Changes in cardiac Na_v1.5 expression, function, and acetylation by pan-histone deacetylase inhibitors

Therapeutic histone deacetylase inhibition with vorinostat or romidepsin significantly reduces ventricular sodium current density and Na_V1.5 protein expression. A slight positive shift in the voltage activation curve, presumably the result of lysine acetylation, decreases the inactivation rate at low activating potentials but does not increase the late sodium current amplitude.

Abstract

Histone deacetylase (HDAC) inhibitors are small molecule anticancer therapeutics that exhibit limiting cardiotoxicities including QT interval prolongation and life-threatening cardiac arrhythmias. Because the molecular mechanisms for HDAC inhibitor-induced cardiotoxicity are poorly understood, we performed whole cell patch voltage-clamp experiments to measure cardiac sodium currents (I_Na) from wild-type neonatal mouse ventricular or human-induced pluripotent stem cell-derived cardiomyocytes treated with trichostatin A (TSA), vorinostat (VOR), or romidepsin (FK228). All three pan-HDAC inhibitors dose dependently decreased peak I_Na density and shifted the voltage activation curve 3- to 8-mV positive. Increases in late I_Na were not observed despite a moderate slowing of the inactivation rate at low activating potentials (<−40 mV). Scn5a mRNA levels were not significantly altered but Na_V1.5 protein levels were significantly reduced. Immunoprecipitation with anti-Na_V1.5 and Western blotting with anti-acetyl-lysine antibodies indicated that Na_V1.5 acetylation is increased in vivo after HDAC inhibition. FK228 inhibited total cardiac HDAC activity with two apparent IC₅₀s of 5 nM and 1.75 μM, consistent with previous findings with TSA and VOR. FK228 also decreased ventricular gap junction conductance (g_j), again consistent with previous findings. We conclude that pan-HDAC inhibition reduces cardiac I_Na density and Na_V1.5 protein levels without affecting late I_Na amplitude and, thus, probably does not contribute to the reported QT interval prolongation and arrhythmias associated with pan-HDAC inhibitor therapies. Conversely, reductions in g_j may enhance the occurrence of triggered activity by limiting electrotonic inhibition and, combined with reduced I_Na, slow myocardial conduction and increase vulnerability to reentrant arrhythmias.

NEW & NOTEWORTHY

Therapeutic histone deacetylase inhibition with vorinostat or romidepsin significantly reduces ventricular sodium current density and Na_V1.5 protein expression. A slight positive shift in the voltage activation curve, presumably the result of lysine acetylation, decreases the inactivation rate at low activating potentials but does not increase the late sodium current amplitude.

histone deacetylase (HDAC) inhibitors represent a novel class of small molecule therapeutics with indications for the treatment of numerous human disorders including cancer, neurodegenerative diseases, type 2 diabetes, and autoimmunity (20, 22, 24, 28, 58). HDAC inhibitors are divided into four major chemical subgroups, the hydroxamic acid derivatives, benzamides, cyclic peptides, and short chain fatty acids (50). The naturally occurring prototypical hydroxamic acid and cyclic peptide HDAC inhibitors trichostatin A (TSA) and romidepsin (depsipeptide, FK228) inhibit all 11 Zn²⁺-dependent human HDAC isoforms (1–11) with high (nM-μM) affinity and are considered to be pan-HDAC inhibitors (5). In contrast, the benzamide derivatives like entinostat (MS-275) and mocetinostat (MGCD-0103) are high affinity inhibitors of only the class I HDACs (HDAC1, 2, 3, and 8). Class IIb inhibitors target HDAC6 with relative specificity and class IIa (HDACs 4, 5, 7, and 9) are under development (11, 34). Four pan-HDAC inhibitors have received Food and Drug Administration (FDA) approval since 2006, all for the treatment of cutaneous and/or peripheral T-cell lymphoma (CTCL and/or PTCL) or multiple myeloma (MM) (14). Cardiac arrhythmias have been associated with the clinical use of HDAC inhibitors since 2006 and the reported global incidence of HDAC inhibitor-induced adverse cardiac side effects is 28.6% (17, 50, 55, 57). Two of the FDA approved HDAC inhibitors, romidepsin (Istodax) and panobinostat (LBH589, Farydak), carry warnings for severe ECG changes and cardiac arrhythmias including ventricular tachycardias, although grade 3 or higher QT interval prolongations and ventricular fibrillation have occasionally occurred with vorinostat (SAHA, VOR, Zolinza) and belinostat (PXD101, Beleodaq) (14, 15, 36).

