Detrimental proarrhythmogenic interaction of Ca2+/calmodulin-dependent protein kinase II and NaV1.8 in heart failure

An interplay between Ca2+/calmodulin-dependent protein kinase IIδc (CaMKIIδc) and late Na+ current (INaL) is known to induce arrhythmias in the failing heart. Here, we elucidate the role of the sodium channel isoform NaV1.8 for CaMKIIδc-dependent proarrhythmia. In a CRISPR-Cas9-generated human iPSC-cardiomyocyte homozygous knock-out of NaV1.8, we demonstrate that NaV1.8 contributes to INaL formation. In addition, we reveal a direct interaction between NaV1.8 and CaMKIIδc in cardiomyocytes isolated from patients with heart failure (HF). Using specific blockers of NaV1.8 and CaMKIIδc, we show that NaV1.8-driven INaL is CaMKIIδc-dependent and that NaV1.8-inhibtion reduces diastolic SR-Ca2+ leak in human failing cardiomyocytes. Moreover, increased mortality of CaMKIIδc-overexpressing HF mice is reduced when a NaV1.8 knock-out is introduced. Cellular and in vivo experiments reveal reduced ventricular arrhythmias without changes in HF progression. Our work therefore identifies a proarrhythmic CaMKIIδc downstream target which may constitute a prognostic and antiarrhythmic strategy.

V oltage-gated sodium channels (Na V ) play a critical role in physiological cardiac conduction. Na V channels become inactive within a few milliseconds after activation under physiological conditions. However, in cardiac pathologies such as ischemia, hypoxia, oxidative stress, and heart failure (HF), some Na V remain persistently open or reopen, generating a small but persistent Na + current, referred to as the late Na + current (I NaL ) [1][2][3][4][5] . This current slows the repolarisation rate and thereby prolongs the action potential duration (APD). Augmented I NaL may additionally cause Na + -dependent Ca 2+ overload in cardiomyocytes, thereby playing an essential role for arrhythmogenesis and diastolic dysfunction 1,2,6,7 . Furthermore, Na + /Ca 2+ overload caused by augmented I NaL can give rise to early afterdepolarizations (EADs), delayed afterdepolarizations (DADs), and hence sustained triggered arrhythmias [8][9][10][11] . In the failing heart, increased I NaL induces an influx of Na + into the cardiomyocyte, which in turn stimulates Ca 2+ influx via the reverse mode of Na + /Ca 2+ exchanger (NCX) 12,13 . Cytosolic Ca 2+ may now bind to calmodulin (CaM), forming a Ca 2+ /CaM complex, which activates Ca 2+ /calmodulin-dependent protein kinase IIδ (CaMKIIδ), a multifunctional serine/threonine protein kinase 8,10,14,15 . CaMKII is expressed in four isoforms α, β, γ, and δ. CaMKIIδ is the predominant isoform in heart 16 while the δc isoform is mainly located in the cytosol. Once CaMKIIδc is activated, it may cause hyperphosphorylation of the ryanodine receptor 2 (RyR2) residing within the sarcoplasmic reticulum (SR)-sarcolemma junction, leading to spontaneous proarrhythmogenic SR-Ca 2+ release events in HF 13,[17][18][19][20] . Further, this augmented CaMKIIδc activity can also induce I NaL augmentation by phosphorylating Na V channels 10,14,21,22 leading to a vicious cycle between I NaL and CaMKIIδc.
Besides Na V 1.5 other Na V isoforms have been reported to be present in the heart. Theoretically, they could also generate I NaL , induce APD prolongation, and spontaneous SR-Ca 2+ release. In the previous few years, different reports have been published on the existence of noncardiac Na V in the heart. Na V 1.8, a noncardiac tetrodotoxin-resistant Na V channel, is encoded by the SCN10A gene and was originally reported to be expressed in the dorsal root ganglion 23 . Certain genome-wide association studies reported an association of SCN10A with changes in ECG parameters but most importantly with cardiac arrhythmias such as atrial fibrillation and sudden cardiac death [24][25][26][27] . Later, the presence of Na V 1.8 was detected in atria and further evidence came from studies conducted in mouse and rabbit cardiomyocytes investigating its involvement in cardiac electrophysiology [28][29][30] . Moreover, we recently reported that Na V 1.8 mRNA and protein expression are upregulated in tissue from human hypertrophied and failing ventricles and that Na V 1.8 contributes to I NaL generation in the human heart under these pathological conditions 31,32 . However, Na V 1.8 regulation, a potential interplay with pathologically increased CaMKIIδc activity in HF, and the role of Na V 1.8 on HF progression and arrhythmias in vivo and in vitro remain elusive.
In this work, we describe a detrimental interaction of Na V 1.8 with CaMKIIδc in human and murine failing cardiomyocytes. Moreover, we investigate the contribution of Na V 1.8 to cellular electrophysiology in relation to enhanced CaMKIIδc activity and consequently show a reduction of arrhythmias which is paralleled by an improved survival due to Na V 1.8 deletion in CaMKIIδc transgenic HF mice.

Results
Na V 1.8 and CaMKIIδc interaction in human HF. Since interaction between CaMKIIδc and Na V 1.5 was shown previously, we hypothesized that CaMKIIδc interacts also with Na V 1.8 and therefore performed co-immunoprecipitation using human ventricular tissue homogenates. Indeed, we found that CaMKIIδc associates with Na V 1.8 in human non-failing as well as HF myocardium (Fig. 1a). Using immunofluorescence stainings, CaMKIIδc and Na V 1.8 were found to be co-localized in isolated human cardiomyocytes (Fig. 1b). Of note, SCN10A mRNA expression in tissue from non-failing and HF hearts, as well as isolated cardiomyocytes from human HF hearts was found to be much lower than SCN5A. Further, RT-qPCR experiments revealed that a relevant part of SCN10A mRNA in the heart originates from cardiomyocytes (Supplementary information, Supplementary Figs. 1, 2). Na V 1.8 inhibition reduces I NaL in human failing, mouse CaMKIIδc transgenic cardiomyocytes, and in SCN10A knockout iPSC-cardiomyocytes. In functional experiments, we could show that I NaL , induced by increased CaMKIIδc activity, was significantly reduced following pharmacological inhibition and genetical knockout of Na V 1.8 in human and murine cardiomyocytes. In isolated murine CaMKIIδc +/T cardiomyocytes, I NaL was augmented to~150% compared to wildtype (WT), whereas the specific Na V 1.8 blockers A-806734 or PF-01247324 reduced I NaL by~40% in the same background (CaMKIIδc +/T ) (Fig. 2a,  b). These results illustrate that CaMKIIδc-induced I NaL can be clearly ameliorated by inhibiting Na V 1.8. A current-voltage relationship of Na V 1.8-dependent I NaL in isolated ventricular myocytes from CaMKIIδc +/T mice is presented in the Supplementary information ( Supplementary Fig. 3).
