Synergistic effects of Inward Rectifier (I_K1) and Pacemaker (I_f) Currents on the Induction of Bioengineered Cardiac Automaticity

Abstract

Introduction

Normal heart rhythms originate in the sinoatrial node. HCN-encoded funny current (I_f) and the Kir2-encoded inward rectifier (I_K1) counteract each other by respectively oscillating and stabilizing the negative resting membrane potential, controlling action potential firing. Therefore, I_K1 suppression and I_f overexpression have been independently exploited to convert cardiomyocytes (CMs) into AP-firing bioartificial pacemakers. Although the two strategies have been largely assumed synergistic, their complementarity has not been investigated.

Methods and Results

We explored the inter-relationships of automaticity, I_f and I_K1 by transducing single left ventricular (LV) CMs isolated from guinea pig hearts with the recombinant adenoviruses Ad-CMV-GFP-IRES-HCN1-ΔΔΔ and/or Ad-CGI-Kir2.1 to mediate their current densities via whole-cell patch clamp technique at 37°C. Results showed that Ad-CGI-HCN1-ΔΔΔ- but not Ad-CGI-Kir2.1-transduction induced automaticity (181.1±13.1 bpm). Interestingly, Ad-CGI-HCN1-ΔΔΔ/Ad-CGI-Kir2.1 cotransduction significantly promoted the induced firing frequency (320.0±15.8 bpm; p<0.05). Correlation analysis revealed that the firing frequency, phase 4 slope and APD₉₀ of AP-firing LV CMs were correlated to I_f (R²>0.7) only when -2>I_K1>-4 pA/pF but not to I_K1 over the entire I_f ranges examined (0.02<R²<0.4). Unlike I_f, I_K1 displayed correlation to neither the phase 4 slope (R²=0.02) nor phase 4 length (R²=0.04) when -2>I_f>-4 pA/pF. As anticipated, however, APD₉₀ was correlated to I_K1 (R²=0.4).

Conclusion

We conclude that an optimal level of I_K1 maintains a voltage range for I_f to operate most effectively during a dynamic cardiac cycle.

Introduction

Normal heart rhythms originate in the sinoatrial node (SAN), a specialized cardiac tissue consisting of only a few thousand pacemaker cells. Funny current (I_f), encoded by the hyperpolarization-activated cyclic nucleotide-modulated (HCN) channel gene family plays a pivotal role in the generation of cardiac rhythms in nodal pacemaker cells. It serves as an intrinsic oscillator that drives diastolic depolarization to the action potential (AP) threshold after each excitation cycle.^1-5 By contrast, the Kir2-encoded inwardly rectifying potassium current (I_K1) has the dual function of stabilizing a negative resting membrane potential (RMP) and facilitating repolarization.⁶ Indeed, the robust expression of I_f and the lack of I_K1 are a signature of rhythmically-firing nodal pacemaker cells.^3,7 Conversely, I_K1 but not I_f is robustly expressed in silent-yet-excitable adult atrial and ventricular cardiomyocytes (CMs).^8,9 As such, the approaches of I_K1 suppression^10,11 and I_f over-expression,^4,5,12-14 have been independently and, most commonly, exploited to convert normally quiescent CMs into spontaneously AP-firing cells as bioartificial pacemakers. Although the two strategies appear to be conceptually synergistic, their complementarity has not been experimentally investigated. Indeed, single-sided reproduction of either the I_f or I_K1 profile of nodal cells in the atrial or ventricular CMs may not be optimal for inducing and modulating automaticity.¹⁵ Atrial and ventricular CMs have repertoires of ionic currents that are different from those of nodal pacemaker cells. Since I_f is activated by hyperpolarization and deactivated by depolarization, here we hypothesize that I_K1, a presumptive opponent of I_f, synergistically enhances I_f by maintaining the voltage changes within a range where HCN channels can most effectively operate during a dynamic cardiac cycle.

