Introduction

Top Abstract Introduction Materials & Methods Results Discussion References

Levosimendan is a pyridazinone-dinitrile derivative belonging to a new class of cardiac inotropic drugs, Ca⁺⁺ sensitizers. It binds to troponin C in a Ca⁺⁺-dependent manner and is reported to facilitate a Ca⁺⁺-induced conformational change of troponin C necessary for activation of the contractile apparatus in cardiac muscle (Pollesello et al., 1994; Haikala et al., 1995a). In contrast to other Ca⁺⁺-sensitizing drugs, levosimendan does not slow cardiac relaxation, presumably due to its detachment from troponin C at low Ca⁺⁺ (Haikala et al., 1995b; Haikala and Linden, 1995). In addition to its cardiac effects, levosimendan is also a vasodilator both in vitro (Rump et al., 1994; Gruhn et al., 1998) and in vivo, (Udvary et al., 1995; Pagel et al., 1996). The vasodilatation by levosimendan may partially be due to a lowering of intracellular free calcium ([Ca⁺⁺]_i) through potential inhibition of phosphodiesterase III (PDE III) (Haikala and Linden, 1995; Vegh et al., 1995) and/or through opening of ATP sensitive potassium channels (Yokoshiki et al., 1997). However, the mechanism of this vasodilator action is not yet known with certainty (Gruhn et al., 1998).

The cardiac target protein of levosimendan, troponin C, belongs to the family of the so called Ca⁺⁺-binding EF-hand proteins. Although troponin C is not found in smooth muscle, EF-hand proteins are expressed in vascular smooth muscle cells and may play a role in regulation of contraction (Persechini et al., 1989; Schafer and Heinzmann, 1996). This raises the possibility that levosimendan may induce vasodilatation not only by lowering [Ca⁺⁺]_i but also by interacting with Ca⁺⁺-binding EF-hand proteins. Targets for levosimendan could include, for example, calmodulin, the regulatory myosin light chains, and S100 proteins. The vasodilatory action of levosimendan through Ca⁺⁺-binding proteins would lead to Ca⁺⁺ desensitization, i.e., a decrease in force without a proportionate decrease in [Ca⁺⁺]_i. Therefore, one of the aims of the present study was to investigate the effects of levosimendan on [Ca⁺⁺]_i and to correlate these results with contractile force in porcine coronary arteries. Because intracellular pH (pH_i) can modulate vascular smooth muscle Ca⁺⁺ sensitivity (Wray, 1988; Nagesetty and Paul, 1994), we also studied the effects of levosimendan on pH_i. In addition, the potential role of the PDE III inhibition in the vasodilatory effect of levosimendan was indirectly evaluated with a PDE III inhibitor, milrinone, as a reference compound.

	Materials and Methods

Top Abstract Introduction Materials & Methods Results Discussion References

Preparation of Arterial Rings. Porcine hearts obtained shortly after slaughter were rinsed of blood and placed in a cold (4°C) bicarbonate-buffered, physiological salt solution (PSS). The distal portions of the left anterior descending coronary artery were dissected and placed in cold PSS. Arteries were then cleaned of fat and connective tissue and cut into 5-mm segments. The arterial wall thickness was between 300 and 500 µm. The segments were everted to expose the luminal side for fluorescence measurements and de-endothelized by rolling gently on filter paper.

Isometric Force Measurements. Arterial rings were suspended isometrically on two stainless steel posts, one of which was attached to a Kistler-Morse DSK force transducer (Kistler-Morse, Bellvue, WA). The rings were placed inside a water-jacketed chamber filled with 15 ml of PSS that was maintained at 37°C. The chambers were bubbled with 95% O₂/5% CO₂ to maintain a physiological pH of 7.4. Tension was recorded with a BioPac digital acquisition system (Goleta, CA) and an IBM compatible computer.

