Deletion in mice of X-linked, Brugada syndrome– and atrial fibrillation–associated Kcne5 augments ventricular K_V currents and predisposes to ventricular arrhythmia

Abstract

KCNE5 is an X-linked gene encoding KCNE5, an ancillary subunit to voltage-gated potassium (K_V) channels. Human KCNE5 mutations are associated with atrial fibrillation (AF)– and Brugada syndrome (BrS)–induced cardiac arrhythmias that can arise from increased potassium current in cardiomyocytes. Seeking to establish underlying molecular mechanisms, we created and studied Kcne5 knockout (Kcne5^−/0) mice. Intracardiac ECG revealed that Kcne5 deletion caused ventricular premature beats, increased susceptibility to induction of polymorphic ventricular tachycardia (60 vs. 24% in Kcne5^+/0 mice), and 10% shorter ventricular refractory period. Kcne5 deletion increased mean ventricular myocyte K_V current density in the apex and also in the subpopulation of septal myocytes that lack fast transient outward current (I_to,f). The current increases arose from an apex-specific increase in slow transient outward current-1 (I_Kslow,1) (conducted by K_V1.5) and I_to,f (conducted by K_V4) and an increase in I_Kslow,2 (conducted by K_V2.1) in both apex and septum. Kcne5 protein localized to the intercalated discs in ventricular myocytes, where K_V2.1 was also detected in both Kcne5^−/0 and Kcne5^+/0 mice. In HL-1 cardiac cells and human embryonic kidney cells, KCNE5 and K_V2.1 colocalized at the cell surface, but predominantly in intracellular vesicles, suggesting that Kcne5 deletion increases I_K,slow2 by reducing K_V2.1 intracellular sequestration. The human AF-associated mutation KCNE5-L65F negative shifted the voltage dependence of K_V2.1-KCNE5 channels, increasing their maximum current density >2-fold, whereas BrS-associated KCNE5 mutations produced more subtle negative shifts in K_V2.1 voltage dependence. The findings represent the first reported native role for Kcne5 and the first demonstrated Kcne regulation of K_V2.1 in mouse heart. Increased K_V current is a manifestation of KCNE5 disruption that is most likely common to both mouse and human hearts, providing a plausible mechanistic basis for human KCNE5-linked AF and BrS.—David, J.-P., Lisewski, U., Crump, S. M., Jepps, T. A., Bocksteins, E., Wilck, N., Lossie, J., Roepke, T. K., Schmitt, N., Abbott, G. W. Deletion in mice of X-linked, Brugada syndrome– and atrial fibrillation–associated Kcne5 augments ventricular K_V currents and predisposes to ventricular arrhythmia.

The potassium voltage-gated channel, subfamily E (KCNE) family of ion channel β subunits consists of 5 members, KCNE1–5, also called minK and minK-related peptides 1–4 respectively. They are single-pass membrane proteins with an extracellular N terminus and an intracellular C terminus and act as ancillary subunits for various voltage-gated potassium (K_V) and other channels (for reviews, see refs, 1–4). Expression profiles vary, yet all KCNE genes are reportedly expressed in the human heart (5–7). Likewise, all KCNE family members have been linked to cardiac arrhythmias, including atrial fibrillation (AF), Brugada syndrome (BrS), and long QT syndrome (8). Best characterized is the role of KCNE1 in the heart, where it is considered the primary β-subunit for the K_V7.1 (KCNQ1) channel in the human heart. Together, K_V7.1 and KCNE1 are thought to be the predominant molecular correlate of the slow delayed rectifier potassium current (I_Ks) which is a major contributor to the repolarization of human ventricular myocyte action potentials (9–11).

Whereas KCNE1-4 are relatively well studied, knowledge of the X-linked KCNE5 (also termed KCNE1L) gene is limited. KCNE5 was originally identified as one of several deleted genes in patients with the rare Alport syndrome, mental retardation, midface hypoplasia, and elliptocytosis contiguous gene syndrome characterized by a large deletion of the X chromosome (Xq22.3) (12). Notable features of the disease are Alport syndrome, mental retardation, midface hypoplasia, and elliptocytosis (13, 14). In one study, affected men show a more complex phenotype including mild abnormalities of the heart: 1 patient displayed a slow right atrial rhythm and short PR interval and also had some minor cardiac morphologic abnormalities (12). Several KCNE5 variants are associated with AF or BrS, and different ion channels have been shown to be misregulated by the various KCNE5 variants, including the K_V7.1 and K_V4 channels (15–17). In addition, we have demonstrated that the KCNE5 protein interacts with K_V2.1 in a heterologous expression system, suppressing K_V2.1 current density without changing kinetic parameters (18).

Despite the association of genetic variation in KCNE5 with disease, knowledge regarding the underlying molecular mechanisms is sparse. Therefore, for the present study, we generated Kcne5 knockout (Kcne5^−/0) mice to determine the implications for cardiac function. Moreover, we examined how human KCNE5 mutations that are linked to AF and BrS affect the K_V2.1 current.

MATERIALS AND METHODS

Kcne5 knockout mice

Kcne5 knockout mice were generated by the commercial provider Taconic (Oxnard, CA, USA) from a cryopreserved Kcne5 knockout line (Taconic TF0342, background: 129/SvEv-C57BL/6) in their knockout repository. In these mice, the Kcne5 gene is completely deleted. Positive clones were screened by Southern blot for correct targeting events. Genotyping was performed by PCR with primers 18 (5′-TGTATGCTTCATTCAGGGCC-3′) and 27 (5′-GGGTTCAAATGATCTTCCTGCC-3′) to yield products of 313 bp for wild-type (WT) or 376 bp for mutant animals. Mice were bred and genotyped at Taconic and delivered to our university facilities at least 2 wk before the experiments. The mice were provided food and water ad libitum and housed in a room kept at 22°C with a 12-h light-dark schedule. Knockout (hemizygous, Kcne5^−/0)) male mice were bred from WT (Kcne5^+/0) or hemizygous males (Kcne5^−/0) with heterozygous (Kcne5^−/+) female mice. Kcne5^−/0 mice are viable and fertile. We also generated a second Kcne5^−/0 mouse line, on a C57BL/6NCrl background, in collaboration with Dr. Boris Jerchow [Transgenic Core Facility, Max Delbrück Center for Molecular Medicine (MDC), Berlin, Germany], using a zinc finger approach (Supplemental Fig. S1). To verify the successful homology-directed repair in founder animals, we used PCR and enzyme digestion, as well as sequencing. Specific primers (Supplemental Fig. S1A, yellow and blue) were used for PCR analysis of genomic DNA. Positive founder animals, the DNA of which contained the new restriction site, displayed 2 fragments instead of 1 (Supplemental Fig. S1B). In addition, the positive founder animals were examined by sequencing (Supplemental Fig. S1C, primers yellow and gray). Sequence analysis showed correct homology-directed repair with the 3 stop codons (Supplemental Fig. S1C, underlined red) and the new TfiI restriction site (underlined green) in the investigated animal. The breeding strategy for the generation and maintenance of Kcne5^−/0 mice was as follows: y^/+ (male) x^−/+ (female) to receive male y^/− (Kcne5^−/0 mice) and y^/+ (Kcne5 WT littermates). Genotyping was performed by PCR (Supplemental Fig. S1D). Neither Kcne5^−/0 mouse line exhibited cardiac hypertrophy, other overt structural defects, or K_Vα subunit transcript remodeling; each exhibited similar ventricular myocyte K_V current upregulation. We conclude that the functional effects we observed in this study are specific to Kcne5 deletion, as they arose in 2 different strains generated by different methodologies.

