Morpholino kcnh2 Knockdown Disrupts Cardiac Repolarization

Kcnh2 is expressed very early in cardiac development in the bilateral cardiac primordia before the onset of cardiac contraction and specifically localized to the cardiac chambers at 48-hours postfertilization (Figure 1A). MO-knockdown of kcnh2 caused a spectrum of repolarization defects, similar to those observed in the zebrafish knch2 mutants, breakdance, and silent ventricle.^7,8 Ninety-one percent of MO-injected embryos (n=210) displayed a repolarization-deficient phenotype (2:1 AVB or ventricular asystole), confirming the robust nature of the MO-induced kcnh2-knockdown. WT hERG1 RNA injection restored normal repolarization in 55% of MO-injected embryos, and shifted the severity of the repolarization defect from ventricular asystole to 2:1 AVB (Figure 1B). To define the cellular correlates of our observed repolarization phenotypes, we directly measured V_m in explanted embryonic ventricle. Spontaneous ventricular action potentials recorded from embryos manifesting 2:1 AVB were significantly prolonged compared with embryos that displayed normal repolarization (1:1 AV conduction; Figure 1C and the online-only Data Supplement Table I). The resting ventricular V_m recorded from embyros displaying a ventricular asystole phenotype was markedly depolarized and spontaneous action potentials were not observed (Figure 1C), similar to that recorded in the complete loss-of-function knch2 mutant, silent ventricle.⁷ MO-injected embyros displaying a normal repolarization phenotype captured the pacing stimulus in a 1:1 fashion, whereas those displaying a 2:1 AVB phenotype were unable to pace in a 1:1 manner as a consequence of action potential duration prolongation (Figure 1D). Taken together, these data demonstrate that kcnh2 MO-knockdown produces repolarization-deficient phenotypes that can be rescued by injection of WT hERG1 RNA.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Embryonic knch2 expression and the effects of kcnh2-knockdown on cardiac repolarization. A, Left, Whole-mount in situ hybridization reveals kcnh2 expression in the bilateral cardiac primordia (12 somite stage) before the onset of cardiac contraction and specific expression in the heart at 48-hours postfertilization. Right, B,The percentage of embryos displaying normal (1:1 atrioventricular [AV] conduction) or abnormal repolarization (2:1 atrioventricular block [AVB]), or ventricular asystole) 48 hours after injection of anti-kcnh2 morpholino (MO) alone or together with wild-type (WT) human ether-a-go-go related gene [hERG] RNA. C, Examples of V_m recordings from explanted embryonic ventricle of MO-injected embryos that displayed normal 1:1 AV conduction, 2:1 AVB, or ventricular asystole phenotypes. D, External pacing (cycle length 500 ms) of explanted ventricle from MO-injected embryos displaying 1:1 or 2:1 AV conduction. AVB indicates atrioventricular block; and MO, morpholino.

An In Vivo Cardiac Assay to Determine the Functional Consequences of hERG1 Variants

Next, we tested the ability of previously described LQTS hERG1 mutations (n=40) and hERG1 polymorphisms (n=10) to restore normal repolarization in the kcnh2-knockdown embryonic zebrafish (Figure 2). A spectrum of normal repolarization rescue was observed for the 40 putative LQTS hERG1 mutants, but the degree of rescue was statistically lower than WT hERG1 rescue (the online-only Data Supplemental Table II). All 10 hERG1 single nucleotide polymorphisms rescued the kcnh2 MO-knockdown to a similar degree as WT hERG1 (P>0.05, generalized linear mixed model). To confirm the electrophysiological basis for the repolarization phenotypes observed, we measured ventricular V_m in kcnh2-MO knockdown embyros injected with WT, N470D, A614V, and A1116V hERG1 (the online-only Data Supplemental Figure I). These data confirm that 2:1 AVB is the result of marked action potential duration prolongation, whereas ventricular asystole is because of marked depolarization.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Functional effects of putative long QT (LQT) 2 mutations and polymorphisms, as determined by zebrafish cardiac assay. Percentage of embryos manifesting normal repolarization (1:1 atrioventricular [AV] conduction) plotted for anti-kcnh2 morpholino (MO)-injected (MO alone), wild-type (WT) human ether-a-go-go related gene [hERG]+MO, and mutant hERG+MO. The location of the variant on the channel is listed below for the putative LQT-2 variants. Note that error bars represent +95% confidence intervals. N=152 to 488 in each group. C-term indicates C-terminus; N-term, N-terminus; polymorph, poly-morphism; and TM, transmembrane domain.

Forty-nine of 50 hERG1 variants have been functionally characterized using the current gold-standard in vitro heterologous expression system. When compared with the benchmark assay, the cardiac assay correctly identified a nondisease-causing variant in 9 of 10 cases (negative predictive value 90%), whereas correctly identifying a disease-causing variant in 39/39 cases (positive predictive value 100%). The only variant incorrectly assigned by the in vivo assay was T436M, which was predicted to be a pathogenic variant.

As an alternative comparison between the 2 assays, we normalized the values obtained using the in vivo zebrafish cardiac assay to the gold-standard in vitro assay as follows. For each hERG1 variant, the fraction of embryos with normal repolarization was normalized to WT value and plotted against the fraction of WT current measured in vitro, using published values (Figure 3). We used hierarchal cluster analysis to group the values into 2 or more unique subsets that represent concordance between the 2 assays. The data were best fit by 2 distinct clusters that distinguish pathogenic from nonpathogenic variants (the online-only Data Supplemental Figure II). For example, variants with current magnitude <0.5 of WT clustered tightly with variants, whose repolarization rescue was <0.5 of WT. Variants with current magnitude >0.7 of WT clustered with a broader range of rescued repolarization values, but were generally concordant with rescue values >0.67 of WT. One variant fell outside of the 95% confidence interval ellipse of either cluster (T474I). Although T474I generates functional channels in vitro, it was considered a pathogenic mutation based on altered kinetic properties⁹ and was correctly identified as disease-causing by the in vivo cardiac assay.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Comparison of in vivo zebrafish cardiac assay with in vitro mammalian cell assay. For each human ether-a-go-go related gene [hERG] variant, the fraction of embryos with normal repolarization is normalized to wild-type (WT) value and plotted against fraction of WT current assayed in mammalian cells, using published values. Cluster analysis identified 2 distinct subsets that represent concordance between the assays to distinguish pathogenic and benign variants. T474I fell outside the 95% confidence interval (CI) ellipses (dotted gray lines) of the 2 clusters. Although T474I generates functional channels in vitro, it was considered a pathogenic mutation based on altered kinetic properties⁹ and was correctly identified as disease-causing by the in vivo cardiac assay.

Finally, we tested the ability of the in vivo cardiac assay to corroborate the clinical observation that inheriting a common polymorphism (K897T) and an LQT-2 variant (A490T) in cis orientation reduces the severity of the LQTS phenotype compared with subjects with the A490T mutation alone.¹⁰ The A490T single mutant behaved as a pathogenic variant in the in vivo assay, whereas the double mutant A490T-K897T rescued repolarization similar to WT hERG1 (the online-only Data Supplemental Figure III). Taken together, our results suggests that replacing the endogenous kcnh2 gene with the human orthologue has predictive value for testing putative disease-causing LQT-2 variants, and corroborates an interesting clinical observation that a common polymorphism modifies the severity of a specific LQTS phenotype.