Cardiac arrhythmias are among the leading causes of mortality. They often arise from alterations in the electrophysiological properties of cardiac cells and their underlying ionic mechanisms. It is therefore critical to further unravel the pathophysiology of the ionic basis of human cardiac electrophysiology in health and disease. In the first part of this review, current knowledge on the differences in ion channel expression and properties of the ionic processes that determine the morphology and properties of cardiac action potentials and calcium dynamics from cardiomyocytes in different regions of the heart are described. Then the cellular mechanisms promoting arrhythmias in congenital or acquired conditions of ion channel function (electrical remodeling) are discussed. The focus is on human-relevant findings obtained with clinical, experimental, and computational studies, given that interspecies differences make the extrapolation from animal experiments to human clinical settings difficult. Deepening the understanding of the diverse pathophysiology of human cellular electrophysiology will help in developing novel and effective antiarrhythmic strategies for specific subpopulations and disease conditions.
Cardiac arrhythmias are major causes of mortality. They most often arise from pathological changes in the electrophysiological properties of myocardial cells. First, this review summarizes the physiology of cardiac action potentials, their regional and species differences, the underlying transmembrane ionic currents, and transporters, with special focus on their human relevance.
Progress in computer modeling and vast quantities of experimental data made computerized replication of the action potential, impulse conduction, and simulation of cardiac electrophysiology possible. Current computer models offer improved observability and controllability via utilization of different modeling scales and can assist future individualized anti-arrhythmic therapy as well as drug electrophysiological safety assessment.
A number of diseases evoke changes in the configuration of the action potential caused by altered function and/or densities of transmembrane ion channels and transporters, collectively termed electrical remodeling. Initially, these alterations are often compensatory; however, remodeling significantly contributes to increased arrhythmia susceptibility by impairing impulse generation, conduction, and myocardial refractoriness in these clinical settings. Electrical remodeling in atrial fibrillation, heart failure, hypertrophic cardiomyopathy, myocardial infarction, and advanced age are discussed. Better understanding of the cellular basis of cardiac electrophysiology, electrical remodeling, and mechanisms of arrhythmias has important implications for future clinical therapeutic strategies.
The heart is a mechanical pump with the vital role of supplying blood to other organs. In humans, it contracts and relaxes in a regular fashion ∼60 times per minute. If the regular heartbeat is interrupted for more than a couple of minutes, the lack of oxygen supply causes irreversible damage to vital organs, including the heart itself, potentially causing sudden cardiac death (SCD). Cardiac contractions, the most important function of the heart, are initiated by a bioelectrical signal, the action potential (AP) (1), via a process called excitation-contraction coupling (2). The action potential originates in the sinus node cells, propagates through the whole heart via an active electrophysiological process called impulse conduction, and can be measured as the electrical potential difference between the intra- and extracellular space. The stimulus spreads through both atria, causing their contraction. The next stage of propagation is the atrioventricular (AV) node, which passes the signal to the ventricles with a slight delay to provide enough time for the atria to contract. From the AV node, the bundle of His conducts the stimulus along the septal wall to the subendocardial Purkinje fibers, which then stimulate the ventricles, allowing their synchronized contraction. The action potential is determined by the opening and closing of various complex transmembrane proteins, which consist of ion channels and transporters, i.e., pumps or exchangers. Disturbances of action potential generation and/or conduction can lead to changes in the regular heart rhythm called arrhythmias. These disturbances can impair contraction to such a degree that thromboembolic stroke of atrial origin or sudden cardiac death may eventually occur. Therefore, understanding the function and regulation of transmembrane ion channels and transporters, as well as their impact on the cardiac action potential, is essential to understand arrhythmia mechanisms and treat life-threatening cardiac arrhythmias. Arrhythmias are usually diagnosed based on the analysis of electrocardiogram (ECG) recordings, which represent the electrical activity of the heart as measured on the body surface. The ECG is determined by many variables, including the function of transmembrane ion channels and transporters in the different heart cells and the consequent changes in the membrane potential The P wave corresponds to the activation (depolarization) and early repolarization of the atrial cells. The QRS complex reflects the time course of the depolarization of the ventricles caused mainly by the activation of the fast sodium channels. The PQ segment mainly indicates the impulse conduction from the atria to the ventricles. The PQ segment also contains the HQ interval, which reflects fast propagation due to the function of the fast sodium current (INa). In addition, cell-to-cell coupling is low in the AV node (8), which makes impulse propagation through the AV node relatively unsafe. The isoelectric ST segment reflects the plateau phase of the ventricular action potentials. In this phase, membrane potential hardly changes at the cellular level because of the fine balance of opening/closing of different ion channels. The configuration of the T wave shows the repolarization time course of the ventricles, and it reflects the balance between the slowly activating repolarizing potassium and chloride currents and the depolarizing steady-state, so-called “window” sodium (9) and “window” calcium. and the slowly decaying, often called “late,” sodium (INaLate) and slowly inactivating calcium currents. Analysis of the PP intervals yields important information regarding heart rate and its regularity.
Cardiac arrhythmia mechanisms are still the subject of intensive research. Because of the large variability in appearances, types (e.g., bradycardia and different types of tachycardia), locations (supraventricular or ventricular), and underlying diseases, it is widely accepted that there is not a single mechanism to explain how arrhythmias originate. Therefore, patients are often treated with little knowledge regarding the mechanisms and/or causes of the arrhythmia.
The majority of cardiac arrhythmias are the result of an enhanced proarrhythmic substrate combined with a trigger. Enhanced heterogeneity of repolarization and impaired impulse conduction represent typical arrhythmia substrates (conditions that are prerequisites for arrhythmia development) for severe tachyarrhythmia. Impairment of impulse conduction can be caused by anatomical (FIGURE 2, A and B) or functional (FIGURE 2C) alterations. The process was described long ago, first in the early twentieth century. Impulse conduction critically depends on the density and kinetics of inward transmembrane ionic currents. Depolarization of the resting membrane potential (RMP), for example, reduces sodium and calcium inward currents and strongly influences their kinetic properties. This can thus slow impulse conduction and cause unidirectional or bidirectional conduction block, and potentially reentry, underpinning a wide range of cardiac arrhythmias.
As mentioned above, reentrant arrhythmias can be caused by functional causes too, without an anatomically well-defined myocardial damage (FIGURE 2C). This form of reentry is more complex and could involve both impulse conduction and repolarization heterogeneities (arrhythmia substrate) as well as enhanced normal or abnormal automaticity (trigger).
Reentrant arrhythmias are often initiated by an extrasystole formed anywhere in the heart, acting as an arrhythmia trigger. Both arrhythmia substrate and trigger, like an extrasystole [premature ventricular complex (PVC)], can be promoted by pathological cardiac conditions (FIGURE 2), e.g., myocardial ischemia, heart failure, and genetic diseases, or by adverse drug reactions. General arrhythmia mechanisms include various cellular aspects, e.g., transmembrane ionic currents, transporters, action potential properties, and automaticity, which are the subjects of this review. However, arrhythmia mechanisms at the whole heart level are more complex, since they are also determined by anatomical and structural properties, impulse conduction, and intercellular communication between myocardial and nonmyocardial cells, like fibroblasts. These factors are beyond the scope of the present work, and the interested reader is referred to other reviews.
