cGMP signals modulate cAMP levels in a compartment-specific manner to regulate catecholamine-dependent signalling in cardiac myocytes

Abstract

Rationale

3′-5′-cyclic adenosine monophosphate (cAMP) and 3′-5′-cyclic guanosine monophosphate (cGMP) are intracellular second messengers involved in heart pathophysiology. cGMP can potentially affect cAMP signals via cGMP-regulated phosphodiesterases (PDEs).

Objective

To study the effect of cGMP signals on the local cAMP response to catecholamines in specific subcellular compartments.

Methods and results

We used real-time FRET imaging of living rat ventriculocytes expressing targeted cAMP and cGMP biosensors to detect cyclic nucleotides levels in specific locales. We found that the compartmentalized, but not the global, cAMP response to isoproterenol is profoundly affected by cGMP signals. The effect of cGMP is to increase cAMP levels in the compartment where the PKA-RI isoforms reside but to decrease cAMP in the compartment where the PKA-RII isoforms reside. These opposing effects are determined by the cGMP-regulated PDEs, namely PDE2 and PDE3, with the local activity of these PDEs being critically important. The cGMP-mediated modulation of cAMP also affects the phosphorylation of PKA targets and myocyte contractility.

Conclusions

cGMP signals exert opposing effects on local cAMP levels via different PDEs the activity of which is exerted in spatially distinct subcellular domains. Inhibition of PDE2 selectively abolishes the negative effects of cGMP on cAMP and may have therapeutic potential.

INTRODUCTION

cAMP and cGMP play a key role in heart function both in normal and pathological conditions. cAMP mediates the chronotropic, inotropic and lusitropic effects of catecholamines and is involved in the pathogenesis of a number of conditions including hypertrophy, arrhythmia and heart failure (HF)¹. cGMP mediates the cardiac effects of nitric oxide (NO) and natriuretic peptides and has been involved both in negative metabolic and inotropic effects² as well as in cardioprotective mechanisms³. Depending on the underlying disease, treatment for acute HF relies on inotropes, acting on cAMP signaling, and vasodialtors, acting on cGMP signalling. However, lack of a full understanding of the complexity of these pathways limits therapeutic effectiveness⁴ and HF mortality remains high.

The reduced intracellular content of cAMP and PKA substrate phosphorylation that normally associates with HF is thought to contribute to the pathophysiology of the disease. However therapy aimed at increasing cAMP levels, although effective in increasing myocardial contractility in the short term, results in increased mortality in the long term⁵ and chronic treatment with β-adrenergic receptor antagonists, which act to decrease cAMP generation, results in improved survival ⁶. To explain this apparent paradox the view is emerging that cAMP signalling is compartmentalized^7,8 and spatial control of signal propagation is paramount for specificity of signalling⁹. Thus, depending on their location, cAMP signals may have different functional effects and changes in the phosphorylation of individual PKA substrates may be either beneficial or harmful, depending on the specific target involved¹⁰.

cAMP is generated by adenylyl cyclases (ACs) upon activation of G protein coupled receptors (GPCRs) in response to a variety of hormones and neurotransmitters. Its main effector is protein kinase A (PKA), a holotetrameric enzyme composed of a dimer of regulatory (R) and two catalytic (C) subunits. Cardiac myocytes express two isoforms of PKA, PKA-RI and PKA-RII, which localize to different subcellular compartments. PKA-RII is mainly found in the particulate fraction of myocyte lysates and PKA-RI, although being mainly recovered in the soluble fraction¹¹, has recently been shown to be anchored to specific subcellular sites in intact cardiac myocytes, albeit with a lower binding affinity than PKA-RII¹². Localization of PKA is achieved via binding to A Kinase Anchoring Proteins (AKAPs)¹³ of the amino-terminal dimerization/docking (D/D) domain of the R subunit¹⁴ and serves to anchor PKA in proximity of its targets thus favoring selective phosphorylation. In a previous study we found that PKA-RI and PKA-RII define two distinct signalling domains within which GPCR agonists generate unique cAMP signals leading to phosphorylation of distinct downstream targets¹². Thus β-adrenergic receptor (β-AR) stimulation generates a pool of cAMP that selectively activates PKA-RII leading to phosphorylation of phospholamban (PLB) and troponin I (TnI). Prostaglandin receptor stimulation generates a different pool of cAMP that selectively activates PKA-RI and does not increase phosphorylation of these targets¹².

cGMP is generated by NO-mediated activation of soluble guanylyl cyclases (sGC) or activation of plasma membrane-bound, particulate guanylyl cyclases (pGC) by natriuretic peptides. pGC and sGC have been shown to mediate different functional effects¹⁵ and, as demonstrated for cAMP, there is evidence suggesting that cGMP may also be compartmentalized in cardiac myocytes¹⁶.