We previously reported that TSA and VOR dose dependently decreased connexin43 (Cx43) expression and electrical coupling between ventricular cardiomyocytes (71). The Duchenne muscular dystrophic (mdx) mouse model exhibits a hyperacetylated protein state resulting in acetylated Cx43, a decrease in Na_V1.5 cardiac sodium channel protein expression, and an increased incidence of restraint-induced cardiac arrhythmias (8, 9). The N^ε-lysine acetylated Cx43 dissociates from ventricular gap junctions, which may account for the stress-induced increased arrhythmogenesis in mdx mice (9). Treatment of mdx mice with VOR reduced the incidence of stress-induced arrhythmias and improved Cx43 localization to the intercalated disc. Conversely, mdx mice exhibited a lower incidence of arrhythmias than wild-type mice when treated with aconitine, a sodium channel opener, and VOR administration increased the incidence of these arrhythmias in mdx mice, suggesting functional differences in the cardiac sodium current between control and mdx mice (8). Since loss of Cx43 myocardial gap junctions is associated with reductions in cardiac sodium current (I_Na) carried by the Na_V1.5 channel protein, we chose to study the effect of pan-HDAC inhibition on cardiac I_Na (25, 46). In this study we report our findings of TSA, VOR, and FK228 effects on the cardiac sodium current (I_Na). This is the first reported study to examine the effects of pan-HDAC inhibition on cardiac I_Na in wild-type cardiomyocytes and these three HDAC inhibitors all decreased I_Na density in a dose-dependent manner. Additionally, a slight 3- to 8-mV positive shift in the steady-state voltage-dependent activation curve was observed with high doses of TSA and VOR without a concomitant shift in the inactivation curve. No increases in late I_Na were observed, suggesting that the only contribution of pan-HDAC inhibition to I_Na-dependent cardiac arrhythmias may be attributed to decreased excitability and slowed conduction, especially when combined with decreased myocardial gap junction coupling.

MATERIALS AND METHODS

Cell culture.

Newborn C57BL/6 mice were anesthetized with isoflurane and the hearts excised in accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol (IACUC #263) was approved by the Institutional Animal Care and Use Committee (IACUC) of SUNY Upstate Medical University. The atria and ventricles were dissociated separately and cultured in Medium 199 (M199) supplemented with 10% FBS as previously described (71). Neonatal mouse ventricular myocytes (NMVMs) were cultured at low (≈2 × 10⁺⁵ cells/35 mm culture dish) or high (≈10⁺⁷ cells/35 mm culture dish) density for patch-clamp and real-time PCR (RT-PCR) or Western blot procedures, respectively. Cryopreserved human-induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CM, iCell cardiomyocytes) were purchased from Cellular Dynamics (CDI, Madison, WI) and plated and grown in culture media and prepared for electrophysiological experiments per manufacturer's instructions (37).

HDAC inhibitors.

Trichostatin A (TSA) was purchased from Calbiochem or ENZO Life Sciences and 1 mg was dissolved in DMSO and stored at −20°C. Vorinostat (VOR) and romidepsin (FK228) were purchased from Selleck Chemicals, dissolved in DMSO, and stored at −20°C. The 100-mM DMSO stock solutions of TSA, VOR, and FK228 were diluted to the desired experimental test concentrations in M199 or maintenance media and applied to NMVM or hiPSC-CM cultures overnight for 18–24 h before experimental procedures. Final DMSO levels were <0.005% (vol/vol).

Whole cell patch clamping.

Single whole cell patch electrode voltage-clamp experiments were performed on NMVMs and hiPSC-CMs using conventional procedures with an Alembic Instruments VE-2, Axopatch 1D, or Axopatch 200B patch-clamp amplifier, Digidata 1320A or 1440 A/D converter, and pClamp8.2 or 10.1 software (Molecular Devices). Voltage-gated sodium currents (I_Na) were elicited from a holding potential (V_h) of −120 mV during voltage steps from −90 to +50 mV in 5-mV increments for 150 ms using reduced NaCl solutions (Table 1). For the I_Na inactivation protocol, V_h was −120 mV and the prepulse voltage increased from −130 to −30 mV in +5-mV increments for 150 ms followed by a 30-ms activation step to −40 mV. In some experiments, 30 μM tetrodotoxin (TTX) was added to the bath to verify that the inward current was TTX sensitive. Gap junction conductance (g_j) measurements were obtained using conventional dual whole cell patch-clamp procedures as previously described (64, 71).

Table 1.

Electrophysiological solutions

Composition, mM	Normal Saline	Pipette Solution	Low Na+ Saline	INa Pipette Solution	hiPSC-CM Saline	CM Pipette Solution
NaCl	140	—	20	5	30	5
KCl	1.3	140	—	—	—	—
TMACl	—	—	100	115	—	—
TEACl	—	—	20	20	110	135
CsCl	—	—	—	—	5	5
NaH2PO4	1.0	—	—	—	—	—
CaCl2	1.8	3.0	1.8	3.0	1.8	3.0
MgCl2/(MgSO4)	(0.8)	1.0	1.0	1.0	1.0	1.0
Glucose	5.5	—	5.5	—	5.5	—
BaCl2 and CdCl2	—	—	0.1 and 0.1	—	—	—
BAPTA/(EGTA)	—	5.0	—	5.0	—	(5)
HEPES*	10	25	10	20	20	20

I_Na sodium current; hiPSC-CM, human-induced pluripotent stem cell-derived cardiomyocyte; TMACl, tetramethylammoniumCl; TEACl, tetramethylammoniumCl.

^*pH was titrated to 7.4 with 1 N NaOH, KOH, TEAOH, or CsOH.

Real-time PCR.

