Cardiac I_K1 Underlies Early Action Potential Shortening During Hypoxia in the Mouse Heart

Abstract

It is established that prolonged hypoxia leads to activation of K_ATP channels and action potential (AP) shortening, but the mechanisms behind the early phase of metabolic stress remain controversial. Under normal conditions I_K1 channels are constitutively active while K_ATP channels are closed. Therefore, early changes in I_K1 may underlie early AP shortening. This hypothesis was tested using transgenic mice with suppressed I_K1 (AAA-TG). In isolated AAA-TG hearts AP shortening was delayed by ∼24 s compared to WT hearts. In WT ventricular myocytes, blocking oxidative phosphorylation with 1 mM cyanide (CN; 28°C) led to a 29% decrease in APD90 within ∼3-5 min. The effect of CN was reversed by application of 100 μM Ba²⁺, a selective blocker of I_K1, but not by 10 μM glybenclamide, a selective blocker of K_ATP channels. Accordingly, voltage-clamp experiments revealed that both CN and true hypoxia lead to early activation of I_K1. In AAA-TG myocytes, neither CN nor glybenclamide or Ba²⁺ had any effect on AP. Further experiments showed that buffering of intracellular Ca²⁺ with 20 mM BAPTA prevented I_K1 activation by CN, although CN still caused a 54% increase in I_K1 in a Ca²⁺-free bath solution. Importantly, both (i) 20 μM ruthenium red, a selective inhibitor of SR Ca²⁺-release, and (ii) depleting SR by application of 10 μM ryanodine+1 mM caffeine, abolished the activation of I_K1 by CN. The above data strongly argue that in the mouse heart I_K1, not K_ATP, channels are responsible for the early AP shortening during hypoxia.

1. Introduction

It is firmly established that K_ATP channels [1] are critical sensory and regulatory elements in the response of the heart to various metabolic challenges, including ischemia and hypoxia. According to a widely accepted paradigm, K_ATP-mediated cardioprotection results from a reduced calcium load due to shortening of the action potential (AP) caused by K_ATP channel activation. Nevertheless, whereas the K_ATP-mediated cardioprotection is beneficial, the resulting AP shortening is generally proarrhythmic. Consistent with this, it has been found that activation of K_ATP channels by K⁺ channel openers is also proarrhythmic [2].

However, under normal conditions, cardiac I_K1, not K_ATP channels, dominate the K⁺ conductance near the resting membrane potential. Therefore, any variations of I_K1 during an initial period of metabolic inhibition would be dominant over that produced by K_ATP channels which activate after a significant delay. Consistently, several studies strongly support the view of I_K1 being a critical player during the early phase of response to hypoxia. For example, Petrich et al (1991) [3] have shown that in the rabbit heart, hypoxia-induced early AP shortening and speeding up of the late phase of repolarization is sensitive to low concentrations of Ba²⁺, a specific blocker of I_K1. The early increase in I_K1, preceding the activation of K_ATP channels, was also observed in guinea-pig cardiomyocytes during metabolic inhibition with cyanide (CN), a blocker of oxidative phosphorylation [4]. An increase in I_K1 is also associated with atrial fibrillation [5], where the metabolic status of cardiac myocytes is probably also affected. The latest data also strongly implicate activation of Kir2.1 channels, the major isoform underlying I_K1 in cardiac myocytes, in response to hypoxia in rabbit arterial smooth muscle cells [6].

In this study, we take advantage of transgenic mice with suppressed I_K1 (AAA-TG) in order to investigate the early effects of CN and true hypoxia on the activity of I_K1 at the level of the whole heart and isolated cardiac myocytes. Our data convincingly show that in the mouse heart the early AP shortening during hypoxia is mediated by I_K1, not K_ATP channels, thus pointing to activation of I_K1 as the primary fast component of electrical response to metabolic stress.

2. Methods

2.1. Expression of Kir2.1 subunits in the mouse heart

Transgenic mice (TG) with suppressed I_K1 in the heart were produced as described previously [7]. Briefly, dominant-negative Kir2.1-AAA channel subunits were fused with Green Fluorescent Protein (GFP) at the C terminus and expressed in the heart under control of the α-myosin heavy chain promoter to produce AAA-TG mice. For whole-heart experiments AAA-TG mice were identified by PCR analysis of genomic tail DNA using primers corresponding to the sequence of GFP. Specifically, in whole-heart experiments we used a transgenic line where ∼95% of cardiac myocytes express a dominant negative transgene, assuring a global suppression of I_K1 throughout the heart (line 2 in ref [7]). In single-cell experiments, myocytes were isolated from both available AAA-TG lines, and cells expressing Kir2.1-AAA subunits were identified by GFP fluorescence [7].

2.2. Composition of solutions

Krebs–Henseleit bicarbonate (KHB) solution (mM)

NaHCO₃ 25, NaCl 118, KCl 4.7, MgSO₄ 1.2, NaH₂PO₄ 1.2, CaCl₂ 2.5, EDTA 0.5, Glucose 10, with pH adjusted to 7.4 using HCl.

Tyrode solution (mM)

NaCl 137, KCl 5.4, MgCl₂ 0.5, NaH₂PO₄ 0.16, NaHCO₃ 3, HEPES 5, Glucose 10, CaCl₂ 1, with pH adjusted to 7.4 with NaOH.

Extracellular solution for Ca²⁺ current (I_Ca) recordings (mM)

tetraethylammonium chloride 137, MgCl₂ 1, glucose 10, CaCl₂ 2, HEPES 10, with pH adjusted to 7.4 using NaOH.

Intracellular solution for I_Ca recordings (mM)

CsCl 111, tetraethylammonium chloride 20, glucose 10, EGTA 14, HEPES 10, MgATP 5, with pH adjusted to 7.2 using CsOH.

