The V2475F CPVT1 mutation yields distinct RyR2 channel populations that differ in their responses to cytosolic Ca²⁺ and Mg²⁺

Abstract

Catecholaminergic polymorphic ventricular tachycardia type 1 (CPVT1) is a lethal genetic disease causing arrhythmias and sudden cardiac death in children and young adults and is linked to mutations in the cardiac ryanodine receptor (RyR2). The effects of CPVT1 mutations on RyR2 ion-channel function are often investigated using purified recombinant RyR2 channels homozygous for the mutation. However, CPVT1 patients are heterozygous for the disease, so this approach does not reveal the true changes to RyR2 function across the entire RyR2 population of channels in the heart. We therefore investigated the native cardiac RyR2 single-channel abnormalities in mice heterozygous for the CPVT1 mutation, V2475F(+/−)-RyR2, and applied molecular modelling techniques to investigate the possible structural changes that could initiate any altered function. We observed that increased sensitivity of cardiac V2475F(+/−)-RyR2 channels to both activating and inactivating levels of cytosolic Ca²⁺, plus attenuation of Mg²⁺ inhibition, were the most marked changes. Severity of abnormality was not uniform across all channels, giving rise to multiple sub-populations with differing functional characteristics. For example, 46% of V2475F(+/−)-RyR2 channels exhibited reduced Mg²⁺ inhibition and 23% were actually activated by Mg²⁺. Using homology modelling, we discovered that V2475 is situated at a hinge between two regions of the RyR2 helical-domain-1 (HD1). Our model proposes that detrimental functional changes to RyR2 arise because mutation at this critical site reduces the angle between these regions. Our results demonstrate the necessity of characterising the total heterozygous population of CPVT1-mutated channels in order to understand CPVT1 phenotypes in patients.

Graphical Abstract

(Left) RyR2 Helical Domain-1 with the V2475F mutation site indicated by the red circle. (Top right) Typical RyR2 single-channel recordings highlighting the increased opening burst durations of RyR2 isolated from V2475F(+/−) hearts compared to wild type hearts. (Bottom right) Possible RyR2 tetramer formations in V2475F(+/−) hearts. White (V) and green (F) circles represent wild type and mutated monomers, respectively.

Introduction

The cardiac isoform of the sarcoplasmic reticulum (SR) ryanodine receptor (RyR2) is the primary Ca²⁺-release channel in the heart (Rousseau et al., 1986). The unique gating and ion conducting properties of RyR2 are essential for maintaining appropriate SR Ca²⁺ release during cardiac excitation-contraction (EC) coupling so that the heart can beat in a rhythmic fashion. The signal that triggers the coordinated opening of RyR2 channels across all the individual muscle cells of the heart is a rise in cytosolic [Ca²⁺] (Fabiato, 1983; Näbauer et al., 1989; Bers, 2002). However, the resultant increase in open probability (Po) of RyR2 is dependent on many other physiological factors, not least on the many ligands that, in synergy with cytosolic Ca²⁺, regulate RyR2 gating (Meissner & Henderson, 1987; Sitsapesan & Williams, 1994a; Kermode et al., 1998). Anything that perturbs this complex regulatory system has the potential to disrupt cardiac SR Ca²⁺ release with lethal consequences. This is exemplified in catecholaminergic polymorphic ventricular tachycardia (CPVT), a disease caused by mutations to RyR2 or associated proteins which can be acquired in an autosomal dominant or recessive manner (Lahat et al., 2001; Priori et al., 2001; Laitinen et al., 2001). When the disease is linked to mutations in RyR2, it is termed CPVT type 1 (CPVT1) and is autosomal dominant (Priori et al., 2001). Typically, patients have a normal baseline ECG, sometimes exhibiting bradycardia, but without any evidence of cardiac structural abnormality (Priori et al., 2002). Upon physical exertion or during emotional stress, premature ventricular beats can arise leading to polymorphic ventricular tachycardia and ventricular fibrillation (Napolitano et al., 2016). Early diagnosis is important because mortality among children and young adults is high when untreated (Swan et al., 1999; Priori et al., 2002).

There is growing awareness that symptoms additional to adrenergic stress-induced ventricular arrhythmias are linked to mutations in RyR2. These include supra-ventricular arrhythmias (Else et al., 2012) and structural changes to the heart that are associated with specific types of cardiomyopathy (Bhuiyan et al., 2007; Denegri et al., 2014; Lavorato et al., 2017; Roston et al., 2017; Bongianino et al., 2017; Alvarado et al., 2019; Nozaki et al., 2020). That such a range of symptoms arise from mutations to RyR2 suggests that different mutations produce distinct functional changes to the gating and/or conductance of RyR2. In turn, this leads to different abnormalies of SR Ca²⁺ release and the development of divergent cardiac problems and there are an increasing number of reports detailing the clinical phenotypes, for example Priori & Chen (2011) (Priori & Chen, 2011). Altered patterns of SR Ca²⁺ release and the characteristics of catecholamine-induced arrhythmias in mouse models of CPVT are also frequently studied. However, there is a paucity of information describing the detailed alterations to RyR2 single-channel function that occur in the various RyR2 mutations that cause life-threatening disease. CPVT1 is generally thought to be caused by ‘leaky’ RyR2 channels that have an increased sensitivity to cytosolic or luminal Ca²⁺ (Jiang et al., 2005; Liu et al., 2006; Xiao et al., 2016; Uehara et al., 2017) or which are ‘hyperphosphorylated’ (Marx et al., 2000; Wehrens et al., 2003; Lehnart et al., 2004) but these generalised descriptions are too vague to enable the design of effective treatments that are suitable to treat the divergent symptoms of the patients.

The problem is partly due to the difficulties of isolating sufficient RyR2 protein from the tiny hearts of CPVT mouse models to incorporate into planar phospholipid bilayers for single-channel studies. Most of our understanding of CPVT-induced changes to RyR2 function has been obtained using membrane preparations enriched with RyR2 from mouse hearts using detergent or, more commonly, using recombinantly expressed RyR2 (Liu et al., 2013; Uehara et al., 2017). It is known that use of detergent can affect the gating of RyR channels (Beard et al., 2002; Györke et al., 2004) and so this should be avoided for functional studies. Using recombinant RyR2 has additional problems. Importantly, the mutation will be present in all of the four monomers that make up the RyR2 tetrameric ion channel. However, CPVT1 is an autosomal dominant condition. Patients are heterozygous for the mutation and so only a small fraction of the total population of channels present in the heart will have four mutated monomers. Fig. 1 illustrates the expected likelihood of various RyR2 subunit compositions for a patient with CPVT1.

Figure 1. Likelihood of formation of RyR2 tetramers with different compositions of V2475F mutated monomers.

White (V) and green (F) circles represent WT and V2475F(+/−) mutated monomers, respectively. Assuming random assembly and identical surface expression, the probabilities of tetramers with the indicated numbers of mutant or WT subunits per tetrameric RyR2 channel is indicated as a percentage (%).

To begin addressing the above problems, we have isolated SR vesicles from the hearts of mice, heterozygous for the V2475F mutation which is associated with CPVT1 in humans, and incorporated these into bilayers to record single RyR2 channel currents. We have investigated if the effects of the four most important RyR2 regulatory ligands (cytosolic Ca²⁺, luminal Ca²⁺, cytosolic Mg²⁺ and ATP) are altered. We have also examined whether monovalent and divalent cation conductance and the relative Ca²⁺: K⁺ permeability is affected. We have recorded an unprecedented number of single-channel experiments which has allowed us to investigate changes in function among the different sub-populations of the V2475F(+/−) RyR2 channels in the heart. V2475 resides in the RyR2 helical domain 1 (HD1) and neighbours R2474, a residue also known to be a CPVT1 mutation site (Jiang et al., 2005; Lehnart et al., 2008). Peng et al. (2016) show that this is a flexible region with the potential to influence neighbouring domains and regulate distant parts of the RyR2 protein (Peng et al., 2016). Using molecular modelling we have therefore compared the possible structural changes induced by the V2475F and R2474S mutations. Our study shows that the V2475F-RyR2 mutation causes very specific alterations to single-channel conductance and ligand regulation. Furthermore, by characterising the overall population of channels derived from the hearts of V2475F(+/−) mice we have identified sub-populations of RyR2 that display graded changes in their response to cytosolic Ca²⁺ and Mg²⁺.

Methods

Ethical Approval

All procedures have been approved by the University of Oxford Ethical Review Committee (Pharmacology and Biochemistry subcommittee, approval ref. PPL: 30–3340) in strict accordance with the relevant United Kingdom Animals (Scientific Procedures) Act 1986 Home Office and institutional guidelines and regulations. The authors confirm that they have taken all steps to minimise the animals’ pain and suffering. This work complies with the Journal’s animal ethics checklist (Grundy, 2015).

