Loss-of-function mutations in cardiac ryanodine receptor channel cause various types of arrhythmias including long QT syndrome

Abstract

Aims

Gain-of-function mutations in RYR2, encoding the cardiac ryanodine receptor channel (RyR2), cause catecholaminergic polymorphic ventricular tachycardia (CPVT). Whereas, genotype–phenotype correlations of loss-of-function mutations remains unknown, due to a small number of analysed mutations. In this study, we aimed to investigate their genotype–phenotype correlations in patients with loss-of-function RYR2 mutations.

Methods and results

We performed targeted gene sequencing for 710 probands younger than 16-year-old with inherited primary arrhythmia syndromes (IPAS). RYR2 mutations were identified in 63 probands, and 3 probands displayed clinical features different from CPVT. A proband with p.E4146D developed ventricular fibrillation (VF) and QT prolongation whereas that with p.S4168P showed QT prolongation and bradycardia. Another proband with p.S4938F showed short-coupled variant of torsade de pointes (scTdP). To evaluate the functional alterations in these three mutant RyR2s and p.K4594Q previously reported in a long QT syndrome (LQTS), we measured Ca²⁺ signals in HEK293 cells and HL-1 cardiomyocytes as well as Ca²⁺-dependent [³H]ryanodine binding. All mutant RyR2s demonstrated a reduced Ca²⁺ release, an increased endoplasmic reticulum Ca²⁺, and a reduced [³H]ryanodine binding, indicating loss-of-functions. In HL-1 cells, the exogenous expression of S4168P and K4594Q reduced amplitude of Ca²⁺ transients without inducing Ca²⁺ waves, whereas that of E4146D and S4938F evoked frequent localized Ca²⁺ waves.

Conclusion

Loss-of-function RYR2 mutations may be implicated in various types of arrhythmias including LQTS, VF, and scTdP, depending on alteration of the channel activity. Search of RYR2 mutations in IPAS patients clinically different from CPVT will be a useful strategy to effectively discover loss-of-function RYR2 mutations.

RyR2, Ryanodine receptor, Ventricular fibrillation, Arrhythmias, QT prolongation, Mutation, Loss of function

What’s new?

• A novel RYR2 mutation, p.S4168P, and reported ones, p.E4146D, p.K4594Q, and p.S4938F were identified in patients showing phenotypes different from catecholaminergic polymorphic ventricular tachycardia (CPVT). Functional analyses showed loss-of-function type channel properties.

• Loss-of-function RyR2 mutants might cause various types of arrhythmias, including QT prolongation, ventricular fibrillation (VF), and short-coupled variant of torsade de pointes, unlike gain-of-function type RyR2 mutants that exclusively cause CPVT.

• According to functional analyses, RYR2 mutations that moderately suppress RyR2 activity with frequent localized Ca²⁺ waves in HL-1 cells may be related with lethal arrhythmias such as VF following short-coupled premature ventricular contractions (early afterdepolarizations), whereas ones that severely suppress RyR2 activity without inducing Ca²⁺ waves might cause milder clinical features.

Introduction

Cardiac ryanodine receptor (RyR2), encoded by RYR2, is a huge subunit for Ca²⁺ release channel expressed on the sarcoplasmic reticulum (SR). In cardiomyocytes, RyR2 is activated by Ca²⁺ influx through sarcolemmal L-type Ca²⁺ channels (LTCC), which is followed by massive Ca²⁺ release from the SR through RyR2.¹ Gain-of-function type RyR2 mutants have been shown to be associated with catecholaminergic polymorphic ventricular tachycardia (CPVT). CPVT is a rare arrhythmogenic disease characterized by polymorphic ventricular tachycardia following adrenergic stimulation and cardiac sudden death in children or young adults without structural heart diseases and electrocardiogram (ECG) abnormalities at rest.² In CPVT, the mutant RyR2s release Ca²⁺ from the SR even during the diastolic phase without Ca²⁺ influx from LTCC, thereby evoking delayed afterdepolarizations (DADs).^3–6 The onset of CPVT is generally at age of 5–10 years.^2,7

More recently, several loss-of-function type RYR2 mutations have been shown to be pathogenic as well. These mutations exhibit various clinical phenotypes, including idiopathic ventricular fibrillation (VF),^8–10 short-coupled variant of torsade de pointes (scTdP),¹¹ and atypical CPVT with left ventricular non-compaction.⁹ Furthermore, in the presence of β-adrenergic stimulation, transgenic model mice carrying a loss-of-function mutant RyR2 (p.A4860G) exhibited VF and early afterdepolarizations (EADs).¹⁰ Up to date, there are however few studies reporting the genotype-phenotype correlation on other loss-of-function type RYR2 mutations.

Inherited primary arrhythmia syndromes (IPAS) is a useful resource for searching genetic mutations¹² that may result in ventricular tachycardia or VF, leading to sudden cardiac death. We previously identified an RYR2-S4938F mutation in a proband with scTdP and demonstrated that this was a loss-of-function mutation.¹¹ In the present study, by using the next generation sequencing technique, we searched genetic variants in cases presenting clinical features different from CPVT among IPAS patients and identified additional two probands carrying RYR2 mutations.

These mutations were p.E4146D and p.S4168P. By using HEK293 cell heterologous expression system, we examined the functional effect of these two mutations as well as those of p.S4938F and p.K4594Q, a previously-reported mutation associated with long QT syndrome (LQTS).¹³ All the mutants have loss-of-function type changes. [³H]Ryanodine binding assay classified the mutants into two types; functionally moderate (p.E4146D and p.S4938F) and severe (p.S4168P and p.K4594Q) suppression groups. Interestingly, heterologous expression of the former group in HL-1 cardiomyocytes caused frequent localized Ca²⁺ waves, whereas that of the latter group showed only suppression in amplitude of action-potential induced Ca²⁺ transients. Consistently, in patients with moderately suppressed RyR2 function, lethal arrhythmias were documented; in contrast, no fatal events were observed in those with a severely suppressed function. Taken together, loss-of-function RYR2 mutations may cause various types of arrhythmias, depending on the alteration level of the channel activity.