Extracellular solution for Na⁺/Ca²⁺ exchange current (I_NCX) recordings (mM)

NaCl 140, CsCl 4, CaCl₂ 2.5, MgCl₂ 1.2, glucose 10, HEPES 5, with pH adjusted to 7.4 using NaOH. Solution was supplemented with 10 μM nicardipine and 10 μM ouabain in order to inhibit I_Ca and Na⁺/K⁺ ATPase, respectively. The putative I_NCX was calculated as a difference between the current measured in the presence of extracellular Ca²⁺ and that measured in the presence of 5 mM Ni²⁺/ 0 Ca²⁺.

Intracellular solution for I_NCX recordings (mM)

CsCl 110; NaCl 20; HEPES 10; MgCl₂ 0.4; glucose 5; tetraethylammonium chloride 20; EGTA 5; MgATP 4; CaCl₂ 1, with pH adjusted to 7.2 using CsOH.

Stock solutions of glybenclamide (10 mM), pinacidil (100 mM) and nicardipine (10 mM) were prepared in dimethylsulfoxide (DMSO). A 10 mM stock solution of oubain was prepared in H₂O. All stocks were kept frozen at −20 °C. A 2 M Sodium Cyanide (CN) stock solution in H₂O was prepared fresh every four days and kept at +4 °C. pH of CN-containing solutions was adjusted to 7.35 with HCl just before the experiment.

KINT (mM)

KCl 130, EGTA 2, CaCl₂ 0.2, HEPES 10, K₂ATP 5, with pH adjusted to 7.35 with KOH.

2.3. Isolated heart perfusion and MAP recordings

Mice were anesthetized with Avertin as previously reported [7]. Hearts were rapidly excised, rinsed in Tyrode solution and immediately perfused with the same solution for 1–2 min at room temperature. Hearts were then mounted on a modified perfusion apparatus (Kent Scientific, USA) for retrograde aortic perfusion with oxygenated KHB solution (95% O₂+5% CO₂) at a constant pressure of 75–85 cm H₂O (37 °C). The PO₂ in the solution was measured using an automated blood gas analyzer (IL 1640, Instrumentation Laboratory, Warrington) and was above 580 mmHg in all experiments. The AV node was ablated by applying a gentle pinch using forceps, and hearts were then paced at a basic cycle length of 150 ms with pacing electrodes placed at the base of the left ventricle. Hearts were allowed to stabilize for 10-20 min. Hypoxia was induced by switching to a KHB solution equilibrated with nitrogen (95% N₂+5% CO₂).

Monophasic action potentials (MAPs) from beating hearts were recorded using an 0.25 mm (diameter) silver wire insulated with Teflon sleeves by gently placing it on the left ventricle [8]. The signals were recorded using a DP-304 differential amplifier (Warner Instruments, USA), Digidata 1322A and Clampex 8.2 software (Molecular Devices, USA). Signals were filtered at 0.1-1 kHz, sampled at 3 kHz, and analyzed off-line using Clampfit 8.2 (Molecular Devices, USA) and Microsoft Excel, 2003.

Quantification of MAP shortening was carried out at the level of 75% (MAPD75) rather than 90%, as in single cell experiments. One of the reasons is that in this type of experiment the magnitude of the observed changes and the reliability of measurements are most optimal with MAPD75. Although, the magnitude of MAPD changes is obviously increased at 90%, and even further at 95%, the well-known ‘motion artifact’ [9], characteristic for MAP recordings, makes MAPD measurements at these levels less reliable. Additionally, due to a highly ‘triangular’ shape of the mouse AP, and the fact that I_K1 contributes primarily to the late (shallow) phase of the AP, MAPD changes at lower levels would be diminished and eventually lost, for example, at MAPD50.

2.4. Preparation of isolated cardiomyocytes and patch-clamp recordings

Cardiomyocytes were isolated from the left ventricular wall of the heart excised from 2–5 month old mice using collagenase treatment as described previously [7]. AP and whole-cell currents from single isolated cardiomyocytes were recorded in current-clamp or voltage-clamp configuration, respectively, using Axopatch 200B amplifier (Molecular Devices, USA). Signals were digitized with Digidata 1322A and Clampex 8.2 software, and analyzed off-line using Clampfit 8.2 (Molecular Devices, USA) and Microsoft Excel, 2003.

Series resistance compensation was routinely set above 80% to produce an access resistance less than 2 MΩ. Composition of solutions for patch-clamp recordings can be found above. All patch-clamp recordings were obtained at 28 °C.

For recordings of APs and I_K1 myocytes were bathed in Tyrode solution modified as indicated in the text, and KINT served as a pipette solution.

I_K1 was measured as a Ba²⁺-sensitive current in response to a 4 sec long voltage ramp from −100 mV to + 10 mV (reduced voltage range is presented in figures). Holding potential was set to −75 mV.

I_Ca was recorded in the following way: Membrane potential was first stepped from a −50 mV holding potential to −75 mV for 100 ms (to remove Ca²⁺ channel inactivation), then returned to the holding potential for 20 ms (to inactivate Na⁺ channels), and then stepped to a series of membrane potentials between −50 and +50 mV for 250 ms. Currents were recorded in the presence of 2 mM extracellular Ca²⁺, and the current-voltage relationship for I_Ca was then calculated as a difference between the peak current and the current remaining at the end of the voltage step.

I_NCX was measured as a Ni-sensitive current at the end of 300 ms long voltage steps from a −40 mV holding potential to potentials between −80 mV and +80 mV.

The density of K_ATP channels and their sensitivity to ATP were measured using inside-out patches at a +50 mV membrane potential. KINT served as both the pipette and bath solution. In patches, membrane capacitance cannot be accurately measured, therefore the K_ATP channel density is calculated per patch (not per pF). In contrast to whole-cell measurements, experiments using inside-out patches (i) are fast, allowing to test a great number of myocytes, (ii) permit measurements of ATP sensitivity, and (iii) permit recordings of IK_ATP which are not compromised by voltage-clamp errors caused by an enormous conductance of fully activated channels.

2.5. Application of hypoxic solution in patch-clamp experiments

In order to minimize the contamination of the perfusing hypoxic solution with atmospheric oxygen in patch-clamp experiments several precautions have been taken.

