Heterogeneity and Function of K_ATP Channels in Canine Hearts

Abstract

Background

The concept that pore-forming Kir6.2 and regulatory SUR2A subunits form cardiac ATP-sensitive potassium (K_ATP) channels is challenged by recent reports that SUR1 is predominant in mouse atrial K_ATP channels.

Objective

To assess SUR subunit composition of K_ATP channels and consequence of K_ATP activation for action potential duration (APD) in dog heart.

Methods

Patch-clamp techniques were used on isolated dog cardiomyocytes to investigate K_ATP channel properties. Dynamic current-clamp, by injection of a linear K⁺ conductance to simulate activation of the native current, was employed to study consequences of K_ATP activation on APD.

Results

Metabolic inhibitor (MI)-activated current was not significantly different from pinacidil (SUR2A-specific)-activated current, and both currents were larger than diazoxide (SUR1- specific)-activated current, in both atrium and ventricle. Mean K_ATP conductance (activated by MI) did not differ significantly between chambers although, within the ventricle, both MI-induced and pinacidil-induced currents tended to decrease from epicardium to endocardium. Dynamic current-clamp results indicate that myocytes with longer baseline APDs are more susceptible to injected “K_ATP” current, a result reproduced in silico using a canine AP model to simulate Epi and Endo (HRd).

Conclusions

Even a small fraction of K_ATP activation significantly shortens APD in a manner that depends on existing heterogeneity in K_ATP current and APD.

INTRODUCTION

K_ATP channels are formed by four pore-forming subunits (Kir6.1 or Kir6.2) and four regulatory sulfonylurea receptors (SUR1, SUR2A, or SUR2B)¹. SUR subunits cause channel opening when bound to MgADP or channel openers (eg, pinacidil and diazoxide), and transduce channel inhibition by sulfonylureas (eg, glibenclamide and tolbutamide). Different combinations of expressed subunits result in qualitative and quantitative differences in functional properties in specific tissues. For example, in the pancreas, the Kir6.2/SUR1 complex regulates insulin secretion in response to glucose-dependent changes in metabolism². Conversely cardiac K_ATP channels contain Kir6.2 and SUR2A subunits³, which are less responsive to nucleotide activation than SUR1-containing channels⁴ and remain predominantly closed under physiological conditions⁵, explaining lack of cardiac abnormalities in patients with neonatal diabetes or hyperinsulinism due to Kir6.2-dependent K_ATP mutations. However, recent studies reveal that cardiac K_ATP channels in mice differ in a chamber-specific fashion: Kir6.1-containing channels may be prominent in conduction system myocytes⁶, and in particular, atrial K_ATP channels are formed by SUR1/Kir6.2 while ventricular channels contain SUR2A/Kir6.2⁷.

Even moderate K_ATP channel opening can cause dramatic action potential (AP) shortening^8–10. As a consequence of higher sensitivity to activation by intracellular nucleoside diphosphates⁴, cardiac SUR1 expression could therefore implicate K_ATP channel participation in cardiac excitation under nominally physiological conditions; recent studies have found that cardiac K_ATP channels may open during strenuous exercise¹¹ and stress¹². On one hand, AP shortening due to K_ATP channel opening may provide cardioprotection by reducing contractile force and energy consumption in response to ischemia or other stresses. On the other hand, K_ATP channel opening might predispose to arrhythmias due to AP shortening, external potassium accumulation, and dispersion of transmural repolarization, particularly if localized regions of the heart respond differentially. Because K_ATP activation depends on channel composition, differential K_ATP channel subunit expression may lead to distinct consequences of K_ATP activation throughout the heart, such as chamber-specific response to ischemia or other stresses¹³. To date, potential subunit heterogeneity in large animals is unexplored. In the present study, we performed a detailed examination of the regional heterogeneity of canine cardiac sarcolemmal K_ATP pharmacology, as well as the consequences of heterogeneous cellular responses by injecting simulated K⁺ current into native myocytes via dynamic clamp.

