The inotropic effect of cardioactive glycosides in ventricular myocytes requires Na⁺–Ca²⁺ exchanger function

Abstract

Glycoside-induced cardiac inotropy has traditionally been attributed to direct Na⁺–K⁺-ATPase inhibition, causing increased intracellular [Na⁺] and consequent Ca²⁺ gain via the Na⁺–Ca²⁺ exchanger (NCX). However, recent studies suggested alternative mechanisms of glycoside-induced inotropy: (1) direct activation of sarcoplasmic reticulum Ca²⁺ release channels (ryanodine receptors; RyRs); (2) increased Ca²⁺ selectivity of Na⁺ channels (slip-mode conductance); and (3) other signal transduction pathways. None of these proposed mechanisms requires NCX or an altered [Na⁺] gradient. Here we tested the ability of ouabain (OUA, 3 μm), digoxin (DIG, 20 μm) or acetylstrophanthidin (ACS, 4 μm) to alter Ca²⁺ transients in completely Na⁺-free conditions in intact ferret and cat ventricular myocytes. We also tested whether OUA directly activates RyRs in permeabilized cat myocytes (measuring Ca²⁺ sparks by confocal microscopy). In intact ferret myocytes (stimulated at 0.2 Hz), DIG and ACS enhanced Ca²⁺ transients and cell shortening during twitches, as expected. However, prior depletion of [Na⁺]_i (in Na⁺-free, Ca²⁺-free solution) and in Na⁺-free solution (replaced by Li⁺) the inotropic effects of DIG and ACS were completely prevented. In voltage-clamped cat myocytes, OUA increased Ca²⁺ transients by 48 ± 4% but OUA had no effect in Na⁺-depleted cells (replaced by N-methyl-d-glucamine). In permeabilized cat myocytes, OUA did not change Ca²⁺ spark frequency, amplitude or spatial spread (although spark duration was slightly prolonged). We conclude that the acute inotropic effects of DIG, ACS and OUA (and the effects on RyRs) depend on the presence of Na⁺ and a functional NCX in ferret and cat myocytes (rather than alternate Na⁺-independent mechanisms).

Cardioactive glycosides selectively bind to and inhibit the Na⁺–K⁺-ATPase, which regulates the trans-sarcolemmal Na⁺ and K⁺ concentration gradients. Cardioactive glycosides have been important in treating congestive heart failure for more than 200 years, in large part because of a positive inotropic effect. The now traditional view is that direct Na⁺–K⁺-ATPase inhibition leads to elevation of [Na⁺]_i with consequent gain of cellular and sarcoplasmic reticulum (SR) [Ca²⁺] as a result of shifts mediated via the Na⁺–Ca²⁺ exchanger (NCX; Bers, 2001). That is, with higher [Na⁺]_i, Ca²⁺ extrusion via NCX is reduced (Satoh et al. 2000) and Ca²⁺ influx via NCX may be enhanced. Thus, there can be higher diastolic [Ca²⁺]_i (closer to threshold for contractile activation), higher SR Ca²⁺ content (causing greater SR Ca²⁺ release during contraction) and even greater Ca²⁺ influx via NCX during the action potential (AP). All of these mechanisms lead to increased Ca²⁺ transient amplitude and inotropy, but they all depend upon increased [Na⁺]_i (even if only locally). However, recent reports have challenged this view, suggesting that Na⁺-independent mechanisms contribute significantly to the cardioactive glycoside-induced inotropy. These include glycosides: (a) directly activating SR Ca²⁺ release (via ryanodine receptors; RyRs); (b) increasing Ca²⁺ selectivity and Ca²⁺ influx via TTX-sensitive Na⁺ channels; and (c) other signalling mechanisms.

Regarding the first Na⁺ independent mechanism, it has been reported that cardioactive glycosides activate RyR opening in cardiac SR vesicles and lipid bilayers (Wasserstrom et al. 1991; McGarry & Williams, 1993; Sagawa et al. 2002) and skeletal muscle vesicles (Sarkozi et al. 1996). As relatively lipophilic cardioactive steroids, such as digoxin (DIG) or acetylstrophanthidin (ACS) can cross the sarcolemma, they could directly activate RyRs in intact myocytes. Hydrophilic glycosides (e.g. ouabain; OUA) are not expected to cross the plasma membrane (Jones et al. 1980; Bers et al. 1980), but might gradually gain intracellular access via receptor-mediated endocytosis (Nuñez-Durán et al. 1988), non-specific membrane transporters in hepatocytes (Eaton & Klaassen, 1978) or by limited diffusion. However, receptor-mediated internalization probably represents natural Na⁺–K⁺-ATPase turnover (Pollack et al. 1981), and hepatocyte OUA uptake probably reflects cell-specific drug metabolism and degradation. In isolated myocytes, OUA entry may depend mainly on simple trans-sarcolemmal diffusion, which should be extremely low. Thus, the effects of OUA on intracellular targets (e.g. RyRs) should be very slow (compared with those of DIG) in intact myocytes.

