Long QT Syndrome 2 PAS Domain Variant Induces hERG1a/1b Subunit Imbalance in Patient-specific iPSC-cardiomyocytes

Abstract

Background -

Inherited long QT syndrome type 2 (LQT2) results from variants in the KCNH2 gene encoding the hERG1 potassium channel. Two main isoforms, hERG1a and hERG1b, assemble to form tetrameric channel. The N-terminal Per-Arnt-Sim (PAS) domain, present only on hERG1a subunits, is a hotspot for pathogenic variants, but it is unknown whether PAS domain variants impact hERG1b expression to contribute to the LQT2 phenotype. We aimed to use patient-specific induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) to investigate the pathogenesis of the hERG1a PAS domain variant hERG1-H70R.

Methods -

Human iPSCs were derived from a LQT2 patient carrying the PAS domain variant hERG1-H70R. CRISPR/Cas9 gene editing produced isogenic control iPSC lines. Differentiated iPSC-CMs were evaluated for their electrophysiology, hERG1a/1b mRNA expression, and hERG1a/1b protein expression.

Results -

Action potentials from single hERG1-H70R iPSC-CMs were prolonged relative to controls, and voltage clamp studies showed an underlying decrease in I_Kr with accelerated deactivation. In hERG1-H70R iPSC-CMs, transcription of hERG1a and hERG1b mRNA was unchanged compared to controls based on nascent nuclear transcript analysis, but hERG1b mRNA was significantly increased as was the ratio of hERG1b/hERG1a in mRNA complexes, suggesting post-transcriptional changes. Expression of complex glycosylated hERG1a in hERG1-H70R iPSC-CMs was reduced due to impaired protein trafficking, whereas the expression of the complex glycosylated form of hERG1b was unchanged.

Conclusions -

Patient-specific hERG1-H70R iPSC-CMs reveal a newly appreciated mechanism of pathogenesis of the LQT2 phenotype due to both impaired trafficking of hERG1a and maintained expression of hERG1b that produces subunit imbalance and reduced I_Kr with accelerated deactivation.

Introduction

Inherited long QT syndrome (LQTS) is characterized by delayed myocardial repolarization resulting in a prolonged QT interval on the ECG with an increased propensity for the ventricular arrhythmia torsades de pointes (TdP) and sudden cardiac death (SCD).¹ Variants in the KCNH2 gene, also known as human Ether-à-go-go Related Gene (hERG), cause LQTS type 2 (LQT2) and account for 35–40% of the diagnosed cases of LQTS.² In human heart, the two main isoforms of hERG1, hERG1a and hERG1b, form the tetrameric voltage-activated potassium channel responsible for the rapidly activating delayed rectifier potassium current, I_Kr.^3–5 hERG1a and hERG1b differ only in the N-terminus with hERG1b having a short 36 amino acids N-terminus which lacks the Per-Arnt-Sim (PAS) domain present in hERG1a (Figure 1A,B). Previous studies proved that hERG1a and hERG1b associate in native myocardium and form heteromeric channels.^{6, 7} In heterologous expression systems, hERG1a/1b heteromeric channels exhibit function distinct from either hERG1a or hERG1b homomeric channels. hERG1a/1b heteromers have larger currents, activate and recover from inactivation faster than hERG1a homomers, and exhibit accelerated deactivation kinetics.^{8, 9} hERG1b homomeric channels generate very small amplitude currents with the fastest deactivation. Both hERG1a-specific and hERG1b-specific missense variants can reduce I_Kr and cause LQT2,^{2, 10} indicating that hERG1a and hERG1b are both critically important molecular constituents of cardiac I_Kr in human heart.

Figure 1.

hERG1-H70R patient-specific iPSCs. A. Schematic representation of hERG1a and hERG1b isoforms with the H70R variant located in the N-terminal PAS domain of hERG1a. B. Schematic of genomic structure of KCNH2 encoding hERG1a and hERG1b with sequence analysis of PCR-amplified genomic DNA from iPSCs (hERG1-H70R, hERG1-H70R^corr, and DF 19–9-11 T). The heterozygous missense variant c.209 A>G is detected in exon 2 of KCNH2 gene (c.209A>G, NM_000238.3), resulting in an H70R substitution only in the hERG1-H70R iPSCs line. C. Electrocardiogram of the hERG1-H70R LQT2 patient showing prolonged QTc interval (507 ms). D. Ventricular arrhythmia torsades de pointes recorded by ICD from hERG1-H70R patient. E. Immunofluorescence analysis of pluripotency markers SSEA4 (red), OCT4 (green) and NANOG (magenta) in a representative hERG1-H70R clone and hERG1-H70R^corr, with nuclear staining (DNA, blue). Scale bars: 100 μm.

More than 700 rare sequence variants, mostly missense, in KCNH2 have been associated with LQT2 (ClinVar database). The pathogenic variants in KCNH2 cause loss of function of hERG1 leading to reduced I_Kr and prolonged cardiac repolarization with an autosomal dominant inheritance pattern. Extensive research over more than two decades using primarily heterologous expression systems has described multiple mechanisms by which pathogenic variants can lead to loss of function of hERG1. Class 1 variants disrupt the synthesis/translation; Class 2 variants reduce the protein trafficking (intracellular transport); Class 3 variants alter channel gating; and Class 4 variants interfere with permeation.¹¹ Most heterologous expression studies have typically expressed solely the hERG1a subunit with the variant under study in HEK293 cells, CHO cells or Xenopus laevis oocytes. Therefore, the functional impact of variants in heteromeric hERG1a/1b channels has rarely been assessed, and even less is known about the impact of variants on the expression of hERG1a and hERG1b in human cardiomyocytes (CMs).

