Title: SUR Domains That Associate with and Gate KATP Pores Define a Novel Gatekeeper
Abstract: Structure-function analyses of K+ channels identify a common pore architecture whose gating depends on diverse signal sensing elements. The "gatekeepers" of the long, ATP-inhibited KIR6.0 pores of KATP channels are ABC proteins, SURs, receptors for channel opening and closing drugs. Several competing models for SUR/KIR coupling exist. We show that SUR TMD0, the N-terminal bundle of five transmembrane helices, specifically associates with KIR6.2, forcing nearly silent pores to burst like native KATP channels and enhancing surface expression. Inclusion of adjacent submembrane residues of L0, the linker between TMD0 and the stimulatory nucleotide- and drug-binding ABC core, generates constitutively active channels, whereas additional cytoplasmic residues counterbalance this activation establishing a relationship between the mean open and burst times of intact pores. SUR fragments, lacking TMD0, fail to modulate KIR. TMD0 is thus the domain that anchors SUR to the KIR pore. Consistent with data on chimeric ABCC/KIRs and a modeled channel structure, we propose that interactions of TMD0-L0 with the outer helix and N terminus of KIR bidirectionally modulate gating. The results explain and predict pathologies associated with alteration of the 5′ ends of clustered ABCC8 (9)/KCNJ11 (8) genes. Structure-function analyses of K+ channels identify a common pore architecture whose gating depends on diverse signal sensing elements. The "gatekeepers" of the long, ATP-inhibited KIR6.0 pores of KATP channels are ABC proteins, SURs, receptors for channel opening and closing drugs. Several competing models for SUR/KIR coupling exist. We show that SUR TMD0, the N-terminal bundle of five transmembrane helices, specifically associates with KIR6.2, forcing nearly silent pores to burst like native KATP channels and enhancing surface expression. Inclusion of adjacent submembrane residues of L0, the linker between TMD0 and the stimulatory nucleotide- and drug-binding ABC core, generates constitutively active channels, whereas additional cytoplasmic residues counterbalance this activation establishing a relationship between the mean open and burst times of intact pores. SUR fragments, lacking TMD0, fail to modulate KIR. TMD0 is thus the domain that anchors SUR to the KIR pore. Consistent with data on chimeric ABCC/KIRs and a modeled channel structure, we propose that interactions of TMD0-L0 with the outer helix and N terminus of KIR bidirectionally modulate gating. The results explain and predict pathologies associated with alteration of the 5′ ends of clustered ABCC8 (9)/KCNJ11 (8) genes. ADP/ATP-dependent (SUR/KIR6.0)4, KATP channels encoded by the ABCC8(SUR1)/KCNJ11(KIR6.2) and ABCC9(SUR2)/KCNJ8(KIR6.1) gene clusters (1Aguilar-Bryan L. Nichols C.G. Wechsler S.W. Clement IV, J.P. Boyd III, A.E. Gonzalez G. Herrera-Sosa H. Nguy K. Bryan J. Nelson D.A. Science. 1995; 268: 423-426Crossref PubMed Scopus (1304) Google Scholar, 2Inagaki N. Gonoi T. Clement IV, J.P. Namba N. Inazawa J. Gonzalez G. Aguilar-Bryan L. Seino S. Bryan J. Science. 1995; 270: 1166-1170Crossref PubMed Scopus (1630) Google Scholar, 3Babenko A.P. Aguilar-Bryan L. Bryan J. Annu. Rev. Physiol. 1998; 60: 667-687Crossref PubMed Scopus (487) Google Scholar) present the puzzle of how disparate subunits, from two large superfamilies of membrane transport proteins, are coupled. KIR6.2 pores, normally retained in the ER 1The abbreviations used are: ER, endoplasmic reticulum; TMD, transmembrane domain; NBD, nucleotide-binding domain. (4Zerangue N. Schwappach B. Jan Y.N. Jan L.Y. Neuron. 1999; 22: 537-548Abstract Full Text Full Text PDF PubMed Scopus (912) Google Scholar), display a low maximal open probability in ligand-free solutions (P O(max)), when forced to the cell surface by deleting the C-terminal RKR retention signal (5Tucker S.J. Gribble F.M. Zhao C. Trapp S. Ashcroft F.M. Nature. 1997; 387: 179-183Crossref PubMed Scopus (689) Google Scholar). Assembly with SURs increases surface expression, the P O(max), and sensitivity to inhibitory ATP, and induces responsiveness to stimulatory MgADP, KATP openers, inhibitory sulfonylureas, and other insulin secretagogues (6Babenko A.P. Gonzalez G. Aguilar-Bryan L. Bryan J. FEBS Lett. 1999; 445: 131-136Crossref PubMed Scopus (51) Google Scholar). ATP acts through the KIR cytoplasmic domains to close the pore, while MgADP acts through the nucleotide-binding domains (NBDs) in the ABC core to stimulate the channel (5Tucker S.J. Gribble F.M. Zhao C. Trapp S. Ashcroft F.M. Nature. 