Characteristics of nitric oxide-mediated cholinergic modulation of calcium current in rabbit sino-atrial node

Abstract

1. We have previously shown that nitric oxide (NO) production is essential for cholinergic inhibition of the β-adrenergic stimulated L-type calcium current (I_Ca-L) in rabbit pacemaker (sino-atrial node (SAN)) cells. The present experiments demonstrate the presence of constitutive nitric oxide synthase (cNOS) in SAN cells, and characterize the NO-mediated cholinergic response.

2. Immunohistochemical staining, using an antibody prepared against endothelial cNOS, demonstrated that this enzyme was present in single myocytes obtained from the SAN.

3. The activation of cNOS is known to be Ca²⁺ and calmodulin dependent. Strongly buffering intracellular Ca²⁺ with the membrane-permeable chelator BAPTA-AM (10 μM) significantly reduced (and in some cases abolished) the attenuation of I_Ca-L by the muscarinic agonist carbamylcholine (CCh). In contrast, the CCh-induced activation of an outward K⁺ current, I_K,ACh, was unaffected by buffering of [Ca²⁺]_i. The calmodulin inhibitor 48/80 (20 μM) also abolished the attenuation of I_Ca-L by CCh, with no change in the activation of I_K,ACh.

4. Neither thapsigargin nor ryanodine (5-10 μM), agents which deplete intracellular Ca²⁺ stores, significantly changed the attenuation of I_Ca-L by CCh.

5. Pertussis toxin (PTX) completely abolished both the inhibitory action of CCh on I_Ca-L and the activation of I_K,ACh. This establishes that a PTX-sensitive GTP-binding protein links the muscarinic receptor to NO synthase activation in SAN cells.

6. Our hypothesis is that NO leads to activation of a cyclic GMP (cGMP)-activated phosphodiesterase (PDE II) as a mechanism for enhanced cyclic AMP breakdown and I_Ca-L attenuation. This was supported by showing that a specific inhibitor of PDE II, erythro-9-(2-hydroxy-3-nonyl) adenine (EHNA), blocks the effect of CCh on I_Ca-L, but not on I_K,ACh. Using reverse transcriptase-polymerase chain reaction techniques, we have established that PDE II is the dominant cyclic nucleotide phosphodiesterase isoform in SAN cells.

Nitric oxide (NO), a recently discovered second messenger, is increasingly becoming recognized as a major physiological and pathophysiological regulator of cardiac function (for recent review see Kelly, Balligand & Smith, 1996). We have recently identified a novel role for NO, showing it to be an obligatory mediator of the action of muscarinic agonists in attenuating the β-adrenergic-activated L-type calcium current, I_Ca-L, in rabbit sino-atrial nodal (SAN) cells (Han, Shimoni & Giles, 1995). NO plays no role in the muscarinic activation of a potassium current (I_K,ACh) which is also implicated in cholinergic regulation of cardiac pacing (Irisawa, Brown & Giles, 1993). Our findings, along with recently reported results in the atrioventricular node (AVN: Han, Kobzik, Balligand, Kelly & Smith, 1996), extend earlier work suggesting an important role for NO in cardiac ventricular cell function (see Kelly et al. 1996) to mammalian cardiac pacemaker tissues.

Our working hypothesis (Han et al. 1995) proposes that occupation of muscarinic receptors can activate a constitutive nitric oxide synthase (cNOS) in rabbit SAN cells, resulting in the production of NO. NO then activates a soluble guanylyl cyclase (Schmidt, Lohmann & Walter, 1993); the ensuing elevated cyclic GMP (cGMP) levels activate a cGMP-dependent cyclic AMP (cAMP)-selective phosphodiesterase (PDE II) (Schmidt et al. 1993; Beavo, 1995), resulting in accelerated breakdown of cAMP. Since elevated cAMP levels cause the augmentation of I_Ca-L by β-adrenergic agonists (Hartzell, 1988), the enhanced cAMP breakdown results in the attenuation of I_Ca-L in the presence of muscarinic agonists.

Many of the functional components of this NO-triggered biochemical cascade have been elucidated in a variety of non-cardiac cell systems (Moncada, Palmer & Higgs, 1991). The activation of cNOS is calcium and calmodulin dependent (Marletta, 1993). It is known that cNOS activation is initiated by various ligand-receptor complexes (as well as by mechanical deformation) through heterotrimeric GTP-binding (G)-proteins, some of which are sensitive to pertussis toxin (PTX) (Marletta, 1993). In addition to the sensitivity of cNOS to PTX, cholinergic effects on ionic currents have until recently also been considered to be totally PTX sensitive. However, Li, Hanf, Otero, Fischmeister & Szabo (1994) have reported a PTX-resistant component of muscarinic action in frog cardiac myocytes.