All animal work conformed to national guidelines for the protection of vertebrate animals used for experimental and other scientific purposes. All mice were housed in pathogen-free facilities, and experiments were approved by the Institutional Animal Care and Use Committees at University of California, Irvine, and at MDC-Berlin. Studies were performed during the light cycle, in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (NIH), Bethesda, MD, USA. For all experiments requiring postmortem harvest of Kcne5^−/0 mouse tissue, each mouse was euthanized by CO₂ inhalation before cervical dislocation.

Heart size analysis

To compare the size of the hearts of adult (8-mo-old) male Kcne5^−/0 mice with age-matched control (WT), the mice were euthanized with CO₂ before cervical dislocation. After opening of the chest cavities, their hearts were transected below the major vessels and rinsed free of blood in normal saline. The saline was removed from the hearts with a napkin before being weighed. Furthermore, we isolated the right tibia, which was rinsed to remove skin and muscle tissue and measured from the crest to the medial malleolus. The size of the hearts was normalized to the length of the tibia and reported as the ratio of heart size (mg)/tibia length (mm). The unpaired Student’s t test with Welch’s correction was performed with Prism (GraphPad Software, La Jolla, CA, USA) for statistical analysis. Values of P < 0.05 were considered statistically significant. Data are expressed as means ± sem.

Body surface electrocardiograph recordings

Mice were anesthetized with 2% isoflurane in oxygen and were spontaneously breathing. They were placed on a heating pad to avoid a dramatic drop in body temperature, while the subcutaneous needle electrodes on shield cables were placed on the left side near the middle part of the rib cage (plus electrode), in the right foreleg pit (minus electrode), and in the right hind leg (reference electrode), to record a single-lead surface ECG. Before signals were recorded, the isoflurane in the oxygen was lowered to 1.75%, and the heating pad was disconnected because of an increase in electrical noise. Signals were recorded for 5 min with Powerlab 8/35 (ADInstruments, Dunedin, New Zealand) at 2 kHz and filtered with a high-pass setting of 0.3 Hz and a low-pass setting of 1 kHz. Beats within min 1 and 4 of the recording were used to generate a mean-signal ECG and were analyzed manually with LabChart 7 (ADinstruments). The isoelectric level was calculated by the program from the pre-P wave segment or from the interval between the end of the P-wave and the beginning of the Q-wave, if possible. Because of electrical noise, the end of the T-wave was difficult to determine on the mean signal. Hence, the QT interval could not be determined accurately. Therefore, for each ECG recording, the mean QT interval was manually determined from 5 single heart beats showing a minimum of electrical noise. Unpaired Student’s t test with Welch’s correction was performed with Prism 5.01 for statistical analysis and values of P < 0.05 were considered statistically significant. Pooled data are expressed as means ± sem.

Invasive electrophysiology studies

For invasive electrophysiology studies, we used a digital electrophysiology lab (EP Tracer; CardioTek, Maastricht, The Netherlands) and octapolar 2-French electrode catheters (CIBermouse Cath; Numed, Cross Roads, TX, USA), to determine standard parameters of electrical conduction and refractoriness and to test for inducibility of arrhythmias. Mice were lightly anesthetized with isoflurane (1.6 vol. % isoflurane/air) at a constant body temperature. The catheter was placed via the right jugular vein into the right atrium and ventricle. We performed programmed electrical stimulation with a standardized protocol with documentation of occurrence and duration of induced arrhythmias (19).

RNA extraction and reverse transcription

For expression analysis, mouse atria and the walls of each ventricle were isolated without excising the whole heart, frozen in liquid nitrogen, and kept at −80°C until further use. RNA was extracted from the atria and ventricles of male Kcne5^−/0 and control (Kcne5^+/0) mice by using the RNeasy RNA Purification Kit (Qiagen, Germantown, MD, USA), according to the manufacturer’s protocol. The RNA was reverse-transcribed with the QuantiTect Reverse Transcription Kit (Qiagen) according to the manufacturer’s instructions.

Human cardiac tissue samples from donor hearts that were unsuitable for transplantation were kindly provided by Dr. András Varró (University of Szeged, Szeged, Hungary). The procedure was approved by the national ethics committee, ethics approval number 4991-0/2010-1018EKU (339/PI/010), and informed consent was given to use the tissue for research purposes. The investigation conformed to the principles outlined in the Declaration of Helsinki. All donors received 25,000 Na-heparin, flushing solution (Custodiol; Dr. Franz Köhler Chemie, Bensheim, Germany), methylprednisolone sodium succinate (Solu-Medrol; Pfizer, New York, NY, USA), norephnephrine (Arterenol; MilliporeSigma, Burlington, MA, USA), and hydroxyethyl starch (Voluven; Pfizer) before heart explantation. The hearts were placed in ice-cold cardioplegic solution, cut into pieces, frozen, and stored in liquid nitrogen until further use. RNA extraction and cDNA generation from human heart tissues were performed as described by Soltysinska et al. (20).

Quantitative PCR

Quantitative analyses of the KCNE5/Kcne5 gene in the cDNA samples from human and mouse hearts were performed in duplicate reactions of 20 μl volumes containing 10 μl PrimerDesign 2X Precision Mastermix (Primerdesign, Southampton, United Kingdom) with the 7300 qPCR System (Thermo Fisher Scientific, Waltham, MA, USA) and the CFX96 Real-Time PCR System (Bio-Rad, Hercules, CA, USA), respectively. Data were obtained with SDS1.2 software and CFX96 software, respectively. Double-dye assays were designed and synthesized by Primerdesign (primer sequences for mouse, sense: 5′-GCACGAAGAGACCTCAGACAT-3′, antisense: 5′-GGACAGGAAAACAAGAACACCAT-3′; and for human, sense: 5′-TACTTCCAAATGCCTCTCCTCTAT-3′, antisense: 5′-AACAACCTTTATTACCTGCCTCTG-3′). The amplification protocol used for each reaction was as follows: a 10 min initiation step at 95°C before 40 cycles of 15 s at 95°C and 1 min at 60°C. For optimal normalization analysis of the cDNA samples, we used the geNorm Reference Gene Selection Kit (Primerdesign) (21). β-Actin (ACTB) and topoisomerase-1 (TOP1) were chosen as optimal reference genes for the human cDNA samples, and for mouse cDNA samples, glyceraldehyde phosphate dehydrogenase, and ActB were used (Primerdesign). The 2^−ΔCt method (22) was used to determine the relative expression of KCNE5/Kcne5 in both human and mouse cDNA samples. A 2-way ANOVA with Bonferroni posthoc and an unpaired Student’s t test with Welch’s correction were performed with using Prism 5.01 (GraphPad) for statistical analysis. Values of P < 0.05 were considered statistically significant. Data are expressed as means ± sem.