Cellular RNA was extracted from cultured NMVMs and 10-μl RT-PCR reactions were carried out on 384-well plates using LightCycler 480 Real-Time PCR System (Roche) using previously published procedures (71). All results were normalized to Gapdh and control samples. Cycle threshold (C_T) values were determined by the apparatus and the quality of the PCR product was confirmed by analyzing the melt curve. Since HDAC inhibition is known to alter housekeeping gene expression (45), the control and VOR Gapdh C_T and relative (to control) expression level values were analyzed statistically and found to not be significantly different (paired t-test, P > 0.25). Primer sets for murine Cdh2, Gja1, Scn5a, and Slc8a, and Cdh2 were designed to span exon-intron regions of the gene of interest. The murine forward and reverse primer sequences were as follows: Cdh2, 5′-TATGTGATGACGGTCACTGC-3′ and 5′-GAAAGGCCATAAGTGGGATT-3′; Gapdh, 5′- TGCCACTCAGAAGACTGTGG-3′ and 5′-AGGAATGGGAGTTGCTGTTG-3′; Gja1, 5′-GAGAGCCCGAACTCTCCTTT-3′ and 5′-TGGAGTAGGCTTGGACCTTG-3′; Scn5a, 5′-CTTCACCAACAGCTGGAACA-3′ and 5′-GACATCATGAGGGCGAAGAG-3′; and Slc8a, 5′-TGAATCTTGCACGCTCATTA-3′ and 5′-CCAGGCACATCCAAAGTATC-3′.

Western blotting.

NMVMs were cultured for 3 days in 200 μM bromodeoxyuridine (BrdU)/M199 media supplemented with 10% fetal FBS and 10/10 U/μg penicillin/streptomycin. TSA, VOR, or 3 ml of control BrDU/M199 media was added overnight. On the fourth day, the NMVM cultures were harvested, lysed, and 15-μg protein samples underwent immunoblot procedures as previously described using EC Western Blot Detection Reagents (Bio-Rad Laboratories, Hercules, CA; Ref. 71). The rabbit polyclonal anti-Na_V1.5 (cat. no. ASC-005) primary antibody was purchased from Alomone Labs (Jerusalem, Israel), the mouse monoclonal anti-α-tubulin antibody (cat. no. T8203, clone AA13) was purchased from Sigma-Aldrich (St. Louis, MO), and the rabbit polyclonal anti-acetylated-α-tubulin and HDAC1 (cat. no. BML-SA452 and BML-SA401) antibodies were purchased from Enzo Life Sciences (Farmingdale, NY).

Immunoprecipitation.

Six adult C57BL/6 mice were intraperitoneally injected with either 10 mg/kg TSA (1 mg/ml in sterile saline: 5% DMSO, n = 3) or saline:DMSO (10 μl/gm body weight, n = 3) daily for 5 consecutive days and killed 4–6 h after the last injection as approved by the SUNY Upstate Medical University IACUC. The hearts were excised under ketamine/xylazine anesthesia and atria and ventricles were separated, and lysed in 1% Triton X-100 extraction buffer (50 mM Tris pH 8.0, 150 mM NaCl, 0.02% sodium azide, 1.0 mM PMSF, 1 μg Aprotinin, 1% Triton X-100, 1 mM Na₃VO₄, and 50 mM NaF) with protease inhibitors (Roche). The ventricular heart lysates were subjected to immunoprecipitation (IP) and Western blot procedures using primary anti-Na_V1.5 and anti-acetyl-lysine (Ac-K, Abcam, cat. no. ab21623) rabbit polyclonal antibodies and the Dynabeads Protein G (Thermo Fisher Scientific) immunoprecipitation procedures. The primary antibody was bound to the protein G magnetic beads according to manufacturer's instructions and the target antigen (Na_V1.5) was immunoprecipitated in an IP buffer containing 1% Triton X-100, 0.5% NP-40, 20 mM HEPES, 50 mM NaCl, and protease inhibitors (Roche), pH 7.4, using a magnet. The immunoprecipitate was washed with washing buffer, eluted with elution buffer, and subjected to SDS-PAGE electrophoresis and Western blot procedures with anti-Ac-K and anti-Na_V1.5 antibodies as previously described. The Na_V1.5 immunoprecipitation experiments were performed in triplicate.

HDAC activity assay.

Aliquots of 6 × 10⁺⁵ NMVMs per well (96-well plate) were grown in 200 μl of 200 μM BrdU/M199, exchanged daily. Cell wells were incubated with 2,000 pmoles of the acetylated Fluor-de-Lys substrate for 6 h during continuous (overnight) HDAC inhibition (71). Cell, media, and standard curve deacetylated substrate sample wells were developed according to manufacturer's directions (BML-AK503 HDAC fluorometric cellular activity assay kit; Enzo Life Sciences) and background subtracted relative fluorescence unit (RFU) counts were acquired with a BIO-TEK Synergy H1 plate reader (360-nm excitation, 460-nm emission).

Statistics.

Averaged values are presented as the mean ± SE. Statistical analyses were performed with the normality and one-way ANOVA tests using the Bonferroni method in Origin 8.6. Statistical comparisons of the late (100 ms) I_Na datasets were performed using the F-test comparison of two datasets function in Origin7.5.

RESULTS

Effects of pan-HDAC inhibition on cardiac Na⁺ current density.

