[ Dario DiFrancesco is Professor of Physiology at the University of Milano where he heads The PaceLab in the Department of Biosciences. His research focuses on basic and clinically-relevant properties of the cardiac pacemaker (funny) current, which he and coworkers first described in 1979. Decades of investigation have established the role of f-channels in cardiac pacemaking and heart rate control. Recent interest in f-channels has grown because of their relevance to development of “biological” pacemakers, pharmacological control of heart rate by rate-reducing agents such as ivabradine, marketed as therapeutic tool against angina and heart failure, and familial types of f-channel-dependent arrhythmias.]
Abstract Many diverse data support a role of the funny current (If) in pacemaking and heart rate control. Among them, clinically relevant applications have special impact, since they translate the concept of funny channel-based pacemaking into practical developments of potential therapeutic value. For example, the If role in pacemaking is the basis for a pharmacological approach to heart rate control. Ivabradine, a specific f-channel blocker acting as a ‘pure heart rate reducing’ agent, is used today as a therapeutic tool in chronic stable angina and heart failure. Also, the pacemaking capability of funny channels has been the main rationale behind the development of ‘biological’ pacemakers, whose aim is to eventually replace the electronic pacemakers implanted today. Finally, the role of If in pacemaking implies that dysfunctional funny channels can contribute to rhythm disorders. This consideration has led to the search for inheritable forms of cardiac arrhythmias potentially linked to functional changes of HCN4 channels, the molecular correlates of funny channels. This review addresses recent reports where investigation of families with arrhythmias has allowed the identification of specific HCN4 channel mutations associated with different types of rhythm disorders, specifically with sinus bradycardia. The perspective of future research in this field is also addressed.
cyclic nucleotide-binding domain
It is well established that the role played by the funny current in the heart relates principally to the generation and control of pacemaker activity in the sinoatrial node (SAN) region, although the precise extent to which f-channels contribute to determine pacemaking and rate control within the framework of the more complex set of participating mechanisms is still fiercely debated (Lakatta & DiFrancesco, 2009).
Among a whole variety of experimental results and clinically relevant applications demonstrating the contribution of If to pacemaking (Barbuti et al. 2007; DiFrancesco, 2010; DiFrancesco & Noble, 2012), genetic evidence has grown rapidly in the last few years that mutations in HCN4, the hyperpolarization-activated, cyclic-nucleotide-gated (HCN) channel isoform expressed in pacemaker tissue, are associated with rhythm disorders.
It also appears from experimental data that the role of f-channels need not be confined to the control of normal spontaneous activity in the primary pacemaker region of the heart. For example, progressive reduction of f-channel expression in pacemaker cells of cardiac-specific, inducible HCN4 knockout mice eventually leads to lethal atrioventricular block, as well as the expected slowing of SAN rate, suggesting a previously unsuspected role of HCN4 in atrioventricular node function (Baruscotti et al. 2011). Also, overexpression of If in ventricular cells from human failing hearts in dilated or ischaemic cardiomyopathy represents a potentially important pro-arrhythmogenic mechanism (Cerbai et al. 2001; Stillitano et al. 2008). Finally, it is known that ectopic beats originating in pulmonary veins are important focal sources that can initiate atrial fibrillation (Haissaguerre et al. 1998; Chen et al. 1999); pulmonary veins contain f-channel-expressing pacemaker cells which contribute to ectopic beat generation and arrhythmogenesis (Chen et al. 2009; Suenari et al. 2012), suggesting that dysfunctional f-channels can be involved in atrial fibrillation.
It is therefore important to investigate by candidate- gene screening if mutations of the HCN4 gene affecting channel function are associated with arrhythmias. Here I discuss recent reports showing that this association exists, and that identified HCN4 mutations may indeed contribute to rhythm disorders such as asymptomatic and symptomatic sinus bradycardia, tachycardia–bradycardia syndrome, atrial fibrillation, chronotropic incompetence and more complex arrhythmias.
