OUP user menu

Genetic variation in KCNA5: impact on the atrial-specific potassium current IKur in patients with lone atrial fibrillation

Ingrid E. Christophersen, Morten S. Olesen, Bo Liang, Martin N. Andersen, Anders P. Larsen, Jonas B. Nielsen, Stig Haunsø, Søren-Peter Olesen, Arnljot Tveit, Jesper H. Svendsen, Nicole Schmitt
DOI: http://dx.doi.org/10.1093/eurheartj/ehs442 1517-1525 First published online: 21 December 2012


Aims Genetic factors may be important in the development of atrial fibrillation (AF) in the young. KCNA5 encodes the potassium channel α-subunit KV1.5, which underlies the voltage-gated atrial-specific potassium current IKur. KCNAB2 encodes KVβ2, a β-subunit of KV1.5, which increases IKur. Three studies have identified loss-of-function mutations in KCNA5 in patients with idiopathic AF. We hypothesized that early-onset lone AF is associated with high prevalence of genetic variants in KCNA5 and KCNAB2.

Methods and results The coding sequences of KCNA5 and KCNAB2 were sequenced in 307 patients with mean age of 33 years at the onset of lone AF, and in 216 healthy controls. We identified six novel non-synonymous mutations [E48G, Y155C, A305T (twice), D322H, D469E, and P488S] in KCNA5 in seven patients. None were present in controls. We identified a significantly higher frequency of rare deleterious variants in KCNA5 in the patients than in controls. The mutations were analysed with confocal microscopy and whole-cell patch-clamp techniques. The mutant proteins Y155C, D469E, and P488S displayed decreased surface expression and loss-of-function in patch-clamp studies, whereas E48G, A305T, and D322H showed preserved surface expression and gain-of-function for KV1.5.

Conclusion This study is the first to present gain-of-function mutations in KCNA5 in patients with early-onset lone AF. We identified three gain-of-function and three loss-of-function mutations. We report a high prevalence of variants in KCNA5 in these patients. This supports the hypothesis that both increased and decreased potassium currents enhance AF susceptibility.

  • Lone atrial fibrillation
  • Genetics
  • KV1.5
  • KCNA5
  • KCNAB2
  • IKur


Atrial fibrillation (AF) is the most common cardiac arrhythmia.1 Today AF affects 2.6 million Americans2 and 6 million Europeans.1 The mechanisms underlying AF are not fully understood, but multiple pathophysiological pathways have been suggested. Atrial fibrillation is often developed secondary to cardiovascular disease, endocrine disorders, and lifestyle factors.3 A subgroup of patients diagnosed with AF in the absence of predisposing factors, a condition called ‘lone AF’, accounts for 10–20% of the total number of patients with AF.4 Genetic factors are believed to be important in the development of the disease in this group. This can be ascribed to (i) the relatively low age at diagnosis, (ii) the lack of obvious biological causative factors, and (iii) familial clustering of the disease. Epidemiological studies have shown clustering of AF in families,5,6 and our group has demonstrated a substantial heritability of AF in a twin study.7 After Chen et al. identified a mutation in KCNQ1 associated with AF in a Chinese kindred in 2003,8 the presence of genetic components in AF and the importance of single-nucleotide polymorphisms (SNPs) have been shown in candidate gene studies and in large genome-wide association studies. Direct sequencing of genes encoding cardiac ion channels, which carry the currents forming the cardiac action potential (AP), has resulted in the identification of mutations associated with AF.9

Most of these studies show that gain- or loss-of-function mutations in these ion channel genes can lead to increased susceptibility to AF. This supports the two current conceptual models for AF: (i) shortening of the effective refractory period (ERP) functioning as a substrate for re-entry wavelets in the atria10,11; and (ii) prolongation of the cardiac AP, which increases the susceptibility to AF12 through enhanced propensity for early after-depolarization (EAD).

The atrial-specific potassium current IKur is carried by the voltage-gated potassium channel KV1.513,14 and plays an essential role in the repolarization of human atrial cardiomyocytes.15 KV1.5 is encoded by KCNA5, and loss-of-function mutations have been associated with AF.12,16,17 Co-expression of KV1.5 with the regulatory β-subunit Kvβ2.1, encoded by KCNAB2, alters the voltage-dependency of activation, which has been proposed to be a regulatory mechanism controlling AP duration in cardiac cells.18,19 This led us to hypothesize that early-onset lone AF is associated with a high prevalence of mutations in KCNA5 and KCNAB2.