Animal Information

Heterozygote RyR2-V2475F Knock-In (RyR2-V2475F(+/−)) C57BL/6J mice and wild-type (WT) littermates were kindly provided by Hector Valdivia from the University of Wisconsin (Loaiza et al., 2013), and were housed in individually ventilated cages in a pathogen-free facility at the University of Oxford. Mice were maintained in a 12 h light/dark cycle and given ad libitum access to normal chow and water. The use of mice was shared between members of the laboratory, in which cardiac and other tissues were taken from the same mouse wherever possible to comply with the “three Rs” principles (Fenwick et al., 2009). Isolated mixed membrane vesicles were prepared from 5 female and 5 male hearts from RyR2-V2475F(+/−) or WT littermates between 10–13 weeks old. Mice were humanely killed by cervical dislocation in line with ASPA Schedule 1 and the Journal’s ethical standards. The use of sedatives or anaesthetic agents were avoided to minimise confounding effects on the tissues obtained from the mice. For each intervention, data were collected using at least 3 membrane preparations (therefore 30 animals total).

Isolation of mixed membrane from mouse cardiac tissue

Isolated mixed membrane vesicles were prepared using methods described previously (El-Ajouz et al., 2017). Briefly, mouse cardiac tissue was dissected and snap frozen. Frozen tissue was finely homogenised in a buffer containing 300 mM sucrose, 20 mM K⁺ piperazine-N,N′-bis(2-ethanesulfonic acid) (KPIPES), 5 mM sodium fluoride (NaF), 2.5 mM dithiothreitol (DTT), 1 mM phenylmethylsulfonyl fluoride (PMSF), supplemented with a protease inhibitor cocktail, pH 7.2. The tissue homogenate was centrifuged at 6,000 g for 20 min at 4°C. The supernatant obtained was centrifuged at 120,000 g for 1 h at 4°C to pellet the membrane fraction. This was resuspended in 400 mM sucrose, 5 mM Tris/HEPES, 5 mM NaF at pH 7.2. The preparation was snap frozen and stored at −80°C.

Single-channel recordings

RyR2 channels were incorporated into planar phospholipid bilayers as previously described (Sitsapesan et al., 1991) in a 1:1 mixture of phosphatidylethanolamine: phosphatidylserine (PE:PS) (Avanti Lipids). With Ca²⁺ as the permeant ion, current fluctuations through RyR2 channels were recorded under voltage-clamp conditions in solutions of 250 mM HEPES, 80 mM Tris, pH 7.2, buffered with 1 mM EGTA to 100 nM – 10 mM Ca²⁺ on the cytosolic side, and 250 mM glutamic acid, 10 mM HEPES, pH to 7.2 with Ca(OH)₂ (free Ca²⁺ approximately 50 mM) on the luminal side. The luminal chamber was voltage-clamped at ground and the cytosolic chamber was clamped at various potentials relative to ground. In experiments with K⁺ as the permeant ion, symmetrical 210 mM KPIPES, pH 7.2 was used at a holding potential of −30 mV to minimise openings of SR K⁺ channels, and to create a physiological Ca²⁺ ion movement of luminal to cytosolic direction. In relative permeability measurements, the luminal channel side was perfused with 210 mM CaCl₂, 40 mM HEPES, pH 7.2. Corrections were made for the measured liquid junction potential. Where required, a Ca²⁺ electrode (Orion 93–20; Thermo Fisher Scientific) was used to buffer Ca²⁺ to the appropriate free [Ca²⁺] and all solutions were buffered to pH 7.2 with a Ross-type pH electrode (Orion 81–55; Thermo Ficher Scientific).

Single-channel analysis

Single-channel recordings were digitised at 20 kHz and low-pass filtered at 800 Hz or 2 kHz in experiments with Ca²⁺ or K⁺ as the permeant ion, respectively. Probability of opening (Po) was determined over 3 min of continuous recording using 50% threshold analysis (Colquhoun & Sigworth, 1995) in Clampfit (Molecular Devices, USA). The Po values indicated above the representative traces shown in the figures were calculated from the full 3 min recording for that channel. Where >1 channel (N) incorporated into the bilayer, Po is reported as average Po as indicated in the equation below:

$$\mathit{\text{avgPo}}=\frac{NPo}{N}=\frac{P{o}_1+2P{o}_2+3P{o}_3+\cdots nP{o}_n}{N}$$

Where only a single channel incorporated into the bilayer, lifetime analysis was performed using Clampfit. With Ca²⁺ as the permeant ion, events shorter than 1 ms were not fully resolved and were stripped from the idealisation. Lifetime distributions were fitted to a probability density function (pdf) by the method of maximum likelihood fitting (Colquhoun & Sigworth, 1995) as described previously (Sitsapesan & Williams, 1990). To estimate Po populations and identify peaks, kernel density estimations (KDE) were calculated in MatLab 2018 using a bandwidth of 0.07 (Hill, 1985). For burst analysis, the interval between bursts was defined according to the method of Magleby and Pallotta (1983) (Magleby & Pallotta, 1983). The Mann-Whitney U test was used to compare burst parameters as the number of events in a burst are described by geometric distributions (Colquhoun & Hawkes, 1981). Single events were considered as part of a burst initially but were also subsequently separated into their own group in order to compare bursts with one or more than one event.

Western blots of RyR2

50 μg of protein from mixed membrane preparations were suspended in Laemmli buffer. Samples were resolved on a 6% reducing SDS polyacrylamide gel and transferred to nitrocellulose membrane in transfer buffer with 10% methanol. Membranes were blocked for 1 h at room temperature in 5% low-fat milk in Tris-buffered saline and 0.1% Tween-20 (TBS-T). Primary antibody for RyR2 was incubated overnight at 4°C (34C, Abcam, 1:1000 dilution). Membranes were washed with TBS-T and incubated for 1 h with anti-mouse HRP-conjugated secondary antibody (GE Healthcare Lifesciences). Antibody-antigen complexes were visualised by Pierce^™ enhanced chemiluminescent substrate with a Bio-Rad ChemiDoc^™ Imaging System. For loading control, total protein was quantified per lane with a Coomassie stain. Densities of protein were quantified on ImageLab 6.1 software.

Statistical analysis

For each intervention, data were collected using at least 3 membrane preparations. Statistical analyses were carried out using GraphPad Prism 8 or MatLab 2018. Data were expressed as the mean ± SD and n ≥ 5 unless otherwise stated. A p value of < 0.05 was considered statistically significant. To compare two groups, an unpaired t-test was used. A two-way ANOVA was used to compare the mean differences between groups that were split into two independent variables (e.g. genotype or [Ca²⁺]). The p value for the main effect and the interaction is described in each figure legend. Where an interaction was found to be significant, the p value for the post-hoc multiple comparisons with Sidak’s correction was indicated on the graph above the relevant point to compare the two genotypes. Where all individual points could not be shown on the figure, the data is included as supporting information.

Molecular modelling

Homology modelling of the HD1 domain of RyR2 was performed using MODELLER (Šali & Blundell, 1993) through the Chimera (UCSF) software (Pettersen et al., 2004). The amino acid sequences of pig and human RyR2 were obtained and extracted from RCSB PDB (Berman et al., 2000) and UniProt (The UniProt Consortium, 2017) (PDB ID: 5GOA; UniProt entry: Q92736), respectively, and residues corresponding to human residues 2111–2679 (the HD1 domain, 2112–2680 in pig numbering) were used for further modelling. The HD1 domains of two WT pig RyR2 structures, open and closed states (PDB ID: 5GOA and 5GO9 (Peng et al., 2016); chain A in both cases), were used as templates for the modelling. Using Chimera, small molecules were deleted from the template files. PDB files of just the HD1 of chain A were saved, followed by alignment of the sequences of HD1 for pig, human WT as well as human with either of the two mutations. Homology models were generated using the default parameters for MODELLER in Chimera except that 50 models were generated for each round of modelling and that the “thorough optimisation” procedure was applied. The resulting HD1 homology models included a total of 300 structural models of six different constructs: human WT, the V2475F mutant and the R2474S mutant, each modelled on both open and closed state pRyR2 template structures. Two lobes were defined for HD1; lobe A (residues 2111–2473) and lobe B (residues 2476–2679), and these lobes were connected by the mutation sites. The angle between these lobes was calculated by defining two vectors, both originating from centre of mass of the mutation site with one of them going to the centre of mass of lobe A and the other one to the centre of mass of lobe B. The angle between these vectors is considered the hinge angle. One-way ANOVA with Fisher’s LSD test was used to compare the calculated hinge angles between the different constructs.

