Calmodulin kinase II inhibition limits the pro-arrhythmic Ca 2 1 waves induced by cAMP-phosphodiesterase inhibitors

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - Keywords pharmacological inhibition of PDE4 shown the pro-arrhythmic effect of b -AR stimulation in mouse ventricular and atrial strips. In genetic ablation of Pde4b or Pde4d the suscepti-bility stress-induced tachycardia. These phenotypes with exacerbated b of Ca 2 + inﬂux in Pde4b mice and PKA-dependent hyperphosphorylation of RyR2 in mice. In the hyperphosphorylation PLB


Introduction
Upon stress or during exercise, norepinephrine released by sympathetic nerve terminals activates the b-adrenergic receptors (b-ARs) to elicit positive inotropic, chronotropic, and lusitropic effects. b-ARs couple primarily to G as proteins, leading to stimulation of adenylyl cyclases and cyclic adenosine monophosphate (cAMP) production. In turn, cAMP activates the cAMP-dependent protein kinase A (PKA), which phosphorylates key proteins of the cardiac excitation -contraction coupling (ECC) process, including the sarcolemmal L-type Ca 2+ channels (LTCCs), the ryanodine receptors (RyR2) of the sarcoplasmic reticulum (SR), and phospholamban (PLB), a constitutive inhibitor of the SR Ca 2+ pump, SERCA2. As a result of PKA phosphorylation, Ca 2+ entry through the sarcolemma is increased, as well as Ca 2+ release and refill from the SR, leading to enhanced Ca 2+ cycling and consequently to inotropic and lusitropic effects. 1 A tight control of Ca 2+ homeostasis is essential as perturbations such as diastolic SR Ca 2+ leak via RyR2 increase occurrence of spontaneous Ca 2+ waves (SCWs) leading to electrogenic sodium-Ca 2+ exchanger currents, causing delayed after depolarizations which are able to trigger action potentials when the threshold for Na + channel activation is reached. 2 Along with PKA, the Ca 2+ /Calmodulin-dependent kinase II (CaMKII) has been identified over the past years as a contributor to b-AR regulation of cardiac function. 3 Indeed, LTCC, RyR2, and PLB are also substrates for CaMKII, which contributes to the inotropic and lusitropic effects of b-AR agonists. This is especially true upon excessive b-AR activation as occurring under pathological conditions, notably in heart failure (HF), in which chronic b-AR stimulation is accompanied by perturbations of the cAMP signalling pathway 4 and increased CaMKII activity. 5 Interestingly, CaMKII has been identified as the main suspect to provoke the Ca 2+ handling disturbances observed upon excessive b-AR stimulation under physiological conditions and in HF. 6 -8 Thereafter, Epac (exchange protein directly activated by cAMP), a direct target of cAMP, has emerged as a link between b-AR/cAMP signalling and CaMKII activation to promote, independently of PKA, a pro-arrhythmogenic SR Ca 2+ leak. 9 -11 The cAMP concentration in cells is orderly regulated not only by its synthesis but also through its degradation by cyclic nucleotide phosphodiesterases (PDEs). 12 Among the 11 PDE families identified in mammals, PDE3 and PDE4 represent 70 -90% of the total cAMP hydrolytic activity in heart. 13 -15 Given that PDE3 dominates in human heart 16 and b-ARs are desensitized in HF, 4 PDE3 inhibitors were identified as useful inotropes to boost the failing pump 17 and to prevent post-operative low cardiac output syndrome after cardiac surgery. 18 Mechanistically, PDE3 inhibitors enhance either the basal or the pre-stimulated Ca 2+ current (depending on species and/or cardiac territory) and Ca 2+ reuptake by the SR. 16,19 However, their use is now limited to advance disease states, mainly acute HF, as chronic treatment with these agents has been incriminated in causing arrhythmias, hence augmenting mortality of treated patients. 18,20 PDE4 is overriding in rodent cardiomyocytes to control ECC, especially upon b-AR stimulation, 13 but also contributes to cAMP degradation in larger species as shown in dogs 21 and human cardiomyocytes. 22 The pharmacological inhibition of PDE4 was shown to enhance the pro-arrhythmic effect of b-AR stimulation in mouse ventricular 23 and human atrial strips. 22 In mice, genetic ablation of Pde4b or Pde4d genes enhances the susceptibility to stress-induced ventricular tachycardia. 