Modernized Classification of Cardiac Antiarrhythmic Drugs

Background: Among his major cardiac electrophysiological contributions, Miles Vaughan Williams (1918–2016) provided a classification of antiarrhythmic drugs that remains central to their clinical use. Methods: We survey implications of subsequent discoveries concerning sarcolemmal, sarcoplasmic reticular, and cytosolic biomolecules, developing an expanded but pragmatic classification that encompasses approved and potential antiarrhythmic drugs on this centenary of his birth. Results: We first consider the range of pharmacological targets, tracking these through to cellular electrophysiological effects. We retain the original Vaughan Williams Classes I through IV but subcategorize these divisions in light of more recent developments, including the existence of Na+ current components (for Class I), advances in autonomic (often G protein–mediated) signaling (for Class II), K+ channel subspecies (for Class III), and novel molecular targets related to Ca2+ homeostasis (for Class IV). We introduce new classes based on additional targets, including channels involved in automaticity, mechanically sensitive ion channels, connexins controlling electrotonic cell coupling, and molecules underlying longer-term signaling processes affecting structural remodeling. Inclusion of this widened range of targets and their physiological sequelae provides a framework for a modernized classification of established antiarrhythmic drugs based on their pharmacological targets. The revised classification allows for the existence of multiple drug targets/actions and for adverse, sometimes actually proarrhythmic, effects. The new scheme also aids classification of novel drugs under investigation. Conclusions: We emerge with a modernized classification preserving the simplicity of the original Vaughan Williams framework while aiding our understanding and clinical management of cardiac arrhythmic events and facilitating future developments in this area.


APPROACHES TO DEVELOPMENTS OF NEW DRUG CLASSIFICATION SCHEMES: THE SICILIAN GAMBIT
A review article published simultaneously in European Heart Journal and Circulation in 1991 represents an important step in the integration of these developments into guidelines for antiarrhythmic drug therapy. 6 The meeting in Taormina, Sicily, sought to furnish opening moves toward new classifications of antiarrhythmic drug therapy, akin to the Queen's Gambit representing a particularly aggressive option in chess, and this new approach was called the Sicilian Gambit. This more complete and flexible framework adopted a pathophysiological foundation identifying vulnerable parameters reflecting electrophysiological properties or events with pharmacological modifications that would terminate or suppress the arrhythmia with minimal undesirable cardiac effects. [7][8][9] It correlated information on molecular targets, cellular mechanisms, functional targets, and clinical arrhythmias for individual drugs with similarities and differences in their effects, accommodating their multiple actions. Although not then seeking a completed formal classification system, it furnished an accurate and comprehensive updated analysis of antiarrhythmic drugs. Although this analysis increased our understanding of drug action, the revised approach has not won widespread acceptance by clinicians and educators, possibly owing to its inevitable complexity. The Sicilian Gambit requires detailed knowledge of cellular and molecular targets of drugs under consideration. This may have made it intimidating or impractical for regular clinical use.

MODERNIZED SCHEME BASED ON THE VAUGHAN WILLIAMS APPROACH
The Vaughan Williams scheme, for all its limitations in light of subsequent developments in the cardiac electrophysiological field, thus remains the most useful, clinically and pedagogically popular approach to categorizing antiarrhythmic drugs. Table 1 summarizes a pragmatic development and expansion of that original classification encompassing principal actions of both current and potential antiarrhythmic agents, retaining the original Classes I through IV as its central core (Table I in the online-only Data Supplement). We thereby address interests and requirements of current workers in the field, mainly citing major reviews rather than original research articles, emphasizing broad principles and generalizations. We first identify major pharmacological targets, whether specific membrane ion channels, transporters, cytosolic biomolecules, or regulators ( Figure 1A) strategic to cardiac electrophysiological activity ( Figure 1B). Most therapeutic agents either block

