Due to the lack of suitable substitutes, Na⁺/K⁺ pump inhibitors remain a valuable therapeutic tool in the management of heart failure (Eichhorn and Gheorghiade, 2002); nonetheless, their use is strongly limited by their well known proarrhythmic effect. Inhibition of Na⁺/K⁺-ATPase and the resulting change in intracellular Ca²⁺ are commonly held to underlie both inotropy and proarrhythmia (Bers, 2002b), thus justifying the claim that these actions may be inseparable. This view is challenged by the observation that the ratio between inotropic and toxic effect (therapeutic index) varies among different Na⁺/K⁺ pump inhibitors (Wasserstrom et al., 1991), possibly due to modulation of targets other than the Na⁺/K⁺ pump. For instance, commonly used glycosides also affect sarcoplasmic reticulum (SR) function (McGarry and Williams, 1993; Sarkozi et al., 1996; Sagawa et al., 2002) with a potential impact on the relation between total intracellular Ca²⁺ content and its effects on cell function.

A sharp dissociation between inotropy and proarrhythmia has been recently observed with the novel nonglycoside Na⁺/K⁺ pump inhibitor (E,Z)-3-((2-aminoethoxy)imino)androstane-6,17-dione hydrochloride (PST2744) (Micheletti et al., 2002). In single-cell, tissue, and whole animal studies, the incidence of arrhythmias observed at similar inotropy (increase in unloaded cell shortening in single myocytes) was markedly lower for PST2744 than for digoxin (Micheletti et al., 2002; Rocchetti et al., 2003). Cellular electrophysiology studies showed that the changes in action potential contour and relevant membrane currents caused by equi-inotropic PST2744 and digoxin concentrations were roughly similar (Rocchetti et al., 2003). The only effect potentially accounting for the different toxicity was the inhibition of the Na⁺/Ca²⁺ exchanger current by PST2744 through a mechanism independent of Na⁺/K⁺ pump blockade (Rocchetti et al., 2003). Nonetheless, those studies could not discriminate whether such an effect resulted from PST2744 action on the exchanger itself or was secondary to changes in the dynamics of intracellular Ca²⁺.

The present work compares digoxin and PST2744 effects on SR function in isolated vesicle preparations and on Ca²⁺ dynamics in intact myocytes. The results obtained provide a basis for the differences between the two agents and a link, of more general interest, between modulation of specific SR targets and overall Ca²⁺ handling by ventricular myocytes.

Materials and Methods

The investigation conforms to the Guide of the Care and Use of Laboratory Animals published by the National Institutes of Health (NIH publication 85-23, revised 1996) and to the guidelines for animal care endorsed by the participating institutions. Hartley guinea pigs (3–4 months old) were killed by cervical dislocation under ether (vesicle preparations) or ketamine-xylazine (myocyte preparations) anesthesia.

Studies on Isolated SR Vesicles

Vesicle Preparation. Cardiac SR vesicles were isolated according to Sitsapesan et al. (1991). Microsomes were prepared at 2°C by homogenization in a medium containing 300 mM sucrose and 20 mM imidazole (pH 7.4). Cell fragments were removed by centrifugation (6800g for 25 min) followed by filtration. The following protease inhibitors were included in all solutions: 200 μM pefabloc, 1 mM benzamidine, 100 nM aprotinin, 1 μM leupeptin, and 1 μM pepstatin A.

Crude microsomes were centrifuged at 120,000g for 90 min; the pellet was resuspended in a medium containing 100 mM NaCl, 10 mM K⁺-PIPES, and 500 mM sucrose (pH 7.0). The suspension was layered onto a 20–45% linear sucrose gradient, spun for 16 h at 90,000g, and the lower visible protein ring was collected. Aliquots from parallel tubes were pooled and homogenized in a medium containing 400 mM KCl, 20 mM imidazole, 0.5 mM MgCl₂, 0.5 mM CaCl₂, and 0.5 mM EGTA. Heavy SR vesicles were then collected by centrifugation at 124,000g for 60 min, resuspended in a medium containing 300 mM sucrose and 25 mM MOPS (pH 7.0), aliquoted, and stored at –70°C until use. Protein concentration was determined by Lowry (1951) assay using bovine serum albumin as standard, and molecular purity was checked using SDS-polyacrylamide gel electrophoresis.