24,25 These phenotypes were associated with exacerbated b-AR stimulation of Ca 2+ influx in Pde4b-deficient mice 24 and PKA-dependent hyperphosphorylation of RyR2 in Pde4d-deficient mice. 25 In the latter, hyperphosphorylation of PLB was also reported. 24,26 Although the contribution of PDEs to confine PKA activity in cardiac cells is well documented, no study reported on the implication of CaMKII in the arrhythmias evoked by PDE inhibitors. Recently, a signalosome including b-AR/Epac/CaMKII organized around the scaffold protein b-arrestin was found to include a PDE4 isoform, 27 suggesting a possible role for PDE4 in controlling this pro-arrhythmic signalling pathway. A link between PDE4 and CaMKII is also suggested by the recent finding that CaMKII can phosphorylate and activate PDE4 activity in cardiomyocytes. 28 Collectively, these studies suggest intricate entanglement of the cAMP/PKA and cAMP/Epac/CaMKII pathways, both of which are under PDE4 control. Particularly, they suggest that Ca 2+ handling alterations observed upon PDE inhibition may involve CaMKII activation via the classical cAMP/PKA pathway and/or Epac. This prompted us to delineate in more detail the cellular mechanisms responsible for the pro-arrhythmic effects of PDE4 and PDE3 inhibitors and to elucidate the underlying signalling pathways of these Ca 2+ disturbances.

Methods
An expanded methods section is provided in the Supplementary material online.
All experiments were carried out according to the European Community guiding principles in the Care and Use of Animals (2010/63/UE, 22 September 2010), the local Ethics Committee (CREEA Ile-de-France Sud) guidelines, and the French decree no. 2013-118, 1 February 2013 on the protection of animals used for scientific purposes (JORF no. 0032, 7 February 2013, p2199, text no. 24). Authorizations to perform animal experiments according to this decree were obtained from the Ministère français de l'Agriculture, de l'Agroalimentaire et de la Forêt (agreement no. B 92-019-01).
Adult rat ventricular myocytes (ARVMs) were obtained using retrograde Langendorff perfusion and 1 mg/mL of collagenase A (Roche Diagnostics GmbH, Mannheim, Germany) (378C). An IonOptix system was used to record intracellular Ca 2+ and sarcomere shortening (SS) simultaneously in ARVMs loaded with 1 mM Fura-2 and paced at 1 Hz, as described previously. 29

PDE4 inhibition potentiates the inotropic effects of b-AR stimulation but promotes SCWs in ARVMs
To investigate the role of PDE4 in controlling the ECC and pro-arrhythmic effects of its inhibition, Ca 2+ transients (CaT) and SS were simultaneously recorded in ARVMs loaded with 1 mM Fura-2 and paced at 1 Hz ( Figure 1A). Under control conditions (Ctrl), average diastolic sarcomere length was 1.75 + 0.01 mm. Mean CaT amplitude was 38.9 + 4.0% above the basal Fura-2 ratio, and SS was 2.1 + 0.6%. CaT and SS declined to diastolic levels with time constants (t) estimated at 0.33 + 0.02 and 0.29 + 0.03 s, respectively ( Figure 1B and C ). Although PDE4 inhibition with Ro 20-1724 (Ro, 10 mM) affected neither the basal amplitude of CaT nor SS, it accelerated the relaxation kinetics of both Ca 2+ and shortening twitches to 0.26 + 0.02 and 0.17 + 0.02 s, respectively (P , 0.05 vs. ctrl, Figure 1C). b-AR stimulation by isoproterenol (Iso, 1 nM) increased the amplitude of CaT and SS by approximately 3-and 6-fold, respectively (P , 0.001 vs. Ctrl, Figure 1B). Iso also strongly accelerated the relaxation rates of both parameters by 54.5 + 6.1 and 86.2 + 1.4% (P , 0.001 vs. Ctrl, Figure 1C). These inotropic and lusitropic effects were potentiated by PDE4 inhibition: CaT was further increased by 19.5 + 4.3% and SS by 39.7 + 4.8% in Iso + Ro (P , 0.001 vs. Iso, Figure 1B). Decay time constants also tended to decrease, but this did not reach statistical significance in comparison with Iso alone ( Figure 1C). When PDE4 was inhibited in the absence of b-AR agonist, the diastolic Fura-2 ratio remained unchanged and none of the cells exhibited SCWs upon cessation of stimulation ( Figure 1D). When cells were subjected to Iso, only sparse SCWs were observed (0.4 + 0.3 per 10 s) in 20% of the cells. However, when Ro was applied in combination with Iso, a 14.7 + 1.2% elevation of the diastolic Fura-2 ratio was observed (P , 0.001 vs. Iso), and all cells exhibited pro-arrhythmogenic SCWs at a frequency of 2.3 + 0.2 per 10 s (P , 0.001 vs. Iso, Figure 1D and Supplementary material online, Table S1).