Both cAMP and cGMP levels are under the tight control of the cyclic nucleotide degrading enzymes phosphodiesterases (PDEs)¹⁷. PDEs are a large superfamily of enzymes including 11 families (PDE1-11) with a number of different genes and splice variants generating a plethora of PDE subtypes showing unique kinetic, regulatory and localization properties¹⁸. Cardiac myocytes express several PDEs (PDE1, 2, 3, 4, 5, 8 and 9)¹⁹, the activity of which has been shown to play a pivotal role in the spatial control of cyclic nucleotide signals in these cells^20,21. PDEs also provide a means by which cGMP signals can modulate cAMP signals²². For example, by binding to the regulatory GAF-B domain at the amino-terminus of PDE2, cGMP potently activates the cAMP hydrolyzing activity of this²³. Through such a regulatory mechanism, stimuli that elevate cGMP may attenuate cAMP signals²⁴. Conversely, cGMP acts effectively as a competitive inhibitor of PDE3 cAMP-degrading activity by virtue of the much higher catalytic rate of this enzyme for cAMP than for cGMP and similarly high affinity for both cAMP and cGMP²⁵. As a consequence, PDE3 provides a means by which an increase in cGMP may lead to an increase in cAMP.

Although interplay between cAMP and cGMP signals through PDEs has been suggested to occur in a number of cell types²⁶, the impact of cGMP on local cAMP in ventriculocytes has not been addressed.

Here we investigate whether the β-adrenergic response in the PKA-RI and PKA-RII compartments of rat ventriculocytes is affected by cGMP signals. Using real-time FRET imaging²⁷ and targeted fluorescent reporters we find that defined, localized cAMP responses to isoproterenol are profoundly modulated by cGMP. Importantly, the effect of cGMP is strikingly different in distinct subcellular compartments and depends on the source of cGMP. We identify PDE2 and PDE3 as the effectors of such modulation and provide evidence that cGMP alters the local activation of PKA isoforms affecting the phosphorylation of downstream targets and myocytes contractility. These novel findings unravel new aspects of cAMP/cGMP cross-talk in the heart and suggest previously unrecognized possibilities for therapeutic intervention.

METHODS

A detailed description of the generation of constructs, primary culture preparation, cell transfection, Western blotting and FRET imaging is reported in the expanded Materials and Methods section in the online data supplement to this article.

RESULTS

sGC and pGC generate spatially distinct cGMP signals

To assess the amplitude of the cGMP signals generated by different stimuli in selected subcellular locales we used FRET imaging and cGMP sensors targeted to the subcellular compartments where PKA-RI and PKA-RII normally reside¹². For this we modified the cGMP sensor Cygnet-2.1²⁸ by fusing the D/D domain of either RIα or RIIβ at its amino-terminus, as previously described¹². As expected¹², the resulting sensors RI_cygnet-2.1 and RII_cygnet-2.1 showed a distinct subcellular localization when expressed in neonatal rat ventriculocytes (NRVMs) (Fig 1A and Online Figure I). Activation of sGC with the NO donor S-Nitroso-N-acetyl-D,L-penicillamine (SNAP, 100 μmol/L) generated a comparable rise in cGMP in the PKA-RI and PKA-RII compartments (Fig 1B). In contrast, upon treatment with the pGC activator atrial natriuretic peptide (ANP, 100 nmol/L) a significantly larger cGMP signal was detected by RII_cygnet-2.1 as compared to RI_cygnet-2.1 (Figure 1B). Thus, different local cGMP signals are generated upon activation of sGC and pGC.

Fig. 1. Activation of sGC or pGC generates different cGMP signals in the PKA-RI and PKA-RII compartments.

(A) Cardiac myocytes co-expressing the targeted cGMP FRET sensor RI_cygnet-2.1 (upper left panel) or RII_cygnet-2.1 (lower left panel) and protein ZASP_RFP¹² (middle panels). The overlay between sensor localization and ZASP-RFP is shown in the right panels. Reported on the right is the intensity profile of the probe signal (in blue) and of the ZASP_RFP signal (in red) in the region indicated by the white line. RI_cygnet-2.1 and RII_cygnet-2.1 show the same localization previously described for RI_Epac and RII_Epac, respectively¹². Scale bars are 10 μm.

