Cyclic GMP regulation of the L-type Ca²⁺ channel current in human atrial myocytes

Abstract

1. The regulation of the L-type Ca²⁺ current (I_Ca) by intracellular cGMP was investigated in human atrial myocytes using the whole-cell patch-clamp technique.

2. Intracellular application of 0.5 μm cGMP produced a strong stimulation of basal I_Ca (+64 ± 5%, n = 60), whereas a 10-fold higher cGMP concentration induced a 2-fold smaller increase (+36 ± 8%, n = 35).

3. The biphasic response of I_Ca to cGMP was not mimicked by the cGMP-dependent protein kinase (PKG) activator 8-bromoguanosine 3′,5′ cyclic monophosphate (8-bromo-cGMP, 0.5 or 5 μm), and was not affected by the PKG inhibitor KT 5823 (100 nm).

4. In contrast, cGMP stimulation of I_Ca was abolished by intracellular perfusion with PKI (10 μm), a selective inhibitor of the cAMP-dependent protein kinase (PKA).

5. Selective inhibition of the cGMP-inhibited phosphodiesterase (PDE3) by extracellular cilostamide (100 nm) strongly enhanced basal I_Ca in control conditions (+78 ± 13%, n = 7) but had only a marginal effect in the presence of intracellular cGMP (+22 ± 7% in addition to 0.5 μm cGMP, n = 11; +20 ± 22% in addition to 5 μm cGMP, n = 7).

6. Application of erythro-9-[2-hydroxy-3-nonyl]adenine (EHNA, 30 μm), a selective inhibitor of the cGMP-stimulated phosphodiesterase (PDE2), fully reversed the secondary inhibitory effect of 5 μm cGMP on I_Ca (+99 ± 16% stimulation, n = 7).

7. Altogether, these data indicate that intracellular cGMP regulates basal I_Ca in human atrial myocytes in a similar manner to NO donors. The effect of cGMP involves modulation of the cAMP level and PKA activity via opposite actions of the nucleotide on PDE2 and PDE3.

The cardiac L-type Ca²⁺ channel current (I_Ca) is an important determinant of myocardial contractility. Its regulation by neurotransmitters, hormones, and paracrine factors contributes to the control of cardiac output to meet the demands of the body. A large number of these extracellular first messengers, acting on specific membrane receptors in cardiac myocytes, regulate the activity of adenylyl cyclase which in turn controls the intracellular concentration of cAMP, the activity of the cAMP-dependent protein kinase (PKA), and the degree of phosphorylation and stimulation of L-type Ca²⁺ channels (Hartzell, 1988; McDonald et al. 1994; Hove-Madsen et al. 1996; Striessnig, 1999). A typical example of such regulation is the control of heart function by the sympathetic and parasympathetic nervous systems, which act via adrenoceptors and muscarinic receptors (Brodde & Michel, 1999). In addition to the cAMP cascade, other factors regulate heart function by acting primarily on the cGMP cascade; these include atrial and brain natriuretic peptides (de Bold et al. 1996) and nitric oxide (NO) (Paulus & Shah, 1999; Shah & MacCarthy, 2000).

NO modulates cardiac contractility and rhythm in part via its ability to control the amplitude of I_Ca (for reviews see Fischmeister & Méry, 1996; Kelly et al. 1996; Kojda & Kottenberg, 1999; Paulus & Shah, 1999; Shah & MacCarthy, 2000). Classically, this regulation is mediated through the generation of cGMP by NO-stimulated soluble guanylyl cyclase activity. But, NO can also regulate cardiac contraction (Chesnais et al. 1999; Sandirasegarane & Diamond, 1999) or Ca²⁺ channel activity (Campbell et al. 1996; Hu et al. 1997) through cGMP-independent effects, so that the relative contribution of cGMP-dependent or -independent mechanisms to the overall effects of NO in the heart remains unresolved (Shah & MacCarthy, 2000).

Several studies in various animal species (reviewed in Lohmann et al. 1991; Fischmeister & Méry, 1996) have shown that exogenous cGMP can both stimulate or inhibit I_Ca (Hartzell & Fischmeister, 1986; Levi et al. 1989; Ono & Trautwein, 1991; Méry et al. 1991; Shirayama & Pappano, 1996; Han et al. 1998) and contractility (Nawrath, 1976; Trautwein & Trube, 1976; Endoh & Yamashita, 1981; Smith et al. 1991; Brady et al. 1993; Mohan et al. 1995; Kojda et al. 1996). These opposite effects can be explained by the presence of three different targets for cGMP with different affinities for the nucleotide (Lohmann et al. 1991; Butt et al. 1992): (1) the cGMP-inhibited phosphodiesterase (PDE3); (2) the cGMP-stimulated phosphodiesterase (PDE2); (3) the cGMP-activated protein kinase (PKG). The stimulatory effects on I_Ca or contractility observed during modest activation of the NO-cGMP pathway are best explained by cAMP elevation following PDE3 inhibition (Ono & Trautwein, 1993; Méry et al. 1993; Wahler & Dollinger, 1995; Kojda et al. 1996). But, the inhibitory effects of a strong activation of this pathway can be attributed either to PDE2 stimulation (in frog, Hartzell & Fischmeister, 1986; Méry et al. 1995) or to PKG activation (in embryonic chick heart, Wahler et al. 1990; Haddad et al. 1995; in adult mammalian heart, Levi et al. 1989; Méry et al. 1991; Wahler & Dollinger, 1995; Sumii et al. 1995; Kojda et al. 1996). Surprisingly, PKG was also reported to stimulate I_Ca in ventricular myocytes from newborn (Kumar et al. 1997) and young rabbit (Han et al. 1998). Altogether, these data indicate that the relative contribution of the different cGMP targets, as well as their final downstream modulation of I_Ca and heart function, may vary depending on the species, the developmental stage, and the region of the heart.

