Introduction

Sinus node dysfunction (SND) is characterized by the slow heart rate accompanied by rhythm disturbance. Patients with SND usually show a benign clinical course, but some patients with severe dysfunction require an artificial pacemaker implantation. Most cardiac arrhythmias are associated with acquired pathological conditions of the heart, including ischemic heart disease, hypertensive heart disease and congestive heart failure. Mutations in genes for cardiac ion channels are known to cause SND1, 2 and other arrhythmias, including long QT syndrome (LQTS),1, 3, 4, 5, 6, 7, 8, 9 short QT syndrome,10, 11 familial atrial fibrillation,12, 13 Brugada syndrome (BrS),14 progressive cardiac conduction disturbance15 and polymorphic ventricular tachycardia (PVT).16 Functional alterations caused by the channel gene mutations have clarified the role of each channel in cardiac excitation and conduction in normal and abnormal conditions.17

We earlier reported a mutation of gene for hyperpolarization-activated channel 4 (HCN4 channel or If channel) in a patient suffering from SND, who also showed QT prolongation in ECG and developed PVT.1 The If channel is preferentially expressed in cells with automaticity controlling the heart rate, and it is called as a pacemaker channel. A functional study of the mutation showed a severe loss of function of HCN4 channel, which may explain the SND phenotype. The If channel is also expressed in the working myocardium,1, 18 raising a question whether there is another role in the working myocardium other than that involved in the automaticity of atrioventricular node and His-Purkinje cells. The PVT observed in the patient with the mutation implied that the If channel dysfunction might be associated with provoking ventricular arrhythmia.

It is well known that bradycardia (lower heart rate) is associated with a higher incidence of sudden death during sleep in LQTS type 3 (LQTS3) and BrS. LQTS3 and about 30% of BrS are caused by SCN5A mutations, and compound heterozygotes of SCN5A mutations were reported to express SND, suggesting that the SCN5A-associated SND was a recessive disorder. On the other hand, in the majority of BrS patients who do not carry the SCN5A mutations, it remains unknown which genetic factor confers a risk of bradycardia-induced ventricular arrhythmias. Concerning the LQTS, it was reported that double mutation cases of two arrhythmia-associated genes were 7.9%, relatively common than formerly expected, and that cumulative effects of disease-related genes would increase the risk of arrhythmia in severe cases.19

In this study, we searched for mutations in the HCN4 channel gene (HCN4) in patients with ventricular arrhythmia, analysed the functional alteration caused by the mutation and investigated how the pacemaker channel was involved in the bradycardia-induced ventricular arrhythmia by using a simulation model.

Materials and methods

Subjects

Fifty-four and thirty-three genetically unrelated patients with LQTS and BrS, respectively, were the subjects in this study. Control subjects were 190 genetically unrelated healthy individuals. A blood sample was obtained from each subject after informed consent was obtained. Genomic DNAs were extracted from the blood samples and analysed for mutations in the known disease genes for cardiac arrhythmia, including KCNQ1, KCNH2, SCN5A, KCNE1, KCNE2 and KCNJ2 by PCR and single-strand conformational polymorphism (SSCP) method,20 using primers reported in the literatures.7, 21, 22, 23 The research protocol was approved by the Ethics Reviewing Committee of Medical Research Institute, Tokyo Medical and Dental University.

Mutational screening in HCN4

Exons of HCN4 were separately amplified using pairs of primers (Table 1). The PCR products from patients were searched for sequence variations by the PCR-SSCP method20 and subsequent sequencing of PCR products cloned into pCRII vectors (Invitrogen, Carlsbad, CA, USA). Several clones were sequenced to ensure that the sequence variation was not the PCR artefacts.

