Protein Kinase a Type I and Type II Define Distinct Intracellular Signaling Compartments

Protein Kinase A Type I and Type II Define Distinct Intracellular Signaling Compartments

Giulietta Di Benedetto, Anna Zoccarato, Valentina Lissandron, Anna Terrin, Xiang Li, Miles D. Houslay, George S. Baillie, Manuela Zaccolo

Abstract—Protein kinase A (PKA) is a key regulatory enzyme that, on activation by cAMP, modulates a wide variety of cellular functions. PKA isoforms type I and type II possess different structural features and biochemical characteristics, resulting in nonredundant function. However, how different PKA isoforms expressed in the same cell manage to perform distinct functions on activation by the same soluble intracellular messenger, cAMP, remains to be established. Here, we provide a mechanism for the different function of PKA isoforms subsets in cardiac myocytes and demonstrate that PKA-RI and PKA-RII, by binding to AKAPs (A kinase anchoring proteins), are tethered to different subcellular locales, thus defining distinct intracellular signaling compartments. Within such compartments, PKA-RI and PKA-RII respond to distinct, spatially restricted cAMP signals generated in response to specific G protein–coupled receptor agonists and regulated by unique subsets of the cAMP degrading phosphodiesterases. The selective activation of individual PKA isoforms thus leads to phosphorylation of unique subsets of downstream targets. (Circ Res. 2008;103:836-844.) Key Words: cAMP Ⅲ compartmentalization Ⅲ compartmentation Ⅲ adrenergic stimulation Ⅲ prostaglandin Ⅲ protein kinase A

rotein kinase A (PKA) is a key regulatory enzyme in the coupled to specific GPCRs to degrade cAMP selectively in Pheart that is involved in the catecholamine-mediated response to a given stimulus.8 control of excitation–contraction coupling, as well as in a Cardiac myocytes express all four types of PKA isozymes, myriad of other functions including activation of transcrip- PKA-RI␣, PKA-RII␣, PKA-RI␤, and PKA-RII␤.9 PKA iso- tion factors and control of metabolic enzymes. The second forms show different subcellular localization, with PKA-RII messenger cAMP activates PKA by binding to the regula- being mainly associated with the particulate fraction of cell tory (R) subunits, causing release of the activated catalytic lysates whereas PKA-RI has been found preferentially in the (C) subunits. cytosol.10,11 PKA isoforms also show different biochemical The fact that, following cAMP-engagement, PKA mediates properties. PKA-RI is more readily dissociated by cAMP than a plethora of cellular responses has raised the question of how PKA-RII,12,13 and the recent structure solution of holoenzyme specificity is maintained. In recent years, features of this complexes14,15 shows critical isoform-specific features that pathway that contribute to specificity have been uncovered.1 specifically regulate inhibition and cAMP-induced activation A key role is played by AKAPs (A kinase anchoring of PKA-RI and PKA-RII. Given the distinct biochemical proteins), a family of proteins that act as molecular scaffolds properties and the specific subcellular localization of PKA to anchor PKA in the vicinity of specific substrate mole- isozymes, it is not surprising that the biological role of cules,2 thus focusing PKA activity toward relevant substrates. PKA-RI and PKA-RII is nonredundant, as demonstrated by A second mechanism contributing to specificity revolves genetic and biochemical studies (reviewed elsewhere16). around the spatial control of the cAMP signal itself. Restric- However, how individual PKA isoforms serve to deliver a tion of intracellular diffusion of cAMP has been shown by specific response remains unknown. In particular, it remains using a variety of approaches,3–5 including direct imaging of to be established how spatial control of the cAMP signal and gradients of cAMP in response to activation of various G activation of individual PKA isoforms are coordinated to protein–coupled receptors (GPCRs).6 A key role in shaping perform a specific biological function. cAMP intracellular pools is played by phosphodiesterases Here, we set out to answer the question of whether (PDEs), the enzymes that hydrolyze cAMP.7 Indeed, individ- confined pools of cAMP elicited in response of specific ual PDE isoforms have been shown to be functionally extracellular stimuli selectively activate individual PKA iso-

Original received February 28, 2008; revision received August 14, 2008; accepted August 19, 2008. From the Dulbecco Telethon Institute (G.D.B., A.Z., V.L., M.Z.), Venetian Institute of Molecular Medicine, Padova, Italy; and Neuroscience and Molecular Pharmacology (A.T., X.L., M.D.H., G.S.B., M.Z.), Faculty of Biomedical & Life Sciences, University of Glasgow, Scotland, United Kingdom. Correspondence to Dr Manuela Zaccolo, Neuroscience and Molecular Pharmacology, Faculty of Biomedical and Life Sciences, University Avenue, G12 8QQ, Glasgow, UK. E-mail [email protected] © 2008 American Heart Association, Inc. Circulation Research is available at http://circres.ahajournals.org DOI: 10.1161/CIRCRESAHA.108.174813 836 Di Benedetto et al Compartmentalized Signaling by PKA Isoforms 837 forms. By using FRET- and FRAP-based imaging approaches we show that, in cardiomyocytes, PKA-RI and PKA-RII, by anchoring to endogenous AKAPs, define distinct compartments within which cAMP is specifically controlled by different subsets of PDEs. In addition, we demonstrate that cAMP levels rise selectively in the PKA-RI and PKA-RII compartments in a stimulus-specific manner, leading to the phosphorylation of unique subsets of downstream PKA targets. The generation of distinct pools of cAMP within cells that allows for the selective activation of individual PKA isoforms is instrumental for the cell to modulate specific physiological functions and points to means for developing strategies for selective pharmacological intervention.