The variability in the results obtained in laboratory animals makes it difficult to extrapolate to humans and compelled us to directly assess the effects of NO and cGMP in human heart. Patch-clamp experiments performed in isolated human atrial myocytes demonstrated that NO donors and cGMP also regulate I_Ca in this preparation. At nanomolar concentrations, the NO donors SIN-1 and SNAP produced a stimulation of basal I_Ca (Kirstein et al. 1995; Vandecasteele et al. 1998a). This effect was blocked by intracellular methylene blue (Vandecasteele et al. 1998a), mimicked by PDE3 selective inhibitors (Kirstein et al. 1995; Kajimoto et al. 1997) or by an intracellular perfusion with cGMP (Rivet-Bastide et al. 1997). Although the molecular mechanisms involved have not been fully elucidated yet, these experiments suggested that in human atrial myocytes low concentrations of NO stimulated I_Ca via cGMP production and cGMP-inhibition of PDE3 (Kirstein et al. 1995). Surprisingly, when used at micromolar concentrations, the stimulatory effect of SIN-1 on I_Ca was strongly attenuated, suggesting the development of a secondary inhibitory effect at higher concentrations (Kirstein et al. 1995). Whether this secondary effect is also mediated by cGMP or results from a direct effect of NO or some of its by-products (e.g. resulting from the chemical reactions between NO, superoxide, and peroxynitrite) on L-type Ca²⁺ channels remains unknown.

In the present study, our aim was to dissect the mechanisms involved in the regulation of I_Ca by intracellular cGMP in human atrial myocytes. More specifically, we tried to address two questions: (1) to what extent can changes in intracellular cGMP mimic the bimodal regulation of I_Ca by NO?; (2) what are the respective contributions of the cGMP targets (PDE2, PDE3 and PKG) in the effect of exogenous cGMP on I_Ca?

A preliminary report of some of these results has appeared elsewhere (Vandecasteele et al. 1998b).

METHODS

Surgery

All protocols for obtaining human cardiac tissue were approved by the ethics committee of our institution (GREBB, Hôpital de Bicêtre, Université de Paris-Sud). Specimens of right atrial appendages were obtained from 33 patients (aged 10-83 years) undergoing heart surgery for congenital defects (n = 2), coronary artery diseases (n = 25) or valve replacement (n = 6). All patients but five received a pharmacological pre-treatment (Ca²⁺ channel blockers, digitalis, β-adrenergic antagonists, diuretics, ACE inhibitors, NO donors and/or anti-arrhythmic drugs). In addition, all patients received sedatives, anaesthesia and antibiotics prior to surgery. But, we found no obvious correlation between the Ca²⁺ current density or the effects on I_Ca of the drugs tested here and the long-term therapy received (if any) by the patient. Dissociation of the cells was realised immediately after surgery.

Human atrial cell dissociation

Myocytes were isolated as described previously (Kirstein et al. 1995) with some modifications. Briefly, quickly after excision, the tissue was washed and cut in small pieces in a Ca²⁺-free Tyrode solution supplemented with 30 mm 2,3-butanedione monoxime (BDM). Small (≈1 mm³) pieces of atria were then incubated in a BDM- and Ca²⁺-free Tyrode solution containing 40 i.u. ml⁻¹ collagenase, 15 i.u. ml⁻¹ protease and 5 mg ml⁻¹ BSA. After 30 min, this solution was removed and replaced by fresh enzymatic solution containing only collagenase (200 i.u. ml⁻¹) for 10-20 min until a satisfactory cell yield was obtained. All steps were carried out at 37 °C, with continuous stirring at 200 r.p.m. and gassing with 95 % O₂-5 % CO₂. The cell suspension was filtered, centrifuged (for 1 min at 600-700 r.p.m.) and the pellet resuspended in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10 % fetal calf serum, non-essential amino acids, 1 nm insulin and antibiotics (penicillin, 100 i.u. ml⁻¹ and streptomycin, 0.1 μg ml⁻¹). For patch-clamp experiments, 20-100 μl of this cell suspension was added to a control extracellular solution in a Petri dish.

Electrophysiological experiments

The whole-cell configuration of the patch-clamp technique (Hamill et al. 1981) was used to record the high-threshold L-type Ca²⁺ current (I_Ca) on Ca²⁺-tolerant human atrial myocytes. In the routine protocols the cells were depolarised every 8 s from a holding potential of -50 mV to 0 mV for 400 ms. This holding potential was chosen to completely inactivate the fast Na⁺ current. K⁺ currents were blocked by replacing all K⁺ ions with intracellular and extracellular Cs⁺. Voltage-clamp protocols were generated by a challenger/09-VM programmable function generator (Kinetic Software, Atlanta, GA, USA). The cells were voltage clamped using a patch-clamp amplifier (model RK-400; Biologic, Claix, France). Currents were sampled at a frequency of 10 kHz using a 12-bit analog-digital converter (DT2827; Data Translation, Marlboro, MA, USA) connected to a PC-compatible computer (386/33 System-pro; Compaq, Houston, TX, USA). All experiments were done at room temperature (19-25 °C) and the temperature varied by < 2 °C during the course of an experiment.