Nishio et al. (2002) suggested that a positive inotropy evoked by OUA in ventricular myocytes in the absence of external Na⁺ could be explained by the OUA-induced increase in RyR sensitivity to cytosolic Ca²⁺. They note that the apparent RyR sensitization could synergize with the elevated SR Ca²⁺ content as a result of Na⁺–K⁺-ATPase inhibition in intact cells (Sagawa et al. 2002). However, the same group reported reduced RyR activity (as Ca²⁺ spark frequency) in intact and permeabilized myocytes upon exposure to OUA in the absence of a Na⁺ gradient (Nishio et al. 2004). Thus, there is a dichotomy in their results. It has also been shown that altered RyR Ca²⁺ sensitivity can only cause transient (not steady-state) inotropy (Eisner & Trafford, 2000; Lukyanenko et al. 2001). This raises the question of whether OUA-induced alteration of RyR sensitivity could produce sustained inotropy.

With regard to the second mechanism, the ionic selectivity of the cardiac TTX-sensitive Na⁺ channel was reported to be drastically altered by cardioactive glycosides such as OUA and DIG, allowing substantial Ca²⁺ influx via Na⁺ channels (slip-mode conductance; Santana et al. 1998; Cruz et al. 1999). The mechanism by which glycosides alter Na⁺ channel selectivity is not known but it was suggested that a direct interaction between Na⁺–K⁺-ATPase (upon glycoside binding) and the Na⁺ channel could mediate this process. Although slip-mode conductance of Na⁺ channels has been challenged (Chandra et al. 1999; Nuss et al. 1999; DelPrincipe et al. 2000; Piacentino et al. 2002), the hypothesis remains controversial.

Finally, the third possibility involves the parallel activation of additional signalling pathways that do not require inhibition of the Na⁺–K⁺-ATPase enzymatic activity or altered [Na⁺]_i upon glycoside binding (Tian et al. 2001; Xu et al. 2001; Xie & Askari, 2002). Tian et al. (2001) suggested that OUA binding to the Na⁺–K⁺-ATPase leads to the activation of protein tyrosine kinases and mitogen-activated protein kinases (MAPKs), which causes increased Ca²⁺ transients and also regulates the transcription of growth-related factors. However, the pathway that links activation of tyrosine kinases and Ca²⁺ increase remains to be determined. Xu et al. (2001) demonstrated that a Na⁺–K⁺-ATPase site-specific antibody competes with OUA binding and induces inotropy without inactivating the enzymatic activity of the pump, suggesting that the binding of the antibody or glycoside triggers alternate signalling pathways.

None of these three alternative mechanisms involve alterations in [Na⁺]_i or require NCX participation, but could work in concert with the classic Na⁺–K⁺-ATPase–NCX pathway discussed above. Our aim was to test whether cardioactive steroids (OUA, DIG and ACS) cause positive inotropy in intact ferret and cat ventricular myocytes in the absence of Na⁺ (in Na⁺-free intracellular and extracellular conditions), when NCX is also blocked. To eliminate Na⁺, cells were first depleted of Na⁺_i in Na⁺-free, Ca²⁺-free solution by the combined action of Na⁺–K⁺-ATPase and passive Na⁺ efflux (Bassani et al. 1994a,b; Ginsburg et al. 1998; Despa et al. 2004). Then Ca²⁺ was returned in a Na⁺-free solution containing Li⁺ (or N-methyl-d-glucamine; NMDG) in place of Na⁺. Intact ventricular myocytes were studied with electrical field stimulation and under voltage clamp. We also tested whether OUA could alter spontaneous Ca²⁺ sparks in permeabilized cat ventricular myocytes in Na⁺-free solution (using confocal microscopy). Although glycosides produced the expected inotropy in control Na⁺-containing solutions, we found no evidence for inotropy in the absence of Na⁺ or NCX function. We conclude that Na⁺-independent pathways make a negligible contribution to the acute inotropic effect of cardioactive glycosides.

Methods

Cardiac myocyte isolation

Ferret and cat ventricular myocytes were isolated as previously described (Nuss & Houser, 1992; Wang & Lipsius, 1995; Piacentino et al. 2000; Weber et al. 2001), and experiments were carried out according to the guidelines of the Loyola University Chicago and the Temple University School of Medicine animal welfare committees. Briefly, adult ferrets and cats were anaesthetized with pentobarbital sodium (50 mg kg⁻¹i.p.) or ketamine/acepromazine (40/0.6 mg kg⁻¹i.m.). Hearts were quickly excised, mounted on Langendorff apparatus and perfused retrogradely with nominally Ca²⁺-free solution for 5 min, and then perfused with a collagenase-containing solution (Yakult Corp. or Boehringer) at 37°C until soft. Left ventricular tissue was gently minced, filtered and washed. Dissociated cells were kept at 23°C and used within 8 h of isolation.

