Molecular dissection of the inward rectifier potassium current (I_K1) in rabbit cardiomyocytes: evidence for heteromeric co-assembly of K_ir2.1 and K_ir2.2

Abstract

Cardiac inward rectifier K⁺ currents (I_K1) play an important role in maintaining resting membrane potential and contribute to late phase repolarization. Members of the K_ir2.x channel family appear to encode for I_K1. The purpose of this study was to determine the molecular composition of cardiac I_K1 in rabbit ventricle. Western blots revealed that K_ir2.1 and K_ir2.2, but not K_ir2.3, are expressed in rabbit ventricle. Culturing rabbit myocytes resulted in a ∼50% reduction of I_K1 density after 48 or 72 h in culture which was associated with an 80% reduction in K_ir2.1, but no change in K_ir2.2, protein expression. Dominant-negative (DN) constructs of K_ir2.1, K_ir2.2 and K_ir2.3 were generated and tested in tsA201 cells. Adenovirus-mediated over-expression of K_ir2.1dn, K_ir2.2dn or K_ir2.1dn plus K_ir2.2dn in cultured rabbit ventricular myocytes reduced I_K1 density equally by 70% 72 h post-infection, while AdK_ir2.3dn had no effect, compared to green fluorescent protein (GFP)-infected myocytes. Previous studies indicate that the [Ba²⁺] required for half-maximum block (IC₅₀) differs significantly between K_ir2.1, K_ir2.2 and K_ir2.3 channels. The dependence of I_K1 on [Ba²⁺] revealed a single binding isotherm which did not change with time in culture. The IC₅₀ for block of I_K1 was also unaffected by expression of the different DN genes after 72 h in culture. Taken together, these results demonstrate functional expression of K_ir2.1 and K_ir2.2 in rabbit ventricular myocytes and suggest that macroscopic I_K1 is predominantly composed of K_ir2.1 and K_ir2.2 heterotetramers.

The inward rectifier K⁺ current (I_K1) in cardiac myocytes plays a significant role in maintaining the resting membrane potential and in shaping the late repolarization phase of the action potential (Shimoni et al. 1992; Lopatin & Nichols, 2001). The molecular composition of cardiac I_K1 remains incompletely understood. On the basis of sequence homology, inward rectifier K⁺ channels have been classified into seven subfamilies (K_ir1 to K_ir7) (Doupnik et al. 1995; Nichols & Lopatin, 1997). Transcriptional analysis coupled with functional characterization of cloned K_ir channels suggest that members of the K_ir2 subfamily underlie cardiac I_K1 (Dixon & McKinnon, 1994; Barry et al. 1995; Brahmajothi et al. 1996). Isolated cardiac membrane patches show single channel I_K1 conductances in the range 9–41 pS (Sakmann & Trube, 1984; Burnashev & Zilberter, 1986; Josephson & Brown, 1986; Wahler, 1992; Wible et al. 1995), possibly suggesting that multiple channels contribute to I_K1. In guinea-pig cardiomyocytes three different inward rectifier single channel conductances (≈34 pS, ≈24 pS and ≈11 pS) have been identified and linked to homotetrameric channels formed of K_ir2.2, K_ir2.1 and K_ir2.3 subunits, respectively (Liu et al. 2001). However, K_ir2.1 channels display unitary currents in the range 2–33 pS (Picones et al. 2001). Furthermore, knockout mice lacking the K_ir2.1 gene displayed no detectible I_K1, whereas I_K1 was reduced by ≈50% in K_ir2.2 knockout mice (Zaritsky et al. 2001), suggesting that I_K1 is formed by obligate heterotetrameric assembly between K_ir2.1 and K_ir2.2 channel proteins. This conclusion contradicts an earlier oocyte expression study, which showed that K_ir2.x subunits cannot form heterotetramers, although the same study provided co-immunoprecipitation evidence for co-assembly between K_ir2.1 and K_ir2.2 in mammalian cells (Tinker et al. 1996). Further evidence for heteromeric assembly of K_ir2 subunits was provided by a recent study combining co-immunoprecipitation experiments, yeast two-hybrid assays and co-expression studies of K_ir2.x-K_ir2.y concatemers (Preisig-Muller et al. 2002). Since the molecular identity of cardiac I_K1 remains unclear, we created adenoviruses carrying dominant-negative (DN) mutants of candidate I_K1 genes (K_ir2.1, K_ir2.2 or K_ir2.3) and examined their ability to down-regulate I_K1 in cultured rabbit ventricular myocytes.