In order to validate further our hypothesis for NO-mediated cholinergic actions on I_Ca-L in SAN cells we set out to determine that cNOS is indeed present in SAN cells. Since cNOS activity is calcium and calmodulin dependent, we investigated whether the attenuation of I_Ca-L by a cholinergic agonist in SAN cells also shows this dependence, as would be expected if cNOS is an obligatory mediator. We then examined the nature of the G-protein linking NOS to the muscarinic receptor in these cells, in terms of being completely or partially PTX sensitive. In addition, we have also investigated an involvement of protein kinase C (PKC), since there is abundant evidence for its involvement in NOS activity (Bredt, Ferris & Snyder, 1992). However, some studies show that PKC activation inhibits cNOS activity (Bredt et al. 1992) as well as muscarinic action (Prestwich & Bolton, 1995), while others show that PKC activation increases NOS activity (Nakane, Mitchell, Forstermann & Murad, 1991). In line with the latter is a recent report showing that PKC inhibitors reduce carbachol (CCh) action and NOS activity in rat atria (Sterin-Borda, Echague, Leiros, Genaro & Borda, 1995). Finally, our working hypothesis suggests that muscarinic agonists produce NO and cGMP, activating the cGMP-stimulated PDE II, which leads to enhanced cAMP breakdown and I_Ca-L attenuation (Han et al. 1995). Since we propose that PDE II involvement is a key element in the cascade of events linking muscarinic receptor binding to I_Ca-L attenuation, we attempted to demonstrate this using two separate methodologies. In electrophysiological recordings we have used the compound erythro-9-(2-hydroxy-3-nonyl)adenosine (EHNA), which has been demonstrated recently (Mery, Pavoine, Pecker & Fischmeister, 1995) to be quite specific as a PDE II inhibitor, in comparison to the non-specific phosphodiesterase inhibitor isobutylmethyl xanthine (IBMX), which we used previously (Han et al. 1995). In addition, we used reverse transcriptase-polymerase chain reaction (RT-PCR) methods to identify PDE isoforms in single SAN myocytes, in order to establish the relative abundance of PDE II and PDE III (cGMP-inhibited) isoforms.

Our results confirm that cNOS is present in SAN cells, and add important new insights into the biochemical cascade following NO production by muscarinic agonists in the mammalian SAN. The results support a key role for the cyclic GMP-dependent phosphodiesterase PDE II in the muscarinic attenuation of I_Ca-L in rabbit primary pacemaker cells.

METHODS

Cell preparation

All electrophysiological measurements were done on single, enzymatically dispersed cells from the rabbit SAN which were prepared as described previously (Han et al. 1995). In brief, rabbits (1.5-2 kg) were anaesthetized with pentobarbitone (60 mg kg⁻¹i.v.) and heparinized (200 units kg⁻¹i.v.). After cervical dislocation, the hearts were removed and perfused retrogradely through the coronary arteries on a Langendorff apparatus (at 37°C, 80 cmH₂O pressure). The perfusate (Tyrode solution) contained (mM): 121 NaCl, 1 CaCl₂, 2.8 sodium acetate, 5 KCl, 1 MgCl₂, 1 Na₂HPO₄, 24 NaHCO₃, 5.5 glucose, equilibrated with 95 % O₂-5 % CO₂ to yield a pH of 7.4. The heart was initially perfused for 5 min with this Tyrode solution to rinse out remaining blood. This was followed by 10 min in Ca²⁺-free Tyrode solution, and then 7-8 min in Tyrode solution containing 11-12 mg (500 ml)⁻¹ (12 U ml⁻¹) collagenase (Yakult Honsha, Tokyo), 5-6 mg (500 ml)⁻¹ protease (Type XIV; Sigma), and 20 mM taurine. The central SAN region was then dissected out, and further incubated (at 37°C) in Ca²⁺-free Hepes-buffered Tyrode solution (composition, mM: 145 NaCl, 5.4 KCl, 1.0 MgCl₂, 1.0 Na₂HPO₄, 5.0 Hepes, 1.8 CaCl₂, 10 glucose, pH 7.4) containing 0.8-1.2 mg ml⁻¹ collagenase (Type I; Sigma) and 0.4-0.6 mg ml⁻¹ elastase (Type III; Sigma). The softened SAN tissue was stirred gently, and single cells were harvested from the superfusate over the next 10-40 min. After centrifugation and removal of the supernatant, the cells were stored (at 4°C) in a storage solution containing (mM): 90 potassium glutamate, 10 potassium oxalate, 25 KCl, 10 KH₂PO₄, 5 Hepes, 0.5 EGTA, 1 MgCl₂, 10 glucose, adjusted to pH 7.2 using KOH.

Drugs

Isoprenaline (Iso), pertussis toxin, thapsigargin and compound 48/80 were purchased from Sigma. BAPTA-AM was purchased from Molecular Probes. All drugs were freshly prepared before use.

Electrophysiological recordings

For current and action potential measurements, aliquots of cells were placed in a 1 ml recording chamber, and superfused with Hepes-buffered Tyrode solution, containing (mM): 140 NaCl, 5.4 KCl, 1.8 CaCl₂, 1 MgCl₂, 0.4 Na₂HPO₄, 5 Hepes, 10 glucose, adjusted to pH 7.4 with NaOH. All recordings were made at 32°C. Only spontaneously beating cells from the central SAN region were used.

Currents and membrane potential were measured using whole-cell voltage- and current-clamp methods, respectively. Since I_Ca-L exhibits rapid ‘run-down’ in these small cells, we used the perforated patch method, in which nystatin (0.4 mg ml⁻¹) is included in the pipette solution. The pipette also contained (mM): 6 NaCl, 140 KCl, 1 MgCl₂, 5 Hepes, 5 glucose, adjusted to pH 7.2 with KOH. I_Ca-L was measured, along with I_K,ACh which is activated by muscarinic agonists, e.g. carbachol (Irisawa et al. 1993). This experimental design enabled us to continuously monitor the sustained action of CCh, even when I_Ca-L was inhibited. This approach was essential for demonstrating that the agents used did not block muscarinic receptors. In most experiments, the cells were held at -40 mV, and depolarizing pulses to 0 mV were applied every 15 s to elicit I_Ca-L. Since I_K,ACh activation by CCh shifted the holding current (and thus the peak inward I_Ca-L in the outward direction), I_Ca-L was measured as the difference between the peak inward and the holding current.