Dysfunctional HCN4 mutations and rhythm disorders
L573X Early evidence indicating a potential association between HCN4 mutations and irregular cardiac rhythm involved a 66-year-old woman with severe sinus bradycardia, intermittent atrial fibrillation (AF) and chronotropic incompetence (Schulze-Bahr et al. 2003). Genetic screening identified a heterozygous mutation in HCN4 consisting of a single base deletion (1631 delC), which generated a truncated protein (HCN4-573X). Residue 573 is in the C-linker (Fig. 1), implying that mutated proteins lack the cyclic nucleotide-binding domain (CNBD). As expected, mutated channels did not respond to cAMP (Fig. 2). Surprisingly, heteromeric wild-type-mutant channels were also cAMP insensitive. This result contrasts with previous data indicating that cotransfection of cAMP-responding and cAMP-non-responding subunits leads to an intermediate cAMP response; this has been shown for example with the cotransfection of HCN4 and HCN1 isoforms (Altomare et al. 2003) or of wild-type and cAMP-insensitive, R591E mutant HCN2 channel subunits (Ulens & Siegelbaum, 2003).
A limit of this study is that an extensive pedigree investigation of the family of this patient was not performed. With data referring to a single patient, familial inheritance could not be examined and it was not possible to positively correlate the arrhythmic phenotype with the HCN4 mutation.
D553N A second report investigated a small family with a complex set of cardiac disorders (Ueda et al. 2004). Three members of this family were found to carry the HCN4 heterozygous mutation D553N (Fig. 1). The proband suffered from recurrent syncope, severe bradycardia, long QT (LQT), polymorphic ventricular tachycardia and torsade de pointes. Two other family members with the mutation had LQT, but whether they also shared other symptoms among the array of disorders manifest in the proband is not clear. Functional studies showed that mutant channels have a reduced membrane expression relative to wild-type channels, suggesting that D553N is a loss-of-function mutation interfering in a dominant-negative manner with the correct protein trafficking to the plasma membrane. This interpretation was supported by the decreased current recorded when transfecting heterozygous wild-type/mutant channels (Fig. 3A). Surprisingly, the mutation caused a rightward shift in the time constant curve, but did not cause any shift in the voltage dependence of activation (Fig. 3C).
A limit of this study is that in the presence of the complex array of symptoms of the proband and in the absence of a clear indication of which specific aspect was common to all mutation carriers, it is difficult to establish a direct correlation between the D553N mutation and the severity and complexity of the rhythm disorders described. Thus, while the loss-of-function change attributed to a trafficking defect can rationally be linked to bradycardia, it is not obviously linked to other symptoms involving ventricular features like LQT, polymorphic ventricular tachycardia or torsade de pointes. These more complex arrhythmias might arise, for example, from a mutation in an additional ion channel.
S672R According to Buraei & Yang, 2012, ‘the first HCN4 mutation to be unequivocally identified as a direct cause of arrhythmia’ was the heterozygous mutation S672R described in 2006 in a family with asymptomatic sinus bradycardia (Milanesi et al. 2006; Fig. 4). The mutation was expressed in a substantial fraction of the family (15 out of 27 members); importantly, all the individuals with the mutation had low resting heart rates (≤60 beats min−1, mean rate = 52.2 beats min−1), while all the wild-type individuals had normal rates (>60 beats min−1, mean rate = 73.2 beats min−1) (Fig. 4A). These results indicated that expression of the mutation and the bradycardic phenotype were highly correlated; the LOD (logarithm of odds) score of the correlation was indeed very high (5.47).
S672 is a highly conserved residue of the CNBD of the channel, located near the cAMP-binding pocket (Fig. 4C). This suggested that the mutation could affect cAMP-induced activation (DiFrancesco & Tortora, 1991) of the channel. However, functional studies showed that wild-type and mutant channels had similar responses to cAMP, implying a similar cAMP dependence; instead, mutant channels had a more negative activation voltage range than wild-type channels (Fig. 4B), indicating intrinsic new gating properties of the mutated channel. S672R can be considered, therefore, as a loss-of-function mutation which causes a decreased inward current flow during diastole, explaining the slower than normal heart rate of individuals with the mutation.
Shifting of the voltage dependence of current activation to more negative voltages is an action similar to that induced by ACh and consequent inhibition of adenylate cyclase and reduction of cAMP synthesis (DiFrancesco & Tromba, 1988; DiFrancesco et al. 1989). In fact, it is possible to verify that the S672R-induced negative shift of 4.9 mV is adequate to account for the slowing observed (29%). This can be done by comparing these data with existing data on the effects of ACh on the If activation curve and rate slowing in single cells (DiFrancesco et al. 1989). In Fig. 4D, the fractional rate slowing is plotted against the shift of the If activation curve corresponding to the same ACh concentration (range 10 nm to 10 μm, grey bars). Plotting in the same graph the mutation-induced shift and the corresponding slowing measured in the bradycardic family (red bar) shows that these values fit well the trend of the curve; these values are in fact intermediate between those caused by 10 and 100 nm ACh, a range of concentrations within which slowing is attributable to If inhibition (DiFrancesco et al. 1989). The plot of Fig. 4 supports the view that the mutation-associated shift is fully reponsible for the bradycardic phenotype.
