Arrhythmia formation in subclinical (“silent”) long QT syndrome requires multiple insults: Quantitative mechanistic study using the KCNQ1 mutation Q357R as example

Abstract

BACKGROUND

In subclinical or silent long QT syndrome, the QT interval is normal under basal conditions. The hypothesis that insults to the repolarization reserve may cause arrhythmias in silent mutation carriers but not in noncarriers has been proposed as a general principle, yet crucial aspects remain descriptive, lacking quantification.

OBJECTIVE

To utilize accurate mathematical models of the human action potential and β-adrenergic stimulation to quantitatively investigate arrhythmia-formation mechanisms peculiar to silent long QT syndrome, using mutation Q357R in KCNQ1 (α subunit of slow-delayed rectifier I_Ks) as a paradigm.

METHODS

Markov models were formulated to account for altered I_Ks kinetics in Q357R compared with wild type and introduced into a detailed model of the human ventricular myocyte action potential.

RESULTS

Dominant negative loss of I_Ks available reserve accurately represents Q357R. Action potential prolongation with mutant I_Ks was minimal, reproducing the silent phenotype. Partial block of rapid delayed rectifier current (I_Kr) was needed in addition to fast pacing and isoproterenol application to cause early after-depolarizations (EADs) in epicardial cells with mutant I_Ks, but this did not produce EADs in wild type. Reduced channel expression at the membrane, not I_Ks kinetic differences, caused EADs in the silent mutant. With mutant I_Ks, isoproterenol plus partial I_Kr block resulted in dramatic QT prolongation in the pseudo-electrocardiogram and EADs formed without I_Kr block in mid-myocardial cells during simulated exercise onset.

CONCLUSION

Multiple severe insults are needed to evince an arrhythmic phenotype in silent mutation Q357R. Reduced membrane I_Ks expression, not kinetic changes, underlies the arrhythmic phenotype.

Introduction

Long QT syndrome (LQTS) can cause cardiac arrhythmia and sudden death in the absence of structural heart disease.¹ LQTS affects as many as 1 in 2000 among Caucasians.² The most common form of LQTS is type 1 (LQT1), with KCNQ1 as the locus of mutations,³ a gene that transcribes the Kv7.1 protein, forming the α subunit of the repolarizing slow delayed rectifier K⁺ current I_Ks.

Although LQT1 is less deadly than other forms of LQTS, lethality increases considerably with emotional/physical stress or exercise, conditions associated with β-adrenergic stimulation and fast heart rate.⁴ These conditions and also certain drug effects (eg, block of rapid-delayed rectifier I_Kr) increase the arrhythmia risk by challenging the genetically compromised repolarization reserve due to LQT1 mutation.⁵ It has been conceptualized that physiological or external insults to the repolarization reserve, individually innocuous, combine synergistically to create arrhythmia substrates and cause fatal arrhythmias. As stated by A. Wilde, “a double hit hurts more.”⁶

An enigmatic case of subclinical or “silent”⁷ LQT1 was identified, missense mutation Q357R near the S6–C-terminal junction of Kv7.1 (Boulet et al⁸). Q357R was discovered in a 40-year-old female proband with a history of syncope, but with the corrected QT interval within the normal range (430 ms).⁹ The absence of electrocardiogram (ECG) phenotype associated with this silent mutation obscures the connection between phenotype and arrhythmia. Exhaustive measurements by Boulet et al⁸ showed that Q357R is characterized by partial loss of I_Ks function, a current that is of small magnitude in the human ventricle under basal conditions.

Observations of Boulet et al at the level of I_Ks current suggest only a vague possibility of minor action potential (AP) repolarization changes and marginally increased arrhythmia risk for this silent mutation carrier. Thus, several clinical questions are raised: Is it reasonable to suspect that the silent mutation was in fact the cause of syncope in the proband? That is, can known LQT1 insults (emotional/physical stress or exercise) account for arrhythmia formation in this case of silent LQT1? Moreover, is the consequence of the silent mutation so mild that these insults affect mutant and wild-type (WT) I_Ks similarly, causing a similar phenotype? If so, the silent mutation may not be the cause of syncope. Finding answers to these questions requires quantification beyond qualitative predictions. Motivated to investigate and explain the possibility of arrhythmia formation in humans with silent LQTS, we investigated Q357R as an instructive paradigm, using newly available, quantitatively accurate mathematical models.