Solutions

Control extracellular solution contained (mm): NaCl 107.1, Hepes 10, CsCl 40, NaHCO₃ 4, NaH₂PO₄ 0.8, CaCl₂ 1.8, MgCl₂ 1.8, d-glucose 5 and sodium pyruvate 5; pH 7.4 adjusted with NaOH. Patch electrodes (0.8-1.5 MΩ) were filled with control GTP-free intracellular solution that contained (mm): CsCl 119.8, EGTA (acid form) 5, MgCl₂ 4, creatine phosphate disodium salt 5, Na₂-ATP 3.1, Hepes 10, and CaCl₂ 62 μm (pCa 8.5); pH 7.3 adjusted with CsOH. In some experiments Na₂-GTP (420 μm) was added to the GTP-free intracellular solution, and the pH was readjusted. Control or drug-containing solutions were applied to the exterior of the cell by placing the cell at the opening of 250 μm inner diameter capillary tubing flowing at a rate of ≈10 μl min⁻¹. Intracellular perfusion of the cell with cyclic nucleotides or the PKA inhibitor PKI during whole-cell recording was made possible by the use of a microcapillary inside the patch-clamp pipette, as already described (Hartzell & Fischmeister, 1986). This capillary was connected to little tanks containing intracellular solutions supplemented with cyclic nucleotides or PKI at different concentrations. Application of a modest negative pressure inside the patch electrode allowed flowing of the desired solution to the tip of the pipette and inside the cell by passive diffusion.

Materials

Collagenase type V and protease type XXIV, used for dissociation of human atrial cells, and Na₂-cGMP, Na₂-8-bromo-cGMP, erythro-9-[2-hydroxy-3-nonyl]adenine (EHNA), cAMP-dependent protein kinase inhibitor (PKI, rabbit sequence) used in patch-clamp experiments were from Sigma-Aldrich (L'Isle d'Abeau Chesnes, France). Cilostamide was from Tocris Cookson (Bristol, UK) and KT 5823 was from Calbiochem-France Biochem (Meudon, France). Cilostamide was dissolved at 10 mm in ethanol. KT 5823 was dissolved at 100 mm in DMSO. An equal amount of ethanol and/or DMSO corresponding to the concentration present in the final dilutions was added to all other solutions. All other drugs were dissolved in ionic aqueous solutions, made fresh daily and kept at 4 °C until use.

Data analysis

The maximal amplitude of I_Ca was measured as the difference between the peak inward current and the leak current (I₄₀₀), which was the current amplitude at the end of the 400 ms duration pulse (Kirstein et al. 1995). Currents were not compensated for capacitive and leak currents. Cell membrane capacitance and series resistances were measured by exponential analysis of current responses to 1 mV step changes in membrane potential. Membrane capacitance was 74.2 ± 21 pF (mean ±s.e.m.) and series resistance was 3.2 ± 0.2 MΩ (n = 82). On-line analysis was performed by programming a PC-compatible computer in PASCAL to determine peak and steady-state current values for each depolarisation.

The results are expressed as means ±s.e.m. In each experimental condition, the effects of the drugs tested on I_Ca are expressed as percentage change with respect to the values of the current under basal conditions, that is, in the absence of any hormonal stimulation. The variations in I_Ca induced by the different drugs were tested for statistical significance by Student's t test. Statistically significant differences between different conditions are indicated in the figures as: * P < 0.05; ** P < 0.01; *** P < 0.005.

RESULTS

Biphasic effect of cGMP on I_Ca

I_Ca was recorded in human atrial myocytes using the whole-cell configuration of the patch-clamp technique (Hamill et al. 1981). Basal I_Ca amplitude was measured 3-5 min after patch break to allow for equilibration between intracellular and pipette solutions. Basal I_Ca amplitude at a membrane potential of 0 mV was 241.0 ± 16.7 pA and I_Ca density, which represents the ratio of I_Ca amplitude to membrane capacitance, was 3.2 ± 0.2 pA pF⁻¹ (n = 82). As in our previous studies (Kirstein et al. 1995; Rivet-Bastide et al. 1997; Vandecasteele et al. 1998a), I_Ca densities showed a large scatter between different patients and between individual cells from the same patient, with no obvious correlation with the diagnosis, sex, age or pretreatment of the patients. Figure 1A illustrates a typical experiment showing the effect of intracellular perfusion with cGMP on the time course of I_Ca amplitude measured at 0 mV from a holding potential of -50 mV. Two concentrations of cGMP (0.5 and 5 μm) were successively dialysed into the human atrial myocyte (see Methods). At the beginning of the experiment, the cell was dialysed with control intracellular (GTP-free) solution. After a stable baseline was achieved, the control solution was changed to a solution containing 0.5 μm cGMP (first arrow), which produced about a 2-fold increase in I_Ca amplitude. This effect was nearly abolished when the cGMP concentration was increased to 5 μm (second arrow). Washout of cGMP (third arrow) resulted in a rebound stimulation of I_Ca before it returned slowly to the control level. As summarised in Fig. 1B, on average 0.5 μm cGMP stimulated I_Ca by 64 ± 5 % above control level (n = 60, P < 0.001 vs. control) and subsequent application of 5 μm cGMP resulted in an ≈50 % attenuation of this effect (36 ± 8 % above basal level, n = 35, P < 0.005 vs. control and cGMP 0.5 μm). A lower concentration of cGMP (0.15 μm) was tested in six other cells, but I_Ca increased in only two of these cells, with an overall non-significant effect of the nucleotide (23 ± 20 %, data not shown). Application of a single high concentration of cGMP (5 or 50 μm) stimulated basal I_Ca by 76 ± 16 % (n = 5) and 53 ± 15 % (n = 5), respectively. But these effects were only transient and the current amplitude returned to the basal level after a few minutes.