Clinical Perspective
What Is New? • We develop a modernized comprehensive classification of both established and potential antiarrhythmic drugs that preserves the basic simplicity of the widely accepted classic Vaughan Williams framework. • This incorporates advances in our understanding made over the past half-century, covering all the major currently known classes of antiarrhythmic mechanisms What Are the Clinical Implications?
• It will provide a valuable guide to our basic understanding of the principal and subsidiary categories of antiarrhythmic and proarrhythmic drug actions in terms of their electrophysiological actions on specific currently known and potential targets bearing on cardiac excitation. • It will facilitate therapeutic decisions in current clinical practice and aid in the development of future novel antiarrhythmic drugs.    ) or open specific ion channels or, in the case of particular signaling molecules and receptors, activate or inhibit the relevant pathway. We then summarize the corresponding principal electrophysiological effects of target modification, including actions progressively investigated at the level of single cells, particular cardiac regions, or the entire heart. [7][8][9] These are illustrated by clinically used drugs, their clinical indications, and likely therapeutic mechanisms of action, acknowledging their additional, often multiple, actions (Table 2), selecting these from the wide range of clinically approved available agents (Table II in the online-only Data Supplement) and, in some cases, the numerous investigational agents under development (Table III in the online-only Data Supplement). Our approach retains but modifies Vaughan Williams Class I, adding a Class Id to include actions on recently reported late Na + current (I NaL ) components, recognizing their importance in long-QT syndrome (LQTS) type 3 (LQTS3). Class II conserves the β-adrenergic inhibitors but now captures subsequent advances in our understanding of autonomic, often G protein-mediated, signaling. Class III is expanded to take into account the large number of subsequently discovered K + channel species determining APD and subsequent refractoriness. Class IV now encompasses recently demonstrated and characterized molecu- Reduction of structural and electrophysiological remodeling changes that compromise AP conduction and increase reentrant tendency AP indicates action potential; APD, action potential duration; AVN, atrioventricular node; CaMKII, calcium/calmodulin kinase II; DAD, delayed afterdepolarization; EAD, early afterdepolarization; ERP, effective refractory period; HCN, hyperpolarization-activated cyclic nucleotide-gated; RyR2, ryanodine receptor 2; SAN, sino-atrial node; SQTS, short-QT syndrome; and SR, sarcoplasmic reticulum.   Table 1. B through E, Activation and inactivation of ion channels, currents, underlying proteins, and encoding genes and their contributions to (B) inward depolarizing and (C) outward repolarizing currents bringing about cardiac action potentials (APs). Ventricular (D) and atrial (E) APs comprise rapid depolarizing (phase 0), early repolarizing (phase 1), brief (atrial) or prolonged (ventricular) phase 2 plateaus (phase 2), phase 3 repolarization, and phase 4 electric diastole. In these, inward Na + or Ca 2+ currents drive phase 0 depolarization and Ca 2+ current maintains the phase 2 plateau (B), and a range of outward K + currents (C) drive phase 1 and phase 3 repolarization. Phase 4 resting potential restoration is accompanied by a refractory period required for Na + channel recovery. The resulting wave of electric activity and refractoriness is propagated through successive sino-atrial node, atrial, atrioventricular, Purkinje, and endocardial and epicardial ventricular cardiomyocytes. CaMKII indicates calcium/calmodulin kinase II; Cx, connexin; G i , inhibitory G protein; G s , stimulatory G-protein; HCN, hyperpolarization-activated cyclic nucleotide-gated channel; Nav1. 5 ating electrotonic coupling between cells (new Class VI). Finally, a range of signaling processes exert longer-term effects on arrhythmic tendency through modifying structural remodeling (Class VII). These classification activities are now discussed briefly in turn. These diverse drug actions converge on a defined set of distinct cellular and tissue electrophysiological end effects bearing on the respective cardiac properties of automaticity, AP generation, and AP conduction ( Table 1, right column). 7 Changes in the automaticity responsible for spontaneous, rhythmic cardiac activity can arise from abnormalities in the repetitive SAN activity, depending on its pacemaker currents. It can also arise from subsidiary pacemaker formation in specialized conducting AVN or Purkinje fibers and from normally nonautomatic atrial and ventricular cardiomyocytes when they are depolarized by some pathological processes. In the case of triggered activity that generates ectopic APs, early afterdepolarization phenomena occur during late phase 2 or early phase 3 of often prolonged APs. The latter are particularly associated with prolonged clinical QT intervals observed under conditions of increased inward late Na + , L-type Ca 2+ , decreased repolarizing outward K + currents, or increased depolarizing Na + /Ca 2+ exchange current arising from spontaneous sarcoplasmic reticular (SR) Ca 2+ release ( Figure 1B). In contrast, delayed afterdepolarizations after full AP repolarization are associated with situations of intracellular Ca 2+ overload, resulting in elevated SR Ca 2+ or increased cytosolic Ca 2+ sensitivity in the cardiac ryanodine receptor (RyR2). Either situation predisposes to spontaneous SR Ca 2+ release. This increases cytosolic [Ca 2+ ], which in turn transiently increases inward depolarizing Na + /Ca 2+ exchange current.
These triggering events may produce persistent arrhythmia whether through their perpetuating further such events or through arrhythmic substrate facilitating reentry of excitation from active into recovered myocardial regions. This takes place in the presence of heterogeneities that generate obstacles to AP conduction, around which the AP circulates with slowed conduction velocities that may reflect altered ion channel or myocardial tissue electric properties. The result is a formation of multiple, heterogeneous pathways of impulse propagation between anatomically or functionally defined points in the heart. 62 Whereas an anatomic  Class reentry of excitation takes place around a central inexcitable anatomic obstacle, a functional reentry of excitation involves a functional central obstacle. Reentrant excitation is also facilitated by abnormalities leading to heterogeneities in AP recovery arising from relative changes in ERP and APD, whether early (phase 2) [63][64][65] or late in the time course of AP repolarization. 19,23,66,67

