Systolic [Ca2+]i regulates diastolic levels in rat ventricular myocytes

Key points For the heart to function as a pump, intracellular calcium concentration ([Ca2+]i) must increase during systole to activate contraction and then fall, during diastole, to allow the myofilaments to relax and the heart to refill with blood. The present study investigates the control of diastolic [Ca2+]i in rat ventricular myocytes. We show that diastolic [Ca2+]i is increased by manoeuvres that decrease sarcoplasmic reticulum function. This is accompanied by a decrease of systolic [Ca2+]i such that the time‐averaged [Ca2+]i remains constant. We report that diastolic [Ca2+]i is controlled by the balance between Ca2+ entry and Ca2+ efflux during systole. The results of the present study identify a novel mechanism by which changes of the amplitude of the systolic Ca transient control diastolic [Ca2+]i. Abstract The intracellular Ca concentration ([Ca2+]i) must be sufficently low in diastole so that the ventricle is relaxed and can refill with blood. Interference with this will impair relaxation. The factors responsible for regulation of diastolic [Ca2+]i, in particular the relative roles of the sarcoplasmic reticulum (SR) and surface membrane, are unclear. We investigated the effects on diastolic [Ca2+]i that result from the changes of Ca cycling known to occur in heart failure. Experiments were performed using Fluo‐3 in voltage clamped rat ventricular myocytes. Increasing stimulation frequency increased diastolic [Ca2+]i. This increase of [Ca2+]i was larger when SR function was impaired either by making the ryanodine receptor leaky (with caffeine or ryanodine) or by decreasing sarco/endoplasmic reticulum Ca‐ATPase activity with thapsigargin. The increase of diastolic [Ca2+]i produced by interfering with the SR was accompanied by a decrease of the amplitude of the systolic Ca transient, such that there was no change of time‐averaged [Ca2+]i. Time‐averaged [Ca2+]i was increased by β‐adrenergic stimulation with isoprenaline and increased in a saturating manner with increased stimulation frequency; average [Ca2+]i was a linear function of Ca entry per unit time. Diastolic and time‐averaged [Ca2+]i were decreased by decreasing the L‐type Ca current (with 50 μm cadmium chloride). We conclude that diastolic [Ca2+]i is controlled by the balance between Ca entry and efflux during systole. Furthermore, manoeuvres that decrease the amplitude of the Ca transient (without decreasing Ca influx) will therefore increase diastolic [Ca2+]i. This identifies a novel mechanism by which changes of the amplitude of the systolic Ca transient control diastolic [Ca2+]i.


Introduction
For the heart to function as a pump, the intracellular Ca concentration ([Ca 2+ ] i ) must increase during systole to activate contraction and then fall, during diastole, to levels that are sufficiently low to allow the myofilaments to relax and the heart to refill with blood. Although the mechanisms that control the amplitude of the systolic rise of [Ca 2+ ] i are well understood (Bers, 2001), the regulation of the diastolic level is obscure (Louch et al. 2012). One problem is that much of the work studying [Ca 2+ ] i has used unphysiologically slow rates of stimulation. Under these conditions, a steady-state is reached where Ca entry into the cell balances efflux and there is no flux into and out of the sarcoplasmic reticulum (SR) (Allen et al. 1984;Rios, 2010). By contrast, at faster rates, fluxes of Ca into and out of the SR will play an important role. Indeed, diastolic [Ca 2+ ] i increases with increasing rate (Layland & Kentish, 1999;Dibb et al. 2007).
An elevation of diastolic [Ca 2+ ] i has been reported in some studies of heart failure (Gwathmey et al. 1987;Beuckelmann et al. 1992;Sipido et al. 1998;Fischer et al. 2013). The frequency-dependent increase of diastolic [Ca 2+ ] i (Gwathmey et al. 1991) and force (Pieske et al. 2002) is increased in heart failure in humans and this may contribute to the phenomenon of 'diastolic heart failure' (Selby et al. 2011). Raised diastolic [Ca 2+ ] i also increases Ca leak from the SR by increasing efflux of Ca through the ryanodine receptor (RyR) (Bovo et al. 2011) and raised levels as a result of RyR leak have been suggested to initiate a vicious cycle by increasing this leak further (Louch et al. 2012).
End diastolic [Ca 2+ ] i presumably depends on a combination of fluxes as a result of both sarcolemmal and SR Ca handling proteins. Heart failure impairs SR function as a result of a combination of a decrease of SERCA activity and increased RyR leak (Lou et al. 2012). Previous work has shown that decreasing SERCA activity elevates diastolic Ca (Negretti et al. 1993). In addition, making the RyR leaky with ryanodine elevates [Ca 2+ ] i , particularly at high stimulation rates (Gao et al. 1995). However, the exact mechanism by which alterations of SR function affect diastolic [Ca 2+ ] i is unresolved. The aim of the present study was to investigate, quantitatively, the effects of interfering with the SR on diastolic [Ca 2+ ] i . We find that decreasing SR function decreases systolic and increases diastolic [Ca 2+ ] i . Importantly, there is no effect on the average level of [Ca 2+ ] i . We conclude that the increase of diastolic [Ca 2+ ] i is a consequence of the decrease of systolic [Ca 2+ ] i decreasing the efflux of Ca from the cell, and therefore that systolic [Ca 2+ ] i plays a major role in controlling diastolic [Ca 2+ ] i .

