Introduction

Top Abstract Introduction Methods Results Discussion References

Despite knowledge about the Ca⁺⁺ pumping activity of intracellular Ca⁺⁺ stores for several decades, for a long period of time there was surprisingly little attention devoted to determining specific inhibitors of the Ca⁺⁺ ATPase. Perhaps this was a result of the fact that some inhibitors of Ca⁺⁺ uptake (such as caffeine) proved instead to be activators of Ca⁺⁺ release channels in the same membranes. Only more recently have a number of pump inhibitors been identified and to a lesser extent characterized.

The most frequently used pump inhibitor is thapsigargin (Sagara et al., 1992; Davidson and Varhol, 1995), but it is quite expensive, sometimes difficult to titrate and may require unusually high concentrations on certain intact, multicellular preparations (Baudet et al., 1993). Another commonly used pump inhibitor is cyclopiazonic acid (Goeger et al., 1988; Goeger and Riley, 1989; Seidler et al., 1989). The third frequently used SERCA pump inhibitor is 2',5'-di(tert-butyl)-1,4-benzohydroquinone (Nakamura et al., 1992), now even modified for use as a caged inhibitor (Rossi and Kao, 1997).

These pump inhibitors have been used in numerous studies on intact cells and tissues attempting to discern roles for these intracellular Ca⁺⁺ pumps and/or the stores that contain them. Such studies have involved cell types and tissues as diverse as skeletal muscle (Westerblad and Allen, 1994; Du et al., 1994), cardiac muscle (Hove-Madsen and Bers, 1993; Rogers et al., 1995), smooth muscle (Shima and Blaustein, 1992; Kasai et al., 1994; Fukao et al., 1995), endothelial cells (Vaca and Kunze, 1994), gastric parietal cells (Negulescu and Machen, 1995), pancreatic acinar cells (Kwan et al., 1990; Toescu and Petersen, 1994), hepatocytes (Llopis et al., 1991), T lymphocytes (Premack et al., 1994), oocytes (Petersen and Berridge, 1994), colonic epithelium (Bischof et al., 1995) and brain (Reyes and Stanton, 1996), to name just a few.

The studies that have used these inhibitors have tended to assume that the only effects they were having were to prevent Ca⁺⁺ loading of the stores. Possible side effects on fluxes through release channels or other proteins were generally not taken into consideration. The aim of our study was to determine which of several pump inhibitors had separate direct side effects on release channels. We were interested in identifying the best pump inhibitor(s) that could be used in studies examining the role of the SR Ca⁺⁺ pump, for our purposes particularly in the phenomenon of quantal Ca⁺⁺ release (Dettbarn et al., 1994).

	Methods

Top Abstract Introduction Methods Results Discussion References

Heavy microsomes and purified terminal cisternae were prepared from rabbit skeletal muscle as described in Dettbarn et al. (1994). Uptake and release determinations were carried out with 190 or 380 µg heavy microsomes, as indicated. ⁴⁵Ca efflux determinations were carried out using 318 µg of purified terminal cisternae. Cerebellar microsomes were prepared from frozen canine cerebella in a fashion described in Dettbarn et al. (1995). Determinations of uptake and release were carried out essentially as described in Palade et al. (1989) and Dettbarn et al. (1995) with 1.47-mg cerebellar microsomes. Protein concentrations were determined using Coomassie Protein Assay Reagent (Pierce, Rockford, IL) using BSA as a standard.

