Unique Properties of the ATP-Sensitive K⁺ Channel in the Mouse Ventricular Cardiac Conduction System

Abstract

Background

The specialized cardiac conduction system (CCS) expresses a unique complement of ion channels that confer a specific electrophysiological profile. ATP sensitive potassium (K_ATP) channels in these myocytes have not been systemically investigated.

Methods and Results

We recorded K_ATP channels in isolated CCS myocytes using Cntn2-EGFP reporter mice. The CCS K_ATP channels were less sensitive to inhibitory cytosolic ATP compared to ventricular channels and more strongly activated by MgADP. They also had a smaller slope conductance. The two types of channels had similar intraburst open and closed times, but the CCS K_ATP channel had a prolonged interburst closed time. CCS K_ATP channels were strongly activated by diazoxide and less by levcromakalim, whereas the ventricular K_ATP channel had a reverse pharmacological profile. CCS myocytes express elevated levels of Kir6.1, but reduced Kir6.2 and SUR2A mRNA compared to ventricular myocytes (SUR1 expression was negligible). SUR2B mRNA expression was higher in CCS myocytes relative to SUR2A. Canine Purkinje fibers expressed higher levels of Kir6.1 and SUR2B protein relative to the ventricle. Numerical simulation predicts a high sensitivity of the Purkinje action potential to changes in ATP:ADP ratio. Cardiac conduction time was prolonged by low-flow ischemia in isolated, perfused mouse hearts, which was prevented by glibenclamide.

Conclusions

These data imply a differential electrophysiological response (and possible contribution to arrhythmias) of the ventricular CCS to K_ATP channel opening during periods of ischemia.

Introduction

The electrophysiological diversity of the heart is best illustrated by the unique action potential profiles within the specialized pacemaker and cardiac conduction systems. For example, myocytes from the sino-atrial (SA) node, atrio-ventricular (AV) node, bundle of His, and Purkinje network all display various degrees of pacemaking activity, and distinct levels of maximum diastolic potential and action potential waveform profiles¹. Before the advent of patch clamping, the electrophysiological properties of Purkinje strands have been studied extensively because their cable-like morphological structure was well suited for the prevailing recording techniques (sucrose gap and two-electrode voltage clamping techniques). Recent patch clamp data have further clarified this picture and we now know that Purkinje myocytes express a unique complement of ion channels (several types of Na⁺ channels, including a large non-inactivating component, L- and T-type Ca²⁺ channels, a variety of K⁺ channels and a robust pacemaker current)². The interaction of these various ion channels, exchangers and pumps is responsible for the characteristic electrophysiological properties of the Purkinje myocyte, including a rapid upstroke, action potential notch, long duration, negative maximum diastolic potential and automaticity. Little data are available regarding the properties of the K_ATP channel in the specialized conduction system, which comes to play during ischemic events. Given the contribution of the ventricular conduction system and Purkinje-muscle junctions in the generation of ischemia-induced arrhythmias³, it is important to understand the biophysical, regulatory and pharmacological properties of K_ATP channels in this tissue.

K_ATP channels are ubiquitously expressed in the heart. Since their original description in cardiac ventricular myocytes⁴, they have also been identified in the atrium⁵, as well as in the specialized cardiac pacemaker/conduction system, including the sino-atrial (SA) node⁶, atrio-ventricular (AV) node⁷ and myocytes from Purkinje fibers⁸. Structurally, K_ATP channels are octaheteromeric proteins that are composed of four inwardly rectifying pore-forming subunits (Kir6.1 or Kir6.2) and four regulatory, sulfonylurea receptor subunits (SUR1, SUR2A and SUR2B)⁹. Their potassium selectivity, inward rectification, and unitary conductance are determined primarily by the Kir6.x subunit, whereas the nucleotide sensitivity and pharmacology of the channel depend largely on the type of SURx subunit present in the channel protein⁹. In this study, we characterized the biophysical and pharmacological properties, regulation by cytosolic nucleotides, and the molecular composition of K_ATP channels in the mouse ventricular cardiac conduction system (CCS) and compared these to the better characterized ventricular K_ATP channel. We also investigated the pathological consequences of CCS K_ATP channels in conduction slowing during ischemia and our results suggest that when K_ATP channels open during ischemia, electrophysiological effects in the CCS may be especially severe. These studies were facilitated by the use of Cntn2-EGFP reporter mice in which EGFP expression is restricted to the cardiac conduction system¹⁰.

Methods

A complete description of the methods used is available on the online supplemental information.

Cardiac myocyte isolation

All animal procedures were in accordance with NIH guidelines and were approved by the Institutional Animal Care and Use Committees of New York University School of Medicine. The methods used were described previously¹¹.