Ethical approval
Animals were cared for and used in accordance with The UK Animals (Scientific Procedures) Act, 1986 and Directive 2010/63/EU of the European Parliament. The experiments were approved by the University of Manchester Ethical Review Board. Male Wistar rats (weighing ß200-250 g) were killed by stunning and cervical dislocation. Single ventricular myocytes were isolated by digestion with collagenase and protease as described previously (Eisner et al. 1989).
Isolated myocytes were superfused with a solution (control) consisting of (in mM) 135 NaCl, 11.1 glucose, 1 CaCl 2 , 10 Hepes, 1 MgCl 2 and 4 KCL. Then, 4-aminopyridine (5 mM) and BaCl 2 (0.1 mM) were added to inhibit K + currents and the solution was titrated to pH 7.4 using NaOH. Probenecid (2 mM) was added to reduce loss of indicator from the myocytes. Micropipettes (< 5 M ) were filled with a solution consisting of (in mM): 125 KCH 3 O 3 S, 12 KCl, 10 NaCl, 10 Hepes, 5 MgCl 2 and 0.1 EGTA; titrated to pH 7.2 with KOH; and a final concentration of amphotericin B of 240 g ml −1 . Cells were voltage clamped with the perforated patch clamp technique using the discontinuous switch clamp mode (frequency 1-2 kHz and gain 0.3-0.7 nA mV −1 ) of an Axoclamp 2A voltage clamp amplifier (Molecular Devices, Union City, CA, USA). Cells were voltage clamped and stimulated at a range of frequencies (0.2 to 3 Hz) with a 40 mV, 100 ms duration pulse from a holding potential of −40 mV. All experiments were performed at room temperature.

[Ca 2+ ] i measurements
Cells were incubated with the acetoxymethyl ester of Fluo-3 (5 μM for 10 min) and allowed to de-esterify before use. An aliquot was then placed in a superfusion chamber mounted on the stage of an inverted fluorescence microscope. To measure changes of [Ca 2+ ] i , at the end of each experiment, the maximum fluorescence (F max ) was measured by damaging the cell with the patch pipette. The dissociation constant of Fluo-3 (K d ) was taken as 864 nM (Cheng et al. 1993) and [Ca 2+ ] i , calculated as described previously (Trafford et al. 1999 ] i was calculated as the mean level from one stimulus to the next. In some experiments, SR Ca content was estimated by releasing Ca from the SR using a mixture of 5 mM caffeine and 20 mM 2,3-butanedione monoxime (Kashimura et al. 2010).
The Ca influx through the L-type Ca current was calculated by integrating the Ca current (Venetucci et al. 2007). All analysis was performed using custom-written software (Greensmith, 2014).
All chemicals were obtained from Sigma-Aldrich (Poole, UK), R&D Systems (Abingdon, UK) or Fisher Scientific (Loughborough, UK). Caffeine was added as required. Ryanodine and thapsigargin were both stored as 1 mM stock solutions in DMSO and made up to a concentration of 1 μM before use. Thapsigargin was dissolved in DMSO and stored as a 1 mM stock solution and used at a final concentration of 1 μM.

