Conduction abnormalities and ventricular arrhythmogenesis: The roles of sodium channels and gap junctions

Ventricular arrhythmias arise from disruptions in the normal orderly sequence of electrical activation and recovery of the heart. They can be categorized into disorders affecting predominantly cellular depolarization or repolarization, or those involving action potential (AP) conduction. This article briefly discusses the factors causing conduction abnormalities in the form of unidirectional conduction block and reduced conduction velocity (CV). It then examines the roles that sodium channels and gap junctions play in AP conduction. Finally, it synthesizes experimental results to illustrate molecular mechanisms of how abnormalities in these proteins contribute to such conduction abnormalities and hence ventricular arrhythmogenesis, in acquired pathologies such as acute ischaemia and heart failure, as well as inherited arrhythmic syndromes.


Introduction
Cardiac arrhythmias arise from disruptions in the normal orderly sequence of electrical activation and recovery of the heart. On the one hand, ventricular arrhythmias can lead to sudden cardiac death (SCD), which accounts for around 400,000 deaths in the United States [1][2][3]. This represents a prevalence of 0.1% in the population. On the other hand, atrial arrhythmias are the most common arrhythmias observed in clinical practice [4], affecting around 2 million people in the U.S. [5]. It is a major contributor to cardiovascular morbidity. For example, up to 15% of all strokes in the U.S. can be attributed to atrial fibrillation [6].
Ventricular arrhythmias can be categorized into disorders affecting cellular depolarization or repolarization [7], but can also be caused by abnormalities in action potential conduction [8]. The commonest underlying mechanism is the formation of re-entry circuits [9]. Re-entry occurs when an action potential fails to extinguish itself and re-activates a region that has recovered from refractoriness [10,11]. It may occur in the presence of an obstacle, around which an action potential can travel (circus-type) [12], or without an obstacle (reflection or phase 2) [13,14]. For circus re-entry to occur, three criteria must be met. First, there must be an obstacle around which an action potential (AP) can circulate. This need not to be a structural abnormality, but can be a functional core of refractory tissue. Secondly, conduction velocity (CV) must be sufficiently reduced so that the myocardial tissue ahead of the excitation wavefront remains excitable. Finally, unidirectional conduction block must be present to prevent waves from selfextinguishing. These three conditions form part of the arrhythmogenic substrate, which upon a trigger, for example from premature generation of an AP, can serve to sustain re-entry. Recent review articles have focused on the mechanisms of cardiac conduction [15] and the factors governing myocardial CV [8,16]. This article briefly discusses the factors causing conduction abnormalities, as well as the roles that sodium channels and gap junctions play in AP conduction and how abnormalities in these proteins contribute to ventricular arrhythmogenesis.

Reduced conduction velocity and unidirectional conduction block
CV of the propagating APs depends on both active and passive properties of the cell membrane (Fig. 1). Active properties refers to voltage-dependent conductances, mainly mediated by I Na , which determines the AP upstroke (phase 0) [18]. A faster CV can arise from two situations. Firstly, a higher maximum upstroke velocity, dV/dt max brings the membrane to threshold much more quickly. Secondly, increased myocardial excitability given by 1/(threshold potential-resting membrane potential) means that smaller inward currents are required to reach threshold. The resting membrane potential also influences I Na through regulation of the inactivation kinetics of sodium channels [19][20][21][22][23]. In contrast, passive properties depend on capacitive and resistive components of the cell membrane as well as the myocardial architecture [24]. Axial resistance (r i ) depends on the resistance of both the myoplasm [25] and the gap junctions between myocytes [26]. Decreased r i would increase CV. Membrane capacitive load is

Contents lists available at ScienceDirect