Materials and Methods Primary cultures of neonatal cardiac ventriculocytes from 1- to 3-day old rats were prepared as described.6 All the details concerning Figure 1. Targeted FRET-based cAMP sensors. Confocal generation of constructs, cells transfection, Western blotting, immu- images of CHO cells expressing either RI_epac or RII_epac nostaining and confocal imaging, FRAP experiments, FRET imaging alone (left panels) or in combination with ezrin and AKAP 79, and RT-PCR are described in the expanded Materials and Methods respectively (right images). The middle images show the local- section in the online data supplement, available at http://circres. ization of green ﬂuorescent protein (GFP)-tagged ezrin and ahajournals.org. GFP-tagged AKAP79 in CHO cells. Scale barsϭ10 ␮m.

Results RII_epac with AKAP79 results in the relocalization of the Generation of cAMP Sensors Selectively Targeted sensor at the plasma membrane, confirming that the modified to the PKA-RI and PKA-RII Compartments sensors localize within the cell where AKAPs are present. We set out to assess whether PKA-RI and PKA-RII are To verify if RI_epac and RII_epac are targeted to different selectively and independently activated by specific extracel- subcellular compartments in neonatal cardiac myocytes, the lular stimuli and generated 2 FRET-based probes that, by localization of the sensors was analyzed by confocal micros- selectively targeting to the same subcellular compartments as copy. As a localization marker, the Z-line protein Zasp fused the endogenous PKA isoforms, monitor the cAMP signals to the red fluorescent protein mRFP (zasp-RFP) was coex- generated at these sites. We took advantage of the unique pressed in combination with either RI_epac or RII_epac. As dimerization/docking domain sequences that have been illustrated in Figure 2A and 2I, RI_epac shows a tight striated shown17 to mediate anchoring of PKA-RI and PKA-RII pattern overlaying with both the Z and the M sarcomeric lines subunits to AKAPs. Thus, by fusing the dimerization/docking (see line intensity profiles at the bottom of Figure 2). In domain from either RI␣ (amino acids 1 to 64) or RII␤ (amino contrast, the distribution of RII_epac shows a very strong acids 1 to 49) to the N terminus of the soluble Epac-1 localization that corresponds to the M line and a much weaker sensor,18 we generated the sensors RI_epac and RII_epac localization overlaying the Z line (Figure 2B and 2L). Such (Figure IA in the online data supplement). For both sensors, localization is identical to the localization of overexpressed it is believed that the binding of cAMP to the cAMP-binding full-length RI and RII subunits and corresponds to the domain will result in a conformational change that causes an localization of endogenous RI and RII subunits (supplemental increase in the distance of the cyan fluorescent protein and Figure II). To assess whether the differences in localization yellow fluorescent protein moieties, with a consequent reduc- were attributable to anchoring of the sensors to endogenous tion of the FRET signal, as shown for the parent sensor.18 The AKAPs, we used the AKAP-competing peptides RIAD21 and modified sensors have been tested with respect to maximal SuperAKAP-IS.22 These peptides have been shown to com- FRET response and to dose-response behavior (supplemen- pete selectively with the binding of PKA-RI and PKA-RII to tary materials and supplemental Figure I). We found that endogenous AKAPs. In particular, the RIAD peptide displays RI_epac and RII_epac are equally sensitive to cAMP more than 1000-fold selectivity for RI over RII,21 whereas the changes. peptide SuperAKAP-IS is 10 000-fold more selective for the RII isoform relative to RI.22 Challenge of cells expressing RI_epac and RII_epac Show a Different and RI_epac with RIAD and of cells expressing RII_epac with AKAP-Mediated Localization in Cardiac Myocytes SuperAKAP-IS completely abolished the striated pattern of To assess the ability of the modified sensors to effectively localization of the sensors (Figure 2C and 2D), whereas bind to AKAPs, we coexpressed RI_epac with ezrin and RIAD and SuperAKAP-IS did not have any effect on the RII_epac with AKAP79. Ezrin is a dual-specificity AKAP localization pattern of, respectively, RII_epac and RI_epac localized at the cortical cytoskeleton and at microvilli,19 (not shown). whereas AKAP79 localizes at the plasma membrane (Figure The accepted paradigm for the PKA signaling pathway 1).20 Coexpression in CHO cells of RI_epac with ezrin results is that type II PKA is associated with particulate subcel- in the relocalization of the sensor to the sub–plasma mem- lular fractions via binding to AKAPs, whereas type I PKA brane region and to microvilli, whereas coexpression of is primarily cytoplasmic.11 Our confocal imaging studies 838 Circulation Research October 10, 2008

Figure 2. RI_epac and RII_epac localize in different subcellular compartments via binding to AKAPs. Confocal images of cardiomyocytes expressing either RI_epac or RII_epac alone (A and B) or in combination with the competing peptides RIAD (C) or SuperAKAP-IS (D). Localization of the marker zasp-RFP in the same cell is shown in E through H. The overlay between probe localization and zasp-RFP is shown in I through N. The bottom images show, for each cell, the intensity proﬁle of the probe signal (in blue) and of the zasp-RFP signal (in red) in the region indicated by the black line. Scale barsϭ10 ␮m.

