Impact of phosphodiesterases PDE3 and PDE4 on 5-hydroxytryptamine receptor4-mediated increase of cAMP in human atrial fibrillation

Atrial fibrillation (AF)–associated remodeling includes contractile dysfunction whose reasons are only partially resolved. Serotonin (5-HT) increases contractile force and causes arrhythmias in atrial trabeculae from patients in sinus rhythm (SR). In persistent atrial fibrillation (peAF), the force responses to 5-HT are blunted and arrhythmic effects are abolished. Since force but not arrhythmic responses to 5-HT in peAF could be restored by PDE3 + PDE4 inhibition, we sought to perform real-time measurements of cAMP to understand whether peAF alters PDE3 + PDE4-mediated compartmentation of 5-HT4 receptor-cAMP responses. Isolated human atrial myocytes from patients in SR, with paroxysmal AF (paAF) or peAF, were adenovirally transduced to express the FRET-based cAMP sensor Epac1-camps. Forty-eight hours later, cAMP responses to 5-HT (100 μM) were measured in the absence or concomitant presence of the PDE3 inhibitor cilostamide (0.3 μM) and the PDE4 inhibitor rolipram (1 μM). We successfully established real-time cAMP imaging in AF myocytes. 5-HT increased cAMP in SR, paAF, and peAF, but in line with previous findings on contractility, this increase was considerably smaller in peAF than in SR or paAF. The maximal cAMP response to forskolin (10 μM) was preserved in all groups. The diminished cAMP response to 5-HT in peAF was recovered by preincubation with cilostamide + rolipram. We uncovered a significantly diminished cAMP response to 5-HT4 receptor stimulation which may explain the blunted 5-HT inotropic responses observed in peAF. Since both cAMP and force responses but not arrhythmic responses were recovered after concomitant inhibition of PDE3 + PDE4, they might be regulated in different subcellular microdomains. Electronic supplementary material The online version of this article (10.1007/s00210-020-01968-1) contains supplementary material, which is available to authorized users.


