Evidence for a β2-Adrenergic/Arachidonic Acid Pathway in Ventricular Cardiomyocytes

The signaling pathway mediating the contractile effect of β2-adrenergic receptors (β2-AR) in the heart is still matter of debate. By using embryonic chick ventricular cardiomyocytes that express both functional β1-and β2-ARs, we show here that the specific β2-AR agonist, zinterol, increases the amplitude of Ca2+ transients and cell contraction of electrically stimulated cells. Zinterol, up to 10 μm, did not stimulate adenylyl cyclase activity, and its effect on Ca2+transients was unmodified by the specific cAMP antagonist, (Rp)-cAMPS. In contrast, zinterol (10–100 nm) triggered arachidonic acid (AA) release from [3H]AA-loaded cells via the activation of the cytosolic phospholipase A2 (cPLA2). Stimulation of the Ca2+ transients by zinterol was abolished by the cPLA2 inhibitor, AACOCF3, and was mimicked by AA (0.3–3 μm). Both stimulations of [3H]AA release and of [Ca2+] i cycling by zinterol were abolished after treatment of the cardiomyocytes with pertussis toxin. Although cell responses to β2-AR stimulation were mediated by AA, they were under cAMP control as follows: (i) the β1-AR stimulation exerted a cAMP-mediated negative constraint on the β2-AR/cPLA2 pathway; (ii) cAMP potentiated AA action downstream β-AR stimulation. We conclude that, in cardiomyocytes, β2-AR is coupled to cPLA2 activation via a pertussis toxin-sensitive G protein. These results demonstrate the involvement of the cPLA2/AA pathway in mediating positive inotropic effects, which could potentially compensate for a defective cAMP pathway.

␤ 1 -and ␤ 2 -Adrenergic receptors (␤ 1 -and ␤ 2 -ARs) 1 coexist in the hearts of various animal species, including humans. However, their relative amount and their respective participation in the positive chronotropic and inotropic effects of adrenaline and noradrenaline vary depending on the cardiac tissue, the animal species, and/or the pathophysiological state (1,2). In the non-failing human left ventricle, ␤ 1 -ARs represent 80% of the total ␤-ARs but mediate about 60% only of ␤-adrenergicinduced ventricular contractility (3). In the human failing heart, the ␤ 1 /␤ 2 -AR ratio decreases, and the contribution of ␤ 2 -AR to the contractile responses becomes predominant over that of ␤ 1 -AR, in particular at low adrenaline concentrations (3,4). For these reasons, the potential role of ␤ 2 -AR for improving cardiac performance has received considerable attention. In fact, the myocardial-targeted overexpression of ␤ 2 -ARs in transgenic mice significantly enhanced myocardial left ventricular contractility (5).
It is well documented that ␤ 1 -AR and ␤ 2 -AR subtypes are coupled to adenylyl cyclase activation and that stimulation of both receptors generally leads to an increase in cellular cAMP (4,6,7). In human healthy heart, ␤ 2 -ARs are more efficiently coupled to adenylyl cyclase than ␤ 1 -ARs (6 -10). However, during cardiac failure, ␤ 2 -AR subtypes are partially uncoupled from adenylyl cyclase (6,7), whereas their contribution to the positive inotropic effects of adrenaline and noradrenaline is increased to 63% (7). In addition, studies in the rat heart (11,12) and in the non-failing and failing canine heart (13) have demonstrated a dissociation between the inotropic effect of ␤ 2 -AR and cellular cAMP increase. Based on those observations, Xiao et al. (12) proposed that unidentified signal transduction pathway(s), other than adenylyl cyclase and cAMP, could be involved in the cardiac inotropic response to ␤ 2 -AR stimulation.
Angiotensin II (14,15), bradykinin (16,17), and endothelin (15,18), which exert positive inotropic responses, evoke AA release in heart. Furthermore, in a recent study, we have demonstrated that glucagon action relies not only on cAMP but also on the synergistic support of AA, by activation of the cPLA 2 which hydrolyzes the sn-2 fatty acyl ester bonds of membranous phospholipids (15).
