Generation of global homozygous NCX overexpressor (OE) (Adachi-Akahane et al., 1997) and heterozygous knockout (hetKO) (Jordan et al., 2010) mouse model has been described previously. The genetic background of OE mice is C57BL/6 (Pott et al., 2012), whereas hetKO mice are on a combined genetic background of C57BL/6 and CD-1 due to embryonic transfer (Bögeholz et al., 2016). Animals entering experimentation were <12.5 weeks of age, since up to this age there is no evidence for structural remodeling or heart failure in OE (Pott et al., 2012) or hetKO mice (Jordan et al., 2010; Bögeholz et al., 2015). All experiments were approved by the local animal welfare authorities (Landesamt für Natur, Umwelt, und Verbraucherschutz NRW, permission number: 8.84-02.05.20.11.098) and conform to the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes.

Ventricular cardiomyocytes were isolated from OE, hetKO and respective wild-type mice (WT_OE/WT_hetKO) as previously described with minor alterations (Schulte et al., 2012; Bögeholz et al., 2015). Mice were sedated and euthanized by carbon dioxide inhalation. Hearts were explanted immediately after decease and perfused retrogradely for 1 min via the aorta with a perfusion solution containing in mM: NaCl 113, KCl 4.7, KH₂PO₄ 0.6, Na₂HPO₄ 0.6, MgSO₄ 1.2, NaHCO₃ 12, KHCO₃ 10, taurine 30, HEPES 10, glucose 11.1, BDM 10, heparin 14.3 IE/ml, pH = 7.4. To isolate rod-shaped viable ventricular myocytes, the extracellular connective tissue was degraded by perfusion for 6 min ± 10 s on 37.1°C with the perfusion solution containing Collagenase type 2 (Worthington Biochemical Corporation) and Protease from streptomyces griseus type XIV (Sigma Aldrich).

Technical and procedural aspects of Ca²⁺ epifluorescence measurement have been described in detail before (Pott et al., 2012; Bögeholz et al., 2015). Data acquisition was performed using a Ca²⁺ imaging system, composed of a DeltaRAM monochromator and two Type-710 photomultiplier (Photon Technology International, USA) mounted to an inverse microscope (Diaphot 200, Nikon, Japan). Cardiomyocytes were transferred to a custom-made recording chamber and field-stimulated by platinum wires (Type 223 Stimulator CS, Hugo Sachs Elektronik- Harvard Apparatus, Germany). Fluorescence signals were recorded by FeliX 1.42 software (Photon Technology International, USA). The following external solution (Tyrode's solution) has been applied; in mM: NaCl 140, KCl 5.8, KH₂PO₄ 0.5, Na₂HPO₄ 0.4 MgSO₄ 0.9, HEPES 10, glucose 11.1, CaCl₂ 2, pH = 7.3. Experiments were performed at room temperature. The NCX function was assessed by the decay of caffeine (10 mM) induced Ca²⁺ transients (Bassani et al., 1994; Bögeholz et al., 2015, 2016) using fluo-4-AM (excitation: 488 nm, emission: 522 nm). The ratiometric fluorescent Ca²⁺ indicator indo-1-AM was employed to record field-stimulated Ca²⁺ transients and proarrhythmic spontaneous Ca²⁺ activity (excitation: 344 nm, emission-ratio: 405/495 nm). The amplitude of field-stimulated Ca²⁺ transients served as a proxy measure for inotropy. The most chosen SEA0400 concentration is 1 μM (Farkas et al., 2008; Namekata et al., 2009; Bourgonje et al., 2013; Nagy et al., 2014). Since previous studies demonstrated a relevant inhibitory effect of SEA0400 on other molecular targets next to NCX, such as the L-type Ca²⁺ channel, at concentrations higher than 1 μM (Birinyi et al., 2005) we decided not to exceed this threshold. Since we could observe an effect of SEA0400 on Ca²⁺ transient dynamics in initial dose-response experiments in a range between 0.3 and 1 μM, we decided to apply both concentrations.

Statistical calculations were conducted with SigmaPlot 11.2 (Systat Software; Inc.). Numerical results are expressed as mean value ± standard error of the mean (SEM). Number of replicates are indicated as n = number of independently analyzed cardiomyocytes/number of mice used for cardiomyocyte isolation. Data were analyzed by ANOVA on ranks followed by Dunn's post hoc test or repeated measures ANOVA followed by Holm-Sidak post hoc test. Rates and proportions were quantified using the Fisher's exact test with R.