(B) Average FRET changes upon treatment with 100 μmol/L SNAP or 100 nmol/L ANP, as indicated. n≥9. Throughout, error bars indicate SEM, unless otherwise stated. *0.01<p<0.05.

cGMP affects the cAMP response to isoproterenol in a compartment-selective manner

Using the targeted cAMP sensors RI_epac or RII_epac¹² we have previously shown that in ventriculocytes PKA-RI and PKA-RII isoforms localize to spatially distinct subcellular compartments within which cAMP signals are generated selectively in response to the activation of different GPCRs¹². Activation of β-ARs generates a pool of cAMP that preferentially activates PKA-RII over PKA-RI¹². To investigate whether cGMP signals may affect the compartment-selective cAMP response to catecholamines we measured the cAMP response elicited by isoproterenol (ISO) in NRVMs expressing either RI_epac or RII_epac¹² in the absence or in the presence of cGMP elevating stimuli. As shown in Fig 2, and in agreement with our previous findings¹², a compartmentalized cAMP response was elicited by ISO (10 nmol/L) with a significantly larger cAMP increase in the PKA-RII compartment as compared to the PKA-RI compartment. Pre-treatment of NRVMs with SNAP (100 μmol/L) led to an inversion of this pattern, such that treatment with ISO led to a preferential increase of cAMP content in the PKA-RI compartment (Fig 2 A, B). The same inversion was observed when cGMP levels were raised by selective inhibition of PDE5 with sildenafil (supplementary information and Online Figure II). A similar effect of SNAP on local cAMP signals was observed in adult rat ventricular myocytes (ARVMs) (Fig 2C).

Fig. 2. Effects of cGMP signals on cAMP levels.

(A) Representative kinetics of cAMP changes detected in NRVMs by the targeted cAMP sensors RI_epac (left panels) or RII_epac (right panels) upon stimulation with 10 nmol/L ISO and subsequent addition of 100 μmol/L IBMX. Where indicated cells were pre-incubated for 10 min with SNAP (100 μmol/L) or SNAP and ODQ (10 μmol/L). (B) Summary of all experiments performed as in A. n≥13 (C) cAMP changes detected in ARVMs expressing RI_epac and RII_epac upon stimulation with 100 nmol/L ISO with or without 100 μmol/L SNAP, as indicated. n≥7. *0.01<p<0.05, **0.001<p<0.01, ***<0.001.

The SNAP-induced inversion is cGMP-dependent as it was completely abolished by the sGC inhibitor 1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one (ODQ, 10μmol/L) (Fig 2A, B). Analysis of the localization of RI_epac and RII_epac presented in the supplementary results excludes that the difference detected by the two probes may be due to a prevalent cytosolic signal revealed by RI_epac as compared to a more specific and local signal revealed by RII_epac (see supplementary results and Online Figure III and IV). In addition, we found that when cAMP levels were detected using the non-targeted cytosolic cAMP sensor Epac1-camps²⁹ no effect of SNAP on the cAMP response to ISO was detectable (Online Figure V). Thus, although cGMP signals dramatically affect the local cAMP response to ISO with opposing effects in the PKA-RI and PKA-RII compartments, such an effect is not apparent when simply measuring global cAMP.

cGMP generated by pGC selectively modulates cAMP levels in the PKA-RII compartment

We next assessed the effect of activation of pGC on local cAMP signals. As shown in Fig 3, and in sharp contrast to SNAP, exposure to 100 nmol/L ANP reduced the cAMP response to ISO in the PKA-RII compartment without affecting the response in the PKA-RI compartment. As before, no significant change in the ISO-dependent cAMP response was detected upon ANP treatment when the untargeted cAMP sensor Epac1-camps was used (Online Figure V) Collectively, these observations show that cGMP affects the cAMP response to ISO, that such an effect is compartment-specific and that this selectivity is attributable to the mechanism through which (and hence the compartment in which) cGMP content is raised.

Fig. 3. ANP reduces the cAMP response to ISO selectively in the PKA-RII compartment.