Spectrophotometric measurements of SR Ca⁺⁺ uptake and release were performed at 33°C in a diode array spectrophotometer (HP 8451A, Hewlett Packard, Palo Alto, CA). The standard assay medium included 100 mM KCl, 20 mM KMOPS, 2.5 to 10 mM K-phosphate, 1 mM MgATP, 5 mM Na₂ phosphocreatine, 20 µg/ml creatine phosphokinase and 0.2 mM antipyrylazo III, pH 6.8. SR Ca⁺⁺ uptake determinations were performed in the presence of 10 mM phosphate, and SR Ca⁺⁺ release measurements in the presence of 2.5 mM phosphate. Rates of uptake in the presence of drugs were normalized by dividing by the prior rate of uptake by the same sample in the same cuvette in the absence of drug. Cerebellar microsome Ca⁺⁺ uptake and release measurements were performed in the presence of 50 mM KCl, 10 mM KMOPS, 56 mM K-phosphate, 1 mM MgATP, 5 mM Na₂phosphocreatine, 20 µg/ml creatine phosphokinase and 0.2 mM antipyrylazo III, pH 6.8 at 30°C. Extravesicular Ca⁺⁺ concentration changes were followed by measuring antipyrylazo III absorbance at A₇₁₀-A₇₉₀.

⁴⁵Ca efflux measurements on terminal cisternae were carried out by spiking the cuvette with ⁴⁵Ca before addition of 318 µg terminal cisternae. The ⁴⁵Ca was then taken up by the sample at the same time as endogenous unlabeled Ca associated with the sample. When the spectrophotometric trace became flat, indicating completion of uptake, 50 µl of the mixture were removed and added to 1.4 ml of a chilled quench solution consisting of 100 mM KCl, 20 mM KMOPS, 5 mM MgCl₂ and 20 mg/liter ruthenium red, at t = 15 sec. At t = 0 sec, a 50 µl addition was made of a buffered Ca solution consisting of 100 mM KCl, 20 mM KMOPS, 20 mM EGTA, 1 or 6 mM CaCl₂, pH readjusted to 6.9. At t = 1 sec, 3 µM thapsigargin, 100 µM cyclopiazonic acid or 100 µM di-TBQ were added. Controls had no addition at t = 1 sec. At t = 15, 30, 45, 60, 75, 90 and 105 sec, 50-µl aliquots were withdrawn and added to 1.4 ml quench medium on ice. The quenched aliquots were then individually filtered through Millipore (Bedford, MA) HAWP filters (0.45 µm effective pore size) and counted to assess ⁴⁵Ca remaining in the vesicles.

	Results

Top Abstract Introduction Methods Results Discussion References

Three pump inhibitors widely used in physiological studies (thapsigargin, cyclopiazonic acid and 2',5'-di(tert-butyl)-1,4-benzohydroquinone) were analyzed. We began by first working with concentrations of each pump inhibitor which caused >50% inhibition of net active Ca⁺⁺ uptake by isolated skeletal muscle sarcoplasmic reticulum in the presence of 10 mM phosphate (to precipitate Ca⁺⁺ inside the SR and thus prolong the linear phase of net uptake) and 2.2 µM ruthenium red (to inhibit release channels). We wished to avoid concentrations that would completely inhibit the pump because we anticipated that such concentrations would inevitably induce release on their own through any release channels or leak pathways normally open at rest. Traces demonstrating the effects of 1 µM thapsigargin, 10 µM CPA and 10 µM di-TBQ on net uptake are shown in the left set of panels of figure 1. The rises in each trace represent additions of 37.5 nmol Ca. The first decline in each trace represents a control Ca⁺⁺ uptake, and the second decline, the slowed Ca⁺⁺ uptake in the presence of the inhibitor under investigation.

View larger version (24K): [in this window] [in a new window]   Fig. 1.   Effects of moderate concentrations of pump inhibitors on uptake and release of Ca++ from isolated skeletal muscle heavy microsomes. Left, Effects of pump inhibitors on uptake of 37.5 nmol additions of Ca () by 380 µg microsomes. Right, Microsomes (380 µg) were preloaded with six additions of 12.5 nmol Ca () and then challenged with the pump inhibitor (arrow), followed by 5 mM caffeine (triangle) and then a recalibration with 12.5 nmol Ca (). Note different time calibrations for uptake and release experiments. Traces representative of experiments performed in quadruplicate.

