Region-specific mechanisms of corticosteroid-mediated inotropy in rat cardiomyocytes

Regional differences in ion channel activity in the heart control the sequence of repolarization and may contribute to differences in contraction. Corticosteroids such as aldosterone or corticosterone increase the L-type Ca2+ current (ICaL) in the heart via the mineralocorticoid receptor (MR). Here, we investigate the differential impact of corticosteroid-mediated increase in ICaL on action potentials (AP), ion currents, intracellular Ca2+ handling and contractility in endo- and epicardial myocytes of the rat left ventricle. Dexamethasone led to a similar increase in ICaL in endocardial and epicardial myocytes, while the K+ currents Ito and IK were unaffected. However, AP duration (APD) and AP-induced Ca2+ influx (QCa) significantly increased exclusively in epicardial myocytes, thus abrogating the normal differences between the groups. Dexamethasone increased Ca2+ transients, contractility and SERCA activity in both regions, the latter possibly due to a decrease in total phospholamban (PLB) and an increase PLBpThr17. These results suggest that corticosteroids are powerful modulators of ICaL, Ca2+ transients and contractility in both endo- and epicardial myocytes, while APD and QCa are increased in epicardial myocytes only. This indicates that increased ICaL and SERCA activity rather than QCa are the primary drivers of contractility by adrenocorticoids.

www.nature.com/scientificreports/ such as dog or humans, I to magnitude exerts less influence on total APD, but rather controls the membrane voltage level during the early part of the plateau phase [17][18][19] . Since the AP constitutes the driving force for the ionic currents flowing during the AP, differences in AP waveform directly affect ion fluxes across the membrane. This is particularly important for the L-type Ca 2+ current, as the amount of Ca 2+ entering the myocytes during an AP controls excitation-contraction-coupling. For example, despite a similar I CaL magnitude in endo-and epicardial myocytes, the longer AP in endocardial myocytes leads to an increased AP-induced Ca 2+ influx 20 . Differences in duration and shape of the AP, therefore, not only control the sequence of repolarization but also affect Ca 2+ influx, intracellular Ca 2+ cycling, contractility and cardiac remodeling [21][22][23] . The particular relevance of the AP shape is highlighted by studies, in which epicardial myocytes displayed increases in Ca 2+ influx, Ca 2+ transients and contractility, when clamped on an endocardial AP and vice versa 24 . Moreover, alterations in AP waveform in ferret myocytes caused desynchronized SR Ca 2+ release which is typically observed in myocytes originating from failing hearts 25 . Although the magnitude of I to is the major parameter responsible for differences in AP shape and duration among different regions of the ventricular wall, it is important to note that the AP waveform of individual myocytes is a function of the cell's magnitude of both, I to and I CaL . In experiments, in which I to , I CaL and APD were measured in the same cells, Gomez and coworkers showed that early repolarization of the rat AP (APD 20 ) neither correlated with the magnitude of I CaL nor with I to alone, but only with both currents together 26 . This suggests that changes not only in the magnitude of I to , but also in I CaL might substantially affect AP waveform and in turn Ca 2+ influx, Ca 2+ transients and contractility.
In the present study, we therefore hypothesized that the corticosteroid induced increase in I CaL affects AP shape, Ca 2+ influx, Ca 2+ transient and contractility differently in endo-and epicardial myocytes. To address this question, we investigated the influence of mineralocorticoid receptor activation on I CaL , AP-induced Ca 2+ influx, contractility, Ca 2+ transient and SR Ca 2+ handling in endo-and epicardial myocytes isolated from the rat left ventricle. We show that MR activation substantially increases contractility to a similar extent in both endo-and epicardial myocytes but modulates intracellular Ca 2+ handling differently.