To test the inhibitory effect of SEA0400 on the NCX forward mode, caffeine application experiments were conducted as reported previously (Bassani et al., 1994; Bögeholz et al., 2015). Before caffeine was applied, cells were field-stimulated in absence or presence of SEA0400, respectively, with 1 Hz for 1 min followed by 10 s of rest. A faster caffeine-induced Ca²⁺ transient decay indicates a higher NCX function.

As demonstrated previously, in absence of SEA0400 (basal conditions), the time to 50% decay of caffeine-induced Ca²⁺ transients (T₅₀) was significantly prolonged in heterozygous hetKO and shortened in OE as compared to the corresponding WT confirming a reduced/enhanced NCX-mediated Ca²⁺ removal rate, respectively (T₅₀ in s: OE: 1.2 ± 0.2⁺; n = 37/5; WT_OE: 2.0 ± 0.1; n = 42/5; hetKO: 3.3 ± 0.2⁺; n = 26/4; WT_hetKO: 2.1 ± 0.2; n = 24/4; ⁺p < 0.05 OE vs. WT_OE or hetKO vs. WT_hetKO; ANOVA on ranks test) (Figures 1A–D). When caffeine was applied in the presence of SEA0400 (1 μM), T₅₀ was significantly increased vs. basal in all mouse lines indicating the inhibitory effect of SEA0400 on NCX forward mode (T₅₀ in s; 1 μM SEA0400: OE: 2.9 ± 0.4^*; n = 21/3; WT_OE: 4.1 ± 0.4^*; n = 25/3; hetKO: 5.8 ± 0.4^*; n = 24/3; WT_hetKO: 4.3 ± 0.4^*; n = 24/3; ^*p < 0.05 vs. basal; ANOVA on ranks test) (Figures 1B,D). In presence of SEA0400, T₅₀ still tended to be shortened in OE and prolonged in hetKO compared to WT_OE and WT_hetKO, respectively, however this finding was not statistically significant (p > 0.05 each; ANOVA on ranks test). Notably, the mean T₅₀ value of OE in presence of SEA0400 was larger than in WT_OE/WT_hetKO but still smaller than in hetKO under basal conditions indicating a remaining NCX function in OE ranging between the levels of WT_OE/WT_hetKO and hetKO.

To test the effects of SEA0400 on the steady-state amplitude of field-stimulated Ca²⁺ transients, Ca²⁺ transients were measured in absence and presence of 0.3 and 1 μM SEA0400. Cells were field-stimulated at 1 Hz for 4 min followed by 20 sweeps on 0.125 Hz. This protocol was repeated in presence of 0.3 and 1 μM SEA0400 within the same cell. The amplitude of the Ca²⁺ transient was quantified at the end of the 1 Hz pacing phase after reaching a steady-state in Ca²⁺ transient kinetics. The Ca²⁺ transient amplitude in presence of SEA0400 is given in % of basal conditions.

SEA0400 led to a similar relative increase of the mean Ca²⁺ transient amplitude in WT_OE (n = 27/7), WT_hetKO (n = 19/4), and hetKO (n = 19/5). However, in OE (n = 27/8) no increase of the Ca²⁺ transient amplitude was observed under application of SEA0400 (amplitude of field stimulated Ca²⁺ transients in % of basal; 1 μM SEA0400: OE: 98.7 ± 5.1; n = 23/8; WT_OE: 137.8 ± 5.2^*; n = 23/7; hetKO: 140.6 ± 12.8^*; n = 9/3; WT_hetKO: 126.3 ± 6.0^*; n = 16/4; ^*p < 0.05 vs. basal; repeated measures ANOVA test) (Figures 2A,B).

To investigate the dependence of the antiarrhythmic effects of SEA0400 on NCX expression level, we monitored the occurrence of spontaneous Ca²⁺ releases (sCR), defined as spontaneously occurring Ca²⁺ transients between the field stimulated Ca²⁺ transients, during the 0.125 Hz pacing period in the above stated protocol before and after SEA0400 administration. An increase of sCR suggests an increase of proarrhythmia.

Under basal conditions, the occurrence of sCR (Figure 3) was similar to previously reported findings for OE (Pott et al., 2012) and hetKO (Bögeholz et al., 2015). The fraction of cells exhibiting spontaneous Ca²⁺ transients was significantly increased in OE compared to WT_OE (Figure 3C) (OE: 63%⁺, WT_OE: 7%; ⁺p < 0.05 OE vs. WT_OE, Fisher's exact test with R). The average number of sCR per cell (Figure 3B) was likewise increased in OE vs. WT_OE (number of sCR/all investigated cells: OE: 12.5 ± 3.8⁺; n = 27/8; WT_OE: 0.2 ± 0.1; n = 27/7; ⁺p < 0.05 OE vs. WT_OE; repeated measures ANOVA).