[Ca²⁺]_i measurements and cell shortening

Isolated myocytes (plated on laminin-treated glass coverslips) were loaded with 4–10 μm membrane permeable (acetoxymethyl ester; AM) fluo-3 (Molecular Probes, Eugene, OR, USA) at 23°C. After 15–20 min, cells were superfused with fresh Tyrode solution for 40 min to allow de-esterification of fluo-3 AM.

Calcium measurements were performed on the stage of an inverted microscope (Zeiss, Germany or Olympus, Japan) with excitation at 480 or 495 ± 5 nm and fluorescence emission recorded at 530 or 535 ± 20 nm. The background fluorescence in unloaded cells was subtracted off-line and fluorescence (F) was normalized to the resting fluorescence (F₀) and reported as F/F₀. For field stimulation experiments (0.2 Hz), twitches were evoked by 2 to 4 ms rectangular voltage pulses through platinum electrodes. Cell shortening was measured continuously, using red light and a video edge-detection system (Crescent electronics, Sandy, UT, USA).

These experiments were performed in a normal Tyrode solution (NT) containing (mm): NaCl 140, KCl 4, MgCl₂ 1, CaCl₂ 2, glucose 10 and Hepes 5; with pH adjusted to 7.4 at 23°C. For Na⁺-free solution (0 Na⁺) NaCl was replaced by LiCl. For Ca²⁺-free solution (0 Ca²⁺), CaCl₂ was replaced by MgCl₂, and 1 mm EGTA was added. ACS (Sigma, St Louis, MO, USA) was added from a 5 mm stock in ethanol. DIG (Sigma) was added from a 20 mm stock in DMSO and OUA (Sigma) was directly dissolved in the recording solution.

Electrophysiology

Voltage-clamp experiments were performed in fluo-3-loaded cells at 37°C. Micropipettes (resistance, 1–3 MΩ) were used in perforated-patch mode to preserve the intracellular milieu (Rae et al. 1991; Piacentino et al. 2000). Membrane capacitance and series resistance were compensated up to 65%. Membrane voltage was controlled by an Axoclamp 2 amplifier in discontinuous mode (6–8 kHz; Axon Instruments, Union City, CA, USA) as previously described (duBell & Houser, 1989). The sample clock (indicative of clamp speed) was monitored on a separate oscilloscope to ensure adequate settling of the switch clamp circuit. The amplifier was controlled by pClamp 8.0 software and data were acquired by a Digidata 1200 analog–digital converter and stored on a PC. Data were analysed with Clampfit 8.0 software. The NT for these experiments contained (mm): NaCl 140, sodium pyruvate 2, KCl 5.4, MgCl₂ 1.2, CaCl₂ 1, glucose 10 and Hepes 5; pH 7.4. In the 0 Na⁺ solution, NaCl was replaced by NMDG and no sodium pyruvate was added. Pipette solution contained (mm): potassium glutamate 120, KCl 25, Hepes 10, MgCl₂ 1 and CaCl₂ 1, and 240 μg ml⁻¹ amphotericin B (from a 60 mg ml⁻¹ stock solution in DMSO); pH adjusted to 7.2 with KOH.

Ca²⁺ sparks in streptolysin O-permeabilized myocytes

We used streptolysin O (SLO) permeabilization and confocal microscopy as previously described (Li et al. 2002). Briefly, myocytes on the microscope stage were superfused with relaxing solution containing (mm): EGTA 0.1, MgATP 5, Hepes 10, potassium glutamate 150, MgCl₂ 0.25 and glutathione (reduced form) 10; pH 7.4. Myocytes were then permeabilized (23°C) by addition of SLO (∼300 U ml⁻¹) in the presence of 2 μm fluo-3 (free acid). The permeabilization time varied from 10 to 25 min (assessed by the increase in fluo-3 fluorescence; F). When cell and bath F were comparable, SLO solution was replaced by internal solution containing unless indicated (mm): EGTA 0.5, free Ca²⁺ 50 nm, caesium glutamate 200, Hepes 10, glucose 5, MgATP 5, phosphocreatine di-tris 5, MgCl₂ 0.5, glutathione (reduced form) 10 and fluo-3 (free acid) 0.05, and 5 U ml⁻¹ creatine phosphokinase and 8% dextran (molecular weight, 40 000). Ca²⁺ sparks were measured using a confocal microscope (LSM 410, Zeiss) equipped with a 40 × oil immersion objective (Zeiss, Plan-Neofluar; NA, 1.3). Fluo-3 was excited at 488 nm and emitted fluorescence was collected at >515 nm. Images were recorded in line-scan mode with 512 pixels per line at 250 Hz. [Ca²⁺]_i was calculated by the pseudo-ratio method (Cheng et al. 1993): [Ca²⁺]_i = K_d(F/F₀)/{(K_d/[Ca²⁺]_rest+ 1) −F/F₀} where K_d = 700 nm and the resting Ca²⁺ concentration ([Ca²⁺]_rest) = 50 nm. F₀ refers to the fluorescence level at resting [Ca²⁺]_i.

