Gi Protein-mediated Functional Compartmentalization of Cardiac β2-Adrenergic Signaling*

In contrast to β1-adrenoreceptor (β1-AR) signaling, β2-AR stimulation in cardiomyocytes augmentsl-type Ca2+ current in a cAMP-dependent protein kinase (PKA)-dependent manner but fails to phosphorylate phospholamban, indicating that the β2-AR-induced cAMP/PKA signaling is highly localized. Here we show that inhibition of Gi proteins with pertussis toxin (PTX) permits a full phospholamban phosphorylation and a de novo relaxant effect following β2-AR stimulation, converting the localized β2-AR signaling to a global signaling mode similar to that of β1-AR. Thus, β2-AR-mediated Gi activation constricts the cAMP signaling to the sarcolemma. PTX treatment did not significantly affect the β2-AR-stimulated PKA activation. Similar to Gi inhibition, a protein phosphatase inhibitor, calyculin A (3 × 10−8 m), selectively enhanced the β2-AR but not β1-AR-mediated contractile response. Furthermore, PTX and calyculin A treatment had a non-additive potentiating effect on the β2-AR-mediated positive inotropic response. These results suggest that the interaction of the β2-AR-coupled Gi and Gs signaling affects the local balance of protein kinase and phosphatase activities. Thus, the additional coupling of β2-AR to Giproteins is a key factor causing the compartmentalization of β2-AR-induced cAMP signaling.

The classical view of ␤-AR 1 signal transduction is that agonist-bound ␤-AR selectively interact with stimulatory G proteins (G s ), which, in turn, activate adenylyl cyclase to enhance cAMP formation. Subsequently, PKA phosphorylates a multitude of regulatory proteins involved in cardiac excitation-contraction coupling, including L-type Ca 2ϩ channels (1,2), the sarcoplasmic reticulum Ca 2ϩ pump regulator PLB (3,4), and myofilament proteins (5,6), producing positive inotropic and lusitropic effects. In addition, PKA also phosphorylates and activates the endogenous protein phosphatase inhibitor 1 (7), which further ensures the action of protein kinases by protein phosphatase inhibition.
Although cardiac ␤ 1 -AR signaling follows the scheme described above, recent studies have revealed a dissociation of ␤ 2 -AR-mediated positive inotropic as well as lusitropic effects from global cAMP accumulation in several mammalian species (8 -11). In addition, it has been demonstrated that in contrast to ␤ 1 -AR, ␤ 2 -AR stimulation fails to induce a cAMP-dependent phosphorylation of non-sarcolemmal proteins involved in excitation-contraction coupling and energy metabolism (e.g. phospholamban, the myofilament proteins, troponin I, C protein, and the cytosolic protein glycogen phosphorylase kinase), but it does activate sarcolemmal L-type Ca 2ϩ channels (10 -12), suggesting that ␤ 2 -AR signaling is localized. More direct evidence supporting the localized ␤ 2 -AR signaling has emerged from single L-type Ca 2ϩ channel recordings. Employing cell-attached patch clamp technique, the activity of single L-type Ca 2ϩ channels has been measured in response to specific ␤-AR subtype agonist outside (remote) or inside (local) the patch pipette. In contrast to the diffusive effect of ␤ 1 -AR stimulation, ␤ 2 -AR stimulation by zinterol only locally activates the L-type Ca 2ϩ channel but has no remote effect (13). Taken together, these previous studies have led to the hypothesis that the ␤ 2 -ARinduced cAMP signaling is compartmentalized to a subsarcolemmal space and cannot be transmitted to cytoplasmic and SR PKA target proteins. Alternatively, the signal is transmitted to cytosolic proteins but local inactivation occurs at these sites.
The mechanism for the local control of ␤ 2 -AR mediated signaling remains unclear. In many biological systems, G s and G i proteins cross-talk and operate as a complementary system. This balance system is usually regulated through different receptor families. For instance, activation of muscarinic receptors or adenosine receptors, prototypic G i -coupled receptors, markedly antagonizes the positive inotropic effect of ␤-AR stimulation (14,15). Interestingly, promiscuous G protein coupling of ␤ 2 -AR to both G s and PTX-sensitive G proteins (G i2 and G i3 ) has been demonstrated in intact cardiomyocytes (16,17). This coupling of ␤ 2 -AR to G i proteins and its downstream pathway might interplay with the ␤ 2 -AR/G s signaling and contribute to the localization of ␤ 2 -AR signaling near the sarcolemmal membrane. Thus, in the present study, we intended to determine whether the localized ␤ 2 -AR signaling is mediated by the additional coupling of ␤ 2 -AR to G i proteins and, if so, to elucidate potential underlying mechanisms.