Figure 1. Effect of intracellular application of cGMP on basal I_Ca in human atrial myocytes.

A, each symbol corresponds to a measure of I_Ca at 0 mV obtained every 8 s. The cell was first dialysed with control GTP-free intracellular solution and perfused with two concentrations of cGMP, first 0.5 μm and then 5 μm, at the time indicated by the arrows. After the applications of 0.5 and 5 μm cGMP the cell was dialysed with control intracellular solution (washout). The individual current traces shown in the upper part were obtained at the times indicated by the corresponding letters in the graph below. The horizontal line indicates the zero current level. B, mean stimulatory effect of cGMP (0.5 μm, □; 5 μm, ▪), obtained using the same experimental protocol as in A. The size of the bars indicates the mean effects expressed as percentage increase over basal I_Ca, and the lines the s.e.m., with the number of experiments indicated above. *** P < 0.005, statistically significant difference between both conditions using Student's t test.

As illustrated by the individual current traces shown in Fig. 1A, the stimulatory effect of cGMP was not accompanied by any significant modification in the kinetics of I_Ca. This suggests that cGMP did not modify the voltage dependence of the Ca²⁺ channel gating but to examine this further, the effect of intracellular cGMP on the I_Ca current-voltage (Fig. 2a) and inactivation (Fig. 2B) relationships was investigated. The U-shape of both curves for basal I_Ca (▪ in Fig. 2A and B), as well as their respective positions on the voltage axis, are characteristic of the high-threshold L-type Ca²⁺ current in this preparation (Kirstein et al. 1995). As shown, 0.5 μm cGMP increased I_Ca by a similar amount at every membrane potential (Fig. 2A) and did not modify the inactivation curve of the current (Fig. 2B). Thus, cGMP modifies I_Ca in an essentially voltage-independent manner.

Figure 2. Voltage dependence of the stimulatory effect of cGMP on basal I_Ca in human atrial myocytes.

Current-voltage relationships (A) and inactivation curves of I_Ca (B) in control condition (▪) and in the presence of 0.5 μm intracellular cGMP (A, ⋄; B, ○). Inactivation curves were obtained using the double-pulse protocol as indicated in the inset (see also Methods).

So far, our results indicate that cGMP activates two sequential mechanisms which affect I_Ca in an antagonistic manner; a stimulatory mechanism activated at concentrations below 0.5 μm and an inhibitory mechanism activated at concentrations above 5 μm. It should be noted that these cGMP concentrations do not necessarily reflect the actual concentrations of the cyclic nucleotide inside the cell. Indeed, access resistance to the cell and the presence of cyclic nucleotide phosphodiesterases are likely to lower drastically the intracellular cGMP concentration effectively used by the cell. The above experiments were performed in GTP-free intracellular solution, to limit the extent of endogenous cGMP synthesis by the myocytes (Rivet-Bastide et al. 1997) which might interfere with the effects of exogenous cGMP introduced through the patch pipette. Therefore, it was conceivable that the absence of intracellular GTP might lead to a progressive reduction of I_Ca due to a loss in a constitutive G protein activation of adenylyl cyclase activity (Skeberdis et al. 1997; Vandecasteele et al. 1998a), which could explain the biphasic effect of cGMP. To examine this, we tested the effect of cGMP in cells dialysed with GTP (420 μm). At this high concentration of GTP, application of 0.5 μm cGMP (n = 19), followed by an application of either 5 μm (n = 11) or 50 μm cGMP (n = 7), modified I_Ca by +36 ± 4 % (at 0.5 μm, P < 0.001 vs. control), +26 ± 9 % (at 5 μm, P < 0.05 vs. control) and -2 ± 6 % (at 50 μm, not significant). Thus, the biphasic effect of cGMP was still observed in the presence of GTP but addition of GTP reduced about 2-fold the stimulatory effect on I_Ca of the lowest cGMP concentration (0.5 μm; P < 0.005 vs. GTP free). Moreover, in the presence of intracellular GTP, it was necessary to increase the concentration of cGMP to higher levels (50 μm instead of 5 μm) to activate the secondary inhibitory mechanism. Thus, the presence of intracellular GTP, most probably via activation of endogenous cGMP synthesis (Rivet-Bastide et al. 1997), attenuated the stimulatory effect of cGMP on I_Ca and reduced the sensitivity of the inhibitory effect to the nucleotide. For a better dissection of the mechanisms involved in these two opposite effects, we returned to GTP-free conditions for all subsequent experiments.

Role of cGMP-dependent protein kinase (PKG)

Our first goal in this study was to determine the molecular mechanism by which cGMP stimulates I_Ca in human atrial myocytes. Since PKG was shown to be responsible for the stimulatory effect of cGMP on I_Ca in rabbit ventricle (Han et al. 1998), we examined whether 8-bromo-cGMP, a potent activator of this enzyme (Butt et al. 1992), could mimic the effect of cGMP. Figure 3A shows a typical experiment in which a human atrial myocyte was first dialysed with control (GTP-free) intracellular solution, and, after a few minutes (first arrow), was challenged with 5 μm 8-bromo-cGMP added to the patch pipette. As shown, 8-bromo-cGMP had no effect on basal I_Ca at this concentration. However, when used at 100 μm, 8-bromo-cGMP clearly enhanced I_Ca, an effect which amounted to ≈50 % of the maximal stimulation of the current obtained when the cell was dialysed with 50 μm cAMP (fourth arrow). Figure 3B summarises the results of several similar experiments in which three concentrations of 8-bromo-cGMP were tested (0.5, 5 and 100 μm). Whereas 0.5 and 5 μm of the cGMP derivative had no effect on basal I_Ca (+2 ± 5 %, n = 3, and -8 ± 11 %, n = 4, respectively), 100 μm increased the current to 87 ± 18 % above basal level (n = 4, P < 0.05).