An expanded Methods section is available in Supplementary material online.

Study population

Scandinavian patients with early-onset lone AF were recruited from Copenhagen, Denmark, and from Oslo and Vestre Viken, Norway. Early-onset was defined as diagnosis of AF before age 50. Lone AF was defined as AF in the absence of clinical or echocardiographic findings of cardiovascular disease, hypertension, metabolic or pulmonary disease. Healthy controls were recruited from blood donors at the Copenhagen University Hospital, who had undergone clinical evaluation and ECG recording. The study conforms to the principles of the Declaration of Helsinki and was approved by the Scientific Ethics Committee of Copenhagen and Frederiksberg (Protocol reference number: H-KF-01313322) and the Regional Ethics Committee (REK) in Norway (Protocol reference number: 2009/2224-5). All included patients gave written informed consent.

Mutational screening

Oligonucleotide primers for exons and splice junctions were designed using the known sequence of human KCNA5 (GenBank Acc. No. NG_012198). Bidirectional sequencing of genes previously associated with AF was performed (Supplementary material online).

Molecular biology

A detailed description is available in Supplementary material online.

Cell culture and transient transfections

Human embryonic kidney (HEK) 293 cells were transfected with plasmids carrying WT or mutant KV1.5, using Lipofectamine and Plus Reagent according to the manufacturer's protocol (Invitrogen, Glostrup, Denmark) (Supplementary material online).

Confocal microscopy and imaging

For imaging experiments, the following antibodies were used: rabbit anti-Kv1.5 (1:50, APC-004, Alomone, Jerusalem, Israel), Alexa Fluor®568-conjugated donkey anti-rabbit IgG (1:200) or Alexa Fluor®488-conjugated donkey anti-rabbit IgG (1:200), Alexa®Fluor 647 Phalloidin (1:200, Invitrogen), and 4,6-diamidino-2-phenylindole (DAPI, 1:300, Invitrogen) (Supplementary material online).


Patch-clamp experiments were performed at room temperature 2–3 days after transfection. Currents were recorded by the application of voltage-step protocols detailed in the figure legends. For a detailed description, see Supplementary material online.

Cardiac action potential modelling

All simulations were performed with the COR program (http://cor.physiol.ox.ac.uk, version,20 using the built-in CVODE integrator with a time step of 0.01 ms. The human atrial cell model described by Courtemanche et al.21 was used for simulations. For a detailed description, see Supplementary material online.

Data analyses

The potentials where half of the channels are activated or inactivated (V1/2) were determined by fitting the current curves with a Boltzmann function: I/Imax = 1/(1 + exp((VV1/2)/k)) (I: current recorded at an applied potential; Imax: maximal current; V: applied potential; V1/2: potential where half of the channels are activated or inactivated; k: slope factor). All data are presented as mean ± standard error of the mean. For statistical analyses (GraphPad Prism 5.0 and Stata 11 statistical software), unpaired t-tests and two-way analysis of variance combined with a Bonferroni post hoc test were used. P < 0.05 was considered statistically significant.

The prevalence of rare ‘deleterious’ variants in cases and controls was compared using χ2 analysis. A ‘rare variant’ was defined as a variant present in <0.1% of the alleles in the exome variant server (EVS).22 ‘Deleterious’ was defined as (i) giving rise to changes in protein function in functional studies, or where functional studies were not available; (ii) predicted to be possibly or probably damaging using Polyphen-2. The statistical significance of differences in the number of alleles between these two groups was determined using Fisher's exact test. P-values <0.05 were considered statistically significant. Statistical analyses were performed using the Stata 11 software (StataCorp).



The study population consisted of 307 unrelated patients of Scandinavian ethnicity with the onset of lone AF before the age of 50 (median age at inclusion 44 years, inter-quartile range 38–48 years). There was a male predominance (82%). We included 216 controls (52% men, median age 39 years, inter-quartile range 30–48 years). Thirty-five per cent of the patients had familial AF defined as ≥1 first-degree relative diagnosed with AF. The women had a significantly higher prevalence of familial AF than men (55 vs. 30%, OR 2.8; 95% CI 1.4–5.4; P = 0.0008). Clinical characteristics are shown in Supplementary material online, Table S1.