IJC Heart & Vasculature j o u r n a l h o m e p a g e : h t t p : / / w w w . j o u r n a l s . e l s e v i e r . c o m / i j c -h e a r t -a n d -v a s c u l a t u r e determined by the surface area and volume ratio [24]. Increased membrane capacitance (C m ) prolongs the time needed to bring membrane to threshold, thereby reducing CV. These processes determining CV can be described mathematically by the cable theory [27]. However, recent experimental data have challenged the applicability of this theory to cardiac conduction [28], and have attracted ongoing debate [29,30]. Traditional cable theory does not explicitly consider junctional clefts in the extracellular space and is therefore incompatible with the notion of ephaptic coupling: interested readers should be directed to this article here [31]. As shown in Fig. 1, acquired and congenital heart diseases influence various determinants of CV, which will be discussed in turn. Additional factors can modulate the CV of the AP wave. Firstly, gap junctions, high-resistance ion channels located between successive myocytes [32,33] cause conduction slowing and discontinuous propagation [34,35]. Secondly, the complex tissue architecture of the myocardium leads to anisotropic conduction, with greater velocity in the longitudinal compared to the transverse direction [36,37]. This is in part due to higher density of gap junctions at the ends of myocytes than in the lateral margins [38]. Finally, the source and sink are determined by the plasma membrane excitability and tissue-structure properties, respectively [8]. Interactions between these factors are recognized to cause reduced CV, conduction block and fractionation of the excitation wave [39]. Finally, it is also important to appreciate that conduction and repolarization are not independent processes, but one can affect the other. For example, I to inhibition enhances conduction between cell pairs, suggesting that repolarization currents are able to influence CV [40,41]. The accepted view that voltage-gated sodium channels, gap junctions and desmosomes are separate entities with distinct functions has been challenged. Sodium channels were assumed to be separate from gap junctions but were later found in intercalated disks [42], co-localizing with gap junctions and playing an important role in cardiac conduction [43]. This co-localization may constitute a cardiac ephapse, thereby contributing to ephaptic coupling [44]. Its discussion is beyond the scope of the current paper and interested readers are referred to review articles here [45,46].
Unidirectional conduction block can occur in homogeneous cardiac tissues in the presence of functional asymmetries [8]. Firstly, it can arise from interaction between the activation wavefront and the refractory tail [47]. This can occur with premature initiation of an AP, e.g. from early or delayed afterdepolarization phenomena, or co-existence of interacting wavefronts during multiple wavelet re-entry [8]. Secondly, heterogeneities in membrane excitability increase the vulnerability to unidirectional block. This could occur in regional ischaemia in which there is local accumulation of potassium ions [48], causing both a positive shift in the resting membrane potential and inhomogeneities in sodium channel recovery [49]. Thirdly, increased spatial heterogeneities in refractoriness can arise from alterations in depolarizing or repolarizing currents; for example, this can occur with regional ischaemia where extracellular accumulation of potassium produces shortening of the AP [49].

Sodium channels
Voltage-gated sodium channels are responsible for the initiation and propagation of APs in excitable cells. Each channel consists of a pore-forming α-subunit, a modulatory β-subunit and other regulatory proteins. Of these, the Na V 1.5 α-subunit is encoded by the SCN5A gene [50] and is made of four domains (I to IV) with each domain containing six transmembrane segments (S1 to S6). Upon depolarization, the S4 segments, which are positively charged, undergo outward movement [51,52]. This opens the channel pore and allows the influx of sodium ions. The resulting transmembrane current, I Na , therefore determines myocardial excitability and CV of the APs. Depolarization also initiates fast inactivation [53,54]. The mechanism involves the linker region between domains III and IV, which functions as a 'lid' to occlude the pore [55][56][57][58]. Slow inactivation involves the P-segment linker sequence between S5 and S6 bending back into the membrane and lining the outer pore [59,60]. Binding sites for Ca 2+ and the Ca 2+ -binding protein calmodulin are present at the carboxyl terminus [65,66], allowing modulation of channel function [67,68]. Ca 2+ /calmodulin-dependent Kinase II phosphorylates the sodium channel, causing a negative shift in the voltage-dependence of steady-state inactivation without altering the voltage-dependence of steady-state activation or the peak I Na [69].