Methods

Using experiments from Boulet et al,⁸ human I_Ks models for WT and Q357R were developed. AP simulations were conducted by using the O'Hara Rudy dynamic human ventricular action potential model (ORd).¹⁰ Simulations were performed in epicardial (epi) and mid-myocardial (M) cells and in a transmural wedge model from which the pseudo-ECG was computed¹¹ mimicking experiments.¹² Further details are given in the Supplement. Simulation results for APs or pseudo-ECGs were always preceded by a train of pacing that is not shown. For plotting purposes, time is reset to begin at t = 0, though the true simulation time is much larger.

To determine β-adrenergic stimulation effects on I_Ks and resulting changes to its role in the AP, we adapted the model by Heijman et al,¹³ including signaling cascade from isoproterenol (ISO) application to compartmentalized protein kinase A concentration and fractional phosphorylation of targets. Equations, validation, and details are given in the Supplement (Supplement Figures S1 and S2).

Unless otherwise stated, black, dashed black, and gray lines are simulation results for WT, Q357R, and the heterozygous (het) case, respectively. Current symbols: fast Na⁺ (I_Na), L-type Ca²⁺ (I_CaL), Na⁺/K⁺ ATPase (I_NaK), ultrarapid K⁺ (I_Kur, represented by K⁺ background, I_Kb in ORd).

Results

Q357R I_Ks kinetics and behavior

WT I_Ks was represented by the Markov model developed by Silva and Rudy¹⁴ shown in Figure 1A, left. The activation of WT channels involves slow transitions from left to right in zone 2 of deep closed states, followed by rapid transitions only, from top to bottom, in zone 1 of closed states. Accumulation in zone 1 states, which are kinetically near the open states, provides an available reserve¹⁴ of channels that can activate rapidly. By contrast, we hypothesized that Q357R mutant channels do not develop an available reserve. That is, Q357R channels activate with 2 identical slow zone 2–like transitions between all closed states. To implement this, we replaced top-to-bottom transition rates (“g” and “d” in WT) with the left-to-right transition rates (“a” and “b” in WT, slightly modified to “A” and “B” in Q357R). In addition, compared with WT, the Q357R model includes depolarization-directed shifts in the voltage dependence of transition rates. Conductance was reduced by 50% to represent the observed lower expression at the membrane.

Figure 1.

Markov I_Ks. A: Schematic diagrams and transition rate equations. The WT model is on the left; Q357R is on the right. Equations are below the diagrams (differences from WT are in large bold gray type). WT model activation, as proposed by Silva and Rudy,¹⁴ represents 2 voltage sensor transitions. First transitions are from left to right (large black arrow) and second transitions are from top to bottom (large gray arrow). Channel kinetic states are divided into 2 zones. Different from WT where zone 1 (gray circles) transitions are rapid, for Q357R these transitions are the same as the slower zone 2 (black circles) transitions. B: WT (white) and Q357R (black) subunits combine to form 6 tetramer permutations in the heterozygous (het) case (gray). The model considers that the mutation was dominant negative. Thus, the behavior of the average het channel is the average behavior of the permutations: het = (1 × WT + 5 × Q357R)/6. WT = wild type.

Boulet et al created the het case by injecting cells with equal amounts of Q357R and WT cDNA. There are 6 permutations of WT and Q357R subunit pairings that can form channel tetramers (Figure 1B). We tested whether dominant negative mutant behavior could account for het results. That is, the model assumes that WT and Q357R tetramer assembly is random (but lumping together degenerate cases); presence of ≥1 Q357R subunit(s) slows gating for the entire channel, and presence of ≥1 Q357R subunit(s) attenuates membrane expression. These assumptions (schematized in Figure 1B) allowed us to develop and test a putative model for average het channel behavior. Accordingly, het current calculation was the weighted average of computed Q357R and WT currents, added together in a 5:1 mixture. We tested whether allowing only subsets of all heterotetramer permutations would affect the comparison with experiments. The model that allowed all permutations provided the best overall fit to the experimental data (Supplement Figure S3).