Measurement of SERCA Activity. ATPase activity was determined at 37°C by a coupled enzyme assay in a medium containing 100 mM KCl, 20 mM Tris-HCl, 5 mM MgCl₂, 5 mM ATP, 0.42 mM phosphoenolpyruvate, 0.001 mM calcimycin (a Ca²⁺ ionophore), 0.2 mM NADH, 7.5 U/ml pyruvate kinase, and 18 U/ml lactate dehydrogenase (pH 7.5). Total hydrolytic activity was measured as the decrease of optical density at the NADH absorbance wavelength (340 nm) and is expressed in micromoles of inorganic phosphate per milligrams of protein per minute, abbreviated as A.U. throughout the text and figures. Total hydrolytic activity ranged from 2.53 to 2.76 A.U. SERCA-dependent ATPase activity was identified as the portion of total hydrolytic activity inhibited by thapsigargin (5 μM) (Lytton et al., 1991).

PST2744 or digoxin was added to the preparation 5 min before starting measurements of enzyme activity. The assay was performed in the presence of Ca²⁺ and EGTA mixtures producing estimated free Ca²⁺ concentrations ranging from 0.1 to 2.9 μM (Fabiato and Fabiato, 1979); the latter concentration was found to induce maximum SERCA activation.

Calcium Efflux from SR. Calcium efflux from actively loaded vesicles was determined at 37°C by monitoring time-dependent changes in free Ca²⁺ concentration of the extravesicular solution. The latter was determined by sampling (1 Hz) antipyrylazo-III (APIII) absorbance difference at wavelengths of 710 and 790 nM (A₇₁₀–A₇₉₀). The APIII signal was calibrated by successive additions of CaCl₂ to vesicle samples in which Ca²⁺ uptake was prevented by 100 nM cyclopiazonic acid (Sarkozi et al., 1996) (see Fig. 1 in Supplemental Data). Vesicles were suspended in the cuvette up to a final protein concentration of 260 μg/ml (final volume 2 ml) in a medium containing 92.5 mM KCl, 1 mM MgCl₂, 7.5 mM sodium pyrophosphate, 18.5 mM K⁺-MOPS, 1 mM ATP, and 0.18 mM APIII. Active loading of vesicles by CaCl₂ addition (0.054 mM) resulted in an average intravesicular Ca²⁺ content of 240 ± 22 nmol per assay. After completion of the ATP to ADP conversion and stabilization for 1 min, Ca²⁺ release was initiated by addition of the compound under study (Szentesi et al., 2001); a known amount of Ca²⁺ was added at the end of each experiment to confirm signal calibration. Background Ca²⁺ efflux, i.e., that occurred over the same time period in the absence of drugs (see Fig. 2 in Supplemental Data) was evaluated for each measurement on a separate sample of vesicles from the same preparation. Data are expressed as the amount of Ca²⁺ released during 20 min of exposure to experimental solutions.

Studies on Intact Myocytes

Myocyte Preparation and Recording Solutions. Guinea pig ventricular myocytes were isolated by using a retrograde coronary perfusion method previously published (Zaza et al., 1998) with minor modifications. Rod-shaped, Ca²⁺-tolerant myocytes were used within 12 h from dissociation.

During measurements, myocytes were superfused at 2 ml/min with Tyrode's solution containing 154 mM NaCl, 4 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 5 mM HEPES/NaOH, and 5.5 mM d-glucose, adjusted to pH 7.35. A thermostated manifold, allowing for fast (electronically timed) solution switch, was used for cell superfusion. All measurements were performed at 35 ± 0.5°C.