Methods

Molecular biology

PCR-based mutagenesis of mouse HCN1 (generously provided by Dr. Steve Segalbaum, Columbia University) of the bicistronic adenovirus shuttle vector pAdCMV-GFP-IRES (or pAdCGI) was performed with overlapping oligos as described previously.^16,17 The internal ribosomal entry site (IRES) allows the simultaneous translation of 2 transgenes, green fluorescence protein (GFP) and an engineered-HCN1 construct, with a single transcript. HCN1 was chosen since its structure-function properties have been extensively investigated in our previous studies.^15-25 Adenoviruses were generated by Cre-lox recombination of purified ψ5 viral DNA and shuttle vector DNA using Cre4 cells.²⁶ The recombinant products were plaque purified, amplified, and purified again by CsCl gradients, yielding concentrations on the order of 10¹⁰ plaque-forming units (PFU) ml^-1.

Adenovirus-mediated gene transfer and isolation of LV CMs

Adult female guinea pigs (∼250-300g) were euthanized by intraperitoneal injection of pentobarbital (80mg/kg). The hearts were quickly excised, followed by perfusion with enzymatic solutions using a customized Langendorff apparatus. Left ventricular (LV) CMs were plated at 5×10⁵ per laminin-coated glass coverslip in medium containing: 5 mM carnitine, 5 mM creatine, 5 mM taurine, 100 μg ml^-1 penicillin-streptomycin and 10% fetal bovine serum in Medium 199 (Sigma-Aldrich Corp., St. Louis, MO, USA) in a 37°C incubator with 5% CO₂ for 2 hrs. For transduction, LV CMs were incubated for 1 hour in serum-free medium containing adenoviral particles at a concentration of ∼2×10⁹ PFU. A transduction efficiency of ∼70-80% could typically be achieved with this protocol. Such an in vitro system of adult guinea pig LV CMs has been previously used by both us ²⁷ and others.²⁸

Electrophysiology and data analysis

Electrical recordings were performed using the whole-cell patch-clamp technique at 37°C (HEKA Instruments Inc. Southboro, MA, USA). Successfully transduced cells were selected based on their epifluorescence at 488/530 nm (excitation/emission). Patch pipettes were prepared from 1.5 mm thin-walled borosilicate glass tubes using a Sutter micropipette puller P-97 and had typical resistances of 3-5 MΩ when filled with an internal solution containing (mM): 110 K⁺ aspartate, 20 KCl, 1 MgCl₂, 0.1 Na-GTP, 5 Mg-ATP, 5 Na₂-phosphocreatine, 1 EGTA, 10 HEPES, pH adjusted to 7.3 with KOH. The external Tyrode’s bath solution was composed of (mM): 140 NaCl, 5 KCl, 1 MgCl₂, 1 CaCl₂, 10 Glucose, 10 HEPES, pH adjusted to 7.4 with NaOH. Cell capacitance was measured by using the internal circuit for capacitance-current compensation, in which both cell capacitance and series resistance were compensated. Voltage-and current-clamp recordings were simultaneously performed on the same cells at 37°C within 24 to 36 hours after transduction. Absolutely no automaticity was detected from cultured control LV CMs recorded over the same period.

To elicit inward currents, cells were held at -30mV and pulsed from 0mV to -140mV with 10mV increments for 3 seconds. I_f was defined as 1mM Ba²⁺-insensitive, time-dependent currents. For recording action potentials, cells were held at 0pA without (for electrically-active cells) or with a stimulation of 0.1-1nA for 5ms to elicit a response. The steady-state current-voltage (I-V) relationship of I_f was determined by plotting the currents measured at the end of the 3-second test pulses of the above-mentioned protocol against the test potentials in the presence of 1mM BaCl₂. Data were not corrected for the junction potential.

Statistical analysis

Correlation analysis was performed by linear regression. The larger the value of R², the better correlation of the two parameters in comparison. . All data reported were means ± S.E.M. with P<0.05 indicating statistical significance as determined using an unpaired Student’s t test.

Results

Adenovirus-mediated overexpression of I_f and/or I_K1 in adult LV CMs

For I_f and/or I_K1 overexpression, we generated the recombinant adenoviruses Ad-CGI-HCN1-ΔΔΔ and Ad-CGI-Kir2.1 that mediate ectopic expression of HCN1-ΔΔΔ and Kir2.1 channels, respectively. The engineered HCN1 construct, HCN1-ΔΔΔ whose S3-S4 linker has been shortened by deleting residues 235-237 to favor channel opening,^16,17,23 was chosen to best mimic native heteromultimeric I_f in adult LV CMs without having to consider poorly defined factors such as accessory subunits and cellular context as we previously described.^4,5 Figure 1A shows that all control LV CMs were electrically quiescent but capable of generating a single AP upon stimulation by a depolarizing current pulse (800 pA for 2 to 5 ms; n=17). Consistent with our previous results,^4,5 robust I_K1 (-32.1±1.2 pA/pF at -140mV) but no detectable I_f (-2.5±1.0 pA/pF at -140mV) could be recorded from the same control LV CMs.

