The Cytosolic Phospholipase A2 Pathway, a Safeguard of β2-Adrenergic Cardiac Effects in Rat*

We have recently demonstrated that in human heart, β2-adrenergic receptors (β2-ARs) are biochemically coupled not only to the classical adenylyl cyclase (AC) pathway but also to the cytosolic phospholipase A2 (cPLA2) pathway (Pavoine, C., Behforouz, N., Gauthier, C., Le Gouvello, S., Roudot-Thoraval, F., Martin, C. R., Pawlak, A., Feral, C., Defer, N., Houel, R., Magne, S., Amadou, A., Loisance, D., Duvaldestin, P., and Pecker, F. (2003) Mol. Pharmacol. 64, 1117–1125). In this study, using Fura-2-loaded cardiomyocytes isolated from adult rats, we showed that stimulation of β2-ARs triggered an increase in the amplitude of electrically stimulated [Ca2+]i transients and contractions. This effect was abolished with the PKA inhibitor, H89, but greatly enhanced upon addition of the selective cPLA2 inhibitor, AACOCF3. The β2-AR/cPLA2 inhibitory pathway involved Gi and MSK1. Potentiation of β2-AR/AC/PKA-induced Ca2+ responses by AACOCF3 did not rely on the enhancement of AC activity but was associated with eNOS phosphorylation (Ser1177) and l-NAME-sensitive NO production. This was correlated with PKA-dependent phosphorylation of PLB (Ser16). The constraint exerted by the β2-AR/cPLA2 pathway on the β2-AR/AC/PKA-induced Ca2+responses required integrity of caveolar structures and was impaired by Filipin III treatment. Immunoblot analyses demonstrated zinterol-induced translocation of cPLA and its cosedimentation with MSK1, eNOS, PLB, and sarcoplasmic reticulum Ca2+ pump (SERCA) 2a in a low density caveolin-3-enriched membrane fraction. This inferred the gathering of β2-AR signaling effectors around caveolae/sarcoplasmic reticulum (SR) functional platforms. Taken together, these data highlight cPLA as a cardiac β2-AR signaling pathway that limits β2-AR/AC/PKA-induced Ca2+ responses in adult rat cardiomyocytes through the impairment of eNOS activation and PLB phosphorylation.

the identified cardiac catecholamine receptors, ␤ 1 -adrenergic receptors (␤ 1 -AR) 1 and ␤ 2 -AR received the most interest as mainstay regulators of cardiac performance. Noteworthy, ␤ 1 -AR and ␤ 2 -AR functionally diverge because of clear discrepancies in signaling transduction pathways (2). Chronic overstimulation of cardiac ␤ 1 -AR, associated with exclusive activation of the G s /adenylyl cyclase (AC) pathway, is detrimental and leads to hypertrophy, fibrosis, and heart failure (3). In sharp contrast, sustained ␤ 2 -AR stimulation protects cardiomyocytes against apoptosis. In addition, transgenic mice with cardiac moderate ␤ 2 -AR overexpression display not only long term improved cardiac function but also normal life expectancy (4 -6). The beneficial effect of ␤ 2 -AR stimulation appears to rely on the ability of ␤ 2 -AR to couple with pertussis toxin (PTX)sensitive G proteins (G i2 and G i3 ), leading to a ␤ 2 -AR/G s /AC signaling spatially confined to the caveolar structures and functionally moderated, as compared with the ␤ 1 -AR/G s /AC pathway (7)(8)(9). Identification of ␤ 2 -AR/G i -downstream effectors is of major interest to elucidate, and to further exploit the cardiac protective ␤ 2 -AR potential.
Recently, on human cardiac ventricular and auricular biopsies, we identified a new ␤ 2 -AR-stimulated pathway, namely the G i /cytosolic phospholipase A 2 (cPLA 2 ) pathway, in addition to the classical ␤ 2 -AR/G s /AC pathway (10). cPLA 2 is a high molecular mass enzyme (85 kDa), activated by submicromolar concentrations of Ca 2ϩ , that displays a unique selectivity for arachidonyl in the sn-2 position of phospholipids (11). cPLA 2 is fully activated by both phosphorylation by mitogen-activated protein kinase and increases in [Ca 2ϩ ] i . cPLA 2 activation is associated with its translocation from the cytosol to intracellular membranes, such as endoplasmic/sarcoplasmic reticulum, Golgi apparatus, and nuclear envelope. The essential role of cPLA 2 in inflammation, asthma, neurodegenerative diseases, and bleomycin-induced pulmonary fibrosis is now well documented, and cPLA 2 has been identified as an attractive therapeutic target in the development of new inflammatory drugs (12)(13)(14)(15). In contrast, a dearth of information exists on the possible functional impact of the cPLA 2 and its role in the development of cardiac diseases. Only one recent study has established the antihypertrophic potential of the cPLA 2 pathway in cardiac and skeletal muscles, using the model of the cPLA 2 knockout mice (16). In the present study, we examined the impact of ␤ 2 -AR-G i /cPLA 2 and ␤ 2 -AR-G s /Ac pathways on Ca 2ϩ signaling and contraction in isolated adult rat cardiomyocytes. We show that ␤ 2 -AR agonists trigger an increase in the amplitude of electrically stimulated [Ca 2ϩ ] i transients that relies on the ␤ 2 -AR-G s /AC/PKA pathway and is under the negative control of the ␤ 2 -AR-G i /MSK1/cPLA 2 pathway. We provide evidence that upon ␤ 2 -AR stimulation cPLA 2 translocates to caveolae/ sarcoplasmic reticulum (SR)-functional platforms and bridles phosphorylation of endothelial nitric-oxide synthase (eNOS) and phospholamban (PLB). Using fluorescence imaging, we show that the ␤ 2 -AR/cPLA 2 pathway neutralizes NO production. Our study highlights the essential physiological role of the ␤ 2 -AR/G i /cPLA 2 pathway in limiting ␤ 2 -AR/AC/ PKA-induced Ca 2ϩ responses.