METHODS

Site-directed mutagenesis and subcloning

Site-directed mutagenesis for K_ir2.2 and K_ir2.3 was performed using the method developed by Kunkel (1985), as described previously (Cho et al. 2000).

Heterologous expression of K_ir2 subunits in tsA201 cells

Constructs for heterologous expression of K_ir channels in tsA201 cells were subcloned into pBi-G (Clontech) which contains a bi-directional promotor. A gene encoding enhanced green fluorescence protein (EGFP) was inserted on the other side of the bi-directional promotor as an indicator for successful transfection using Lipofectamine 2000 (Gibco-BRL) according to the manufacturer's instructions.

Preparation of rabbit cardiomyocytes

Ventricular myocytes were isolated by enzymatic dissociation using a method previously reported in detail (Diaz et al. 1999). Purification of the cells obtained after digestion for cardiac myocytes was achieved by employing a gravity albumin gradient. Visual inspection of the cultured cells at different time points (2 h, 24 h and 48 h) revealed no significant contamination with non-cardiomyocytes. Hearts were excised from rabbits anaesthetised with pentobarbital (60 mg kg⁻¹). All procedures conformed to the guidelines provided by the Canadian Council on Animal Care. The myocytes were cultured on laminin-coated culture dishes using M199 (Sigma) supplemented with 4 mM L-carnitine, 5 mM taurine, 5 mM creatine, 2 mg ml⁻¹ bovine serum albumin (BSA), 0.1 μM insulin, 0.1 nM triiodothyronine, 50 μg ml⁻¹ gentamycin, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin. The medium was changed 2 h after plating to remove loosely attached cells (i.e. ‘differential attachment') and thereafter at 24 h intervals.

Immunoblotting

The myocytes were prepared as described above and plated at a density of approximately 8 × 10⁵ cells per 10 cm dish. After 2 or 48 h in culture, the myocytes were washed three times with cold phosphate-buffered saline before they were harvested in a solution containing 25 mM Tris, 5 mM EGTA, 2 mM EDTA, 0.2 mM PMSF, 5 μg ml⁻¹ pepstatin A, 5 μg ml⁻¹ leupeptin and 50 μg ml⁻¹ aprotinin; pH 7.5. Rabbit brain membranes were prepared as described previously (Hartshorne & Catterall, 1984). After protein determination, myocyte and brain membranes were frozen in liquid nitrogen. For Western blot analysis, either 50 μg total myocyte protein or 50 μg brain membrane was diluted with a 4 × sample buffer containing 8% SDS, 0.25 M Tris–HCl, 30% glycerol, 0.4 mM DTT and 0.001% bromophenol blue before fractionation on a 10% polyacrylamide-SDS gel. After electrophoretic transfer to a polyvinyl difluoride membrane (Bio-Rad), samples were incubated with K_ir2.1 antibody (1:2000) (Tinker et al. 1996), polyclonal K_ir2.2 antibody (1:300) (Raab-Graham & Vandenberg, 1998) or polyclonal K_ir2.3 antibody (1:300) (Chemicon International, Inc.). Bound primary antibody was detected with horseradish peroxidase-conjugated goat anti-rabbit IgG (1:10000) (Amersham). Immunoreactivity was visualised using the enhanced chemiluminescence (ECL) reagent (Amersham).