NEW CLASS 0 OF DRUGS ACTING ON SINO-ATRIAL AUTOMATICITY
Detailed characterizations of the properties of SAN cells postdated the original Vaughan Williams classification. 10 SAN cells are exceptional in showing automaticity under normal physiological conditions, with contributions from a "membrane clock" giving rise to a spontaneous diastolic depolarization described as the pacemaker potential. This is driven by a net inward current, to which the most important contribution may be the "funny current" (I f ) carried by hyperpolarization-activated cyclic nucleotide-gated channels, particularly during the initial phase of the diastolic depolarization. 12,14,68 The only currently clinically adopted Class 0 agent, ivabradine, is used to reduce heart rates in situations of inappropriate sinus tachycardia 69,70 or when sinus tachycardia accompanies cardiac failure. 71 It likely acts through hyperpolarizationactivated cyclic nucleotide-gated channel block, with possible additional effects on intracellular Ca 2 ⁺ cycling. 72 Future investigations may explore the extent to which diastolic depolarization is further augmented by deactivation of outward delayed rectifier K + current and activation of inward currents, including Na + -dependent background current (I bNa ), T-and L-type Ca 2+ currents (I CaL and I CaT ), 73 and possibly sustained inward current (I st ). Inward voltage-dependent Na + current (I Na ) has also been recorded from SAN pacemaker cells, although it may be inactivated at the relatively positive potentials during the pacemaker potential in the SAN. 13,14 In addition, intracellular signaling involving SR Ca 2+ stores, cellular cAMP levels, and consequent phosphorylation of their signaling proteins has recently been implicated in a "Ca 2+ clock" in which spontaneous RyR2-mediated Ca 2+ release enhances electrogenic Na + /Ca 2+ exchanger activity during both SAN 14,74,75 and Purkinje cell diastolic depolarization. 76