Patient-specific human induced pluripotent stem cell-derived cardiomyocytes (iPSC-CMs) provide a new model system for investigating inherited arrhythmia syndromes including LQTS that overcome some of the limitations of heterologous expression systems.¹² For example, hERG1a and 1b mRNAs are co-translated to form heteromeric channels in iPSC-CMs,¹³ enabling characterization of variants in human cardiomyocytes with the full complement of subunits and associated proteins like native heart. At least eight different LQT2-associated missense variants in KCNH2 have been characterized using the iPSC-CM model and demonstrated increased action potential duration (APD) with decreased I_Kr current.^14–21 Furthermore, gene editing in iPSC lines has been used to correct some variants in KCNH2 to provide fully isogenic control lines enabling robust assessment of the functional consequence of the single sequence variant under study.^{17, 20}

Here, we use the iPSC-CM model to investigate the mechanism by which hERG1 channel PAS domain LQT2 pathogenic variants, which are present only in hERG1a, cause disease. The N-terminal PAS domain is a “hot spot” for LQT2 pathogenic variants and contains about 30% of all LQT2 mutations.¹¹ We studied the PAS domain LQT2 variant hERG1-H70R. The evidence for the pathogenicity of the H70R variant is strong with it present in multiple unrelated patients with LQT2 and absent in large control genome/exome datasets such as the 141,456 sequences in gnomAD v2.1.1 along with functional evidence of pathogenicity from multiple heterologous expression studies. Nevertheless, the precise mechanism of pathogenicity of this variant is unclear given conflicting evidence from heterologous expression studies. For example, the H70R variant accelerated deactivation of the channels when expressed in Xenopus laevis oocytes but not in HEK293 cells.^{22, 23} hERG1-H70R is also variably associated with defective protein trafficking.^23–25 Because hERG1-H70R, like all PAS domain variants, is expressed only on hERG1a, it raises questions about secondary effects on hERG1b expression and the overall composition of channel tetramers in native heart. Therefore, we generated iPSCs from a LQT2 patient with the c.209A>G heterozygous missense variant resulting hERG1-H70R and used the clustered regularly interspaced short palindromic repeat (CRISPR)/ CRISPR-associated protein 9 (Cas9) gene editing to correct the variant and generate isogenic control iPSC line (hERG1-H70R^corr). The differentiated iPSC-CMs from the hERG1-H70R variant compared to the gene corrected line showed a prolonged APD associated with not only a reduction in I_Kr amplitude but also faster deactivation. Analysis of mRNA and protein expression show no change in basal transcription of hERG1a and hERG1b in hERG1-H70R iPSC-CMs but a relative increase in hERG1b mRNA level togetherwith impaired hERG1a trafficking in the hERG1-H70R lines. Our data support a model in which impaired trafficking of the hERG1a-H70R subunit with maintained hERG1b levels leads to subunit imbalance resulting in reduced I_Kr with accelerated deactivation leading to action potential and QT prolongation.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Human iPSC generation, characterization and maintenance

Somatic reprogramming was used to generate iPSC lines from skin fibroblasts from the patient with c.209A>G KCNH2 H70R by expression of the reprogramming factors SOX2, OCT4, NANOG, LIN28, KLF4, SV40LT and c-MYC, using nonintegrating episomal vectors.²⁶ Written informed consent was obtained from the patient in accordance with the last version of the Declaration of Helsinki and with approval by the University of Wisconsin Health Sciences institutional review board. Three iPSC clones were generated, karyotyped, characterized and cryopreserved (Cellular Dynamics International; Madison, WI). We further confirmed the pluripotency of the iPSCs by immunolabeling with the pluripotency markers of OCT4, NANOG and SSEA4. Human iPSC lines derived from foreskin fibroblasts without integration of vector and transgene sequences (DF19–9-11T) were used as a healthy unrelated control.²⁶ iPSCs were maintained on Matrigel (GFR, BD Biosciences) in mTeSR1 medium (WiCell) and passaged every 4–5 days.

CRISPR/Cas9 gene correction of hERG1-H70RiPSCs

The CRISPR/Cas9 system was used to perform genome editing in hERG1-H70R iPSCs and modify c.209A>G KCNH2 to a wild type sequence by single-strand oligonucleotide (ssODN) method (Waisman Center, Human Stem Cell Core, University of Wisconsin-Madison). Detailed methods in supplementary material.

Human iPSC-CM differentiation

The iPSCs from hERG1-H70R, hERG1-H70R^corr and DF19–9-11T were differentiated into iPSC-CMs using a monolayer-based, small molecule protocol modified from a previously published method and as described in detain in supplemental material. ^{27, 28}

Whole-cell patch -clamp of iPSC-CMs

The iPSC-CMs were dissociated after 30 days of differentiation, plated, and studied 10–20 days after replating. Recordings were performed on single iPSC-CMs, with the ruptured whole cell patch-clamp technique at 36 ± 1°C. Data were collected from at least two or three independent differentiations per line. I_Kr and APs were recorded with voltage-clamp and current-clamp separately using an Axopatch 200B amplifier (Molecular Devices). Voltage control and data acquisition of I_Kr and APs were performed with pClamp10.1 (Axon Instruments), while analysis of was performed with Origin, v7.5. Detailed methods are in supplementary material.

hERG1 single-molecule RNA Fluorescence In Situ Hybridization (smFISH) and image analysis

The smFISH procedure was modified from the method previously published.^{29, 30} Briefly, Custom Stellaris smFISH probes (Biosearch Technologies, Inc., Petaluma, CA) were designed against the exons of KCNH2 utilizing the smFISH Probe Designer available at www.biosearchtech.com/stellaris designer. The hERG1a-specific probe set (PSa) includes 48 unique oligonucleotides labeled with TAMRA. The hERG1a and hERG1b probe set (PSab) include 48 unique oligonucleotides tagged with Quasar 670. iPSC-CMs or HEK293 were fixed on the coverslip and incubate with smFISH probes at 37°C. The smFISH image were captured by Leica TCS SP5 confocal laser scanning microscope with hypersensitive detectors (HyDs). All analyses were done in MATLAB 2019b using custom MATLAB codes (available at www.github.com/chleelab/hERGsmFISH). Detailed methods are in supplementary material.

