Evidence against Functional Heteromultimerization of theK ATP Channel Subunits Kir6.1 and Kir6.2*

K ATP channels consist of pore-forming potassium inward rectifier (Kir6.x) subunits and sulfonylurea receptors (SURs). Although Kir6.1 or Kir6.2 coassemble with different SUR isoforms to form heteromultimeric functionalK ATP channels, it is not known whether Kir6.1 and Kir6.2 coassemble with each other. To define the molecular identity of K ATP channels, we used adenoviral gene transfer to express wild-type and dominant-negative constructs of Kir6.1 and Kir6.2 in a heterologous expression system (A549 cells) and in native cells (rabbit ventricular myocytes). Dominant-negative (DN) Kir6.2 gene transfer suppressed current through heterologously expressed SUR2A + Kir6.2 channels. Conversely, DN Kir6.1 suppressed SUR2B + Kir6.1 current but had no effect on coexpressed SUR2A + Kir6.2. We next probed the ability of Kir6.1 and Kir6.2 to affect endogenous K ATP channels in adult rabbit ventricular myocytes, using adenoviral vectors to achieve efficient gene transfer. Infection with the DN Kir6.2 virus for 72 h suppressed pinacidil-inducible K ATP current density measured by whole-cell patch clamp. However, there was no effect of infection with the DN Kir6.1 on theK ATP current. Based on these functional assays, we conclude that Kir6.1 and Kir6.2 do not heteromultimerize with each other and that Kir6.2 is the sole K ATPpore-forming subunit in the surface membrane of heart cells.

Whereas the gating of classical ion channels is regulated by membrane potential, K ATP channels respond instead to changes in cellular energy metabolism. K ATP channels are present in the surface membrane of various excitable cells, including cardiac myocytes (1,2) and pancreatic ␤ cells (3). These channels play an important role in several cellular responses, notably insulin secretion and hypoxic vasodilation, by linking the metabolic status of the cell to its membrane potential (4).
Given the important physiological and pathophysiological roles of these channels, there is good reason to elucidate their molecular composition. There are two types of K ATP channels in cardiac muscle (1,2,5), the sarcolemmal channel and the mitochondrial channel (mitoK ATP ) 1 (6 -10). Although the molecular identity of the mitoK ATP channel is not known, the sarcolemmal K ATP channel consists of pore-forming potassium inward rectifier (Kir6.x) subunits and sulfonylurea receptors (SURs). Heterologous expression studies suggest that cardiac sarcolemmal channels are heteromultimers of pore-forming Kir6.2 subunits and SUR2A subunits (5,(11)(12)(13)(14), a prediction confirmed by Kir6.2 knockout studies in mice (15). However, a possible contribution from products of the Kir6.1 gene, which is richly expressed in the normal heart and elevated after ischemia (16,17), has not been excluded. Previous work has shown that smooth muscle K ATP channels consist of Kir6. To investigate the assembly of this channel, we used viral vectors designed to overexpress wild-type or dominant-negative Kir6.1 or Kir6.2 constructs driven by inducible ecdysone promoters in a heterologous expression system (A549 cells) and in adult rabbit ventricular myocytes. The electrical properties of K ATP channels in these cells lead us to conclude that Kir6.1 and Kir6.2 do not form functional heteromultimers.
EXPERIMENTAL PROCEDURES Viral Vectors-The coding sequence for mouse Kir6.1 (or rabbit Kir6.2) was amplified using a polymerase chain reaction and cloned into pGFP-IRES (20) vector to generate polycistronic expression constructs. Site-directed mutagenesis was performed using a QuikChange TM kit (Stratagene, La Jolla, CA) to modify codons 142 and 144 of wild-type mouse Kir6.1 (Kir6.1WT) to code for alanines instead of glycines to generate Kir6.1AFA (Fig. 1A). Similarly, wild-type rabbit Kir6.2 (Kir6.2WT) codons 131 and 133 were modified to code for alanines instead of glycines to generate mutant Kir6.2AFA. The wild-type and mutant Kir genes were cloned into a polycistronic adenovirus shuttle vector with ecdysone-inducible promoter coexpressing enhanced green fluorescent protein (pAdEGI) or human CD8 protein (pAdECD8I) to generate the modified vectors (21). Recombinant adenoviruses containing the various Kir6.1 and Kir6. Cell Isolation-Rabbit ventricular myocytes were isolated by enzymatic dissociation of adult rabbit hearts as described previously (9,10,23) and washed several times with calcium-free solution. Calcium concentration was gradually brought back to 1 mM. Cells were cultured on laminin-coated coverslips in M199 culture medium (CellGro; Mediatech, Herndon, VA) with 2% fetal bovine serum and 1% penicillin/ streptomycin at 37°C.
