Activated Cardiac Adenosine A1 Receptors Translocate Out of Caveolae*

The cardiac affects of the purine nucleoside, adenosine, are well known. Adenosine increases coronary blood flow, exerts direct negative chronotropic and dromotropic effects, and exerts indirect anti-adrenergic effects. These effects of adenosine are mediated via the activation of specific G protein-coupled receptors. There is increasing evidence that caveolae play a role in the compartmentalization of receptors and second messengers in the vicinity of the plasma membrane. Several reports demonstrate that G protein-coupled receptors redistribute to caveolae in response to receptor occupation. In this study, we tested the hypothesis that adenosine A1 receptors would translocate to caveolae in the presence of agonists. Surprisingly, in unstimulated rat cardiac ventricular myocytes, 67 ± 5% of adenosine A1receptors were isolated with caveolae. However, incubation with the adenosine A1 receptor agonist 2-chlorocyclopentyladenosine induced the rapid translocation of the A1 receptors from caveolae into non-caveolae plasma membrane, an effect that was blocked by the adenosine A1 receptor antagonist, 8-cyclopentyl-1,3-dipropylxanthine. An adenosine A2areceptor agonist did not alter the localization of A1receptors to caveolae. These data suggest that the translocation of A1 receptors out of caveolae and away from compartmentalized signaling molecules may explain why activation of ventricular myocyte A1 receptors are associated with few direct effects.

Adenosine has multiple and profound effects on cardiac function. Adenosine increases coronary blood flow and exerts direct negative chronotropic and dromotropic effects (1). There is also significant evidence that adenosine reduces both reversible and irreversible myocardial ischemic-reperfusion injury (2). The cardiac effects of adenosine are mediated by the activation of specific extracellular adenosine receptor subtypes, A 1 , A 2a , and A 3 , located on vascular smooth muscle, coronary endothelial cells, and cardiac myocytes (1). The adenosine A 1 receptor subtype is localized primarily on cardiac myocytes (1).
The adenosine A 1 receptor is a 36-kDa, seven-transmembrane domain protein localized to the sarcolemma (3). Agonist binding induces a conformational change in the receptor, which facilitates its interaction with an inhibitory guanine nucleotide-binding protein (G i ) (3). With the exception of the ferret, adenosine A 1 receptor occupation does not significantly alter contractility, intracellular calcium, or cAMP concentration in normal mammalian ventricular myocardium (4). However, A 1 receptor occupation does reduce ␤-adrenergic receptor-induced increases in all of the above parameters. Although the antiadrenergic effect of adenosine is well studied, evidence of other signal transduction mechanisms mediated by adenosine A 1 receptor occupation in ventricular myocardium is lacking. The fact that other receptors in ventricular myocytes, which presumably couple to the same G i protein, produce direct effects (5)(6)(7)(8) raises the question of why adenosine A 1 receptor activation is associated with so few effects. One explanation for the lack of direct adenosine A 1 receptor effects may be the relatively low receptor density in ventricular myocardium compared with atrial tissue (9). Another possibility is that the receptor is inefficiently coupled to G i proteins and second messengers. Coupling efficiency may be due, in part, to the grouping of receptors and second messengers in microdomains for rapid and selective activation or deactivation of intracellular signaling mechanisms.
Caveolae are cholesterol/sphingomyelin-rich plasma membrane microdomains that have been shown to serve as a scaffold for receptors and second messengers (10). Caveolae are present in most cell types and are abundant in cardiomyocytes (11). There are several reports that G protein-coupled receptors redistribute to caveolae in response to receptor occupation (11)(12)(13). In addition, other studies have documented that specific second messenger systems, including heterotrimeric G proteins, are concentrated in caveolae (for review, see Ref. 10). Although the role of caveolae in cardiac signal transduction has not been well studied, it has been reported that muscarinic M 2 cholinergic receptors in rat ventricular cardiomyocytes localize to caveolae after agonist stimulation (11).
