The cardiac persistent sodium current: an appealing therapeutic target?

Abstract

The sodium current in the heart is not a single current with a mono-exponential decay but rather a mixture of currents with different kinetics. It is not clear whether these arise from distinct populations of channels, or from modulation of a single population. A very slowly inactivating component, [_INa(P)] I_Na(P) is usually about 1% of the size of the peak transient current [I_Na(T)], but is enhanced by hypoxia. It contributes to Na⁺ loading and cellular damage in ischaemia and re-perfusion, and perhaps to ischaemic arrhythmias. Class I antiarrhythmic agents such as flecainide, lidocaine and mexiletine generally block I_NA(P) more potently than block of I_Na(T) and have been used clinically to treat LQT3 syndrome, which arises because mutations in SCN5A produce defective inactivation of the cardiac sodium channel. The same approach may be useful in some pathological situations, such as ischaemic arrhythmias or diastolic dysfunction, and newer agents are being developed with this goal. For example, ranolazine blocks I_Na(P) about 10 times more potently than I_Na(T) and has shown promise in the treatment of angina. Alternatively, the combination of I_Na(P) block with K⁺ channel block may provide protection from the induction of Torsades de Pointe when these agents are used to treat atrial arrhythmias (eg Vernakalant). In all of these scenarios, an understanding of the role of I_Na(P) in cardiac pathophysiology, the mechanisms by which it may affect cardiac electrophysiology and the potential side effects of blocking I_Na(P) in the heart and elsewhere will become increasingly important.

Background

The existence of a persistent component of the cardiac sodium current was suggested almost 30 years ago. Tetrodotoxin (TTX) was shown to shorten action potential in Purkinje fibres at concentrations (3.3 × 10⁻⁸ M), which had little or no effect on upstroke (10⁻⁶ M) (Coraboeuf et al., 1979). A slow TTX-sensitive current was subsequently shown directly by voltage clamp (Gintant et al., 1984; Carmeliet, 1987a), and the channels responsible for several time components of I_Na inactivation shown by patch clamp (Patlak and Ortiz, 1985; Kiyosue and Arita, 1989). Similar observations have been made in rat ventricle cells (Kirsch and Brown, 1989), dog Purkinje fibres (Fozzard et al., 1987) and mouse (Bohle and Benndorf, 1995) and rabbit (Grant and Starmer, 1987) ventricle cells and human ventricle cells (Maltsev and Undrovinas, 2006). More recently, approaches such as voltage ramps (Clancy et al., 2003) or ‘synthetic' action potential clamp protocols (Magyar et al., 2004) and a ‘dynamic action potential clamp' technique (Berecki et al., 2006), have been developed which can directly demonstrate the magnitude of I_Na(P) or the channel activity during the action potential and the influence of channel kinetics on this.

The potential of this persistent sodium current (I_Na(P)) as a therapeutic target was initially highlighted by reports that mutations in Na⁺ channels that produced a slowing of inactivation were responsible for some serious clinical arrhythmias (Bennett et al., 1995; Marban, 2002), and by demonstrations that I_Na(P) was enhanced in hypoxia (Hammarstrom and Gage, 1998, 2002). Later work has shown that I_Na(P), is important in myocardial ischaemic damage (or, more accurately, ischaemia–reperfusion damage) and some types of arrhythmias. Recently, compounds which block this current have reached the later stages of clinical development: ranolazine (Ranexa; Cardiovascular Therapeutics) was approved by the US FDA for the treatment of chronic angina in January 2006 and Vernakalant (RSD1235-Cardiome) is undergoing phase III trials for the treatment of atrial fibrillation. Both block I_Na(P) to some extent, although both have other actions also, which makes ascribing their clinical effectiveness to block of I_Na(P) somewhat contentious.

Properties of the persistent sodium current

Biophysically, I_Na(P) has similar properties to I_Na(T) except that the inactivation component is missing or greatly slowed. Hence, at potentials more positive than resting potential the transient current inactivates but I_Na(P) does not. I_Na(P) deactivates very rapidly on repolarization (Saint et al., 1992) and has slightly different activation properties to I_Na(T) at the macroscopic level. In rat ventricular cells I_Na(P) has a V₅₀ for activation about 20 mV more negative than I_Na(T) (V₅₀ of –52 and −34 mV) (Saint et al., 1992). I_Na(P) in cardiac cells has the same ionic selectivity as I_Na(T) and the single channels have the same conductance, mean open time and reversal potential as the transient channel (Gintant et al., 1984). I_Na(P) is not a ‘window' current since it can be recorded at very positive membrane potentials (Saint et al., 1992). I_Na(P) should also not be confused with the background or ‘leak' sodium current, which is TTX insensitive and not voltage dependent (Spindler et al., 1998). I_Na(P) usually has a magnitude of about 1% of the transient component; in rat myocytes: Saint et al. (1992) reported peak I_Na(T) of 53 nA and I_Na(P) of about 300 pA at 140 ms after a step depolarization to –40 mV. In guinea-pig ventricular myocytes, Kiyosue and Arita (1989) reported a value of 12–50 pA for the sustained I_Na at −40 mV (26±14 pA, mean ±s.d., n=5). Although it is small, since it is non-inactivating the amount of charge (and sodium) that can be carried by it over the time course of an action potential can be of the same order as that carried by the transient current. Block of I_Na(P) can therefore considerably reduce the sodium load on the cell.