Methods
Cardiomyocyte Isolation-The care and the use of animals were in accordance with institutional guidelines. Adult male Wistar rats (180 -250 g, Janvier, Le Genest St Isle, France) were used. Calcium-tolerant myocytes were isolated by cardiac retrograde aortic perfusion as previously described, but using liberase blendzyme III (30 g/ml) instead of collagenase (17,18). Freshly isolated cardiomyocytes were plated on laminin (3 g/ml; Sigma) and cultured up to 3 days according to Sambrano et al. (19).

Measurement of [Ca 2ϩ ] i Transients and Cell Fractional
Shortening-Plated cardiomyocytes were loaded with Fura 2-AM (1.5 M) for 20 min and submitted to electrical stimulation (square waves, 0.5 Hz). [Ca 2ϩ ] i imaging experiments were performed, as previously described in a saline buffer containing 10 mM glucose, 130 mM NaCl, 5 mM KCl, 10 mM Hepes pH 7.4, 1 mM MgCl 2 , 2 mM CaCl 2 (BSS buffer) (18,20,21). Results are shown as mean Ϯ S.E. for 5 to 15 cells obtained from three different isolations or as a typical representation.
Cell fractional shortening was determined, as described previously (18,20), from the fluorescence images recorded to measure F 360 /F 380 , with Scion Image Software (Scion, Frederick, MD). Results are shown as a typical representation.
Inside cells, this dye is hydrolyzed by cytosolic esterases to the nonpermeable DAF-FM. In the presence of nitric oxide and oxygen, the relatively non-fluorescent DAF-FM is converted into the highly fluorescent and photostable triazole form, DAF-FM T, whose fluorescence intensity is directly proportional to the NO concentration. This dye was chosen as the most sensitive fluorescent probe allowing NO imaging (detection from 3 nM [NO] i ), and displaying the important advantages of a fluorescent spectrum independent of pH variations above pH 5.5 and a resistance to bleaching (22). Experiments were performed at room temperature, and cells were exposed to 1 M DAF-FM diacetate and 1.5 M Fura 2-AM, for 20 min, to combine [NO] i and [Ca 2ϩ ] i imaging. Cells were washed in BSS solution containing 100 M L-arginine and submitted to electrical stimulation. Because of Fura-2 loading, cells were checked for appropriate [Ca 2ϩ ] i transient responses to electrical stimulation, as well as to agonists (zinterol or zinterol plus AACOCF 3 ), before and after the NO imaging experiment. DAF-FM diacetate fluorescence was excited at 480 nm, and emitted cellular fluorescence was recorded at 540 nm. Changes in cellular DAF-FM diacetate fluorescence intensities (F) in each experiment were normalized to the level of fluorescence recorded before agonist application (F 0 ), and changes in [NO] i are expressed as F/F 0 . L-Arginine was omitted when L-NAME was used to block NO synthase. Images were recorded every 6 s. Results are described as mean Ϯ S.E. for 10 -15 cells obtained from three different isolations or shown as a typical representation.
Adenylyl Cyclase Assay-A particulate fraction was obtained from freshly isolated cardiomyocytes suspended in lysis buffer containing 50 mM Hepes, pH 7.4, 2 mM EDTA, 1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride, disrupted by sonication, and centrifuged for 30 min at 20,000 ϫ g. The pellet was resuspended in 50 mM Hepes, pH 7.4 and stored in liquid nitrogen. Adenylyl cyclase was measured as described previously (10,23). Results were obtained from triplicate determinations.
Drug Treatment-Note that a maximal 1 M zinterol concentration was chosen in order to focus on selective ␤ 2 -AR-mediated effects. AA-COCF 3 (10 M), RO 318220 (1 M), L-NAME (1 mM), and H89 (3 M) were added 10 min, or 30 min concerning H89, before addition of other stimuli, respectively. PTX treatment (500 ng/ml for 24 h) was carried out as described previously (23). Adult rat cardiomyocytes were incubated with Filipin III (0.2 g/ml) at room temperature for 30 min before zinterol (1 M) exposure. It is noteworthy that this dose of Filipin III allowed us to perform experiments on Fura-2-loaded cells that were devoid of any sign of Fura-2 leakage and were responsive to electrical stimulation for at least an additional 1 h experimental duration, thus arguing for peripheral cellular membrane integrity.