Introduction
Atrial fibrillation (AF) is the most frequent arrhythmia but its treatment remains a challenge. Conventional antiarrhythmic drugs have limited efficacy and increase the risk of ventricular arrhythmia (Nattel et al. 1994;Dobrev and Nattel 2010). Modest progress with anti-AF drugs is mostly due to the fact that compounds were produced in the absence of specific molecular targets. Refinement of conventional ion channel blocker therapy as an atrial-selective approach has not yet led to an improvement. Therefore, there is an urgent need to better understand the molecular pathophysiology of AF.
One important mechanism in the regulation of excitationcontraction coupling in cardiomyocytes is the phosphorylation of key proteins by activation of the cAMP and the cAMP-dependent protein kinase (PKA) signaling cascade. Phosphodiesterases (PDEs) are hydrolytic enzymes which degrade cAMP and limit the phosphorylation of PKA targets in specific myocyte compartments. PDEs comprise a large group of isoenzymes that are divided into 11 PDE families (Conti and Beavo 2007). Of these, PDE4 contributes most of the cAMP-hydrolytic activity in rodent ventricle (Rochais et al. 2006), while in human atria, cAMP is mainly hydrolyzed by PDE3 (Galindo-Tovar et al. 2009;Molina et al. 2012). Inhibition of either PDE3 or PDE4 leads to increase of the propensity of 5-HT-evoked arrhythmias in human atrial trabeculae from patients in sinus rhythm (SR) and with paroxysmal AF (paAF) but not with persistent AF (peAF) (Berk et al. 2016). Furthermore, it has been shown that the activity of PDE4 is reduced in peAF (Molina et al. 2012). FRET experiments in human atrial myocytes (HAMs) from patients in SR have identified PDE3 and to a smaller extent PDE4 to control both basal cAMP levels and cAMP responses under βadrenergic stimulation (Molina et al. 2012).
5-HT increases force of contraction (Kaumann et al. 1990) and causes arrhythmias through 5-HT 4 receptors in human atrial trabeculae from patients in SR (Kaumann 1994;Sanders and Kaumann 1994;Sanders et al. 1995;Pau et al. 2007). However, in peAF patients, positive inotropic effects of 5-HT were markedly reduced and 5-HT-induced arrhythmias were abolished (Christ et al. 2014). In SR and paAF, combined inhibition of PDE3 and PDE4 increased force responses of 5-HT as well as arrhythmic responses (Berk et al. 2016). In peAF, this treatment reversed the blunted force responses but did not restore arrhythmic responses to 5-HT (Berk et al. 2016). While in peAF force responses to 5-HT are mainly controlled by PDE3 (Berk et al. 2016), the abolished 5 HT-induced arrhythmias are not affected by PDE3 or PDE4, suggesting that force and arrhythmic responses may be regulated by distinct subcellular compartments.
Since none of these effects has been correlated with actual intracellular cAMP levels, we sought to establish an imaging approach for real-time monitoring of this crucial second messenger in primary HAMs from AF patients. To understand whether the blunted force responses or abolished arrhythmic responses to 5-HT in peAF result from a decrease of cAMP triggered by the 5-HT4 receptor stimulation, increased activity of PDEs or both, we established live cell imaging of cAMP and measured the effects of 5-HT and PDE inhibition on cAMP levels in HAMs from patients with SR, paAF, and peAF. Here, we demonstrate that cAMP responses to 5-HT are strongly reduced in peAF but they can be restored by concomitant PDE3 and PDE4 inhibition. Since similar effects have previously been shown for positive inotropic but not for arrhythmia responses to 5-HT, we suggest that these two effects might be regulated in different subcellular microdomains.
Drugs 5-Hydroxytryptamine (5-HT) hydrochloride, forskolin (FSK), rolipram (Rol), and cilostamide (Cil) were purchased from Sigma. Rolipram and cilostamide were solubilized in dimethyl sulfoxide and in K + -Ringer solution. The final concentration of dimethyl sulfoxide was less than 0.1%, which by itself did not affect cAMP levels. All the drugs were dissolved in a K + -Ringer solution previously described reaching the final concentration of 100 μM 5-HT, 1 μM rolipram, and 300 nM cilostamide.
Statistics Values are expressed as mean ± SEM. D'Agostino and Pearson omnibus normality test was used to test for normal distribution. If normally distributed, statistical significance was evaluated using a paired Student's t test or oneway ANOVA followed Tukey's multiple comparisons post hoc test when more than two groups were compared. In case of skewed distribution, data were analyzed by non-parametric Mann-Whitney test when two groups or by non-parametric Kruskal-Wallis test when more than two groups were compared. Differences were considered statistically significant when p < 0.05. All statistical tests were done with GraphPad Prism 6.0 (GraphPad Software, San Diego, USA).

Real-time monitoring of cAMP in HAMs
To perform real-time cAMP measurements in HAMs, we cultured isolated myocytes for 48 h and transduced them with an adenovirus to express the FRET-based cAMP biosensor Epac1-camps (Fig. 1a) (Nikolaev and Lohse 2006). Although we previously used this protocol for SR cells and a similar biosensor for cytosolic cAMP called Epac2-camps (Molina et al. 2012), it has not been applied to AF cells so far. Even after 2 days in culture, HAMs isolated from AF patients and expressing the biosensor still showed a normal morphology and membrane structure (Fig. 1b). They also responded to 5-HT and FSK application with a change of FRET (Fig. 1c, d).
Basal FRET ratios were similar in all studied groups, although paAF and specially peAF myocytes showed a tendency towards increased basal FRET signal which might indicate a slightly higher basal cAMP levels in AF (Supplemental Fig. 1a). Importantly, no obvious differences in the subcellular sensor localization could be observed between SR and AF cells by confocal microscopy (Supplemental Fig. 1b).

Reduced 5-HT triggered cAMP levels in peAF
We first investigated whether the reduced inotropic effect of 5-HT in peAF could be due to a reduced increase in cAMP. Myocytes were stimulated with saturating concentrations of 5-HT, and 10 μM FSK was added in the presence of 5-HT at the end of each experiment to measure the maximal cAMPgenerating capacity (as in Fig. 1). Typical time courses of the FRET signals are shown in Figs. 2 a-c. 5-HT (100 μM) increased cytosolic cAMP levels and FSK (10 μM) increased cAMP even further in SR, paAF, and peAF. However, cAMP responses to 5-HT alone were 63% smaller in peAF than in SR (Fig. 2d), suggesting either a diminished ability of the 5-HT 4 receptor to activate adenylyl cyclase or diminished ability of adenylyl cyclase to produce cAMP. However, in paAF cells, 5-HT/cAMP responses were comparable with those of SR ( Fig. 2a-d). The FSK effects in the presence of 5-HT were not different between the three groups, demonstrating intact ability of adenylyl cyclase to produce cAMP in myocytes from patients with peAF.