To examine the effect of pan-HDAC inhibition on functional cardiac I_Na, cultured NMVMs were incubated overnight with 50 or 100 nM TSA, 1 or 5 μM VOR, or 10 nM FK228 and whole cell patch-clamp experiments were performed the next day (18–24 h later). To activate I_Na, NMVM membrane potential (V_m) was depolarized between −90 and +60 mV in 150 ms and 5-mV increments from a V_h of −120 mV. Examples of I_Na current traces elicited from control, 100 nM TSA-, 1 μM VOR-, and 10 nM FK228-treated NMVMs are illustrated in Fig. 1, A–D. Peak I_Na density decreased from control values of −36.2 ± 2.8 to −13.7 ± 2.5 pA/pF with 100 nM TSA, −17.6 ± 1.7 pA/pF with 5 μM VOR, and −15.1 ± 2.0 pA/pF with 10 nM FK228 (Figs. 1, F–H). Cellular input capacitance values were not significantly changed from control values (22.5 ± 1.4 pF, n = 20) except at the highest concentrations of TSA (100 nM; 15.3 ± 0.7 pF, n = 17) and VOR (5 μM; 16.2 ± 0.7 pF, n = 24) tested. The results are consistent with a pan-HDAC inhibitor-induced reduction in peak cardiac I_Na irrespective of whether the HDAC inhibitor belongs to the hydroxamate or cyclic peptide chemical group of HDAC inhibitors.

Fig. 1.

Reduction of cardiac sodium current I_Na by pan-histone deacetylase (HDAC) inhibition. Families of cardiac Na⁺ current traces are shown for 10-mV depolarizing voltage-clamp steps from −90 to +50 mV under control (A), 100 nM trichostatin A (TSA; B), 5 μM vorinostat (VOR; C), and 10 nM romidepsin (FK228; D). E: diagram of the V_m pulse protocol for activating I_Na. F–H: I_Na current density (pA/pF)-V_m relationships under control conditions and after an 18- to 24-h treatment with 50 or 100 nM TSA (F), 1 or 5 μM VOR (G), and 10 nM FK228 (H) illustrate dose-dependent decreases in peak I_Na with pan-HDAC inhibitor treatments.

Persistent activation of I_Na, caused by mutations in the Na_V1.5 sodium channel protein that inhibit inactivation (e.g., ΔKPQ) or the α-sodium channel toxins like aconitine, is known to cause cardiac arrhythmias via induction of early afterdepolarizations (EADs) (8, 65). Thus we measured the steady state current values at 100 ms for all control, 100 nM TSA, 1 μM VOR, and 10 nM FK228 experiments. Linear regression analyses of the whole cell current measured at the 100 ms time point of the voltage-clamp steps from −120 mV to −65 through 0 mV for all four experimental groups revealed no significant differences in the slopes (+0.37, +0.24, +0.34, and +0.44 pA/mV for control, TSA, VOR, and FK228, respectively) of all four datasets. The y-intercepts were significantly (P < 0.05) more negative (19 pA inward) for TSA and significantly more positive (6 pA outward) for FK228 relative to Control and VOR values. Taken together, there is no evidence of an increase in the late I_Na current in ventricular cardiomyocytes with either of the clinically approved HDAC inhibitors tested.

I_Na activation and inactivation.

A slight positive shift in the activation voltage is apparent in the I_Na current-voltage relationships in Fig. 1, F–H, so the peak sodium conductances (g_Na) were calculated by dividing the normalized peak I_Na density values by (V_m − E_Na) for each I–V curve in Fig. 1 and normalized by dividing by the average maximum g_Na (ḡ_Na) for each experimental group. The calculated Nernst potential for Na⁺ (E_Na) was +35 mV and the experimental reversal potential (E_rev) values were within 5 mV or ± 1 pA/pF of E_Na for all experimental groups. The V_m-dependent activation curves for TSA, VOR, and FK228 (Fig. 2, A–C) illustrate a +3- to +8-mV shift in the half activation voltage (V_½act) for all three pan-HDAC inhibitor experimental groups. Since I_Na activation was slightly affected by the HDAC inhibitor treatments, I_Na inactivation was examined using a prepulse protocol illustrated in Fig. 2E. The highest inhibitory concentrations of TSA (100 nM) and VOR (5 μM) tested shifted the control I_Na half-inactivation voltage (V_½inact) of −83 mV by less than ±1 mV (Fig. 2D). The effect of FK228 on I_Na inactivation was not examined. The small positive shift in the I_Na activation curves without an accompanying change in the inactivation curve decreases the V_m-dependent “window” for persistent I_Na activation (Fig. 2F).

Fig. 2.

Shift of I_Na activation by pan-HDAC inhibition. A–C: the apparent shift in V_m-dependent activation present in the I–V curves in Fig. 1, F–H, was examined by calculating the Na⁺ conductance (g_Na) and normalizing the data to the maximum g_Na (ḡ_Na) for each experimental group and plotting the activation curves for TSA (A), VOR (B), and FK228 (C) relative to the control activation curve. The data points were fitted with a Boltzmann function with a half-activation voltage (V_½act) of −49.8 mV for control, −46.6 and −47.0 mV for 50 and 100 nM TSA, −46.0, and −42.3 mV for 1 and 5 μM VOR, and −46.2 mV for 10 nM FK228, respectively. D: V_m-dependent inactivation was assessed using a prepulse protocol illustrated in E and no shift in the half-inactivation voltage (V_½inact) of −83.0 ± 0.5 mV was observed between control, 100 nM TSA, and 5 μM VOR conditions. F: comparison of the overlap between the g_Na inactivation and activation curves under control and 5 μM VOR conditions illustrating the decrease in the V_m-dependent “window” for persistent activation of I_Na.