Figure 3. Effect of intracellular 8-bromo-cGMP and cAMP on basal I_Ca in human atrial myocytes.

A, the cell was superfused with control extracellular solution throughout the experiment. Each symbol corresponds to a measure of I_Ca at 0 mV obtained every 8 s. At the beginning of the experiment, the cell was intracellularly dialysed with control GTP-free intracellular solution. The same solution containing 5 μm 8-bromo-cGMP was allowed to diffuse into the cytoplasm, without any notable effect on I_Ca (first arrow). In contrast, 100 μm of the cGMP derivative (second arrow) roughly doubled the current amplitude. After washout of 8-bromo-cGMP (third arrow), a maximal stimulation of I_Ca was obtained by perfusing the myocyte with 50 μm cAMP (last arrow). B, summary of the effects of 8-bromo-cGMP (0.5, 5 and 100 μm) on I_Ca in human atrial myocytes. The size of the bars indicates the mean effects expressed as percentage variation relative to basal I_Ca, and the lines the s.e.m., with the number of experiments indicated above. * P < 0.05, statistically significant difference from control I_Ca amplitude using Student's t test.

The above results indicate that 8-bromo-cGMP is 200-fold less potent than cGMP in stimulating I_Ca. Since the cGMP-derivative is more potent than the native nucleotide in activating PKG (Butt et al. 1992), these results argue against an involvement of this enzyme in the stimulatory effect of cGMP on I_Ca. However, to examine this hypothesis further, we tested the effect of KT 5823, a highly selective (Komalavila & Lincoln, 1996) and commonly used PKG inhibitor (Wahler & Dollinger, 1995; Kumar et al. 1997). KT 5823 was used at 100 nm, a concentration which significantly reduced the inhibitory effect of NO donors on I_Ca in guinea-pig (Wahler & Dollinger, 1995) and rat ventricular myocytes (Abi-Gerges et al. 2001) but which is unlikely to inhibit PKA (Kase et al. 1987). As shown in Fig. 4, extracellular application of KT 5823 on human atrial myocytes did not modify the stimulatory effect of an intracellular application of either 0.5 μm cGMP (Fig. 4A) or 100 μm 8-bromo-cGMP (Fig. 4B) on I_Ca. As summarised in Fig. 4C, 0.5 μm cGMP produced a 72 ± 15 % (n = 6) and 92 ± 24 % (n = 6) stimulation of basal I_Ca in the absence or presence, respectively, of KT 5823. Similarly, in three cells, intracellular perfusion with 100 μm 8-bromo-cGMP induced a 102 ± 21 % increase in basal I_Ca, and this effect remained unchanged after a subsequent application of 100 nm KT 5823 (92 ± 15 %, Fig. 4C). Altogether, these results exclude the possibility that PKG plays a determinant role in the stimulatory effect of cGMP on I_Ca.

Figure 4. Role of cGMP-dependent protein kinase (PKG) in the stimulatory effect of cGMP on I_Ca in human atrial myocytes.

A and B, two representative experiments showing the absence of effect of PKG inhibition by KT 5823 (100 nm) on I_Ca stimulated by cGMP (0.5 μm; A) and 8-bromo-cGMP (100 μm; B), in human atrial myocytes. Each symbol corresponds to a measure of I_Ca at 0 mV obtained every 8 s. In both experiments the cell was initially perfused with the usual GTP-free control intracellular solution, then dialysed with cGMP or 8-bromo-cGMP. As can be seen, both cyclic nucleotides clearly enhanced I_Ca. Superfusion of the cell with 100 nm KT 5823 in addition to cGMP or 8-bromo-cGMP failed to modify the amplitude of I_Ca. During the interruption of the curve in A, current-voltage (I–V) relationships were obtained, like those illustrated in Fig. 2A. Then the routine stimulation protocol was applied again. C, comparison of the mean effect of cGMP (0.5 μm, □) and 8-bromo-cGMP (100 μm, ▪) in the absence and presence of KT 5823 (100 nm). The size of the bars indicates the mean effects expressed as percentage increase over basal I_Ca, and the lines the s.e.m., with the number of experiments indicated above.

Role of cAMP-dependent protein kinase (PKA)

To examine the participation of PKA in the stimulatory effect of cGMP on basal I_Ca in human atrial myocytes, we performed experiments in which cGMP stimulation was followed by intracellular perfusion with PKI, a highly selective peptide inhibitor of PKA (Walsh et al. 1990). In four cells, intracellular perfusion with 0.5 μm cGMP increased I_Ca by 30 ± 4 % above the control value. After stabilisation of I_Ca amplitude, intracellular perfusion was switched to a solution containing 10 μm PKI added to the cGMP (0.5 μm)-containing solution. This quickly resulted in abolition of the cGMP stimulation of I_Ca and in a decrease in the calcium current amplitude below the initial baseline (54 ± 6 % below control value, see also Skeberdis et al. 1997). Thus, the stimulatory effect of cGMP on I_Ca is likely to be mediated by activation of PKA.