Patch clamp recording techniques

Single-channel recordings were performed in the inside-out configuration at room temperature using standard patch-clamp techniques. Patch pipettes (2–4 MΩ) were filled with (in mmol/L) 150 KCl, 2 CaCl₂, 1 MgCl₂, and 10 HEPES, pH 7.4. The bath solution consisted of (mmol/L) 150 KCl, 1 EGTA, 10 HEPES, 1.2 MgCl₂ and pH 7.2. Currents were filtered (−3 dB at 1 kHz) and digitized (5 kHz) for computer storage and analysis. Unless otherwise indicated, the pipette potential was +80mV. If needed, rundown was corrected. Dwell times and burst analysis were estimated using patches containing a single active channel. Following baseline subtraction, event detection was performed with 50% single channel current amplitude threshold. A minimum resolution was imposed by ignoring events shorter than 0.18 ms¹².

Real-time and conventional RT-PCR

Real-time semi-quantitative RT-PCR (qRT-PCR) was used to quantify mRNA expression, as described previously¹³. EGFP-positive myocytes (originating from the CCS) or non-GFP expressing cells (ventricular myocytes) were enzymatically isolated from the ventricles of the Cntn2-EGFP reporter mouse strain¹⁰. Individual cells were manually selected by epifluorescence microscopy and RNA was isolated from 80–100 pooled cells (PicoPure™, Arcturus). Reverse transcription (RT; Superscript III, Invitrogen) was performed using random hexamer primers. The PCR threshold cycle (Ct) values were determined and data were analyzed using custom routines written in the R programming language (available upon request). Gene expression was calculated relative to the following reference genes: cyclophilin B (PPIB) and hydroxymethylbilane synthase (HMBS). Conventional RT-PCR was performed using the same RT reactions described above. Primer sequences are given as Supplementary Information

Membrane preparation and Western blotting

Membrane fractions were prepared as described¹¹ with modifications. Free running canine Purkinje strands and left ventricular free wall tissue were snap-frozen, pulverized in liquid nitrogen and homogenized using 30 strokes of a glass-glass homogenizer followed by 30 strokes in a Dounce homogenizer in (in mmol/L) 250 sucrose, 1 EDTA, 10 HEPES, 1 DTT and pH 7.4 supplemented with protease inhibitor cocktail (Roche Applied Science). After a brief centrifugation (1,000 g for 5 min at 4°C), the supernatant was loaded on top of an 18% Optiprep (Sigma-Aldrich) layer (with in mmol/L: 150 sucrose, 1 EDTA, 10 HEPES, 1 DTT and pH 7.4 supplemented with protease inhibitor cocktail) and subjected to ultracentrifugation (200,000 g for 1 h at 4°C). The membrane fraction was collected from the interphase and membranes were recovered by centrifugation (after dilution with 10mM Tris HCl, 1 mM EDTA, pH 7.5). After overnight solubilization at 4°C in 20 mmol/L HEPES, pH 7.4 and 0.5% v/v Triton X100, unboiled proteins were resolved by 10% SDS-PAGE and immunoblotted using standard techniques¹¹. Antibodies used were: rabbit anti-Kir6.1 (NAF-1), chicken anti-Kir6.2 (C-62), goat anti-SUR2B (C-15, Santa Cruz) and GAPDH (Millipore). The specificity of these antibodies is illustrated in Supplemental Figure S1.

Isolated, perfused hearts: low-flow ischemia

Following sacrifice, mouse hearts were excised and the right atrial free wall was removed. Hearts were cannulated by the aorta and retrogradedly perfused at 74 mmHg with Tyrode's solution (in mM: 1.8 CaCl₂, 1.0 MgCl₂, 1.2 KH₂PO₄, 130 NaCl, 4.7 KCl, 11.1 glucose, 24 NaHCO₃, 0.52% albumin w/v, pH 7.4, equilibrated with 95% O₂/5% CO₂). Throughout the experiment, the left atrium was electrically paced (ms stimuli 1.5× above threshold) at a cycle length of 300 ms, while the volume conducted local cardiac electrical response (‘electrocardiogram’ or ECG) was recorded using Ag-AgCl electrodes flanking the heart. The time from the end of atrial depolarization to the beginning of ventricular depolarization (PR segment) was measured as an index of overall conduction time. The experimental protocol consisted of a stabilization time of 20 min before the hearts were divided into two groups. For experimental group, hearts were perfused with the K_ATP channel blocker, glibenclamide (2 µM), for 5 minutes followed by 20 minute low-flow ischemia, which was induced by decreasing the perfusion pressure to 25% of the pre-ischemic period. The control group was identical, except that glibenclamide was substituted by the solvent (DMSO; 0.017% v/v final concentration). In other experiments, rat hearts (Sprague-Dawley, 250–300g) were perfused in Langendorff as described above and the coronary artery flow was monitored (Transonic Systems). The perfusion pressure was decreased to achieve low-flow ischemia. The volume conducted ECG was recorded and the time from the beginning of ventricular depolarization to partial ventricular repolarization (QRS duration) was measured as an index of distal conduction time.

