Caveolin-3 Regulates Protein Kinase A Modulation of the CaV3.2 (α1H) T-type Ca2+ Channels*

Voltage-gated T-type Ca2+ channel Cav3.2 (α1H) subunit, responsible for T-type Ca2+ current, is expressed in different tissues and participates in Ca2+ entry, hormonal secretion, pacemaker activity, and arrhythmia. The precise subcellular localization and regulation of Cav3.2 channels in native cells is unknown. Caveolae containing scaffolding protein caveolin-3 (Cav-3) localize many ion channels, signaling proteins and provide temporal and spatial regulation of intracellular Ca2+ in different cells. We examined the localization and regulation of the Cav3.2 channels in cardiomyocytes. Immunogold labeling and electron microscopy analysis demonstrated co-localization of the Cav3.2 channel and Cav-3 relative to caveolae in ventricular myocytes. Co-immunoprecipitation from neonatal ventricular myocytes or transiently transfected HEK293 cells demonstrated that Cav3.1 and Cav3.2 channels co-immunoprecipitate with Cav-3. GST pulldown analysis confirmed that the N terminus region of Cav-3 closely interacts with Cav3.2 channels. Whole cell patch clamp analysis demonstrated that co-expression of Cav-3 significantly decreased the peak Cav3.2 current density in HEK293 cells, whereas co-expression of Cav-3 did not alter peak Cav3.1 current density. In neonatal mouse ventricular myocytes, overexpression of Cav-3 inhibited the peak T-type calcium current (ICa,T) and adenovirus (AdCav3.2)-mediated increase in peak Cav3.2 current, but did not affect the L-type current. The protein kinase A-dependent stimulation of ICa,T by 8-Br-cAMP (membrane permeable cAMP analog) was abolished by siRNA directed against Cav-3. Our findings on functional modulation of the Cav3.2 channels by Cav-3 is important for understanding the compartmentalized regulation of Ca2+ signaling during normal and pathological processes.

T-type Ca 2ϩ channels (TTCC) 2 are low voltage-activated Ca 2ϩ channels, expressed in various tissues including brain and heart and contribute to a variety of physiological functions such as neuronal excitability, hormone secretion, muscle contraction, and pacemaker activity (1)(2)(3). Molecular cloning studies have identified three different TTCC isoforms, Ca V 3.1 (␣ 1G ), Ca V 3.2 (␣ 1H ), and Ca v 3.3 (␣ 1I ), which functionally can be distinguished by their electrophysiological properties (4 -7). Ca v 3.1 and Ca v 3.2 are the most commonly expressed subunits generating T-type Ca 2ϩ current (I Ca,T ) in brain and heart. Ca v 3.2 T-type Ca 2ϩ channels are involved in neurological disorders such as epilepsy and pain (8). In the heart, the I Ca,T participates in Ca 2ϩ entry and Ca 2ϩ -dependent hormonal secretion, pacemaker activity, and arrhythmia (9,10). The Ca v 3.1 and Ca v 3.2 isoforms are normally expressed in embryonic hearts (11, 12), but postnatal expression of these isoforms diminishes with almost no expression in normal adult ventricular myocytes. However, the TTCCs are re-expressed during conditions of cardiac hypertrophy and heart failure and are reported to be associated with decreased cardiac function (13)(14)(15). Ca 2ϩ influx through the re-expressed voltage-gated Ca v 3.2 (␣ 1H ) TTCC is indicated to be responsible for inducing pathological cardiac hypertrophy in a pressure overload model (16).
Although we are beginning to understand the physiological role of Ca v 3.2 channels in Ca 2ϩ cycling, the precise regulation of these channels has not been defined. To clearly understand the regulation and physiological function of these proteins in normal and diseased stages it is important to define the exact subcellular localization of this protein. Caveolae are specialized membrane microdomains enriched in cholesterol and sphingolipids, contain the signature scaffolding protein caveolins, and provide temporal and spatial regulation of intracellular Ca 2ϩ in many cell types (17)(18)(19). Three different caveolin isoforms are identified, of which Cav-1 and Cav-2 are ubiquitously expressed (20,21), whereas Cav-3 is widely expressed in muscle (cardiac, skeletal, and smooth muscle) cells (22) and neuronal tissue (23). A number of ion channels and transporters have been localized to caveolae and associate with Cav-3, including L-type Ca 2ϩ channels (Ca v 1.2) (24), the Na ϩ channel (Na v 1.5) (25), pacemaker channels (HCN4) (26), Na ϩ / Ca 2ϩ exchanger (27), and others (18). Closely associated with these channels are specific macromolecular signaling complexes containing a variety of regulatory proteins including the G protein-coupled receptors and kinases such as PKC and PKA that provide highly localized regulation of the channels (18). The present study therefore was designed to determine the precise subcellular localization and regulation of the Ca v 3.2 subunit of the T-type Ca 2ϩ channels. Our results dem-* This work was supported by Scientist Development Grant 0730010N from the American Heart Association (to R. C. B.). □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Table S1 and Figs. S1 and S2. 1  onstrate that the Ca v 3.2 channel protein is localized to caveolar microdomains in the ventricular myocytes and Cav-3 interacts with Ca v 3.2 channel and regulates its function.

EXPERIMENTAL PROCEDURES
Materials-All chemicals and reagents were procured from Sigma unless otherwise stated. Mouse monoclonal antibodies to Cav-3, cardiac actin, and biotin were obtained from BD Biosciences; rabbit polyclonal antibody to Cav-3 and control IgG antibodies were from Santa Cruz Biotechnology. Rabbit polyclonal antibodies to Ca v 3.1 and Ca v 3.2 channel were from Alamone Labs, Jerusalem, Israel; rabbit polyclonal antibody to Ca v 3.1 was from Millipore. Tetradotoxin was obtained from Calbiochem. PKA inhibitor peptide 14-22, myristoylated (amino acid sequence: Myr-Gly-Arg-Thr-Gly-Arg-Arg-Asn-Ala-Ile-NH 2 ), was from Sigma.