The time dependence of I_Na inactivation was also examined by curve-fitting the decaying phase of the I_Na traces with a first-order exponential function. The V_m-dependent inactivation time constants (τ_h) were significantly slower in the −60- to −45-mV range for 100 nM TSA and 1 μM VOR but not for 10 nM FK228 (Fig. 3, A–C). There was no difference in the inactivation rates between control and HDAC inhibitor-treated ventricular cardiomyocytes within the −40- to −10-mV range, suggesting that the slow τ_h values at low activating voltages seen with pan-HDAC inhibition result from the positive shift in the activation curves (Fig. 2, A–C) and that I_Na inactivation rates were otherwise unaffected.

Fig. 3.

I_Na inactivation time constants. A–C: the average V_m-dependent inactivation time constants (τ_h) from the control, 100 nM TSA (A), 1 μM VOR (B), and 10 nM FK228 (C) I_Na current traces illustrate only a slight slowing of the inactivation rates at low activating potentials of −60 to −40 mV with no slowing of inactivation at fully activated potentials > −40 mV.

hiPSC-CMs.

To assess whether the pan-HDAC inhibitor-induced decrease in cardiac I_Na occurs in human ventricular myocytes, hiPSC-CMs were treated with 1 μM VOR and I_Na whole cell patch-clamp experiments were performed (Fig. 4). In hiPSC-CMs, the recommended human therapeutic dose of VOR reduced cardiac I_Na density by 52% relative to control experiments, indicating that NMVMs are a valid experimental model for studying the effects of HDAC inhibitors on cardiac I_Na. The mean whole cell current, measured between 100 and 150 ms, did not reveal any increase in late inward currents in the VOR hiPSC-CM group relative to the control group, consistent with our findings in NMVMs suggesting that late I_Na was not induced by clinically relevant HDAC inhibitor treatments.

Fig. 4.

Effect of pan-HDAC inhibition on human I_Na. A and B: examples of a family of I_Na traces obtained from a human-induced pluripotent stem cell-derived cardiomyocyte (hiPSC-CM) under control (A) and 1 μM VOR (B) conditions. C: I_Na current density (pA/pF)-V_m relationships under control conditions and 1 μM VOR conditions illustrate a significant reduction in I_Na density after pan-HDAC inhibitor treatment, verifying previous results obtained from neonatal mouse ventricular myocyte (NMVM) cultures.

Na_V1.5 expression.

Experiments in NMVMs and hiPSC-CMs consistently indicate that pan-HDAC inhibition reduced I_Na density, but the molecular basis for the decrease in cardiac I_Na is unknown. The shift in the I_Na activation curve is not sufficient to decrease peak I_Na levels and only slowed the inactivation kinetics at low activating potentials (less than −40 mV). To test the hypothesis that pan-HDAC inhibition may be decreasing the Na_V1.5 protein expression level, real-time PCR and Western blot experiments were performed on NMVM cultures under control and pan-HDAC inhibitor-treated conditions. The relative Scn5a mRNA abundance was less than that observed for Gja1, Cdh2, or Slc8 and all were <50% of Gapdh levels (Fig. 5A). As previously reported, the Gja1 and Cdh2 mRNA levels were significantly reduced by 1 μM VOR (71), but Scn5a and Slc8a mRNA levels were not significantly different from control values (Fig. 5B). 100 nM TSA did reduce Scn5a mRNA levels by 50% (n = 2) but did not reach statistical significance (P > 0.05). However, Western blot analyses, performed in triplicate, revealed that Na_V1.5 protein levels were significantly reduced at all concentrations of TSA and VOR tested. The decrease in I_Na density correlates closely (within 15–25% deviation from unity) with the reduction in Na_V1.5 protein levels for both TSA and VOR, suggesting that the molecular basis for the pan-HDAC inhibitor-induced decrease in cardiac I_Na is the dose-dependent reduction in Na_V1.5 protein expression. Real-time PCR results with 10 nM FK228 exhibited qualitatively similar abundances for the Chd2, Gja1, Scn5a, and Slc8a genes and a significant reduction in their mRNA levels relative to control values for all except the Scn5a gene transcript levels (Fig. 5E).

Fig. 5.

Effects of pan-HDAC inhibition on Na_V1.5 expression. A: the abundance of mRNA levels were measured by RT-PCR from high density NMVM cultures under control conditions for the Cx43 (Gja1), Na_V1.5 (Scn5a), N-cadherin (Cdh2), and the sodium-calcium exchanger (NCX, Slc8a) genes relative to Gapdh. B: the relative mRNA levels for these same cardiac genes after 1 μM VOR treatment were compared with control values and only Gja1 and Cdh2 levels were significantly reduced as previously reported. The Gapdh levels were not significantly different from control values (P > 0.25, paired t-test). C: representative Western blots for Na_V1.5 protein from TSA- and VOR-treated NMVM samples. HDAC1 was used as loading control and acetylated α-tubulin (Ac-α-t) as an indicator of increased protein acetylation. The experiments were performed in triplicate. D: densitometric scans of the Western blot data were performed to quantify the Na_V1.5 protein levels at low and high doses of TSA and VOR. Na_V1.5 protein levels were significantly reduced at both doses of TSA and VOR. E: the relative mRNA levels for the Slc8a, Gja1, and Cdh2 genes were significantly reduced by 10 nM FK228 treatment compared with control values but the Scn5a levels were not significantly affected.