Plasmids

Human K_V2.1 (GenBank Accession number NM_004975) in the eGFP-N1 vector (Takara Bio USA, Mountain View, CA, USA) and human KCNE5 (hKCNE5, NM_012282) in pXOOM are described in refs. 18 and 20. Mutant constructs P33S (c.97C > T), L65F (c.193C > T), R85H (c.254G > A), Y81H (c.241T > C) or D92E/E93X (c.276-277CG > AT) were generated by mutated oligo extension PCR with hKCNE5 in pXOOM as template using PfuUltraII HS Fusion Polymerase (Agilent Technologies, Santa Clara, CA, USA) and subsequent restriction with DpnI (Thermo Fisher Scientific). Reaction samples were transformed into Escherichia coli XL1 Blue cells applying a standard heat-shock procedure. After plasmid DNA preparation, constructs were verified by complete DNA sequencing of the cDNA insert (Macrogen, Seoul, South Korea).

Ventricular tissue immunofluorescence

Cryosections (8 mm) from snap-frozen ventricular tissue were fixed with ice-cold acetone for 10 min at 4°C, blocked, and permeabilized in 10% goat serum, 0.3% Triton X-100 and 0.2% bovine serum albumin for 30 min. The following primary antibodies were used and incubated overnight at 4°C: rabbit anti-Kcne5 (1:20; MilliporeSigma, Burlington, MA, USA), mouse anti-β-catenin (1:20; MilliporeSigma), rabbit anti-K_V2.1 (1:20; Thermo Fisher Scientific), and mouse anti-Cx43 (1:20; Thermo Fisher Scientific). As secondary antibodies we used AlexaFluor 555 goat anti-rabbit and AlexaFluor 647 goat anti-mouse (1:1000; Thermo Fisher Scientific), incubated for 60 min at room temperature. Kcne5 staining was amplified with biotin-conjugated anti-rabbit (1:500; Jackson ImmunoResearch Laboratories, West Grove, PA, USA) for 30 min followed by incubation with streptavidin-conjugated AlexaFluor 555 (1:1000; Thermo Fisher Scientific) for 60 min at room temperature. Stained cryosections were mounted with fluorescence mounting medium (Agilent Technologies) and imaged on a confocal laser scanning microscope (TSC SP5 with software, ver. LAS AF 2.632; Leica Microsystems, Wetzlar, Germany). Images were assembled in Photoshop CS4 (Adobe, San Diego, CA, USA) and CorelDraw 34 (Corel, Ottawa, ON, Canada).

Electrophysiological analysis of adult mouse cardiomyocytes

Ventricular cardiomyocytes were isolated from 7-mo-old male mice, as described in detail by Hu et al. (23). Only rod-shaped, quiescent cardiomyocytes with clear striations were used for recording. Whole-cell patch-clamp recordings from dispersed ventricular myocytes (derived from 2 to 8 mice of each genotype) were obtained at room temperature with an IX50 inverted microscope (Olympus, Center Valley, PA, USA) equipped with an FHD chamber from IonOptix (Westwood MA, USA) and a Multiclamp 700A Amplifier, a Digidata 1300 Analog/Digital converter, and a PC with pClamp9.2 software (all from Molecular Devices, Sunnyvale, CA, USA). For whole-cell patch-clamp recordings, the bath solution contained (in mM) 117 NaCl, 4 KCl, 1.7 MgCl₂, 10 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 1 KH₂PO₄, 4 NaHCO₃, 3 CoCl₂, and 10 d-glucose (pH 7.4 with NaOH). Bath solutions were treated with either no drug, 50 μM 4-aminopyridine (4-AP; ICN Biomedicals, Irvine, CA, USA), 25 mM tetraethylammonium (TEA; MilliporeSigma), or 500 nM heteropodatoxin (HpTx2) for K_V current inhibition. Pipettes were of 2.0–3.1 MΩ resistance when filled with intracellular solution containing (in millimolars) 130 KCl, 2 MgCl₂, 20 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid, 11 EGTA, 5 Na₂ATP, 0.4 Na₂GTP, and 5 Na₂CP (pH 7.4 with KOH). Outward K⁺ currents were evoked during 4-s voltage steps to test potentials between −60 and +60 mV in 20 mV increments from a holding potential of −70 mV after a 100-ms prepulse to −40 mV. Uncorrected leak currents were always <100 pA. Data were analyzed with pClamp9.2 software (Molecular Devices), and statistical analysis (ANOVA) was performed with Origin 6.1 (Microcal, Northampton, MA, USA).

Human embryonic kidney cell culture, transient transfection, and patch-clamp electrophysiology

Human embryonic kidney (HEK)293 cells were cultured as previously described (18). HEK293 cell cultures (70–80% confluence) were cotransfected with 250 ng K_V2.1 and 1 µg WT or mutant hKCNE5 with Lipofectamine 2000 reagent (Thermo Fisher Scientific). Green fluorescent protein cDNA (0.5 µg) was cotransfected as a transfection marker. Transfected cells were dissociated with trypsin and used for electrophysiological analysis 16–24 h after transfection. Patch-clamp recordings from HEK293 cells were performed as previously described (18). Series resistance was compensated (80%) for, and cells with a voltage error exceeding 5 mV after compensation were excluded from analysis. The Boltzmann equation, y = 1/{1+exp[−(V−V_1/2)/k]}, where V is the voltage applied, V_1/2 is the voltage at which 50% of the channels are (in)activated, and k is the slope factor, was used to fit the voltage-dependence of activation and inactivation. Single and double exponential functions were used to fit the activation and deactivation kinetics, respectively. Student’s t test or Mann-Whitney rank sum test was used for statistical analysis, and values of P < 0.05 were considered statistically significant.

HL-1 cell culture, transient transfection, and imaging

HL-1 cells were cultured and transfected in gelatin/fibronectin-coated T25 flasks (Nunc, Roskilde, Denmark) with Claycomb medium (MilliporeSigma), supplemented with 10% fetal bovine serum (Batch 11A568; MilliporeSigma), 100 U/ml penicillin (Thermo Fisher Scientific), 100 mg/ml streptomycin (Thermo Fisher Scientific), 0.1 mM norepinephrine (MilliporeSigma), and 2 mM l-glutamine (MilliporeSigma) at 37°C in 5% CO₂. At 90–100% confluence, the cells were trypsinized and resuspended in 5 ml medium, and 1.25 ml cell solution was transfected with 2 μg DNA in total, with 3.5 μl Silentfect Lipid Reagent (Bio-Rad) according to the manufacturer’s protocol. Transfected cells were plated into a 35-mm tissue culture dish (Greiner Bio-One, Frickenhausen, Germany) containing 4 glass cover slips (diameter, 12 mm; Thermo Fisher Scientific) and coated with gelatin/fibronectin. The cells were incubated at 37°C in 5% CO₂ for 16–24 h before they were fixed by adding 4% paraformaldehyde in PBS for 10 min at room temperature.