PDE3 and PDE4 inhibition restores the 5-HT-triggered increase of cAMP in peAF
Increased PDE activity could result in reduced cAMP responses upon 5-HT 4 receptor activation. To test this possibility, we measured cAMP levels under concomitant cilostamide (Cil, 0.3 μM) and rolipram (Rol, 1 μM) treatment of HAM from the three groups of patients. Typical time courses are shown in Fig. 3 a-c. Interestingly, the relative increase of cytosolic cAMP caused by the two PDE inhibitors was not greater in peAF than in SR or paAF (Fig. 3d, p = 0.34, oneway ANOVA), refuting the hypothesis that a higher PDE activity per se might reduce the cAMP increase after 5-HT 4 receptor stimulation in peAF. Next, we calculated the effect of concomitant PDE3 and PDE4 inhibition on 5-HT-evoked cAMP accumulation measured in Fig. 3 a-c. The magnitude of the 5-HT-induced cAMP levels after Cil + Rol preincubation was no longer smaller in peAF than in SR and paAF (Figs. 3 and 4), and the further increase by FSK (10 μM) was not statistically significant between the groups (Fig. 3).

Discussion
Classical antiarrhythmic drugs were developed without specific molecular targets and there is an urgent need to better understand AF molecular pathophysiology. We performed this study to establish live cell imaging of cAMP in AF myocytes and used this approach to understand the molecular mechanisms of 5-HT/cAMP response and its regulation by PDEs in paAF and peAF. 5-HT caused a smaller increase of cAMP in HAMs from peAF patients than from SR patients and paAF patients. This is in agreement with the depressed force response to 5-HT in trabeculae from peAF patients (Christ et al. 2014). Inhibition of cAMP hydrolysis with the concomitant administration of the PDE3-selective inhibitor cilostamide (0.3 μM) and the PDE4-selective inhibitor rolipram (1 μM) restored cAMP levels in peAF patients. This agrees with the recovery of the force response after treatment with the two PDE inhibitors (Berk et al. 2016). In contrast, concomitant application of cilostamide and rolipram failed to restore the arrhythmic responses to 5-HT in peAF. Of interest, either cilostamide or rolipram caused only marginal increases of 5-HT-induced arrhythmias in trabeculae of SR patients. However, cilostamide and rolipram administered together enhanced four-fold the 5-HT-induced arrhythmias in trabeculae from SR patients but not at all in trabeculae from peAF patients (Berk et al. 2016), despite the robust increases in cAMP in myocytes from both SR and peAF (Figs. 3 and 4). It could be proposed that the subcellular microdomain through which PDEs enhance 5-HTinduced arrhythmias is disrupted in peAF.
The reduced cAMP responses to 5-HT in peAF are associated with preserved I Ca,L responses but blunted force (Berk et al. 2016). In HAM from patients with peAF, cAMP responses by 5-HT were reduced by 63% while increases by FSK were preserved. However, it should be noted that despite the preserved cAMP responses to FSK in peAF (Fig. 3), the increase of I Ca,L with FSK in peAF was not larger than with 5-HT (Christ et al. 2014). This finding indicates that even if the cytosolic cAMP response upon 5-HT is reduced in peAF, it is possible to activate I Ca,L maximally compared with FSK. In contrast to virtually maximum I Ca,L increases observed with 5-HT and FSK, the 5-HT-induced FRET responses were reduced by 63%, and force responses to 5-HT were reduced by 85% in peAF (Berk et al. 2016;Christ et al. 2014). We suggest that the preserved effects of 5-HT on I Ca,L in peAF are due to restricted pool of cAMP within the L-type Ca 2+ channel Fig. 2 cAMP responses to 5-HT but not to FSK are reduced in peAF. Time courses of representative FRET signals indicating changes in cytosolic cAMP in 3 HAMs from patients with SR (a), paAF (b), and peAF (c) exposed to 100 μM 5-HT and in the continuous presence of 5-HT to 10 μM forskolin (FSK). d Summary of the results from panels a-c. Depicted are FRET responses to both 5-HT and forskolin in the presence of 5-HT (+FSK) in individual myocytes. Mean values ± SEM are indicated by the circles. n = number of myocytes/number of patients. *p < 0.05 vs. 5-HT in SR (one-way ANOVA; Tukey's multiple comparisons test, based on myocytes); # p < 0.05 FSK vs. 5-HT (paired t test). Values for FSK were not significantly different between groups (one-way ANOVA; Tukey's multiple comparisons test) microdomain. However, cytosolic cAMP levels strongly affect various intracellular targets phosphorylated by PKA in distinct subcellular microdomains to control contractile force and relaxation upon 5-HT stimulation. As an alternative to the subcellular compartmentation, the putative mechanism mediating 5-HT-induced arrhythmias, which are significantly reduced in peAF, could be independent of cAMP, PDE3, and PDE4. Nevertheless, arrhythmias in human atrial trabeculae evoked by 5-HT can be abolished by 5-HT receptor antagonists (Kaumann 1994). 5-HT induces arrhythmias in mice only when the 5-HT receptor is overexpressed (Gergs et al. 2010). Both findings argue against receptor-independent arrhythmia induction by 5-HT. More importantly, application of an inhibitor of PKA stops 5-HT-induced arrhythmias in atria from mice transgenic for the 5-HT receptor (Gergs et al. 2013).
In the future, it will be important to study cAMP response in distinct subcellular microdomains using targeted cAMP biosensors. For example, microdomains which are important for Ca 2+ handling and contractility (i.e., L-type Ca 2+ channel, SERCA2A, ryanodine receptors, or troponin I-associated microdomains) can be analyzed using recently described targeted biosensors (Surdo et al. 2017) to dissect the molecular mechanisms of the compartmentalized 5-HT/cAMP responses.