Construction of recombinant adenoviruses

Four first generation type 5 recombinant adenoviruses were constructed by subcloning the following genes into an adenoviral shuttle vector downstream of a cytomegalovirus promoter (Johns et al. 1997): AdGFP, AdGFPK_ir2.1C122S, AdGFPK_ir2.2C123S and AdGFPK_ir2.3C113S (GFP, green fluorescent protein). The dominant negative adenoviruses expressed mutant Kir2.x channels along with GFP. The AdGFP virus was purchased from Quantum Biotechnologies Inc., Montreal, Canada, while AdGFPK_ir2.1C122S (AdK_ir2.1dn), AdGFPK_ir2.2C123S (AdK_ir2.2dn) and AdGFPK_ir2.3C113S (AdK_ir2.3dn) were generated using the pAdeasy system (He et al. 1998) with pAdtrackCMV as a shuttle vector and pAdeasy-1 as adenoviral backbone. All viruses were plaque-purified and virus titres were determined using the 50% tissue culture infectious dose method (TCID₅₀): AdGFP, 3.6 × 10⁹ TCID₅₀ ml⁻¹; AdK_ir2.1dn, 1.4 × 10¹⁰ TCID₅₀ ml⁻¹; AdK_ir2.2dn, 1.4 × 10⁹ TCID₅₀ ml⁻¹; and AdK_ir2.3dn, 2.8 × 10⁹ TCID₅₀ ml⁻¹.

Electrophysiology in cultured rabbit myocytes and tsA201 cells

The myocytes were prepared as described above and then cultured on laminin-coated glass coverslips at a density of 4–5 × 10⁴ cells per 35 mm dish. After allowing 2 h for attachment of the myocytes, the solution was changed and the cells were either used directly for patch-clamp recordings (see below) or infected with the different viruses: AdGFP, 25 TCID₅₀ per myocyte; AdK_ir2.1dn, 5 or 10 TCID₅₀ per myocyte; AdK_ir2.2dn 5 or 10 TCID₅₀ per myocyte; or AdK_ir2.1dn and AdK_ir2.2dn, 5 TCID₅₀ per myocyte. The amount of virus was sufficient to allow for an infection efficiency of approximately 95%. In agreement with the results from a previous publication (Rust et al. 1998) infected myocytes did not display morphological alterations when compared to time-matched non-infected myocytes. After 48 h or 72 h in culture, glass coverslips with myocytes were transferred into a small recording chamber mounted on the stage of an inverted microscope (Olympus IX 70). The chamber was perfused with physiological salt solution containing (mM): NaCl 140, KCl 4, MgCl₂ 1, CaCl₂ 2, Hepes 10, glucose 10 and CdCl₂ 0.5. Whole-cell voltage-clamp recordings were made using an Axopatch 200A amplifier (Axon Instruments). Recording pipettes were prepared from thin-walled borosilicate glass (1.5 mm diameter, World Precision Instruments) using a Flaming-Brown micropipette puller (Sutter Instruments). The pipette resistance was in the range 2–4 MΩ. The pipette solution contained (mM): potassium aspartate 90, KCl 20, MgCl₂ 1, Hepes 10, EGTA 10 and Na₂ATP 5; pH was adjusted to 7.3 with 30 mM KOH. Myocytes suitable for patch clamping were initially identified based on their morphological appearance (rod-shaped and absence of blebs), not on their fluorescence properties. Once identified, electrical recordings were only made on myocytes showing GFP fluorescence (95% of myocytes). After membrane rupture, the cell capacitance was estimated by integrating the capacity currents, followed by compensating cell capacitance and pipette series resistance.

To measure I_K1, myocytes were held at a membrane potential of −40 mV and voltage steps from −130 to −10 mV in 10 mV increments were applied for 500 ms with an interval of 1 s. To determine the dependence of current block on Ba²⁺ concentration at steady state level the pulse duration was extended to 2–4 s. The transient outward K⁺ current (I_to) was measured from a holding potential of −80 mV and a 100 ms prepulse to −40 mV was used to inactivate the Na⁺ current followed by a 1000 ms test pulse to +60 mV. The data were acquired with pCLAMP6 software (Axon Instruments) and analysed with pCLAMP version 8 (Axon Instruments). The data were not corrected for the liquid junction potential, which was 7.2 mV.