Table 1 PCR primers used in the analysis of genomic gene for human HCN4a (5′ → 3′)
Full size table

Functional analysis of a splice site mutation

To investigate the functional changes caused by the splicing donor site mutation, exon 2 and its flanking intron sequences of HCN4 were amplified by PCR with primers ET2-F (5′-GAATTCTCTGTAACTTGTCCTCTCTC-3′) and either ET2-WR (5′-GATATCTTACCTCTTCCCACTGGTG-3′) or ET2-MR (5′-GATATCTTACTCACCTCTTCCCACT-3′) from the patient carrying the mutation. The PCR products were cloned into pSPL3 at EcoRI and EcoRV sites. Several independent clones were isolated and sequenced to ensure that the fragment was free from errors in PCR. COS7 cells were cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal calf serum (GIBCO BRL) and antibiotics. The cells were transfected with either wild type or mutant construct (2 μg per 10 cm dish) using SuperfectAMINE (Qiagen Japan, Tokyo, Japan) according to the manufacturer's instructions. Total mRNA from the transfected cells was extracted and used for generating cDNA templates. SSCP analysis was performed on the reverse transcription (RT)-PCR products amplified with an EX2S primer (5′-CGATATATTCACCAGTGGGAA-3′) and a VA primer (5′-CACCTTCTGATAGGCAGCCT-3′); the former primer corresponded to exon 2 of HCN4 and the latter primer was corresponded to the pSPL3-encoded exon. An aberrant splicing was detected by using sequence-specific primers. When the natural splice donor site was used, the NS (5′-TTCACCAGTGGGAAGAGACC-3′) and VA primers would generate a PCR product. On the other hand, when the mutant sequence could work as a splice donor, the aberrant splicing product should be amplified by the IS (5′-CCAGTGGGAAGAGGTGAACC-3′) and VA primers. The sequence-specific primers, NS and IS, span the junction of the HCN4 exon and the vector-derived exon.

Electrophysiological analysis of HCN4 channel

COS7 cells were co-transfected with 0.4 μg pEGFP-C1 (Clontech, Mountain View, CA, USA) and 1.6 μg rabbit HCN4 cDNA (GenBank Accession no. AB022927) cloned into PCI (Promega, Madison, WI, USA). The transfected cells were removed from the 35 mm culture dish 36–48 h after transfection and placed into a chamber on the stage of an inverted microscope (Diaphoto TMD, Nikon). Thereafter, currents were recorded by a whole-cell patch-clamp configuration using an amplifier (Axopatch 200B, Axon Instrument) from cells expressing co-transfected pEGFP-C1, as described previously.1 After forming the whole cell configuration, cell membrane capacitance was estimated by analysing the capacity transient elicited by 5 mV hyperpolarizing pulses. Series resistance compensation was done by 50–70% using the circuit built into an amplifier (Axopatch 200B, Axon Instruments, Foster City, CA USA). The voltage-dependence of tail currents was analysed by fitting to the Boltzmann equation of

according to the method of Chen J et al.,24 where V1/2, κ and min P0 are half maximal activation voltage, slope factor and the minimum open probability (defined as the minimum value of relative tail current), respectively. Software (pCLAMP8, Axon Instruments) was used to generate voltage protocols, to acquire data and to analyse current signals. All experiments were done at 34±1.0 °C. Data were obtained from eight cells and represented by mean±s.e.m.

Computer simulation of If current on ventricular action potential

On the basis of our experimental data, we implemented the current through the If channel into action potential models, including two human ventricular action potential models (the Priebe–Beuckelmann model for heart failure25 and the Tusscher–Noble–Noble–Panfilov model26) and the Luo–Rudy model27 that was originally proposed for guinea-pig ventricular myocytes. If was formulated by a classical Hodgkin–Huxley-type gating scheme, using the variable for activation gate, y,

in which gf,Na and gf,K represent the conductance of Na+ and K+ components of If, respectively, and ENa and EK represent equilibrium potentials for Na+ and K+. The time constant (τ) and activation curve (yinf) obtained in this study were used for the implementation of the If current. The value of gf,Na and gf,K, and the current density of If were from a report analysing human ventricular myocytes.28 gf,Na was set to 0.0032 mS μF−1 and gf,K to 0.011 mS μF−1, consistent with a value of PK/PNa=3.3 determined by the reversal potential measurement in this study. The outcome of If incorporation was qualitatively similar in different action potential models.