Intracellular Calcium Measurements. [Ca⁺⁺]_i was measured with the fluorescent probe fura-2 AM as previously described (Nagesetty and Paul, 1994). Briefly, arterial rings were everted, rolled on filter paper to remove the endothelium, and mounted isometrically on a stainless steel bracket. Arteries were then incubated for 3 h at 25°C in a well-stirred and aerated PSS containing 12.5 µM fura-2 AM, 25 µg/ml pluronic F-127 (BASF, Wyandotte, MI)and approximately 2 mg/ml bovine serum albumin. After incubation, the tissues were rinsed in PSS for 20 to 30 min to remove any residual dye. The stainless steel bracket and artery assembly was then attached to a Teflon mount with an inflow and outflow port and fitted into an acrylic cuvette; the final chamber volume was 2.4 ml. The cuvette was connected to a Cole-Palmer circulating pump (Cole-Palmer Instruments, Chicago IL) via polyethylene tubing in which 37°C solutions (PSS) could be perfused (10 ml/min). The cuvette was placed in a water-jacketed holder maintained at 37°C of a PTI Delta Scan-1 (Photon Technology International, South Brunswick, NJ) dual wavelength spectrofluorimeter, configured for front face measurements. The cuvette was aligned such that the artery was placed perpendicular to the path of the excitation light beam. Fluorescence was excited at 340 and 380 nm and emission measured at 510 nm.

Intracellular pH Measurements. pH_i was monitored by measuring fluorescence with the pH indicator dye 2',7'-bis(carboxyethyl-5(6')carboxyfluorescein) (BCECF), as previously described (Nagesetty and Paul, 1994). After equilibration, an everted ring was mounted as described above for Ca⁺⁺ measurements. Background tissue autofluorescence was measured at 505 and 439 nm excitation wavelengths and at 523 nm emission. After obtaining a baseline fluorescence, tissues were loaded with 5 µM BCECF-AM in PSS for up to 60 min at 37°C or until the dye signal was at least 10 times the baseline tissue fluorescence at 505 nm and 3 to 5 times the signal at 439 nm. Subsequently, the tissue was washed in PSS for approximately 30 min to eliminate any unesterified BCECF. Absolute values of pH_i were calibrated by the high-K⁺ nigericin technique (Thomas et al., 1979). At the end of each experiment, nigericin (8 µM) was added to the cuvette, then PSS-buffered, high K⁺ solutions of known pH's (pH 6.8-7.8) were perfused through the cuvette. The ratio (minus background) of the fluorescence intensities was near linearly related to the pH. PTI software, using their look-up tables, were used to convert the ratios to pH_i.

Reagents. All reagents were of the highest purity and purchased from Sigma Chemical Co. (St. Louis, MO) except as noted. Fura-2 and BCECF were purchased from Molecular Probes (Eugene, OR). Levosimendan was a gift from Orion Pharma Research (Espoo, Finland). U46619 was dissolved in ethanol, and levosimendan and milrinone in dimethyl sulfoxide; no effects of vehicle were noted if total vehicle was 0.1% or less.

Analysis of Data

The values given are arithmetic means ± S.E.M. N-values for sample size represent the number of hearts from which arteries were taken. Differences between groups were assessed by standard analysis of variance or two-tailed Student's t test for paired data as appropriate. A significance level of P < .05 was chosen for rejection of the null hypothesis.

	Results

Top Abstract Introduction Materials & Methods Results Discussion References

Contractility Studies

Fig. 1 shows a record of isometric force using a protocol typical for these studies. After an equilibration period, at least two contraction/relaxation challenges to KCl were performed until reproducible contractions were achieved. The effect of levosimendan or milrinone on baseline force was then measured. Cumulative additions of either agonist at 1 to 10 µM produced a modest but dose-dependent relaxation, limited largely to denuded vessels. This is likely related to the observation that endothelium-denuded vessels are known to develop considerable active force ("basal tone") in contrast to intact arteries (Ngai et al., 1990). The maximum reductions were small <2% and <10% of the maximal developed isometric force for endothelium-intact and denuded arteries, respectively. This reduction in basal tone was similar for both levosimendan and milrinone; however, the small reduction precluded precise analysis.