Methods

Study subjects and genetic analysis

We screened RYR2 gene for 710 IPAS patients less than 16-year-old, who were registered at Shiga University of Medical Science or Kyoto University Graduate School of Medicine between 2002 and 2017 for genetic analysis. We obtained written informed consent from their guardians in accordance with the study protocol approved by the ethics committees in both institutes. Genomic DNA was isolated from peripheral blood lymphocytes of patients, and we performed a genetic analysis by targeted gene sequencing of 60 genes (for detailed gene names, Supplementary material online, Table) using the MiSeq System (Illumina, CA, USA). Detected variants were confirmed by the Sanger method. Cardiac events were defined as syncope, ventricular tachycardia/VF, or aborted cardiac event.

Generation of stable inducible HEK293 cells

RyR2 cDNA was amplified by PCR as five fragments (F1, 1–2351; F2, 2351–4404; F3, 4404–8861; F4, 8861–10185; and F5, 10185–14901) using cDNA from murine ventricular myocytes as previously reported.^6,14,15 The PCR products were cloned into the expression vector pcDNA5/FRT/TO (Life Technologies, CA, USA). Ligation of these fragments using specific restriction enzymes was performed to construct full-length RyR2. Three mutations, p.E4146D, p.S4168P, and p.K4594Q, were introduced by inverse PCR, and confirmed by DNA sequencing. HEK293 cells expressing wild-type (WT) or mutant RyR2 were generated using the Flp-In T-REx system (Life Technologies, CA, USA). Clones with suitable expression levels of RyR2s were selected and used in the experiments.

Single-cell Ca²⁺ imaging in HEK293 cells

Single-cell Ca²⁺ imaging in HEK293 cells was performed 26–30 h after induction with doxycycline as described previously.^6,11,14 For cytoplasmic Ca²⁺ monitoring, cells were loaded with fura-2 AM (Molecular Probes, OR, USA) for 30 min at 37°C and then incubated with normal Krebs solution (140 mM NaCl, 5 mM KCl, 2 mM CaCl₂, 1 mM MgCl₂, 11 mM glucose, 5 mM HEPES, pH 7.4) for 30 min at room temperature to allow de-esterification of the indicator. Fura-2 was excited with an alternating beam at 340 and 380 nm, and fluorescence signals at wavelengths longer than 420 nm were acquired using an inverted microscope (TE2000E, Nikon, Tokyo, Japan) equipped with sCMOS camera (Zyla4.2, Andor, Belfast, UK) and MetaMorph software (Molecular Devices, CA, USA). Ca²⁺ signals were obtained in normal Krebs solution and then in caffeine containing Krebs solution. Ratiometric signals (F340/F380) in individual cells were determined using region of interest (ROI) analysis.

Endoplasmic reticulum (ER) luminal Ca²⁺ signals were obtained using genetically encoded Ca²⁺ indicator R-CEPIA1er¹⁶ (a gift from Dr Masamitsu Iino, The University of Tokyo, K_D = 565 nM, n = 1.71) using inverted microscope (TE2000E, Nikon, Tokyo, Japan) equipped with EM-CCD camera and AquaCosmos software (Hamamatsu Photonics, Hamamatsu, Japan). Fluorescent Ca²⁺ signals (F) were measured in normal Krebs solution and then in caffeine containing Krebs solution. At the end of each measurement, F_min and F_max of Ca²⁺ indicator were obtained with 0Ca Krebs solution containing 20 µM ionomycin (Calbiochem/Merck, Darmstadt, Germany), 5 mM BAPTA, and 20 µM cyclopiazonic acid and 20Ca Krebs solution containing 20 µM ionomycin and 20 mM CaCl₂, respectively.⁶ Fluorescence intensities in individual cells (F) were determined using ROI analysis with AquaCosmos software. Ca²⁺ signals were expressed as (F-F_min)/(F_max-F_min).

Ca²⁺ imaging in HL-1 cells expressing mutant RyR2

HL-1 cardiac cells,¹⁷ derived from mouse atrial cardiac muscle cells, were maintained in the supplemented Claycomb medium (JRH Biosciences, KS, USA) with 10% foetal bovine serum, 100U/mL penicillin and 100 µg/mL streptomycin, 2 mM glutamine, and 0.1 mM norepinephrine. Recombinant WT or mutant RyR2 was expressed with baculovirus expression system as previously reported.⁶ Near-confluent HL-1 cells were infected with RYR2-IRES-mCherry baculovirus. The cytosolic Ca²⁺ imaging of beating HL-1 cells was recorded 48 h after baculovirus infection using the Ca²⁺ indicator Cal520-AM (AAT Bioquest, Sunnyvale, CA, USA) at 30°C. Cal520 images were obtained every 48 ms for 48 s in one measurement. Fluorescent Ca²⁺ signals (F) were expressed as F/F₀ where F₀ was the minimal signal at rest. To analyse Ca²⁺ waves, time-based scan images were obtained from lines drawn on individual cells, and Ca²⁺ wave frequency, propagation distance, and wave duration were determined. The Ca²⁺ wave duration was obtained as the time required to return to 10% of the peak from the time of wave onset.