Electrophysiology Techniques. Ventricular myocytes were voltage-clamped in the whole-cell configuration (Axopatch 200-A, Axon Instruments Inc., Union City, CA). The pipette solution contained 110 mM K⁺-aspartate, 23 mM KCl, 0.2 mM CaCl₂ (calculated free-Ca²⁺ = 10^–7 M), 3 mM MgCl₂, 5 mM HEPES-KOH, 0.5 mM EGTA-KOH, 0.4 mM GTP-Na⁺ salt, 5 mM ATP-Na⁺ salt, and 5 mM creatine phosphate Na⁺ salt, pH 7.3. Pipette solution was modified in specific cases, as detailed in the section on Experimental Protocols (below). Membrane capacitance and series resistance were measured in every cell but left uncompensated. Current signals were filtered at 2 KHz and digitized at 5 KHz (Axon Digidata 1200). Trace acquisition and analysis was controlled by dedicated software (Axon pClamp 8.0). As preliminarily tested under the conditions used in the present experiments, peak inward current closely reflects I_CaL. K⁺ replacement with Cs⁺, commonly used to isolate I_CaL, was avoided because SR function may be affected by this ion (Kawai et al., 1998).

Measurements of Intracellular Ca²⁺and Cell Shortening. Single myocyte intracellular Ca²⁺ activity was measured fluorimetrically using the membrane permeable dye Indo1-AM (Molecular Probes, Leiden, The Netherlands). After 30 min of incubation in Indo1-AM (8 μM), myocytes were washed for 30 min at 36°C. Indo1-AM fluorescent emission was measured at two wavelengths (410 and 490 nm) (Grynkiewicz et al., 1985). The signals at the two wavelengths (F₄₁₀ and F₄₉₀) were separately low-pass filtered (200 Hz) and digitized at 2 KHz. Cytosolic Ca²⁺ activity was calculated from the F₄₁₀/F₄₉₀ ratio after low-pass digital filtering (fast Fourier transform, 100 Hz) and subtraction of background luminescence. F₄₁₀/F₄₉₀ ratio values were converted to free cytosolic Ca²⁺ concentration ([Ca]_f) by the formula described in Table 1 using parameters determined by calibration in ionomycin permeabilized myocytes (Sipido and Callewaert, 1995).

Total cytosolic Ca²⁺ concentration ([Ca]_tot) and content (Ca_tot), the amounts of Ca²⁺ moved across the sarcolemma through I_CaL (Ca_L) or the Na⁺/Ca²⁺ exchanger (Ca_NCX), and the gain of Ca²⁺-induced Ca²⁺ release (CICR) were estimated as described in Table 1. In SR reloading experiments, a qualitative estimate of CICR gain was obtained simply by dividing the peak value of the Ca²⁺ transient by peak I_CaL (Bers, 2002c); although expressed in different units, the results of this procedure were highly correlated with those obtained through the quantitative one.

Total SR Ca²⁺ content normalized to cell volume ([Ca]_SR-tot, micromoles per liter of cytosol) was measured during caffeine pulses (protocol 1 below). [Ca]_SR-tot estimate can be biased by removal of Ca²⁺ by the Na⁺/Ca²⁺ exchanger, which may be differently affected by the agents under the study (Rocchetti et al., 2003). To obtain an unbiased estimate of [Ca]_SR-tot, Ca_NCX during the ascending phase of the caffeine-induced Ca²⁺ transient was added back to Ca_tot measured at its peak (Table 1, Fig. 2, A and B). In control conditions, Ca_NCX during the ascending phase of the transient was 12 ± 1% of total SR Ca²⁺.

Intracellular caffeine concentration was monitored by measuring caffeine-induced quenching of Indo1-AM fluorescence according to the procedure described by O'Neill et al. (1990). Cell shortening was measured in intact myocytes and field-stimulated at a rate of 2 Hz by a fast acquisition camera (sampling rate 250 Hz) and a video-edge detector (Crescent Electronics, Sandy, UT).

Experimental Protocols.1. Ca²⁺dynamics and SR Ca²⁺content under physiological conditions. Transmembrane current and cytosolic Ca²⁺ were simultaneously measured during a train of depolarizing (200 ms) pulses to 0 mV (0.37 Hz) applied from a holding potential (E_H) of –40 mV. E_H was then changed to –80 mV, and a 5-s caffeine pulse (10 mM) was applied after 10 s (electronically timed).