View larger version (9K): [in this window] [in a new window]   Fig. 1.   Isometric force records for porcine coronary artery. Force was elicited in response to 0.1 µM U46619. Cumulative concentration-responses were generated to levosimendan (left) and milrinone (right). In separate time-controls, a steady state force of 90% of the initial peak was maintained after 10 min of addition of U46619. On the right, dimethyl sulfoxide at the highest concentration used (0.1%) was added before the milrinone as a control for vehicle effects. This was then rinsed out and U46619 was added back to the bath. Averaged data for eight arteries are summarized graphically in Fig. 2.

We studied the relaxation effects of levosimendan and milrinone on contractions activated by the receptor-mediated, thromboxane A₂ analog, U46619. At 0.1 µM, this agonist produces a steady state force of approximately 90% of the maximum isometric force within 10 min. Concentrations of these vasodilators in the range from 0.01 to 10 µM were used. Higher concentrations of levosimendan (10⁴ - 3 × 10^-3 M) have been reported (Gruhn et al., 1998) to elicit complete relaxation. In contrast to other vasodilators such as isoproterenol, the relaxations appeared to continuously increase once a threshold was attained, in contrast to achieving a definite steady state. The rates were clearly more rapid with larger doses. An apparent concentration-relaxation relation was generated by terminating a measurement at a fixed time, approximately 10 min after addition of vasodilator. Data from such cumulative concentration-response curves for eight coronary arteries are shown in Fig. 2. At any given concentration, the decrease in force was greater in denuded than in intact arteries for either levosimendan or milrinone (Fig. 2). The degree of relaxation was similar, although milrinone was marginally more potent than levosimendan (apparent ED₅₀ 12 versus 24 µM, respectively).

View larger version (14K): [in this window] [in a new window]   Fig. 2.   Averaged concentration-relaxation relations for levosimendan (left) and milrinone (right) for porcine coronary artery. Upper record in each case (squares) are the relations for the intact artery, while the lower lines (circles) represent endothelium-denuded arteries. Error bars represent ± S.E.M. for n = 8 hearts each.

Intracellular Calcium Ion Concentration

To aid in limiting the mechanisms underlying the observed inhibition of force, we measured [Ca⁺⁺]_i by ratiometric technique with fura-2 in porcine coronary artery. Figure 3 shows a typical emission (510 nm) ratio for fura-2 fluorescence excited at 340 and 380 nm.

View larger version (20K): [in this window] [in a new window]   Fig. 3.   Effects of levosimendan (upper) and milrinone (lower) on [Ca++]i in porcine coronary artery. After a control contraction/relaxation cycle with 30 mM KCl, the agents were added under unstimulated conditions. Then, U46619 (0.1 µM) was added and the pharmaceuticals were added cumulatively. Average data are presented in Fig. 4.

Fig. 4 shows the averaged data for the absolute 340/380 ratio for five arteries. This index of [Ca⁺⁺]_i is expressed in terms of the absolute ratio for a control KCl contracture taken as 100% with unstimulated conditions set to 0%. The effects of levosimendan were to decrease the 340 and 380 fluorescence in a concentration-dependent manner, as expected based on the absorbance spectra of levosimendan. The 380 value decreased more than that of the 340, leading to an increase in the ratio. Interestingly, the absorption effects of levosimendan were similar in fura-2 loaded or unloaded arteries, suggesting that the effect was solely attributable to a reduction in the intensity of the exciting light. Levosimendan alone was not fluorescent. We chose a correction based on both unloaded arteries and loaded tissue to which levosimendan had been added under unstimulated conditions. Under stimulated (0.1 µM U46619) conditions, the increment added to the 340/380 ratio by levosimendan was corrected for that added under unloaded or unstimulated conditions. The latter is based on the assumption that the apparent increase in [Ca⁺⁺]_i with 10 µM levosimendan, if real, would have activated force under unstimulated conditions. The uncorrected ratio and corrected ratios are summarized in Fig. 4.