Ca²⁺-dependent [³H]ryanodine binding analysis

[³H]Ryanodine binding analysis was performed as described previously,^11,14,15 and some parameters were calculated by the results. Microsomes (50–100 μg of protein) prepared from HEK293 cells were incubated with 5 nM [³H]ryanodine for 1 h at 25°C in a solution containing 0.17 M NaCl, 20 mM MOPSO, pH 7.0, 2 mM dithiothreitol, 1 mM AMP, 1 mM MgCl₂, and various concentration of free Ca²⁺ buffered with 10 mM EGTA. The protein-bound [³H]ryanodine was filtered with polyethyleneimine-treated Whatman GF/B glass filters. The presence of 20 μM unlabelled ryanodine was included to confirm the specificity of binding. The [³H]ryanodine binding data (B) were normalized to the maximal binding of [³H]ryanodine (B_max), which was separately determined by the Scatchard plot analysis using various concentrations (3–20 nM) of [³H]ryanodine in a high-salt buffer containing 1 M NaCl. The resultant B/B_max represents the average activity of each channel.

The parameters of Ca²⁺-dependent [³H]ryanodine binding were obtained by fitting the data to the following equation:

$$A=A_{\max }\times f_{\mathrm{A}}\times \left(1–f_1\right)$$

(1)

where A is the activity at the specified Ca²⁺, A_max is the gain that determines the maximal attainable activity, and f_A and f_I are fractions of the activating Ca²⁺ site (A-site) and inactivating Ca²⁺ site (I-site) occupied by Ca²⁺, respectively.^11,14,15f_A and f_I at the specified Ca²⁺ concentration ([Ca²⁺]) are expressed as:

$$f_{\mathrm{A}}={\left[{\text{Ca}}^{2+}\right]}^{n_{\mathrm{A}}}/\left({\left[{\text{Ca}}^{2+}\right]}^{n_{\mathrm{A}}}+K_{\mathrm{A}}{}^{n_{\mathrm{A}}}\right)$$

(2)

$$f_{\mathrm{I}}={\left[{\text{Ca}}^{2+}\right]}^{n_{\mathrm{I}}}/\left({\left[{\text{Ca}}^{2+}\right]}^{n_{\mathrm{I}}}+K_{\mathrm{I}}{}^{n_{\mathrm{I}}}\right)$$

(3)

where K_A and K_I are dissociation constants, and n_A and n_I are Hill coefficients for Ca²⁺ of A- and I-sites, respectively. The Hill coefficients were fixed at 2.0 and 1.0 for n_A and n_I, respectively, which maximized the sum of the R² values for curve fitting. Curve fitting was performed using the Prism 8 software (GraphPad Software, CA, USA).

Statistics

In each of the experiments, n represents number of performed experiments. Data are expressed as mean ± standard deviation. Statistical significance was analysed using Student’s t-test or one-way analysis of variance. P-values of <0.05 were considered significant.

Results

Genetic analysis and patients’ characteristics

In 710 IPAS probands, we identified 63 RYR2 mutation carriers (8.9%), and three of them did not show the typical CPVT phenotypes (Table 1, Patients 1, 2, 4); c.12438g>c, p.E4146D (Figure 1A), c.12502t>c, p.S4168P (Figure 1B), and c.14813c>t, p.S4938F. E4146D was reported previously¹⁸ and S4168P was a novel one. We previously reported S4938F and demonstrated that it caused a loss-of-function.¹¹ Except for the RYR2 mutations, the three probands carried no other variants in the analysed genes, including the common LQTS-related genes KCNQ1, KCNH2, and SCN5A (Supplementary material online, Table).

Figure 1

Genetic analyses of two probands. (A) The electropherogram and DNA sequence showing a RYR2 mutation in a female baby with RYR2-E4146D. (B) The electropherogram and DNA sequence in a male baby with RYR2-S4168P. (C) Topology of the RyR2 and the locations of four loss-of-function mutations (red circles). Black circles indicate previously described loss-of-function mutations; I4855M in ref.⁹ and A4860G in ref.⁸

Table 1

Clinical characteristics of four index patients

Characteristic	Patient 1	Patient 2	Patient 3	Patient 4
Sex	Female	Male	Female	Male
Clinical diagnosis	LQTS, scPVC	LQTS	LQTS	IVF, scTdP
Age of onseta (years)	0.2	Before birth	10	13
Family history	Unknown	+	−	−
Cardiac events
Aborted cardiac arrest	VF	−	−	VF
VT/TdP	TdP	−	−	−
Syncope	+	−	+	+
Condition of cardiac events	Crying	−	Running, Swimming	Rest after exercise
Gene	RYR2	RYR2	RYR2	RYR2
Electrocardiogram
QT prolongation	+	+	+	−
QTc duration (ms)	506	514	485	401
Bradycardia	–	+	–	–
Short-coupled PVC	+	−	–	+
Pathogenic mutation
Nucleotide	12438 g>c	12502 t>c	13780 a>c	14813 c>t
Amino Acids	E4146D	S4168P	K4594Q	S4938F
SNP ID	rs796052206	–	rs796052207	–
Previous reports	Shigemizu et al.18	Novel	Taniguchi et al.13	Fujii et al.11

IVF, idiopathic ventricular; LQTS, long-QT syndrome; scPVC, short coupled premature ventricular contract; scTdP, short coupled torsades de pointes; TdP, torsades de pointes; VF, ventricular fibrillation; VT, ventricular tachycardia.

a

Age of lethal arrhythmic event.

In addition to these three mutations, we noticed an RYR2 mutation, K4594Q, reported in a patient presenting the overlapping phenotype of LQTS and CPVT¹³ (Table 1, Patient 3), but the variant was not yet functionally analysed. Figure 1C shows the location of the four mutations (red dots). E4146D and S4168P are located in the central domain of RyR2, whereas K4594Q and S4938F in the S2-S3 region and the C-terminal domain, respectively.