PDE4 inhibition upon b-AR stimulation leads to diastolic SR Ca 21 leak
To further delineate the cellular mechanisms underlying the SCWs evoked by PDE4 inhibition upon b-AR stimulation, we measured the SR Ca 2+ leak using a 0Na + /0Ca 2+ solution to prevent Ca 2+ extrusion by the Na + /Ca 2+ exchanger, and tetracaine (1 mM) to block RyR2s. A rapid application of 10 mM caffeine at the end of the experiment was used to evaluate SR Ca 2+ load ( Figure 2A). Although negligible under basal conditions, an SR Ca 2+ leak appeared upon b-AR stimulation with 1 nM Iso (P , 0.05 vs. Ctrl, Figure 2B), which correlated with an increased SR Ca 2+ load (+33 + 5.6%, P , 0.001 vs. Ctrl, Figure 2A -C ). Interestingly, additional PDE4 inhibition drastically increased SR Ca 2+ leak (+147.6 + 33.3%, P , 0.001 vs. Iso, Figure 2A and B), whereas modestly increasing SR Ca 2+ load (+20.6 + 7.8%, P , 0.05 vs. Iso, Figure 2C). Fractional release (the ratio of Ca 2+ released during a twitch divided by SR Ca 2+ load) was increased from 52% in Ctrl to 62% in Iso (P , 0.05 vs. Ctrl) and up to 72% in Iso + Ro (P , 0.05 vs. Iso, Figure 2D). , and proportion of cells with SCWs under Ctrl, Ro, Iso, and Iso + Ro. Number of cells obtained from three to four rats is indicated inside each bar representing the mean. Statistical significance is indicated as: *P , 0.05; **P , 0.01; ***P , 0.001 (vs. Ctrl); $$ P , 0.01; $$$ P , 0.001 (Ro vs. Iso + Ro); ### P , 0.001 (Iso vs. Iso + Ro). Analysis of variance (ANOVA) followed by a Tukey post hoc test was used to analyse CaTs, sarcomere amplitude, and diastolic Fura-2 ratio. ANOVA followed by a Dunn post hoc test was used to analyse the occurrence of SCWs, and x 2 test followed by a Fisher's exact test was used to compare the per cent of cells with SCWs.

