It is now well known that carbon monoxide (CO) has significant physiological and pathophysiological effects in a number of different organ systems in mammals (Wu and Wang, 2005). This paper considers CO induced ventricular rhythm disturbances in the human heart. Recently, attention has been drawn to the fact that CO can induce cardiac arrhythmias. This is based on a number of studies in animal models and literature regarding emergency hospital admissions due to life-threatening cardiac rhythm disturbances triggered by increases in ambient levels of CO (Goldstein, 2008; Bell et al., 2009; Dallas et al., 2012). More generally, substances somewhat similar to CO, including nitric oxide (NO), and hydrogen sulfide (H₂S) are now often referred to in the neurophysiological literature as gaseous transmitters (Leffler et al., 2006). These compounds are receiving increased attention as (i) mediators or modulators of significant, sometimes acute pathophysiological challenges, and as (ii) potential targets for therapeutic interventions (Motterlini and Otterbein, 2010).

The ability of CO to alter electrophysiological activity in the heart and nervous system has led to quite broad-ranging examinations of the ability of CO to regulate specific ion channels (Wilkinson and Kemp, 2011). These effects can occur either as direct actions of CO, or (and more commonly) CO effects on closely related intracellular signaling pathways, often involving nitric oxide signaling, metabolism, and/or downstream targets for chemical modifications of individual amino acids in functional proteins. In the cardiovascular system, specific attention has been drawn to S-nitrosylation of, specific ion channel residues (Jaffrey et al., 2001; Haldar and Stamler, 2013).

In this paper, our focus will be on the cellular electrophysiological effects of NO and related S-nitrosylation of specific sites on ion channels, ion transporters, or directly related signaling molecules (Jaffrey et al., 2001; Tamargo et al., 2010; Haldar and Stamler, 2013). This is because many cellular effects initiated by increased CO levels are mediated by downstream changes in NO and ultimately through targeted nitrosylation of defined sites (residues). At present, it is known that NO/S-nitrosylation can: (i) increase the background K⁺ current that generates the resting potential in heart (Gómez et al., 2009) (ii) activate ATP sensitive K⁺ channels in neurons (Kawano et al., 2009), and (iii) reduce a major repolarizing current (K_v1.5) in mammalian atria (Núñez et al., 2006). In addition, a similar CO and/or NO induced reaction mechanism, as S-nitrosylation, can modulate (iv) Na⁺/K⁺ ATPase turn over in the hypoxic heart (Yakushev et al., 2012), and (v) alter release of Ca²⁺ from the sarcoplasmic reticulum (SR) by targeting the ryanodine receptor complex in heart and skeletal membranes (Gonzalez et al., 2010; Wang et al., 2010). There is also evidence that NO can increase the slowly inactivating or late Na⁺ current in nerve, muscle and heart preparations (Ahern et al., 2000; Evans and Bielefeldt, 2000).

Two recent papers (Abramochkin et al., 2011; Dallas et al., 2012) have reported CO induced changes in electrical activity and contractions, as well as induction of cardiac arrhythmias in a rat ventricle model. Evidence that a major ion channel mediated effect is an augmentation of slow inactivation of the cardiac Na⁺ current is provided. These results also emphasize an essential involvement of NO as a second messenger of this CO-induced effect. CO-induced increases in this late Na⁺ current, I_Na−L, can significantly lengthen the action potential (AP) and result in abnormal electrical activity (early after-depolarization, EADs) characterized by repetitive firing, even after application of only one stimulus (Dallas et al., 2012). This arrhythmia can be reduced substantially following superfusion of this experimental preparation with the compound ranolazine (Dallas et al., 2012). Ranolazine, originally developed as a coronary vasodilator, is now known to be a potent and quite selective inhibitor of the I_Na-L in a number of different mammalian preparations (Makielski and Valdivia, 2006; Zaza et al., 2008), and also in human ventricle (Moss et al., 2008). Ranolazine effects are significant in the settings of stable angina and also in heart failure where, in both cases, it is known that I_Na-L is increased (c.f. Trenor et al., 2012).

The main goal of our mathematical simulations is to illustrate, using a current mathematical model of the AP in human ventricle, the consequences of a CO induced: (i) increase in I_Na-L, (ii) NO-S-nitrosylation induced changes in I_Ca-L, and (iii) this combination. For this purpose, CO-induced changes in I_Na-L that are similar to those reported in the primary experimental data (Dallas et al., 2012) were introduced into two different models of the ventricular AP and the consequences were explored systematically.

The initial set of computations was done for the purpose of illustrating the consequences of the CO induced changes in mammalian heart Na⁺ current, I_Na that were described in a primary experimental study. This paper (Dallas et al., 2012) reported that CO acting through an intracellular NO-mediated signaling cascade can (i) reduce peak I_Na by as much as 50% while also (ii) shifting the inactivation curve for the large transient component of I_Na in hyperpolarizing direction, and (iii) significantly increasing the slowly inactivating component of I_Na (which is denoted I_Na-L).