Because adenosine and acetylcholine exert similar effects in ventricular myocardium (14), we hypothesized that stimulation of adenosine A 1 receptors would promote translocation of the receptors into caveolae. Surprisingly, we found that in rat ventricular cardiomyocytes, adenosine A 1 receptors were associated with caveolae in unstimulated cells, whereas activation of the adenosine A 1 receptors induced the receptors to translocate out of caveolae. Ventricular myocytes were dissociated from adult, male Sprague-Dawley rats (350 -400 g) using collagenase and hyaluronidase as described previously (15). The resulting cell suspension was washed twice with enzyme-free buffer, and suspended in normal HEPES buffer (1.0 mM CaCl 2 , pH ϭ 7.4) to a final concentration of ϳ1 mg/ml. Typically, over 70% of the cells were rod-shaped, trypan blue-excluding ventricular myocytes.

Materials-Mouse
Isolation of Caveolae-Caveolae were isolated as described previously (16). Ten to 15 g of purified caveolae were isolated from three rat heart preparations with this method.
Enzyme Assays-Alkaline phosphatase was measured by the method of Engstrom (17). Galactosyltransferase and NADPH-cytochrome c reductase were assayed using methods adapted from Graham (18). Lactate dehydrogenase was measured with a standard kit from Sigma.
Cholesterol Mass Measurements-The mass of cholesterol was determined as described previously (19).
Treatment Protocols-Three groups (n ϭ 2-3 experiments/group) of cells were studied. To provide enough protein sample for the fractionation of cell lysates, the cardiomyocyte yield from three hearts was combined in a single experiment. All myocyte suspensions were performed in the presence of adenosine deaminase (2 units/ml) to degrade endogenous adenosine. Control myocytes were suspended at room temperature in 7 ml of HEPES buffer (ϳ3-4 mg of protein/ml) with periodic mixing. A second group of myocytes was incubated with the adenosine A 1 receptor agonist 2-chloro-N 6 -cyclopentyladenosine (CCPA, 200 nM) for 15 min. The suspension was then centrifuged (1000 ϫ g), and the pellet was washed three times with normal HEPES buffer (10 ml). After the final wash, the myocyte pellet was processed to isolate caveolae as described above. Group 3 myocytes were initially pretreated with the adenosine A 1 receptor antagonist 8-cyclopentyl-1,3-dipropylxanthine (DPCPX, 200 nM) for 5 min prior to the addition of CCPA (200 nM). One additional experiment was performed in which myocytes were incubated with the adenosine A 2a receptor agonist 2-p-(2-carboxyethyl)phenethylamino-5Ј-N-ethylcarboxyamidoadenosine (CGS 21680, 100 nM) as described for CCPA.
Immunoelectron Microscopy-Immunogold localization of caveolin-3 (20 nm gold) and A 1 receptors (5 nm gold) was performed using whole mount plasma membrane preparations as described previously (20).

Isolation of Caveolae from Rat Ventricular Myocytes-We
defined the presence of caveola in ventricular myocytes by an established isolation procedure (16). In brief, ventricular myocytes were lysed and a postnuclear supernatant generated. The postnuclear supernatants were separated into cytosol, plasma membrane, and intracellular membrane (endoplasmic reticulum, Golgi, etc.) fractions by centrifugation in Percoll. The plasma membranes were then sonicated and fractionated by density gradient centrifugation to isolate caveolae. Similar to other cell type studies, the caveola fraction from ventricular myocytes contained less than 0.4% of the protein found in the initial postnuclear supernatant fraction (Table I).
The most probable contaminates of the plasma membrane and caveola fractions are the endoplasmic reticulum and Golgi. Therefore, we measured the amount of NADPH-cytochrome reductase (endoplasmic reticulum) and galactosyltransferase (Golgi) activity in each of the subcellular fractions ( Table I).
The majority of the activities were associated with the intracellular membranes. The plasma membrane contained less than 2% of the total postnuclear supernatant activities, and the caveola fraction did not contain any detectable NADPH cytochrome reductase or galactosyltransferase activity.