First, we used a glass syringe as the main container for Tyrode solution equilibrated with 100% nitrogen gas. Solution was bubbled continuously for >20 minutes at room temperature and pH adjusted to ∼7.3. Solution was then quickly loaded into the glass syringe just before the experiment, and a 1/4″ of mineral oil poured over the top of the solution to prevent any contact with atmospheric oxygen. Second, since the flow rate of solution in the patch-clamp experiment is rather slow, and the distance between the main solution container and perfusion chamber is quite long (∼2′), even a connection with a teflon tube may potentially be insufficient to prevent contamination with atmospheric oxygen. Thus, we used a second larger outer teflon tube ventilated continuously at a high flow rate with 100% nitrogen to further disrupt contact with air. Solution then entered an in-line heater mounted on the stage of the microscope near the perfusion chamber. The interior of the heater is made of stainless steel and thus the possibility of contamination with air is excluded. Third, in order to avoid contact of the solution with the air right at the chamber, myocytes, voltage-clamped in the whole-cell configuration on the floor of the chamber, were lifted up and moved into the center of the laminar stream of O₂-free solution.

The total length of teflon tube in direct contact with the air is ∼1-1.5″, including the last ½″ segment. Experimental results show that the level of O₂ depletion is sufficient to cause metabolic inhibition comparable to that produced by 1 mM CN.

2.6. Data analysis

The data are represented as a mean±S.E. (Standard Error). Statistical significance was estimated using a two-tailed t-test with equal variances. *, ** and *** indicate p values <0.05, <0.01 and <0.005, respectively.

3. Results

3.1. Early MAP shortening in response to hypoxia is abolished in TG hearts lacking I_K1

The effects of hypoxia, mimicked by application of an O₂-free solution, were first studied in Langendorff perfused hearts from WT and AAA-TG mice using MAP recordings (Fig. 1). In WT hearts, the response to hypoxia displays two distinct components. Early, presumably I_K1-dependent, MAP shortening occurred nearly concurrently with the replacement of O₂ by N₂ in the perfusate (Fig. 1A; 2). This was then followed by a more rapid, presumably K_ATP-dependent phase (Fig. 1A; 3), which led to short ‘neuron-like’ MAPs (Fig. 1A; 4). This biphasic MAP shortening was observed in all four WT hearts tested. In contrast, only monophasic MAP shortening was observed in AAA-TG hearts (n=4). The early phase was absent, and only the second markedly rapid MAP shortening (Fig. 1A; 3) occurred at the same time as the rapid phase of MAP shortening observed in WT hearts.

Fig. 1. Early MAP shorting induced by hypoxia is abolished in AAA-TG mice.

A. A representative time course of MAP duration during hypoxia. In WT hearts, removal of O₂ from the perfusate (application of N₂; ) leads to an early, presumably I_K1-dependent, abbreviation of MAP (2; ) followed by a further rapid MAP shortening (3; ), presumably due to the activation of K_ATP channels. The time for early MAP shortening was defined as the time when MAPD75 falls by ∼10% of the Ctrl value measured just before application of N₂ (1; ). The early phase of MAP shortening is absent in AAA-TG hearts, and only the rapid phase is observed, which starts at the same time as the K_ATP-dependent phase of MAP shortening in WT hearts (3; ). B. Representative MAPs from WT (gray) and AAA-TG (black) hearts recorded at the times as indicated in A. MAP amplitudes were normalized to each other to highlight the differences in shape. C, D. MAPD75 measured at the times as indicated in A. At the time when the early MAP shortening has just begun in AAA-TG hearts (3), in WT hearts MAPD75 has been already reduced by 28%. Statistical significance is indicated compared to MAPD75 before application of N₂ (1). D. The time delay (D_t) between the application of N₂ and the time when MAPD75 is decreased by 10% in WT and AAA-TG hearts. n=4 for both WT AAA-TG hearts.

Fig. 2. CN-induced early AP shortening is abolished in isolated myocytes with suppressed I_K1.

Examples of APs in WT (A) and AAA-TG (B) myocytes. A. In WT cells, 10 μM glybenclamide (Gb) does not significantly affect AP. Application of 1 mM CN in the presence of Gb leads to AP shortening within ∼3 min. Further addition of 100 μM Ba²⁺ completely reverses the effect of CN. (Insert) AP shortening caused by 100 μM pinacidil is not blocked by 100 μM Ba²⁺. B. Application of CN, Gb or Ba²⁺ exerts no effect on the AP in AAA-TG myocytes within the same period of time (∼3-5 min). C. Summarized data show that in WT myocytes 10 μM Gb has virtually no effect on APD90 while 1 mM CN reduces APD90 by nearly 30% within 3 min (n=10). Further application of 100 μM Ba²⁺ increases APD90 by ∼17% above the control level (n=10). In AAA-TG myocytes with suppressed I_K1, CN (3 min application), Gb or Ba²⁺ have no affect on APD90 (n=7). For each of the three conditions, a relative change in APD90 values is presented; therefore a dashed line at 100% represents control data.

Fig. 3. Time course of the CN-induced current at −60 mV in WT myocytes.

An example of the time course of the whole-cell current amplitude measured at a −60 mV membrane potential during application of 1 mM CN at 28 °C. (Insert) −60 mV roughly corresponds to the peak of outward I_K1. (1) Currents are stable for a prolonged period of time until the application of CN. (2) CN leads to a quick increase in current amplitude which remains quasi-stable until (3) a sharp increase after >15 min due to activation of K_ATP channels.

Fig. 4. Application of CN leads to early upregulation of I_K1 in WT myocytes.

A and B. Representative Ba²⁺-sensitive (A) and Ba²⁺-insensitive (B) currents recorded before and after a 3 min application of 1 mM CN at 28 °C. In A, the dotted trace represents a difference, or CN-activated, current. Currents were recorded in response to a 4 s voltage ramp from −100 mV to 10 mV (reduced voltage range is shown) in the absence and presence of 100 μM Ba²⁺. C. Current densities at −60 mV. Ba²⁺-sensitive currents are increased by ∼50%, and Ba²⁺-insensitive currents are not affected by CN treatment (n=11, p<0.05 vs Ctrl, respectively).