The two superimposed human ventricular APs in Figure 1A illustrate the control AP waveform at 1 Hz, and the AP elicited following the changes in I_Na and I_Na-L described above. For these computations, the steady-state inactivation curve was shifted in the hyperpolarizing direction by 5 mV and I_Na-L was increased 2-fold. Figure 1B shows the changes in I_Na-L; Figure 1C illustrates the indirect effects on I_Ca-L; that is, the AP waveform lengthened, and this changed the I_Ca-L; and Figure 1D shows the computed intracellular Ca²⁺ transient, [Ca²⁺]_i.

In summary the reported changes in I_Na and I_Na-L, when combined, result in predictable decreases in AP rate-of-rise (not shown) and small increase in AP duration. However, these changes fail to elicit spontaneous firing (EADs on DADs) or produce any correlates of arrythmogenesis. Although this pattern of results was somewhat unexpected, it is important to recall that the primary experimental data on CO were obtained in rat hearts (Ahern et al., 2000; Dallas et al., 2012), and not from human ventricular tissue or myocytes. We therefore, continued to attempt to illustrate and further understand the electrophysiological effects of CO on the human ventricular myocardium by considering additional variables.

In fact, studies of NO-mediated effects in heterologous expression systems utilizing the alpha subunit of the human heart Na⁺ channel, Na_v1.5 (Ueda et al., 2008), reveal a somewhat different pattern of results. These include (i) an (approximately 30%) increase in peak inward Na⁺ current, a (ii) small hyperpolarizing shift in the steady-state inactivation curve for I_Na, (iii) a marked enhancement (approximately 2-fold) in I_Na-L and (iv) an essential NO-induced S-nitrosylation of a defined residue on Na_v1.5. These changes, when introduced into the Grandi model, also failed to result in increased arrhythmogenesis (results not shown).

We were aware of the detailed studies of Campbell et al. (1996) on the effects of NO on I_Ca-L, in the mammalian (ferret) heart. These investigators reported a significant increase in I_Ca-L (30–50%) under conditions of NO induced S-nitrosylation of this Ca²⁺ channel α-subunit. Importantly, this paper also reported a nitrosylation-induced change in I_Ca-L gating, specifically, a small but significant shift in the voltage dependent activation relationship—approximately 6 mV in the hyperpolarizing direction. The superimposed APs in Figure 2 illustrate the consequences of introducing these changes into the original Grandi et al. model (Grandi et al., 2010). Panel (A) shows the AP waveforms (black, baseline, or control) (red, following I_Ca-L changes). I_Na-L and I_Ca-L records are shown in Panels (B,C), respectively. The L-type Ca²⁺ current is shown in Panel (C). As shown in Panel (D), these changes in I_Ca-L also resulted in a large increase in [Ca²⁺]_i, and the [Ca²⁺]_i exhibited pronounced frequency dependence or alternans (see Discussion and Clark et al., 1996; Bouchard et al., 2004).

Detailed inspection of Figure 2 and comparison of this pattern of results with those in Figure 1 reveal (Panel B) that I_Na-L now shows pronounced changes in peak amplitude, even in response to the fixed stimulus rate at 1 Hz. Presumably, this arises from the markedly prolonged AP and intrinsic voltage-dependence of reactivation of I_Na-L (see Discussion).

The third set of computations in this study combined the reported CO induced changes in I_Na and I_Na-L (Figure 1), with those for I_Ca-L (Figure 2). The resulting very significant changes in AP duration (Panel A), I_Na-L (Panel B), I_Ca-L (Panel C), and [Ca²⁺]_i (Panel D) are shown in Figure 3. This pattern of results was similar to, but not identical with Figure 2. In particular, the majority of the APs elicited at a physiological stimulus rate (1 Hz) exhibited marked prolongation (approximately 100%), with the appearance of an EAD late in the prolonged plateau phase. Note that under these starting conditions, the combined inward currents due to I_Na-L and I_Ca-L “held” the membrane potential of the plateau near 0 mV in contrast with the approximately −10 mV plateau level in Figure 2.

The pattern of changes shown in Figure 3 is not surprising. It is now well known that abnormal, slow depolarizations, which are denoted EADs, can be elicited by changes in I_Ca-L (Marban et al., 1986; January and Riddle, 1989, for review see Clusin, 2003). In fact, in rabbit ventricle a recent report from the Weiss laboratory (Madhvani et al., 2011) has illustrated in detail ways in which even very small changes in the voltage dependence of the activation of I_Ca-L can give rise to EADs. This is mainly, but not entirely due to the fact that there is an increased inward current at or very near the range of membrane potentials of the plateau of the normal or prolonged AP. The underlying biophysical mechanism(s) involves the non-linear interactions of the small overlapping Ca²⁺ channel—and Na⁺/Ca²⁺—exchanger mediated currents in this range of membrane potentials (Madhvani et al., 2011, for review see Fink et al., 2006). The altered AP waveform and related pro- or anti-arrhythmic effects are sometimes described in terms of an altered “repolarization reserve” at defined time-points of early or late repolarization (Wu et al., 2009; Varró and Baczkó, 2011; Trenor et al., 2013).