2. SR reloading during Na⁺/Ca²⁺exchanger inhibition. The Na⁺/Ca²⁺ exchanger was inhibited by incubation of the cells for 30 min in aNa⁺- and Ca²⁺-free solution (replaced by equimolar Li⁺ and 1 mM EGTA). During measurements, the perfusate contained 0 mM Na⁺ and 1 mM Ca²⁺. The pipette solution was Na⁺-free (Na⁺ salts were replaced by K⁺ or Tris salts). SR was initially depleted by a brief caffeine pulse (with 154 mM Na⁺ to allow Ca²⁺ extrusion through the Na⁺/Ca²⁺ exchanger) and then progressively reloaded by a train of depolarizing pulses (0.25 Hz, pulse parameters as in protocol 1) in the presence of 1 mM Ca²⁺ (protocol outline in the inset of Fig. 5).

Substances

Stock Indo1-AM solution (1 mM in dry DMSO) was diluted in Tyrode's solution. PST2744 was dissolved in water; digoxin was dissolved in DMSO or ethanol. The final DMSO or ethanol concentrations did not exceed 0.1%. PST2744 was synthesized at Prassis Sigma-Tau (Settimo Milanese, Italy), antipyrylazo-III was purchased from MP Biomedicals (Irvine, CA), Indo1-AM from Molecular Probes, protease inhibitors from Roche Diagnostics (Mannheim, Germany) and Merck (Darmstadt, Germany), and all other chemicals from Sigma-Aldrich (St. Louis, MO).

Statistical Analysis

Least-square fitting was used to estimate the kinetics of twitch and Ca²⁺ decays. Means were compared by paired or unpaired t test and, wherever appropriate, by two-way analysis of variance (paired measurements). Data are expressed and plotted as the mean ± S.E. value obtained from a number of independent determinations. Statistical significance was defined as p < 0.05 (N.S. = not significant). Absolute values of the variables and sample size for each experimental condition are provided in the figures and the respective legends.

Results

Studies on Isolated SR Vesicles

Effects on SERCA Enzymatic Activity. Dependence of SERCA activity on extravesicular Ca²⁺ (0.1–2.3 μM) was tested in the absence (control) and in the presence of PST2744 and digoxin (Fig. 1A).

Under control conditions, increments in Ca²⁺ enhanced SERCA activity with an asymptote (V_max) at 2.5 ± 0.12 A.U. and K_m of 209.2 ± 21.9 nM. PST2744 (100 nM) significantly reduced K_m to 143.9 ± 6.3 nM (p < 0.05) and left V_max unchanged (2.74 ± 0.05; N.S.). Conversely, 100 nM digoxin left V_max (2.3 ± 0.11 A.U.) and K_m (193.0 ± 17.1 nM) unchanged.

Concentration-response curves for the two agents were then obtained at submaximal (500 nM) and saturating (2300 nM) Ca²⁺ concentrations (Fig. 1B). At submaximal Ca²⁺, PST2744 exerted clear-cut stimulatory effects over the range 30 to 300 nM. Its concentration-response curve was bell-shaped, with maximal effect occurring at 100 nM, and decreasing thereafter. When Ca²⁺ was increased to the saturating level, basal activity increased and the PST2744 stimulatory effect, even if still significant, was largely hindered. At variance with PST2744, digoxin failed to stimulate SERCA activity at both Ca²⁺ levels.

Effects on Spontaneous Ca²⁺Efflux from the SR. The effect of low and high drug concentrations on spontaneous Ca²⁺ efflux from SR vesicles is illustrated in the sample tracings of Fig. 1C. Concentration-response curves for PST2744 and digoxin were constructed by measuring the amount of Ca²⁺ released by the vesicles 20 min after compound addition. Figure 1D compares curves obtained with the two agents to the confidence limits for drug-independent Ca²⁺ efflux. Digoxin enhanced Ca²⁺ efflux in a concentration-dependent way with a threshold concentration at 0.1 nM and peak effect at 1 μM (40.5 ± 0.3 nmol at 20 min, exceeding the upper confidence limit of background by approximately eight times). One hundred nanomolars of PST2744, i.e., the most effective concentration on SERCA, failed to enhance Ca²⁺ release; an effect appeared only at concentrations ≥300 nM and even at 10 μM, it amounted to 12.9 ± 4 nmol only (p < 0.05 versus digoxin 1 μM).

