Sudden death–associated KCNH2 variants exert opposing effects on a nuclear subdomain of the cardiac potassium channel hERG1

TL;DR

This study shows that genetic variants in a nuclear subdomain of a cardiac potassium channel can cause sudden death in the young by either reducing or increasing the channel's function.

Contribution

The study reveals that hERG1NP variants can cause both loss-of-function and gain-of-function effects, offering new insights into sudden death mechanisms.

Findings

Two hERG1NP variants (R885C and R1047L) disrupted hERG1NP activity by altering nuclear transport or abolishing current suppression.

Two variants (G1036D and Q1068R) enhanced hERG1NP activity by altering the voltage dependence of hERG1a channels.

Two variants (R1035W and R1069S) had no effect on hERG1NP activity.

Abstract

Human Ether-à-go-go–Related Gene 1 (hERG1) nuclear peptide (hERG1NP) is a recently discovered nonconducting subdomain of the full-length hERG1 channel that is expressed in the nuclei of developing cardiomyocytes. From the nucleus, hERG1NP modulates the gating and expression of the full-length hERG1 channel, introducing an unexplored regulatory mechanism of cell physiology. hERG1NP variants are linked with sudden death in the young, but the impact of these variants on hERG1NP activity is unknown. To determine the effect of hERG1NP variants on hERG1NP activity, we measured hERG1NP intracellular localization and modulation of membrane current from the full-length hERG1a channel in human embryonic kidney 293 cells. Wildtype hERG1NP suppressed both hERG1a current (IhERG) and protein. We then screened six hERG1NP variants for changes in either intracellular targeting or modulation of hERG1a current. Two variants, R885C and R1047L, disrupted hERG1NP activity by altering nuclear transport or abolishing both IhERG current suppression and nuclear targeting, respectively. Two additional variants, G1036D and Q1068R, enhanced hERG1NP activity by depolarizing hERG1a's voltage dependence of activation via accelerated channel deactivation and recovery from inactivation. Last, two variants had no effect: R1035W and R1069S. This work demonstrates that hERG1NP variants associated with sudden death in the young can trigger both loss-of-function and gain-of-function effects on hERG1NP. Additional work is needed to identify specific pathogenic mechanisms of hERG1NP variants, but these data demonstrate that hERG1NP variants may represent a novel mechanism of disease.

Figures

Full text

KCNH2 encodes the primary subunits of the voltage-gated potassium channel, human Ether-à-go-go–Related Gene 1 (hERG1), which conducts the rapid delayed rectifier potassium current (IKr). In cardiac tissue, reduced IKr caused by loss-of-function KCNH2 variants or off-target pharmacological block causes the cardiac disorder long QT (LQT) syndrome (1, 2). Patients with LQT syndrome are at increased risk of syncope, cardiac arrhythmia, and sudden cardiac death (1, 2). LQT syndrome–associated KCNH2 variants are also linked with sudden infant death syndrome (SIDS), intrauterine fetal death, and sudden unexplained death in epilepsy (SUDEP), yet the mechanistic link between sudden death in the young and KCNH2 variants is not fully understood (3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16).

hERG1 subunit abundance critically regulates human cardiac excitability. In human cardiomyocytes, at least two hERG1 subunits combine to conduct cardiac IKr, hERG1a, and hERG1b (17, 18, 19, 20, 21). hERG1a and hERG1b are identical, save for their N termini (17, 18). The hERG1a N-terminal domain includes a Per–Arnt–Sim (PAS) domain that interacts with the cytoplasmic S4–S5 linker, along with the cyclic nucleotide–binding homology domain (CNBHD) of the C terminus, to slow channel gating (22, 23, 24, 25, 26). The hERG1b N-terminal domain is unique and lacks a functional PAS domain. In a heterologous expression system, coexpressing hERG1a with hERG1b significantly accelerates hERG1 gating and increases hERG1 current compared with homomeric hERG1a channels. In human cardiomyocytes, changes in the relative abundance of hERG1a and hERG1b modulate IKr magnitude, action potential morphology, and susceptibility to arrhythmia (21, 27, 28, 29).

The C-terminal cytoplasmic domain of hERG1 can be divided into proximal and distal halves. The proximal half is highly structured and includes the CNBHD that critically regulates channel gating (22, 30, 31). KCNH2 variants within the proximal C-terminal domain disrupt hERG1 gating and surface expression, leading to LQT syndrome (31, 32). The distal half of the C-terminal domain is unstructured, save for a predicted coiled-coil domain spanning residues 1035 to 1073 (hERG1a numbering) (33, 34, 35, 36). KCNH2 variants within the distal half of the C-terminal domain are linked with SIDS and SUDEP, yet they minimally affect full-length channel function, and the mechanism linking them with sudden death is unclear (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 37, 38).

Previous work from our laboratory identified a KCNH2-encoded polypeptide, hERG1 nuclear peptide (hERG1_NP_). hERG1_NP_ is a subdomain of the full-length hERG1 channel that is targeted almost exclusively to the cell nucleus (39). Native hERG1_NP_ is enriched in the nuclei of immature cardiomyocytes and maps to the distal half of the C-terminal domain of the full-length hERG1 channel. Like native hERG1_NP_, the distal C-terminal domain is targeted to the cell nucleus, where it reduces hERG1 current (IhERG) by roughly half in human embryonic kidney 293 (HEK293) cells stably expressing hERG1a (HEK293-hERG1a cells). This initial study identified hERG1_NP_ as a novel modulator of hERG1 activity and a potential modifier of cardiac excitability. From the nucleus, hERG1_NP_ has the capacity to broadly regulate cellular physiology. Variants within hERG1_NP_ may therefore represent a novel trigger of cellular dysfunction that is distinct from the well-established role of full-length hERG1 in congenital and acquired LQT syndromes.

We hypothesized that KCNH2 variants within the distal C-terminal domain may selectively disrupt hERG1_NP_ activity. Here, we identify four hERG1_NP_ variants associated with sudden death in the young that trigger loss-of-function (R885C, R1047L) or gain-of-function (G1036D, Q1068R) effects on hERG1_NP_ using hERG1a surface membrane current as a functional measure of activity. These data demonstrate for the first time that KCNH2 variants associated with sudden death disrupt hERG1_NP_ activity and its regulation of the full-length hERG1 channel and may represent a novel mechanism of cellular dysfunction.

Experimental procedures

HEK293-hERG1a cell culture

We maintained cells at 37 °C with 5% CO_2_ in Heracell incubators (Thermo Fisher). We cultured HEK293 cells stably expressing hERG1a channels (HEK293-hERG1a) in Dulbecco’s modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Gibco, catalog no.: 26140079) and 1% penicillin/streptomycin (P/S; 10,000 U/ml, Gibco, catalog no.: 15140122). We split cells every 3 to 5 days at 60% to 80% confluency with PBS (Gibco, catalog no.: 10010023) and 0.05% trypsin–EDTA (Gibco, catalog no.: 25300054). We added 50 μg/ml geneticin antibiotic (Gibco, catalog no.: 10131035) to all freshly split cells immediately after passaging to maintain stable expression of hERG1a channels.