Next, the same amount of SR vesicles was loaded with a larger amount of Ca⁺⁺ in the absence of ruthenium red or the pump inhibitor. After preloading in a medium containing a more physiological concentration of 2.5 mM phosphate, the vesicles were challenged with 5 mM caffeine to give the control Ca⁺⁺ release shown in the upper right trace in figure 1. In this trace, the first rise in the trace represents the addition of vesicles, and the next six rises in the trace represent six additions of 12.5 nmol Ca each. After uptake of these additions had been completed, the addition of caffeine is indicated by triangles.

The effect of pump inhibitors on such releases was then assayed in the remaining traces of the right panel of figure 1 using the same concentration of pump inhibitor as used in the left panel. After uptake of Ca additions had been completed, the pump inhibitor was added, in all cases inducing a temporary rise in the traces. The rise shown in the figure was greater with thapsigargin than the other inhibitors, but the degree of pump inhibition was also greater. Unless there is substantial leak, a compound that solely inhibited pumping partially should not cause net release. As seen in the set of traces in the right panel of figure 1, however, thapsigargin, cyclopiazonic acid and di-TBQ all performed as expected of release channel activators. Such releases were followed shortly by slowed reuptake, suggesting that the inhibitors had the dual effect of decreasing uptake and at least transiently increasing leak as well. In the case of the thapsigargin trace, 5 mM caffeine were subsequently added, eliciting a release smaller than the control, demonstrating that thapsigargin had released Ca from the caffeine-sensitive pool.

After addition of the pump inhibitor, the vesicles were challenged with 5 mM caffeine to activate Ca⁺⁺ release channels. None of the compounds interfered with the ability of caffeine to release Ca⁺⁺. In fact caffeine administered when di-TBQ or CPA-induced release was ongoing resulted in a potentiated caffeine-induced release of Ca⁺⁺ that was not observed if caffeine was administered after a baseline in the trace had been reestablished (table 1, second column). The rate of reuptake of Ca⁺⁺ after the release was clearly slowed relative to that of the control caffeine-induced release. At the end of all traces, a further 12.5 nmol Ca addition was made for recalibration purposes in the presence of drug.

Still higher concentrations of certain pump inhibitors are known to cause release in other systems, presumably by unmasking a resting leak of Ca⁺⁺. We wished to determine if this was an effect held in common by all pump inhibitors and to determine the leak pathway involved. Accordingly, we preloaded the vesicles as before and used higher concentrations of each pump inhibitor, concentrations that inhibited net uptake in the presence of ruthenium red by more than 95%. These experiments are shown in figure 2 using a similar format as in figure 1. The left panel demonstrates the greater inhibition of uptake at the higher drug concentrations. The middle panel then shows the same heavy microsomes preloaded with 6 × 12.5 nmol Ca in the absence of ruthenium red and subsequently challenged with the same high concentration of pump inhibitor, in all cases now leading to a sustained release of Ca⁺⁺. To determine which pathway was involved in this release, the experiment was repeated in the right panel, this time with 2.2 µM ruthenium red added just before the addition of pump inhibitor. The releases induced by all three pump inhibitors were markedly inhibited by ruthenium red, indicating that much of the release of Ca⁺⁺ was mediated by ryanodine receptors.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Effects of higher concentrations of pump inhibitors on uptake and release of Ca++ from isolated skeletal muscle microsomes. Left, Effects of pump inhibitors on uptake of 37.5 nmol additions of Ca () by 190 µg microsomes. Middle, Microsomes (380 µg) were preloaded with six additions of 12.5 nmol Ca () and then challenged with the pump inhibitor (upward arrow), followed by a recalibration with 12.5 nmol Ca (). Right, Microsomes (380 µg) were treated exactly as in the middle set of traces except that 2.2 µM ruthenium red (RR) was added (downward arrow) just before addition of the pump inhibitor. Traces representative of experiments performed in quadruplicate.

Quantitation of these results and others at different inhibitor concentrations is provided in table 1. All three pump inhibitors inhibited Ca⁺⁺ uptake in a dose-dependent manner, although thapsigargin's dose-dependence was significantly steeper than those of the other inhibitors. None of the pump inhibitors had significant effect on caffeine's ability to release Ca except at an inhibitor concentration high enough to cause significant release by itself. All three pump inhibitors caused significant release at several of the concentrations tested.