Detailed clinical characteristics of the patients

Patient 1: a female LQTS infant with RYR2-E4146D

A 2-month-old girl baby suddenly started crying and became pale after taking milk, and mandibular breathing appeared (Figure 2). She was immediately taken to her doctor, and VF was recorded by the monitor ECG. Cardiopulmonary resuscitation was promptly initiated, and her heart rhythm was restored by electrical defibrillation. After admission, ECG showed moderate QT prolongation (QTc: 482 ms; Figure 2A), and she was diagnosed with LQTS. Combination therapy with propranolol (9 mg per day) and mexiletine (180 mg per day) was then started.

Figure 2

Clinical features of the proband with RYR2-E4146D and her parents. (A) A 12-lead ECG recorded at the 3 months of age. (B) Holter ECG traces showing a short-coupled PVC. Arrowhead, bidirectional arrows, and horizontal line indicate as described in the text. RR intervals (ms) were displayed at the bottom of the trace. (C) Electrograms of her parents. The pedigree of the family is shown in the centre. The arrow indicates the proband.

After hospitalization, she suffered from repetitive VF events after crying. ECG monitors showed a short-coupled premature ventricular contraction (PVC), which often preceded a TdP with a further QT prolongation (QTc: 506 ms). ECG trace in Figure 2B demonstrated that the QRS complex (indicated by arrowhead) showed a sudden and bizarre QT prolongation, and then the next QRS complex also showed a very different T-wave morphology, which was immediately followed by a PVC with an extremely-shortened coupling interval (260 ms, bidirectional arrows), suspecting the presence of EAD. This PVC failed to induce the TdP but caused a marked change in T-wave morphology of the next sinus complex with QT prolongation (QT 385 ms, QTc 574 ms, indicated by the horizontal bar). In addition, RR intervals and T-wave morphology before and after the PVC were unstable, which might reflect the heterogeneity of the action potentials among the cardiomyocytes.

As she had repeated similar life-threatening events even after starting the medication, an implantable cardiac defibrillator (ICD) was implanted at the age of 3 months. The combination therapy then suppressed her cardiac events. However, at the age of 14 years, she received the appropriate ICD therapy due to recurrent arrhythmic events. Genetic analysis was then performed and identified RYR2-E4146D (Figure 1A). As her events were suspected to be due to the mutation of RYR2, a major causative gene of CPVT, her therapeutic regimen was changed to a combination of carvedilol (50 mg per day) and flecainide (100 mg per day). Life-threatening events were prevented after modification of her medications. She had no family history of lethal arrhythmias, and her parents’ ECGs showed no QT prolongation (insets of Figure 2C). We did not detect the same mutation in her mother, and her asymptomatic father refused the genetic analysis (Figure 2C).

Patient 2: a male LQTS infant with RYR2-S4168P

The patient first presented severe bradycardia at 20 weeks of gestation (Figure 3). Due to the bradycardia, he underwent a Caesarean section at 37 weeks of gestation. Sinus bradycardia continued after birth (heart rate: ∼75 b.p.m.),¹⁹ indicating the possibility of congenital heart disease.¹⁹ His ECG also showed remarkable QT prolongation (QTc: 514 ms; Figure 3A). LQTS type 3 was then suspected, and mexiletine was administered. After starting oral mexiletine, his heart rate gradually increased, and the QT interval was shortened from 514 ms to 433 ms at 30 days after birth (Figure 3B).

Figure 3

Clinical features of the proband with RYR2-S4168P. (A) A 12-lead ECG of the proband (A) at 1 day after birth and (B) at 30 days after mexiletine administration. (C) Pedigree of the proband’s family. The arrow indicates the proband. (D) A 12-lead ECG of proband’s mother.

Familial genetic analysis revealed that the proband’s mother also carried the same mutation (Figure 3C). At the age of 14 years, she had an episode of syncope during swimming. Due to the absence of abnormal ECG findings (Figure 3D), she was then diagnosed with epilepsy and had started anti-epileptic for short term, and no episode of syncope recurred even after its cessation. His father had no syncope, and his ECG was normal. No other family members had syncopal episodes.

Patient 3: a 12-year-old LQTS girl with RYR2-K4594Q

The proband suffered repeated exercise-induced syncopal events since she was 10 years old as previously reported.¹³ Her ECG showed QT prolongation at rest (QTc: 485 ms). The QT interval was remarkably prolonged after exercise and during the administration of epinephrine, like type 1 long-QT syndrome.

Patient 4: a 13-year-old boy with RYR2-S4938F

We previously reported that the patient suffered VF events induced by short coupled PVC.¹¹ He was treated by verapamil for the prevention of VF, however, it was gradually ineffective, and appropriate ICD shocks for VF were recorded. Therefore, he was given flecainide, and it prevented VF for 1.5 years.

Summary of the clinical phenotypes

As summarized in Table 1, the age of onset and clinical severity varied among four probands. The onset of symptoms was <1 year old in Patients 1 and 2. QT interval prolongation was observed in three of four patients except Patient 4. Patients 1 and 4 received ICD implantation due to repeated VF, which were eventually suppressed by flecainide. Exercise-related cardiac events were documented in three of four patients except Patient 2, but his genotype-positive mother experienced syncope during swimming.