I Na consists mainly of a tetrodotoxin-insensitive component, attributable to the cardiac isoform Na V 1.5 [50,70,71]. It also contains a persistent, tetrodotoxin-sensitive component, as suggested initially by voltage clamp technique [61,62] and later confirmed by patch clamping [63,64]. This is mediated by neuronal isoforms Na V 1.1, Na V 1.3, and Na V 1.6 [72][73][74]. Recent immunohistochemical experiments have demonstrated neuronal isoforms in cardiac tissue of many species, in keeping with the electrophysiological findings [75]. They were found in pacemaker cells with whole cell current and voltage clamp experiments demonstrating a contributory role in pacemaker activity [76]. Furthermore, they were localized to the transverse tubular system of cardiomyocytes, and tetrodotoxin application resulted in desynchronization of excitation-contraction coupling and a decrease in cardiac contractility [77]. However, this is a contentious area as another study showed no significant effects of tetrodotoxin on shortening of rat ventricular myocytes [78].

Abnormalities in sodium channel structure and function
Two main types of autosomal dominant inherited arrhythmias resulting from mutations in SCN5A have been described [79]. Firstly, gain-of-function mutations are found in Long QT Syndrome (LQTS) type 3 [80]. These produce disruptions in fast inactivation and allow repeated channel opening during sustained depolarization [81][82][83]. Normally, the repolarization time course is determined by the balance between the inward currents mediated by the voltage-gated L-type calcium channel (I Ca,L ) and sodium-calcium exchanger (I NCX ), and the outward currents mediated by the voltage-gated delayed rectifier potassium channels (I K ) [84]. The rapid and slow currents (I Kr and I Ks , respectively) make up I K . Persistent sodium current activation leads to delayed cellular repolarization and the development of early after-depolarization phenomena by the two mechanisms. Firstly, depolarizing shifts in the membrane potential can reactivate the L-type calcium channels [85], resulting in increased I Ca,L that further depolarizes the membrane. This sets up a positive feedback loop, triggering an AP. Secondly, at membrane potentials negative to the threshold of I Ca,L activation (but before full repolarization), spontaneous calcium release from the sarcoplasmic reticulum can activate I NCX [86], resulting in membrane depolarization. Aberrant initiation of APs themselves could produce all three conditions for re-entry: unidirectional block, slowed conduction and a refractory core around which an AP can circulate [87]. Interestingly, in sudden unexplained death in epilepsy (SUDEP), a possible explanation is ventricular arrhythmias [88,89]. The proposed mechanism is AP prolongation due to increased contributions of neuronal sodium channel isoforms to the late component of I Na (I Na, L ), similar to that of LQTS type 3 [88,89].
Secondly, loss-of-function mutations in the SCN5A gene are observed in Brugada syndrome [90]. These have opposing effects on the fast and slow inactivation of sodium channels [91]. Fast inactivation is disrupted, culminating in a sustained sodium current, which prolongs repolarization at slow heart rates. However, the intermediate kinetic component of slow inactivation is augmented, delaying sodium channel recovery and reducing the sodium current at fast heart rates. Normally, I Na is balanced by the transient outward potassium current, I to in the early plateau phase of the AP. Because higher I to density is observed in the epicardium compared to the endocardium, it has been postulated that repolarization occurs more readily in the epicardium, resulting in a loss of dome morphology [92]. Propagation of the AP dome from regions where it is maintained to regions where it is abolished can produce an extrasystole that initiates ventricular tachycardia (VT) [93,94]. Although traditionally linked to abnormal repolarization, slowed AP and discontinuous conduction have recently been recognized as contributors towards arrhythmogenesis in Brugada syndrome [93,95].