Immunofluorescence staining and imaging with the LSM710 or LSM780 laser scanning confocal microscope (Zeiss, Jena, Germany) were performed with published methods (18, 24). Primary antibodies were: rabbit polyclonal anti-KCNE5 (1:100, in-house) (18), mouse monoclonal anti-K_V2.1 (1:100, clone K39/25; NeuroMab, University of California, Davis/NIH NeuroMab Facility, Davis, CA, USA), and rat monoclonal anti-hemagglutinin (HA) (1:50, clone 3F10; Roche Diagnostics, Indianapolis, IN, USA). Alexa-Fluor 647–conjugated and rhodamine-conjugated phalloidin (1.5 U/ml; Thermo Fisher Scientific) were used as plasma membrane markers in HL-1 cells. Rat monoclonal anti-lysosomal-associated membrane protein (LAMP)-2 (1:100, clone GL2A7; Abcam, Cambridge, United Kingdom) was used as a lysosomal marker. Depending on the antibodies used, the plasma membrane was visualized with either rhodamine-conjugated phalloidin or Alexa Fluor 647 phalloidin to identify the submembranous F-actin in HL-1 cell secondary antibodies: AlexaFluor 488 donkey anti-rabbit IgG [heavy (H)+light (L) chain] (1:200), AlexaFluor 488 donkey anti-mouse IgG (H+L) (1:200), AlexaFluor 488 donkey anti-rat (H+L) (1:200), AlexaFluor 568 donkey anti-rabbit IgG (H+L) (1:500), and AlexaFluor 555 donkey anti-mouse IgG (H+L) (1:500). All secondary antibodies were purchased from Thermo Fisher Scientific.

RESULTS

The KCNE5 gene is expressed in both mouse and human hearts

In WT (Kcne5^+/0) male mice, we detected Kcne5 transcript in both the atria and ventricles (Fig. 1A). Because the level of KCNE5 expression in the human heart is still a matter of debate (5–7), we also analyzed the relative expression in cardiac tissue samples from 6 male donor hearts, and found that the KCNE5 gene was expressed in all human heart regions analyzed (Fig. 1B).

Figure 1 .

Kcne5 is expressed in mouse and human hearts. A) Quantification of Kcne5 transcript expression in atria and ventricles (n = 5 and n = 6, respectively) for male Kcne5^+/0 (WT) and Kcne5^−/0 (knockout) mice by real-time qPCR, normalized to the reference genes ActB and glyceraldehyde phosphate dehydrogenase. Data are means ± sem. B) Relative KCNE5 regional expression in male human hearts (n = 6): left (LA) and right (RA) atria and the left ventricular epi-, mid- and endomyocardial wall. No significant differences in expression levels between the regions were found. Quantification was by real-time qPCR with ACTB and TOP1 as reference genes. C) Body weight, heart size, and tibia length of ∼8-mo-old male mice (n = 20 in each group; means ± sem shown). Kcne5^−/0 mice showed a significant increase in body weight (44.8 ± 1.0 g) compared to control mice (40.1 ± 1.1 g). P = 0.0030. Heart size (0.147 ± 0.003 vs. 0.143 ± 0.003 g, respectively) and the length of tibia (11.8 ± 0.1 vs. 11.9 ± 0.1 mm, respectively) did not show any differences. Likewise, there was no difference in heart size when normalized to tibia length [(mg/mm) 12.5 ± 0.2 vs. 12.0 ± 0.3, respectively]. **P < 0.01.

Kcne5 deletion predisposes to ventricular arrhythmia

Because we found that Kcne2^−/− pups from homozygous crosses show cardiac hypertrophy as early as 3 wk (25), we investigated whether deletion of Kcne5 had a similar effect (Fig. 1C). Kcne5 deletion did not alter heart weight, normalized to either body weight or tibia length, in either of the strains we assessed (Fig. 1C and Supplemental Table S1). In addition, echocardiographic analysis uncovered no evidence of cardiac hypertrophy (Supplemental Table S2).

Noninvasive body surface ECG analysis in anesthetized male mice (Supplemental Fig. S2), aged ∼5 mo (n = 9–10 in each group) (Supplemental Fig. S2A) and 8 mo (n = 16–20 in each group) (Supplemental Fig. S2B), did not reveal any genotype-dependent changes. In contrast, invasive ECG uncovered ventricular premature beats (VPBs), which were followed by postextrasystolic compensatory pauses, in male Kcne5^–/0 mice (Fig. 2A). In 1 Kcne5^–/0 mouse, we observed a spontaneously occurring episode of polymorphic ventricular tachycardia VT. The episode could not be terminated by burst pacing, but eventually converted into an atrial arrhythmia with a 2:1 and 3:1 atrioventricular conduction, (Fig. 2B). The Kcne5^–/0 mice also had increased susceptibility to polymorphic VT induction with a defined pacing protocol followed by 3 extra stimuli (Fig. 2C, D). Furthermore, Kcne5 deletion shortened the ventricular effective refractory period (Fig. 2E).

Figure 2 .

Kcne5 deletion causes ventricular arrhythmias. A) Spontaneously occurring monomorphic VPBs were detectable during continuous ECG monitoring in sedated live Kcne5^−/0 mice. No such VPBs were found in Kcne5^+/0 mice. Upper ECG curves represent body surface ECGs, middle ECG curves represent intracardiac ECG from the ventricular catheter, and lower ECG curves represent intracardiac ECG registration from the atrial catheter. Note postextrasystolic compensatory pauses after VPBs are pathognomonic for ventricular extrasystolic beats. B) A spontaneously occurring episode of polymorphic VT was noted in 1 Kcne5^−/0 mouse. The episode could not be terminated by burst pacing (pacing spikes on the left). The VT finally converted into an atrial arrhythmia with a 2:1 and 3:1 atrioventricular conduction. C) Polymorphic VTs were inducible during continuous ECG monitoring of sedated live Kcne5^−/0 mice. After a defined pacing protocol followed by 3 extrastimuli polymorphic VTs were highly inducible in Kcne5^−/0 mice as compared to WT controls. Data for ECG curves were obtained as described in A. D) Percentage of inducible protocols was increased in Kcne5^−/0 mice vs. WT controls. E) The ventricular effective refractory period was reduced in Kcne5^−/0 mice vs. Kcne5^+/0 controls. ***P < 0.001.

Kcne5 gene deletion augments specific ventricular K_V currents

We next used whole-cell patch-clamp electrophysiology to quantify K_V currents in ventricular myocytes from adult male mice, to examine possible mechanisms for arrhythmogenesis in Kcne5^–/0 mice. In previous studies of adult mice, it was discovered that their ventricular myocyte K_V current is composed primarily of the fast outward transient current (I_to,f), which is generated by K_V4.2, K_V4.3, or both channels and is sensitive to heteropodotoxin (HpTx)-2; the slow transient outward current (I_to,s), probably generated by K_V1.4 (I_K,slow); and a steady-state component (I_ss) (26). I_K,slow comprises the 50 μM 4-AP-sensitive K_V1.5 current (I_K,slow1) (27, 28) and the 25 mM TEA-sensitive K_V2.1 current (I_K,slow2) (27, 29). We studied ventricular myocytes isolated from the apex (Fig. 3) and the septum (Fig. 4). Similar to our prior work with the Kcne2^−/− mouse line (30), we found that apical myocytes isolated from either Kcne5^–/0 or Kcne5^+/0 mice lacked I_to,s, whereas septal myocytes were divided roughly equally into those with I_to,f and those without (as assessed by curve fitting the current decay at +40 mV) (Table 1).