Ca²⁺ signals in HEK293 cells expressing WT or mutant RyR2

To evaluate the functional changes caused by the above RyR2 mutations of unknown functions, we generated stable and inducible HEK293 cells expressing RyR2-WT, E4146D, S4168P, K4594Q, and S4938F. We initially recorded cytoplasmic Ca²⁺ signals using fura-2. HEK293 cells harbouring WT RyR2 exhibited spontaneous cytoplasmic Ca²⁺ oscillations in normal Krebs solution (Figure 4Aa, arrows). In contrast, no spontaneous Ca²⁺ oscillations were observed in cells expressing mutant RyR2 (E4146D, S4168P, K4594Q, and S4938F) (Figure 4Ab–e and B). The resting cytoplasmic Ca²⁺ level was significantly lower in the mutant cells than in the WT ones (Figure 4C). Application of 30 mM caffeine (30 Caf), a potent RyR2 activator, caused a transient increase in Ca²⁺ in both WT and the mutant cells (Figure 4Aa–e). The amplitudes of the caffeine responses in E4146D, K4594Q, and S4938F cells were significantly larger than that in WT cells (Figure 4D). These findings are consistent with the idea of higher ER Ca²⁺ level in these mutant cells due to suppressed Ca²⁺ release. On the other hand, S4168P cells exhibited smaller increase in Ca²⁺ with slow rising and decay phases in response to caffeine application (Figure 4Ac and D). This can be explained by either an increased Ca²⁺ release activity of S4168P causing reduction in the releasable ER Ca²⁺, or a reduced caffeine sensitivity of S4168P mutation.

Figure 4

Functional properties of mutant RyR2s characterized by single-cell Ca²⁺ imaging in HEK293 cells expressing WT or mutant RyR2. (A) Representative fura-2 cytosolic Ca²⁺ [Ca²⁺]_cyt signals from RyR2-WT (a) and mutant (b–e) cells in normal Krebs solution followed by 30 mM caffeine-containing Krebs solution (grey bar). Arrows indicate spontaneous Ca²⁺ oscillation (a). Thin lines indicate 30 s. (B) Oscillation frequency in cells with WT or mutant RyR2 in normal Krebs solution. Data are means ± SD (n = 69–71). (C) Resting [Ca²⁺]_cyt signals in WT and mutant cells in normal Krebs solution. Data are means ± SD (n = 66–70). (D) Peak [Ca²⁺]_cyt changes on application of 30 mM caffeine containing solution. Data are means ± SD (n = 66–70). (E) Representative ER Ca²⁺ ([Ca²⁺]_ER) signals from WT (a) and mutant (b–f) RyR2 cells. [Ca²⁺]_ER signals were obtained with R-CEPIA1er in normal Krebs solution followed by the addition of 10 mM (a–c, e, f, open bar) or 30 mM (d, grey bar) caffeine. Fluorescence signal values are expressed as (F-F_min)/(F_max-F_min) where F_min and F_max were determined in the presence of ionomycin with BAPTA plus cyclopiazonic acid and CaCl₂, respectively. Thin lines indicate 1 min. (F) [Ca²⁺]_ER in cells expressing WT and mutant RyR2s in normal Krebs solution. Both upper (open column) and lower (hatched column) levels are shown for WT cells, and only upper [Ca²⁺]_ER are shown for the four mutant cells because no spontaneous Ca²⁺ oscillations were observed. Data are means ± SD (n = 62–103). NA, not applicable. *** indicates statistical significance with P < 0.001 vs. WT using one-way ANOVA followed by Dunnett’s multiple comparison test. (G) Fraction of cells showing Ca²⁺ release in response to 10 mM (open column) and 30 mM caffeine (hatched column) (n = 48–70).

To know whether ER Ca²⁺ was increased in these mutant cells, we next monitored ER Ca²⁺ level with ER Ca²⁺ sensor protein, R-CEPIA1er, which has been proved to be a useful tool to estimate the activity of RyRs expressed in HEK293 cells.^6,11,14–16 HEK293 cells harbouring WT RyR2 exhibited spontaneous cyclic decrease in ER Ca²⁺ which corresponds to spontaneous cytoplasmic Ca²⁺ oscillations as previously reported^6,11,14–16 (Figure 4Ea). The R-CEPIA1er signals increased to the upper level (shown by dashed line in Figure 4Ea) before Ca²⁺ release, and then fell to the lower level (dotted line in Figure 4Ea) by Ca²⁺ release, and then increased again to the upper level. Therefore, the upper level corresponds to the threshold for spontaneous Ca²⁺ release. In contrast to WT, all four cells expressing the mutant RyR2 consistently showed significantly higher ER Ca²⁺ signals, without any cyclic Ca²⁺ changes, in normal Krebs solution (Figure 4Eb–f and F). Taken together with the results of cytoplasmic Ca²⁺ signals, these findings suggest that the four RyR2 mutants have reduced Ca²⁺ release activity compared to WT.

Application of 10 mM caffeine triggered massive Ca²⁺ release with almost complete depletion of ER Ca²⁺ in E4146D, K4594Q, and S4938F cells which is comparable to WT (Figure 4Eb, e, f and G). In S4168P cells, 10 mM caffeine failed to evoke Ca²⁺ release (Figure 4Ec and G) but 30 mM caffeine exerted a moderate Ca²⁺ release in most S4168P cells (83%), which was substantially higher than in control HEK293 cells with no exogenous RyR2 (21%) (Figure 4Ed and G). Thus, S4168P appears less sensitive to caffeine than WT and other three mutations.

[³H]ryanodine binding

Next, we measured Ca²⁺-dependent [³H]ryanodine binding to microsomes prepared from HEK293 cells expressing WT or mutant RyR2 to quantitatively evaluate the activity of the RyR2 channels.^6,11 RyR2-WT channels demonstrated biphasic Ca²⁺-dependent [³H]ryanodine binding (Figure 5A). E4146D and S4938F showed a reduced [³H]ryanodine binding with a rightward shift of Ca²⁺ dependence. In contrast, both S4168P and K4594Q channels almost lost the [³H]ryanodine binding at all the Ca²⁺ concentrations examined.