Results

Identification of HCN4 mutation in patients with ventricular arrhythmia

To identify arrhythmia-related mutations in patients with idiopathic ventricular arrhythmia, we analysed 54 LQTS patients and 33 BrS patients for mutations in the known arrhythmia-causing genes, including KCNQ1, KCNH2, SCN5A, KCNE1, KCNE2, and KCNJ2. We identified Gly229Asn (G-to-A transition at position 686) and Phe275del (3 nucleotides deletion at position 824–826) as LQTS-associated KCNQ1 mutations. In addition, Phe627Ile (T-to-A transition at position 1879), Phe640del (3 nucleotides deletion at positions 1918–1920), Leu987fs (2 nucleotides deletion at positions 2959 and 2960), Arg1033fs (C insertion at position 3098), Val1074fs (T deletion at position 3221) were found in KCNH2 from five different LQTS patients. A Gly146Ser (G-to-A transition at position 436) was found in KCNJ2 from an LQTS patient with periodic paralysis (LQTS7 phenotype). Regarding SCN5A, Met1296fs (T insertion at position 3888) was found in a BrS patient. These mutations were not found in 380 control chromosomes and no compound heterozygotes of mutations were found in this study.

We next investigated for HCN4 mutations in the patients. The analysis revealed an HCN4 mutation in heterozygous state in a symptomatic idiopathic ventricular tachycardia patient who had no SCN5A mutation. This mutation, IVS2DS, was an insertion of four bases at a splicing junction of exon 2 and intron 2 (Figure 1b), which was not found in the controls. The patient with IVS2DS, a 41-year-old man who suffered from recurrent episodes of syncope at rest, was admitted to a hospital, and his ECG showed typical ST-elevation in V1-2 (saddle-back type as shown in V2, Figure 1c) and incomplete right bundle branch block. The QT intervals corrected by the heart rate, QTc, were between 0.450 and 0.475 s, prolonged to an upper limit of normal range. His 24-h ECG recording showed PVT during sleep. An electrophysiological study showed that ventricular fibrillation was reproducibly induced by programmed ventricular stimulation without any drug provocation. From the clinical manifestation and response to electrophysiological study, he was diagnosed as having BrS and was implanted with an implantable cardioverter defibrillator. He had no apparent family history of sudden cardiac death, but genetic analysis of his family members could not be carried out because they refused genetic testing.

Figure 1
figure 1

Genetic analysis of patients with idiopathic arrhythmia for HCN4 mutation. (a) The variant, IVS2DS, is represented on the genomic structure of HCN4. (b) Sequence profiles of plasmid clones containing normal and IVS2DS-derived exon 2 and intron 2. IVS2DS was an insertion of four nucleotides (enclosed within square) at the splicing donor site. An arrowhead and an arrow indicate a normal exon–intron boundary and a putative splicing position in the variant allele, respectively. (c) ECG of patient with the HCN4 mutation showed typical ST-elevation in V1-2 and incomplete right bundle branch block.

Full size image

Functional analysis of IVS2DS

As the four-base insertion was found at the splicing donor site, Inserted bases might be expressed as a part of mRNA (Figure 1b). When the insertion was transcribed as a part of mRNA, it was predicted to cause a frameshift with an Ile-to-Val change at codon 404 and following non-relevant 43 amino acids, that is, HCN4 carrying IVS2DS might encode a truncated HCN4 channel. As heart biopsy sample was not obtained from the patient, the possible functional alteration owing to the splice donor site mutation was investigated by using a transfection experiment with artificial mini-gene constructs. PCR fragments containing the splicing junction of wild type (WT) and variant (or insertion, IS) alleles from the patients were cloned into pSPL3 (Figure 2a) and transfected into COS7 cells to analyse the splicing products by the RT-PCR method. The IS construct was found to generate an additional RT-PCR product not present in cells transfected with the WT construct (Figure 2b), suggesting that the additional 4 bases after the natural splice donor site were incorporated into mRNA. The aberrant splicing was detected by the RT-PCR method using combinations of sequence-specific primers, as the insertion-specific RT-PCR product was found only in the cells transfected with the IS construct (Figure 2c). The splice product using the natural donor site was also found with IS (Figure 2c), and the SSCP profile suggested that almost equal amount of normal and abnormal splicing products were expressed (Figure 2b).

