|Home | About | Journals | Submit | Contact Us | Français|
Variability in delivery of oxygen can lead to electric instability in the myocardium and the generation of arrhythmias. In addition ischemic heart disease and angina are associated with an increase in circulating catecholamines that further increases the risk of developing ventricular tachyarrhythmias.
We investigated the net effects of acute hypoxia and catecholamines on the cardiac action potential.
We incorporated all published data on the effects of hypoxia on the late Na+ current (INa-L), the fast Na+ current (INa), the basal L-type Ca2+ channel current (ICa-L), and the slow (IKs) and rapid components of the delayed rectifier K+-current (IKr) in the absence and presence of β-adrenergic receptor (β-AR) stimulation into the Luo–Rudy model of the action potential. Hypoxia alone had little effect on the action potential configuration or action potential duration. However in the presence of β-AR stimulation, hypoxia caused a prolongation of the action potential and early afterdepolarizations (EADs) and spontaneous tachycardia were induced. Experiments performed in guinea pig ventricular myocytes confirmed the modeling results.
EADs occur predominantly because of the increased sensitivity of ICa-L to β-AR stimulation during hypoxia. β-AR stimulation is necessary to induce EADs as EADs are never observed during hypoxia in the absence of β-AR stimulation.
Ventricular tachycardia and ventricular fibrillation are a major cause of death in patients with myocardial infarction and a reduced left ventricular ejection fraction.1 Typically arrhythmias occur as a result of re-entrant excitation or increased automaticity. Early afterdepolarizations (EADs) are depolarizations of the membrane potential that occur predominantly during phase 2 or 3 of the cardiac action potential and can degenerate to polymorphic ventricular tachycardia.2,3 EADs and triggered activity can induce reentrant arrhythmias. Generation of EADs requires an inward current that is large enough to depolarize the membrane potential.4,5
Variability in delivery of oxygen can lead to electric instability in the myocardium and the generation of arrhythmias.6 The cellular consequences of temporary acute hypoxia (seconds to minutes) differ significantly from chronic hypoxia (hours to days) or anoxia. A rapid decrease in oxygen supply to cardiac myocytes from 150 to 15 mm Hg is not energy limiting and does not deplete ATP7 but can alter the function of a number of cardiac ion channels.8–17 Under these conditions hypoxia increases late Na+ current (INa-L) while decreasing fast Na+ current (INa) in rat ventricular myocytes.14–16 It has been proposed that the increase in INa-L may be arrhythmogenic.18 In addition, acute hypoxia decreases the basal current through L-type Ca2+ channels (ICa-L)8,9,11–13,19,20 and the slow component of the delayed rectifier K+ channel (IKs) without affecting the rapid component (IKr).10 However, the net effects of acute hypoxia on action potential (AP) configuration in cardiac myocytes are not known.
Ischemic heart disease and angina are also associated with an increase in circulating and tissue catecholamines that increases the risk of developing ventricular tachyarrhythmias and sudden cardiac death.21 Hypoxia decreases the K0.5 for activation of ICa-L by the β-adrenergic receptor (β-AR) agonist isoproterenol (Iso).11 However, hypoxia also increases the sensitivity of IKs to β-AR stimulation without altering IKr and this could counteract the effects of hypoxia on ICa-L.10 In this study we used the Luo–Rudy model of a ventricular myocyte22 to determine the effects of acute hypoxia on the AP in the absence and presence of β-AR stimulation. By incorporating all published data on the effects of acute hypoxia (po2 of 15 to 20 mm Hg) on Na+, Ca2+, and K+ currents, we find that in the absence of β-AR stimulation, hypoxia has little effect on the AP configuration and duration. However, in the presence of β-AR stimulation, hypoxia causes a prolongation of the AP and triggers EADs. We produce experimental data in guinea pig ventricular myocytes that support these theoretical findings and determine that EADs are generated predominantly because of hypoxia-induced increased sensitivity of ICa-L to β-AR activation.