(A) Representative kinetics of cAMP changes detected by the targeted cAMP reporters RI_epac or RII_epac in NRVMs upon stimulation with 10 nmol/L ISO and 100 μmol/L IBMX. Where indicated cells were pre-treated for 10 min with 100 nmol/L ANP. (B) Summary of all the experiments performed as in A. n≥15. ***p<0.001.

PDE2 and PDE3 mediate the cGMP-dependent modulation of cAMP

cGMP can affect cAMP-mediated signalling by modulating the cAMP-hydrolytic activity of myocardial PDEs: cGMP inhibits the cAMP-hydrolytic activity of PDE3 but stimulates the cAMP-hydrolytic activity of PDE2²². If the PDE3 and PDE2 activity is exerted in the PKA-RI and PKA-RII compartments, respectively, the stimulatory effect of cGMP on cAMP content in the PKA-RI compartment and the inhibitory effect on cAMP content in the PKA-RII compartment could be explained through this mechanism. To test this notion, NRVMs expressing either RI_epac or RII_epac were treated with 10 nmol/L ISO in the presence of 100 μmol/L SNAP and of selective pharmacological inhibitors of either PDE2 or PDE3. As shown in Fig 4A, the PDE2-specific inhibitor BAY 60-7550 (BAY, 10 μmol/L) significantly reduced the effect of SNAP on the ISO-induced cAMP response in the PKA-RII compartment, without affecting the cAMP response in the PKA-RI compartment. These findings support a mechanism in which PDE2 activity is preferentially coupled to the PKA-RII compartment, such that activation of PDE2 by cGMP leads to a selective reduction of cAMP content in this compartment.

Fig. 4. Effect of selective PDE2 and PDE3 inhibition.

(A) cAMP changes induced by ISO in NRVMs expressing the targeted cAMP sensors RI_epac or RII_epac. Where indicated myocytes were pre-treated for 10 min with 100 μmol/L SNAP, 10 μmol/L BAY or 10 μmol/L CILO. n≥15. (B) cAMP changes induced by ISO in NRVMs expressing RI_epac or RII_epac. Where indicated myocytes were pre-treated for 10 min with 100 nmol/L ANP, 10 μmol/L BAY or 10 μmol/L CILO. n≥7. *0.01<p<0.05, **0.001<p<0.01, ***<0.001.

Selective inhibition of PDE3 with cilostamide (CILO, 10 μmol/L) did not affect the SNAP-induced inversion of the cAMP gradient in response to ISO in either compartments (Fig 4A). This finding is in keeping with the hypothesis that PDE3 activity is preferentially coupled to the PKA-RI compartment, such that inhibition of PDE3 by cGMP leads to a selective increase in cAMP content in this locale. In agreement with the notion that, in shaping the cAMP response to ISO, PDE2 and PDE3 exert a predominant activity in the RII and RI compartments respectively, selective inhibition of PDE2 with BAY generates a larger cAMP response to ISO in the PKA-RII compartment whereas inhibition of PDE3 with CILO generates a larger cAMP response in the PKA-RI compartment (Online Figure VIB, C).

In another set of experiments (Fig 4B) we assessed how selective inhibition of PDE2 and PDE3 affects the ANP-mediated modulation of the cAMP response to ISO. Pre-treatment of NRVMs with BAY (10 μmol/L ) completely abolished the effect of ANP in the PKA-RII compartment without affecting the cAMP response in the PKA-RI compartment. Inhibition of PDE3 with 10 μmol/L CILO had no effect on the cAMP response in either compartment. The above data confirm that the effect of ANP on the cAMP response to ISO is dependent on a PDE2 hydrolytic activity that is confined to the PKA-RII compartment.

A localized pool of PDE2 is required for the compartment-specific effects of cGMP on the cAMP signals

The above data suggest that the activity of spatially confined PDEs is critical for the effects of cGMP on cAMP. To test this hypothesis we generated a catalytically inactive mutant of PDE2A (dnPDE2A) by introducing two D to A mutations at positions 685 and 796. These mutations are located in the catalytic site of the enzyme and completely abolish the ability of PDE2 to degrade cAMP (Online Figure VII). When overexpressed in NRVMs the catalytically dead dnPDE2A will displace the cognate endogenous active PDE2A enzymes from specific signalling complexes to which it is sequestered³⁰. Thus, contrary to pharmacological inhibition with BAY, overexpression of dnPDE2A is expected to specifically affect the local activity of PDE2 without affecting the overall PDE2 activity in the cell.

