The cardiac inward rectifier K+ channel subunit, CIR, does not comprise the ATP-sensitive K+ channel, IKATP.

Cardiac IKACh is comprised of two inwardly rectifying K+ channel subunits, CIR and GIRK1 (Krapivinsky, G., Gordon, E. G., Wickman, K., Velimirovic, B., Krapivinsky, L., and Clapham, D. E.(1995) Nature 374, 135-141). A cardiac [Medline] protein virtually identical to CIR, termed rcKATP-1 (Ashford, M. L. J., Bond, C. T., Blair, T. A., and Adelman, J. P.(1994) Nature 370, 456-459), was reported to form an ATP-sensitive [Medline] inwardly rectifying K+ channel, IKATP. We attempted to determine whether CIR alone or together with an unknown protein(s) participated in the formation of cardiac IKATP. Expression of CIR in insect, oocyte, and mammalian cell systems did not increase the appearance of ATP-sensitive currents, but rather gave rise to unique strongly inwardly rectifying, G protein-regulated K+ currents. CIR protein is found exclusively in atria, in contrast to the predominance of IKATP functional activity in ventricle. Also, CIR was completely depleted from heart membrane after immunodepletion of GIRK1. We conclude that CIR/rcKATP-1 is not a subunit of cardiac IKATP and that GIRK1 is the only channel protein coassociating with CIR in heart.

Inwardly rectifying K ϩ channels (IRKs) 1 maintain resting membrane potential and control cell excitability. Two of the most highly regulated IRK channels are the cardiac atrial muscarinic-gated channel, I KACh , and the ATP-sensitive channel, I KATP (3)(4)(5)(6). I KACh is in part responsible for the vagally mediated slowing of heart rate. Characteristic features of I KACh are its gating by the G protein ␤␥ dimer (G ␤␥ ), single channel conductance of 35-40 picosiemens in symmetrical 140 mM K ϩ , sharp inward rectification in the presence of internal Mg 2ϩ , and a mean single channel open time of ϳ1 ms (3,(7)(8)(9)(10)(11). Recently, it was shown that cardiac I KACh was comprised of two homologous subunits, termed GIRK1 and CIR (1). GIRK1 and CIR are both members of a subfamily of IRKs regulated by G proteins (GIRKs or Kir 3.0 (12)). In brain, GIRK1 may combine with a CIR homolog, GIRK2, to form a channel with properties similar to I KACh (13). 2 I KATP is a K ϩ -selective, inward rectifier found in pancreatic ␤ cells, heart, smooth muscle, skeletal muscle, brain, and kidney. Evidence suggests that I KATP channels play roles in the early repolarization of ischemic cardiac cells and insulin secretion. Cardiac I KATP is blocked by intracellular ATP and sulfonylureas, is modulated by intracellular nucleotide diphosphates and activated ␣ subunits of G proteins (G ␣ ), rectifies weakly, and has a characteristic single channel bursting profile (4 -6). A recently cloned cDNA, termed rcKATP-1, was reported to form I KATP when expressed in the human embryonic kidney cell line HEK293 (2). rcKATP is virtually identical to CIR. The sequence for rcKATP-1 reported in GenBank™ differs by only two amino acids from CIR (I188V, a conserved substitution, and Q375E). Each difference can be explained by a single base polymorphism. To reconcile the demonstration that CIR is part of cardiac I KACh (1) and the demonstration that CIR expression generates I KATP (2), it was suggested that CIR does not form I KATP alone, but may coassemble with an unknown subunit(s) expressed in cardiac and HEK293 cells to form I KATP . In this report we demonstrate that CIR is not part of cardiac I KATP and appears to be wholly associated with GIRK1 in heart.