The cardiac action potential is a transmembrane potential change, with an amplitude ranging between 60 and 120 mV. It starts from a negative value, i.e., the resting membrane potential (RMP) in working myocardial cells or maximal diastolic potential in spontaneously beating cells (1), ranging from −95 to −40 mV. As in other excitable cells, the RMP is mainly defined by the conductance of inwardly rectifying K+ currents and can be roughly estimated by the Nernst equation from the uneven distribution of mainly K+ ions across the cell membrane. The electrogenic ATP-dependent Na+-K+ pump also contributes to the RMP, by exporting 3 Na+ and importing 2 K+ (35–38). In healthy conditions, the duration of the action potential (APD) determines the effective refractory period (ERP), defined as the shortest time interval needed before a new stimulus, or an early extrasystole, can elicit another action potential. The relationship between APD and ERP can be disrupted in pathological conditions, for example, in hyperkalemia, resulting in postrepolarization refractoriness (39).
In the context of the cardiac action potential, two aspects should be emphasized. First, there is no such uniform entity as “the cardiac action potential,” since its shape, i.e., the time course of the transmembrane potential changes, differs in the various regions of the heart (FIGURE 1), and therefore different action potentials should be considered and discussed separately. Second, there are significant interspecies differences (40), even when action potentials are recorded from similar regions of the heart. This is an important, and often overlooked, issue, since many experimental results have been obtained in small rodents, recently particularly in transgenic mice.
In general, the cardiac action potential is divided into five distinct phases (FIGURE 1). Phase 0 is the fast depolarization due to an abrupt increase in sodium influx, and it is characterized by the upstroke velocity and can result in an overshoot, i.e., the rapid change of potential from the negative RMP to positive voltage values, reaching a peak of up to +30 to +40 mV. The overshoot is followed by a return to negative values, in a process called repolarization, which includes phases 1, 2, and 3. Phase 1 is characterized by a transient and relatively fast repolarization brought about by a decrease in sodium influx and a transient increase in potassium efflux and chloride influx. Phase 2 consists of a long-lasting plateau, still at depolarized voltage, during which the membrane potential remains almost constant or decreases slowly, caused by a small net transmembrane current carried by simultaneous calcium (and some sodium) influx and potassium efflux. Phase 3 represents the large repolarization toward the diastolic potential, mostly due to increased potassium efflux and decreased calcium and sodium influx. Phase 4 represents the resting membrane potential in diastole in working myocytes and the spontaneous depolarization in pacemaker cells. In cardiac myocytes that do not beat spontaneously the voltage remains stable at the RMP, whereas in cells exhibiting automaticity the potential gradually changes toward the positive values, in a process called spontaneous diastolic depolarization. When the threshold potential is reached, a new spontaneous action potential is generated, with a certain cycle length.
There are four common methods to record cardiac action potentials:
Weidmann and Coraboeuf were the first to record cardiac action potentials in dog ventricular muscle (41) and later in dog Purkinje fibers, using the sharp glass capillary-based microelectrode technique (42). The classical cardiac cellular electrophysiological knowledge gained by using this technique was elegantly summarized in an early monograph entitled Electrophysiology of the Heart by Hoffman and Cranefield (1), published in 1960, which is still useful today. This technique is still considered one of the best for accurate cardiac action potential recordings and can be used for both single-cell and tissue recordings. Its major advantages include 1) the ability to accurately record very fast voltage changes and 2) because of the very fine tip of the pipette, very little diffusion takes place out of the pipette solution, having negligible effects on the intracellular milieu. However, since this technique has some limitations (e.g., difficulties in maintaining a stable impalement for extended periods of time), other methods have also been developed and used widely.
In intact hearts, in in vivo animal experiments, or in clinical studies, where the microelectrode technique is difficult to apply, monophasic action potential recording can also be used, with either a suction electrode (43, 44) or a Franz catheter (45). With this technique, recordings can be easily performed from multiple sites simultaneously, and impalements/attachments are not lost because of vigorous contractions, even in in vivo or ex vivo conditions. However, rapid voltage changes or action potential amplitudes and shapes cannot be determined accurately.
Since the introduction of the patch clamp by Neher and Sakmann (46–48), the whole cell configuration of this technique has been widely used. In the current-clamp mode, it can record action potentials from isolated myocytes. Despite its widespread use, this technique has important limitations that should be emphasized. First, measurements are performed in single isolated myocytes or, occasionally, cell pairs, and it is uncertain how, and to what degree, different ion channels are influenced by individual enzymatic digestion during the isolation procedure (49). Therefore, even if the recordings show single-cell action potentials with a normal shape, the function of the finely regulated ion channels can be drastically altered from their original condition. Also, the cell is dialyzed with the pipette contents, and its intracellular composition will change. When carefully and deliberately applied, however, this point can also be considered an advantage, as it allows control of the intracellular milieu. It should also be emphasized that single isolated myocytes are devoid of electrotonic interactions from neighboring cells. Therefore, the stochastic opening/closing behavior of ion channels has a more profound effect on membrane potential than in well-coupled tissue preparations (50). In addition, in multicellular preparations part of the ionic currents are utilized to depolarize neighboring cells during impulse propagation, and this can considerably reduce the action potential peak compared with single cells (51). Hence, action potential measurements in single isolated myocytes obtained with the patch-clamp technique should be interpreted with caution and not directly extrapolated to intact tissue.
The latest approach to recording cardiac action potentials is the optical mapping technique, which uses voltage-sensitive dyes and allows simultaneous recordings from multiple sites (52, 53). This technique is also excellent for dynamic studies and for investigations of arrhythmia mechanisms (54, 55). Disadvantages of this method include the difficulty of calibration to millivolts, phototoxicity, photodegradation, and photon scattering effects (56). Also, the application of excitation-contraction uncoupling compounds, e.g., blebbistatin, is necessary to avoid motion artifacts (57). These compounds may interfere with the experiments, since, e.g., blebbistatin was reported to elicit anomalous electrical activities (58) and prolongation of action potential duration (59), and inhibition of contraction will also decrease metabolic rate at the concentrations needed for motion artifact reduction.
Figure 4 : Action potential and underlying ionic currents recorded from human ventricular myocytes with the patch-clamp technique applying human ventricular action potential as command pulses at 1 Hz stimulation frequency, in the absence of any sympathetic effects. Inward rectifier potassium current (IK1), rapid (IKr) and slow (IKs) components of delayed rectifier potassium current, transient outward current (Ito), and L-type calcium current (ICa,L) were measured as difference current following application of selective channel inhibitors. INaL, late sodium current. Unpublished data from our laboratory at the Department of Pharmacology and Pharmacotherapy,
Figure 6 : Tissue-specific (human) cardiac atrial, Purkinje fiber, and ventricular action potentials and the underlying ionic currents in different action potential phases, indicating their pharmacology and modulation. Black arrows indicate inward and yellow arrows indicate outward current. The contributions of different currents to the action potentials are indicated below, with a time course adjusted to the action potential. CaM, calmodulin; CaMKII, Ca2+-calmodulin kinase II; hERG, human ether-à-go-go-related gene; IK1, inward rectifier potassium current; IK,Ach, acetylcholine-activated potassium current; INa, sodium current; ICaL, L-type calcium current; ICaT, T-type calcium current; If, funny/pacemaker current; Ito, transient outward current; IKCa, calcium-activated potassium current; IKr, IKs, and IKur, rapid, slow, and ultrarapid components of delayed rectifier potassium current; Kir, inward rectifier potassium channel; KV, voltage-gated potassium channel; NaV, voltage-gated sodium channel; TASK, Tandem of pore domains in a weak inward rectifying potassium channel (TWIK)-related acid-sensitive potassium channel; TTX, tetrodotoxin.