Numerical model of the K_ATP channel

We developed numerical models of the ventricular and CCS K_ATP channels, based upon our experimental data and a previously described model¹⁴. Full details of the models are available in the Supplementary information. In brief, the K_ATP channel current is expressed as

$$I_{K,\mathit{\text{ATP}}}=\mathrm{N}\times {\mathrm{P}}_{\mathrm{o}}\times \gamma \times {\mathrm{f}}_{\mathrm{K},\text{ATP}}\times ({\mathrm{E}}_{\mathrm{m}}-{\mathrm{E}}_{\mathrm{K}})$$ (1)

where N represents the channel density (channels/µm²), P_o the intrinsic open probability in the absence of modulation by Na⁺, Mg²⁺, γ the unitary conductance (pS), and nucleotides, f_K,ATP the fraction of open channels, E_m the membrane potential (mV) and E_K the equilibrium potential for K⁺ ions (mV). The number of functional K_ATP channels and intrinsic open probabilities were estimated from our patch clamp data. Inhibition by ATP and stimulation by MgADP were modeled by experimental data obtained. To simulate effects of K_ATP channel on the action potential duration, we used published models of either the human ventricular¹⁵ or the human Purkinje² myocyte (see online supplement for full information).

Statistical analysis

The Kolmogorov-Smirnov test was used to test whether data followed a normal distribution. Assumption of equal variance per group was confirmed with the use of Bartlett's test. Data are given as mean ± SEM (where n indicates number of patches or number of animals per group). Statistical tests used included unpaired or paired Student’s t-test. With small sample sizes or when the normality or equal variance test failed, the non-parametric Mann-Whitney rank sum test was used. Statistical significance was assumed at a p-value <0.05.

Results

Given the importance of the specialized cardiac conduction system in electrical propagation and arrhythmogenesis, we set out to characterize the K_ATP channel of the mouse cardiac conduction system (CCS). This study was facilitated by the use of a Cntn2-EGFP BAC transgenic reporter mouse strain¹⁰, which allowed us to visualize isolated cells of the CCS by their EGFP epifluorescence. Atria were removed before cell isolation and EGFP-positive myocytes may originate from both proximal and distal elements of the CCS (the ventricular Purkinje network). We examined the biophysical and regulatory properties, as well as the molecular composition, of ventricular CCS K_ATP channels in comparison with non-EGFP expressing ventricular myocytes. We also investigated the role of K_ATP channels during ischemia-induced conduction slowing.

Unitary conductance

The unitary conductance was determined using patches with only a few K_ATP channels per patch, using a ramp voltage step (Figure 1). After baseline subtraction, an all-points histogram was constructed to obtain the unitary current at different voltages (Figure 1B). The current-voltage relationships of CCS and ventricular K_ATP channels were both weakly inward rectifying in the presence of ‘intracellular’ Mg²⁺. The unitary conductance was respectively 57.1±3.20 pS (n=8) and 75.9±1.32 pS (n=6; p=0.001, Mann-Whitney rank sum test) for CCS and ventricular K_ATP channels (Figure 1C). Most patches recorded in the inside-out patch clamp configuration of CCS or ventricular myocytes contained a large number of channels (pipette resistance 2.4±0.04 MΩ, n=57, Student’s t-test, p = 0.572)). Although the number of channels per patch was variable, the data suggested a similar channel density for the two types of myocytes. The estimated median K_ATP channel number was 28.6 (n=28) and 31.7 channels/patch (n=29) respectively for CCS and ventricular myocytes (Mann-Whitney rank sum test, p = 0.549).

Figure 1.

The unitary conductance of K_ATP channels recorded from mouse cardiomyocytes originating from the cardiac conduction system (CCS) and the ventricle. A) A representative trace of CCS K_ATP channel unitary events recorded during a voltage clamp ramp (111 mV/s) from −100 to 100 mV. B) Following baseline subtraction, the region of the current marked by the box (membrane potential around −80mV) is shown. Also shown is an all-points histogram for this current section. C) Current-voltage relationship of the unitary current amplitude for K_ATP channels recorded from CCS or ventricular myocytes. Inset: slope conductances (mean±SEM), measured between −80 and −20 mV. *p<0.05; Student’s t-test.

Single channel kinetics

Superficially, CCS and ventricular K_ATP channels displayed similar dwell characteristics in the absence of nucleotides. Dwell times were difficult to analyze given the high channel densities. We were able to obtain recordings with a single active channel in a few patches when the pipette resistance was increased to ~10 MΩ. We analyzed the mean open and closed times within bursts over a 30 s period, using recordings from patches containing a single active channel (Figure 2). Open and closed time histograms were constructed and data were subjected to curve fitting with a single exponential function (long closing events were ignored). The mean open times were respectively 2.2±0.16 ms (n=6) and 2.9±0.95 ms (n=5; Student’s t-test, p = 0.931) for CCS and ventricular K_ATP channels, whereas the mean closed times were respectively 0.4±0.02 ms (n=6) and 0.4±0.02 ms (n=5; Mann-Whitney rank sum test, p = 0.208). Both channel types displayed bursting behavior. The intraburst and interburst kinetics were similar for the two types of K_ATP channels, with the exception of a longer interburst closed time observed with the CCS K_ATP channels (Table 1).