Statistical analysis
Data are reported as the mean ± SEM where applicable for descriptive analysis. Statistical comparisons were made using two-way ANOVA. Regression lines were compared with an F test. P < 0.05 was considered statistically significant.

Effects of RyR leak on diastolic and systolic [Ca 2+ ] i
The first series of experiments examined the effect of increased RyR leak on the response to an increased stimulation rate. The grey traces in Fig. 1A

The frequency-dependence of average [Ca 2+ ] i
Because the data of Fig. 1 show that the effects of caffeine on diastolic [Ca 2+ ] i were more prominent in the presence of ISO, all subsequent experiments were performed in the presence of ISO. The experiment shown in Fig. 2 was designed to investigate Ca handling over a wider range of frequencies. Figure 2A shows the data obtained in the presence and absence of caffeine. As the frequency of stimulation increased, the Ca transient amplitude decreased slightly and diastolic [Ca 2+ ] i increased. The decrease in Ca transient amplitude was accompanied by (and is presumably at least in part caused by) a decrease in the amplitude of the L-type Ca current (Fig. 2C) (Antoons et al. 2002;Dibb et al. 2007). The effects of frequency were increased by caffeine. The continuous line in Fig. 2A shows, again, that average [Ca 2+ ] i was unaffected by caffeine and was increased by increasing frequency. Notably, the increase of average [Ca 2+ ] i was a saturating function of frequency as shown by the fact that raising rate from 0.5 to 1 Hz had a larger effect than that from 2 to 3 Hz. The above observations of the frequency dependence of average [Ca 2+ ] i are emphasized by the mean data shown in Fig. 2B. Caffeine decreased the amplitude of the Ca transient at all frequencies. An increase of frequency increased diastolic [Ca 2+ ] i ; this effect is much more obvious in the presence of caffeine. By contrast to the marked effects of caffeine on both diastolic and systolic [Ca 2+ ] i , average [Ca 2+ ] i was unaffected by caffeine (ANOVA, P > 0.5 at all frequencies). Average [Ca 2+ ] i did, however, increase in a saturating manner with an increasing frequency of stimulation. Figure 2B sheds light on this saturation of average Ca. The Ca influx via the L-type Ca current on each pulse decreased with an increasing stimulation rate (Fig. 2C) as a result of increasing inactivation (Antoons et al. 2002;Dibb et al. 2007). Consequently, the Ca influx per unit time (Fig. 2B, bottom) was a saturating function of frequency and therefore paralleled the frequency dependence of average [Ca 2+ ] i . Data obtained in the presence of caffeine (Fig. 2B, red symbols) were identical to those in its absence (ANOVA, P > 0.5 at all frequencies). The correlation between Ca entry per unit time and average [Ca 2+ ] i is emphasized by Fig. 2D, which shows a clear linear relationship between these two parameters that is not statistically different (F test; P = 0.064) in the absence and presence of caffeine.

The effects of thapsigargin and ryanodine
The purpose of the experiments shown in Figs 3 and 4 was to examine whether the above observations were specific to caffeine or were a general consequence of interfering with RyR function. In the experiment shown in Fig. 3A, the application of thapsigargin decreased systolic [Ca 2+ ] i and increased diastolic [Ca 2+ ] i during stimulation at 0.5 Hz.
On increasing the stimulation rate to 2 Hz, there was a more marked increase of diastolic and decrease of systolic [Ca 2+ ] i . When stimulation was stopped, [Ca 2+ ] i declined to a level similar to the original diastolic one. The mean data of Fig. 3B show the effects of thapsigargin at 0.5 and 2 Hz. (Because thapsigargin is irreversible, it was not feasible to study the full range of frequencies used for caffeine). Thapsigargin decreased the amplitude and   (to 273 ± 3.1%, P < 0.001), whereas there is no effect on average [Ca 2+ ] i (102.8 ± 1.4%; P = 0.17).