Using the assumptions described, models reproduced the kinetics of the experimentally recorded Q357R and het currents compared with WT (Figure 2). Specifically, for mutant current simulation compared with WT, the (1) steady-state activation was shifted to more depolarized potentials, (2) activation rate was slowed, (3) deactivation rate was unaffected, and (4) tail current was reduced. The correspondence between simulations and experiments supports the hypothesis that Q357R can be considered as dominant negative loss of available reserve.

Figure 2.

Validation for WT, Q357R, and het I_Ks. Experiments are from Boulet et al⁸ (squares: WT is black, Q357R is white, and het is gray). A: Steady-state activation of tail currents. B: Time constant for step current activation. C: Time constant for deactivation. D: Relative current, after 5-second steps to the potentials shown along the abscissa. Simulated protocols were the same as those used in Boulet experiments. Exp, experiments; sims, simulations. het = heterozygous; WT = wild type.

Model design was hypothesis driven, guided by previous simulations¹⁴ and experimental observations.^8,15 We did not consider other models of the mutant channel. Five free parameters were needed to capture essential behaviors. Parameter values were determined by fitting the simulated kinetics to experimental data. Parameter sensitivity analysis is provided in Supplement Figure S4.

Phenotype is silent under basal conditions

We tested the effect of replacing WT with Q357R or het I_Ks in different cell types at normal heart rate under basal conditions (cycle length [CL] = 1000 ms, no ISO, Figure 3). In epi cells, Q357R and het had very little effect on the AP duration (APD). Prolongation was 14 and 12 ms, respec tively. For M cells, it was 30 and 23 ms, respectively. These are small increases (≤8.5% increase), though larger than those for epi cells (≤6.0% increase). Such minor AP prolongation under basal conditions reproduces the silent phenotype. Note that I_Ks was reduced significantly by the mutation, though it is relatively small to begin with in WT.

Figure 3.

Basal conditions and normal pacing rate (CL = 1000 ms). Epi (left) versus M cell (right) under basal conditions (no isoproterenol). The top row shows APs. I_Ks is the middle row. Closed state zone occupancy is below. Time = 0 is the start of the final paced AP. AP = action potential; CL = cycle length; epi = epicardial; het = heterozygous; M = mid-myocardial; WT = wild type.

Somewhat surprisingly, after many beats of pacing (t = 0 in Figure 3 is the start of the final paced AP), occupancy in rapidly activating zone 1 states of I_Ks was actually higher for Q357R compared with WT in both epi and M cells (Figure 3, bottom); that is, zone 1 accumulation¹⁶ during pacing was greater in the mutant (explained later).

Multiple insults to the repolarization reserve cause early afterdepolarizations in the silent mutant, not in WT

In Figure 4, insults to the repolarization reserve, representing known arrhythmia triggers in LQT1, were added sequentially in epi cell simulations. They included (1) fast pacing (CL = 300 ms), (2) β-adrenergic stimulation (1 μM ISO), and (3) partial I_Kr block (30%). With fast pacing, Q357R and het prolonged the AP by only negligible amounts (8 and 7 ms, respectively). Note that the prolongation was less than that observed at the slower resting heart rate (CL = 1000 ms, far left column, 14 and 12 ms prolongation for Q357R and het, respectively). This is typical of the LQT1 clinical phenotype, where prolongation is more severe at slower rates, though arrhythmic events are associated with fast heart rates.^1,4 When ISO was added to fast pacing, the APD was increased by 27 and 21 ms for Q357R and het respectively, relative to WT at basal conditions (CL = 1000 ms). The increase was 15 ms (Q357R) and 10 ms (het) compared with that without ISO at CL = 300 ms. It was due to a larger inward I_CaL that was not countered sufficiently by an increase in the mutation-compromised I_Ks. By contrast, ISO slightly decreased the WT APD (4 ms, shown in Supplement Figure S2).

Figure 4.

Effect of multiple insults on the AP in epi cells. Descending rows show the AP, I_CaL, I_Ks, and closed state zone occupancy. Columns, from left to right, show normal pacing under basal conditions, fast pacing under basal conditions, fast pacing with application of ISO, and 30% I_Kr block in addition to fast pacing and ISO. EADs in the rightmost column (top row, arrows) were caused by I_CaL reactivation (second row, arrows) due to prolongation of AP plateau (deficient I_Ks, third row). The AP following the EAD is short because of the residual activation of I_Ks (incomplete deactivation, arrows). AP = action potential; CL = cycle length; EAD = early afterdepolarization; epi = epicardial; ISO = isoproterenol; het = heterozygous; WT = wild type.