The results presented thus far suggested that all three pump inhibitors were able to activate ryanodine receptors even at concentrations at which some Ca pumping activity remained. To verify this, ⁴⁵Ca efflux experiments were performed. To visually enhance leak through the ryanodine receptors, a purified terminal cisternae preparation was used. This had the added advantage of reducing background counts remaining in membrane vesicles containing no ryanodine receptors. As seen in figure 3A, at two different free Ca concentrations, 100 µM cyclopiazonic acid had no significant effect on efflux of ⁴⁵Ca. Indeed, thapsigargin appeared to reduce ⁴⁵Ca efflux. Only di-TBQ exhibited the anticipated effect of increasing ⁴⁵Ca efflux (fig. 3B), and even this effect was modest. These results suggest that the rapid release seen in spectrophotometric studies had a very strong degree of amplification due to CICR.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Effects of pump inhibitors on unidirectional efflux of 45Ca from purified terminal cisternae. Measurements were performed as described in "Methods," with Ca/EGTA buffered solution added at t = 15 sec and pump inhibitor added at t = 16 sec. Efflux was then followed by Millipore filtration to assess 45Ca remaining in the vesicles at different time points. Experiments performed in the final presence of 1 mM EGTA/0.05 mM Ca (free Ca++ estimated to be 52 nM: "Low Ca2+") or 1 mM EGTA/0.3 mM Ca (free Ca++ estimated to be 426 nM: "High Ca2+"). A, Effects of 3 µM thapsigargin and 100 µM cyclopiazonic acid on 45Ca efflux. B, Effects of 100 µM di-TBQ on 45Ca efflux. Experiments performed at least in triplicate and expressed as mean ± S.D.

Because skeletal muscle, in contrast to most cell types with intracellular stores, has few InsP₃ receptors, we also attempted to determine the effects of these same pump inhibitors on uptake into InsP₃-sensitive stores of brain cerebellar microsomes. Unfortunately, under the conditions we worked we were unable to generate dose-response curves because we found that the rate of uptake of Ca⁺⁺ by the microsomes was inhibited less than 50% by concentrations of thapsigargin, cyclopiazonic acid or 2',5'-di(tert-butyl)-1,4-benzohydroquinone that completely inhibited uptake by muscle microsomes. Evidently, much of the Ca⁺⁺ uptake measured under our measurement conditions was into InsP₃-insensitive stores, possibly inside-out plasma membrane vesicles, lacking SERCA pumps. However, 1 to 30 µM thapsigargin, 10 to 200 µM CPA or 10 to 100 µM di-TBQ present during Ca⁺⁺ loading of brain microsomes all inhibited InsP₃-induced Ca⁺⁺ release, whereas the same inhibitor concentrations were essentially ineffective when added before InsP₃ but after Ca⁺⁺ loading (fig. 4). As a consequence, we could determine that none of the pump inhibitors had measurable direct side effects on cerebellar InsP₃ receptors.

View larger version (26K): [in this window] [in a new window]   Fig. 4.   Effects of pump inhibitors on uptake into InsP3-sensitive stores of brain cerebellum and on InsP3-induced release of Ca++ from those stores. Cerebellar microsomes (1.47 mg) were loaded with four additions of 12.5 nmol Ca each (not shown). In each trio of traces, the uppermost trace represents a control in which loading was performed in the absence of pump inhibitors, with 10 µM InsP3 added subsequently. In the middle trace of each trio, loading was performed in the absence of pump inhibitors, which were added 30 sec before addition of 10 µM InsP3. In the bottom, essentially flat traces, loading was performed in the presence of pump inhibitors, with 10 µM InsP3 added subsequently. Maximum release amplitudes were approximately equal to 12.5 nmol Ca. Traces representative of determinations performed in triplicate with one sample and replicated once with another sample. Trace durations at bottom right: 170 sec.