In hetKO the fraction of cells with sCR (Figure 3C) (hetKO: 5%; n = 19/5 WT_hetKO: 21%; n = 19/4; p > 0.05 hetKO vs. WT_hetKO, Fisher's exact test with R) as well as mean number of sCR per cell (Figure 3B) tended to be reduced vs. WT_hetKO but the results were not statistically significant (number of sCR/all investigated cells: hetKO: 0.2 ± 0.2; WT_hetKO: 1.3 ± 1.0; p > 0.05 hetKO vs. WT_hetKO; repeated measures ANOVA).

Application of SEA0400 altered the occurrence of sCR, but the specific effect depended upon the basal NCX expression level. In OE, SEA0400 significantly decreased the mean number of sCR per cell (Figure 3B) (number of sCR/all investigated cells; 1 μM SEA0400; OE: 3.7 ± 1.0^*; n = 23/8; ^*p < 0.05 vs. basal; repeated measures ANOVA) without altering the number of sCR-positive cells (Figure 3C). In contrast, we observed an increase in the occurrence of sCR in groups with normal or reduced NCX expression level. In WT_OE and WT_hetKO SEA0400 significantly increased the number of sCR-positive cells to 41 and 71% (Figure 3C). At the same time the occurrence of sCR per cell was increased significantly in WT_hetKO and in tendency in WT_OE in response to SEA0400 (Figure 3B) (number of sCR/all investigated cells; 1 μM SEA0400: WT_OE: 2.4 ± 0.8; n = 23/7; WT_hetKO: 8.8 ± 1.9^*; n = 16/4; ^*p < 0.05 vs. basal; repeated measures ANOVA). In hetKO application of 1 μM SEA0400 significantly increased both, the occurrence of sCR per cell (Figure 3B) (number of sCR/all investigated cells; 1 μM SEA0400: hetKO: 5.5 ± 2.2^*; n = 8/3; ^*p < 0.05 vs. basal; repeated measures ANOVA) and the number of sCR-positive cells (Figure 3C) (hetKO: 88%^*; ^*p < 0.05 vs. basal, Fisher's exact test with R). Moreover, SEA0400 drastically impaired Ca²⁺ homeostasis in hetKO as reflected by an abrupt elevation of the diastolic Ca²⁺ level, incessant Ca²⁺ oscillations and subsequent cell death, defined as Ca²⁺ hypercontraction (Figure 4), so that only 42% of all measured hetKO cells could be analyzed. The fraction of cells suffering from Ca²⁺ hypercontraction was significantly increased in hetKO as compared to WT_hetKO (Figure 4B) (hetKO: 58%⁺, WT_hetKO: 11 %; ⁺p < 0.05 hetKO vs. WT_hetKO, Fisher's exact test with R) but unaltered between OE and WT_OE (OE: 15%, WT_OE: 19%; p > 0.05 OE vs. WT_OE, Fisher's exact test with R).

SEA0400 reduced the NCX forward mode function to a similar extent in all mouse lines thus confirming inhibitory efficacy. At the same time, SEA0400 mediated a significant increase of the Ca²⁺ transient suggesting a positive inotropic effect in both wild-types (WT_OE; WT_hetKO) and hetKO but interestingly not in OE.