Adding a partial I_Kr block to ISO at a fast rate caused WT epi APD to increase by 31 ms. For Q357R and het, epi APD was severely prolonged under these conditions; extended time at depolarized plateau voltages allowed I_CaL reactivation, triggering early afterdepolarizations (EADs) (mechanism shown previously¹⁰). Supplement Figure S5 shows that no EADs formed when insults were applied individually (not layered sequentially).

In M cells, with the first insult alone (fast pacing), the Q357R and het I_Ks changes had a strong effect on the AP (Supplement Figure S6).

I_Ks gating kinetic changes in the silent mutant were not responsible for EADs

The I_Ks zoom inset in Figure 4 (third row) shows that mutant I_Ks was actually larger than WT I_Ks for the first approximately 40 ms of the AP. However, mutant current amplitude was approximately half that of WT once peak values were reached during the critical phase 3 of the AP, when repolarizing current can effectively shorten the APD. Pacing for many beats caused accumulation in zone 1 for both WT and Q357R, as anticipated^14,16 (Figures 3 and 4, bottom rows). However, as mentioned earlier, zone 1 accumulation was greater for Q357R than for WT, explaining the appearance of early current in the mutant. Greater zone 1 accumulation with pacing was due to slowed “deep deactivation” in the mutant, defined as the transition from shallow zone 1 closed states to deep zone 2 closed states (see Supplement Figure S7; a detailed explanation is provided in the figure caption). This result is a model prediction, not tested experimentally by Boulet et al, whose protocols did not include the response of I_Ks to pacing.

Although interesting, the I_Ks kinetic differences associated with the Q357R mutation were not responsible for EAD formation, as described below. To test whether altered gating kinetics of Q357R might have contributed to EAD generation after multiple insults, component mutants were investigated: one with altered conductance alone (type A) and another with altered gating kinetics alone (type B). Qualitative results for component mutant type A were the same as for Q357R and het (Figure 5). The AP was prolonged, and then EADs formed after the insults were layered. Although mutant type B showed noticeable AP prolongation compared with WT after insults, EADs did not form. These results signify that the more arrhythmic consequence of Q357R is due to reduced channel expression at the membrane, represented by reduced channel conductance. Kinetic differences were not a determining factor in EAD formation.

Figure 5.

Effect of reduced conductance versus kinetic changes in the silent mutation. WT, WT with Q357R conductance (mutant type A), and WT with Q357R kinetics (mutant type B) are black, dashed black, and gray lines, respectively. Descending rows show the AP, I_CaL, and I_Ks. The four columns, from left to right, show the cases for normal pacing under basal conditions, fast pacing under basal conditions, fast pacing with application of ISO, and 30% I_Kr block in addition to fast pacing and ISO. Insults of fast pacing with ISO caused AP prolongation relative to WT. With addition of I_Kr reduction, kinetic changes alone did not cause EAD formation, but reduced conductance alone did (arrows, rightmost column). AP = action potential; CL = cycle length; EAD = early afterdepolarization; ISO = isoproterenol; WT = wild type.

Heterogeneous transmural wedge simulations

To determine how the Q357R mutation affects repolarization in the more realistic context of heterogeneous tissue, we performed transmural wedge (pseudo-ECG) simulations (Figure 6). Under basal conditions, the silent mutation barely affected the QT interval (15 and 13 ms longer for Q357R and het, respectively, compared with WT). With ISO application, prolongation was 24 and 23 ms for Q357R and het relative to WT, and T-wave amplitude increased. With I_Kr block added, QT prolongation relative to WT became severe for Q357R and het (>80 ms). ISO plus I_Kr block-induced prolongation for Q357R and het was nearly twice (1.8-fold) that for WT. Dispersion of repolarization, measured as T-peak to T-end duration, was 2-fold larger for the mutant cases. However, the WT T-wave was also markedly affected by these conditions.

Figure 6.

Effects of ISO and I_Kr reduction on the pseudo-ECG. The simulated pseudo-ECG is plotted by using a shared, normative scale. On the left, middle, and right, the cases of control, ISO, and 30% I_Kr block plus ISO are shown. CL = cycle length; ECG = electrocardiogram; het = heterozygous; ISO = isoproterenol; WT = wild type.