EXTENSION OF VAUGHAN WILLIAMS CLASS I
Our revised classification system retains the 3 original Class I subcategories listing cardiac Na + channel (Na v 1.5) blockers. 16,30 However, it incorporates recent biophysical findings bearing on gating transitions regulated by voltage sensing components of Nav1.5 (Table 1). 77 Nav1.5 is preferentially expressed in atrial, Purkinje conducting, and ventricular as opposed to SAN and AVN cardiomyo-cytes. AP initiation then involves regenerative transitions from the resting state of Nav1.5 to its active state that permits the inward Na + current (I Na ), responsible for phase 0 rapid depolarization (Figure 2). The depolarization also causes the subsequent transition of Nav1.5 into an inactivated state, resulting in channel refractoriness. Channel recovery from the inactivated to the resting state then requires membrane repolarization and takes place over a finite time course. 17,22 Class Ia drugs subsequently proved to show concomitant effects on other, particularly K + , channel species, with potential consequences for Class III actions related to late depolarizing events. 78 Nevertheless, we retain their original Class Ia subclassification as Na + channel blockers, with all drugs in Classes Ia through 1c reducing AP maximum upstroke rates (dV/dt) max and AP conduction in atria, ventricular, and conducting tissue despite different effects on APD (Figure 2).
Class Ia drugs preferentially bind to the open state of Nav1.5 with dissociation time constants (τ) of ≈1 to 10 seconds. They thus reduce AP conduction velocity and increase ERP. Concomitant K + channel block by Class Ia drugs also increases APD. Together, these properties reduce reentrant tendency. In contrast, Class Ib drugs bind preferentially to the Nav1.5 inactivated state from which they dissociate relatively rapidly with a τ of ≈0.1 to 1.0 second. This minimizes perturbations of processes in the remaining cardiac cycle and explains the effectiveness of Class 1b drugs in preventing arrhythmias, particularly in ventricular tissue, where Nav1.5 channels remain inactive for the longest duration. Class Ib drugs result in shortening of both APD and ERP in normal ventricular muscle and Purkinje cells 16 but cause prolongation of ERP and consequently prolongation of postrepolarization refractoriness in ischemic, partially depolarized, cells. 79 Class Ic drugs similarly bind to the inactivated Nav1.5, from which, however, they dissociate more slowly, over τ >10 seconds. Use-dependent channel block in Classes Ia through 1c arises from accumulation of blocked channels during repetitive stimulation at high frequencies and accordingly occurs to extents in the sequence Class Ic>Class Ia>Class Ib. This results in a generalized reduction in cardiac excitability with nonspecific and widespread effects. These include slowed AP conduction with increased APD at high heart rates and possible reductions in cardiac automaticity. 16 These differing dissociation rates shown by Class Ia, Ib, and Ic agents also result in contrasting effects on AP conduction reflected in the associated normal and prolonged QRS durations, at least under conditions of normal cardiac rhythm. The different properties of Class I drug subgroups thus result in differing clinical effects, varying with the particular electrophysiological conditions underlying the targeted arrhythmias. The relatively slow dissociation of the Class Ic agent flecainide, from its binding to the inactivated state of Nav1.5, compromises AP initiation and conduction. Flecainide is thus antiarrhythmic, with the compromised AP recovery, gain of Na + channel function, and increased I NaL in both clinical LQTS3 and genetically modified murine Scn5a +/∆kpq hearts experimentally modeling this condition. 23 In contrast, flecainaide was clinically proarrhythmic under conditions of compromised postinfarct AP generation and propagation (Table 3) 80 and the Brugada syndrome and in murine Scn5a +/− models that replicate its associated loss of Nav1.5 function and age-dependent fibrotic changes. [18][19][20][21] This also contrasts with the respective antiarrhythmic actions of the more rapidly dissociating Class Ia and Class Ib agents quinidine and lidocaine in situations of compromised AP generation and propagation. Quinidine is additionally proarrhythmic under conditions of prolonged AP recovery, at least partially reflecting its additional I K -blocking effects. The different Class I actions also influence their clinical indications for arrhythmias affecting different regions of the heart. Finally, because atrial Nav1.5 channels remain open for longer than in the ventricles, Class Ia (exemplified in Table 1 by quinidine, ajmaline, and disopyramide) and Ic (exemplified A, Initial depolarization activates the Nav1.5 voltage sensor, in turn causing a transition from its resting, closed, to its open state, permitting extracellular Na + influx through the selectivity filter of the channel. The resulting regenerative depolarization results in slower transitions into an inactivated state, causing channel closure, from which recovery to the resting state requires membrane potential repolarization. The different Class Ia through Id drugs act at different stages in this reaction cycle and on differing early, I Na , and late, I NaL , Na + current components. This results in (B) differential actions on action potential (AP) conduction, triggering, and duration, with proarrhythmic or antiarrhythmic effects, depending on the background clinical conditions. BrS indicates Brugada syndrome; DAD, delayed afterdepolarization; EAD, early afterdepolarization; and LQTS, long-QT syndrome.

STATE OF THE ART
here by propafenone and flecainide) drugs are useful in preventing supraventricular arrhythmias. 30 Finally, actions of drugs such as ranolazine, GS-458967, and F15845 in the new Class Id differ sharply from those in Classes Ia through 1c. They inhibit the relatively small but persistent late Na + current (I NaL ) that follows the principal rapidly inactivating I Na decay and influences AP shape and duration. This increases in ac-quired or congenital proarrhythmic conditions, including hypoxia, heart failure, and LQTS3. These drugs thus shorten AP recovery and increase refractoriness and repolarization reserve. Both clinical and experimental reports suggest that they have potential antiarrhythmic effects in I NaL -related arrhythmia. 32,81,82 Class Id effects may also contribute to multiple drug actions. This effect is found with mexiletine, originally placed in Class

STATE OF THE ART
Ib, and makes it useful in management of not only LQTS3 78,83,84 but also Timothy syndrome, associated with L-type Ca 2+ channel abnormality. 85

