Empagliflozin Inhibits Cardiac Late Sodium Current by Ca/Calmodulin-Dependent Kinase II

Recently, it was published in Circulation that empagliflozin inhibits H₂O₂-induced cardiac late sodium current (late I_Na).¹ Using computational modeling and point mutagenic approaches, Philippaert et al suggested a possible site of empagliflozin-binding within Na_V1.5 similar to that of local anesthetics, supportive of direct drug binding to Na_V1.5, although this remains to be determined conclusively and alternative mechanisms may exist.¹ We have previously shown that CaMKII (Ca/calmodulin-dependent kinase II) binds to Na_V1.5, stimulates late I_Na, and affects its H₂O₂-dependent regulation.^2,3 We also demonstrated that empagliflozin inhibits CaMKII in failing human and murine cardiomyocytes.⁴

Here we show that inhibition of H₂O₂-induced late I_Na by empagliflozin cannot solely be mediated by direct drug binding but depends on CaMKII-dependent phosphorylation of Na_V1.5 at serine 571. We demonstrate that empagliflozin inhibits late I_Na in patients with aortic stenosis (AS) and phenotypic features of heart failure (HF) with preserved ejection fraction.

Raw data/analytic methods can be made available for purposes of reproducing results or replicating procedures. Human tissue/proprietary antibodies cannot be made available because of legal constraints. Experiments conform to the Declaration of Helsinki. Human/murine studies were approved by institutional committee. Written informed consent was obtained from patients before tissue donation. Left ventricular samples were obtained from septal resections of 11 patients (8 male/3 female, age 69.3±2.6 years) with AS undergoing valve replacement. Patients had a HF with preserved ejection fraction–like phenotype with hypertrophy and preserved ejection fraction (59.4±1.7%). Murine models of CaMKIIδ knock-out (CaMKIIδ^-/-),³ inhibition of CaMKII-dependent Na_V1.5 phosphorylation at serine 571 (S571A), and with CaMKII phosphomimetic Na_V1.5 S571E mutation were tested for involvement of CaMKII-Na_V1.5 phosphorylation. Isolated ventricular myocytes were incubated (30 min) with empagliflozin (1 µmol/L) or control (dimethyl sulfoxide). Some cardiomyocytes were incubated with inhibitors of open-state Na channel inactivation (ATX-II or veratridine) or lidocaine (100 µmol/L, 30 min) for direct Na channel inhibition. H₂O₂ (100 µmol/L, 5 min) was used to induce reactive oxygen species, which stimulate late I_Na in HF via CaMKII³ (tested with CaMKII-inhibitor myristoylated-autocamtide-2-related inhibitory peptide (AiP); 2 µmol/L, 30 min). For some experiments, empagliflozin was washed in to ATX-II or H₂O₂ preincubated myocytes.

Late I_Na was measured as described previously.^2,3 Resting membrane potential was held at –120 mV and I_Na elicited by depolarizing to –20 mV for 1000 ms, quantified by integrating from 100 to 500 ms of the start of depolarization (normalized to membrane capacitance). Western blots used human ventricular tissue exposed to empagliflozin/vehicle (30 min).⁴ Data were analyzed using mixed-effects analysis with Holm-Sidak, linear mixed model with random factor “individual” and Sidak correction, or paired t test (GraphPad Prism 9).

We demonstrate that late I_Na can be reduced by empagliflozin in ventricular myocytes from patients with AS similar to CaMKII-inhibitor AiP (Figure [A]). ATX-II–dependent (Figure [B]) enhancement of late I_Na in murine wild-type cardiomyocytes was not affected by empagliflozin (not even at 10 and 100 µmol/L), or after wash-in (at 1 µmol/L) to ATX-II preincubated myocytes, which would be expected if empagliflozin were a direct Na_V1.5 inhibitor. Moreover, wash-in of empagliflozin (up to 10 µmol/L) also did not inhibit late I_Na in myocytes preincubated with a moderate concentration of veratridine (16 nmol/L, experimentally determined as EC₅₀ by dose–response; data not shown). In sharp contrast, both veratridine and ATX-II–enhanced late I_Na were blocked by lidocaine (not shown). Empagliflozin robustly inhibited H₂O₂-induced late I_Na (Figure [C]), with maximal efficacy at 6 minutes but not at 2 minutes after onset of exposure (late I_Na integral during wash-in: 0 minutes, –50.8±4.3 A*F^-1 [ampere*farad]*ms; 2 minutes, –39.9±4.6 A*F^-1*ms, P=0.0934 versus 0 minutes; 4 minutes, –24.2±4.6 A*F^-1*ms, P=0.0007 versus 0 minutes; 6 minutes, –17.2±4.0 A*F^-1*ms, P<0.0001 versus 0 minutes, n=6). No additional effect of empagliflozin on late I_Na was observed with AiP (not shown) or in myocytes lacking either CaMKIIδ (CaMKIIδ^−/−) or CaMKII-dependent Na_V1.5 phosphorylation at serine 571 (S571A, Figure [C]). Accordingly, the enhanced late I_Na in mice with CaMKII phosphomimetic Na_V1.5 S571E was blocked by neither empagliflozin nor AiP (Figure [D]). In contrast, lidocaine inhibited late I_Na in S571E cells, underscoring that empagliflozin primarily acts by CaMKII-Na_V1.5 phosphorylation. Empagliflozin dose-response revealed an IC₅₀ for inhibition of H₂O₂-dependent late I_Na of 0.086 µmol/L in murine myocytes (not shown). Empagliflozin inhibited CaMKII autophosphorylation and CaMKII-dependent phosphorylation of Na_V1.5 in AS and HF (Figure [E and F]).