METHODS

Cardiomyocyte isolation

All protocols were approved by the Animal Studies Committee at Washington University School of Medicine. Hearts were harvested from 9–12 month old mongrel canines (7 males, 10 females) weighing 19–30 kg. All animals were anesthetized with intravenous propofol (7 mg/kg), intubated with a cuffed endotracheal tube and mechanically ventilated with a pressure-controlled ventilator. An adequate level of anesthesia was maintained by inhaled isoflurane (1%–3%) and a limb-lead ECG was monitored. After hypothermic cardioplegic arrest, the explanted heart was place in cold saline solution.

Atrial myocyte isolation

In cold saline solution, ventricular branches from the right coronary artery were ligated with silk sutures. The right coronary artery was cannulated at the right coronary cusps and perfused with 2 mM Ca²⁺-Tyrode solution (in mM: 136 NaCl, 5.4 KCl, 2 CaCl₂, 1 MgCl₂, 0.33 NaH₂PO₄, 5 HEPES, 10 glucose, pH 7.35). The tissue was digested by perfusing the following solutions at 40 mL/min at 37°C: 2 mM Ca²⁺-Tyrode’s for 10 min, Ca²⁺- free Tyrode’s for 10 min, and 0.05% collagenase + 0.1% Albumin in Ca²⁺ free Tyrode’s for 60 min, while bubbling with 100% O₂. After digestion, the tissue lateral to the crista terminalis was excised and gently triturated to tease apart the tissue and to suspend isolated myocytes cells in 200 µM Ca²⁺-Tyrode’s solution.

Ventricular myocyte isolation

After the right atrium was removed, the remainder of the heart was immediately transferred to cold, oxygenated Ca²⁺-free modified Krebs solution (in mM; 118 NaCl, 4.8 KCl, 1.2 MgCl₂, 11 glucose, 1.2 KH₂PO₄, 25 NaHCO₃, 0.03 EGTA, 1× MEM amino acids, 1.5 nM insulin, pH 7.30), and a branch of the left anterior descending artery was cannulated, to start perfusion with a syringe. Cannulated tissue was then attached to perfusion tubing and maintained at 37°C in an incubator. Tissue was perfused with Ca²⁺-free modified Krebs solution for 15 min and 0.08% collagenase solution (200 U/ml collagenase in 35 µM Ca⁺ modified Krebs solution) for another 15 min. Epicardial (Epi) and endocardial (Endo) surfaces were separated from mid-myocardium (Mid) and transferred to individual petri dishes, each with 20 ml 0.08% collagenase solution. These tissue sections were gently chopped into small pieces, and incubated in a shaking bath for 10 min at 37° and 80 rpm. After incubation, the supernatant was discarded, and 20 ml 0.08% collagenase solution was added to each flask for 15 min. The supernatant was then collected and filtered and additional collagenase solution was added for a further 15 min. This process was repeated and supernatants acquired from all three incubations. Supernatants were centrifuged at 45 × g for 2 min and myocytes were re-suspended in 50 µM Ca²⁺-HEPES solution (in mM; 133.5 NaCl, 4.8 KCl, 1.2 MgCl₂, 1.2 KH₂PO₄, 10 HEPES, 10 glucose, 0.05 CaCl₂, 10 taurine, 1× MEM amino acids, 0.5 pyruvate, 1.5 nM insulin, 0.5% BSA, pH 7.30).

Electrophysiological recordings

Cardiomyocytes were voltage- or current-clamped in whole-cell or excised patch mode using an Axopatch 200A amplifier (Molecular Devices, Sunnyvale, CA), and signals were digitized using a Digidata 1322A digitizer board (Molecular Devices, Sunnyvale, CA) after 4-pole low-pass Bessel filtering at 2 kHz. Signals were digitized at 10 kHz using pCLAMP 8 (Molecular Devices, Sunnyvale, CA) and analyzed using pCLAMP 9 software (Molecular Devices, Sunnyvale, CA).