To ensure that the caveola fraction was not contaminated with other plasma membrane domains, 5 g (Fig. 1A) or 3% (Fig. 1B) of each subcellular fraction was resolved by SDS-PAGE and immunoblotted for various caveola and non-caveola proteins. The equal protein loads in Fig. 1A illustrate the relative enrichment of each protein in the caveola fraction, whereas the proportional protein loads in Fig. 1B illustrate total protein distribution. Caveolin-3, a marker for muscle caveolae, was highly enriched in the caveola fraction. In addition, eNOS, which has been shown to directly interact with caveolin (21), was also enriched in the caveola fraction. The non-caveola proteins, clathrin and transferrin, were excluded from the caveola fraction. The yield of caveolin-3 and eNOS in the caveola fraction was determined by immunoblot analysis in the linear range of detection (data not shown). The estimated yield, with respect to the plasma membrane fraction, of caveolin-3 was 61 Ϯ 6% and eNOS was 69 Ϯ 8%.
Caveolae are highly enriched in cholesterol (16). To further characterize the myocyte caveola fraction, we used an enzymatic kit to determine the mass of cholesterol associated with each fraction. Fig. 2 demonstrates that the caveola fraction, which contained 24 g of cholesterol, was 7.4-fold more enriched in cholesterol than the plasma membrane fraction, which contained 79 g of cholesterol.
Subcellular Localization of Adenosine A 1 Receptors-The subcellular distribution of A 1 receptors under basal conditions was determined in ventricular myocytes incubated in HEPES buffer supplemented with adenosine deaminase. Three percent of each subcellular fraction was analyzed for the presence of A 1 receptors by SDS-PAGE and immunoblotting. Under these conditions, A 1 receptors were highly concentrated (67 Ϯ 5% with respect to the plasma membrane) in the caveola fraction (Fig.  3A, Buffer). Similar results were obtained in two other groups of control myocytes.
We next determined the subcellular localization of A 1 receptors after incubating myocytes with the A 1 receptor agonist CCPA. Treatment with CCPA caused the quantitative removal of A 1 receptors out of the caveola fraction (Fig. 3A, CCPA). Similar results were obtained when this treatment was repeated in additional myocytes. Quantification of the immunoreactive bands demonstrated that greater than 90% of the signal originally detected in the caveola fraction was now detected in the plasma membrane fraction (data not shown). To verify that this caveola to plasma membrane translocation was selective for A 1 receptors, two additional protocols were performed. In the first, pretreatment of myocytes with the A 1 receptor antagonist DPCPX prior to addition of CCPA prevented the redistribution of the receptor (Fig. 3A, CCPA ϩ DPCPX). Finally, treatment of cells with the A 2a agonist CGS 21680 did not affect the caveola localization of the A 1 receptor (Fig. 3A, CGS). Immunoblots for caveola and non-caveola marker proteins demonstrated that CCPA treatment did not alter the yield or purity of caveolae (Fig. 3B).
Recently, Stan et al. (22) suggested that subcellular fractionation methods to isolate caveolae are contaminated with vesicles with similar biophysical properties as caveolae. Therefore, to verify that A 1 receptors are associated with caveolae, we used caveolin-3 IgG to immunoprecipitate caveola membranes from isolated caveolae. Isolated ventricular myocytes were incubated in control or CCPA-supplemented HEPES buffer for 15 min, then processed to immunoprecipitate caveolae with caveolin-3 IgG. The precipitated material and the remaining supernatants were resolved by SDS-PAGE and immunoblotted with IgGs for eNOS, A 1 receptor, and caveolin-3. Caveolin-3 coimmunoprecipitated eNOS and the A 1 receptor from caveolae isolated from cells treated with buffer (Fig. 4, Buffer). However, caveolin-3 only co-immunoprecipitated eNOS from caveolae isolated from cells treated with CCPA (Fig. 4, CCPA). Approximately 10% of caveolin-3, eNOS, and A 1 receptor was not precipitated with caveolin-3 IgG.