The aim of the present study was to investigate the respective role of cAMP and AA in the cardiac response to ␤-adrenergic agonists. We used the model of embryonic chick ventricular cardiomyocytes that has been widely exploited for studies on metabolism, contractile physiology, electrophysiology, and examination of pathophysiologic states such as ischemia (19). We show that those cells, in addition to expressing ␤ 1 -AR (19,20), also respond to ␤ 2 -AR stimulation. We compared the ␤ 1 -and ␤ 2 -AR-mediated effects on adenylyl cyclase, [Ca 2ϩ ] i transients, cell contraction, and AA release. Our results demonstrate that cAMP is the messenger of ␤ 1 -AR responses. In contrast, cell responses to ␤ 2 -AR stimulation were mediated by AA but under cAMP control.
Fura-2 Loading and [Ca 2ϩ ] i Imaging-Cells were plated on plastic dishes, the bottom of which was replaced by a glass coverslip coated with laminin (1 g/ml), and were incubated at 37°C in humidified 5% CO 2 , 95% air for 17-24 h.
Cells, attached to laminin, were bathed in 2 ml of saline buffer B (10 mM glucose, 130 mM NaCl, 5 mM KCl, 10 mM Hepes buffered at pH 7.4 with Tris base, 1 mM MgCl 2 , 2 mM CaCl 2 ) and were incubated for 20 min at 25°C with 1.5 M Fura-2/AM (3 l of 1 mM Fura-2/AM in Me 2 SO), in the presence of 1 mg/ml bovine serum albumin to improve Fura-2 dispersion and facilitate cell loading. Cells were then washed with saline buffer B (2 ϫ 2 ml) and allowed to incubate in the same buffer for 15 min at 25°C to facilitate hydrolysis of intracellular Fura-2/AM. The concentration of Fura-2 in myocytes was estimated as described previously (15,21,22), according to the procedure of Donnadieu et al. (23). Under usual loading conditions, the average intracellular concentration of Fura-2 was 15 M. Ca 2ϩ imaging, developed by A. Trautman in collaboration with the IMSTAR (Paris, France), was essentially as described by Sauvadet et al. (21). Field electrical stimulation (square waves, 10-ms duration, amplitude 20% above threshold, 0.5 Hz) was supplied through a pair of platinum electrodes connected to the output of an HAMEG stimulator (Paris, France). Cells were perifused with saline buffer B containing 2 mM CaCl 2 and stimulated until a steadystate level of the Ca 2ϩ transients was achieved, before addition of drugs and peptides to the perfusion medium.
Contractility Measurements-Experiments were performed in conditions similar to Ca 2ϩ imaging, but cells were illuminated with visible light and images transmitted through a solid-state camera (CCD, black and white, 0.847-cm high sensitivity) connected to the sideport of the microscope, as described previously (15). Contractions of single stimulated (0.5 Hz) myocytes were displayed on a video monitor, and the corresponding images (pixel ϫ pixel) were recorded at a frequency of 9/s. Contractility measurements were determined by assessing changes in cell length using the Morphostar II software, developed by the IM-STAR (Paris, France).
Adenylyl Cyclase Assay-A particulate fraction of embryonic chick ventricular cardiomyocytes was obtained from cells washed twice in saline buffer B, disrupted by sonication, and centrifuged for 30 min at 30,000 ϫ g. The pellet was resuspended in 50 mM Hepes, pH 7.4, and stored in liquid nitrogen. Adenylyl cyclase activity was measured as described previously (24). The assay medium contained, in a final volume of 60 l, 50 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 1 mM EDTA, 1 mM [␣-32 P]ATP (10 6 cpm), 1 mM [8-3 H]cAMP (20,000 cpm), 50 M GTP, 0.2 mM methylisobutylxanthine, 25 mM creatine phosphate, 1 mg/ml creatine kinase. The incubation was initiated by the addition of 20 -70 g of proteins and run at 37°C. The reaction was terminated by adding 0.2 ml of 0.5 N HCl. Samples were boiled for 6 min and thereafter buffered with 0. H]AA-labeled cells were exposed to various peptides and/or enzymatic inhibitors and incubated for various periods at 37°C. Incubation was terminated by the addition of ice-cold EGTA (2 mM final), and the media were immediately transferred to microcentrifuge tubes. Centrifugation at 17,600 ϫ g for 20 min in a Sigma centrifuge (model 2K15) at 4°C was performed to pellet any cells or debris inadvertently collected with the extracellular medium. The amount of radioactivity in the supernatant was quantitated by liquid scintillation counting.