In the human failing heart, CaMKIIδc and I NaL are upregulated in parallel 2,33 . Therefore, we inhibited Na V 1.8 by using PF-01247324 and compared its ability to reduce I NaL to CaMKIIδc inhibition using autocamtide-2-related inhibitory peptide (AIP) in human failing ventricular cardiomyocytes. In addition, we blocked Na V 1.8 and CaMKIIδc in parallel by exposing human failing cells simultaneously to PF-01247324 and AIP. I NaL measurements demonstrated that Na V 1.8 inhibition alone (PF-01247324) led to a~40% decrease and CaMKIIδc inhibition Fig. 1 CaMKIIδc interacts with Na V 1.8 in human myocardium and isolated cardiomyocytes. a Co-immunoprecipitation of CaMKIIδc and Na V 1.8 from left ventricular homogenates of human non-failing and failing hearts (NF: n = 7; HF: n = 7). b Co-localization of CaMKIIδc and Na V 1.8 in human failing cardiomyocytes with immunofluorescence staining. Scale bar: 10 µm (staining was performed in cardiomyocytes isolated from five heart failure patients).
by AIP to a~53% reduction of I NaL in human failing cardiomyocytes (Fig. 2c, d). However, preincubation with AIP and PF-01247324 together decreased I NaL comparable to AIP alone suggesting that CaMKIIδc inhibition might already suppress Na V 1.8-driven I NaL . Further, peak I Na measurements revealed, that I NaL reduction due to Na V 1.8 inhibition is not caused by a reduction of overall Na + current (Supplementary information, Supplementary Fig. 4).
As the existence of Na V 1.8 and its role in cardiomyocytes is still a matter of debate, we generated homozygous Na V 1.8 knockout (KO) lines by using CRISPR-Cas9 in human induced pluripotent stem cells (iPSC) and differentiated these cells into 2-month-old cardiomyocytes. Sanger sequencing demonstrated frameshifts and premature stop codons on both alleles (Supplementary information, Supplementary Fig. 5). As the amplitude of I NaL is relatively small under healthy conditions we used isoproterenol (Iso, 50 nmol/l) to enhance I NaL . Control iPSC-cardiomyocytes treated simultaneously with Iso and PF-01247324 exhibited~60% less I NaL compared to Iso alone, (Fig. 2e, f). Most importantly, in Na V 1.8 KO iPSC-cardiomyocytes I NaL was reduced by~70% compared to the control iPSC-cardiomyocytes. PF-01247324 did not exert any further impact on I NaL in Na V 1.8 KO iPSCcardiomyocytes compared to untreated Na V 1.8 KO cells, underlining the specificity of the drug. Na V 1.8 modulates Ca 2+ homeostasis under enhanced CaM-KIIδc activity. In HF, enhanced I NaL can potently induce proarrhythmic SR-Ca 2+ release events 8,11 . Accordingly, we investigated whether inhibition of the Na V 1.8-mediated I NaL could attenuate the increase of the proarrhythmogenic SR-Ca 2+ spark frequency (CaSpF) caused by overexpression of CaMKIIδc ( Fig. 3a, b). We incubated CaMKIIδc +/T mouse cardiomyocytes with Na V 1.8 inhibitors and measured the CaSpF. A~50% reduction of CaSpF was observed in both Na V 1.8 inhibitor groups compared to untreated cells (Fig. 3a, b). These results display that SR-Ca 2+ leak due to increased CaMKIIδc expression and activity can be reduced by inhibiting Na V 1.8.
It is well known that inhibition of CaMKIIδc can attenuate SR-Ca 2+ leak 18,34 . However, therapeutic general inhibition of CaMKIIδc in humans may not be suitable because of its pivotal involvement in different vital pathways such as learning processes 35 . We explored whether the Na V 1.8 inhibitor PF-01247324 exerts similar effects comparable to the inhibition of CaMKIIδc. Incubation of human failing cardiomyocytes with either the CaMKIIδc inhibitor AIP or the Na V 1.8 inhibitor PF-01247324 resulted in a similar reduction of CaSpF compared to untreated cells (Fig. 3c, d). Furthermore, blocking CaMKIIδc and Na V 1.8 in parallel resulted in a significant reduction of CaSpF, comparable to AIP or PF-01247324 alone in human failing cardiomyocytes (Fig. 3c, d).
We further investigated whether Na V 1.8 inhibition modulates the Ca 2+ transient amplitude and SR-Ca 2+ load in cardiomyocytes isolated from CaMKIIδc +/T mice. Na V 1.8 inhibition using PF-01247324 did not pose any effect on either the Ca 2+ transient  (Fig. 3e-h). Furthermore, diastolic Ca 2+ , SR-Ca 2+ content, and SERCA2a activity were not affected by PF-01247324 ( Fig. 3i-k). Similar effects of PF-01247324 on Ca 2+ transient kinetics and SR-Ca 2+ content were observed in cardiomyocytes isolated from WT mice (Supplementary information, Supplementary Fig. 6).