Immunohistochemical detection of cNOS

These experiments were done using procedures which have recently been described in detail (Han et al. 1996). Cryostat sections of SAN tissue and freshly isolated (by enzymatic dispersion, Han et al. 1995) spontaneously beating SAN cells were fixed in buffered 2 % paraformaldehyde for 5 min, followed by 10 min in 100 % methanol. After rinsing, immunostaining was performed by sequential application of primary mouse monoclonal antibody raised against a polypeptide (residues 1030-1209) of the human endothelial cNOS (5 μg ml⁻¹; Transduction Laboratories), followed by goat anti-mouse IgG (1/50; Steinberger Monoclonals, Inc.), and mouse peroxidase-anti-peroxidase complex (1/100; Steinberger Monoclonals, Inc.). This was followed by labelling with the chromogen diaminobenzidine (Sigma) and H₂O₂. The slides were washed with water, counterstained with Haematoxylin, dehydrated, and mounted for light microscopy. Samples for the negative control were treated the same way, either without the primary antibody or with a mouse myeloma IgG (Sigma) as an alternative primary antibody.

RT-PCR and cloning

Amplification by PCR was carried out in a total volume of 50 μl. The template cDNA was obtained by reverse transcription of RNA isolated from SAN cells (pooled from 3-5 rabbits). Three pairs of primers were used according to the manufacturer's (Gibco-BRL) specification. The sequences of the primers were: for PDE II, CCCATCAAGAACGAGAACCAGGAG (sense) and GAAAAGGCGTGCATCCAGTTGTGG (antisense); for PDE III, GATATAGGGATATTCCTTATCATAAC (sense) and TGCAGCAGCTGCGTGATGATTCTC (antisense); for glyceraldehyde 3-phosphate dehydrogenase (GAPDH), CAGGGTCGACCTTGCCCACAGCC (sense) and GCC-AAGGATATCCATGACAACT (antisense). The volumes of the reagents were the following: for PDE II and GAPDH, 3 μl cDNA, 5 μl of ×10 PCR buffer (Gibco-BRL), 1.5 μl 50 mM MgCl₂, 1 μl of 10 mM dNTP, 1.5 μl sense primer, 1.5 μl antisense primer, and 0.6 μl Taq DNA polymerase; for PDE III, 10 μl cDNA, 5 μl of ×10 PCR buffer, 1.5 μl 50 mM MgCl₂, 1 μl of 10 mM dNTP, 1 μl sense primer, 1 μl antisense primer, and 0.6 μl Taq DNA polymerase. Denaturation was carried out at 94°C for 1 min, annealing for thirty-five cycles at (a) 94°C for 30 s, (b) 60°C for 1 min 15 s, and (c) 72°C for 1 min, and extension for one cycle at 72°C, 10 min. PCR products were separated on a 2.0 % agarose gel with Tris acetate-EDTA buffer. The band sizes were verified, extracted and purified (QIA Quick Gel Extraction Kit, Qiagen). Eighty femtomoles of each sample plus 6.4 pmol of each specific primer were used for sequencing.

Statistics

Changes in I_Ca-L and/or I_K,ACh elicited by different protocols were averaged. Students t test was used where appropriate for statistical comparison and evaluation of differences, with P < 0.05 set as the level of significance.

RESULTS

Presence of cNOS in SAN cells

The effects of NO may not be restricted to the cell in which it is generated; NO can be produced in one cell type, and then diffuse to sites of action in adjacent cells (Moncada et al. 1991). To evaluate our hypothesis, it is therefore essential to demonstrate the existence of cNOS in isolated SAN cells.

NOS has been detected by immunolabelling techniques in a variety of heart cells, including intracardiac neurons and both atrial and ventricular myocytes (Klimaschewski et al. 1992; Balligand et al. 1995). Recently, cNOS was shown to be present in AVN as well (Han et al. 1996). There are two known cNOS enzymes (Forstermann, Nakane, Tracey & Pollock, 1993), endothelial and neuronal isoforms. In this study, we find that SAN cells express the endothelial cNOS isoform of this enzyme (ecNOS, or NOS III). Sections from the central SAN region show prominent staining for cNOS (Fig. 1A); similar staining was observed with isolated SAN cells (Fig. 1C and D). No staining was detectable above background when the primary antibody was omitted, or when mouse myeloma IgG was applied (Fig. 1B and E).

Figure 1. Immunohistochemical detection of endothelial cNOS protein in sections of rabbit SAN (A and B) and in freshly isolated SAN cells (C, D and E).

A, SAN cells in situ (arrow) show clear immunostaining with anti cNOS monoclonal antibody. Note the strong staining in adjacent capillaries as well. B, no labelling was seen in a section of SAN when non- specific mouse myeloma IgG was used. C and D, positive staining for cNOS is also seen in isolated SAN cells incubated with the NO endothelial cNOS antibody. E, isolated SAN cells show no staining with non-specific myeloma IgG. The bars represent 40 μm in A and B and 20 μm in C, D and E.

Dependence of muscarinic attenuation of I_Ca-L on intracellular calcium and calmodulin

A functional characteristic of cNOS is dependence on intracellular calcium ([Ca²⁺]_i) (Moncada et al. 1991). We therefore performed a series of experiments in which [Ca²⁺]_i was strongly buffered and then examined whether the attentuation of I_Ca-L by CCh was prevented. Since our experiments were done with the perforated-patch method (preventing the introduction of large molecules through the pipette), we used a membrane-permeable esterified chelator of [Ca²⁺]_i, BAPTA-AM. This molecule is de-esterified by intracellular esterases, resulting in chelation of [Ca²⁺]_i by BAPTA.