Analysis of the lipids released in the incubation medium was performed as described (26). At the end of the incubation period, the reaction mixture was acidified to pH 3.0 with HCl, and the products were extracted twice with ethyl acetate. The dried extracts were dissolved in ethanol/chloroform (1:2, v/v) and chromatographed on silica gel thin layer plate (Whatman LK5) in ethyl acetate/isooctane/water/ acetic acid (11:5:10:2, v/v) as the solvent system. Standard concentrations of AA, prostaglandins E, and hydroxyeicosatetraenoic acids were co-chromatographed and visualized by exposing the plates to ultraviolet light. The area corresponding to each visualized spot was carefully extracted, and the radioactivity was determined by liquid scintillation counting.

␤ 1 -and ␤ 2 -AR Stimulations Increase [Ca 2ϩ ] i Cycling and Contractility in Electrically Stimulated Embryonic Chick Ventricular Myocytes-
The effect of increasing concentrations of isoproterenol was examined on [Ca 2ϩ ] i cycling of electrically stimulated embryonic chick ventricular myocytes. A dose-dependent increase in the amplitude of [Ca 2ϩ ] i transients was observed, reaching a maximal (210 Ϯ 9%) stimulation at 1-10 M isoproterenol (Fig. 1). Preincubation for 10 min with 100 nM of the selective ␤ 2 -AR antagonist, ICI 118551 (7), significantly reduced the stimulation evoked by 10 M isoproterenol (26% inhibition) but was poorly effective in inhibiting the effect of lower concentrations. In contrast, 300 nM of the selective ␤ 1 -AR antagonist, CGP 20712A (7), markedly blocked the effect of low isoproterenol concentrations, leading to a rightward-shifted dose-response curve of isoproterenol effect. CGP 20712A also reduced by 54% the maximal effect of 10 M isoproterenol ( Fig.  1). It thus appeared that isoproterenol behaved as a ␤ 1 -AR agonist at concentrations below 100 nM, and as a mixed ␤ 1 / ␤ 2 -AR agonist in the micromolar range of concentrations.
Both ␤ 1 -and ␤ 2 -AR stimulatory effects on [Ca 2ϩ ] i cycling were correlated with increases in the amplitude of cell contraction; 30 nM zinterol (␤ 2 -AR agonist) and 100 nM isoproterenol (at a concentration at which the agonist functioned as a ␤ 1 -AR agonist) increased the amplitude of cell contraction by 80 and 150% over basal, respectively (Fig. 3, A and B). Furthermore, as shown in the normalized and superimposed tracings of contraction (Fig. 3C), zinterol, like isoproterenol, markedly accelerated the kinetics of relaxation. Isoproterenol, at 10 M, elicited a maximal 1.8-fold stimulation of adenylyl cyclase activity, the half-maximal effect being obtained at 0.15 M isoproterenol (Fig. 4A). This effect was totally blocked by 300 nM of the ␤ 1 -AR antagonist, CGP 20712A. Under the same assay conditions, zinterol had no effect on adenylyl cyclase activity (Fig. 4A). These results suggest that, in embryonic chick ventricular cardiomyocytes, adenylyl cyclase is specifically coupled to ␤ 1 -ARs but not to ␤ 2 -ARs.