Respective role of PKA and CaMKII in the dysregulation of Ca 21 homeostasis induced by PDE4 inhibition
When cells were challenged with Iso (1 nM) alone, an increase in the phosphorylation of RyR2 at Ser2808 and PLB at Ser16 (PKA sites) assessed by western blot was observed, which became significant with concomitant PDE4 inhibition (Supplementary material online, Figure  S1A and B). Similarly, phosphorylation of both RyR2 at Ser2814 and PLB at Thr17 (CaMKII sites) became significant only when cells were stimulated with a combination of Iso + Ro (Supplementary material online, Figure S1C and D). These results indicate that both PKA and CaMKII are activated upon PDE4 inhibition. We next evaluated the relative contribution of both kinases in the pro-arrhythmogenic effects of Iso + Ro. For this, cells were first pre-incubated for 10 min prior to the experiment and throughout data acquisition in a Ringer solution containing the PKA inhibitor H-89 at 10 mM. This concentration was chosen as it abolishes ≈75% of the kinase activity induced by Iso + Ro in ARVMs, whereas 1 mM had minor effects as determined by fluorescence resonance energy transfer using the PKA reporter AKAR3 30 (Supplementary material online, Figure S2). H-89 abolished the increase in both CaT and SS induced by b-AR stimulation alone or with concomitant PDE4 inhibition ( Figure 3A and B, P , 0.001 vs. without inhibitor), whereas under basal conditions, H-89 did not affect the mean CaT amplitude but slowed the return to diastolic Ca 2+ levels (P , 0.05 vs. Ctrl) without significantly affecting SS ( Figure 3B and C ). The lusitropic effects of Iso + Ro were also suppressed in these conditions ( Figure 3C, P , 0.001 vs. without inhibitor). The increase in diastolic Ca 2+ levels observed in Iso + Ro was blunted by PKA inhibition, and none of the cells exhibited SCWs (Figure 3 and Supplementary material online, Table S2, P , 0.001 vs. Ctrl). Similarly, adenoviral transduction of ARVMs with PKI, a highly specific inhibitory peptide of PKA, 31 did not modify basal CaT and SS when compared with control ARVMs infected with an adenovirus encoding b-galactosidase (b-Gal), but abolished the inotropic and lusitropic effects of Iso + Ro, as well as the diastolic Ca 2+ overload and the occurrence of SCWs (P , 0.001 vs. b-Gal, Supplementary material online, Figure S3 and Table S2).
In order to test the contribution of CaMKII, the effects of Iso and Iso + Ro were analysed in cells pre-incubated for 10 min with either 10 mM of the CaMKII inhibitor, KN-93 or 10 mM of its inactive analogue, KN-92 ( Figure 4). In the presence of KN-93, Iso increased CaT amplitude by 113.8 + 37.5% and SS by 392.6 + 113.1%, similar to what was recorded in cells pre-incubated with KN-92. PDE4 inhibition potentiated the effect of Iso on CaT amplitude and SS, which were additionally increased by 87.7 + 23.5 and 91.3 + 7.0%, respectively, upon Iso + Ro (P , 0.001 vs. Iso, Figure 4B). Decay times of Ca 2+ and contraction twitches were similarly decreased by Iso and further accelerated by Ro in cells pre-incubated with KN-93 or KN-92 ( Figure 4C). Interestingly, although KN-93 did not prevent the inotropic and lusitropic effects of the PDE4 inhibitor, it reduced the number of cells exhibiting SCWs by 50% and the occurrence of SCWs by 71.6 + 12.0% ( Figure 4D, P , 0.05 vs. KN-92). KN-93 also prevented the increase in diastolic Ca 2+ levels observed in cells pre-incubated with KN-92 (P , 0.05 vs. Ctrl, Figure 4D and Supplementary material online, Table S3). Finally, pre-incubation of the cardiomyocytes with 100 nM of the autocamtide-2-related inhibitory peptide (AIP), a highly potent and specific substrate competitive inhibitor of CaMKII, recapitulated the inhibitory effects of KN-93 on the pro-arrhythmogenic Ca 2+ events (Supplementary material online, Figure S4 and Table S3). The relative contribution of PKA and CaMKII to the setting of the SR Ca 2+ leak and the increase of the SR Ca 2+ load promoted by PDE4 inhibition were then assessed using the same protocol, as described in Figure 2. As shown in Figure 5, H-89 abolished the SR Ca 2+ leak and decreased the SR Ca 2+ load measured upon Iso + Ro by 47.1 + 3.0% (P , 0.001, Figure 5B and C ). Fractional release was drastically diminished by PKA inhibition from 69 + 3% in Iso + Ro alone to  Figure 5B). However, SR Ca 2+ load was slightly increased in KN-93 ( Figure 5C), and consequently, fractional Ca 2+ release was decreased ( Figure 5D, P , 0.01  vs. KN-92).