Figure 5

Ca²⁺-dependent [³H]ryanodine binding assay. (A) Ca²⁺-dependent [³H]ryanodine binding to microsomes from HEK293 cells expressing WT or mutant RyR2 channels. Data are means ± SD (n = 3). All four mutant channels exhibited a reduced channel activity. (B) Three parameters for CICR activity. CICR is determined by three independent parameters: gain (A_max) and the sensitivities to activating Ca²⁺ (K_A) and inactivating Ca²⁺ (K_I), and expressed by Equations (1)–(3) (see Methods). (C) Three parameters (A_max, K_A, and K_I) for WT and the mutant channels obtained from data in A. Data are mean ± SD (n = 3). *P < 0.05, ***P < 0.001 from WT using one-way ANOVA followed by Dunnett’s multiple comparison test. (D) Calculated CICR activity of mutant RyR2s at resting Ca²⁺ (pCa 7). Note that the CICR activities for E4146D and S4938F were less than 1/10 of WT, and those for S4168P and K4594Q were too low to be reliably detected. ND, not determined.

The biphasic property of Ca²⁺-induced Ca²⁺ release (CICR) activity is explained by three parameters, i.e. the gain (A_max) and dissociation constants for activating Ca²⁺ (K_A) and inactivating Ca²⁺ (K_I)^11,14 (Figure 5B). We obtained these parameters by fitting Equations (1)–(3) to the [³H]ryanodine binding data (see Methods). E4146D and S4938F reduced A_max and increased K_A and K_I compared to WT (Figure 5C). We failed to calculate reliable parameters for S4168P and K4594Q due to extremely low [³H]ryanodine binding activity. The CICR activity at resting [Ca²⁺]_i is useful to interpret the phenotype of cellular Ca²⁺ homeostasis of the mutant channel. Since the [³H]ryanodine binding at resting [Ca²⁺]_i (100 nM) is too low to be accurately determined, we instead calculated the value by substituting the obtained parameters (A_max, K_A and K_I) into Equations (1)–(3) in which free [Ca²⁺]_i was set at 100 nM. The calculated CICR activity at resting [Ca²⁺]_i of RyR2-E4146D and -S4938F was less than 1/10 of WT (Figure 5D). These findings suggest that all four mutations cause loss-of-function phenotype and that E4146D and S4938F channels show milder phenotypes than S4168P and K4594Q.

Ca²⁺ signals in HL-1 cardiomyocytes expressing WT or mutant RyR2

In general, patients with RyR2 mutation heterozygously express WT and mutant RyR2s. To investigate the properties of mutant RyR2 channels in the cardiac environment, recombinant RyR2s were exogenously expressed in HL-1 cardiomyocytes by the baculovirus system.⁶ mCherry was used as a reporter for transduction. Since HL-1 cells endogenously express RyR2, mCherry-positive cells are expected to have both endogenous WT and exogenous mutant RyR2s. Ca²⁺ measurements were carried out in the beating confluent HL-1 cells, where clustered cells displayed spontaneous and synchronized action potential-induced Ca²⁺ transients.

Control HL-1 cells without baculovirus infection showed rhythmic Ca²⁺ transients but rarely displayed Ca²⁺ waves (Figure 6Aa and B). The expression of exogenous RyR2-WT caused occasional global Ca²⁺ waves that propagate throughout the cells (Figure6Ab, B, and F, Supplementary material online, Movie S1). Interestingly, cells expressing RyR2-E4146D and S4938F mutants showed localized Ca²⁺ waves that propagated only a short distance (Figure 6A c and f, B, and F, Supplementary material online, Movie S2) and duration of the Ca²⁺ waves were significantly shorter than that in RyR2-WT cells (Figure 6G). Although the percentage of cells showing the Ca²⁺ waves was lower in RyR2-S4938F and E4146D than in RyR2-WT, the frequency of the waves was much higher than that in RyR2-WT (Figure 6C and E). In RyR2-S4168P and K4594Q cells, in contrast, there were few cells showing Ca²⁺ waves during beating (Figure 6C). Ca²⁺ transient amplitudes were significantly reduced in all the infected cells compared with control (Figure 6D). Application of 10 mM caffeine caused larger Ca²⁺ release in E4146D, K4594Q, and S4938F cells than in RyR2-WT cells (Figure 6H and I), suggesting that ER Ca²⁺ levels in mutant cells remained high despite the small Ca²⁺ transients. Although caffeine response was small in the RyR2-S4168P cells (Figure 6H and I), this does not mean reduction in ER Ca²⁺ levels, because the mutant is less sensitive to caffeine in HEK293 cells (see Figure 4Ec and d).