suggest, however, that RI is also anchored in neonatal both the RI_epac and RII_epac being restricted, we noted cardiac myocytes. To assess the extent of anchoring of the that RI_epac was 1.6 times as mobile as RII_epac. RI_epac and RII_epac sensors, we conducted fluorescence To assess the extent to which the reduced mobility of RI_epac recovery after photobleaching (FRAP) experiments. As and RII_epac was attributable to anchoring to endogenous shown in Figure 3A, following fluorescence bleaching in AKAPs, we repeated the FRAP experiments in cardiomyocytes cells expressing either the untagged, cytosolic cAMP expressing each of the targeted sensor in combination with sensor Epac-1 or 1 of the targeted sensors, we found that specific competing peptides. Figure 3B shows a significant fluorescence recovery occurred with a t1/2 of 5.7Ϯ0.4 increase in the fractional recovery both for cells expressing seconds (nϭ12) for Epac1, 14.3Ϯ1.9 seconds (nϭ10) for RI_epac in combination with RIAD (85.3Ϯ2.3% [nϭ10]) and RI_epac, and 21.6Ϯ1.7 seconds (nϭ10) for RII_epac. for cells expressing RII_epac in combination with Fluorescence recovery was almost complete for the parent SuperAKAP-IS (80.1Ϯ2.5% [nϭ11]). Accordingly, we found a

Epac-1 probe (fractional recovery of 94.2Ϯ1.2%). How- decrease in the recovery times with a t1/2 of 8.4Ϯ0.8 seconds ever, this was reduced to 73.1Ϯ3.4% for RI_epac and to (nϭ10) in cells expressing RI_epacϩRIAD and of 8.9Ϯ0.9 45.4Ϯ5.0% for RII_epac. These results confirm that al- seconds (nϭ11) in cells expressing RII_epacϩSuperAKAP-IS. though the Epac-1 parent probe is free to diffuse in the These results demonstrate that RI_epac and RII_epac specifi- cytosol, the diffusion of both RI_epac and RII_epac is cally localize to different intracellular compartments primarily substantially constrained. However, despite the mobility of by binding to specific endogenous AKAPs.

Figure 3. AKAP binding limits the intracellular mobility of RI_epac and RII_epac. Mean FRAP curves recorded in cardiomyocytes expressing Epac-1 (white), RI_epac (red), or RII_epac (green) in the absence (A) and in the presence (B) of the competing peptides RIAD and SuperAKAP-IS, as indicated. Error bars indicate SEM. Di Benedetto et al Compartmentalized Signaling by PKA Isoforms 839

Figure 4. Cyclase and PDE activity in the PKA-RI and PKA-RII compartments. Representative kinetics of FRET changes recorded in cardiomyocytes expressing either RI_epac (gray circles) or RII_epac (black circles) in response to the application of 5 ␮mol/L forskolin (A), 100 ␮mol/L IBMX (B), 10 ␮mol/L cilostamide (C), 10 ␮mol/L EHNA (D), and 10 ␮mol/L rolipram (E). F, Summary of all the experiments performed in the above conditions. Error bars indicate SEM. *0.01 Ͻ PϽ0.05. ns indicates not signiﬁcant.

cAMP Levels Are Specifically Regulated in the itor, EHNA (10 ␮mol/L) resulted in a ⌬R/R0ϭ2.2Ϯ0.8% PKA-RI and PKA-RII Compartments (nϭ15) when detected by RI_epac and in a ⌬R/

In a first set of experiments, we wanted to assess whether the R0ϭ0.3Ϯ0.2% (nϭ12) when detected by RII_epac PKA-RI and PKA-RII compartments have equal access to (Pϭ0.04), indicating that, in basal conditions, PDE2 ac- cAMP. Cardiac myocytes expressing either RI_epac or RII_epac tivity is prominent in the PKA-RI compartment but very were challenged with 5 ␮mol/L forskolin. As shown in Figure low in the RII_epac compartment (Figure 4D and 4F). 4A and 4F, the 2 sensors detected a comparable rise in [cAMP] PDE4 inhibition, assessed using the selective inhibitor ⌬ ϭ Ϯ ϭ ⌬ with R/R0 2.1 0.4% (n 7) for RI_epac and R/ rolipram (10 ␮mol/L), generated a ⌬R/R0ϭ0.7Ϯ0.2%

R0ϭ2.9Ϯ0.4% (nϭ14) for RII_epac (Pϭ0.24), indicating that (nϭ13) when detected by RI_epac and a ⌬R/ the compartments hosting PKA-RI and PKA-RII are equally R0ϭ2.6Ϯ0.7% (nϭ16) when detected with RII_epac associated to adenylyl cyclases and have potentially access to (Pϭ0.02) (Figure 4E and 4F), indicating that, contrary to comparable cAMP levels. PDE2, PDE4 exerts its activity mainly in the PKA-RII Next, we wanted to assess the role of PDEs in the domain and to a much lower extent in the PKA-RI control of cAMP levels in both PKA compartments. When compartment. These results show that cAMP levels are myocytes were challenged with isobutylmethylxanthine differently regulated in the PKA-RI and PKA-RII compart- (IBMX) (100 ␮mol/L), no significant difference was ments and point to a specific association of individual PKA detected in the level of cAMP in the 2 compartments isoforms with selected subsets of PDEs in these cells.