As we have shown previously, suppression of I_K1 leads to a significant prolongation of AP in AAA-TG hearts [7]. Here, under control conditions, MAPD75 in AAA-TG hearts was larger by 44% when compared to WT hearts: 49.4±4.1 ms and 34.3±2.4 ms, respectively (p<0.005, n=4, (Fig. 1 C&D). Upon application of hypoxic solution, in WT hearts a 10% decrease in MAPD75 was observed 30.0±5.4 s after application of N₂, while in AAA-TG hearts the MAPD75 decrease was significantly delayed to 53.8±7.5 s (p<0.05, n=4, Fig. 1E). This ∼24 s delay led to a situation such that at the time when the MAP shortening had just begun in AAA-TG hearts, MAPD75 has been already decreased by 28% in WT hearts (Fig. 1; comp. C and D).

Once the quick late phase of MAP shortening began, the rates of rapid decrease were not different between WT and AAA-TG hearts. After 150 s of hypoxia, MAPD75 reached small steady state values: 11.3±0.5 ms and 17.5±4.2 ms in WT and AAA-TG hearts, respectively (n=4 each; not statistically different). The above results strongly support the hypothesis that I_K1 plays a significant role in the early stages of electrical response to metabolic stress.

3.2. Suppression of I_K1 abolishes CN-induced early AP shortening in isolated myocytes

To further address the above hypothesis, the effects of metabolic stress on AP were studied in isolated myocytes from WT and AAA-TG hearts. Oxidative phosphorylation was blocked with 1mM CN in the presence of 10 mM glucose, and APs recorded using the current-clamp technique (T=28°C). For recordings of AP smaller pipettes were used allowing for higher quality gigaseals and minimal depolarization of membrane potential in AAA-TG myocytes essentially lacking stabilizing I_K1 [7]. Specifically, the resting potentials were −72.7±0.4 (n=8) and −71.6±0.5 (n=7) in WT and AAA-TG myocytes, respectively (not significantly different; p>0.1). As shown in Fig. 2A, application of 10 μM glybenclamide, a potent blocker of K_ATP channels, did not significantly affect AP in WT myocytes, suggesting that K_ATP channels do not contribute to AP repolarization, and that the metabolic status of the myocytes is not affected by the isolation procedure. Addition of 1 mM CN caused significant AP shortening within 3-5 min, primarily in the range of membrane potentials negative to −40 mV (Fig. 2A), which corresponds to a slow phase of repolarization where the contribution of I_K1 is most significant. Specifically, APD90 was decreased by 29.1±3.5% of the initial value (p<0.005, n=10, Fig. 2C). Further addition of 100 μM Ba²⁺, a specific blocker of I_K1, not only completely reversed the effect of CN (Fig. 2A), but also increased APD90 by 16.9±8.3% compared to control (p<0.05, n=10, Fig. 2C). As expected, prolonged application of CN (>15-30 min) ultimately led to a sudden dramatic shortening of the AP followed by a complete loss of excitability due to activation of K_ATP channels (not shown).

To exclude the possibility of K_ATP channel contribution to early AP shortening, we took advantage of AAA-TG mice with suppressed I_K1. In AAA-TG myocytes, CN had no effect on the AP within 5 min of treatment (n=7, Fig. 2B,C), and neither glybenclamide nor Ba²⁺ affected the AP within the same period of time (5 min). As with WT myocytes, in AAA-TG cells, AP shortening caused by the opening of K_ATP channels ultimately occurred after 15-30 min of treatment with CN (not shown).

3.3. The early K⁺ current underlying CN-induced AP shortening is I_K1

All together, the above data show that the 29% reduction in APD90 induced by CN (Fig. 2A,C) is due to the activation of I_K1. To directly test this hypothesis we have characterized the properties of the early CN-activated current in the range of hyperpolarized potentials where the AP is primarily affected, i.e. negative to −40 mV (Fig. 2).

Fig. 3 shows a typical time course of the development of the whole-cell current at −60 mV (a membrane potential close to the peak of outward I_K1) during CN treatment. Application of 1 mM CN led to a fast current increase, reaching a quasi-steady state level within 3-5 min at 28 °C (Fig. 3; phase 2). After that time, currents remained stable for several minutes until a sudden manifold increase in conductance due to activation of K_ATP channels occurred (Fig. 3; phase 3). While in all experiments the early effects of CN (phase 2) occurred within 3-5 minutes, the time delay before the activation of K_ATP channels varied significantly, ranging from 15 min to more than 30 min (was not quantified). In the initial control experiments in the absence of CN, currents remained stable for an extended period of time (>30 min).

It should be noted that the time course of response to metabolic inhibition was faster in experiments with isolated hearts (Fig. 1A) than that observed in isolated cells (Fig. 3). We think that the differences in (a) temperature − 37 °C vs 28 °C, and (b) pacing rates − 150 ms BCL vs none, in isolated hearts and single cells, respectively, contributed most to the observed discrepancy.

The relative stability of phase 2 provided us with the opportunity to characterize the early CN-activated current in more detail (Fig. 4). Currents were recorded in response to a slow voltage ramp in the absence and presence of 100 μM Ba²⁺. Fig. 4A shows that application of 1 mM CN for ∼3 min leads to an increase in Ba²⁺-sensitive current in both the inward and outward direction, negative and positive to the reversal potential, respectively. At −60 mV the amplitude of outward current was increased from 0.74±0.07 pA/pF to 1.11±0.08 pA/pF (p<0.05, n=11, Fig. 4C), a 50% increase. In contrast, no significant changes were observed in the Ba²⁺-insensitive component of the CN-activated current (Fig. 4B&C). Importantly, the Ba²⁺-sensitive difference current displayed strong inward rectification, undoubtedly confirming it to be I_K1.