Cell Culture and Transfection-Wild-type human Ca v 3.1, Ca v 3.2 channels, or Cav-3 protein were expressed in human embryonic kidney 293 (HEK293) cells as previously described (28). The HEK293 cells were maintained in modified DMEM (Invitrogen) and cultured at 37°C in 5% CO 2 . HEK293 cells were transfected using Lipofectamine 2000 (Invitrogen) as described previously (29). 48 h after transient transfection the cells were used for experiments. Neonatal or adult mouse ventricular myocytes were enzymatically isolated as previously described (24,30). The neonatal myocytes were transfected by the electroporation method by a Nucleofector device (AMAXA) as described earlier (24)  Immunoprecipitation-Isolated neonatal mouse ventricular cardiomyocytes from four 100-mm diameter dishes or adult mouse myocytes (ϳ2 mg of protein) were rinsed with ice-cold TBS (pH 7.4) and lysed in ice-cold solubilization buffer containing 25 mM Tris-HCl (pH 7.4), 150 mM NaCl, 60 mM noctyl D-glucoside, 1% Triton X-100, 2 mM phenylmethylsulfonyl fluoride, 5 g/ml of aprotinin, 5 g/ml of benzamidin, 5 g/ml of leupeptin, and 5 M pepstatin A. The lysate was centrifuged at 10,000 ϫ g for 10 min to remove insoluble debris, and the soluble supernatant was pre-cleared using protein G-Dynabeads (Invitrogen), followed by incubation for 4 h at 4°C with anti-Cav-3 (5 g) or anti-Ca v 3.1 or Ca v 3.2 (5 g) antibodies or control IgG in a total of 450 l. 50 l of 1:1 slurry of protein G-Dynabeads was added to the sample and further incubated for 1 h at 4°C. Beads were washed four times with solubilization buffer on a magnetic stand, and bound proteins were eluted with SDS-PAGE sample buffer by boiling for 5 min. Immune complexes were analyzed by SDS-PAGE (4 -15% gradient gels, Bio-Rad) and Western blot by probing with antibodies to Ca v 3.1, Ca v 3.2, and Cav-3.
Cav-3 Subdomain Plasmids, GST-Cav-3 Fusion Plasmids, and Pulldown Assays-Full-length and different subdomains of Cav-3 were generated by standard polymerase chain reaction strategies utilizing the human Cav-3 as template. cDNA clones were constructed of the Cav-3 N terminus domain (amino acids 1-54, Cav-3 1-54 ), scaffolding domain (amino acids 55-74, Cav-3 55-74 ), membrane domain (amino acids 75-107, Cav-3 75-107 ), and C-terminal domain (amino acids 108 -151, Cav-3 108 -151 ). The amplified product was cloned into pcDNA3.1 with a TOPO directional cloning system reagent kit from Invitrogen according to the manufacturer's instructions and transformed into Escherichia coli BL21 strain. The plasmid DNA was purified using a Plasmid Maxi kit (Qiagen). DNA sequencing was used to confirm the correctness of the domains. For generations of Cav-3-GST fusion proteins the PCR products were subcloned into Pme1 and Sgf1 sites of pFN2A (GST) Flexi vector (Promega) by restriction digestion. The resulting wild-type and truncated Cav-3 constructs were confirmed by sequencing and transformed into E. coli BL21(DE3) strain. The GST fusion protein was purified from E. coli following induction by 0.1 mM isopropyl 1-thio-␤-Dgalactopyranoside and linked to glutathione-agarose beads as described earlier (31,32). For pulldown assays, Ca v 3.2 channel protein was expressed in HEK293 cells, 48 h after transfection the cells were lysed and solubilized on ice using ice-cold solubilization buffer (10 mM Tris, 150 mM NaCl, 5 mM EDTA, 1% Nonidet P40, 1% Triton X-100, 60 mM N-octylglucoside, 2 M phenylmethylsulfonyl fluoride, 1 g/ml of aprotinin, 2 g/ml of benzamidin, 1 g/ml of leupeptin, and 1 g/ml of pepstatin A), the lysate was then centrifuged at 10,000 ϫ g for 10 min. The soluble supernatant was collected and then allowed (2 mg of protein) to bind to the MagneGST-agarose beads (Promega) linked with different Cav-3 GST fusion protein constructs. After a 4-h incubation at 4°C, the sample was washed in 20 mM Tris, 150 mM NaCl, 5 mM EDTA, 0.1% Triton X-100, 0.5% sodium deoxycholate. The eluted sample was then separated and analyzed by SDS-PAGE and Western blot, respectively, by probing with anti-Ca v 3.2 (1:100, rabbit polyclonal, Alomone Labs) and anti-GST antibody (1:500, mouse monoclonal; BD Biosciences).
SDS-PAGE and Western Blot Analysis-Myocyte lysates, or the immune complexes were separated by SDS-PAGE (4 -15% gradient acrylamide gel) and transferred to polyvinylidene difluoride membranes. Nonspecific binding sites were blocked by 5% (w/v) dried skim milk in TBS detergent (0.1% Tween 20). Membranes were then probed with specific primary antibodies (Ca v 3.1 or Ca v 3.2, 1:100; Cav-3, 1:1000) followed by a washing step (four times for 10 min each) with TBS detergent. The membranes were then incubated with either goat anti-mouse Ig conjugated to horseradish peroxidase (1:10,000) or goat anti-rabbit IgG to horseradish peroxidase (1:10,000; Bio-Rad) for 1 h and washed (six times for 10 min) with TBS detergent. Immunoreactivity was visualized using peroxidase-based chemiluminescent detection by ECL (Amersham Biosciences).
Biotinylation of Cell Surface Protein and Immunoblotting-Surface biotinylation studies were performed as described earlier (33) in HEK293 cells stably expressing Ca v 3.2 channel protein. Briefly, HEK293 cells stably expressing Ca v 3.2 channels were grown in 100-mm culture dishes and transfected with Cav-3 or GFP. After 48 h, cells were washed three times with PBS and incubated with sulfo-NHS-(LC)-biotin (0.5 mg/ ml; Pierce) in PBS at 4°C for 1 h. Cells were then washed five times with ice-cold PBS to remove residual biotin reagent and solubilized in lysis buffer with protease inhibitors. Lysate pro-teins were quantified with a bicinchoninic acid assay (Pierce). Proteins (0.5 mg/reaction) were mixed with anti-Cav3.2 antibody (5 g) immunoprecipitated using the protein G-Dynabeads as described above. The immunoprecipitated sample was analyzed by Western blotting by probing with anti-biotin antibody (mouse monoclonal, 1:5,000, BD Biosciences). Immunoblot membrane was then stripped and re-probed with anti-Ca v 3.2 antibody (1:100) to detect the Ca v 3.2 channel protein signal.