Scn10a knockout improves survival in CaMKIIδc +/T mice in the absence of structural ventricular changes. To study whether inhibition of Na V 1.8 influences the development of HF, arrhythmogenesis, or survival in CaMKIIδc +/T mice, we crossbred these mice with Na V 1.8 knockout mice. Interestingly, SCN10A −/− /CaMKIIδc +/T mice showed a significantly improved survival compared to CaMKIIδc +/T (median survival 98 vs. 72 days, 64 vs. 43 animals, Hazard Ratio 0.6) as assessed in blinded investigations. Specifically, CaMKIIδc +/T mice showed only a 37% survival at 12 weeks, whereas SCN10A −/− /CaMKIIδc +/T died at a slower rate, with 67% survival at this age (Fig. 4a). These data illustrate that Na V 1.8 knockout is capable to counteract the lethal phenotype of CaMKIIδc overexpression to a relevant extent.
Detailed investigations of hearts from SCN10A −/− /CaMKIIδc +/T double-mutant and CaMKIIδc +/T mice by the age of 12 weeks exhibited comparably enlarged heart chambers (Fig. 4b). Heart weight to tibia length ratio was similarly increased in double-mutant and CaMKIIδc +/T mice (Fig. 4c). To investigate whether Na V 1.8 untreated. e Representative Ca 2+ transients stimulated at 1 Hz and caffeine-induced Ca 2+ transients in ventricular cardiomyocytes from CaMKIIδc +/T under untreated conditions. f Mean data ± SEM show no effect of PF-01347324 treatment on Ca 2+ transient amplitude at 1.0, 2.0, and 4.0 Hz stimulation (n = 13 cells/5 mice) compared to untreated (n = 10 cells/5 mice). g Ca 2+ -transient decay (90% of Ca 2+ -removal) RT90 was unchanged in PF-01347324-treated cells (n = 11 cells/5 mice) compared to untreated (n = 8 cells/5 mice). Data were presented as mean values ± SEM. h Representative Ca 2+ transients stimulated at 1 Hz and caffeine-induced Ca 2+ transients in ventricular cardiomyocytes from CaMKIIδc +/T treated with PF-01247324. i Diastolic Ca 2+ after addition of PF-01347324 (n = 13 cells/5 mice) compared to untreated cells (n = 10 cells/5 mice) at different stimulation frequencies was unchanged (one-way ANOVA with post hoc Bonferroni's correction, Fig. 3f, g, i). Data were presented as mean values ± SEM. j Mean and individual values ± SEM of caffeine-induced Ca 2+ transients (untreated: n = 7 cells/4 mice, PF-01247324: n = 11 cells/5 mice) did not differ between the groups (Student's t-test). k Ca 2+ -reuptake into the SR was not affected by inhibition of Na V 1.8 (untreated: n = 7 cells/4 mice, PF-01247324: n = 11 cells/5 mice), analyzed by Student's t-test. Data were presented as mean values ± SEM. knockout influences CaMKIIδc-mediated hypertrophy we prepared histological heart stainings. CaMKIIδc overexpression led to an increase of cardiomyocyte cross-sectional area (CSA) compared to WT and SCN10A −/− in mouse left ventricles. However, in doublemutant mice CSA did not differ from CaMKIIδc +/T alone (Fig. 4d, e). In addition, Scn10a knockout did not change the expression of Na V 1.5 (Scn5a) and CaMKIIδc on protein and mRNA levels in hearts from WT and CaMKIIδc +/T mice (Supplementary information, Supplementary Figs. 7, 8).
Reduction of proarrhythmic activity in SCN10A −/− /CaMKIIδc +/T cardiomyocytes. To test whether enhanced I NaL in CaMKIIδc +/T mice can be ameliorated by genetic knockout of Na V 1.8, we measured I NaL in isolated mouse cardiomyocytes (Fig. 5a, b). While CaMKIIδc +/T cardiomyocytes showed significantly enhanced I NaL ,  knockout of Na V 1.8 resulted in a~45% decrease in I NaL in SCN10A −/− /CaMKIIδc +/T mice compared with CaMKIIδc +/T . Of note, at basal/unstimulated conditions, I NaL did not differ between WT and SCN10A −/− . To evaluate whether chronic ablation of Na V 1.8 in CaMKIIδc +/T mice may reduce proarrhythmic cellular activity, we performed electrophysiological measurements (Fig. 5c-f). Patch-clamp experiments revealed that CaMKIIδc +/T cardiomyocytes exhibited approximately fivefold more delayed afterdepolarizations/min (DADs) compared to WT and SCN10A −/− . Scn10a knockout reduced the fraction of CaMKIIδc-induced DADs/min by~60% (Fig. 5c, d). Comparable observations were made regarding the occurrence of early afterdepolarizations (EADs, Fig. 5e, f). While WT and SCN10A −/− cardiomyocytes developed almost no EADs, 13.6 ± 3.2 EADs/min were recorded in CaMKIIδc +/T . Na V 1.8 knockout caused an 80% reduction of EADs/min in these cells. A detailed description of the action potential characteristics of these measurements is provided in the Supplementary information (Supplementary Table 1).
Furthermore, to evaluate whether inhibition of Na V 1.8 may decrease the number of cardiomyocytes exhibiting Ca 2+ -derived proarrhythmic events, we quantified the occurrence of Ca 2+ waves in mouse ventricular cardiomyocytes. The fraction of cardiomyocytes developing Ca 2+ waves was significantly less in SCN10A −/− /CaMKIIδc +/T versus CaMKIIδc +/T (14.8 vs 27.8%, Fig. 5g, h). However, some cardiomyocytes showed more than one Ca 2+ wave. Therefore, we calculated the occurring events per minute. In CaMKIIδc +/T the frequency of Ca 2+ waves was~2.5fold higher compared to WT and SCN10A −/− and was reduced by~55% in cardiomyocytes from SCN10A −/− /CaMKIIδc +/T compared to CaMKIIδc +/T (Fig. 5i). Of note, Ca 2+ transient amplitude and Ca 2+ extrusion from cytosol was unchanged between SCN10A −/− /CaMKIIδc +/T and CaMKIIδc +/T (Supplementary information, Supplementary Fig. 9). In summary, we describe a cellular rescue of the proarrhythmic CaMKIIδc +/T phenotype due to genetic ablation of Na V 1.8, which is associated with improved animal survival.