DNA constructs

The distal C-terminal domain (873–1159-mCitrine.pcDNA3) plasmid was kindly provided by Professor Matthew Trudeau at the University of Maryland Medical School. To generate the R1047L mutant, we designed forward and reverse primers with QuikChange Primer Design (agilent.com) and synthesized the primers through Eurofins (forward: 5′-cctgttgagctggagctggagggcatc-3′, reverse: 5′-gatgccctccagctccagctcaacagg-3′). Mutants R885C, R1035W, G1036D, Q1068R, and R1069S were inserted into the distal plasmid backbone by GenScript. We verified all construct sequences with DNA sequencing provided by Eurofins.

Quantitative RT–PCR

To measure hERG1a mRNA levels in HEK293-hERG1a cells, total RNA was prepared using the RNeasy Mini Kit (Qiagen). We synthesized complementary DNA (cDNA) by reverse transcribing 300 ng of RNA with M-MLV reverse transcriptase (Invitrogen, catalog no.: 28025-013) and oligo(dT)12–18 primers. We used IDT Mastermix (Thermo Fisher, catalog no.: 1055772) and TaqMan assay primers (10 μM, Thermo Fisher, catalog no.: 4331182 and 4351372) for GAPDH and hERG1a, respectively. We ran samples at 95 °C for 30 s, then 39 cycles of 95 °C for 3 s and 60 °C for 20 s. Melting-curve analysis verified the Amplicon correctness. We analyzed samples in technical triplicates using a Bio-Rad C1000 Touch Thermal Cycle CFX96 (Applied Biosystems). The expression of hERG1a mRNA relative to GAPDH in cells transfected with either GFP or the distal C-terminal domain (hERG1_NP_) was calculated by the ΔΔCT method, based on the threshold cycle (CT), as fold change = 2^−(ΔΔCT)^, where ΔCT = CT_hERG1a −_ CT_GAPDH_ and ΔΔCT = ΔCT_hERG1NP cells_ – ΔCT_GFP cells_. From each experiment, the cDNA of three cell culture wells was measured as biological replicates.

Western blot

For protein extraction and quantification, we collected experimental cells from plastic 6-well plates by washing with ice-cold PBS (Gibco, catalog no.: 10010023) and then adding 100 μl radioimmunoprecipitation assay lysis and extraction buffer (Thermo Fisher, catalog no.: 89900). We scraped cells into 1.5 ml microcentrifuge tubes and sonicated all samples before centrifuging at 12,000 RPM for 15 min at 4 °C. After centrifugation, we collected sample supernatants and diluted each 1:20 and added 5 μl per sample to a 96-well plate. We mixed DC Protein Assay Reagents A (Bio-Rad, catalog no.: 5000113) and B (Bio-Rad, catalog no.: 5000114) at a 1:50 dilution and added 200 μl of the A + B mixture to each sample well before incubating the plate at 37 °C for 30 min. We read sample wavelengths at 490 nm to determine protein loading before diluting the sample 1:1 in Laemmli sample buffer (Bio-Rad, catalog no.: 1610737). We ran 40 μg of total sample protein on a 10% SDS-PAGE gel (deionized water, 30% acrylamide, 1.5 M Tris buffer at pH 8.8, 10% SDS, and N,N,N′,N′-tetramethylethylenediamine) topped by a stacking gel (deionized water, 30% acrylamide, 1 M Tris buffer at pH 6.8, 10% SDS, and N,N,N′,N′-tetramethylethylenediamine). Gels ran at 80 V for 20 min and then 120 V for 40 min before transfer. We transferred protein onto nitrocellulose membranes using 200 mA in a Bio-Rad adaptor cell at 4 °C. After 4 h, we blocked with 5% nonfat dry milk in 1X Tris-buffered saline with Tween-20 at room temperature for 1 h on a countertop shaker. We then incubated nitrocellulose membranes in blocking buffer containing a 1:1000 dilution of pan-hERG C-terminal domain primary antibody (Enzo Life Sciences, catalog no.: ALX-215-049-R100) and a 1:1000 dilution of GAPDH primary antibody (Fitzgerald Industries International, catalog no.: 10RG109a) overnight at 4 °C. The next day, we washed each membrane three times for 5 min per wash with 1X Tris-buffered saline with Tween-20 before incubating for 1 h at room temperature in goat anti-rabbit secondary antibody (LI-COR Biosciences, catalog no.: D00115-06) and goat anti-mouse secondary antibody (LI-COR Biosciences, catalog no.: 926-32210), both diluted 1:10,000 in blocking buffer. We washed membranes with 1X TSBT three times for 5 min per wash before developing and quantifying blot protein signals using a LICOR Odyssey instrument.

Immunocytochemistry

We seeded HEK293-hERG1a cells at 60% confluency on 12 mm circular glass coverslips in plastic 24-well plates. We transfected cells with 800 ng DNA 24 h after plating using Lipofectamine 3000, P3000 (Invitrogen, catalog no.: L3000008) and Opti-MEM I Reduced Serum Medium (Gibco, catalog no.: 31985062). We replaced transfected media with fresh DMEM (10% FBS, 1% P/S) 4 to 6 h post-transfection. After 48 h, we fixed cells with 4% paraformaldehyde in PBS for 15 min. We washed the fixed cells three times using PBS (Gibco, catalog no.: 10010023) and stained cell nuclei using 1 μg/ml 4′,6-diamidino-2-phenylindole (Thermo Fisher, catalog no.: 62248) for 15 min. For untransfected cells, prior to 4′,6-diamidino-2-phenylindole staining, we immunolabeled with a 1:150 dilution of pan-hERG primary antibody (Enzo Life Sciences, catalog no.: ALX-215-049-R100) and a 1:250 dilution of goat anti-rabbit secondary antibody AF647 (Southern Biotech, catalog no.: 4050-31) to stain for the hERG1a C-terminal domain. We then washed three more times with PBS (Gibco, catalog no.: 10010023) and mounted the coverslips on microscope slides using ProLong Gold Antifade Reagent (Thermo Fisher, catalog no.: P36930). We completed all imaging using a Zeiss 880 confocal microscope and quantified pan-hERG antibody staining, GFP, or mCitrine localization using FIJI (Fiji Is Just ImageJ), Fiji/ImageJ development community, open-source project, (https://fiji.sc) software.

Electrophysiology

We seeded HEK293-hERG1a cells at ∼20% confluency as single cells in plastic 6-well plates containing 8 to 10 rectangular glass coverslips per well hand-cut using a glass cutter. After 24 h, we transfected cells with 800 ng DNA using Lipofectamine 3000, P3000 (Invitrogen, catalog no.: L3000008) and Opti-MEM I Reduced Serum Medium (Gibco, catalog no.: 31985062). We replaced transfected media with fresh DMEM (10% FBS, 1% P/S) 4 to 6 h post-transfection. We measured IhERG using whole-cell patch-clamp at room temperature, a minimum of 48 h post-transfection.