The theoretical dynamic model of a mammalian ventricular AP, the Luo–Rudy model, provides the basis for the simulations.23 The model is predominantly based on guinea pig experimental data. The membrane ionic channel currents are formulated mathematically using Hodgkin–Huxley formalism. Ionic pumps and exchangers are also included in the model. The model accounts for processes that regulate intracellular ionic concentration changes of Na+, K+, and Ca2+. Intracellular processes represented in the model include Ca2+ uptake and Ca2+ release by the sarcoplasmic reticulum (SR) and the buffering of Ca2+ by calmodulin and troponin (in the myoplasm) and calsequestrin (in the SR). For the Na+–Ca2+ exchanger, the model uses a formulation based on conservation principle.24 Experimental data on voltage dependence of conductance and open time duration14,16 were used to formulate and include a model of INa-L in the model. β-AR effects were included in the model by using the K0.5 for enhancement of ICa-L and IKs caused by Iso as observed experimentally10,11 and by upregulation of SR Ca2+ uptake. Iso effect on inward rectifying potassium current (IK1)25 was also considered in simulations of the progressive effect of hypoxia on APs. Details of the model are provided in the Online Data Supplement and the research section of http://rudylab.wustl.edu.
Hypoxia decreases INa and increases INa-L in ventricular myocytes.14–16 The effect of hypoxia on the sodium current was modeled by reducing conductance of INa by 10% and increasing the conductance of INa-L so that current was in the range of 0.1 to 0.5% of INa as seen experimentally.15 Hypoxia decreases basal ICa-L and IKs in the absence of β-AR stimulation.10,11 Hypoxia also decreases K0.5 for activation of ICa-L and IKs by Iso.10,11 Both of these effects were included in the model. These responses are reversible with an increase in oxygen tension to normoxia (room oxygen). During pacing, a stimulus of −80 μA/μF is applied for a duration of 0.5 ms. The model is paced with a conservative current stimulus carried by K+.26 A variable adaptive time step algorithm was used to simulate action potentials.22 Steady state was reached after 88 simulated beats. The ventricular action potential cell model was coded in C++ and run on Linux cluster nodes. For further details, see the Online Data Supplement.
We incorporated in the model the experimentally measured effects of hypoxia on INa, INa-L, ICa-L, IKr and IKs in the absence of β-AR stimulation. In experiments, exposure to hypoxia reversibly reduces peak ICa-L by approximately 25% at 0 mV8,9,11–13,19,20 without shifting the current–voltage (I-V) relationship (Figure 1A).11,12 Figure 1B shows the simulated I-V relationship for ICa-L during control (normoxic) conditions and during hypoxia in the absence of Iso. There is a reduction in current amplitude but no shift in the I-V relationship, in good agreement with the experimental data. Acute hypoxia also causes a 21.9±1.8% reversible decrease in steady state basal IKs current during a voltage-step to +50 mV (Figure 1C).10 Simulations demonstrate that hypoxia is associated with a 26.5% decrease in current at +50 mV, in good agreement with experimental data (Figure 1D).
We modeled the effects of hypoxia in the absence of β-AR stimulation on the AP at 2 pacing cycle lengths (CLs) of 300 and 1000 ms (Figure 2). As expected, the action potential duration (APD) decreases with increase in frequency of stimulation, consistent with rate-dependent shortening of the APD.27 The APD90 for CL of 300 ms is 160 ms and the APD90 for CL of 1000 ms is 225 ms. When the effects of hypoxia on INa-L alone were incorporated in the model, the APD90 for CL of 300 ms is 162 ms and for CL of 1000 ms is 233 ms, with no change in resting membrane potential (RMP) or action potential peak (APP). Incorporating the effects of hypoxia on INa-L and INa in the model did not alter APD90 at CL of 300 ms (162 ms) or CL of 1000 ms (232 ms), and there was no change in RMP or APP. We conclude that neither INa-L nor INa alters the AP parameters during hypoxia. The effects of hypoxia on INa-L and INa were included in all the hypoxia simulations reported below.
When the effects of hypoxia on ICa-L, INa, and INa-L, but not on IKs, are incorporated in the model, the APD90 at CL of 300 ms is reduced by 4% to 153 ms and at CL of 1000 ms by 9% to 205 ms (Figure 2B). In some studies, a slight leftward shift in the voltage dependence of ICa-L during hypoxia has been reported.8,9 This shift was incorporated in the model but did not significantly alter the AP morphology although there was a slight decrease in APD (APD90=151 ms at CL of 300 ms and APD90=200 ms at CL of 1000 ms).