The sinoatrial node (SAN) is located in the upper part of the right atrium, and it has a special role in the heart, serving as the natural pacemaker. SAN cells have a relatively low maximal diastolic potential (less than −60 mV), that, after the termination of repolarization, gradually becomes less negative during the so-called spontaneous diastolic depolarization—until it reaches the potential range of L-type Ca2+ channel activation, thereby generating a new action potential. The exact nature of the pacemaker function in the SAN is still under debate (515–519). Originally, it was thought that the slow diastolic depolarization was due to a hyperpolarization-activated inward current. This current was first described in cardiac Purkinje fibers and was thought to be carried by Na+, based on its dependence on extracellular [Na+] (520). Later on, other studies suggested that the slow diastolic depolarization was the result of a decaying K+ current called IK2 (521, 522). These data were subsequently reinterpreted in terms of ionic currents resulting from extracellular accumulation of K+, and a hyperpolarization-activated inward current, called “funny current” (If), carried by both Na+ and K+, was described by Di Francesco (524) and others (523, 525). Despite the fact that other currents, such as IKr, T-type ICa and NCX, were also suggested to play a role in the SAN pacemaker activity (526–529), for more than two decades since its discovery the major mechanism underlying the pacemaker function was generally considered to be If. However, in an early study of Noma, Morad, and Irisawa (530), the significance of If in the SAN pacemaker function was questioned. These authors showed that cesium, an inhibitor of If, did not influence the SAN spontaneous frequency and did not eliminate the epinephrine-induced increase in SAN frequency. The dominant role of If in SAN pacemaker function was later also challenged by studies showing that a cycling spontaneous release of Ca2+ happens during diastole, and this can cause spontaneous depolarization by activating forward NCX as an alternative mechanism for the pacemaker function (“calcium clock hypothesis”) (203, 531–534). This controversy seems to be settled by the “coupled clock hypothesis” (FIGURE 11), suggesting an important role for both the “calcium clock” and the “membrane clock” (If) (528, 535). Also, in a recent study, Morad and Zhang (517) demonstrated and pointed out that If was very small and very slowly activating at the range of maximal diastolic potential of the SAN cells (−60 mV), therefore not enough to generate a significant amount of pacemaker current. These authors suggested that expression and function of inward If in SAN cells can counterbalance the electrotonic interaction of the more negative resting potential of the surrounding atrial cells. According to this suggested function, If is important since it insulates SAN cells from the hyperpolarizing influence of the atria, thus allowing a proper SAN function. Consistent with this speculation, Boyett et al. (536) found higher HCN (If) expression in peripheral rabbit SAN tissue than in the central core.
A recent study in human induced pluripotent stem cell (HiPSC)-derived myocytes suggested that mitochondria could also play a role in the spontaneous activity, setting the rhythm of the “calcium clock” by taking up and releasing Ca2+ from and to the cytosol (517). However, other studies suggested that mitochondrial Ca2+ transient decay is slow, and that their Ca2+ efflux is relatively small during one cardiac cycle (463), thus raising questions about a significant role of mitochondrial Ca2+ release on SAN frequency. Further studies in adult SAN cells are necessary to confirm the possible role of mitochondria in pacemaking.
In general, SAN frequency can be slowed down by inhibiting If, NCX, ICa,L, or IKr in feline, rabbit, and porcine SAN myocytes (537–540). Inhibition of If and NCX decreases the slope of diastolic depolarization, while inhibition of ICa,L would shift the threshold potential toward more positive values, and thereby the cycle length is increased in all cases. Inhibiting IKr prolongs repolarization of SAN cells and lengthens their spontaneous cycle lengths. It has to be considered, however, that loss-of-function mutation of If results in sinus bradycardia, arguing for some role of If in SAN pacemaking (541).
SAN frequency can also be affected by intracellular cAMP levels. Elevation of intracellular cAMP increases If and shifts its activation to more negative potentials (542), which in turn enhances ICa and intracellular Ca2+, further favoring the increase of SAN frequency. Vagal stimulation has an opposite effect on intracellular cAMP and SAN frequency. Data from SAN cells isolated from Girk4 (Kir3.4)-knockout mice lacking IK,ACh suggest that it may also activate IK,ACh, which can cause hyperpolarization and may also contribute to SAN bradycardia (543).
SAN cells lack cardiac type Nav1.5 fast Na+ channels, and therefore their depolarization is caused by ICa,L (544). Recently, neuronal Nav1.6 Na+ channels were reported in SAN cells (90, 545), but their role in SAN function is not well understood (529). The SAN action potential does not show a distinct plateau phase with relatively weak IK1 current.
Experimental evidence obtained in spontaneously beating guinea pig sinoatrial cells suggests that If function decreases SAN frequency variability, which is the intrinsic behavior of the calcium clock function (546).
Elucidating the exact mechanisms underlying pacemaker function and its regulation still remains an important task for the future, since it provides the regular heartbeat but it can also act as a possible trigger for serious atrial and ventricular arrhythmias (547, 548).
Large species-dependent variations make it difficult to describe the general shape of cardiac action potentials, including atrial action potentials (FIGURE 12 and FIGURE 13). There is also significant diversity within the same heart (549), at least in humans (3, 550), and this results in a relatively large dispersion of repolarization, which favors the development of atrial fibrillation.
In most species, the atrial action potential lacks a long and stable plateau phase. In human atria, most of the atrial action potential recordings show a plateau phase, but at a more negative voltage range (−10 to −30 mV) than observed in the ventricle. This is due to the function of the abundantly expressed Ito and IKur potassium currents (FIGURE 6). Ito in the atria has slower inactivation kinetics than in the ventricles, having a robust slowly deactivating component (τ = 91 ms at −20 mV), with slower recovery (551) from inactivation (τ = 125 ms) than the ventricular one. Therefore, frequency-dependent changes of APD and restitution are different in atrium and ventricle. In the atria IKur is large (85) and contributes to repolarization, in contrast with the ventricle, where it is absent or very weakly expressed (85). Therefore, IKur inhibition would be expected to prolong the AP. However, inhibition of IKur shortens repolarization by shifting the plateau voltage to the positive voltage range (309), thus changing the activation and deactivation/inactivation of other plateau currents, such as IKr, IKs, ICa, and INaLate. IKur inhibition has also been shown to slightly prolong human atrial APD in tissue obtained from chronic AF patients (309, 552). However, in this case, phase 1 and 2 repolarization were delayed and shifted to positive potentials, because of electrophysiological remodeling, thus allowing for a larger IKur contribution compared with normal conditions. It is interesting that, in rabbit (unlike in human) atrial muscle, sustained Cl− current was reported after Ito inactivation (324, 553). This should be taken into consideration when drugs are studied in rabbit atrial preparations, and it highlights the importance of species-dependent electrophysiological differences. A distinct, cellular swelling-induced Cl− current has also been described in human atrial myocytes (554), which is also modulated by PKA-independent, cAMP-mediated β-adrenoceptor signaling (555). The existence and potential role of Nav1.8 channels in INaLate in the atria represents an interesting issue; however, it is still the subject of debate (556) and requires further studies. L-type and T-type ICa have similar properties in atria and ventricle (557). IK1 current density is relatively small in the atria, especially in the voltage range between −80 and 0 mV. This is consistent with the finding of lower Kir2.1 mRNA expression in right atrial compared with right ventricular human tissues (85). Recently, it has also been reported that neurokinin-3 receptor activation induced prolongation of atrial refractoriness, which was attributed to the inhibition of a nonspecific K+ background current (297). It is difficult to measure IKr and IKs in atrial myocytes, most likely because of the cell isolation techniques (252). Regardless of that, IKr block significantly lengthens atrial repolarization, in both animal and human studies (558, 559). On the other hand, the role of IKs in atria is not well explored. Experiments with chromanol 293B on atrial action potential are not conclusive, since this drug inhibits both IKs and Ito (560). It must be emphasized that, in physiological conditions, the atrial plateau voltage is more negative than the activation threshold of IKs. Therefore, IKs is not expected to contribute to atrial repolarization. However, it has been reported that in atrial tissue LQT1, MinK, and MiRP levels are similar to those in ventricular tissues (85). On the basis of these results, it can be speculated that, at fast heart rate (enhanced sympathetic tone) and in situations where the atrial plateau is shifted to positive voltage, IKs may have a role in atrial repolarization. Therefore, its modulation could influence arrhythmogenesis. Indeed, it has been shown that two gain-of-function mutations in KCNQ1 (S140G and V141M), detected in patients with AF (275, 277), markedly slowed deactivation of IKs and contributed to the development of AF (561). Unlike in the ventricles, most studies showed the existence and function of small-conductance, apamin-sensitive and calcium-dependent potassium current (SK2) in normal atria, but its significance seems to be far more pronounced in diseased tissue (5, 314, 316, 317, 562). The increased apamin-sensitive SK current was found along with decreased mRNA and protein levels of SK1, SK2, and SK3 channels in human atrial cardiomyocytes isolated from patients with AF (319). However, in the same experiments CaMKII was increased and its inhibition by KN-93 reduced the apamin-sensitive SK currents to a higher degree in myocytes isolated from patients with AF compared with those in sinus rhythm (319), suggesting that SK channels are more sensitive to Ca2+ in AF patients and CaMKII modulation may represent a pharmacological target in the management of AF. Neurohormonal modulation of the atria is particularly important. Acetylcholine (358, 563), catecholamines (555, 563), substance P (297), adenosine (564, 565), and serotonin (566) have been reported to influence atrial action potential and its underlying currents. In the atria, IK,ACh is robust and even has a small but persistent constitutively active component, which operates without parasympathetic stimulation and seems greatly augmented during chronic AF (358). Accordingly, atrial tissue expresses mRNAs for Kir3.1 and GIRK channels abundantly. Of note, TWIK TASK channels are also relatively abundantly expressed in the atria (85), but their roles in atrial electrophysiology are not well explored yet.