MATERIALS AND METHODS
Plasma membranes from bovine atria and ventricle were isolated as described (15). Two to 10 g of membrane protein was separated by SDS-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride (Millipore) film for Western blots. aCIRN2 antibodies raised against an N-terminal peptide (amino acids 19 -32) were affinitypurified as described (1). aCIRN2 specifically recognized CIR expressed in Sf9 cells and immunoprecipitated in vitro translated CIR (data not shown). Western blots were developed with anti-rabbit peroxidase (Pierce) and enhanced chemiluminescence (ECL; Amersham Corp.).
The CIR coding region was subcloned into pCDNA1/Amp (Invitrogen) downstream from the cytomegalovirus promoter for use in mammalian transfection experiments. To maximize the chance of assaying HEK293 and CHO cells transiently expressing CIR, we included a construct driving expression of murine CD4, a T cell-specific cell-surface antigen, in all transfections (1). CHO and HEK293 cells plated on 10-cm Petri dishes at 50% confluence were transfected via calcium phosphate precipitation with CIR or rcKATP-1 (10 g) and 5 g of the CD4 construct. 24 h post-transfection, CHO cells were detached, concentrated, and incubated at 4°C for 10 -15 min with 5 l of (R)-phycoerythrin-conjugated anti-mouse CD4 monoclonal antibodies (Pharmingen, 0.4 mg/ml). Unbound antibody was removed by a 5-ml wash with phosphate-buffered saline. Cells were replated on coverslips in their respective media and allowed to reattach for 1-2 h prior to electrophysiological experi- A CIR-containing baculovirus construct was produced in the nonfusion baculovirus expression vector, pBlueBac III (Invitrogen). The recombinant virus was generated, isolated, and amplified as described (MaxBac, Invitrogen). For infections, Sf9 cells grown in suspension were plated at ϳ50% confluence in 35-mm Petri dishes containing coverslips. Cells were infected at an multiplicity of infection of ϳ100 plaque-forming units/cell and assayed for channel currents 24 -40-h post-infection.
The pipette and bath solutions for most Sf9, CHO, and HEK293 patch clamp experiments were identical: 140 mM K ϩ , 140 mM Cl Ϫ , 10 mM HEPES, 5 mM EGTA, and 2 mM Mg 2ϩ , pH 7.2. In several Sf9 experiments, bath Cl Ϫ was reduced to 2 mM by replacement with methanesulfonate. Pipettes were pulled from KG-12 glass, coated with Sylgard (Dow Corning), and had resistances of 2-4 m⍀ in the above solution. Currents were recorded using an Axopatch 200A patch clamp amplifier, filtered at 5 kHz using an 8-pole Bessel filter, and digitized at 25 kHz using PCLAMP6 software for analysis of amplitude histograms.

RESULTS
When CIR was infected or transfected into several host cells, increased I KATP channel activity was never observed. Sf9 cells infected with a CIR-containing baculovirus or mammalian cells transfected with CIR, however, display novel currents unlike those of I KATP . CIR in all expression systems (and rcKATP-1 in HEK293 cells) generates flickery, short-lived channels with variable conductance (Fig. 1A; see also Ref. 1). Amplitude histograms taken from data filtered at 5 kHz show a distribution of amplitudes consistent either with many openings that are too brief to be resolved or by a rapid block mechanism. The signal-to-noise ratio did not allow us to determine the single channel conductance at larger bandwidth where these brief openings might be resolved. However, within many recordings are occasional longer openings of 32-36 picosiemens. Despite the fact that the range of conductances and brief openings resulting from CIR expression are atypical of currents expected in native tissues, the current fluctuations have properties con-sistent with ion channels. The channels are K ϩ -selective, rectify sharply in the presence of Mg 2ϩ (Fig. 1), are blocked by 1 mM external Cs ϩ and Ba 2ϩ , and are stimulated by GTP␥S, G ␤␥ , and ATP (1). Similar results are observed in Xenopus oocytes injected with CIR mRNA. 3 In control cells a channel resembling that described as I KATP by Ashford et al. (2) was recorded. The channel was present in 1-3% (CHO; n Ͼ 100) to 18% (HEK293; n ϭ 78) of control or mock-transfected cells. These endogenous channels were not noted by Ashford et al. (2). I KATP activity is higher in ventricular than atrial cells (4,11,17,18), while I KACh is found exclusively in atria. We used antibodies specific for CIR to examine the distribution of this inward rectifier subunit in bovine ventricle and atrial tissues. A unique peptide corresponding to amino acids S19-Y32 of CIR (identical in rcKATP-1) was synthesized to raise a CIR/rcK-ATP-1-specific polyclonal antibody. In contrast to the functional distribution of I KATP , CIR (rcKATP-1) was not detected in ventricular plasma membranes. CIR, however, was readily apparent in atria, consistent with I KACh expression (Fig. 2).
CIR was shown to coimmunoprecipitate with GIRK1 in atria (1). To determine whether CIR associates with another protein(s) in heart tissue, we immunoprecipitated CIR directly. aCIRN2 immunoprecipitated the 45-kDa CIR as well as 58and 72-kDa proteins from atrial membranes, but none were immunoprecipitated from ventricular plasma membranes (Fig.  3). In control experiments, aCIRN2 did not immunoprecipitate in vitro translated GIRK1 (data not shown). The proteins migrating at 72 and 58 kDa were identified as glycosylated and unglycosylated forms, respectively, of GIRK1 by the specific anti-GIRK1 antibody, aCsh (Fig. 3, see also Ref. 1). Thus, we did not find novel proteins interacting with CIR in atria. However, we could not rule out the possibility that other proteins comigrated with, and therefore could not be discriminated from, CIR and GIRK1 polypeptides.
To address the possibility that other proteins comigrated with CIR and GIRK1 polypeptides, we immunoprecipitated 3 N. Lim, unpublished observations.