I_Na(P) is more sensitive to block by agents such as lidocaine and TTX than is I_Na(T) (Nilius et al., 1987; Josephson and Sperelakis, 1989; Ju et al., 1992; Saint et al., 1992). Channels which have been modified by the action of toxins to slow or disable inactivation, or which have mutations affecting inactivation, have similar enhanced susceptibility to block by TTX, lidocaine and related drugs such as flecainide (Liu et al., 2002) and mexiletine (Sicouri et al., 1997).

Mutations in SCN5A leading to LQT3

I_Na(P) is implicated in many genetic abnormalities leading to arrhythmias, such as some forms of long Q-T and Brugada syndrome (Rook et al., 1999; Baroudi and Chahine, 2000; Deschenes et al., 2000; Makita et al., 2000; Huang et al., 2006). A bewildering array of mutations in SCN5A (the cardiac sodium channel) have now been documented (for a review see Tan et al., 2003), most of which lead to a gain of function in the channel in LQT3 (Clancy and Kass, 2005) but a loss in function in Brugada syndrome (Brugada and Brugada, 1992). LQT3 mutations in SCN5A generally lead to an enhanced action potential shortening with increased rate, and this is the mechanism for the pause-induced Torsades seen with these mutations (Viswanathan and Rudy, 1999; Oginosawa et al., 2005; Fredj et al., 2006a). Block of I_Na(P) may be a useful therapeutic intervention in some LQT3 cases. For example, chronic low-dose flecainide significantly shortened the QTc interval in LQT-3 subjects with the DeltaKPQ mutation (Moss et al., 2005), mexiletine was used to treat a patient with SCN5A single-nucleotide substitution from arginine to glutamine at position 1623 (R1623Q) (Miura et al., 2003) and lidocaine has been used clinically to differentiate LOT3 from Herg mutations (Schwartz et al., 1995). However, the detailed pharmacology of LQT3 is beyond the scope of this review.

Enhanced I_Na(P) with toxins

There are a variety of toxins and other agents which can interact with the sodium channel and produce persistent activation (Wang and Wang, 2003). By and large the toxins affect inactivation of the channel, or shift the voltage dependence of activation and so produce enhanced persistent currents. Hence, many of the toxins have been used to mimic the effect of enhanced I_Na(P) seen in LQT3 (Sicouri et al., 1997). However, it should be borne in mind that modulation with toxins is not necessarily a perfect mimic of the modulation of I_Na that may occur in ischaemia or hypoxia, in the same way that mutations of the channel in LQT3 do not reproduce the details of enhanced I_Na(P).

Hypoxia and I_Na(P)

Perhaps, a closer mimic to the modulation of ‘native' I_Na(P) may be obtained by use of compounds which alter the redox state of the channel. DPI [(±)-4-(3-(4-(diphenylmethyl)-1-piperazinyl)-2-hydroxyperoxy)-1H-indole-2-carbonitrile: Sandoz] appears to switch the channel to a bursting mode (Nilius, 1987). Importantly, hypoxia enhances I_Na(P) (Ju et al., 1996b). Somewhat paradoxically, hydrogen peroxide also enhances I_Na(P) (Song et al., 2006) while 1 mM reduced glutathione can reverse the effects of hypoxia (Wang et al., 2007). (It has been suggested that the redox effects may be mediated by the action of a closely associated regulatory protein, which may explain the apparently paradoxical effects of redox agents; Hammarstrom and Gage, 1999).

Is the persistent current carried by a distinct channel type?

Apart from the mutations in SCN5A or modification of sodium channels by toxins, in which the underlying mechanisms producing I_Na(P) are clear, there are two possible ways in which a non-inactivating component of I_Na might arise: (1) there could be only one isoform of the channel, but a proportion of the channels could at any given time display different inactivation kinetics because of, for example, modification by phosphorylation, or the action of a regulatory subunit and (2) there could be different isoforms of the channel present, each with different inactivation kinetics.