Figure 6

Functional properties characterized by Ca²⁺ imaging in beating HL-1 cells expressing WT and mutant RyR2s. (A) Representative traces of Ca²⁺ signals in non-infected control (a) and exogenous-RyR2 expressing mCherry(+) cells (b–f). In b–f, Ca²⁺ signals from mCherry(−) cells (grey line) together with mCherry(+) cells (red line) in the same cluster are shown to indicate autogenic action potential pacing. Blue ticks below indicate the onset of synchronized action potential-induced Ca²⁺ transients (a–f), and purple arrows indicate the onset of Ca²⁺ waves in mCherry(+) cells (b, c, and f). Thin lines indicate 2 s. (B) Representative time-based scan images of Ca²⁺ signals in control and mCherry(+) cells expressing exogenous RyR2. The time-based scan images were obtained from red lines drawn on single cells, of which cell boundaries are indicated with yellow dotted lines. Blue ticks indicate the onset of action potential-induced Ca²⁺ transients. Purple arrows indicate centre of localized Ca²⁺ waves in E4146D and S4938F cells. (C) Fraction of cells showing Ca²⁺ waves during beating in control and exogenous RyR2 expressing mCherry(+) cells (n = 82–163 cells). (D) Peak amplitudes of Ca²⁺ transients in control and mCherry(+) exogenous-RyR2 expressing cells. Data for the Ca²⁺ transients were obtained from mCherry(+) cells both with and without showing Ca²⁺ waves. But we did not collect data for Ca²⁺ transients in cells with too many Ca²⁺ waves but without clear Ca²⁺ transients (n = 15–30 cells). *P < 0.05, **P < 0.01, ***P < 0.001 from Control, ^#P < 0.05, ^###P < 0.001 from WT using one-way ANOVA followed by Dunnett’s multiple comparison test. (E–G) Frequency (E), propagation distance (F), and duration (G) of Ca²⁺ waves in mCherry(+) WT, E4146D and S4938F cells (n = 20–21 cells). Data are given as box and whisker plots with minimum and maximum (n = 20–21 cells). ***P < 0.001 from WT using one-way ANOVA followed by Dunnett’s multiple comparison test. (H) Representative traces of Ca²⁺ signals obtained in normal Krebs solution following the addition of 10 mM in control (left) and mCherry(+) WT (left), S4168P (right) and K4594Q cells (right). Note that S4168P showed slow onset of caffeine response, indicating poor sensitivity to caffeine. (I) Amplitude of caffeine response in control and exogenous-RyR2 expressing mCherry(+) cells. *P < 0.05, **P < 0.01, ***P < 0.001 from control using one-way ANOVA followed by Dunnett’s multiple comparison test.

Discussion

The advent of next generation sequencing enabled us to identify many RYR2 mutations, and most of them were gain-of-function mutations. Recently, several RYR2 mutations were found to show loss-of-functions.^8–11 However, the genotype–phenotype correlations on loss-of-function type RYR2 mutations remain unclear due to the small number analysed so far. In this study, we searched for arrhythmias different from CPVT in IPAS patients linked to RyR2 mutations and effectively found loss-of-function mutations with extremely high probability (3 out of 3 probands = 100%). To examine the genotype–phenotype correlations, we performed functional analyses of four RYR2 mutations displaying non-CPVT phenotypes: QT prolongation, VF, short-coupled PVC, and bradycardia with early onset. The functional analyses using HEK293 cells and HL-1 cardiomyocytes, together with clinical investigations, suggest that RyR2 loss-of-function mutations may be divided into two groups: mutations that moderately suppress RyR2 activity is associated with VF and scTdP, and those severely suppress RyR2 activity mainly caused QT prolongation and bradycardia.

Infantile fatal arrhythmias have been linked to cardiac channelopathies.¹² In 2007, Tester et al. reported two RYR2 mutations, p.R2267H and p.S4565R, in patients with SIDS.¹² The functional analysis of these mutant RyR2 channels displayed an increased open probability compared with WT. They concluded that the gain-of-function of RYR2 mutation is one of the major causes associated with SIDS.^3,4 We here identified two loss-of-function mutations in infant patients. The baby with RYR2-E4146D suffered from VF at 2 months and might have been diagnosed with SIDS if the resuscitation were unsuccessful. The baby carrying RYR2-S4168P was found to present severe foetal bradycardia and QT prolongation immediately after birth. Their onsets were much earlier than that of CPVT. Our data suggest that loss-of-function RYR2 mutations may also be associated with SIDS.

The cellular mechanism of VF by one of loss-of-function RYR2 mutations (RYR2-A4860G) has been previously reported.^8,10 Single channel recordings, Ca²⁺ imaging and [³H]ryanodine binding data all indicated that RYR2-A4860G mutation dramatically suppressed the channel activity.^8,10 Zhao et al.¹⁰ engineered a transgenic mouse model carrying RyR2-A4860G^+/− and demonstrated that an adrenergic stress induced randomly-occurring EADs, the prolongation of action potential duration (APD) and VF. The proposed mechanisms for the EADs, APD prolongation and VF are as follows: as RyR2-A4860G^+/− cardiomyocyte expresses low activity RyR2 channels, LTCC first elicits a smaller amplitude of Ca²⁺ transient than that in WT cells and leaves a residual amount of Ca²⁺ inside the SR, which gradually builds up at each action potential generation. When the SR Ca²⁺ load reaches a threshold sufficient to activate hyporesponsive RyR2 channels, prolonged Ca²⁺ release occurs. It in turn activates the electrogenic Na⁺/Ca²⁺ exchanger (NCX) during phase 2 and 3 of the action potential, thereby causing APD prolongation and triggering EADs. The reduced amplitude of Ca²⁺ transient was also suggested to contribute to QT prolongation by suppressing inactivation in LTCC. In RYR2-E4146D and K4594Q, the prolongation of QT interval was induced by adrenergic stress: crying in E4146D and treadmill exercise test in K4594Q.¹¹ Therefore, the changes in QT intervals in these patients seem to occur by a mechanism similar to that with RyR2-A4860G mutation.

More interestingly, the phenotypes of loss-of-function mutations in our study seemed to fall into two types: one with severe VF, and the other with QT prolongation without lethal arrhythmias. Our baby patient with RYR2-E4146D showed a short-coupled PVC during a Holter recording (Figure 2B). Though this PVC failed to trigger VF on this occasion, it mimicked the ECG features of a short-coupled torsade de pointes carrying the loss-of-function RyR2 mutation (S4938F, Figure 1C).¹¹ The short-coupled PVC was preceded by sudden further prolongation of QT interval with bizarre change in T-wave morphology. These features of loss-of-function type RyR2 mutants were unique and differed from those observed in gain-of-function RYR2 mutants (typical CPVT).