Resolution on thin layer chromatography (TLC) of the 3 Hlabeled material, released in cell supernatants, identified [ 3 H]AA as the major product, both in control and zinteroltreated cells (75 and 79%, respectively) ( Table I). Non-enzymatic degradation or contaminants of standard [ 3 H]AA represented 4 -19% of the total radioactivity recovered in supernatants; lipoxygenase and cycloxygenase products represented 4 -24 and 1-8%, respectively (Table I).
Since AA formation in the heart is essentially attributed to PLA 2 activity, we examined the effect of AACOCF3, a specific inhibitor of the cPLA 2 . The addition of 10 M AACOCF3 to the perfusion medium dramatically reduced [ 3 H]AA release evoked by zinterol (from 206 to 136% of control [ 3 H]AA release, Table  I). This inhibitory effect correlated with a blockade of the stimulatory effects on [Ca 2ϩ ] i transients of both specific ␤ 2 -AR agonists zinterol and fenoterol (Fig. 5B). In contrast, AA-COCF3 did not affect the ␤ 1 -AR-mediated increase in [Ca 2ϩ ] i cycling triggered by either prenalterol or isoproterenol at 100 nM (Fig. 5B). Taken together, these findings further supported the notion that ␤ 2 -AR stimulation elicited AA release by stimulating the cPLA 2 , sensitive to AACOCF3. It may be noted that AACOCF3 completely inhibited ␤ 2 -AR-mediated effects on [Ca 2ϩ ] i cycling, whereas it had only a partial effect on ␤ 2 -ARstimulated AA release (Table I). This may suggest that the onset of the Ca 2ϩ response requires the cellular AA level to reach a threshold.
Importantly, exogenous application of micromolar concentrations of AA reproduced the effect of ␤ 2 -AR agonists on [Ca 2ϩ ] i transients; at 3 M, AA evoked a 140% increase in amplitude of [Ca 2ϩ ] i transients (Fig. 6). The activating effect of AA on [Ca 2ϩ ] i cycling was potentiated by 8-Br-cAMP (Fig. 6).
The ␤ 1 -AR/cAMP Pathway Occludes Cell Responses to ␤ 2 -AR Stimulation-Next, we looked for a possible cross-talk between ␤ 1 -and ␤ 2 -AR responses. In a first series of experiments, cells were electrically stimulated and exposed to 300 nM of the ␤ 1agonist, prenalterol. The time course of the amplitude of [Ca 2ϩ ] i transients is illustrated in Fig. 7. [Ca 2ϩ ] i transient amplitude increased for the first minutes of exposure to prenalterol, reaching a maximal 50% increase over basal at 10 min. After 15 min, a decline in stimulation of [Ca 2ϩ ] i cycling occurred, and after 30 min, the ␤ 1 -AR-mediated effect was no more detectable (Fig. 7). Such a waning of a stimulated response in the face of continuous agonist exposure is typical of a desensitization phenomenon (29).
In a second series of experiments, we examined the response to ␤ 2 -AR stimulation of cells under two extreme conditions: (i) after 3 min exposition to prenalterol, when ␤ 1 -ARs are fully activated; (ii) after 45 min exposition to prenalterol, when ␤ 1 -ARs are desensitized. As shown in the insets of Fig. 7, after 8-Br-cAMP reproduced prenalterol effect and inhibited the ␤ 2 -AR-mediated effects on [Ca 2ϩ ] i cycling (Fig. 8A). In addition, the cAMP antagonist, (Rp)-cAMPS, as well as the PKA inhibitor, H89, blocked the inhibitory effect of the ␤ 1 -AR agonist, prenalterol, on the cell response to ␤ 2 -AR stimulation (Fig. 8, B and C). Taken together, those data suggest that cAMP, via PKA activation, exerts an inhibitory constraint on ␤ 2 -AR stimulation. cPLA 2 Activation by ␤ 2 -AR Agonists Is Sensitive to Pertussis Toxin Treatment-␤ 2 -AR can couple to both G s and G i proteins (30,31). Thus, we investigated the possible role of G i in the specific coupling of ␤ 2 -AR to cPLA 2 . Treatment of the cells with PTX totally abolished the stimulatory effects of zinterol on both [ 3 H]AA release and [Ca 2ϩ ] i cycling (Fig. 9, A and B). The efficiency of PTX treatment was checked by the blockade of G i -mediated acetylcholine inhibition of isoproterenol effect on Ca 2ϩ cycling (Fig. 9B). It should be noted that treatment with pertussis toxin was without detectable impact on basal or isoproterenol-stimulated [Ca 2ϩ ] i transients suggesting the absence of a tonic control by G i , in particular over G s . DISCUSSION In the present study, we show that ␤ 1 -and ␤ 2 -ARs are both expressed in embryonic chick ventricular cardiomyocytes, and this model allowed us to demonstrate the following: (i) ␤ 2 -ARs are specifically coupled to cPLA 2 via a G i protein; (ii) cAMP exerts a dual tuning on cell responses to ␤ 2 -AR stimulation.