Although they have similarities, AV node cells differ from the SAN cells (593–595), with a lower spontaneous frequency and a more negative maximal diastolic potential (596, 597). The structure of the AV node, like the SAN, includes a large variety of different cell types (598, 599), with different channel expression (600–603). In the rat AV node, high expression levels of HCN4, Cav3.1, Cav3.2, Kv1.5, Kir3.1, and Kir3.4 and low expression of Nav1.5 and Kir2.1 mRNA were measured compared with those observed in the ventricle (600). In rabbit AV node, no or low expression of Nav1.5, Cav1.2, Kv1.4, KChIP2, and RYR3 and high expression of Cav1.3 and HCN4 mRNA were reported (601). Like SAN cells, AV nodal cells do not have functioning fast Nav1.5 INa channels, and therefore their depolarization and impulse conduction depend on the function of ICa,L (565, 603). In AV nodal cells, the expression and function of IK,ACh is particularly important (604, 605), since their activation via the adenosine 1 or muscarinic receptors (606) hyperpolarizes the AV nodal cells, which slows or blocks impulse conduction in the AV node (607, 608). In addition, adenosine and acetylcholine decrease ICa,L via Gi protein signaling pathway, further decreasing the safety of impulse propagation through the AV node. These are the principal cellular mechanisms that make intravenous adenosine so useful in stopping AV nodal reentry tachycardia (604). The T-type Ca2+ channel is expressed and is functional in the AV node as in the SAN (607). Indeed, mibefradil, a potent ICa,T inhibitor, increased AV nodal conduction time and even elicited second- or third-degree AV block in isolated, blood-perfused dog hearts (607). The slow conduction through the AV node is physiological and provides a time lag for contraction between the atria and the ventricles. AV nodal tissue has diverse and different connexin 40, 43, and 45 distribution compared with other parts of the heart (8), with a complex and diverse structure, containing a dual faster and slower impulse conduction pathway (594, 603, 609). This latter can provide the basis of fixed-rate supraventricular reentry tachycardia as was elegantly demonstrated in an early study in rabbits by Janse et al. (610), and this tachycardia can be terminated by blocking ICa,L and IK,ACh and by adenosine (604, 611, 612).
AVN cells lack Ito and have background sodium inward current flowing through a nonselective cation channel (605, 613). These cells express functioning IKr, IKs, and If, but If is not required for their pacemaking (614).
Purkinje fibers play a pivotal role in impulse conduction and propagation in the ventricles (615, 616). Purkinje cells can also act as subsidiary pacemakers, and they display a spontaneous diastolic depolarization, although their frequency is normally inhibited by the higher-frequency discharges of the SAN, causing overdrive suppression. Purkinje fibers have a longer APD than do ventricular cells, and this may have a protective function against retrogradely propagating stimuli (615) but it can also represent a source of arrhythmogenic repolarization inhomogeneity, even in the normal heart, which can increase in diseased conditions or after drug exposure. In certain pathophysiological situations, Purkinje fibers can show triggered activities (617), such as EADs, DADs, and cellwide ectopic Ca2+ waves, in surviving tissue in the border zone of an infarct (618). It was recognized long ago that Purkinje strand fibers, which run close to or on the surface of the endocardium, are less affected by ischemia-induced tissue damage compared with ventricular cells (118, 619). Purkinje fibers have been extensively studied with the conventional microelectrode (620) and the two-electrode voltage-clamp (621) techniques. However, since cell isolation from cardiac Purkinje tissue is particularly difficult, research in Purkinje fibers has benefited less from the introduction of the patch-clamp technique, despite its importance (622).
Purkinje fibers have a special role in impulse conduction, with their depolarization being about two to three times faster (300–750 V/s) than in atrial or ventricular (100–250 V/s) muscle preparations (623). The fast depolarization is thought to be due to the abundant expression of the TTX-sensitive Nav1.5 channel isoform (85), but low density of TTX-sensitive neuronal Nav1.1 and 1.2 Na+ channel isoforms may also significantly contribute to INa in Purkinje fibers (624, 625). Importantly, the relation between impulse conduction, upstroke velocity (Vmax), and ionic currents is not linear (626–628), and Vmax is not a direct measure of ionic currents. INa in Purkinje fiber also seems to have a particular impact on repolarization, due to its relatively large slowly inactivating component (12, 13, 82). Some authors suggested that the skeletal muscle Na+ channel isoform Nav1.4 (78) while others suggested that Nav1.7 subunits (85) were responsible for this slowly inactivating component in Purkinje fibers; however, this issue is not clarified and needs further investigations. In canine Purkinje fibers, connexin 40 is more abundantly whereas connexin 43 is similarly expressed compared with ventricular myocardium (629): this can play a role in differences between their conduction properties. The large phase 1 repolarization in Purkinje fibers (FIGURE 6) is caused by the fast inactivation of INa and activation of Ito. Not only is Ito larger than in ventricular muscle, but it inactivates somewhat slower and recovers from inactivation >10 times slower than that in the ventricular muscle cells (630). This behavior of Ito is attributed to the different expression of Ito α- and β-subunits in dog ventricular muscle and Purkinje fibers (141, 157). In human ventricular myocytes Kv4.3 channel subunits dominate, whereas in human Purkinje fibers abundant Kv3.4 and Kv4.3 expressions were reported, with marked differences also in β-subunit (KChIP2, KChaP, KCNE2) expression patterns (158). The slow Ca2+-dependent Ito2 Cl− current was also described in rabbit Purkinje fibers (631) and implicated in phase 1 repolarization and DAD formation in sheep Purkinje cells (632), since at potentials more negative than the chloride equilibrium potential chloride channels conduct depolarizing current. The L-type ICa is well expressed in Purkinje fibers and, unlike in ventricular myocytes, it is carried not only by Cav1.2 but also by Cav1.3 channels (633), and this may result in some differences in the properties of ICa,L between ventricular and Purkinje myocytes. In the canine Purkinje fiber, there is larger T-type ICa, with a higher Cav3.2 expression than in ventricular and atrial myocytes (633). Based on this, it was speculated that T-type ICa has an important role in Purkinje fibers, by contributing to both depolarization and pacemaker function (634). IKr, IKs, calcium-activated K+ currents, as well as IK1 have been described in Purkinje fiber (323, 635, 636). Accordingly, inhibition of IKr and IK1 significantly lengthens repolarization, even more so than in the ventricular muscle (615). However, the inhibition of IKs does not change repolarization of Purkinje fibers in the normal situation (256). This is not surprising, since Purkinje fibers have a plateau voltage less positive then 0 mV, which is below the activation threshold of IKs. However, at high frequencies and a high level of sympathetic activation, the plateau level is shifted toward more positive values, increasing both IKs amplitude and speed to such a level to influence both repolarization and pacemaker function (256).