Ca²⁺ sparks were characterized by an algorithm in Interactive Data Language (IDL 5.3, Research Systems Inc., Boulder, CO, USA; Song et al. 1997; Cheng et al. 1999). The algorithm detects Ca²⁺ sparks as areas of increased fluorescence versus standard deviation (s.d.) of the background of the fluorescence image. We used a threshold of 3.8 ×s.d. for Ca²⁺ sparks. Peaks were normalized to F/F₀. Ca²⁺ spark duration was taken as the time above 50% level of the peak (full-duration half-maximum; FDHM). Width of Ca²⁺ sparks was indicated by the spatial size at 50% level of the peak (full-width half-maximum; FWHM). Ca²⁺ spark frequency (CaSpF) was normalized to cell volume and time (pl⁻¹ s⁻¹); assuming voxel length and width of 0.2 μm and depth of 1 μm.

Statistics

All data were stored on a PC for off-line analysis and are reported as means ± s.e.m. Data were statistically analysed by paired Student's t test or ANOVA followed by all pairwise multiple comparison (Holm-Sidak method; SigmaStat 3.0; SPSS Inc., Chicago, IL, USA). P < 0.05 was considered significant.

Results

Cell shortening

The effects of the cardioactive glycosides DIG and ACS were evaluated in freshly isolated, intact ferret ventricular myocytes. Fluo-3-loaded cells were stimulated at 0.2 Hz and [Ca²⁺]_i and cell shortening were measured. The inset in Fig. 1 shows that in control NT (where NCX was functional), 4 μm ACS (or 20 μm DIG, not shown) produced the expected increase in cell contraction, which was accompanied by an increase in fluo-3 Ca²⁺ signal within 5 min (not shown). In order to determine whether a functional NCX is necessary for the inotropic response, we exposed ferret cardiac myocytes to ACS or DIG under Na⁺-free conditions, where NCX activity is prevented.

Figure 1. Protocol for assessing inotropy under Na⁺-free conditions in ferret ventricular myocytes.

The cell was field stimulated at a relatively low frequency (0.2 Hz). Shortening was measured continually using a video edge-detector system. Upon steady state, the external NT was replaced by a Na⁺-free, Ca²⁺-free solution. The cell was kept in this solution for about 10 min until the cytosol was depleted of Na⁺. Then, normal external [Ca²⁺] (2 mm) was restored to the bath and electrical stimulation proceeded until a new steady state in shortening and [Ca²⁺]_i transients (not shown) was reached. ACS (4 μm) or DIG (20 μm) was then applied in Na⁺-free conditions to evaluate inotropy. Inset shows the normal inotropic effect of 4 μm ACS on cell shortening in the presence of external Na⁺.

To deplete intracellular [Na⁺] and block NCX (Fig. 1), we used the procedure developed and validated by Bassani et al. (1994a,b) as follows. First, NT was switched to a Na⁺-free, Ca²⁺-free solution for 10 min. This allows the Na⁺–K⁺-ATPase and passive Na⁺ efflux to completely deplete [Na⁺]_i (in agreement with direct [Na⁺]_i measurements; Despa et al. 2004). It is important that Ca²⁺ is absent in this external solution to avoid progressive cellular Ca²⁺ loading during depletion of [Na⁺]_i. After 10 minutes in 0 Na⁺/0 Ca²⁺ solution, 2mm Ca²⁺ was added to the Na⁺ -free bath solution (Li⁺-substituted NT solution). We previously demonstrated the efficacy of this approach in preventing Ca²⁺ entry via NCX upon readmission of Ca²⁺_o in rabbit, rat and ferret ventricular myocytes, even when [Ca²⁺]_o was raised to 50 mm (Bassani et al. 1994a,b). To appropriately test for Na⁺-independent effects of cardiac glycosides it is crucial to obtain a true steady state in Na⁺-free solution prior to drug addition (in these studies we waited 10–20 min).

In Na⁺-free NT (2 mm[Ca²⁺]_o), field stimulation was resumed at 0.2 Hz (as in NT). It is important to appreciate that Li⁺ produces Na⁺ channel current comparable to that produced by Na⁺, allowing APs to occur. The frequency of stimulation must be kept low to prevent progressive cellular Ca²⁺ overload in the absence of NCX. This is important because NCX is normally the main route of Ca²⁺ efflux from the cell and so the Ca²⁺, which enters via the Ca²⁺ current during each beat, must be extruded by the sarcolemmal Ca²⁺-ATPase, which is much slower. The relatively powerful sarcolemmal Ca²⁺-ATPase in ferret ventricular myocytes (Bassani et al. 1994b, 1995; Ginsburg et al. 1998) makes these experiments especially practical in this species.