Scn10a knockout reduces ventricular arrhythmias in CaM-KIIδc +/T mice. To further investigate whether the proarrhythmic potential of Scn10a knockout on the cellular level is translatable into the in vivo situation, we implanted telemetric monitors into 8 week old SCN10A −/− /CaMKIIδc +/T and CaMKIIδc +/T mice. After 10 days of recovery from surgery, telemetric measurements were performed twice a week for 24 h over a period of 2 weeks. Baseline ECG characteristics are presented in the Supplementary information (Supplementary Table 2). There was no relevant difference in overall physical animal activity between the groups (Fig. 6b). Blinded analysis revealed that SCN10A −/− /CaMKIIδc +/T showed a strong trend towards a reduction of premature ventricular contractions (PVC) bỹ 93% compared to CaMKIIδc +/T (Fig. 6a, c, p = 0.08). Most importantly, the incidence of ventricular tachycardia (VT) was significantly lower (91%) in SCN10A −/− /CaMKIIδc +/T (Fig. 6a, d). Therefore, the observed improved survival of SCN10A −/− /CaM-KIIδc +/T -animals is associated with reduced ventricular arrhythmias.

Discussion
In our present study, we demonstrate that CaMKIIδc interacts with the neuronal sodium channel Na V 1.8 in human ventricular cardiomyocytes. Using different approaches in human and mouse cardiomyocytes, we demonstrated the relevance of Na V 1.8 for I NaL generation in HF and that an enhanced CaMKIIδc, indeed, regulates this Na V 1.8-driven I NaL . Isolated cardiomyocytes from SCN10A −/− /CaMKIIδc +/T compared to CaMKIIδc +/T mice exhibit reduced cellular arrhythmic events. While there was no change with respect to HF progression, i.e., similar left ventricular ejection fraction and chamber diameters, we found reduced ventricular arrhythmias and an improved animal survival of SCN10A −/− /CaMKIIδc +/T animals. Thus, we identified a modifiable proarrhythmic CaMKIIδc downstream target in the failing heart.
We recently found that Na V 1.8 is upregulated and thereby contributes to I NaL under conditions of HF and cardiac hypertrophy where CaMKIIδc activity is known to be enhanced 31,32 . Therefore, the aim of the current study was to investigate a potential crosstalk between Na V 1.8 and CaMKIIδc and its consequences on I NaL generation and cellular arrhythmogenety. Accordingly, we revealed an interaction of Na V 1.8 and CaMKIIδc in human ventricular myocardium of both non-failing and HF samples. CaMKIIδc is known to also interact with Na V 1.5 channels at the intercalated disc where Na V 1.5 and CaMKIIδc are part of a macro complex with Ankyrin-G and ßIV-spectrin 36 . The interaction with CaMKIIδc in this complex influences Na V 1.5 channel function in several cardiac disease states such as HF 14,21 . As Na V 1.8 was previously shown to interact with Ankyrin-G in neurons 37 in a similar fashion as it is known for Na V 1.5 in cardiomyocytes, similar mechanisms of Na V 1.8 interaction with CaMKIIδc like known for Na V 1.5 are conceivable. Several Serin residues as well as CaMKII-binding consensus motifs were found to be conserved between Na V 1.5 and Na V 1.8 (for details see Supplementary Fig. 10).
In a variety of cardiac pathologies, enhanced CaMKIIδc activity and expression is a key contributor to maladaptive electrical remodeling and thereby promotes arrhythmias 9,38-40 . CaMKIIδc can influence I NaL magnitude by phosphorylating Na V , which has been exclusively investigated for cardiac Na V 1.5 before, whereas a possible CaMKIIδc-dependent regulation of Na V 1.8 has not been investigated yet 14,21,22 . In our study, we investigated the contribution of Na V 1.8 to I NaL in an HF model that was exclusively induced by chronic CaMKIIδc overexpression. As previously shown, I NaL was augmented in cardiomyocytes from CaMKIIδc +/T compared to WT 21,40 , while these effects were ameliorated by the application of specific Na V 1.8 blockers. Therefore, at least a relevant part of CaMKIIδc-induced I NaL appears to be driven by Na V 1.8. Additional support for this conclusion comes from our I NaL measurements from human failing cardiomyocytes where CaMKIIδc activity and I NaL are both known to be increased in parallel 2,13,16,21,41 . Inhibition of either CaMKIIδc or Na V 1.8 reduced I NaL as separately demonstrated before 21,31 . However, simultaneous inhibition of Na V 1.8 and CaMKIIδc had no additive effect compared to CaMKIIδc inhibition alone. Therefore, it can be assumed that the majority of Na V 1.8-driven I NaL was already abolished by CaMKIIδc inhibition and hence, seems to be CaMKIIδc-dependent. In addition, Scn10a knockout in CaMKIIδc +/T mice resulted in a reduction in I NaL comparable to Na V 1.8 inhibition upon specific blockers.
In previous studies, the functional relevance of Na V 1.8 expression in cardiomyocytes was questioned as an application of specific blockers had no effects on peak I Na and I NaL in healthy and unstimulated cardiomyocytes 42,43 . These data are not in conflict with our findings, as we also did not observe differences in I NaL magnitude in cardiomyocytes from healthy mice, while clear effects were evident under conditions of enhanced I NaL either by chronic CaMKIIδc overexpression or isoproterenol treatment of iPSC-cardiomyocytes. Therefore, the interaction of Na V 1.8 with enhanced CaMKIIδc activity might be necessary to generate meaningful effects on cardiomyocyte electrophysiology while Na V 1.8 appears to play a negligible role in healthy cardiomyocytes. This establishes Na V 1.8 to be a disease-specific target.