Next, we modeled the effect of hypoxia on INa, INa-L, and IKs but not on ICa-L (Figure 2C). Under these conditions, hypoxia causes a prolongation of APD90 by 8% to 173 ms at CL=300 ms and by 14% to 256 ms at CL=1000 ms. Finally, we modeled the known effects of hypoxia on all depolarizing and repolarizing currents (INa, INa-L, ICa-L, IKs, and IKr) (Figure 2D). There was very little effect on APD compared to normoxic controls (Figure 2A and Online Table I). To further examine a possible arrhythmogenic role for INa-L, we modeled the effect of hypoxia on INa-L at twice the value reported in published studies.14,16 There was no significant effect on APD or AP morphology (at CL of 1000 ms: APD90=234 ms, RMP=−88 mV, and APP=45 mV). We conclude that hypoxia alone does not significantly alter the action potential.
The effects of β-AR stimulation on ICa-L are well documented.28 Binding of the β-AR leads to activation of cAMP and protein kinase A–dependent phosphorylation of the channel protein that then increases current magnitude and mode 2 open time. β-AR stimulation also leads to an increase in the magnitude of IKs as a result of direct phosphorylation of the channel.29 Neither the function of INa-L nor INa is regulated by protein kinase A or β-AR stimulation at concentrations less than 1μmol/L.30 In the absence of hypoxia, 10 nmol/L Iso increases ICa-L to 72.1% of the current produced by a maximally stimulating concentration of Iso (1μmol/L) in the same cell11 (Figure 3A). The peak current is increased 2.8-fold and is shifted 10.8 mV in the negative direction relative to the peak current recorded in control (no hypoxia). Corresponding simulations in Figure 3B are in good agreement with the experimental data in Figure 3A. Exposure to 10 nmol/L Iso in the presence of hypoxia (Figure 3C) increases the magnitude of the peak current an additional 22.2% without further shifting the I-V relationship (−11.7 mV). Corresponding simulations in Figure 3D are in good agreement with the experimental data in Figure 3C.
In the presence of a saturating concentration of Iso (1μmol/L), there is ≈3-fold increase in steady-state IKs (Figure 3E).10 In the absence of hypoxia, 10 nmol/L Iso increases IKs to ≈45.5% of the current produced by 1 μmol/L Iso within the same cell (Figure 3E), while in the presence of hypoxia, 1 nmol/L Iso increases IKs ≈56.5% of the current produced by 1 μmol/L Iso in the same cell (Figure 3G).10 The magnitude of current density produced by 1 nmol/L Iso during hypoxia was comparable to the current density produced by 10 nmol/L Iso in room oxygen over the entire voltage range. Figure 3F demonstrates the effect of 10 nmol/L Iso on IKs during normoxia predicted by the model. Figure 3H demonstrates simulations for the effect of 1 nmol/L Iso on IKs in the presence of hypoxia.
We modeled the concentration dependence of ICa-L and IKs on Iso. In the absence of hypoxia, Iso increases ICa-L with a concentration that produces half-maximal activation (K0.5) at 5.3±0.7 nmol/L.11 When cells were exposed to hypoxia, the K0.5 for ICa-L was significantly decreased to 1.6±0.1 nmol/L and the current was maximally stimulated with 10 nmol/L Iso.11 In the model, we used a K0.5 value of 5.3 nmol/L under normoxic conditions and 1.6 nmol/L Iso during hypoxia for enhancement of ICa-L. Figure 4A shows that the model prediction of ICa-L enhancement by Iso is in good agreement with experimental data. In the absence of hypoxia, exposure of myocytes to 1 nmol/L Iso produces a subthreshold response for IKs and the current is maximally stimulated in the presence of 1μmol/L Iso.10 The K0.5 for enhancement of the current in the absence of hypoxia is 18.3±3.9 nmol/L. In the presence of hypoxia, 0.1 nmol/L Iso produced a threshold response and IKs was near-maximally activated at 10 nmol/L Iso. The K0.5 for enhancement of the current was significantly decreased to 1.88±0.43 nmol/L under hypoxic conditions. Hypoxia did not alter the response to 1 μmol/L Iso, a maximally stimulating concentration of the agonist. In the model we used the K0.5 value of 15 nmol/L during normoxia (control) and a value of 1.5 nmol/L during hypoxia. These values are within the range of observed experimental values (Figure 4B).