The first of these possibilities is given credence by the observation that a persistent component of I_Na is present even when a single isoform of the Na⁺ channel is expressed in heterologous systems (Fearon and Brown, 2004). There have been reports of the sodium channel undergoing changes in gating mode, where ‘burst' mode or non-inactivating modes can arise spontaneously (Patlak and Ortiz, 1985; Bohle et al., 2002). Since the slow Na⁺ channels had the same conductance, reversal potential and TTX sensitivity as fast Na⁺ channels in the same cells, Patlak and Ortiz argued that the fast and the slow currents result from a single class of Na⁺ channels with two or more kinetic modes. Presumably, a small subpopulation of channels in a cell can be in these modes at any given time, giving rise to a persistent current at the macroscopic level. This model for the genesis of I_Na(P) has the appeal of simplicity, but seems, at first sight, difficult to reconcile with the apparent different sensitivity to blocking agents such as TTX and lidocaine of I_Na(P) and I_Na(T).

The obvious way to account for this different pharmacology is to propose that I_Na(P) is due to a distinct channel. The Na⁺ channel has many different isoforms; it has long been known that cardiac channels are substantially different from neuronal channels in their kinetics and sensitivity to TTX. The molecular basis of this diversity of sodium channels is now well understood (Leffler et al., 2005). The dominant form of the channel in the heart is Nav 1.5 or Nav 1.1, with associated β-1 or β-2 subunits. Of course, there is no reason why a given cell should not express a mixture of isoforms, and this could give rise to mixed Na⁺ currents. Indeed, as long ago as 1985 it was suggested that the slowly inactivating current (in sensorimotor neurones) was due to a different channel, based on block by QX-314 (Stafstrom et al., 1985). If I_Na(P) in cardiac cells does arise from a different isoform of the channel, this channel is apparently identical to the I_Na(T) channel in conductance, mean open time and selectivity (Ju et al., 1992), suggesting that the only difference is in kinetic properties and apparent affinity of some blockers.

Its important to realize, though, that this difference in pharmacology (at the macroscopic level) is not inconsistent with I_Na(T) and I_Na(P) arising from one population of channels. It has been suggested that the kinetics of the channel can interact with the use-dependence of blockers such as TTX or lidocaine to produce an apparent preferential block (Carmeliet, 1987b; Josephson and Sperelakis, 1989; Saint, 2006). A similar mechanism has been proposed for flecainide (Liu et al., 2002) and may hold for other similar types of blockers. Theoretical models of channel block by mexiletine and lidocaine predict an enhanced block of LQT3, or burst mode, channels compared to the transient channels, because of their different kinetics, not because of intrinsic differences in the channels themselves (Clancy et al., 2007).

Although I_Na(P) is small compared to I_Na(T) under most conditions, there are situations in which I_Na(P) is enhanced and which have physiological (or pathophysiological) implications; for example reactive oxygen species, such as H₂O₂, activate I_Na(P) (Meng and Nie, 2004; Ma et al., 2005; Song et al., 2006), as does NO (Hammarstrom and Gage, 1999; Ahern et al., 2000), some phospholipids (Undrovinas et al., 1992) and hypoxia (Ju et al., 1996b; Hammarstrom and Gage, 2002; Saint, 2006). While the mechanisms by which these operate is not fully understood, it's likely that they mimic some of the mechanisms that operate in vivo (for example in ischaemia). Its not known whether these interventions produce their effects by enhancing the activity of a susceptible subpopulation of distinct channels, or whether they are converting some I_Na(T) channels into I_Na(P) channels.

In summary, the jury is still out on whether the co-presence of I_Na(T) and I_Na(P) is a consequence of distinct channel isoforms being co-expressed, or merely due to a kinetically distinct subpopulation of an otherwise heterogeneous isoform. Either hypothesis is consistent with the data available at present.