View larger version (20K): [in this window] [in a new window]   Fig. 4.   Effects of levosimendan on [Ca++]i, estimated from the 340/380 ratio. Left panel, experimental values. KCl was 30 mM and U46619 was 0.1 µM. Peak and ss refer to maximum and steady state ratio values. Right panel, ratio corrected for the effects of levosimendan absorption; n = 6; bars, ± S.E.M. Despite relaxing force, levosimendan does not markedly decrease intracellular Ca++.

There are a number of notable features of these data. First, the increase in force with KCl or U46619 was similar, but the increase in [Ca⁺⁺]_i with the receptor-mediated agonist was significantly less. This is in agreement with the literature, suggesting that the apparent Ca⁺⁺ sensitivity is lower for depolarization. The most significant aspect is that despite the decrease in force, [Ca⁺⁺]_i remains elevated in the presence of inhibitory concentrations of levosimendan. As seen in Fig. 4, the corrections for levosimendan absorption at 1 and 3 µM is relatively minor, so that the precision and accuracy of fura-2 ratio as an index of [Ca⁺⁺]_i is likely the highest. Under these conditions, levosimendan decreases force by 20 to 25% (Fig. 2), with little change in intracellular [Ca⁺⁺]. At 10 µM, the corrected fura-2 signal is only 13% below that of the U46619 stimulated value despite a 40% decrease in force; however, the size of the correction makes the precision of this measurement less than at the lower concentrations of levosimendan.

Milrinone also relaxes force in a dose-dependent manner, but in contrast to levosimendan, milrinone also lowers the fura-2 ratio in a parallel fashion as shown in Fig. 5. Milrinone does not affect any optical artifacts and the fura-2 ratio is a direct index of [Ca⁺⁺]_i. Thus the primary mechanism of the milrinone vasodilatation appears to be a decrease of intracellular Ca⁺⁺ . 

View larger version (18K): [in this window] [in a new window]   Fig. 5.   Effects of milrinone on intracellular Ca++ in porcine coronary artery. Note that increasing the concentration of milrinone decreases [Ca++]i, whereas the opposite is true for levosimendan (Fig. 4) despite the decrease in force (Fig. 1). KCl was 30 mM and U46619 was 0.1 µM. Peak and ss refer to maximum and steady state ratio values.

To further investigate the hypothesis that levosimendan is a Ca⁺⁺-desensitizer in smooth muscle we used an alternative experimental protocol, designed to maximize the inhibition of levosimendan at concentrations which did not appreciably interfere with fura-2 measurements of [Ca⁺⁺]_i. As shown in Fig. 6, low doses of U46619 (1-3 nM) produce a slowly developing contraction, which develops about 80% of the maximal isometric force. Addition of 1 µM levosimendan reduced this contraction by 70%, as summarized in Fig. 7. Simultaneous measurement of [Ca⁺⁺]_i (Fig 6) showed a small increase averaging 32% of the KCl control (Fig. 7) with low concentrations of U46619. Addition of 1 µM levosimendan decreased [Ca⁺⁺]_i from 32% to 20% of the KCl control. Thus part of its reduction of force can be attributed to a decrease in [Ca⁺⁺]_i. Although this decrease was statistically significant in pairwise comparisons, the small increase in [Ca⁺⁺]_i with low concentrations of U46619 does not permit the highest precision in these experiments. In addition, contractions in response to U46619 are near irreversible; therefore, repeated dose-response measurements in the presence and absence of drug are not possible. Although levosimendan is less potent against KCl stimulation, KCl provides greater increases in [Ca⁺⁺]_i and is reversible, facilitating the following experimental design to further study Ca⁺⁺-dependent and -independent effects of levosimendan.