Materials

All chemicals were purchased from VWR or Sigma or as otherwise stated. All solutions were prepared in deionised water and filtered through a 0.45 μM pore diameter filter (Millipore).

Results

Single-channel recordings

Typical single-channel recordings of native RyR2 channels derived from the hearts of WT or V2475F(+/−) mice are shown in Fig. 2A, illustrating the similar fast gating kinetics and absence of resolvable sub-conductance state gating. The V2475F mutation produced a small but significant reduction in single-channel conductance when Ca²⁺ was the permeant ion as shown in Fig. 2B (Note that in (A), recordings from WT and V2475F(+/−) channels of similar conductance were used to aid comparison of gating behaviour). In contrast, no deviation in K⁺ conductance was observed (698.4 ± 49.2 pS; n = 7 and 703.0 ± 58.6 pS; n = 8; SD in symmetrical 210 mM K⁺ for WT and V2475F(+/−) channels, respectively). Under bi-ionic conditions with cytosolic 210 mM K⁺ and luminal 210 mM Ca²⁺, we also found no difference in reversal potential (see Fig. 2C), indicating that the mutation does not alter the relative Ca²⁺: K⁺ permeability ratio (PCa²⁺/PK⁺) which was 7.13 ± 0.67 and 7.15 ± 1.39 (mean ± SD; n = 5 and n = 7) for WT and V2475F(+/−) channels, respectively.

Figure 2. Typical gating and conduction properties of native RyR2 channels derived from the hearts of WT and V2475F(+/−) mice.

(A) Representative single-channel traces in Tris/HEPES, Ca²⁺ glutamate solutions (see methods), illustrating the effects of varying cytosolic [Ca²⁺] on gating and conductance for WT (left panel) and V2475F(+/−) (right panel) channels. Luminal [Ca²⁺] is approximately 50 mM and the cytosolic free [Ca²⁺] is indicated above each trace along with the average Po (measured over 3 min). O and C represent the fully open and closed levels, respectively. For comparative purposes, WT and V2475F(+/−) channel traces from channels with similar single-channel conductances are shown. (B) Single-channel conductance of WT (black) and V2475F(+/−) (green) channels in the same solutions as in (A) with cytosolic Ca²⁺ maintained at 1 μM. Bars indicate mean ± SD; unpaired t-test, p = 0.0306. (C) Single-channel current-voltage relationships with luminal 210 mM K⁺ and cytosolic 210 mM Ca²⁺. Mean ± SD, n ≥ 5, for WT (black squares) and V2475F(+/−) (green circles). The third order polynomial lines plotted for each genotype are not different, p = 0.4060. Where not shown, error bars are within the symbol.

Loaiza et al. 2013 reported that purified recombinant V2475F(+/+) channels are more sensitive to cytosolic Ca²⁺ than WT channels. However, they suggest that there is wide variability in gating when native channels, derived from the hearts of V2475F(+/−) mice, are incorporated into bilayers (Loaiza et al., 2013). It is essential to investigate this variability because V2475F CPVT1 patients are heterozygous for the mutation, expressing both WT and mutated subunits. Hence, the RyR2 tetramers of V2475 CPVT1 patients will be constructed from varying numbers of mutated monomers, combined with WT monomers (Fig. 1). We have therefore compared the cytosolic Ca²⁺ sensitivity of native cardiac RyR2 channels from WT and V2475F(+/−) mice. Fig. 2A compares the characteristic gating responses of WT and V2475F(+/−) channels to key cytosolic [Ca²⁺]. Most channels in both groups are completely closed (Po = 0) at diastolic free [Ca²⁺] (100 nM) as shown in the top traces. Importantly, 14−20% of the channels do open and we illustrate the gating properties of this critical subset of channels in subsequent figures. For V2475F(+/−) RyR2, the maximum increase in Po for the total population of channels occurs at approximately 1 μM Ca²⁺ (middle traces) and channels are almost fully inactivated by 1 mM (bottom traces). The full [Ca²⁺]-Po relationship highlights these important alterations (Fig. 3A). While the V2475F(+/−) channels are opened more readily by the increase in cytosolic [Ca²⁺] from 100 nM to 1 μM (EC₅₀ values are 0.82 μM for WT and 0.12 μM for V2475F(+/−)), they are also more readily inactivated (IC₅₀ values are 5.18 mM for WT and 0.27 mM for V2475F(+/−)). It is significant that although the V2475F(+/−) channels appear to be more sensitive to cytosolic Ca²⁺, the average peak Po that the total V2475F(+/−) channel population achieves is no higher than that of WT channels. Thus, if the overall population of V2475F(+/−) channels is considered, cytosolic Ca²⁺ appears to remain a partial agonist for this mutant RyR2.

Figure 3. Effects of the V2475F mutation on the response of RyR2 channels to cytosolic Ca²⁺.

(A) Relationship between [Ca²⁺] and Po values for RyR2 derived from WT (black) and V2475F(+/−) (green) mice. Two-way ANOVA: Ca²⁺-by-genotype interaction, p < 0.0001; main effect of Ca²⁺, p <0.0001; main effect of genotype p = 0.0325. Post-hoc multiple comparisons with Sidak’s correction comparing WT and V2475F(+/−): p < 0.0001 at 1 μM, p = 0.0291 at 1 mM Ca²⁺. Mean ± SD; n for WT = 8 – 110, n for VF = 5 – 95. Lines display non-linear bell-shaped fitted curves. EC₅₀ values are 0.82 μM for WT and 0.12 μM for V2475F, and IC₅₀ values are 5.18 mM for WT and 0.27 mM for V2475F. (B) Percentage of channels that open at 100 nM Ca²⁺ (n = 110 and n = 95 for WT and V2475F(+/–), respectively). (C) Po of single channels that open at 100 nM Ca²⁺. Only channels that do open at 100 nM Ca²⁺ are included for WT (n = 15) and V2475F(+/–) (n = 19) mean Po values ± SD. Unpaired t-test; p = 0.0046. (D) Representative recordings from WT and V2475F(+/–) mice at 100 nM cytosolic Ca²⁺. Po displayed is for the full 3 min recording. C and O indicate the closed and open states, respectively.

As shown earlier in the top traces of Fig. 2A, at 100 nM Ca²⁺, most WT (86%; n = 115) and mutant (80%; n = 95) channels are silent (Fig. 3B) (although the presence of channels in the bilayer at 100 nM Ca²⁺ was confirmed because they open when cytosolic [Ca²⁺] is raised). However, the V2475F(+/−) channels that do open at 100 nM Ca²⁺, exhibit significantly higher Po than that of WT channels (Fig. 3C). Representative single channel recordings (Fig. 3D) unmistakably demonstrate the typically longer openings of the V2475F(+/−) channels. Lifetime analysis provides further insight into this altered gating behaviour of mutant channels, however, only experiments where a single channel was present in the bilayer could be used for this analysis. The mean open times were significantly longer in V2475F(+/−) channels (26.63 ± 24.52 ms, n = 13, SD) than WT channels (10.64 ± 6.73 ms, n = 12, SD, p = 0.0021). The open and closed lifetime distributions for a typical WT and V2475F(+/−) channel are shown in Fig. 4A. The open lifetime distributions of WT channels at this low free [Ca²⁺] were always best described by one or two short time constants. In contrast, the distribution of openings of V2475F(+/−) channels was shifted such that there was a greater proportion of longer events, and the best fit to the data required at least three (as in Fig. 4A) and sometimes four time constants. The closed lifetime distributions of both WT and V2475F(+/−) channels were more variable and dependent on Po (as previously described for cytosolic Ca²⁺-regulated gating of RyR2 (Sitsapesan & Williams, 1994b)) and could usually be best described by three time constants (Fig. 4A, right). A large proportion of the V2475F(+/−) channel closings occurred to the shortest time constant, close to the minimum resolvable event duration. Looking back at the single channel recordings (Fig. 3D), it appears that these very short closings occur within bursts of openings. We therefore performed burst analysis and the key findings are summarised in Fig. 4B and Table 1. What was especially interesting was that the interburst interval was similar for WT and V2475F(+/−) channels (left bar chart), whereas there was a 6.15-fold elongation of burst duration (middle bar chart) in V2475F(+/−) channels. Those bursts with >1 openings were 6.14 times longer in V2475F(+/−) channels compared to WT channels (right bar chart). Importantly, the fact that the interburst interval was similar in WT and V2475F(+/−) channels suggests, that while Ca²⁺ must bind to the closed channel state in both WT and mutant channels to drive channel openings, the higher Po of the V2475F(+/−) channels at 100 nM Ca²⁺ is not primarily due to greater affinity of Ca²⁺ for V2475F(+/−) channel closed states compared to WT channel closed states. Instead, the main difference between V2475F(+/−) and WT channel gating at 100 nM Ca²⁺, is that openings are longer in V2475F(+/−) channels, suggesting that the affinity of Ca²⁺ for the open state is increased.