The physiological role of the persistent sodium current

The physiological role of I_Na(P) in the heart is unclear. It certainly contributes to the action potential plateau, as demonstrated by the effect of low doses of TTX to shorten cardiac action potentials noted above, although this effect does not seem large enough to produce contractile changes under physiological conditions (Brette and Orchard, 2006). Some groups have reported a gradient of expression of I_Na(P) across the ventricular wall; in guinea-pig, mid-myocardial cells had a smaller I_Na(P) (0.23±0.02 pA pF⁻¹) than epi- or endo-cardial (0.40±0.04 and 0.36±0.03 pA pF⁻¹ respectively) and computer modelling predicted that this could lead to action potential heterogeneity (Sakmann et al., 2000). These results were not consistent with those of Main et al. (1998), who observed that 0.1 μmol l⁻¹ TTX shortens APD₉₀ in guinea-pigs by about 36% in subendocardial myocytes, 16% in mid-myocardial myocytes, and 20% in subepicardial myocytes. In myocytes isolated from different regions of the dog heart, late sodium current density at 0 mV was 47% greater in mid-myocardial cells and averaged −0.532±0.058 pA pF⁻¹ in endocardial, −0.463±0.068 pA pF⁻¹ in epicardial and −0.785±0.070 pA pF⁻¹ in mid-myocardial cells (Zygmunt et al., 2001). Although the gradient of I_Na is thought to be important in the pro-and anti-arrhythmic effect of class 1 agents, particularly in models of LQT3, the existence of a distinct population of mid-myocardial cells (‘M cells') has not been verified in other species, for example, pig (Rodriguez-Sinovas et al., 1997); rabbit (Idriss and Wolf, 2004). Their existence and possible physiological significance in human is unclear (Drouin et al., 1995; Antzelevitch et al., 1999; Taggart et al., 2003).

Therapeutic potential of I_Na(P)

As noted above, block of I_Na(P) may be a worthwhile therapeutic intervention in some LQT3 syndromes. A far more common clinical application may arise from the observation that I_Na(P) may be upregulated in cardiac hypertrophy and failure, and may contribute to the action potential prolongation in these conditions. In addition, enhanced I_Na(P) is almost certainly involved in reperfusion injury, some arrhythmias and possibly other disease states.

Protection in ischaemia

When myocardial tissue becomes ischaemic intracellular pH falls due to lactic acid production. Normally the increased [H⁺]i is extruded via the sodium–hydrogen exchanger, NHE-1. There is some evidence that NHE is inhibited during ischaemia (Xiao and Allen, 2000), which results in an uncontrolled increase in [H⁺]_i. Upon re-perfusion the inhibition of NHE is removed and consequently it acts to produce a large extrusion of H⁺ at the expense of allowing a large Na⁺ influx. The increased [Na]i reduces the electrochemical energy gradient available to the Na–Ca exchanger (NCX), leading to decreased Ca²⁺ extrusion from the cell or, in extreme cases, reversal of Na–Ca exchange; either situation leads to Ca²⁺ overload (the ‘coupled exchanger' hypothesis (Lazdunski et al., 1985)) (Figure 1).

Figure 1.

Diagram of coupled exchanger model. Ischaemia or hypoxia results in enhanced glycolysis, producing lactate and H⁺. The H⁺ ions are pumped out of the cell by NHE1 in exchange for sodium influx. Na efflux by the Na-KATPase is reduced because of low ATP. Additional Na⁺ influx occurs via I_Na(P). The Na⁺ load results in reduction or reversal of NCX, leading to calcium overload (Note that NCX is shown here in reverse mode. From Saint, 2006.).

In animal experiments, NHE blockers are effective at ameliorating re-perfusion damage (Hartmann and Decking, 1999), although clinical trials have been disappointing (Theroux et al., 2000; Zeymer et al., 2001). An important clue as to their failure may lie in the observation that in animal models block of NHE does not provide complete protection, and protection can equally and synergistically be provided by blocking voltage-dependent sodium channels, for example with TTX (Eigel and Hadley, 1999) (Figure 2).

Figure 2.

Fluorescence measurements of cytosolic [Na⁺] were made in isolated guinea-pig ventricular myocytes [loaded with SBFI-AM] exposed to anoxia beginning at the ‘initial' time point and continuing until 60 min post-rigor contracture. Either HOE-642 (10 μM) or TTX (30 μM) reduced the anoxic Na⁺ load by about 50% while a combination of both almost eliminated Na⁺ load. Each time point represents mean six s.e. of eight cells. ^*P=0.05 versus drug-free value at same time point. From Eigel and Hadley (1999).

Hence, influx of Na⁺ through I_Na(P) channels contributes at least part of the [Na⁺]_i load in ischaemia. This effect is likely to be linked to the fact that I_Na(P) is enhanced by hypoxia (Ju et al., 1996b; Hammarstrom and Gage, 2002; Fearon and Brown, 2004; Saint, 2006). Hence, agents which block I_Na(P) should be protective in ischaemia, and this seems to be true (Haigney et al., 1994; Belardinelli et al., 2006). Indeed, over 10 years ago, Le Grand et al. (1995) showed that low concentrations of TTX (320 nM) can alleviate the contractile dysfunction caused by ischaemia in isolated guinea-pig heart, and suggested that this would be a useful target for therapeutic intervention.