EXPERIMENTAL PROCEDURES
Measurements of Cell Length-Single ventricular myocytes were isolated from rat hearts by a standard enzymatic technique (18). The cells were suspended in HEPES pH 7.4 buffer containing (in mmol/liter) 20 HEPES, 1 CaCl 2 , 137 NaCl, 5 KCl, 15 dextrose, 1.3 MgSO 4 , 1.2 NaH 2 PO 4 . In some experiments, cells were separately or simultaneously treated with 1.5 g/ml PTX (Sigma) for 3 h at 37°C to block G i protein activation (15) or 3 ϫ 10 Ϫ8 M calyculin A (Calbiochem) for 20 min at 23°C to inhibit protein phosphatases. Cells were stimulated at 0.5 Hz at 23°C, and cell length was monitored from the bright-field image by an optical edge-tracking method using a photodiode array (model 1024 SAQ, Reticon) with a 3-ms time resolution (18). * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed: Laboratory of Cardiovascular Science, Gerontology Research Center, NIA, National Institutes of Health, 5600 Nathan Shock Dr., Baltimore, MD 21224. Tel.: 410-558-8662; Fax: 410-558-8150; E-mail: XiaoR@grc.nia.nih.gov. 1 The abbreviations used are: ␤ 1 -AR, ␤ 1 -adrenoreceptor; PLB, phos-L-type Ca 2ϩ Current Measurements-Whole cell L-type Ca 2ϩ current (I Ca ) was measured via patch clamp technique using an Axopatch 1D amplifier (Axon Instruments Inc.) (12). To activate I Ca selectively, cells were voltage-clamped at Ϫ40 mV to inactivate Na ϩ and T-type Ca 2ϩ channels. K ϩ currents were inhibited by 4 mmol/liter 4-aminopyridine and 5.4 mmol/liter CsCl instead of KCl in the HEPES buffer and the pipette solution containing (mmol/liter) CsCl 100, NaCl 10, tetraethanolamine Cl 20, HEPES 10, MgATP 5, EGTA 5, pH 7.2, adjusted with CsOH. I Ca was elicited by 300-ms pulses from a holding potential of Ϫ40 to 0 mV at 0.5 Hz at 23°C. I Ca was measured as the difference between the current at the peak and the end of the 300-ms pulse.
Site-specific Phospholamban Phosphorylation-The detection of sitespecific PLB phosphorylation was performed as described recently (11). Briefly, cardiomyocytes were treated for 10 min with specific ␤-AR subtype agonists as indicated and solubilized prior to electrophoresis at 95°C for 5 min to dissociate fully PLB into its monomeric form. Following electrophoresis, proteins were transferred to a polyvinylidene difluoride membrane (Serva), which was probed with the phosphorylation site-specific Ser 16 PLB antibody (PhosphoProtein Research). Following incubation with a peroxidase-conjugated antibody (Dianova), the immunoreaction was detected with ECL (Amersham Pharmacia Biotech) and quantified with a video documentation system (Bio-Rad).
Protein Kinase A Activity-The activation of PKA in soluble and particulate fractions was analyzed by a modified method of Murray et al. (19). Rat cardiomyocytes were homogenized and centrifuged at 6,000 ϫ g for 5 min. The resulting supernatant was taken to represent the soluble protein kinase activity and the resuspended pellet, the particulate fraction. The PKA activity is expressed as the activity ratio of malantide (Bachem) 32 P incorporation in the absence and presence of cAMP (2.8 mol/liter).
Statistics-Results are presented as means Ϯ S.E. Statistical significance was determined by Student's t test or analysis of variance when appropriate. Values with p Ͻ 0.05 were considered to be statistically significant.

RESULTS
The ␤ 2 -AR agonist, zinterol (10 Ϫ5 M), increased the whole cell L-type Ca 2ϩ current (I Ca ) to 161 Ϯ 8.8% (n ϭ 7, p Ͻ 0.05) of control in single rat cardiomyocytes (Fig. 1a), which was completely abolished by the ␤ 2 -AR antagonist ICI 118,551 (10). To delineate a role of cAMP-dependent PKA activation in the modulation of L-type Ca 2ϩ channels, an inhibitory cAMP analog, (R p )-cAMPs (10,11), was used to specifically block PKA activation. In the presence of (R p )-cAMPs (10 Ϫ4 M) zinterol failed to augment I Ca (Fig. 1b), indicating that the cAMP-dependent PKA activation is obligatory for ␤ 2 -AR-mediated modulation of L-type Ca 2ϩ channels.
