Measurement of Free Ca 2 (cid:49) in Sarcoplasmic Reticulum in Perfused Rabbit Heart Loaded with 1,2-Bis(2-amino-5,6-difluorophenoxy)ethane- N,N,N (cid:42) ,N (cid:42) -tetraacetic Acid by 19 F NMR*

,

The sarcoplasmic reticulum (SR) 1 plays an important role in regulation of mammalian cardiac muscle contraction. It is generally accepted that contraction is activated by Ca 2ϩ influx through the sarcolemmal L-type channel, which subsequently releases SR Ca 2ϩ via the Ca 2ϩ -induced Ca 2ϩ -release mechanism (1). During relaxation, Ca 2ϩ is resequestered into the SR by the SR Ca 2ϩ -ATPase and extruded by the sarcolemmal Na ϩ -Ca 2ϩ exchanger (2,3). The SR Ca 2ϩ content available for release is an important determinant of contractile state. In spite of the importance of SR Ca 2ϩ , little is known about the levels of free SR Ca 2ϩ concentration ([Ca 2ϩ ] SR ) and how it is altered by physiologic or pathologic perturbations. Although previous studies (4,5) using rapid cooling contractures have provided a valuable index of SR Ca 2ϩ load in cultured myocytes, there are currently no direct measurements of [Ca 2ϩ ] SR . It is generally agreed that the SR Ca 2ϩ -ATPase maintains a calcium gradient between the SR matrix and the cytosol which is close to the theoretical limit based on the free energy available from ATP hydrolysis. However, the exact degree of efficiency is debated due to the lack of precise knowledge of [Ca 2ϩ ] SR (see Ref. 6). Values for free SR calcium have been estimated (based on binding to calsequestrin) to be in the range of 0.3-5 mM (6 -8). Furthermore, if the SR Ca 2ϩ -ATPase is operating near its theoretical limit, a fall in phosphorylation potential, for example under conditions of ischemia (9,10), could affect the Ca 2ϩ gradient across the SR.
The present study provides direct measurements of [Ca 2ϩ ] SR using 19 F NMR in Langendorff perfused rabbit hearts loaded with acetoxymethyl ester of 1,2-bis(2-amino-5,6-difluorophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid (TF-BAPTA). Our laboratory has recently developed TF-BAPTA and applied it to measurements of cytosolic [Ca 2ϩ ] in perfused rat hearts (11,12). TF-BAPTA has a high K D for Ca 2ϩ (65 M) and combines both a large shift sensitivity with fast-intermediate exchange kinetics at typical magnetic field strengths (11,12). Such an indicator offers the potential for simultaneous determinations of Ca 2ϩ concentrations in different cellular compartments, contingent on the degree of indicator loading into these compartments. For TF-BAPTA, the chemical shift rather than the ratio of resonance intensities provides information regarding [Ca 2ϩ ] (11,12). The 6-fluorine resonance in TF-BAPTA is insensitive to calcium binding and serves as a shift reference, and the 5-fluorine resonance shifts downfield on Ca 2ϩ complexation. Using this approach we directly measure an ionized calcium concentration in the SR of ϳ1.5 mM, a value in good agreement with estimates obtained using calsequestrin binding constants (7). In addition we calculate the free energy required for the Ca 2ϩ gradient across the SR under control conditions, under conditions of cardiac arrest induced by high extracellular potassium concentration, and under conditions of reduced phosphorylation potential.

EXPERIMENTAL PROCEDURES
Isolated Rabbit Heart Preparations-Male New Zealand White rabbits (ϳ1-1.5 kg) were used and received humane care in accordance with National Institutes of Health standards (35). Rabbits were anesthetized by intravenous injection of pentobarbitone (ϳ100 mg) into a marginal ear vein. The heart was excised rapidly, and the aorta was cannulated. Retrograde perfusion was begun under constant pressure * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Hearts were placed in a standard 30-mm NMR tube. After 10 min of control perfusion, loading with 1000 ml of 5 M acetoxymethyl ester of TF-BAPTA was started. With typical flow rates of 30 -50 ml/min, loading took about 30 min. To monitor contractility, a latex balloon was inserted into the left ventricle. The balloon was inflated to give an end diastolic pressure of 5-10 cm H 2 O. As observed in our previous study of perfused rat hearts (12), TF-BAPTA did not cause a significant reduction in contractility.