FIG. 1. Heterologous expression of CIR or rcKATP-1 in Sf9, CHO, and HEK293 expression systems gives similar channel currents.
Recordings are from excised patches at the voltages shown. An "all-point" amplitude histogram was constructed for channels appearing upon infection of Sf9 cells with a recombinant CIR baculovirus (a), following transfection of CHO cells with CIR (b), or rcKATP-1 in HEK293 cells (c). Holding membrane potential at positive voltages reveals the strong inward rectification of these channels. With the addition of 5 mM EDTA, which chelates the 2 mM bath Mg 2ϩ , inward rectification was abolished. In all three systems, amplitude histograms were generated from inside-out patches held at Ϫ80 mV, stimulated by GTP␥S (Sf9 cells) or ATP and GTP␥S (HEK293 and CHO cells). GIRK1 and measured depletion of CIR. If other proteins interacted with CIR, full depletion of GIRK1 would not eliminate all detectable CIR. GIRK1 was immunoprecipitated with an increasing amount of aCsh. The relative amounts of GIRK1 and CIR remaining in the supernatant and after immunoprecipitation were determined by Western blot with aCsh and aCIRN2 antibodies. As GIRK1 was depleted, CIR was exhausted; no CIR remained in the supernatant after removal of GIRK1/CIR complexes (Fig. 4). Therefore, we conclude that GIRK1 is the only channel protein associated with CIR in heart. CONCLUSION We have shown that the inward rectifier subunit CIR (rcK-ATP-1) initially proposed by Ashford et al. (2) to comprise I KATP does not participate in formation of cardiac I KATP . Our evidence is summarized as follows. 1) Expression of CIR alone in Sf9, CHO, and HEK293 expression systems did not yield channels with properties similar to I KATP . Interestingly, CIR channels rectify strongly, but have an asparagine, not an aspartic acid residue, in their putative second transmembrane segment. This contradicts the hypothesis that asparagine is a common feature of weak inward rectifiers (19). 2) CIR was present only in cardiac atria, not ventricle, whereas I KATP is abundant in both tissues. 3) CIR protein does not associate with inward rectifier subunits other than GIRK1 in heart. We speculate that the channels reported by Ashford et al. (2) in HEK293 cell were actually endogenous channels, not novel currents appearing as a result of CIR or rcKATP-1 expression. It seems likely that an as yet undefined but homologous inward rectifier subunits may form cardiac I KATP , perhaps in conjunction with the sulfanylurea receptor (20). In fact, an IRK (uKATP-1) from a rat pancreatic cDNA library with only 43-46% identity to any of ROMK1, GIRK1, CIR/rcKATP-1, and IRK1 has been reported to constitute I KATP in HEK293 cells (14).