Statistics

Data were analysed using ANOVA followed by the Student-Newman-Keuls test. All values are given as means ± s.e.m.

RESULTS

Molecular composition of I_K1 in rabbit ventricular myocytes

In agreement with previously published results (Mitcheson et al. 1996; Christe, 1999; Veldkamp et al. 1999), Fig. 1 shows that the density of I_K1 in adult rabbit myocytes was reduced by approximately 50% after 48 h in culture with no further reductions occurring after 72 h in culture. In association with these reductions of I_K1 in ventricular myocytes after 48 h in culture, there was an 80% reduction in K_ir2.1 expression and no change in K_ir2.2 protein levels when compared to myocytes cultured for 2 h (Fig. 2). No band was detected at the expected molecular weight of K_ir2.3 (i.e. ≈50 kD) in rabbit ventricular myocytes using an antibody for K_ir2.3 at either time point despite strong expression in rabbit brain tissue or in rabbit myocytes infected with the K_ir2.3dn in adenoviral vector.

Figure 1. Effect of culture on I_K1 in rabbit ventricular myocytes.

A, typical Ba²⁺-subtracted membrane currents elicited by 500 ms voltage steps ranging from −130 mV to −10 mV (by 10 mV) from a holding potential of −40 mV. B, peak I_K1 densities from myocytes cultured for 2 h (□) were significantly decreased (P < 0.05) after culturing for 48 h (▵) but stayed unchanged upon further culturing (72h, •) for membrane potentials between −130 and −80 mV.

Figure 2. Western blots for native K_ir2.1, K_ir2.2 and K_ir2.3 proteins in rabbit ventricular myocytes.

Western blots showing typical signals obtained by probing rabbit ventricular myocyte preparations harvested after 2 or 48 h in culture along with rabbit brain controls using specific antibodies for K_ir2.1 (A), K_ir2.2 (B) or K_ir2.3 (C). The bar graphs show summarized results of densitometric measurements for K_ir2.1 (n = 3) and K_ir2.2 (n = 7) after 48 h in culture normalized to expression levels after 2 h in culture.

The Western blotting results above suggest that K_ir2.1 and K_ir2.2 channels, but not K_ir2.3 channels, are expressed in rabbit ventricular myocytes. In order to investigate further the molecular underpinnings of rabbit cardiac I_K1, we created dominant-negative (DN) adenoviruses for transfection of cultured rabbit ventricular myocytes. The DN properties of a mutant K_ir2.1 channel (i.e. K_ir2.1dn) lacking a critical disulfide bond in the P-loop have been described previously (Cho et al. 2000). The equivalent absolutely conserved cysteine was converted to serine (i.e. C123S) in the K_ir2.2 gene to generate a DN gene (K_ir2.2dn). As shown in Fig. 3, co-expression of K_ir2.1dn with wild-type (WT) K_ir2.1 or K_ir2.2 in tsA201 cells resulted in similar conductance reductions compared to expression of WT channels alone. Similar reductions in conductance were observed when K_ir2.2dn was co-expressed with WT channels. These reductions in conductance observed in tsA201 cells co-expressing WT along with DN genes were not related to the transfection levels since doubling the amount of K_ir2.1 or K_ir2.2 DNA yielded significantly higher slope conductances (Fig. 3). Collectively, these results suggest that K_ir2.1 and K_ir2.2 can co-assemble to form heterotetramers in tsA201 cells.

Figure 3. Validation of dominant-negative (DN) genes in tsA201 cells.