Six novel missense mutations in KCNA5

DNA sequencing of KCNA5 in the 307 patients revealed six novel non-synonymous mutations [E48G, Y155C, A305T (twice), D322H, D469E, and P488S] in seven patients (Figure 1) and seven previously reported SNPs (R87Q, A251T, G309D, P310L, G558R, G568V, and R578K) in 12 patients (Supplementary material online, Table S2). No novel mutations were found in the controls. All mutation carriers were heterozygous, and in one of these, we also identified a variant (Y25C) in KCNAB2, but this was also found in one control. No additional mutations were identified in KCNAB2. The patients were screened in genes previously associated with AF (Supplementary material online), with no mutational findings.

Figure 1

Genetic screening. (A) Chromatograms from the six identified novel mutants showing the amino acid substitutions. (B) Molecular model of KV1.5 (modified from Yang et al.17). Red: mutations; green: rare variants.

The amino acid positions of all novel mutations are highly conserved across species, suggesting a functional importance (Supplementary material online, Figure S1). The mutations E48G and Y155C reside in the N-terminus of the channel protein, A305T and D322H are located in the extracellular S1–S2 loop, whereas D469E and P488S are located in the pore module S5–S6 (Figure 1B).

High frequency of rare variants in KCNA5 in patients with early-onset lone atrial fibrillation

Besides screening the controls for mutations found in the patients, we also screened the entire coding sequence of KCNA5, and found no mutations or deleterious variants. This resulted in a significantly higher frequency of both novel mutations (1.14 vs. 0%; P = 0.046) and deleterious variants (1.79 vs. 0%; P = 0.004) in the patients than in the controls.

Family history

The female patient carrying A305T in KCNA5 and Y25C in KCNAB2 had onset of palpitations at age 16 and documented persistent AF at age 18. ECG analysis at inclusion showed sinus rhythm with severely prolonged PR interval (360 ms), incomplete right bundle branch block, and prolonged QTc (490 ms).23 She has a structurally normal heart. Her son carries the same mutation in KCNA5 and has experienced paroxysmal palpitations and tachycardia since age 17. The symptoms have been increasing in severity and frequency in his 20s, yet event recordings and an exercise test have shown intermittent supraventricular extrasystoles and sinus tachycardia, but no documented AF. His 12-lead ECG was normal.

The other mutation carriers had onset of AF at age 22–35, but did not have family history of AF (Supplementary material online, Table S2).

Effects of mutations on subcellular localization

To analyse whether the mutations affected expression of the channels at the cell surface, we expressed KV1.5 WT, novel mutations, and rare variants in HEK293 cells and performed confocal imaging. Representative images are shown in Figure 2. The channel protein is expressed both in the endoplasmatic reticulum (ER) and, importantly, at the surface of the cells in KV1.5-WT, E48G, A305T, and D322H (Figure 2A). In contrast, the mutants Y155C and P488S appear to be completely retained in the ER, and surface expression of D469E is largely impaired, suggesting loss-of-function for these mutants (Figure 2B). We also analysed the subcellular localization of the rare variants G309D and G558R, as they were novel at the time of these experiments. They behaved as WT (Figure 2C).

Figure 2

Subcellular localization. Kv1.5 WT or mutations/variants were expressed in HEK293 cells. The cell surface was visualized by actin staining by phalloidin (red), the nucleus was visualized by DAPI (blue), and KV1.5 channels were stained green. Both WT and mutants E48G, A305T, and D322H demonstrated membrane expression as they were found in the cell surface co-localizing with the actin marker (A; far right; yellow staining and arrows). Y155C, D469E, and P488S (B) were retained in the ER (D469E only partially). G309D and G558R (C) behave like WT. Scale bar = 10 µm.

E48G, A305T, and D322H exert gain-of-function effects

To address functional effects of the mutants with preserved surface expression, we heterologously expressed KV1.5-E48G, KV1.5-A305T, and KV1.5-D322H in mammalian HEK293 cells and performed whole-cell patch-clamp experiments. Applying a standard voltage-step protocol, we measured steady-state and tail currents as a function of the voltage command (Figure 3A).