Under control conditions, adding CCh in the presence of Iso caused a large reduction in I_Ca-L, as well as activation of an outward K⁺ current, I_K,ACh. An example of these effects is shown in Fig. 2A. However, preincubation with 10-20 μM BAPTA-AM for 10-20 min greatly reduced (n = 7) or completely abolished (n = 9) the attenuation of I_Ca-L by 1 μM CCh. Importantly, the activation of I_K,ACh was not changed, showing that CCh binding to the M₂ muscarinic receptor was not altered. BAPTA did not increase the Iso-stimulated calcium current, with a 2.28 ± 0.41 (mean ± s.e.m.)-fold increase, compared with control (Iso alone) cells (2.30 ± 0.09-fold increase, n = 27). This suggests that a potentiation of adenylyl cyclase activity by BAPTA, as observed by You, Pelzer & Pelzer (1997), does not contribute to the anti-muscarinic effect of BAPTA on I_Ca-L (however, see Discussion). A representative data set from one of these SAN cells is shown in Fig. 2B. In sixteen control cells CCh (1 μM) reduced I_Ca-L to 50.7 ± 3.0 % of the amplitude with Iso alone. In contrast, after incubation in BAPTA (n = 16), I_Ca-L was reduced to only 75.7 ± 3.8 % of the amplitude with Iso alone, a value which is significantly (P < 0.05) different from the control (without BAPTA). I_K,ACh values, measured as the current activated at -40 mV (the holding potential), did not differ significantly (93.8 ± 13.2 pA in control, and 68.9 ± 15.4 pA in BAPTA). These results are summarized in Fig. 3.

Figure 2. CCh actions in the presence of intracellular Ca²⁺ ([Ca²⁺]_i) buffers.

A, in a control cell, isoprenaline (Iso, 1 μM) increased I_Ca-L (b vs. a); CCh (1 μM) attenuated I_Ca-L (c), as well as activating I_K,ACh. These effects were reversible and could be repeated. The superimposed current traces on the right are from times a, b and c, as indicated. Here and in subsequent figures the horizontal arrow to the left of the traces indicates the zero current level, and I_Ca-L magnitude (○) and the holding current (•) are plotted against time. B, SAN cell was pre-exposed to 10 μM BAPTA-AM for 20 min prior to recording, CCh still activated I_K,ACh, but the inhibition of I_Ca-L was reduced. This attenuation of I_Ca-L was smaller during the second exposure, presumably due to increased buffering by BAPTA, as it de-esterifies further from BAPTA-AM. Superimposed current traces on the right were recorded at the times indicated in the left panel.

Figure 3. Summary of muscarinic effects on I_Ca-L and I_K,ACh in the presence of [Ca²⁺]_i buffering.

A, the mean (+s.e.m.) magnitude of I_Ca-L (as a percentage of the amplitude with Iso alone, prior to addition of CCh) is shown for control cells (□, n = 16), and in cells pre-exposed to BAPTA (10-20 μM) for 10-20 min (▪, n = 16). B, the mean +s.e.m. of I_K,ACh for the same groups as in A are shown, illustrating the lack of any significant effect when [Ca²⁺]_i is strongly buffered. *P < 0.05.

The source of the calcium required for cNOS activity in SAN cells is unknown. There have been reports showing that depletion of thapsigargin-sensitive stores can inhibit agonist-induced release of endothelial autacoids (MacArthur, Hecker, Busse & Vane, 1993). We attempted to eliminate the muscarinic action on I_Ca-L in rabbit SAN cells by depletion of intracellular Ca²⁺ stores. In four cells, 10 μM thapsigargin (TG) did not alter the usual effects of CCh; I_Ca-L was reduced to 47.3 ± 11.6 % of its amplitude with Iso alone, and I_K,ACh was 133.3 ± 37 pA. A representative result is shown in Fig. 4. In two other cells, CCh produced the same effects on I_Ca-L and I_K,ACh in the presence of 10 μM ryanodine (not shown). Both TG and ryanodine abolished the spontaneous contractions of these cells, indicating their effective access to the interior of the cells.

Figure 4. Lack of effect of thapsigargin (TG, 10 μM) on CCh actions.

A SAN cell was pre-treated with TG and then the normal protocol was carried out - Iso (1 μM), followed by CCh (1 μM). I_Ca-L is augmented by Iso and strongly attenuated by CCh, which also activates an outward holding current due to I_K,ACh, in the presence of TG.

The Ca²⁺-dependent activation of cNOS also depends on the formation of a calcium-calmodulin complex (Nakane et al. 1991; Bredt et al. 1992). Thus, calmodulin antagonists have been shown to block cNOS activation (Wolff, Datto, Samatovicz & Tempsick, 1993), as well as inhibiting NO-dependent actions such as relaxation of smooth muscle (Schini & Vanhoutte, 1992) and the effects of CCh (Chen, Yu, de Petris, Biancani & Behar, 1995). We therefore examined whether calmodulin antagonists could interfere with the action of CCh on I_Ca-L in SAN cells. An obvious prerequisite for these experiments is to demonstrate that these agents do not directly affect I_Ca-L.

Several calmodulin antagonists were found to affect transmembrane ionic currents in different cells, including a voltage-gated Ca²⁺ current (Li, Satoh, Ginsburg & Bers, 1997). Such antagonists also affect I_K,ACh desensitization (Kim, 1991). In our experiments, two calmodulin antagonists (W-7 and calmidazolium) significantly reduced I_Ca-L (results not shown) and so were unsuitable. However, a third compound, 48/80, with potent calmodulin-inhibitory properties (Ning & Sanchez, 1995), did not affect I_Ca-L. As shown in Fig. 5, addition of 2.5 μM 48/80 did not change the augmentation of I_Ca-L by Iso. In eight cells the Iso-induced augmentation of I_Ca-L in the presence of 48/80 was 1.65 ± 0.07-fold. Subsequent addition of 1 μM CCh resulted in an outward shift in the holding current (reflecting activation of I_K,ACh), but I_Ca-L was not decreased significantly even at 3 μM CCh. In eight cells which were pre-incubated in 48/80 and then exposed to CCh (1 μM), I_Ca-L was 97.0 ± 2.6 % of its value with Iso alone. In these cells I_K,ACh had a mean value of 55.7 ± 9.1 pA (see Discussion). This result indicates that calmodulin is required for the attenuation of I_Ca-L by CCh in SAN cells, presumably due to the calmodulin dependence of cNOS activation.