Unlike ventricular but similarly to atrial muscles, Purkinje fibers express IK,Ach and Kir3.1 GIRK channel subunits, thus responding with an APD shortening upon acetylcholine administration (637). The pacemaker If current is robust in Purkinje fibers (638–640), where it was first discovered (524), and it plays an important role in the pacemaker function. Later, a K+ current, IKdd, was also described in Purkinje fibers (640), deactivating at more positive potentials than If and thus having an additional role in the pacemaker function in Purkinje fibers. In addition, spontaneous Ca2+ release-induced intracellular Ca2+ waves can also modulate normal Purkinje fiber pacemaker activity (641). This may have particular importance in ectopic automaticity of Purkinje fibers surviving myocardial infarction (618). Recently, significant SK2 current and channel expression were described in rabbit Purkinje fibers, and an important role for SK2 current in Purkinje fiber repolarization was suggested (323).
Free-running Purkinje strands emerging distally into ventricular muscle constitute a relatively large-resistance connection (642), and a high degree of sink for current flow, and a more favorable site of conduction block than other parts of the healthy myocardium. Also, because of the weaker electrotonic coupling, the dispersion of repolarization here can be far greater than in other places (201). Indeed, it was experimentally proven that EADs in free-running Purkinje strands could elicit extrasystoles in ventricular muscle (643) by electrotonic interaction (FIGURE 15), which is unlikely to happen in healthy and well-coupled regions within the ventricular wall.
Since Purkinje fibers are considered particularly important in arrhythmogenesis (644), further research studying Purkinje fiber ventricular muscle junctions or subendocardial layer containing a mixture of Purkinje fibers and ventricular muscle would offer promising results.
Ventricular action potentials (FIGURE 6) in humans and in most mammals have a positive (+20 to +30 mV) and relatively long plateau phase, with a small or pronounced phase 1 repolarization and notch afterwards, depending on their transmural site of origin (subendocardial, midmyocardial, or subepicardial) (143, 645, 646). Regardless of their origin, they express large INa and ICa,L and relatively weak but persistent INaLate (80). These currents provide robust depolarization and thereby secure impulse conduction. In addition, by counterbalancing outward potassium currents, e.g., Ito, IKr, IKs, and IK1, they participate in the maintenance of the plateau phase (FIGURE 6). As mentioned above, IKur and IK,Ach are expressed weakly or not at all in the ventricles.
There are several reports indicating electrophysiological differences between the right and left ventricle. In dog hearts, it was found that left ventricular myocytes had longer APD than those in the right ventricle (647, 648). Also, right ventricular myocytes exhibited more pronounced phase 1 repolarization and larger Ito (648, 649) and increased IKs (648). It was also reported that canine subepicardial and subendocardial myocytes in the left ventricle possessed larger INa with higher Vmax compared with those in the right ventricle (650). The authors suggested that these differences provided a mechanism for the right ventricular manifestation of Brugada syndrome (650).
Almokalant, a drug with APD-prolonging effects, increased interventricular dispersion of repolarization that was associated with the occurrence of EADs and torsade de pointes arrhythmia in dogs with chronic AV block (647). In these dogs, the APD prolongation was larger in the left ventricle than in the right ventricle (647). The opposite was observed in guinea pigs, i.e., the APD-prolonging drugs dofetilide and quinidine lengthened APD more in the right ventricle (651). These findings highlight important differences among species in their responses to drugs with repolarization-prolonging effects. In humans, similarly to dogs, the APD is longer in the left ventricle with slower adaptation to increased heart rate than in the right ventricle (652). These interventricular differences in repolarization are not sufficient to cause arrhythmias in the normal heart; however, in the presence of an ischemic region at the left ventricle (LV)-right ventricle (RV) junction, the interventricular APD heterogeneity and different AP rate adaptation promote reentry arrhythmias (652).
Important regional electrophysiological differences were described in the dog ventricle long ago (653), transmurally (654), regionally (655), and in the basoapical direction (143). These differences are due to substantial differences in the density of expression of various transmembrane ion channels, shaping the action potentials accordingly in the different regions of the ventricles (645). Transmural differences in repolarization have been extensively studied and are well explored (656). It has been found that subepicardial cardiomyocytes exhibit a large phase 1 repolarization and Ito compared with those in the subepicardium (657). Myocytes isolated from the midmyocardium, called M cells (654), are variable in this respect, but they are still characterized by a distinct phase 1 repolarization and a relatively large Ito (654), INaLate (658), and NCX (659) and small IKs (660). All these ion current characteristics contribute to the longer APD of M cells compared with subepicardial and subendocardial myocytes leading to a substantial transmural dispersion of repolarization (656). Augmented transmural dispersion of repolarization has been considered a contributor to ventricular tachycardia and fibrillation development in patients with Brugada syndrome, acquired and congenital LQT syndromes, short QT syndrome, and catecholaminergic polymorphic ventricular tachycardia (CPVT) (656). However, these differences in transmural repolarization are relatively small in the intact, undiseased ventricles (642, 661), since the neighboring myocytes are well coupled. Accordingly, experimental evidence suggests that in the intact heart the transmural dispersion of repolarization is considerably less (642, 661, 662) than that reported in tissue slices and perfused wedge preparations (663, 664). Therefore, it was argued that it is unlikely that a large, significant repolarization gradient between the subendocardium, M cells, and subepicardium existed, contributing to EADs or arrhythmias (665). However, in pathological settings and/or drug treatment and at Purkinje fiber ventricular junctions (FIGURE 15), APDs can lengthen in a nonuniform manner and cellular coupling can deteriorate, thus resulting in a substrate for serious ventricular arrhythmias.
Apico-basal electrophysiological differences have also been described: the APD was shorter and phase 1 repolarization was markedly larger in canine cardiomyocytes isolated from the apical region than in those from the basal region (655). In the same study, larger Ito and IKs were observed in apical than in basal cardiomyocytes (655). In this context, the apico-basal and the antero-posterior, and not the transmural, repolarization gradients have been considered to contribute to the generation of the T wave (665, 666).
Since idiopathic ventricular arrhythmias often originate from the right and left ventricular outflow tracts (RVOT and LVOT), the electrophysiological properties of these regions have also been investigated (667–670). It was found that rabbit right ventricular myocytes from the apex had longer APD, larger Ca2+ transients, higher Ca2+ stores, increased INaLate and Ito, but smaller IKr, L-type Ca2+ current, and NCX than RVOT myocytes (671). These differences were associated with increased incidence of DADs induced by pacing (671). Myocytes from the LVOT exhibited longer APD, larger INaLate and NCX, and smaller Ito and IKr than those from the RVOT (669).