Figure 3

Steady-state activation of KV1.5-WT, E48G, A305T, or D322. (A) Representative recordings of HEK293 cells expressing WT or mutants (protocol shown in inset). Note that inserted tail-current traces of WT recording are shown at different scale. (B) Steady-state current–voltage relationship. Measurements were performed at the end of the current step. WT was significantly different from E48G, WT/E48G, A305T, WT/A305T, D322H, or WT/D322H by two-way ANOVA with a Bonferroni post hoc test (*P < 0.05; E48G, WT/E48G, A305T, WT/A305T, D322H, or WT/D322H vs. WT). (C) Normalized tail current–voltage relationship measured at −40 mV. Tail current is illustrated as a function of the voltage applied during the 250 ms activation steps. Measurements were performed at the start of the tail current. The absolute V1/2,act values are listed in Table 1. All mutants behaved significantly different from WT by one-way ANOVA with a Bonferroni post hoc test.

Both wild-type (WT) and mutant channels activated at around −30 mV, yet the mutants showed a significant increase in total current compared with WT (Figure 3B, Table 1). The potential where half of the channels are activated (V1/2,act) was determined by fitting the tail current with the Boltzmann equation. For WT, V1/2,act was −5.52 ± 1.12 mV, whereas all mutants displayed negative voltage shifts in the activation curves and significantly decreased V1/2,act values (E48G: −9.29 ± 0.89 mV; A305T: −10.88 ± 2.93 mV; D322H: −10.22 ± 1.90). This effect remained when the mutants were co-expressed with WT to mimic the heterozygous state in the patient (E48G/WT: −10.94 ± 2.94 mV; A305T/WT: −11.66 ± 2.58 mV; WT/D322H: −10.11 ± 1.27) (Figure 3C, Table 1).

View this table:
Table 1

Biophysical parameters for KV1.5 gain-of-function mutants associated with atrial fibrillation

Peak current at 20 mV (pA/pF)nV1/2 of steady-state activation (mV)knV1/2 of steady-state inactivation (mV)kn
WT493.0 ± 46.3625−5.52 ± 1.128.81 ± 0.6923−12.99 ± 2.316.09 ± 1.0510
E48G668.2 ± 71.53*18−9.29 ± 0.90*8.65 ± 0.7318−10.66 ± 2.506.39 ± 1.335
WT/E48G706.8 ± 62.66*8−10.94 ± 2.94*6.28 ± 0.547n.d.n.d.
A305T726.5 ± 80.97*9−10.88 ± 2.93*8.27 ± 1.109−8.53 ± 1.595.45 ± 0.395
WT/A305T745.7 ± 120.1*6−11.66 ± 2.58*6.27 ± 0.846n.d.n.d.
D322H931.9 ± 130.8***8−10.22 ± 1.90*6.81 ± 0.948−5.14 ± 2.12*5.26 ± 0.335
WT/D322H888.8 ± 107.5***11−10.11 ± 1.27*7.92 ± 1.049−5.45 ± 2.72*5.32 ± 0.588
  • n.d., not determined.

  • *P < 0.05, ***P < 0.001 (significantly different from WT hKV1.5).

Inactivation properties were quantified by comparing the current amplitude of the second +40 mV step with the initial +40 mV step. The potential where half of the channels are inactivated (V1/2,inact) was −16.10 ± 2.88 mV for WT. The mutants E48G, A305T, and D322H displayed positive voltage shift in the inactivation curves, with increased V1/2,inact for steady-state inactivation, but this was only significant for D322H (E48G: −10.66 ± 2.5 mV; A305T: −8.53 ± 1.59 mV; D322H: −5.14 ± 2.12) (Figure 4B). The decreased inactivation for D322H was still significant, when co-expressed with WT (−5.45 ± 2.72 mV). Hence, these three mutations led to gain-of-function compared with WT.