NRVMs co-expressing either RI_epac or RII_epac and dnPDE2A tagged with a red fluorescent protein (mRFP) were pre-treated with 100 μmol/L SNAP and challenged with 10 nmol/L ISO and the cAMP response was detected by FRET analysis. Overexpression of the mRFP-tagged dnPDE2A was found to reverse the effect of SNAP on the PKA-RII compartment (Fig 5A).These results indicate that an active PDE2 associated with the PKA-RII compartment is responsible for the cGMP-dependent reduction of the cAMP response in this locale. In the presence of overexpressed dnPDE2A-RFP the cAMP response in the PKA-RI compartment was significantly lower than observed in the absence of dnPDE2A-RFP (Fig 5A), suggesting that the displaced endogenous PDE2A enzyme diffuses to the PKA-RI compartment where the effects of its cGMP-mediated activation are observed. Interestingly, genetic knock down of PDE2A by small interference RNA blocks the selective effect on local cAMP levels elicited by cGMP in the PKA-RII compartment, confirming that this effect is mediated by PDE2 activity, however does not affect the response in the PKA-RI compartment (Online Figure VIIIB). Overexpression of mRFP-tagged dnPDE2A also abolished the effect of ANP on the cAMP response to ISO in the PKA-RII compartment whereas no effect was detected in the PKA-RI compartment in the same conditions (Fig 5B). Similar results were obtained when PDE2A was knocked down by siRNA (Online Figure VIIIC).

Fig. 5. Effects of displacing endogenous PDE2A.

cAMP changes induced by 10 nmol/L ISO in NRVMs expressing RI_epac or RII_epac with or without dnPDE2A-mRFP or PDE2A_D485A, as indicated. Where indicated cells were pre-treated for 10 min with 100 μmol/L SNAP (A) or 100 nmol/L ANP (B). n≥9, *0.01<p<0.05, **0.001<p<0.01, ***<0.001.

To further confirm the role of a spatially confined PDE2 enzyme in the cGMP mediated modulation of the local cAMP response to ISO, we generated a PDE2A mutant (PDE2A_D485A) in which a D to A mutation was introduced at position 485. This mutation has been shown to abolish the ability of the GAF-B domain to bind cGMP³¹. Overexpression of PDE2A_D485A can be expected to displace the endogenous, cGMP-sensitive enzyme from its intracellular anchor sites replacing it with a cGMP-insensitive version of the enzyme. As shown in Fig 5A, overexpression of mRFP-tagged PDE2A_D485A completely counteracts the effects of SNAP on the cAMP response to ISO in the PKA-RII compartment. Similarly, the expression of mRFP-tagged PDE2A_D485A completely abolishes the effect of ANP on cAMP levels in PKA-RII compartment (Fig 5B). Taken together, these data provide strong evidence that PDE2 is instrumental in effecting the cGMP-mediated control on cAMP levels and that sequestered PDE2 localization in the PKA-RII compartment is required for achieving the observed compartment-selective modulation of cAMP signals by cGMP.

cGMP signals affect local, isoform-specific PKA activation and the phosphorylation of downstream targets

Our aim here was to assess whether the effect of cGMP signals on the local cAMP response to ISO would also impact on the downstream activation of individual PKA isozymes. To address this, PKA isoform-selective activity was measured using the FRET sensors RI_AKAR3 or RII_AKAR3. These are genetically modified versions of the FRET-based PKA activity reporter AKAR3³² in which the D/D domain from either RIα or RIIβ was fused at the amino-terminus of the sensor in order to achieve selective targeting (Fig 6A and Online Figure IX). In agreement with our previous findings¹², and in keeping with our present results, Fig 6B shows that NRVMs expressing either RI_AKAR3 or RII_AKAR3 and challenged with 0.5 nmol/L ISO have a significantly higher PKA-RII activity compared to PKA-RI. In striking contrast, when cells were pretreated with 100 μmol/L SNAP, ISO selectively increased the activity of PKA-RI over PKA-RII. The effect of ISO on local PKA activity was completely abolished when the sGC was inhibited with 10 μmol/L ODQ (Fig 6B). In keeping with these findings, SNAP treatment also significantly reduced the phosphorylation level of the PKA-RII selective targets TnI and PLB (supplementary information and Online Figure X). Similarly, treatment with 100 nmol/L ANP selectively reduced the ISO-induced phosphorylation activity of PKA-RII, with no effect on the activity of PKA-RI (Fig 6C) and reduced the phosphorylation level of TnI and PLB (supplementary information and Online Figure XI). In agreement with these findings, selective pharmacological inhibition of PDE2 with BAY resulted in an increase in the ISO-induced phosphorylation of TnI and PLB, whereas PDE3 inhibition with CILO showed a smaller and not significant effect (Online Figure XII).