After 10–20 min in this Na⁺-free, 2 mm Ca²⁺ solution, the Ca²⁺ transients and cell shortening reached a new steady state. Then ACS (or DIG) was added to the bath. In the absence of external and internal Na⁺, ACS did not alter twitch contractions (e.g. Figure 1, with ACS exposure for 40 min) or Ca²⁺ transients (not shown). Figure 2A shows pooled data for the percentage change in shortening upon addition of ACS from six experiments in NT and six experiments in Na⁺-free conditions. Differences are significant for all NT data points (P < 0.05), but none of the Na⁺-free time points. Figure 2B shows similar results obtained with DIG.

Figure 2. Pooled data of the effect of 4 μm acetylstrophanthidin (ACS, A) and 20 μm digoxin (DIG, B) on shortening in ferret ventricular myocytes.

Data were measured at the indicated time after exposure to ACS or DIG in the presence or absence of Na⁺. Each data point represents the average of six cells from six different animals for ACS, and four cells from three different animals for DIG. Shortening is expressed as percentage change normalized to the recordings before addition of glycoside. *P < 0.05 compared with control.

Steady-state twitch amplitude in Na⁺-free solution was higher than in control solution (by ∼28%; Fig. 3). This is probably due to the weaker ability of the sarcolemmal Ca²⁺-ATPase to compete with the SR Ca²⁺-ATPase than is the case for NCX. This results in a new steady state with higher SR Ca²⁺ load and Ca²⁺ transient amplitude. However, there is still capacity for substantial inotropy because, in the presence of Na⁺, cell contractions can be increased by ∼100% by many inotropic mechanisms (Bers, 2001). Indeed, in the same cells, the reintroduction of millimolar levels of [Ca²⁺]_o in Na⁺-free solution typically caused a transient increase in twitch amplitude (of ∼50%versus NT), which recovered to a steady-state level ∼28% above control. This probably reflects the transient extrusion of the last small amount of Na⁺ from the cell (and consequent Ca²⁺ gain). Thus the failure of ACS or DIG to induce inotropy in Na⁺-free conditions cannot be due to a ‘ceiling’ effect prior to addition of the drug.

Figure 3. Changes in shortening of ferret ventricular myocytes exposed to Na⁺-free solution, before and after addition of 4 μm acetylstrophanthidin (ACS, A) and 20 μm digoxin (DIG, B).

Data were normalized to shortening measured in NT. Each bar represents the pooled data for six and four different cells exposed to ACS and DIG, respectively. Peak shortening was measured in [Na⁺]_i depleted cells (as illustrated in Fig. 1) 2–5 min after external Ca²⁺ had been readmitted. This transient increase in shortening probably represents the removal of remaining intracellular Na⁺, with the consequent increase in Ca²⁺. Steady-state (SS) shortening was measured ∼20 min after external Ca²⁺ readmission. ACS and DIG data were measured 20 min after glycoside exposure in Na⁺-free, 2 mm Ca²⁺ solution.

Ca²⁺ transients under voltage-clamp conditions

The experiments described so far were performed in field-stimulated ferret ventricular myocytes. Nishio et al. (2002) suggested a direct Na⁺-independent positive inotropic effect of glycosides in voltage-clamped cat ventricular myocytes. In addition, experiments in Figs 1 and 2 were done without control of membrane potential, such that serendipitous offsetting effects on AP duration cannot be unequivocally ruled out. To address both of these issues, a series of experiments was performed in voltage-clamped cat ventricular myocytes at 37°C.

Figure 4 shows the effect of OUA (3 μm) on Ca²⁺ transients during steady-state 0.5 Hz depolarizations (to +10 mV from a holding potential of −70 mV, 250 ms, pulse duration). Figure 4A shows the expected increase in Ca²⁺ transient in the presence of Na⁺ in a representative cell. This normal inotropic response was seen regardless of pulse duration (50–500 ms).

Figure 4. Ca²⁺ transients in voltage-clamped cat ventricular cells exposed to 3 μm ouabain (OUA) in control (A and C) and Na⁺-free solution (B and D).

Cells were depolarized to +10 mV (from a holding potential of −70 mV) at 0.5 Hz. A, shows the expected increase in [Ca²⁺]_i in a representative cell in the presence of external Na⁺. B, cells were incubated previously in a Na⁺-free solution until [Na⁺]_i had been depleted and then exposed to OUA in a Na⁺-free, 1 mm Ca²⁺ solution. [Ca²⁺]_i transients were not affected by exposure to OUA in a Na⁺-free solution. However, rapid increase in external [Ca²⁺] to 5 mm (inset) or the β adrenergic agonist isoprenaline (1 μm, not shown) could still substantially increase the amplitude of the [Ca²⁺]_i transient in Na⁺-free conditions. C and D, pooled data for the normalized Ca²⁺ transients in NT and Na⁺-free solutions, respectively (numbers in parentheses indicate numbers of cells studied from two and three animals for control and Na⁺-free experiments, respectively). There was no significant increase in the amplitude of the Ca²⁺ transient in Na⁺-free solutions, whereas in NT it significantly increased (*P < 0.05 compared with control).