Results
Incubation of isolated left ventricular cardiomyocytes for 24 h with corticosteroids, such as the mineralocorticoid aldosterone or the glucocorticoid corticosterone, has been shown to increase the L-type Ca 2+ current (I CaL ) via activation of the mineralocorticoid receptor (MR) 8,27 . In the present study, we used dexamethasone as corticosteroid to stimulate I CaL . Moreover, in a previous study 27 , insulin was part of the primary culture conditions and we later learned that it is a prerequisite for I CaL regulation by corticosteroids: Fig. 1A displays typical recordings of I CaL obtained from a control myocyte, a myocyte incubated with 1 µM dexamethasone, a myocyte incubated with 100 nM insulin and a myocyte incubated with the combination of both dexamethasone and insulin. 24 h incubation with dexamethasone substantially increased I CaL only in the presence of insulin. The current-voltage (I-V) relations shown in Fig. 1B summarize similar experiments. On average, dexamethasone + insulin increased I CaL at 0 mV by 46% from 7.0 ± 0.5 pApF −1 (n = 21) to 10.2 ± 0.8 pApF −1 (n = 19, p < 0.01) while in the presence of dexamethasone (7.7 ± 0.6 pApF −1 , n = 19) or insulin (7.9 ± 0.8 pApF −1 , n = 15) alone, no significant difference was observed. For that reason we combined 1 µM dexamethasone with 100 nM insulin (DI) in all following experiments. To address the question whether DI increases I CaL via the MR or the glucocorticoid receptor (GR), we incubated myocytes with DI alone and with DI and the MR antagonist spironolactone (10 µM). 24 h incubation with DI substantially increased I CaL , while in the presence of spironolactone, I CaL remained at the level of the control myocytes (see Fig. 2A). The current-voltage (I-V) relations shown in Fig. 2B summarize similar experiments. Similar to the experiments shown in Fig. 1, DI increased I CaL by 46% from 7.6 ± 0.4 pApF −1 (n = 21) to 11.1 ± 0.6 pApF −1 (V Pip = 0 mV, n = 31, p < 0.001). In the presence of spironolactone, DI did not significantly affect I CaL , indicating that dexamethasone, like aldosterone or corticosterone 27 , increases I CaL via the MR. To investigate regional differences in the effects of corticosteroids on I CaL , we isolated endo-and epicardial myocytes from the left ventricular free wall and investigated I CaL after 24 h incubation with DI. Figure 3A shows average I-V relations of I CaL recorded from endo (left panel) and epicardial (right panel) myocytes. Compared to control, DI increased I CaL at 0 mV in endocardial and in epicardial myocytes to a similar extent by 45% (n = 37, p < 0.001) and 49% (n = 36, p < 0.001), respectively. We also investigated the effect of DI on repolarizing K + currents. The transient outward K + current (I to ) displayed the typical gradient among the left ventricular free wall with much larger currents in epi-compared to endocardial myocytes. However, in both endo-and epicardial myocytes, I to magnitude was not affected by 24 h incubation with DI (Fig. 3B). To address a potential effect of DI on the delayed rectifier group of K + currents (I K ), we used the current at the end of a 600 ms voltage pulse 15 . Figure 3C displays average I-V relations obtained by plotting the current at the end of a 600 ms voltage pulse versus the pulse potential. We did neither observe regional differences nor an effect of DI on I K .
Next, we recorded action potentials (AP) from endo-and epicardial myocytes. Figure 4A (upper panels) shows representative AP recordings obtained under control conditions (blue) and after 24 h incubation with DI (green). As reported previously, APs were much shorter in epicardial than in endocardial myocytes. DI substantially increased APD in epicardial but not in endocardial myocytes. On average, in epicardial myocytes APD at 90% repolarization (APD 90 ) increased from 90.2 ± 11.4 to 220.2 ± 31.4 ms (24 ≤ n ≤ 28, p < 0.001, Fig. 4C). In endocardial myocytes, the small increase in APD 90 did not reach statistical significance. Similar results were observed with respect to the APD at 0 mV (Fig. 4B). The substantial increase in epicardial APD in the absence of alterations in endocardial APD completely abolished the large difference in APD between endo-and epicardial myocytes in the presence of DI.