In contrast to reports on many cell types containing InsP₃ receptors and our own studies with isolated SR, even high concentrations of the pump inhibitors produced no rapid release of Ca⁺⁺ from cerebellar microsomes (not shown). Possibly the leak through InsP₃ receptors here is much lower than in other preparations where InsP₃ may also be present under basal conditions. Alternatively, there may be a slower non-SERCA uptake mechanism associated with the InsP₃-sensitive store (e.g., Thevenod and Schulz, 1988), one insufficient to load the stores under our experimental conditions but perhaps capable of maintaining them in a loaded state.

	Discussion

Top Abstract Introduction Methods Results Discussion References

Our spectrophotometric results with isolated SR membranes may appear to be inconsistent with our ⁴⁵Ca efflux determinations, in that the former suggest significant increases in Ca⁺⁺ efflux not seen with the isotope determinations. However, the cytoplasmic Ca⁺⁺ was buffered in the case of the isotope determinations, whereas it was allowed to rise in the spectrophotometric determinations. A modest efflux at low (resting) free Ca⁺⁺ concentration could easily be magnified as the Ca⁺⁺ concentration increased outside the vesicles, due to the intrinsic positive feedback of Ca⁺⁺-induced Ca⁺⁺ release.

Yet how could there be even a modest efflux of Ca⁺⁺ as long as the pump was active, and how could such a net efflux be followed by resequestration? The only reasonable answer is to assume vesicle heterogeneity. Thus, the ratio of ryanodine receptors to pump molecules is not constant from one vesicle to another. As long as there is some low-level unidirectional efflux of Ca⁺⁺ through ryanodine receptors at rest, a pump-leak situation dictates that vesicles with a higher ratio of ryanodine receptors to pumps will need to have a smaller proportion of their pumps inhibited before they begin to release Ca⁺⁺ than vesicles with fewer ryanodine receptors and more pumps. As a consequence, the vesicles with the higher ryanodine receptor to pump ratio will be prone to release their contents at a lower concentration of pump inhibitor than vesicles with a low ratio of ryanodine receptors to pumps. Thus, the true effect of pump inhibitors on release channels is reflected far better in the isotope measurements than in the spectrophotometric measurements.

Among the pump inhibitors tested, thapsigargin has proved to be quite selective in its ability to inhibit the SR Ca⁺⁺ pump with no activatory and only modest inhibitory effect on fluxes through SR release channels. This result confirms conclusions drawn by Kirby et al. (1992), who found no inhibitory effect of passively loaded vesicles at still higher thapsigargin and Ca⁺⁺ (0.1 mM each). Our results also suggest that it has minimal side effects on InsP₃ receptors, in accord with the conclusions of Missiaen et al. (1992). Thapsigargin has been reported to induce Ca⁺⁺ release in various preparations (Razani-Boroujerdi et al., 1994; Smith and Gallacher, 1994) and unmask fluxes mediated by InsP₃ receptors (Toescu and Petersen, 1994). However, in skinned fibers (Du et al., 1994) and in multicellular preparations (Baudet et al., 1993), such high thapsigargin concentrations are required to inhibit the pump fully that its utility is compromised and nonspecific side effects could occur. Finally, it must also be mentioned that thapsigargin is not without side effects on other membrane transporters, as it has been shown to inhibit voltage gated Ca⁺⁺ channels in surface membranes (Nelson et al., 1994; Buryi et al., 1995).

Our findings confirm the unmasking by thapsigargin of a significant Ca⁺⁺ efflux from isolated SR that was reported by Pessah et al. (1997). They reported a significant ryanodine- and ruthenium red-insensitive efflux which could be rendered sensitive to both ryanodine- and ruthenium red by the addition of bastadin 5, suggesting that even the ryanodine-insensitive leak originated from ryanodine receptors operating in a new mode. The observation that the release of Ca⁺⁺ induced by pump inhibitors was only partially inhibited by ruthenium red might involve this ryanodine- and ruthenium red-insensitive mode of the ryanodine receptor (Pessah et al., 1997), but it could also involve the SR Ca⁺⁺ pump itself or a sphingolipid gated release channel, SCaMPER (Mao et al., 1996; Betto et al., 1997). Our results suggest that there is a significant ruthenium red-sensitive component to the efflux as well, although this may only be manifest at higher cytoplasmic [Ca⁺⁺]. We were unable to draw similar conclusions regarding resting efflux from InsP₃-sensitive stores (e.g., Toescu and Petersen, 1994) in our cerebellar microsomes due to the presence of significant Ca⁺⁺ uptake into thapsigargin- and InsP₃-insensitive stores.