View larger version (13K): [in this window] [in a new window]   Fig. 6.   Simultaneous measurement of isometric force and [Ca++]i in porcine coronary artery. The fura-2 ratio is the slightly broader line. Major scale unit equals 20 mN for force and 0.04 for the 340/380 ratio. Note that 2 nM U46619 yields a contraction of near magnitude to the KCl contracture, but with a much smaller increase in [Ca++]i. One micromolar levosimendan after initial rapid increase, due to its absorbance, causes a small decrease in the ratio. Data from this type of experiment are summarized in Fig. 7.

View larger version (33K): [in this window] [in a new window]   Fig. 7.   Summary of the effects of preincubation with 1 µM levosimendan on force and [Ca++]i measured simultaneously in porcine coronary artery.

After stable isometric contraction to 80 mM had been achieved, a cumulative concentration-force relation to KCl was generated. The artery was then relaxed and 3 µM levosimendan or milrinone added. After 15 min, a second concentration-force relation was measured. In control experiments, a similar protocol was followed, except for omission of the pharmaceuticals. The results of these simultaneous measurements of force and [Ca⁺⁺]_i are shown in Fig. 8. Both levosimendan and milrinone elicited rightward shifts in the KCl-force relations. Inhibition was primarily seen at lower KCl concentrations with only marginal decreases noted at 30 mM KCl. [Ca⁺⁺]_i measurements on the other hand showed considerable divergence. Milrinone decreased [Ca⁺⁺]_i at every KCl concentration, whereas levosimendan increased [Ca⁺⁺]_i. These data are graphically presented in parametric form to further highlight these differences in Fig. 9.

View larger version (28K): [in this window] [in a new window]   Fig. 8.   The dependence of isometric force (top) and [Ca++]i (bottom) on KCl concentration in porcine coronary artery. Force and [Ca++]i were measured simultaneously for each of four arteries, Ca++ was calibrated with a modified protocol after Grynkiewicz et al. (1985, see Materials and Methods). LEVO and MILRI indicate 3 µM levosimendan and milrinone, respectively. Two cumulative dose-response curves were measured for each artery, a control curve was generated followed by a 15-min incubation with the drug and then the second measurement. Control-L and Control-M refer to the initial control for levosimendan and milrinone and Control-T1 and Control-T2 represent a time control in which the second dose-response relation was measured after a 15-min incubation but in the absence of drug. Both levosimendan and milrinone inhibit force, but whereas milrinone decreases [Ca++]i, levosimendan was associated with an increase at each concentration of KCl.

View larger version (23K): [in this window] [in a new window]   Fig. 9.   Relation between isometric force and [Ca++]i. Data and symbols from experiments described in Fig. 8. In the presence of 3 µM levosimendan (filled squares), the relation is shifted to the right of that for the controls. This operationally defines levosimendan's action as that of "Ca++ desensitization". Data derived in the presence of 3 µM milrinone, on the other hand, lie within the range of relations obtained for the initial controls and time (2nd dose-response) controls.

The control and milrinone data can be seen to cluster, whereas the levosimendan data are clearly right-shifted. A similar although smaller rightward shift in the force-[Ca⁺⁺]_i relation was also observed in experiments with 1 µM levosimendan (data not shown). The rightward shift of the levosimendan data indicate that much lower forces are produced at any given level of [Ca⁺⁺]_i, the operational definition for Ca⁺⁺ desensitization. In contrast, Milrinone lies in the region of the control Ca⁺⁺-force relations, supporting the contention that its mechanism of action involves reduction of [Ca⁺⁺]_i rather than Ca⁺⁺ desensitization.

One potential mechanism of Ca⁺⁺ desensitization is acidification of intracellular pH. This hypothesis is tested in the following section.

Effects of Levosimendan and Milrinone on pH_i

pH_i can be an important effector of contractility in vascular smooth muscle. We have previously shown that force is inhibited at acidic pH_is and enhanced at alkaline pH_is (Nagesetty and Paul, 1994). Thus the effects of levosimendan and milrinone on pH_i are of importance to our understanding of the mechanism underlying their vasodilatory effects. We have undertaken these pH_i measurements with BCECF, a ratiometric fluorescent dye sensitive to pH_i, in intact porcine coronary artery.