In contrast, autosomal recessive mutations of SCN5A have been associated with the development of Sick Sinus Syndrome (SSS) [96]. Progressive cardiac conduction defect (PCCD), also called Lenègre disease, has been linked to mutations of SCN5A [97][98][99]. This is characterized by progressive alterations in conduction with bundle branch blocks potentially leading to complete atrio-ventricular block. Scn5a +/− mice modelling Lenègre disease showed progressive conduction defects associated with myocardial fibrosis [100]. Finally, overlap disorders have been described in several SCN5A mutations [101]. Most recently, a case involving p.Y1449C mutation was reported with phenotypes of conduction disease, Brugada syndrome and atrial flutter [102]. Other phenotypes have also been described, which can be explained by the biophysical effects produced by the particular mutation in the SCN5A gene [103].
In addition to the above congenital abnormalities in SCN5A, acquired alterations in channel properties have been observed. Experimental models of heart failure have demonstrated positive shifts in the resting membrane potential, which causes partial inactivation of sodium channels [104]. Furthermore, deglycosylation of the sodium channels has been detected, in turn causing a positive shift in the voltage-dependence of steady-state activation and a negative shift in the voltage dependence of steady-state inactivation [105]. The net effect is a reduced transient component of the sodium current (I Na, T ), peak I Na leading to a decrease in dV/dt max . Increased late component of I Na (I Na, L ) has been observed in heart failure and also in post-infarction modelling, leading to action potential prolongation [106][107][108][109]. In the failing myocardium, calcium handling is abnormal [110], with reduced amplitudes of Ca 2+ transient, sarcoplasmic reticular Ca 2 + content and Ca 2 + removal [111]. Calmodulin Kinase II is upregulated and its activity increases during heart failure [112], which would be expected to increase I Na, L [110]. This persistent activation of I Na is likely to be a contributing factor for sudden death in patients with failing ventricles.
A particular type of potassium channel, K ATP , is regulated by ATP levels inside the cells [113]. Under normoxic conditions, they are strongly inhibited by intracellular ATP [114]. In acute ischaemia, the oxygen supply cannot match the metabolic demand of the myocardial tissue, resulting in depletion of ATP and switch to anaerobic pathway for substrate utilization. K ATP channels are activated upon ATP depletion and ADP accumulation [114], producing action potential shortening [115]. However, other experiments have demonstrated no action potential shortening during ischaemia, suggesting that K ATP channels might not have opened at least during early ischaemia [116]. Furthermore, a number of extracellular and intracellular changes are observed during myocardial ischaemia. Extracellularly, [K + ] o is increased and pH is reduced. A rise in [K + ] o would initially cause membrane hyperpolarization, but as steady state is reached, there is a positive shift in the resting membrane potential as described by the Goldman field relationship [117]. It also increases the threshold potential to a smaller extent, thereby increasing myocardial excitability [117,118]. This may explain why CV increases with mild increases in [K + ] o [119]. At higher concentrations, [K + ] o increases the proportion of inactivated sodium channels, which would reduce dV/dt max and therefore the CV of the propagating AP [118,120]. This is represented in Fig. 1 by an arrow originating from RMP and pointing to I Na . Thus, the effects of high [K + ] o on CV are biphasic, first increasing then decreasing the speed of conduction [119]. Experimental observations using microelectrode recordings indeed demonstrate reduced dV/dt max during combined hyperkalaemia, acidosis and hypoxia in guinea pig ventricular muscle [49]. Extracellular acidification also reduces CV [121]. Intracellularly, increases in cyclic AMP and Ca 2+ have been described. Beta adrenergic stimulation increases cAMP and activation of protein kinase A [122,123]. PKA in turn phosphorylates serine residues at 525 and 528 positions within the linker region between domains I and II [124], causing trafficking of sodium channels to the plasma membrane [125] and increase in I Na . However, there may also be a PKA-independent mechanism that reduces I Na , likely through a G-protein coupled signaling pathway. In contrast, calcium binds to the conserved C2 domain of protein kinase C [126], thereby activating it. PKC in turn phosphorylates the serine residue at 1505 position, located at the inactivation gate between domains III and IV, thereby reducing I Na [127]. Together, all of the above changes produce decreases in both CV and action potential duration (APD), promoting arrhythmogenesis through a reduction in excitation wavelength given by the product of CV and effective refractory period (ERP, approximated by APD).