Figure 5. The calmodulin inhibitor 48/80 blocks CCh actions on I_Ca-L in rabbit SAN.

A, the cell was exposed to 2.5 μM 48/80 along with 1 μM Iso. The subsequent addition of CCh (1 and 3 μM) caused an outward shift in the holding current, but failed to decrease I_Ca-L. B, superimposed current traces corresponding to times a, b and c indicated in A.

Involvement of GTP-binding proteins

In many other systems, cNOS activation by different ligand-activated receptors is mediated by a heterotrimeric G-protein (Marletta, 1993). The activation of M₂ muscarinic receptors is coupled to the binding of GTP to G_i; both α_i- and βγ-subunits are functionally important for inhibiting adenylyl cyclase and activating I_K,ACh, respectively (Meij, 1996). The G_i-protein involved in transduction of M₂ muscarinic effects in mammalian heart can be inhibited by PTX treatment (Meij, 1996). In addition, acetylcholine (ACh) inhibition in rabbit SAN was shown to be inhibited by PTX (Molyvdas & Sperelakis, 1989). However, recently Li et al. (1994) showed that there was a selective effect of PTX on the actions of ACh on I_K,ACh and on I_Ca-L in frog ventricular cells. Their observation of a residual, PTX-insensitive action of ACh on I_Ca-L suggested that different G-proteins may be involved in the transduction mechanisms. Since GDPβS, which abolishes all G-protein activity (Li et al. 1994), caused total suppression of cholinergic effects, contribution of a non-G-protein-mediated process was ruled out (Li et al. 1994). For these reasons, we decided to determine the PTX sensitivity of the CCh-induced block of I_Ca-L in SAN pacemaker cells under our experimental conditions. SAN cells were incubated for 6 h in 1 μg ml⁻¹ PTX, either at room temperature (21-23°C) or at 4°C; thereafter electrophysiological tests of the effects of CCh were performed.

Figure 6 shows spontaneous pacemaker activity and action potentials from a PTX-treated cell, compared with a control cell. In the control cell (Fig. 6A), isoprenaline (Iso, 1 μM) enhanced the rate of pacemaking whereas CCh (0.2 μM) caused a significant slowing of the rate, as well as hyperpolarization due to activation of I_K,ACh. Following PTX pre-treatment (Fig. 6B), Iso was still capable of enhancing the rate, since the G_s-protein linking the β-adrenergic receptor to adenylyl cyclase is PTX insensitive. However, CCh (2 μM) was without any effect on either the pacemaker rate or the maximum diastolic potential. In untreated cells 2 μM CCh totally blocks spontaneous activity. A similar pattern of results was seen in three other cells. Under voltage-clamp conditions, it was established that neither I_K,ACh activation nor I_Ca-L attenuation is achieved by addition of CCh to a PTX-treated cell. This result, shown in Fig. 7, was seen in four other cells. The augmentation of I_Ca-L by Iso was unchanged in the presence of PTX (2.66 ± 0.73-fold).

Figure 6. PTX blocks the effects of CCh on SAN action potentials.

A, spontaneous action potentials from a central SAN cell were recorded in control conditions (C) and following addition of 1 μM Iso, and subsequently 0.2 μM CCh (still in the presence of Iso). Note the slowing of the rate and the marked hyperpolarization following CCh. B, the same protocol was repeated in a cell which had been exposed to 1.0 μg ml⁻¹ PTX for 6 h at room temperature. Note that Iso still enhanced the pacemaker rate, but 2 μM CCh (10-fold higher than in A) did not slow the rate or hyperpolarize the membrane potential.

Figure 7. The effects of PTX pretreatment on CCh action on membrane currents in rabbit SAN.

A, a SAN cell was pretreated with 1.0 μg ml⁻¹ PTX for 6 h. The cell was then voltage clamped and currents recorded. Pulses from -40 mV to 0 were applied every 15 s; the calcium current (I_Ca-L) magnitude (○) and holding current (•) are plotted vs. time. Iso (1 μM) increased I_Ca-L, but CCh (2 μM) was without effect (compare with normal action in Fig. 2). B, superimposed current traces obtained (different cell) for control (C), and following addition of Iso and then CCh (still in Iso).

Thus, in rabbit SAN cells, there is no evidence for a residual PTX-insensitive action of CCh on either I_Ca-L or I_K,ACh. No PTX-insensitive attenuation of the calcium current was evident even at 2 μM CCh, ruling out a shift in concentration dependence such as found by Li et al. (1994), where 1 μM ACh still showed a 50-60 % attenuation of I_Ca-L in cells treated with PTX. Our results demonstrate that in SAN cells, the activation of cNOS is mediated by a PTX-sensitive G-protein, presumably G_i (Meij, 1996).

Possible involvement of protein kinase C in muscarinic effects in SAN

In the heart, muscarinic M₂ receptor activation leads to activation of a phosphoinositide-specific phospholipase C (PLC-β₃ isoform), in addition to inhibition of adenylyl cyclase (Meij, 1996). PLC-β₃ activation by βγ-subunits derived from G_i results in the hydrolysis of a phosphoinositide (PIP₂), leading to production of inositol 1,4,5-trisphosphate (IP₃) and diacylglycerol (DAG); IP₃ releases calcium from intracellular stores, whereas DAG activates PKC.