Adenoviral Infection of Ca v 3.2 Channel-Adenoviruses expressing either Ca v 3.2 channel protein (AdCav3.2) or green fluorescent protein (AdGFP) were amplified as described by Brueggemann et al. (34). Isolated neonatal mouse ventricular myocytes were plated on laminin-coated coverslips in a 35-mm Petri dish at 1 ϫ 10 5 cells and infected with AdCa v 3.2 or AdGFP at a multiplicity of infection of 5-10. More than 90% cells infected with Ad-GFP expressed GFP 48 h after infection. Negligible cell death was observed in cultures infected with adenovirus. Experiments were performed 72-96 h after adenoviral infection.
siRNA-mediated Cav-3 Knockdown-siRNA-mediated knockdown of Cav-3 in isolated neonatal mouse cardiomyocytes was archived by transfecting three pairs of pre-validated Cav-3 specific siRNAs as described earlier (24). Cav-3 sequences targeting different regions for mouse Cav-3 (GenBank TM accession number NM_007617; sense, GCUUC-GACGGUGUAUGGAAtt, and antisense, UUCCAUACACC-GUCGAAGCtg; sense, GGUUCCUCUCAAUUCCACCtt, and antisense GGUGGAAUUGAGAGGAACCtc; sense, CGUUCACCGUCUCCAAGUtt, and antisense, UACUUGG-AGACGGUGAACGtg) were used for Cav-3 knockdown. Nonspecific siRNA oligos to mouse GAPDH (Applied Biosciences) was used as controls. Briefly, freshly isolated neonatal mouse myocytes were transfected by the electroporation method with the desired oligos (10 nM) and 0.5 g of cDNA for GFP as previously described (24). Forty-eight to 72 h after transfection myocytes was evaluated for GFP expression and immunostained for Cav-3 (rabbit polyclonal antibody; Santa Cruz Biotechnology 1:500 dilution) and cardiac-specific actin (mouse monoclonal antibody; BD Biosciences; 1:1000 dilution) expression to determine Cav-3 knockdown in the myocytes. The number of GFP-expressing cells divided by the total number of cells in the field of view determined transfection efficiency (data not shown) and was estimated to be between 50 and 60%. Neonatal myocytes expressing GFP fluorescence were used for patch clamp experiments.
Transthoracic Aortic Constriction (TAC)-induced Pressure Overload and Echocardiography Analysis-TAC was performed on 8 -12-week-old C57BL/6 male mice as described previously (35). Sham-operated mice were subjected to identical interventions except for the constriction of the aorta. Noninvasive transthoracic echocardiography (35) was performed before and after 4 weeks of TAC surgery in mice, and LV wall thickness, chamber dimension, and contractility were evaluated. The pressure gradients across the aortic constriction were measured to ensure similar pressure overload in the TAC mice.
Immunogold Labeling and Electron Microscopy-Double immunogold labeling co-localization studies and electron microscopy (EM) was performed on isolated and cultured neonatal myocytes as described by Balijepalli et al. (24) using anti-Ca v 3.2 (rabbit polyclonal) and anti-Cav-3 (mouse monoclonal) antibodies and a double silver enhancement technique.
Electrophysiology-Electrophysiological recordings were made from HEK293 cells transiently expressing human cardiac T-type calcium channels (I Cav3.1 and I Cav3.2 ) and Cav-3 as described previously (28). T-and L-type calcium currents were recorded from cultured mouse neonatal cardiomyocytes (3-5 days in vitro). Calcium currents were measured from isolated cells with bright GFP fluorescence with either ruptured or perforated patch configuration. The patch pipettes were pulled from thin walled borosilicate glass capillaries (World Precision Instruments, Inc., Sarasota FL) and filled with an intracellular solution containing the following (in mM) for both HEK293 cells and neonatal cardiomyocytes: 114 CsCl, 10 EGTA, 10 HEPES, 5 MgATP (pH 7.2) adjusted with CsOH. Extracellular buffer for HEK293 cells contained the following (in mM): 128 CsCl, 2 CaCl 2 , 1.5 MgCl 2 , 10 HEPES, 25 D-glucose (pH 7.4), adjusted with CsOH. For neonatal cardiomyocytes (in mM): 145 tetramethylammonium chloride, 5 CaCl 2 , 1 MgCl 2 , 5 CsCl, 1,4-aminopyridine, 0.01 tetrodotoxin, 10 HEPES, 5 D-glucose (pH 7.4), adjusted with tetramethylammonium OH. The current-voltage relationship was evoked by step depolarization from Ϫ90 to ϩ60 mV with 10-mV step voltage for 200 ms. For perforated patch clamp measurements the pipette solution contained (in mM): 140 Cs-glutamate, 10 HEPES, 0.5 CaCl 2 , and 400 g/ml of amphotericin B (pH 7.2). The external buffer consists of (in mM): 145 tetramethylammonium chloride, 5 CaCl 2 , 1 MgCl 2 , 10 HEPES, and 5 D-glucose, 5 CsCl, 1 aminopyridine, and tetrodotoxin (1 M) bath applied. The peak I Ca,T were measured at a holding potential at Ϫ90 mV, a test pulse of Ϫ30 mV for 200 ms was applied at 15-s time intervals. After the initial basal I Ca,T measurement, 3-5-min myocytes were perfused with 100 M 8-Br-cAMP (in the bath solution) to activate the protein kinase A (PKA)-dependent stimulation of the I Ca,T . Following stimulation with 8-Br-cAMP, 10 M nitrendipine and 0.5 mM NiCl were included in the bath solution to block I Ca,L and I Ca,T , respectively. In some experiments neonatal myocytes were pretreated with 10 M myristoylated PKA inhibitor peptide (overnight incubation) to inhibit the PKA activity. The data were collected from a minimum of three different transfections. All experiments were carried out at room temperature with pipette resistance of 1.5-2.5 M⍀. The data were acquired using Axopatch 200B amplifier (Axon Instruments, Foster City, CA) with pCLAMP 10.2. The data were filtered at 5 kHz and digitized at 50 kHz. The current traces were corrected for linear capacitance and leak using -P/4 subtraction protocol.