Inhibition of cardiac HDAC activity by romidepsin.

To correlate the decreases of cardiac I_Na and Cx43 mRNA expression levels with the HDAC inhibitory activity of FK228, the total NMVM HDAC activity was measured using a fluorimetric HDAC activity assay in the presence of increasing [FK228] to detect the amount of deacetylated Fluor-de-Lys substrate using previously published procedures (71). As reported for TSA and VOR, inhibition of total cardiac HDAC activity by FK228 exhibited two affinities with the higher affinity site possessing at least twice the capacity of the lower affinity site (Fig. 6A). The IC₅₀ values for FK228 were 5.4 ± 0.6 nM and 1.75 ± 0.47 μM with capacities of 63.5 and 27.7% of total HDAC activity, respectively. Thus, 10 nM FK228 corresponds to twice the IC₅₀ of the high affinity HDAC inhibitory site and is far below the threshold concentration for the secondary low affinity site previously associated with significant reductions in Cx43 expression and function (71). Since TSA and VOR reduced cardiac g_j in a dose-dependent manner, we assessed the dose-dependent effects of FK228 on initial g_j values of NMVM cell pairs (Fig. 6B). Consistent with our previous findings, FK228 concentrations >1 nM significantly decreased g_j.

Fig. 6.

Romidepsin inhibition of cardiac HDAC activity and gap junction conductance. A: total HDAC activity was measured in NMVM cultures exposed to increasing concentrations of FK228 using the deacetylated Fluor-de-Lys fluorescence method. All data were normalized to the background subtracted maximum relative fluorescence of the control well. The data from three experiments were averaged and fitted with a second-order exponential decaying function and equilibrium inhibition constants (K_i) were calculated from the decay constant for each exponential component. B: the gap junction conductance (g_j) was measured in NMVM cell pairs from control and FK228-treated culture dishes and a dose-dependent decrease in g_j was observed, consistent with published results with TSA and VOR (71).

DISCUSSION

This study consistently found that pan-HDAC inhibition reduced peak I_Na density and Na_V1.5 protein levels (Figs. 1, 4, and 5). The only other functional alteration of cardiac I_Na observed with pan-HDAC inhibition was a slight (3–8 mV) shift in the V_m-dependent activation curve and a subsequent slowing of the inactivation time constants at low activating potentials (< −40 mV). Scn5a mRNA levels were not decreased by 1 μM VOR yet there was a 40% reduction in Na_V1.5 protein levels (Fig. 5, B–D). The 400 mg daily maximum recommended human therapeutic dose of VOR correlates to a maximum plasma concentration (C_max) value equal to 1.2 μM (320 ng/ml) (29). Reduced Na_V1.5 protein levels and I_Na density without a concomitant decrease in Scn5a mRNA levels have also been reported in mdx mice (16). TTX-sensitive sodium channel isoforms Na_V1.3, Na_V1.4, Na_V1.2, Na_V1.1, and Na_V1.6 (in order of abundance) have been detected in neonatal and adult mouse and rat ventricular cardiomyocytes (21, 27, 38, 39). These neuronal and skeletal muscle Na_V isoforms, along with the β₁- and β₃-auxiliary subunits, are primarily located in the transverse tubules whereas the predominantly more abundant cardiac Na_V1.5 along with the β₁-, β₂-, and β₄-subunits preferentially localize to the intercalated disc. The noncardiac Na_V isoforms may account for up to 30% of cardiac I_Na in neonatal rat ventricular myocytes (NRVMs) but account for <10% of total cardiac I_Na in isolated adult mouse ventricular myocytes (21, 27). We did not assay for noncardiac Na_V or β-subunit isoforms in the present study and will consider these transcripts and proteins in future experiments, but the close correlation between the decrease in Na_V1.5 protein levels and peak I_Na density in our experiments (Figs. 1 and 5) is consistent with the interpretation that the decrease in peak cardiac I_Na observed with pan-HDAC inhibition is due to dose-dependent reductions in Na_V1.5 protein levels.