Figure 1.

Representative action potential waveforms and averaged steady-state I-V relationships of control (A, E), Ad-CGI-Kir2.1- (B, F), Ad-CGI-HCN1ΔΔΔ- (C, G), as well as Ad-CGI-HCN1ΔΔΔ/Ad-CGI-Kir2.1 (D, H)-tranduced LV CMs.

We next studied the effect of transducing LV CMs with Ad-CGI-Kir2.1 (Figure 1B), Ad-CGI-HCN1-ΔΔΔ (Figure 1C) and both (Figure 1D). For Ad-CGI-Kir2.1-transduced LV CMs, there was a significant increase in I_K1 relative to control (-101.2±3.2 pA/pF at -140 mV; p<0.05), but I_f was still absent (-1.8±1.1 pA/pF at -140 mV; p<0.05). By contrast, I_K1 in Ad-CGI-HCN1-ΔΔΔ-trasnduced LV CMs (-34.8±5.5 pA/pF at -140 mV; n=5) was not different from control (p>0.05), but hyperpolarization-activated time-dependent I_f could be readily recorded (-28.5±3.9 pA/pF at -140 mV; p<0.05). For LV CMs simultaneously cotransduced by Ad-CGI-HCN1-ΔΔΔ and Ad-CGI-Kir2.1-WT, both I_K1 (-46.6±5.5 pA/pF at -140 mV; n=9) and I_f (-40.7±6.4 pA/pF at -140 mV; n=9) were substantially expressed. The corresponding current-voltage relationships are given (Figure 1E-H).

Coexpression of I_f and I_K1 promoted induced automaticity

To directly probe the functional consequences of overexpressing I_f or I_K1 alone or both in the AP waveform, we subjected the same cells presented above to current-clamp recordings. While Ad-CGI-Kir2.1-transduction significantly hyperpolarized the RMP (-69.7±0.7 mV vs. -66.0±0.7 mV of WT; p<0.05) (Figure 2A) and shortened the duration of AP elicited upon a depolarizing current stimulus (by ∼65%, n=7; p<0.05) (Figure 2B and Table 1), Ad-CGI-HCN1-ΔΔΔ uniquely induced ventricular automaticity (n=10), even without stimulation, with a firing rate of 181.1±13.1 bpm, a relatively depolarized maximum diastolic potential (MDP = -55.8±1.8 mV) and phase-4 depolarization (180±20 mV/s), consistent with what we recently reported.^4,5 Spontaneous APs were never observed in either control or Ad-CGI-Kir2.1-transduced LV CMs. Interestingly, simultaneous overexpression of I_f and I_K1 significantly promoted the induced automaticity by increasing the AP-firing frequency by ∼1.8-fold to 320.0±15.8 bpm (n=9; p<0.05) (Figure 2C). Relative to single Ad-CGI-HCN1-ΔΔΔ transduction, cotransduced LV CMs had significantly (p<0.05) hyperpolarized MDP (-57.1±0.9 mV, Figure 2A), hastened phase 4 depolarization (280±16 mV/s, Figure 2D) and shortened APD₉₀ (by 56% to 97.3±7.9 ms from 199.5±8.7 ms). All these AP parameters as well as cycle length and APD₅₀ of control and transduced LV CMs are summarized in Table 1.

Figure 2.

Bar charts showing (A) RMP or MDP; (B) APD₉₀; (C) beating rate; and (D) phase 4 slope of control, Ad-CGI-Kir2.1-transduced, Ad-CGI-HCN1ΔΔΔ-transduced, as well as Ad-CGI-HCN1ΔΔΔ/Ad-CGI-Kir2.1-cotranduced LV CMs. *, P<0.05.

Table 1.