PLB, the main modulator of cardiac relaxation, is phosphorylated following cardiac ␤ 1 -AR stimulation at two adjacent phosphorylation sites, Ser 16 (Fig. 2a) and Thr 17 , catalyzed by PKA and Ca 2ϩ /calmodulin-dependent kinase, respectively (3,4,20). In contrast, the ␤ 2 -AR agonist, zinterol, even at a maximal concentration (10 Ϫ5 M for 10 min), had only a very minor effect on the PKA-mediated Ser 16 phosphorylation of PLB, as detected with phosphorylation site-specific PLB antibodies (21) in the Western blot ( Fig. 2, a and b). The dose-response relation and the time course of Ser 16 PLB phosphorylation are shown in Fig. 2b and Fig. 3c, respectively. A maximal concentration (10 Ϫ5 M) of the ␤ 2 -AR agonist, zinterol, only induced a minor increase in Ser 16 PLB phosphorylation even if the incubation time was extended to 20 min (Fig. 3c). Concomitantly, the ␤ 2 -AR-mediated increase in contractility occurred in the absence of a significant relaxant effect (Fig. 3a). Thus, the failure of ␤ 2 -AR stimulation to induce PLB phosphorylation is the apparent mechanism for the absence of a lusitropic effect (Fig.  3c). These data illustrate that whereas both ␤ 1 -and ␤ 2 -AR share the common second messenger cAMP, they exhibit differences with respect to PKA-mediated protein phosphorylation and relaxant effects (8 -12).
Based on our recent finding that ␤ 2 -AR dually couples to G s and G i proteins (16,17), we hypothesized that the additional coupling of ␤ 2 -AR to G i might interfere with the ␤ 2 -AR/G s signaling, contributing to the restriction of ␤ 2 -AR-mediated cAMP signaling to a subsarcolemmal domain. To test this hypothesis, we examined the effect of G i protein inhibition by PTX on the PLB phosphorylation following ␤ 2 -AR stimulation and its functional consequences. Whereas PTX itself had only a negligible effect on the basal Ser 16 PLB phosphorylation (Fig.  2b), ␤ 2 -AR stimulation with zinterol in PTX-treated cardiomyocytes markedly increased PLB phosphorylation in a dose-dependent manner (EC 50 ϭ 48.6 Ϯ 1.8 nM) (Fig. 2b), with a maximal increase of 6.5-fold, comparable with that induced by the ␤ 1 -AR agonist norepinephrine (NE at 10 Ϫ7 M) (37.9 Ϯ 4.7 and 36.7 Ϯ 7.7 in arbitrary units, respectively). Fig. 3a shows the time courses of ␤ 2 -AR effects on contraction amplitude and duration (t1 ⁄2 ) in both PTX-treated and non-treated cells. In addition to the 1.5-fold potentiation of the ␤ 2 -AR inotropic response, inhibition of G i function allowed zinterol to induce a de novo relaxant effect in rat cardiomyocytes. The ␤ 2 -AR-induced lusitropic effect in PTX-treated cells is highly comparable with that of ␤ 1 -AR stimulation in control cells (Fig. 3, a and  b). Furthermore, the time course of the ␤ 2 -AR-induced Ser 16 PLB phosphorylation was tightly correlated to the time course of the relaxant effect in PTX-treated cardiomyocytes, both reaching a steady state within 5 min (Fig. 3c). In contrast, neither the ␤ 1 -AR-mediated contractile nor its relaxant response was affected by PTX (Fig. 3b). These results strongly suggest that the ␤ 2 -AR/G i coupling functionally compartmentalizes the ␤ 2 -AR/G s -mediated cAMP signaling, altering the quality as well as the magnitude of its cellular response.