Since under physiological conditions the 'command potential' for the I CaL current is not a rectangular voltage pulse such the ones used in Figs. 1, 2 and 3, but rather the myocyte's own AP, we conducted AP-clamp experiments to address the question whether the DI-induced changes in AP shape and duration affect the AP induced Scientific RepoRtS | (2020) 10:11604 | https://doi.org/10.1038/s41598-020-68308-4 www.nature.com/scientificreports/ Ca 2+ influx. We therefore recorded APs from myocytes and subsequently used their own individual AP as voltage template for the following voltage clamp experiments, in which we clamped the membrane voltage of each myocyte on its own AP in the absence and in the presence of 100 µM Cd 2+ to inhibit I CaL . The resulting current is a good estimate of the AP-induced Ca 2+ current 20 . Figure 4A (lower panels) displays the corresponding APinduced Ca 2+ currents obtained from individual myocytes of endo-and epicardial origin recorded under control conditions (blue) and after incubation with DI (green). Incubation with DI increased the peak AP-induced Ca 2+ current in endo-and epicardial myocytes to a similar extent (Fig. 4D), reflecting the increase in I CaL shown in Figs. 1, 2 and 3. The area under the AP-induced Ca 2+ current equals the Ca 2+ charge (Q Ca ), i.e. the total amount of Ca 2+ , entering the myocyte via L-type channels during an AP. As we have published previously 20 , the long endocardial AP leads to a significantly larger Q Ca in endo-compared to epicardial myocytes under control conditions. Incubation with DI dramatically increased Q Ca by 301% (n = 18 vs. n = 20, p < 0.001) in epicardial myocytes, while in endocardial myocytes the increase was much smaller (84%, n = 19 vs. n = 20, n.s.) and did not reach statistical significance (Fig. 4C,E). Given the substantial increase in AP-induced Ca 2+ influx, we combined Ca 2+ imaging with simultaneous sarcomere length measurements to investigate intracellular Ca 2+ transients and myocyte contractility. Figure 5A shows representative recordings of individual Ca 2+ transients (given as relative Fura ratio) obtained from an endo-and an epicardial myocyte under control conditions (blue) and after 24 h incubation with DI (green), while Fig. 5B shows representative simultaneous recordings of sarcomere length. Despite the longer APD and the larger Q Ca in endocardial cardiomyocytes, the amplitude of Ca 2+ transients (endo: 0.20 ± 0.01, n = 48; epi: 0.23 ± 0.02, n = 52, n.s.) as well as fractional unloaded sarcomere shortening (endo: 5.0 ± 0.5%, n = 48; epi: 5.1 ± 0.5%, n = 52, n.s.) were similar in endocardial and epicardial myocytes after 24 h of incubation under control conditions (Fig. 6C+D). Similarly, diastolic Ca 2+ level and diastolic sarcomere length were not significantly different between endo-and epicardial myocytes under control conditions (see Table 1). After 24 h incubation with DI, a substantial increase in the systolic Ca 2+ transient was observed in endo-(+ 70%, 0.34 ± 0.02, n = 48, p < 0.001 vs. control) as well as in epicardial cardiomyocytes (+ 43%, 0.33 ± 0.02, n = 52, p < 0.001 vs. control). Accordingly, unloaded www.nature.com/scientificreports/ fractional shortening increased by 74% (8.7 ± 0.7%, n = 48, p < 0.001 vs. control) in endocardial cells and by 45% (7.4 ± 0.5%, n = 52, p < 0.05 vs. control) in epicardial myocytes ( Fig. 5C and D). Diastolic Ca 2+ levels and diastolic sarcomere length remained unaffected by DI (see Table 1).