Conventional whole-cell recording

Isolated myocytes were transferred to the recording chamber using a Pasteur pipette and continuously perfused at room temperature with extracellular solution containing (in mM): 137 NaCl, 5.4 KCl, 0.5 MgCl₂, 3 NaHCO₃, 0.2 NaH₂PO₄, 5 HEPES, 10 Glucose, pH 7.40. Electrodes had 1.5–2.0 MΩ resistance when filled with pipette solution (in mM): 130 K-aspartate, 20 KCl, 4 K₂HPO₄, 1 MgCl₂, 10 EGTA, 0.1 K₂ATP, and 10 HEPES, pH 7.20–7.30. K_ATP currents were activated by applying metabolic inhibitors (MI, 2 mM NaCN plus 10 mM Deoxyl-glucose replacing glucose), 100 µM pinacidil or 100 µM diazoxide to the bath solution (test solution). K_ATP currents were continuously assessed using a 4-s voltage ramp from −120 mV to +40 mV from a holding potential of −70 mV. K_ATP current amplitude was typically measured at +40 mV, and reported conductances (used in simulations) were measured as the slope conductance between −75 to −65 mV.

Perforated whole-cell recording

Isolated myocytes were continuously perfused at 35–37°C with extracellular solution containing (in mM): 137 NaCl, 5.4 KCl, 1.5 CaCl₂, 0.5 MgCl₂, 3 NaHCO₃, 0.2 NaH₂PO₄, 5 HEPES, 10 Glucose, pH 7.40. Patch clamp electrodes were tip-filled with pipette solution (in mM: 110 K-asparate, 20 KCl, 1 MgCl₂, 0.5 CaCl₂, 5 EGTA, 5 K₂ATP, 10 HEPES, pH 7.2–7.3), and back-filled with the same pipette solution containing 250 µg/ml nystatin. Whole-cell configuration was typically achieved within 5 min after forming a gigaohm seal. To examine the physiological consequences of differential K_ATP expression, we used a dynamic clamp circuit (Fig. 1)¹⁴ in which we injected a simulated K_ATP current (KATP-INJ) via an analog circuit (Fig. 1) to simulate a progressively increasing linear K⁺ conductance. The analog circuit generated a voltage-independent conductance reversing at manually set E_K (−80 mV) that could be manually altered to apply currents up to 50 pA.

Figure 1.

Dynamic Clamp Circuit for applying K_ATP-INJ. The goal was to add the result of the following equation I_KATP-INJ = G_K(V_m-E_K) to the stimulus, I_Stim. G_K is the K_ATP-INJ conductance, V_m is the transmembrane potential measured by the amplifer and E_K is the reversal potential of K⁺, −80 mV set by the D/A amplifier. (A) The circuit relied on a series of inverting operational amplifiers (A), which were combined (B) to generate the real time K_ATP-INJ simulated conductance applied to the current-clamped cell.

Excised inside-out patch-clamp recording

Pipettes had resistances of 1.0–1.5 MΩ when filled with ATP-free Kint (in mM, 140 KCl, 10 HEPES, 1 EGTA, 1 K₂-EDTA, and 4 K₂HPO₄, pH 7.40). Following forming of an on-cell gigaohm seal, the patch was excised and responses to various dugs examined as described. Patch current was recorded at a pipette potential of +50 mV at pipette potential, i.e. −50 mV membrane potential. Diazoxide or pinacidil-induced current was normalized as the percentage of ATP-free Kint current.

Simulations

Action potentials were simulated with the 2010 Hund-Rudy canine ventricular cell model¹⁵. To better match observed AP durations, the magnitudes of the repolarizing I_Ks and I_Kr conductances were increased to 220 and 380 pS/pF, respectively and the magnitude of the late Na⁺ current was decreased to 1.3 pS/pF. To simulate shorter action potentials (<140 ms) I_Ks and I_Kr conductance magnitudes were further increased to 1240 and 2100 pS/pF, respectively. Sets of ordinary differential equations were solved with the forward Euler method and a fixed time step of 5 µs. Simulations were executed on a Beowulf cluster. Each node utilized 12, 2.3 GHz, Intel Xeon cores, allowing simulations to complete within minutes.