To elucidate further the mechanism underlying the G i -mediated spatial control of ␤ 2 -AR signaling, we measured the PKA activity following ␤-AR subtype stimulation. Similar to ␤ 1 -AR stimulation, ␤ 2 -AR activation also significantly increased the PKA activity ratio in both soluble and particulate fractions (Fig. 4). This suggests that, unlike ␤ 1 -AR, ␤ 2 -AR-mediated increases in cAMP accumulation (8,10) and PKA activation (Fig. 4a) are dissociated from Ser 16 PLB phosphorylation (Fig.  2). Surprisingly, PTX did not significantly affect the response of PKA in either fraction (Fig. 4), suggesting that the cross-talk of G s and G i signaling following ␤ 2 -AR stimulation may occur downstream of PKA (see "Discussion"). We therefore examined the potential involvement of protein phosphatases in ␤ 2 -ARmediated G i signaling. Rat cardiomyocytes were treated with calyculin A (3 ϫ 10 Ϫ8 M) for 20 min to inhibit protein phosphatases. Control experiments showed that calyculin A at this concentration had only a marginal effect on contraction amplitude (127.1 Ϯ 18.6% of control, n ϭ 4), but induced a time-and dose-dependent augmentation in contraction at higher concentrations (data not shown). Interestingly, preincubation of cells with calyculin A at this near-threshold concentration markedly and selectively potentiated the submaximal contractile response to the ␤ 2 -AR agonist zinterol (10 Ϫ6 M), whereas it had no effect on the submaximal ␤ 1 -AR (NE, 10 Ϫ8 M)-stimulated contractile response (Fig. 5). Thus, the effects of protein phosphatase inhibition are similar to that of G i inhibition by PTX, enhancing the contractile response in a ␤ 2 -AR-specific manner (Fig. 3). This result strongly suggests that protein phosphatases are likely to be involved in the ␤ 2 -AR/G i -directed signaling. This conclusion was further substantiated by the observation that calyculin A failed to potentiate further the ␤ 2 -AR-mediated contractile response if the G i pathway is disrupted by PTX treatment (Fig. 5). Therefore, protein phosphatases may serve as a key element of the ␤ 2 -AR-coupled G i signaling cascade to spatially control the G s -mediated cAMP/PKA signaling. DISCUSSION Recent advances in ␤ 2 -AR signaling have provided evidence for a novel subcellular compartmentalization of cAMP signaling. Specifically, although both ␤ 1 -and ␤ 2 -AR stimulation enhance cAMP accumulation (8,10) and PKA activity and modulate I Ca via cAMP/PKA-dependent signaling, the ␤ 2 -AR stimulation is uncoupled from the phosphorylation of more remote proteins (8,9,11). This indicates that the signaling may be highly localized to sarcolemmal microdomains or that it can be transmitted to cytoplasmic sites but locally inactivated there. The key question then is what mechanism enforces the tight spatial control of ␤ 2 -AR-mediated cAMP signaling. In principle, a localized cAMP signaling could arise from localization of signaling components, e.g. localization of cAMP by phosphodiesterases (22,23) or PKA by specific anchoring proteins of PKA (24,25). A close spatial association of Ca 2ϩ channels with adenylyl cyclase and PKA (26, 27) could provide a structural basis for the localized cAMP-dependent modulation of L-type Ca 2ϩ channels following ␤ 2 -AR stimulation. However, since both cAMP and catalytic subunits of PKA (released following activation) are diffusive molecules, additional mechanisms should be involved to restrict actively their signaling to certain subcellular domains.
In the present study, we demonstrated that, apart from localization of signaling molecules of the cAMP/PKA cascade (22)(23)(24)(25), an interaction between functionally opposing signal transduction pathways can also create compartmentalization of receptor-mediated signaling. In particular, following inhibition of G i function by PTX treatment, ␤ 2 -AR stimulation markedly increased Ser 16 PLB phosphorylation and elicited a de novo lusitropic effect in rat ventricular myocytes, which is highly comparable with that following ␤ 1 -AR stimulation. Thus, inhibition of G i proteins converts the ␤ 2 -AR signaling mode to that of ␤ 1 -AR-like signaling. These results strongly suggest that the ␤ 2 -AR/G i coupling effectively compartmentalizes the ␤ 2 -AR/G s -mediated cAMP signaling, altering not only the magnitude but also the quality of its cardiac response.
It is well established that activation of protein phosphatases functionally counterbalances cellular effects of protein kinases. Recently, it has been reported that the inhibitory effect of the G i protein-coupled muscarinic receptor stimulation on the ␤-AR-induced cAMP signaling is largely mediated by activation of protein phosphatases (28). Furthermore, angiotensin II increases protein phosphatase 2A activity in cultured neurons through a G i protein-dependent mechanism (29). We therefore investigated the potential involvement of protein phosphatases in the cross-talk of the ␤ 2 -AR-stimulated G s /G i signaling. Interestingly, calyculin A, a non-selective protein phosphatase inhibitor, mimics the PTX effect. Both interventions, G i inhibition by PTX or protein phosphatase inhibition by calyculin A, had a non-additive potentiating effect on ␤ 2 -AR-mediated contractile response when applied together, suggesting that PTX and calyculin A act on a common signaling pathway. Therefore, the negating and spatial restricting effects of ␤ 2 -AR-activated G i proteins on the G s -directed signaling might be mediated by a modulation of the protein phosphatase/kinase balance. For instance, a high dephosphorylation rate of non-sarcolemmal proteins following ␤ 2 -AR stimulation might negate the PKAmediated phosphorylation of PLB (and other cytoplasmic proteins).