The exchange protein directly activated by cAMP Epac2 participates to the arrhythmias induced by PDE4 inhibition
To precise the signalling pathway leading to CaMKII activation upon PDE4 inhibition, we pre-incubated the cardiomyocytes with either SCWs in Ctrl, Iso, and Iso + Ro + H-89. Statistical significance is indicated as: **P , 0.01; ***P , 0.001 (vs. Ctrl); ### P , 0.001 (Iso vs. Iso + Ro); $ P , 0.05; $$ P , 0.01; and $$$ P , 0.001 (without inhibitor vs. H-89) and was tested using an ANOVA followed by a Tukey test for CaT and SS, or by a Dunn test to analyse the occurrence of SCWs and a x 2 test followed by a Fisher's exact test for differences in the percentage of cells with SCWs. NS, non-significant.
10 mM CE3F4 to inhibit Epac1 32 or 5 mM ESI-05 to inhibit Epac2 33 ( Figure 6A). None of the two Epac inhibitors affected the amplitude of CaT or SS under different experimental conditions ( Figure 6B and C and Supplementary material online, Table S4). However, although the occurrence of SCWs was not impacted by Epac1 inhibition ( Figure 6D), ESI-05 diminished SCWs occurrence in Iso + Ro by 41.2 + 8.8% ( Figure 6E, P , 0.05 vs. vehicle) and decreased SR Ca 2+ leak by ≈32%, whereas neither SR Ca 2+ load nor fractional release was affected by Epac2 inhibition (Supplementary material online, Figure S5).

CaMKII inhibition prevents the SR Ca 21 leak promoted by PDE3 inhibition upon b-AR stimulation
The aforementioned results clearly identify CaMKII as a mediator of the pro-arrhythmic effects elicited by PDE4 inhibition. In order to know whether this mechanism is specific to PDE4, we studied the effects of PDE3 inhibition on Ca 2+ homeostasis ( Figure 7). As expected, PDE3 inhibition with cilostamide (Cil, 1 mM) enhanced the positive inotropic effects of Iso (data not shown, but see Mika et al. 29 ). Cil potentiated SR Ca 2+ leak by 52.8 + 15.7% and SR Ca 2+ load by 23.0 + 6.8% ( Figure 7B, P , 0.05 vs. Iso) and significantly promoted fractional release ( Figure 7B, P , 0.01 vs. Iso). To assess whether CaMKII inhibition prevents the arrhythmogenic SR Ca 2+ leak induced by Iso + Cil, cells were preincubated with KN-93 or KN-92. CaMKII inhibition did not modify the stimulatory effect of Iso + Cil on CaT amplitude and SS ( Figure 7C), but largely prevented SR Ca 2+ leak ( Figure 7D, P , 0.001 vs. KN-92) and slightly increased the SR Ca 2+ load, and these effects were accompanied with decreased fractional release ( Figure 7D, P , 0.001 vs. KN-92).

Discussion
PDE3 and PDE4 are the main PDE families degrading cAMP generated upon b-AR stimulation in the heart. Cardiotonic drugs such as milrinone target these enzymes and more specifically the PDE3 family to improve cardiac output in HF or after surgery for congenital cardiac diseases. 18 By elevating cAMP, these treatments have beneficial haemodynamic actions but promote sudden cardiac death due to arrhythmias, which seriously compromise their use. 18,20,34 PDE4 downregulation also enhances ECC and abnormal pro-arrhythmic spontaneous Ca 2+ release events. 22,24,25 This study provides new insights into the underlying cellular mechanisms of these electrophysiological perturbations associated with the use of PDE inhibitors. We show that upon b-AR stimulation, the positive inotropic effects of PDE4 inhibition are accompanied by increased SR Ca 2+ load and leak, leading to elevated diastolic Ca 2+ levels and increased occurrence of SCWs. We demonstrate that PDE4 inhibition exerts positive inotropic effects via PKA but induces a pro-arrhythmogenic SR Ca 2+ leak via both PKA and CaMKII, the latter being activated via the classical cAMP/PKA pathway and to some extent via Epac2. Finally, we show that CaMKII inhibition can also prevent the SR Ca 2+ leak promoted by PDE3 inhibitors. Thus, our results show that inhibiting CaMKII could prevent the pro-arrhythmic effects of PDE inhibitors while preserving their beneficial inotropic effects.