Statistical analysis

Data are presented as mean ± standard error of the mean (S.E.M.) or mean ± standard deviation (S.D.) as indicated. Statistical significance was determined by one-way ANOVA for three groups with post-hoc test using Tukey method. Statistical analysis was performed using Microcal Origin, v7.5, P < 0.05 was considered statistically significant.

Results

H70R KCNH2 LQT2 patient

Dermal fibroblasts were collected from a 29 year-old Caucasian female who suffered a syncopal event with seizures at the age of 19 years. She was found to have normal cardiac structure and function by echocardiography, but a prolonged QTc (Bazett) of 507 ms was present on her ECG (Figure 1C). Genetic testing identified c.209A>G heterozygous missense variant in KCNH2 resulting in an H70R substitution in the PAS domain of hERG1a (Figure 1B), and the patient was diagnosed with LQT2. She has been managed with an implantable cardioverter-defibrillator (ICD), and subsequently experienced exercise-triggered recurrent syncope and near syncope associated with ICD shocks. Interrogation of the ICD revealed polymorphic ventricular tachycardia (Figure 1D). She has been managed with 60 mg of nadolol daily and now remains free of syncope and ICD shocks including two recent successful pregnancies with uneventful post-partum periods.

Characterization of hERG1-H70R and hERG1-H70R^corr iPSC lines

Patient-specific iPSCs were generated by somatic reprogramming of skin fibroblasts with episomal vectors encoding SOX2, OCT4, NANOG, LIN28, KLF4, SV40LT, c-MYC as previously described²⁶. The iPSC clones displayed normal karyotypes (Supplemental Figure IA,B), and 2 clones were randomly chosen for this study. Sequence analysis of genomic DNA verified that both iPSC clones from the LQT2 patient carried the heterozygous missense variant c.209 A>G in exon 2 of KCNH2 (c.209A>G, NM_000238.3) (Figure 1B, Supplemental Figure IIA). Subsequently, genome editing was performed using CRISPR/Cas9 on hERG1-H70R iPSCs to correct the c.209 A>G variant as demonstrated by genomic DNA sequence analysis (Figure 1B). The genomic corrected line (hERG1-H70R^corr) maintained a normal karyotype (Supplemental Figure IC). The generated iPSC lines and the isogenic control line express pluripotency markers of OCT4, SSEA4 and NANOG demonstrated by immunofluorescence (Figure 1E,Supplemental Figure IIB). The previously described iPSC line DF19–9-11T²⁶ was used as an unrelated iPSC control line in this study and verified as not carrying c.209 A>G variant in KCNH2 gene (Figure 1B).

hERG1-H70R iPSC-CMs exhibit prolonged action potential duration

We differentiated the hERG1-H70R, hERG1-H70R^corr and DF19–9-11T iPSCs to iPSC-CMs using a monolayer-based small molecule protocol as previously described.^{27, 28} Eight to twelve days after the induction, spontaneous contracting iPSC-CMs were observed. The iPSC-CMs after 30 days differentiation were dissociated into single cells, plated on glass coverslips and characterized by immunofluorescence labeling for the cardiac-specific myofilament markers, cardiac Troponin T (cTnT) and sarcomeric α-actinin (Figure 2A, Supplemental Figure IIC). No differences were detected in the expression of cardiac myofilament proteins or overall cell morphology between iPSC-CMs from hERG1-H70R, hERG1-H70R^corr and DF19–9-11T.

Figure 2.

Characterization of iPSC-CMs from hERG1-H70R, hERG1-H70R^corr and DF19–9-11T. A. Immunofluorescence images of hERG1-H70R, hERG1-H70R^corr, DF19–9-11T iPSC-CMs immunolabeled with antibodies to α-actinin (red) and cardiac troponin (cTnT, green). Scale bars: 100 μm. B. Representative action potentials from single iPSC-CM paced at 1Hz (Temp 36 ± 1°C). C. D. Action potential parameters for action potential amplitude (APA), maximum diastolic potential (MDP), action potential duration at 30, 50, 70, 80, and 90% of repolarization (APD30, APD50, APD70, APD80 and APD90) for individual cells from the three groups as indicated; mean (thick bar) ± S.D. (thin bars); * p<0.05 for indicated comparisons.

We performed current-clamp experiments to record action potentials from single spontaneous beating iPSC-CMs. Ventricular, atrial and pacemaker types of APs were all observed in all the groups based on previously described criteria.³¹ Ventricular-like action potentials paced at 1Hz were chosen for detailed comparison between groups (Supplemental Table I). hERG1-H70R iPSC-CMs demonstrated a significantly prolonged APD80 and APD90 compared to hERG1-H70R^corr and DF19–9-11T consistent with the LQT2 phenotype (Figure 2B,D). The hERG1-H70R iPSC-CMs also exhibited a more depolarized maximum diastolic potential (MDP) relative to hERG1-H70R^corr with reduced action potential amplitude (APA) (Figure 2C).