Figure 3 .

Kcne5 gene deletion augments ventricular apical myocyte K­_V currents. A) Representative native whole-cell K_V currents in Kcne5^+/0 (left) and Kcne5^−/0 (right) ventricular apex cardiomyocytes (voltage protocol inset) (n = 27–31). B) Mean current density vs. voltage characteristics (I-V curve) from recordings as in A for Kcne5^+/0 (filled squares) and Kcne5^−/0 (open squares) apical myocytes (n = 27–31). ***P < 0.001 at +60 mV. C) Typical digitally subtracted 25 mM TEA-sensitive current (I_K,slow2) for Kcne5^+/0 (left) and Kcne5^−/0 (right) ventricular myocytes. D) Mean I-V relationships for TEA-sensitive currents as in C (n = 9–10 cells/genotype, derived from 3 mice/genotype). **P = 0.01 at +60 mV. E) Typical digitally subtracted current traces for 50 μM 4-AP-sensitive current (I_K,slow1) for Kcne5^+/0 (left) and Kcne5^−/0 (right) ventricular myocytes. F) Mean I-V relationships for myocytes as in E (n = 9–10 cells/genotype, derived from 5 mice/genotype). P < 0.05 at +60 mV. G) Typical digitally subtracted 500 nM HpTx-sensitive current (I_K,slow2) for Kcne5^+/0 (left) and Kcne5^−/0 (right) ventricular myocytes. H) Mean I-V relationships for HpTx-sensitive currents as in G (n = 6 cells/genotype, derived from 4 to 5 mice per genotype). P = 0.45 at +60 mV.

Figure 4 .

Kcne5 gene deletion augments I_K,slow2 in ventricular septum myocytes lacking I_to,f. A) Typical current traces from Kcne5^+/0 and Kcne5^−/0 adult ventricular septum myocytes (voltage protocol inset). B) Mean I-V relationships for Kcne5^+/0 (solid) and Kcne5^−/0 (open) septal myocytes with (squares) or without (circles) I_to,f (n = 15–19 cells per group). *P < 0.05 at +60 mV. C) Typical digitally subtracted 25 mM TEA-sensitive K_V2.1 current (I_K,slow2) for Kcne5^+/0 (left) and Kcne5^−/0 (right) ventricular septum myocytes. D) Mean I-V relationships for TEA-sensitive currents as in C (n = 6 cells/genotype, derived from 2 mice/genotype). **P = 0.009 at +60 mV. E) Typical digitally subtracted current traces for 50 μM 4-AP-sensitive K_V1.5 current (I_K,slow1) for Kcne5^+/0 (left) and Kcne5^−/0 (right) ventricular myocytes. Dashed lines: 0 current level. F) Mean I/V relationships for myocytes as in E (n = 8 cells/genotype, derived from 2 mice/genotype). P = 0.984 at +60 mV. Dashed lines: 0 current level.

TABLE 1.

Fitted K_V current densities and inactivation kinetics in ventricular apex and septum myocytes from Kcne5^+/0 and Kcne5^−/0 mice at +40 mV

Genotype and myocyte type	I to,f	I to,s	I K,slow	n
Kcne5+/0 apex	68.6 ± 2.5	NA	1351 ± 59.5	27
	14.7 ± 0.9	NA	8.5 ± 0.6	27
Kcne5+/0 septum, with Ito,f	52.1 ± 3.3	526.3 ± 35	3869 ± 943	19
	14.3 ± 1.6	9.2 ± 0.5	11.2 ± 1.0	19
Kcne5+/0 septum, no Ito,f	NA	165.7 ± 11.6	1601.8 ± 76	16
	NA	5.8 ± 0.8	9.5 ± 1.1	16
Kcne5−/0 apex	66.7 ± 2.6	NA	1134 ± 31.4***	31
	17.5 ± 0.9 *	NA	11.6 ± 0.7***	31
Kcne5−/0 septum, with Ito,f	40.8 ± 1.8**	383.9 ± 22.8***	1857 ± 135	15
	16.3 ± 1.4	8.6 ± 1.1	11.2 ± 1.1	15
Kcne5−/0 septum, no Ito,f	NA	159.2 ± 15	1250 ± 88***	16
	NA	8.5 ± 1.4	12.9 ± 1.7 *	16

Values are means ± se. Current properties: τ_decay (ms); I (pA/pF). I, current density; τ_decay, inactivation time constant; n, number of cells; NA, not applicable (current not present). Statistically significant changes in current density are underlined.

^*P < 0.05, **P < 0.01, ***P < 0.005 vs. equivalent criteria in WT mice.

In apical myocytes, Kcne5 deletion increased peak total K_V current density by 22% (Fig. 3A, B). As assessed by curve fitting, the primary effect in apical myocytes of Kcne5 deletion was a 36% increase in I_K,slow density. Kcne5 deletion also increased apical myocyte I_to,f density, by 20% (Table 1). We next used pharmacological analysis as an additional methodology for quantifying effects on myocyte K_V currents. In apical myocytes, Kcne5 deletion increased the density at +60 mV of both components of I_K,slow: the 50 μM 4-AP-sensitive K_V1.5 current (I_K,slow1) by 78% and the 25 mM TEA-sensitive K_V2.1 current (I_K,slow2) by 30%. The apical myocyte HpTx2-sensitive K_V4 current (I_to,f) was increased by 13% by Kcne5 deletion, although the increase did not reach statistical significance (Fig. 3C–H). Thus, the results of curve fitting and pharmacological analyses were essentially congruent.

In the septum, Kcne5 deletion increased the myocyte peak total K_V current density at +40 mV by 32% in myocytes lacking I_to,f, but left the peak total K_V current unchanged in septal myocytes with I_to,f (Fig. 4A, B). As quantified by curve fitting, the sole effect of Kcne5 deletion on specific currents in the septum was an increase in I_K,slow density by 36% in septal myocytes lacking I_to,f. In contrast, Kcne5 deletion left I_K,slow, I_to,s, and I_to,f density unchanged in septal myocytes with I_to,f (Table 1). Pharmacological analysis in a population of septal myocytes including both those with I_to,f and those without (as assessed by curve fitting) showed that mean Kcne5^–/0 septum myocyte K_V2.1 (25 mM TEA-sensitive) current (I_K,slow2) amplitude was 37% greater at +40 mV than that of Kcne5^+/0 myocytes (Fig. 4C, D). In contrast, K_V1.5 (50 µM 4-AP-sensitive) current (I_K,slow1) amplitude at +40 mV was unchanged by Kcne5 deletion (Fig. 4E, F). HpTx2 sensitivity was not quantified in septal myocytes because of the challenges of pharmacologically assessing I_to,f with a toxin in this mixed cell population. As described above, curve-fitting data indicated that Kcne5 deletion did not alter I_to,f in the septum (Table 1). Finally, Kcne5 deletion did not alter ventricular K_Vα subunit transcript expression (Supplemental Fig. S3), suggesting that the effects on currents were primarily from loss of Kcne5 interaction rather than from transcript remodeling.