Effects of decreasing Ca influx
The above experiments suggest that the level of diastolic [Ca 2+ ] i depends on a balance between Ca influx and efflux. If this is the case, diastolic [Ca 2+ ] i would be expected to be decreased by reducing Ca influx. In the experiment shown in Fig. 5A, caffeine had been added when stimulating at either 0.5 or 3 Hz. In agreement with the results reported above, the rise of diastolic [Ca 2+ ] i was greater at the higher frequency. Addition of cadmium chloride (50 μM) to decrease the L-type Ca current was observed to decrease diastolic [Ca 2+ ] i at both stimulation rates (Fig. 5B). On average, cadmium decreased the L-type Ca current to 33 ± 5% (n = 5 cells). Figure 5C shows the mean data obtained from five cells (at 0.5 Hz).  (Hansford & Lakatta, 1987;Negretti et al. 1993). In the present study, we clarified the mechanism of this effect and found that the reduction in Ca transient caused by increasing leak with caffeine or ryanodine was always associated with an increase in diastolic [Ca 2+ ] i . The same behaviour was observed after the application of thapsigargin to inhibit SERCA. In other words, decreased systolic [Ca 2+ ] i associated with increased diastolic [Ca 2+ ] i is a distinguishing feature of impaired SR function, no matter whether this results from increased RyR leak or decreased SERCA activity. In the absence of stimulation, interfering with the SR had no effect on the level of resting [Ca 2+ ] i (Figs 1, 2 and 4). This is to be expected because, under these conditions, [Ca 2+ ] i is presumably determined solely by the surface membrane (Allen et al. 1984;Rios, 2010). The increase of diastolic [Ca 2+ ] i at elevated frequencies results from the fact that the next transient begins before the previous one has relaxed. Increasing leak or decreasing SERCA slows the rate of relaxation of the Ca transient (Negretti et al. 1993;Belevych et al. 2007;Sankaranarayanan et al. 2016   We suggest that the answer resides in the mechanisms that maintain Ca flux balance. In the steady-state, during each cycle of stimulation, the influx mediated by the Ca current must be precisely equal to the Ca efflux, largely via sodium-calcium exchange (NCX) (Eisner et al. 2013). The activity of NCX depends on [Ca 2+ ] i . Increasing SR leak or decreasing SERCA activity will decrease the SR Ca content and therefore the amplitude of the systolic Ca transient. This, in turn, will decrease Ca efflux to a level less than the influx. Consequently, the cell will gain Ca and, because the SR is compromised, much of this Ca will remain in the cytoplasm, increasing diastolic [Ca 2+ ] i . This elevated diastolic [Ca 2+ ] i produces more Ca efflux and compensates for the loss of efflux associated with the systolic transient (Dibb et al. 2007). If we assume that NCX activity is proportional to [Ca 2+ ] i then the Ca efflux per cycle will be proportional to average [Ca 2+ ] i . If Ca influx is unaffected, then the need for constant efflux requires that average [Ca 2+ ] i be constant and therefore the decrease of systolic [Ca 2+ ] i must be balanced by an increase of diastolic such that average [Ca 2+ ] i remains constant.
This consideration of flux balance is a more complicated and general version of previous work showing that potentiating the opening of the RyR with low concentrations of caffeine leads to a transient increase of the amplitude of the systolic Ca transient (Trafford et al. 2000;Greensmith et al. 2014). In the steady-state, however, the amplitude of the Ca transient was the same as in control. Under the conditions of those experiments (performed at low rates of stimulation), diastolic [Ca 2+ ] i did not change and therefore maintenance of flux balance required that systolic [Ca 2+ ] i was constant. In the present experiments, the changes of systolic [Ca 2+ ] i required that diastolic [Ca 2+ ] i change to maintain flux balance.

