β-Adrenergic Receptor Activation Induces Internalization of Cardiac Cav1.2 Channel Complexes through a β-Arrestin 1-mediated Pathway*

Voltage-dependent calcium channels (VDCCs) play a pivotal role in normal excitation-contraction coupling in cardiac myocytes. These channels can be modulated through activation of β-adrenergic receptors (β-ARs), which leads to an increase in calcium current (ICa-L) density through cardiac Cav1 channels as a result of phosphorylation by cAMP-dependent protein kinase A. Changes in ICa-L density and kinetics in heart failure often occur in the absence of changes in Cav1 channel expression, arguing for the importance of post-translational modification of these channels in heart disease. The precise molecular mechanisms that govern the regulation of VDCCs and their cell surface localization remain unknown. Our data show that sustained β-AR activation induces internalization of a cardiac macromolecular complex involving VDCC and β-arrestin 1 (β-Arr1) into clathrin-coated vesicles. Pretreatment of myocytes with pertussis toxin prevents the internalization of VDCCs, suggesting that Gi/o mediates this response. A peptide that selectively disrupts the interaction between CaV1.2 and β-Arr1 and tyrosine kinase inhibitors readily prevent agonist-induced VDCC internalization. These observations suggest that VDCC trafficking is mediated by G protein switching to Gi of the β-AR, which plays a prominent role in various cardiac pathologies associated with a hyperadrenergic state, such as hypertrophy and heart failure.

Regulation of voltage-dependent calcium channels (VDCCs) 3 plays a pivotal role in excitation-contraction coupling in cardiac myocytes. During the action potential upstroke, membrane depolarization causes the opening of VDCCs, encoded by the pore-forming ␣ 1 subunit, Ca v 1.2 (1). Ca 2ϩ entry through VDCCs triggers the release of Ca 2ϩ from the sarcoplasmic reticulum via ryanodine receptors. Although the regulation of VDCCs in the heart has been extensively studied, key molecular mechanisms underlying channel function, trafficking, membrane targeting, retention, and internalization remain unknown. Activation of the ␤-AR, a G protein-coupled receptor (GPCR), leads to positive inotropic effects mediated by phosphorylation of the VDCC via cAMP-dependent protein kinase A (2). This, however, is a transient phenomenon since persistent activation of the receptor causes its subsequent phosphorylation by GPCR kinases (GRKs) (3), causing the ␤-AR to become a target for arrestin (4), which mediates the recruitment of the receptor into clathrin-coated vesicles (5).
In addition to decreasing single channel permeability, persistent membrane depolarization can regulate the number of Ca v 1.2 channels at the plasma membrane. For example, sustained KCl-induced depolarization of rat cortical neurons effectively decreases Ca v 1.2 channel activity (6). Ca v 1.2 channels have been proposed to contain a membrane-targeting domain within their calmodulin (CaM)-binding domain in the C terminus (7). Pitt and colleagues (8) showed that Ca 2ϩ -CaM interaction with this domain accelerated the rate of trafficking of Ca v 1.2 channels to distal regions of the dendritic arbor. CaM imparts Ca 2ϩ -dependent regulation of not only mature Ca v 1.2 channels at the cell surface but also during channel biosynthesis.
Mechanisms underlying Ca v 1.2 channel trafficking and retention at the plasma membrane have not been studied in cardiomyocytes, where these channels play a pivotal role in excitation-contraction coupling. Moreover, it is not known whether receptor activity that modulates I Ca-L can also contribute to Ca v 1.2 targeting and retention at the plasma membrane.
Our results show that sustained activation of the ␤-AR induces internalization of VDCCs. This observation raises the possibility that during desensitization, not only does the ␤-AR need to be recycled, but also that the recovery of the channel from the response might require the recycling of the effector, i.e. the calcium channel itself. The time course of channel internalization and its prevention by pertussis toxin raise the possi-bility that the internalization of calcium channels is a result of ␤-AR switch in coupling from G s to G i . Our results represent a new mechanism of cellular adaptation during hyperadrenergic simulation, which might have implications to a host of cardiac pathologies, including hypertrophic remodeling.