Figure 4

Steady-state inactivation of KV1.5-WT and D322H. (A) Representative recordings (protocol shown in inset). The cells were depolarized briefly at 40 mV from a holding potential of –80 mV and hyperpolarized for 2 s at −100 mV, then depolarized again for 5 s from −70 to 30 mV in 10 mV increments, and finally the current was measured at 40 mV, where the cell was held for 1 s. (B) Normalized peak current at 40 mV plotted as a function of the voltage applied during the 5 s pulse. Measurements were performed at the start of the 1 s pulse. Absolute V1/2,inact values are listed in Table 1. D322H and WT/D322H were significantly different from WT, as shown by one-way ANOVA with a Bonferroni post hoc test.

The D322H mutant appeared to be the most severe mutation, resulting in the largest increase in current density and changes in both activation and inactivation parameters. The principal effect of KV1.5 gain-of-function on atrial AP duration was modelled by introducing the effects of the D322H mutation into the human atrial cell model developed by Courtemanche et al.21 The model's IKur formulation was modified to qualitatively reflect the changes caused by the D322H mutation (a 1.9-fold increase in conductance and −4.7 and 7.9 mV shifts in V1/2 for activation and inactivation respectively; Supplementary material online, Table S3). Figure 5 depicts the effect of the mutation on the simulated atrial AP. The D322H mutation resulted in the triangulation of the AP morphology and shortened the APD from 280 to 222 ms compared with WT.

Figure 5

Modelling of the effect of the D322H mutation in a human atrial cell model. Simulated action potentials are shown for WT (blue) and D322H (green). The last action potentials in a train of 50 beats at a pacing rate of 1 Hz are shown.

We also assessed the mutations Y155C, D469E, and P488S in heterologous expression experiments. Compared with WT, these mutants showed a significant decrease in total current (Supplementary material online, Table S4 and Figure S2) in line with the previous notion that loss-of-function mutations of KV1.5 are associated with AF.12


We have investigated the prevalence of mutations in KCNA5 in early-onset lone AF patients and identified six novel mutations and several previously reported rare variants. We observed significantly higher prevalence of rare deleterious variants in KCNA5 in these patients, compared with our healthy control population and what is reported in the EVS. No other gene has been reported to have such a high frequency of rare variants associated with AF, suggesting that it is one of the most important genes involved in early-onset lone AF.

Three previous studies have reported loss-of-function mutations in KCNA5, leading to a decrease in IKur in patients with lone AF.12,16,17 This has given rise to the hypothesis that the prolongation of the atrial AP and the ERP lead to increased propensity for EADs, which increases the susceptibility to AF. Our findings regarding the loss-of-function mutations Y155C, D469E, and P488S support this notion. However, this is the first report on AF-associated KCNA5 mutations which lead to increased peak current densities and changed biophysical properties of the KV1.5 channel. These changes are likely to give rise to gain-of-function for IKur in affected individuals. As shown for D322H, the simulation of the effects in a human atrial cell model led to the triangulation of the AP morphology and shortening of the AP duration compared with WT. The resulting shortening of the ERP is likely to increase excitability in atrial tissue, suggesting a potential mechanism behind early-onset lone AF in patients harbouring KV1.5 gain-of-function mutations.

Association with disease susceptibility was underscored in several ways. First, the patients in our study were younger than in previous studies, rendering a genetic component likely.24 Second, we did not find any mutations in the entire coding region of KCNA5 in healthy controls. Third, the mutations found in our patients were absent in the EVS, which holds data on 6503 exomes, and the frequency of rare deleterious variants identified in our patients was significantly higher than what was found in the EVS. This is supported by Mann et al.,25 who recently reported a higher number of rare variants in KCNA5 in patients with AF compared with controls. However, it should be noted that the data on the rare variants in our study are results of analyses of different Caucasian populations, using different screening techniques and that they thus need to be replicated.

Finally, there has been reported limited variability in the coding region of this gene in a polymorphism screening study from 2005.26 We ruled out functional hemizygosity, a potentially silent condition, by demonstrating preserved trafficking of the channels harbouring gain-of-function mutations to the cell surface, and potential loss-of-function was shown for the mutated channels that were retained intracellularly. Altogether, this excludes that these mutations are common polymorphisms, and underscores that variations in this gene most likely have functional influence on the protein product.