Preparation of Low Density Caveolin-3-enriched Membrane Fractions-Fractions enriched in the muscle-specific caveolin-3 isoform were prepared from isolated adult rat cardiomyocytes according to a detergent-free purification scheme as described previously (24,25). Freshly isolated cardiomyocytes (1.5 ϫ 10 6 cells/condition) were submitted to a 10 min of pretreatment with or without 10 M AACOCF 3 followed by a 10-min incubation with or without 1 M zinterol in BSS buffer at 37°C. All subsequent steps were carried out at 4°C. Cells suspended in 1 ml of solution A (0.5 M Na 2 CO 3 , pH 11, containing phosphatase inhibitors (10 mM NaF, 100 M Na 3 VO 4 , and 5 mM Na 4 P 2 O 7 )) were sequentially disrupted by homogenization with a loose fitting Dounce homogenizer (10 strokes), a Polytron tissue grinder (three 10-s bursts), and a bath sonicator (one 5-min burst). 1 ml of homogenate was then adjusted to 45% sucrose (% Brix measured with a refractometer) by adding 3 ml of 60% sucrose prepared in 2 volumes of MES-buffered saline (MBS: 25 mM MES, pH 6.5, and 0.15 M NaCl) and 1 volume of solution A. The 45% sucrose fraction was placed on the bottom of an ultracentrifuge tube, overlaid with a 5-30% continuous sucrose gradient (sucrose solutions were prepared using equal volumes of MBS and solution A) and centrifuged at 39,000 rpm for 18 h at 4°C in a SW41Ti rotor (Beckman Coulter). After centrifugation, eleven 1-ml fractions were collected from the top of the gradient, concentrated by precipitation with trichloroacetic acid (addition of 75 l of a 100% solution). After centrifugation at 10,000 ϫ g for 15 min at 4°C, pellets were resuspended in 50 mM Tris, pH 8.8 added with phosphatase inhibitors (10 mM NaF, 100 M Na 3 VO 4 , and 5 mM Na 4 P 2 O 7 ) and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 2 g/ml leupeptin, 1 g/ml pepstatin, and 2 g/ml aprotinin). Following determination of the protein content, fractions were dissolved in Laemmli loading buffer.

Immunocytochemistry and Confocal Laser Scanning Microscopy
Immunofluorescence Staining-Indirect immunofluorescence was performed on freshly isolated myocytes fixed with 4% formaldehyde at room temperature as described previously (26). Briefly, myocytes were incubated in phosphate-buffered saline containing 5% bovine serum albumin for 30 min to block nonspecific binding sites, followed by overnight incubation with a solution of mouse monoclonal antibodies against caveolin-3 (clone 26, 1:100) and rabbit polyclonal antibodies against SERCA 2a (1:100, see Ref. 27). After washes, cells were first incubated for 1 h with an excess of the secondary antibody goat antimouse IgG (HϩL) highly cross-absorbed coupled to Alexa Fluor® 594 (1:100). Myocytes were then washed and treated with the secondary antibody goat anti-rabbit IgG (HϩL) highly cross-absorbed coupled to Alexa Fluor® 647 (1:100) for 1 h. This was followed after washes by incubation with FITC-conjugated-phalloidin (1/30) for 90 min. After a final wash, coverslips were mounted in Vectashield mounting medium containing DAPI. Labeling of cells with secondary antibodies alone was carried out as a negative control.
Confocal Microscopy and Image Processing-Images were collected with a Zeiss LSM-510 multitracking laser scanning confocal microscope (Carl Zeiss SAS, Frankfurt, Germany). Fluorochromes were detected sequentially using laser lines 488 nm (FITC), 543 nm (Alexa Fluor® 594) and 633 nm (Alexa Fluor® 647). DAPI was visualized by UV excitation. Offset and gain for each channel was set in order to avoid any cross-talk between the three fluorochrome emission spectra. The images were coded white (FITC), red (Alexa Fluor® 594), and green (Alexa Fluor® 647) giving yellow co-localization in merge images (Alexa Fluor® 594 and Alexa Fluor® 647). The oil objective used was ϫ63 (NA 1.4), giving a resolution of 100 nm in the x,y plane and 300 nm in the z axis (pinhole, 72 m).
Quantification of Overlapping-We studied 10 individual myocytes and analyzed 55 Ϯ 2 successive single sections acquired in each image stack. Overlapping was quantified with the LSM 510 3 software (Carl Zeiss). The overlapping of SERCA 2a and caveolin-3 stainings in the x,y plane was determined owing to analysis of fluorescence intensity profiles as previously described by Bolte et al. (28). Positive structures were evaluated using threshold values of 152 Ϯ 17 (caveolin-3) and 144 Ϯ 5 (SERCA 2a) pixel units on a grayscale of 0 -255. A scatterplot of the individual pixels from each paired images was generated, and overlapped pixels were characterized by an overlap coefficient value of 1. The number of overlapped pixels was compared with the number of total positive pixels for each labeling (caveolin-3 or SERCA 2a).
Statistical Analysis-Results were analyzed by the unpaired twotailed Mann-Whitney test. Differences were considered statistically significant at a value of p Ͻ 0.05.

RESULTS
The ␤ 2 -AR/cPLA 2 Pathway Restrains the ␤ 2 -AR/AC/PKA-induced Ca 2ϩ and Contraction Responses in Isolated Cardiomyocytes-The amplitude of [Ca 2ϩ ] i transients was measured in electrically stimulated adult rat cardiomyocytes and loaded with Fura 2-AM, in response to the ␤ 2 -AR selective agonist, zinterol. Zinterol induced a dose-dependent stimulatory effect on the amplitude of [Ca 2ϩ ] i transients (maximal response of 182 Ϯ 11% with zinterol (1 M)) ( Fig. 1, A and B). The cPLA 2 inhibitor, AACOCF 3 (10 M), had no effect on basal [Ca 2ϩ ] i transient amplitude (Fig. 1B), but enhanced zinterol-induced Ca 2ϩ responses (maximal response of 307 Ϯ 28% with 1 M zinterol) (Fig. 1, A and B). Typical traces of [Ca 2ϩ ] i transients and fractional shortening showed the correlation between the increase in the amplitude of [Ca 2ϩ ] i transients and that of contraction in cardiomyocytes in response to zinterol added alone, or after treatment with AACOCF 3 (Fig. 1A). Inhibition of PKA with H89 (3 M) blunted the zinterol effect on the amplitude of [Ca 2ϩ ] i transients and contraction, in the absence (not shown) as well as in the presence of AACOCF 3 (Fig. 1A). These results identified PKA activation as an essential feature of zinterol-induced Ca 2ϩ and contraction responses, and demonstrated that the cPLA 2 pathway exerted a constraint on PKAdependent events.