We completed all recordings at room temperature using whole-cell patch clamp with an IPA Integrated Patch Amplifier run by SutterPatch (Sutter Instrument) within Igor Pro 8 (Wavemetrics). Data were sampled at 5 kHz and low-pass filtered at 1 kHz. We recorded cells in extracellular solution containing (in millimolar): 150 NaCl, 5.4 KCl, 1.8 CaCl_2_, 1 MgCl_2_, 15 glucose, 10 Hepes, 1 sodium pyruvate, and titrated to pH 7.4 using NaOH. Recording pipettes had resistances between 2 and 5 MΩ when backfilled with intracellular solution containing (in millimolar): 5 NaCl, 150 KCl, 2 CaCl_2_, 5 EGTA, 10 Hepes, 5 MgATP, and titrated to pH 7.2 using KOH. We stored single-use 1 ml aliquots of intracellular solution at −20 °C until the day of recording and kept thawed aliquots on ice throughout all recordings. All HEK293-hERG1a cells were visually confirmed to emit GFPmCitrine fluorescence prior to single-cell recording.

To activate IhERG, we stepped cells from a holding potential of −80 mV for 1 s to a 3 s prepulse between −80 mV and +50 mV in 10 mV increments. Tail currents were then measured during a −50 mV, 6 s test pulse, followed by a 1 s postpulse at −80 mV. We normalized peak tail current to cellular capacitance, plotted current density as a function of prepulse potential in millivolts, and fitted the data with the following Boltzmann equation:

$$[eqn]$$

where A1 and A2 represent the maximum and minimum of the fit, respectively, V is the membrane potential, V0 is the midpoint, and k is the slope factor. The time course of IhERG deactivation was assessed by fitting current decay during the test pulse with a double exponential function:

$$[eqn]$$

where Y0 is the asymptote, A1 and A2 are the relative components of the fast and slow time constants, τ_1_ and τ_2_, respectively.

To evaluate the time course of inactivation recovery, we stepped cells from a holding potential of −80 mV for 1 s to a 3 s prepulse at +30 mV to inactivate the channels. We then measured IhERG using 3 s test pulses stepping from −120 mV to −50 mV in 10 mV increments before stepping cells back to −80 mV for a 1 s postpulse. We assessed IhERG inactivation recovery by fitting an exponential function to the current immediately elicited by each respective test pulse.

Statistical analysis

We completed data analysis using Prism 8 (GraphPad Software, Inc) and Igor Pro 8. We assessed values for normality (Shapiro–Wilk test) and for outlier identification (ROUT and Grubbs’ tests) before all statistical evaluations. Outliers for inactivation recovery time constants were determined as ± 2 SDs from the mean. Confocal imaging data are displayed as the median. Data table values are reported as mean ± SD or mean (lower limit, upper limit) 95% confidence intervals where applicable, as indicated in table legends. All other data are reported as mean ± SD. Data were compared using the following statistical analysis methods, where applicable: nonparametric two-tailed Mann–Whitney test, nonparametric Kruskal–Wallis test with a Dunn's multiple comparisons correction, or two-way ANOVA with Bonferroni’s, Dunnett’s, or Tukey’s multiple comparisons tests. We noted statistical significance at p < 0.05. Unless stated otherwise, the number “N“ of observations indicates the number of transfection repeats, and the number “n“ of observations reflects the number of individual HEK293-hERG1a cells evaluated. We completed all experiments as single-blind studies to minimize bias.Figure 1hERG1_NP_ decreases IhERG and protein. A, sample IhERG traces elicited by the protocol in the inset from HEK293-hERG1a cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). The scale bar indicates 2 s by 5 pA/pF. B, steady-state IhERG plotted as a function of test potential from cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). C, peak hERG1a tail current density plotted as a function of prepulse potential and fitted with a Boltzmann equation (Equation 1) from HEK293-hERG1a cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). D, sample Western blot of HEK293-hERG1a cells transfected with either GFP or hERG1_NP_ displaying bands for GAPDH, hERG1_NP_, core-glycosylated hERG1a, and fully glycosylated hERG1a protein. The unlabeled band at 75 kDA is a nonspecific band associated with the anti-hERG1 antibody in HEK293 Western blots. E, total hERG1a protein (core glycosylated + fully glycosylated) normalized to GAPDH from cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). F, fully glycosylated hERG1a protein (155 kDa) normalized to GAPDH from cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). G, core-glycosylated hERG1a protein (135 kDa) normalized to GAPDH from cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). H, quantification of 155 kDa hERG1a protein relative to 135 kDa hERG1a protein from cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). I, quantitative RT–PCR of KCNH2 transcripts normalized to GAPDH from HEK293-hERG1a cells transfected with either GFP (green circles) or hERG1_NP_ (blue squares). Data in B and C were compared using a two-way ANOVA and post hoc Bonferroni's multiple comparisons test. Data in E–I were compared using a nonparametric two-tailed Mann–Whitney test. All error bars represent mean ± SD. N = 2, n = 11 for patch clamp. N = 3, n = 6 for Western blot. N = 3, n = 3 for quantitative RT–PCR. ∗ indicates p < 0.05. HEK293, human embryonic kidney 293 cell line; hERG1, human Ether-à-go-go–Related Gene 1; hERG1NP, hERG1 nuclear peptide; IhERG, hERG1a current.Figure 2hERG1_NP_KCNH2 variants differentially modulate IhERG. A, linear schematic depicting the relative location of KCNH2 variants within hERG1_NP_. B, sample IhERG traces from HEK293-hERG1a cells transfected with either GFP (green) or hERG1_NP_ (blue). C–H, patch-clamp data from HEK293-hERG1a cells expressing GFP (green circles), hERG1_NP_ (blue squares), or hERG1_NP_ carrying one of the following KCNH2 variants (red triangles): (C) R885C_NP_, (D) R1035W_NP_, (E) G1036D_NP_, (F) R1047L_NP_, (G) Q1068R_NP_, or (H) R1069S_NP_. Each panel displays a sample IhERG trace recorded in the presence of the variant (left, red), steady-state current–voltage plot (middle), and peak tail current–voltage plot fitted with a Boltzmann equation (right). All variant data are compared with the same GFP and hERG1_NP_ groups but visually separated for clarity. Data in C–H were compared using a two-way ANOVA and post hoc Dunnett's multiple comparisons test with hERG1_NP_ set as the control group. Error bars represent the mean ± SD. N ≥ 3. N ≥ 7. ∗ indicates p < 0.05 for GFP compared with hERG1_NP_. ^#^ indicates p < 0.05 for KCNH2 variant compared with hERG1_NP_. See Table 1 for specific experimental values. HEK293, human embryonic kidney 293 cell line; hERG1, human Ether-à-go-go–Related Gene 1; hERG1_NP_, hERG1 nuclear peptide; IhERG, hERG1a current.Figure 3G1036D_NP_ and Q1068R_NP_ depolarize the voltage dependence of hERG1a activation. A–F, peak tail current–voltage relationships normalized to the maximum peak tail IhERG for HEK293-hERG1a cells transfected with GFP (green circles), hERG1_NP_ (blue squares), or hERG1_NP_ carrying one of the six KCNH2 variants (red triangles): (A) R885C_NP_, (B) R1035W_NP_, (C) G1036D_NP_, (D) R1047L_NP_, (E) Q1068R_NP_, and (F) R1069S_NP_. Data are plotted as a function of prepulse potential and fitted with a Boltzmann equation (Equation 1). All variant data are compared with the same GFP and hERG1_NP_ groups but visually separated for clarity. Data in A–F were compared using a two-way ANOVA and post hoc Dunnett’s multiple comparisons test with hERG1_NP_ set as the control group. Error bars represent the mean ± SD. N ≥ 3. N ≥ 7. ∗ indicates p < 0.05 for GFP compared with hERG1_NP_. ^#^ indicates p < 0.05 for any given KCNH2 variant compared with hERG1_NP_. See Table 1 for specific experimental values. hERG1, human Ether-à-go-go–Related Gene 1; IhERG, hERG1a current.Figure 4G1036D_NP_ and Q1068R_NP_ accelerate hERG1a′s time course of deactivation. A, sample IhERG traces recorded from HEK293-hERG1a cells transfected with GFP (green), hERG1_NP_ (blue), G1036D_NP_ (red), or Q1068R_NP_ (purple) elicited by the displayed voltage-step protocol and scaled as indicated. Only sample traces for statistically significant variants are represented for clarity. (B) Fast and (C) slow deactivation time constants recorded from HEK293-hERG1a cells transfected with GFP (green circles), hERG1_NP_ (blue squares), R885C_NP_ (yellow upward triangles), R1035W_NP_ (aqua downward triangles), G1036D_NP_ (red diamonds), Q1068R_NP_ (purple circles), or R1069S_NP_ (gray squares). Data in B and C were compared using a nonparametric Kruskal–Wallis test with a post hoc Dunn's multiple comparisons correction. Comparisons included GFP versus hERG1_NP_ and all variants versus both GFP and hERG1_NP_. Error bars represent the mean ± SD. N ≥ 3. N ≥ 7. ∗∗∗∗ indicates p < 0.0001, ∗∗∗ indicates p < 0.0002, ∗∗ indicates p < 0.0021, and ∗ indicates p < 0.0332. See Table 2 for specific experimental values. hERG1, human Ether-à-go-go–Related Gene 1; hERG1_NP_, hERG1 nuclear peptide.Figure 5G1036D_NP_ and Q1068R_NP_ accelerate the time course of hERG1a inactivation recovery. A, sample IhERG traces recorded from HEK293-hERG1a cells transfected with GFP (green), hERG1_NP_ (blue), G1036D_NP_ (red), or Q1068R_NP_ (purple) elicited by the displayed voltage-step protocol and scaled as indicated. B, time constants of inactivation recovery recorded from HEK293-hERG1a cells transfected with GFP (green circles), hERG1_NP_ (blue squares), G1036D_NP_ (red upward triangles), or Q1068R_NP_ (purple downward triangles). Data in B were compared using a two-way ANOVA and post hoc Tukey's multiple comparisons test. Error bars represent the mean ± SD. N = 3. N ≥ 12. ∗∗∗∗ indicates p < 0.0001, ∗∗∗ indicates p < 0.0002, ∗∗ indicates p < 0.0021, and ∗ indicates p < 0.0332. HEK293, human embryonic kidney 293 cell line; hERG1, human Ether-à-go-go–Related Gene 1; hERG1_NP_, hERG1 nuclear peptide; IhERG, hERG1a current.