In electrically stimulated embryonic chick ventricular cardiomyocytes, 30 -100 nM zinterol, a specific partial ␤ 2 -AR agonist, elicited a 40 -50% increase over basal in the amplitude of [Ca 2ϩ ] i transients (Fig. 2), which correlated with increases in twitch amplitude and twitch velocity (Fig. 3). Such a positive inotropic effect of ␤ 2 -AR agonists is undisputed. Nevertheless, in contrast to ␤ 1 -AR-mediated positive inotropic effect, which definitely relies on a rise in intracellular cAMP, the contribu-tion of cAMP to the positive inotropic effect of ␤ 2 -AR agonists, and the possible coupling of ␤ 2 -AR to cAMP-independent pathways, are still a matter of debate. According to Bristow et al. (6), ␤ 2 -ARs in the non-failing human heart are tightly coupled to adenylyl cyclase since a numerically small ␤ 2 -AR fraction (19% of the total ␤ 1 -and ␤ 2 -ARs) accounts for the majority of adenylyl cyclase stimulation. Such an inherent efficacy for the human ␤ 2 -AR in activating adenylyl cyclase, compared with that of its ␤ 1 counterpart, has been confirmed by expression of those receptors in fibroblast cell lines (9,10). However, Kaumann and Lemoine (7) have compared the relative contribution of ␤ 1 -and ␤ 2 -ARs to adenylyl cyclase stimulation and positive inotropic effects of adrenaline and noradrenaline in pathological human heart. They concluded that the positive inotropic response was not straightforwardly correlated to adenylate cyclase stimulation. These authors were also the first to suggest compartmentation of cAMP since cAMP produced upon ␤ 2 -AR stimulation was less efficiently used than cAMP produced upon ␤ 1 -AR stimulation by cellular effectors involved in contractility. More recently, the group of Lakatta (12) suggested that, in addition to coupling to adenylyl cyclase, ␤ 2 -AR stimulation activates other signal transduction pathways to produce changes in [Ca 2ϩ ] i and contraction. This proposal relies on two observations. First, in rat ventricular cells ␤ 2 -AR stimulation elicits a positive inotropic response that is dissociated from cAMP increase (12). Evidence for the involvement of cAMP is given only for high ␤ 2 -agonist concentrations; indeed, activation by 10 M zinterol of both contraction (32) and L-type Ca 2ϩ current (2) is blocked by (Rp)-cAMPS, the specific inhibitory cAMP analog. Second, in electrically stimulated dog myocytes, ␤ 2 -AR activation is ineffective in stimulating adenylyl cyclase, whereas it produces increases in [Ca 2ϩ ] i transient and twitch amplitudes (13). In this regard, we show here that, in embryonic chick heart cells, ␤ 2 -AR stimulation by zinterol triggers a positive inotropic effect, independent of adenylyl cyclase activation (Figs. 2, 3, and 4A). The absence of participation of cAMP in this effect of zinterol is further confirmed by the fact that, in contrast to the actions of ␤ 1 -AR agonists, it is not blocked by either (Rp)-cAMPS (Figs. 4B and 8B) or the PKA inhibitor, H89 (Fig. 8C). Thus, we conclude that cAMP does not support the inotropic effect of low ␤ 2 -AR agonist concentrations although it could contribute in the effects of high ␤ 2 -AR agonist concentrations.