The effects of β-adrenergic stimulation on [Ca 2+ ] i
The above analysis also explains why ISO increases average [Ca 2+ ] i . ISO will increase Ca entry through the L-type current and this will have to be balanced by increased efflux on NCX. This increased efflux can be achieved by an increase of average [Ca 2+ ] i . The exact circumstances will determine whether the increase of average [Ca 2+ ] i results primarily from a rise of diastolic as opposed to systolic [Ca 2+ ] i . For example (Fig. 1), with normal SR function, the increase of the amplitude of the systolic Ca transient is sufficiently large that diastolic [Ca 2+ ] i does not increase. By contrast, when the SR is partly disabled, systolic [Ca 2+ ] i cannot increase sufficiently and a rise of diastolic [Ca 2+ ] i ensues (Fig. 1D). It is also possible that an increase of Ca leak from the SR, possibly via a Ca/calmodulin-dependent kinase II (CaMKII) mechanism (Curran et al. 2007), contributes to the increase of diastolic [Ca 2+ ] i . In this context, it is worth noting that, in mice overexpressing CaMKII, the rise of systolic [Ca 2+ ] i produced by ISO was less than in wild-type. However, ISO produced a larger increase of diastolic [Ca 2+ ] i in the overexpressing mice compared to the control (Sag et al. 2009). Given that these transgenic mice have elevated SR Ca leak, the reciprocal effect on diastolic and systolic [Ca 2+ ] i is consistent with the conclusions of the present study. It should also be noted that the normal inotropic response to ISO, resulting from an increase of systolic [Ca 2+ ] i with no change of diastolic [Ca 2+ ] i , requires a normal, functional SR. Because SR activity is compromised, an increase of diastolic [Ca 2+ ] i will occur.

The effects of stimulation frequency
An increase of stimulation frequency will increase Ca influx per unit time, thereby requiring an increase of average [Ca 2+ ] i to maintain flux balance. From first principles, this can be achieved by an increase of either or both diastolic or systolic [Ca 2+ ] i . In the present experiments, performed on rat myocytes, increased frequency decreases systolic [Ca 2+ ] i and therefore systolic efflux. As the frequency is increased, the cell therefore faces three challenges: (i) increased Ca influx per unit time (Fig. 2B); (ii) decreased systolic efflux; and (iii) decreased diastolic time for efflux to occur in. Ca flux balance can therefore only be established with an increase of diastolic [Ca 2+ ] i . If SR function is depressed, increased frequency produces a larger decrease of systolic [Ca 2+ ] i and therefore a larger rise of diastolic [Ca 2+ ] i would be expected. In agreement with these predictions, we find that increasing frequency increases diastolic [Ca 2+ ] i and this increase is potentiated by increasing SR leak or decreasing SERCA activity. Average [Ca 2+ ] i is a saturating function of frequency (Fig. 2B) presumably because the Ca entry per unit time also saturates with frequency as a result of increased inactivation of the L-type Ca current (Fig. 2C), (Antoons et al. 2002;Dibb et al. 2007). Consistent with this, average [Ca 2+ ] i is a linear function of Ca influx per unit time through the L-type Ca current (Fig. 2D). It should be noted that, if the L-type Ca current did not decrease at higher frequencies, the rise of diastolic [Ca 2+ ] i would be even greater.
One final conclusion can be derived from Fig. 2D. It is clear that, even with zero influx through the L-type Ca current, [Ca 2+ ] i has a finite value. This has been accounted for by a background Ca entry (Choi et al. 2000;Kupittayanant et al. 2006) which is unaffected by stimulation rate. The value of this background flux, estimated from the horizontal intercept of Fig. 2D is of the order of 4 μmol l −1 s −1 . The existence of this background flux may also be relevant to the effects of cadmium. We found that 50 μM cadmium decreased Ca influx to 33% but average [Ca 2+ ] i fell to only 70%. This discrepancy can be accounted for if the background flux is unaffected by cadmium.