(⌬R/R0ϭ4.5Ϯ0.3%; nϭ20 for RI_epac and ⌬R/

R0ϭ4.9Ϯ0.4%; nϭ15 for RII_epac; Pϭ0.45) (Figure 4B Individual GPCRs Generate a cAMP Signal and 4F), suggesting that the basal level of cAMP is under Selectively in the PKA-RI or comparable levels of PDE control in both PKA-RI and PKA-RII Compartments PKA-RII compartments in these cells. We next asked whether the PKA-RI and PKA-RII compart- To determine whether specific subsets of PDEs control the ments may be coupled to specific GPCRs such that individual cAMP signal in the 2 PKA compartments, cardiac myocytes PKA isoforms respond selectively to cAMP signals elicited expressing either RI_epac or RII_epac were treated with by different agonists. As shown in Figure 5A, the application selective PDE inhibitors. As shown in Figure 4C and 4F, of isoproterenol (10 nmol/L), a specific activator of ␤-adre- PDE3 inhibition assessed using the selective inhibitor cilos- noreceptors, generated a rise in [cAMP] more pronounced in tamide (10 ␮mol/L) generated a small and similar cAMP rise the PKA II domain as compared to the PKA I domain ⌬ ϭ Ϯ ϭ ⌬ in both PKA compartments (⌬R/R0ϭ0.8Ϯ0.2%; nϭ9 for ( R/R0 2.5 0.5% [n 13] for RI_epac and R/

RI_epac and ⌬R/R0ϭ0.6Ϯ0.2%; nϭ11 for RII_epac; R0ϭ4.5Ϯ0.7% [nϭ9] for RII_epac; Pϭ0.007). Addition of Pϭ0.58). PDE2 inhibition assessed using the selective inhib- IBMX (100 ␮mol/L) abolishes such a difference. This result 840 Circulation Research October 10, 2008

Figure 5. Effect of GPCR stimulation on the cAMP signal in the PKA-RI and PKA-RII compartments. Representative kinetics of FRET change on application of 10 nmol/L isoproterenol (A), 100 nmol/L GLP-1 (B), 300 nmol/L glucagon (C), and 1 ␮mol/L PGE1 (D) recorded in cardiomyocytes expressing RI_epac (gray circles) of RII_epac (black circles). E, Summary of the experiments performed in the above conditions. F and G, FRET change detected in cardiomyocytes coexpressing either RI_epac in combination with the competing peptide RIAD (gray bars) or RII_epac in combination with SuperAKAP-IS (black bars) and stimulated with 10 nmol/L isoproterenol (F) or with 1 ␮mol/L PGE1 (G). Error bars indicate SEM. *0.01ϽPϽ0.05; **0.001ϽPϽ0.01; ***PϽ0.001.

is in agreement with our previous studies showing that alone; Figure 5F). Coexpression of SuperAKAP-IS and ␤-adrenergic stimulation generates a restricted pool of cAMP RII_epac did not affect RII_epac ability to detect the that activates selectively AKAP-anchored PKA-RII.6 By isoproterenol-induced cAMP signal (Figure 5F). Similarly, contrast, challenging the myocytes with GLP-1 (100 nmol/L) when RII_epac was coexpressed with the selective competing resulted in a larger [cAMP] increase in the PKA-RI compart- peptide SuperAKAP-IS, the amplitude of the cAMP signal ment as compared to the PKA-RII compartment (⌬R/ generated by application of 1 ␮mol/L PGE1 was now clearly

R0ϭ3.7Ϯ0.8%, nϭ15 for RI_epac and ⌬R/R0ϭ1.0Ϯ0.4%, detected by RII_epac (⌬R/R0ϭ2.8Ϯ0.7% [nϭ17]; Pϭ0.0001 nϭ12, for RII_epac; Pϭ0.008) (Figure 5B). Similarly, 300 versus RII_epac alone; Figure 5G), whereas coexpression of nmol/L glucagon (Figure 5C) and 1 ␮mol/L prostaglandin E1 RIAD and RI_epac had no effect on the amplitude of the

(PGE1) (Figure 5D) generated a higher [cAMP] increase in signal detected on PGE1 application (Figure 5G). In addition, the PKA-RI associated compartment than in the PKA-RII we found that treatment with RIAD significantly reduced the compartment, showing ⌬R/R0ϭ1.3Ϯ0.4% (nϭ7) for RI_epac velocity of the FRET change of RI_epac to PGE1 (supple- and ⌬R/R0ϭ0.2Ϯ0.1% (nϭ6) for RII_epac (Pϭ0.017) in the mental Figure III). case of glucagon and ⌬R/R0ϭ1.4Ϯ0.2% (nϭ35) for RI_epac Taken together, these results demonstrate that individual and ⌬R/R0ϭ0.6Ϯ0.2% (nϭ30) for RII_epac (Pϭ0.01) in the GPCR agonists generate spatially restricted pools of cAMP case of PGE1. In the presence of the specific competing that selectively activate individual AKAP-anchored PKA peptide RIAD, the cAMP signal generated by the application isoforms. Removal of PKA isoforms from their anchoring of isoproterenol (10 nmol/L) was now clearly detected by sites allows them to diffuse and to be activated by pools of

RI_epac (⌬R/R0ϭ5.0Ϯ1.0% [nϭ9]; Pϭ0.008 versus RI_epac cAMP that would not normally affect them. Di Benedetto et al Compartmentalized Signaling by PKA Isoforms 841

Figure 6. Speciﬁc GPCR agonist stimulation results in a different pattern of downstream PKA targets phosphorylation. A, Western blots of cardiac myocyte lysates probed for PLB, TnI, and the ␤2AR with corresponding phosphoblots after treatment with 10 nmol/L isoproterenol and 1 ␮mol/L PGE1. Quantiﬁcations are means of at least 3 separate experiments. Error bars indicate SEM. **0.001ϽPϽ0.01; *0.01ϽPϽ0.05. B, Representative Western blots of cardiac myocyte lysates probed for PLB and TnI with corresponding phosphoblots after treatment with isoproterenol 10 nmol/L with and without pretreatment with KT5720 (2 ␮mol/L).