Arrhythmia research is often performed in different animal models, but its ultimate goal is to understand the mechanisms in humans and to prevent, or successfully treat, arrhythmias in patients. Therefore, it should always be kept in mind how experimental results can be extrapolated to humans. It is often overlooked that rodents (rats and mice) have ion channel expression profiles distinctly different from humans (40, 672). This also results in different cardiac electrophysiology properties (672, 673), especially during repolarization (FIGURE 12). Mice and rats have a high heart rate (600 and 400 beats/min, corresponding to cycle lengths of 100 and 150 ms), ∼10-fold faster than in humans. As a consequence, mice and rats have very short ventricular and atrial action potentials, to provide enough time for diastole. These short action potentials result from the presence of transmembrane currents like Ito and IKur (144). In the ventricles of larger mammals (guinea pig, rabbit, dog, etc.) or humans, these channels are expressed less or not at all and also have a different molecular background and functional role. Action potentials in mice or rats lack a plateau phase (144); therefore IKr and IKs are not likely to operate despite the fact that expression of mRNA for these channels has been reported (674). Notably, both IKr and IKs were observed in neonatal mouse ventricle, but after further development neither IKr nor IKs was detected in adult mouse ventricular myocytes (675). Also, inward currents like ICa and INa have different impact on ventricular repolarization than in other mammals. Consequently, drugs can have marked species-dependent effects on action potentials as demonstrated with an example of IKr inhibition in FIGURE 13. These fundamental differences mean that mice and rats can only be properly used in arrhythmia and related pharmacological research if the limitations of the models are described. As FIGURE 13 illustrates, pharmacological inhibition of specific potassium channels, such as IKr, Ito, and IKur, elicits strikingly different effects on ventricular repolarization in the rat, a commonly used laboratory experimental animal, compared with humans. Despite this, hundreds or thousands of papers using mice or rats have been published on this topic, because these animals are relatively cheap and easy to house. In addition, transgenic manipulations of transmembrane ion channels are almost entirely applied in mice (674), with very few exceptions (676–678). This makes the mouse a favorable target, despite the fact that the mouse can be useful to study sodium and calcium channels and connexins but not potassium channels.
There are far less consistent data on characteristic species-dependent differences in atrial action potentials and the underlying specific transmembrane currents. As FIGURE 10 shows and several papers indicate (297, 645, 679, 680), most of the species commonly used in experimental laboratories exhibit similar action potentials and underlying currents, with the atrial action potentials lacking a plateau phase, except in humans (556, 681). Human atrial tissue samples, unlike ventricular samples, can be obtained from cardiac surgery departments (from patients undergoing open heart surgery for coronary artery bypass grafting, heart valve repair or replacement); therefore, human atrial cellular electrophysiological data are abundant, somewhat limiting the need for such studies in experimental animals.
Although the species differences in the shape of ventricular action potentials are most striking between small rodents and humans, important species differences also exist in the action potentials between humans and other mammals. Guinea pig ventricular muscle, unlike human, does not express Ito and does not exhibit a prominent phase 1 repolarization (144, 145); however, it expresses large IKs with distinct gating properties compared with human (3, 214, 249). Rabbit Ito, unlike human, is conducted mainly by Kv1.4 channels, and as a consequence cycle length-dependent APD is markedly different from that observed in human ventricle (682). Pig ventricular muscle exhibits Ca2+-activated Ito chloride current that shapes phase 1 repolarization (683) but lacks 4-AP-sensitive Ito despite abundant expression of Kv4.2 and KChIP2 mRNA and proteins (684). The dog ventricular muscle was found to express a considerably higher density of IK1 compared with human (685). This results in a stronger repolarization reserve and consequently less APD prolongation upon IKr inhibition in the dog ventricle compared to human. All of these differences have a particular significance when drug effects and pathophysiological electrophysiological alterations are extrapolated from animal models to humans.
Human induced pluripotent stem cell-derived cardiac myocytes (HiPSCs) are increasingly used in cellular arrhythmia research (686, 687). This new approach is promising and expanding rapidly (688). At the present stage, however, it seems that HiPSCs have some important limitations (689). Although in HiPSCs experiments can be performed relatively fast, at present they cannot provide a substitute for carefully applied animal preparations. The so-far unresolved problems with HiPSCs are the following: data include cardiomyocytes that have not fully differentiated, still showing an immature phenotype (690, 691), and the cells spontaneously beat and often have a relatively low resting potential because they lack IK1 and also have low upstroke velocity (692–694). However, an interesting study suggests that the low resting potential and reduced IK1 are not necessarily inherent characteristics of HiPSC-derived cardiomyocytes; rather, these observations might be due to technical issues related to performing patch clamp on the relatively smaller cells (692). Also, HiPSC sarcomeres are disorganized, and their shapes are different from those of the adult cells (691). So far, atrial and ventricular-like HiPSC cells have been successfully generated, whereas SAN, AV node, or Purkinje-like stem cell generation has been unsuccessful (691). However, HiPSC-derived myocytes from patients with defined mutations using CRISPR/Cas9-edited cells can mimic diseases (688, 695–698) that are often hard or impossible to recreate properly in animal experiments. Recent efforts to culture and continuously pace HiPSC-derived cardiomyocytes cultured in collagen gels as “engineered heart tissue” represent a new two-dimensional (2-D) and three-dimensional (3-D) approach (699–701), since it resembles more the mature myocardium. Future research in this area, however, may revolutionize the field, opening new horizons for arrhythmia research.
Impaired depolarization capability in cardiomyocytes can reduce conduction safety, providing proarrhythmic substrate and potentially contributing to conduction block and reentry arrhythmia. One example of depolarization abnormality leading to reentry arrhythmia is the Brugada syndrome, caused by the loss of function of SCN5A (778, 779). It should be noted, however, that in patients with Brugada syndrome the mechanisms of arrhythmia are more complex, since decreased function of L-type ICa and enhanced function of Ito also contribute to the development of phase 2 reentry (780), due to enhancement of dispersion of repolarization across the ventricular wall (779).
Ischemia can cause regional membrane depolarizations, which indirectly decrease the strength of INa by eliciting partial or full channel inactivation and result in slowing of impulse conduction or unidirectional or bidirectional conduction block. All of these factors are considered important in arrhythmogenesis. However, their detailed discussion is beyond the scope of this review, since they are described by others in great detail (31, 781–785). Drugs that inhibit INa can have similar effects and can also cause reentry arrhythmias (202, 786, 787).
Upregulation of the If current (788) and HCN channels (789) in the ventricles and atria (790) was reported in HF (789) and HCM (791, 792). These changes can cause abnormal myocardial depolarizations, and they can relate to increased incidence of ectopic beats, providing possible triggers for arrhythmias in an environment where dispersion of repolarization (arrhythmia substrate) is already augmented by structural heart disease.
Abnormal myocardial automaticity (formation of propagating spontaneous action potential) is an established trigger contributing to arrhythmia onset (26). One particular type of such automaticity closely linked to disturbance of normal cellular electrophysiology is the so-called “triggered automaticity,” which requires preceding action potentials that are essential for the subsequent spontaneous firing (793). Depending on the temporal relationship between such depolarization and the preceding action potential, triggered automaticity is typically separated into early and delayed afterdepolarizations.