Application of CN had no effect on currents in transgenic AAA-TG myocytes (Fig. 4C). We have shown previously that in AAA-TG myocytes (identified using the intense green fluorescence of Kir2.1-AAA-GFP transgene) I_K1 is suppressed >95% and, therefore, small remaining currents consist primarily of minor leak and other minor ion channel or pump currents (not characterized) [7]. Nevertheless, we have further confirmed essential suppression of I_K1 in the current study. Currents measured in AAA-TG myocytes at −60 mV before and after application of 100 μM Ba²⁺ were indistinguishable: −0.59±0.04 pA/pF and −0.56±0.04 pA/pF, respectively (n=6).

Similar to experiments with AP, K_ATP channels were finally activated after >15 min of CN perfusion in both WT and AAA-TG myocytes (n>10, for each type), leading to a linearized current-voltage relationship with high slope conductance (not shown).

3.4. I_K1 is activated during true hypoxia

To exclude the possibility that the early activation of I_K1 is simply due to peculiar effects of CN, we have carried out experiments using true hypoxic conditions. Specifically, whole-cell currents were measured in WT myocytes during application of nominally O₂-free (N₂ saturated) solution in the presence of 10 mM glucose (see Methods). Similar to the effect of CN, hypoxia also activated a strongly inwardly rectifying current (Fig. 5), although the time for the current amplitude to reach a quasi steady state level was somewhat delayed compared to CN treatment (4-6 min vs 3-5 min, respectively). Specifically, at −60 mV the Ba²⁺-sensitive current (I_K1) was increased from 0.56±0.06 pA/pF to 0.89±0.12 pA/pF, a nearly 60% increase (p<0.05, n=4, Fig. 5), and no significant change was observed in Ba²⁺-insensitive currents (not shown). Again, as expected, prolonged (15-30 min) perfusion with O₂-free solution ultimately led to activation of K_ATP channels (not shown). Thus the data strongly support the hypothesis that early AP shortening originally observed in isolated hearts (Fig. 1) is indeed due to an increase in I_K1.

Fig. 5. True hypoxia leads to early activation of I_K1.

A. Representative Ba²⁺-sensitive current traces recorded in a WT myocyte before and after a 4 min application of nominally O₂-free solution (N₂) at 28 °C. The current activated by hypoxia displays strong inward rectification, confirming it is I_K1. B. At −60 mV the Ba²⁺-sensitive current is increased by 60% when compared to control (n=4, p<0.05).

3.5. K_ATP channels are not involved in early AP shortening

In WT myocytes CN-induced early AP shortening might still be caused by activation of K_ATP channels, which can potentially be affected at the concentration of Ba²⁺ used to block I_K1 in the above experiments. The data in Fig. 2A (insert) exclude this possibility since 100 μM Ba²⁺ does not reverse AP shortening caused by 100 μM pinacidil, a potent K_ATP channel opener (p<0.05, n=6, Fig. 2A; insert).

Although experiments with AAA-TG myocytes provide the most compelling evidence that I_K1, not K_ATP channels, underlie early AP shortening, the conclusions are based on the assumption that K_ATP channels are not affected by transgenic manipulation of I_K1. For example, it may be argued that overexpression of mutant Kir2.1-AAA subunits in AAA-TG hearts may significantly downregulate native K_ATP channels by some unknown mechanism, thus explaining a lack of the effect of CN in AAA-TG myocytes. The data in Fig. 6 argue against this possibility. First, we compared the density of K_ATP channels using inside-out patches excised from WT and AAA-TG myocytes. Although a 25% decrease in K_ATP channel density was indeed observed in AAA-TG myocytes compared to WT cells (p<0.01; n=91 and n=57, respectively), the effect was clearly too small to account for the inefficiency of CN to cause early AP shortening in these cells. Additionally, the sensitivity of K_ATP channels to ATP (K_1/2) was not affected in AAA-TG myocytes (Fig. 6): 37.8±1.7 μM (n=91) and 38±2.4 μM (n=57) in WT and AAA-TG myocytes, respectively. Therefore, the data strongly suggest that remodeling of K_ATP channels does not underlie the differential response of WT and AAA-TG myocytes to CN application.

Fig. 6. K_ATP channels in WT and AAA-TG myocytes.

K⁺ currents were measured at +50 mV in high symmetrical K⁺ using excised membrane patches, and K_ATP current amplitudes calculated as a difference between currents in 0 and 1 mM ATP. A. In AAA-TG myocytes, K_ATP current density (per patch) is reduced by ∼25% compared to that in WT cells (p<0.01). B. A representative example of an ATP dose-response relationship in WT cells obtained under the same conditions as above. C. Normalized IK_ATP (I_REL) from individual patches were fit with a Hill equation (insert) to estimate the slope (h) and concentration of half maximum block (K_1/2). The data at each concentration show only a minor variation, thus standard error (SE) bars are too small compared to symbols. Neither the Hill coefficient (h) nor the sensitivity to ATP (K_1/2) are affected in any significant way in AAA-TG myocytes, therefore, for visualization purpose, the averaged data were fit again with the Hill equation and the fit is plotted as indicated (continuous line). n=91 and n=57 for WT and AAA-TG cells, respectively, in all experiments.

3.6 I_Ca and I_NCX are not changed in AAA-TG myocytes

I_Ca and I_NCX also contribute to AP repolarization. In order to check whether the remodeling of these currents is responsible for a lack of AP shortening in AAA-TG myocytes, we compared the current densities of I_Ca and I_NCX in WT and AAA-TG myocytes.

As shown in Fig. 7A, neither the peak amplitude of I_Ca nor its voltage dependence are affected in any significant way in AAA-TG myocytes, thus ruling out the contribution of I_Ca to early AP shortening.

Fig. 7. I_Ca and I_NCX in WT and AAA-TG myocytes.

A. (1) Representative Ca²⁺ currents in WT myocytes in response to a voltage protocol shown in (2). (3) Current-voltage relationships for peak I_Ca in WT and AAA-TG myocytes. There is no significant difference between the two groups (n=11, n=10, respectively). B. (1) Representative background currents in WT myocytes in response to voltage steps to potentials between −80 mV and +80 mV. (2) and (3) Current-voltage relationships for total, Ni²⁺-insensitive and Ni²⁺-sensitive background currents in WT and AAA-TG myocytes measured at the end of voltage steps. There is no significant difference between any corresponding currents in the two groups (n=14, n=9, respectively). See Methods for solution composition.