Onset of emotional/physical stress or exercise

Arrhythmia in LQT1 occurs during transition to a state of emotional/physical stress (eg, exercise onset⁴ or diving into a pool¹). To simulate this transient scenario, we accelerated the pacing rate and applied a bolus of ISO (using the more mutation-sensitive M cell, Figure 7). The pacing rate was doubled, from CL = 1000 to 500 ms, in a graded fashion over approximately 7 seconds, as observed in human subjects during exercise onset.¹⁷ For I_CaL, the fraction of channels phosphorylated by protein kinase A increased more rapidly than for I_Ks (validated previously¹³). Eventually, the resulting imbalance between inward and outward current ISO responses led to pronounced EADs, even without I_Kr block, in the silent mutant cases.

Figure 7.

Exercise onset. Simulations are in M cells. Shown from top to bottom: APs, I_Ks, I_CaL, and fraction of I_CaL (f_ICaL, gray) and I_Ks (f_IKs, black) channels phosphorylated by PKA. Bars along the top show the accelerating stimulus delivery. An ISO bolus of 1 μM was applied at t = 1000 ms (arrow in bottom panel). Time axis is broken (dots), skipping an uneventful midsection. The box on the far right shows the final paced beats on an expanded time scale. Arrows show EADs caused by I_CaL reactivation. AP, action potential; CL, cycle length; EAD, early afterdepolarization; het, heterozygous; ISO, isoproterenol; PKA, protein kinase A; WT, wild type.

Other exercise-onset simulation results are worth mentioning. With the addition of 30% I_Kr block, EADs appeared in WT as well as Q357R and het in M cells. In epi cells, EADs appeared in Q357R and het only with 30% I_Kr block. However, under these conditions, they also appeared in WT.

Discussion

As shown by Jost et al,¹⁸ the small-magnitude human I_Ks is augmentable, helping to shorten the AP as rate increases and terminate the AP when the repolarization reserve is compromised. Silva and Rudy¹⁴ simulations showed that I_Ks augmentation can overcome an EAD due to I_Kr block and restore normal repolarization. These concepts explain how I_Ks loss of function leads to arrhythmia formation in patients with LQT1. However, with silent mutant Q357R, the phenotype is so mild that the applicability of these concepts is suspect. Is it plausible to assume that the silent mutation is the sole cause of arrhythmia? To answer this question, rigorous quantitative analysis is needed. In guinea pig simulations, arrhythmic Q357R and het phenotypes were revealed under basal conditions alone, without any repolarization reserve insults (Supplement Figure S8). This consequence of species-dependent ionic profiles (large I_Ks in guinea pig) highlights the fact that quantitative accuracy is paramount when investigating silent mutations where arrhythmia formation is marginal.

Because of the mild phenotype, multiple insults to the repolarization reserve were required to reveal the abnormal behavior of the silent mutation compared with WT in humans. Fast pacing and ISO application alone substantially prolonged the APD (LQT phenotype) in this particular mutation, but the external factor of partial I_Kr block was needed to cause EADs in epi cells. In M cells during exercise onset, EADs formed without I_Kr block. These results are consistent with the accepted dogma for arrhythmia formation in LQT1. Relatively simple computer models helped develop these basic concepts. However, silent mutation effects are subtle and demand more sophisticated models for quantitative accuracy. The new simulations presented here showed that (1) separate insults are insufficient to differentiate silent mutant phenotype from WT (multiple insults are needed) and (2) multiple severe insults to the repolarization reserve may also affect WT. Therefore, a silent mutation must be investigated quantitatively to determine whether the mutation alone is indeed responsible for arrhythmia formation; the simulation results suggest that it may not be for Q357R.

Note that when EADs formed in silent mutant Q357R and in het, the WT was not unaffected. That is, multiple insults caused minor voltage deflections even in WT. If the final insult of partial I_Kr block was increased from 30% to 40%, EADs appeared in epi cells from WT as well (not shown). Also, during the characteristic arrhythmogenic LQT1 challenge of exercise onset, it was difficult to find conditions that revealed hidden arrhythmia propensity in silent mutant cases without also affecting WT behavior. The possibility of EAD generation in the WT model was validated previously¹⁰ based on experiments showing EAD formation after nearly complete I_Kr block in slowly paced nonfailing isolated human ventricular myocytes.¹⁹ However, in the intact heart, cells are electrically loaded and EAD formation is less likely. Taken together, the results cast doubt as to whether the silent mutation alone sufficiently explains syncope in the Q357R proband.