Data Analysis-Data were analyzed using the Clampfit and Origin 7.5 software programs. The curves for steady state activation and inactivation were fitted using Boltzmann equation. The current-voltage curves were fitted using the Boltzmann function: All the results are presented as the mean Ϯ S.E. and the significance of the observed difference was evaluated by unpaired Student's t test. p value of Ͻ0.05 was considered statistically significant.

T-type Ca 2ϩ Channel Isoforms Associate with Cav-3 in Ventricular Myocytes and HEK293 Cells-T-type
Ca 2ϩ channel isoforms Ca v 3.1 and Ca v 3.2 as well as Cav-3 are known to be expressed in neonatal cardiomyocytes. To determine whether Cav-3, Ca v 3.1, and Ca v 3.2 channels are associated in ventricular myocytes, we performed co-immunoprecipitation experiments. Lysates from isolated neonatal mouse ventricular myocytes were solubilized in Triton X-100 and N-octyl Dglucoside-containing buffer and subjected to immunoprecipitation with anti-Cav-3 or control mouse IgG. Western blot analysis shows that Ca v 3.1 and Ca v 3.2 protein bands were detected from the mouse neonatal myocytes lysate (Fig. 1, A  and B). Both TTCC isoforms co-immunoprecipitated with Cav-3 from neonatal mouse ventricular myocytes lysates. In a converse experiment we used either anti-Ca v 3.1 or Ca v 3.2 or a control rabbit IgG antibody for immunoprecipitation from neonatal ventricular myocyte lysates and found that Cav-3 co-immunoprecipitated with either of the TTCC isoform antibodies but not with control IgG. These results suggested that the Ca v 3.1 and Ca v 3.2 channels associate with Cav-3 in ventricular myocytes. To further confirm these results in a heterologous expression system we transiently expressed Cav-3 and Ca v 3.1 or Ca v 3.2 channel proteins in HEK293 cells that do not express either of these proteins endogenously. 48 h post-transfection cell lysates were subjected to immunoprecipitation with anti-Cav-3, or control mouse IgG. As demonstrated in Fig. 1C, the T-type Ca 2ϩ channel isoform Ca v 3.1 and Ca v 3.2 proteins co-immunoprecipitated with anti-Cav-3 antibody. These results confirmed that the T-type Ca 2ϩ channels associate with Cav-3. Non-transfected HEK293 cells did not show protein bands for T-type Ca 2ϩ channel isoforms or Cav-3.
Ca v 3.2 Channel Protein Co-localized with Cav-3 Relative to Caveolae in Ventricular Myocytes-To determine the precise localization of Ca v 3.2 channels we used the immunogold labeling technique combined with electron microscopy. To determine localization of the Ca v 3.2 channels, we initially used isolated mouse neonatal ventricular myocytes that were fixed and immunogold co-labeled with anti-Cav-3 and anti-Ca v 3.2 antibody. Transmission electron micrographs revealed two distinct populations of different sized gold particles (Fig. 2) present on surface membrane invaginations typical of caveolae. The small gold particles identify anti-Cav-3 labeling (arrows) and large gold particles (as a result of double silver enhancement, arrowhead) identify anti-Ca v 3.2 channel protein labeling suggesting the co-localization of the Ca v 3.2 with Cav-3 relative to caveolae (Fig. 2, a and c). Immunogold labeling showed labeling for Cav-3 but did not show Ca v 3.2 channel staining in the ventricular myocytes of normal mouse adult heart tissue as, cardiac T-type Ca 2ϩ channel isoforms are normally expressed during development but not expressed in the adult hearts (Fig. 2, e and g). However, these channels are known to re-express during pressure overloadinduced cardiac hypertrophy and heart failure (11-15). Adult mice were subjected to the TAC procedure to generate pressure overload-induced cardiac hypertrophy. Echocardiography analysis (supplemental Table S1) confirmed induction of cardiac hypertrophy. Adult hearts were fixed by perfusion fixation and co-immunogold labeling was performed. As shown in the transmission electron micrograph (Fig. 2, b and d), Ca v 3.2 channel protein (large gold particle) and Cav-3 (small gold particle) were co-localized relative to caveolae. Also, the gold particle distribution was restricted to sarcolemma regions in the ventricular myocytes suggesting specific caveolar localization of the protein. In control samples (hypertrophic adult mouse heart) from which the primary antibodies had been omitted, did not show gold particle staining.
Co-expression of Cav-3 Inhibits I Cav3.2 but Not I Cav3.1 in HEK293 Cells-To determine the functional impact of Cav-3 association on T-type Ca 2ϩ channel currents we used a heterologous system of HEK293 cells, which provides a convenient expression system to study specific transfected T-type Ca 2ϩ channels and Cav-3, given the lack of endogenous expression of these proteins. HEK293 cells have been used by us and others to study the heterologous expression of T-type Ca 2ϩ channels and a wide array of other ion channels (28, 29, 36 -39). Human TTCC isoforms Ca v 3.1 and Ca v 3.2 were separately co-expressed with Cav-3 in HEK293 cells and the respective currents were measured using the whole cell patch clamp technique. Currents were measured from a holding potential of Ϫ90 mV and stepped to 60 mV in 10-mV increments for 200 ms (as shown in the Fig. 3, B and inset D). Fig.  3, B and D, shows the current-voltage relationship of I Cav3.1 and I Cav3.2 , respectively, with and without co-expression of Cav-3. Co-expression of Cav-3 did not effect the I Cav3.1 (-32.8 Ϯ 5 pA/pF, n ϭ 11) compared with expression of Ca v 3.1 with GFP (Ϫ33.48 Ϯ 4 pA/pF, n ϭ 11). On the other hand co-expression of Cav-3 with Ca v 3.2 significantly reduced I Cav3.2 (Ϫ11.48 Ϯ 3 pA/pF, n ϭ 11) compared with Ca v 3.2 ϩ GFP (Ϫ31 Ϯ 4 pA/pF, n ϭ 11). We also investigated the effect of co-expression of Cav-3 on the biophysical properties of the I Cav3.2 in HEK293 cells. Co-expression of Cav-3 did not affect Ca v 3.2 channel activation and inactivation properties. These data are presented in supplemental Fig. S1.