Similarly, the data in Fig. 7B show that both the Ni²⁺-sensitive current (putative NCX), and the total current recorded in the presence of extracellular Ca²⁺ are also not significantly different between WT (n=14) and AAA-TG (n=9) myocytes. Therefore, the contribution of NCX to early AP shortening can also be excluded.

3.7 SR Ca²⁺ release is involved in the upregulation of I_K1 by hypoxia

It has been shown that a significant increase in intracellular Ca²⁺ (Ca²⁺_i) during hypoxia may occur rather early, well before myocyte shortening [1]. Therefore, we sought to determine whether the early activation of I_K1 is linked to changes in Ca²⁺_i.

Since the Ba²⁺-insensitive current is not affected by CN (Fig. 4), changes in the total current represent changes in I_K1. Therefore, to quantify the effect of CN we measured only the total outward current density at −60 mV.

As shown in Fig. 8, in the presence of 2 mM EGTA and 200 μM Ca²⁺ (pCa∼8.0) in the pipette solution, the total outward currents were increased from 0.33±0.05 pA/pF to 0.54±0.08 pA/pF (p<0.05, n=11). Elevating Ca²⁺ pCa∼5.0 did not significantly influence the effect of CN on total currents: 0.23±0.06 pA/pF and 0.60±0.10 pA/pF (p<0.05, n=4) before and after CN application, respectively. In contrast, buffering of Ca²⁺_i with 20 mM BAPTA (nominally zero Ca²⁺; estimated pCa_i>12) prevented CN-induced activation of I_K1 (n=6), strongly suggesting that Ca²⁺_i signaling is involved in the activation of I_K1 by hypoxia.

Fig. 8. SR Ca²⁺ release is involved in the activation of I_K1 during hypoxia.

A. Strong intracellular Ca²⁺ buffering abolishes activation of I_K1 by CN. Current densities at −60 mV were measured before and after application of 1 mM CN (∼3 min) using different concentrations of Ca²⁺ and types of Ca²⁺ buffer in the pipette solution (n=4, 11 and 6 from left to right). pCa_i values were calculated using the following concentrations of Ca²⁺-binding components and Ca²⁺. pCa²⁺_i ∼5 − 5 mM ATP, 2 mM EGTA 190 μM Ca²⁺; pCa²⁺_i ∼8 − 5 mM ATP, 170 μM Ca²⁺; pCa²⁺_i >12 − 5 mM ATP, 20 mM BAPTA, 0.1 μM Ca²⁺; pH 7.3, T=25 °C. Calculations were performed using WinMaxc programs, http://www.stanford.edu/∼cpatton/maxc.html. B. Removal of extracellular Ca²⁺_o does not affect the activation of I_K1 by CN (n=6). In contrast, depletion of Ca²⁺ stores by extracellular application of 10 mM caffeine (Caff) plus 10 μM ryanodine (Rya) or the inhibition of Ca²⁺ release by 10 μM ruthenium red (RR; included into pipette solution) abolishes I_K1 activation (n=7 and 5, respectively).

We then investigated what the potential source of Ca²⁺_i could be. In the absence of Ca²⁺ in the bath solution (containing 0.5 mM EGTA), CN still caused an increase in the total currents from 0.27±0.04 pA/pF to 0.58±0.05 pA/pF (Fig. 8B, p<0.05, n=6), further excluding I_Ca in the early activation of I_K1. Surprisingly, the effect of CN was abolished after the myocytes were briefly treated (2-3 min) with 10 mM caffeine and 10 μM ryanodine to deplete Ca²⁺_i stores (Fig. 8B, n=7). Similarly, inclusion of 10 μM ruthenium red, an inhibitor of Ca²⁺ release, in the pipette solution also blocked the increase of I_K1 by CN (Fig. 8B, n=5), strongly arguing that SR Ca²⁺ release is involved in the activation of I_K1 by hypoxia.

4. Discussion

This study uncovers the mechanism of early AP shortening during metabolic stress. Specifically, the data clearly show that in the mouse heart early AP shortening induced by CN poisoning or true hypoxia is mediated by a significant increase in I_K1, preceding the activation of K_ATP channels.

4.1. Can I_K1 compete with K_ATP?

Since the discovery of an ATP-dependent K⁺ channel by Noma (1983) [1], a wealth of experimental data has been collected which undoubtedly implicate these channels as critical players in the response of cardiac muscle to metabolic challenges [10]. The power of K_ATP channel activation originates from a high density of these channels in the sarcolemmal membrane, such that the opening of less than 1% of them is sufficient to significantly abbreviate the AP [11], while their full activation leads to complete loss of excitability.

Importantly, 1% of K_ATP activation mentioned above corresponds to only a few nS of conductance [11], while the conductance of I_K1, a major constitutively active K⁺ conductance, is estimated to be about 10-15 nS (at the resting membrane potential; mouse myocytes [12]). Thus, potential changes in I_K1 during metabolic stress may lead to effects comparable to what is considered significant in the K_ATP arena, and would even dominate over K_ATP channels during the early phase of stress.

4.2. Previous evidence for I_K1 activation by hypoxia

There are several studies which point to the involvement of I_K1 in the electrical response of the heart to metabolic stress, but the biggest challenge in all of them was to differentiate between the effects produced by K_ATP and other ion channels. For example, Ruiz-Petrich (1991) [3] showed that in the rabbit heart the acceleration of final AP repolarization during hypoxia is Ba²⁺ sensitive, implicating activation of I_K1. A number of follow-up studies also suggest that channels other than K_ATP may contribute to early AP shortening [13-15]. Specific conclusions, however, were largely based on the efficiency of K_ATP specific agents, such as glybenclamide, to block AP shortening caused by metabolic inhibition. The latter should be taken with caution, including this study, since it has been long established that the efficacy of glybenclamide, and other relevant drugs, may be strongly compromised by prolonged metabolic inhibition [16].