Dominant negative loss of available reserve

Missense mutation Q357R is an exchange of neutral glutamine for positively charged arginine at the junction between S6 and the C-terminus. Boulet et al's⁸ experiments showed that because of the influence of the positive arginine, voltage-dependent gating was affected. We speculate that placement of mutant arginine adjacent to the positive–negative charge interactions between S4 and S2 segments²⁰ could disrupt gating. However, in the absence of further experimental analysis based on structural detail, it remains unclear how this might amount to the observed slowing of activation.

Boulet et al⁸ were not able to determine whether Q357R reduced membrane expression because of trafficking or assembly abnormalities. Simulations involved lumping together degenerate tetramer assemblies, which could reflect assembly abnormalities. Thus, our analysis cannot determine that Q357R-reduced membrane expression could be due to impaired trafficking alone. Additional experiments are needed to clarify this issue.

Limitations

The effects of β-adrenergic stimulation were validated on a target-by-target basis by using the ORd. The overall result of β-adrenergic stimulation on the AP and APD restitution are in agreement with nonfailing human ventricular measurements (Supplement Figures S1 and S2). However, the experimental data set available for β-adrenergic effects on target proteins is nearly all from experiments performed in nonhuman mammals (dog), adapted here for humans.

Data are not available that compare the effects of β-adrenergic stimulation on Q357R versus WT. It is possible that differential effects exist. However, the simulations assume that the influence of ISO application was identical in both. Inherent gating differences could have but in fact did not cause different functional consequences of ISO application; the ISO-induced percentage increase in the charge carried by I_Ks during an AP was the same for WT, Q357R, and het (not shown). Thus, simulations predict that ISO application does not elevate EAD propensity in the silent mutant more than for WT. Moreover, the binding site for the β-adrenergic signaling complex (near residue 589)²¹ is distal to the mutation site.

The pseudo-ECG generated by a 1D transmural fiber is related to but not equated with the clinical ECG, where the QT interval reflects the time between the earliest activation and the latest repolarization over the entire ventricles. Naturally, clinical corrected QT intervals are longer than those for simulated pseudo-ECG. Nevertheless, the pseudo-ECG QT interval for WT can be recalibrated to match normal clinical values (403 ms at 1 Hz). Using the recalibrated model, the het QT interval was 426 ms, similar to the corrected QT interval in the proband (430 ms, 1 Hz) under basal conditions. Importantly, I_Ks elimination (complete block), as in severe nonsilent LQT1, indeed caused QT prolongation consistent with LQT1 diagnosis (1 Hz basal QT ≥ 440 ms¹).

The ORd does not represent gender or age differences; it was validated based on experiments from more than 100 undiseased human hearts, of which 56% were male of average age 41 ± 12 years.¹¹ Recent data showed K⁺ current reduction in females compared with males.²² Female gender may be an additional risk factor for patients with LQTS. The I_Kr block simulations presented relate to, but do not specifically account for, gender differences. Advanced age is also an LQT risk factor that can induce pathology of silent mutations.^23,24

Conclusion

For the Q357R mutation, multiple insults to the repolarization reserve were needed to separate mutant from WT behavior. Yet, the separation was marginal. Simulations demonstrate that simple loss of channels at the membrane, not kinetics alterations, accounts for EAD generation in the silent mutation. The Q357R example admonishes that when considering arrhythmia-formation mechanisms in silent LQTS, it is important to apply quantitative analysis to determine whether the identified mutation, and not some external factor(s), was indeed the true cause of arrhythmia.

ABBREVIATIONS

AP

action potential

APD

action potential duration

CL

cycle length

EAD

early afterdepolarization

ECG

electrocardiogram

epi

epicardial

het

heterozygous

ISO

isoproterenol

LQT1

long QT syndrome type 1

LQTS

long QT syndrome

M

mid-myocardial

ORd

O'Hara Rudy dynamic human ventricular action potential model

WT

wild type