Summary of AP parameters of control, singly- and co-transduced LV CMs. All values are means ± S.E.

	RMP (mV)	MDP (mV)	APD (ms)	APD50 (ms)	APD90 (ms)	Phase 4 Length (ms)	Phase 4 Slope (mV/s)	Cycle Length (ms)	Beating Rate (Beats/ min)
Control (17)	-66.0 ± 0.7	--	387.2 ± 10.4	299.5 ± 12.0	357.3 ± 10.0	--	--	--	--
Kir 2.1 (7)	-69.7 ± 0.6	--	138.6 ± 17.6	116.1 ± 17.2	127.8 ± 17.8	--	--	--	--
HCN1-ΔΔΔ (10)	--	-55.8 ± 1.8	207.4 ± 9.4	166.9 ± 6.9	199.5 ± 8.7	88.3 ± 9.6	180.0 ± 20.0	343.7 ± 24.9	181.1 ± 13.1
HCN1-ΔΔΔ + Kir2.1-WT (9)	--	-57.1 ± 1.0	107.37 ± 7.8	72.7 ± 7.8	97.3 ± 7.9	55.0 ± 4.5	280.0 ± 16.0	182.3 ± 6.8	320.0 ± 15.8

n numbers are given in bracket.

As I_K1 plays important roles in setting the resting membrane potential and in influencing the final phase of repolarization, the accelerated firing rate in cotransduced LV CMs may be a consequence of a greater magnitude of I_f due to the more negative membrane potential, or of a shorter AP duration due to a more rapid repolarization, or both. In order to dissect these I_K1-dependent effects contributing to the accelerated firing rate, we measured firing rate, action potential duration, phase-4 slope and phase-4 duration of automatic cotransduced LV CMs by pharmacological blockade of I_K1 with external Ba²⁺ at concentration of 0.1 mM and 0.3 mM. At baseline, the cycle length was 257.8±6.0 ms, i.e., 233.9±5.4 bpm with APD₉₀ of 94.8±1.0 ms and phase-4 length of 68.7±2.7 ms. As expected, external application of Ba²⁺(0.1 mM) shifted the maximal diastolic potential from -45.9±0.1 mV to -29.1±0.3 mV, which was associated with a reduction of firing rate for 26.8%. Although the APD₉₀ was lengthened by 58 ms, i.e., 61.2% increment, the phase-4 duration were prolonged by67.4 ms (98.2%) due to the flattening of the phase-4 slope by 51.2%. However, further increase in external Ba²⁺ concentration to 0.3 mM resulted in a complete abortion of automaticity. These data suggests that I_K1 accelerates the firing rate of I_f induced automaticity by enhancing the phase-4 slope as well as shortening AP duration.

Inter-relationships between I_K1, I_f and automaticity

For shedding mechanistic insights, we next investigated the intricate inter-relationships among I_K1, I_f and induced automaticity. To obtain a more global and accurate view, we plotted the various biophysical parameters recorded from all of control, singly- and doubly-transduced LV CMs to generate broader ranges for correlation analyses (i.e., the different experimental groups were included on the same plots in different symbols, but the entire data sets were fitted to liner regression). In spontaneously AP-firing cells, the frequency did not depend on the amplitude of I_f at -60 mV (I_{f, -60mV}) when I_K1 at -60 mV (I_{K1, -60mV}) was ≥-2 pA/pF (R²=0.03; Figure 4A, solid squares). However, the induced firing rate displayed a strong positive correlation to I_{f, -60mV} when -2>I_{K1, -60mV} >-4 pA/pF (R²=0.8; Figure 4A, open circles). Such a positive correlation diminished (to R²=0.3) as I_{K1, -60mV} increased to ≤ -4 pA/pF (Figure 4A, solid triangle). In stark contrast to I_f, the induced automaticity did not depend on I_{K1, -60mV} when I_{f, -60mV} was similarly fixed (i.e. I_{f, -60mV} ≤ 2pA/pF, -2 < I_{f, -60mV} ≤ 4pA/pF and I_{f, -60mV} >4pA/pF; Figure 4B).

Figure 4.