Nevertheless, direct evidence for I_K1 activation by hypoxia are available. For example, Muramatsu et al (1990) [4] showed using whole-cell current measurements in guinea-pig myocytes that CN poisoning leads to early activation of a K⁺ current, long preceding K_ATP channel activation, which displays strong inward rectification similar to I_K1. Also, Park et al (2005) [6] recently showed that in rabbit arterial smooth muscle cells, 2 min of hypoxia activated Kir2.1 channels, which are believed to be the major isoform underlying I_K1 in mouse cardiac myocytes.

4.3. Activation of I_K1 underlies early AP shortening during hypoxia

In this study we took advantage of TG mice with suppressed I_K1 to overcome a number of difficulties related to the limited efficiency of a pharmacological approach. The original support for early activation of I_K1 comes from experiments with isolated hearts. A delayed shortening of the MAP in AAA-TG hearts lacking I_K1 during hypoxia is strongly indicative of early activation of I_K1 in WT hearts (Fig. 1), but leaves a lot of room for other explanations.

On the contrary, results of experiments with isolated cells are strongly conclusive.

The first result is that the early AP shortening induced by CN poisoning in WT myocytes is completely eliminated in AAA-TG myocytes (Fig. 2B).

Second, glybenclamide could not prevent early AP shortening, while Ba²⁺ was efficient in doing so. Therefore, the early activation of K_ATP channels can be largely ruled out. However, as mentioned above, the efficiency of glybenclamide has to be treated with caution in these particular experiments. Additionally, in this study we also show that K_ATP channel density and ATP sensitivity is essentially unaffected in AAA-TG hearts (Fig. 6).

Experiments with Ba²⁺ ions also deserve attention. It is known that K_ATP channels are significantly less sensitive to block by extracellular Ba²⁺. For example, the Kd (0 mV) value for Ba²⁺ block of the least sensitive Kir2.1 channels is 62 μM [17], while the Kd (0 mV) value for K_ATP channels is in the mM range (1.18 mM [18]; 3.7 mM [19]). Nevertheless, Ba²⁺ block of K_ATP channels possesses a stronger voltage dependence, therefore complicating the estimation of Ba²⁺ block at hyperpolarized potentials, where most of the AP shortening occurs. In order to experimentally exclude the possibility of K_ATP involvement and to avoid complex calculations of potential block, we show that pinacidil-induced AP shortening (Fig. 2A; Insert) is not affected by 100 μM Ba²⁺.

Importantly, in contrast to ischemia, the level of high energy phosphates may not be significantly affected during early stages of hypoxia. For example, Allen et al (1985) [20] showed that in the ferret heart neither ATP nor ADP levels are affected within a few minutes of hypoxia, again arguing against K_ATP channel activation.

The third, probably most compelling evidence, come from the direct measurements of an early activated K⁺ current in WT myocytes. The mere fact that the early-activated K⁺ current displays strong inward rectification undoubtedly identifies it as I_K1. In addition, this early-activated K⁺ current is strongly Ba²⁺-sensitive, with no effect of CN being observed in the Ba²⁺-insensitive component, thus further eliminating the contribution of K_ATP channels to this phenomenon (Fig. 4).

Also important, the early activated I_K1 is not the peculiar result of poisoning by CN, since the same outcome is observed in myocytes exposed to true hypoxic conditions (Fig. 5).

It should be noted that since the experiments with isolated myocytes were carried out in the whole-cell configuration (T=28 °C), allowing for partial intracellular dialysis of small molecules, the magnitudes of AP shortening and I_K1 activation may be different from that occurring in intact hearts. Nevertheless, even intracellular dialysis with solutions containing moderate concentrations of common buffers such as 2 mM EGTA and 10 mM HEPES cannot prevent changes in the intracellular environment caused by CN or hypoxia, suggesting that the effects observed in isolated cells may be even larger in intact hearts.

Whether I_K1, but not K_ATP, channels are responsible for early AP shortening in other species or experimental models remains highly controversial. In particular, it has been shown that in the guinea-pig heart, selective blockade of K_ATP channels by HMR 1098 eliminates APD shortening in a low flow ischemia model, a condition similar to hypoxia in the current experiments [21]. In contrast, it has been shown that in the rabbit heart, glybenclamide had no effect on AP shortening up to 25 min of hypoxia [22]. There is more controversy with the data obtained in rodents. One of the striking findings is that in rat cardiac myocytes HMR 1098 does not block K_ATP channels or reverse AP shortening caused by metabolic inhibition (CN+ iodoacetic acid, no glucose) [23]. Irrespective of the mechanism of low efficiency of HMR 1098 in the above study, it shows that the data cannot be easily extrapolated from one species to another, or even from one specific technique to a similar one.

4.4. Are other channels or pumps involved in early AP shortening?

Most of the changes in AP occurred at potentials negative to −40 mV where I_K1 channels are most active. This strongly limits a number of other potential players, but does not completely rule them out and make I_K1 exclusive. For example, stress-induced changes in time-dependent currents, such as a Ca²⁺ current, may potentially contribute to early AP shortening in a dynamic manner. The data in Fig. 2B, however, nearly eliminate this possibility since selective inhibition of I_K1 in AAA-TG myocytes completely prevents early AP shortening.

Nevertheless, the validity of the AAA-TG model still relies on the assumption that other major contributors to AP repolarization are not affected by transgenic I_K1 manipulation. Therefore, one needs to be sure that these other conductances are intact in AAA-TG myocytes.

The data in Fig. 7A clearly show that there is no remodeling of the Ca²⁺ current in AAA-TG myocytes, further confirming that Ca²⁺ channels are not involved in early AP shortening. Time-independent currents such as that produced, for example, by the Na⁺/Ca²⁺ exchanger, or other uncharacterized conductances, can be largely excluded based on the results in Fig. 4, which show activation of only the Ba²⁺-sensitive I_K1. Additionally, the data in Fig. 7B show that neither the total time independent current nor the Ni²⁺-insensitive (Na⁺/Ca²⁺ exchanger) conductance are remodeled in the AAA-TG mouse, leaving I_K1 as the exclusive player in early AP shortening during hypoxia. Importantly, the magnitude of early AP shortening and I_K1 upregulation are in a quantitative agreement sufficient to explain most of the results.