CPA appears equally if not more selective than thapsigargin for effects on the pump as opposed to the ryanodine receptor. Although 10 µM CPA in our hands caused release from isolated SR, like thapsigargin, it may be less effective in more intact preparations. For instance, at a concentration of 40 µM, CPA did not inhibit caffeine-induced release from skinned myocardium when applied after loading was complete (Takahashi et al., 1995).

In ryanodine receptor-containing preparations in our hands, 2,5-di-(tert-butyl)-1,4-benzohydroquinone significantly increased unidirectional ⁴⁵Ca efflux in addition to causing net Ca⁺⁺ release. This suggests that there is only a very narrow range of concentrations over which it could be considered a selective inhibitor of the SR Ca⁺⁺ pump. It did not induce Ca⁺⁺ release from cerebellar membranes, which also contain ryanodine receptors, possibly because these membranes fail to respond to most RyR agonists such as caffeine under these experimental conditions (Dettbarn et al., 1995). This hydroquinone also has reported side effects, such as its inhibition of endoplasmic reticulum Ca⁺⁺ permeability (Missiaen et al., 1992), voltage-dependent Ca channels (Nelson et al., 1994) and inward rectifier K channels (Hassessian et al., 1994).

Missiaen et al. (1992) found that 50 µM cyclopiazonic acid or 2,5-di-(tert-butyl)-1,4-benzohydroquinone decreased endoplasmic reticulum Ca⁺⁺ permeability in A7r5 cells lacking ryanodine receptors without affecting InsP₃-induced Ca⁺⁺ release. Our determinations would not have detected such relatively subtle effects, but their results suggest that thapsigargin might be an agent of choice for examination of the role of SERCA pumps not just in ryanodine-sensitive stores but in InsP₃-sensitive ones as well.

Many of the studies referenced here concern the ability of these substances to activate capacitative Ca⁺⁺ entry (Putney, 1986, 1993) through store depletion. The effects of pump inhibitors reported here do not in any way invalidate the conclusions reached from those studies, because stores would have been depleted irrespective of side effects of the pump inhibitors on any release channels affected. The only possible caveat might involve those systems in which a SERCA-independent uptake mechanism is also associated with a particular store.

The concentrations of pump inhibitors used here may not lead to similar effects in other systems. This could be a consequence of the extremely high affinity of thapsigargin and the number of pump sites being titrated, lipophilicity in the case of certain compounds, and even the ratio of pumps to release channels in different preparations. For stores containing only SERCA pumps, any of these three inhibitors could be used to probe the role of SERCA-containing stores in a particular biological response, but, due to possible small activating side effects of di-TBQ on ryanodine receptors, only thapsigargin or cyclopiazonic acid can be relied upon to determine the role of the SERCA Ca⁺⁺ pump itself (e.g., Bassani et al., 1994) in such a response. In preparations where thapsigargin must be used at unusually high concentrations to fully inhibit uptake, cyclopiazonic acid represents the best alternative for testing for the role of the Ca⁺⁺ pump in a particular biological response. In situations where it is important to avoid side effects on L-type Ca⁺⁺ channels, cyclopiazonic acid would also be the agent of choice (Badaoui et al., 1995). Finally, in systems where only InsP₃ receptors are involved, our results suggest that any of these three inhibitors could be used to test for roles of SERCA pumps. This also holds true for use in testing the role of the stores that contain these pumps, provided that emptying of the stores is assured by allowing sufficiently long exposure to the inhibitor. Obtaining similar results in separate experiments with each pump inhibitor would strengthen the conclusions that effects were causally related to pump inhibition as opposed to some other extraneous side effect.