Effect of Co-expression of Cav-3 on Plasma Membrane Expression of Ca v 3.2 Channel
Protein-To understand the mechanism of Cav-3 inhibition of the I Cav3.2 we investigated if Cav-3 co-expression alters trafficking and reduces the surface expression of the Ca v 3.2 channels. One way to determine the number of channels expressed on the plasma membrane is to measure the gating currents, however, given the small I Cav3.2 amplitudes (300 -500 pA) we could not measure the gating currents to estimate the number of plasma membrane-expressed Ca v 3.2 channels with Cav-3 co-expression. We used an alternative approach of cell surface biotinylation of the Ca v 3.2 channels. HEK293 cells were transfected with cDNAs of Ca v 3.2 ϩ GFP or Ca v 3.2 ϩ Cav-3, or GFP alone. Cell lysates were precipitated with neutravidin beads and samples were analyzed by Western blots by probing with anti-Ca v 3.2 or anti-␤-actin antibody. As shown in a representative Western blot (Fig. 3E), no difference was noticed in the biotinylated Ca v 3.2 protein signal intensity when Ca v 3.2 was expressed alone (Ca v 3.2 ϩ GFP) or co-expressed with Cav-3 (Ca v 3.2 ϩ Cav-3) suggesting that co-expression of Cav-3 did not affect the surface membrane expression of Ca v 3.2 channel. These experiments were repeated three times and the signal for the biotinylated Ca v 3.2 band was semi-quantitatively estimated by densitometry. We found that the mean signal density for the biotinylated Ca v 3.2 band was not different between groups (data not shown). The Ca v 3.2 protein signal was absent in the sample that was transfected with GFP alone. We did not detect a signal for ␤-actin, demonstrating the biotinylation of only surface membrane proteins. The signal for Cav-3 as also absent in the neutravidin pulldown as Cav-3 is localized to the inner leaflet of the plasma membrane bilayer and not biotinylated. A portion of the lysate sample from 3 groups was also analyzed by probing with anti-Ca v 3.2 and anti-␤-actin (Fig. 3, panel F) for loading control. Again similar signal intensity for the Ca v 3.2 channel protein was detected with either Ca v 3.2 alone or with co-expression of Cav-3 and a similar ␤-actin signal was detected between the three groups of cells demonstrating identical sample loading.
Ca v 3.2 Interaction Site on Cav-3-We next examined the site of interaction for Cav-3 with the Ca v 3.2 channel protein.
We created five different GST fusion constructs of Cav-3 based on known domains (illustrated in Fig. 4A and see Ref. 18): 1) full-length (Cav-3FL), 2) Cav-3Nterm (Cav-3 1-54 ), 3) Cav-3Scaf (Cav-3 55-73 ), 4) Cav-3Mem (Cav-3 74 -106 ), and 5) Cav-3Cterm domain (Cav-3 107-151 ) using a bacterial system as described under "Experimental Procedures." Ca v 3.2 channel protein was expressed in HEK293 cells, and the lysates were incubated with GST alone or different GST-Cav-3 fusion proteins linked to glutathione beads. Following that procedure, a pulldown assay was performed and samples were analyzed by Western blotting. As demonstrated in Fig. 4B, the Ca v 3.2 channel was found to interact and associate with the full-length GST fusion protein of Cav-3 (GSTCav-3FL) and N terminus GST fusion protein of Cav-3 (GST-Cav3NT; amino acids 1-54) but not associate with GST-Cav-3Scaf  Co-expression of Cav-3 N Terminus Domain Inhibits I Cav3.2 in HEK293 Cells-Our GST pulldown assay demonstrates that the N terminus of Cav-3 interacts with the Ca v 3.2 channel. To determine the functional impact of this interaction on Ca v 3.2 channels, we generated cDNA constructs of the N terminus region (Cav-3  ) and other regions of Cav-3 as described under "Experimental Procedures." These cDNAs along with the Ca v 3.2 cDNA were then transiently transfected into HEK293 cells. I Cav3.2 was then measured by whole cell patch clamp analysis. As demonstrated in Fig. 4B, co-expression of Cav-3Nterm significantly reduced the I Cav3.2 (11.7 Ϯ 1 pA/pF, n ϭ 7) compared with Ca v 3.2 alone (24.1 Ϯ 2 pA/pF, n ϭ 7). On the other hand co-expression of Cav-3Cterm, Cav-3Scaf, or Cav-3Memb with Ca v 3.2 channels did not alter the I Cav3.2 . These data further demonstrate that interaction of the N terminus region of Cav-3 with the Ca v 3.2 channel protein modulates the channel function.