Gap junctions
Gap junctions were first described as hexagonal arrays of protein molecules in the plasma membrane of Mauthner cell synapses of goldfish [128]. They form high-resistance pathways for intercellular coupling, allowing passive, electrotonic spread of ions and also passage of larger molecules such as amino acids and nucleotides [129]. The resistance of gap junctions therefore contributes to r i and determines CV [130,131]. Each gap junction is made of two connexons, which are hexameric proteins of the connexin (Cx) subunit. In cardiac tissue, Cx 30.2, 40, 43 and 45 have been described [132]. Of these, Cx43 is expressed throughout the atria and ventricles [133], whereas Cx40 is expressed only in the atria and His-Purkinje system [134,135]. Cx43 and Cx40 have unitary conductances of about 100 pS [136] and 160 pS [137], respectively. There are two major gating mechanisms, the conventional membrane voltage-dependent gating and also transjunctional voltage-dependent gating [138]. Its function is also regulated by phosphorylation [139][140][141], Ca 2+ i [142][143][144][145], pH i [146,147] and the surrounding lipid environment [148][149][150][151]. Gap junctions are also affected by exogenous substances such as lipophilic agents that uncouple them [152][153][154] or peptides enhancing their ionic conductances [155,156].
Gap junctions represent the electrical coupling pathway between cardiomyocytes, whereas anchoring complexes termed fascia adherens junctions and desmosomes mediate mechanical coupling between them. Together these structures form the highly organized intercalated disk [158]. There is increasing evidence that sodium channels co-localize with gap junctions, and form the connexome, which includes a "protein-interacting network" of desmosomes, gap junctions and sodium channels [159]. What is becoming clear is that altered expression of one component influences the expression or localization of others. Thus, experiments using super-resolution fluorescence localization microscopy demonstrated that truncation of the C-terminus of Cx43 in adult murine cardiomyocytes resulted in reduced I Na at the end of the cells [160]. Conversely, in heterozygous Scn5a-knockout mice with 50% reduction in sodium channel expression, disarrangement of gap junctions was observed [161].

Abnormalities in gap junction structure and function
Inherited mutations in GJA1, the gene that encodes for Cx43, have been associated with an autosomal dominant condition called oculodentodigital dysplasia [162], which predisposes to VT and sudden cardiac death. The effects of Cx43 loss on conduction and arrhythmogenesis have been extensively studied in mouse models [163][164][165][166][167][168][169][170][171][172][173], but not without considerable disagreement, as summarized previously [169]. Cardiac-restricted inactivation of Cx43 followed by crossing with Cre recombinase produced mice with mosacism, in which Cx43 was reduced by 86 to 95% compared to wild-type [163]. Their hearts were structurally normal by echocardiography but showed increased incidence of sudden death. In other studies, heterozygous Cx43 +/− mice showed 45 to 50% decrease in Cx43 expression. In these systems, CV was either unaltered [165][166][167]169,171,172] or decreased by 23 to 44% [164,168,170]. Together, these studies suggest different parameters, such as intracellular calcium levels, interstitial volume [28], perinexal width and perfusate composition and osmolarity [169], influence the net behaviour, conduction, even in mice with identical genotype. In addition to conduction slowing, increased conduction dispersion is also a common abnormality following Cx43 downregulation [171,174,175]. This can refer to phase difference in conduction latencies of neighbouring regions [171], difference in CV across the myocardial wall [174] and dispersion represented by coefficient of dispersion, which uses standard deviation of the mean CV [175]. Finally, a mutation in plaktoglobin, a protein that links adjacent myocytes and anchors sarcomeres, has been observed in the recessive form of arrhythmogenic right ventricular dysplasia (ARVD), called Naxos disease, and has been associated with reduced gap junction expression in the myocardium [176].