Figure 2.

Dwell time kinetics of CCS and ventricular K_ATP channels. Shown are representative traces of K_ATP channel activity. Dwell time histograms were constructed (bin size of 0.2 ms) to obtain the mean open (left) and closed (right) times, obtained by exponential curve fitting. Channel activity was recorded for 30 s at −80 mV.

Table 1.

Properties of K_ATP channels recorded from mouse ventricular myocytes or from myocytes isolated from the specialized conduction system.

	Ventricle	Conduction system	P value
Single channel conductance (pS)	75.8±1.32 (n=6)	57.1±3.20 (n=8)	0.001*
Intrinsic open probability	0.153±0.06 (n=4)	0.167±0.06 (n=5)	0.690
Bursting kinetics
Mean intraburst open time (ms)	2.2±0.95	2.2±0.16	0.931
Mean intraburst closed time (ms)	0.4±0.02	0.4±0.02	0.208
Number of openings per burst	34±3.1	41±5.8	0.324
Burst duration	115±43.7	139+49.1	0.320
Interburst duration (ms)	16±6.1	27±9.5	0.027
	207±43.1	594±230.7	0.063*

The unitary conductance was obtained between −80 and 0 mV. The intrinsic open probability reflects the average behavior of the channel (including interburst closed times). Bursting kinetics were calculated from n=5 and n=4 patches respectively from CCS and ventricular myocytes.

The data marked with * are compared with Mann-Whitney rank sum test, other data are compared with Student’s t-test.

Nucleotide sensitivity

We next examined the sensitivity of K_ATP channels to intracellular nucleotides. A concentration-response curve for ATP inhibition was constructed by exposing patches sequentially to ATP between 1 and 1000 µM (Figure 3A). The concentration of ATP producing half-maximal inhibition (IC₅₀) was 59±12.4 µM (Hill coefficient 1.5±0.14; n=12) for ventricular K_ATP channels, which was significantly lower than that of the CCS K_ATP channel (116±22.8 µM, Hill coefficient 1.4±0.17; n = 12, p=0.04; Figure 3B). Thus, in addition to having a smaller unitary conductance, the CCS K_ATP channel is less sensitive to the inhibitory effect of cytosolic ATP. We also examined the degree of stimulation of K_ATP channel activity by MgADP. For these experiments, channels were half-maximally inhibited by ATP (50 or 100 µM respectively for ventricular or CCS K_ATP channels). ADP was then applied at a ADP:ATP ratio of 0.1–10 (Figure 3C). Consistent with previous reports, ventricular K_ATP channels were stimulated by lower MgADP concentrations¹⁶, but inhibited as the ADP concentration was increased¹⁷. By contrast, the CCS K_ATP channels were stimulated by MgADP, but not inhibited within the concentration range examined (Figure 3D). Half-maximal activation occurred at 84.±3.7 µM MgADP (n=13), which also reasonably describes the MgADP activation phase of ventricular K_ATP channel. A major difference between these channel types appears to be their differential sensitivities to inhibitory ADP.

Figure 3.

Nucleotide sensitivities of CCS and ventricular K_ATP channels. A) Representative traces of the mean patch current (patches contained many active channels). ATP (1 µmol/L to 1 mmol/L) was applied until steady state block occurred. B): Current (normalized to the maximal channel current in the absence of ATP; I_o) is plotted as a function of the ATP concentration. Shown are mean±SEM of cumulative data (n=12 for CCS and ventricle). C) Representative experimental recordings demonstrating the effect of MgADP on K_ATP channels. D) The degree of activation was normalized as (I-I_½)/(I_o-I_½), with I is the mean patch current, I_½ the current recorded in the presence of 50 or 100 µM ATP, and I_o the current measured in the absence of nucleotides (n=13 for CCS and n=12 for ventricle). The solid lines were produced by a K_ATP channel numerical model (see Supplemental Information). The dashed lines represent zero current.