An augmentation of I NaL was demonstrated to cause a Na + -dependent Ca 2+ overload and spontaneous SR-Ca 2+ release providing a substrate for cellular proarrhythmia 8,15,30,44 . We previously reported that inhibition of I NaL by specifically targeting Na V 1.8 is potent enough to reduce NCX reverse mode and thereby diastolic SR-Ca 2+ leak 11,31,32 . In the present work, we inhibited either CaMKIIδc or/and Na V 1.8 in failing humans and CaMKIIδc +/T mouse ventricular cardiomyocytes and correspondingly observed a decrease in I NaL and diastolic SR-Ca 2+ leak. A significant reduction of diastolic SR-Ca 2+ release events was already prominent after Na V 1.8 inhibition alone in failing human cardiomyocytes which was almost comparable to the effect caused by CaMKIIδc inhibition. Moreover, we did not observe a further reduction in SR-Ca 2+ leak when AIP was coadministrated with the Na V 1.8 blocker suggesting that effects of Nav1.8 inhibition on SR-Ca 2+ leak act via indirect inhibition of CaMKIIδc as it was previously shown for I NaL inhibition with TTX or Ranolazine 8,15 . The present findings, therefore confirm that Na V 1.8 inhibition abrogates the vicious proarrhythmic cycle between enhanced I NaL and CaMKIIδc.
Increased expression and enhanced activity of CaMKIIδc is not only known to increase proarrhythmic triggers in HF but is also associated with a strong HF phenotype and increased mortality in mice 9,13,16,45 . Our current findings illustrate that Na V 1.8 plays a relevant role for arrhythmogenesis under conditions of enhanced CaMKIIδc activity. For deeper mechanistic analysis, we crossbred our CaMKIIδc +/T mice with Scn10a knockout mice. In fact, we observed an improved survival in these SCN10A −/− /CaMKIIδc +/T mice but still with an unchanged typical phenotype of dilated cardiomyopathy in these blinded investigations. There may be three potential explanations for the improved survival, which were either reduced lethal arrhythmias, pump failure, or a combination of both. As we could demonstrate that induction and progression of HF is not influenced by the additional Scn10a knockout we propose a reduction of augmented I NaL with subsequent lower proarrhythmic activity to constitute the underlying mechanism of this improved survival. The link between enhanced I NaL and increased arrhythmia risk is rather complex. On the one hand, enhanced I NaL prolongs APD and may therefore give rise to the formation of EADs. On the other hand, enhanced I NaL causes Na + overload of the cardiomyocyte subsequently leading to Ca 2+ overload by activating NCX reverse mode 6,8,11,15 . This may trigger the occurrence of diastolic SR-Ca 2+ release and DADs due to enhanced CaMKIIδc-dependent RyR2 phosphorylation 8,10,15 . In our experiments, a reduction of I NaL in SCN10A −/− /CaMKIIδc +/T compared to CaMKIIδc +/T cardiomyocytes was observed. This was clearly associated with a reduction in the occurrence of EADs, DADs and diastolic SR-Ca 2+ -release events, thereby illustrating that the CaMKIIδc-induced proarrhythmic phenotype can be ameliorated on the cellular level by Na V 1.8 knockout.
In a recent study by our groups, we could demonstrate that selective inhibition of diastolic SR-Ca 2+ leak by the compound Rycal S36 improves survival in an HF mouse model, where I NaL and CaMKIIδc activity were also described to be enhanced 10,46 . As in our study, improved survival was caused by a significant reduction of malignant arrhythmias, while the severity of HF was unchanged. Likewise, we observed less ventricular arrhythmias in SCN10A −/− /CaMKIIδc +/T mice in vivo. It is well known that individuals with structural heart disease are at increased risk for sustained ventricular tachycardia and ventricular fibrillation 47 . Moreover, sustained ventricular tachycardia may degenerate to ventricular fibrillation 47 . In addition, clinical trials have shown an association of PVCs and ventricular tachycardia with adverse outcomes in patients with dilated cardiomyopathy [47][48][49] . The Na V 1.8 knockout significantly reversed a relevant part of the arrhythmogenic CaMKIIδc transgenic substrate in vitro and in vivo which is known to be associated with sudden cardiac death. However, due to the sporadic nature of sudden cardiac death, we were not able to correlate mortality with ventricular arrhythmias in these CaMKIIδc transgenic mice as this is technically and ethically not feasible. We are also not aware of any other experimental trial that investigates such a high number of HF mice with implanted telemetric recorders during death struggle. Finally, a chicken or the egg question would remain, as detected rhythm disorders cannot be differentiated to be either caused primary by arrhythmia or secondary due to hypoxia.
Persistent CaMKIIδc hyperactivity in hearts of HF patients has several detrimental effects such as influencing myocardial remodeling and promoting arrhythmias 33,39 . Despite the fact that several CaMKIIδc inhibitors have been demonstrated to effectively reduce arrhythmogenesis in HF in vitro 46,50 , CaMKIIδc inhibition to manage cardiac arrhythmias has not been established in patients so far. This can be explained by the vast signaling network of this enzyme, which is required to keep a balance between pivotal off-target effects and the therapeutic benefits. Any other alternative downstream or upstream target of the CaMKIIδc pathway, which is involved in promoting arrhythmias may be a more specific therapeutic option. As it has consistently been shown that activation of CaMKIIδc is involved in enhancement of I NaL and vice versa 10,11,21,34,40 our proposed crosstalk between CaMKIIδc and Na V 1.8 causing enhanced I NaL in the presence of hyperactive CaMKIIδc may be a novel strategy to prevent detrimental CaMKIIδc-induced effects on cellular electrophysiology. Most importantly, we and others have shown that selective Na V 1.8 inhibition does not influence either peak Na + current or Na + current kinetics and thereby cardiac conduction as another critical regulator of proarrhythmia 11,29,31,42,51 .
In summary, the results of our current study demonstrate that increased CaMKIIδc activity in HF contributes to enhanced I NaL via interaction with Na V 1.8. This augmented I NaL generated by Na V 1.8 detrimentally influences cellular electrophysiology by increasing RyR2-leakiness and can give rise to cellular proarrhythmic events in HF cardiomyocytes. Importantly, the results of the present study demonstrate that pharmacological inhibition and genetic ablation of Na V 1.8 can specifically reverse these detrimental proarrhythmic effects in the human HF heart and in our SCN10A −/− /CaMKIIδc +/T mouse model. This is associated with improved animal survival. Therefore, targeting Na V 1.8 as a specific substrate of increased CaMKIIδc activity may constitute a promising antiarrhythmic approach in HF which merits further translational investigation.