In several cell types, activation of PKC can lead to a negative feedback regulation of the receptor-G-protein- phospholipase C cascade. PKC activation can directly inhibit phospholipase C (Ryu et al. 1990) or G_i (Katada, Gilman, Watanabe, Bauer & Jakobs, 1985). Thus, some muscarinic actions are subject to negative feedback control by PKC (Prestwich & Bolton, 1995). Furthermore, there is also evidence that PKC activation can directly inhibit endothelial cNOS (Bredt et al. 1992). We attempted to establish whether activation of PKC by phorbol esters (Bogoyevitch, Parker & Sugden, 1993) could interfere with the action of CCh on I_Ca-L in SAN cells. In ten cells exposed to phorbol 12-myristate 13-acetate (PMA; 20-200 nM) and in two cells exposed to phorbol 12,13-dibutyrate (PDBU), no inhibition of the effects of CCh on I_Ca-L was observed. An example is shown in Fig. 8A. A summary of the data (Fig. 8B) from phorbol ester-treated cells, in comparison with control cells, actually shows that the CCh-induced reduction in I_Ca-L in the presence of PMA was significantly larger (P < 0.05) than the control inhibitory effect of CCh. It is unclear if this has physiological relevance, but may indicate that PKC action is synergistic to muscarinic action. Nakane et al. (1991) reported that PKC activation enhances NOS activity, which is consistent with our result. Sterin-Borda et al. (1995) also found that PKC inhibition diminished muscarinic action in rat atria, which suggests that PKC activity maintains NOS function. Whatever the precise mechanism may be, our results indicate that PKC activation in SAN node cells does not inhibit the muscarinic-NOS pathway, as it does in some cell types. The magnitude of I_K,ACh in the presence of PMA (97.9 ± 9.5 pA, n = 10) was similar to the magnitude in control conditions (93.8 ± 13.2 pA, n = 16), and the Iso-induced enhancement of I_Ca-L was not significantly different from control (2.58 ± 0.33-fold).

Figure 8. CCh actions measured following PKC activation by the phorbol ester PMA.

A, in this cell Iso (1 μM) increased I_Ca-L, after which 20 nM PMA was added. Subsequent addition of CCh (1 μM) led to attenuation of I_Ca-L, as well as an activation of I_K,ACh. This protocol was repeated following washout to allow a longer duration of PKC activation. Following a second washout, PMA concentration was increased to 200 nM. A third exposure to CCh in the presence of 200 nM PMA showed that PKC does not alter the cholinergic effects in SAN. B, a summary of the effects of CCh in the presence of PMA (16 cells in each group). The attenuation of I_Ca-L by CCh (left) is somewhat enhanced in the presence of PMA, whereas the effects on the holding current are the same (right).

We also attempted to identify a role for PLC as a mediator of muscarinic action on the calcium current, as is required in our suggested scheme (see Discussion). However, the reportedly most specific PLC inhibitor U-73122 (Smith, Sam, Justen, Bundy, Bala & Bleasdale, 1990) markedly depressed basal I_Ca-L as well as the Iso stimulation. Thus, the role of PLC in the muscarinic action on this current could therefore not be studied using this PLC inhibitor.

In the final set of experiments, we re-examined the target for the cGMP formed by NO action on soluble guanylyl cyclase. We previously suggested (Han et al. 1995) this to be a cGMP-activated phosphodiesterase (PDE II). We have now directly addressed this issue, using two different methodologies. It has recently been shown that EHNA selectively inhibits cGMP-stimulated phophodiesterase in cardiac myocytes (Mery et al. 1995). We therefore tested whether EHNA prevented the attentuation of I_Ca-L by addition of CCh. This was indeed found to be the case, as shown in Fig. 9. The activation of I_K,ACh was unaffected. The inhibition by EHNA was reversible; upon washout of EHNA (in the presence of CCh), the attenuating effect of CCh of I_Ca-L re-developed (Fig. 9Ad). Thereafter, a second application of EHNA blocked CCh action, and I_Ca-L increased. Throughout these procedures, I_K,ACh remained activated, demonstrating that there was no block of muscarinic receptors by EHNA. Figure 9B shows superimposed current traces recorded at different times (denoted a-d) in the protocol. A pattern of results very similar to this one was observed in a total of five cells: the increase in I_Ca-L by Iso in these cells was by 87, 163, 260, 126 and 355 %. Following CCh addition the corresponding current amplitudes were 92, 126, 265, 98 and 311 % above the control values, indicating that CCh does not antagonize Iso action in the presence of EHNA.

Figure 9. CCh actions in the presence of the PDE-II inhibitor EHNA.

A, in this cell Iso (1 μM) led to the usual increase in I_Ca-L (○). CCh (1 μM) attenuated this current, as well as activating I_K,ACh (•). These effects were reversible upon CCh washout, at which time EHNA (30 μM) was introduced. A second exposure to CCh (with EHNA present) activated I_K,ACh again, but failed to attenuate I_Ca-L. Removal of EHNA, still in the presence of CCh, enabled the development of I_Ca-L attenuation. Reapplication of EHNA again blocked this effect. I_K,ACh was activated throughout the exposure to CCh, indicating that EHNA was not interfering with CCh binding to the receptor. B, superimposed current traces taken at the times (a-d) indicated in A.

In a final series of experiments, we used RT-PCR methods to identify the relative abundance of two isoforms of the phosphodiesterases PDE II and PDE III in SAN cells. PDE II and PDE III were detected by using primers from cloned isoforms, with confirmation by subsequent cloning and sequencing of the PCR products. The expression of PDE II in SAN cells was much more abundant than PDE III, as shown in Fig. 10A. This was striking in view of the fact that the cDNA (and reference GAPDH) used for PDE III was 3-fold higher than for PDE II. The PCR product of PDE III (374 bp) is smaller than that of PDE II (427 bp), indicating that the relatively weak PDE III band cannot be due to the size of the PCR product, which is inversely related to the PCR efficiency. A comparison of the homology between the partial nucleotide sequences from the PCR products of the rabbit SAN and the cloned human PDE II showed approximately 91 % identical nucleotides, as illustrated in Fig. 10B.