We have shown by subfractionation and immunoprecipitation that A 1 receptors are localized to caveolae. To confirm FIG. 1. Characterization of rat ventricular cardiomyocyte caveolae. A, 5 g of protein from each subcellular fraction was resolved by SDS-PAGE and immunoblotted for caveolin-3 and eNOS (caveola markers) and transferrin receptor and clathrin (non-caveola markers). B, equal proportions of each subcellular fraction (3%) were resolved by SDS-PAGE and immunoblotted as described above. PNS, postnuclear supernatant (96 g); CYTO, cytosol (45 g); IM, total intracellular membranes (43 g); PM, total plasma membrane (6 g); CM, caveola membrane (0.3 g). The immunoblots were developed by the method of chemiluminescence. The caveolin-3 immunoblot was exposed for 10 s and the eNOS, transferrin receptor (TR), and clathrin immunoblots were exposed for 3 min. Representative data from four independent experiments are shown.  4. CCPA prevents the co-immunoprecipitation of adenosine A 1 receptors with caveolin-3. Ventricular cardiomyocytes were incubated in control buffer or treated for 15 min with buffer containing CCPA (200 nM) at 37°C. The cells were then processed to isolate caveola membranes. IgG directed against caveolin-3 (2 g/ml) was used to immunoprecipitate caveolae from the caveola enriched subcellular fraction. The entire precipitate and the entire supernatant was resolved by SDS-PAGE and immunoblotted for eNOS, A 1 receptor, and caveolin-3. The immunoblots were developed by the method of chemiluminescence (4-min exposures). Representative data from three independent experiments are shown. PEL, pellet; SUP, supernatant. these findings in an independent manner, A 1 receptors were localized by immunoelectron microscopy. Cells were treated with buffer (Fig. 5A) or CCPA (Fig. 5B) as described above. The membranes were then processed to localize caveolin-3 (arrowheads) and A 1 receptors (arrows). Caveolin-3 and A 1 receptors were found over small invaginations on the membrane surface that had the characteristic appearance of caveolae in buffertreated cells. However, after CCPA treatment, only caveolin-3 was associated with these small invaginations.

DISCUSSION
This is the first report of adenosine receptor association with caveolae. Under basal conditions, ventricular myocyte A 1 receptors were localized primarily in caveolae. However, after agonist (CCPA) binding, A 1 receptors translocated from the caveolae to the plasma membrane fraction. This effect was blocked by prior treatment with the selective A 1 receptor antagonist DPCPX, indicating that this phenomenon is agonistspecific. This unique pattern of receptor localization may explain, in part, the lack of direct effects of A 1 receptor in ventricular myocytes.
The A 1 receptor is a member of the family of G proteincoupled receptors containing a seven-transmembrane-spanning domain. There are several reports that similar G proteincoupled receptors, e.g. angiotensin II (12), bradykinin B 2 (13), and endothelin A (23), and muscarinic receptors (13) are present in caveolae. However, the present results indicate that, in contrast to these receptors, the A 1 receptor translocates out of, rather than into, caveolae with receptor activation. There are at least two differences between ventricular myocyte A 1 receptors and these other G protein-coupled receptors, which may explain this unique relationship of A 1 receptors with caveolae. First, angiotensin II, bradykinin, and endothelin have been reported to increase intracellular calcium and exert direct inotropic effects in ventricular myocytes (5)(6)(7). In contrast, A 1 receptor occupation is not associated with these effects (4). Another difference is the phenomenon of receptor desensitization. Angiotensin II, bradykinin B 2 , and endothelin A receptors all desensitize within minutes of receptor activation (24 -26). In fact, it has been hypothesized that receptor movement into caveolae may play a role in this phenomenon (27). Cardiac A 1 receptors desensitize only after several hours to days of continuous agonist treatment (28). It should be noted that movement out of caveolae is also associated with desensitization because agonist-induced translocation of epidermal growth factor receptor out of caveolae is associated with rapid desensitization (29).