Methods
Human tissue samples. The study conforms to the declaration of Helsinki and was approved by the local ethics committee. All participants were informed about the study prior to inclusion. All patients signed informed consent. We obtained left ventricular tissues from explanted hearts of patients with end-stage HF (NYHA HF classification IV) who were undergoing heart transplantation. After explantation, the whole heart or myocardial tissues were immediately placed in a container having precooled cardioplegic solution containing (in mmol/L): NaCl 110, KCl 16, MgCl 2 16, NaHCO 3 16, CaCl 2 1.2, and glucose 11. Myocardial samples for Western blot and co-immunoprecipitation were immediately frozen in liquid nitrogen and stored at −80°C. The heart tissue for cell isolation was stored in cooled cardioprotective solution containing (in mmol/L): 156 Na + , 3.6 K + , 135 Cl − , 25 HCO 3 − , 0.6 Mg 2+ , 1.3 H 2 PO 4 − , 0.6 SO 4 2− , 2.5 Ca 2+ , 11.2 glucose, and 10 2,3-butanedionmonoxime (BDM) aerated with 95% O 2 and 5% CO 2 . We used healthy myocardium from healthy donor hearts that could not be transplanted for technical reasons as controls co-immunoprecipitation. Patient characteristics are presented in the Supplementary information (Supplementary Table 3). The study was approved by the local ethics committee of the University Medicine of Goettingen.
CaMKIIδc transgenic mice. CaMKIIδc transgenic (CaMKIIδc +/T ) mice were generated using an α-MHC promoter 13 . We used 12-to 14-week-old mice for electrophysiology experiments. The animal investigations conform to the "Guide for the Care and Use of Laboratory Animals" published by the US National Institutes of Health (8th edition, revised 2011) and the guidelines from Directive 2010/63/EU of the European Parliament on the protection of animals used for scientific purposes. The mouse study was approved by the local ethics committee of the University Medicine of Goettingen and the public authority on animal welfare.
Generation of SCN10A −/− /CaMKIIδc +/T mice. To generate the SCN10A −/− / CaMKIIδc +/T mouse model we crossbred CaMKIIδc +/T mice from a Black-Swiss strain with Na V 1.8 knockout mice (SCN10A −/− ) from a C57BL/6 strain. Eight weeks old male CaMKIIδc +/T mice were mated with 8 weeks old female SCN10A −/− mice. This breeding resulted in a genotype carrying a WT or transgenic (TG) allele of the CaMKIIδc gene and a heterozygote knockout of Scn10a. From this first generation, 8 weeks old male mice carrying a TG CaMKIIδc allele (CaMKIIδc +/T ) were mated with 8 weeks old female mice carrying homozygous WT CaMKIIδc alleles and heterozygous for Scn10a gene. This breeding produced 25% mice with the desired genotype having heterozygous CaMKIIδc gene (CaM-KIIδc +/T ) and homozygous Scn10a knockout (SCN10A −/− /CaMKIIδc +/T ). Mice carrying a heterozygote Scn10a gene were excluded from survival analysis and experimental studies. A detailed scheme of the crossbreeding is displayed in the Supplementary information ( Supplementary Fig. 11). The breeding rooms were maintained at 20-22°C with 50-60% humidity. Mice were housed in a room with a 12-h light/dark cycle with ad libitum access to water and food.
Survival of CaMKIIδc +/T mice was documented in a database of the local facility of animal experiments. A survival curve was prepared from this database and a comparison of survival was made between the groups CaMKIIδc +/T and SCN10A −/− /CaMKIIδc +/T . Animals used for experiments were excluded from survival analysis. Before scarifying animals, body weight was recorded while after explantation heart weight and tibia length were measured to tabulate animal size. Some of the animals showed signs of heart failure with fluid retention while some had signs of terminal illness therefore, heart weight to tibia length ratio was analyzed instead of heart weight to body weight ratio. The mouse study was approved by the local ethics committee of the University Medicine of Goettingen and the public authority on animal welfare.
Directed cardiac differentiation of iPSCs was done by manipulation of the Wnt signaling pathway 52 . Briefly, iPSCs were cultured in E8 medium as a monolayer on Geltrex-coated 12-well plates to a confluence of 85-95%. Wnt signaling was initiated by a medium change to RPMI1640 GlutaMAX (Thermo Fisher Scientific) supplemented with human recombinant albumin, L-ascorbic acid 2-phosphate including the GSK3β inhibitor CHIR99021 (4 µmol/L, Millipore). After 48 h, cells were treated with fresh medium supplemented with the inhibitor of Wnt production 2 (IWP2) (5 µmol/L, Millipore) for 48 h. On day 6, the medium was changed to RPMI1640 GlutaMAX with 1x B27 with insulin (Thermo Fisher Scientific) with a medium change every 2-3 days. Cardiomyocytes were purified by using a metabolic selection for 2-4 days with 4 mmol/L lactate as carbon source after 15-20 days of differentiation 53 . Cells were cultured for 60 days and passaged onto glass-bottom FluoroDishes (WPI, 30 K/dish) by trypsinizing for 3 min at 37°C. Cells settled for 7 days prior to further measurements with medium change every 2 days. IPSC-cardiomyocytes were analyzed 8-10 weeks after initiation of differentiation except when mentioned otherwise. Following differentiation, purity of iPSC-CM was determined by flow analysis (>90% cardiac TNT+) and qPCR of ventricular cardiomyocytes marker MLC2v. Four to five differentiation experiments into ventricular iPSC-CMs of two Na V 1.8 knockout lines and the corresponding healthy isogenic control line were used.