The cPLA 2 Constraint Is Specific to ␤ 2 -AR and Does Not Affect  (Fig. 2). These results indicated that the constraint exerted by the cPLA 2 pathway selectively targeted the ␤ 2 -AR-induced Ca 2ϩ response, without affecting the ␤ 1 -AR-induced Ca 2ϩ response.
Involvement of a PTX-sensitive G Protein and MSK1 in the ␤ 2 -AR/cPLA 2 Inhibitory Pathway-In heart, PTX treatment ADP-ribosylates the ␣-subunits of G i and G o proteins, leading to their blockade in the trimeric form (␣␤␥) and their inhibition. As shown in Fig. 3, and in accordance with published data (29), PTX treatment potentiated the effect of zinterol (1 M) on the amplitude of [Ca 2ϩ ] i transients (330 Ϯ 32 versus 179 Ϯ 9% in PTX-untreated cells), mimicking AACOCF 3 effect. Furthermore, PTX treatment rendered the ␤ 2 -AR-induced Ca 2ϩ response insensitive to AACOCF 3 (Fig. 3). This suggested, in line with our previous study in human cardiac tissue (10), that a PTX-sensitive G protein-mediated stimulation of the cPLA 2 activity by ␤ 2 -AR agonists in rat.
Previous studies using purine agonists in adult rat cardiomyocytes identified cPLA 2 as a downstream target of the mitogen-activated protein kinase, MSK1, (30,31). To examine the role of MSK1 in ␤ 2 -AR induced Ca 2ϩ responses, we treated cardiomyocytes with a MSK1 inhibitor, RO 318220. RO 318220 did not affect basal amplitude of [Ca 2ϩ ] i transients (not shown) but potentiated the zinterol-induced Ca 2ϩ response, thus mimicking AACOCF 3 action. The effects of RO 318220 and AA-COCF 3 were not additive (Fig. 3). These results suggested that ␤ 2 -AR-induced activation of the cPLA 2 pathway relied on MSK1 stimulation.
The ␤ 2 -AR/cPLA 2 Pathway Restrains NOS Activity and PLB Phosphorylation-In cardiomyocytes isolated from cat atria, ␤ 2 -AR stimulation has been reported to induce an increase in the L-type Ca 2ϩ current (Ica,L) because of NO production (32,33). In the cat atrial cardiomyocytes, NO acts via cGMP-mediated inhibition of the type III phosphodiesterase (PDE) activity, enhancing cAMP-dependent stimulation of Ica,L. Among NO-generating enzymes expressed in the cardiomyocytes, eNOS can be activated upon PKA-or PI3K-dependent phosphorylation of the Ser 1177 residue (34). Because in adult rat cardiomyocytes, both PKA and PI3K activities are activated upon ␤ 2 -AR stimulation (35), we investigated whether zinterol did trigger NO production that could be amplified by AACOCF 3 treatment. Cells were loaded with the NO-sensitive fluorescent indicator, DAF-FM diacetate, and Fura 2-AM, to correlate NO and Ca 2ϩ responses. In rat cardiomyocytes that were electrically stimulated, no change in DAF fluorescence was detected within the first 20 min of incubation time following addition of zinterol alone (not shown). In contrast, exposure to zinterol combined with AACOCF 3 produced a marked increase in DAF fluorescence (Fig. 4A), that reached a mean 2.1 Ϯ 0.2-fold maximal increase in F/F 0 after 8 Ϯ 1 min. The increase in DAF fluorescence was blocked in the presence of the NO synthase inhibitor, L-NAME (not shown), and could thus be attributed to NO production. An increase in the amplitude of [Ca 2ϩ ] i transients occurred together with NO release in cells stimulated with zinterol plus AACOCF 3 . Interestingly, L-NAME treatment did not modify the increase in the amplitude of [Ca 2ϩ ] i transients in response to zinterol alone (Fig. 4B) but partially reversed the potentiation of zinterol-induced Ca 2ϩ responses elicited by AACOCF 3 (Fig. 4B). These results suggested that, in adult rat cardiomyocytes, basal zinterol-induced Ca 2ϩ responses were unrelated to NO production. In contrast, inhibition of the ␤ 2 -AR/cPLA 2 pathway by AACOCF 3 , uncovered NO release in response to zinterol. The latter took part in the mechanism of amplification of PKA-dependent Ca 2ϩ responses.