Figure 4. Effects of the V2475F mutation on the response of RyR2 channels to 100 nM cytosolic Ca²⁺.

(A) Open and closed lifetime distributions. Open (left) and closed (right) lifetime distributions and pdfs for representative single RyR2 channels derived from WT (black) or V2475F(+/−) (green) mice in the presence of 100 nM cytosolic Ca²⁺. The best fits to the data were obtained by the method of maximum likelihood and the resulting time constants (τ, ms) and percentage areas (%) are shown. (B) Burst analysis parameters comparing interburst interval (left, Mann-Whitney test; p = 0.1806), burst duration (middle, Mann-Whitney test; p = 0.0420), and burst duration with >1 event (right, Mann-Whitney test; p = 0.0312) for WT (black) and V2475F(+/−) (green) conducted on single channels at 100 nM cytosolic Ca²⁺. n = 6 for WT, n = 10 for V2475F(+/−), mean ± SD.

Table 1: Burst analysis parameters at 100 nM cytosolic Ca²⁺.

Unless otherwise stated, the parameters listed are calculated from all bursts. The p value was calculated by a Mann-Whitney test. n = 6 for WT, n = 10 for V2475F(+/−).

	WT	V2475F(+/−)
PARAMETER	MEAN ± SD	MEAN ± SD	p value
Interburst interval (ms)	1739.5 ± 1371.5	1377.8 ± 1991.1	0.1806
Burst duration (ms)	72.7 ± 82.7	447.7 ± 497.3	0.0420 *
Burst duration with >1 event (ms)	123.3 ± 118.5	757.5 ± 950.6	0.0312 *
Burst duration of 1 event (ms)	4.1 ± 3.0	4.9 ± 6.8	0.5622
Intraburst open duration (ms)	8.5 ± 3.3	14.3 ± 18.6	0.7128
Intraburst closed duration (ms)	9.6 ± 5.4	18.0 ± 16.3	0.3132
Events in burst (n)	9.4 ± 13.3	19.8 ± 17.0	0.1179
Proportion of bursts with 1 event (%)	55.0 ± 18.4	39.4 ± 14.5	0.1806

We then compared how mean open and closed times changed as cytosolic Ca²⁺ was increased from 100 nM for WT and V2475F(+/−) channels (Fig. 5A and B). For both WT and V2475F(+/−) channels, the main mechanism for the increase in Po at 1 μM Ca²⁺ was an increase in the frequency of channel opening as shown by the massive decrease in mean closed times. In V2475F(+/−) channels, this was more pronounced with a 208-fold decrease in comparison to the 136-fold decrease for WT channels (note that the very brief closings in V2475F(+/−) channels at 100 nM Ca²⁺ occur within the bursts). Ca²⁺ activation was not associated with increases in mean open time. Instead, as Fig. 5A demonstrates, the trend was for reductions in mean open time with increasing [Ca²⁺]. This trend was more obvious for V2475F(+/−) channels than for WT channels. Thus, the longest open times were observed at 100 nM Ca²⁺, and V2475F(+/−) channels were characterised by significantly longer openings than WT channels at this [Ca²⁺] as described in the previous paragraph.

Figure 5. Effects of the V2475F mutation on open and closed lifetimes.

(A) Relationships between mean open lifetimes for WT (black) and V2475F (green). Two-way ANOVA: Ca²⁺-by-genotype interaction, p = 0.0283; main effect of Ca²⁺, p <0.0001; no main effect of genotype, p = 0.5733. Post-hoc multiple comparisons with Sidak’s correction comparing WT and V2475F(+/−): p = 0.0021 at 100 nM. Mean ± SD, n ≥ 4. (B) Relationships between mean closed lifetimes for WT (black) and V2475F (green). At pCa²⁺ = 7, only channels that opened are included. Two-way ANOVA: no Ca²⁺-by-genotype interaction, p = 0.6504; no main effect of Ca²⁺, p = 0.0663; no main effect of genotype, p = 0.5504. Mean ± SD, n ≥ 4. (C) – (F): The fitting of pdfs to open and closed lifetime histograms, examples of which are shown in Fig. 4A, was performed for all experiments (where only a single channel was gating in the bilayer) across the range of cytosolic [Ca²⁺] shown in panels (A) and (B). The symbols (WT = open black circles; V2475F(+/−) = open green circles) in panels (C) – (F) represent the mean time constants (τ) of the pdfs to open (C) and closed lifetime distributions (E) and their respective % areas (D and F). The numbers within the symbols refer to the 1st (shortest), 2nd, 3rd and 4th time constants. To aid visualisation of the Ca²⁺-dependent changes, the same time constants and their areas are joined with a dashed line. Where the longest time constants do not exist in every [Ca²⁺], the points are not joined with a dashed line. Mean ± SD, n ≥ 4 (except for long open 3rd or 4th components which did not occur in all channels and where n ≥ 2). The entire raw data set is provided in the Supporting Information.

As for 100 nM Ca²⁺ (Fig. 4A), we performed lifetime analysis across the range of [Ca²⁺] for all experiments where just a single channel was present in the bilayer. The pdfs that best fit the open and closed lifetime distributions were calculated for each channel, and the mean values for the open and closed time constants and percentage areas are shown in Fig. 5C, D, E and F. The trends for decreasing open time constants with increasing cytosolic [Ca²⁺] can be clearly seen for both WT and V2475F(+/−) channels (Fig. 5C) and the two shortest open time constants (1 and 2) are similar for WT and V2475F(+/−) channels. 1 μM Ca²⁺ gives the peak Po for V2475F(+/−) channels with the channels still gating into the long third and fourth open states. In contrast, Po does not peak for WT channels until cytosolic Ca²⁺ is in the 10–100 μM range and, again, the WT channels also gate into the longer third and fourth open states. Thus, these longer open states appear to be important for enabling the channels to open with maximum effectiveness. Comparison of the closed lifetime distributions (Fig. 5E and F) shows that as Ca²⁺ is raised from 100 nM to 1 μM, the time constants get shorter for both WT and mutant channels (Fig. 5E) but additionally, mutant channels only transition between the two shortest closed states (time constants 1 and 2) whereas WT channels can still enter into the longer third closed state (time constant 3). Fig. 5F demonstrates that at 1 μM Ca²⁺, 95% of closings in mutant channels are to the shortest closed state whereas in WT channels, only approximately 80% of closings are to the shortest closed state (Fig. 5F). This greater increase in frequency of opening of the mutant channels is the major factor contributing to their higher Po at 1 μM Ca²⁺ (note the log y-axis scale for open (Fig. 5C) and closed (Fig. 5E) lifetime constants as open lifetime constants do not change by the same magnitude). These gating changes cause V2475F(+/−) channels to transition rapidly between open and closed states at peak Po levels when the cytosolic [Ca²⁺] is 1 μM, as can be seen in the representative trace in Fig. 2A.

Fig. 3A also highlights that Ca²⁺-inactivation is greatly affected by the V2457F mutation, causing a 19-fold increase in affinity of Ca²⁺ for inactivation sites evidenced by the shift in IC₅₀ from 5.18 mM for WT to 269 μM for V2475F(+/−) channels.

Fig. 5A & B demonstrates that for V2475F(+/−) channels, mean open time decreases when cytosolic Ca²⁺ is raised above 1 μM, chiefly because the channel no longer gates into the two longest open states (Fig. 5C). The primary mechanism for inactivation, however, is the increase in the duration of closings (Fig. 5B) that clearly takes place as Ca²⁺ is raised above 10 μM. As shown in Fig. 5E and F for V2475F(+/−) channels at 1 mM Ca²⁺, the dominance of brief closings is lost and the V2475F(+/−) channels also gate frequently into the longer closed states (time constants 2, 3 and 4). In contrast, inactivation of WT channels takes place at much higher concentrations of cytosolic Ca²⁺. At 10 μM Ca²⁺, the long open states (time constants 3 and 4, Fig. 5C) are still present but disappear by 1 mM Ca²⁺ when the shortest open state is the most frequently visited (>90% of openings) (Fig. 5D). Fig. 5B suggests that by 10 mM Ca²⁺, mean closed time is starting to increase in WT channels but because the current amplitude is greatly reduced under these conditions (luminal to cytosolic driving force is reduced), the resolution of brief events is lost and so we cannot perform accurate lifetime analysis to examine this properly.