␤ 2 -AR-mediated G i protein activation could either directly modulate protein phosphatase activity or inhibit PKA activation at subcellular compartments and secondarily modulate the protein phosphatase inhibitor 1 and therefore protein phosphatase activity (7). In other words, the G i protein activation could negate the ␤ 2 -AR/G s -mediated PKA activation in certain subcellular microdomains, resulting in a higher protein phosphatase activity and a lower protein phosphorylation in those subcellular microdomains, which is functionally indistinguishable from a G i -mediated direct activation of protein phosphatases. Previous studies in frog and canine hearts have provided evidence for the involvement of phosphodiesterases in the compartmentalization of ␤-AR signaling (22,23). However, our previous studies have demonstrated that PTX treatment has no significant effect on ␤ 2 -AR-mediated global cAMP accumulation (10). The present results show further that ␤ 2 -AR-stimulated increase in PKA activity in two different subcellular fractions is insensitive to PTX. Although the non-additive effect of protein phosphatase and G i protein inhibition suggests, but does not prove, that ␤ 2 -AR stimulation is modulating protein phosphatase activity through a PKA-independent pathway, these standard measures of global cAMP levels and PKA activity in cardiomyocytes provide no insight into the highly localized signaling. Although we are unable to distinguish if ␤ 2 -AR/G i signaling modulates protein phosphatases PKA-independent or -dependent due to technical limitation, the present results strongly suggest that protein phosphatases are critically involved in the ␤ 2 -AR/G i signaling, contributing to the functional compartmentalization of the ␤ 2 -AR/G s signaling in rat ventricular myocytes.
It is noteworthy that the ␤ 2 -AR-mediated cardiac response and the extent of the G i protein coupling may vary substantially among species, resulting in an enormous diversity in cardiac ␤ 2 -AR-mediated responses and its sensitivity to PTX. In mouse cardiomyocytes ␤ 2 -ARs are not functional unless G i function is inhibited by PTX treatment, indicating a high level of G i protein coupling (16). In contrast, ␤ 2 -AR stimulation does induce positive inotropic and lusitropic effects as well as phosphorylation of regulatory proteins in the failing human heart  (30,31). Between these extremes, ␤ 2 -AR stimulation in rat cardiomyocytes induces significant increases in I Ca and contractility in the absence of phosphorylation of cytoplasmic regulatory proteins. PTX treatment further enhances the ␤ 2 -AR contractile response (17) and restores its ability to phosphorylate cytoplasmic regulatory proteins in this species. The situation in canine myocytes is similar to that of rat myocytes, except that ␤ 2 -AR does induce lusitropic as well as inotropic effects in the absence of cytoplasmic protein phosphorylation (9,11).
The aforementioned data also illustrate that the same signaling molecule, cAMP, mediates remarkably different cardiac functional responses following ␤ 1 -and ␤ 2 -AR stimulation (8 -12). Analogously, it has been shown that intracellular Ca 2ϩ located in different subcellular compartments may mediate distinctly different and sometimes even opposing cellular functions. For instance, a global elevation in cytosolic Ca 2ϩ in arterial smooth muscle cells causes vasoconstriction, but Ca 2ϩ sparks near the sarcolemma induce relaxation (32). In this case, local Ca 2ϩ gradients are possible, because various endogenous binding sites buffer Ca 2ϩ of discrete origins. Thus, physical or functional compartmentalization of ubiquitous intracellular messengers, such as cAMP and Ca 2ϩ , creates specificity and diversity of a given receptor-mediated signaling.
In summary, we have demonstrated that, in addition to the potentiation of the inotropic response, inhibition of G i function by PTX induces a de novo lusitropic effect and PLB phosphorylation following ␤ 2 -AR stimulation in rat ventricular myocytes. These results suggest a contribution of ␤ 2 -AR/G i -coupled signaling to the compartmentalization of ␤ 2 -AR/G s -stimulated cAMP/PKA signaling, possibly through a protein phosphatasedependent mechanism. In addition, the present study demonstrates that compartmentalization of a common second messenger-directed signaling allows for selective modulation of a variety of target proteins and cellular processes, creating signaling specificity and versatility among closely related G protein-coupled receptors.