These phenomena suggested the presence of repolarization abnormality, triggered activity and EADs by a mechanism similar to that proposed in RyR2-A4860G^+/− mice. In such situation, EAD evoked by irregular Ca²⁺ release from Ca²⁺-saturated SR could lead to lethal arrhythmias. Interestingly, HL-1 cells expressing RyR2-E4146D and S4938F evoked high-frequency Ca²⁺ waves with short-distance propagation (Figure 6), suggesting that these mutants are prone to cause abnormal Ca²⁺ release and facilitate EAD. Because [³H]ryanodine binding of RyR2-E4146D is similar to that of RyR2-S4938F, a partially decreased peak activity with reduced sensitivity to activating Ca²⁺ (Figure 5),¹¹ mutations that moderately suppress RyR2 activity might cause short-coupled PVC related arrhythmias due to the facilitated abnormal Ca²⁺ release during repolarization phase in cardiomyocytes. It is noteworthy that arrhythmias in the RyR2-E4146D and S4938F patients were sensitive to flecainide, similar to those seen in patients with gain-of-function mutations. The inhibitory effect of RyR2 by flecainide has been still under debate. What is certain is that flecainide can suppress the triggered activity generated from DAD and EAD by Na channel blocking action.²⁰ Because E4146D and S4938F are likely to cause EAD and scTdP, it is reasonable to expect that those arrhythmias can be suppressed by flecainide.

In contrast, the ECG in patients with RyR2-S4168P and K4594Q mutations showed QT prolongation at rest without lethal arrhythmic events. These mutants displayed almost no [³H]ryanodine binding activity (Figure 5). HL-1 harbouring RyR2-S4168P and K4594Q showed few Ca²⁺ wave and substantial reduction in Ca²⁺ transient amplitude (Figure 6). The suppression of RyR2 activity by the mutations was seemingly too severe to evoke random Ca²⁺ release after Ca²⁺ overloading of the SR, but only to decrease Ca²⁺ transient amplitude. The reduction in Ca²⁺ transients may cause the impairment of Ca²⁺ dependent inactivation in LTCC. However, severely defeated RyR2 activity would not induce afterdepolarizations. As a consequence, severely depressed RyR2s might cause milder clinical features, such as QT prolongation alone, without provoking lethal arrhythmias. In addition, severely-reduced activity of RyR2 may result in foetal bradycardia because RyR2 is known to play an important role in the Ca²⁺ clock that regulates pacemaker activity in sinoatrial nodal cells.²¹

Regarding to the Ca²⁺ oscillation in mutant RyR2s which were observed in HL-1 cells but not in HEK293 cells, the phenomenon may be explained as follows. In HEK293 cells, there is no endogenous RyR2 in HEK293 cells, the mutant RyR2s in the experiments using HEK293 cells are all homo-tetramers. Therefore, the activities of the mutant RyR2s were not observed, even if the functional changes were moderate. In contrast, in the cells with endogenous RyR2s including human cardiomyocytes and HL-1 cells, mutant RyR2s are supposed to be hetero-tetramers or the mix of homo-tetramers of WT and mutants, and these diverse RyR2s would lead to the functional difference among the mutant RyR2s.

Limitations and future research

We used HEK293 cells and HL-1 cells to measure Ca²⁺ dynamics and ryanodine bindings. These experimental methods are excellent way to evaluate the mutant RyR2s. Especially in HL-1 cells, we could detect Ca²⁺ kinetics in the coexistence with WT and mutant RyR2s. However, the expression ratios of WT and mutant RyR2s are not controlled as with in the patients’ ones. Therefore, our results might not reflect the exact kinetics of the mutant RyR2s. Future experiments using induced pluripotent stem cell-derived cardiomyocytes or animal models will elucidate the proarrhythmic mechanisms caused by these mutant RyR2s. In silico studies of computational modelling would also help understand the mechanistic insights of these mutations into cardiac action potentials and arrhythmogenesis.

Conclusion

We demonstrated four loss-of-function RYR2 mutations. According to those phenotypes, loss-of-function RyR2 mutants might cause a very early onset of arrhythmias, including QT prolongation, VF and short-coupled PVC, unlike gain-of-function type RyR2 mutants that induce a classical type of CPVT. Those moderately suppressing RyR2 activity cause short-coupled PVCs (EADs), VF and/or QT prolongation, whereas those severely suppressing RyR2 milder arrhythmic events without fatal arrhythmias. Genetic screening of patients with those arrhythmias functional analyses might unveil mechanisms underlying the arrhythmias caused by RYR2 mutations.

Supplementary material

Supplementary material is available at Europace online.

Acknowledgements

We thank Ms Arisa Ikeda, Kazu Toyooka, Madoka Tanimoto, Ikue Hiraga, and the Laboratory of Radioisotope Research, Research Support Center, Juntendo University Graduate School of Medicine for excellent technical assistance.

Funding

This work was supported by the AMED [JP18ek0109202 to S.O.], Basis for Supporting Innovative Drug Discovery and Life Science Research (BINDS) [JP20am0101080 to T.M.], a Health Science Research Grant from the Ministry of Health, Labour and Welfare of Japan for Clinical Research on Measures for Intractable Diseases [H27-032 to M.H.]; Grants-in-Aid for Scientific Research from the Japan Society for the Promotion of Science [15K09689 to S.O., 15H04818 to M.H., 15K08243, 19K07105 to N.K., and 19H03404 to T.M.]; an Intramural Research Grant for Neurological and Psychiatric Disorders of National Center of Neurology and Psychiatry [29-4 and 2-5 to T.M.]; and the Vehicle Racing Commemorative Foundation [6114 to T.M.].

Conflict of interest: none declared.

Data availability

The data underlying this article will be shared on reasonable request to the corresponding author.