(A) Correlations between I_f and the frequency of spontaneously AP-firing cells when I_{K1, -60mV} ≥-2 pA/pF (■), -2>I_{K1, -60mV} >-4 pA/pF (○) and I_{K1, -60mV} ≤-4 pA/pF (▲). (B) Correlations between I_K1 and the firing frequency rate when I_{f, -60mV} ≥-2 pA/pF (■),-2>I_{f, -60mV} >-4 pA/pF (○) and I_{f, -60mV} ≤-4 pA/pF (▲).

To further dissect the inter-relationship, we similarly plotted the Phase 4 slope, Phase 4 length, and APD₉₀ against I_{f, -60mV} with fixed ranges of I_{K1, -60mV}. Similar to the firing frequency, all these AP parameters were also correlated to I_{f, -60mV} only when -2>I_K1>-4 pA/pF. Figure 5A shows that steeper Phase 4 slopes could be attained with higher I_f densities (R²=0.7). Both the Phase 4 length (Figure 5C; R²=0.6) and APD₉₀ (Figure 5E; R²=0.7) were negatively correlated to I_{f, -60mV}. Unlike I_{f, -60mV}, I_{K1, -60mV} displayed correlation to neither the Phase 4 slope (to R²=0.02) nor Phase 4 length (to R²=0.04) when -2> I_{f, -60mV} >-4 pA/pF (Figures 5B & D). As anticipated, however, APD₉₀ was correlated to I_{K1, -60mV} (to R²=0.4; Figure 5F).

Figure 5.

(A) Correlations between I_f and the phase 4 slope of spontaneously AP-firing cells when I_{K1, -60mV} ≥-2 pA/pF (■), -2>I_{K1, -60mV} >-4 pA/pF (○) and I_{K1, -60mV} ≤-4 pA/pF (▲). (B) Correlations between I_K1 and the phase 4 slope when I_{f, -60mV} ≥-2 pA/pF (■), -2>I_{f, -60mV} >-4 pA/pF (○) and I_{f, -60mV} ≤-4 pA/pF (▲). (C) Correlations between I_f and the phase 4 length of spontaneously AP-firing cells when I_{K1, -60mV} ≥-2 pA/pF (■), -2>I_{K1, -60mV} >-4 pA/pF (○) and I_{K1, -60mV} ≤-4 pA/pF (▲). (D) Correlations between I_K1 and the phase 4 length when I_{f, -60mV} ≥-2 pA/pF (■), -2>I_{f, -60mV} >-4 pA/pF (○) and I_{f, -60mV} ≤-4 pA/pF (▲). (E) Correlations between I_f and the APD₉₀ of spontaneously AP-firing cells when I_{K1, -60mV} ≥-2 pA/pF (■), -2>I_{K1, -60mV} >-4 pA/pF (○) and I_{K1, -60mV} ≤-4 pA/pF (▲). (F) Correlations between I_K1 and the APD₉₀ when I_{f, -60mV} ≥-2 pA/pF (■), -2>I_{f, -60mV} >-4 pA/pF (○) and I_{f, -60mV} ≤-4 pA/pF (▲).

Isoproterenol accelerates firing rate of Ad-CGI-HCN1-ΔΔΔ transduced LV CMs

To test the functionality of automatic Ad-CGI-HCN1-ΔΔΔ-transduced LV CMs as a pacemaker, we investigated their response to adrenergic stimulation by superfusion of 1 μM isoproterenol for 5 minutes. The cycle length reduced from 377±8.7 to 318±10.1 ms (p=0.001; n=3) representing 18% increase in the firing rate (Figure 6). This is largely due to the increase of phase-4 slope (166±9.5 mV/s to 265±11.9 mV/s, p=0.003), while the APD₉₀, maximal diastolic potential and take-off potential were largely unchanged after isoproterenol treatment.

Figure 6.

Effects of adrenergic stimulation by superfusion of 1 μM isoproterenol for 5 minutes (3 cells from a single animal) on HCN1-ΔΔΔ-transduced LV CMs.