NMR Measurements-19 F NMR measurements were performed on a Varian 400 wide-bore NMR spectrometer at 376.27 MHz at 37°C. We shimmed on the proton signal from the unbathed heart, and we routinely obtained a non-spinning line width at one-half height of ϳ0.25 ppm. Ungated spectra were acquired every 5 min using 0.26-s intervals between scans with a pulse angle of 40°(20 s). The spectral width was Ϯ 7060 Hz, and 4000 data points were collected. The free-induction decay was multiplied by an exponential function corresponding to 100-Hz line broadening before Fourier transformation.
To examine whether [Ca 2ϩ ] SR varies during systole and diastole, the NMR pulse was gated to the contraction cycle through a homemade system, which provides a precise, adjustable trigger at the desired point on the pressure wave. The gating system utilized the dP/dt signal to generate a standard pulse signal which initiated the NMR pulse. A delay relative to the standard pulse signal was included in the pulse sequence so that the NMR pulse could be imposed at the desired time during the cardiac cycle. At exactly the same time as the NMR pulse, a triggering signal (TS) was sent out. We recorded simultaneously left ventricular pressure (LVP), dP/dt, and TS to verify the timing of the NMR acquisition. Once the NMR pulse was generated, a free induction decay was acquired. Since the peak of the cytosolic Ca 2ϩ transient occurs near the beginning of the upstroke of LVP wave (13,14), one NMR pulse was gated at the time shown as TS 1 to measure systolic [Ca 2ϩ ] SR ; and another NMR pulse was delayed by 150 ms (TS 2) to measure diastolic [Ca 2ϩ ] SR . The time interval between the scans depended on the heart rate. For a typical heart rate of ϳ150 -200 beats/ min, the interval was ϳ300 -400 ms. 1000 -2000 consecutive gated scans were acquired to achieve an acceptable signal-to-noise ratio. The other parameters were the same as those used in the ungated study.
For the gating data to be valid, the kinetics of calcium binding to the indicator should be fast and the relaxation time should be short on the NMR time scale. These conditions were met in these experiments. First, the lifetime of the Ca 2ϩ -chelator complex is very short (ϳ33 s) (11). Second, the NMR pulse is also negligibly short (20 s). Third, the apparent transverse relaxation time is ϳ1.2 ms, so the acquisition time for the [Ca 2ϩ ] measurements was only 2 ms. Thus the time resolution should be Ͻ3 ms, which is Ͻ1% of the typical cardiac cycle.
Calculation of Free Energy (⌬G) for SR Ca 2ϩ -ATPase-We calculated the ⌬G required for the Ca 2ϩ -ATPase using Equation 2 (15), where R and T are the gas constant and temperature and we assume no membrane potential (16).
Calculation of Free Energy for ATP hydrolysis (⌬G ATP )-Three groups of rabbit hearts were studied. One group received 30 min of control aerobic perfusion (n ϭ 3); the second group was subjected to 15 min of control perfusion, followed by 15 min of high [K ϩ ] 0 perfusion (n ϭ 3); and the third group was subjected to 30 min of normothermic global ischemia (n ϭ 4). The high energy phosphates, P i and pH i , were monitored by 31 P NMR. At the end of the experiment, the heart was immediately frozen by freeze-clamping at the temperature of liquid nitrogen and ATP, creatine phosphate, and creatine contents were measured enzymatically following perchloric acid extraction (17).
The following equation was used to calculate ⌬G ATP , with ⌬G 0 ϭ -30.5 kJ/mol.
By thermodynamic convention values for ⌬G 0 and ⌬G ATP are negative for exergonic reactions, but after calculating ⌬G ATP we refer to the values in the text as absolute values. The [ATP] f /[ADP] f ratio was calculated by assuming that the creatine kinase reaction is at equilibrium. The apparent equilibrium constant (KЈ ck ) was calculated according to Equation 4 (18). 31 (Eq. 4) PCr and Cr contents were measured biochemically and converted to concentration by assuming that these metabolites are entirely cytosolic and that cytosolic volume equals 2.3 ml/g (dry weight). [P i ] was obtained by comparing the NMR peak area to that of the basal PCr peak, after correction for NMR saturation, and assuming that the P i peak is entirely cytosolic. Materials-Cyclopiazonic acid (CPA; Sigma) was dissolved in dimethyl sulfoxide (16.8 mg/200 l) and diluted in 1 liter of perfusate to a final concentration of 50 M. Caffeine (1.94 g/liter) and BDM (1.01 g/liter) were dissolved directly in the perfusate to a final concentration of 10 mM each.