The precise molecular basis for the decrease in Na_V1.5 protein expression remains to be determined by future investigations, though downregulation of Scn5a expression cannot be ruled out at higher inhibitory concentrations of TSA and VOR. Possible mechanisms for the downregulation of cardiac I_Na include variations in the Scn5a transcripts that do not equally translate into functional Na_V1.5 protein, microRNA regulation of Scn5a transcripts, increased Na_V1.5 degradation by Nedd4-2-mediated ubiquitination, and reduced expression of the MOG1 protein that directly interacts with and increases the functional surface expression of Na_V1.5 channels (10, 52, 63, 69). The surface expression and localization of Na_V1.5 is also modulated by direct interactions with α-actinin-2, ankyrin-G, glycerol-3-phosphate dehydrogenase 1 like gene (GPD1-L), SAP-97, and tyrosine-phosphorylated β1 (pYb1) and indirect protein-protein interactions with Cx43, N-cadherin (Ncad), the syntrophin-dystrophin complex including calcium/calmodulin-dependent serine kinase (CASK), and plakophilin-2 (PKP2), all of which can result in functional reductions in Na_V1.5 expression (12, 25, 35, 41, 46, 48, 51, 52, 54, 68, 74). Thus our findings of reduced Ncad and Cx43 protein levels observed with TSA, VOR, and FK228 are consistent with decreases in I_Na (71).

Besides the many possible direct and indirect protein interactions with Na_V1.5, numerous posttranslational modifications alter Na_V1.5 function including glycosylation, phosphorylation by numerous kinases (e.g., PKA, PKC, CaMKII, and Fyn), ubiquitination, S-nitrosylation, and methylation (1, 3, 7, 52, 63). N^ε-lysine acetylation directly or indirectly impacts protein methylation, neddylation, ubiquitylation, and phosphorylation either by mutually exclusive posttranslational modification (PTM) of lysine residues or lysine residues within concensus protein kinase phosphorylation sites (72). NH₂-terminal acetylation of Na_V1.5 was recently discovered in end stage human heart failure and is associated with specific protein degradatory pathways that likely results in the downregulation of cardiac I_Na (2, 4, 23, 42). NH₂-terminal (Nt- or N^α) acetylation is irreversibly catalyzed by ribosome-associated NH₂-terminal acetyltransferases (NATs) and is distinct from the N^ε-lysine acetylation reaction catalyzed by the histone acetyltransferases (HATs) and reversed by the HDACs and sirtuins (SIRTs, class III NAD⁺-dependent HDACs) (2, 73). Classical HDAC inhibition will not affect NAT or SIRT activities, although HDAC6 and SIRT2 colocalize to form the tubulin deactylase complex (47). Whether or not Na_V1.5 is a substrate for N^ε-lysine acetylation is currently unknown, so we subjected mouse and human Na_V1.5 sequences to N^ε-lysine acetylation site analysis using the PHOSIDA website (18). This bioinformatic analysis predicted two high probability (>90%) cytoplasmic acetylation sites near the domain II and IV S4 segments known to be involved in V_m-dependent activation (Fig. 7A). To determine if Na_V1.5 is acetylated, we injected six adult C57BL/6 mice with 10 mg/kg TSA or DMSO/saline control intraperitoneal injections for 5 days and immunoprecipitated the ventricular heart lysates with anti-Na_V1.5 antibodies. Subsequent immunoblotting for acetylated protein with an anti-acetyl-lysine (Ac-K) antibody revealed an increase in acetylated Na_V1.5 protein with TSA treatments (Fig. 7, B and C). The human and murine Na_V1.5 proteins possess 94% sequence identity; hence we hypothesize that acetylation of K830 and K1644 (K1641 in the human sequence) may account for the small depolarizing shift in the I_Na activation curve and the slowing of the inactivation rate at low activation potentials seen with pan-HDAC inhibition (Figs. 2 and 3). This postulated acetylation-dependent shift of Na_V1.5 activation decreased the V_m-dependent window for prolonged activation of I_Na (window current) and is unlikely to account for the observed decrease in results in I_Na density, which correlates closely with the decreased levels of Na_V1.5 protein. Identification of the actual Na_V1.5 acetylation sites will require future liquid chromatography-electron spray ionization tandem mass spectroscopic analyses of IP-enriched and/or SDS-PAGE proteolytic Na_V1.5 peptide digests (32).

Fig. 7.

A: putative N^ε-lysine acetylation sites on the Na_V1.5 Na⁺ channel subunit. The full length mouse Na_V1.5 amino acid sequence was analyzed for possible lysine acetylation (Ac-K) sites using the prediction algorithm available at the PHOSIDA Posttranslational Modification database (http://www.phosida.org; Ref. 18). Eight putative acetylation sites (probability >90%) were identified and mapped onto the membrane topological model for the Na_V1.5 channel protein. Six of the eight predicted Na_V1.5 Ac-K sites (K158, K175, K767, K863, K1362, and K1617) mapped to transmembrane or extracellular domain locations and were discounted as possible acetylation sites (lined circle). Two possible Ac-K sites, K830 and K1644, mapped to cytoplasmic domains located near the cytoplasmic COOH-terminal side of the repeat domain II and IV S4 domains (star circle). We hypothesize that acetylation of these 2 sites, which are conserved in the human Na_V1.5 sequence, may account for the slight positive voltage shift (< 10 mV) in the cardiac I_Na activation curve associated with pan-HDAC inhibitor treatments (Fig. 2, A-C). B, left: an example of an immunoprecipitation (IP) performed on saline:DMSO (control, C) or TSA-injected (T) mouse heart ventricular lysates with anti-Na_V1.5-protein G magnetic beads and immunoblotted (IB) with an anti-acetyl-lysine (Ac-K) antibody to detect protein acetylation. TSA injection for 5 days caused a dramatic increase in the acetylated Na_V1.5 band. Rabbit IgG controls for control (IgGc) and TSA heart samples (IgGt) are shown in lanes 3 and 4. B, right: lanes 1 and 2 are the Na_V1.5 immunoprecipitate input controls for the control and TSA heart lysates and the lanes 3 and 4 are the rabbit IgG controls for each sample. C: Normalized mean Ac-K Na_V1.5 band densities for the upper [high molecular weight (MW), upper arrow] and lower (lower MW, lower arrow) from 3 control:TSA immunoprecipitation experiments as illustrated in B. Each Ac-K TSA band density was divided the control band density for each experiment and this TSA/control band ratio was normalized by dividing by the Na_V1.5 input TSA/control band ratio.