Fluoresceinated 894 -929 and 920 -944 peptides used in this study were based on the Ca v 2.2 ␣ 1 sequence from chick dorsal root ganglion neurons (CDB1, GenBank TM AAD51815). Peptides were synthesized by FastMoc chemistry at the Tufts University Core Facility (Boston, MA) and purified by high pressure liquid chromatography with Ͼ97% purity as determined by mass spectrometry. The N terminus included the sequence of the Penetratin domain of the Drosophila protein Antennapedia. Peptides were dissolved in 5 mM acetic acid at 1 mg/ml.
Cardiomyocyte Isolation-Cardiomyocytes were isolated from adult rat hearts as described previously (9). After the cardiomyocyte isolation steps, the pellet was collected and resuspended in plating medium containing Dulbecco's modified Eagle's medium, low glucose 1ϫ (MEM), 10% fetal calf serum, 10 mM 2,3-butanedione, 100 units/ml penicillin, 2 mM glutamine, and gentamicin. Cardiomyocytes were plated on laminin-coated dishes and allowed to attach in a 2% CO 2 incubator at 37°C. After attachment, the cells were washed with plating medium and cultured for 3-18 h with a medium containing: MEM, 42.5 ml; bovine serum albumin, 0.1%; 2,3-butanedione, 10 mM; penicillin, 100 units/ml; glutamine, 2 mM; gentamicin, 0.1 mg. No differences were found in the data obtained from rat cardiac myocytes plated for different time intervals.
Immunohistochemistry-Cardiomyocytes were exposed to control solution (culture medium) or solution containing 100 M isoproterenol (ISO) with 1 mM ascorbic acid. Cells were fixed and permeabilized in methanol at Ϫ20°C for 5 min as described previously (10). Cells were incubated at 4°C overnight with the primary antibody and for 1.5 h at room temperature with the secondary antibody.
Confocal Imaging-The Zeiss LSM-510 Meta (UV) microscope was used to perform confocal laser-scanning microscopy on the fixed cells. For the images taken by the Zeiss Meta microscope, the UV ϫ63 1.4NA oil objective lens was used with a pinhole setting of 1.0. The Zeiss Meta software was used to calculate the number of sections based on acquisition of sections at 380-nm intervals in the Z-plane. The Zeiss Meta microscope settings were kept the same for all scans. Metamorph Analysis software (Universal Imaging Corp., West Chester, PA) was used for all morphometric measurements. For the inte-grated density analysis and the correlation analysis, random cardiomyocytes were selected and manually traced, and background staining was subtracted. The average intensity of fluorescence signal was measured and expressed in arbitrary units of fluorescence per square area.
The integrated values were obtained by measuring the fluorescence values as a function of area. The correlation coefficient (Pearson's coefficient) was calculated using Metamorph (Molecular Devices, Inc.) to measure the degree of coincidence between the two signals.
Co-precipitation-Cardiomyocytes, isolated and cultured in laminin-coated dishes as described above, were used for each condition. Cells were exposed to control solution (culture medium) or control solution containing 100 M ISO plus ascorbic acid. After treatment with agonist, cells were lysed. VDCC was immunoprecipitated from 2 mg of lysate as described previously in Schiff et al. (11).

Cardiac Ca v 1.2/␤-Arr1 Complexes Are Internalized upon
Sustained ␤-AR Activation-Indirect immunofluorescence imaging shows that under basal conditions (treatment with saline), the vast majority of pore-forming calcium channel subunits (Ca v 1.2) are located at the plasma membrane of the adult rat cardiomyocyte (Fig. 1, A and E). Furthermore, these surfacetargeted channels are preassociated with ␤-Arr1 as evidenced from the Z-stack optical slices, which show a high degree of overlap between Ca v 1.2 (labeled with a green fluorescent probe) and ␤-Arr1 (labeled with a red fluorescent probe), yielding a merged yellow hue, indicating co-localization of the two proteins. The relative abundance of the merged signal in the top slice and its relative absence except for a discrete ring around the cell periphery in deeper layers further underscores the selective localization of both Ca v 1.2 and ␤-Arr1 at the cell membrane (Fig. 1A).