RESULTS
Ionized free calcium in compartments that contain different calcium concentrations can be measured from the 19 F NMR spectra of TF-BAPTA loaded hearts from the shift difference between the Ca 2ϩ -insensitive 6F resonance and the Ca 2ϩ -sensitive 5F resonance of TF-BAPTA. In rat heart we observe one Ca 2ϩ -sensitive peak (ϳ5 ppm) corresponding to a time-average cytosolic free Ca 2ϩ concentration ([Ca 2ϩ ] c ) of ϳ600 nM (12). However, in rabbit heart, as shown in Fig. 1, the 19 F spectrum shows two calcium-sensitive resonances: a cytosolic free calcium resonance at ϳ 5 ppm, and an additional resonance peak at ϳ14 ppm that corresponds to a [Ca 2ϩ ] of ϳ1.5 mM. To assess if the ϳ14 ppm peak represents SR Ca 2ϩ , hearts were treated with a SR Ca 2ϩ -ATPase inhibitor or a SR Ca 2ϩ release channel activator or were arrested with 30 mM [K ϩ ] o to determine if these manipulations would alter the measured [Ca 2ϩ ] as would be expected if the peak reflected indicator in the SR.
FIG. 1. 19 F NMR spectrum (addition of spectra from four experiments), using 0.26-s intervals between scans with a pulse of 40 0 (20 s). The spectral width was Ϯ7060 Hz, and 4000 data points were collected. The free-induction decay was multiplied by an exponential function corresponding to 100 Hz-line broadening.
If the resonance at ϳ14 ppm originates from the SR, then perfusion with 50 M CPA, a specific SR Ca 2ϩ -ATPase inhibitor (19), should decrease [Ca 2ϩ ] SR . Consistent with our expectation, after ϳ10 min of perfusion with CPA, there was an upfield shift of the resonance near 14 ppm, indicating a decrease in [Ca 2ϩ ] SR (Fig. 2a). As shown in the inset, the shift and broadening observed with CPA addition is consistent with heterogeneity of the SR Ca 2ϩ pool. The spectrum at 20 -25 min of CPA perfusion could be modeled, assuming that 66% of the SR pool had a [Ca 2ϩ ] of 100 M and 33% of the SR had a [Ca 2ϩ ] of 260 M. The change in contractility observed on addition of CPA is typical of SR Ca 2ϩ -ATPase inhibition, i.e. a moderate decline in left ventricular developed pressure to 80 Ϯ 14 mm Hg compared to the initial level of 106 Ϯ 16 mm Hg, and a moderate increase in end diastolic pressure (see Fig. 2b).
We further tested whether SR Ca 2ϩ could be depleted by the addition of caffeine, which activates the SR Ca 2ϩ -dependent Ca 2ϩ release channel (20). To minimize the caffeine-induced contracture so that we could maintain tissue perfusion in the presence of caffeine, 10 mM BDM was administered prior to caffeine (21). Subsequent combined perfusion with 10 mM caffeine ϩ BDM resulted in a moderate increase in end diastolic pressure, while left ventricular developed pressure was reduced to 16.2 Ϯ 3.1 mm Hg compared to the initial level of 101 Ϯ 6 mm Hg. As shown in Fig. 3, addition of caffeine resulted in a loss of the resonance at 14.3 ppm, presumably due to lowering of [Ca 2ϩ ] SR and exchange broadening of the peak. Also shown in Fig. 3, when caffeine and BDM were washed out, the resonance at ϳ14 ppm returned, as did contractility, consistent with an SR location for the calcium pool at ϳ14 ppm. This broadening of the ϳ14 ppm resonance during caffeine perfusion followed by the reappearance of a sharp resonance during caffeine washout indicates that the ϳ14 ppm resonance is not due to extracellular indicator, and that the caffeine effect is not due to loss of indicator from the SR.