Overexpression of Cav-3 Inhibits I Ca,T in Mouse Neonatal Ventricular Myocyte-Our data from the heterologous expression system of HEK293 cells demonstrated that co-expression of Cav-3 specifically inhibited the I Cav3.2 . Next we investigated if Cav-3 will have a similar effect on the native T-type Ca 2ϩ current (I Ca,T ) using isolated neonatal cardiomyocytes where Cav-3 and the Ca v 3.2 channel isoform are endogenously expressed. Neonatal ventricular myocytes express both isoforms of TTCC as well as Ca v 1.2, L-type Ca 2ϩ chan-

Caveolin-3 Regulates Ca v 3.2 Channels
nels. T-type Ca 2ϩ channels can be distinguished from L-type Ca 2ϩ channels on the basis of their conductance and gating properties (40). The TTCCs are known to activate at significantly more negative membrane voltage potentials with a threshold for activation of I CaT about Ϫ60 mV and peak I CaT between Ϫ30 mV at physiological Ca 2ϩ concentrations (1,9,16), whereas the Ca v 1.2 channels begin to activate at about Ϫ30 mV and the peak I Ca,L is elicited at more positive potentials (10 mV) (24). To isolate the pure I Ca,T from high voltageactivated L-type currents we used a dual pulse protocol as described earlier (16). Using a whole cell patch clamp technique (Fig. 5) the total I Ca (I Ca,Tot ) was recorded from myocytes using a holding potential of Ϫ90 mV and pulsed to 60 mV in 10-mV steps for 200 ms, followed by a brief holding potential of Ϫ50 mV and further pulsed to 70 mV in a 10-mV steps for 200 ms to record the I Ca,L . To obtain the I Ca,T , traces of I Ca,L (holding potential Ϫ50 mV; Fig. 5, B and C) were subtracted from the corresponding trace of I Ca,Tot with a holding potential of Ϫ90 mV. Thus we found that peak I Ca,T was at Ϫ30 mV in these cells. Cav-3 or GFP alone were overexpressed in the isolated mouse neonatal ventricular myocytes by electroporation as described earlier (24). Using the same dual pulse protocol we measured the I Ca,Tot , I Ca,L , and then obtained I Ca,T by subtraction as demonstrated in Fig. 5, and found that the average peak I Ca,T density for nontransfected cells (Ϫ5 Ϯ 0.7 pA/pF, n ϭ 11) or GFP-transfected cells (Ϫ4.7 Ϯ 1 pA/pF, n ϭ 11) was similar (Fig. 5E). On the other hand overexpression of Cav-3 significantly reduced the I Ca,T 42% (Ϫ2.1 Ϯ 1 pA/pF, n ϭ 8) compared with GFP-transfected control cells (Fig. 5, D and E). Overexpression of Cav-3 had no impact on the average peak I Ca,L density (Fig. 5F).
Neonatal ventricular myocytes express Ca v 3.1 and Ca v 3.2 and kinetic properties of I Cav3.1 and I Cav3.2 closely resemble native I CaT . Thus it is difficult to differentiate between the two different current components in a native cardiomyocyte system based on their biophysical properties. Studies have shown that I Cav3.1 and I Cav3.2 can be differentiated by their sensitivity to block by Ni 2ϩ . Recombinant Ca v 3.2 channel is shown to be blocked by low concentrations of Ni 2ϩ (IC 50 ϭ 12 M), whereas the Ca v 3.1 channel is blocked by higher concentrations of Ni 2ϩ (IC 50 ϭ 250 M) (41). However, in our conditions we did not find it practical to use Ni 2ϩ to differentiate between the components of TTCCs because a higher concentration of Ni 2ϩ is needed to block I Cav3.1 . So we used an adenoviral overexpression model of I Cav3.2 and studied the effect of Cav-3 on the I Ca,T . Isolated neonatal myocytes were transfected with either Cav-3 or GFP alone as above and then infected with AdCa v 3.2 or AdGFP. Overexpression of Ca v 3.2 proteins was confirmed by Western blot analysis (data not shown) and by measuring I Cav3.2 . AdCa v 3.2-treated cells showed significantly increased I Cav3.2 (-32.4 Ϯ 7 pA/pF) compared with AdGFP-treated control cells (Ϫ4.9 Ϯ 1 pA/pF) or non-treated control cells (Fig. 6A). Co-expression of Cav-3 significantly reduced (89%) the AdCa v 3.2-induced I Cav3.2 (Ϫ4.37 Ϯ 1pA/pF), suggesting that Cav-3 inhibits and regulates that I Cav3.2 in neonatal cardiomyocytes. AdCa v 3.2 or AdGFP treatment did not alter the average I Ca,L density in the neonatal ventricular myocytes (Fig. 6B).
To determine the involvement of Cav-3 on the modulation of Ca v 3.2 channels in neonatal cardiomyocytes, we investigated the impact of the specific inhibition of Cav-3 expression using siRNA-mediated gene silencing as described earlier (24). Three different siRNA oligos specific to mouse Cav-3 or control siRNA to GAPDH were co-transfected with GFP into isolated mouse neonatal cardiomyocytes. 72 h after transfection knockdown of Cav-3 was established by Western blot analysis and immunofluorescence imaging for Cav-3, which confirmed that Cav-3 siRNA-transfected cells (GFP-expressing) exhibited near complete knockdown of Cav-3 (see supplemental Fig. S2). Whole cell electrophysiology was performed on GFP expressing cells (green fluorescence) to measure I Ca,T . siRNA-mediated knockdown of Cav-3 did not

Caveolin-3 Regulates Ca v 3.2 Channels
alter the I Ca,T densities (Ϫ4.5 Ϯ 0.8 pA/pF n ϭ 22), and was similar to control (GAPDH) siRNA-transfected myocytes (Ϫ5.4 Ϯ 1 pA/pF, n ϭ 21) or nontransfected neonatal cardiomyocytes (5 Ϯ 0.7 pA/pF, n ϭ 9) (Fig. 6C). siRNAmediated knockdown of Cav-3 also had no effect on the average I Ca,L density as shown in Fig. 6D. A similar effect of siRNA-mediated knockdown on the I Ca,L was observed in an earlier study from our group when siRNA-mediated knockdown of Cav-3 did not alter the mean I Ca,L current densities compared with control siRNA-treated neonatal ventricular myocytes (24).