Pharmacological properties

K_ATP channels from diverse tissues have unique pharmacological profiles. For example, the ventricular K_ATP channel is strongly activated by levcromakalim¹⁸, but is less sensitive to diazoxide¹⁹. We examined the effects of these compounds on ventricular and CCS K_ATP channels (Figure 4). Experiments were performed in the absence of ADP and in the presence of an ATP concentration close to the respective IC₅₀ values (see above) to avoid confounding influences of nucleotides. Diazoxide (200µM) had a relatively small stimulatory effect on ventricular K_ATP channels (16±4.4%; n=12). By contrast, diazoxide induced a significantly larger increase in the CCS K_ATP channel activity (48±11.9%; n=7; p = 0.011) (Figure 4B). The activation of the CCS K_ATP channel by diazoxide was readily reversed by tolbutamide (200 µM). The opposite result was observed with levcromakalim (30 µM), which activated the ventricular K_ATP channel more readily (40±5.8%; n=12) than the CCS K_ATP channel (15±4.3%; n=7; p = 0.001; Figure 4B). The effect of levcromakalim was reversed by glibenclamide (2 µM) in both cell types. These data demonstrate that mouse CCS K_ATP channels have a different pharmacological sensitivity compared to the ventricular K_ATP channel.

Figure 4.

Pharmacological profiles of CCS and ventricular K_ATP channels recorded in the inside-out patch clamp configuration. A) Representative recordings of a CCS or ventricular K_ATP channel demonstrating the effects of diazoxide (Diaz, 200 µM), diazoxide plus tolbultamide (Tolb, 200µM), levcromakalim (Lev, 30 µM) and levcromakalim plus glibenclamide (Glib, 2 µM). B) The fractional currents (relative to the maximum current in the absence of ATP; I₀) in the presence of diazoxide or levcromakalim are depicted as box and whisker plots. *p<0.05; Student’s t-test.

K_ATP channel subunit mRNA expression levels

We next evaluated K_ATP channel subunit mRNA expression by real-time qRT-PCR. As expected, Kir6.2 and SUR2A subunit mRNA expression was robust in mouse ventricular myocytes (Figure 5A). The Kir6.2 and SUR2A mRNA expression levels were lower in the CCS myocytes. The Kir6.1 mRNA expression was low, but was double in the CCS myocytes relative to the ventricular myocytes. SUR1 mRNA expression was negligible in both cell types. Due to primer design considerations, it was not possible to discriminate between the major SUR2 splice variants by qRT-PCR. We therefore used conventional RT-PCR with primers that can discriminate between SUR2A and SUR2B²⁰. As expected, these primers only amplified the SUR2B isoform in mouse aortic smooth muscle, whereas both SUR2A and SUR2B amplicons were observed in mouse left ventricle (Figure 5B). Both CCS and ventricular myocytes expressed SUR2A and SUR2B mRNA. However, CCS myocytes contained higher levels of SUR2B (Mann-Whitney rank sum test, p = 0.005 for SUR2A vs. SUR2B), whereas SUR2A was expressed at elevated levels in ventricular myocytes (p = 0.001 SUR2A vs. SUR2B, Figure 5B). These data suggest that the molecular composition of CCS K_ATP channels may be heterogeneous and include contributions from Kir6.1, Kir6.2 and SUR2B.

Figure 5.

K_ATP channel subunit expression in the CCS and ventricle. A) Real-time semi-quantitative RT-PCR (qRT-PCR) analysis of indicted K_ATP channel subunit mRNA expression in mouse CCS and ventricular myocytes. Data are expressed relative to reference genes (HKG) and represent mean±SEM of 3 experiments (*p<0.05; paired t-test). B) Conventional RT-PCR discriminates between SUR2 spice variants. A primer pair was used that resulted in a 471 bp amplicon for SUR2A and a 276 bp band for SUR2B. Shown are reactions performed with RNA isolated from mouse left ventricle (LV), aorta (AO), enzymatically isolated ventricular myocytes (VM) and CCS myocytes (CCS). The PCR products were sequenced to confirm the identity of the amplicons. The relative expression of SUR2A/SUR2B is plotted as a bar graph (mean±SEM; n=3 per group; *p<0.05). C) K_ATP channel subunit protein expression in canine cardiac Purkinje fibers and left ventricle. Western blotting was performed using membrane fractions and antibodies against Kir6.1 (NAF-1, 1:500), Kir6.2 (C-62, 1:200), SUR2B (C-15, 1:200) subunits and GAPDH (1:5000). D) Bar graphs represent K_ATP channel subunit expression levels, normalized by GAPDH expression. Data are mean±SEM (n=2–5).

K_ATP channel subunit protein expression levels

Due to limitations in obtaining sufficient quantities of CCS tissue from the mouse heart, we performed Western blotting using free-running Purkinje strands from canine hearts. The Kir6.2 subunit was lower (albeit not statistically significant) in Purkinje fibers, whereas both Kir6.1 and SUR2B subunits were expressed at significantly elevated levels in Purkinje fibers relative to the canine left ventricle (Figure 5), consistent with the measurements of mRNA levels. Immunohistochemistry performed with cryosections from Cntn-2 EGFP mouse hearts confirmed the expression of Kir6.1, Kir6.2 and SUR2B protein in CCS myocytes (Supplemental Figure S4).