We were also interested in testing whether this resonance at ϳ14 ppm could be shifted by perfusion with high [K ϩ ] o (30 mM), which depolarizes the myocytes, arrests the heart, and increases diastolic [Ca 2ϩ ] c (22). If the peak at ϳ14 ppm arises from SR Ca 2ϩ , one might expect a downfield shift on perfusion with high [K ϩ ] o . As demonstrated in Fig. 4 After confirming that the resonance at ϳ14 ppm originates from SR, we investigated the magnitude of the change in [Ca 2ϩ ] SR during the cardiac cycle, which we thought might have a similar time course to the cytosolic Ca 2ϩ transient. This is done by gating the NMR pulse and acquisition to the contraction cycle. As shown in Fig. 5a, [Ca 2ϩ ] SR was measured at the start of systole (TS 1) and after the heart had fully relaxed (TS 2). The corresponding spectra are shown in Fig. 5b 2. a, typical 19 F NMR spectra from a heart perfused with 50 M CPA, a Ca 2ϩ -ATPase inhibitor. This is representative of three experiments. During 5-10 min of CPA perfusion, the peak that was initially located at 14.4 ppm is inhomogeneously shifted upfield to ϳ13 ppm with a marked broadening, corresponding to a decline in [Ca 2ϩ ] SR from 1.5 mM to ϳ300 M. As shown in the inset, this could be modeled assuming that there are two separate pools of calcium, as discussed in the text. b, LVP of a perfused rabbit heart before and during perfusion with 50 M CPA.
FIG. 3. Typical 19 F NMR spectra from a heart perfused with 10 mM caffeine, a SR Ca 2؉ release channel activator, in the presence of 10 mM BDM. This is representative of three experiments. The peak at 14.3 ppm becomes so broad that it is virtually undetectable after 25 min of caffeine perfusion. During the washout period without drugs, the peak returns to very near the original position, corresponding to a [Ca 2ϩ ] SR (22), and assuming no membrane potential across the sarcoplasmic reticulum (16), the free energy (⌬G) required for the SR Ca 2ϩ -ATPase is ϳ49.5 kJ/mol. Using metabolites measured in parallel perfused rabbit hearts (Table I), we have calculated the ⌬G for ATP hydrolysis (⌬G ATP ) to be 59 kJ/mol (Table II), consistent with the reported values in the range of 55-60 kJ/mol (9,23). After 30 min of total ischemia, we find that ⌬G ATP falls to Ͻ48 kJ/mol, as shown in Table II. A ⌬G ATP of Ͻ48 kJ/mol would not be sufficient to support the SR/cytosol calcium gradient of 1.5 ϫ 10 4 measured under basal conditions; this might result in reversal of the Ca 2ϩ -ATPase and net calcium release from the SR during ischemia. To investigate how the decrease in the free energy of ATP hydrolysis that occurs during ischemia affects the SR calcium gradient, we subjected hearts to ischemia and determined the effect on [Ca 2ϩ ] SR . As illustrated in Fig. 6a, even at 25-30 min of ischemia, the chemical shift of the SR Ca 2ϩ peak is unchanged. If we assume that pH in the SR is the same as pH i , which declined progressively (to be 6.08 Ϯ 0.10 at the end of 30 min of ischemia, n ϭ 4), the calculated [Ca 2ϩ ] SR was not changed significantly (Fig. 6b), despite a marked reduction in ⌬G ATP , which we measured to be Ͻ48 kJ/mol. However, there is a substantial decrease in the SR/cytosol calcium gradient, due to the increase in [Ca 2ϩ ] c to a value of ϳ 3 M (Fig. 6c), which reduces the energy requirement (⌬G) for the Ca 2ϩ -ATPase during ischemia, to a value of 32 kJ/mol after 30 min of ischemia. Thus with a decrease in the energy available from ATP hydrolysis during ischemia, the SR/cytosol calcium gradient is not maintained as [Ca 2ϩ ] c increases. This is in contrast to the effect of cardiac arrest with high [K ϩ ] o perfusion, where an increase in [Ca 2ϩ ] c (from ϳ100 nM to ϳ350 nM) is accompanied by an increase in [Ca 2ϩ ] SR (from ϳ1.5 mM to ϳ5 mM). The ⌬G for ATP hydrolysis does not decrease under these conditions, and therefore the SR/cytosol calcium gradient can remain constant. These data suggest that a rise in cytosolic free calcium would normally be associated with an increase in [Ca 2ϩ ] SR ; however, this does not occur during ischemia, possibly due to the decline in ⌬G ATP . Although the free energy available to pump calcium into the SR is reduced during ischemia, the ⌬G ATP is adequate to prevent net release of SR calcium during 30 min of total ischemia in the isolated rabbit heart. 