4).

DISCUSSION
The voltage-sensing L-type Ca 2ϩ channels in T-tubules play a major role in E-C coupling in both heart and skeletal muscle (1,28). The close apposition of the L-type channels and RyRs in specific spatial relationships is required to accurately bridge FIG. 3. Sorcin antibody recovered HIS 6 -tagged ␣ 1C C terminus fragments. COS-1 cells transiently co-transfected with full-length sorcin and ␣ 1C fragments A, B, or C and HEK cells (expressing endogenous sorcin, Fig. 1B) transfected with fragments D and E were lysed in buffer A. Aliquots containing 30 g of the soluble materials were fractionated on 11% acrylamide gels along with proteins immunoprecipitated from 100 g of soluble cell proteins with antibody raised against peptides from the N or C terminus of sorcin (C terminus peptide antibody immunoprecipitation is shown here). After electrophoresis, proteins were transferred to nitrocellulose, and the membranes were incubated with anti-polyhistidine antibody as described above. Direct Western blot is shown in the first set of lanes the gap between SR and the sarcolemma for interchannel cross-talk (12)(13)(14). Indeed, a change in the relationship between SR Ca 2ϩ -release channels and sarcolemmal Ca 2ϩ channels may be a defect in cardiac hypertrophy that contributes to contractile failure (53). Functional coupling between L-type Ca 2ϩ channels and RyRs in neurons has also been demonstrated (54), and sensors for synaptic signaling just beneath sites of Ca 2ϩ entry in the hippocampus have been proposed (55). In normal cardiac myocytes, the opening of voltage-activated L-type channels results in the development of a high local intracellular [Ca 2ϩ ] in the microenvironment of the junctional space surrounding the sarcolemmal and SR membranes (56 -59). How the local L-type channel-generated signals are transmitted to and activate RyRs is incompletely understood as is the potential retrograde signal (60) by which RyRs may affect L-type channel function. Proteins or factors interposed between the L-type Ca 2ϩ channels and RyRs that might facilitate the relay of signals between them have been postulated (1,8,12). The results of the present study indicate that the Ca 2ϩ -binding protein, sorcin, associates with voltage-dependent Ca 2ϩ channels. These new data, together with previous studies, which established that sorcin binds to and modulates RyR, suggest a role for sorcin in interchannel communication and E-C coupling.
Sorcin binds Ca 2ϩ (K d ϳ1 M) and undergoes both a Ca 2ϩdependent decrease in intrinsic fluorescence and a Ca 2ϩ -mediated intracellular translocation from soluble to membrane components (41,61). The protein inhibits binding of ryanodine to cardiac RyR and is a modulatory ligand for RyR gating (22). In contrast to sorcin's effect on cardiac RyR, sorcin increases binding of ryanodine to skeletal muscle RyRs, perhaps reflecting structural differences between the cardiac and skeletal muscle RyR isoforms (22).
Our finding that sorcin associates with the cytoplasmically oriented C-terminal domains of L-type Ca 2ϩ channel ␣ 1 subunits together with our previous demonstration that the addition of sorcin to the cytoplasmic side of cardiac RyRs in planar bilayers decreases the single channel open probability (22) places sorcin within the sarcolemmal/SR junctional space, a strategic location for facilitation of interchannel communication. The ␣ 1C domain from amino acids 1622-1748 appeared to be sufficient for sorcin binding in vitro; however, the sorcin/␣ 1 interaction may be more complex in vivo. Shortened forms of ␣ 1 subunits, reportedly cleaved at C-terminal residues distal to the delineated sorcin binding domain (43, 44), were not immu-noprecipitated by sorcin antibody. This suggests that an intact C-terminal is required for sorcin interaction with the subunits in vivo and that the distal C-terminal regions may participate in stabilization of the sorcin/␣ 1 interaction.
Sorcin modulates RyR gating in planar bilayers, but whether it affects L-type Ca 2ϩ channel function is not known. We have no evidence that sorcin antibody recovered functional channels. The cytoplasmically oriented C terminus of ␣ 1C has been shown to be involved in Ca 2ϩ -sensitive channel inactivation, and a 142-amino acid segment from amino acids 1572-1717 in the ␣ 1C C terminus has been suggested to be required for inactivation (62). The putative sorcin-binding domain delineated in this report is within the margins of the potential inactivation domain and suggests that sorcin's involvement in this aspect of L-type Ca 2ϩ channel regulation should be investigated.
We found that sorcin expression increases in abundance in both developing heart and differentiating muscle cells. The contractile machinery in developing rodent heart is activated largely by Ca 2ϩ entering the cell through L-type channels (1,49). As the T-tubule/SR system gradually develops, contraction becomes more dependent on SR Ca 2ϩ release activated by Ca 2ϩ entry through sarcolemmal channels (49). The increase in sorcin expression in postnatal rat heart coincides with the developing Ca 2ϩ -induced Ca 2ϩ release mechanism, consistent with a putative role for sorcin in L-type channel/RyR interchannel communication. The increase in sorcin expression during C2C12 myotube formation, commensurate with L-type Ca 2ϩ channel and contractile apparatus development, is in accord with a role for sorcin in interchannel communication in skeletal muscle as well. A program of coordinate expression of myocyte proteins involved in contraction and Ca 2ϩ regulation has been suggested (51). Treatment of heart cells with verapamil initiates divergent expression of proteins that may be involved in this program along with contractile arrest (51). Whether sorcin, whose expression was substantially reduced in verapamil-arrested myocytes, is a component of that Ca 2ϩ regulation program remains to be determined.
Sorcin may undergo Ca 2ϩ -mediated dynamic changes in structure and subcellular localization, allowing it to either simultaneously or sequentially interact with L-type Ca 2ϩ channels and RyR in response to changing Ca 2ϩ levels. [Ca 2ϩ ] changes in the sarcolemmal/SR microenvironment may result in conformational changes and translocation of sorcin from cytoplasmic to membrane locations (41,61) in a manner analogous to the Ca 2ϩ -mediated conformational modification of recoverin (63). Ca 2ϩ -mediated dimerization of sorcin, shown to occur in vitro (60), may also occur in vivo. In addition, the recent demonstration of sorcin binding to annexin VII (64), thought to play a role in E-C coupling in skeletal muscle (65), and our demonstration that calmodulin exerts an additive effect on sorcin's inhibition of ryanodine binding to cardiac RyR (22) suggests that sorcin may function in conjunction with other channel accessory proteins. Whether these proteins might include L-type Ca 2ϩ channel subunits other than ␣ 1 is speculative; direct interaction between sorcin and other channel subunits was not in evidence.
Our studies demonstrating that sorcin interacts with both sarcolemmal L-type Ca 2ϩ channels and SR Ca 2ϩ release channels suggest a role for sorcin in interchannel communication. Sorcin may act as a sensor of the T-tubule junctional environment to ultimately participate in regulation of intracellular Ca 2ϩ mobilization and E-C coupling.   2 and 3, respectively), and fully differentiated myotubes 48 h after mitogen depletion (lane 4) were analyzed. RyR, MHC, and ␣ 1S were analyzed on 6% gels, and sorcin was analyzed on 11% gels. The legend to Fig. 1A