The activity of SERCA is essentially controlled by its inhibitor, PLB. Phosphorylation of PLB reverses this inhibition, thereby accelerating Ca 2ϩ uptake into the SR (for a review see Ref. 36). In adult rat cardiomyocytes, enhancement of ␤ 2 -AR activation of Ca 2ϩ signaling has been observed in conditions that promote PKA-dependent PLB phosphorylation on Ser 16 (35,37). We evaluated PLB phosphorylation on Ser 16 in quiescent cardiomyocytes exposed to zinterol, in the absence or in the presence of AACOCF 3. As reported previously, zinterol alone did not induce PLB phosphorylation (Fig. 5). But interestingly, we detected an AACOCF 3 -dependent phosphorylation of Ser 16 -PLB residue in response to zinterol (Fig. 5). This result indicated that cPLA 2 activation hindered PKA-dependent phosphorylation of PLB, thus limiting SERCA activation and ␤ 2 -AR-induced Ca 2ϩ and contraction responses. In contrast, control experiments performed with a ␤ 1 -AR selective stimulus, namely 100 nM isoproterenol added with 100 nM ICI 118,551, showed a marked PLB phosphorylation on Ser 16 , that was insensitive to AACOCF 3 treatment (not shown). Thus, the cPLA 2 constraint was specific to ␤ 2 -AR-and did not affect ␤ 1 -AR-induced Ser 16 -PLB phosphorylation. It should be noted that, in quiescent rat ventricular myocytes, nonspecific ␤-AR stimulation (in response to norepinephrine) has been shown to increase PKA-dependent phosphorylation of PLB at Ser 16 with little impact on the CaMKII-dependent phosphorylation of Thr 17 . In fact, in contrast with our previous observations in isolated spontaneously beating heart (18), we did not detect herein any phosphorylation of PLB on Thr 17 in response to either ␤ 2 -or ␤ 1 -AR stimulation, in the absence as well as in the presence of AACOCF 3 (not shown). Such an observation agrees with the reported pacing-dependence of the phenomenon (36, 38).

FIG. 4. The enhancing effect of AACOCF 3 on zinterol-induced
Ca 2؉ response is associated with NO production and abrogated by L-NAME treatment. Cardiomyocytes, isolated from adult rats, were loaded with Fura 2-AM and DAF-FM (A) or Fura2-AM alone (B), pretreated for 10 min, and maintained in cell medium with or without AACOCF 3 (10 M), and/or L-NAME (1 mM), as described under "Experimental Procedures." Cardiomyocytes were electrically stimulated at 0.5Hz and exposed to the

Integrity of Caveolae Is Required to Observe the cPLA 2 -dependent Constraint of ␤ 2 -AR-induced Ca 2ϩ
Responses-In adult rat cardiomyocytes, ␤ 2 -ARs are preferentially concentrated in caveolae, together with G proteins (i.e. G s , G i ), effectors (i.e. AC), kinases (i.e. PKA, protein kinase C, mitogenactivated protein kinase) and adaptor proteins (i.e. AKAP, NHERF). Importantly, such a localization has been shown to dictate ␤ 2 -ARs downstream signaling pathways (25,39,40). Caveolae are characteristic flask-shaped invaginations of the plasma membrane that are distinctively enriched in cholesterol (41). Next, experiments were designed to evaluate the impact of caveolar structures on the constraint exerted by the cPLA 2 pathway on ␤ 2 -AR/AC/PKA-induced Ca 2ϩ responses. We used Filipin III, a detergent that disrupts caveolar structures by binding to cholesterol. Cardiomyocytes loaded with Fura 2-AM were pretreated with Filipin III (0.2 g/ml) for 30 min, including a pretreatment with or without AACOCF 3 (10 M) during the last 10 min, and then exposed to zinterol (1 M). Filipin III by itself produced a 149 Ϯ 7% increase in the amplitude of basal [Ca 2ϩ ] i transients, that might reflect the accessibility of new targets to messengers (i.e. cAMP) or effectors (i.e. PKA) previously sequestrated inside caveolar structures (Fig. 6). Subsequent addition of zinterol (1 M) led to a final 198 Ϯ 10% increase. However, zinterol action was no more sensitive to activation by AACOCF 3 treatment (Fig. 6). These results suggested that, in adult rat cardiomyocytes, the integrity of caveolar structures was necessary for the cPLA 2 pathway to exert a constraint on zinterol-induced Ca 2ϩ responses. cPLA 2 Recruited to Caveolae/SR Platforms Hinders Zinterolinduced Phosphorylation of eNOS and PLB-We sought to further investigate the role of caveolae in ␤ 2 -AR signaling pathways. We incubated isolated adult cardiomyocytes with or without zinterol (1 M), in the presence or in the absence of AACOCF 3 . We prepared a low density caveolin-3-enriched membrane fraction using extraction in detergent-free alkaline sodium carbonate buffer followed by centrifugation in a continuous sucrose gradient, as described recently (25). As expected caveolin-3-enriched fractions were recovered in light sucrose gradient fractions that contained only a minor portion of total cellular protein (Fig. 7A).