Myocytes were depleted of cytosolic Na⁺ by incubation in a Na⁺-free (replaced by NMDG) and Ca²⁺-free solution for 15 min at 23°C prior to the voltage-clamp study, and then transferred to the microscope stage and perfused with Na⁺-free, 1 mm Ca²⁺ solution at 37°C. Figure 4B shows no increase in [Ca²⁺]_i upon exposure to OUA in Na⁺-free solutions (in this case for 50 ms pulses). As a positive control for remaining inotropic capacity, in some cells after 5 min of exposure to OUA in Na⁺-free solution, external [Ca²⁺] was rapidly increased to 5 mm. The Ca²⁺ transients showed the expected increase in amplitude (Fig. 4B, inset). Other cells were exposed to the β-adrenergic agonist isoprenaline/isoproterenol (1 μm), which also evoked a substantial increase in the Ca²⁺ transient (not shown). Thus, as above there is still capacity for substantial inotropy. However, these two experimental manoeuvres (which dramatically increase Ca²⁺ influx) caused Ca²⁺ overload to occur rather rapidly in cells in which NCX was blocked, making these tests usually the experimental endpoint. Pooled data for the normalized Ca²⁺ transients in NT and Na⁺-free solutions are shown in Fig. 4C and D. Ca²⁺ transient amplitude was not significantly altered in the Na⁺-free solutions (n = 6 cells from three cats), whereas in NT it increased by 48 ± 4% (n = 4, P < 0.05).

Ca²⁺ sparks in permeabilized cells

Ca²⁺ sparks are fundamental SR Ca²⁺ release events underlying excitation–contraction coupling but they also occur spontaneously at low frequency in quiescent cells (Cheng et al. 1993). Nishio et al. (2004) reported that OUA caused a decrease in Ca²⁺ spark frequency in cat ventricular myocyte (intact or permeabilized). This is opposite to the effect expected based on their voltage-clamp data and some single-channel data, so we also re-examined whether OUA could directly affect RyR gating properties in Na⁺-free solutions in situ. We measured resting Ca²⁺ spark frequency (CaSpF) in cat ventricular myocytes permeabilized with SLO before and after exposure to OUA. Under these conditions, Na⁺ is absent, cytosolic [Ca²⁺] is constant (50 nm) and hydrophilic cardioactive glycosides, such as OUA, can freely diffuse into the cytosol, allowing us to test relatively high concentrations of this glycoside. Upon SLO permeabilization, spark frequency and properties remain relatively constant (Li et al. 2002) and exposure to 3 μm OUA had no effect on CaSpF. Figure 5A shows confocal images from a representative cell before and 5 min after continuous exposure to OUA. Figure 5B illustrates the time course of the mean values of CaSpF from nine different cells before and after OUA exposure. In addition, Fig. 5A shows that there was no significant increase in background fluo-3 fluorescence upon exposure to OUA (similar results were obtained in nine cells). This ensures that the free [Ca²⁺]_i remained constant at the control level. It is noteworthy that in similar experiments Nishio et al. (2004) reported a sustained increase in [Ca²⁺] in the presence of OUA.

Figure 5. Effect of ouabain (OUA, 3 μm) on resting [Ca²⁺]_i sparks in streptolysin-O-permeabilized cat ventricular myocytes.

A, spontaneous sparks measured with line-scan confocal microscopy in a Na⁺-free, 50 nm Ca²⁺ solution before and 5 min after exposure to OUA. B, shows the time course of the pooled data from nine cells from three animals for the resting [Ca²⁺]_i spark frequency (CaSpF) before and during exposure to OUA. C, illustrates typical spontaneous [Ca²⁺]_i sparks before and during OUA exposure. Both [Ca²⁺]_i sparks were measured in the same cell and at the same site in order to avoid out of focus problems and cell–cell variability in [Ca²⁺]_i sparks. There was a significant increase in spark duration (FDHM, P < 0.05). However, neither Ca²⁺ spark amplitude nor spatial spread (FWHM) was significantly affected (see Table 1).

Table 1 shows that there was no change in either Ca²⁺ spark amplitude or spatial spread (FWHM), as seen for CaSpF. However, there was a slight (19%) prolongation of Ca²⁺ spark duration (FDHM, P < 0.05). These Ca²⁺ spark characteristics can be appreciated in the examples in Fig. 5C. These two Ca²⁺ sparks were recorded in the same cell and site, before and after OUA exposure, minimizing site variability and out of focus fluorescence problems.

Table 1.

[Ca²⁺]_i spark properties in streptolysin-O (SLO)- permeabilized cat ventricular myocytes in the presence and absence of ouabain (OUA 3 μm) in Na⁺-free solutions

	Control (n = 9 cells, 3 cats)	OUA (n = 9 cells, 3 cats)
Frequency (CaSpF)	91 ± 8	89 ± 7
(Ca2+ sparks pl−1 s−1)
Amplitude	117 ± 1	112 ± 1
([Ca2+]i; nm)
Spatial spread	1.79 ± 0.02	1.80 ± 0.02
FWHM (μm)
Duration	30.2 ± 0.6	36.0 ± 0.9*
FDHM (ms)

^*P < 0.05 versus control group.