Levosimendan. Fig. 10 shows an example of such an experiment, the design of which parallels that used for force (Fig. 1). Under basal conditions there was little effect of either levosimendan or milrinone on pH_i. The addition of U46619 was consistently associated with a small decrease in pH_i (0.05-0.08). The cumulative addition (1 and 3 µM) of levosimendan was associated with slight increases in pH_i (~0.02). These data are summarized in Fig. 12.

View larger version (12K): [in this window] [in a new window]   Fig. 10.   Effects of levosimendan on pHi of porcine coronary artery. Right panel, enlargement of the left panel, showing the effects of levosimendan. Stimulation with U46619 slightly decreases pHi, whereas levosimendan elicited a small alkalinization.

Milrinone. The effects of milrinone on pH_i are shown in Fig. 11. Again, the effects were minor. The behavior was similar to that of levosimendan in that there was a small alkalinization noted in a dose-dependent fashion. These data are summarized in Fig. 12.

View larger version (10K): [in this window] [in a new window]   Fig. 11.   The effects of milrinone on pHi in porcine coronary artery. Right panel is an enlargement of the left panel. Milrinone had little effect on pHi

View larger version (15K): [in this window] [in a new window]   Fig. 12.   The average effects of levosimendan (left) and milrinone (right) on pHi in porcine coronary artery (n = 4-5). Bars indicate the averaged (±S.E.M.), steady-state values for the conditions indicated above, corresponding to the experimental protocol shown in Figs. 4 and 5. As a whole, the effects of these pharmaceutical agents (1 and 3 µM) were similar but quite small. Note that the ordinate scales are not identical.

	Discussion

Top Abstract Introduction Materials & Methods Results Discussion References

Both levosimendan and milrinone decreased active isometric force and inhibited the development of force in a dose-dependent manner in porcine coronary arteries. This was true for either receptor-mediated stimulation by the thromboxane analog U46619, or depolarization (KCl). The evidence presented here, however, suggests that the mechanisms underlying these effects differ significantly between these two pharmaceuticals.

The relaxation elicited by vasodilator drugs often varies with the mode of stimulation. Interpretation of our data in terms of vasodilator mechanism is dependent on our understanding of the mechanisms underlying KCl and U46619 contractures. In this context, it is perhaps easier to hypothesize for KCl stimulation, for which there appears to be a general consensus (de Lanerolle and Paul, 1991) for its mechanism of activation of contraction. Depolarization leads to an opening of voltage-dependent Ca⁺⁺ channels and an increase of [Ca⁺⁺]_i, although effects of Ca⁺⁺-induced Ca⁺⁺ release by the sarcoplasmic reticulum might also be involved. After preincubation with levosimendan, [Ca⁺⁺]_i attained at each level of KCl (Fig. 8, bottom) was greater than control values. Some caution must be given to this interpretation as for any study using indicator dye-based measurement. Levosimendan decreased the fura-2 fluorescence at both 340 and 380 nm, although this interference was small at 1 and 3 µM. Our correction for this initial absorbance is not unreasonable, but additional intracellular interaction between levosimendan and fura-2 fluorescence cannot be unambiguously ruled out. However, preincubation with levosimendan did not result in any change in the 340/380 ratio after the correction for the initial absorbance. This would argue against the possibility of some time-dependent interference.

The mechanism by which levosimendan increased [Ca⁺⁺]_i under depolarization is speculative at this point, but it could be explained by interaction of the compound with Ca⁺⁺-binding proteins. The increase in [Ca⁺⁺]_i elicited by levosimendan was potentiated, in absolute terms, when [Ca⁺⁺]_i was elevated at higher KCl concentrations. This phenomenon may suggest that the binding of levosimendan to its target protein occurred in a calcium-dependent manner as it does in cardiac cells to troponin C (Pollesello et al., 1994; Haikala et al., 1995a). Therefore, the levosimendan-induced increase in [Ca⁺⁺]_i under KCl depolarization may be ascribable to the dissociation of Ca⁺⁺ ions from EF-hand Ca⁺⁺-binding proteins upon binding by this compound. Because levosimendan simultaneously inhibited the development of contraction, it may dissociate calcium from the proteins which act as an important link in the cascade of the contraction process in vascular smooth muscle.