Bar graphs showing the membrane conductance measured in tsA201 cells transfected with either wild-type K_ir2.1 (A) or K_ir2.2 (B) genes. Equimolar co-transfection with either K_ir2.1dn or K_ir2.2dn constructs significantly diminished the slope conductance of the wild-type channels. Doubling the amount of wild-type construct (‘K_ir2.1 double’ or ‘K_ir2.2 double’) increased the slope conductance correspondingly. Membrane potentials were held at −40 mV and voltage steps from −130 to −10 mV in 10 mV increments were applied for 500 ms with an interval of 1 s. Slope conductance was calculated from the linear range of the current–voltage relationship between −130 and −80 mV). * Significantly different at P < 0.05 relative to K_ir2.1 (A) or K_ir2.2 (B).

To dissect the contribution of K_ir2.1 and K_ir2.2 channels to I_K1, rabbit ventricular myocytes were transfected with recombinant adenoviruses containing DN K_ir2.1dn or K_ir2.2dn genes plus GFP (i.e AdK_ir2.1dn and AdK_ir2.2dn) or GFP alone (AdGFP). Expression of the DN genes was confirmed by Western blotting analysis (Fig. 2) and expression of GFP was verified using fluorescence microscopy. For the transfection studies using DN AdK_ir2.1dn and AdK_ir2.2dn, infection with AdGFP alone was used as the control group. After transfection with AdGFP, I_K1 densities were not different from non-infected cultured myocytes (data not shown) establishing that viral infection per se does not affect cardiac I_K1. In contrast, 72 h after transfection cultured rabbit ventricular myocytes expressing AdK_ir2.1dn or AdK_ir2.2dn showed a marked (≈70%) reduction (P < 0.05) of I_K1 density, measured at −130 mV (Fig. 4). Interestingly, Fig. 4 shows that co-infection with both AdK_ir2.1dn and AdK_ir2.2dn caused reductions in I_K1 to levels indistinguishable from those observed in myocytes infected with AdK_ir2.1dn or AdK_ir2.2dn alone. Moreover, the extent of reduction of I_K1 observed 48 h after co-transfection with AdK_ir2.1dn and AdK_ir2.2dn was identical (i.e. ≈70%) to that observed after 72 h (data not shown), suggesting that maximal I_K1 reductions of about 70% were achieved under our experimental conditions by over-expression of DN K_ir2.1 and/or K_ir2.2 genes.

Figure 4. Effect of infection of different AdK_ir2.xdn constructs on I_K1 in cultured rabbit ventricular myocytes.

A, original traces showing typical Ba²⁺-substracted I_K1 recorded from myocytes 72 h after infection with the indicated constructs. B, Ba²⁺-subtracted peak currents as a function of voltage in cultured rabbit myocytes 72 h after infection. I_K1 density (from −130 mV to −90 mV) decreased significantly in myoctyes infected with AdK_ir2.1dn, AdK_ir2.1dn or AdK_ir2.1dn plus AdK_ir2.1dn compared to the control myocytes (AdGFP, P < 0.05). No significant difference was detected between I_K1 density recorded after infection with AdK_ir2.1dn, AdK_ir2.2dn or AdK_ir2.1dn plus AdK_ir2.2dn.

The results above show that over-expression of dominant-negative K_ir2 genes could not completely suppress I_K1 even after 72 h in culture. Despite the absence of a K_ir2.3-specific band of the appropriate size in the immunoblotting studies, it remains conceivable that the residual current is generated by K_ir2.3 channels possibly as a result of up-regulation due to prolonged culture or due to K_ir2.1 and K_ir2.2 current suppression. To investigate this possibility, a DN K_ir2.3 adenovirus (AdK_ir2.3dn) expressing K_ir2.3dn plus GFP was created. Figure 5A depicts the DN effects of AdK_ir2.3dn in tsA201 cells. Infection of cultured myocytes with the AdK_ir2.3dn virus had no effect on I_K1 densities (Fig. 5B). Taken together with the results of the DN K_ir2.1dn and K_ir2.2dn experiments, these findings suggest that rabbit cardiac I_K1 is largely generated by heterotetramers of K_ir2.1 and K_ir2.2 channel proteins.

Figure 5. Effects of a dominant-negative K_ir2.3 construct.