Fig. 6. Effect of SNAP on PKA isoforms activity.

(A) Cardiac myocytes expressing the targeted PKA activity reporter RI_AKAR3 or RII_AKAR3 (left panels) and the marker ZASP_RFP (middle panels). The overlay between probe localization and ZASP-RFP is shown in the right panels. On the right the intensity profile of the probe signal (in blue) and of the ZASP_RFP signal (in red) in the region indicated by the white line is reported. RI_AKAR3 and RII_AKAR3 show the same localization as RI_cygnet-2.1 and RII_cygnet-2.1, respectively. Scale bars are 10 μm.

(B) cAMP changes detected by RI_AKAR3 or RII_AKAR3 upon application of 0.5 nmol/L ISO. Where indicated cells were pre-incubated with 100 μmol/L SNAP or SNAP and 10 μmol/L ODQ. n≥7, *0.01<p<0.05, **0.001<p<0.01, ***<0.001.

The above data demonstrate that the compartment-selective modulation of cAMP levels by cGMP signals propagates to the downstream effectors of cAMP, PKA-RI and PKA-RII, resulting in a dramatic effect on the selective activation of PKA isoforms and target phosphorylation.

cGMP activation of PDE2 affects the contractile response to ISO

To assess the physiological relevance of the above findings we measured the effect of raising cGMP levels with SNAP on the ISO-induced contractility response of ARVM. We found that treatment with 100μmol/L SNAP significantly reduced myocytes shortening induced by 10 nmol/L ISO (Fig 7A), an effect that was ablated by selective inhibition of PDE2 with 10μmol/L BAY (Fig 7B).

Fig. 7. Functional effects of cAMP/cGMP interplay.

(A) Example tracing showing the fractional shortening recorded in isolated ventriculocytes obtained from adult hearts in the absence or in the presence of 10 nmol/L ISO in control cells and cells treated with 100 μmol/L SNAP. (B) Summary of the results obtained from three independent experiments performed as in (A). (C) Summary of the results obtained in three independent experiments in which cells were treated as in A but pretreated with BAY 10 μmol/L. n≥11. *=p<0.05, ns = non significant.

DISCUSSION

In the present study we examined the effect of cGMP signals on the cAMP response to ISO in rat ventricular myocytes. We found that, although cGMP signals do not change the overall cellular cAMP response, they dramatically affect the cAMP signals locally. Specifically, we found that an increase in cGMP leads to a significant increase in the cAMP pool that activates PKA-RI and to a significant reduction in the cAMP pool that activates PKA-RII. The specific local effect of cGMP on cAMP levels depends on the mechanism through which cGMP content is raised and requires local PDE2 and PDE3 activities. The effect of cGMP propagates to the downstream activation of the cAMP effectors PKA-RI and PKA-RII, leading to altered phosphorylation of PKA targets and affecting myocytes contractility. Based on these findings, we propose a model for cardiac myocytes in which, in the presence of catecholamines, PDE2 activity is preferentially coupled to the PKA-RII compartment while PDE3 activity is preferentially coupled to the PKA-RI compartment and in which cGMP-mediated activation of PDE2 and cGMP-mediated inhibition of PDE3 are responsible for the opposing effects of cGMP on local cAMP signals (Fig 8). This model may also be relevant for the in vivo functions of other cell systems, such as smooth muscle cells and endothelial cells, in which PDE2 and PDE3 enzymes are co-expressed ³³.

Fig. 8. Model of cGMP-mediated modulation of local cAMP signals.

(A) β-AR stimulation generates a spatially restricted pool of cAMP that preferentially activates PKA-RII over PKA-RI. (B) Activation of sGC generates a uniform increase in cGMP in both PKA-RI and PKA-RII compartments. Increased cGMP levels inhibit PDE3, which is mainly confined to the PKA-RI compartment, and at the same time activate PDE2 associated with the PKA-RII compartment, thereby leading to an inversion of the cAMP gradients in response to catecholamines. (C) Activation of pGC generates a local pool of cGMP that specifically affects the PKA-RII compartment and blunts the cAMP response to catecholamines via PDE2 activation selectively in this locale.