Kcne5 and K_V2.1 colocalize at ventricular myocyte intercalated discs

We analyzed the K_V2.1-Kcne5 interaction in more detail, because it occurred in both apical and septal myocytes. Using immunofluorescence, we found that Kcne5 localizes strongly to the intercalated discs (ICDs) in ventricular myocytes, with Kcne5^–/0 tissue serving as a negative control for antibody specificity (Fig. 5A) and β-catenin as a marker for the ICDs (Fig. 5B). Similarly, we detected K_V2.1 strongly localized to the ICDs, using Connexin 43 as an ICD marker in these experiments (Fig. 5C). There was no evidence that Kcne5 deletion altered the ability of K_V2.1 to specifically target the ICDs and no evidence of increased K_V2.1 protein expression overall.

Figure 5 .

Kcne5 and K_V2.1 colocalize in the ICD of mouse ventricular cardiomyocytes. A) Immunofluorescence staining of ventricular heart sections of Kcne5^+/+ and Kcne5^−/− mice with an anti-Kcne5 antibody. Kcne5 is detectable in ICDs of cardiomyocytes (arrows) whereas Kcne5 signal is absent in Kcne5^−/− hearts. Yellow framed areas are shown in higher magnification as indicated. B) Coimmunofluorescence staining with antibodies against Kcne5 (red) and β-Catenin (green) for ICD visualization. No Kcne5 signal was detectable in Kcne5^−/− hearts. C) Coimmunofluorescence staining with antibodies against K_V2.1 (red) and Connexin-43 (green) as ICD marker proteins. Scale bars, 20 µm.

K_V2.1 and KCNE5 colocalize at the plasma membrane and in intracellular vesicles in HL-1 cells

To facilitate more detailed analysis of KCNE5-K_V2.1 localization and examination of both mouse and human subunits, we used transient expression of epitope-tagged KCNE5 and K_V2.1 in the mouse atrial cardiomyocyte tumor lineage HL-1. When individually expressed, detection with an in-house KCNE5-specific antibody revealed that human KCNE5 was located both in small, dense clusters, and in some vesicle-like structures (Fig. 6A). The ratio between these 2 different entities varied. To address possible species differences, we also transfected HL-1 cells with mouse Kcne5. The mouse protein localized similar to the human protein in these cells (Supplemental Fig. S4). To deduce whether these KCNE5⁺ clusters were at the membrane or inside the cell, we generated an N-terminal, HA-tagged KCNE5 construct. The tag did not disturb the trafficking of the subunit, as the HA-KCNE5 localized similarly to the untagged KCNE5 (Fig. 6B), nor did the tag affect biophysical properties when coexpressed with K_V2.1 (Supplemental Fig. S5). When a specific HA-tag antibody was applied without permeabilizing the cell (Fig. 6C, surface), the same dense clusters appeared, indicating plasma membrane expression. Permeabilization of the same cells and exposure to the anti-KCNE5 antibody revealed additional KCNE5⁺ intracellular vesicles (whole cell).

Figure 6 .

KCNE5 and K_V2.1 colocalize in HL-1 cells. A, B) Confocal images of HL-1 cells heterologously expressing WT KCNE5 (A) or HA-tagged WT KCNE5 (B). Phalloidin (Pha) labeled the submembranous actin cytoskeleton and was used to visualize the cell membrane. C) Nonpermeabilized HA-KCNE5–expressing HL-1 cells stained with a specific HA-tagged antibody (surface staining, green), permeabilized, and then stained with a KCNE5-specific antibody (red) and Phalloidin (blue). D) Colabeling of single transfected K_V2.1 (green) and endogenous Phalloidin (red) in HL-1 cells. E) Colabeling of cotransfected KCNE5 (green) and K_V2.1 (red) in HL-1 cells.

Next, we evaluated colocalization of KCNE5 with K_V2.1 in mouse atrial cardiomyocyte tumor lineage HL-1 cells. When expressed alone, K_V2.1 was predominantly located at the cell surface, overlapping phalloidin, which labels the actin cytoskeleton just beneath the plasma membrane. In most cells K_V2.1 could also be found in dense clusters and vesicles, similar to KCNE5 and consistent with results from studies of HEK293 cells (20, 30–33) (Fig. 6D). Upon coexpression of KCNE5 with K_V2.1, the channel subunits colocalized in the same clusters and vesicles in most of the cells (Fig. 6E), also in line with our recent results in HEK293 cells (18). Upon costaining with LAMP-2, we observed that KCNE5 and K_V2.1 localized to intracellular compartments that were in proximity to the lysosomes, indicative of late endosome compartments (Fig. 7A–C).

Figure 7 .

KCNE5 and K_V2.1 colocalize in intracellular compartments in proximity to the lysosomes. A, B) Laser confocal microscopy was used to acquire images of KCNE5 and K_V2.1 subunits singly transfected into HL-1 cells, respectively. Both were found to be located in dense clusters and vesicles for which some appear very close to the lysosomal marker LAMP2. Phalloidin (Pha) was used to identify the cell membrane. C) When K_V2.1 and KCNE5 were coexpressed in the same HL-1 cells, they colocalized in vesicles and dense clusters, similar to the above-mentioned structures, which again were located in proximity to the lysosome maker, LAMP2—possibly indicative of late endosome compartments.

Arrhythmia-associated KCNE5 mutations confer functional changes on the K_V2.1-KCNE5 channel complex

Although K_V2.1 is primarily characterized in rodent cardiac electrophysiology (31, 32), both K_V2.1 mRNA (33) and protein (34) are reportedly expressed in human atria and ventricles. With further evidence of an interaction between KCNE5 and K_V2.1 from our previous work (18) and the current study, we next investigated the effect of human KCNE5 genetic variants that had been linked to AF and BrS (15–17, 35). We coexpressed the KCNE5 AF variants P33S, L65F, or R85H (Fig. 8) and the KCNE5 BrS variants Y81H or D92E/E93X (Fig. 9) with K_V2.1 in HEK293 cells. Currents generated by K_V2.1 with KCNE5-P33S and Y81H were similar to those generated by K_V2.1 and WT KCNE5. In contrast, KCNE5-L65F shifted both the K_V2.1 voltage dependence of activation (Fig. 8A–C) and the voltage-dependence of inactivation (Fig. 8D) ∼10 mV toward negative potentials. KCNE5-L65F also modified the K_V2.1 activation kinetics significantly (with a ∼2-fold acceleration at 0 mV) without affecting the deactivation kinetics (Fig. 8E). In addition, KCNE5-L65F increased the K_V2.1 current density (∼2-fold) compared with KCNE5-WT (243 ± 55 and 100 ± 25 pA/pF, respectively) (Fig. 8F). Coexpression of K_V2.1 with KCNE5-R85H changed neither the voltage-dependence of inactivation (Fig. 8D) nor the activation or deactivation kinetics (Fig. 8E). Yet, KCNE5-R85H tended to shift the voltage dependence of activation toward positive potentials (Fig. 8C) and to further reduce K_V2.1 current density (Fig. 8F), albeit not reaching statistical significance. KCNE5-D92E/E93X shifted the K_V2.1 voltage-dependence of activation (Fig. 9A–C) ∼6 mV toward more negative potentials. Data for channels containing mutant KCNE5 are summarized in Table 2. The mutant forms of KCNE5 did not perturb KCNE5 trafficking in mouse atrial cardiomyocyte tumor lineage HL-1 cells (36) (Fig. 10A–C). Thus, the KCNE5-L65F and D92E/E93X mutations exerted their effects by their impact on the electrical properties of K_V2.1.