The modeling results in Figures Figures1,1, ,3,3, and and44 serve as validation that the model reproduces the experimental effects on individual membrane currents. In this section, we explore the effects of hypoxia and β-AR stimulation on the whole cell AP. Because hypoxia affects both the plateau-forming depolarizing current ICa-L and repolarizing current IKs, in addition to affecting their sensitivity to Iso, it is unclear which of these currents plays a dominant role in shaping the AP morphology during hypoxia. Therefore, we first investigated the role of each channel in isolation and then combined their effects.
We investigated the effect of hypoxia on the AP while varying the amount of Iso to determine whether and at what Iso concentration arrhythmic disturbances occur. At low concentrations of Iso, we observed the generation of EADs. The threshold for EAD generation in the ICa-L hypoxic cell was 0.6 nmol/L at 1000-ms CL (Figure 5B and Online Table I). In the absence of Iso or at the same level of Iso (0.6 nmol/L), there was no evidence of rhythm disturbance in a control normoxic cell (Figure 5A and 5D; Online Table I). The APD90 in a control cell, with the addition of 0.6 nmol/L Iso, increased 8% to 244 ms at CL of 1000 ms and by 4% to 166 ms at CL of 300 ms with no changes in RMP or APP. When ICa-L was made hypoxic, the APD90 at CL 1000 ms was prolonged extensively by 82% (Online Table I) and EADs were observed (Figure 5B). At CL of 300 ms, APD90 was increased by 13% (relative to control) to 180 ms, but EADs were not generated.
At 1000-ms CL, prolongation of the AP led to the generation of an EAD. This is in contrast to the reduction of APD90 measured during hypoxia in the absence of Iso (Figure 2B). The concentration of Iso that induced EADs in the model (0.6 nmol/L) is close to K0.5 for activation of ICa-L by Iso during hypoxia (1.6 nmol/L) and to the threshold for activation of IKs under normoxic conditions (Figure 4). The prolongation of APD was not sufficient to generate EADs at 300 ms at any concentration of Iso.
Next, we investigated the effect of hypoxic IKs on the AP while varying the amount of Iso. We kept the values of other ion channels and transporters at normoxic levels in the model, except INa-L and INa, which were set at hypoxic values. We examined the effect of 0.6 nmol/L Iso in the presence of hypoxic IKs at CL of 1000 ms. From Figure 5E, it is seen that EADs were not observed for either CL of 1000 or 300 ms. The APD90 for CL of 1000 ms was increased by 26% compared to control (from 225 to 284 ms; Online Table I) and for CL of 300 ms 11% (from 160 to 177 ms). EADs were not generated over a wide range of Iso concentrations when only IKs channels were made hypoxic. This is attributable to the early enhancement of IKs as a result of increased sensitivity to Iso during hypoxia (K0.5=1.9 nmol/L).
Next, we modeled the combined effects of hypoxic ICa-L and IKs in the presence of β-AR stimulation. We progressively increased the concentration of Iso in the cell until EADs were observed at CL of 1000 ms. The level of Iso at which EADs first occurred was 0.5 nmol/L. The APD90 at CL of 300 ms was increased by 14% to 182 ms and for CL 1000 ms by 103% to 457 ms (Figure 5F; Online Table I). The concentration of Iso that induces EADs is lower than K0.5 for activation of IKs during hypoxia (1.9 nmol/L), implying that IKs current magnitude remains close to the basal hypoxic level and is not sufficient to counter the proarrhythmogenic effects of ICa-L.
In summary, the simulations predict that during periodic pacing in the presence of β-AR stimulation, the primary reason for EAD generation during hypoxia is increased influx of calcium through ICa-L. This effect may be augmented by the reduced outward current produced by IKs during hypoxia. However, the effect of hypoxia on IKs alone is insufficient to generate EADs. In addition, β-AR stimulation is necessary to induce EADs, because EADs are never observed under hypoxic conditions in the absence of β-AR stimulation.