Studies on Intact Myocytes

With the aim of obtaining a functionally relevant comparison, PST2744 and digoxin were compared at equi-inotropic concentrations, as previously assessed under the same experimental setting (Micheletti et al., 2002). The concentrations used throughout the present study were 4 μM for PST2744 and 1 μM for digoxin, corresponding to the steep portion of the relation between concentration and unloaded cell shortening of isolated myocytes (Micheletti et al., 2002).

Effects on Resting Ca²⁺, CICR Gain, and Total SR Ca²⁺Content. Transmembrane current and cytosolic Ca²⁺ concentration were simultaneously measured under physiological ionic conditions according to experimental protocol 1 (see Materials and Methods).

Resting Ca²⁺ (at E_H =–80 mV) was similarly increased by digoxin (from 46.5 ± 5.7 to 70.4 ± 6.1 nM; +60 ± 14%, n = 8, p < 0.05) and PST2744 (from 61.3 ± 6.4 to 92.4 ± 11.1 nM, +51 ± 12%, n = 8, p < 0.05).

Total SR Ca²⁺ content was estimated from the amplitude of the caffeine-induced transient after correction for the Ca²⁺ extruded by the Na⁺/Ca²⁺ exchanger (Fig. 2; see Materials and Methods) and normalized to cytosolic volume ([Ca]_SR-tot, moles per liter of cytosol) (Shannon et al., 2000b). Digoxin increased [Ca]_SR-tot by 28 ± 10% (p < 0.05, Fig. 2, B and C) and PST2744 by 47 ± 8% (p < 0.05, Fig. 2, A and C). The effect of PST2744 on [Ca]_SR-tot was significantly larger than that of digoxin (p < 0.05, Fig. 2C). Both agents increased the amount of Ca²⁺ extruded by the Na⁺/Ca²⁺ exchanger (Ca_NCX) during caffeine-induced transients (digoxin +127 ± 38%, PST2744 +130 ± 46%, Fig. 2D).

As shown in Fig. 2, A and B, both digoxin and PST2744 shortened the time elapsing between the start of the (electronically timed) caffeine pulse and SR Ca²⁺ release. This was not associated with a matching effect on the time course of intracellular caffeine concentration (see Fig. 3 in Supplemental Data). Thus, anticipation of the Ca²⁺ transient may reflect drug-induced facilitation of caffeine action on RyR channels, probably a consequence of increased [Ca]_SR-tot.

The gain of CICR was estimated during the last precaffeine voltage pulse as the ratio between SR-released Ca²⁺ and transmembrane Ca²⁺ influx. As previously shown (Shannon et al., 2000a), CICR gain was directly related to [Ca]_SR-tot (r ≥ 0.72; p < 0.05). Digoxin and PST2744 increased CICR gain to a similar extent (+157 ± 39% versus +139 ± 24%; N.S.) (Fig. 3).

Effects on the Rate of Clearance of Cytosolic Ca²⁺. The rate of Ca²⁺ clearance from the cytosol was measured by the exponential fitting of Ca²⁺ decays following voltage- and caffeine-induced transients, respectively (Fig. 4). The former reflects operation of all Ca²⁺ handling mechanisms (mainly SERCA and the Na⁺/Ca²⁺ exchanger); the latter is largely accounted for by the Na⁺/Ca²⁺ exchanger operation (Bers, 2002d).