Results

hERG1NP reduces IhERG and hERG1a protein levels

The molecular mechanisms linking KCNH2 variants of the hERG1 C-terminal domain with sudden death are poorly understood. Here, we measured the impact of several KCNH2 variants associated with SIDS and/or SUDEP on hERG1_NP_ activity. We first validated our previous finding that hERG1_NP_ reduces IhERG amplitude in HEK293 cells stably expressing hERG1a (HEK293-hERG1a) (39). We transfected HEK293-hERG1a cells with constructs encoding either GFP or the distal C-terminal domain (hERG1_NP_) fused with mCitrine. As previously reported, hERG1_NP_ significantly reduced both steady-state and peak tail IhERG compared with currents recorded from GFP-transfected cells (Fig. 1, A–C).

The reduced hERG1a tail current in the presence of the polypeptide suggests that hERG1_NP_ reduces the total protein of the full-length hERG1a channel. To determine if hERG1_NP_ affected protein abundance, we measured hERG1a protein levels by Western blot in HEK293-hERG1a cells transfected with either GFP or hERG1_NP_. Both core-glycosylated and fully glycosylated hERG1a protein bands were visible at 135 kDa and 155 kDa, respectively (Fig. 1D). Total hERG1a protein, measured as the 135 kDa signal and 155 kDa signal combined, was significantly reduced in hERG1_NP_-transfected cells compared with GFP-transfected cells (Fig. 1E). Fully glycosylated hERG1a is associated with mature hERG1 channels localized to the surface membrane, whereas core-glycosylated hERG1 represents immature hERG1 channels within the protein trafficking pathway (40, 41, 42). To determine if hERG1_NP_ affected hERG1a trafficking, we measured the abundance of the mature and immature hERG1a bands independently. In the presence of hERG1_NP_, the mature hERG1a protein band showed a downward trend that was not statistically significant (p = 0.065, Fig. 1F), whereas immature hERG1a protein levels were significantly reduced (p = 0.004, Fig. 1G). Next, we assessed the impact of hERG1_NP_ on the relative abundance of mature and immature hERG1a. We measured relative mature hERG1a as mature/(mature + immature). The relative abundance of mature protein was not significantly different compared with GFP-transfected cells (p = 0.132, Fig. 1H), suggesting that both mature and immature hERG1 are downregulated by hERG1_NP_.

To determine if the reduced hERG1a protein by hERG1_NP_ was due to a change in mRNA levels, we measured hERG1a mRNA by quantitative RT–PCR in GFP and hERG1_NP_-transfected cells. Unlike hERG1a protein, hERG1_NP_ did not affect hERG1a transcript levels, compared with GFP controls (Fig. 1I). Collectively, the data in Figure 1 demonstrate that hERG1_NP_ reduces IhERG and total hERG1a protein without affecting hERG1a mRNA levels.

KCNH2 variants alter hERG1NP modulation of IhERG

While clinical studies have identified KCNH2 variants in cases of SIDS and SUDEP, many variants associated with sudden death do not significantly affect full-length hERG1 channel activity (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 37, 38). We hypothesized that variants of the distal C-terminal domain may instead selectively impact hERG1_NP_ activity. To determine if SIDS–SUDEP-associated variants of the distal C-terminal domain disrupt hERG1_NP_ activity, we transfected HEK293-hERG1a cells with hERG1_NP_-encoding constructs carrying one of the following KCNH2 variants: R885C, R1035W, G1036D, R1047L, Q1068R, and R1069S (hERG1a numbering). Variants R885C, G1036D, R1047L, and Q1068R were identified in cases of SIDS and/or SUDEP, but they each have minimal effects on the gating or expression of the full-length hERG1 channel (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 37, 38). R1035W and R1069S were identified in clinical cohorts but are not associated with cardiac disease (7, 43, 44, 45, 46, 47, 48). R1035W and R1069S, therefore, served as negative controls. Each variant is located within one of two hERG1_NP_ structural regions of interest. R885C resides in the hERG1_NP_ nuclear localization sequence (NLS), a canonical sequence motif responsible for tagging proteins for nuclear transport. The remaining five variants sit within the predicted coiled-coil structure (Fig. 2A) (33, 39).