23 Compartmentalized PKA Isozymes Phosphorylate coupled to G␣i. We asked whether the reduction in the

Selected Subsets of Downstream Targets phosphorylation level of PBL and TnI observed on PGE1

We next wanted to assess whether the selectivity of the cAMP stimulation may be attributable to activation of an EP3 signals generated in the PKA-RI and PKA-RII compartments receptor and subsequent G␣i-mediated inhibition of cyclase on stimulation of individual GPCRs has a functional rele- activity. RT-PCR analysis of mRNA extracted from neo- vance and results in a specific pattern of PKA-mediated natal rat cardiomyocytes confirmed that these cells express phosphorylation of downstream targets. To this aim, we high levels of message for the G␣i-coupled EP3 receptor, as studied the phosphorylation level of several PKA targets after well as message for the G␣s-coupled EP2 and EP4 receptors stimulation with either isoproterenol or PGE1. The phosphor- (Figure 7A). ylation level of phospholamban (PLB), troponin (Tn)I, and To evaluate the occurrence and the relevance of G activa- ␤ ␤ i the adrenergic receptor type 2 ( 2AR) was markedly tion on the local control of the cAMP signal generated by increased (Figure 6A) on stimulation with isoproterenol (1 PGE1, we performed imaging experiments applying PGE1 on nmol/L) but not on stimulation with PGE1 (10 ␮mol/L). cardiomyocytes that had been pretreated with the Gi-inhibitor Stimulation of myocytes with isoproterenol in the presence of pertussis toxin (PTX). As shown in Figure 7B, application of the PKA inhibitor KT5720 (2 ␮mol/L) completely abolished 1 ␮mol/L PGE following 2 to 4 hours pretreatment with 2 the increased phosphorylation level of both PLB and TnI 1 ␮g/mL PTX generated a comparable cAMP increase in the (Figure 6B), confirming the involvement of PKA. Unexpect- PKA-RI and PKA-RII compartments. In particular, PTX edly, PGE -stimulated cells showed a reduction in the phos- 1 pretreatment results in a much-increased cAMP signal in the phorylation level of PLB and TnI as compared to untreated PKA-RII compartment (⌬R/R ϭ1.7Ϯ0.4% [nϭ11] versus cells. Further analysis indicated that the reduced phosphory- 0 0.6Ϯ0.2% [nϭ30] in the absence of PTX; Pϭ0.006). Indeed, lation of these targets was dependent on the activity of phosphatases (supplemental Figure IV). this equals the amplitude of the signal generated in the ⌬ ϭ Ϯ ϭ Based on these results we can conclude that individual PKA-RI compartment ( R/R0 1.4 0.2%; P 0.54; see Fig- agonists, via activation of specific PKA isoforms, affect ure 7B). No significant effect of PTX pretreatment was distinct subsets of downstream targets. observed on the amplitude of the cAMP signal observed in the PKA-RI compartment (⌬R/R0ϭ2.1Ϯ0.5% [nϭ8] as com- ⌬ ϭ Ϯ ϭ AGi-Mediated Mechanism Contributes to Control pared to R/R0 1.4% 0.2% in control cells; P 0.13). the cAMP Signal, Leading to the Phosphorylation To assess the functional relevance of such a G␣i-mediated of PLB and TnI control of the cAMP signal, we measured the level of

Four different receptors for PGE have been described (EP1 to phosphorylation of PLB and TnI in cardiomyocytes pre-

EP4) with EP2 and EP4 being coupled to G␣s and EP3 being treated with PTX for 1, 2, or 4 hours before the application of 842 Circulation Research October 10, 2008

Figure 7. AGi-mediated mechanism contributes to the control of the cAMP signal in the PKA II compartment. A, RT-PCR analysis of the EP receptors expressed in neonatal rat cardiomyocytes. Hypoxanthine phosphoribosyltransferase (HPRT) is used as a control for input mRNA concentration. B, FRET changes recorded in cardiomyocytes expressing either RI_epac (gray bars) or RII_epac (black bars) with or without pretreatment with PTX (2 ␮g/mL for 2 to 4 hours) and challenged with 1 ␮mol/L PGE1. C, Western blots of cardiac myocyte lysates probed for PLB and TnI with corresponding phosphoblots after treatment with 1 or 10 nmol/L isoprenaline and 1 ␮mol/L PGE1 with and without pretreatment with PTX (2 ␮g/mL for 1, 2, and 4 hours) Quantiﬁcations are means of at least 3 separate experiments. Error bars indicate SEM. *0.01ϽPϽ0.05.