hERG1-H70R iPSC-CMs show decreased I_Kr with altered gating

To evaluate the impact of the hERG1-H70R variant on I_Kr, we performed voltage-clamp recordings of single iPSC-CMs. I_Kr is defined as the E-4031-sensitive current (Figure 3A). Average peak I_Kr current density was significantly less in hERG1-H70R iPSC-CMs compared with hERG1-H70R^corr and DF19–9-11T iPSC-CMs at test potentials of 0mV to +20mV (Figure 3B). In addition, the peak-tail I_Kr density upon repolarization was significantly reduced in hERG1-H70R iPSC-CMs following depolarizations to 0 mV to +20 mV (Figure 3C). Furthermore, we analyzed the voltage-dependence of I_Kr activation, and I_Kr activated at more negative voltages in the hERG1-H70R iPSC-CMs than the hERG1-H70R^corr and DF19–9-11T iPSC-CMs (Figure 3D). Evaluation of deactivation of I_Kr revealed that the fast (τ_fast) and slow (τ_slow) deactivation time constants were significantly decreased in hERG1-H70R iPSC-CMs and the contribution of fast deactivation component for tail current was significantly increased (Figure 3E,F,G,H). Thus, the H70R variant not only reduces I_Kr current density but also changes the voltage-dependence of activation and accelerates deactivation (Table 1). These changes in I_Kr in aggregate are consistent with a loss of repolarizing current that can lead to action potential and QT prolongation.

Figure 3.

Reduced I_Kr with accelerated deactivation in hERG1-H70R iPSC-CMs. A. Representative whole-cell voltage clamp current traces from hERG1-H70R, hERG1-H70R^corr and DF19–9-11T iPSC-CMs elicited upon 4 s depolarizing voltage steps to −40mV, −20mV, 0mV, +10mV and +20mV before and after the application of 1 μM E-4031. The E-4031-sensitive current is defined as I_Kr. Inset: voltage protocol. B. C. D. Average current–voltage relationships for I_Kr measured at the end of the test pulses (B), peak tail I_Kr resulting from repolarization to −40 mV (C), and peak tail I_Kr normalized to the maximal current following repolarization to −40mV from iPSC-CMs. Values are represented as mean ± S.D, hERG1-H70R(n=14), hERG1-H70R^corr (n=9), DF19–9-11T (n=9); Temp 36 ± 1°C, * p<0.05 for hERG1-H70R relative to hERG1-H70R^corr or DF19–9-11T. E. Representative tail I_Kr at −40 mV measured after a depolarizing step to +20mV from iPSC-CMs. F. G. H. Fit I_Kr tail current deactivation at −40 mV following depolarizing step to +20 mV to a bi-exponential decay function for fast time constant, τ_fast (F); slow time constant, τ_slow (G); and fraction of fast deactivating I_Kr (H) from hERG1-H70R (n=9), hERG1-H70R^corr (n=8), and DF19–9-11T (n=7) iPSC-CMs. Values are for single cells with mean (thick bar) ± S.D. (thin bars); Temp 36 ± 1°C; ‡ p<0.005.

Table 1.

Summary of Mean Tail I_Kr Fit Parameters From iPSCCM Groups

	hERG1-H70R	hERG1-H70Rcorr	DF19–9-11T
Activation
V1/2	−18.3 ± 2.8 mV*	−11.3 ± 2.6 mV	−12.0 ± 1.4 mV
K	5.6 ± 0.7	6.1 ± 1.0	5.5 ± 0.4
Deactivation
τfast at +20mV†	69.9 ± 12.0 ms*	200.8 ± 42.8 ms	153.6 ± 33.8 ms
τslow at+20mV†	696.0 ± 130.5 ms*	1451.9 ± 349.9 ms	1313.1 ± 313.5 ms

All data are listed as mean ± S.E.M.. hERG-H70R: n= 9, hERG-H70R^corr: n=8, DF19–9-11T: n=7; Temp 36 ± 1°C

^*Indicates a significant difference from controls (p<0.05)

^†The τ value of the deactivation corresponds to a bi-exponential fit to the tail current (Figure 4 A)

Heterologous expression of hERG1-H70R reduces I_hERG without changing deactivation kinetics

To probe for the contribution of hERG1a and hERG1b subunits to the changes in I_Kr observed in the hERG1-H70R in iPSC-CMs, we performed heterologous expression studies in HEK293 cells comparing expression of WT hERG1a to that of hERG1a-H70R in homomeric hERG1a channels and heteromeric hERG1a/1b channels. The H70R variant significantly decreased I_hERG peak and tail current density when expressed as homomeric hERG1a or heteromeric hERG1a/1b channels (Figure 4A,B,C). However, expression of hERG1a-H70R did not alter the deactivation time constants (τ_fast and τ_slow) relative WT homomeric or heteromeric channels (Figure 4E,F,G,H). In agreement with prior studies, coexpression of hERG1b with hERG1a did result in acceleration of deactivation kinetics relative to hERG1a homomeric channels.^{8, 10} These results suggest that direct changes in gating properties by the hERG1a-H70R variant alone cannot fully recapitulate the findings found in the iPSC-CMs given the lack of effect of the variant on deactivation.

Figure 4.

Heterologous expression of hERG1-H70R shows reduced I_hERG without changing deactivation kinetics. A. Representative current traces for I_hERG from HEK293 transfected with equal total plasmid amounts encoding hERG1a (WT), hERG1a(H70R), and hERG1a (WT+H70R) homomeric channels as well as hERG1a (WT)/1b and hERG1a (H70R)/1b heteromeric channels. B. C. D. Average current–voltage relationships for I_hERG, measured at the end of the test pulses (B), peak tail I_hERG resulting from repolarization to −50 mV (C), and peak tail current normalized to the maximal current following repolarization to −50mV. Values are represented as mean ± S.D. E. Representative deactivation current traces I_hERG from hERG1a (WT), hERG1a (H70R), hERG1a (WT+H70R) homomeric channels and hERG1a (WT) /1b, hERG1a (H70R)/1b heteromeric channels. F. G. H. Fit I_hERG tail current deactivation at −40 mV following depolarizing step to +40 mV to a bi-exponential decay function for fast time constant, τ_fast (F); slow time constant, τ_slow (G); and fraction of fast deactivating I_hERG (H) from hERG1a (WT), hERG1a (H70R), hERG1a (WT+H70R) homomeric channels and hERG1a (WT)/1b, hERG1a (H70R)/1b heteromeric channels. Individual cell values are plotted with mean (thick bar) ± S.D. (thin bars). hERG1a (WT): n=5, hERG1a(H70R): n=7, hERG1a(WT+H70R): n=7, hERG1a(WT)/1b: n=6; hERG1a(H70R)/1b: n=8, * p<0.05.