Glucagon action in heart relies on the synergistic actions of glucagon itself and its metabolite (19 -29), mini-glucagon (15,22). We have demonstrated that cAMP mediates glucagon action and that AA is the second messenger of mini-glucagon (15). In the present study, several lines of evidence support the proposal that AA is also the second messenger in response to stimulations by ␤ 2 -AR agonists: 1) zinterol increases AA release from [ 3 H]AA-prelabeled myocytes in a dose-dependent manner from 3 to 100 nM (Fig. 5A); 2) AA, added to the cell medium at concentrations as low as 1-3 M, reproduces the effect of zinterol on [Ca 2ϩ ] i cycling in electrically stimulated cardiomyocytes (Fig. 6). AA release results from ␤ 2 -AR activation of the cPLA 2 via a pertussis toxin-sensitive G protein (Fig.  8). Such a coupling of ␤ 2 -ARs to pertussis toxin-sensitive G protein(s) has been already reported in rat cardiomyocytes (30) and in cells sur-expressing ␤ 2 -ARs (31).
cAMP exerts a dual tuning on cell responses to ␤ 2 -AR stimulation. On the one hand, we show that cAMP, produced upon ␤ 1 -AR stimulation, evokes a quenching of cell responses to ␤ 2 -AR stimulation; thus, after 3 min exposure to prenalterol, cells did not respond further to ␤ 2 -AR stimulation, whereas following complete desensitization of ␤ 1 -ARs, ␤ 2 -AR stimulation was restored (Fig. 7). The negative constraint exerted by cAMP is likely to rely on PKA stimulation since H89, the PKA inhibitor, hampers it. It could be due to phosphorylation, and inhibition, by protein kinase A of the PTX-sensitive G protein coupling cPLA 2 to ␤ 2 -AR (33). On the other hand, downstream cPLA 2 activation, cAMP potentiates AA-mediated stimulation of [Ca 2ϩ ] i cycling (Fig. 6). Those synergistic actions of cAMP and AA would rely on the ability of AA to accumulate Ca 2ϩ into the sarcoplasmic reticulum stores and that of cAMP to induce "Ca 2ϩ -induced Ca 2ϩ release" from these stores (15).
In conclusion, we show that, at low concentrations, ␤ 2 -AR agonists elicit a positive inotropic effect via cPLA 2 activation and AA release. Contrary to the ␤ 1 -AR/cAMP pathway, ␤ 2 -AR/ cPLA 2 pathway involves a pertussis toxin-sensitive G protein. cAMP exerts a dual regulation on the ␤ 2 -AR/AA pathway; it inhibits the cell response to ␤ 2 -AR stimulation but potentiates AA-mediated stimulation of [Ca 2ϩ ] i cycling.
There is now accumulating evidence that hydrolytic products derived from membrane phospholipids play important roles in cardiovascular signaling (34). Originally, attention mainly focused on diacylglycerol and eicosanoids (prostaglandins, prostacyclins, thromboxanes, leukotrienes, etc.) (35,36). However, the list of bioactive lipidic molecules now includes AA, the precursor of eicosanoids. Studies on mice deficient in cPLA 2 have demonstrated the major role of this enzyme in allergic responses, reproductive physiology, and pathophysiology of neuronal death (37,38). The participation of cPLA 2 and/or AA in mediating positive inotropic response to various agents was suspected (14 -18). Our results unequivocally establish that, at low concentrations of agonist, the ␤ 2 -AR-mediated inotropic effect relies on the selective activation of cPLA 2 and AA release.
Although this remains to be demonstrated, it is tempting to speculate that the ␤ 2 -AR/cPLA 2 /AA pathway could be determinant in failing hearts that have lost 50% of ␤ 1 -ARs and show a parallel decrease in agonist-stimulated adenylyl cyclase activity (4, 6, 7).