A, slope conductance measured in tsA201 cells expressing K_ir2.3 or K_ir2.3 plus K_ir2.3dn. Cells were held at a membrane potential of −40 mV and voltage steps from −130 to −10 mV in 10 mV increments were applied for 500 ms with a 1 s interval. Slope conductance was calculated from the linear range of the current–voltage relationship (−130 to −80 mV). B, Ba²⁺-subtracted peak I_K1 as a function of voltage measured in rabbit ventricular myocytes expressing either AdGFP or AdK_ir2.3dn. * Significant difference at P < 0.05 between K_ir2.3 alone and K_ir2.3 plus K_ir2.3dn.

To determine whether the observed reductions in I_K1 density by the DN actions of K_ir2.1dn or K_ir2.2dn are specific to I_K1 channels and not caused by non-specific actions, we also measured the transient outward K⁺ current (I_to) in myocytes infected with AdK_ir2.1dn or AdK_ir2.2dn. Figure 6A shows typical recordings of I_to from a control myocyte infected with AdGFP as well as myocytes co-infected with AdK_ir2.1dn and/or AdK_ir2.2dn. The pooled data in Fig. 6B demonstrate that no significant difference in I_to density could be observed under any of the conditions studied, suggesting that the effects of AdK_ir2.1dn and AdK_ir2.2dn are specific for I_K1 channels.

Figure 6. Effects of infection with AdK_ir2.X genes on transient outward K⁺ currents in cultured rabbit ventricular myocytes.

A, typical recordings of I_to measured at +60 mV with a prepulse at +40 mV to inactivate sodium currents in myocytes infected with AdGFP or AdK_ir2.1dn plus AdK_ir2.2dn 72 h post-infection. B, summary of the effects of infection on I_to in cultured myocytes following infection with the indicated AdK_ir2.X genes.

Characteristics of I_K1 block by Ba²⁺ in cultured ventricular rabbit myocytes

The results above suggest that I_K1 in rabbit ventricular myocytes is generated by heterotetrameric co-assembly of K_ir2.1 and K_ir2.2 channel genes. Previous studies have established that homomeric K_ir2.1 channels are about 5-to 10-fold less sensitive to blockade by Ba²⁺ than K_ir2.2 channels (Liu et al. 2001; Preisig-Muller et al. 2002), and we have confirmed this (data not shown). Armed with this information, we explored the blockade of I_K1 by Ba²⁺ in isolated rabbit ventricular myocytes. Figure 7A shows the effects of various externally applied Ba²⁺ concentrations on I_K1 in rabbit ventricular myocytes cultured for 2 h or 48 h. The dependence of I_K1 on different Ba²⁺ concentrations was accurately fitted using a mono-sigmoidal binding isotherm function. It was interesting that neither the estimations of IC₅₀ nor the Hill coefficients (HC) for block of I_K1 by Ba²⁺ differed (P > 0.05) between myocytes cultured for 2 versus 48 h (2 h: IC₅₀, 1.8 ± 0.4 μM; HC, 1.0 ± 0.2; n = 8; 48 h: IC₅₀, 1.7 ± 0.2 μM; HC, 1.3 ± 0.1, n = 8), despite the large reduction in K_ir2.1 expression observed after culturing. In addition, Fig. 7B shows that block of cardiac I_K1 in myocytes infected with AdGFP (control), AdK_ir2.1dn, AdK_ir2.2dn, AdK_ir2.3dn or AdK_ir2.1dn plus AdK_ir2.2dn as a function of the Ba²⁺ concentration could also be well described by a single binding curve with no significant difference in IC₅₀ values and Hill coefficients (P > 0.05) among the different groups (IC₅₀: AdGFP, 1.6 ± 0.1; AdK_ir2.1dn, 1.8 ± 0.2; AdK_ir2.2dn, 1.1 ± 0.1; AdK_ir2.3dn, 1.6 ± 0.2; AdK_ir2.1dn + AdK_ir2.2dn, 1.6 ± 0.2). These results establish that the Ba²⁺ blocking properties of I_K1 currents are invariant with current density, suggesting an invariant molecular composition of I_K1 channels.