Like the wildtype polypeptide, R885C_NP_, R1035W_NP_, G1036D_NP_, Q1068R_NP_, and R1069S_NP_ all significantly reduced steady-state and peak tail current compared with recordings from GFP-transfected controls (Fig. 2, B–E, G and H). In addition to reducing current amplitude, G1036D_NP_ and Q1068R_NP_ both depolarized the voltage dependence of IhERG activation by 10 mV, compared with both wildtype hERG1_NP_ and GFP-transfected cells (Table 1; Fig. 2, B, E and G). To highlight the effects of G1036D_NP_ and Q1068R_NP_ on the voltage dependence of activation, we normalized tail current amplitude to the maximum current recorded for each cell. We then plotted those data as a function of prepulse potential and fitted the data with a Boltzmann function (Equation 1; Fig. 3). These data demonstrate that G1036D_NP_ and Q1068R_NP_ have a gain-of-function effect on hERG1_NP_. In contrast, R1047L_NP_ had no effect on IhERG magnitude or voltage dependence (Fig. 2, B and F), suggesting that R1047L is a loss-of-function hERG1_NP_ variant. R1035W_NP_ and R1069S_NP_, which were not associated with sudden death, displayed hERG1a currents that were identical to currents recorded in the presence of wildtype hERG1_NP_. Thus, only variants associated with sudden death altered hERG1_NP_ modulation of IhERG: loss of function (R1047L_NP_) or gain of function (G1036D_NP_ and Q1068R_NP_).Table 1. Biophysical parameters of transfected HEK293-hERG1a cellsConstructPeak tail IhERG at +30 mV (pA/pF)Steady-state IhERG at 0 mV (pA/pF)V1/2 (mV)knGFP30.1 ± 13.0a16.8 ± 10.5a−13.9 (−14.9, −12.9)8.6 (7.8, 9.5)27hERG1_NP_18.7 ± 11.99.3 ± 6.0−14.8 (−16.5, −13.1)9.3 (7.9, 10.8)12R885C_NP_16.7 ± 10.39.1 ± 5.5−18.0 (−19.8, −16.2)9.4 (7.9, 11.2)8R1035W_NP_20.0 ± 6.712.2 ± 5.5−11.9 (−13.3, −10.4)8.5 (7.3, 9.8)7G1036D_NP_20.6 ± 11.410.2 ± 6.9−4.2 (−6.9, −1.4)a12.2 (9.9, 14.9)16R1047L_NP_27.0 ± 15.3a15.7 ± 10.2a−12.3 (−13.6, −11.0)9.1 (8.1, 10.3)20Q1068R_NP_18.8 ± 14.99.0 ± 3.3−4.0 (−6.0, −2.0)a9.4 (7.8, 11.2)15R1069S_NP_15.5 ± 5.011.0 ± 6.7−14.4 (−15.9, −12.9)8.3 (7.1, 9.6)8Two-way ANOVA and post hoc Dunnett’s multiple comparisons test with hERG1_NP_ set as the control group. Mean ± SD for peak tail and steady-state IhERG magnitudes. Mean (lower limit, upper limit) for V1/2 and slope factor (k) 95% confidence intervals as calculated by GraphPad Prism. N ≥ 3.aIndicates p < 0.05 compared with hERG1_NP_.

G1036DNP and Q1068RNP accelerate the time course of deactivation

Our previous work showed that wildtype hERG1_NP_ nuclear activity slows the time course of hERG1a deactivation (39). We hypothesized that currents recorded in the presence of G1036D_NP_ and Q1068R_NP_, variants that depolarized IhERG activation, would display accelerated deactivation compared with wildtype hERG1_NP_-transfected cells (Fig. 3, C and E). To test this hypothesis, we measured the time course of hERG1a deactivation by fitting current decay at −50 mV with a double exponential function (Equation 2). Consistent with previous work, currents recorded from wildtype hERG1_NP_-transfected cells exhibited significantly larger time constants for both the fast and slow components of deactivation, compared with GFP-transfected cells (Table 2; Fig. 4, A–C). Both G1036D_NP_ and Q1068R_NP_ displayed accelerated rates of current decay compared with wildtype hERG1_NP_ (Table 2; Fig. 4, A–C). The accelerated deactivation time constants from G1036D_NP_ and Q1068R_NP_ are consistent with the concomitant depolarization in hERG1a′s voltage dependence of activation (Fig. 3, C and E). R885C_NP_, R1035W_NP_, and R1069S_NP_, which suppressed IhERG magnitude but did not affect channel activation, displayed deactivation time constants that were comparable to wildtype hERG1_NP_ (Table 2; Fig. 4, B and C).Table 2. Deactivation parameters of transfected HEK293-hERG1a cellsConstructFast tau at +30 mV (ms)nSlow tau at +30 mV (ms)nGFP233.8 ± 50.3a121727 ± 280.2a12hERG1_NP_393.8 ± 65.1b112355 ± 387.2b10R885C_NP_334.2 ± 45.172481 ± 412.97R1035W_NP_313.2 ± 84.871867 ± 285.47G1036D_NP_226.1 ± 51.2a151423 ± 261.0a15Q1068R_NP_237.0 ± 98.2a151467 ± 490.2a15R1069S_NP_322.1 ± 86.281914 ± 701.88Nonparametric Kruskal–Wallis test with a post hoc Dunn's multiple comparisons correction. Comparisons included GFP versus hERG1_NP_ and all variants versus both GFP and hERG1_NP_. Mean ± SD. N ≥ 3.aIndicates p < 0.05 compared with hERG1_NP_.bIndicates p < 0.05 compared with GFP.

G1036DNP and Q1068RNP accelerate the time course of inactivation recovery

Having established that G1036D_NP_ and Q1068R_NP_ display altered time courses of deactivation, we tested if G1036D_NP_ and Q1068R_NP_ similarly affected hERG1a inactivation. We measured hERG1a inactivation recovery by first depolarizing the membrane with a 3-s conditioning pulse at 30 mV to fully activate and inactivate the channels. Cells were then stepped to a test pulse between −50 mV and −120 mV in 10 mV increments, and we fitted the rebound current with a single exponential function (Equation 2). Wildtype hERG1_NP_ minimally affected inactivation recovery (Fig. 5, A and B). Compared with GFP controls, hERG1_NP_ only accelerated inactivation recovery at −50 mV (Fig. 5B). In contrast, the kinetics of inactivation recovery in the presence of either G1036D_NP_ or Q1068R_NP_ were significantly faster than both GFP and wildtype hERG1_NP_, at −90 mV through −50 mV (Fig. 5, Aand B). The accelerated inactivation kinetics parallel the accelerated deactivation and depolarizing shifts in activation.