CPVT1 is a disease caused by mutations to RyR2 that manifests as an autosomal-dominant (heterozygous) disease in patients (Priori & Chen, 2011). This suggests that if all four monomers of every RyR2 molecule contained the mutation, the functional change would be too severe overall and, certainly, our V2475F(+/−) mouse model of CPVT1 is homozygous lethal (Loaiza et al., 2013). The mouse model therefore mimics the human disease and our cardiac membrane preparations will contain populations of channels that have varying numbers of mutated monomers (Fig. 1). CPVT mutations can sometimes lead to a reduction in the expression levels of RyR2 protein (for example, (Zhong et al., 2021)) and so we examined V2475F(+/−) RyR2 levels by immunoblot analysis (Fig. 6A and B). The V2475F mutation did not cause a reduction in RyR2 levels suggesting that all possible combinations of mutated monomers shown in Fig. 1 may be viable ion-channels. We have collected extremely large Po data sets and so we plotted Po frequency histograms for WT and V2475F(+/−) channels at the cytosolic [Ca²⁺] of 1 μM (Fig. 6C and D). The distribution of WT channels is clearly weighted towards lower Po values which is not the case for V2475F(+/−) channels. We used the method of kernel density estimation (KDE) to highlight the possible Po populations of the WT and V2475F channels. (Fig. 6E). The main peak for WT channels data was at 0.07. For V2475F(+/−) mutated channels, the KDE suggests the presence of at least four different populations of channels identified by four peak values of 0.08, 0.34, 0.59 and 0.87. Examples of the type of gating exhibited by the V2475F(+/−) channels at these activity levels are shown in Fig. 6F. For V2475F(+/−) channels, the lowest Po component (0.08) occurs at a similar Po to that of the WT channels but three other higher Po components make up the bulk of the distribution.

Figure 6. Effects of the V2475F mutation on expression levels of RyR2 and on the population frequency of high Po RyR2 channels in mouse cardiac muscle.

(A) Representative western blot of RyR2 derived from WT or V2475F(+/−) hearts. (B) Band intensity of RyR2 normalised to total protein per lane comparing RyR2 derived from WT or V2475F(+/−) hearts (mean ± SD). Each point represents a sample from an independent mixed membrane preparation (n = 5). Unpaired t-test; p = 0.9519. Po values from individual experiments were plotted as frequency distributions for (C) WT (n = 59) and (D) V2475F(+/−) (n = 62) experiments in the presence of 1 μM cytosolic Ca²⁺. (E) A Kernel Density Estimate (KDE) was fit to the WT and V2475F(+/−) Po values. For the V2475F(+/−) Po values, the KDE estimated 4 peaks and therefore suggests the presence of at least four different populations of activity at Po values of 0.08, 0.34, 0.59 and 0.87. (F) The gating behaviour of typical V2475F(+/−) channels drawn from these populations peaks are shown in the four traces labelled for the corresponding component.

RyR2 is a ligand-gated channel modulated by many soluble ligands in addition to cytosolic Ca²⁺. To ascertain whether the mutation specifically altered cytosolic Ca²⁺ regulation of RyR2 or whether other aspects of gating were also dysfunctional, we examined the response of the channel to the three other key physiological regulators of RyR2 activity: adenine-based ligands, luminal Ca²⁺ and cytosolic Mg²⁺. ATP was very effective at activating channels from the V2475F(+/−) mice in the presence of 1 μM cytosolic Ca²⁺ and Fig. 7A indicates the expected action of ATP. We also investigated the effects of adenosine. This ligand is less effective than ATP at activating RyR2 at the adenine nucleotide binding site because the triphosphate group is important for inducing long open states (Kermode et al., 1998; Chan et al., 2000; Lindsay et al., 2018). However, unlike ATP, adenosine will not induce phosphorylation and does not chelate Ca²⁺, and so is a useful test of activity that is caused purely via the RyR2 adenine nucleotide binding site. Again, as shown in Fig. 7B, adenosine activated the V2475F(+/−) channels in a manner expected for a ligand acting at the RyR2 adenine nucleotide binding site and was able to increase Po above control values by a similar increment for both WT and V2475F(+/−) channels. Compared to the increment caused by ATP, adenosine increased Po by 81% and 87% in the WT and V2475F(+/−) channels, respectively.

Figure 7. ATP and adenosine activation of RyR2 from WT or V2475F(+/−) mouse hearts.

(A) shows the activation of WT (black) and V2475F (green) channels by ATP. Two-way ANOVA: no ATP-by-genotype interaction, p = 0.8203; main effect of ATP, p = 0.0002; main effect of genotype, p = 0.0226. Mean ± SD, n = 7 for WT and n = 5 for V2475F. (B) shows the activation of WT (black) and V2475F (green) channels by adenosine. Two-way ANOVA: no adenosine-by-genotype interaction, p = 0.9379; main effect of Ca²⁺, p < 0.0001; main effect of genotype, p < 0.0001. Mean ± SD; n = 21 for WT and n = 33 for V2475F. In both cases, control channels were activated solely by 1 μM cytosolic Ca²⁺ and the free [Ca²⁺] was maintained at 1 μM after ATP or adenosine addition by buffering with EGTA where required.

Luminal Ca²⁺ regulation of V2475F channels was investigated using K⁺ as the permeant ion with increasing permeation by Ca²⁺ as the luminal [Ca²⁺] was raised. Recordings were made at the holding potential of −30 mV so that current flowed in the physiological luminal to cytosolic direction. We performed the experiments in the presence of 1 μM free cytosolic Ca²⁺ and 1 mM ATP as we have previously shown that luminal Ca²⁺ alone does not activate RyR2 over physiological luminal [Ca²⁺] and that an activator in addition to cytosolic Ca²⁺ is required for maximum response (Sitsapesan & Williams, 1994a, 1997; Lukyanenko et al., 1996). Fig. 8A shows typical recordings to illustrate the potentiating effects of luminal Ca²⁺ on WT (left) and V2475F(+/−) (right) single-channel gating when Ca²⁺ is increased from 10 μM free luminal Ca²⁺ (top traces) to 1 mM free luminal Ca²⁺. The luminal Ca²⁺-induced increments in Po are similar to those observed with sheep cardiac RyR2 under these experimental conditions (Ching et al., 2000). Fig. 8B illustrates the mean data showing the relationships between luminal [Ca²⁺] and Po. It can be seen that while both groups of channels are stimulated by raising luminal Ca²⁺, there is only a 2.18 fold increase in Po for V2475F(+/−) channels when luminal Ca²⁺ is raised from 10 μM Ca²⁺ to 1 mM in comparison to a 4.48-fold increase for WT channels.

Figure 8. Luminal Ca²⁺ regulation of WT and V2475F(+/−) channel gating.

(A) Representative recordings of WT and V2475F(+/−) channels in symmetrical 210 mM K⁺ solutions in the presence of cytosolic activators 1 μM Ca²⁺ and 1 mM ATP at the indicated luminal [Ca²⁺]. The holding potential was −30 mV and current flow was in the luminal to cytosolic direction. C indicates the closed channel level and O1, O2 and O3 illustrate the levels where one, two and three channels, respectively, are open. (B) Relationship between Po and luminal [Ca²⁺] for WT and V2475F(+/−) channels activated by cytosolic 1 μM Ca²⁺ and 1 mM ATP. Mean ± SD, n ≥ 5. Two-way ANOVA: no Ca²⁺-by-genotype interaction, p = 0.8766; main effect of Ca²⁺, p = 0.0006 (indicated on the figure); no main effect of genotype p = 0.1476.

The effects of the V2475F(+/−) mutation on Mg²⁺ inhibition are shown in Fig. 9. The addition of 500 μM cytosolic free Mg²⁺ to WT channels activated by 1 μM free Ca²⁺ and 1 mM adenosine, consistently decreased Po by >80%, causing reductions in both the duration and frequency of openings. This typical pronounced inhibition of WT channel gating by 500 μM cytosolic Mg²⁺ is shown in Fig. 9A (left recordings). Under identical experimental conditions, Mg²⁺ exerted a variable and much weaker inhibitory action on the gating of V2475F(+/−) channels and, as shown in Fig. 9A (right recordings), some channels could still open frequently and with long duration. The average reduction in Po for V2475F(+/−) channels was only 64%. The relationship between Po and free cytosolic [Mg²⁺] yielded an approximate IC₅₀ of 86 μM for WT channels (Fig. 9B). However, the variable effect of Mg²⁺ on V2475F(+/−) channels did not allow us to compute an accurate IC₅₀ value for these mutant channels as can be seen from the large error bars on the plotted points. On closer inspection of the individual V2475F(+/−) channel recordings, there appeared to be at least three types of response to Mg²⁺: 1. Those channels that were inhibited by >80% (4 of 13 channels) and thus appeared similar to WT channels in their response. 2. Those channels that were inhibited by <80% (6 of 13 channels) and 3. Those channels that were activated by 500 μM or 1 mM Mg²⁺ (3 of 13 channels). Fig. 9C illustrates the mean values of Po within these groups and Fig. 9D shows representative gating of V2475F(+/−) channels within the three groups.