Cardiomyocyte isolation
Human. Human myocardium was rinsed, cut into small pieces, and incubated at 37°C in a spinner flask filled with Joklik-MEM solution (JMEM; AppliChem, Darmstadt, Germany) containing 1.0 mg/mL collagenase (type 1, 185 U/mg; Worthington Biochem, New Jersey, USA) and trypsin (2.5 g/L, Life Technologies, Carlsbad, California, USA). After 45 min of digestion, the supernatant was discarded, and the tissue was incubated with fresh JMEM solution containing only collagenase. The solution was incubated for 10 to 20 min until cardiomyocytes were disaggregated using a Pasteur pipette. The supernatant containing disaggregated cells was removed and centrifuged (700 r.p.m., 5 min). Fresh JMEM with collagenase was added to the remaining tissue. This procedure was repeated 4 to 7 times. After every step, the centrifuged cells were resuspended in JMEM medium containing (in mmol/L): 10 taurine, 70 glutamic acid, 25 KCl, 10 KH 2 PO 4 , 22 dextrose, 0.5 EGTA, and 10% bovine calf serum (pH 7.4 adjusted with KOH at room temperature). Only cell solutions containing elongated, non-granulated cardiomyocytes with cross-striations were selected for experiments, plated on laminin-coated recording chambers, and allowed to settle for 30 min.
Immunofluorescence staining of human isolated cardiomyocytes. Isolated human ventricular cardiomyocytes were allowed to attach the poly-L-lysine coated glass chambers for 30 to 45 min at room temperature. Cardiomyocytes were fixed in ice-cold 99.2% ethanol for 20 min at −20°C. After fixation, cardiomyocytes were washed with phosphate buffer saline (1X PBS) and subsequently blocked and permeabilized with 0.5% Triton X-100 and 5% BSA (bovine serum albumin) in PBS at room temperature.
After blocking, cells were washed with PBS (3х 5 min) and incubated with a primary antibody: mouse monoclonal anti-Na V 1. Patch-clamp experiments I NaL measurements. To measure I NaL in human and mouse ventricular cardiomyocytes, a ruptured whole-cell patch-clamp was performed at room temperature. The resistance of the pipette was between 2 and 3 mega-Ohm (MΩ) when filled with pipette solution containing (in mmol/L): 95 CsCl, 40 Cs-glutamate, 10 NaCl, 0.92 MgCl 2 , 5 Mg-ATP, 0.3 Li-GTP, 5 HEPES, 0.03 niflumic acid (to block Ca 2+activated chloride current), 0.02 nifedipine (to block Ca 2+ current), 0.004 strophanthidin (to block Na + /K + ATPase current) 1 EGTA, and 0.36 CaCl 2 (free [Ca 2+ ] i ,100 nmol/L) (pH 7.2 with CsOH at room temperature). Cardiomyocytes were maintained in the bath solution containing (in mmol/L): 135 NaCl, 5 tetramethylammonium chloride, 4 CsCl, 2 MgCl 2 , 10 glucose, 10 HEPES (pH 7.4 with CsOH at room temperature). To minimize contaminating Ca 2+ currents during I NaL measurements, Ca 2+ was omitted from the bath solution. I NaL was measured only in those cardiomyocytes where a seal of more than 1 giga-Ohm was achieved and the access resistance remained <7 MΩ. When whole-cell patch configuration was achieved, cardiomyocytes were given a period of 3 min to be stabilized before conducting measurements. Then cardiomyocytes were held at −120 mV before depolarization to −35 mV for a duration of 1000 ms with 10 pulses and a basic cycle length of 2 s. I NaL was measured as integral current amplitude between 100 and 500 ms and was normalized to the membrane capacitance. In some experiments, cardiomyocytes were incubated with PF-01247324 (1 μmol/L, Sigma) for 15 min before initiation of I NaL measurements. After patch rupture cardiomyocytes were allowed to stabilize for at least 3 min. Currents were elicited using a voltage step protocol with 10 mV increments ranging from −120 to +30 mV (step duration: 1 s, holding potential: −120 mV). I NaL was determined as the mean current density between 180 to 190 ms of each depolarization step. PF-sensitive I NaL was calculated as the difference between the mean I NaL densities of the vehicle and PF-treated cardiomyocytes at each membrane potential.
Membrane potential recordings. The whole-cell patch-clamp technique in currentclamp configuration was used to measure membrane potential in single isolated cardiomyocytes. Microelectrodes after cell stabilization, which was~5 min after rupture. Data were collected using Patchmaster 2.0 (HEKA Elektronik) and was analyzed using LabChart 7.37 (ADInstruments). Afterdepolarizations such as EADs and DADs were counted under continuous stimulation with 0.5 Hz. An EAD was counted as an EAD when a renewed depolarization of at least 1 mV occurred before complete membrane repolarization. Criteria for counting DADs were similar, but after achieving complete membrane repolarization.
Ca 2+ spark and Ca 2+ wave measurement. Human and mouse ventricular cardiomyocytes were loaded with a Fluo-4AM (10 µmol/L for 15 min, Molecular Probes) at room temperature. For some experiments, either A-803467 (30 nmol/L for 15 min, Sigma) or PF-01247324 (1 µmol/L for 15 min, Sigma) was added to the loading buffer to inhibit Na V 1.8. Cells were perfused with Tyrode's solution containing (in mmol/L): 136 NaCl, 4 KCl, 0.33 NaH 2 PO 4 , 4 NaHCO 3 , 2 CaCl 2 , 1.6 MgCl 2 , 10 HEPES, 10 glucose (pH 7.4 adjusted with NaOH at room temperature) and the respective active agent. The cardiomyocytes were washed with Tyrode's solution for 15 min for de-esterification of the Ca 2+ dye. Ca 2+ spark measurements were performed with a laser scanning confocal microscope (LSM 5 Pascal, Zeiss) using a 40× oil immersion objective. Fluo-4AM was excited by an argon ion laser (488 nm), and emitted fluorescence was collected through a 505-nm long-pass emission filter. Fluorescence images were recorded in the line scan mode with 512 pixels per line (width of each scanline: 38.4 µm) and a pixel time of 0.64 µs. One image consists of 10,000 unidirectional line scans, equating to a measurement period of 7.68 s. Experiments were conducted under resting conditions after loading the sarcoplasmic reticulum with Ca 2+ by field stimulation at 1 Hz and 20 V. Ca 2+ sparks were analyzed with the program SparkMaster for ImageJ. The mean spark frequency of the respective cell (CaSpF) resulted from the number of sparks normalized to cell width and scan rate (100 µm/s). Cardiomyocytes exhibiting major arrhythmic events (Ca 2+ waves, spontaneous Ca 2+ transients, and spark clouds) were excluded from the quantification of the Ca 2+ spark. Cardiomyocytes showing Ca 2+ waves were analyzed separately. The percentage of cardiomyocytes exhibiting Ca 2+ waves was calculated and compared between the groups. Some cardiomyocytes showed more than one event therefore we calculated the events per time and represented this ratio as waves per minute. The time to the first arrhythmic event was measured in seconds.