5. a, simultaneous recording of LVP, dP/dt, and TS. Panel A shows the location of the NMR triggering signal at the start of systole (TS 1) and panel B shows the location of the NMR triggering signal during diastole (TS 2). b, typical 19 F NMR spectra from a heart triggered at systole (TS 1) and at diastole (TS 2). This is representative of three experiments. The chemical shift of the SR Ca 2ϩ resonance at diastole is ϳ0.2 ppm upfield compared to the chemical shift at systole, corresponding to a decrease in [Ca 2ϩ ] SR of ϳ0.5 mM (ϳ30%) at the start of systole.  4. Typical 19 F NMR spectra from a heart perfused with 30 mM [K ؉ ] 0 in the perfusate. This is representative of three experiments. The peak at 14.4 ppm is shifted to 14.7 ppm, corresponding to an increase in [Ca 2ϩ ] SR from 1.5 mM to 5 mM. DISCUSSION We report a value for time-average [Ca 2ϩ ] SR of 1.5 mM measured in the beating perfused rabbit heart under basal conditions. When the [Ca 2ϩ ] SR measurement was gated to the contraction cycle, the diastolic value was essentially the same as the time-average value. This value agrees well with calculated values based on estimates of total calcium and calsequestrin binding sites (7). Calsequestrin, the major calcium-binding protein in the SR has a K d for Ca 2ϩ in the range of 1 mM (24 -26). Thus, the [Ca 2ϩ ] SR values measured in this study suggest that [Ca 2ϩ ] SR is being maintained near the K d of calsequestrin. These data also allow calculation of the free energy requirement of the SR Ca 2ϩ -ATPase. The data in this report suggest a calcium gradient across the SR of 1.5 ϫ 10 4 at diastole, and assuming no membrane potential, this would correspond to a ⌬G for the Ca 2ϩ -ATPase of 49.5 kJ/mol, a value that agrees well with that estimated previously (6 -8). We further show that high [K ϩ ] o perfusion, which we have shown previously (22) to increase diastolic [Ca 2ϩ ] c from ϳ100 nM to 350 nM, increases the [Ca 2ϩ ] SR to ϳ5 mM; this corresponds to a ⌬G for the Ca 2ϩ -ATPase of 49.3 kJ/mol. Thus, perfusion with 30 mM potassium elevates both cytosolic and SR [Ca 2ϩ ], but the ⌬G for the Ca 2ϩ -ATPase is unchanged. We also show that during ischemia there is a parallel decline in the free energy of ATP hydrolysis and the energy requirement of the SR Ca 2ϩ -ATPase. These data suggest that there is tight coupling between the SR calcium gradient and the free energy of ATP hydrolysis, and that the Ca 2ϩ -ATPase works at near its theoretical limit.
SR calcium appears to be well buffered during a normal calcium transient. The difference in [Ca 2ϩ ] SR between the start of systole and diastole suggests that calcium release from the SR to trigger contraction causes a transient decrease in [Ca 2ϩ ] SR of ϳ30%. It has been reported that the fractional SR calcium release varies with SR calcium load and trigger calcium, but an estimate obtained under physiologic conditions suggests that ϳ35% of the SR calcium content may be released with each twitch (27). This is similar to an earlier report that only about half of SR Ca 2ϩ is released during a twitch (5). The relatively small and transient nature of the changes in [Ca 2ϩ ] SR during the contraction cycle is also indicated by the sharpness of the ϳ14 ppm resonance observed in the timeaverage measurement of [Ca 2ϩ ] SR in the beating heart, which is similar to that observed in hearts arrested with high [K ϩ ] o . The sharp resonance also suggests that under basal conditions there is little spacial heterogeneity in [Ca 2ϩ ] SR .