Our observation of A 1 receptor translocation from caveolae to sarcolemma after agonist treatment is the opposite of that reported for the ventricular myocyte muscarinic M 2 receptor (11). This is surprising, given the fact that A 1 and muscarinic M 2 receptors exert a nearly identical pertussis toxin-sensitive anti-adrenergic effect in ventricular myocytes. However, mus-carinic receptors have several properties unique from A 1 receptors, which may explain these different associations with caveolae. The muscarinic agonist carbachol, at doses similar to that used by Feron et al. (11), increases inositol phosphate turnover, intracellular Ca 2ϩ , and contractility in ventricular myocytes (5). In addition, it appears that muscarinic M 2 receptors may exert direct effects in ventricular myocytes via coupling to eNOS (11). Similar to the observations of Feron et al. (11), we observed in the present study that eNOS is localized in caveolae. Finally, it has been reported that muscarinic M 2 receptors can be desensitized with brief agonist exposure (30).
The A 1 receptor antibody used in this study (Chemicon International) was raised against the third extracellular domain of the rat adenosine A 1 receptor gene (amino acids 163-176). In addition to the 36-kDa band (the A 1 receptor is a 36-kDa protein), the antibody also cross-reacted with a 74-kDa band. Ciruela et al. (31) also observed a 74-kDa band in several tissues (brain, kidney, lung) in several species (rat, pig, lamb) using a rabbit polyclonal antibody raised against the same peptide sequence of the rat A 1 receptor. They reported that this band was converted to a 39-kDa form after agonist binding. We observed similar results in the present study, although agonist binding did not convert all of the higher molecular weight band to the lower molecular weight form. The mechanistic explanation for this observation is not known. Identical results were obtained with an Alpha Diagnostic International antibody generated against the same epitope (data not shown).
Several lines of evidence support the hypothesis that A 1 receptor movement between caveolae and sarcolemma is dependent on occupation of the receptor by an A 1 agonist. First, all studies were performed in the presence of adenosine deaminase to degrade endogenously released adenosine. The subsequent addition of the A 1 receptor agonist CCPA, which is relatively selective for A 1 receptors, resulted in the movement of the receptor from caveolae to sarcolemmal membranes. At the concentration used in the present study, there is no evidence that CCPA activates cardiac myocyte A 2a or A 3 receptors. In addition, the A 2a agonist CGS 21680 did not induce A 1 receptor translocation from caveolae. Finally, DPCPX, which is a highly selective adenosine A 1 receptor antagonist (32), completely blocked the effects of CCPA.
Since this is the first report of the localization of the A 1 receptor in caveolae, it is not known whether A 1 receptor translocation out of caveolae activates or terminates signaling. There are reports that inhibitory G protein subunits are localized in caveolae (33), and cardiac A 1 receptors couple primarily to G␣ i subunits (34). Two components of the only known pathway that A 1 receptor activation modulates in ventricular myocardium, ␤-adrenergic receptors and adenylyl cyclase, have been reported to be localized in light vesicular fractions which may be caveolae (35,36). Furthermore, cardiac adenylyl cyclase can be inhibited in vitro by the caveolin-3 scaffolding domain peptide (37). Therefore, the location of A 1 receptors is likely to significantly influence the ability of the receptors to couple to downstream signaling events. It has recently been reported that stimulation of rat cardiac myocytes with endothelin is associated with the recruitment of PKC-␣ and -⑀ isoforms to the caveolae fraction (38). Henry et al. (39) previously reported that A 1 receptor stimulation of ventricular myocytes induced a transient cytosol to membrane translocation of PKC-␦. Since endothelin, but not A 1 , receptor stimulation is associated with direct effects in myocytes, it is possible that A 1 receptor stimulation may recruit PKC-␦ out of caveolae, where PKC signaling may be physiologically important.
In summary, we have demonstrated that, in contrast to other G protein-coupled receptors, A 1 receptors translocate out of FIG. 5. Localization of caveolin-3 and A 1 receptors in plasma membranes by immunoelectron microscopy. Immunogold labeling was performed using monoclonal anti-caveolin-3 IgG and polyclonal anti-A 1 receptor IgG. Caveolin-3 staining is marked with arrowheads, and A 1 receptor staining is marked with arrows. Bar ϭ 0.2 m.