Discussion

A main mechanism of cardioactive glycoside action has long been known. It is well accepted that they selectively bind to and inhibit the Na⁺–K⁺-ATPase, and in this traditional view, NCX has a pivotal role (Bers, 2001). Na⁺–K⁺-ATPase inhibition leads to Na⁺_i accumulation and impairs Ca²⁺ extrusion by NCX (and if [Na⁺]_i gets high enough, NCX can cause net Ca²⁺ influx; Satoh et al. 2000). This results in an increased diastolic [Ca²⁺]_i and higher SR Ca²⁺ content, causing greater SR Ca²⁺ release and stronger contraction. Bulk cytosolic [Na⁺] may not have to increase during Na⁺–K⁺-ATPase inhibition. Locally elevated [Na⁺]_i between the sarcolemma and adjacent SR may suffice for enhanced contraction in arterial smooth muscle (Arnon et al. 2000) and cardiac myocytes (Su et al. 1998; Terracciano, 2001). Reuter et al. (2002) showed that cardioactive glycosides have no inotropic effect in heart tubes from embryonic mice in which the NCX gene has been knocked out. This is consistent with our results that Na⁺ is essential for acute glycoside-induced inotropy in adult ventricular myocytes. Together these results demonstrate the essential role of [Na⁺]_i and NCX in glycoside-induced inotropy.

Alternative inotropic pathways of cardiac glycosides in Na⁺-free solutions

This traditional view has been challenged recently by the suggestion of alternative inotropic mechanisms, not involving altered [Na⁺]_i or NCX function. These possible mechanisms include glycosides: (a) directly binding and activating RyRs; (b) increasing Ca²⁺ selectivity and Ca²⁺ influx via TTX-sensitive Na⁺ channels; and (c) other signalling mechanisms. We tested whether glycosides induce positive inotropy in ferret and cat ventricular myocytes in Na⁺-free conditions (where all of these alternate mechanisms could still have occurred). However, our data confirmed that Na⁺ gradients and NCX are essential for acute glycoside-induced inotropy.

Cell shortening and Ca²⁺ transients in Na⁺-free solution

After [Na⁺]_i depletion, field-stimulated ferret cells showed no positive inotropic effect in response to ACS and DIG (Figs 1–3), despite clear inotropy when NCX was functional. In these cells, steady-state contractions and Ca²⁺ transients were achieved within 20 min of perfusion with Na⁺-free solution and remained constant for up to 40 min of continuous perfusion (Fig. 1) without spontaneous Ca²⁺ waves or twitches. This indicates that even if SR Ca²⁺ content was higher at the new Na⁺-free steady state, the myocytes could still regulate Ca²⁺ levels in the absence of NCX and exogenous Ca²⁺ buffers. In addition, steady-state twitch amplitude in the Na⁺-free solution was significantly smaller (P < 0.05) than that immediately after readmitting external Ca²⁺ (Fig. 3), indicating that the cells could still respond to inotropic agents.

These results contrast with a report (Nishio et al. 2002) in which ouabain induced positive inotropy in Na⁺-free solution in cat ventricular myocytes. To rule out species-dependent differences, we also performed experiments in cat ventricular myocytes under voltage-clamp conditions at 37°C (Fig. 4) and still found no positive inotropy in Na⁺-depleted cells. We do not know why the results of Nishio et al. (2002) differ from ours. However, they did not attain a true steady state prior to OUA addition (contractions were already increasing). This might have been due to progressive Ca²⁺ loading during NCX inhibition (possibly independent of OUA effects), because the protocol did not allow sarcolemmal Ca²⁺-ATPase to keep up with Ca²⁺ influx (or possibly [Na⁺]_i depletion was incomplete).

Our inability to detect OUA-induced inotropy in Na⁺-free solutions cannot be due to an already maximal SR Ca²⁺ content or inotropic state, because further inotropic capacity was readily demonstrable (both in field-stimulated and voltage-clamped myocytes). We cannot be sure why Nishio et al. (2002) observed OUA-induced inotropic effects in Na⁺-free solutions, but we saw no hint of this under conditions where we were especially careful to attain a true steady state in entirely Na⁺-free conditions prior to application of OAU (or DIG).

RyR as an intracellular glycoside target

Several glycosides have been reported to increase open probability (P_open) of single RyR channels in lipid bilayers (Wasserstrom et al. 1991; McGarry & Williams, 1993; Sagawa et al. 2002). We observed a very small, but significant prolongation of Ca²⁺ spark duration with OUA (19%), which could be consistent with these bilayer results and results of Nishio et al. (2004). However, whether this small increase in Ca²⁺ spark duration is functionally important is not clear (as noted by Nishio et al. 2004).