Statistics

Data are presented as the mean ± standard error of the mean (S.E.M.) with n (n = cell or patch number, heart number) being the number of cells and hearts in whole-cell recording or patches and hearts in excised patch recording. All data were first analyzed using a linear mixed model (SAS) with “heart” as a random effect to detect the significance of “heart” in all the measurements. The results show that the “heart” was not significant, so myocytes were treated as independent samples in the analysis. Statistical analysis was performed using a two-tailed student t-test, one-way ANOVA followed by Tukey’s post hoc comparison of means or two-way ANOVA followed by Bonferroni’s post-tests, as appropriate. Statistical significance was set as p< 0.05.

Materials

Collegenase was purchased from Worthington Biochemical Corporation (Lakewood, NJ). All other chemicals were obtained from Sigma (St. Louis, MO) unless indicated otherwise.

RESULTS

Atrial K_ATP channels are fully activated by pinacidil and slightly activated by diazoxide

After stable whole-cell patch-clamp configuration was achieved, the bath solution was switched to a test solution containing MI, diazoxide, or pinacidil (Fig. 2A), to assess K_ATP densities in atrial myocytes (cell capacitance 86 ± 7 pF, n = 19, 5). Pinacidil activates SUR2A-containing channels, while diazoxide activates channels that contain SUR1^{16, 17}. Pinacidil-induced current (40 ± 4 pA/pF, n = 14, 5) did not significantly differ from MI-induced current (49 ± 9 pA/pF, n = 11, 5; Fig. 2B). Diazoxide induced a much smaller, but still significant K_ATP current (8 ± 5 pA/pF, n = 11, 5; P<0.05 by ANOVA, Fig. 2B). These results suggest that ~20% of dog atrial K_ATP channels may be regulated by SUR1, whereas almost all channels are regulated by SUR2A.

Figure 2.

Whole-cell patch-clamp recordings of K_ATP currents from dog atrial myocytes. (A) Time courses of whole-cell K_ATP currents induced by metabolic inhibition (above, MI, 10 mM deoxyglucose + 10 mM NaCN), 100 µM pinacidil, and 100 µM diazoxide from right atrial myocyte. (below) Raw traces from the time points indicated above. (B) Summary of K_ATP currents in dog atrial myocytes. MI and pinacidil (Pina) activate comparable currents from dog atrial myocytes.

Ventricular K_ATP channels are predominantly pinacidil-activated

Ventricular K_ATP channel composition and heterogeneity was independently assessed in epicardial (Epi), mid-myocardial (Mid), and endocardial (Endo) myocytes. No statistical difference in capacitance was detected between myocytes from these regions (148 ± 25 pF in Epi, n = 9, 3; 102 ± 25 pF in Mid, n = 7, 3; and 139 ± 21 pF in Endo, n = 12, 4; p>0.05 by one-way ANOVA). Pinacidil-induced current was not significantly different from MI-induced current in Epi (62 ± 16 pA/pF, n = 5, 3; versus 53 ± 11 pA/pF, n = 7, 3; p>0.05), Mid (43 ± 11 pA/pF, n = 6, 3; versus 46 ± 10 pA/pF, n = 7, 3; p>0.05), or Endo (37 ± 7 pA/pF, n = 5, 4;versus 29 ± 5 pA/pF, n = 6, 4; p>0.05) myocytes, suggesting that SUR2A is also the predominant subunit across the ventricular wall (Fig. 3). Pinacidil-induced currents as well as MI-induced currents were non-significantly smaller in Endo than Epi and Mid (P>0.05 by two-way ANOVA), implying a weakly heterogeneous transmural K_ATP density across the ventricle. Consistently, diazoxide-induced current was negligible compared with MI- or Pinacidil-induced current in Epi (6 ± 5 pA/pF, p<0.05 by two-way ANOVA), Mid (1 ± 3 pA/pF, p<0.05 by two-way ANOVA), and Endo (2 ± 1 pA/pF, p<0.05 by two-way ANOVA) (Fig. 3).

Figure 3.