Discussion

In recent years, efforts to generate bioartificial pacemakers as potential alternatives or supplements to electronic devices for treating cardiac rhythm disorders have been pursued. Experimentally, the one-sided approaches of maximizing I_f ^4,12 or minimizing its “antagonizing” I_K1 ^10,29 to convert nonpacing CMs to spontaneously firing pacemaker cells have been investigated. Our previous in silico and in vivo HCN-gene-transfer experiments of LV CMs demonstrate that both I_f and I_K1 play critical roles in the induction of automaticity ^{4, 5, 21}. While the two strategies appear to be conceptually synergistic, their complementarity has not been investigated. Indeed, although the primary functions of I_K1 and I_f have been largely presumed to antagonize each other, our present study provides novel experimental supports that they do not necessarily counter-balance each other. Firstly, adenovirus-mediated expression of I_f (or I_K1) in LV CMs did not alter I_K1 (or I_f). These observations suggest that the two “opposing” currents do not undergo compensatory electrophysiological changes against each other when their relative expression is altered. As such, the resultant AP phenotypes can be principally attributed to the changes of either I_f or I_K1 alone. Secondly, simultaneous overexpression of I_f and I_K1 in LV CMs generates the highest oscillation frequency of automaticity. Correlation analysis to dissect the intricate mechanistic relationships among various I_f, I_K1 and AP biophysical parameters indicates that, mechanistically, the frequency of induced AP-firing cells was dependent upon I_f when I_{K1, -60mV} was between -2 and -4 pA/pF. However, the induced automaticity did not rely on I_K1 when I_{f, -60mV} was similarly fixed. Thus, the data were in accordance with the notion that I_K1 functions only as a binary switch for automaticity without providing a direct means for frequency modulation^4,5, and further indicate that the strategies of I_K1 suppression and I_f overexpression are not necessarily synergistic in conferring automaticity on quiescent cardiac muscles.

While the level of induced automaticity by simultaneous I_f and I_K1 overpression is tachycardic and, as such, not desirable for bioartificial pacing, the findings have significant implications. Clinically, the implantation site for bioartificial pacemaker needs to be carefully considered because the different extents of innervations and the differential responses of I_K1 and I_f to endogenous sympathetic and parasympathetic inputs might significantly alter the I_f/I_K1 ratio from the original level, thereby leading to unanticipated tachyarrhythmias. Mechanistically, it is clear that the I_f/I_K1 ratio is a more effective parameter to engineer for automaticity induction to generate bioartificial pacemakers than manipulating either I_f or I_K1 per se. As a complex and heterogeneous tissue, the SAN displays gradual changes in phenotypic properties such as ionic current densities, potential profile and gap junction expression from the center (dominant pacemaker range) to the periphery (subsidiary pacemaker range) of the nodal cells.³⁰ While the maximum diastolic potential is more negative, the AP is shorter with faster spontaneous rate in the center than in the periphery, AP is first initiated in the center and then transmitted to the periphery of the SAN.³¹ Such gradients ensure normal AP propagation in appropriate directions. Our present findings therefore raise the possibility that bioengineered cells that mimic the central and peripheral SA nodal cells showing native AP gradient can be best engineered by targeted firing rate engineering via virus cotransduction to improve the functional efficacy.

Of note, the current study has certain limitations. The choice of a gate-engineered version of HCN1 instead of wild-type HCN4 (the predominant HCN in genuine biological pacemakers) appears nonphysiological. However, this is because overexpression of wild-type HCN4 alone does not suffice to cause automaticity in adult LV CMs, probably due to the vast differences in ion channel panoplies as well as certain contentious β-subunits between sinoatrial pacemaker cells and normally quiescent adult LV CMs.¹³ Consequently, the use of this gate-engineered version of HCN1 may compensate for any context-dependent gating effects such as other modifying subunits and factors that may be present in SA nodal cells but not CMs. Second, the expression of I_f and I_K1 in the present work were sporadic in nature due to the available transduced method, which may render a large variation in certain physiological measurement such as APD.

Conclusion

Based on the present results, we conclude that the induction of automaticity in LV CMs requires a fine balance of I_K1 and I_f. Indeed, I_K1 synergistically interacts with I_f by maintaining the voltage changes within a range where HCN channels can most effectively operate during a dynamic cardiac cycle.

Figure 3.

The effect of external application of Ba²⁺ on the automaticity in Ad-CGI-HCN1ΔΔΔ/Ad-CGI-Kir2.1-cotranduced LV CMs. Representative action potential waveforms at baseline, with 0.1 mM BaCl₂, and 0.3 mM BaCl₂ are shown, respectively.