To examine whether the increase in Ca 2+ transient was due to an increased Ca 2+ content of the SR, we analyzed Ca 2+ transients evoked by rapid application of 10 mM caffeine after 60 s of steady-state pacing at 1 Hz. Figure 6A displays typical caffeine-induced Ca 2+ transients recorded from endo-and epicardial myocytes under control conditions (blue) and after incubation with DI (green). Since caffeine locks the ryanodine-receptor in an open state, Ca 2+ release from the SR is maximal 28 . Accordingly, the Ca 2+ transients were substantially larger than those observed in response to pacing. SR Ca 2+ content was assessed as the amplitude of the Ca 2+ transients and was similar in endo-and epicardial myocytes under control conditions. Interestingly, DI did not significantly alter SR Ca 2+ content (see Fig. 6B). However, in the presence of DI, the amplitude of the Ca 2+ transients during regular pacing reached ~ 70% of the caffeine-induce Ca 2+ transients, while under control conditions Ca 2+ transients during pacing reached only ~ 30% of the caffeine-induced Ca 2+ transients (Fig. 6C). This suggests that incubation with DI leads to a substantial increase in fractional release of Ca 2+ from the SR.
To further address this question, we assessed Ca 2+ transients under steady-state pacing at 1 Hz and then blocked Ca 2+ uptake into the SR by inhibiting the SERCA using thapsigargin (1 µM) leaving residual Ca 2+ transients arising from Ca 2+ influx via L-type Ca 2+ channels only. Figure 6D displays a typical recording of Ca 2+ transients stimulated by 1 Hz pacing. The arrow indicates the application of thapsigargin to the bath solution. Within 3-5 min, Ca 2+ transients had decreased to a residual amplitude reflecting Ca 2+ influx from the extracellular space only. Figure 6E summarizes similar experiments and displays the difference between the amplitude of Ca 2+ transients before and after the application of thapsigargin thus equaling the amount of Ca 2+ released from the SR. In both, endo-and epicardial myocytes, DI substantially increased SR Ca 2+ release. Since the total amount of Ca 2+ in the SR was not affected (see Fig. 6B), this confirms that dexamethasone treatment increased the fractional release of Ca 2+ from the SR during each cardiac cycle. Average current-voltage relations of currents similar to those shown in (A). I CaL was quantified by subtracting the peak current from the current at the end of the voltage pulse (at 600 ms). *p < 0.05, **p < 0.01, ***p < 0.001, dexamethasone + insulin versus control; # p < 0.05, ## p < 0.01, ### p < 0.001 dexamethasone + insulin versus dexamethasone + insulin + spironolactone. 21 ≤ n ≤ 32. www.nature.com/scientificreports/ In light of increased Ca 2+ release from the SR and an increased Ca 2+ influx from the extracellular space, both Ca 2+ extrusion via the Na/Ca exchanger (NCX) and Ca 2+ reuptake into the SR via SERCA should be increased in myocytes treated with DI. We therefore assessed time and rate constants of the decline of Ca 2+ transients during regular pacing and in response to application of caffeine. In DI treated myocytes the time constant of the Ca 2+ transient decay was accelerated in both endo-and epicardial myocytes (Fig. 7) indicating increased rates of Ca 2+ removal from the cytoplasm. The time constant of the caffeine-induced Ca 2+ transient decay was much slower, since in the presence of caffeine, Ca 2+ extrusion from the myocytes via the NCX is the only significant pathway left. In myocytes incubated with DI, the NCX time constants were accelerated compared to control which is consistent with the increased AP-induced Ca 2+ influx upon DI treatment. The SERCA time constant (Fig. 7C, calculated as the reciprocal value of the difference in total and NCX-dependent rate-constants of the decay of the Ca 2+ transient) was also substantially accelerated by DI treatment.