There are nearly 500 Na_V1.5 mutations associated with cardiac arrhythmias including long QT (LQT3) syndrome, Brugada syndrome (BrS), cardiac conduction disease (CCD), sick sinus syndrome (SSS), and atrial fibrillation (AF) (13, 40, 51, 52, 65, 75). Na_Vβ1–4 subunits also contribute to LQT and AF cardiac arrhythmias (43, 66, 67). Approximately 75% of these mutations cause loss of Na_V1.5 function except those involved in LQT and some AF arrhythmias. Gain-of-function Na_V1.5 mutations typically increase activation or decrease inactivation, resulting in a net increase of a persistent (sustained) or late I_Na that prolongs action potential duration and enhances the occurrence of early afterdepolarizations (EADs) (43, 65). For example, expression of the Na_V1.5F1486del LQT mutation increased late I_Na and produced EADs in NRVMs (59). Recently, increased expression of Na_V1.1 and Na_V1.3 isoforms in Scnb1^−/− mice and ischemic heart failure dog hearts produced increases in late I_Na and arrhythmias, thus indicating a role for neuronal isoforms in cardiac arrhythmogenesis (33, 44). However, we found no evidence for pan-HDAC inhibitor-induced increase in late I_Na in NMVMs or hiPSC-CMs. By contrast, loss-of-function mutations shorten the cardiac refractory period and slow conduction velocity leading to arrhythmias including BrS, SSS, and CCD (6, 56, 61, 67). Peak I_Na density and gap junction conductance (g_j) are primary determinants of myocardial conduction velocity and reductions in both of these factors will slow conduction and increase the vulnerability to reentrant arrhythmias (30, 62). BrS arrhythmias frequently develop in the right ventricle, which is more susceptible to conduction slowing secondary to Na_V1.5 dysfunction (49, 51).

All three pan-HDAC inhibitors we have tested thus far caused significant reductions in g_j and I_Na, consistent with an increased susceptibility to cardiac arrhythmias due to slowed or discontinuous propagation (71). Reduced gap junction coupling also enhances the ability of triggered afterdepolarizations to initiate an arrhythmia including LQT-induced Torsades de Pointes (TdP) arrhythmias by modulating the source-sink interactions of an ectopic focus with the surrounding normal myocardium (53, 70). Enhancement of late I_Na currents and blockade of repolarizing delayed outward K⁺ currents account for >90% of all congenital and drug-induced LQT syndromes (26). However, the IC₅₀ for VOR blockade of the rapid delayed outward rectifying K⁺ current (I_Kr) formed by the hERG and MiRP1 channel and accessory proteins is 300 μM, orders of magnitude higher than the recommended therapeutic dose of 1 μM (29). Thus direct hERG blockade is unlikely to account for the observed QT interval prolongation and TdP arrhythmias occasionally observed with this HDAC inhibitor (36, 50). Two recent studies have concluded that the cardiotoxic effects of HDAC inhibitors likely results from transcriptional changes in genes affecting ion channel function owing to their delayed onset of cardiotoxicity and lack of acute ion channel blockade (31, 60). Owing to the present and recent findings, identification of the ionic mechanisms responsible for HDAC inhibitor-induced QT interval prolongation and TdP arrhythmias merits further investigation. In conclusion, pan-HDAC-induced reductions in peak cardiac I_Na and Na_V1.5 protein levels and shifted the activation curve positive, possibly due to direct acetylation of two S4 segment adjacent cytoplasmic lysine residues. These effects on I_Na should decrease cardiac excitability, refractoriness, and conduction velocity in a manner that will increase myocardial vulnerability to reentrant arrhythmias but not LQT-associated TdP arrhythmias that occasionally occur with pan-HDAC inhibitory treatments.

GRANTS

The work was supported by National Heart, Lung, and Blood Institute Grant HL-042220, the Joseph C. Georg Fund of the Central New York Community Foundation/Upstate Medical University Foundation, and the Hendricks Fund of the Health Science Center Foundation at Syracuse (to R. D. Veenstra).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Q.X. and R.D.V. conception and design of research; Q.X., D.P., and X.Z. performed experiments; Q.X., D.P., X.Z., and R.D.V. analyzed data; Q.X., D.P., X.Z., and R.D.V. interpreted results of experiments; Q.X., D.P., X.Z., and R.D.V. prepared figures; Q.X., D.P., X.Z., and R.D.V. edited and revised manuscript; Q.X., D.P., X.Z., and R.D.V. approved final version of manuscript; R.D.V. drafted manuscript.