PGE1. Figure 7C shows that, on inhibition of G␣i, PGE1 regulated by different subsets of PDEs, although the exact stimulation induce a significant increase in the phosphoryla- contribution of individual PDEs would require simultaneous tion level of both these PKA-RII–specific targets (compare inhibition of combinations of different isoforms. We show

Figures 6 and 7C). Thus, a G␣i-mediated mechanism appears that the cAMP signal is specifically generated either in the to contribute significantly to the control of the cAMP signal PKA-RI or in the PKA-RII compartment depending on the generated by PGE1 stimulation in the PKA-RII domain. GPCR agonist applied and that cAMP does not diffuse from one compartment to the other so as to cross-activate PKA Discussion isozymes, allowing fidelity of the response (Figure 8). One The organization of the signaling machinery in discrete functional consequence of such compartmentalization is that compartments is increasingly recognized as a critical feature isoproterenol stimulation leads to the specific phosphoryla- for the specificity of the cAMP/PKA system in cardiomyo- tion of PLB, TnI, and ␤2AR, whereas PGE1 stimulation does cytes.24 Here, we used real-time imaging of cAMP to study not affect these substrates, demonstrating that individual PKA how the diversity of PKA isoforms may contribute to the isoforms are coupled with defined subsets of targets and that transduction of specific downstream signals. We did this by PKA isoforms activity is not promiscuous. targeting FRET-based reporters of cAMP concentration at Our findings are consistent with previous work25–27 show- intracellular sites where endogenous PKA-RI and PKA-RII ing that, in cardiac myocytes, there is a clear dichotomy isoforms localize. We demonstrate that PKA-RI and PKA-RII between the effects of PGE1 and isoproterenol on a variety of reside in physically segregated and distinct compartments cAMP-dependent events. These studies demonstrated that within which the level of cAMP appears to be selectively isoproterenol generates an increase in contractile force25 Di Benedetto et al Compartmentalized Signaling by PKA Isoforms 843