Diastolic dysfunction in ischaemia

Post-ischaemic rat myocardium displays an elevated diastolic calcium (to about 0.5 μM from control values of about 0.3 μM) accompanied by an increased LVEDP (to about 80 mm Hg from control of about 10 mm Hg) and a prolonged calcium transient (Meissner and Morgan, 1995). These observations have subsequently generally been confirmed (for example, Saini and Dhalla, 2005), although other groups have disputed whether the increased intracellular calcium is directly related to the increased diastolic tension (Eberli et al., 2000), pointing out that the situation is complicated by changes in myofilament calcium sensitivity. Since it appears to be a persisting defect in Ca²⁺ handling that is induced by ischaemia and reperfusion, rather than a persisting enhancement of I_Na(P) itself, its likely that blockers of I_Na(P) may be useful during reperfusion, but not post-reperfusion. This is still a matter to be resolved.

Protection against arrhythmias

Since I_Na(P) carries a depolarizing current, it tends to prolong the action potential. Normally this contributes only slightly to the plateau of the AP but, in conditions where repolarization is defective, I_Na(P) may produce arrhythmias. Even if I_Na(P) channel activity is constant during the plateau, as the action potential starts to repolarize the driving force for Na⁺ increases, and so I_Na(P) also increases. If repolarizing currents are less powerful than normal (because of drug-induced block of K⁺ channels, for example), the increase in I_Na(P) can produce a regenerative depolarization and early afterdepolarization (EAD) (Noble and Noble, 2006). A similar effect will also occur if I_Na(P) is increased more than usual, for example by hypoxia or by LQT3 mutations. Hence, block of I_Na(P) should be antiarrhythmic in these situations.

A longer term effect of I_Na(P) on arrhythmias is related to the coupled exchanger scheme noted above. Increased Ca²⁺ loading increases SR calcium to the point where the SR ‘overflows' and produces spontaneous oscillatory Ca²⁺ release, which activates NCX. Since NCX is electrogenic (Kang and Hilgemann, 2004) under these conditions it can produce membrane currents powerful enough to generate arrhythmias. This appears to be the mechanism behind delayed afterdepolarizations (DADs) (Verkerk et al., 2000) and some EADs (Szabo et al., 1994). Following the logic above, blocking I_Na(P) should be effective in preventing arrhythmias of these types by reducing [Na⁺]_i load, and the consequent [Ca²⁺]_i load, with the caveat that the block must be specific; some coincidental block of I_Na(T) would be expected to be pro-arrhythmic due to slowing of conduction velocity facilitating re-entry.

There may be other pathological conditions where block of I_Na(P) might provide an appealing antiarrhythmic target. It has been well documented that arrhythmias are more common in heart failure and hypertrophy. The mechanism is unclear, but it has recently been suggested that I_Na(P) contributes to these arrhythmias (Maltsev and Undrovinas, 2006). Block of I_Na(P) in myocytes from dogs with failure indeed appears to be antiarrhythmic (Undrovinas et al., 2006), although this result must be interpreted with the reservation that it is difficult to extrapolate effects in isolated cells to the level of the whole heart. A number of experiments on ventricular samples from human heart failure patients have shown that I_Na(P) is upregulated in failure (Undrovinas et al., 1999; Maltsev et al., 2001), although this is by no means universally accepted (Goldman and Balke, 2002). If I_Na(P) is indeed upregulated in failure, this would provide an appealing therapeutic target.

Blockers of I_Na(P)

Ranolazine

Ranolazine (Ranexa; Cardiovascular Therapeutics) was approved by the US FDA for the treatment of chronic angina in January 2006 (Abrams et al., 2006). Ranolazine has been shown to improve exercise performance and increase the time to angina/ischaemia in at least five trials (Chaitman, 2006). Interestingly, ranolazine seems to produce an additional improvement when added to existing antianginal agents (diltiazem, amlodipine and atenolol) (Morrow et al., 2006). Patient recruitment into a Phase III trial for ranolazine in non-S-T segment elevation acute coronary syndrome [(MERLIN)-TIMI] was started in October 2004 and the results reported in early 2007 (see below).

The proposed mechanism of action of ranolazine to produce this anti-anginal effect is block of I_Na(P). However, ranolazine was not developed as an I_Na(P) blocker, and initially it was thought that ranolazine acted by shifting metabolic substrate utilization from fatty acids to glucose (McCormack et al., 1996, 1998; Sabbah et al., 2002). Ranolazine also attenuates α1 and β1 adrenergic receptor-mediated responses and has 5-HT1A receptor agonist activity (Allely et al., 1993). This sympatholytic action may underlie some of the bradycardia seen with ranolazine in animal models, although the β-agonist effect is weak and there is an additional bradycardic effect not related to β-receptor block (Letienne et al., 2001).