Transfection and Infection-A human epithelial cell line A549 (CCL-185, American Type Culture Collection, Manassas, VA) was cotransfected with 1 g of rabbit Kir6.2 and rat SUR2A (a gift of Dr. S. Seino, Chiba University School of Medicine, Chiba, Japan) using Lipo-fectAMINE (Life Technologies, Inc., Gaithersburg, MD) before infecting with various virus constructs. The virus constructs were coinfected with either the Kir or reporter-only virus and the receptor virus AdVgRXR at a ratio of about 10:1 as described previously (21). After about 1 h the cells were washed with virus-free culture medium. Expression was induced by the addition of 1 M ponasterone A (Invitrogen, San Diego, CA) for up to 72 h before electrophysiological studies were conducted. We also transfected A549 cells using 6 g of mouse Kir6.1 and mouse SUR2B (both gifts from Dr. Y. Kurachi, University of Osaka, Osaka, Japan) and LipofectAMINE before infecting as described above. Myocytes were cultured in 2% fetal bovine serum-containing M199 media and infected with various reporter and Kir constructs and induced for 72 h after infection before measuring the currents.
Electrophysiology-To quantify K ATP channel activity, we measured agonist-induced membrane current using the whole-cell patch clamp technique (24). Cells were superfused with external solution containing (in mM) NaCl, 140; KCl, 5; CaCl 2 , 1; MgCl 2 , 1; and HEPES, 10 (pH 7.4) with NaOH at room temperature (ϳ22°C). The internal pipette solution contained (in mM) potassium glutamate, 120; KCl, 25; MgCl 2 , 0.5; EGTA, 10; HEPES, 10; and MgATP, 1 (pH 7.2) with KOH. Currents were elicited by ramp pulses in A549 cells (between Ϫ110 and 50 mV from a holding potential of Ϫ80 mV). The current amplitude at 0 mV was measured at steady state after the application of the K ATP channel agonist P1075 (pinacidil analog, 30 M). In rabbit ventricular myocytes, currents were elicited every 6 s by square pulses (in consecutive steps from a holding potential of Ϫ80 mV to Ϫ40 mV for 100 ms and 0 mV for 380 ms), and I K, ATP was quantified 15 and 30 min after exposure to 100 M pinacidil. To be sure that the currents measured reflected agonistinduced K ATP channels, we excluded experiments in which pinacidilinduced currents were irreversible after washout of drugs (8 cells), although including those data would not have changed the conclusions.
Preliminary experiments revealed that 48 -72 h were required to obtain robust dominant-negative effects; therefore, electrophysiological recordings were made 72 h after infection in both A549 cells and rabbit ventricular myocytes.
Statistical Analysis-All data are presented as mean Ϯ S.E. Statistical analysis was performed by analysis of variance combined with the Bonferroni pairwise comparison test when needed. p values Ͻ0.05 were considered significant.

RESULTS AND DISCUSSION
We have previously shown that dominant-negative Kir6.2 constructs suppress K ATP current in heterologous expression systems and in rat neonatal cardiomyocytes (11). Thus, we predicted that infection with AdKir6.2AFA would suppress the cardiac sarcolemmal K ATP current (Fig. 1A). To address the question of heteromultimerization of the Kir6.1 and Kir6.2 subunits, we examined the effects of infection with various Kir and control virus constructs on I K, ATP A549 cells cotransfected with 6.2WT and SUR2A subunits. I K, ATP was elicited pharmacologically by addition of the pinacidil derivative P1075, which potently activates these channels (25) . Fig. 2 fection, in the absence (control) and presence of P1075. Panel A shows the wild-type current expressed by Kir6.2 and SUR2A. Fig. 2, B-D shows currents after cotransfection and infection with 6.1AFA (B), CD8 (C), or 6.2AFA (D) viral constructs. Only the 6.2AFA cell (D) shows notably suppressed drug-activated currents. Fig. 2E summarizes the data for P1075-induced current in the various groups of A549 cells. The uninfected cells had current densities similar to those infected with CD8 reporter virus, indicating that infection itself does not alter the membrane currents. Significant dominant-negative suppression of I K, ATP through Kir6.2-SUR2A channels was achieved by infection with the Kir6.2AFA virus (p Ͻ 0.001 versus control) but not with Kir6.1AFA.