Figure 2
figure 2

Splicing analysis of IVS2DS. (a) Schematic representations of wild type and variant constructs cloned into pSPL3. Exon 2 of HCN4 is represented by a shaded box and additional four nucleotides of GTGA in IVS2DS are indicated by a filled box. Normal splicing events are shown by α, whereas a predicted aberrant splicing is denoted by β. Positions of primers used in the RT-PCR analysis are shown: EX2S,exon2-specific primer; NS, normal splicing-specific primer; IS, insertion-associated splicing-specific primer; and VA, vector-associated primer. (b) SSCP pattern of RT-PCR products obtained with EX2S and VA primers from COS7 cells transfected with either wild-type (left) or variant (right) constructs. An additional RT-PCR product (arrowhead) was detected only for the variant. (c) RT-PCR analysis to detect the aberrant splicing. Lanes 1, 3 and 5 are derived from cells transfected with the WT construct, and lanes 2, 4 and 6 are derived from that with the variant construct. The primers used are EX2S and VA in lanes1 and 2, NS and VA in lanes 3 and 4, and IS and VA in lanes 5 and 6.

Full size image

Electrophysiological analysis of HCN4 channel

To explore the possible contribution of the If current to the action potential of ventricular cells, we used computer simulation studies. As there was no previous data for incorporating If to be used in the simulation study, we conducted the whole-cell patch-clamp analysis of COS7 cells transfected with a rabbit HCN4 construct to obtain the functional characteristics of the HCN4 channel. When hyperpolarizing test pulses were applied from a holding potential of −30 mV, voltage- and time-dependent inward currents were activated at potentials negative to −60 mV (Figures 3a and c), which were similar to the If current recorded in native ventricular cardiomyocytes except for the less negative activation voltage.28 Tail currents were obtained at membrane potential shifted to −100 mV after various test hyperpolarizations (Figure 3b). The time-course of current activation became faster (Figure 3d) with hyperpolarization of test pulses negative to −60 mV. Voltage-dependence of If current was analysed by fitting to the Boltzmann equation to obtain various parameters of the HCN4 channel function; V1/2=−77.98±1.06, κ=9.11±1.03, and min P0 was 0.109±0.029.

Figure 3
figure 3

Electrophysiological properties of hyperpolarization-activated current (If) generated by HCN4 channel. Representative records of HCN4 channel currents and data analysis are shown. (a) The If current through HCN4 channel was activated during hyperpolarizing test pulses of 3 s between −40 and −120 mV in 10 mV increments from a holding potential of −30 mV, and tail currents were obtained at −100 mV after the test pulses. The data were analysed to show If–V relationship curve (b) and normalized tail current–V relationship curve (c). V1/2, κ and min P0 were −77.98±1.06, 9.11±1.03 and 0.109±0.029, respectively. (d) Deactivation currents were recorded during test voltages between −60 and 30 mV in 10 mV increments after the conditioning pulse to −100 mV from the holding potential of −30 mV. The data for activation and deactivation currents were used for drawing a time constant curve in logarithmic scale (e). Absence of error bars indicates errors smaller than the symbol size.

Full size image

Effects of If current on action potential in a simulation model

The If channels expressed in the heart are mainly composed of HCN2 and HCN4 channels. According to the previous reports,24, 29 electrophysiological properties, V1/2 and τ, of HCN2 channel suggested that more HCN4 channel could be open than the HCN2 channel under the physiological condition, and min P0 of the HCN2 channel was similar to that obtained in our study. Thus, V1/2, τ, min P0 and time constant of HCN4 channel obtained in this study were used as parameters of the If channel in the simulation models.30

Figure 4 illustrates the effects of If current introduced into a human ventricular action model proposed by Tusscher et al. (Tusscher–Noble–Noble–Panfilov model).26 As shown in Figure 4a, incorporation of If current shortened the action potential duration (APD) and depolarized the resting membrane potential minimally. Being a hyperpolarization-activated current, If was activated during the diastole and deactivated following the upstroke of the action potential. This gating produced a decaying outward current at the plateau phase, as If had mixed the conductance for Na+ and K+ leading to a reversal potential around −30 mV.