Conclusions
We conclude that there are profound differences in the cAMP regulation by PDEs in peAF compared with SR or paAF. Effects of PDE inhibition on 5-HT-evoked cAMP and force response are increased in peAF while 5-HT-mediated arrhythmogenic effects are not restored by PDE inhibition. Taking together, our results suggest that PDE3 and PDE4 control cytosolic 5-HT/cAMP-stimulated inotropic responses, whereas arrhythmic responses may be regulated by another pool of cAMP which is compartmentalized in a distinct subcellular microdomain.

Limitations
The conclusion that 5-HT triggers less cytosolic cAMP in peAF is only valid as long as the affinity of 5-HT to its receptor(s) is comparable between different groups and 100 μM of 5-HT would lead to full occupancy of 5-HT 4 receptors. Although there are no radioligand binding data available for 5-HT receptors in peAF, from concentration-response dependencies for 5-HT on force, it can be concluded that 100 μM 5-HT used in our study is indeed a saturating agonist concentration for both in SR and peAF (Berk et al. 2016). Regarding the extent of PDE inhibition, similar to our former work (Berk et al. 2016), we used 0.3 μM cilostamide to inhibit PDE3 and 1 μM rolipram to inhibit PDE4. These rather low concentrations were chosen to avoid cross-reactivity by the inhibitors. It should be noted that the actual extent of inhibition of PDE3 and PDE4 is expected to be different with less inhibition of PDE4 by 1 μM rolipram than inhibition of PDE3 by 0.3 μM cilostamide. As a result, effects of concomitant inhibition of PDE3 and PDE4 demonstrated here may depend more on PDE3 than on PDE4. We are aware of that limitation. However, these concentrations were chosen on purpose to directly compare cAMP data with our previously published results on force and arrhythmias.
Authors' contribution AJK, TC, CM, and VON conceived the research. CM, NGP, and BD performed experiments. VN provided setups and reagents/samples. HR provided samples and YY analyzed patient data. CM, BD, and TC analyzed the data. TC, TE, and VON handled funding and project management. BD, CM, TC, and AJK drafted and revised the manuscript. All authors edited and approved the final version of the manuscript. The authors declare that all data were generated in-house and that no paper mill was used.
Funding Open Access funding provided by Projekt DEAL. The work was supported by the European Union's Horizon 2020 research and innovation program under the Marie Sk1odowska-Curie grant agreement no. 675351 and the German Centre for Cardiovascular Research (DZHK).

Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict of interest.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.