Figure 9. Effect of cytosolic Mg²⁺ on the gating of WT and V2475F channels.

(A) Recordings from WT (left) and V2475F (right) channels before and after adding 500 μM free cytosolic [Mg²⁺] in the presence of 1 mM adenosine and a maintained 1 μM free cytosolic [Ca²⁺] (control). In order to directly compare the ability of Mg²⁺ to inhibit WT and V2475F channels, recordings from channels with a similar Po under control conditions were chosen. O and C indicate the open and closed states, respectively. (B) Relationship between Po and free [Mg²⁺] for WT (black) and V2475F (green) channels in the presence of 1 μM free Ca²⁺ and 1 mM adenosine. Mean ± SD, n ≥ 5. (C) Left of the dashed line: comparison of the percentage inhibition of Po induced by 500 μM Mg²⁺ in all WT (black) and all V2475F (green) channels. Unpaired t-test, p = 0.0255. Right of the dashed line: percentage change in Po of V2475F channels after the channels are separated into 3 groups: those channels inhibited to within the range of inhibition expected for WT channels (>80% inhibition, n = 4), those channels inhibited to a reduced or intermediate extent (<80% inhibition, n = 6) and those channels that were activated (increase in Po, n = 3) by Mg²⁺. Mean ± SD, n numbers for each group are indicated on the bars. (D) Typical examples of the gating behaviour of V2475F channels showing WT-like inhibition (top trace), intermediate inhibition (middle trace) and activation (bottom trace) in response to 500 μM Mg²⁺.

Alterations to the effects of cytosolic Mg²⁺ are not totally unexpected since Ca²⁺ regulation is altered and Mg²⁺ is known to compete with Ca²⁺ for both activation and inhibition sites (Meissner et al., 1986; Ashley & Williams, 1990; Laver et al., 1997). The affinity of Mg²⁺ for inactivation sites appears reduced, but interestingly, in a sub-population of channels, the efficacy of Mg²⁺, once bound, is also affected. Instead of acting purely as an antagonist of RyR2, as occurs with WT channels (Fig. 9), there is a change in the interaction of Mg²⁺ with the V2475F(+/−) channels such that Mg²⁺ behaves as an agonist in 23% of experiments. The greatest proportion of V2475F(+/−) channels (46%) respond to Mg²⁺ with ‘intermediate’ inhibition and in these experiments it appears that the Mg²⁺ affinity of the channel is reduced. It is interesting that the proportion of V2475F(+/−) channels that respond to Mg²⁺ in the same manner as WT channels (i.e. ‘full’ inhibition) is 31% which is equivalent to the predicted proportion of V2475F(+/−) RyR2 tetramers that contain zero or one mutated monomer (Fig. 12B). It is therefore tempting to speculate that the degree of Mg²⁺ dysfunction is related to the number of mutated monomers such that if two monomers are mutated, then Mg²⁺ becomes less effective as an antagonist but when three or four monomers are mutated, Mg²⁺ can even induce channel activation. Overall, 69% (46% + 23%) of all V2475F(+/−) channels exhibited an altered response to Mg²⁺ which would be expected if this behaviour required the presence of two or more mutated monomers per tetramer (37.5% + 25% + 6.25% = 68.75%).

Figure 12. Likelihood of formation of RyR2 tetramers with different compositions of V2475F mutated monomers.

(A) White and green (F) circles represent WT and V2475F(+/−) mutated monomers, respectively. Assuming random assembly and identical surface expression, the probabilities of tetramers with the indicated numbers of mutant or WT subunits per tetrameric RyR2 channel is indicated as a percentage. (B) Data from Fig. 9, indicating the percentage of V2475F(+/–) channels exhibiting full inhibition (31%), intermediate inhibition (46%) or activation (23%) in response to cytosolic Mg²⁺. (C) Data from Fig. 3, displaying (i) a frequency histogram for Po at 1 mM cytosolic Ca²⁺ and (ii) indicating the percentage of V2475F(+/–) channels exhibiting high sensitivity to inhibition by 1 mM cytosolic Ca²⁺. (Po< 0.1 in the presence of 1 mM cytosolic Ca²⁺).

Mg²⁺ and Ca²⁺ are thought to inhibit RyR2 through common binding sites. Thus, it might be expected that Ca²⁺ inhibition of V2475F(+/−) would also be affected in the same population of channels that exhibit reduced Mg²⁺ inhibition. We therefore compared the frequency of Po values obtained in WT and V2475F(+/−) channels at 1 mM Ca²⁺ (Fig. 12Ci). It can be seen in Fig. 12Cii that whereas no WT channels (n = 10) exhibited a Po < 0.1 at 1 mM Ca²⁺, 73% (8 of 11 channels) of V2475F(+/−) channels gated with Po < 0.1. Thus, 73% of V2475F(+/−) channels responded in an abnormal manner to 1 mM Ca²⁺, a percentage that is very similar to the percentage of channels that respond abnormally to Mg²⁺ (68.74%), and also similar to the percentage of V2475F(+/−) RyR2 channels expected to be expressed with two or more mutated monomers (69%). This is summarised in Fig. 12B and suggests that if only one monomer is mutated, divalent cation inhibition of RyR2 is not altered.

Molecular modelling

The single point V2475F mutation caused marked functional changes despite the fact that structurally, it seems like a relatively slight change from a smaller to a larger hydrophobic sidechain. To explore whether structural models could point towards an explanation for these changes, we constructed homology models of the RyR2 HD1 in which the mutation is found. Another CPVT1 gain-of-function mutation is located next to V2475F, namely R2474S (Lehnart et al., 2008). We speculated that, given the proximity of these mutations, perhaps they would have similar structural effects, and for this reason we also constructed models of the R2474S single point mutant. Fig. 10 illustrates the locations of these mutations within HD1 of RyR2.

Figure 10. The location of the CPVT1 mutations R2474S and V2475F within the RyR2 structure.

(A) Domain organisation of RyR2 and, below, side view (left) and cytosolic view (right) of the structure. (B) Zoom-in view of the HD1 domain indicating the location of R2474 and V2475. (C) Sequence alignment including the pig WT sequence corresponding to the structure (human sequence is identical in this region) and the R2474S and V2475F mutated sequences. The mutation sites are highlighted within the blue box and the secondary structure associated with the WT sequence is highlighted in yellow.

We used structures of the pig RyR2 in open and closed states as templates for our homology modelling (PDB ID: 5GOA, 5GO9 (Peng et al., 2016) and our models were restricted to HD1 (residues 2111 to 2679). Models for human WT, human R2474S and human V2475F were constructed for both the open and the closed state. 50 models were generated for each of these six constructs, and the five best models from each group, based on their MODELLER scores, were chosen for analysis. Given the homology modelling approach applied, we did not expect major conformational changes, although the ‘thorough optimization’ protocol used does allow for some flexibility. However, when comparing the models, we noticed what appeared to be different hinge motions in the HD1 (Fig. 11). To analyse these differences, we defined two lobes of HD1 (Fig. 11A), linked by the mutation sites, and calculated the hinge angle between these lobes as described in the methods section. Interestingly, this hinge angle was significantly decreased for both V2475F and R2474S mutations relative to the WT models and was especially decreased when the channel was in the open state. The mean data are shown in Fig. 11B and Fig. 11C shows a cartoon illustrating the structural change.

Figure 11. Modelling the effects of mutating R2474S and V2475F on the HD1 domain.

(A) The HD1 domain is shown to demonstrate the approximate location of the R2474S and V2475F mutations (blue circle) and how the hinge angle was defined. The angle was calculated between the two vectors originating from the mutation site by dividing the HD1 domain into two parts, lobe A (residues 2111–2473) and lobe B (residues 2476–2679), connected by the mutation sites. The centre of mass (yellow circles) was calculated for each lobe as well as for the mutation site. (B) Bar chart illustrating the effects of the V2475F (VF) and R2474S (RS) mutations on the hinge angle. One-way ANOVA, p = 0.0002. Post hoc multiple comparisons with Tukey’s correction comparing: WT and R2474S in open state p = 0.0003; WT and V2475F in open state p = 0.0010. Mean ± SD, n = 5. (C) Cartoon illustrating the bending of the RyR2 HD1 domain and reduced hinge angles of the mutants.