KCNH2 variants modulate hERG1NP nuclear trafficking

Having established that R1047L_NP_, G1036D_NP_, and Q1068R_NP_ alter hERG1_NP_ modulation of IhERG, we next tested the impact of all six variants on hERG1_NP_ nuclear localization. We previously demonstrated that hERG1_NP_ modulation of hERG1a current only occurred when hERG1_NP_ was targeted to the cell’s nucleus (39). We therefore predicted that R1047L_NP_, which does not modulate IhERG, would not traffic into the nucleus. The remaining variants were predicted to retain wildtype-like nuclear localization. We therefore transfected HEK293 cells stably expressing hERG1a with GFP, hERG1_NP_, or one of the six KCNH2 variant constructs.

First, to demonstrate that full-length hERG1a cDNA does not generate a nuclear polypeptide, we stained untransfected HEK293-hERG1a cells with an antibody that targets the hERG1 distal C-terminal domain. The epitope of this antibody is identical in both the full-length hERG1 channel and hERG1_NP_ (39). As predicted, the anti-hERG1a antibody showed robust immunofluorescence at the surface membrane and throughout the cytoplasm but was absent from the nucleus (Fig. 6, A and B). These data demonstrate that the hERG1a-encoding construct does not generate the hERG1_NP_ polypeptide in HEK293 cells.Figure 6KCNH2 variants alter hERG1_NP_ nuclear localization. A, sample images of HEK293 cells stably expressing hERG1a. Top left, the distal C-terminal domain of full-length hERG1a is immunolabeled (magenta) and only appears in the cytoplasm. The remaining images depict cells transfected with either GFP or hERG1_NP_-encoding constructs fused with mCitrine (green). For all images, DAPI delineates the nuclei (blue). The scale bars indicate 20 μm. B, quantification of nuclear fluorescence intensity relative to cytoplasmic fluorescence intensity (FNucleus/FCytoplasm) for cells depicted in A. Data in B were compared using a nonparametric Kruskal–Wallis test with a post hoc Dunn's multiple comparisons correction and hERG1_NP_ set as the control group. Black bars represent the median. N = 3. N ≥ 15. ∗∗∗∗ indicates p < 0.0001, ∗∗∗ indicates p < 0.0002, ∗∗ indicates p < 0.0021, and ∗ indicates p < 0.0332. DAPI, 4′,6-diamidino-2-phenylindole; HEK293, human embryonic kidney 293 cell line; hERG1, human Ether-à-go-go–Related Gene 1; hERG1_NP_, hERG1 nuclear peptide.

We next measured the relative distribution of GFP and hERG1_NP_. Consistent with previous reports, GFP was expressed throughout both the cytoplasm and nucleus (Fig. 6, A and B). In contrast, wildtype hERG1_NP_ targeted almost exclusively to the nucleus and displayed significantly greater nuclear targeting compared with GFP-transfected cells (Fig. 6, A and B). These data demonstrate that nuclear targeting of the fluorophore is dependent upon the hERG1_NP_ polypeptide and are consistent with our previous publication (39).

Finally, we quantified the intracellular distribution of each of the six KCNH2 variant constructs. R1047L_NP_, which had no effect on IhERG, displayed almost exclusive cytoplasmic targeting. Surprisingly, R885C_NP_ displayed a bimodal distribution, with a roughly 50:50 split of nuclear-targeted and cytoplasm-targeted cells (Fig. 6, A and B). R1047L_NP_’s loss of current suppression reaffirms that nuclear localization is critical for hERG1_NP_ modulation of hERG1 current. R885C_NP_‘s bimodal distribution suggests that direct disruption of the hERG1_NP_ NLS causes a loss of nuclear targeting and is therefore a loss-of-function variant. The remaining constructs, R1035W_NP_, G1036D_NP_, Q1068R_NP_, and R1069S_NP_, did not modify IhERG suppression and were targeted to the nucleus at levels comparable to wildtype hERG1_NP_ (Fig. 6, A and B). These results demonstrate that a near total loss of nuclear targeting may be required to alter hERG1_NP_’s suppression of IhERG, whereas a partial loss of nuclear targeting is not sufficient to modify hERG1_NP_'s IhERG suppression.

Discussion

The present study demonstrates that SIDS-associated and SUDEP-associated KCNH2 variants exert distinct effects on the nuclear-targeted hERG1_NP_ polypeptide. Consistent with our previous findings, wildtype hERG1_NP_ localizes to the nucleus and suppresses IhERG (39), an effect associated with reduced protein levels but unchanged transcript abundance. Here, we extend these observations by showing that variants produce both gain-of-function and loss-of-function effects on hERG1_NP_ activity: G1036D_NP_ and Q1068R_NP_ enhanced gating modulation, whereas R885C_NP_ and R1047L_NP_ impaired nuclear targeting and/or current suppression. Together, these data provide evidence that hERG1_NP_ dysfunction may represent a novel pathogenic mechanism linking KCNH2 variants to SIDS and SUDEP.

KCNH2 variant identification

The variants reported in this study were chosen to identify a potential link between hERG1_NP_ dysfunction and sudden death. R885C, which maps to the NLS of hERG1_NP_, was identified in multiple cases of SIDS but does not affect the biophysical properties of full-length hERG1 (5, 6, 12). The remaining variants sit within a predicted coiled-coil domain (33). R1035W was identified in several LQT patient cohorts and a nonlethal case of idiopathic ventricular fibrillation, yet R1035W has no known link with sudden death (7, 43, 44, 45). G1036D was identified in several LQT study cohorts (8, 43, 44, 49) and a case of resuscitation after cardiac arrest. In addition, two groups found that it causes a small reduction in full-length hERG1 current, but G1036D is not predicted to be pathogenic (8, 37). Like G1036D, R1047L was found in multiple LQT study cohorts (43, 44, 45) and is linked to cases of SIDS (5, 9, 11, 15, 16), SUDEP (10), and drug-induced Torsades de pointes (38). R1047L is also present in a case of cardiac arrest resuscitation for a patient with a family history of sudden death and a separate case of posthumous sudden unexplained death in the young (13, 14). The effects of R1047L on full-length hERG1 and its potential as a disease-causing variant are conflicting. In some hands, R1047L diminishes hERG1 current magnitude (10); in others, it modifies hERG1a voltage dependence but not current magnitude (38), or the variant carries no phenotype (16, 37). Bioinformatic models used for in silico analysis of R1047L are similarly inconclusive (9). Q1068R was present in multiple LQTS study cohorts (43, 44). It was identified in multiple cases of SIDS, and functional evaluations in full-length hERG1a display small acceleration of inactivation and inactivation recovery with no change to current magnitude (11, 15, 16, 37). Last, R1069S is a clinical variant with no association to sudden death (46, 47, 48). In this study, we only observed changes in hERG1_NP_ activity in variants associated with sudden death, supporting the conclusion that sudden death–associated hERG1_NP_ variants alter hERG1_NP_ activity and may represent a novel mechanism of KCNH2-related pathophysiology.