Finally, to address mechanisms underlying the increased SERCA transport rate, we performed western blot experiments in cardiomyocytes incubated for 24 h under control conditions and after treatment with DI to quantify SERCA and phospholamban expression as well as phospholamban phosphorylation. Since SERCA activity was similar in endo-and epicardial myocytes of both, control and DI groups, and in order to increase the total protein yield, we used isolated myocytes of the whole left ventricle for western blot experiments. Figure 8A-D  Fig. 1 and 2, obtained from endo-and epicardial myocytes of the left ventricular free wall incubated for 24 h under control conditions (blue) and with DI (green). *p < 0.05, **p < 0.01, ***p < 0.001, DI versus control. (B and C) Average current-voltage relations of I to and I K recorded from endo-and epicardial myocytes of the left ventricular free wall incubated for 24 h under control conditions (blue) and DI (green). Myocytes were clamped for 600 ms from the holding potential of V Pip = − 90 mV to test potentials between V Pip = + 60 mV to − 80 mV in steps of −20 mV. Basic cycle length was 3,000 ms. I to was quantified by subtracting the peak current from the current at the end of the voltage pulse (at 600 ms), I K was estimated as the current at the end of the voltage pulse (600 ms). ***p < 0.001, epi-versus endocardial myocytes incubated under control conditions, ### p < 0.001, epi-versus endocardial myocytes incubated with DI. www.nature.com/scientificreports/ display typical western blots stained against SERCA (8A), phospholamban (8B) and, using phospho-specific antibodies, against pSer16 (8C) and pThr17 (8D). Figure 8E summarizes the results of similar western blots and shows that SERCA expression is unaffected by treatment with dexamethasone while phospholamban expression was decreased. Moreover, phosphorylation of phospholamban increased at the Thr17 site thereby further reducing the inhibitory action of phospholamban on SERCA. This is consistent with increased SERCA activity.

Discussion
In the present study, MR activation increased I CaL to a similar extent in endo-and epicardial myocytes, demonstrating that, independent of regional origin, I CaL is not only similar in magnitude, but is also identically affected by MR agonists. Although treatment with DI led to a much larger increase in APD and AP-induced Ca 2+ influx in epicardial myocytes, the increase in intracellular Ca 2+ transient and, hence, contractility was similar in both regions. Moreover, SR Ca 2+ content was similar in both regions and not affected by DI treatment. We discovered that in both regions the fractional release of Ca 2+ from the SR was increased by DI treatment. Our results suggest that the increase in I CaL density rather than APD or AP-induced Ca 2+ influx controls the increase in Ca 2+ transient and contractility observed by DI treatment.
In the present study, we confirm regional differences in AP shape and duration as well as in underlying ionic currents that we and others have previously described for the left ventricular free wall of the rat 15,16,20 . Specifically, APD was substantially shorter in epicardial than in endocardial myocytes. This difference is attributable at least to a large extent to the larger epicardial I to current density, while other ionic currents, such as I K and I CaL , were similar in both regions. Since the AP waveform constitutes the voltage driving force for ionic currents including I CaL , the longer AP in endocardial myocytes led to an increased AP-induced Ca 2+ influx in endocardial compared to epicardial myocytes, despite of a similar current density of I CaL in both regions. This is in line with previous observations 20,29,30 and underlines the importance of AP shape for control of the AP-induced Ca 2+ influx. Moreover, all recordings in the present study were performed after at least 24 h incubation, hence, our data show that the regional differences in APD, underlying ionic currents and AP-induced Ca 2+ influx remain well preserved even after a prolonged (24-36 h) period of primary culture.  37 . We also found no significant difference in Ca 2+ transient amplitude or sarcomere shortening. Moreover, in our hands SR Ca 2+ content was similar in endo-and epicardial myocytes, suggesting that the larger AP-induced Ca 2+ influx in endocardial myocytes does not result in an increased SR Ca 2+ filling state. In a state of intracellular Ca 2+ homeostasis, a larger Ca 2+ influx from the extracellular space into the cytoplasm during each AP must be matched by an increased Ca 2+ extrusion into the extracellular space, carried predominantly by the NCX 38 . Consistently, we found a trend towards a shorter time constant of NCX Ca 2+ removal in endocardial compared to epicardial myocytes (Fig. 7B), possibly indicating an increased NCX activity. This trend did not reach statistical significance, however, one should keep in mind that the difference in NCX time constant necessary to match the increased Ca 2+ influx in endocardial myocytes without an increase in Ca 2+ transient might be small and below our detection threshold, since the amount of Ca 2+ entering via L-type current and exiting via the NCX is only a small fraction of the total Ca 2+ transient. Moreover, by using EGTA in the pipette solution in our AP-clamp experiments we might have somewhat overestimated AP-induced Ca 2+ influx since EGTA moderately ameliorates 39 Ca 2+ -induced Ca 2+ -inactivation of I CaL .