Unlike levosimendan, the PDE III inhibitor milrinone inhibited KCl-induced increases in force and [Ca⁺⁺]_i in a parallel fashion (Fig. 8, bottom). The lack of effects of milrinone on the relationship between force and [Ca⁺⁺]_i (Fig 9) is consistent with the major mechanism of milrinone inhibition being reduction of [Ca⁺⁺]_i. It has been suggested (Silver et al., 1988) that the decrease in [Ca⁺⁺]_i can be attributed to the inhibition of PDE by milrinone and increase in cAMP. However, the increase in cAMP in porcine coronary arteries reported for the concentrations of milrinone used in this study is marginal at best (Gruhn et al., 1998).

An understanding of potential vasodilatory mechanisms for U46619 contractures is more difficult due to the controversy concerning the mechanism of this receptor-mediated agonist. U46619, as any receptor-mediated stimulus, is purportedly linked to inositol 1,4,5-triphosphate and diacylglycerol production (Dorn et al., 1992). 1,4,5-Triphosphate causes a release of Ca⁺⁺ from the sarcoplasmic reticulum and diacylglycerol is an activator of protein kinase C. In earlier studies (Bradley and Morgan, 1987), it was suggested that this agonist elicited a contraction in the absence of an increase in [Ca⁺⁺]_i. Our previous studies (Nagesetty and Paul, 1994) and others (Sumiyoshi et al., 1997) show an increase in [Ca⁺⁺]_i which can be as large as that of KCl at concentrations of U46619 at 0.1 µM or greater. On the other hand, we are in agreement with the literature in that at lower concentrations, considerable force can be elicited by U46619 with only modest increases in [Ca⁺⁺]_i (Figs. 6 and 7). Interestingly, we found (unpublished observations) that nifedipine can inhibit 63.5 ± 3.5% (n = 4) of the 0.1 µM U46619 contraction and nearly 100% (n = 3) of that elicited by 2 nM. This suggests that a large fraction of the increase in [Ca⁺⁺]_i is due to activation of Ca⁺⁺ channels. What is clear is that U46619 contractures can require significantly less Ca⁺⁺ than that associated with KCl. This has been associated with Ca⁺⁺-sensitization mechanisms attributable to activation of protein kinase C and the potential involvement of thin filament proteins caldesmon (Walsh, 1990) and/or calponin (Marston, 1995).

At 0.1 µM U46619, levosimendan (0.1-10 µM) decreased contraction without any significant changes in [Ca⁺⁺]_i (Figs. 3 and 4). The caveat here is that the relaxations were relatively small and the largest relaxation (~40%) with 10 µM levosimendan required a significant correction due to the interference of levosimendan with fura-2 reducing the precision of this measurement. At the lower concentrations of levosimendan there is a higher confidence in the lack of change in [Ca⁺⁺]_i concomitant with the observed relaxation. At lower doses of U46619 (2-3 nM), significant forces were developed, with only modest increases in [Ca⁺⁺]_i. Low concentrations of levosimendan (1-2 µM) elicited significant relaxations (70%) which were accompanied by a 35% decrease in [Ca⁺⁺]_i. This suggests that part of the effect of levosimendan could be attributable to decreasing [Ca⁺⁺]_i, under these conditions. Milrinone, in contrast to levosimendan, reduced [Ca⁺⁺]_i in a concentration-dependent manner for contractions elicited by 0.1 µM U46619. This is similar to that for KCl stimulation and again supports a mechanism for milrinone of a reduction in [Ca⁺⁺]_i leading to the relaxation,.