Simulation of K_ATP channel activity: effects on the action potential

We adapted an empirical model of the cardiac K_ATP channel¹⁴ by incorporating data obtained in our study to produce K_ATP channel models for the CCS and ventricular K_ATP channels. Minor changes to the model included alterations in the parameters for channel density, unitary conductance, intrinsic open probability and inhibitory ATP. We adapted the model to account for the degree of MgADP activation (especially pronounced for the CCS K_ATP channel) and inhibition (most pronounced for the ventricular K_ATP channel). The solid lines in Figure 6 are produced by these K_ATP channel models, which approximate the experimental data. Some of the characteristics of the CCS K_ATP channel (such as the reduced unitary conductance) predict a smaller net effect on the action potential compared to the ventricle, whereas others (reduced ATP sensitivity and increased ADP sensitivity) suggest the opposite. Since it may be difficult to address this issue experimentally (e.g. obtaining equivalent degrees of nucleotide alterations in different tissue types), we incorporated the CCS and ventricular K_ATP channel models respectively into action potential simulations of the human Purkinje fiber² and human ventricle¹⁵. To simulate ischemia, we progressively decreased intracellular [ATP] while increasing [ADP], thereby keeping total nucleotide concentration constant (see Supplement for details). Each level of ATP:ADP ratio produced a different degree of K_ATP channel activation, and the steady-state effects on action potentials (1 Hz pacing) were computed. The analysis predicts that K_ATP channel activation, caused by alterations in cytosolic nucleotide levels, confers greater sensitivity in CCS compared to the ventricle (Figure 6). A systematic analysis of changes in ATP and ADP concentrations on action potential duration in the two tissue types is shown in Supplemental Figures S2 and S3.

Figure 6.

Simulation of the effect of K_ATP channel opening on the action potential duration. A) A numerical model was developed to represent the measured activities of CCS or ventricular K_ATP channels. Tissue-specific K_ATP channel models were incorporated into an action potential model of the human ventricle or Purkinje fibers. Shown are effects on the action potential durations caused by K_ATP channel activation, induced by changing the ATP: ADP ratio (ATP was decreased from 6.4 to 1 mM and ADP was increased from 35 to 180 µM in 10 steps). B) Plot of the action potential duration (measured at −90% repolarization) of the data in Panel A against the ATP:ADP ratio.

CCS K_ATP channels contribute to conduction slowing during ischemia

We investigated the potential role of CCS K_ATP channels in electrical conduction during cardiac ischemia, during which these channels are expected to open. During low-flow ischemia of isolated, Langendorff perfused mouse hearts (the perfusion pressure was decreased to 25% of the pre-ischemic level) the PR segment progressively prolonged from 19.8±3.12 ms to 26.3±1.22 ms at 20 minutes ischemia (n=4; p=0.045; Figure 7A). In the presence of the K_ATP channel blocker, glibenclamide (2 µM), the PR segment was unchanged during ischemia (19.6±0.41 ms and 19.7±0.56 ms respectively at the beginning and at 20 minutes of ischemia, n=5; p=0.574, Figure 7A). Hearts developing AV conduction block (n=3 and n=4 respectively in control and treatment groups) were excluded from this analysis. In separate experiments using perfused rat hearts, the QRS duration progressively prolonged during low-flow ischemia (n=4; p=0.019; Figure 7B). This QRS widening was not observed in the presence of glibenclamide (n=4; p=0.327; Figure 7B), suggesting a contribution of K_ATP channels within the distal Purkinje fibers and the Purkinje-ventricular junctions in electrical propagation and conduction disturbances during ischemia. No AV conduction block occurred in this group of experiments.

Figure 7.

Blocking K_ATP channels improves cardiac conduction during ischemia. A) The ECG was measured in isolated, perfused mouse hearts (top), paced at the right atrium (arrows depict stimulation artifacts). The PR segment is plotted as a function of time (bottom). The solid horizontal bar represents low-flow ischemia (25 cm H₂O perfusion pressure). Shown are data with glibenclamide (2 µM, open symbols, n=5) or vehicle (DMSO, filled symbols, n=4). B) The volume-conducted ECG was also recorded from isolated rat hearts (top) and the QRS duration was measured as an index of distal Purkinje fiber function (bottom). Prior to the introduction of ischemia, hearts were perfused with glibenclamide (2 µM; n=4) or vehicle (DMSO; n=4). Ischemia was introduced by lowering the perfusion height, which caused the flow rate to decrease from 9.6±0.39 to 1.9±0.29ml/min (7.4±0.45 to 1.5±0.15 in the experimental group).

Discussion

Our data demonstrate that K_ATP channels in the ventricular CCS share certain properties with their ventricular counterparts (including the open probability and rapid dwell time kinetics). However, the CCS K_ATP channel is also conferred with a unique set of biophysical, regulatory and pharmacological properties. These include a lower unitary conductance, an altered sensitivity to intracellular nucleotides, and a different pharmacological profile. Data obtained using RT-PCR, Western blotting and immunohistochemistry demonstrate that Purkinje myocytes express Kir6.2, Kir6.1, SUR2B and SUR2A subunits, with little evidence of SUR1 expression. We found that ischemia-induced cardiac conduction delay is prevented by blocking K_ATP channels. Numerical simulation of the properties of CCS K_ATP channels suggests that the consequences of K_ATP channel opening may be more severe in the Purkinje system than in the ventricle.