It is well established that MR activation for > 18-24 h increases I CaL in vitro in isolated cardiomyocytes 8,27 as well as in vivo in mice with an increased plasma aldosterone concentration 40 or in transgenic mice overexpressing www.nature.com/scientificreports/ the MR 11 . In the present study, we demonstrate that MR activation increases I CaL in endo-and in epicardial myocytes to a similar extent. Accordingly, regulation of I CaL magnitude by corticosteroids per se does not directly contribute to regional differences among the left ventricular free wall. However, especially in the early phase of the AP, the relation of the magnitudes of the repolarizing K + -current I to and the depolarizing I CaL not only sets the level of the plateau potential and controls early repolarization and the APD 26 but is also a potent modulator of the Ca 2+ -influx during the AP. For example, an acute delay in early repolarization (e.g. caused by a decrease in  www.nature.com/scientificreports/ I to ) increased AP-induced Ca 2+ influx 20,41 . One would therefore expect that an acute delay in early repolarization caused by an increase in I CaL should even further increase the AP-induced Ca 2+ influx, since not only the driving force (i.e. the membrane potential set by the AP) changes, but also the Ca 2+ conductance of the membrane. Indeed, the increase in I CaL we observed in response to MR activation substantially delayed early repolarization and increased APD, predominantly in epicardial myocytes, to such an extent that the endo-epicardial differences in APD disappeared. As a consequence, AP induced Ca 2+ -influx increased substantially in epicardial and to a lesser (but not significant) extent in endocardial myocytes. The relatively small increase of the AP induced Ca 2+ -influx in endocardial myocytes is consistent with their per se longer AP and delayed early repolarization, which leaves less room for an increase in APD upon an increase in I CaL .
MR activation led to a substantial increase in the intracellular Ca 2+ transient in both endo-and epicardial myocytes. A prolonged AP with a resulting increase in sarcolemmal Ca 2+ influx might alter several factors contributing to the intracellular Ca 2+ transient: an increase in Ca 2+ influx can trigger an increased release from the SR and might also increase the SR Ca 2+ content which in turn increases SR Ca 2+ release. Moreover, fractional release of Ca 2+ from the SR is controlled by two mechanisms, the SR Ca 2+ content and the magnitude of the trigger Ca 2+ , i.e. the AP-induced Ca 2+ influx 38,42 . In the present study, we found no increase in SR Ca 2+ content despite a substantial increase in AP-induced Ca 2+ influx and in SERCA activity. The increase in amplitude of the Ca 2+ transient can be explained by an increase in fractional release from the SR and (to a lesser extent) by an increase in trans-sarcolemmal Ca 2+ influx. This finding is supported by previous observations. Trafford et al. showed that an increase in Ca 2+ influx can substantially increase Ca 2+ release from the SR (and contractility) in the absence of effects on SR Ca 2+ content 43,44 . In animal models of heart failure with