Previous studies found a positive inotropic effect in response to SEA0400 in murine, rat and pig models (Tanaka et al., 2005; Farkas et al., 2008; Ozdemir et al., 2008). Other studies reported rather neutral effects on contractility in rabbit, guinea pig and dog models (Nagy et al., 2004; Tanaka et al., 2007; Bourgonje et al., 2013). These diverse effects may be mostly attributable to differences in Ca²⁺ handling and basic electrophysiological properties between the various species. Analogous to cardiac glycosides, NCX inhibition can result in a net gain of intracellular Ca²⁺, which augments the Ca²⁺ transient amplitude and consequently inotropy. SEA0400 seems to mediate such a positive inotropic effect in smaller rodents (mouse and rat) but not in higher mammals, with the exception for pigs. In smaller rodents, the intracellular Na⁺ level is elevated compared to higher mammals which results in a reversal potential for NCX (E_Na/Ca) close to the diastolic membrane potential (E_m) reducing the driving force for the NCX forward mode (E_m-E_Na/Ca) (Bers, 2002). Moreover, Ca²⁺ removal from the cytosol in these species largely depends on sarco(endo-)plasmic reticulum Ca²⁺-ATPase (SERCA) which mediates the re-uptake of Ca²⁺ ions into the sarcoplasmic reticulum while NCX mediates Ca²⁺ extrusion to the extracellular compartment. In higher mammals, the competition between SERCA and NCX shifts in favor of NCX, although SERCA still removes the major proportion of cytosolic Ca²⁺ (Puglisi et al., 1996). To assure a balanced Ca²⁺ homeostasis, the exact amount of Ca²⁺ entering the cell mainly via L-type Ca²⁺ channels has to be extruded to the extracellular compartment via NCX. Thus, SEA0400-mediated reduction of a comparatively low NCX-mediated Ca²⁺ removal rate as observed in smaller rodents may more likely result in intracellular Ca²⁺ accumulation than in higher mammals which might explain the observed different inotropic responses in different studies. However, Ozdemir et al. found a positive inotropic response of SEA0400 in pigs as an example for higher mammals (Ozdemir et al., 2008). But they also found a lower NCX current in pig compared to mice, which might explain the positive inotropy via Ca²⁺ accumulation.

In contrast to both wild-types and hetKO, cardiomyocytes from OE mice did not exhibit a significant increase in the Ca²⁺ transient amplitude in response to SEA0400. In OE 1 μM SEA0400 reduced the NCX-mediated Ca²⁺ removal rate on average to a level ranging between the basal hetKO rate and the rate of both wild-types (Figures 1B,D). A NCX-mediated Ca²⁺ removal rate as low as in hetKO mice under basal conditions was not associated with a measurable increase of the amplitude of Ca²⁺ transients as reported previously (Bögeholz et al., 2015). Hence, the remaining NCX function in OE in presence of SEA0400 seems to be still high enough to cope with Ca²⁺ removal and thus to avoid net gain of Ca²⁺ and consequently an increase in Ca²⁺ transient amplitude. In this regard, the inhibitory effect of SEA0400 on L-type Ca²⁺ current has also to be taken into account (Birinyi et al., 2008). It has been shown that SEA0400 inhibits the L-type Ca²⁺ current either directly via pleiotropic effects, or indirectly by slowing the NCX-mediated Ca²⁺ removal leading to an increase of subsarcolemmal Ca²⁺ concentration which in turn promotes Ca²⁺ dependent inactivation of L-type Ca²⁺ current (Pott et al., 2005; Bögeholz et al., 2015).

Interestingly, Ozdemir et al. similarly found absence of a positive inotropic effect in MLP-/- mice, a murine model of heart failure (Ozdemir et al., 2008). Due to the heart failing phenotype NCX upregulation is likely, which may prevent from Ca²⁺ accumulation and positive inotropy. Taken together within the same species the effect of the same concentration of SEA0400 on calcitropy and thus inotropy seems to depend on the NCX expression level and corresponding function.

In previous studies, SEA0400 exhibited antiarrhythmic efficacy on cellular level by suppressing afterdepolarizations (Nagy et al., 2004; Zhao et al., 2012), on whole heart level in vitro (Milberg et al., 2008, 2012) and even in vivo in a canine model (Nagasawa et al., 2005; Bourgonje et al., 2013). On the other hand, proarrhythmic effects mediated by SEA0400 have been reported as well (Ozdemir et al., 2008). The respective NCX function might be considerably heterogeneous in the investigated animal models, i.e., increased in the case of heart failure, or effectively decreased in the setting of cardiac glycoside-induced proarrhythmia. We here demonstrate that the effects of SEA0400 on cellular proarrhythmia again depend on the underlying baseline NCX function within the same species. Considering NCX inhibition as a promising therapeutic strategy, this might be of translational relevance, since major cardiac diseases like heart failure or atrial fibrillation are accompanied by NCX upregulation, whereas other specific cardiac disease entities, like the postinfarction (Zhang et al., 1996; Quinn et al., 2003) and diabetic cardiomyopathy (Schaffer et al., 1997; Hattori et al., 2000) exhibit a reduced NCX function, at least in animal models. There also might be further specific disease entities with altered NCX function that are yet to be further evaluated (Primessnig et al., 2016).