Next, experiments were performed using the most enriched caveolin-3 fraction, namely fraction 7, which was separated because of a 32 Ϯ 1.5% sucrose density and corresponded to a mean 14 Ϯ 3% of the total cellular proteins (Fig. 7A). Identical quantities of proteins from fraction 7, isolated from cells submitted to the four different treatments (described above), were separated on preparative gels. Following electroblotting, membranes were cut in small pieces and analyzed, in parallel, for the presence of caveolae markers (caveolin-3 and eNOS), ␤ 2 -AR signaling effectors (MSK1, P-(Thr 581 )-MSK1 and cPLA 2 ), targets of the cPLA 2 constraint (P-(Ser 1177 )-eNOS and P-(Ser 16 )-PLB) and sarcoplasmic reticulum markers (PLB and SERCA 2a). As shown in Fig. 7, equal amounts of caveolin-3 were recovered in the four of the fraction 7 samples. cPLA 2 was exclusively detected in fraction 7 obtained from cardiomyocytes that have been treated with zinterol alone, illustrating the zinterol-induced translocation of the enzyme to the low density caveolin-3-enriched membrane fraction. Treatment with AA-COCF 3 hindered cPLA 2 translocation to such a fraction. Similar levels of MSK1 were detected in each fraction 7, but only cells incubated with zinterol, in the absence or in the presence of AACOCF 3 , displayed the MSK1-Thr 581 -phosphorylated form, a prerequisite for MSK1 activation. Taken together, these results supported the role of MSK1 in the ␤ 2 -AR/cPLA 2 pathway, upstream cPLA 2 activation in view of the insensitivity of the ␤ 2 -AR induced MSK1 phosphorylation toward AA-COCF 3 (see Fig. 3). FIG. 6. Treatment with Filipin III abrogates the enhancing effect of AACOCF 3 on zinterol-induced Ca 2؉ response. Cardiomyocytes, isolated from adult rats, were loaded with Fura 2-AM and electrically stimulated at 0.5 Hz. They were treated for 30 min with Filipin III (0.2 g/ml), and with or without AACOCF 3 (10 M) during the last 10 min of incubation. Cardiomyocytes were then exposed for 3 min to the ␤ 2 -AR agonist zinterol (1 M). The amplitude of [Ca 2ϩ ] i transients was normalized to control values determined at time 0, before the addition of Filipin III. Values are mean Ϯ S.E. of effects observed on at least six cells obtained from two different isolations. *, p Ͻ 0.05 versus control eNOS is a well known marker of caveolae, and the phosphorylation of eNOS at Ser 1177 is usually correlated with eNOS activation. Similar levels of eNOS were detected in the four fraction 7 samples. Interestingly, a large increase in eNOS phosphorylation level was observed, in response to zinterol added together with AACOCF 3 . Thus, zinterol-induced AA-COCF 3 -dependent production of NO (Fig. 4) could be related with phosphorylation and activation of eNOS, sedimented in the caveolin-3-enriched membrane fraction.
Interestingly, all caveolin-3-enriched fraction 7 obtained from the different treatments contained comparative amounts of PLB, with phosphorylation of PLB on Ser 16 detected only in response to zinterol added with AACOCF 3 , in accordance with results described in Fig. 5. Similarly, SERCA 2a was also detected in equal amounts in the different caveolin-3-enriched fraction 7. These results argued for an association between caveolae and a portion of SR membranes in adult rat cardiomyocytes, close enough to lead to their pull-down together during cellular fractionation.
We performed confocal microscopy in isolated adult rat cardiomyocytes immunostained with SERCA 2a and caveolin-3 antibodies in order to compare intracellular distribution of the two proteins (Fig. 8). Immunostained cardiomyocytes displayed a striated and regular actin pattern as illustrated by phalloidin staining (Fig. 8A, panel b). As previously reported in adult mouse cardiomyocytes (27,42), SERCA 2a exhibited a longitudinal and transverse distribution and was also concentrated around the nucleus (Fig. 8A, panel c). Caveolin-3 was present all along the peripheral cellular membrane, the intercalated disks and exhibited a punctuated distribution within the Ttubules (Fig. 8A, panel d). The merge image (Fig. 8A, panel a) showed partial overlapping of the two fluorescent stainings (yellow). Analysis of the overlapping of SERCA 2a and caveolin-3 stainings, using distribution of fluorescence intensities, as shown in the typical histogram in Fig. 8B, demonstrated a mean overlapping distance of 640 Ϯ 36 nm largely above the limits of resolution of the confocal microscope (100 nm). Overlapping pixels represented 25 Ϯ 5% of green pixels (SERCA 2a) FIG. 7. Zinterol induces translocation of the cPLA 2 to platforms integrating caveolin-3-enriched membranes, closely associated with sarcoplasmic reticulum membranes and allowing functional interaction between upstream activators (MSK1) and downstream targets (eNOS and PLB) of the cPLA 2 . Cardiomyocytes, isolated from adult rats, were pretreated (or not) for 10 min, maintained with AA-COCF 3 (10 M), and treated for 10 min with or without zinterol (1 M). A, cardiomyocytes obtained from each of four distinct conditions of incubation (control, zinterol, AACOCF3, AACOCF3 ϩ zinterol) were homogenized and fractionated according to a detergent-free purification scheme, as described under "Experimental Procedures." Samples (5 g of each gradient fraction) were subjected to SDS-PAGE. Membranes were immunostained with antibodies against caveolin-3, as described under "Experimental Procedures." The presence of caveolin-3, and the protein and sucrose distribution patterns were analyzed in parallel in the fractions corresponding to each gradient. Note that these patterns were equivalent under the four experimental conditions. A representative immunoblot is presented from experiments that have been repeated four times. B, caveolin-3-enriched fraction 7 of cardiomyocyte fractionations obtained from each of four distinct conditions of incubation was loaded on a preparative gel, as described under "Experimental Procedures." Each of the four membranes was cut in small pieces and analyzed, in parallel, for the presence of caveolae markers (caveolin-3, eNOS), upstream activators of the cPLA 2 (MSK1, P-(Thr 581 )-MSK1), downstream effectors of the cPLA 2 (P-(Ser 1177 )-eNOS, P-(Ser 16 )-PLB), and sarcoplasmic markers (SERCA 2a, PLB). Representative immunoblots are presented from experiments that have been repeated three times. and 43 Ϯ 7% of red pixels (caveolin-3) (Fig. 8C, area 3). In addition to the well known major distribution of the SERCA 2a along the longitudinal SR and around the nucleus, these results strongly supported the close vicinity of a significant portion of SERCA 2a-enriched membranes (SR) with a portion of caveolin-3-enriched membranes (caveolae).