Figure 4
figure 4

Ventricular action potential model incorporating the If current. Properties of the If currents were incorporated into the TNNP human ventricular action potential model based on the results shown in Figure 3. (a) Action potentials with and without If incorporation were shown with red and black lines, respectively. Models were paced with a basic cycle length (BCL) of 1000 ms. (b) Temporal profiles of current systems during the action potential with If incorporation. Inward rectifier K current (IK1): blue; delayed rectifier K currents (IKs and IKr): green; Na–Ca exchange current (INa-Ca): purple. (c) Temporal profiles of If current (red line). Current amplitudes with BCL of 500 ms (dotted black line) and 2000 ms (thin black line) were superposed. (d) Values of activation gate (red line). Values with BCL of 500 ms (dotted black line) and 2000 ms (thin black line) were superposed.

Full size image

In the Tusscher–Noble–Noble–Panfilov model, the amplitude of If was comparable with other outward currents, including IKr and IKs. It should be noted that the amplitude of If was stimulus-frequency dependent, as the degree of activation was determined during diastolic interval (Figure 4d). Hence, the extent of APD shortening was augmented when the basic cycle length (BCL) was increased. It was found that, with BCL of 1000, 2000 and 4000 ms, inclusion of If current shortened the APD90 by 9.6 ms (3.5% of the control APD90), 10.2 ms (3.7%) and 10.6 ms (3.8%), respectively. The APD shortening by the incorporation of If was a consistent finding in all action potential models, whereas the extent of shortening showed variations. When If was introduced into the Luo–Rudy model,27 decrease in APD90 was 1.9 ms or 1% of the control value when BCL was set to 1000 ms.

Effects of If current on action potentials under pathological conditions

Triggering of arrhythmias in the inherited arrhythmogenic diseases generally requires particular situations, such as physical or emotional stress and additional APD prolongation associated with specific medications or bradycardia. The underlying mechanisms of these events may be analogous, for example, with the onset of lethal arrhythmias in patients with cardiac failure or myocardial ischemia. Therefore, we were concerned with the effects of If current when cardiomyocytes were under pathological or unusual situations.

Figure 5a shows the effect of If on abnormal ventricular automatic activities constructed in the Tusscher–Noble–Noble–Panfilov model. In this case, the automaticity was elicited by the application of a depolarizing current following the augmentation of ICa-L and IKs currents, which mimics the marginal area of myocardial infarction under sympathetic stimulation. As expected for the pacemaker current, implementation of If accelerated the rhythm because the current was inward, below the reversal potential. Figure 5b, on the other hand, shows the effect of plateau outward If current on the action potential repolarization. In this case, the Priebe–Beuckleman action potential model25 was used, as it was mainly concerned with the modelling of cardiac failure. In the Priebe–Beuckleman ‘heart failure’ model, reduction in the IKr current produced early after depolarizations (EADs).25 As shown in Figure 5b, unusual APD prolongation owing to EAD was prevented by the inclusion of the If current.

Figure 5
figure 5

Effects of If current on (a) ventricular abnormal automaticity, and (b) EAD in heart failure model. (a) Application of depolarizing injury current (1.8 μA μF−1) into the TNNP model produced repetitive automatic activity. To simulate β-adrenergic stimulation observed in myocardial infarction, the amplitude of ICaL and IKs were increased by factors of 4 and 2, respectively. Incorporation of If into this situation accelerated the automaticity as shown by a red line. (b) The PB human ventricular action potential model was set to ‘heart failure’, characterized by long APD and slowed decay of Ca transient.25 Reduction of IKr in the PB heart failure model elicited EADs as shown by a black line. Incorporation of If prevented EAD production (red line).

Full size image

Discussion

In this study, we found several mutations in the channel genes from patients with ventricular arrhythmia, especially a novel HCN4 mutation in a symptomatic BrS patient, which caused abnormal splicing. In addition, we showed from the computer simulations that the If channel produced a background current contributing to the action potential repolarization of ventricular cardiomyocytes. The background current was generated by a mixed ion-selectivity for Na+ and K+, and incomplete closure for deactivation gate of If channels. This is the first report suggesting that mutation in HCN4 may render the bradycardia-induced ventricular arrhythmias.