We saw no effect of OUA on Ca²⁺ spark frequency, amplitude or spatial spread nor on resting [Ca²⁺] in permeabilized cat ventricular cells in Na⁺-free solution (where NCX function is eliminated, and hydrophilic glycosides can freely diffuse into the cytosol). This does not agree with the inhibition of Ca²⁺ spark frequency and spontaneous Ca²⁺ waves, but enhanced resting [Ca²⁺] reported by Nishio et al. (2004) in intact and permeabilized cat ventricular myocytes. We cannot explain the difference, but it is hard to understand why they saw a dramatic rise in tonic myoplasmic [Ca²⁺] with OUA, despite strong exogenous Ca²⁺ buffering in permeabilized myocytes (even during sustained caffeine exposure). Their reduced CaSpF is also hard to reconcile with previous reports of increased RyR P_open in bilayers by the same group (Sagawa et al. 2002). In our hands, 3 μm OUA neither increased nor decreased CaSpF, nor did it modify the steady-state [Ca²⁺] in permeabilized cells.

Nishio et al. (2004) also found that a higher OUA concentration was required in rat ventricular myocytes to produce Ca²⁺ spark inhibition and tonic [Ca²⁺]_i elevation, consistent with the well-known low OUA affinity of the Na⁺–K⁺-ATPase in rat (Langer et al. 1975). The simplest interpretation is that the species-dependent differences seen by Nishio et al. (2004) are because the primary effect is on the Na⁺–K⁺-ATPase (a well-defined OUA receptor with species-dependent affinity).

OUA is also used as a membrane impermeant probe for Na⁺–K⁺-ATPase in studies of membrane sidedness (e.g. Jones et al. 1980; Bers et al. 1980, 1986; Caroni & Carafoli, 1983). Given that OUA is highly hydrophilic (with a negligible oil–water partition coefficient), extracellular OUA is not expected to have very rapid effects on intracellular targets, so that the rapid OUA effects reported by Nishio et al. (2002, 2004) might be mediated by a cell surface receptor (such as Na⁺–K⁺-ATPase). Lipophilic cardioactive steroids such as DIG or ACS can more readily diffuse into intact cells, so this limitation would not apply to these agents.

RyR sensitization as an inotropic mechanism

It has been demonstrated experimentally and theoretically that simple RyR sensitization (or desensitization) to Ca²⁺, which could increase P_open in bilayer studies, can only cause transient inotropic effects in myocytes (Eisner & Trafford, 2000; Lukyanenko et al. 2001; Shannon et al. 2005). Acute RyR sensitization causes an initially larger SR Ca²⁺ release, but more Ca²⁺ is extruded from the cell (via NCX) and the SR Ca²⁺ content decreases. That reduces the amount of SR Ca²⁺ release at the next beat (due to both reduced Ca²⁺ availability for release and the effects of luminal Ca²⁺ on RyR gating). Thus even if OUA sensitized or inhibited RyR opening, sustained steady-state alteration in inotropy is not expected.

Slip-mode conductance of the Na⁺ channel

Our experiments also do not support slip-mode conductance of TTX-sensitive Na⁺ channels (Santana et al. 1998; Cruz et al. 1999). The slip-mode hypothesis, by which glycosides (and β-adrenergic agonists) alter Na⁺ channel selectivity, allowing Ca²⁺ influx (with a permeability ratio of Ca²⁺ to Na⁺ (P_Ca/P_Na) of ∼1) has been challenged. Nuss et al. (1999) found no evidence for altered P_Ca/P_Na in heterologously expressed cardiac Na⁺ channel subunits, although Cruz et al. (1999) in similar experiments found effects which were similar to their results in cardiac myocytes (Santana et al. 1998). A large change in Na⁺ channel selectivity induced by cAMP or OUA has also eluded detection by other groups using optical measurement of Ca²⁺ influx (DelPrincipe et al. 2000) or measuring P_Ca/P_Na (0.017) directly for cardiac I_Na with/without β-adrenergic stimulation (Chandra et al. 1999). Piacentino et al. (2000) argued that β-adrenergic slip-mode conductance could be explained by spurious activation of L-type Ca²⁺ channels, as it could be abolished by nifedipine and Cd²⁺ but not by TTX.

Slow parallel signalling pathways

Glycoside binding to Na⁺–K⁺-ATPase could also trigger the activation of additional signalling pathways which lead to protein tyrosine phosphorylation, cytosolic Ca²⁺ increase and gene regulation (Tian et al. 2001; Xie & Askari, 2002). However, our data lead us to conclude that their contribution to the acute positive inotropic effect is very small compared with the well-known Na⁺–K⁺-ATPase–NCX pathway. We did not find any positive inotropic effect in any of the different protocols applied, where those pathways could still be active. Nevertheless, we cannot rule out longer term inotropic effects due to slower secondary regulation of gene expression.

In conclusion, our data provide evidence for the importance of a Na⁺ gradient and functional NCX in the acute glycoside-induced inotropy in cardiac myocytes.