Because of concerns about the discrepancy between the EC₅₀ for these effects and the therapeutic plasma level, attention has latterly turned to the electrophysiological actions of Ranolazine. In addition to block of I_Na(P) (EC₅₀=5.9 μM), ranolazine inhibits a number of other ionic currents, including IK_r (12 μM), late I_Ca (50 μM), peak I_Ca (296 μM), I_(Na–Ca) (91 μM) and IK_s (17% at 30 μM) (Antzelevitch et al., 2004). Importantly, the EC₅₀ for block of I_Na(P) can vary from 5 to 21 μM, depending on the voltage protocol and frequency. In addition, a simple statement of the EC₅₀ for block of IK_r, late I_Ca and I_Na(P) does not give a true reflection of the degree of block in the therapeutic range—because of the different apparent Hill coefficients, each of these currents is blocked to a similar extent (by about 25%) in the therapeutic range of 2–6 μM (Antzelevitch et al., 2004) (Block of I_Na(T) was not specifically reported, but little block occurs at 10 μM (inferred from reduction in dv/dt_max). In two different mutations of SCN5A, ranolazine was about 10 times more potent to block I_Na(P) than I_Na(T) (Fredj et al., 2006b) (Figure 3).

Figure 3.

Summary of potency of ranolazine to block various currents. From Antzelevitch et al. (2004) (Circulation).

Hence, although ranolazine blocks I_Na(P) with high potency, it blocks other currents just as well in the therapeutic range. Prediction of the effects of block of these currents at the tissue level is not a trivial exercise, since it depends on the relative size of each of the currents at each phase of the action potential. For example, ranolazine prolongs APD₉₀ in guinea-pig isolated perfused hearts (Wu et al., 2004), prolongs APD in canine epicardial cells but produces a bi-phasic effect on M cells (prolongation at low concentration, shortening at high), and shortens canine Purkinje fibre APD (Antzelevitch et al., 2004).

Consistent with other I_Na(P) blockers, ranolazine is protective in a rabbit model of ischaemia–reperfusion (Hale et al., 2006), suggesting that it may be a I_Na(P) blocker in this situation (although an earlier study in dogs found no reduction in infarct size (Black et al., 1994)). Direct evidence that this is the mechanism of protection has been sparse, although a recent study comparing the effect of ranolazine on Ca²⁺ accumulation and ventricular dysfunction in rat heart produced by either ATX-II or ischaemia–reperfusion did directly show that ranolazine produced a reduction in diastolic Ca²⁺ accumulation and an improvement in function in both situations (Fraser et al., 2006). However, the proposal that ranolazine is effective in angina by blocking I_Na(P) seems to be based on the logic that if ranolazine blocks I_Na(P) in reperfusion experiments and is effective in angina, then I_Na(P) must be implicated in angina. This is something of a circular argument, and needs more investigation. The difficulty is that neither the levels of hypoxia produced in angina or the degree to which I_Na(P) is increased by this are well understood—the levels of hypoxia in ischaemia–reperfusion experiments, or those used to enhance I_Na(P) in vitro are comparatively severe (pO₂ around 10–20 mm Hg) and it seems unlikely that pO₂ can fall to this level in angina, particularly non-ST-segment-elevation angina. There may of course be synergistic effects: it is possible that formation of reactive oxygen species, release of palmitoyl carnitine or lysophosphatidylcholine may occur in angina and these agents have been shown to increase I_Na(P) (Undrovinas et al., 1992; Wu and Corr, 1994; Ward and Giles, 1997), but whether this actually occurs in angina is again unknown. In this case it is perhaps unsurprising that the results of the MERLIN-TIMI trial showed that ranolazine provided no improvement in the primary efficacy endpoint (a composite of cardiac death, myocardial infarction or recurrent ischaemia), although recurrent ischaemia was reduced.

QTc prolongation requiring a reduction in the dose of intravenous drug occurred in 31 patients (0.9%) receiving ranolazine, compared with 10 patients (0.3%) receiving placebo. There was actually a significant reduction in arrhythmias detected on Holter monitoring during the first 7 days of treatment. The authors stated that the results of the MERLIN trial ‘do not support the use of ranolazine for acute management of ACS or as disease-modifying therapy for secondary prevention of cardiovascular death or MI.'

Hence, although there is clinical evidence that ranolazine improves ventricular function in angina (Hayashida et al., 1994; Rousseau et al., 2005) the idea that it does this via I_Na(P) block is unproven and it may be that its other pharmacological actions underlie this effect. Indeed, block of I_Na(P) was being proposed only as a potential mechanism of action of ranolazine as late as 2004 (Chaitman, 2004).

Nevertheless, it may be that this complexity of action is an advantage; it has been noted that because of these mixed electrophysiological actions, ranolazine, even though it may prolong QT in some situations, does not produce large disturbances in trans-mural dispersion of repolarization, and hence is less arrhythmogenic (particularly for Torsade de pointes) than comparable agents (Antzelevitch and Belardinelli, 2006).