Our study suggested that IVS2DS would partially affect the splicing. The presence of unaffected splicing may explain why SND was not observed in the patient. Although we did not prove the presence of truncated HCN4 channel in the patient, such channel should not be functional. When the abnormal splicing had occurred after the 4-bp insertion (at the abnormal donor site) and ended at the normal accepter site before exon 3, the inserted four nucleotides were included into mRNA, which was predicted to encode for a truncated channel of 447 residues containing non-relevant 44 amino-acid sequences at the C-terminal side. The predicted truncated channel should not form functional tetramers, and hence would be defective. It could be speculated that the abnormal splicing product might be unstable owing to the nonsense-mediated decay of mRNA. On the other hand, a cyclic nucleotide binding a domain-lacking mutation, P544fs, in HCN4 was associated with idiopathic SND, and the phenotype was a defective response of heart rate to exercise.2 The mutations IVS2DS and G553N1 were found in patients with frequent recurrence of ventricular tachycardia or syncope but were found neither in healthy controls nor in asymptomatic arrhythmia patients who had never experienced any life-threatening arrhythmia. These observations may provide an additional clue linking the HCN4 mutations and arrhythmogenesis. HCN4 mutations would affect channel property even if they were not responsible for any overt phenotypes, and exposing arrhythmic patients to slow heart rate was considered to put them into danger.

The If channels encoded by HCN2 and HCN4 in the heart are called as pacemaker channels, and expressed preferentially in cells with automaticity, thus playing a crucial role in impulse formation in the sinus node cells and cells in the specialized conduction system. However, it remains to be clarified whether and how the If channels would function in the working cardiomyocytes because of its ubiquitous expression in the heart.1, 18 We incorporated the If current into three ventricular action potential models (the Tusscher–Noble–Noble–Panfilov model for human ventricular cardiomyocytes, the Priebe–Beuckelmann model for ventricular myocytes adopted for human heart failure and the Ruo–Rudy model that is widely used for ventricular cardiomyocytes) based on the kinetic parameters experimentally determined in this study (Figure 3) and on the current amplitude from the previous report using human ventricular cardiomyocytes.28 Our studies showed the inward If current during the diastole and the outward If current during the plateau of action potentials. At the physiological conditions, however, the effect of If on action potential was not prominent. The If inward current depolarized the resting potential a little, whereas the plateau outward current contributed to slight a abbreviation of APD (Figure 4). This was a common finding of If incorporation in all three simulation models tested in this study, although the effects on action potential parameters showed quantitative variations.

If repeats the cycle of activation at the diastole phase and deactivation during the action potential. When the action potential was elicited, the outward current amplitude during the early phase of repolarization was determined by the duration and the voltage of resting state, and also by the persistent component of the activation curve, minP0 (Figure 4). Therefore, the effects of If should be exaggerated during bradycardia, which may provide with long diastolic intervals. In other words, If may contribute to breaking down the excessive prolongation of APD when heart rates were decreased.

As the membrane conductance of ventricular cardiomyocytes is extremely low at the plateau phase, only a slight change in current amplitude may affect the membrane potential to determine APD. We showed here an HCN4 mutation that would reduce the If current owing to the possible formation of a truncated channel. Therefore, there might be circumstances where this type of HCN4 mutation contributed to arrhythmogenesis through the prolongation of APD, as it ‘pushed off the safety catch’. Our simulation results in the Priebe–Beuckelmann model (Figure 5) support this assumption. In fact, it was possible to produce situations where the presence of If current played critical roles to suppress EAD in all three models.

In conclusion, we identified a novel HCN4 mutation in 1 out of 33 (3%) patients with BrS. The mutation was showed to cause splicing abnormality in transfected cells. Simulation studies suggested that the If currents produced by the HCN4 channel played a preventive role against the bradycardia-induced ventricular arrhythmia.