Figure 8 .

AF-associated KCNE5-L65F causes a gain-of-function of K_V2.1-KCNE5 channels. A–D) Biophysical properties of K_V2.1 coexpressed with WT or mutant human KCNE5. All recordings were performed on cells from 3 to 5 repeated transfections of 2–5 independent cell cultures. Representative current recordings determined the activation (A) and inactivation (B) properties of K_V2.1+KCNE5 and K_V2.1+KCNE5-L65F. The applied pulse protocols are given on top. Voltage-dependence of activation (C) and of inactivation (D) of K_V2.1+KCNE5, K_V2.1+KCNE5-P33S, K_V2.1+KCNE5-L65F, and K_V2.1+KCNE5-R85H. Activation and inactivation curves were obtained by plotting the normalized current amplitudes at −25 and +60 mV as a function of the 500-ms and 5-s prepulse potentials, respectively. Solid lines: Boltzmann fits. Coexpression of KCNE5-L65F shifted both the voltage dependence of activation and of inactivation. P < 0.05. E) Time constants of activation and deactivation of K_V2.1 coexpressed with KCNE5 variants as indicated; symbols as in C. Activation and deactivation constants were derived from single and double exponential fits of the raw current recordings, respectively. F) Current densities at 0 mV of K_V2.1 coexpressed with KCNE5 variants, as indicated. The numbers above each bar represent the number of cells analyzed. The L65F mutation increased current density compared with WT K_V2.1-KCNE5 channels. *P < 0.05.

Figure 9 .

BrS-associated KCNE5 mutations cause a negative shift in the voltage dependence of K_V2.1-KCNE5 channels. All recordings were performed on cells from 3 to 5 repeated transfections of 2–5 independent cell cultures A–D) Biophysical properties of K_V2.1 coexpressed with WT or mutant human KCNE5. Representative current recordings determined the activation (A) and inactivation (B) properties of K_V2.1+KCNE5-Y81H and K_V2.1+KCNE5- D92E/E93X. The applied pulse protocols are given on top. Voltage-dependence of activation (C) and of inactivation (D) of K_V2.1+KCNE5, K_V2.1+KCNE5-Y81H, and K_V2.1+KCNE5- D92E/E93X. Activation and inactivation curves were obtained as in Fig. 4. Solid lines: the Boltzmann fits. Coexpression of KCNE5-D92E/E93X shifted the voltage dependence of activation. P < 0.05. E) Time constants of activation and deactivation of K_V2.1 coexpressed with KCNE5 variants as indicated; symbols as in C. Activation and deactivation constants as in Fig. 4E. F) Current densities at 0 mV of K_V2.1 coexpressed with KCNE5 variants as indicated. The numbers above each bar represent the number of analyzed cells.

TABLE 2.

Biophysical properties of human K_V2.1 alone or coexpressed with WT or mutant human KCNE5

Subunits	Activation				Inactivation			Deactivation
	V 1/2	k	τ (at +60 mV)	n	V 1/2	k	n	τ1 (at −40 mV)	τ2 (at −40 mV)	n
KV2.1	0.0 ± 1.9	9.4 ± 0.5	15.7 ± 1.6	6	−26.5 ± 1.6	5.2 ± 0.5	5	15.3 ± 1.1	73.1 ± 11.1	4
KV2.1+
WT	1.3 ± 2.2	9.7 ± 0.7	17.4 ± 0.9	7	−24.3 ± 2.8	5.7 ± 0.2	8	15.6 ± 1.0	61.4 ± 5.1	8
P33S	−0.2 ± 2.5	10.0 ± 1.1	16.0 ± 1.2	6	−23.9 ± 2.1	5.6 ± 0.4	6	16.2 ± 2.6	68.0 ± 7.6	5
L65F	−10.4 ± 2.4	9.1 ± 1.3	13.9 ± 1.1	7	−34.8 ± 3.3	5.6 ± 0.2	7	17.0 ± 1.1	75.3 ± 5.2	4
R85H	7.0 ± 1.8	10.4 ± 1.0	16.4 ± 0.7	6	−24.1 ± 2.4	6.4 ± 0.3	6	16.0 ± 1.1	63.9 ± 7.7	5
Y81H	−2.4 ± 2.1	9.6 ± 0.6	14.8 ± 0.3	6	−26.9 ± 3.9	5.4 ± 0.3	5	18.0 ± 1.5	67.2 ± 10.7	6
D92E/E93X	−5.2 ± 1.4	8.8 ± 0.5	13.4 ± 0.9	7	−26.4 ± 1.3	5.8 ± 0.3	5	16.3 ± 1.8	84.4 ± 3.5	5

Values are means ± sem. V_1/2, midpoint (mV) of activation or inactivation; k, slope factor; τ, time constant (ms); n, number of cells. Values underlined showed a statistically significant difference compared to K_V2.1+KCNE5 values. P < 0.05.

Figure 10 .

AF-associated KCNE5 mutants locate similarly to WT KCNE5 in HL-1 cells. KCNE5-P33S (A), KCNE5-L65F (B), and KCNE5-R85H (C) expressed in HL-1 cells alone were found in compartments similar to WT KCNE5 (compare to Fig. 6) visualized with confocal microcopy. Thus, the mutations did not seem to confer a trafficking deficiency. Phalloidin (Pha) was used to mark the cell membrane.