PDE3 or PDE4 inhibition promotes the inotropic effects of a nonmaximal b-AR stimulation by increasing PKA phosphorylation of LTCC, RyR2, PLB, and contractile proteins. 29 The increase in CaT amplitude is consistent with an enhanced Ca 2+ entry via the LTCC 35 and the increased SR Ca 2+ load. 36 PLB phosphorylation by PKA at Ser16, which is essential for b-AR stimulation of ECC, 37 is potentiated by PDE4 inhibition (as shown in this study) as well as by PDE3 inhibition. 29 PKA phosphorylation of RyR2 at Ser2808, even though controversial, 38 has been shown to promote its sensitivity to Ca 2+ . 39 Altogether, these mechanisms can explain the increased SR Ca 2+ load and fractional release observed upon PDE4 and PDE3 inhibition. Accordingly, H-89 at 10 mM, the concentration required to inhibit most of the PKA activity in ARVMs under our experimental conditions (Supplementary material online, Figure S2), prevented all the inotropic and lusitropic effects of PDE4 inhibition. However, H-89 at this concentration can also inhibit SERCA activity. 40 This could explain the slower relaxation of CaT observed in H-89, and by limiting the SR Ca 2+ load, it could have precluded the PKA-independent stimulatory effects of Epac on ECC reported earlier. 11 Nonetheless, overexpression of PKI, a more specific PKA inhibitor but equally effective to inhibit the kinase in cardiomyocytes, 30 consistently abolished the inotropic and lusitropic effects of b-AR stimulation with or without concomitant PDE4 inhibition (Supplementary material online, Figure S3). These results and the lack of effect of Epac and CaMKII inhibitors on CaT amplitude and shortening shown in Figures 4 and 6 further demonstrate that PKA is responsible for the cardiotonic effects of PDE4 inhibition, in accordance with the preserved b-AR stimulation of ECC observed in mice invalidated for Epac 10 or CaMKII. 41 Upon submaximal b-AR stimulation, PDE4 inhibition increases diastolic Ca 2+ levels and the occurrence of SCWs, suggesting SR Ca 2+ leak via RyR2. This was corroborated by the acute block of RyR2 with tetracaine, resulting in increased shift of Ca 2+ back to the SR. 8 It is known that upon b-AR stimulation, not only PKA but also CaMKII is activated to promote SR Ca 2+ leak. 3 Alongside PKA and in agreement with a recent report, 28 CaMKII is also activated upon PDE4 inhibition as demonstrated by PLB and RyR2 phosphorylation on Thr17 and Ser2814, respectively (Supplementary material online, Figure S1). Unlike what has been reported for the SR Ca 2+ leak promoted by Iso in rabbit cardiomyocytes, 6 -8 inhibition of PKA abolished the increase in diastolic Ca 2+ levels, the SR Ca 2+ leak, and SCWs, demonstrating that PKA is critical for the pro-arrhythmic effects of the PDE4 inhibitor. This discrepancy could be due either to species differences or to the concentration of H-89 used to block PKA. Indeed, H-89 was used at 1 mM concentration in previous studies, 6 -8 which was clearly insufficient in our hands to fully block PKA activity in intact cells (Supplementary material online, Figure S2). Interestingly, CaMKII  inhibition diminished the number of cells exhibiting SCWs and drastically blunted the SR Ca 2+ leak. Furthermore, the increase in fractional release induced by Iso and concomitant PDE4 inhibition was abolished by H-89 and decreased by KN-93, suggesting that both kinases participate in the Ca 2+ sensitization of RyR2. Whether PKA phosphorylation of RyR2 participates in this process remains controversial 38,42 and cannot be ruled out from our experiments. Nonetheless, our results corroborate that both kinases must be activated to increase RyR2 sensitivity, as recently proposed. 43 Furthermore, because PDE3 or PDE4 inhibition increases SR Ca 2+ load via PKA phosphorylation of PLB, this could by itself explain the RyR2 sensitization to Ca 2+ , given that luminal Ca 2+ determines RyR2 open probability. 