Discussion

We find that the V2475F mutation impairs ligand regulation of RyR2 in addition to altering the Ca²⁺ ion flux through the pore. The damage to function is severe yet specific. Homology modelling proposes that these detrimental functional changes arise because of the critical location of valine 2475 at a hinge between two regions of the HD1 domain. Our HD1 models show that both the V2475F and the neighbouring CPVT1 mutation, R2474S (Jiang et al., 2005; Lehnart et al., 2008), reduce the angle between the two regions. From the structures of pig RyR2 in open and closed states, it has previously been suggested that the motions of the helical domain and the central domain (including the pore region) are coupled and, therefore, that shifts in the helical domain may be translated into changes in the central domain (Peng et al., 2016). Thus, despite the distant location of the HD1 “hinge” region harbouring the mutation site from the affected ligand binding sites and pore region of the channel, it seems likely that the structural change proposed by our models is effectively transmitted to these regions in a manner that produces the unique alterations to conduction and gating that we observe.

By individually examining the effects of critical physiological regulators of RyR2, cytosolic and luminal Ca²⁺, cytosolic Mg²⁺ and adenine-based ligands, we have isolated the major gating defects as specific to the interactions of cytosolic Ca²⁺ and Mg²⁺ with the channel. A decrease in single-channel Ca²⁺ conductance was evident, indicating that modifications within the conduction pathway may take place. However, there was no significant change in K⁺ conductance or in the permeability of RyR2 to Ca²⁺ relative to K⁺, suggesting that alterations to SR Ca²⁺ flux during EC-coupling may be minimal.

Our results demonstrate that a particularly harmful change to the mechanism of Ca²⁺ regulation of V2475F(+/−) channels operates at the cytosolic free [Ca²⁺] of 100 nM and we suggest that this will be key to fuelling the delayed afterdepolarisations (DADs) that are characteristic of this CPVT1 mutation (Loaiza et al., 2013). Most WT (86%) and V2475F (80%) channels remain closed at 100 nM Ca²⁺. However, the crucial mechanistic change to gating is that any V2475F(+/−) channels that open do so with much longer open times than WT channels. It is well documented that cytosolic Ca²⁺ activates normal native RyR2 channels primarily by binding to closed states of the channel to increase the frequency of opening (Ashley & Williams, 1990; Sitsapesan & Williams, 1994b). However, at 100 nM cytosolic Ca²⁺, the main gating change to V2475F(+/−) channels is elongation of individual openings and bursts of openings rather than an increase in frequency of opening. Thus, at this low [Ca²⁺] (100 nM), the main effect of the V2475F(+/−) mutation is not to increase the on-rate of Ca²⁺ binding to closed channel states and thereby ‘destabilise’ the closed channel state, a phrase often coined as a global description of the effects of CPVT mutations. Instead, the rate of closing is reduced in the Ca²⁺-bound V2475F(+/−) channels causing prolonged open states.

As the [Ca²⁺] is increased, deviation from normal Ca²⁺-induced RyR2 gating regulation is further extended. At 1 μM Ca²⁺, as expected, the main mechanism for the increase in Po of WT channels is the increase in frequency of channel opening. For V2475F(+/−) channels, this is also true but the effect is greater, bringing the mean closed lifetimes down to only 1.96 ± 2.18 ms (SD; n = 13) and thus causing the significantly higher overall Po at 1 μM than occurs in WT channels. The Po frequency histograms and KDEs shown in Fig. 6 demonstrate that at this [Ca²⁺], a large proportion of the total population of RyR2 channels from WT mouse hearts have a low Po with a KDE Po peak of 0.07. In contrast, the Po values of RyR2 channels derived from V2475F(+/−) hearts fall into four distinct peaks indicating that there are at least three additional groups of channels with higher Po.

We suggest that these additional Po peaks arise due to the presence of different numbers of mutated monomers within in RyR2 tetramer. RyR2 clusters in the heart vary in size up to 100 RyR2 per cluster (Franzini-Armstrong et al., 1999; Asghari et al., 2020). It seems likely therefore, that Ca²⁺-release from a cluster, especially from the larger clusters, will predominantly reflect the activity of the RyR2 channels of intermediate phenotype. Thus, those channels with either no mutated monomers or 4 mutated monomers may exert little effect overall on Ca²⁺-release. These data highlight the necessity of using native channels from heterozygous tissue to study the effect that a CPVT1 mutation will exert in the heart.

It is noteworthy that Ca²⁺-induced inactivation starts to kick in for V2475F(+/−) channels even at 1 μM Ca²⁺, as evidenced by the sharp decrease in mean open times at 1 μM Ca²⁺ in addition to the gradual decrease in frequency of opening as [Ca²⁺] further increases (whereas inactivation of WT channels only starts to be observed at approximately 1 mM). Thus, the increased affinity of Ca²⁺ for V2475F(+/−) channel inactivation sites ensures that the channel transitions only briefly to the open state. Overall, these changes in Ca²⁺ regulation can be summed up by saying that the V2475F mutation seems to cause an intrinsic change in gating such that the open channel state is more stable at very low [Ca²⁺] than WT channels. In addition, although the actual mechanisms by which Ca²⁺ activates and inactivates mutant channels is similar to those of WT channels, the affinity of Ca²⁺ for both activation and inactivation sites is increased.

The second major impairment to ligand regulation of V2475F(+/−) channels relates to Mg²⁺ inhibition. This is the first time, in single-channel experiments, that Mg²⁺ has been shown, not only to exhibit reduced ability to close RyR2 channels, but to behave as an agonist and activate RyR2. These changes in Mg²⁺ regulation effectively lift the brake on Ca²⁺ activation during diastole to enhance the probability of RyR2 opening to elicit DADs. For CPVT1, this is an important step forward in understanding the mechanisms driving the arrhythmic tendencies of this disease. It is notable also that the V2475F mutation is not the only CPVT1 mutation to be associated with altered Mg²⁺ regulation. Using recombinantly expressed channels, the P2328S, Q4201R, and V4653F mutations were shown to reduce the inhibitory ability of Mg²⁺ (Lehnart et al., 2004) in single-channel experiments, while the R2474S and K4750Q mutations decreased the ability of Mg²⁺ to inhibit ryanodine binding (Uehara et al., 2017). Regarding the R2474S mutation; as arginine2474 neighbours valine2475 within the HD1 domain, and homology modelling predicts similar structural changes when these two residues are mutated to serine or phenylalanine (Fig. 11), respectively, we would anticipate that both of these CPVT1 mutations would produce similar defects in Mg²⁺ regulation of RyR2 channels.

It has been reported that the V2475F mutation produces heightened sensitivity to the activating effects of luminal Ca²⁺ (Loaiza et al., 2013), however, this was not observed in the current experiments. Both our WT and V2475F(+/−) channels were highly sensitive to micromolar concentrations of luminal Ca²⁺, exhibiting sensitivity similar to that of native channels from sheep hearts (Sitsapesan & Williams, 1997; Ching et al., 2000). In contrast, Loaiza et al. (2013) found that WT channels were not stimulated by luminal Ca²⁺ until concentrations exceeding 1 mM were reached (Loaiza et al., 2013). Our different observations could be due to ATP being additionally present in our experiments as we have previously shown that the potentiating effects of luminal Ca²⁺ are minimal in the absence of another activating agent apart from cytosolic Ca²⁺ (Sitsapesan & Williams, 1994a). The reduced sensitivity to luminal Ca²⁺ could also be due to the fact that Loaiza et al. (2013) used recombinant channels for these experiments. Since use of recombinant channels ensures that all monomers of all RyR2 tetramers contain the mutation, the data provide information concerning only 6.25% of channels expressed in V2475F(+/−) hearts and these channels are likely to represent the most extreme deviation from normal function. We therefore suggest that the response of V2475F(+/−) channels to luminal Ca²⁺ that we observed, over the range of luminal [Ca²⁺] likely to occur physiologically during the diastolic and systolic phases of the cardiac cycle, may be more representative of the changes that operate in the hearts of patients with CPVT1.