siRNA-mediated Knockdown of Cav-3 Eliminates Protein Kinase A Regulation of the I Ca,T in Mouse Neonatal Myocytes-
Overexpression of Cav-3 inhibited the I Cav3.2 but the siRNA-mediated inhibition of Cav-3 expression did not affect the I Ca,T density in neonatal myocytes. We reasoned that Cav-3 may play an important role in regulation of Ca v 3.2 channels and Cav-3 knockdown may alter regulation of the I Ca,T in neonatal mouse cardiomyocytes. The I Ca,T and I Cav3.2 are reported to be augmented by cAMP-dependent protein kinase (PKA) (42)(43)(44). We investigated the PKA regulation of the I Ca,T in neonatal myocytes using 8-bromo-cAMP (membrane permeable cAMP-a known activator of PKA) by perforated clamp analysis. Initially a dose-dependent stimulation of I Ca,T was performed with 8-Br-cAMP and we found a maximal stimulation of I Ca,T at 100 M concentrations (data not shown). In the myocytes that were transfected with control siRNA, 8-Br-cAMP resulted in a PKA-mediated increased stimulation of the peak I Ca,T to 126 Ϯ 6% (Fig. 7, A and D). 8-Br-cAMP is also known to stimulate I Ca,L through activation of the high voltage Ca V 1.2 channels in ventricular myocytes at a much higher concentration (1 (45) and 2 mM (46)). To eliminate the possible involvement of I Ca,L in the 8-Br-cAMP-induced I Ca response, the cells were perfused with 10 M nitrendipine, a specific blocker of the I Ca,L . Perfusion with 10 M nitrendipine did not affect the peak I Ca,T . However, FIGURE 5. Overexpression of Cav-3 inhibits I Ca,T in isolated mouse neonatal ventricular myocytes. Whole cell patch clamp analysis was performed in neonatal myocytes transfected with Cav-3 or GFP (control) using a protocol as described under "Results" and shown in the inset on top. A and C show representative current traces of GFP and Cav-3-transfected cells, respectively, using a holding potential of Ϫ90 mV (left) and Ϫ50 mV (middle) to a step depolarization to the respective indicated command potential. The traces on the right are obtained by subtracting the middle trace from the left current trace and represent the T-type Ca 2ϩ current. Panels C and D, representative current density plot generated by a series of step depolarization potentials of 10 mV at different holding potentials, i.e. Ϫ90 mV (I Ca,Total , f) and Ϫ50 mV (I Ca,L , F) the difference is the T-type current (I Ca,T , OE) in control (GFP) and Cav-3 overexpression cells, respectively. E, peak current density of I Ca,T with or without overexpression of Cav-3. F, peak I Ca,L densities with and without overexpression of Cav-3. The data represent mean Ϯ S.E. (n ϭ 8 -11 cells, *, p Ͼ 0.05) from 3 different transfections.
upon perfusion with 0.5 mM NiCl, the I Ca,T was completely abolished, suggesting that 8-Br-cAMP stimulation of the I Ca is specifically through activation of the T-type Ca 2ϩ channel. To confirm that 8-Br-cAMP stimulation of the I Ca,T is dependent on PKA, we pretreated neonatal myocytes (8 -12 h incubation) with the myristoylated PKA inhibitor peptide fragment 14-22 (10 M). Pretreatment of cells with the specific PKA inhibitor peptide failed to evoke 8-Br-cAMP stimulation of the I Ca,T in the control siRNA-transfected myocytes, suggesting a PKA-dependent augmentation of I Ca,T by 8-Br-cAMP (Fig. 7, C and D). In the next set of experiments, we used myocytes that were transfected with siRNA to Cav-3. In contrast with the control siRNA-treated cells the siRNA-mediated Cav-3 knockdown almost completely abolished 8-Br-cAMP stimulation of the I Ca,T in the myocytes (Fig. 7, B and D). These findings confirm that Cav-3 is required for PKA-mediated regulation of the Ca v 3.2 T-type Ca 2ϩ channels in the mouse neonatal ventricular myocytes.

DISCUSSION
In the present study we describe regulation of the Ca v 3.2 subunit of T-type Ca 2ϩ channels by Cav-3 in the mouse ventricular myocytes. Double immunogold labeling and electron microscopy imaging data clearly demonstrate co-localization of the Ca v 3.2 channel with Cav-3 relative to caveolae in the mouse ventricular myocytes. Co-immunoprecipitation analysis suggested an association of Ca v 3.1 and Ca v 3.2 channel isoforms with Cav-3, the GST pulldown assay confirmed a close interaction between the Ca v 3.2 channel protein and Cav-3. Functional analysis using whole cell patch clamp in the heterologous expression system of HEK293 cells demonstrated Cav-3 inhibition of the Ca v 3.2 channel but not Ca v 3.1 channel. In addition, we show that the N terminus region of Cav-3 interacts with the Ca v 3.2 channel and modulates the I Cav3.2 . Whole cell patch clamp studies using the neonatal cardiomyocytes demonstrated specific inhibition of I Cav3.2 by Cav-3. On the other hand, siRNA-mediated knockdown of Cav-3 eliminated PKA regulation of the I Ca,T .
T-type Ca 2ϩ channels are known to modulate Ca 2ϩ influx, membrane potential and hormone secretion. Of the three different T-type Ca 2ϩ channel isoforms reported, alternative splicing of the Ca v 3.2 channels results in functional diversity of the channel during cardiac development (47). Recent findings using a heterologous expression system of HEK293 cells suggest a complex regulation for the Ca v 3.2 channels. It is well known that G-protein-dependent signaling pathways modulate native T-type Ca 2ϩ channels (2). Extensive studies   (48), Ca 2ϩ /calmodulin-dependent protein kinase II (49,50), and inhibition by G␤␥ (51,52). Another study demonstrated a selective inhibition of the Ca v 3.2 channel by corticotropin releasing factor receptor-1 is likely mediated by ␤␥ (53). On the other hand, a recent study demonstrated phospholipase C␤, and protein kinase C modulation of the Ca v 3.2 channel mediated by G q/11 in a voltage and G␤␥ independent fashion (54). In addition in DRG neurons, monocyte chemoattractant protein-1 (a cytokine) was shown to inhibit Ca v 3.2 channels (55). The present work demonstrates that Cav-3 modulates transiently expressed Ca v 3.2 in HEK293 cells and native Ca v 3.2 current in neonatal mouse cardiomyocytes. Co-expression of Cav-3 inhibited I Cav3.2 without altering the biophysical properties of the Ca v 3.2 channel, suggesting that Cav-3 modulates Ca v 3.2 current possibly in a voltage independent fashion. Our surface biotinylation data demonstrate that Cav-3 co-expression did not alter plasma membrane expression of the Ca v 3.2 channels, which rules out the possibility of internalization of the Ca v 3.2 channels. Thus, Cav-3 inhibition of I Cav3.2 could possibly result through different signaling molecules associated with Cav-3.
Although it is clear from our GST-Cav-3 pulldown experiments that the N terminus region of Cav-3 interacts with the Ca v 3.2 channel and modulates the I Cav3.2 , there exists a likely possibility of direct interaction between the Ca v 3.2 channel and Cav-3. The intracellular linker region between domains II and III for the Ca v 3.2 channel contains clusters of serine and threonine residues and has been identified as an important region for interaction and regulation of the channel by serine/ threonine kinases and G-protein signaling pathway proteins (3). In this context, it may not be unreasonable to hypothesize an interaction between the N terminus region of Cav-3 and the intracellular loop between domains II and III of the Ca v 3.2 channel. Further studies are needed to identify the site of Ca v 3.2 channel involved in the interaction with Cav-3, which may also help us understand dynamic regulation of the Ca v 3.2 channels.