Taken together, these results show that cPLA 2 limits phosphorylation of eNOS and PLB via its translocation toward caveolae/SR functional platforms, gathering ␤ 2 -AR signaling effectors. DISCUSSION The new findings of this study are 3-fold. First, we show that ␤ 2 -AR, through stimulation of a PTX-sensitive G protein and MSK1, evokes cPLA 2 activation. Second, we give evidence that ␤ 2 -AR-induced cPLA 2 activation is linked to the translocation of the enzyme to low density caveolin-3-enriched membrane fractions that are closely associated with SR membranes and constitute functional signaling platforms. Inside those platforms we identified two components of the ␤ 2 -AR/cPLA 2 pathway, namely eNOS and PLB, showing that the blockade of cPLA 2 unmasks zinterol-induced phosphorylation of eNOS and PLB. Finally, we demonstrate that the ␤ 2 -AR/cPLA 2 pathway, through inhibition of NOS phosphorylation and NO production, associated with the inhibition of PLB phosphorylation, selectively constrains the ␤ 2 -AR/AC/PKA-induced Ca 2ϩ response.
In rat cardiomyocytes, cPLA 2 has been identified as a target of MSK1 in response to purinergic agonists stimulation (43,44). We report for the first time that ␤ 2 -AR agonists trigger phosphorylation of MSK1 that we find associated within caveolin-3/SR platforms.
Caveolae constitute a unique endocytic and exocytic compartment of the cell membrane, capable of importing molecules, delivering them to specific locations within the cell, and compartmentalizing a variety of signaling activities (41). Caveolae are a site of Ca 2ϩ storage and entry into the cell. Our study provides evidence for ␤ 2 -AR-induced translocation of the cPLA 2 to platforms constituted of caveolae tightly connected to SR membranes. PLA 2 activity has been implicated in constitutive membrane trafficking (45); however, only one recent study reported the presence of the cPLA 2 in caveolin 1-enriched membrane fractions isolated from hippocampal preparations (46). Most caveolin-interacting proteins identified so far contain a caveolin-binding motif located within their enzymatically active catalytic site (41). As a matter of fact, such a caveolinbinding motif is present within the catalytic domain of the cPLA 2 . Nevertheless, up to date, activation of cPLA 2 has been essentially associated with its Ca 2ϩ -dependent translocation from the cytosol to intracellular membranes, including perinuclear, endoplasmic, or SR membranes, but not to plasma membranes (47). We observe that treatment with AACOCF 3 abrogates the association of the cPLA 2 to caveolin-3/SR-platforms. One possible explanation relies on recent data showing that the cPLA 2 catalytic domain modulates membrane association and membrane residence time of the enzyme (48). Because AA-COCF 3 is expected to target the catalytic site of the enzyme, competing with endogenous substrates, it might also affect interaction between caveolin-3 or SR membranes and cPLA 2 .
This study describes the cosedimentation of caveolae markers (caveolin-3 and eNOS) with SR markers (PLB and SERCA  5 m). C, typical scatterplot of the individual pixels from paired images. The threshold levels of green (139) and red (169) signals determined the overlapping region (area 3); areas 2 and 1 corresponding to green and red pixels, respectively, with no color mixing. 2a), suggesting the possible close interactions between the two types of membranes in cardiomyocytes. It is noteworthy that SR Ca 2ϩ uptake is reported to take place mainly in the longitudinal SR (away from caveolae), which contains the highest SERCA 2a density (49). In contrast, most of the SR Ca 2ϩ release would occur in proximity of the terminal cisternae (junctional and corbular SR), close to the T-tubules (49), where ryanodine receptors are concentrated (50). We provide confocal microscopic support for the major distribution of the SERCA 2a along the longitudinal SR and around the nucleus in adult rat cardiomyocytes. However, our experiments also demonstrate the close vicinity of a pool of SERCA 2a with caveolin-3-enriched membranes. These results are in agreement with the confocal microscopic study of Vangheluwe et al. (27), in adult mice cardiomyocytes, describing a close vicinity between SERCA and ryanodine receptor molecules, in an area close to the T-tubules. A similar localization of SERCA has been previously reported by Greene et al. (42). Interestingly, recent ultrastructural studies also describe the vicinity of the SR membranes with caveolae in airway smooth muscle cells (51), and caveolae are proposed to provide a platform of interaction between SR and plasmalemmal ion channels (52).
This study suggests that all components of the ␤ 2 -AR-cAMP pathway are located in caveolin-3/SR-interacting platforms where they can also be regulated by other signaling pathways, such as cPLA 2 . The potentiating effect of AACOCF 3 , which neutralizes cPLA 2 action, seems to rely on phosphorylating of target proteins already present inside such platforms rather than on a recruitment of effectors toward platforms. It is noteworthy that caveolae have been reported to retain eNOS in the inactive state, because eNOS activity is inhibited by binding to the caveolin scaffold domain. However, our results identify the phosphorylated form of eNOS in the caveolin-3-enriched membrane fraction, which argues for its activation inside caveolae. In fact, parallel imaging studies show that adult rat cardiomyocytes treated with AACOCF 3 plus zinterol produce NO.