Role of cGMP-inhibited phosphodiesterase (PDE3)

One possible way by which cGMP could stimulate I_Ca in a PKA-dependent manner is through an increase in cAMP concentration due to inhibition of the cGMP-inhibited phosphodiesterase (PDE3). PDE3 was shown to regulate basal I_Ca in human atrial myocytes (Kirstein et al. 1995) and to be implicated in the stimulatory effect of cGMP on the isoprenaline-stimulated I_Ca in guinea-pig ventricular myocytes (Ono & Trautwein, 1991; Shirayama & Pappano, 1996). To examine a possible role of PDE3 in the effect of cGMP, we compared the effect of cGMP with that of cilostamide, a selective PDE3 inhibitor (Stoclet et al. 1995). As shown in Fig. 5, extracellular application of cilostamide (100 nm; Fig. 5A) and intracellular dialysis with cGMP (0.5 μm; Fig. 5B) produced comparable stimulatory effects on I_Ca. Moreover, when cilostamide was added as well as cGMP, it had only a marginal additional effect on I_Ca (Fig. 5B). The summary data of Fig. 5C allow for a comparison of the mean effects on basal I_Ca of cilostamide (100 nm), cGMP (0.5 μm) and of both compounds applied together. As shown, cilostamide and cGMP used alone produced very similar stimulatory effects on I_Ca, 78 ± 13 % (n = 7) and 64 ± 8 % (n = 11), respectively. When cilostamide was added to cGMP, the effect of the nucleotide was increased by 22 ± 7 % (n = 11, P < 0.05, paired t test). These results indicate that the stimulatory effect of cGMP was mimicked by a selective PDE3 inhibitor, and the effect of PDE3 inhibition was greatly reduced in the presence of cGMP. Therefore, partial PDE3 inhibition is likely to be responsible for the cGMP-induced stimulation of I_Ca in human atrial myocytes.

Figure 5. Effect of PDE3 inhibition by cilostamide on basal and cGMP-stimulated I_Ca in human atrial myocytes.

A and B, each symbol corresponds to a measure of I_Ca at 0 mV obtained every 8 s. Both experiments were carried out using a GTP-free intracellular solution. A, after a few minutes during which the cell was superfused with control extracellular solution, extracellular application of 100 nm cilostamide during the period indicated by the horizontal line produced an ≈60 % increase of I_Ca above control level. B, the same extracellular application of 100 nm cilostamide had only a modest effect when I_Ca had been previously increased by intracellular perfusion with 0.5 μm cGMP (first arrow). The second arrow indicates the washout of intracellular cGMP. C, summary of the effect of cilostamide (100 nm, □), cGMP (0.5 μm, ) and the combination of both (▪) on basal I_Ca. The size of the bars indicates the mean effects expressed as percentage increase over basal I_Ca, and the lines the s.e.m., with the number of experiments indicated above. * P < 0.05, statistically significant differences using Student's t test.

Role of cGMP-stimulated phosphodiesterase (PDE2)

As shown above, the stimulatory effect of cGMP on I_Ca is reduced when the concentration of cGMP is increased. This suggests the development of a secondary inhibitory mechanism at higher cGMP concentrations. As stated in the Introduction, two possible mechanisms could account for a cGMP-dependent inhibition of I_Ca: activation of PKG or activation of the cGMP-stimulated phosphodiesterase (PDE2). In order to differentiate between these two possibilities, we tested the effects of KT 5823, the PKG inhibitor, and EHNA, a selective PDE2 inhibitor (Méry et al. 1995; Rivet-Bastide et al. 1997), for their ability to reverse the effect of 5 μm cGMP on I_Ca. In the experiment shown in Fig. 6A, a human atrial myocyte was first dialysed with 0.5 μm cGMP, resulting in approximately 60 % stimulation of basal I_Ca. This effect was strongly reduced when the concentration of cGMP was increased to 5 μm. During the decrease in I_Ca, the cell was successively exposed to KT 5823 (100 nm), cilostamide (100 nm) and EHNA (30 μm). As shown, KT 5823 did not antagonise the inhibitory effect of cGMP, and cilostamide induced only a 15 % increase in I_Ca. By contrast, application of EHNA induced a strong and reversible stimulation of I_Ca which recovered to the amplitude obtained in the presence of 0.5 μm cGMP. The results of several similar experiments are summarised in Fig. 6B. In 14 cells, 0.5 μm cGMP increased I_Ca by 70 ± 10 % and a subsequent increase in cGMP concentration to 5 μm reduced this stimulation to 25 ± 11 % above control level (P < 0.01 vs. 0.5 μm cGMP). In seven individual cells, extracellular application of either KT 5823 (100 nm) or cilostamide (100 nm) in the continuous presence of intracellular cGMP (5 μm) had no significant effect on the current amplitude (+17 ± 14 %, P = 0.73, and +20 ± 22 %, P = 0.83, respectively). This indicates that activation of PKG does not account for the inhibitory effect of cGMP on I_Ca and that PDE3 is already fully inhibited at 5 μm cGMP. Exposure of the cells to EHNA (30 μm) induced a strong stimulation of I_Ca, to 99 ± 16 % above the control level (n = 7, P < 0.005 vs. 5 μm cGMP), an effect which was similar to that seen with 0.5 μm cGMP. These results demonstrate that the inhibitory effect of cGMP on I_Ca in human atrial myocytes involves an activation of PDE2.

Figure 6. Contribution of PKG, PDE3 and PDE2 to the inhibitory effect of cGMP on I_Ca in human atrial myocytes.