A recent report (Xu et al. 2012) has confirmed the shifting effect of the mutation, in agreement with Milanesi et al. 2006 data, but has proposed that the shift is caused by a reduced cAMP binding to the CNBD and consequent decreased cAMP sensitivity, rather than by constitutive structural modifications. The results from the two studies cannot apparently be reconciled, although several significant differences exist in the experimental settings. A major difference concerns the types of channels investigated, since while Milanesi et al. used full-length human HCN4, Xu et al. used a chimaera obtained by replacing the C-terminus of mouse hcn2 with a portion of the human HCN4 C-terminus; these latter channels therefore did not faithfully reproduce HCN4 channels.
Differences in the results obtained in the two studies indeed support the notion of a different behaviour of the two types of channels. For example, activation curves of wild-type channels have substantially more negative V1/2 values in Xu et al.'s than in Milanesi et al.'s study (by some 10–15 mV in whole-cell and 20–35 mV in inside-out patches). Part of this difference may derive from the different protocols used to measure activation curves. While Milanesi et al. apply steps of variable duration, lasting longer at depolarized than hyperpolarized voltages, so as to reach steady state at all levels, Xu et al. apply, instead, steps of fixed 3 s duration, which may not be sufficient in the more depolarized range. This protocol may lead to underestimation of the degree of activation at depolarized voltages, with a resulting negative shift of the activation curve.
However, this is unlikely to explain other discrepancies. Considering for example the cAMP dependence of the activation curve, K1/2 and the maximal shift from Hill fitting were 0.2 μm and 17.5 mV, respectively, for the chimaeric channel of Xu et al. and 1.53 μm and 9.8 mV, respectively, for HCN4 of Milanesi et al. It is also worth noting that the cAMP dependence of S672R-mutated mouse hcn2-h4 channels, by predicting significant activation only at high concentrations (10–100 μm), would imply a nearly complete lack of sensitivity at physiological cAMP levels. The clinical consequence of this feature would be chronotropic incompetence, a feature not found in the asymptomatic patients of Milanesi et al.'s study.
Taken together, these data point to substantial functional differences between the two types of channels. Clearly more experimental data are required to investigate these differences.
G480R and A485V The heterozygous mutation G480R was found in a family with asymptomatic sinus bradycardia evident from a young age (Nof et al. 2007); affected members of the family did not have other rhythm or conduction abnormalities and had normal chronotropic competence. The mutation co-segregated with the bradycardic phenotype in most patients of the family.
Unlike the HCN4 mutations reported earlier, this was the first to occur in the extracellular pore (P)-loop of the channel (Fig. 1). More specifically, the mutation involves a glycine residue belonging to the GYG selectivity filter motif typical of K+-selective channels. This might suggest that the G480R mutation affects channel permeabilty and/or conductance. Indeed the mutation was found to cause a reduced current in heterologous expression experiments. However, the current carried by mutant G480R channels had the same reversal potential than wild-type channels, suggesting no significant change in the Na+/K+ permeability ratio; possible changes in the channel conductance were not investigated. Analysis by Western blot and electrophysiology indicated that the current decrease was attributable to a negative shift of the voltage dependence and slowing of current activation, and to a decreased synthesis and trafficking to the plasma membrane of mutated channels. The effect of the loss-of-function G480R mutation therefore confirms the idea that reduced HCN4 contribution to the activity of pacemaker cells slows the heart.
The same team (Laish-Farkash et al. 2010) later identified a second extracellular single-point mutation (A485V), located in the loop connecting P to S6 (Fig. 1), potentially associated with sinus bradycardia in unrelated individuals belonging to the same ethnic group of Moroccan Jews (Fig. 5). Unlike carriers of the G480R mutation, carriers of the A485V mutation were symptomatic with dizziness, presyncopal episodes and cardiac arrest, the latter occurring in one index patient.