Our study highlights the need to characterise the function of native RyR2 channels derived from cardiac muscle in order to understand the phenotype of CPVT1 that emerges in patients due to the heterozygous nature of this disease. We have demonstrated significant differences between the function of recombinantly-expressed channels described previously (Loaiza et al., 2013), where every RyR2 tetramer harbours the single-point mutation in every monomer, and that of native cardiac RyR2 channels where binomial distribution predicts that only 1 in 16 RyR2 channels expressed in the heart will contain four mutated monomers. In fact, there could be even fewer than this if the mutation affects oligomerisation, membrane trafficking or the structural stability of the tetrameric protein. In agreement with our current work, Loaiza et al. (2013) noted the variable nature of gating when RyR2 from the hearts of V2475F(+/−) mice were incorporated into bilayers. In another study, Li & Chen (2001) found that heterologous co-expression of WT and CPVT1-E3987A mutated channels gave rise to channels with Po values intermediate to those of purely WT or E3987A mutated channels (Li & Chen, 2001). A necessary drawback of our experiments is that we have removed the RyR2 channels from their cellular milieu and so the combined changing influences of every interacting protein and post-translational modification is absent. Thus, the total cellular CPVT1 phenotype cannot be represented in the bilayer chamber. However, by isolating RyR2 channels in a bilayer, we have been able to exactly control the biochemical and ionic environment of the channels so that, step-by-step, we have examined the response of the channel to specific ligands in a bimolecular fashion, something that is impossible in a cell. Our results already provide information that is helpful for understanding why this particular RyR2 mutation leads to CPVT1 but they will also provide a framework for future work examining the effects of other changes to RyR2 that can happen in CPVT such as phosphorylation or the loss/gain of a binding protein.

In summary, our homology models propose that the CPVT1-V2475F mutation causes structural deformation to the HD1 domain due to the position of the mutation at the hinge region. HD1 has been suggested to be capable of transmitting signals to distant regions of the channel (Peng et al., 2016) and this is confirmed by our experiments, demonstrating that highly specific changes in gating and ion conduction result from the V2475F mutation. We suggest that these gating changes would destabilise normal mechanisms of EC-coupling and increase episodes of excess SR Ca²⁺ leak to induce life-threatening arrhythmias in the manner suggested in Fig. 13. Briefly, in a V2475F CVPT1 patient who is not experiencing physical or emotional stress (Fig. 13A), our data predicts that there will be few cardiac RyR2 channels opening (20%) during diastole when free cytosolic Ca²⁺ is 100 nM (Fig. 3B) and that those openings will occur with low frequency as in people without CPVT1. The diastolic RyR2 openings will be of longer duration (Fig. 3D) in V2475F CPVT1 patients but the resultant increase in SR Ca²⁺ release may be transient (Kashimura et al., 2010) as more Ca²⁺ is extruded from the cell and SR content lowered. These important mechanisms compensating for changes in RyR2 Po (Díaz et al., 2005) enable net SR Ca²⁺ release to fall towards normal levels, thus preventing the generation of DADs.

Figure 13. Model of the changes in diastolic SR Ca²⁺ release in V2475F(+/–) cardiac muscle cells.

1. No stress Similar low numbers (14–20%) of WT or V2475F(+/−) RyR2 are activated during diastole at 100 nM cytosolic Ca²⁺ (Fig. 3B). V2475F(+/−) RyR2 channels open for longer durations (Fig. 3D) but it is expected that the increase in SR Ca²⁺ release will be transient (Kashimura et al., 2010) as SR Ca²⁺ content will be lowered until equilibrium is achieved and net SR Ca²⁺ release reverts back to WT levels. Thus, DADs are not triggered.

2. Stress Adrenergic stimulation activates SERCA, causing increased SR Ca²⁺ content (Kashimura et al., 2010) in both WT and V2475F(+/−) cells. SR Ca²⁺ binds to sites on the luminal side of RyR2 and potentiates the effects of ligands activating RyR2 from the cytosolic side (Sitsapesan & Williams, 1994a). At 100 nM Ca²⁺, WT channels open with very low Po, and the potentiating effects of luminal Ca²⁺ are not significant until cytosolic Ca²⁺ is increased when Ca²⁺ enters through L-type Ca²⁺ channels during the action potential. In contrast, at 100 nM Ca²⁺, V2475F(+/−) channels are already gating into long open events (Fig. 3D). In addition to this defect, V2475F(+/−) RyR2 are also exquisitely sensitive to slight increases in cytosolic Ca²⁺ (EC₅₀ = 120 nM) (Fig. 3A) and a large proportion of V2475F(+/−) channels are insensitive to the inhibiting effects of the millimolar cytosolic Mg²⁺ present in cardiac cells (Fig. 9). It is therefore likely that the increased SR [Ca²⁺] will potentiate the heightened stimulatory effects of cytosolic Ca²⁺ and Mg²⁺ to increase SR Ca²⁺ release causing Ca²⁺ waves and DADs. In spite of the increased number of spontaneous Ca²⁺-release events in the V2475F(+/−) cardiac cells, there is no evidence that the SR Ca²⁺ content of the V2475F(+/−) cells is significantly lower than that of WT cells as evidenced by caffeine-induced SR Ca²⁺ release experiments by Loaiza et al. (2013) (Fig. 6C).

The situation is different during periods of adrenergic stress (Fig. 13B) because phosphorylation of phospholamban stimulates sarcoendoplasmic reticulum Ca²⁺ ATP-ase (SERCA) activity (Tada et al., 1974) leading to increased SR Ca²⁺ content (Kashimura et al., 2010). The increased [Ca²⁺] in the SR increases not only the Po of RyR2 (Sitsapesan & Williams, 1994a, 1997; Lukyanenko et al., 1996) but will also increase the Ca²⁺ current (Tinker et al., 1993) through each open channel. In physiological conditions, the amount of Ca²⁺-induced- Ca²⁺ released during EC-coupling is related to the amount of Ca²⁺ entering the cell during the action potential and the SR Ca²⁺ content (Bassani et al., 1995). The sensitivity of RyR2 to cytosolic activators is such that during diastole, when cytosolic Ca²⁺ falls to low levels, the openings of RyR2 are brief and rare and the SERCA-driven increase in SR Ca²⁺ content does not potentiate channel activity to the extent required to elicit DADs. In contrast to this normal physiological regulation of RyR2, Figs. 3 & 9 demonstrate the pathological changes to RyR2 gating that occur in channels derived from V2475F(+/−) hearts. Not only do V2475F(+/−) channels gate into longer open states at 100 nM Ca²⁺ (Fig. 3D), but they are also extremely sensitive to slight increases in cytosolic Ca²⁺ (EC₅₀ = 120 nM) (Fig. 3A) and are not inhibited effectively by cytosolic Mg²⁺ (Fig. 9). As suggested in Fig. 13, this increased sensitivity of RyR2 to cytosolic activators is likely to be potentiated by the adrenergic-induced increase in SR [Ca²⁺], so that Ca²⁺ waves and DADs result. Where more Ca²⁺ is released from the SR in cardiac cells from V2475F(+/−) hearts than in cells from WT hearts, the SR content might be expected to be lower after adrenergic stimulation. However, Loaiza et al. (2013) did not find evidence for this (Fig. 6C). SERCA was still able to compensate for the increase leak and SR content was not lowered below the threshold required for spontaneous Ca²⁺ release events.

Our study has exposed a number of mutation-driven changes to RyR2 function that were unexpected given the general current understanding of CPVT1. As CPVT1 can result from many different RyR2 mutations, determining the molecular changes to RyR2 ion-channel behaviour as we have done, provides the first step towards specialised treatments for patients with this severe cardiac disease.

Key Points summary.

1. RyR2 mutations can cause type-1 catecholaminergic polymorphic ventricular tachycardia (CPVT1), a lethal, autosomal-dominant arrhythmic disease. However, the changes in RyR2 ion-channel function that result from the many different patient mutations are rarely investigated in detail and often only recombinant RyR2, homozygous for the mutation, are studied. As CPVT1 is a heterozygous disease and the tetrameric RyR2 channels expressed in the heart will contain varying numbers of mutated monomers, we have investigated the range of RyR2 single-channel abnormalities found in the hearts of mice heterozygous for the CPVT1 mutation, V2475F(+/−)-RyR2.

2. Specific alterations to ligand regulation of V2475F(+/−)-RyR2 were observed. Multiple sub-populations of channels exhibited varying degrees of abnormality. In particular, an increased sensitivity to activating and inactivating cytosolic [Ca²⁺], and reduced sensitivity to Mg²⁺ inhibition were evident.

3. Our results provide mechanistic insight into the changes to RyR2 gating that destabilise SR Ca²⁺-release causing life-threatening arrhythmias in V2475F(+/−)-CPVT1 patients.

Data Availability Statement

All data is included in figures apart from Fig 3A and Fig 5C–F in which the entire raw data set is included in the Supporting Information file.