an increase in APD, Sah 41 and Kaprielian 45 observed an increased Ca 2+ release from the SR in the absence of an increased SR Ca 2+ content. In our study, fractional Ca 2+ release averaged ~ 40% under control conditions which is similar to the ~ 50% observed by Picht et al. 46 . In cardiomyocytes from mice overexpressing the MR or after 48 h of incubation with aldosterone, SR Ca 2+ content was also unaltered compared to control myocytes 7 . Since Bassani et al. 42 found only a 4% increase in SR Ca 2+ content in response to a switch to high loading condition, one could speculate that already under control conditions, SR Ca 2+ content might be nearly maximal. Moreover, Bode et al. 47 could show that in rat cardiomyocytes, SR Ca 2+ content only weakly depends on SERCA activity, when the SR Ca 2+ content is high. This was explained by an increase in SR Ca 2+ leak. Interestingly, aldosterone has been shown to increase SR Ca 2+ leak by downregulation of FKBP12 and 12.6 expression, thereby further limiting a potential increase in SR Ca 2+ load 7 . SERCA activity is in addition modulated by SUMOylation 48,49 . SUMOylation of SERCA by SUMO1 is decreased in heart failure and contributes to decreased SERCA activity under this condition. Moreover, β-arrestin-2 enhances SERCA SUMOylation 49 . Interestingly, glucocorticoids decrease β-arrestin-2 expression at least in human lung adenocarcinoma cells 50 , could thus indirectly decrease SUMOylation of SERCA and thereby reduce SERCA activity. Taken together, modulation of I CaL density via the MR appears as a potent regulator of intracellular Ca 2+ transient magnitude and contractility in both, endo-and epicardial myocytes. www.nature.com/scientificreports/ Methods Isolation of myocytes. Cardiomyocytes were isolated from the left ventricular free wall of female Wistar rats (~ 220 g) as described previously 51 . After induction of deep anesthesia by intraperitoneal injection of thiopental-sodium (100 mg kg −1 body mass), the heart was quickly excised and placed into cold (4 °C) Tyrode's solution. Subsequently, the aorta was retrogradely perfused for 5 min with modified Tyrode's solution containing 4.5 mM Ca 2+ and 5 mM EGTA (~ 1 μM free Ca 2+ concentration) supplemented with 1 μM insulin (Sigma-Aldrich Chemie GmbH, Taufkirchen, Germany). The perfusion was continued for 19 min, recirculating 25 ml of the same solution containing collagenase (CLS type II, 160 U/ml, Biochrom KG, Berlin, Germany) and protease (type XIV, 0.6 U/ml, Sigma). Then, the heart was perfused for another 5 min with storage solution 8 containing 100 μM Ca 2+ . Using fine forceps, tissue portions of the subendocardial (endocardial) and subepicardial (epicardial) layers were taken and placed in separate cell culture dishes containing the same solution at 37 °C. Tissue pieces were minced and gently agitated to obtain single cardiomyocytes. Myocytes were stepwise adapted to physiological Ca 2+ levels, transferred to cell culture dishes containing storage solution supplemented with 1 g l −1 BSA, 100 IU ml −1 penicillin and 0. Patch-clamp technique. The ruptured-patch whole-cell configuration was used as described previously 15,52 .