Figure 10. RT-PCR identification of transcripts for PDE II and PDE III in rabbit SAN cells and homology comparison between the RT-PCR products of rabbit SAN and the cloned human PDE II.

A, both PDE II and PDE III were confirmed by sequencing the RT-PCR products. B, rabbit SAN PDE II shares approximately 91 % identical nucleotides with human PDE II. Three times as much cDNA was required for amplifying PDE III than for PDE II (see Methods).

DISCUSSION

Summary and relation to previous work

The results presented here reveal several novel aspects of the M₂ receptor-mediated muscarinic regulation of I_Ca-L in mammalian primary pacemaker SAN cells. The direct immunohistochemical detection of cNOS in SAN myocytes (Fig. 1) is consistent with our earlier proposal (Han et al. 1995) that NO is an obligatory mediator of CCh action on I_Ca-L in these cardiac pacemaker cells.

Our electrophysiological experiments were designed to evaluate whether features of the NO signalling pathway, which have been demonstrated by a variety of techniques in many cell types, are also present in the muscarinic regulation of I_Ca-L in the rabbit SAN. The [Ca²⁺]_i chelator, BAPTA-AM, significantly diminished the action of CCh on I_Ca-L (Figs 2 and 3).

The effect of BAPTA was not examined with sub-maximal Iso concentrations, so the possibility that part of the ability of BAPTA to block the inhibitory effect of CCh on I_Ca-L is due to an increase in adenylyl cyclase activity (You et al. 1997) cannot be excluded.

The [Ca²⁺]_i dependence of CCh action on I_Ca-L, but not on I_K,ACh, is consistent with a requirement for an increase in [Ca²⁺]_i for cNOS activity. A similar dependence of NOS on internal calcium, based on the abolition of NOS function by BAPTA, has also been reported in rat ventricular myocytes, where NO is proposed to regulate contractile function (Kaye, Wiviott, Balligand, Simmons, Smith & Kelly, 1996). A CCh-induced increase in [Ca²⁺]_i may be very localized in the myoplasm of SAN cells since cellular contractility is not increased by cholinergic agonists. A spatial heterogeneity in calcium compartments may also be part of the cause for the observed variability in results with BAPTA. This variability may be due to limited accessibility to the calcium stores, as well as to the requirements for intracellular de-esterification of the membrane-permeable chelator. Compartmentation and co-localization of transducing elements have recently been suggested for cAMP (Jurevicius & Fischmeister, 1996).

The source of the augmented [Ca²⁺]_i for cNOS remains unknown. Presumably, CCh increases phosphoinositide turnover (PLC-β₃ activation by G_i-derived βγ-subunits) which produces IP₃ and DAG (Meij, 1996). IP₃ then presumably releases calcium from some internal stores, which in these cells are not depleted by thapsigargin or ryanodine (Fig. 4). The calcium required for cNOS activation may be (at least partly) derived from influx through the I_Ca-L channel, with an efficient buffering by BAPTA preventing the association with calmodulin and NOS activation.

The ability of the calmodulin antagonist, 48/80, to selectively abolish the inhibitory effect of CCh on I_Ca-L (Fig. 5), also supports the NO hypothesis, since cNOS activity also requires calmodulin. However, this statement must be made with caution because compound 48/80 is not completely specific. For example, 48/80 has been reported to be a phosphodiesterase inhibitor (Veigl, Klevitt & Sedwick, 1989), as well as an inhibitor of PLC (Bronner et al. 1987). These non-specific effects would also inhibit the cholinergic cascade that we have proposed (Han et al. 1995, and below). In addition, the outward current elicited by CCh in the presence of 48/80 was smaller than in control conditions (barely significant, at the 0.05 level). This may suggest an effect of this compound at the G-protein level. A partial inhibition of the cholinergic receptor cannot be ruled out, although a substantial amount of outward current is activated. Furthermore, the Iso-induced augmentation of I_Ca-L may have been smaller in the presence of 48/80, although this was not significant with the small number of cells used. Although the absence of an effect of 48/80 on basal I_Ca-L permitted the use of this agent as a putative calmodulin antagonist, it must be acknowledged that compound 48/80 is a mixture of molecules with uncertain selectivity. Nevertheless, the cholinergic effect on I_Ca-L is completely abolished by 48/80, supporting the suggestion that calmodulin is required. In future, when better calmodulin antagonists (that do not directly affect the calcium channel) become available, this issue will need to be further addressed. The effects of 48/80 in conjunction with those of BAPTA are consistent with the involvement of the Ca²⁺- and calmodulin-dependent NOS in this complex signalling mechanism.

The effect of CCh on both I_Ca-L and I_K,ACh in rabbit SAN cells was completely blocked by PTX pretreatment (Figs 6 and 7), in contrast to the partially PTX-insensitive action of ACh on I_Ca-L in frog myocytes (Li et al. 1994). Therefore, G_i probably couples the M₂ receptor to multiple effectors in rabbit SAN cells, as has been suggested (Meij, 1996).

Our study involved an attempt to address the issue of a possible PKC regulation of the cholinergic action on I_Ca-L. CCh activates several PKC isoforms (Puceat, Hilal-Dandan, Strulovitch, Brunton & Heller-Brown, 1994). However, there are also many indications that in some cell types PKC activation acts as a negative feedback regulator of NOS and/or of cholinergic action (Katada et al. 1985; Ryu et al. 1990; Prestwich & Bolton, 1995). In contrast, there are also reports of enhancement of NOS activity by PKC activation (Nakane et al. 1991), as well as of attenuated cholinergic action when PKC is suppressed (Sterin-Borda et al. 1995). We used phorbol esters to activate PKC (Bogoyevitch et al. 1993), and measured cholinergic effects under these conditions. The attenuation of I_Ca-L was not inhibited, but was actually larger in the presence of PMA (Fig. 8); I_K,ACh was unchanged. Thus, in SAN cells, PKC does not seem to inhibit cholinergic action on I_Ca-L (see below).