The difference between the effects of levosimendan in KCl and U46619 experiments may be important to differentiate between various mechanisms. Because the effects of potassium channels are known to be attenuated at high extracellular KCl, our results are consistent with recent studies indicating that levosimendan increases ATP-sensitive potassium channel currents (Yokoshiki et al., 1997). There is, however, indirect evidence suggesting potassium channels may not be involved (Gruhn et al., 1998). Because milrinone lowered [Ca⁺⁺]_i under KCl depolarization, it likely is operating through a mechanism differing from levosimendan.

It is worth noting that at similar concentrations of levosimendan, but with KCl stimulation instead of U46619, fura-2 measurements of [Ca⁺⁺]_i showed an increase. This again argues that the estimates of [Ca⁺⁺]_i using fura-2 were not affected by a systematic error due to the correction for the absorbance of levosimendan.

A pathway involving Ca⁺⁺ desensitization was most clearly shown for KCl contractures, because under these conditions any action of levosimendan on potassium channels would be minimized. One possible mechanism for the Ca⁺⁺ desensitizer action might be attributable to pH. pH_i can be an important effector of contractility in vascular smooth muscle (Wray, 1988). We have previously shown that steady-state force in porcine coronary artery is inhibited at acidic pH_is and enhanced at alkaline pH_is (Nagesetty and Paul, 1994). We thus undertook to measure the effects of these pharmaceuticals on pH_i with BCECF as the indicator dye. In this study, neither levosimendan nor milrinone markedly affected pH_i. Under basal conditions there was little effect of either agent. To parallel the stimulated conditions under which relaxation was measured, we also measured the effects in the presence of 0.1 µM U46619. The addition of U46619 was consistently associated with a small decrease in pH_i (0.05-0.08). The cumulative addition (1 and 3 µM) of either levosimendan or milrinone was associated with slight increases in pH_i (~0.02). Given that alkalinization was associated with an increase in force, the small alkalinizations seen (Fig. 6) are very unlikely to underlie the relaxing effects of these pharmaceuticals, and in particular, the Ca⁺⁺ desensitization observed with levosimendan.

The mechanism(s) involved in this Ca⁺⁺ desensitization is not clear, but given the known binding of levosimendan to troponin C (Haikala et al., 1995a), primary candidates would likely include Ca⁺⁺ -binding EF-hand proteins. Binding to putative thin filament regulatory proteins in smooth muscle is speculative, but an intriguing possibility. As Ca⁺⁺-binding EF-hand proteins, the regulatory myosin light chains may also serve as targets for levosimendan. Clearly, more experimental evidence is needed to understand the mechanism of the observed Ca⁺⁺ desensitization.

A final observation of significance is that in arteries with an intact endothelium, the relaxation after 10 min for each cumulative dose of levosimendan (as well as milrinone) was less than that in denuded arteries (Fig. 2). Although in our studies a true steady state was not achieved, these results are in agreement with other investigators (Gruhn et al., 1998), who reported that levosimendan was a less effective vasorelaxant in endothelium-intact arteries. The effects of levosimendan on endothelial cell function are not known and may be of future importance to potential clinical effects.

In conclusion, of highest interest in terms of the basic science of smooth muscle is that levosimendan can act in the coronary artery as a "calcium desensitizer". This is of potential relevance, because levosimendan's clinical action on coronary artery is the opposite to its Ca⁺⁺-sensitizing action in cardiac muscle. Evidence has been reported (Gruhn et al., 1998) suggesting that the mechanism of levosimendan vasodilitation is not mediated Ca⁺⁺ entry blockade, release of cyclooxygenase products, or -adrenoceptor stimulation. Our results indicate a lowering of [Ca⁺⁺]_i by levosimendan consistent with opening of potassium channels and a relaxation that is independent of [Ca⁺⁺]_i and is not likely attributable to effects of pH_i. Taken as a whole, the evidence points to a mechanism that might involve the direct effect of levosimendan on the smooth muscle contractile or regulatory proteins themselves. Additional experimentation is needed to resolve this interesting mechanism.