Our present findings clearly demonstrate that cGMP-mediated activation of a spatially confined PDE2 activity is responsible for the cGMP-mediated regulation of cAMP levels in the PKA-RII compartment. Consistent with this model, even in the absence of cGMP signals, selective inhibition of PDE2 generates a cAMP response to ISO that is larger in the PKA-RII than in the PKA-RI compartments. With respect to the compartmentalization of PDE3, the situation is more complex. The observation that pharmacological inhibition of PDE3 does not significantly affect the cAMP response in the PKA-RII compartment suggests that the contribution of PDE3 to shaping the cAMP response to ISO in this locale is marginal. It must be noted, however, that while low cGMP concentrations (< 50 nmol/L) exclusively inhibit PDE3, at higher concentrations - between 200 to 500 nmol/L - cGMP also activates PDE2³⁴. At high cGMP concentrations, therefore, a possible inhibitory effect of cGMP on PDE3 activity in the PKA-RII compartment may be masked by a prevalent activating effect on PDE2 in this compartment.

The functional consequences of raising cGMP content by activation of pGC and or sGC are different. One possible explanation for this is that cGMP levels increase in specific subcellular compartments depending on the stimulus, thus regulating different phosphodiesterases. A previous study assessing cGMP in the sub-plasma membrane compartment of rat ventriculocytes showed that the particulate cGMP pool is readily accessible at the plasma membrane, whereas the soluble pool is not¹⁶. Our study, which extends the analysis to compartments localized deep inside the myocytes, confirms that the diffusion of cGMP within the cell is spatially regulated, and the nature of the cGMP-raising stimulus determines which compartment is affected: NO donors, which activate sGC, affect both the PKA-RI and PKA-RII compartments, whereas ANP, which activates pGC, generates a rise in cGMP that is restricted to the PKA-RII compartment. It is worth noting that endogenous NO is generated in a temporally and spatially restricted manner via activation of different NO synthase (NOS) isoforms with distinct subcellular localization, activation mechanisms and regulation³⁵, with a potential to generate local cGMP signals that are more complex than those revealed by treatment with exogenous NO donors.

The modulatory effects of NO on cardiac myocytes function are complex and still controversial³⁵. For example, NO exerts a positive lusitropic effect that has been attributed to cGMP/PKG-mediated phosphorylation of TnI³⁶. On the other hand, NO appears to have mainly negative inotropic and chronotropic effects³⁷. NO has been shown to regulate contractility in a bimodal fashion, both in basal conditions³⁸ and under β-adrenergic drive³⁹, with low concentrations of NO exerting a positive effect and high concentrations of NO exerting a negative effect⁴⁰. The specific molecular mechanisms underlying these opposing functions remain unclear, and a variety of possibilities have been implicated, mainly involving PKG-dependent mechanisms³⁶ or protein nitrosylation⁴¹. Here we confirm that treatment with high concentration (100 μmol/L) SNAP results in reduced myocytes contractility and our findings indicate that the effects of NO on contractility may be due, at least in part, to the PDE2-mediated modulation of local cAMP levels.

In most studies, the effects of ANP on contractility have been negative, both in vitro⁴² and in vivo⁴³, although these observations are not universal⁴⁴. The effects of natriuretic peptides on cardiac contractility are important as these peptides become available for clinical use to treat HF. Our results show that ANP, via cGMP-mediated activation of a spatially confined PDE2, significantly reduces cAMP levels in the PKA-RII compartment, a domain that is involved in the control of excitation-contraction coupling⁶, suggesting that cGMP-mediated activation of PDE2 may contribute to the negative inotropy associated with ANP.