Finally, we have made use of a recent report describing EHNA as a specific inhibitor of PDE-II (Mery et al. 1995) to further address the issue of the mechanism of action of NO and the cGMP it generates. Previously (Han et al. 1995), we used IBMX, which blocks all isoforms of PDE. By using EHNA as a more selective inhibitor of the cGMP-activated isoform (PDE-II), and confirming that CCh no longer attenuates I_Ca-L, we can strengthen the hypothesis that muscarinic agonists attenuate I_Ca-L by activating this PDE. The RT-PCR experiments used to evaluate the relative abundance of PDE II and PDE III in SAN cells clearly showed that PDE II, the cGMP-activated isoform, is by far the more abundant one (Fig. 10). This greatly supports the proposed chain of events leading to I_Ca-L attenuation by muscarinic agonists in pacemaker cells.

A summary of our current conception of the signalling pathway by which muscarinic agonists attenuate I_Ca-L in SAN cells is presented in Fig. 11. The components which were manipulated in this study (by PTX, EHNA, BAPTA, 48/80) are marked {font AaSans}X. The possible effects of PKC on either cNOS itself, or on the G_i-protein linking the receptor (M₂-R) to the PLC or to the I_K,ACh channel, have been marked with question marks.

Figure 11. A schematic illustration of the pathways by which CCh affects I_Ca-L and I_K,ACh (boldly outlined boxes) in SAN cells, in the presence of Iso-stimulated cAMP levels.

Experimental interventions used in this study, PTX, BAPTA and 48/80, as well as EHNA, are labelled {font AaSans}X. The participation of PKC as a regulator of G_i or NOS is indicated by question marks. The involvement of PLC as a possible source of diacylglycerol (DG) and IP₃ is based upon work by others (Meij, 1996).

Limitations of this study

The evidence for our working hypothesis has been greatly strengthened by the present work. However, although the multiple protocols used all give consistent results, the evidence does remain indirect. Unfortunately, we were unable to test directly our hypothesis in experiments using flash photolysis with caged NO or caged cGMP. Our interpretation for this failure is that the efficiency of our apparatus was too low to produce amounts of either NO or cGMP which would be sufficient to attenuate I_Ca-L. Higher flash intensities appeared to damage these fragile cells.

Nevertheless, despite this limitation, our results are the first to demonstrate the calcium and calmodulin dependence of cholinergic action in pacemaker cells, as well as its total PTX sensitivity. One must bear in mind, however, that the calmodulin inhibitor 48/80 may not be completely selective (as mentioned in the appropriate section), as well as its possible effects on I_K,ACh.

The issue of PKC involvement in this signalling pathway is clearly complex, and must await future additional experiments. Using phorbol esters has limitations, as some PKC isoforms are insensitive to phorbol esters (Bogoyevitch, et al. 1993) which may also have other effects in addition to PKC stimulation. It is unknown which isoforms exist in the SAN, although phorbol esters affect the major PKC isoforms detected in heart (Bogoyevitch et al. 1993). The conflicting reports in the literature regarding the effects of PKC inhibition or stimulation on cholinergic effects or NOS activity may relate to species or tissue differences. Our results indicate that in SAN, concentrations of phorbol esters used to activate PKC in other cardiac tissues (Bogoyevitch et al. 1993) do not inhibit the cholinergic attenuation of I_Ca-L. All these studies are hampered by the limitations of using different inhibitors, which may have additional unknown effects. Thus, the involvement of PLC as a component linking the M₂ receptor to NOS stimulation could not be established, due to direct effects of the PLC inhibitor U-73122 (Smith et al. 1990) on the current.

The present experiments showed that buffering internal calcium prevented the attenuation of I_Ca-L by CCh, although attempts to deplete internal stores were ineffective. Thus, the source of the calcium required to activate NOS is unknown, but must be very localized, as no contraction is elicited by CCh. Cholinergic action on I_Ca-L was previously considered to act by an inhibition of adenylyl cyclase (see Fig. 11). It is known that the isoforms of adenylyl cyclase present in the heart (V and VI) are inhibited by calcium (Iyengar, 1993). If this is part of the mechanism of CCh action, the buffering of calcium may prevent this inhibition and thus the cholinergic action. This is unlikely, however, since the same adenylyl cyclase isoforms are insensitive to calmodulin. The combination of calcium and calmodulin sensitivity, in addition to our previous work, strongly support the NO-mediated pathway.

In summary, our scheme suggests a remarkably complex pathway by which the calcium current is regulated in SAN cells. It is important to acknowledge, however, that although our results provide strong evidence for an obligatory involvement of cNOS in the M₂-mediated cholinergic modulation of I_Ca-L in rabbit SAN cells, this mechanism may not be operative in other cardiac tissues. Thus, although similar effects have now been found in the AVN of rabbit heart (Han et al. 1996), findings from the atria and ventricles of adult and neonatal mammals are at variance. For example, in rat atria (Nawrath, Baumner, Rupp & Oelert, 1995), effects of cholinergic stimulation are not altered by inhibition of the NO pathway. Moreover, in guinea-pig ventricle, cholinergic responses recorded in the presence of isoproterenol were not significantly affected by nitric oxide synthase inhibitors or nitric oxide donors (Zakharov, Pieramici, Kumar, Prabhakar & Harvey, 1995). Although some of these results may be due to species differences, cell-specific mechanisms responsible for these pharmacological responses also appear to be present in mammalian heart.