Early afterdepolarization (EAD; FIGURE 10, A and B) is characterized by depolarizing potential changes occurring before the termination of the preceding action potential during phase 2 or phase 3 repolarization. EADs are usually generated when action potential duration is excessively prolonged, e.g., when IKr is impaired. As a consequence of the lengthened action potential, those L-type calcium channels that have already recovered from inactivation can reopen, and some of the calcium channels carry a Ca2+ window current (707, 708), causing positive voltage oscillations during the plateau (phase 2 EAD) or terminal repolarization (phase 3 EAD). The Luo–Rudy studies (707, 708) also proposed a second type of EADs, phase 3, resulting from spontaneous calcium release during repolarization, which then translates into depolarization via NCX. The relevance of this mechanism was subsequently demonstrated experimentally in Purkinje fibers (131). This type of EAD strongly resembles delayed afterdepolarizations in its mechanism but differs in the timing (during AP vs. after AP). Both types of EADs are rate dependent, with the reactivation-driven mechanism appearing predominantly at slow pacing, whereas the release-driven EADs occur at fast pacing (794).
In general, both the L-type calcium current and NCX are known to act synergistically in EAD formation (795, 796), and the controllability of computational models has been used to understand their interplay. For example, Kurata et al. (797) demonstrated how NCX contributes to EADs in the popular O’Hara–Rudy model (736) via two distinct mechanisms. First, the influx of calcium via the L-type calcium current upon its reactivation translates into calcium efflux via NCX and thus additional inward current (which can, in turn, promote further activation of L-type calcium current). Second, NCX expressed adjacent to L-type calcium channels acts as a “sanitizer” of calcium in the cellular region, removing calcium ions from the junctional subspace during repolarization. This may subsequently reduce the calcium-dependent inactivation of L-type calcium current, facilitating earlier reactivation. The insight into EAD origins is not limited to L-type calcium current and NCX; nonequilibrium gating of late sodium current was implicated in EAD formations (798). The increased availability of data on signaling pathways has also enabled computationally driven insights on how EADs are facilitated via CaMKII- or PKA-driven pathways (799). This is particularly important for our understanding of EADs in disease conditions such as heart failure, where these pathways are dysregulated.
One interesting involvement of IKr in EAD formation beyond the role in APD prolongation lies in its dynamic of activation and reactivation. Lu et al. (719) used a range of electrophysiological protocols to demonstrate an intriguing interplay of IKr activation, inactivation, and recovery from inactivation, which leads to rapid increase of IKr during late-plateau reactivation, such as during an EAD. Such an increase in repolarizing current would be expected to counteract and potentially outweigh depolarizing currents, potentially preventing the formation of a larger-amplitude EAD.
The delayed afterdepolarization (DAD) is characterized by depolarizing potential changes following the termination of the preceding action potential (129, 130, 632) during diastole (FIGURE 10C). DAD is generally attributed to calcium-sensitive depolarizing currents after spontaneous Ca2+ release from the SR during Ca2+ overload, or CaMKII-dependent phosphorylation (800, 801), which is promoted by diseases like chronic AF, ischemia, heart failure (801–803), and catecholaminergic polymorphic ventricular tachycardia (CPVT) or drugs like digitalis. The most important calcium-sensitive current implicated in DAD formation is the forward-mode NCX, but a role for the calcium-sensitive chloride current has also been suggested (804). The calcium-induced depolarization is opposed primarily by IK1, which tries to maintain resting potential (805). When the depolarization induced by calcium-sensitive currents is of sufficient magnitude, overcoming IK1, it activates INa, triggering a new action potential. Automaticity occurring in the pulmonary veins and in the myocytes represents important abnormal impulse formations, and at present it seems that their cellular mechanisms are complex, including the possibility of DAD and EAD generation as well (FIGURE 14).
The spontaneous calcium release is a stochastic phenomenon, represented by calcium sparks (806, 807) and intracellular calcium waves (807–810), and multiple calcium release sites need to synchronize to produce a cellwide calcium release (811, 812). The stochastic nature of such events as well as generally limited controllability and observability of subcellular calcium handling in the experimental setting, complicate detailed understanding of their origins. On the other hand, computer models (736) offer excellent controllability and observability and thus are a popular tool to understand origins of spontaneous calcium release and ultimately DADs. Increasing availability of subcellular experimental data has enabled the construction of spatially detailed models, where the cell is subdivided into up to hundreds of thousands of subdomains with separate clusters of potentially stochastic ryanodine receptors (760, 813, 814). As a result, such models can give very detailed predictions, elucidating how originally random calcium sparks are recruited into calcium waves and ultimately DADs but also giving explanations for DADs that do not rely on calcium waves (814).
One important aspect of triggered automaticity is that the role of EAD and DADs in arrhythmogenesis is most likely overestimated in single-cell experiments versus the intact heart. Even in relatively poorly coupled tissue, electrotonic interactions with neighboring cells will decrease the depolarization produced by an EAD or a DAD. However, moderate uncoupling will decrease the electrotonic interaction between a focus and the surrounding cells and can actually favor action potential propagation (815).
Computer simulations studies have significantly contributed to the understanding of the conditions under which the afterdepolarizations of single cells may translate into propagation throughout myocardium (816). These modeling studies showed that when simulating healthy cells and their afterdepolarizations, ∼70 cells manifesting an afterdepolarization are needed to trigger excitation in a fiber, ∼7,000 in 2-D tissue, and ∼700,000 in 3-D tissue. The specific numbers may very well vary with numerical aspects related to the simulations, such as the mesh discretization, but this nevertheless suggests the unlikeliness of EADs or DADs promoting into tissue reactivation in a healthy tissue. At the same time, however, the study by Xie et al. (816) also investigated the effect of gap junction uncoupling, fibrosis, and heart failure-like remodeling, showing that the combined effect may reduce the number of cells needed for an afterdepolarization-driven propagation by two orders of magnitude. Specific patterns of uncoupling, such as thin strands of myocytes within postinfarction scars (817), which are similar to a fiber with regard to coupling, may increase the relevance of afterdepolarizations of arrhythmia even further. Another type of weakly coupled tissue that might enable synchronization of afterdepolarizations is the endocardial Purkinje ventricular junctions, as mentioned above (FIGURE 15).
In addition to promotion of extrasystolic reactivation via cell decoupling, other mechanisms of synchronization of afterdepolarizations are via stretch-activated channels (818), current flow in the border zone of acute ischemia (706), and partial chaos synchronization (369). In a modeling study of calcium-driven afterdepolarizations, it was suggested that calcium waves also synchronize by calcium flux through gap junctions (819); however, experimental results argued against this possibility (811). Ultimately, simulation studies have shown that the baseline risk of EADs may be considerably increased in diseased conditions (776, 820), further facilitating translation of a depolarization to tissue activation.
In the case of increase in intracellular calcium during elevated sympathetic drive and/or diseases like heart failure or catecholaminergic polymorphic ventricular tachycardia (CPVT) (821), Ca2+ overload may occur and the SR can become leaky and release additional Ca2+ (822). Mutations in the gene encoding the cardiac ryanodine receptor-related Ca2+ release channels (RyRs) can cause extrasystole and serious tachycardia such as CPVT by abnormally releasing Ca2+ from the SR (823) into the cytosol on the response of catecholamines which Ca2+ would activate the electrogenic forward NCX depolarizing cells beyond their threshold of activation. In a recent study, in a new model for CPVT in engineered human tissue fabricated from human pluripotent stem cell-derived cardiomyocytes, high-frequency pacing and isoproterenol administration increased Ca2+ wave propagation heterogeneity and elevated intracellular [Ca2+], leading to local depolarizations and conduction block (creating the arrhythmia substrate), and subsequently resulting in reentry (700). Similarly to CPVT, leaky ryanodine receptor-related Ca2+ release channels were also reported in heart failure (802, 824).