In OE, application of SEA0400 reduced the occurrence of spontaneous Ca²⁺ transients. In a former study the increased NCX expression and function in OE has been identified as an independent proarrhythmic factor (Pott et al., 2012). Thus, the observed reduction of arrhythmogenic spontaneous Ca²⁺ transients in OE can be attributed to the SEA0400-mediated inhibition of NCX function.

The underlying molecular mechanism of SEA0400-mediated decrease in the occurrence of spontaneous Ca²⁺ transients in OE may consist of reduced NCX forward mode with subsequently reduced depolarizing inward current thus avoiding membrane depolarization and the occurrence of EADs and DADs as proarrhythmic triggers. However, in this regard again the inhibitory potency of SEA0400 on L-type Ca²⁺ current has to be taken into account (Birinyi et al., 2008; Bourgonje et al., 2013), which might contribute to the observed reduction of proarrhythmic spontaneous Ca²⁺ transients in OE in presence of SEA0400. Supplementary to these molecular mechanisms of arrhythmia, we have previously demonstrated a significant antiarrhythmic effect of SEA0400 on whole-heart arrhythmia (ventricular tachycardia) in a rabbit model of heart failure (Milberg et al., 2012). SEA0400 reduced the occurrence of EADs due to a reduced monophasic action potential duration, dispersion and QT interval. Thus, the antiarrhythmic effects of SEA0400 may be diverse and apply to diverse functional levels.

In contrast to OE, SEA0400 led to an increased occurrence of spontaneous Ca²⁺ transients in WT_OE, WT_hetKO, and hetKO. Furthermore, Ca²⁺ homeostasis seemed to be drastically impaired in hetKO treated with SEA0400 as reflected in the increase percentage of cardiomyocytes with hypercontraction and subsequent cell death as a consequence of cytosolic Ca²⁺ overload. Based on these observations, it seems likely that a critical level of NCX function must remain after SEA0400 treatment to avoid detrimental intracellular Ca²⁺ accumulation and overload. While the reduced basal NCX function in hetKO led to a phenotype with reduced likelihood of arrhythmia and a balanced Ca²⁺ homeostasis as reported previously (Bögeholz et al., 2015), lowering NCX function below this level with SEA0400 was clearly detrimental in hetKO, WT_hetKO, and WT_OE. In this regard, one has to distinguish between acute reduction of NCX function via SEA0400 and chronically reduced NCX function via transgenic manipulation. The latter offers the opportunity for long-term adaption beginning from the early stages of development to compensate for the reduced Ca²⁺ removal rate mediated via NCX. The most important adaptive mechanism in models with reduced (Bögeholz et al., 2015) or absent NCX function (Pott et al., 2007a,b,c) to keep a balanced Ca²⁺ homeostasis consists of a reduced Ca²⁺ entry via the L-type Ca²⁺ channel as mediated by an enhanced Ca²⁺-dependent inactivation of the L-type Ca²⁺ current. Thus, if less Ca²⁺ enters the cell, the reduced NCX function is still sufficient to keep a balanced Ca²⁺ homeostasis in a chronically adapted system. In contrast, the lack of time for adaptation to acute NCX inhibition by SEA0400 may contribute to the impaired Ca²⁺ homeostasis in WT and most prominently in hetKO. However, in OE acute application of SEA0400 was not detrimental but reduced the cellular proarrhythmia through acute NCX inhibition. Thus, depending on the baseline NCX expression level, acute SEA0400 application can cause destabilization of intracellular Ca²⁺ homeostasis leading to arrhythmia and cell death or mediate antiarrhythmic effects.

In this study, the Ca²⁺ transient amplitude served as a proxy measure for inotropy on the cellular level. Though an acute increase of the Ca²⁺ transient amplitude is usually accompanied by a simultaneous increase in contractility, we did not measure changes in contractility directly. Likewise, although spontaneous Ca²⁺ releases represent a well-known proarrhythmic substrate, it has to be acknowledged, that clinically relevant whole-heart arrhythmia is a more complex result from a magnitude of involved factors. Thus, future studies should investigate the in vivo-relevance of our findings derived from the cellular level.

Moreover, our findings may not be uncritically transferable into the human (patho-)physiology, since there are substantial differences in cardiac Ca²⁺ cycling und electrophysiology compared to a murine model. It remains also to be tested whether more specific NCX inhibitors show comparable effects to SEA0400. However, despite these limitations our data suggest that care should be taken when testing NCX inhibitors in clinical studies where we usually do not know the respective NCX expression and function level of individuals, which might determine beneficial or detrimental effects of the treatment approach with an NCX inhibitor.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.