As expected, under control conditions Ca²⁺ decay time constant (τ_decay) was smaller during voltage- than during caffeine-induced transients (851 ± 104 versus 1415 ± 280 ms; p < 0.05). Digoxin similarly reduced τ_decay during voltage-(–30.1 ± 8.4%, p < 0.05) and caffeine-induced transients (–36.5 ± 7.4%, p < 0.05) (Fig. 4, B and D). PST2744 reduced τ_decay in voltage-induced transients (–41.3 ± 7.0%, p < 0.05) significantly more than in caffeine-induced transients (–20.0 ± 3.9%, p < 0.05) (Fig. 4, A and C). For both compounds, higher resting Ca²⁺ levels were associated with a faster decay of the caffeine-induced transient (r ≥ 0.74, p < 0.05).

Effects on SR Reloading during Na⁺/Ca²⁺Exchanger Inhibition. To evaluate drug effects on SR function in isolation, the Na⁺/Ca²⁺ exchanger was blocked by removing Na⁺ from both sides of the sarcolemma (see protocol 2 under Materials and Methods). After the SR was depleted, all the parameters of Ca²⁺ transients progressively recovered during subsequent depolarizations. A progressive accumulation of resting Ca²⁺ occurred due to the absence of Na⁺/Ca²⁺ exchanger-mediated Ca²⁺ extrusion (Fig. 5A). During SR reloading, PST2744 markedly accelerated the recovery of Ca²⁺ transient amplitude (Fig. 5B), slightly increased decay amplitude (Fig. 5C), and sharply accelerated Ca²⁺ decay (smaller τ_decay values, Fig. 5D). Since I_CaL decreased significantly during both digoxin and PST2744 superfusion, data were also analyzed in terms of CICR gain (Fig. 5E) by using a simplified method (see Materials and Methods). Under control conditions, CICR gain started to increase only after several cycles of the loading train, thus suggesting that a “threshold” Ca²⁺ concentration within the SR is required for efficient Ca²⁺ release (Shannon et al., 2000a). PST2744 markedly increased CICR gain and eliminated the initial increase of τ_decay (Fig. 5D) and the delay in gain changes (Fig. 5E). Effects on SR reloading qualitatively similar to those of PST2744, but of larger magnitude, were observed with the β1-adrenergic agonist isoproterenol (0.1 μM, see Fig. 4 in Supplemental Data); at variance with PST2744, isoproterenol increased I_CaL.

Digoxin did not modify any of the Ca²⁺ transient parameters during early SR reloading cycles and tended to depress Ca²⁺ reuptake in the late ones (Fig. 6C). Thus, digoxin displayed a behavior distinctly different from PST2744 (compare Fig. 6 to Fig. 5).

Since SERCA function is up-regulated by high cytosolic Ca²⁺, the acceleration of Ca²⁺ decays induced by PST2744 might simply reflect drug-induced increase in cytosolic Ca²⁺. If so, PST2744 should not affect the relation between τ_decay and the Ca²⁺ concentration present at the onset of decay (initial Ca²⁺) (Bers and Berlin, 1995). Contrary to this hypothesis, the relation was modified by PST2744, which reduced τ_decay at each at each initial Ca²⁺ level (Fig. 7A). This effect prevailed at low initial Ca²⁺ levels (leftmost part of the relation in Fig. 7A) and was not shared by digoxin (Fig. 7B).

Effects on Twitch Decay Rate. Twitch decay rate was measured by fitting an exponential through the decay phase and defined by its time constant (τ_relax). τ_relax was decreased by digoxin (–9.4 ± 1.4%, n = 5, p < 0.05) and, to a larger extent, by PST2744 (–28.6 ± 6.8%, n = 5, p < 0.05 versus digoxin; Fig. 8). Notably, the present results are fully consistent with those in vivo showing that PST2744 increased maximum relaxation velocity substantially more than digoxin (Micheletti et al., 2002).

Discussion

Studies in Isolated Vesicles. The results obtained in isolated vesicles demonstrate that PST2744 and digoxin modulate SR function in opposite ways. PST2744, but not digoxin, stimulated the ATPase activity of SERCA. Conversely, digoxin enhanced spontaneous Ca²⁺ efflux from the SR, starting at concentrations 1000-fold lower than PST2744.