If environmental or tissue-derived CO can induce life-threatening ventricular rhythm disturbances through a mechanism involving enhancement of I_Na-L, it is plausible that, as mentioned in the Introduction, ranolazine could be effective in Critical Care settings (Dallas et al., 2012, cf. Belardinelli et al., 2006; Bell et al., 2009). The mathematical simulations shown in Figure 4 address this possibility by illustrating the effects of ranolazine “in silico.” These computations were carried out under starting conditions (i.e., an initial parameter set) almost identical to those for Figure 3. However, for these computations it was assumed that ranolazine actions would be the equivalent of a 50% reduction of the CO induced increase in I_Na-L (see Figure 1). The results show that if ranolazine (at an assumed plasma level of 5 μ M) did block approximately 50% of I_Na-L, it could shorten the AP elicited by a 1 Hz train of stimuli. Note, however, that this partial block of I_Na-L reduced APD only (approximately) every 3rd AP in this 1 Hz AP train. An additional effect is also apparent: reducing I_Na-L changes the (i) waveform and (ii) the most negative membrane potential in some repolarizing components within this train of AP's. This primary “effect” appears to be linked functionally to the approximately normal repolarization of the next AP. In summary, when ranolazine is introduced in silico, the resulting AP waveforms show complex alternating patterns. However, there is an increased number of (i) near normal AP waveforms and (ii) monotonic [Ca²⁺]_i records. The incidence of EADs is also reduced, consistent with the previously described actions of this drug in the settings of free radical challenge, genetic abnormalities in Na⁺ channel biophysics, hypoxia, or global ischemia/heart failure (Fink et al., 2006; Trenor et al., 2012).

The simulations in Figure 5 address and attempt to illustrate some of the underlying causative factors for EAD generation within the framework/limitations of this model of the human ventricle AP. EAD generation requires the membrane potential to be relatively depolarized, so that L-type Ca²⁺ channels can be activated and contribute a small but significant inward Ca²⁺ current. We note that the main interacting (or overlapping) ion channel mediated currents at the plateau of the AP are: (i) I_Na-L, (ii) I_Ca-L, (iii) I_K1, a time-independent or background inwardly rectifying current, and (iv) the delayed rectifier K⁺ current, I_K−R (Trenor et al., 2013). In the presence of an enhanced I_Na-L the AP plateau is somewhat depolarized and the AP tends to be prolonged. In addition, under these conditions I_NCX generated by the Na⁺/Ca²⁺ exchange mechanism can be functionally important. A key question is: are the pronounced changes in [Ca²⁺]_i the cause or the consequence of the EAD/CO-induced pro-arrhythmic substrate. A plausible answer can be obtained by consideration of the results in Panels (A–F) of Figure 5.

Panel (A) shows two superimposed pairs of APs, each elicited at steady state (1 Hz) before (black) and after (red) introducing the changes shown in Figure 3. Panels (B,C) show the corresponding I_Na-L and I_Ca-L records. Although this information is redundant with that in Figure 3: here, it provides a means for direct comparison with the new data: Panel (D), [Ca²⁺]_i records; Panel (E), I_NCX; and Panel (F), the Ca²⁺ content of the SR.

Taken together, these results suggest that in this model the CO/NO-induced changes in I_Ca-L are critical. With each stimulus, a large “extra” Ca²⁺ influx prolongs the plateau and results in enhanced Ca²⁺ release and a much larger increase in [Ca²⁺]_i. This change in [Ca²⁺]_i can stimulate the I_NCX. Perhaps more importantly, however, this time-dependent and cumulative change in myocyte Ca²⁺ homeostasis results in markedly increased Ca²⁺ in the SR. Eventually, this causes anomalous Ca²⁺ release from the SR, and the Na⁺/Ca²⁺ exchanger (denoted by arrows) responds by generating an inward current that gives rise to the EAD. Under these conditions, approximately every second AP is associated with spontaneous release at the steady-state stimulus rate of 1 Hz.

This scheme/scenario seems plausible, since similar patterns of changes have been described experimentally. However, it may not be readily apparent how ranolazine, acting only on I_Na-L could be anti-arrhythmic (see Discussion). In brief, the answer appears to be that ranolazine, by blocking I_Na-L, can shorten the AP. This reduction in APD results in reduced Ca²⁺ influx through I_Ca-L during each AP and thus, there is less Ca²⁺-loading of the SR. The end result is that SR Ca²⁺ content does not reach the abnormal levels required for anomalous and spontaneous Ca²⁺ release. Hence, there is no intracellular Ca²⁺ “trigger” for the electrogenic I_NCX, that has an important role in initiating the EAD in this model (Houser, 2000; Clusin, 2003).

Doctors Trenor, Saiz and Giles shared responsibilities for study design. Computations were performed at Universidad Politécnica de Valencia by Doctors Cardona and Trenor, Dr. Giles wrote this manuscript. All authors contributed to final editing and the R-1 revisions.