It was observed long ago that action potential duration and impulse conduction depend on heart rate or on the stimulation frequency. To study frequency-independent repolarization changes caused by disease, drugs, or any other factors, correction formulas have been used to estimate QT intervals corrected for heart rate (QTc) on the ECG. At elevated heart rate, extracellular K+ accumulation may occur in the clefts, slightly depolarizing the resting membrane potential that slows impulse conduction and impairs safety of impulse propagation. At extrasystoles early following the end of ERP, repolarization is still not fully terminated and Na+ and Ca2+ channels are partially inactivated, resulting in less depolarizing current and decreased safety or slowed impulse propagation. Frequency-dependent APD changes show general patterns (682, 825–830) that APD is short at high and longer at slow constant rates (FIGURE 17). The frequency dependence of APD shows substantial species, tissue, and regional variation and has important implication for arrhythmogenesis. At slower heart rate, APD can be markedly prolonged, favoring triggered arrhythmias via EAD formation, and may also result in enhanced substrate for arrhythmias by increasing dispersion of repolarization.
Electrical restitution refers to the recovery of the APD of an interpolated beat as a function of time following the previous beat. This changes in a manner that is somewhat similar to that seen (831) during frequency-dependent steady-state APD changes (FIGURE 17, left). Despite the similarities, there are important differences (FIGURE 17, right) that warrant the restitution being studied as a separate rate-dependent property (826, 827, 832–835), with importance for arrhythmia research (20, 836). According to the restitution hypothesis, as the diastolic interval increases because of the propagation of an extra beat, the second extrasystole would encounter longer APD/ERP and local conduction block may occur. A steeper restitution curve would favor such an effect and is considered proarrhythmic (20, 662, 837, 838), whereas flattened electrical restitution curves would have an opposite consequence (FIGURE 18). Local regional differences in the APD restitution curves (839) may also favor arrhythmogenesis (840). The ion channel background of cycle length-dependent APD changes including APD restitution is attributed to the incomplete recovery and/or deactivation of different inward (INa, ICa,L) or outward (Ito, IKr, IKs, ICl) currents based on the gating behavior of these channels (841, 842). In addition, intracellular ion concentration changes for Ca2+ and Na+ rapidly or slowly would activate electrogenic NCX or Na+-K+ pumps. Also, frequency changes can result in significant alterations in extracellular K+ concentration in the extracellular clefts (843), causing changes not only in depolarizing but also in repolarizing transmembrane ionic currents. Detailed discussion of these mechanisms is beyond the scope of this review.
Repolarization alternans (FIGURE 19) at the cellular level manifests as oscillation of long and short APD at rapid heart rates, typically with concurrent oscillations in calcium transient amplitude (712, 844). It has been demonstrated that alternans can precede the formation of arrhythmia in the heart (845, 846). Multiple studies and reviews explain the mechanisms of arrhythmia induction following alternans, typically linked to increased dispersion of repolarization (712, 847, 848). The spatial pattern of alternans across cardiac tissue is typically “concordant” at submaximal heart rates, i.e., the APD is either simultaneously shortened or prolonged in all sites (FIGURE 19) (712). However, further increase in the pacing rate can elicit so-called discordant APD alternans (FIGURE 19), when APDs at more distant regions can alternate with opposing phases (712), substantially increasing dispersion of repolarization and thereby the substrate for arrhythmias (849).
The fact that repolarization alternans typically occurs together with underlying oscillation of calcium transient amplitude (850, 851) poses the question of which of these two drives the other.
The first hypothesis suggests that the steep slope of APD restitution is the alternans driver (852), and this mechanism of arrhythmia is replicated by the ten Tusscher–Noble–Noble–Panfilov model of human ventricular myocyte (732, 733). As mentioned above, the ion channel background of APD alternans is attributed to the incomplete recovery and/or deactivation of different inward (INa, ICa) or outward (Ito, IKr, IKs, ICl) currents based on the gating behavior of these channels (841, 842). In addition, intracellular ion concentration changes for Ca2+ and Na+ rapidly or slowly would activate electrogenic NCX or Na+-K+ pumps. Also, frequency changes can result in significant alterations in extracellular K+ concentration in the extracellular clefts (843), causing changes not only in depolarizing but also in repolarizing transmembrane ionic currents.
On the other hand, other studies suggested that oscillations in the calcium transient amplitude are the primary alternans driver (851, 853). Such oscillation of calcium transient can be subsequently translated into APD alternans by NCX and other calcium-sensitive currents. Calcium-driven alternans was first suggested to originate from a Ca2+ release-reuptake mismatch due to the steep dependence of Ca2+ release on SR loading (851, 854). While in good agreement with a majority of experimental data, the experiments by Picht et al. (855) suggested that refractoriness of the ryanodine receptor rather than release-reuptake mismatch may underlie at least some of the observed alternans. Such a mechanism is also supported by certain computer models and may be either due to the refractoriness of the channel or due to changes in calsequestrin conformations with subsequent RyR block from within the SR (761). In a recent study utilizing iterated maps as well as a spatially detailed myocyte model, Qu et al. (856) showed that both mechanisms may act synergistically to promote alternans. A third explanation of calcium-driven alternans is the alternans driven by sarcoplasmic reticulum calcium cycling refractoriness (SRCCR) (857, 858). SRCCR alternans arises from a combination of steep load-release relationship (similar to release-reuptake mismatch hypothesis) and refractoriness of the SR release (similar to RyR refractoriness hypothesis). However, the latter does not result from an intrinsic RyR refractoriness but is a result of a limited rate of refilling of releasable calcium in the junctional SR. This mechanism underlies alternans in the Rudy-family models (728, 736, 857).
Alternans typically occurs only at rapid heart rates; however, data from human hearts show that in some cases alternans manifesting at rapid heart rate may cease with a further increase in pacing frequency (“eye-type alternans”) (859). The eye-type pattern was replicated with a populations-of-models approach (859), with the mechanism of the eye closure at rapid pacing being linked to flattening of the SR load-release relationship in certain conditions (858). Using a spatially distributed model of calcium handling, Qu et al. (856) also observed eye-type alternans in simulations where sarcoplasmic reticulum Ca2+-ATPase (SERCA) pumps were downregulated.
One important question pertaining to the spatial pattern of alternans in tissue is what determines whether alternans manifests in the relatively benign spatially concordant pattern or the highly proarrhythmic discordant one. Pastore et al. (849) observed in their experiments that nodal lines (lines in tissue separating areas of opposing alternans phase) were associated with structural abnormalities. However, discordant alternans may arise even in tissue with no obvious structural abnormality. Computer models provide two explanations of this phenomenon. Qu et al. (860) have shown that discordant alternans may emerge as a result of conduction velocity restitution. This explanation is characterized by the radial pattern of nodal lines with regard to pacing site. The second explanation by Sato et al. (762) relies on tissue synchronization of discordant alternans arising at the subcellular level. This explanation does not rely on a particular pattern of nodal lines and can explain experimental observations in Ref. 847.
Spontaneous episodes of ventricular tachyarrhythmia in patients are often preceded by a short-long-short sequence of cardiac cycles or irregular beat-to-beat variation of the QT intervals. This temporal instability of repolarization measured as short-term APD or QT variability similarly to spatial repolarization inhomogeneity is considered as an important marker for proarrhythmia (250, 861) (FIGURE 20). The mechanisms of short-term APD variability are not fully understood; however, they may relate to the stochastic behavior of the ion channels underlying the action potentials or the fluctuation of the intracellular Ca2+ movements and its electrophysiological consequences (822).