Figure.

Late I_Na inhibition by empagliflozin requires CaMKII. A, Original recordings and mean data of empagliflozin- or AiP-mediated inhibition of late I_Na in human ventricular cardiomyocytes from patients with AS (n=patients). B, Original recordings and mean data of late sodium current (late I_Na) in murine cardiomyocytes from wild-type (WT) or CaMKIIδ^-/- mice (n=cells per mice). The ATX-dependent enhancement of late I_Na could not be blocked by empagliflozin. C, In contrast, the H₂O₂-dependent stimulation of late I_Na was blocked by CaMKII inhibition (AiP, CaMKII^−/−), by transgenic inhibition of CaMKII-dependent Na_V1.5 phosphorylation (S571A), or in the presence of empagliflozin. D, In contrast with local anesthetic lidocaine, neither empagliflozin nor AiP could block enhanced late I_Na in mice with phosphomimetic substitution of glutamic acid for serine at 571 (S571E). E and F, Western blots of cardiomyocytes on empagliflozin show reduced CaMKII-autophosphorylation (T287) and reduced CaMKII-dependent Na_V1.5 phosphorylation. For comparison of multiple groups, mixed-effects analysis plus Holm-Sidak (A) or linear mixed model plus Sidak were performed. For comparison of 2 groups, paired t test was done (F). A indicates ampere; AIP, autocamtide-2-related inhibitory peptide; AS, aortic stenosis; ATX II or ATX, Anemonia viridis toxin 2; CaMKII, Ca/calmodulin-dependent kinase II; CaMKIIdelta-/-, CaMKII delta knock out δ; E, Empa; Empa, empagliflozin; F, farad; GAPDH, Glyceraldehyde 3-phosphate dehydrogenase; HF, heart failure; kDa= kilo Dalton; ms, miliseconds; p, phosphorylated; pA, picoampere; S571A and S571E: Nav1.5 with a phosphomimetic mutation at Ser571 (S571E), or Nav1.5 with the phosphorylation site ablated (S571A); V, vehicle; and WT, wildtype.

In conclusion, inhibition of late I_Na by empagliflozin is at least in part caused by inhibition of CaMKII-dependent regulation of Na_V1.5.^2,4 If cardiac Na channels were solely directly inhibited, empagliflozin, like local anesthetics, should have blocked ATX-II/veratridine–stimulated late I_Na, but it did not. Nevertheless, the target of empagliflozin in the heart remains unclear,⁵ and further research is needed to better understand direct versus indirect effects on late I_Na. We demonstrate that empagliflozin also inhibits late I_Na in patients with AS and features of HF with preserved ejection fraction, which may reduce the propensity for arrhythmias and contribute to the positive results of the EMPEROR-Preserved trial (Empagliflozin Outcome Trial in Patients With Chronic Heart Failure With Preserved Ejection Fraction).

Article Information

Acknowledgments

J.M., M.J.B., and S.W. designed experiments, interpreted data, wrote the article, and are responsible for the integrity of the article. J.M., M.J.B., T.S., M.T., Z.P., T.J.H., H.M., and S.P. acquired data and revised the article. S.S., P.J.M., and L.S.M. revised the article for critical intellectual content.

Sources of Funding

J.M. is funded by the German Cardiac Society Clinician Scientist program and by Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) grant MU 4555/2-1. L.S.M. is funded by DFG grant MA1982/5-1. S.W. and L.S.M. are funded by SFB 1350 TPA6 and the University of Regensburg ReForM C program. S.W. is funded by DFG grants WA 2539/4-1, 5-1, 7-1, and 8-1. M.J.B. is supported by German Heart Foundation/German Foundation of Heart Research grant F/50/20. S.P., T.S., and M.T. are funded by the Else-Kröner-Fresenius Stiftung (EKFS, 2019_A84). S.S. is funded by DFG (SO 1223/4-1) and EKFS (2017_A137). P.J.M. is funded by National Institute of Health grant R35 HL135754 and by Leducq Foundation. T.J.H. is funded by National Institute of Health grants R01 HL156652 and R01 HL135096.

Disclosures

None.