A, each symbol corresponds to a measure of I_Ca at 0 mV obtained every 8 s. The cell was first dialysed with intracellular GTP-free solution. Intracellular dialysis with 0.5 μm cGMP (first arrow) resulted in the usual stimulation of I_Ca, and switching to 5 μm cGMP (second arrow) partially reversed this stimulatory effect. Superfusion of the cell with KT 5823 (100 nm) or cilostamide (100 nm) had little or no effect on I_Ca, whereas EHNA (30 μm) restored the amplitude of the current observed in the presence of 0.5 μm cGMP. B, summary of several experiments similar to those shown in A. The effect of cGMP (0.5 μm, □; 5 μm, ▪) on I_Ca is shown in the absence or presence of KT 5823 (100 nm), cilostamide (100 nm) or EHNA (30 μm). The size of the bars indicates the mean effects expressed as percentage increase over basal I_Ca, and the lines the s.e.m., with the number of experiments indicated above. ** P < 0.01, *** P < 0.005, statistically significant differences using Student's t test.

DISCUSSION

In the present study, we examined the effects of cGMP on the L-type Ca²⁺ current (I_Ca) in human atrial myocytes. Several main conclusions can be drawn from our experiments: (1) cGMP activates two sequential mechanisms which affect I_Ca in an antagonistic manner, a stimulatory mechanism activated at concentrations below 0.5 μm and an inhibitory mechanism activated at concentrations above 5 μm; (2) the regulation of I_Ca by cGMP is not accompanied by any modification in the voltage dependence of the Ca²⁺ current; (3) PKG does not seem to play a major role in either of these two opposite mechanisms; (4) the stimulatory effect of cGMP is due to activation of PKA resulting from a cGMP-dependent inhibition of PDE3; (5) the inhibitory effect is due to a reduction in PKA via cGMP-dependent stimulation of PDE2. We conclude that cGMP regulates I_Ca in human atrial myocytes by controlling the intracellular concentration of cAMP through opposing actions on PDE3 and PDE2.

This study follows up and confirms earlier studies from our laboratory on the regulation of basal I_Ca by NO donors (Kirstein et al. 1995; Vandecasteele et al. 1998a) and by PDE2 in human atrial myocytes (Rivet-Bastide et al. 1997). In particular, we found that SIN-1 stimulates I_Ca in the nanomolar concentration range, an effect which is reduced when the concentration of the NO donor is increased in the micromolar range (Kirstein et al. 1995). The present experiments reveal that intracellular perfusion with cGMP, the second messenger of NO, produces very similar effects on I_Ca in human atrial myocytes.

Over the last 15 years, numerous studies have reported opposite and contradictory effects of intracellular cGMP on I_Ca in different cardiac preparations (for review, see Lohmann et al. 1991; Fischmeister & Méry, 1996). Although it is still difficult to draw a clear picture of the effects of the nucleotide on heart function, all these studies have contributed to our understanding that the cGMP signalling pathways are intimately linked to those of cAMP and involve three main enzymes, namely PDE2, PDE3 and PKG (Hove-Madsen et al. 1996). Thus, it is now accepted that cGMP will produce different effects depending on: (1) the presence and relative activities of these three enzymes; (2) their respective location inside the cell; (3) their respective affinities for cGMP; (4) whether adenylyl and/or guanylyl cyclases are constitutively active in the cells under study; (5) the concentration of cGMP used; (6) whether native cGMP or a cGMP analogue is used; (7) whether the effect of cGMP on I_Ca is examined under basal conditions or after the current has been enhanced by activation of the cAMP cascade. While the last three conditions are determined by the experimental conditions, the others are essentially determined by the animal species, the cardiac tissue, the developmental stage, and the pathophysiological condition of the preparation. For instance, cGMP stimulates basal I_Ca via PKG in ventricular myocytes isolated from neonatal (Kumar et al. 1997) and young rabbit hearts (Han et al. 1998), but has no effect on basal I_Ca in adult rabbit heart (Kumar et al. 1997) due to a lower expression of PKG in adult heart (Kumar et al. 1999). Isoprenaline-stimulated I_Ca is inhibited by cGMP in rat (Méry et al. 1991; Sumii & Sperelakis, 1995) and guinea-pig ventricular myocytes via activation of PKG (Levi et al. 1989), while cGMP inhibits the current via PDE2 activation in frog ventricular myocytes (Hartzell & Fischmeister, 1986).

Since PKG has been shown to mediate the stimulatory effect of cGMP on basal I_Ca in rabbit ventricular myocytes (Kumar et al. 1997; Han et al. 1998), we first examined whether PKG contributed to the stimulatory effect of cGMP in human atrial myocytes. However, the negative results obtained with 8-bromo-cGMP or KT 5823 forced us to reject this hypothesis. In the case of 8-bromo-cGMP, a stimulation of I_Ca was observed only at 100 μm, which is a 200-fold higher concentration than necessary when using native cGMP. Nevertheless, since 8-bromo-cGMP is 10-fold more potent than cGMP in activating PKG (Butt et al. 1992), we suspect that the stimulation of I_Ca seen at such a high concentration was not due to PKG but rather to PKA activation. Indeed, 8-bromo-cGMP was shown to activate PKA with a K_d of 12 μm (Butt et al. 1992). In addition, 8-bromo-cGMP was shown to inhibit PDE3 with a K_i of 8 μm (Butt et al. 1992) which may lead to cAMP accumulation and activation of PKA (see below).

The stimulatory effect of cGMP on I_Ca in human atrial myocytes clearly involved activation of a cAMP-dependent phosphorylation process. Indeed, PKA inhibition with PKI completely abolished the stimulatory effect of cGMP. The PKA-mediated activation of I_Ca is most probably due to the phosphorylation of a subunit on the L-type Ca²⁺ channel (Gao et al. 1997; Bünemann et al. 1999; Striessnig, 1999). Interestingly, PKI not only antagonised the cGMP effect but also decreased basal I_Ca amplitude (see also Skeberdis et al. 1997). This indicates that, in human atrial myocytes, a constitutive PKA activity persists in the absence of any cAMP elevating stimulus and contributes to the basal amplitude of I_Ca. This constitutive PKA activity is most probably due to a substantial basal cAMP synthesis resulting from a constitutive activity of adenylyl cyclase. Indeed, acetylcholine decreases (Vandecasteele et al. 1998a) and phosphodiesterase inhibitors increase (Kirstein et al. 1995; Rivet-Bastide et al. 1997) basal I_Ca in this preparation.