EXTENSION OF VAUGHAN WILLIAMS CLASS II
We similarly retain the Vaughan Williams Class II but extend its coverage beyond an updated range of sympathetic β-adrenergic effects to further include parasympathetic targets. 5 This thereby provides more complete coverage of autonomic effects as a whole, including actions through cell surface membrane guanine nucleotide-binding protein (G-protein)-coupled receptors. First, Vaughan Williams would have been aware of the end effects but not the detailed mechanisms of β-adrenergic receptor activation through increased cytosolic [cAMP] i after successive G s -protein and adenylate cyclase activation. The increased [cAMP] i activates protein kinase A, which phosphorylates a wide range of ion channels, including Nav1.5, the K + channel species Kv11.1, Kv7.1 (mediating the rapid and slow K + currents, I Kr and I Ks respectively), Cav1.2, and Cav1.3 (mediating L-type Ca 2+ currents), and RyR2. cAMP also exerts a direct influence on hyperpolarization-activated cyclic nucleotide-gated channel activity and consequently on the pacemaking I f . Finally, exchange proteins directly activated by cAMP have been reported to trigger RyR2-mediated Ca 2+ release. 19 These actions together produce multiple inotropic, chronotropic, and lusitropic effects on cardiac function. 33,53 Table 1 places the clinically used nonselective and selective β 1 -adrenergic receptor inhibitors carvedilol and propranolol (nonselective) and atenolol (selective), indicated in a wide range of tachyarrhythmias, in Class IIa. These often act through inhibiting Ca 2+ entry and SR Ca 2+ release and their consequent proarrhythmic early afterdepolarization-or delayed afterdepolarization-induced triggered activity. The classification also places nonselective β-adrenergic receptor activators, exemplified by isoproterenol, in Class IIb. The latter contrastingly activate Ca 2+ entry and SR Ca 2+ release, potentially accentuating proarrhythmic early afterdepolarization-induced triggered activity. However, their chronotropic effects usefully accelerate rates of ventricular escape rhythm in the management of complete atrioventricular block before pacemaker implantation. 34 Such acceleration of depressed heart rates and relief of prolonged postextrasystolic pauses may additionally suppress bradycardia-dependent early afterdepolarizations. Isoproterenol thus exerts antiarrhythmic effects in bradycardia-dependent, drug-or atrioventricular block-related, and possibly congenital LQTS type 2-and LQTS3-related torsades de pointes but proarrhythmic effects in adrenergic dependent or LQTS type 1-related torsades de pointes. 86 Second, of the large range of further G-protein subtypes, G i proteins mediate parasympathetic cholinergic muscarinic (M 2 ) or adenosine (A 1 ) receptor activation. Their activation and inhibition reduce and increase membrane excitation, respectively, particularly under conditions of preexisting adenylyl cyclase activity, affecting chronotropic and conduction function. Table 1 introduces M 2 inhibitors in the new Class IIc, exemplified by atropine, indicated for relieving sinus bradycardia and supra-His (Table 1), although not degenerative, atrioventricular block at or below the His bundle level (Table  3). Table 1 also illustrates drugs inhibiting G i exemplified by carbachol and adenosine in new Classes IId and IIe, respectively, while bearing in mind the brief period of intravenous adenosine action and its tendency to produce atrial fibrillation. 87 It also cites an action of aminophylline in adenosine receptor block, useful to treat bradycardia associated with sinus node dysfunction. 88 The latter actions take place in the SAN, AVN, or atrial myocardium even in the absence of sympathetic stimulation but in ventricular tissue take place only after adrenergic activation. Thus, drugs activating G i are normally effective in SAN, atria, or AVN tachycardias but are effective only in adrenergically stimulated Purkinje or ventricular cells. G i activation opens inward rectifying I KACh or I KAdo channels mediated by βγ subunits of the G protein, particularly in supraventricular tissue, through actions on their GIRK1 and GIRK4 components. 35,89,90 G i activation also inhibits adenylyl cyclase, which reduces [cAMP] I ; therefore cAMP-associated increases in I CaL and I f . G i activation may also upregulate protein phosphatase 2-mediated dephosphorylation at protein kinase A phosphorylation sites on inwardly rectifying K + channels, L-type Ca 2+ channels, RyR2s, phospholamban, troponin subunit cardiac troponin I, and cardiac-type myosin-binding protein C. 36,37 Finally, ≈150 of the large number of additional potential G protein-coupled receptors remain orphan receptors that might offer potential therapeutic targets.