Acquired abnormalities in gap junctions occur in several pathological conditions, such as ischaemia and infarction [177]. In the failing ventricle, tyrosine phosphorylation of Cx43 by c-Src tyrosine kinase is increased [178], reducing gap junction conductance. Additionally, heterogeneous distribution of Cx43 leads to increased dispersion of conduction and therefore re-entrant arrhythmogenesis [175]. The importance of Cx43 is supported by findings of its deletion following genetic modification, in which CV slowing and increased anisotropic conduction were observed [171]. In acute ischaemia, the following abnormalities have been detected. Increased Ca 2+ i has been associated with reduced conductance as well as uncoupling of gap junctions. PKCmediated phosphorylation, a calcium-dependent process, of the serine residue at the 368 position, has been linked to reduced conductance [136,141]. Dephosphorylation of gap junctions was also described [179], resulting in electrical uncoupling [180] and gap junction lateralization [181,182]. In summary, the consequence of these changes in gap junction structure or function, whether in congenital or acquired disorders, would be to reduce CV and increase heterogeneity of conduction, hence predisposing to re-entrant excitation.

Therapeutic strategies
Understanding the mechanisms of conduction abnormalities allows effective pharmacological therapy to be developed for the treatment of ventricular arrhythmias. Several studies have focused on the arrhythmic effects of gap junction downregulation [169] or uncoupling [183]. However, protective effects of reducing infarct size as well as preventing arrhythmogenesis produced by ischaemia have also been shown for gap junction uncouplers [184,185]. As demonstrated theoretically, in non-uniform tissue, mild loss of gap junction function paradoxically increases CV and improves the safety margin of conduction [16]. This can remove unidirectional blocks and therefore protect against arrhythmogenesis [32,186,187]. Indeed, experiments using cell strands connected to rectangular cell monolayers showed that partial gap junction uncoupling converted unidirectional conduction block to bidirectional conduction [188]. The therapeutic effects of gap junction uncouplers therefore warrant further investigation.
For heart failure, fibrosis is a significant pro-arrhythmic factor. In hypertensive rats, fibrosis can be prevented by the ACE inhibitor enalapril [189], the angiotensin II receptor blocker losartan [190] as well as the calcium channel blocker amlodipine [191]. In mouse hearts, the aldosterone receptor antagonist eplerenone and losartan both reduce fibrosis and arrhythmic tendency [192]. Fibrosis resulting from fibroblast activation is mediated by many protein growth factors, such as transforming growth factor beta [193]. This can be inhibited by the antifibrotic hormone relaxin [194]. To prevent enzymatic degradation, adenoviruses expressing relaxin can be injected into the body, and the relaxin can be detected in plasma [195]. This treatment has demonstrated efficacy in mice with fibrotic cardiomyopathy [195]. It may also be effective in individuals with Lenègre disease, Brugada syndrome and ARVD, in which myocardial fibrosis is present [97][98][99]176,196].

Conclusion
This article reviewed the physiological mechanisms by which alterations in structure and function of gap junctions and sodium channels lead to conduction abnormalities and in turn, to ventricular arrhythmogenesis. It is clear that AP conduction is not a simple phenomenon solely determined by biophysical parameters of ion channel conductances, and resistive and capacitive properties of myocytes. It is also influenced by myocardial cell size, ion channel localization, cellular architecture and source-sink relationships. For example, in heart failure, the extensive electrophysiological and structural remodelling are key factors in arrhythmogenesis. For instance, up-and downregulation of ion channels alter transmembrane currents, and fibrosis affects axial resistance and membrane capacitance. Conduction is a fascinating aspect of cardiac electrophysiology and there is much to be learnt. A systems approach is needed for the development of effective therapeutic agents that can prevent or abolish arrhythmias. Without a full appreciation of the pharmacological effects on the above factors, these agents may be at best, ineffective but at worst, will themselves induce malignant ventricular arrhythmias.