Figure 8. Model of PKA isoforms compartmentalization. A, Activation of the PGE receptor system leads to the generation of a restricted cAMP pool affecting PKAI. The signal in the specific PKA-RII compartments may be switched off with the contribution of an EP3-dependent, Gi-mediated mechanisms. B, Only when the ␤AR is activated via binding of the appropriate agonist a cAMP pool is generated in the PKA-RII compartment and this isoform is activated to phosphorylate a specific subset of downstream targets. enhances ventricular pressure development and induces phos- PKA-RI isozyme may in fact be anchored in the heart in vivo. phorylation of target proteins such as phosphorylase kinase The fact that PKA-RI is found mainly in the supernatant of 26 and TnI, whereas PGE1, although elevating cAMP to heart cell homogenates may reflect relatively low affinity comparable levels and inducing activation of PKA similar to interactions between PKA-RI and the protein binding part- isoproterenol, does not show any effect on contractility or on ners that are lost during the homogenization and washing the phosphorylation of these target proteins. steps. Notwithstanding this, the nature of the PKA-RI anchor- The general notion concerning subcellular localization of ing proteins in cardiac myocytes remains to be established. PKA isoforms in cardiac myocytes is that PKA-RII holoen- Catecholamine-mediated sympathetic control of cardiac stim- zyme is localized whereas PKA-RI is mainly cytosolic. ulation is a major compensatory mechanism that maintains or Biochemical studies on cell homogenates showed that PKA- augments systolic and diastolic ventricular function during RII subunits are bound to particulate material in heart cells, physiological stress or pathological conditions. In particular, whereas RI subunits are largely soluble.10,11 Interestingly, the catecholamines selectively improve diastolic function by reduc- selective effects of ␤-adrenergic stimulation on heart function ing myofilament calcium sensitivity through phosphorylation of has been shown to correlate with activation of a membrane- proteins such as TnI and accelerate sequestration of calcium into bound fraction of PKA, whereas PGE1 stimulation increases the sarcoplasmic reticulum through phosphorylation of PLB and the activity of a soluble pool of PKA.26 Our data now provide release of its inhibitory effect on the SERCA pump.33 Enhanced a mechanism for these observation by showing that catechol- and sustained cardiac adrenergic drive, however, is known to be amines generate restricted pools of cAMP that selectively deleterious and to contribute, in part, to development and engage PKA-RII isoforms. However, the notion of a PKA-RI progression of pathological states such as heart failure.34 Our migrating freely in the cytosol, as suggested on the basis of results indicate that key proteins involved in the catecholamine- previous biochemical studies, is difficult to reconcile with mediated regulation of cardiac contractility are under the control any selectivity in the activation of PKA isozymes, particu- of a restricted pool of cAMP that selectively activates a subset of larly considering that PKA-RI is more readily activated by PKA-RII isozymes, thus leading to a coordinated and specific cAMP as compared to PKA-RII.13 In tissues other than the response. In addition, we show that a supplementary element heart, dual-specificity AKAPs capable of binding PKA-RI contributing to specificity is provided by a Gi-mediated mecha- and -RII have been described,28 and AKAP-mediated local- nism that contributes to keep the PKA-RII compartment clear of ization of PKA-RI to the neuromuscular junction29 or to cAMP unless the appropriate catecholaminergic stimulus im- interphase microtubules and specific regions of the mitotic pinges on the cell. Such a mechanism may indeed serve to spindle has been shown.30 The discovery in Caenorhabditis protect the heart from excessive adrenergic stimulation in patho- elegans of a specific RI-binding AKAP that does not interact logical conditions. Interestingly, PGE1 stimulation is known to with RII subunits31 suggests the potential for nonredundant protect myocardial tissue from injury following ischemia and PKA-RI localization and function and a specific role for reperfusion,35 an effect that has been suggested to depend on localized PKA-RI in modulating T-cell receptor signaling has PGE-mediated stimulation of EP3 receptors in cardiac myo- 32 36 been demonstrated. Thus, by performing FRAP experi- cytes and the consequent Gi-mediated inhibition of cAMP ments on intact cells, here we find that neonatal cardiac synthesis.37 myocytes express a considerable amount of PKA-RI anchor In summary, the present work provides original insight into sites, as shown by the slow fluorescence recovery time and the mechanisms that underpin cAMP/PKA specificity of re- the large immobile fraction detected in cells expressing the sponse by demonstrating that, in cardiac myocytes, both RI_epac sensor. This suggests that a large fraction of the PKA-RI and PKA-RII isoforms subsets anchor to subcellular 844 Circulation Research October 10, 2008 sites via binding to endogenous AKAPs. This defines exclusive 13. Cummings DE, Brandon EP, Planas JV, Motamen K, Idzerda RL, signaling domains within which the cAMP signal is uniquely McKnight GS. Genetically lean mice result from targeted disruption of the RII beta subunit of protein kinase A. Nature. 1996;382:622–626. generated via activation of specific GPCR and their associated G 14. Wu J, Brown SH, von Daake S, Taylor SS. PKA type II alpha proteins and is uniquely modulated by the activity of different holoenzyme reveals a combinatorial strategy for isoform diversity. subsets of PDEs, resulting in stimulus-specific phosphorylation Science. 2007;318:274–279. 15. Kim C, Cheng CY, Saldanha SA, Taylor SS. PKA-I holoenzyme structure of downstream protein targets. These results provide a mecha- reveals a mechanism for cAMP-dependent activation. Cell. 2007;130: nism for the different function of PKA-RI and PKA-RII subsets 1032–1043. and provide a functional rationale for the design of isoform- 16. Skalhegg BS, Tasken K. Specificity in the cAMP/PKA signaling pathway. Differential expression, regulation, and subcellular localization specific activators and inhibitors of PKA. of subunits of PKA. Front Biosci. 2000;5:D678–D693. 