Vernakalant (RSD1235)

Vernakalant (Cardiome Ltd, Vancouver, BC, Canada) has been reported to be a blocker of _INa(P) (Orth et al., 2006). It is in Phase III trials in intravenous form for reversion of atrial fibrillation (Roy et al., 2004) and the oral form is in Phase II trials. As with ranolazine, Vernakalant was not initially developed as an I_Na(P) blocker, but as an atrial-specific K⁺ channel blocker (Kv1.5, or IK_ur). In rats, ischaemic arrhythmias induced by coronary artery occlusion were inhibited with an ED₅₀ of 1.5 μmol kg⁻¹ per min (infusion rate), while 4–8 μmol kg⁻¹ per min gave complete protection (Fedida et al., 2005). Unfortunately, it was not stated whether there was any reduction in the size of the ischaemic zone with Vernakalant. In channels expressed in HEK cells the same paper reported block of K⁺ channels with IC₅₀ values of 13 μM on hKv1.5 (1 Hz) versus 38 and 30 μM on Kv4.2 and Kv4.3, respectively, and 21 μM on hERG channels. Sodium channel block was less potent (IC₅₀=43 μM at 1 Hz), although the potency increased with high frequency (IC₅₀=40 μM at 0.25 Hz, to an IC₅₀=9 μM at 20 Hz), and with depolarization (107 μM at −120 mV to 31 μM at −60 mV (1 Hz)). This paper did not examine I_Na(P) block, but this was subsequently investigated by the same group in 2006 (Orth et al., 2006). Although it blocked I_Na(P), Vernakalant showed almost no selectivity for I_Na(P) over I_Na(T): 30 μM blocked I_Na(P) by 70±4% and I_Na(T) by 61±4%, a difference much less than lidocaine in the same experiments (Orth et al., 2006). Again, these data were obtained with Nav1.5 channels expressed in HEK cells. This lack of block of I_Na(P) and a more potent block of K⁺ channels is consistent with Vernakalant's effects on action potential duration: in isolated rat myocytes the AP was prolonged approximately 75% over control by 10 μM, and at 30 μM APD increased approximately threefold, with prominent elevation of the AP plateau. Studies in patients with atrial fibrillation have shown that QRS duration is increased by 10% by Vernakalant at peak plasma levels of 5.8 μg ml⁻¹ (or ∼15 μM) (Roy et al., 2004). In the Phase I trial a dose of 5 mg kg⁻¹ produced plasma levels of 4 μg ml⁻¹ (about 11 μM), heart rate increased from 61±11 to 70±11 b.p.m., PR interval increased from 169±24 to 184±15 and QRS from 88±8 to 100±5. Slight QT prolongation was seen, from 384±14 to 419±6. These prolongations of QRS and QTc were not seen in the Phase II trial, compared to placebo at up to 3 mg kg⁻¹ (about 16 μM). In Phase III, Vernakalant has shown promising conversion rates (Fedida, 2007). Hence, it is likely that Vernakalant's antiarrhythmic actions stem from block of K⁺ channels, with an additional effect of frequency- and depolarization-dependent block of I_Na(T). I_Na(P) is unlikely to play a direct role, although it is possible that block of I_Na(P) counters to some extent the AP prolonging effect in the ventricle, and hence makes Vernakalant less likely to induce Torsades de Pointe.

Riluzole

Riluzole is a neuroprotective agent which can protect the brain from focal and global ischaemia (Bae et al., 2000). The mechanism by which it does this is uncertain, but it is thought to be due to a reduction of glutamate release (Doble, 1996) or possibly by activation of tandem pore K⁺ channels (Duprat et al., 2000). Recently, it has been suggested that Riluzole blocks I_Na(P) in neurones (St John et al., 2007) and it may be that this is how it confers protection. Data on the effects of Riluzole in cardiac tissue are sparse, but it appears that it does not have antiarrhythmic effects (Mestre et al., 2000).

Other compounds

There are other compounds purported to block I_Na(P) which are in early development as antiarrhythmic or anti-ischaemic agents. AZD7009 (AstraZeneca) is a compound under development for the treatment of atrial fibrillation (Crijns et al., 2006). It is proposed to be atrial-selective, in much the same way as Vernakalant. Its antifibrillatory action is probably due to block of IK_r, IK_ur and I_to, but it has less pro-arrhythmic effect than would be expected from this. It has been proposed, in much the same way as for Vernalkalant, that this is due to a concomitant block of I_Na(P). A recent study examined block of I_Na(P) in CHO cells expressing hNav1.5. AZD7009 blocked I_Na(P) with an EC₅₀ of 11±2 μM, but showed little selectivity over I_Na(P) (16±2.2 μM). Similar results were found in rabbit atrial and ventricular myocytes. AZD7009 also counteracted the AP prolongation and EAD formation in Purkinje fibres induced by E-4031 (Persson et al., 2007).