hERG1-H70R iPSC-CMs show differential hERG1a/1b expression and trafficking

To investigate the molecular basis for the change in I_Kr in the hERG1-H70R iPSC-CMs, we evaluated expression levels of hERG1a/b isoforms. First, we used quantitative RT-PCR with primers specific for hERG1a and hERG1b mRNA to examine the gene expression in iPSC-CMs. hERG1b expression was significantly higher in hERG1-H70R iPSC-CMs than in hERG1-H70R^corr and DF19–9-11T iPSC-CMs (2.2 and 3.5 times, respectively). In contrast, there was no significant difference in hERG1a mRNA across the groups (Figure 5A).

Figure 5.

hERG1a and hERG1b expression in hERG1-H70R, hERG1-H70R^corr and DF19–9-11TiPSC-CMs. A. qRT-PCR analysis of hERG1a, hERG1b and KCNQ1 mRNA from hERG1-H70R, hERG1-H70R^corr, and DF19–9-11T iPSC-CMs. Expression values are relative to those of hERG1-H70R^corr, normalized to GAPDH, mean (thick bar) ± S.D. (thin bars); n = 4, * p<0.05. B. Representative immunoblots of hERG1subunit proteins from hERG1-H70R, hERG1-H70R^corr and DF19–9-11T iPSC-CMs with 3 independent experiments (Set1, Set2, Set3). Complex-glycosylated and core glycosylated hERG1 (hERG1a:155 /135 kDa, hERG1b: 90/80 kDa respectively) are indicated. β-actin is shown as a loading control. C. D. Immunoblot densitometry determined relative expression levels of complex glycosylated 155KDa hERG1a normalized to β-actin (C) and of complex glycosylated 90KDa hERG1b (D) normalized to β-actin mean (thick bar) ± S.D. (thin bars), n = 5, * p<0.05. E. F. Trafficking efficiency of hERG1a (E) and hERG1b (F) calculated from ratio of complex glycosylated immunoblot band intensity to total intensity of complex and core glycosylated proteoforms, mean (thick bar) ± S.D. (thin bars), n = 5, *p<0.05, n.s.= not significant.

To examine the effect of H70R on the protein levels of hERG1a and hERG1b, immunoblots using antibodies labeling the N-terminus and C-terminus of hERG1a and hERG1b were performed (Figure 5B). hERG1a was detected as a core glycosylated form (135 KDa) and a complex glycosylated form (155 KDa). hERG1b was detected as a core glycosylated form (80 KDa) and a complex glycosylated form (90 KDa). These hERG1a/1b proteoforms expressed in the iPSC-CMs are consistent with the proteoforms detected in heterologous expression and native human cardiac tissue.³² Because only the complex glycosylated forms of hERG1a/1b are thought to compose functional surface membrane channels,³³ we compared the average protein levels and found the expression of complex-glycosylated hERG1a was decreased in hERG1-H70R iPSC-CMs relative to both hERG1-H70R^corr and DF19–9-11T but the expression of complex-glycosylated hERG1b was not significantly different (Figure 5B,C,D). Heterologous expression studies have variably reported that the hERG1-H70R impairs trafficking of the protein leading to loss of functional channels, so we compared trafficking efficiency by determining the percent of total immunoreactivity present in the complex-glycosylated forms for hERG1a and hERG1b. Indeed the calculated trafficking efficiency for hERG1a was significantly reduced in the hERG1-H70R iPSC-CMs relative to the controls in contrast to the unchanged trafficking efficiency of hERG1b proteoforms which lacks the N-terminal H70R substitution (Figure 5E,F).

To evaluate further the effect of the hERG1-H70R variant on hERG1 expression and subcellular distribution in iPSC-CMs, we used N-terminal targeted antibodies and immunofluorescence imaging to detect specifically hERG1a and hERG1b. hERG1a immunofluorescence was less in hERG1-H70R iPSC-CMs relative to immunofluorescence in hERG1-H70R^corr and DF19–9-11T iPSC-CMs (Figure 6A) with approximately half the mean fluorescence intensity (Figure 6D). In addition, immunolabeling for hERG1a in hERG1-H70R iPSC-CMs showed greater overlap with the immunolabeling for protein disulfide isomerase (PDI), a marker of the endoplasmic reticulum relative to control iPSC-CM consistent with impaired protein trafficking. In contrast, hERG1b immunolabeling was comparable in all three lines studied (Figure 6B, E). These immunolabeling results in aggregate suggest that hERG1-H70R iPSC-CMs exhibit a reduction in surface membrane hERG1a expression due to impaired protein trafficking without a measurable change in the level or localization in hERG1b.Thus, the increase in the ratio of mature hERG1b to hERG1a expression may impact the composition of tetrameric hERG1 channels.

Figure 6.