To verify that Kir6.1AFA can function as a dominant-negative construct in Kir6.1 homomultimers, we heterologously expressed Kir6.1 ϩ SUR2B channels. Kir6.1 and SUR2B have been reported to form smooth muscle-type K ATP channels (26,27). As with Kir6.2 and SUR2A, we measured membrane currents elicited by the K ATP channel opener P1075. Fig. 3, A-D shows I K, ATP recorded in A549 cells 72 h after transfection, before and during exposure to P1075. Panel A shows the current produced by simple cotransfection with Kir6.1 and SUR2B DNA. Fig. 3, B-D shows the currents after cotransfection and infection with 6.1AFA (B), CD8 (C), or 6.2AFA (D) viral constructs. Here, the only permutation with suppressed P1075inducible current was 6.1AFA (B). The pooled data in Fig. 3E reveal that infection with 6.1AFA significantly inhibited I K, ATP through Kir6.1 ϩ SUR2B channels (p Ͻ 0.005 versus either no virus or CD8), but there was no effect of Kir6.2AFA.
The whole-cell currents produced by heterologously expressed Kir6.1 ϩ SUR2B are about half as large as those produced by Kir6.2 ϩ SUR2A channels (0.75 Ϯ 0.2 versus 1.7 Ϯ 0.3 nA at 0 mV). However, the number of functional channels is likely to be comparable, considering the known difference in single-channel conductance of the two channels (35 versus 80 picosiemens; Refs. 5 and 27) and assuming similar open probabilities. Thus, the lack of suppression by the 6.1AFA construct on 6.2WT ϩ SUR2A coexpressed current (and, conversely, the lack of effect of 6.2AFA on 6.1WT ϩ SUR2B current) is most likely due to the absence of coassembly.
To probe the roles of Kir6.1 and Kir6.2 in native cells, we infected these viral constructs in adult ventricular myocytes in primary culture. Whole-cell patch recordings revealed that K ATP current was greatly suppressed at 72 h in Kir6.2AFAinfected cells compared with green fluorescent protein-, 6.2WT-, 6.1WT-, or 6.1AFA-infected cells in primary culture. Fig. 4 summarizes the current densities measured 15 and 30 min after exposure to pinacidil in cells infected with all the viral constructs tested. Note that 6.2AFA-infected cells displayed pinacidil-induced current densities significantly lower than the others (at 30 min, 11.3 Ϯ 2.7 picoampere/picofarad, n ϭ 7 versus control, 36.8 Ϯ 4.1 picoampere/picofarad, n ϭ 6, p Ͻ 0.01). None of the other constructs had any effect on K ATP current density, indicating that infection with 6.2AFA signifi- cantly suppressed the native sarcolemmal K ATP current in rabbit ventricular cells. Whereas the results are clear-cut at 72 h, no effect was evident at 24 h, and only a trend was apparent at 48 h (data not shown). This time course is consistent with that described for effective suppression of other K ϩ channels by dominant-negative constructs in myocytes and neurons (20). The length of time required for effective suppression presumably reflects the time required for the assembly of the nonfunctional tetramer relative to the degradation of previously synthesized functional channels.
The results are notable in several respects. First of all, they confirm that Kir6.2 is crucial for cardiac surface K ATP channels. Second, the absence of an effect of exogenous wild-type Kir6.2 expression on the surface K ATP channel indicates that the availability of Kir6.2 is not a rate-limiting factor in the biogenesis of these channels (as might be the case if there were a reserve store of presynthesized but unpaired SUR2 proteins). Third, the lack of an effect of Kir6.1WT or Kir6.1AFA on the sarcolemmal K ATP channels, taken together with the lack of suppression of 6.1AFA on the heterologously expressed Kir6.2 ϩ SUR2A channel, and of 6.2AFA on Kir6.1 ϩ SUR2B channels, indicate that Kir6.1 and Kir6.2 do not form functional heteromultimers either in A549 cells or in native cardiac myocytes. The latter finding leaves open the question of what role, if any, Kir6.1 may play in heart cells. Specifically, the present experiments do not exclude a role for Kir6.1 in the formation of mitoK ATP channels (18), which are crucial for the endogenous cardioprotective phenomenon known as ischemic preconditioning (25). Dominant-negative approaches similar to those used here in myocytes, but with mitochondrial flavoprotein fluores- cence as the end point, may help to address possible contributions of Kir6.1 and/or 6.2 to mitoK ATP channels.