During periodic pacing, the simulations did not generate EADs in a cell paced at CL of 300 ms. However, a case of clinical importance is periodic excitation followed by a pause.31 We examined the effect of a 1000-ms pause on a cell paced at a regular rate at 300-ms CL. In the absence of β-AR stimulation, EADs were not induced when the effects of hypoxia on ICa-L (Figure 6B), on IKs (Figure 6C), or the combined effects of hypoxia on ICa-L and IKs were included (Figure 6D).
The concentration of Iso that produced a threshold for EAD generation for hypoxic ICa-L was 0.6 nmol/L Iso (Figure 6B). At this concentration, EADs were not generated in a control or hypoxic IKs cell (Figure 6A and 6C). The combined effect of hypoxic ICa-L and IKs on a postpause AP is shown in Figure 6D. The threshold concentration of Iso for induction of EADs decreases from 0.6 to 0.4 nmol/L. We conclude that in a postpause AP, the primary cause for EAD formation is the effect of hypoxia on ICa-L in the presence of β-AR stimulation. Reduced IKs plays a secondary role and does not by itself induce EADs at Iso concentrations that generate EADs as a result of increased ICa-L.
In the model, when we only include the effect of Iso on inward rectifier K+ current (IK1),25 in the absence of hypoxia, there is no evidence of rhythm disturbance (data not shown). APs follow the periodic pacing and when pacing is stopped after 50 seconds, the cell becomes quiescent. Similar results were obtained for a hypoxic cell. However, in the presence of 1 nmol/L Iso with its time-dependent effect on ICa-L and IKs and instantaneous effect on IK1, the hypoxic cell generates EADs. Shortly thereafter, the cell generates triggered nonpaced beats (Figure 7). The pattern of EADs and triggered beats continues until pacing is stopped after the 50th beat. After cessation of pacing, the cell continues to beat periodically with a CL of 464 ms. There is no evidence of spontaneous activity in a control (nonhypoxic) cell in the presence of 1 nmol/L Iso (data not shown). We find that the spontaneous beats are generated by activation of ICa-L and not INa. The ICa-L during the spontaneous beat upstroke is substantially larger than ICa-L during the paced beat (Online Figure I).
Simulations show that the minimum diastolic cytosolic Ca2+ concentration (Cai) before the EADs occur is lower than the minimum diastolic Cai after EADs and/or sustained triggered activity. INaCa before the EADs occur is smaller than INaCa when EADs and/or spontaneous beats occur. When diastolic INaCa or Cai is clamped to values before EADs occur, the spontaneous beats are abolished (Online Figures II, III, and IV). Similarly, when extracellular Ca2+ is reduced from 1.8 mmol/L to 1.44 mmol/L in the simulations, both EADs and spontaneous beats are abolished. These results imply that Ca2+ plays an important role in the generation of EADs and spontaneous beats.
In the model, when we block the SR Ca2+ release channel to mimic the experimental protocol of ryanodine infusion, the sustained triggered activity persists. However, when we block ICa-L mimicking the action of Nifedipine, it is abolished (Online Figure V). These results imply that ICa-L is responsible for generation of the triggered activity and that spontaneous SR Ca2+ release is not required for its generation.
We investigated the model predictions of the effects of hypoxia in guinea pig ventricular myocytes. Reducing po2 from normoxic (po2 of 150 mm Hg) to hypoxic conditions (po2 of 17 mm Hg) for 60 seconds did not significantly alter RMP, APP, or APD (Figure 8A and Online Table I). We then exposed the myocyte to hypoxia in the presence of 1 nmol/L Iso, a concentration of the β-AR agonist that is subthreshold for activation of ICa-L and IKs during normoxia (see Figure 4). The addition of Iso did not alter RMP (−82±3 versus −81±2 mV) or APP (45±2 versus 44±2 mV) but significantly increased APD by 11% (182±20 versus 201±25 ms, P<0.05, n=5). Increasing the concentration to 3 nmol/L Iso significantly increased APD by 37% without altering RMP or APP (Figure 8B and Online Table I). Application of dantrolene to block ryanodine receptor release of Ca2+ did not prevent the prolongation of AP (see Online Figure VIII). Three of 7 cells exposed to hypoxia +3 nmol/L Iso generated EADs and then started to beat spontaneously at CL of ≈500 ms, close to the CL of 464 ms predicted by the model (Figure 8C and 8D). The upstroke velocity (dV/dt)max of a paced beat was 138 mV/ms and of a spontaneous beat 73.5 mV/ms (Online Table I). This confirms the model prediction that the upstroke in a paced beat is generated by INa and in a spontaneous beat is by ICa-L. Importantly, 3 nmol/L Iso did not alter AP parameters under normoxic conditions (RMP=−78±3 versus −79±4 mV, APP=41±5 mV versus 38±6 m, APD=238±55 versus 233±58 ms, all P>0.05, n=3), indicating that it is the combination of acute hypoxia and β-AR stimulation that is responsible for AP prolongation and EAD formation.