Reduced hERG1a immunofluorescence in hERG1-H70R relative to hERG1-H70R^corr and DF19–9-11T iPSC-CMs in contrast to unchanged hERG1b immunofluorescence. A. B. Immunofluorescence images for (A) hERG1a- and (B) hERG1b-specific N-terminal antibodies (green) in representative iPSC-CMs derived from hERG1-H70R, hERG1-H70R^corr and DF19–9-11T iPSC and co-labeled with antibodies for endoplasmic reticulum marker protein disulfide isomerase (PDI, red) and for the myocyte marker sarcomeric myosin (MF20, cyan). Nuclei are labeled with 4’,6-diamidino-2-phenylindole for DNA (blue). Scale bars: 25 μm. C. Primary antibodies were omitted for background fluorescence signal. D. E. Quantification of hERG1a (D) and hERG1b (E) immunofluorescence signal intensity in iPSC-CM groups. Data points are for single cells with mean (thick bar) ± SD (thin bars), * p<0.05.

hERG1-H70R increases the hERG1b/hERG1a ratio in mRNA complexes

To determine if the observed changes in hERG1a/1b expression are due to changes in the microtranslatome that is responsible for co-translational assembly of hERG1a/1b subunits,¹³ we evaluated the impact of hERG1-H70R on KCNH2 gene expression using smFISH to visualize transcripts.^{30, 34} Two smFISH probe sets were designed with spectrally distinct fluorophores – probe set a (PSa) targeting the 5’ region of KCNH2 ORF that is specific to hERG1a and probe set ab (PSab) targeting the 3’ region of KCNH2 that both hERG1a and hERG1b share (Figure 7A, see Methods). To verify the specificity of the probes, we used PSa and PSab in HEK293 cells stably transfected with constitutively expressed hERG1a, dox-inducible hERG1b or empty plasmid. Both probe sets produced smFISH signal in HEK293 cells expressing hERG1a or hERG1a/1b but not in control HEK293 cells (Supplemental Figure IVA). In hERG1a expressing cells, both PSa and PSab detected overlapping signals; however, in hERG1a/1b expressing cells PSab detected a subset of signals absent from PSa representing hERG1b mRNA (arrow head, Supplemental Figure IVA).

Figure 7.

smFISH analysis reveals increase in hERG1b/1a ratio in mRNA complexes without change in nuclear nascent transcript abundance in hERG1-H70R iPSC-CMs. A. Diagram of hERG1 smFISH probe design. The gene structure of KCNH2 is shown with exons (blue boxes) and introns (black solid lines). To visualize hERG1a and hERG1b and distinguish between them, two smFISH probes sets were used, with spectrally distinct fluorophores. The Probe Set a (PSa) targets exons only in hERG1a (black dashed lines) and thus visualizes only hERG1a. The Probe Set ab (PSab) targets 3’ region that hERG1a and hERG1b share (magenta dashed lines) and thus visualizes both isoforms. Each probe set comprises 48 of 20-nucleotide-long probes. Diagram is not to scale. B. hERG1 smFISH with hERG1-H70R (top), hERG1-H70R^corr (middle) and DF 19–9-11 T (bottom).smFISH were conducted with PSa (left, yellow) and PSab (middle, magenta). DNA is stained with Hoechst (right, cyan). Square boxes (right) are 9X magnification of white boxes. Arrow: hERG1a mRNA (seen both with PSa and PSab); arrowhead: hERG1b mRNA (seen only with PSab); asterisk: hERG1a&1b mRNA complex. Scale bar: 10 μm. C. Left: hERG1a mRNAs can be seen in both PSa and PSab channels whereas hERG1b mRNAs can be seen only in PSab channel. Right: the equations for counting the number of each isoform in the RNA complex. D. hERG1 isoform ratio (1b/1a) per mRNA complex is shown for hERG1-H70R (red, n=324), hERG1-H70R^corr (blue, n=117) or DF 19–9-11 T (black, n=160)., §: p< 0.001. E. hERG1 isoform ratio (1b/1a) of total hERG1 mRNAs in the cytoplasm. (n= 28, 23, and 28 cells from left). F. G. Total number of nascent hERG1a (F) or hERG1b (G) transcripts in each nucleus is estimated based on its intensity relative to the average single mRNA intensity. n= 59, 30, and 39 nuclei from left). D-G. mean (thick middle bar) ± S.D. (thin bars). n.s., not significant.

In iPSC-CMs, multiplexed PSa and PSab distinguished hERG1a mRNA from hERG1b mRNA. As expected, some spots were seen both in PSa and PSab channels (arrows in Figure 7B), which are hERG1a transcripts, and some only in PSab channel (arrowheads in Figure 7B) showing hERG1b. Most smFISH spots in the cytoplasm were similar in their size and intensity, indicative of individual mRNA visualization as previously shown.³⁵ Notably, there were multiple spots that were larger than single mRNA spots in cytoplasm of iPSC-CMs (asterisks in Figure 7B). These mRNA complexes varied in size and intensity. The RNA complexes were located in the cytoplasm near the nucleus, and were seen in both PSa and PSab channels in hERG1-H70R, hERG1-H70R^corr and DF19–9-11T iPSC-CMs. The average intensities of cytoplasmic hERG1a mRNAs (from PSa channel) and hERG1b mRNAs (from PSab channel corrected for hERG1a signal) were used to calculate the number of isoform-specific mRNAs in each complex which could impact the composition of the translated tetrameric channels. Consistent with the qRT-PCR data (Figure 5A), analysis of hERG1 isoforms present in complexes demonstrated significantly more hERG1b mRNAs in hERG1-H70R iPSC-CMs compared to either hERG1-H70R^corr or DF19–9-11T CMs whereas hERG1a expression was unchanged across the cell lines (Supplemental Figure IVB,C). If the analysis expanded to all mRNAs, both single mRNAs and mRNAs in complexes, we did not detect a significant difference in isoform expression, but the high cell-to-cell variability both in total number of mRNAs and the number of mRNA complexes in the cell limited the power to observe differences (Supplemental Figure IVD,E)