Whole-cell patch-clamp recordings of K_ATP current from dog ventricular myocytes. Time courses of whole-cell K_ATP currents induced by metabolic inhibition (MI, 10 mM deoxyglucose + 2 mM NaCN), 100 µM pinacidil, and 100 µM diazoxide on dog myocytes from epicardium (Epi) (A), mid myocardium (Mid) (B), and endocardium (Endo) (C). (D) Summary of K_ATP currents in dog ventricular myocytes. Pinacidil activates large K_ATP currents from Epi, Mid, and Endo, while diazoxide induces negligible current. *p<0.05 vs MI and Pina groups in Epi, Mid or Endo two-way ANOVA followed by Bonferroni’s post-tests

Pinacidil induces K_ATP currents in excised atrial and ventricular membrane patches

K_ATP channel expression and drug sensitivities were also probed with excised inside-out patch recordings from right atrial (Fig. 4A), left ventricular Epi (Fig. 4B top), mid (Fig. 4B middle) and Endo (Fig. 4B bottom) myocyes. Consistent with larger pinacidil-induced whole-cell current, pinacidil-induced patch current (relative to current in zero ATP) was much greater than diazoxide-induced current in atrial myocytes (3.2 ± 1.0 pA versus 0.3 ± 0.08 pA, n = 10, 3; p<0.05 by t-test, Fig. 4A, C), as well as in ventricular Epi (1.8 ±0.7 pA versus 0.2 ±0.05 pA, n = 7, 2), Mid (2.1 ± 0.5 pA versus 0.08 ± 0.01 pA, n = 6, 2), and Endo (1.4 ± 0.5 pA versus 0.17 ± 0.06 pA, n = 5, 2) myocytes (P<0.05 by two-way ANOVA, Fig. 4B, C). In addition, pinacidil-induced patch current tended to be non-significantly smaller in Endo compared to Epi, and Mid.

Figure 4.

Excised inside-out patch recordings of K_ATP currents from dog myocytes. Representative traces showing K_ATP current induced by 100 µM pinacidil and 100 µM diazoxide from excised inside-out patches of dog myocytes from (A) right atrium (RA), (B) ventricular epicardium (Epi, top), mid-myocardium (Mid, middle), and endocardium (Endo, bottom) . Part of recording in A is enlarged as indicated. Diazoxide- or pinacidil-induced current was normalized to ATP-free Kint current. (C) Summary of K_ATP patch currents from atrial (left) and ventricular (right) myocytes. Pinacidil activates larger patch currents from atrial and ventricular myocytes. *p<0.05 vs. pinacidil group two-way ANOVA followed by Bonferroni’s post-tests.

Myocytes with longer APD are more sensitive to injected “K_ATP current”

We examined the physiological consequences of differential K_ATP expression by injecting simulated K_ATP current (K_ATP-INJ) via dynamic clamp (Fig. 1) ¹⁴ into atrial and ventricular myocytes under perforated whole-cell current-clamp (Fig. 5A). After normalizing for cell size, APD (averaged over twenty APs at a cycle length (CL) of 1000 ms), was plotted versus K_ATP-INJ conductance, which ranged from 0 to 4 pS/pF (Fig. 5B, left). Starting APDs for individual cells (i.e. without injected current) are indicated by dashes on the right of each plot. Consistent with previous reports of high APD variability¹⁸, we observed a large range with long APDs in myocytes from the Mid and Epi layers. As a group, Epi and Mid myocytes showed relatively high sensitivity to K_ATP-INJ. In contrast, Endo and atrial myocytes were relatively resistant to K_ATP-INJ. Prior AP shortening, as a result of rapid pacing (CL = 300 ms), reduced the sensitivity to K_ATP-INJ in Epi and Mid, while leaving Endo and right atrial APD relatively unaffected (Fig. 5B right). The level of K_ATP-INJ necessary to cause a 50% reduction in APD was 1.00 ± 0.27 (n = 4, 1), 2.40 ± 0.62 (n = 4, 2), 2.03 ± 0.45 (n = 7, 2), and 3.25 ± 0.66 (n = 6, 2) pS/pF for Epi, Mid, Endo and RA, myocytes, respectively. These averages corresponded to 0.4 – 0.9% of the total K_ATP current in each region, consistent with previous estimates of the relative effectiveness of the available K_ATP conductance in ventricular myocytes from rat or guinea pig^{14, 19}.