44 In addition, promoted SERCA activity will enable sufficient SR Ca 2+ refilling to counterbalance the SR Ca 2+ leak and to maintain luminal Ca 2+ levels close to the threshold necessary to generate Ca 2+ waves, as demonstrated previously. 45 Thus, PKA is the primary mediator of SR Ca 2+ leak promoted by PDE4 inhibition, because it increases the SR Ca 2+ load to reach the necessary threshold to trigger Ca 2+ waves, 45 most likely by increasing the L-type Ca 2+ current and SERCA2 activity via PLB phosphorylation. Indeed, PKA leads to CaMKII activation by the classical Ca 2+ /Calmodulin pathway, a phenomenon which will increase RyR2 phosphorylation by CaMKII which alongside PKA contributes to increase RyR2 sensitivity under b-AR stimulation. 43 CaMKII inhibition limits SR Ca 2+ leak upon b-AR stimulation and concomitant PDE4 inhibition without affecting the SR Ca 2+ content. This could be explained by increased Ca 2+ sensitivity of RyR2 due to the increased phosphorylation at Ser2814 46 as observed here, whereas PKA phosphorylation of PLB at Ser16 rather than by CaMKII at Thr17 would be sufficient to promote SERCA activity as demonstrated previously. 37 Despite the fact that the elevated cytosolic Ca 2+ due to PKA appears as the main contributor to CaMKII activation, the diminished occurrence of SCWs and SR Ca 2+ leak upon PDE4 inhibition when cells were pre-incubated with ESI-05 suggests also a role for Epac2. This result is in line with diminished Ca 2+ sparks observed upon b-AR stimulation in Epac2 knock-out mice 10 and suggests that PDE4 regulates Epac2 to limit CaMKII activation, as suggested recently. 28 However, the effect of Epac2 inhibition was rather modest compared with CaM-KII inhibition with KN-93 or AIP. This may be due to incomplete inhibition of the exchange factor with ESI-05. Alternatively, it may suggest that CaMKII activation via Epac2 is minor when compared with the classical PKA-mediated cytosolic Ca 2+ elevation. Of note, ESI-05 partially diminished the occurrence of SCWs in Iso + Ro, but not the few events observed upon non-maximal b-AR stimulation alone. This could be explained by the higher cAMP levels required for Epac activation compared with PKA activation. 47 Whether activation of Epac2 is under the control of PDE3 remains to be determined. However, CaMKII inhibition prevented the increased SR Ca 2+ leak induced by cilostamide, demonstrating that this kinase participates in the arrhythmias evoked by PDE3 inhibitors.
To conclude, our work provides new insights into the cellular mechanisms underlying the arrhythmias observed upon PDE3 and PDE4 inhibition. It demonstrates that PDE3 and PDE4 inhibitors exert inotropic effects via PKA by increasing SR Ca 2+ load, but shows for the first time that they promote diastolic pro-arrhythmic Ca 2+ waves linked to an SR Ca 2+ leak via both PKA and CaMKII. Interestingly, CaMKII inhibition prevented arrhythmias but did not alter the inotropic effects of the PDE inhibitors. Additional experiments are needed to evaluate whether the present results may be transposable to a pathological state, particularly in the context of HF in which the b-AR/cAMP pathway is altered. 4 However, in the context of acute and decompensated HF in which Epac 48 and CaMKII 5 are upregulated, PDE inhibitors have been shown to exert beneficial effects on haemodynamics. Furthermore, CaMKII is likely to contribute to the pro-arrhythmic effects of these drugs in HF, as its activity is increased by elevated cytosolic Ca 2+ levels and oxidative stress evoked by the pathological activation of b-AR and renin-angiotensin -aldosterone pathways. 49 Interestingly, CaMKII inhibition can also prevent apoptosis, 50 another adverse consequence of PDE inhibition. 20,51 Altogether, our results suggest the potential use of CaMKII inhibitors in adjunct to PDE inhibitors to counteract their side effects while preserving their cardiotonic properties.

Supplementary material
Supplementary Material is available at Cardiovascular Research online.