The crosstalk between NO/cGMP and adrenergic/cAMP signalling is clinically important because these two pathways are often concomitantly targeted to treat cardiovascular decompensation. cGMP-raising agents are valuable as vasodilators in clinical practice, and there is evidence for direct cardioprotective effects in animal models⁴⁵. As NO-mimetic agents, however, nitrates also influence cardiac contractility by blunting the positive response to adrenergic signalling⁴⁶, an effect that is even more pronounced in failing hearts⁴⁷. Our study reveals that the positive and negative effects of cGMP on cAMP levels occur in spatially distinct compartments, via modulation of the activity of spatially confined PDEs. In particular, cGMP-mediated activation of PDE2 in the PKA-RII compartment leads to reduced myocytes contractility. The physical separation of positive and negative effects of cGMP on cAMP levels presented here offers the previously unrecognized possibility that the negative inotropic effects associated with increases in intracellular cGMP could be prevented. From this perspective, inhibition of PDE2 might limit the negative inotropic effects of cGMP-raising agents, increasing their clinical utility. On the other hand, stimulation of the cAMP-hydrolytic activity of PDE2, by blocking rises in cAMP, may contribute to the cardioprotective actions of cGMP-raising agents. Expression, intracellular localization and activity of NOS isoforms have been shown to change in the presence of cardiac disease⁴⁸ potentially affecting the cGMP-mediated regulation of local cAMP signals, however whether such local changes contribute to pathology, or may rather be protective remains to be determined. Further experiments in disease models and ultimately in human cardiac myocytes will be necessary to distinguish between these two possibilities.

NOVELTY AND SIGNIFICANCE

What is known?

• There is evidence that signalling mediated by the cyclic nucleotides cAMP and cGMP is compartmentalized with spatially distinct pools of these second messengers affecting different cellular functions.

• phosphodiesterases (PDEs), the enzymes that degrade cAMP and cGMP, play a key role in spatial control of cyclic nucleotide signals.

• cGMP can potentially modulate cAMP levels by activating PDE2 and inhibiting PDE3.

What new information does this article contribute?

• In neonatal rat ventricular myocytes (NRVMs) cGMP modulates the cAMP response to isoproterenol (ISO) in a compartment specific manner

• The functional impact of cGMP signals depends on the source of cGMP (pGC or sGC) and relies on the specific PDEs associated with each compartment (PDE2 or PDE3).

• cGMP generated by sGC decreases cAMP levels via PDE2 activation and increases cAMP levels via PDE3 inhibition.

• cGMP generated by ANP-mediated activation of pGC selectively decreases cAMP levels via PDE2 activation.

NO/cGMP and adrenergic/cAMP signalling pathways are common targets in the treatment of cardiovascular decompensation. cGMP can activate PDE2 and inhibit PDE3; therefore stimuli that elevate cGMP can either attenuate or enhance cAMP signals with potentially very different functional effects. The interplay between cAMP and cGMP signalling pathways in cardiac myocytes has not been fully elucidated and this knowledge is critical to better target therapy.

Using real-time imaging of intact myocytes we investigated the impact of cGMP signals on the cAMP response to ISO. We found that, depending on the cyclase that generates it, cGMP differently modulates the local cAMP response in distinct subcellular compartments, affecting protein phosphorylation and myocytes contractility. By acting via differently localized PDEs, cGMP can exert opposing effects on local cAMP levels, via activation of PDE2 in one compartment and inhibition of PDE3 in another compartment.

The physical separation of positive and negative modulation of cAMP levels by cGMP revealed in this study offers the previously unrecognized possibility to selectively modulate local cAMP signals to obtain more specific functional effects and, potentially, to reduce undesired effects. For example, inhibition of PDE2 might limit the negative inotropic effects of cGMP-raising agents, increasing their clinical utility.

NON-STANDARD ABBREVIATIONS AND ACRONYMS

AC

Adenylyl cyclase

AKAP

A Kinase anchoring protein

ANP

atrial natriuretic peptide

β-AR

βeta adrenergic receptor

cAMP

3′-5′- cyclic adenosine monophosphate

cGMP

3′-5′- cyclic guanosine monophosphate

D/D

dimerisation/docking domain

FRET

Fluorescence resonance energy transfer

GPCR

G protein coupled receptor

HF

heart failure

IBMX

3-isobutyl-1-methylxanthine

NO

nitric oxide

NOS

nitric oxide synthetase

NRVMs

neonata rat ventricular myocytes

ODQ

1H-[1,2,4]oxadiazolo[4,3-a]quinoxalin-1-one

PDE

phosphodiesterase

pGC

particulate guanylyl cyclase

PKA

protein kinase A

PKG

protein kinase G

PLB

phospholamban

sGC

soluble guanylyl cyclase

SNAP

S-Nitroso-N-acetyl-D,L-penicillamine

TnI

troponin I

AKAR

A kinase activity reporter

Cygnet

cyclic GMP indicators using energy transfer

Epac

Exchange factor activated by cAMP