To gain further insight into the mechanism by which cGMP enhances PKA activity, we tested the hypothesis that cGMP leads to cAMP elevation though an inhibition of PDE3. Our reasoning was that cGMP inhibits PDE3 in a submicromolar range of concentrations (Butt et al. 1992), and also that PDE3 inhibition with milrinone mimicked the stimulatory effect of a low concentration of NO donors on I_Ca in human atrial myocytes (Kirstein et al. 1995; Vandecasteele et al. 1998a). We found indeed that cilostamide, a highly selective PDE3 inhibitor (Stoclet et al. 1995), mimicked the stimulatory effect of 0.5 μm cGMP on I_Ca and induced little additional effect in the presence of the cyclic nucleotide. Thus we conclude that cGMP stimulation of I_Ca in human atrium is due to inhibition of PDE3.

Surprisingly, increasing the concentration of cGMP from 0.5 to 5 μm reduced the stimulatory effect seen at the lowest concentration. This dual effect of cGMP was observed whether GTP was present or not in the pipette solution, refuting the hypothesis that a progressive reduction in the spontaneous activity of G_s proteins and adenylyl cyclase contributed to this phenomenon. But when GTP was present, the stimulatory effect of 0.5 μm cGMP on I_Ca was reduced approximately 2-fold, possibly because in this situation exogenous cGMP has to compete with endogenous cGMP production which is stimulated by the presence of GTP (Rivet-Bastide et al. 1997).

The attenuation of the stimulatory response of I_Ca to cGMP when the concentration of the nucleotide is increased suggests the existence of a secondary inhibitory mechanism with a lower sensitivity to cGMP than PDE3. Two possible candidates for such an inhibitory mechanism exist: (i) cGMP activation of PDE2 and (ii) cGMP activation of PKG (Lohmann et al. 1991; Fischmeister & Méry, 1996). Whereas PKG inhibition by KT 5823 was ineffective, PDE2 blockade by EHNA, a selective PDE2 inhibitor (Méry et al. 1995), fully reversed the inhibitory effect of 5 μm cGMP. Thus, PDE2 rather than PKG was responsible for the secondary inhibitory effect of cGMP on I_Ca in human atrial myocytes. In this respect, human atrial myocytes behave like frog ventricular myocytes where PDE2 activation is responsible for cGMP inhibition of pre-stimulated I_Ca (Hartzell & Fischmeister, 1986; Méry et al. 1995). This result was somewhat surprising because most previous electrophysiological studies suggested a predominant role of PKG in NO donor and/or cGMP inhibition of I_Ca in mammalian species (Thakkar et al. 1988; Levi et al. 1989; Wahler et al. 1990; Méry et al. 1991; Wahler & Dollinger, 1995; for review, see Lohmann et al. 1991). However, all these studies in mammalian heart were performed in ventricular tissues, so the possibility exists that atrial tissue differs from ventricular tissue in the amount of the respective cGMP target enzymes, their localisation within the cell and/or their coupling to L-type Ca²⁺ channels. The resemblance in the action of cGMP on I_Ca in human atrial and frog ventricular myocytes would support this hypothesis, since these two preparations share a number of similarities both on ultrastructural and functional grounds (see e.g. Morad & Cleeman, 1987).

Our demonstration that cGMP, like NO donors (Kirstein et al. 1995), exerts two opposite effects on I_Ca in human atrium may have pathophysiological relevance. Indeed, human cardiomyocytes possess a Ca²⁺-dependent NO synthase (NOS) subtype which is constitutively expressed in endothelium (NOS3) (Wei et al. 1996), and are at reasonable diffusion distances from other NOS sources located in endothelial (NOS3) and neuronal cells (NOS1). Therefore, NO is likely to modulate myocardial contractility under physiological conditions (Shah & MacCarthy, 2000). Moreover, exogenous application of NO by NO donors modulates cardiac contractility (Kojda & Kottenberg, 1999; Paulus & Shah, 1999; Shah & MacCarthy, 2000), including in human atrium (Flesch et al. 1997). Finally, myocardial cells are exposed to high concentrations of NO upon induction of the Ca²⁺-independent NOS (NOS2), which occurs in several pathological states, such as sepsis (Schulz et al. 1992; Thoenes et al. 1996), heart failure (De Belder et al. 1993; Haywood et al. 1996; Vejlstrup et al. 1998) and cardiac allograft rejection (Yang et al. 1994; Paulus et al. 1997). Although it is likely that the cGMP-dependent alterations in I_Ca observed in this study contribute to the alterations in myocardial function in these pathological situations (but see Abi-Gerges et al. 1999), a number of other mechanisms may also be relevant, including a cGMP-dependent reduction of myofilament response to Ca²⁺ (Shah et al. 1994; Shah & MacCarthy, 2000), cGMP-independent effects on L-type Ca²⁺ channels (Campbell et al. 1996; Hu et al. 1997), ryanodine receptors (Xu et al. 1998), creatine kinase (Gross et al. 1996) or mitochondrial respiration (Wolin et al. 1997) and modulation by NO of cardiovascular reflexes (Zanzinger, 1999). Additional studies are needed to evaluate the respective contribution of each of these mechanisms to the overall cardiac response to NO.