EXTENSION OF VAUGHAN WILLIAMS CLASS III
Much progress has also followed the original Vaughan Williams classification 42 resulting from increased knowledge of K + channel subtypes. More is also known about the α and auxiliary β subunits, selective localization of K + channels in particular cardiac regions, and roles of these channels in AP recovery and membrane potential stabilization ( Figure 1B and Table 1). 35,41 After phase 0 depolarization, complex components of transient inward current (I to ) contribute to early rapid phase 1 AP repolarization. These include rapidly activating and inactivating Kv4.3and Kv4.2-mediated fast inactivating I to,f and Kv1.4-mediated and slowly inactivating I to,s , which become activated at potentials of >−30 mV. Atrial myocytes show particu-

STATE OF THE ART
larly prominent I to , and an atrium-specific Kv1.5 (KCNA5) mediates ultrarapid I Kur . In addition, there is a 6-fold greater expression of GIRK1 and GIRK4 proteins that mediate I KACh . These multiple K channel contributions together result in the shorter atrial compared with ventricular APs. Kv11.1 (HERG or KCNH2) mediating I Kr rapidly activates with phase 0 AP depolarization but then rapidly inactivates over AP phases 0 to 2. The onset of phase 3 repolarization reverses this inactivation, reopening the channel leading to outward phase 3 and early phase 4 currents terminating the AP plateau. The channel responsible for I Kr is more greatly expressed in human ventricular than atrial cardiomyocytes and in left than right canine atrial cardiomyocytes. In contrast, Kv7.1 (KCNQ1) mediating I Ks requires depolarization to a more positive potential for activation, which then takes place relatively slowly. I Ks increases over phase 2 to become a major phase 3 K + conductance that barely inactivates. It is expressed uniformly in canine atria but at a higher density in epicardial and endocardial cells than M cells and in right ventricular than left ventricular M cells with possible implications for proarrhythmic LQTS. 40,43 Inward rectifying I K1 , mediated by Kir2.1, Kir2.2, and Kir2.3 (KCNJ2, KCNJ12, and KCNJ4), reflects reductions in K + conductance at membrane potentials more depolarized than ≈−20 mV, as occurs in phases 0 to 2 of the AP. This reduces the net depolarizing inward currents required to maintain the AP plateau phase. In contrast, the K + conductance becomes greater when the AP recovers to membrane potentials more hyperpolarized than ≈−40 mV. This results in the increased K + outward current, which in turn facilitates late phase 3 AP repolarization. This channel also stabilizes phase 4 diastolic resting potentials. It occurs at a higher density in human ventricular than atrial myocytes.
Finally, the metabolically dependent I KATP is normally small but is activated by reduced intracellular ATP levels when it results in triangulation of AP waveforms. 44 The K 2P 2.1 (KCNK2, expressing K 2P currents) and the ATP-sensitive Kir6.2 (KCNJ11) mediating I KATP show little time or voltage dependence but contribute background currents regulating resting membrane potentials and cell excitability.
The more extensive group of clinical Class III agents now includes wider ranges of voltage-dependent K + channel blockers (Class IIIa), including nonselective (ambasilide, amiodarone) and selective (HERG; I Kr ; dofetilide, ibutilide, sotalol) Kv11.1, Kv1.5 (I Kur ; vernakalant), and K V 1.4 and K V 4.2, (I to1 : tedisamil) blockers, as well as important drugs opening metabolically dependent (Kir6.2: I KATP : nicorandil, pinacidil; Class IIIb) and investigational drugs blocking transmitter-dependent (GIRK1 and GIRK4: I KACh ; BMS 914392; Class IIIc) K + channels. These may act directly on the channels concerned or involve further indirect effects such as those exemplified by the inhibitory actions of dofetilide on phosphoinositide 3-kinase signaling, in turn inhibiting I Kr and increasing I NaL . 91,92 In addition, a number of agents with multiple actions are included here (Table 2).
Amiodarone and dronedarone show diverse actions even at therapeutic concentrations and complex therapeutic and toxicity profiles but find widespread use in managing atrial fibrillation (Table 2). Finally, the significant K + channel and therefore Class III actions demonstrated for the original Class Ia agents have been recognized. Thus, although quinidine was originally placed in Class I, its clinical antiarrhythmic effects in Brugada syndrome probably include inhibition of I to . 63 It has been suggested that this involves reductions in the transmural dispersions of ventricular repolarization that arise from the greater epicardial than endocardial expression of I to , which results in the normally shorter epicardial relative to endocardial APDs. 23,64,66 Further examples of agents with such multiple actions are listed in Table 2. Finally, K + itself influences K + channel permeabilities with important effects on resting membrane potential stability and APD. 23