Although it is well recognized that arrhythmias are a significant cause of death in ischemic heart disease, the role of acute hypoxia in induction of arrhythmia is not well understood. We incorporated all published data reporting the effects of acute hypoxia on INa, INa-L, ICa-L, IKs, and IKr into the Luo–Rudy model of a cardiac ventricular AP and determined the effect on AP morphology and APD. In the absence of β-AR stimulation, hypoxia has little effect on the AP (Figures (Figures2,2, ,5,5, and and6)6) even when we modeled the effects of hypoxia on INa-L and INa at twice the rates reported in published studies.14–16 In the presence of β-AR stimulation in a paced cell, EADs are generated only at CL of 1000 ms when the effects of hypoxia on ICa-L are modeled alone or together with hypoxic IKs (Figure 5). Experimental results confirm the modeling results (Figure 8, Online Figure VIII, and Online Table I). Similar results were obtained for pause-induced EADs at a shorter CL of 300 ms (Figure 6). EADs quickly degenerate into spontaneous tachycardia only in a hypoxic cell in the presence of 1 nmol/L Iso when we also include the effect of Iso on IK1 (Figure 7).
Oxygen is the substrate for the production of reactive oxygen species. A rapid decrease in oxygen tension that is not energy limiting (and not ATP depleting; thus, ATP-dependent potassium current [IKATP] is not activated) is associated with a decrease in cellular reactive oxygen species and a more reduced cellular redox state.11,13,32–35 Electrophysiological effects of acute myocardial ischemia where there is complete cessation of perfusion and IKATP plays an important role have been investigated elsewhere.36 The increase in sensitivity of ICa-L to Iso during hypoxia occurs as a result of modification of thiol groups on the channel or a regulatory protein such as protein kinase A because exposing myocytes to dithiothreitol and intracellular perfusion with catalase mimic the effect of hypoxia.11,13 Our results suggest that a reduced redox state is protective with respect to cellular excitability because we could not induce EADs in a native cell during hypoxia or when modeling hypoxia alone. However, in the presence of β-AR stimulation, an increase in calcium influx through ICa-L prolongs APD and triggers EADs. The frequency of EADs is influenced by the modal gating of ICa-L. Increased ratios of channels gating in mode 2 (that occurs with β-AR stimulation) are associated with increased frequency of EADs.37
Sympathetic stimulation increases the risk of arrhythmia. β-Blockers are the only class of antiarrhythmics that have been demonstrated to decrease mortality.21 The results of this study are consistent with previously published data indicating that decreasing calcium influx through the channel or decreasing adrenergic stimulation can reduce the incidence of EADs. Ca2+/calmodulin–dependent protein kinase II inhibitory peptide can eliminate EADs and ventricular tachycardia,38 as can protein kinase A inhibitors and β-AR antagonists.21,39–41 We conclude that ICa-L is the primary initiator of EADs and spontaneous tachycardia occurs during hypoxia as a result of increased sensitivity of the channel to β-AR stimulation.
Sources of Funding
This study was supported by National Health and Medical Research Council of Australia grants 404002 and 513726 (to L.H.) and National Heart, Lung, and Blood Institute (NIH) grants R37-HL33343 and R01-HL49054 (to Y.R.). Y.R. is the Fred Saigh Distinguished Professor at Washington University. L.H. is recipient of a National Health and Medical Research Council of Australia Career Development Award.