R56865 (Janssen Pharmaceuticals, Beerse, Belgium) has been shown to block the late I_Na induced by veratridine (Le Grand et al., 1998) and it was proposed that this is the basis for its protective effects against ischaemic damage (Ravens and Himmel, 1999). A more recent paper cast doubt on this interpretation, showing that R56865 did not prevent sodium overload during ischaemia (Hartmann and Decking, 2003), suggesting that its protective effect may be due to some other action (for example it was suggested some time ago that R56865 blocked calcium release from the SR; Ichikawa et al., 1994). Development of R56865 has apparently not progressed past the preclinical stage. KC12291 (Solvay Pharmaceuticals, Daix, France) is an anti-ischaemic compound thought to act by blocking I_Na(P) and thus preventing sodium overload. (Decking et al., 1998; Tamareille et al., 2002; John et al., 2004). KC12291 is still in the early stages of development.

Some caveats about I_Na(P) block

Although the heart expresses mainly Nav 1.5 (the more TTX-insensitive Na⁺ channel), there is also considerable expression of the more TTX-sensitive Nav 1.1, 1.2 and 1.3 (Brette and Orchard, 2006), particularly in the Purkinje fibres and bundles of His (Haufe et al., 2005). This should sound a note of caution for attempts to develop blockers of I_Na(P)—such a block may lead to a greater than expected disruption of the conducting system of the heart.

In addition, I_Na(P) appears to play a role in pacemaking in cells in the SA node (Muramatsu et al., 1996; Ju et al., 1996a). Consistent with this, KC12291 and R56865, which more selectively block I_Na(P) than TTX or lidocaine, have a much greater bradycardic effect in rat heart (Letienne et al., 2006). There is evidence that ranolazine is similarly bradycardic in animal models (Letienne et al., 2001) and small decreases in heart rate have been reported in the MARISA and CARISA trials, although subsequent trials have not raised concerns in this area. Vernakalant also appears to be bradycardic in humans, but this interpretation is complicated by the number of patients reverting from AF being included in the data (Roy et al., 2004). Paradoxically, LQT3 patients are often bradycardic (Veldkamp et al., 2003) while TTX has been reported to not be bradycardic in rabbit (Muramatsu et al., 1996).

In addition to being present in the heart, I_Na(P) is widely expressed in the CNS, where it seems to underlie repetitive or bursting behaviour for example (Rybak et al., 2003). Hence, a specific I_Na(P) blocker may have CNS effects. Of course, the presence of similar channels in the CNS and the heart has not prevented the development of antiarrhythmic agents blocking those channels (for example lidocaine). In this context, the blood–brain barrier can be an ally.

It was shown almost 20 years ago that a very specific blocker of I_Na(P), that is low dose TTX, is a good antiarrhythmic agent (Abraham et al., 1989). Clearly TTX is not a good candidate for an antiarrhythmic or ischaemia-protective agent. It remains to be seen whether a truly specific blocker of I_Na(P) will be.

Future directions

I_Na(P) blockers certainly show promise in protecting myocardium from ischaemia/reperfusion damage. However, it appears that in post-ischaemic myocardium, the contractile dysfunction is likely due to impaired calcium handling (due to irreversible damage produced during reperfusion) not due to alterations of I_Na(P). The idea that I_Na(P) may be enhanced in post-ischaemic (or ‘stunned') myocardium appears not to have been fully tested, and would be a worthwhile investigation. If it is enhanced, then blockers of I_Na(P) may be useful after reperfusion has occurred, rather than only prior to reperfusion or during ischaemia. Better data on the degree to which I_Na(P) may be involved in angina is also needed.

Paradoxically, targeting of I_Na(P) as a strategy for ischaemic protection may be best served by development of highly specific blockers, whereas for anti-arrhythmic efficacy, the development of deliberately non-specific channel blockers may be more fruitful, that is agents that have a clever combination of block of different ion channels, with attendant frequency dependence, to produce the desired electrophysiological profile. In these agents, block of I_Na(P) would be only one of the suite of blocking actions used to tailor the electrophysiological response. This approach (perhaps serendipitously) seems to have worked well in the case of Vernakalant and Ranolazine, and similar compounds may be on the horizon.

Table 1.

Merlin TIMI-36 outcomes

Abbreviations

I_Na(P)

persistent sodium current

I_Na(T)

transient sodium current

Conflict of interest

The authors state no conflict of interest.