Figure 5.

Dynamic-clamp and computational simulations of K_ATP conductance effects on action potentials. (A) Representative traces of Epi, Mid, Endo and right atrial APs obtained with perforated patch. Injected currents via dynamic clamp are shown below. Shading of currents corresponds with shading of APs. Increasing K_ATP-INJ results in significant APD shortening. (B) APD as a function of conductance, normalized to capacitance, at CL=1000 ms and 300 ms. Epi and mid cells (with longer APs) are more sensitive to K_ATP-INJ, an effect which is blunted as the AP becomes shorter in response to increased rate. (C) Normalizing to starting APD shows that myocytes with longer APDs undergo greater fractional shortening. *p<0.05, APD>140 ms group vs. APD<140 ms group for both CL=1000 ms and CL=300 ms by a two-tailed student t-test. (D) APs simulated with the HRd canine ventricular model (insets show APDs with increasing K_ATP).

The sensitivity of longer APs to injected K_ATP currents is illustrated in Fig. 5C. At a conductance of 3.5 pS/pF, myocytes with baseline APD >140 ms show ~70% shortening, whereas cells with APD <140 ms show only ~40% shortening (p = 0.02). In the same myocytes paced at CL = 300 ms, there is a modest decrease in sensitivity of longer APs, whereas shorter APs are unaffected, but the cells with longer APs still show greater shortening than those with shorter APs (p = 0.048).

To examine whether computational models could account for the observed behavior, increasing levels of K_ATP were simulated with the 2010 HRd canine ventricular AP model¹⁵. Longer APs were simulated by the starting conditions (APD of 195 ms), whereas shorter APs (APD of 96 ms) were obtained by increasing I_Kr and I_Ks conductances, as detailed in the Methods. Approximately 5× more injected K_ATP current is required to cause similar APD shortening in the model than is observed in experiments. This difference in sensitivity was not easily accounted for by varying the conductances of major currents (data not shown), suggesting that the kinetic parameters that describe the currents may be responsible. Nevertheless, the model does qualitatively reproduce differential sensitivity of short and long APs in addition to blunted effect of K_ATP at faster pacing rates (Fig. 5D).

DISCUSSION

Heterogeneity of cardiac K_ATP channels differs between species

SUR subunits, encoded by ABCC8 (SUR1) and ABCC9 (SUR2) ¹ endow K_ATP channels with distinct sensitivity to K_ATP channel inhibitors and activators. The consensus notion that cardiac K_ATP channels are SUR2A-based^{3, 20}, is now questioned by studies that show inhomogeneous SUR distribution in mouse heart: SUR1 predominates in atria, SUR2A in the ventricles, and SUR2B may be expressed in the conduction system^{6, 7}. The present results imply that the SUR distribution in larger hearts is less marked: both atrial and ventricular canine K_ATP appear to be primarily regulated by SUR2A subunits, but small diazoxide-induced currents in both atrial and ventricular myocytes imply the presence of SUR1-dependent channels at low levels. A similar SUR1 and SUR2A distribution seems likely in human atrium and ventricle²¹. Thus, in larger mammals, the distribution of SURx subunits may be predominantly SUR2A in all chambers, with low expression of SUR1 in both atria and ventricles. K_ATP expression levels may vary in different species and regions of the heart. Measured under the same conditions, K_ATP current density in mouse heart is between 100–600 pA/pF²⁴ but only 20–100 pA/pF in dog heart. These points should be carefully considered when using other species to model normal or diseased human hearts.

Studies from heterologous expression systems have shown that SUR1 and SUR2 subunit can coexist within a single functional K_ATP channel^{22, 23} and heterogeneous SURx composition may also be the case in native K_ATP, since pinacidil and diazoxide both induce currents from the same cardiomyocytes and in the same excised patches. It has recently been shown that cardiac K_ATP channels provide protection during adaptation to physiological stress¹¹, which may imply a physiological role for opening of SUR1-based K_ATP channels, since SUR1 channels will open at higher ATP concentrations⁴. Deliberate overexpression of SUR1 together with a gain-of-function mutation of Kir6.2 in mice led to embryonic lethality and arrhythmias^{25, 26}.