Cytosolic Ca²⁺ has been shown to affect SERCA activity both by being its rate-limiting substrate (Bers and Berlin, 1995; Shannon et al., 2000b) and by allosteric modulation through phospholamban (Kranias, 1985). PST2744 markedly increased SERCA Ca²⁺ affinity (lower K_m), thus suggesting that PST2744 may modulate SERCA mainly by affecting its interaction with cytosolic Ca²⁺. The finding that, in the presence of saturating Ca²⁺ concentrations, PST2744 effect was minimal is consistent with this hypothesis. In isolated SR vesicles, the number of targets potentially accounting for such an action is limited to SERCA itself and to those of its regulatory proteins, which remain associated with the SR membrane. Among them, phospholamban may be contained in variable amounts in SR vesicles prepared according to the present technique (R. Micheletti, unpublished data), thus constituting a potential target for PST2744 action.

Ca²⁺ efflux from SR vesicles was sharply enhanced by digoxin. Several cardiac glycosides, including digoxin, were previously shown to increase the open probability of RyR channels in cardiac (digoxin EC₅₀ = 0.20–0.65 nM) (McGarry and Williams, 1993; Sagawa et al., 2002) and skeletal myocytes (Sarkozi et al., 1996). Hence, digoxin effects reported here were not unexpected, and although other mechanisms cannot be ruled out, they may be accounted for by its action on RyR channels. On the other hand, at least at the concentrations acting on SERCA (see above), PST2744 did not affect spontaneous Ca²⁺ efflux significantly.

Studies in Intact Myocytes. Although SERCA stimulation may increase SR Ca²⁺ content, an opposite effect is expected from enhancement of Ca²⁺ leak. In myocytes, such qualitative differences between the two agents were observed when Na⁺/Ca²⁺ exchanger was blocked, thus precluding effects secondary to Na⁺/K⁺ pump inhibition and unveiling direct modulation of SR targets.

Under such conditions, PST2744 effects were all compatible with a sharp enhancement of Ca²⁺ uptake by the SR, as expected from stimulation of SERCA activity. Indeed, PST2744 increased the magnitude and rate of Ca²⁺ reuptake (Fig. 5, C and D), thus stimulating the SR reloading process (Fig. 5B). The marked increase in CICR gain caused by PST2744 (Fig. 5E) may be secondary to larger increments in SR Ca²⁺ content during the reloading protocol (Shannon et al., 2000a). Consistent with Ca²⁺ dependence of SERCA stimulation in SR vesicles, acceleration of cytosolic Ca²⁺ decay by PST2744 prevailed at low Ca²⁺ levels (Fig. 7A).

Similarity between PST2744 and isoproterenol effects might suggest a receptor- or protein kinase A-mediated mechanism (Tsien et al., 1986). However, PST2744 did not enhance I_CaL (Rocchetti et al., 2003), and binding studies showed that this agent is not a ligand of β-adrenergic receptors (Micheletti et al., 2002).

During Na⁺/Ca²⁺ exchanger blockade, digoxin effects were compatible with a slight inhibition of the SR reloading process. In particular, the amount of Ca²⁺ cleared at each cycle was reduced by this agent in the late cycles of the reloading protocol (Fig. 6C). This effect might reflect drug-induced enhancement of Ca²⁺ leak, which would become more significant as SR Ca²⁺ content increases (Sagawa et al., 2002). In the light of this, the tendency of digoxin to increase CICR gain (Fig. 6E) implies the presence of a compensatory mechanism. This could be represented by the facilitation of the RyR channel opening, which may favor CICR (McGarry and Williams, 1993). In other words, in the absence of Na⁺/Ca²⁺ exchanger, digoxin modulation of CICR gain might be the net result of decreased SR Ca²⁺ content and RyR channel sensitization to cytosolic Ca²⁺. However, the latter mechanism may lose its relevance in the presence of normal Na⁺/Ca²⁺ exchanger function (Trafford et al., 2000).