Ca 2+ transient measurements. Cardiomyocytes were isolated from CaMKIIδc +/T mouse ventricles, plated as described above, and incubated with a Fura-2AM loading buffer (10 µmol/L, Molecular Probes, Eugene, Oregon, USA) for 20 min. After staining, the cardiomyocytes were washed with experimental solution (as described above in the Ca 2+ sparks measurement section, at room temperature) for 15 min before measurements were started to enable complete de-esterification of intracellular Fura-2AM and to allow cellular rebalance of Ca 2+ cycling properties. During measurements, cardiomyocytes were continuously superfused with an experimental solution. Measurements were performed with a Motic AE31 microscope provided with a fluorescence detection system (ION OPTIX Corp.) at room temperature. Cells were excited at 340 and 380 nm and the emitted fluorescence was collected at 510 nm. The intracellular Ca 2+ level was measured as the ratio of fluorescence at 340 and 380 nm (F340/F380 nm in ratio units, r.u.). Ca 2+ transients were recorded at steady-state conditions under constant field stimulation (1.0, 2.0, and 4.0 Hz, 20 V). The recorded Ca 2+ transients were analyzed with the software IONWizard R (ION OPTIX Corp.).
Analysis of Ca 2+ -transients from confocal line scans. From confocal line scans that were used for quantification of diastolic Ca 2+ waves in cardiomyocytes from CaMKIIδc +/T and SCN10A −/− /CaMKIIδc +/T mice Ca 2+ -kinetics were analyzed from the Ca 2+ -transients at the beginning of each line scan before 3 Hz steadystate stimulation was paused. Ca 2+ -transient amplitude (F, normalized to diastole, F0) and time to half-maximal decay (RT50), and time to 90% of maximal decay were calculated from confocal line scan images using ImageJ software.
Quantitative real-time PCR (RT-qPCR). Human or mouse cardiac tissues, isolated human cardiomyocytes or iPSC-cardiomyocytes were snap-frozen in liquid nitrogen and stored at −80°C. RNA was isolated by use of the ReliaPrep TM RNA Tissue Miniprep System (Promega). About 200 ng RNA was reverse transcribed into cDNA using standard protocols (QuantiNova Reverse Transcription Kit (QIAGEN) for mouse tissue and iScript cDNA Synthesis Kit (Bio-Rad) for human samples). For qPCR, 10 µL SYBR Green PCR Master Mix (Bio-Rad), 7 µL nucleasefree water, 1 µL forward and 1 µl reverse Primer, and 1 µL of cDNA were mixed. Q-PCR was carried out using the CFX ConnectTM Real-Time System (Bio-Rad). Forty cycles of 15 s at 95°C followed by 1 min of 60°C were used and fluorescence was measured after each cycle. After 40 cycles melt curve analysis was performed to ensure the specificity of the products. Thresholds cycles were evaluated and normalized to housekeeping genes and controls. A list of all primers used is presented in Supplementary Table 4.
Histology. Freshly explanted hearts were fixed in 4% buffered formaldehyde at 4°C for 24 h. LV tissue was harvested, fixed in 4% buffered formaldehyde overnight, paraffin-embedded, sectioned (5 μm), and stained with fluorescein-conjugated wheat germ agglutinin (WGA-Alexa Fluor 594; Invitrogen, USA) for crosssectional area (CSA) assessment. At least 300 randomly selected cardiomyocytes per animal from different sections and different regions (basal, mid-ventricular, apical) of each heart were measured using the ImageJ software (Bethesda, USA).
Mouse echocardiography. For echocardiography, mice were anesthetized using 1.5% isoflurane, and echocardiography was performed using a VS-VEVO 660/230 (VisualSonics, Canada). During this procedure, the core temperature was maintained at 37°C, and heart rates were kept consistent between experimental groups at 400-500 b.p.m. Electrocardiogram monitoring was obtained using hind limb electrodes. LV geometry and systolic function were assessed by using standard 2D parasternal long and short-axis views. The examiner was blinded towards group assignment.
Telemetric ECG recordings in mice. Mice were implanted with an intraperitoneal telemetric ECG transmitter (TA11ETA-F10, Data Sciences International) with its Cable in lead II configuration. After a postoperative recovery period of 10 days, we recorded ECGs during periods of normal activity (24 h continuous recording, twice a week for 2 weeks). ECG parameters were analyzed, and the QT interval was corrected (QTc) using Bazett's formula, QTc = QT/(RR/100)1/2 established for mice with QT and RR expressed in ms. Premature ventricular contractions and ventricular tachycardia (>3 beats) were counted per 24 h and a medium value was per animal was calculated from all measurements. Further, physical activity was recorded by the telemetric monitors.
Statistics. All data were expressed as the mean ± standard error of the mean (SEM). Where appropriate, one-way ANOVA with multiple comparison tests (post hoc Bonferroni's and two-stage step-up methods of Benjamini, Krieger, and Yekutieli correction) was used. Mixed-effects analysis with Holm-Sidak's post hoc test was used to analyze I-V curves. Otherwise, Student's unpaired t-tests were used. The Chi-Square test was used to compare the occurrence of diastolic Ca 2+ waves. Log-rank (Mantel-Cox) test was used to compare survival between CaM-KIIδc +/T and SCN10A −/− /CaMKIIδc +/T mice. Two-sided p ≤ 0.05 was considered significant.
Reporting Summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All data generated or analyzed during this study are available within the Article and its Supplementary Information. All raw data supporting the findings from this study are available from the corresponding author upon reasonable request. Source data are provided with this paper. Any remaining raw data will be available from the corresponding author upon reasonable request.