Upon ␤-AR stimulation with ISO (100 M) for 5 min, the individual green and red fluorescence signals become distinct and more intense in the middle slices, suggesting dissociation of the Ca v 1.2/␤-Arr1 complex and movement of each component to deeper layers away from the cell surface ( Fig. 1, C, E, and F). The ISO-mediated internalization of the channels was observed in 95% of the cells tested with a 40 Ϯ 5% mean loss of fluorescent signal in the top optical slice. This pattern of Ca v 1.2/␤-Arr1 separation and internalization is further amplified following 15 min of exposure to ISO (Fig. 1, D and F). Exposure of cells to a lower concentration of ISO (100 nM) yielded qualitatively similar results, with 55.0 Ϯ 2.of the signal associated with the cytosolic region observed in 93% of the cardiac myocytes.
Dynamic changes in the association of the two proteins were quantitatively measured using the Pearson's correlation coefficient in saline-treated myocytes (sham, control) and following ISO exposure for 20 s, 5 min, and 15 min. In accordance with the qualitative images, the Pearson's coefficient was significantly reduced from 0.89 Ϯ 0.05 in saline-treated cardiomyocytes to 0.77 Ϯ 0.02 (20 s ISO), 0.69 Ϯ 0.03 (5 min ISO), and 0.67 Ϯ 0.05 (15 min ISO) (n ϭ 10). Biochemical experiments provided further evidence of the association between ␤-Arr1

R E T R A C T E D R E T R A C T E D
and Ca v 1.2 under normal conditions. In fact, our studies confirmed that ␤-Arr1 co-precipitates with Ca v 1.2 channel protein in lysates from normal rat cardiac myocytes (Fig. 1G), as well as normal porcine atrial and ventricular tissue (Fig. 1H). Since our imaging results suggested that ␤-AR activation with ISO for 5 min significantly decreased the association between ␤-Arr1 and Ca v 1.2, we tested whether the physical interaction between the two proteins was also decreased in co-precipitation experiments (Fig. 1I). Ca v 1.2 was immunoprecipitated from cardiac myocytes treated with saline or 100 M ISO, and the precipitates were probed for ␤-Arr1. Indeed, the amount of ␤-Arr1 that co-precipitated with Ca v 1.2 from cells treated with ISO was significantly (p Ͻ 0.05) less than that observed for saline treatment cells (Fig. 1J). Taken together, our data suggest that the association between ␤-Arr1 and Ca v 1.2 channels decreases upon sustained receptor activation.
In heterologous expression systems, ␤-Arr1 is thought to serve as an adaptor to components of the endocytotic machinery, such as clathrin (12). To test whether this is indeed the case in cardiac myocytes, immunofluorescence confocal microscopy was used to image the distribution of ␤-Arr1 and clathrin prior to receptor activation. Optical slices from saline-treated myocytes show that both ␤-Arr1 and clathrin are in close proximity, as indicated by the yellow signal ( Fig. 2A). The Pearson correlation coefficient for saline-treated cells was 0.80 Ϯ 0.05 (n ϭ 10). Upon exposure to ISO for 5 min, the fluorescent signal is detected uniformly in deeper layers, suggesting a cytosolic distribution of the proteins with a Pearson correlation coefficient of 0.65 Ϯ 0.02 (n ϭ 10). Indeed, these results indicate that components of the endocytotic machinery are within close proximity of ␤-Arr1.