Gain-of-function KCNH2 variants suggest a mechanism of hERG1a modulation

The regulatory pathways linking hERG1_NP_ nuclear activity with altered hERG1a activity remain unknown. But gain-of-function variants, G1036D_NP_ and Q1068R_NP_, provide a possible insight into one mechanism by which hERG1_NP_ may regulate the full-length channel. Both G1036D_NP_ and Q1068R_NP_ depolarized hERG1a′s voltage dependence of activation while accelerating the time course of deactivation and inactivation recovery. These effects were in addition to the reduced current amplitude seen by the wildtype polypeptide. These effects on hERG1a channel gating are strikingly similar to hERG1a modulation by PKA, and to a lesser extent, PKC.

PKA modulates hERG1 current during acute (within minutes) and chronic (>24 h) stimulation. Multiple groups have reported that PKA phosphorylation of hERG1a initially reduces hERG1a current amplitude, accelerates deactivation, and depolarizes the voltage dependence of activation (50, 51). These effects are blocked by mutating all four of hERG1’s known PKA phosphorylation sites. Sustained PKA stimulation, using either forskolin or chlorophenyl thiol-cAMP, increases hERG1 channel synthesis, leading to increased current (52, 53). The effects of PKA on hERG1 voltage dependence are lost during prolonged PKA stimulation. PKC activity similarly depolarizes the voltage dependence of hERG1 activation and reduces current amplitude (54). Interestingly, PKC activity indirectly controls channel surface expression (52, 53), whereas PKA directly phosphorylates hERG1 (55, 56, 57). In the case of PKC, activation leads to increased phosphorylation of Nedd4-2, the E3 ubiquitin ligase responsible for hERG1 degradation (58, 59). Nedd4-2 phosphorylation inhibits ligase activity, thereby reducing hERG1 degradation (50, 59, 60). It is possible that hERG1_NP_ nuclear activity stimulates PKA and/or PKC to regulate full-length hERG1.

Loss-of-function KCNH2 variants as indicators of hERG1NP structure and function

R885C resides within hERG1_NP_’s NLS (39) and disrupts but does not abolish hERG1_NP_ nuclear targeting. We previously identified the karyopherin-α–β1 complex as the mediator of hERG1_NP_ trafficking (39). The results herein suggest that replacing arginine with cysteine perturbs the binding and/or recruitment of karyopherin-α, without being so disruptive as to completely nullify nuclear localization. Indeed, screening the mutated R885C_NP_ NLS with the open source software “cNLS mapper” predicts reduced but not abolished nuclear activity (61). Crystallography studies have demonstrated that positively charged lysines and arginines within any NLS form hydrogen bonds and salt bridges with the hydrophilic residues that line the karyopherin-α binding pocket (62, 63, 64, 65, 66, 67, 68). Therefore, replacing any one of the six arginines and lysines within the hERG1_NP_ NLS would be predicted to diminish nuclear localization. R885C does not fully disrupt nuclear localization because five additional basic residues are still present within the R885C_NP_ NLS. Interestingly, despite disrupted nuclear transport, R885C_NP_ continued to suppress IhERG. This is likely because of the still substantial nuclear-localized R885C_NP_. We did not directly track the relationship between IhERG density and magnitude of R885C_NP_ nuclear versus cytoplasmic fluorescence for individual cells. However, one might expect a bimodal distribution of IhERG density that mimics the nuclear to cytoplasmic distribution of R885C_NP,_ with high nuclear localization resulting in less current and vice versa.

R1047L resides in the hERG1_NP_ coiled-coil domain and abolishes hERG1_NP_ nuclear targeting and subsequent suppression of IhERG. The four other coiled-coil variants, R1035W, G1036D, Q1068R, and R1069S, did not affect nuclear transport. Coiled-coil domains facilitate protein–protein and protein–DNA interactions to serve an array of cellular functions (69). Many proteins form coil-mediated dimers, rimmers, and tetramers to undergo nuclear trafficking, making it reasonable to hypothesize that the hERG1_NP_’s coiled-coil could play a key role in maintaining the structural integrity of a hERG1_NP_ multipolypeptide formation (70, 71, 72). R1047L replaces a polar and positively charged residue with a hydrophobic residue, which could destabilize coiled-coil assembly. However, a similarly substantial biochemical shift is observed for several additional variants that do not modify nuclear targeting: R1035W, a positive charge to a nonpolar aromatic; G1036D, a nonpolar aliphatic to a negative charge; and R1069S, a bulky positive charge to a smaller uncharged side chain. These results suggest that the location of the variant within the coiled-coil may also play a role in its effect on activity.

The hERG1NP coiled-coil predicts a hERG1NP tetramer

Canonical coiled-coil sequences follow a heptad repeat sequence of abcdefg, where a and d are hydrophobic residues. These residues are spaced by polar residues at positions b, c, e, f, and g, with e and g often carrying charges (Fig. 7A) (73). As our understanding of protein structures has improved, groups have found success identifying specific residue interactions and motifs within coiled-coils that predict or trigger the higher order structural packaging of a protein (73, 74, 75, 76). The hERG1_NP_’s predicted coiled-coil sequence follows the heptad repeat motif with hydrophobic-rich a and d positions and charged e and g positions (Fig. 7A). For both dimeric and trimeric packaging, short coiled-coils such as the hERG1_NP_’s often carry a centrally located R-h-x-x-h-E sequence motif with an arginine residue flanked by a hydrophobic residue denoted h, two filler residues denoted x, another hydrophobic residue, and glutamic acid (74). Interestingly, this motif is selective for dimeric and trimeric assemblies but not tetrameric assemblies. However, mutation of the lead arginine to glutamine imparts tetramer-specific structural packaging, and such a Q-h-x-x-h-E motif is present within the hERG1_NP_ from residues 1048 to 1053 (sequence: Q-L-N-R-L-E) (Fig. 7B), suggestive of tetrameric peptide packaging (77). The trimeric R-h-x-x-h-E motif is thought to impart structural integrity to coil interactions via key salt bridges (77). As a result, it is plausible that the hERG1_NP_’s Q-L-N-R-L-E sequence plays a similar role in tetramer stability needed for trafficking (Fig. 7B). Variant R1047L directly precedes this motif (sequence: L-Q-L-N-R-L-E), whereas all four other coil variants are substantially proximal or distal. Therefore, R1047L may disrupt the tetramer-selective sequence’s interactions or spatial orientation, resulting in a loss of R1047L_NP_ nuclear trafficking.Figure 7hERG1_NP_’s predicted coiled-coil carries a potential tetramer-selective motif. A, individual heptad repeat, labeled a–g, to highlight the basic amino acid patterning for a coiled-coil. B, side view (left) of an individual hERG1_NP_ coil (magenta) from residues R1035 to L1073 (hERG1a numbering). hERG1_NP_’s hypothesized tetramer-selective motif is highlighted in cyan. Zoomed in side view (right) of the tetramer-selective Q-h-x-x-x-h-E motif residues (cyan) with hERG1_NP_-specific labels and displayed side chains. All coiled-coil structural images were created using AlphaFold2 (121) and ChimeraX (122, 123, 124). hERG1, human Ether-à-go-go–Related Gene 1; hERG1_NP_, hERG1 nuclear peptide.