Biophysical properties of Purkinje K_ATP channels

The unitary conductance of mouse CCS K_ATP channel was ~57pS (in symmetrically high K⁺ concentrations), which is close to the 60pS value reported for rabbit Purkinje K_ATP channels⁸, but is smaller than that of the mouse ventricular channel under identical experimental conditions (~76pS). This is a potentially revealing finding, since the unitary conductance is largely dependent on the Kir6.x isoform present. For example, both Kir6.2/SUR1 and Kir6.2/SUR2A channels have unitary conductances of around 75–80pS^21–22, whereas corresponding values for Kir6.1-containing channels are around 35pS^23–24. The conductance of heteromeric Kir6.1/Kir6.2 channels is intermediate between these two values^24–25 and the difference between Kir6.1 and Kir6.2 conductance is determined to a large extent by specific amino acid residues present in the outer pore region²³. Since we observed expression of both Kir6.1 and Kir6.2 in CCS myocytes, the possibility is raised that the CCS K_ATP channel might be composed of both of these subunits (similar to our report for heteromeric K_ATP channels in coronary endothelial cells²⁶). It is also of interest to note that Kir6.1 channels are normally silent after patch excision (unless activated by NDPs) whereas Kir6.2-containing channels (or heteromeric Kir6.1/Kir6.2 channels) exhibit spontaneous openings^{24, 27}, which further suggests that CCS K_ATP channels may be heteromeric Kir6.1/Kir6.2 channels. We were not able to investigate this possibility formally with co-immunoprecipitation approaches due to limited mouse CCS tissue availability.

We found a similar open probability, mean open time and mean closed time when comparing CCS and ventricular K_ATP channels. When excluding long closing events, the rapid dwell times represent the intraburst open/closed events, which were similar for the two types of K_ATP channels. The mean open time of ~2.2 ms, and the mean closed time of ~0.4 ms (at −80mV) of the CCS K_ATP channel were similar to those recorded for the channel in the ventricle [this study; see also^{21, 28}], atrium²⁹, skeletal muscle³⁰, pancreatic β-cells³¹, and heterologously expressed Kir6.2/SUR1³¹ or Kir6.2/SUR2A channels²¹. Both types of channels displayed bursting behavior, with the CCS K_ATP channel having a prolonged interburst closed time relative to the ventricular channel. It is unclear at present to what extent these kinetic differences are due to the diverse molecular composition, which is known to influence bursting behavior³¹.

Nucleotide regulation

The CCS K_ATP channels behaved in many respects the same as ventricular channels. They opened spontaneously upon patch excision in ATP-free solutions and their activity ran down over long periods of recording. They were blocked by cytosolic ATP, but differed from ventricular K_ATP channels in their ATP-sensitivity. The CCS K_ATP channels had an IC₅₀ for inhibitory ATP similar to the value of 119 µmol/L previously reported for rabbit Purkinje K_ATP channels⁸. In contrast, the ventricular K_ATP channel was approximately twice as sensitive to inhibitory ATP. The molecular mechanisms for the differences in ATP sensitivity may be complex and the differences in ATP-sensitivity are not straightforward to interpret. Kir6.1 and Kir6.2 channels have been described to have a similar sensitivity to inhibitory ATP³², but clear differences exist in terms of their regulation by other NDPs, such as UDP²⁷. Moreover, their combination with SURx subunits may alter the nucleotide sensitivity⁹. It is conceivable, for example, that the presence of SUR2B in the CCS confers the higher MgADP sensitivity³³ (see also Suppl Figure S3). Further experiments are underway to fully address this issue.

Pharmacological differences between CCS and ventricular K_ATP channels

A prior report demonstrated that rabbit Purkinje K_ATP channels are activated by 10 µmol/L levcromakalim and are blocked by glibenclamide⁸. Although levcromakalim activated the mouse CCS K_ATP channels, we found diazoxide to be a much more effective opener of these channels. The ventricular K_ATP channel had a reverse pharmacological profile with levcromakalim being a more effective K_ATP channel opener and diazoxide having little effect on channel activity. This profile is similar to that of the corresponding channel subtypes found in the pancreatic β-cell³⁴, the vascular endothelium^35–36, smooth muscle³⁷, atrium^38–39 or post-infarcted ventricular myocytes⁴⁰. The pharmacological profile is determined largely by K_ATP channel molecular composition and the metabolic state of the cell. In particular, SUR1 and SUR2B subunits confer a high sensitivity to diazoxide^{9, 41}. In contrast to atrial K_ATP channels, in which the SUR1 subunit is expressed³⁹, we found little SUR1 expression but robust SUR2B mRNA and protein expression in CCS myocytes. Collectively, these data are in support of the concept that dog and mouse CCS K_ATP channels are SUR2B-based. Assuming that CCS K_ATP channels in other species have similar pharmacological properties, these data may have important implications for pharmacotherapy (e.g. heart disorders in diabetic patients treated with sulfonylureas).