It is somewhat unexpected that despite the different symptoms, the A485V mutation is reported to affect channel function in ways similar to those found with G480R, i.e. a decrease in the synthesis and membrane expression of channel protein, with no change in ion selectivity, and a negative shift of the voltage range of current activation (Laish-Farkash et al. 2010).
To explain the modifications induced by the mutation on cardiac rhythm, including the fact that mutation carriers had a normal chronotropic and exercise capacity despite functional changes of mutated channels, the authors discussed possible effects of heteromeric assembly of wild-type and A485V mutant HCN4 channels, or even of heteromeric assembly comprising HCN2 subunits. A more precise interpretation of the functional effects of the A485V mutation will require a more in-depth analysis of the detailed biophysical and cellular changes associated with the mutation itself.
K530N The screening of patients affected by atrial and/or ventricular tachyarrhythmias has recently led to the identification of a new mutation in the HCN4 C-linker (K530N) associated with familial tachycardia–bradycardia syndrome and atrial fibrillation (Duhme et al. 2012). Following identification of a proband with paroxysmal AF, several members of the same family with similar symptoms were found to carry the same heterozygous mutation, although one individual with the mutation had no symptoms.
Functional studies revealed that K530N is a loss-of-function mutation whose effect is to shift the current activation curve to more negative voltages (Fig. 6). An atypical property of this mutation is that a negative shift was observed in heterologous wild-type/mutant channels, which mimics the condition of individuals heterozygous for the mutation, but not in homomeric mutant channels, whose kinetic behaviour was indisinguishable from that of wild-type channels (Fig. 6B and C). The authors interpreted this surprising finding by assuming that homomeric wild-type or mutant channels have a more stable ‘tetrameric’ arrangement than heteromeric channels, where the ‘dimer of dimers’ configuration is favoured instead, leading to increased tonic inhibition of activity. However, no evidence was provided to support this hypothesis.
As discussed above for the mutation D553N, it is unclear how a simple modification like a more negative range of current activation can be responsible for a complex phenotype such as the one observed in this family, especially because the same type of action exerted by other mutations (for example S672R) is associated with milder arrhythmic phenotypes. According to the authors, slow heart rate increases susceptibility to ectopic activity, particularly during adrenergic stress, and may eventually trigger AF onset by inducing premature beats or pulmonary vein activity and re-entry, especially under conditions of conduction heterogeneity in diseased/ageing heart. However, while highlighting an interesting, potential functional role of the f-channels in AF, this proposed mechanism remains speculative.
Recent studies investigating the association of HCN4 mutations with inheritable forms of arrhythmias have provided growing evidence for an important role of dysfunctional f-channels in rhythm disorders. These studies, along with evidence from the investigation of knockout animals (Baruscotti et al. 2011) and from the clinical use of the If blocker ivabradine (DiFrancesco & Camm, 2004; DiFrancesco & Borer, 2007; Cappato et al. 2012), have shown that as well as being involved in the generation and control of basal physiological rhythm of the SAN and autonomic rhythm control, proper If function is relevant to the maintenance of normal rhythmic activity in a broader context. Dysfunctional f-channels and/or altered f-channel expression can thus be associated with asymptomatic and symptomatic sinus bradycardia, tachycardia–bradycardia syndrome, atrioventricular node block, atrial fibrillation and contribute to more complex types of arrhythmias. For example, If has a pathophysiological role in hypertrophied ventricles of the failing human heart, where increased transcription of HCN channels predisposes to enhanced ventricular automaticity (Cerbai & Mugelli, 2006). Also, dysfunctional If may also have a role in atrial fibrillation originating from pulmonary veins, since f-channel-expressing pacemaker cells are present in pulmonary veins, where they can contribute to ectopic beat generation and atrial arrhythmogenesis (Chen et al. 2009; Suenari et al. 2012). Interestingly, a recent genome-wide association study of atrial fibrillation conducted on European individuals has identified, among others, a susceptibility locus in the first intron of HCN4 (Ellinor et al. 2012).
It is worth noticing that all HCN4 mutations described so far are loss-of-function mutations. It is likely that more mutations, perhaps including gain-of-function mutations, and/or mutations in HCN channel subunits and ancillary proteins, will be found to contribute to other types of arrhythmias in the near future.
This work was supported by the Italian Ministry of Education, University and Research grant PRIN 2010BWY8E9.