G i/o Mediates the Internalization of Cardiac Calcium Channels-The time course of ISO-induced VDCC internalization suggests that this process might contribute at least in part to the desensitization or termination of the ␤-AR response on I Ca-L . ␤-AR-mediated responses can be terminated in two independent ways: 1) through GRK-mediated receptor phosphorylation (13) or 2) through a switch in G protein coupling (14). Hyperadrenergic . Calcium channels and ␤-Arr1 were detected by indirect immunofluorescence using anti-pan ␣1 antibody followed by Oregon Greenconjugated anti-rabbit IgG and anti-␤-Arr1 followed by Cy-3 anti-mouse IgG, respectively. Series of merged images (Ca v 1.2 channel/␤-Arr1, yellow signal) of X-Y optical slices acquired at 0.2 m intervals are shown from the top to bottom of each cell. The scale bar represents 10 m. Data are representative of five independent experiments. E, close view (zoom 150%) of the membrane-associated channel/␤-Arr1 clusters. F, pie charts representing the subcellular distribution of calcium channels expressed as the percentage of total fluorescence signal. Integrated fluorescence was calculated for membrane and cytoplasmic pools of calcium channels. Data represent the mean value of 25 cells. G, calcium channels from adult rat cardiac myocytes were precipitated using anti-pan ␣1 antibody, and the precipitate was probed for ␤-Arr1. Lane 1 is the precipitate, and lane 2 is the total lysate. To avoid the signal from the antibody heavy and light chains, peroxidase-conjugated protein A/G were used. H, calcium channels were precipitated from pig atria and ventricle using anti-Ca v 1.2, and the precipitate was probed for ␤-Arr1. IP, immunoprecipitate; IB, immunoblot. I, adult rat cardiac myocytes were treated with saline or 100 M ISO and then lysed. Ca v 1.2 channel was precipitated, and the precipitate was probed for ␤-Arr1. The membrane was stripped and probed with anti-Ca v 1.2 channel antibodies for normalization. J, the histogram shows quantitation of the density of the ␤-Arr1 band from three independent experiments. Protein density was normalized using the density for the calcium channel band. Error bars represent S.D. Analysis between saline and 5 min of ISO was significant at p ϭ 0.037. JUNE 20, 2008

R E T R A C T E D R E T R A C T E D
stimulation results in the switch of ␤ 2 ARs to G i protein.
To determine whether activation of heterotrimeric G i/o proteins is required for ISO-induced internalization of VDCC/␤-Arr1 complexes in cardiac myocytes, cells were pretreated with 100 ng/ml pertussis toxin (PTx) for 4 h and then exposed to either agonist (ISO, 100 M) or saline. PTx treatment effectively blocked the internalization of VDCC/␤-Arr1 complexes in 95% of the cells tested. Confocal X-Y optical slices of cells pretreated with PTx showed that the VDCC complexes were retained at the plasma membrane. In fact, the yellow signal representing the co-localization of Ca v 1.2 and ␤-Arr1 was limited to the top optical slice and the periphery of the myocyte in the middle slices. Integrated fluorescence signals for cells pretreated with PTx and exposed to ISO for 5 min demonstrate that Ca v 1.2 is located predominantly at the plasma membrane in sharp contrast to the internalized channel distribution in myocytes treated with ISO for 5 min (Fig. 2B). Pretreatment of myocytes with PTx prevents ISO-induced internalization of Ca v 1.2 as the fluorescent signal remains membrane bound, even following 15 min of ISO treatment (Fig. 2, B and C). In addition to its potent inhibitory effect upon channel internalization, PTx also prevents the dissociation between Ca v 1.2 and ␤-Arr1, as evident by a high Pearson coefficient measured in myocytes following exposure to ISO for 5 (0.81 Ϯ 0.04) and 15 (0.82 Ϯ 0.01) min, indicating a high degree of co-localization between then two proteins. These results indicate that ISO-induced internalization of Ca v 1.2/␤-Arr1 complexes requires the activation of G i/o proteins.