Pathophysiological and clinical relevance

Myocytes in the cardiac conduction system have a high membrane resistance with a long action potential duration and it is therefore unlikely that K_ATP channels are constitutively active in these cells under normoxic conditions. Rather, a role for K_ATP channels is visualized under conditions of metabolic stress, such as during high heart rates, hypoxia or cardiac ischemia. Our numerical simulation data are supportive of the notion that K_ATP channel opening contributes to action potential duration shortening in the CCS. We introduced a K_ATP channel model into a numerical simulation of the human ventricle or Purkinje fiber. Interestingly, the Purkinje action potential was more sensitive to alterations in K_ATP channel opening (induced by changing the ATP:ADP ratio) when compared to the ventricle. This was due in part to the inherent differences in properties of the two action potentials (e.g. the total membrane conductance during the action potential). However, there was also a greater contribution of the CCS K_ATP channel itself to action potential shortening, as evidenced by simulations in which CCS K_ATP channels were incorporated into ventricular myocytes, and vise-versa (Supplemental Figure S2). Thus, the specialized cardiac conduction system may be particularly prone to the effects of K_ATP channel opening.

It is likely that K_ATP channel opening in the CCS contribute to electrophysiological responses during ischemia. Free-running Purkinje strands may be oxygenated directly from circulating blood and are probably less susceptible to ischemic events. In contrast, the sub-endocardial Purkinje strands and Purkinje-muscle junctions are strongly affected by myocardial perfusion and local ischemia⁴². Within minutes following the onset of acute myocardial ischemia, conduction velocity is depressed⁴³. This conduction delay coincides with the occurrence of ventricular tachycardia and ventricular fibrillation³. We directly investigated the role of K_ATP channels in ischemia-induced conduction delay. The PR interval, which under non-ischemic conditions is due mainly to conduction delay within the AV node, was lengthened by low-flow ischemia. Without His recordings, it is not possible to determine to what extent the His-Purkinje network contributed to the increased conduction delay during ischemia. Since glibenclamide effectively prevented ischemia-induced PR segment lengthening, our data support the concept that CCS K_ATP channels (AV node and/or His-Purkinje) contribute to ischemia-induced conduction abnormalities under these conditions. The fact that the Purkinje system plays a pronounced role in the development of ventricular fibrillation during myocardial ischemia⁴⁴ further underscores the relevance of our findings with ischemia-induced arrhythmias. This notion is supported by the inhibitory effect of K_ATP channel blockers on ischemia-induced arrhythmias⁴⁵. The sensitivity of Purkinje K_ATP channels to “non-cardiac” pharmacological compounds may therefore offer anti-arrhythmic therapeutic potential. It may be possible that preventing K_ATP channels from opening in the CCS may confer antiarrhythmic properties, but allowing opening in the ventricles may conserve energy and prevent cell death. Our data suggest that a SUR2B-specific K_ATP channel blocker may fulfill these criteria (if one becomes available), with the caveat that this drug may also impair blood flow (by blocking smooth muscle K_ATP channels).

Conclusions

We characterized mouse Purkinje K_ATP channels and found them to have biophysical and pharmacological properties that differ from those of the ventricular K_ATP channel. The molecular composition is also different, with a potential contribution by Kir6.1, Kir6.2 and SUR2B subunits. These properties might confer different electrophysiological responses of Purkinje fibers during metabolic stress conditions, such as hypoxia or ischemia.

Cardiovascular ATP-sensitive K⁺ channels have both protective and deleterious roles in the heart. A clear beneficial role for these channels in cardiac pathophysiological states has long been identified; opening of K_ATP channels protects against ischemic events by reducing the infarct size and participates as one of the triggers for ischemic preconditioning. However, K_ATP channel opening may also be detrimental. For example, their opening during myocardial ischemia may promote K⁺ efflux and reduce the action potential duration. The ensuing electrical heterogeneity creates a substrate for re-entrant arrhythmias. Our present study characterizes the K_ATP channel in the specialized conduction system and found them to have biophysical and pharmacological properties that differ from those of the ventricular K_ATP channel. Their molecular composition is also different and they express subunits normally abundant in non-cardiac cells. Numerical simulation studies predict these channels to be more sensitive to metabolic stress than their ventricular counterparts, suggesting that they may preferentially open during the early phases of cardiac ischemia. We demonstrate that K_ATP channel blockade mitigates electrical conduction slowing during ischemia, suggesting a contribution of the cardiac conduction system K_ATP channels to ischemia-induced conduction disturbances and arrhythmias.