4.5 SR Ca²⁺ release is involved in the activation of I_K1 by hypoxia

This study deals primarily with the cellular mechanism of early AP shortening, and only limited data are available with regard to the potential molecular mechanisms.

For example, while the results in Fig. 8B exclude the role of extracellular Ca²⁺ in I_K1 activation, we found that Ca²⁺_i plays a significant role. In particular, a buffering of Ca²⁺_i with 20 mM BAPTA or pharmacological inhibition of SR release completely abolished the effect of CN on I_K1 (Fig. 8), suggesting an essential role of SR in the phenomenon.

Consistent with our data, Dipp et al (2001) [24] have shown that in pulmonary smooth muscle cells hypoxia leads to a release of Ca²⁺ from the SR without affecting the Ca²⁺ influx from extracellular solution.

It may seem paradoxical that an increase in Ca²⁺_i leads to I_K1 activation since it has been found that Ca²⁺_i blocks Kir channels in a voltage dependent manner in ventricular myocytes [25]. However, the blocking concentrations of Ca²⁺_i are too high to have any physiological significance, with Ca²⁺ block being even less potent than that of Mg²⁺ [25], and therefore, other mechanisms of Ca²⁺_i-dependent regulation of I_K1 will have to be considered in future experiments.

4.6. Other potential mechanisms of I_K1 activation

It is established, for example, that the activity of Kir channels is strongly dependent on the level of membrane PIP₂ [26] and on their phosphorylation state [27], parameters which can easily be affected early in hypoxia. Nevertheless, involvement of free intracellular PAs, the major inwardly rectifying factors, seems more realistic [28]. This hypothesis is based on the facts that (i) I_K1 channels are only partially blocked by PAs, allowing for a significant up or down-regulation, and (ii) the concentration of free, active PAs is determined by the intracellular concentrations of their numerous binding partners [29]. For example, the PAs are part of an ATP-Mg-PA complex [30, 31], and inorganic phosphates are also known chelating agents for PAs. Therefore, changes in the concentration of free PAs, and thus changes in I_K1, can be as quick and versatile as that of their binding components. Consistent with this, it has been shown that in the rabbit heart, hypoxia leads to a manifold increase in the concentration of free inorganic phosphates in less than 3 minutes [20], which inevitably should affect the balance of free PAs, and thus affect I_K1 rectification, within the same fast time frame.

Intracellular pH seems to be another key parameter involved in upregulation of I_K1 by CN [32] or hypoxia, although a pH-dependent mechanism is seemingly against the common view that (i) metabolic inhibition leads to intracellular acidification, and (ii) Kir2 channels are inhibited by low intracellular pH. This contradiction, however, is superficial, since ischemia and hypoxia, especially early hypoxia, are quite different processes. For example, the fact that CN application causes large reductions in intracellular pH is not a universal fact in all animals, and is certainly not a universal fact for different conditions. For example, Allen et al (1985) [33] showed that in the ferret heart, the intracellular pH is actually transiently increased, not decreased, after only 3 min of hypoxia without any change in the level of intracellular ATP/ADP. Eisner et al (1989) [34] observed a similar phenomenon in isolated rat myocytes by mimicking hypoxia with application of CN in the presence of glucose, as in our experiments.

Interestingly, alkalinization leads to deprotonation of PAs and relief of block at depolarized potentials [35], which would be consistent with a PA-dependent mechanism of I_K1 activation. In addition, increase in pH is shown to increase binding of PA to ATP-Mg (and other nucleotides) complex [31], which would lead to a decrease in free PAs and thus increase in I_K1.

4.7. Physiological significance

The significance of I_K1 activation in arrhythmogenesis mirrors in many aspects that of K_ATP channels, although the results of this study place I_K1 as the first responder to hypoxic stress. For example, it has been shown in a number of animal models that arrhythmogenic consequences of metabolic stress, a condition known to activate K_ATP channels, can be prevented by glybenclamide, a potent K_ATP channel blocker [36, 37]. Alternatively, activation of K_ATP channels by K⁺ channel openers has been shown to have a proarrhythmic effect [2].

In line with the above, upregulation of I_K1 in TG mice also leads to multiple abnormalities in excitability [12], and provides a substrate for highly stable reentrant arrhythmias [38]. Consistent with this, it has been shown that in the guinea-pig heart, blockade of I_K1 by Ba²⁺ ions can terminate induced ventricular fibrillation [39].

In humans, a number of gain-of-function mutations in the Kir2.1 gene were recently discovered to be the cause of Short QT Syndrome (SQT3) [40] and familial atrial fibrillation [41]. Thus, early in hypoxia, upregulation of I_K1 may underlie a significant proarrhythmic potential.

Importantly, hypoxia and low-flow ischemia, but not complete ischemia, are the most common conditions experienced by the heart. In ∼90% of cases of sudden cardiac death, coronary arteries are narrowed but not completely blocked. Since even in the normally beating heart the extraction of oxygen is very high, such that 65-70% of it is removed from the blood as it passes through the heart, this may lead to frequent episodes of acute hypoxia/low-flow ischemia due to, for example, release of adrenaline during increased physical activity, which often acts as a trigger for sudden death. In fact, although ischemic conditions are clinically relevant, in most patients with an acute coronary occlusion, myocardial flow in the infarct region is not null but typically ∼15% of normal [42], a condition best mimicked by low-flow ischemia and hypoxia. Therefore, early hypoxia-induced activation of I_K1 may potentially underlie a significant number of unexplained sudden cardiac deaths. If true, understanding the role of I_K1 in excitability may be key to new ways of prevention and treatment of deadly arrhythmias, the most devastating cardiac condition in the world.

In conclusion, our studies show that in the mouse heart early AP shortening during hypoxia is mediated by activation of I_K1.