We observed that siRNA-mediated knockdown of Cav-3 did not affect the I Ca,T density in the mouse neonatal cardiomyocytes but abolished PKA regulation of the I Ca,T . Similar to this observation, in an earlier study we had demonstrated that a subpopulation of the Ca v 1.2 L-type Ca 2ϩ channels are localized to caveolae and siRNA-mediated Cav-3 knockdown did not affect the I Ca,L density in neonatal ventricular myocytes. However, Cav-3 knockdown specifically inhibited the ␤ 2 -adrenergic receptor regulation of the Ca v 1.2 channels (24). Caveolins contain scaffolding domains that interact with multiple signaling molecules including G-proteins and coupled receptors, regulatory proteins such as Ca 2ϩ /calmodulin-dependent protein kinase II, PKA, and PKC, and some ion channels, thereby providing temporal and spatial regulation of cellular signal transduction (18,56,57). The subcellular localization of ion channels to caveolae allows the integration of these channels into specific macromolecular signaling complexes in a distinct lipid microenvironment providing for their precise modulation. Caveolar localization and Cav-3 inhibition of Ca v 3.2 channels in neonatal cardiomyocytes explains the compartmentalized and dynamic regulation of the Ca v 3.2 channels by PKA and possibly other kinases in the native cells. Perhaps, caveolar localization and Cav-3 interaction of the Ca v 3.2 are necessary for precise modulation of this channel not only by PKA but also by different signaling molecules during normal and pathological states.
Expression of Cav-3 is developmentally regulated as it increases postnatally and reaches maximum levels by days 4 -5 followed by a decrease to stable expression levels seen in the adult cardiomyocytes (58,59). The cardiac T-type Ca 2ϩ channels are abundantly expressed during cardiac development but their expression is not detected in normal adult ventricular myocytes. It is well established that moderate to severe ventricular tissue remodeling occurs, with associated changes at the single myocyte level, during pathological states such as cardiac hypertrophy and heart failure (60 -62). During cardiac hypertrophy, Ca v 3.1 and Ca v 3.2 channels are known to be up-regulated in the ventricular myocytes. Caveolae and expression of Cav-3 is also known to be altered during these conditions (63,64). It may also be possible that during the ventricular remodeling process altered Cav-3 and caveolae expression could alter subcellular localization and expression FIGURE 7. siRNA-mediated knockdown of Cav-3 expression eliminated PKA-mediated 8-Br-cAMP stimulation of I Ca,T in neonatal mouse ventricular myocytes. Perforated patch whole cell voltage clamp recordings of I Ca,T were performed by using a holding potential of Ϫ90 mV with 50-ms test pulses to Ϫ30 mV every 15 s in myocytes. A, peak I Ca,T is increased by stimulation with 8-Br-cAMP (100 M) in a representative control siRNAtreated myocyte (whole cell capacitance ϭ 8.6 picofarads); B, siRNAmediated Cav-3 inhibition eliminated 8-Br-cAMP stimulation of I Ca,T in a representative myocyte (whole cell capacitance ϭ 7.9 picofarads). C, pretreatment with 10 M PKA inhibitor peptide completely inhibited the 8-Br-cAMP stimulation of the I Ca,T in a representative control siRNA-treated myocyte (whole cell capacitance ϭ 11.9 picofarads). In all the groups of cells, perfusion with 10 M nitrendipine (Nitrend) did not block I Ca , whereas 0.5 mM Ni 2ϩ completely inhibited I Ca , indicating a stimulation of the T-type Ca 2ϩ current by 8-Br-cAMP. D, average effect of 8-Br-cAMP stimulation on I Ca,T in myocytes with and without Cav-3 siRNA inhibition. The data represent mean Ϯ S.E., number of cells used is indicated in parentheses *, p Ͻ 0.001 relative to control.

Caveolin-3 Regulates Ca v 3.2 Channels
of the Ca 2ϩ channels, including the Ca v 3.2 channels that may result in altered coupling and regulation of the channels. The Ca v 3.2 T-type Ca 2ϩ channels are involved in various diseases such as epilepsy (39,(65)(66)(67), pain (68), hypertension (69), and cardiac hypertrophy (16). In fact, using genetic deletion of CACNA1H, which encodes the Cav3.2 channel, it was demonstrated that the Ca v 3.2 T-type Ca 2ϩ channel is required for induction of pathological cardiac hypertrophy (16). One other report indicated that overexpression of Cav-3 is protective against agonist-induced hypertrophic responses in the rat neonatal cardiomyocytes (70). In our studies overexpression of Cav-3 specifically inhibited the T-type Ca 2ϩ current without affecting the L-type Ca 2ϩ currents, which are also known to localize to caveolae and associate with Cav-3 (24). Our data also demonstrate that Cav-3 specifically inhibited the adenovirus-mediated increased overexpression of I Cav3.2 . In this context our finding may have important therapeutic implications pertaining to the T-type Ca 2ϩ channel block, because, T-type Ca 2ϩ channel blockers have been proposed to be useful in the therapeutics of a variety of conditions including hypertension and heart failure (69,71). Further studies are needed to understand the role of Cav-3 and caveolae in the ventricular remodeling process during cardiac disease conditions such as hypertrophy and heart failure. Regulation of the T-type Ca 2ϩ channels and Ca 2ϩ signaling by caveolae is important for understanding the mechanism of pathological changes in cardiac hypertrophy and heart failure.
In summary, our results demonstrate for the first time a precise caveolar localization of the Ca v 3.2 T-type Ca 2ϩ channel in cardiomyocytes. We show that Cav-3 interacts with the Ca v 3.2 channels and modulates PKA regulation of I Ca,T in the ventricular myocytes. Our studies provide the basis for understanding the compartmentalized regulation of the T-type Ca 2ϩ channel in cardiomyocytes and other cell types in normal and pathological conditions.