Transmural heterogeneity and function of cardiac K_ATP channels

Ventricular myocytes are organized into three layers with distinct electrophysiological profiles: Epi, Mid and Endo^{18, 27}. Epi and Mid tend to have a prominent notch in phase I due to larger I_to. Mid exhibit larger late I_Na, larger sodium-calcium exchange current (I_Na/Ca), and smaller I_Ks, leading to a longer APD. The present study suggests that heterogeneity of K_ATP current, as previously reported in feline Epi- versus Endo-mycoytes ²⁸, may generate distinct pharmacologic profiles and responses to pathophysiologic states. Heterogeneity of AP shortening by K_ATP channel opening between epicardium and endocardium has been reported in canine ventricle²⁹. The AP is preferentially shortened by pinacidil in epicardium³⁰, potentially due to the larger I_to-dependent AP notch, although larger epicardial K_ATP density might also contribute. Greater AP shortening due to K_ATP opening in epicardium could result in transmural dispersion of repolarization, causing ST-segment elevation and phase 2 reentry, precipitating ventricular tachycardia and ventricular fibrillation^{29, 30} during ischemia. AP shortening caused by ischemia is more pronounced in the epicardium ^{28, 31}, and the present results suggest that increased epicardial expression of K_ATP channels may in part be responsible.

In silico analysis using the HRd model confirmed that activation of a small fraction of available K_ATP channels can have dramatic effects on the cellular electrophysiology¹⁹, although the model implies a weaker effect of K_ATP activation than demonstrated by current-injection experiments (Fig. 5C,D). While it is beyond the scope of this study to reformulate the model, these quantitative differences may lie in the description of I_Ca,L, whose activation may in reality be more susceptible to small increases in outward plateau current. Previous studies have shown proarrhythmic consequences when K_ATP is sufficiently activated to cause a loss of the dome in AP morphology, in myocytes that express high levels of I_to ³⁰. By applying our dynamic-clamp methodology to dissociated myocytes we observed enhanced sensitivity of myocytes with longer APD to simulated K_ATP. Greater fractional shortening in myocytes with long APDs is consistent with greater rate-adaptation of APD in mid-myocardial cells with prolonged APD¹⁸, and with the concept of reduced “repolarization reserve” in which patients with previously compromised repolarizing currents are more susceptible to QT prolongation and arrhythmia when repolarizing currents are additionally blocked by drugs³². Physiologically, increased sensitivity of these cells to K_ATP activation would tend to result in enhanced APD uniformity across the ventricular wall, once activation is initiated. This effect on the overall dispersion of APD would likely persist until catastrophic loss of the dome, in a cell with increased I_to. In intact tissue, AP synchronization through gap junction-coupling may minimize the effect of K_ATP channel activation on AP uniformity, but K_ATP channels may provide a strong influence in diseased conditions where APD can show larger dispersion.

The influence of subtle heterogeneities of K_ATP density or properties on APD may be amplified under exercise or mild ischemia. Even if expressed at low levels, the higher sensitivity of SUR1- containing channels to nucleotide activation⁴ could have dramatic consequences. Finally, we detected no major difference in K_ATP density between atrium and ventricle, yet even similar activation may have less effect in atrium due to the intrinsically shorter atrial APD. Previous studies have shown that interaction between I_to and K_ATP can induce a heterogenous response³⁰, but quantitative evaluation of K_ATP effects across the myocardium reveals complex dependence of the physiological response on composition of the channel, APD, and level of channel expression.

Abbreviations

K_ATP

ATP sensitive potassium channel

AP

action potential

APD

action potential duration

EGTA

ethylene glycol-bis(2-aminoethylether)-N,N,N’,N’-tetraacetic acid

HEPES

4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

MI

metabolic inhibitors

INa

sodium current

ICa,L

L-type Ca2+ current

Ito

transient outward current

IKs

slow delayed rectifier current

IKr

rapid delayed rectifier current

IK1

inward rectifier current.