The modeling in the inset of Fig. 2 shows that the spectra observed after addition of CPA are consistent with a heterogeneous loss of calcium from the SR network. One possible explanation for this observation is that there are distinct types of SR, which may respond differently to SR calcium-ATPase inhibitors. Another possibility is that the signals were obtained from the whole heart, which displays a great deal of heterogeneity in cellular structure and function. In support of the first concept, Jorgensen et al. (28) showed that the total SR calcium measured by electron probe microanalysis (EPMA) varied within subpopulations of SR. They suggest that Ca 2ϩ is accumulated into the network SR, which immunoelectron microscopic studies show have Ca 2ϩ -ATPase, which is apparently absent from the junctional and corbular SR. They further suggest that Ca 2ϩ accumulated by the network SR is then transferred and stored by the junctional and corbular SR which contain large amounts of calsequestrin. Further support for distinct populations of SR comes from a study by Kijima et al. (29), which reports different intracellular localization of inositol 1,4,5trisphosphate and ryanodine receptors in cardiomyocytes. The results would also be consistent with different leak rates from different parts of the SR network. The data in this report suggest that different pools of SR respond differently to CPA and caffeine.
One question that we have not yet addressed is the possible contribution of mitochondrial  (30,31). Thus, if TF-BAPTA loads into mitochondria, the mitochondrial peak might not be distinguishable from the cytosolic peak and would be expected to be well separated from the putative SR peak at ϳ14 ppm. While excitation-contraction coupling of normal rabbit myocytes under physiologic conditions is heavily dependent on calcium fluxes across the plasma membrane and the SR membrane, with relatively little contribution from the mitochondria (3), under conditions where SR function is inhibited by caffeine, mitochondria can be involved in calcium removal from the cytosol (3). However, in our experiments, we never observed a separate resonance or a shoulder on the ϳ5 ppm peak that could be attributed to the mitochondria. This reflects either the lack of indicator loading into the mitochondria or very tight coupling between [Ca 2ϩ ] c and [Ca 2ϩ ] m . The total calcium content of the SR and the percentage released during an action potential, tetanus, or other interventions have been measured by EPMA and estimated by caffeine release and rapid cooling contractures, using fluorescent calcium indicators to measure the change in [Ca 2ϩ ] c . The EPMA studies of frog skeletal muscle suggest that the resting content of total calcium in the terminal cisternae (TC) is ϳ120 mmol/kg (dry weight) of TC; a 1.2-s tetanus releases ϳ70 mmol/kg (dry weight) (32). Similarly, addition of caffeine reduces TC calcium to ϳ40 mmol/kg (dry weight) (33). Thus these studies show that greater than half the total TC calcium is released during tetanus or addition of caffeine in skeletal muscle. Using either caffeine or rapid cooling contractures, estimates of the percentage of SR calcium release have also been made in cardiac muscle. Baro et al. (34) report that caffeine released ϳ71% of the total calcium in cardiac muscle. The data presented here are generally in agreement with this estimate of caffeine-releasable calcium. We find that [Ca 2ϩ ] SR decreases by more than 80% upon addition of CPA or caffeine.
The EPMA data also suggest that tetanus or caffeine addition results in a ϳ50% increase in total SR magnesium (16,32,33). The increase in free magnesium is likely to be much less than the increase in total magnesium, due to buffering. However, assuming that caffeine results in a 50% increase in free SR magnesium, this would increase our calculated [Ca 2ϩ ] SR by only 0.01 mM.
In summary, several conclusions can be drawn from the data in this report. First, the time-average [Ca 2ϩ ] SR in the beating rabbit heart is ϳ1.5 mM, which is not significantly different than [Ca 2ϩ ] SR measured at diastole. Second, [Ca 2ϩ ] SR decreases by ϳ30% at the start of systole but this decrease is very brief, with ϳ50% recovery in 10 ms. Third, there is little spacial heterogeneity in [Ca 2ϩ ] SR in the beating heart. Fourth, there may be a heterogeneous response to caffeine and CPA consistent with previous studies suggesting that there are subpopulations of SR with different calcium handling characteristics. Fifth, the data are consistent with tight coupling between the SR Ca 2ϩ gradient and the free energy of ATP hydrolysis.