Myocardial cells were transferred into an elongated chamber (2.5 × 20 mm), mounted on the stage of an inverted microscope (Axiovert 25, Zeiss, Jena, Germany) and initially superfused with control solution. All experiments were performed at room temperature (22-25 °C). Patch pipettes were pulled from borosilicate glass (GC150-15, www.nature.com/scientificreports/ Clark Electromedical Instruments, Reading, UK) using a P-97 Puller (Sutter Instruments, Novato, CA, USA). Pipette resistance (R Pip ) was 1.5-5 MΩ. Currents were recorded using an EPC-10 amplifier (HEKA Elektronik, Lambrecht, Germany), controlled by PULSE-Software (HEKA Elektronik). Membrane voltage (V m ) and APs were recorded in the zero current-clamp mode and ionic currents in the voltage-clamp mode. For AP voltageclamp recordings, APs were recorded at the beginning of the experiments and used as a voltage template in the voltage-clamp mode of the amplifier 20,53 . Membrane capacitance (C m ) and series resistance (R s ) were calculated using the automated capacitance compensation procedure of the EPC-10 amplifier. Series resistance was in the range of ~ 5 MΩ, was not allowed to exceed 10 MΩ and was compensated by 85%. The reference electrode of the amplifier headstage was bathed in pipette solution in a separate chamber and was connected to the bath solution via an agar-agar bridge filled with pipette solution. Pipette potential (V Pip ) and V m were corrected for liquid junction potentials at the bridge-bath junction. Whole-cell currents were low-pass filtered at 1 kHz and sampled at 5 kHz. Action potentials were sampled at 5 kHz.
To assess I CaL , myocytes were clamped for 600 ms from the holding potential of − 90 mV to test potentials between − 60 mV and + 70 mV in steps of 10 mV. Na + currents were inactivated by a prepulse of 70 ms to − 50 mV. Basic cycle length was 3,000 ms. I CaL was quantified by subtracting the current at the end of the test pulse from the peak current. 54 To elicit outward K + currents, myocytes were clamped for 600 ms from the holding potential of − 90 mV to test potentials between 60 mV and − 80 mV in steps of − 20 mV. Na + currents were inactivated by a prepulse of 20 ms to − 50 mV. Basic cycle length was 3,000 ms. I to was quantified by subtracting the current at the end of the test pulse from the peak current. I K was defined as the current at the end of the voltage pulse.
In some experiments, caffeine or thapsigargin were washed in. For all Ca 2+ fluorescence experiments, cells were paced for 1 min at 1 Hz before starting the measurements to ensure that Ca 2+ balance was at steady state.
Protein extraction. Isolated cardiomyocytes were incubated with DI for 24 h while myocytes isolated from the same heart were incubated for 24 h with vehicle and served as paired control. Myocytes were pelleted by centrifugation and dispensed in 1 ml TNE buffer. 40 µl protease inhibitor, 40 µl phosphatase inhibitor, 30 µl triton X-100 (10%), 5 µl PMSF (200 mM in 100% EtOH) and 20 µl sodium deoxycholate (12.5% in H 2 O) were added. Handled on ice at all times, samples were mechanically homogenized for 20 s, sonicated for 3 × 5 s and then centrifuged at 13,000 g and 4 °C for 10 min. The supernatant was used for further studies. To ensure equal protein loading, protein concentration was measured using the BCA Protein Assay Reagent Kit for microplate assay (Pierce, Rockford, USA). BSA in concentrations between 25 and 2000 µg/ml in TNE buffer was used as standard.
Data analysis and statistics. Patch clamp data were analyzed using the PULSE-FIT software (HEKA Elektronik, Lambrecht/Pfalz, Germany), IGOR Pro (WaveMetrics, Lake Oswego, USA), and Microsoft Excel (Microsoft Corporation, Redmond, USA) as described previously 54 . Ca 2+ epifluorescence data were analyzed using Ionwizard 5.0 (IonOptix Corporation, Milton, USA) and Microsoft Excel. Quantitative densitometric analysis of western blots was performed using ImageJ software. The intensity of specific bands was determined after background subtraction and normalized to the total protein content per lane as quantified by densitometric analysis of the corresponding Ponceau stains 58 . Data are given as mean ± SEM. Statistical significance was evaluated by paired or unpaired Student's t test when two groups were compared or one-way ANOVA followed by Newman-Keuls test (with the exception of APD 0mV and Q Ca , which were analyzed by Kruskal Wallis test followed by Dunn's post test because both were not normally distributed) when more than two groups were compared using Prism 5 (GraphPad, San Diego, USA). p < 0.05 was considered statistically significant.