Figure 7. Blockade of cardiac I_K1 by Ba²⁺ in cultured rabbit ventricular myocytes.

A, Ba²⁺ dose–response curves for I_K1 in non-infected myocytes cultured for 2 h or 48 h at −100 mV. The data were fitted with a single binding isotherm I_Ba/I_Con = 1/(1 + [Ba²⁺]ⁿ/IC₅₀ⁿ) where I_Ba is the I_K1 in the presence of Ba²⁺ while I_Con is the I_K1 before application of Ba²⁺. B, Ba²⁺ dose–response curve for I_K1 measured at −100 mV in myocytes cultured for 48 h following over-expression of various AdK_ir2.x genes as indicated. Despite marked differences in the level of I_K1 in the various groups, there was no measurable difference in the IC₅₀ values for Ba²⁺.

DISCUSSION

Previous studies suggested that I_K1 channels in mouse hearts are formed by heteromeric assembly of K_ir2.1 and K_ir2.2 subunits (Zaritsky et al. 2001). Since these mouse studies were complicated by potential compensatory adaptations that occur following gene ablation, we attempted to explore the molecular basis of ventricular I_K1 using virus-mediated over-expression of dominant-negative K_ir2 subunits. Our Western blotting results demonstrated that K_ir2.1 and K_ir2.2, but not K_ir2.3, are expressed in rabbit ventricular myocytes, consistent with previous mRNA measurements in mouse (Kurachi & Takahashi, 1996) and rat (Nagashima et al. 2001). These conclusions are also similar to those based on results in guinea-pig myocytes using unitary I_K1 amplitude distributions (Liu et al. 2001) and in human myocardium (Wang et al. 1998) showing that K_ir2.1 and K_ir2.2 are the primary determinants of cardiac I_K1, with very little, but measurable, contributions from K_ir2.3 channels.

Transfection of rabbit ventricular myocytes with K_ir2.1dn, K_ir2.2dn or K_ir2.1dn + K_ir2.2dn reduced I_K1 levels by the same amount after 72 h of culture, while K_ir2.3dn over-expression had no effect. Reduction of current to the same level would not be expected if significant proportions of I_K1 channels contained K_ir2.1 or K_ir2.2 subunits as either homotetramers or (possibly) heterotetramers with other unidentified K_ir α-subunits. Heterotetrameric assembly is also consistent with our observation that blockade of I_K1 by Ba²⁺ was described by a single, rather than a double, binding isotherm equation since the IC₅₀ for Ba²⁺ differs by 5-to 10-fold between the K_ir2.1 and K_ir2.2 homomers (Liu et al. 2001; Preisig-Muller et al. 2002). Similar heterotetrameric assembly of K_v4.2 and K_v4.3 occurs in the formation of mouse cardiac I_to (Guo et al. 2002). In addition, K_ir2.2 expression was unchanged during culture of rabbit ventricular myocytes despite large reductions in I_K1 and K_ir2.1 expression. Since our Western blotting results cannot distinguish between K_ir2 subunits in the sarcolemma and non-sarcolemmal membranes, it appears that K_ir2.1 assists in translocation of heterotetrameric I_K1 channels to the surface membrane as proposed previously from transgenic mice studies (Zaritsky et al. 2001). Consistent with this suggestion, K_ir2.2 currents expressed in oocytes and tsA201 cells are routinely 10-fold less than K_ir2.1 currents, despite equimolar levels of RNA or DNA (H. C. Cho & P. H. Backx, unpublished data) which could be related to the observation that transfer of the distal C-terminus from K_ir2.2 to K_ir2.1 diminished surface expression (Ma et al. 2001).