␤-Arr1 and Tyrosine Kinase Activity Are Required for Agonist-mediated Internalization of Cardiac Ca v 1.2 Channels-We tested whether arrestin plays a role in agonist-induced internalization of VDCCs. Ca v 1.2 channels at the cell surface were labeled for 5 min at 37°C using 1.5 M DM-BODIPY, a fluorescent dihydropyridine. Because dihydropyridines bind to the extracellular domain of the channel, only the channels at the plasma membrane are labeled. Preincubation of cardiac myocytes with unlabeled dihydropyridine prevents the binding of DM-BODIPY, demonstrating that the probe binds selectively (data not shown). The pattern of VDCC subcellular distribution and the extent of ISO-induced internalization were not different in myocytes in which the channels were labeled with anti-Ca v 1.2 antibodies when compared with the labeling by fluorescent dihydropyridine (data not shown). Image acquisition at 5-s intervals for 15 min produced Ͻ5% loss of signal, suggesting that any loss of fluorescence signal observed in our experiments was not caused by photobleaching.
In live cell imaging experiments, exposure of cardiac myocytes to ISO produced a decrease in the fluorescence signal in the top surface of the cell (Fig. 2D), concurrent with an increase in fluorescence in the middle optical slices. Washout of the agonist (ISO) resulted in the recovery of the fluorescent signal at the top slice and the cell periphery of deeper optical slices to a level comparable with that prior to ISO exposure (supplemental Fig. 1).
To test whether Ca v 1.2-␤-Arr1 interaction is required for ISO-induced internalization of VDCCs, we used cell-permeant peptides based on the sequence of the ␤-Arr1-binding site in Ca v 2.2 channel (Gallus gallus CDB1, aa 894 -944) (15). We have previously used this peptide to interfere with ␤-Arr1-channel interaction in dorsal root ganglion neurons and block agonist-induced VDCC internalization (15). This peptide works as a "sponge" to bind the endogenous ␤-Arr1, which should result in the blockade of responses mediated by this molecule. In control experiments, a wide range of concentrations (100 ng/ml to 100 g/ml) and incubation times (5 min to 1 h) was tested. Data shown in Fig. 2D were obtained by incubating rat cardiac myocytes with a saturating concentration of peptide (1.4 g/ml) for 5 min. In time-lapse experiments, the aa 894 -929 peptide prevents agonist-induced Ca v 2.2 channel internalization without altering their basal distribution. A structurally similar but inactive peptide containing aa 920 -944 was without effect (Fig. 2D).

DISCUSSION
Our results show that ␤-Arr1 interacts with Ca v 1.2 channels in an agonist-independent manner (Fig. 2G). This is in contrast to the interaction of arrestin with receptors, which requires activation of the receptor, and GRK-mediated phosphorylation of the receptor. Upon sustained receptor activation, Ca v 1.2 channels and ␤-Arr1 are internalized, and a decrease in association between the two proteins is observed over time. The decrease in co-localization of the channels and ␤-Arr1 might suggest differential sorting of these proteins in later stages of endocytosis.
In the present study, we demonstrate that GPCRs strongly regulate the trafficking of cardiac calcium channels at the plasma membrane. The internalization of calcium channels is observed upon sustained (Ն5 min) receptor activation, raising the possibility that desensitization might require the recycling of VDCCs in addition to receptor endocytosis. Future experiments will address whether the channel undergoes dephosphorylation during the recycling phase in a manner analogous to that of the receptor. Our results indicate that ␤AR-induced internalization and decreased association between Ca v 1.2 and ␤-Arr1 are prevented by PTx, implicating a role for G i /G o (Fig.  2G). Phosphorylation of the ␤ 2 AR by cAMP-dependent protein kinase A switches its coupling from G s to G i (16). Our observations raise the possibility that VDCC trafficking can be a consequence of G protein switching of the ␤-AR, which might play a prominent role in cardiac hypertrophy and disease progression.
The role that pertussis toxin-sensitive G i/o proteins play in the regulation of cardiac VDCCs remains unclear. Activation of