Limitations
It should be noted that, in these experiments, we used a holding potential of −40 mV to inactivate the Na + current. This holding potential will decrease the L-type Ca current and lead to an underestimate of the effects of frequency on diastolic [Ca 2+ ] i (Dibb et al. 2007). The removal of Na + current might be expected to decrease the frequency-dependent increase of intracellular Na + concentration but, because Na + influx through Na channels is quantitatively smaller than that through NCX , this may not be a major issue. Previous work has measured systolic and diastolic [Ca 2+ ] i in rat ventricular myocytes excited with physiological action potentials. An increase of frequency increased both diastolic and systolic [Ca 2+ ] i indicated that a frequency-dependent increase of average [Ca 2+ ] i is also seen with more physiological stimulation (Dibb et al. 2007). As a more general point, the above discussion assumes that the only factor regulating NCX is [Ca 2+ ] i . It therefore ignores the effects that changes of intracellular sodium concentration ([Na + ] i ) may have on NCX activity. [Na + ] i will be increased by an increase of stimulation rate and as a consequence of the increased NCX activity in response to the Ca loading produced by β-adrenergic stimulation. It will be decreased as a consequence of phosphorylation of phospholemman and stimulation of the sodium pump . It should also be noted that the above discussion is based on the assumption that NCX activity is proportional to [Ca 2+ ] i . This will only be true over a certain range and, at higher [Ca 2+ ] i , efflux will tend towards saturation. In this case, when SR function is decreased, a reduction of average [Ca 2+ ] i might be expected. Any such effect is below the resolution of the experiments. Finally, although, in the present experiments, Ca influx through the L-type current was unaffected by altering SR function with caffeine, more generally, it is possible that changes of Ca-dependent inactivation may affect the L-type Ca current and this would need to be allowed for,

Relationship with disease
Some previous work has shown that heart failure results in an increase of diastolic force and/or [Ca 2+ ] i at elevated rates of stimulation (Sipido et al. 1998;Baartscheer et al. 2003;Selby et al. 2011). One explanation for this is the measured increase of [Na + ] i which will decrease Ca efflux on NCX (Pieske et al. 2002). The results of the present study suggest an additional explanation for the rise of diastolic [Ca 2+ ] i . It is known that heart failure is often associated with increased RyR leak (Marks, 2000;Marx et al. 2000;Shannon et al. 2003;Terentyev et al. 2008;Belevych et al. 2013) and decreased SERCA activity (Nagai et al. 1989;Mercadier et al. 1990;Hasenfuss et al. 1994).
As a consequence of the resulting decrease of SR Ca content and hence systolic [Ca 2+ ] i , these changes would be expected to also elevate diastolic [Ca 2+ ] i . Indeed, both this mechanism and the increase of [Na + ] i will decrease Ca efflux and increase diastolic [Ca 2+ ] i .
A common observation is that, in human heart failure, the reduction of the amplitude of the Ca transient is more marked at higher frequencies (Gwathmey et al. 1990;Mulieri et al. 1992). This is accompanied by removal of the increase of SR Ca content produced by increasing stimulation frequency (Lindner et al. 1998;Pieske et al. 1999). These effects have previously been attributed to a decrease of SERCA activity (Pieske et al. 1995). Our data suggest that Ca leak may also contribute to the loss of this frequency dependence in heart failure. The results of the present study are also relevant to the changes produced by increasing Ca buffering by the myofilaments as occurs in some case of hypertrophic cardiomyopathy. Such increased buffering slows the rate constant of decay of the Ca transient and increases diastolic [Ca 2+ ] i (Schober et al. 2012). The increased buffering will decrease the increase of [Ca 2+ ] i produced by a given total Ca release from the SR, thereby decreasing Ca efflux. This, and the slowed decay will elevate diastolic [Ca 2+ ] i .

Conclusion: systolic [Ca 2+ ] i controls diastolic [Ca 2+ ] i
The results of the present study show that the time-averaged level of [Ca 2+ ] i is an important factor in regulating Ca cycling. This average level determines the Ca efflux from the cell required to balance Ca influx. The total efflux can be considered to comprise two components: (i) a component activated by the diastolic level of [Ca 2+ ] i and (ii) an additional component that occurs during the systolic Ca transient (Dibb et al. 2007). Anything that decreases the amplitude of the systolic Ca transient, without affecting Ca influx, will decrease the systolic efflux, thereby requiring an increase of diastolic [Ca 2+ ] i to maintain Ca flux balance. In this way, the systolic Ca transient plays a vital role in regulating diastolic [Ca 2+ ] i .