EXTENSION OF VAUGHAN WILLIAMS CLASS IV
Much recent physiological progress has broadened the range of drugs included as Vaughan Williams Class IV drugs, originally defined as drugs blocking Ca 2+ entry through specific Ca 2+ channels. Here, we have extended Class IV to include drugs with a variety of actions that can be described as Ca 2+ handling modulators. The L-type voltage-gated Ca 2+ current (I CaL ) emerges with roles in both atrial and ventricular cardiomyocyte function and in AP conduction in the AVN. It thus both contributes to the ventricular and atrial AP plateau phases and initiates excitation-contraction coupling. I CaL brings about an initial cytosolic [Ca 2+ ] elevation that triggers the Ca 2+ -induced release of SR Ca 2+ by intracellular RyR2 Ca 2+ release channels. The resulting further elevations of cytosolic [Ca 2+ ] in turn drive contractile activation. An inositol trisphosphate (IP 3 )-triggered Ca 2+ release that has been implicated in atrial arrhythmia may also exist.
After AP recovery, cytosolic [Ca 2+ ] is returned to resting levels by Ca 2+ transport from cytosol to SR lumen by phospholamban-regulated SR Ca 2+ -ATPase 33 and from cytosol to extracellular space by plasma membrane Ca 2+ -ATPase and by surface membrane ion exchangers, particularly sarcolemmal Na + /Ca 2+ exchange. 53 Of these, Na + /Ca 2+ exchange involves electrogenic entry of 3 Na + for each Ca 2+ extruded. Depending on the membrane potential and submembrane [Ca 2+ ] that determine the driving forces on Na + and Ca 2+ fluxes, this can exert depolarizing effects.
Activity in a significant proportion of these membrane and cytosolic signaling and Ca 2+ transport molecules is altered by kinase-mediated phosphorylation and phosphatase-mediated dephosphorylation. 55,56 These opposing processes are in turn modified by cytosolic, often Ca 2+ -sensing, signaling molecules also offering potential pharmacological targets. Besides protein kinases A and C, these include calmodulin and calcium/ calmodulin kinase II. 44,50,56,57 Modifications in 1 or more of these processes in turn altering cytosolic [Ca 2+ ] have been implicated in both atrial and ventricular clinical arrhythmias. 51,52 In particular, Na + /Ca 2+ exchange exerts electrogenic effects that can increase to become potentially proarrhythmic with cellular Ca 2+ overload. 5,48 The central importance of Ca 2+ homeostasis to cardiac electrophysiological activity with extensive findings after the original Vaughan Williams classification accounts for a wide range of potential applications directed at clinical arrhythmia (Table 1). Besides nonselective (bepridil) and Cav1.2/Cav1.3 (I CaL )-selective (verapamil, diltiazem) Ca 2+ channel blockers (Class IVa), Mg 2+ , although not strictly falling within the category of a drug, also exerts Ca 2+ channel blocking and membrane stabilizing effects, with applications in treatment of torsades de pointes. In recent reports, the Class Ic agent flecainide and the Class IIa agent carvedilol show additional Class IVb actions in reducing RyR2-mediated SR Ca 2+ release. This proved potentially applicable in the management of catecholamine-sensitive polymorphic ventricular tachycardia, whether through reduced triggering activity or reversing associated proarrhythmic reductions in I Na . [93][94][95] Possible clinical applications of decreasing cardiac myosin heavy chain-or SR Ca 2+ reuptake-related ATPase activity (Class IVc) have prompted explorations of the investigational new drugs MYK-461 96 and istaroxime 97 in hypertrophic cardiomyopathy and cardiac failure, respectively. Possible applications will also likely emerge from drugs modifying Na + /Ca 2+ exchange (Class IVd) and phosphorylation of proteins involving Ca 2+ homeostasis, including calcium/calmodulin kinase II (Class IVe) 98 ( Table III in