DISCUSSION

Human KCNE5 sequence variants are associated with both AF (16, 17) and BrS (15), conditions that increase the risk of mortality (35, 36). The functional mechanisms linking KCNE5 to these cardiac arrhythmia syndromes have been suggested to be dysregulation of I_Kslow (16) or I_to (15), respectively, but little is known of the roles of KCNE5 in vivo. To aid in this endeavor, we studied the effects of germline deletion of Kcne5 in mice. Transgenic mouse models have been instrumental in detecting and characterizing novel native functions of KCNE subunits (23, 25, 30, 37–45). By phenotypically studying Kcne5^–/0 mice, we discovered that Kcne5 regulates K_V2.1 current (I_K,slow,2) (27, 29) in mouse ventricles. Furthermore, we demonstrated region- and myocyte-subtype–specific differences in the ventricular role of Kcne5, observing that Kcne5 deletion augments I_K,slow,1 (generated by K_V1.5) (in addition to I_Kslow,2) in the apex but not the septum. Previously, we found that Kcne2 deletion downregulates ventricular I_Kslow,2 by hampering K_V1.5 trafficking to the ICDs (30). We also found that, despite not being expressed in mouse heart, Kcne3 deletion predisposes to ventricular arrhythmias and AF via an autoimmune attack on the adrenal glands that increases serum aldosterone (19, 23). In the atria, this tendency was demonstrated to occur via increased K_V1.5 recycling, leading again to diminished I_Kslow,2 (19). Future studies will focus on elucidating the mechanism for KCNE5 modulation of K_V1.5 current density.

In the present study, Kcne5 deletion increased I_to,f density, but the change was less dramatic than we observed for I_K,slow and was also restricted to the apical myocytes, despite the known capacity for KCNE5 to robustly regulate K_V4 channels in vitro (15, 46, 47). This finding, together with the apical-specific I_K,slow,1 regulation by KCNE5 suggests that myocyte subtype-specific mechanisms are in place to limit K_Vα subunit regulation by KCNE5, despite its capacity to interact in other cell subtypes or contexts.

Our finding that Kcne5 is expressed in both atria and ventricles in mice is congruent with our expression data for human cardiac tissue, where we identified KCNE5 throughout the human heart, in line with prior studies (5–7), with slightly raised expression in the atria. We also detected Kcne5 protein expression specifically at the ventricular ICDs. We observed neither cardiac hypertrophy nor K_Vα subunit transcript remodeling in Kcne5^–/0 mice, suggesting that the arrhythmias we detected by intracardiac ECG arose primarily from a purely electrical defect.

Our findings are the first, to our knowledge, to demonstrate β-subunit regulation of K_V2.1 in mice, and the augmentation of I_K,slow arising from Kcne5 deletion is consistent with current knowledge of the mechanistic bases for human AF and BrS (i.e., pathologic increase in specific K_V currents; the more common basis for BrS is loss of function of voltage-gated sodium channels, but gain in K_V current has a similar effect). The relatively subtle and apically restricted effect of Kcne5 deletion on I_to,f in mouse ventricles is of interest, given that Ohno and colleagues (15) reported that BrS-linked KCNE5 mutants increase the current conducted by human K_V4 channels. K_V4 channels may be regulated differently with respect to KCNEs in mouse vs. human hearts, with more extensive modulation in the latter. Alternatively, the capacity of KCNE5 mutants to alter K_V4-KCNE5 channel function in vitro may not be representative of the primary molecular mechanism underlying human BrS disease pathogenesis. In the mouse model presented herein, Kcne5 deletion gave rise to specific K_V current augmentation, and macroscopically caused VPBs, which were followed by postextrasystolic compensatory pauses and in one case, a spontaneous polymorphic VT. Thus, as is thought to occur in human BrS, augmentation of specific K_V currents in mice via Kcne5 disruption caused ventricular arrhythmia.

Our data present strong evidence of a novel KCNE5-K_V2.1 interaction in mouse cardiac myocytes; however, little is known about how these subunits interact. Expression of the subunits in HL-1 cells revealed that neither subunit affected the trafficking or localization of the other; however, our electrophysiology data suggest that KCNE5 directly regulates K_V2.1 channel attributes at the plasma membrane, in line with our previous work in HEK293 cells (20). We detected most of both human and mouse KCNE5 in dense clusters at the membrane and in some intracellular clusters and vesicles located in proximity to the lysosomal marker LAMP2. K_V2.1 was mainly located at the border of the cell, overlapping the cell membrane marker phalloidin, but could also be found in vesicular structures similar to KCNE5. We speculate that some of the intracellular compartments positive for KCNE5 subunits are compartments of the late endosomal pathway and could serve either as a recycling reserve or as part of an endocytic mechanism regulating the number of K_V2.1 channels at the membrane. These observations are in line with results obtained in HEK293 cells (18). In this model, Kcne5 deletion would prevent this recycling, thereby contributing to augmentation of K_V2.1 current density. A similar model could apply to other channels inhibited by KCNE5.

Finally, in our study, KCNE5-L65F lacked the inhibitory effect on K_V2.1 exhibited by WT KCNE5 and shifted the voltage dependence of activation and inactivation to more negative potentials. To date, no native current within the human heart has been assigned to K_V2.1, yet both K_V2.1 mRNA and protein have been detected in human ventricles and atria (33, 34). Ravn et al. (16) speculated that the L65F mutation in KCNE5 is associated with nonfamilial or acquired forms of AF, given that it causes a gain-of-function of I_Ks. Our findings, at the very least, suggest mechanistic commonalties between KCNE5 regulation of K_V7.1 and K_V2.1 and the possibility that human KCNE5-linked AF is more complex than first thought and perhaps involves K_V2.1, a channel of which little is known in terms of its importance in the human heart.

CONCLUSIONS

We provide the first description of the physiologic roles of KCNE5 in vivo. Mouse Kcne5 functions as a ventricular myocyte subtype-specific inhibitor of I_Kslow,1 and I_Kslow2, which are conducted by K_V1.5 and K_V2.1, respectively, and to a lesser extent I_to,f (K_V4 α subunits). Kcne5 deletion increases ventricular arrhythmia susceptibility. These findings provide potential mechanistic clues to the molecular basis of human KCNE5-linked BrS and AF.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

4-AP

4-aminopyridine

ACTB

β-actin

AF

atrial fibrillation

BrS

Brugada syndrome

ECG

electrocardiograph

HA

hemagglutinin

HEK

human embryonic kidney

HpTx

heteropodatoxin

ICD

intercalated disc

I _K,slow

slow transient outward current generated by K_V1.4

I _K,slow1

slow transient outward current generated by K_V1.5

I _K,slow2

slow transient outward current generated by K_V2.1

I _ss

steady-state current

I _to,s

slow transient outward current

I_to,f

fast transient outward current, Kcne, potassium voltage-gated channel, subfamily E

K_V

voltage-gated potassium channel

LAMP

lysosomal-associated membrane protein

TEA

tetraethylammonium

TOP1

topoisomerase-1

VPB

ventricular premature beat

VT

ventricular tachycardia

WT

wild type

AUTHOR CONTRIBUTIONS

J.-P. David, U. Lisewski, T. K. Roepke, N. Schmitt, and G. W. Abbott designed the research; J.-P. David, U. Lisewski, S. M. Crump, T. A. Jepps, E. Bocksteins, N. Wilck, and J. Lossie performed the research; J.-P. David, U. Lisewski, S. M. Crump, T. A. Jepps, E. Bocksteins, N. Wilck, J. Lossie, and G. W. Abbott analyzed the data; and J.-P. David, S. M. Crump, T. A. Jepps, E. Bocksteins, T. K. Roepke, N. Schmitt, and G. W. Abbott wrote the manuscript.