We next asked how the mRNA expression pattern in mRNA complexes was impacted by the H70R variant. To answer the question, we calculated the hERG1b/1a mRNA ratio in each mRNA complex defined as having an intensity at least equal to 4 single mRNAs. Notably, hERG1-H70R iPSC-CMs shows an increased average hERG1b/1a ratio by about 50 % (Figure 7D). However, the variant does not change the hERG1b/1a ratio of free individual mRNAs in the cytoplasm (Figure 7E) nor the total number of mRNA complexes (Supplemental Figure IVF), suggesting that the H70R variant specifically impacts the mRNA ratio in the mRNA complexes. To determine if the mRNA ratio change is due to altered transcription of hERG1a and/or hERG1b, or due to post-transcriptional regulation, we examined nuclear nascent transcripts which reflect the transcriptional activity of the corresponding locus.^{30, 35, 36} We compared the nuclear smFISH signal using PSa and PSab in hERG1-H70R, hERG1-H70R^corr and DF19–9-11T iPSC-CMs to estimate nascent transcripts of hERG1a and hERG1b. In all three groups of iPSC-CMs, the numbers of nascent transcripts for both hERG1a and hERG1b are comparable (Figure 7F,G). These results suggest that the H70R variant raises the relative number of hERG1b mRNAs in the mRNA complexes without impacting the transcription of hERG1a and hERG1b but by altering the stability or degradation of the transcripts.

Discussion

Here we provide new insights into the pathogenic mechanisms of a PAS domain sequence variant associated with LQT2 using patient-specific hERG1-H70R iPSC-CMs. We demonstrate that hERG1-H70R iPSC-CMs exhibit a prolonged APD due to a reduction in I_Kr compared to isogenic control and unrelated control iPSC-CMs consistent with the LQT2 phenotype. In agreement with our prior heterologous expression studies, we find evidence for impaired trafficking of hERG1a based on a reduction in the relative abundance of the mature, complex glycosylated hERG1a relative to the immature form. In contrast, the protein level of mature hERG1b is unchanged in the H70R variant line. Despite unchanged transcription of hERG1a and 1b in hERG1-H70R iPSC-CMs relative to control lines based on nuclear nascent transcripts, we uncover an unexpected increase in hERG1b mRNA abundance as well as an increase in the hERG1b:1a ratio in mRNA complexes. Our results are consistent with a model in which reduced functional hERG1 channels at the surface membrane and an increases in hERG1b relative to hERG1a in heteromeric channels generate smaller I_Kr with faster deactivation (Figure 8). These changes in I_Kr underlie the prolongation of APD and QT_c typical of this syndrome. The study uncovers a newly appreciated effect of the hERG1a-specific H70R PAS domain pathogenic variant producing subunit imbalance, a mechanism that may be operative in up to 30% of LQT2 patients.

Figure 8.

Model of hERG1-H70R variant pathogenesis of LQT2. Cotranslational assembly of hERG1 complexes occurs at the ER where hERG1a is required for dimer formation initially leading to hERG1a-1a and hERG1a-1b complexes that ultimately form tetrameric hERG1 channels. In this simplified model, the hERG1a-H70R encoded subunit misfolds and undergoes ER-associated degradation (ERAD) along with associated subunits. This disproportionately impacts dimeric hERG1a-1a complexes (75% contain mutant allele) relative to hERG1a-1b complexes (50% contain mutant allele). Post-transcriptional changes in mRNA levels result in increased hERG1b mRNA abundance with an increase in the ratio of hERG1b:1a in mRNA complexes. The net result is to increase the relative ratio of hERG1b:1a in tetrameric channels at the membrane with an overall reduction in the number of total channels relative to control. (RNA-binding proteins, RBP)

The PAS domain of hERG1 channels represents a ‘hotspot’ for LQT2-associated genetic variants with at least 63 pathogenic or likely pathogenic variants at 52 different sites listed in the ClinVar database.¹¹ However, multiple mechanisms have been proposed to cause the prolongation of repolarization from PAS domain variants based on heterologous expression studies in mammalian cultured cells. A class 2 or trafficking deficit of channels was identified for 86% of tested variants based on a reduction in the 155-kDA complex glycosylated channel on immunoblots from HEK293 cells including the H70R variant,²⁴ although another study found fewer PAS domain variants as trafficking deficient.²⁵ Alternatively, altered gating of hERG1 channels leading to faster channel deactivation has been identified for some PAS domain variants, but H70R expressed without hERG1b in heterologous systems does not lead to altered channel deactivation in our study or others.²³ The results from the present study show that H70R human iPSC-CMs do exhibit impaired trafficking of hERG1a leading to reduced I_Kr but in addition demonstrate an unanticipated acceleration of deactivation of I_Kr potentially due an increase in the relative abundance of hERG1b.

Changes in the relative expression of hERG1a/1b have been described during development and in disease. In mouse, ERG1b/1a ratiometric expression is greater in neonatal heart relative to the adult heart.³⁷ Such developmental differences have not been evaluated for human heart. Intriguingly, differences in deactivation kinetics of I_Kr have been observed comparing the epicardium to the midmyocardium of the canine left ventricle suggesting possible differences in hERG1a/1b transmurally.³⁸ In human heart failure, although changes in hERG1a and hERG1b mRNA expression were not detected, a significant reduction at the protein level in the ratio of hERG1a/hERG1b was detected and suggested to contribute to impaired repolarization.³² In tumor cells, dynamic changes in the relative abundance of hERG1a and hERG1b have been demonstrated in a cell cycle-dependent manner.³⁹ These observations suggest that the stoichiometry of hERG1a/1b in functional channels is not fixed and is highly regulated.