NEW CLASS V OF DRUGS ACTING ON MECHANOSENSITIVE CHANNELS
Class V is introduced to include mechanosensitive channel blockers. These are selective for cation-selective and mechanosensitive ion channels, particularly transient receptor potential channels (TRPCs) such as TRPC3 or TRPC6. Multiple subclasses of TRPCs exist in the heart, although their functions are only now beginning to emerge. They potentially suppress abnormal ectopic or triggered activity in cardiac conditions such as cardiac hypertrophy and heart failure. 58 A TRPC subclass may regulate the cardiac hypertrophic response. Although TRPCs allow permeation by a range of different cations, their specific biological functions have generally been attributed to Ca 2+ influx, resulting in signaling within local domains, direct interactions with Ca 2+ -dependent regulatory proteins, or regulation of cardiac fibroblastic Ca 2+ signals in arrhythmic hypertrophic and fibrotic heart disease and cardiac failure. 58 Accordingly, inhibition of TRPC-mediat-ed Ca 2+ influx could potentially both exert direct antiarrhythmic effects and attenuate replacement fibrosis after cardiomyocyte death. Such an approach is being explored with a number of investigational drugs, including ACA [N-(p-amylcinnamoy)anthranilic acid], GSK2332255B, GSK2833503A, pyrazole-3, GsMTx4, and SKF 96365 ( Table III in

NEW CLASS VI OF DRUGS ACTING ON CONNEXIN-ASSOCIATED CHANNELS
AP conduction depends on intercellular local circuit current spread involving gap-junction conductances containing apposed connexin (Cx) hemichannels electrically connecting the intracellular spaces of adjacent cardiomyocytes. 62 This possible therapeutic direction is being investigated with both Cx-blocking and -opening agents, exemplified by carbenoxalone and the peptide analog rotigaptide (ZP-123), respectively, the latter in connection with potential treatments for atrial fibrillation ( Table III in the online-only Data Supplement). Of cardiac Cx isoforms, Cx40 occurs in atrial myocytes, AVN, and the Purkinje conduction system. Cx43 occurs in both atrial and ventricular myocytes and the distal conduction system. Cx45 occurs mainly in the SAN, AVN, and Purkinje conducting system. Blocking gap junction conductance or expression, depending on circumstances, can enhance or reduce arrhythmogenicity. Changes in gap junction function can accompany alterations in other AP conduction determinants such as fibrotic change or other remodeling processes in which these are accompanied by altered excitability. Plasticity reducing and lateralizing Cx43 expression occurs in both hypertrophic and dilated ventricular cardiomyopathies. 18,59

NEW CLASS VII OF DRUGS ACTING ON UPSTREAM MODULATORY TARGETS
The introduction of a Class VII results from the need to encompass tissue structure remodeling processes and their consequently longer-term changes that contrast with the primary preoccupation with the short-term effects of particular drugs on specific ion channels in the original Vaughan Williams classification. In addition, molecular mechanisms influencing longer-term changes upstream of the electrophysiological processes also constitute novel potential therapeutic targets. Fibrotic change is an important accompaniment to postinfarct healing, potentially leading to chronic scarrelated arrhythmogenesis, pressure overload, 99 and the development of atrial fibrillation. 47,60,61 It also accompanies some Na + channelopathies. 18 Experimental studies have demonstrated that renin-angiotensin-aldosterone inhibitors, omega-3 fatty acids, and statins prevent such electrophysiological and/or structural remodeling.

STATE OF THE ART
These drugs are already available for indications such as hypertension, coronary artery disease, and heart failure, which are some of the most frequent causes of atrial fibrillation. Angiotensin-converting enzyme inhibitors or angiotensin-receptor blockers may be useful in modifying the atrial substrate for primary or secondary prevention, reducing susceptibility to or progression of established atrial fibrillation in the presence of cardiac failure and hypertension. Statin therapy may be useful for the primary prevention of new-onset atrial fibrillation after coronary artery surgery. 25,60,61

RECAPITULATION
The revised classification of antiarrhythmic drugs presented here summarizes current views of their electrophysiological effects, which are categorized as principal (Table 1), subsidiary (Table 2), and proarrhythmic (Table 3). It represents a pragmatic development of the Vaughan Williams classification ( Table I in the onlineonly Data Supplement). The revised scheme is consistent with clinical actions of therapeutically established drugs (Table 1 and Table II in the online-only Data Supplement) and provides a classification framework for studies of new drugs under investigation (exemplified in Table III in the online-only Data Supplement).