Our findings that ␤ 2 -AR-induced cPLA 2 activation impairs ␤ 2 -AR-induced NO release agree with the previously reported inhibition of NO release by AA in PC12 cells (53). NO release, evoked by ␤ 2 -AR agonists in the presence of the cPLA 2 inhibitor, AACOCF 3 , is associated with an augmented Ca 2ϩ response. Data from the literature point out the complexity of the modulatory effects of NO on the contractile function. NO influences the positive ␤-adrenergic inotropic effect in a bimodal fashion, depending not only on its concentration but also on that of catecholamines (54). And the group of Lipsius, in particular, reported an activatory role of NO on ␤ 2 -AR-mediated contractile effects (32,33). Several mechanisms can underlie the potentiation of the Ca 2ϩ response by NO, including inhibition of the cG i -PDE that would favor cAMP pathways (55), and nitrosylation of the L-type Ca 2ϩ channel (56) and the ryanodine receptor that would stimulate channel activities (57). Furthermore, through nitrosylation of enzymes of the respiratory complexes, NO enhances mechanoenergetic coupling (58). It should be noted that potentiation by NO of the ␤ 2 -AR/PKA-induced Ca 2ϩ response, through the blockade of the cPLA 2 pathway, described herein in isolated adult rat cardiomyocytes, is not observed in cat atrial myocytes. In the latter, detectable NO production in response to zinterol does not require the blockade of cPLA 2 (32,33).
Xiao and co-workers (35) previously identified PI3K as a downstream target of ␤ 2 -AR signaling, in adult rat cardiomyocyte. Our study highlights clear homology between ␤ 2 -AR/PI3K and ␤ 2 -AR/cPLA 2 pathways since they both confine and negate concurrent ␤ 2 -AR/G s -mediated PKA signaling, without affecting ␤ 1 -AR-induced responses. PI3K inhibition, as well as cPLA 2 inhibition, enables the ␤ 2 -AR/PKA signaling to reach and to phosphorylate intracellular substrates such as phospholamban, and markedly enhance the ␤ 2 -AR-positive contractile response in cardiac myocytes. Likewise, both potentiating effects are mediated through a PTX-sensitive G protein and are not accompanied by an increase in cAMP formation. Because of their similar impact on ␤ 2 -AR/PKA signaling, a link between the two mechanisms cannot be ruled out. One hypothesis would be that both pathways share a common effector. One of the potential targets of arachidonic acid might be the protein phosphatase 5, a Ser/Thr phosphatase expressed in the heart, that is activated by lipids and calyculin A (59 -61). Note that Xiao and co-workers have suggested that ␤ 2 -AR-induced PI3K activation resulted in the stimulation of a calyculin-sensitive protein phosphatase activity. However, a number of studies argue in favor of the divergence of cPLA 2 and PI3K signaling pathways. Thus, studies on transgenic mice have shown that ␤ 2 -AR signaling bifurcates at the level of two different PTX-sensitive G i proteins, G␣ i2 and G␣ i3 , leading to different effects depending of the G␣ isoform. G␣ i2 takes an essential protective part in the chronic signaling of overexpressed ␤ 2 -AR, leading to prolonged survival and delayed cardiac pathology. In contrast, G␣ 3 is supposed to mediate ␤ 2 -AR induced reduction of Ca 2ϩ channel activity (8), potentially through the activation of the PI3K␥ isoform, a downstream target of G i3 -␤␥ signaling (32,56). Knockout of the PI3K␥ gene results in a net increase in cardiac contractility without impact on cardiomyocyte hypertrophy (35,62). In contrast, the major role attributed to the cPLA 2 is a critical regulation of muscular cell size (16), and the knockout of the cPLA 2 gene produces cardiac hypertrophy without change in basal cardiac contractility. Studies performed in cardiomyocytes indicate that activation of PI3K stimulates rather than inhibits NO production (63,64). In this report, we demonstrate that eNOS inhibition is an essential component of the cPLA 2 limiting action.
Our present study in adult rat ventricular cardiomyocytes further substantiate the major role in the heart of the ␤ 2 -AR/ cPLA 2 pathway (65), beyond that of the ␤ 2 -AR/AC/PKA pathway, a role that we previously suspected from studies in human tissues (10) and embryonic chick cardiomyocytes (23,66). It appears that the respective roles of each, ␤ 2 -AR/cPLA 2 and ␤ 2 -AR/AC/ PKA pathways, depend on the species or the pathological state. Thus, in adult rat cardiomyocytes, the ␤ 2 -AR/AC/PKA pathway mediates the Ca 2ϩ response triggered by ␤ 2 -AR agonists, and is constitutively depressed because of the concomitant activation of the ␤ 2 -AR/cPLA 2 pathway. In contrast, in embryonic chick heart cells, cPLA 2 activation exclusively mediates ␤ 2 -AR stimulation of [Ca 2ϩ ] cycling and cell contraction, (23,66). In human tissue, ␤ 2 -AR activates both AC and cPLA 2 activities, the ␤ 2 -AR/cPLA 2 pathway being favored under conditions of altered ␤ 2 -AR/AC/ PKA signaling (10). Taken together, those data pose the question as to the contractile impact of the cPLA 2 in human heart, and its possible evolution in the course of heart failure or aging that both represent physiopathological situations associated with a degradation of the ␤ 2 -AR/AC coupling (1). Another remaining important question is whether activation of the cPLA 2 takes part in the cardiac protective effect of ␤ 2 -AR stimulation, with eNOS inhibition as an essential component of its limiting action on the AC/ PKA pathway.