IhERG as an indicator of hERG1NP activity

In our initial description of hERG1_NP_, we reported that the polypeptide stabilized the hERG1a-inactivated state in addition to reducing hERG1a tail current (39). In the current study, hERG1_NP_ reduced hERG1a tail current but did not affect inactivation. The discrepancy between our first study and this one highlights a limitation of using IhERG as an indicator of hERG1_NP_ activity. In cardiomyocytes, hERG1_NP_ nuclear targeting is upregulated in immature cells (39). This suggests that hERG1_NP_ contributes to developmental pathways and that its activity is dependent upon the physiological state of the cell in which it is expressed. Given the all too common but often unacknowledged variability of HEK293 cells, this may explain the differences between the current study and our previous work. Because hERG1_NP_ is working from within the nucleus, the downstream targets of its activity are likely dynamic and widespread. Nuclear peptides, including those derived from ion channels, are known to broadly affect transcriptional networks, chromatin remodeling, and protein assembly (78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108). We therefore recognize that it is likely that hERG1_NP_ mediates cellular functions undetectable through IhERG specifically. Follow-up studies identifying specific gene targets of hERG1_NP_ may help to identify more direct measures of hERG1_NP_ function and may provide a more reliable prediction of the potential pathophysiological impact of hERG1_NP_ variants.

Ion channel subdomains in the nucleus

R1047L_NP_ is the only hERG1_NP_ construct that failed to traffic to the nucleus. It was also the only hERG1_NP_ construct that did not modify hERG1 current, indicating that hERG1_NP_ regulation of IhERG is nuclear dependent. This was previously demonstrated using a hERG1_NP_ construct, where the NLS was removed by site-directed mutagenesis (39). Our data do not delineate the pathway linking hERG1_NP_ nuclear activity with altered ion channel expression and gating. However, there is precedent for ion channel subdomains serving as regulators of gene expression. The C-terminal domain of the voltage-gated calcium channel, Ca_v_1.3, translocates to the nucleus of atrial myocytes, where it acts as a transcription factor (97). Similarly, CACNA1C generates a nuclear-targeted polypeptide, CCAT, that is homologous to the C-terminal domain of Ca_v_1.2 and directly modulates the expression of various genes, including Ca_v_1.2 (86, 96). The calcium channel subunits, Ca_v_β1–4, are capable of localizing to the nucleus, where they regulate transcription and development (99, 100, 101, 102, 103, 104, 105, 106). The sodium channel β1 subunit generates an intracellular domain that acts as a “molecular brake” on gene transcription related to SCN1B-linked channelopathies (107, 108). Last, the C-terminal domain of TRPM7 is proteolytically cleaved and translocated to the nucleus, where it mediates chromatin remodeling and gene expression (98). These studies collectively suggest that regulation of gene expression by ion channel subdomains may be an overlooked mechanism of intracellular feedback. Variants that map to these polypeptides may thus disrupt these potentially important feedback loops, impairing a cell’s response to external stimuli.

Future research directions

To inform the future identification of a mechanism for hERG1_NP_ activity, additional work is needed using more quantitative techniques that directly link hERG1_NP_ to reduced hERG1a protein. More broadly, our findings build a foundation for the need to better understand the direct physiological and translational ramifications of hERG1_NP_ function. With the knowledge that hERG1_NP_ is highly expressed in immature cardiac cells, the innate immaturity of stem cell–derived cardiomyocytes may act as a key tool to determine the mechanisms and targets of hERG1_NP_ activity. Gene-edited stem cells also provide the opportunity to investigate hERG1_NP_ expression beyond the heart, as KCNH2-encoded channels are expressed throughout the human body (109, 110, 111, 112, 113). The concomitant loss of nuclear targeting and subsequent suppression of IhERG by R1047L_NP_ reported herein is consistent with our previous work that hERG1_NP_ activity is nuclear dependent. Native Western blots and coimmunoprecipitation assays using stem cell–derived and/or native cardiac tissue could identify the nuclear proteins that interact with hERG1_NP_ or if the peptide undergoes post-translational modification. Last, hERG1_NP_ may regulate other ion channels in a similar fashion to hERG1a, making investigation of other currents in the presence of hERG1_NP_ a logical next step. These experiments should provide a more direct understanding of hERG1_NP_’s mechanism(s) of action and targets so that we may ultimately investigate the consequences of disease-associated hERG1_NP_ variants on pathophysiology.

Conclusion

The experimental data presented herein demonstrate that KCNH2 variants associated with sudden death in the young can modify hERG1_NP_ activity. While the mechanism through which hERG1_NP_ engages with the hERG1a channel remains unknown, these findings provide support for the hypothesis that hERG1_NP_ dysfunction causes human disease. hERG1_NP_ has at least one physiological output seen through hERG1a, and it is reasonable to hypothesize that there is potential for hERG1_NP_ to engage with other channels or protein families across tissues or cellular compartments that require additional study.

Limitations

Here, we report findings through experiments exclusively carried out in a line of HEK293 cells stably overexpressing the hERG1a ion channel. While our study demonstrates varied effects of KCNH2 variants on hERG1_NP_ function, HEK293 cells are an intrinsically simplified model system that cannot replicate the complex pathways and protein dynamics that occur in specialized cell types that are known to express hERG1 protein, such as cardiomyocytes or neurons. Furthermore, overexpression of hERG1_NP_ and/or hERG1a could introduce interactions not found at physiological expression levels or could exacerbate the physiological effects of hERG1_NP_ on cellular physiology.

Our system also only evaluates hERG1a homomeric channels. Native hERG1 channels are heterotetramers composed of up to three distinct subunits: hERG1a, hERG1b, and hERG1c (17, 18, 19, 21, 114, 115, 116, 117, 118, 119, 120). hERG1a contains an N-terminal PAS domain and a C-terminal CNBHD (22, 30, 31). Channel gating is strictly regulated by direct interactions between the PAS and the CNBHD. hERG1b and hERG1c, however, carry unique N termini that are shorter than hERG1a and lack functional PAS domains (17, 18, 114, 117, 118, 119, 120). Of these three subunits, hERG1a is the most well characterized and is the only one with intact PAS to CNBHD and PAS to S4–S5 linker interactions. While we chose to focus on hERG1a alone to simplify our early exploration of hERG1_NP_, the complex dynamics of different hERG1 subunits on trafficking and current introduce additional relationships to be explored in the future. Nonetheless, these data demonstrate exciting insights into the functional role and physiological relevance of hERG1_NP_ and act as an important foundation to build upon with future studies.

Data availability

All data generated or analyzed in this study are included in this article and its supporting information.

Conflict of interest

The authors declare that they have no conflicts of interest with the contents of this article.