We found that the risk of having one or more first-degree relative with AF was almost tripled in women with early-onset lone AF compared with men (OR 2.8; 95% CI 1.4–5.4; P = 0.0008). This is consistent with data from Chen et al.,27 who reported similar findings in older lone AF patients, and the Framingham Heart Study, where maternal AF was shown to increase the risk of AF in an individual almost six-fold, compared with a 2.3-fold increase with paternal AF.5 This shows that the male predominance of lone AF is diminished in the setting of inherited AF.

The rare variants G309D, G558R, and G568V are reported in the EVS, and the latter also in the NCBI dbSNP database (G309D: 1/7.014 alleles, G558R and G568V: 2/7.020 alleles each). These are very low frequencies, and the EVS sample consists of patients collected from multiple, diverse, richly phenotyped populations collected throughout the USA. Accordingly, these five alleles might have been identified in patients with AF or other disease, and we cannot exclude them from the spectrum of susceptibility variants for AF. We identified one variant in KCNAB2 which was, however, found in one of our healthy controls and also reported in 4 out of 8600 European alleles in EVS. KCNA2B is highly conserved throughout evolution, and its gene product KVβ2 exerts a variety of functions in human cells, but it has yet not been linked to any disease.19 We did not find any compelling evidence for an association with lone AF.

The current model for induction and maintenance of AF is a shortening of the ERP in the human atria, acting as a substrate for re-entrant arrhythmia.10 This hypothesis is supported by animal models28,29 and reports of gain-of-function mutations in genes encoding cardiac IKs (KCNQ1/KCNE2) and IK1 (KCNJ2).9 Our data regarding KCNA5 mutations support this theory.


We limited our analysis to the coding regions of KCNA5; hence, the possibility of mutations occurring in regions beyond coding regions cannot be excluded. Genetic testing of family members was limited as some were unavailable. Functional studies were performed using conventional heterologous expression systems, in which the environments differ from that in the native cardiomyocyte. However, in vitro studies employing mammalian cells are an acknowledged way of testing functional effects of mutated ion channels.

Population stratification is an important consideration in any genetic study. However, we do not consider this as a major issue in this study, as Oslo and Copenhagen are located within 500 km, and Scandinavian populations have been shown to be genetically similar.30,31 We have characterized the distribution of the seven SNPs most highly associated with AF and the PR interval in the Danish control group. When comparing the minor allele frequencies in the Danish control group with the ones found in the control groups presented in previous GWAS (reviewed in Mahida and Ellinor32), we find that they are similar (Supplementary material online, Table S5). This indicates that our control group is representative with respect to other AF-causing variants. Accordingly, the driving force of the association with AF is not caused by differences in SNP frequencies, and SNP frequencies in AF-associated variants in Caucasians appear similar across widespread geographical areas.


In this study, we have identified six novel mutations in KCNA5 in seven patients with early-onset lone AF. We show for the first time that mutations in this gene can lead to gain-of-function for IKur in vitro. Increased IKur may result in AF through the shortening of the atrial AP and re-entry mechanisms. We also identified loss-of-function mutations associated with early-onset lone AF. Hence, this study supports the hypothesis that both gain- and loss-of-function mutations in potassium currents enhance AF susceptibility and extend this working model to include KCNA5/KV1.5.


The Danish National Research Foundation; The John and Birthe Meyer Foundation; The Arvid Nilsson Foundation; The Danish Heart Foundation (grant no. 11-04-R84-A3333-22660); Fondsbørsvekselerer Henry Hansen og Hustru Karla Hansen, født Westergaards Stipend; The Memorial Fund of Eva and Henry Frænkel; Dagmar Marshalls Foundation; Aase og Ejnar Danielsens Foundation; Odd Fellow's Medical Scientific Research Foundation, Norway; Beckett Foundation; Torben og Alice Frimodts Foundation, and Direktør Ib Henriksens Foundation.

Conflicts of interest: none declared.


We are grateful for the work performed by Katrine Kastberg and Mina Ghasemilee (Laboratory of Molecular Cardiology, Denmark), Nancy Thomsen (Ion Channel Group), and Mona Olufsen, Steve Enger, and Hilde Larhammer (Medical Research Department at Bærum Hospital, Norway). We thank Knut Gjesdal (Oslo University Hospital) for recruiting patients. We thank all patients participating in this study in Norway and Denmark.


  • These authors contributed equally to the work.


View Abstract