When drug effects on Ca²⁺ dynamics were evaluated in the presence of a normal Na⁺/Ca²⁺ exchanger activity, digoxin and PST2744 still differed, but in quantitative rather than in qualitative terms. PST2744 modulation of SERCA was manifested by 1) a larger increment in SR Ca²⁺ content (Fig. 2C), and 2) preferential acceleration of the Ca²⁺ decay following voltage-compared with caffeine-induced transients (Fig. 4C). On the other hand, the observation that digoxin and PST2744 effects were qualitatively similar (i.e., in the same direction) implies the contribution of a mechanism based on Na⁺/Ca²⁺ exchanger and shared by the two agents. Such requirements are satisfied by Na⁺/K⁺ pump blockade, which increases cytosolic Ca²⁺ through a change in the Na⁺/Ca²⁺ exchanger equilibrium potential. Indeed, an increment in cytosolic Ca²⁺ may, on its own, account for increased [Ca]_SR-tot (Bers and Berlin, 1995) and for the resulting increment in CICR gain (Shannon et al., 2000a). Higher Ca²⁺ may also explain acceleration of Ca²⁺ decay and twitch relaxation through allosteric Na⁺/Ca²⁺ exchanger activation (Weber et al., 2001). This view is reinforced by the observation that τ_decay of caffeine-induced transients was inversely related to resting Ca²⁺ levels.

A finding which might appear at odds with the observation that PST2744 and digoxin differently affect Ca²⁺ confinement in the SR is the similar increase in resting Ca²⁺ levels induced by the two agents. However, it should be stressed that diastolic Ca²⁺ levels closely depend on the Na⁺/Ca²⁺ exchanger equilibrium potential (Bers and Bridge, 1989) and, in turn, on Na⁺/K⁺ pump function, which was equally affected by the two drugs at the concentrations used (Rocchetti et al., 2003). Thus, a similar elevation of diastolic Ca²⁺ with the two agents was expected under physiological conditions.

In the presence of Na⁺/Ca²⁺ exchanger blockade, SR reloading was associated with a progressive increment in resting cytosolic Ca²⁺, which is likely to reflect the increase in free SR Ca²⁺ concentration (with the maximum Ca²⁺ gradient across SERCA remaining constant) (Shannon et al., 2000b). The observation that PST2744 enhanced resting Ca²⁺ accumulation (Fig. 5) is compatible with faster SR Ca²⁺ loading through SERCA stimulation. Indeed, recruitment of more SERCA molecules (e.g., by increasing their affinity for cytosolic Ca²⁺) may increase Ca²⁺ flow rate without necessarily affecting the maximum Ca²⁺ gradient (as batteries connected in parallel supply more current with unchanged voltage).

Conclusions and Practical Implications. Although under physiological conditions Na⁺/K⁺ pump inhibition contributed to both inotropy and lusitropy, concomitant SERCA stimulation was associated with quantitative differences in these effects and, most notably, with a widely different therapeutic index (Micheletti et al., 2002; Rocchetti et al., 2003). This suggests that SERCA stimulation may be a key mechanism in tempering the proarrhythmic effect of an increased Ca²⁺ load, such as that caused by many among inotropic interventions. In light of this, SERCA down-regulation in the failing myocardium (Beuckelmann et al., 1992; Movsesian et al., 1994; Mishra et al., 2002, 2003) might sensitize to the detrimental effect of inotropes. Whether PST2744 may be able to revert this condition is unknown and deserves further investigation.

Limitations of the Study. Although the Ca²⁺ handling mechanisms considered in this work are largely prevalent (Bers, 2002a), Ca²⁺ extrusion by the sarcolemmal Ca²⁺ ATPase and mitochondrial Ca²⁺ uptake may also contribute to determine cytosolic Ca²⁺ levels. Based on the present results, it is impossible to rule out interference with such mechanisms as a further cause for the differences between PST2744 and digoxin.

Cytosolic Ca²⁺ buffering parameters (B_max and K_d) and the cell surface/volume ratio used in the estimation of cytosolic Ca²⁺ levels were not available for guinea pig myocytes, and the values reported for rabbit ventricular myocytes have been used instead (Table 1). This might make absolute quantitative estimates imprecise, but it would not affect the comparison between drug effects.