The mechanisms governing the protein levels of hERG1a and hERG1b subunits in functional hERG1 channels in CMs are only beginning to be understood. Regulation at the level of transcription, post-transcription, translation and post-translation ultimately determine the abundance of functional channels. A study of the genomic structure of hERG1 and transcript analysis demonstrated that hERG1a and hERG1b transcripts arise from separate promoters rather than by alternative splicing,⁴⁰ but in our experiments the hERG1-H70R variant did not change nuclear nascent transcript levels for hERG1a or hERG1b suggesting that transcription is unchanged. Coexpression studies of hERG1a/1b in oocytes have demonstrated that titrating the abundance of hERG1b mRNA relative to hERG1a mRNA produced ionic currents with increasingly rapid kinetics of current deactivation that reached a plateau and never became as fast as hERG1b monomeric channels.⁹ This finding and biochemical crosslinking studies provided evidence that hERG1b preferentially interacts with hERG1a subunits rather than hERG1b itself. ⁹ Liu et al. reported that mRNA transcripts encoding hERG1a and 1b subunits are physically associated and undergo co-translation, consistent with the preferred heteromeric association of the corresponding channel subunits,¹³ and our smFISH data directly demonstrate the presence of hERG1a/1b mRNA complexes. We observed that H70R variant resulted in an increase in hERG1b:hERG1a ratio in mRNA complexes and an increase in overall hERG1b mRNA. The H70R variant induced mRNA changes are secondary to alterations in post-transcriptional regulation given unchanged basal transcription, but the mechanistic basis for this change in mRNA levels requires further investigation. The H70R variant also induces a major defect in trafficking of hERG1a to the surface membrane and an associated reduction of mature hERG1a protein levels; whereas, hERG1b subunits do not show a defect in trafficking and maintain unchanged expression levels. The mechanistic link between changes in mRNA levels and protein trafficking defects induced by H70R requires future investigation, but the net effects is to produces reduced I_Kr amplitude with altered gating properties manifest by accelerated deactivation. To integrate these findings, we provide a model in which the preferential association of hERG1b with hERG1a leads to a preferential depletion of hERG1a in the presence of the H70R variant induced trafficking defect resulting in an imbalance in subunits and fewer functional channels at the surface membrane (Figure 8).

Loss of the appropriate stoichiometric subunit balance of multiprotein complexes is recognized as potentially damaging. This is described by the balance hypothesis first developed based on yeast genetics data.⁴¹ In addition, dosage compensation mechanisms exist such as post-translational degradation or other types of nonlinear buffering to maintain subunit stoichiometry.^42–44 Variant-induced imbalances in subunit stoichiometry have been linked to a handful of genetic disease such as COL5A1 variants leading to Ehlers Danlos Syndrome or BMPR2 variants linked to pulmonary hypertension, but such subunit imbalance has not been previously described in the case of LQT2 or other inherited arrhythmia syndromes. The PAS domain LQT2 variants are uniquely positioned to induce subunit imbalance as they are expressed only in hERG1a; whereas, most other LQT2 variants are present in both hERG1a and hERG1b given the largely overlapping exon structure.

There are limitations to the present study. The iPSC-CM model is more representative of fetal human CMs rather than adult CMs. Thus, there may be differences in the relative expression of hERG1a and hERG1b among other maturation changes relative to adult CMs that we cannot directly determine given the lack of adult tissue from patients with this genotype. Secondly, the stoichiometry of individual hERG1 channels remains unknown. Hence, we do not know the precise distribution of subunits in the tetrameric channels present in the cell membrane or the composition of channel proteins undergoing degradation by ER-associated degradation, (ERAD).

What Is Known?

• The KCNH2 gene encodes hERG1a and hERG1b subunits which form the tetrameric voltage-activated potassium channel responsible for IKr.

• The Per-Arnt-Sim (PAS) domain present in hERG1a but not hERG1b is a hot spot for long QT syndrome type 2 (LQT2) pathogenic variants which have an unknown impact on channel subunit composition.

What the Study Adds?

• Patient-specific iPSC-derived cardiomyocytes expressing the H70R PAS domain LQT2 variant exhibit prolonged action potential duration relative to controls due to a reduction in peak IKr and accelerated deactivation.

• H70R iPSC-derived cardiomyocytes exhibit impaired trafficking of hERG1a to the surface membrane and an increase in the relative abundance of hERG1b mRNA and protein levels leading to an imbalance in subunit composition and associated abnormalities in IKr.

Nonstandard Abbreviations and Acronyms

AP

Action Potential

APA

Action potential Amplitude

APD

Action Potential Durations

CHO

Chinese Hamster Ovary cell

CMs

Cardiomyocytes

CRISPR-Cas9

The Clustered Regularly Interspaced Short Palindromic Repeat-associated protein 9

c-TnT

cardiac Troponin T

ECG

Electrocardiogram

ER

Endoplasmic Reticulum

ERAD

Endoplasmic Reticulum Associated Degradation

hERG1

human Ether-à-go-go Related Gene 1

HEK293

Human Embryonic Kidney 293 cell line

ICD

Implantable Cardiodefibrillator

I_Kr

rapid delayed rectifier potassium current

iPSC

induced Pluripotent Stem Cell

iPSC-CMs

induced Pluripotent Stem Cell-derived Cardiomyocytes

LQTS

Long QT Syndrome

MDP

Maximum Diastolic Potential

mRNA

messenger Ribonucleic Acid

PAS

Per/Arnt/Sim domain

PDI

Protein Disulfide Isomerase

RBP

RNA-binding proteins

RT-PCR

Reverse transcriptase-Polymerase chain reaction

SCD

Sudden Cardiac Death

smFISH

single-molecule RNA Fluorescence In Situ Hybridization

ssODN

single-strand oligonucleotide

TdP

Torsades de pointes