It is remarkable that the biophysical properties of I_K1, including blockade by Ba²⁺ (i.e. the IC₅₀ and the Hill coefficient), did not vary when I_K1 was reduced following culturing or in response to transfection with K_ir2.1dn and/or K_ir2.2dn. This observation is similar to the absence of changes in single channel I_K1 conductance or open probability reported previously during culture of rabbit myocytes (Veldkamp et al. 1999). While these findings argue in favour of a fixed stoichoimetry of I_K1 channels in rabbit ventricle, the relationship between channel stoichiometry and the biophysical properties of I_K1 channels is likely to be complex and further studies are clearly necessary to determine the precise stoichiometry of rabbit cardiac I_K1.

The degree of reduction of I_K1 density in rabbit myocytes was the same after 48 or 72 h of co-transfection with AdK_ir2.1dn and AdK_ir2.2dn. Moreover, doubling the level of AdK_ir2.1dn and AdK_ir2.2dn was also unable to induce further I_K1 reductions (C. Zobel, H. C. Cho & P. H. Backx, unpublished data). These observations suggest that the maximal reduction of I_K1 possible in our experimental conditions is about 70%. The molecular composition of the remaining I_K1 is unclear. As a result of the finite turnover rates of endogenous I_K1 channels, the remaining current might still be generated by K_ir2.1 and K_ir2.2 subunits, although, at first glance, this suggestion is inconsistent with the observation that the degree of current reduction was equivalent at 48 and 72 h post-transfection with AdK_ir2.1dn and AdK_ir2.2dn. It is nevertheless conceivable that a dynamic balance between new channel production and DN transgene expression is reached between 48 and 72 h, particularly if the level of DN transgene expression declines after 48 h in culture as has been shown previously (Parks et al. 1999; Seharaseyon et al. 2000). Our inability to fully eliminate I_K1 in myocytes using K_ir2.1dn and K_ir2.2dn could also arise from incomplete suppression of I_K1, despite our previous studies establishing a potent dominant-negative action of K_ir2.1dn in oocytes (Cho et al. 2000). Consistent with this possibility, K_ir2.1dn and K_ir2.2dn were far less effective inhibitors of wild-type K_ir2.X currents in tsA201 cells than in oocytes (authors' unpublished observations). Therefore, it is possible that other dominant-negative channel mutants of K_ir2.1 and K_ir2.2 (GYG to AAA) might more effectively eliminate rabbit cardiac I_K1 under our experimental conditions and this deserves further investigation. Alternatively, I_K1 remaining after infection with AdK_ir2.1dn and/or AdK_ir2.2dn might arise partially (or solely) from channels lacking K_ir2.1 or K_ir2.2 channel protein. Based on previous publications (Liu et al. 2001; Preisig-Muller et al. 2002), one possible candidate for this remaining current would be K_ir2.3 either alone or combined with other members of the K_ir family. This possibility seems unlikely since in our studies transfection with AdK_ir2.3dn did not cause reductions in I_K1 and Western blots did not reveal evidence of K_ir2.3 expression. In addition, transfection with AdK_ir2.1dn and/or AdK_ir2.2dn did not affect the biophysical current–voltage properties, or blockade by Ba²⁺, of I_K1 in the cultured rabbit myocytes, which (as mentioned already) suggests that the molecular composition of I_K1 did not change in response to reductions in I_K1 in these experiments.

It has previously been suggested that the decline in I_K1 following culture is related to the corresponding decreases in t-tubule density induced by culture conditions (Christe, 1999). This assertion is consistent with the observed localization of K_ir2.1 (Clark et al. 2001) and K_ir2.2 (Leonoudakis et al. 2001) in the t-tubules. However, this mechanism for reduced I_K1 in culture would predict parallel declines in the expression levels of K_ir2.1 and K_ir2.2, which was not observed in our studies. Clearly, the identification of mechanisms responsible for reductions in I_K1 following myocyte culturing warrants further investigation.

In conclusion, our results demonstrate that both K_ir2.1 and K_ir2.2 are functionally expressed in ventricular rabbit myocytes and the predominant molecular correlate of macroscopic I_K1 appears to be heteromeric channels assembled from both K_ir2.1 and K_ir2.2 subunits.