17. Carr DW, Stofko-Hahn RE, Fraser ID, Bishop SM, Acott TS, Brennan Acknowledgments RG, Scott JD. Interaction of the regulatory subunit (RII) of cAMP- dependent protein kinase with RII-anchoring proteins occurs through an We thank Martin Lohse for providing Epac-1, John Scott and Kjetil amphipathic helix binding motif. J Biol Chem. 1991;266:14188–14192. Tasken for the constructs for RIAD and for SuperAKAP-IS, Kjetil 18. Nikolaev VO, Bunemann M, Hein L, Hannawacker A, Lhose MJ. Novel Tasken for ezrin and ezrin-GFP, Matteo Vatta for wt-Cypher/ZASP- single chain cAMP sensors for receptor-induced signal propagation. GFP, and Marc Dell’Acqua for AKAP-79 and AKAP-79-GFP. J Biol Chem. 2004;279:37215–37218. 19. Algrain M, Turunen O, Vaheri A, Louvard D, Arpin M. Ezrin contains Sources of Funding cytoskeleton and membrane binding domains accounting for its proposed role as a membrane-cytoskeletal linker. J Cell Biol. 1993;120:129–139. This work was supported by Telethon Italy grant GGP05113; Human 20. Klauck TM, Faux MC, Labudda K, Langeberg LK, Jaken S, Scott JD. Frontier Science Program Organization grant RGP0001/2005-C (to Coordination of three signaling enzymes by AKAP79, a mammalian M.Z.); Fondation Leducq grant O6 CVD 02 (to M.Z.); and the scaffold protein. Science. 1996;271:1589–1592. European Commission grant LSHB-CT-2006-037189 (to M.Z., 21. Carlson CR, Lygren B, Berge T, Hoshi N, Wong W, Taskeń K, Scott JD. G.S.B., and M.D.H.); and Medical Research Council grant Delineation of type I protein kinase A-selective signaling events using an G0600765 (to M.D.H. and G.S.B.). RI anchoring disruptor. J Biol Chem. 2006;281:21535–21545. 22. Gold MG, Lygren B, Dokurno P, Hoshi N, McConnachie G, Taskeń K, Carlson CR, Scott JD, Bardford D. Molecular basis of AKAP specificity Disclosures for PKA regulatory subunits. Mol Cell. 2006;24:383–395. None. 23. Narumiya S, Sugimoto Y, Ushikubi F. Prostanoid receptors: structures, properties, and functions. Physiol Rev. 1999;79:1193–1226. References 24. Bornfeldt KE. A single second messenger: several possible cellular 1. Tasken K, Aandahl EM. Localized effects of cAMP mediated by distinct responses depending on distinct subcellular pools. Circ Res. 2006;99: routes of protein kinase A. Physiol Rev. 2004;84:137–167. 790–792. 25. Keely SL. Prostaglandin E1 activation of heart cAMP-dependent protein 2. Wong W, Scott JD. AKAP signalling complexes: focal points in space kinase: apparent dissociation of protein kinase activation from increases in and time. Nat Rev Mol Cell Biol. 2004;5:959–970. phosphorylase activity and contractile force. Mol Pharmacol. 1979;15:235–245. 3. Jurevicius J, Fischmeister R. cAMP compartmentation is responsible for ϩ 26. Hayes JS, Brunton LL, Mayer SE. Selective activation of particulate a local activation of cardiac Ca2 channels by beta-adrenergic agonists. cAMP-dependent protein kinase by isoproterenol and prostaglandin E1. Proc Natl Acad Sci U S A. 1996;93:295–299. J Biol Chem. 1980;255:5113–5119. 4. Barnes AP, Livera G, Huang P, Sun C, O’Neal WK, Conti M, Stutts MJ, 27. Brunton LL, Hayes JS, Mayer SE. Hormonally specific phosphorylation Milgram SL. Phosphodiesterase 4D forms a cAMP diffusion barrier at the of cardiac troponin I and activation of glycogen phosphorylase. Nature. apical membrane of the airway epithelium. J Biol Chem. 2005;280:7997–8003. 1979;280:78–80. 5. Rich TC, Fagan KA, Nakata H, Schaack J, Cooper DM, Karpen JW. 28. Huang LJ, Durick K, Weiner JA, Chun J, Taylor SS. D-AKAP2, a novel Cyclic nucleotide-gated channels colocalize with adenylyl cyclase in protein kinase A anchoring protein with a putative RGS domain. Proc regions of restricted cAMP diffusion. J Gen Physiol. 2000;116:147–161. Natl Acad Sci U S A. 1997;94:11184–11189. 6. Zaccolo M, Pozzan T. Discrete microdomains with high concentration of 29. Imaizumi-Scherrer T, Faust DM, Benichou JC, Hellio R, Weiss MC. cAMP in stimulated rat neonatal cardiac myocytes. Science. 2002;295: Accumulation in fetal muscle and localization to the neuromuscular 1711–1715. junction of cAMP-dependent protein kinase A regulatory and catalytic 7. Houslay MD, Baillie GS, Maurice DH. cAMP-Specific subunits RI alpha and C alpha. J Cell Biol. 1996;134:1241–1254. phosphodiesterase-4 enzymes in the cardiovascular system: a molecular 30. Imaizumi-Scherrer T, Faust DM, Barradeau S, Hellio R, Weiss MC. Type toolbox for generating compartmentalized cAMP signaling. Circ Res. I protein kinase a is localized to interphase microtubules and strongly 2007;100:950–966. associated with the mitotic spindle. Exp Cell Res. 2001;264:250–265. 8. Mongillo M, McSorley T, Evellin S, Sood A, Lissandron V, Terrin A, 31. Angelo R, Rubin CS. Molecular characterization of an anchor protein Huston E, Hannawacker A, Lhose MJ, Pozzan T, Houslay MD, Zaccolo (AKAPCE) that binds the RI subunit (RCE) of type I protein kinase A M. Fluorescence resonance energy transfer-based analysis of cAMP from Caenorhabditis elegans. J Biol Chem. 1998;273:14633–14643. dynamics in live neonatal rat cardiac myocytes reveals distinct functions 32. Skalhegg BS, Tasken K, Hansson V, Huitfeldt HS, Jahnsen T, Lea T. of compartmentalized phosphodiesterases. Circ Res. 2004;95:67–75. Location of cAMP-dependent protein kinase type I with the TCR-CD3 complex. Science. 1994;263:84–87. 9. Scholten A, van Veen TA, Vos MA, Heck AJ. Diversity of cAMP- 33. MacLennan DH, Kranias EG. Phospholamban: a crucial regulator of dependent protein kinase isoforms and their anchoring proteins in mouse cardiac contractility. Nat Rev Mol Cell Biol. 2003;4:566–577. ventricular tissue. J Proteome Res. 2007;6:1705–1717. 34. Sabbah HN. Biologic rationale for the use of beta-blockers in the 10. Corbin JD, Sugden PH, Lincoln TM, Keely SL. Compartmentalization of treatment of heart failure. Heart Fail Rev. 2004;9:91–97. Ј Ј Ј Ј adenosine 3 :5 -monophosphate and adenosine 3 :5 -monophosphate- 35. Hohlfeld T, Zucker TP, Meyer J, Schro¨r K. Expression, function, and dependent protein kinase in heart tissue. J Biol Chem. 1977;252: regulation of E-type prostaglandin receptors (EP3) in the nonischemic 3854–3861. and ischemic pig heart. Circ Res. 1997;81:765–773. 11. Brunton LL, Hayes JS, Mayer SE. Functional compartmentation of cyclic 36. Hohlfeld T, Meyer-Kirchrath J, Vogel YC, Schro¨r K. Reduction of infarct AMP and protein kinase in heart. Adv Cyclic Nucleotide Res. 1981;14: size by selective stimulation of prostaglandin EP(3)receptors in the 391–397. reperfused ischemic pig heart. J Mol Cell Cardiol. 2000;32:285–296. 12. Dostmann WR, Taylor SS, Genieser HG, Jastorff B, Doskeland SO, 37. Yamamoto T, Habuchi Y, Tanaka H, Suto F, Morikawa J, Kashima K, Ogreid D. Probing the cyclic nucleotide binding sites of cAMP-dependent Yoshimura M. EP receptor-mediated inhibition by prostaglandin E(1) of protein kinases I and II with analogs of adenosine 3Ј,5Ј-cyclic phospho- cardiac L-type Ca(2ϩ) current of rabbits. Am J Physiol. 1999;277: rothioates. J Biol Chem. 1990;265:10484–10491. H1369–H1374.