1997; 387: 179-183Crossref PubMed Scopus (689) Google Scholar, 7Nichols C.G. Shyng S.L. Nestorowicz A. Glaser B. Clement IV, J.P. Gonzalez G. Aguilar-Bryan L. Permutt M.A. Bryan J. Science. 1996; 272: 1785-1787Crossref PubMed Scopus (476) Google Scholar). In vivo, SUR/KIR6.0 coupling machinery can overcome the power of saturating concentrations of inhibitory ATP (8Babenko A.P. Bryan J. J. Biol. Chem. 2001; 276: 49083-49092Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar), indicating a novel "gatekeeper" (9Armstrong, C. M. (2003) Science's STKE http://stke.sciencemag.org/cgi/contnt/full/sigtrans;2003/188/re10?fulltext.Google Scholar) that exerts activating forces at multiple points on the long (10Nishida M. MacKinnon R. Cell. 2002; 111: 957-965Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar, 11Kuo A. Gulbis J.M. Antcliff J.F. Rahman T. Lowe E.D. Zimmer J. Cuthbertson J. Ashcroft F.M. Ezaki T. Doyle D.A. Science. 2003; 300: 1922-1926Crossref PubMed Scopus (738) Google Scholar) KATP pore. The recent structure of a related ABC transporter shows the SUR core (gray in Fig. 1A) consists of two bundles of six transmembrane helices (TMDs) fused to NBDs (12Chang G. J. Mol. Biol. 2003; 330: 419-430Crossref PubMed Scopus (248) Google Scholar). SURs, and certain ABCC proteins (13Dean M. Rzhetsky A. Allikmets R. Genome Res. 2001; 11: 1156-1166Crossref PubMed Scopus (1530) Google Scholar), have an additional TMD0-L0 module, for which there is no three-dimensional template. Several competing models for SUR/KIR coupling have been suggested (14Giblin J.P. Leaney J.L. Tinker A. J. Biol. Chem. 1999; 274: 22652-22659Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar, 15Schwappach B. Zerangue N. Jan Y.N. Jan L.Y. Neuron. 2000; 26: 155-167Abstract Full Text Full Text PDF PubMed Scopus (149) Google Scholar, 16Mikhailov M.V. Mikhailova E.A. Ashcroft S.J. FEBS Lett. 2000; 482: 59-64Crossref PubMed Scopus (26) Google Scholar, 17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar), but attempts, using chimeric subunits, to discover which domains of SUR interact with KIR have been unsuccessful. Here we used SUR fragments to generate "mini-KATP" that validate our original model (17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar) and reveal the mechanism of SUR/KIR coupling. Molecular Biology—The N-terminal HA-tagged and C-terminal myc-tagged SUR1 constructs were generated starting with hamster SUR1 (1Aguilar-Bryan L. Nichols C.G. Wechsler S.W. Clement IV, J.P. Boyd III, A.E. Gonzalez G. Herrera-Sosa H. Nguy K. Bryan J. Nelson D.A. Science. 1995; 268: 423-426Crossref PubMed Scopus (1304) Google Scholar) using standard subcloning methods. Two copies of the myc epitope were inserted in the extracellular segment between SUR1 transmembrane helices 12 and 13 using a unique Bpu1102 site as described previously (18Sharma N. Crane A. Clement IV, J.P. Gonzalez G. Babenko A.P. Bryan J. Aguilar-Bryan L. J. Biol. Chem. 1999; 274: 20628-20632Abstract Full Text Full Text PDF PubMed Scopus (132) Google Scholar). The constructs were co-expressed with human KIR6.2ΔC35, and green fluorescent protein as a transfection marker, in COSm6 cells as described previously (17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar). Human MRP1 was a kind gift from Susan Cole and Roger Deeley. Immunoprecipitation—COSm6 cells were transfected by electroporation with HA-TMD0, KIR6.2-myc, or both plasmids. After 24 h cells were starved for 1.5 h in methionine-free medium then grown in medium containing 200 μCi/ml [35S]methionine for 3 h before lysis in 1% digitonin in phosphate-buffered saline plus protease inhibitors. Clarified lysates were prepared by centrifugation at 10,000 × g for 15 min then immunoprecipitated with either anti-HA or anti-myc antibodies using protein A-agarose. Following electrophoresis gels were treated with Amplify™ fluorographic reagent (Amersham Bioscience) for 30 min, stained, dried, and visualized using autoradiography. Membranes were isolated from COSm6 cells transfected with SUR1, Ct196, or Ct288, incubated with [γ-32P]8-azido-ATP (1 μCi/ml), plus or minus 100 μm ATP, then irradiated at 254 nm. Labeled membranes were solubilized with 1% digitonin and immunoprecipitated using an anti-myc mouse monoclonal antibody. The immunoprecipitates were separated by electrophoresis, and blotted using an anti-myc mouse monoclonal antibody visualized by chemiluminescence using a peroxidase-labeled secondary antibody. The 32P label was visualized by autoradiography. Electrophysiology—Non-tagged subunits were used for the electrophysiological experiments. Patch clamp recordings, single channel kinetics, and macro-currents noise analysis were done as described previously (17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar). The pipette solution contained (in mm): KCl, 145; MgCl2,1; CaCl2, 1; HEPES, 10; pH 7.4 (KOH). The bath, "intracellular" solution contained (in mm): KCl, 140; MgCl2, 1; EGTA, 5; HEPES, 5; KOH, 10; pH 7.2 (KOH). The [Mg2+] i was kept at a quasi-cytosolic level of ∼0.7 mm by adding MgCl2. The Mg2+-free internal solution contained (in mm): KCl, 140; EDTA, 5; HEPES, 5; KOH, 10; pH 7.2 (KOH). The holding potential was –40 mV. Burst analysis was done using a burst delimiter of 1 ms. ATP dose responses were taken at eight concentrations. In Figs. 1B and 3C, the arrowhead marks isolation of an inside-out patch; the horizontal dotted line shows the level where all KIR6.2-based channels are closed. The differences in averaged values (mean ± S.D.) with p < 0.05 were considered significant. Molecular Modeling—Homology models of the KIR6.1 and KIR6.2 pore, and the SUR core, were generated using KIR3.1 (1N9PA (10Nishida M. MacKinnon R. Cell. 2002; 111: 957-965Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar)) and KirBac1.1 (1P7BA/B (11Kuo A. Gulbis J.M. Antcliff J.F. Rahman T. Lowe E.D. Zimmer J. Cuthbertson J. Ashcroft F.M. Ezaki T. Doyle D.A. Science. 2003; 300: 1922-1926Crossref PubMed Scopus (738) Google Scholar)), and VC-MsbA (1PF4A/B (12Chang G. J. Mol. Biol. 2003; 330: 419-430Crossref PubMed Scopus (248) Google Scholar)) structures, respectively (for details, see Supplement Material). Models giving minimal Cα root mean square deviation values from the three-dimensinal templates were selected. Their quality was evaluated using ProCheck and WhatCheck available via www.expasy.org. Energy minimization did not lower the scores of the selected models. Fig. 1B shows that N-terminal fragments (Nt) of SUR1, including TMD0 and increasing lengths of the L0 linker, have dramatic effects on the gating of KATP pores lacking the C-terminal ER retention signal. TMD0 (Nt196) markedly stabilized the burst and conducting states of the poorly active KIR raising its PO(max) 5-fold, suggesting interactions of transmembrane helices are able to force KIR openings in novel mini-KATP channels. Nt232, including an amphipathic helix (19Bakos E. Evers R. Calenda G. Tusnady G.E. Szakacs G. Varadi A. Sarkadi B. J. Cell Sci. 2000; 113: 4451-4461Crossref PubMed Google Scholar) within L0 made the pore constitutively active. In the Nt232-gated channels, when (Vm – E K) → 0 the mean open time of the KIR, τO → ∞, the intraburst closed time, τCf → 0 (20Babenko A.P. Gonzalez G. Bryan J. Biochem. Biophys. Res. Commun. 1999; 255: 231-238Crossref PubMed Scopus (70) Google Scholar), and PO → τO/(τO + τCf) → 1. The result argues that both trans- and submembrane contacts contribute additively to the opening of long K+ pores and suggest that Nt232 can lock pores in a conducting state under biochemical conditions (i.e. no K+ electrochemical gradient) without resorting to mutation. KATP channels display similar saturation of their PO when stimulated with nucleotide diphosphates (8Babenko A.P. Bryan J. J. Biol. Chem. 2001; 276: 49083-49092Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar), implying the locked state resembles the native open state. The inclusion of additional L0 residues, i.e. Nt256, Nt288, and Nt297, attenuated the PO(max) by destabilizing burst and open states in ATP-free solution. The results reveal that TMD0-L0 possesses separable stimulatory and inhibitory structures, consistent with the original observation that SUR1 TMD0-L0 specifies the β-cell channel-like (lower) spontaneous activity of chimeric SUR1–2/KIR6.2 channels (21Babenko A.P. Gonzalez G. Bryan J. J. Biol. Chem. 1999; 274: 11587-11592Abstract Full Text Full Text PDF PubMed Scopus (66) Google Scholar). While the inhibitory effect saturates with Nt288, all of the mini-KATP are more active than KIR alone, and the PO(max) of the fully attenuated channel is equivalent to the ∼35% inhibition of β-cell KATP channels, in nucleotide-free conditions, upon saturation of SUR1 with inhibitory sulfonylureas (22Babenko A.P. Gonzalez G. Bryan J. FEBS Lett. 1999; 459: 367-376Crossref PubMed Scopus (57) Google Scholar). Single-channel kinetics data from ∼700 mini-KATP patches revealed a positive correlation between the mean burst (TB) and open (τO) times of KIR pores in ATP-free solution with data for full SUR-containing channels, without pore mutations, falling on the same curve (Fig. 2A). This relationship is intrinsic to KIR structure, reflecting the increased stability of the K+ driving force-dependent conducting state when pores are in the longer lived active (burst) conformation. The relationship is consistent with the possibility that the dynamics of the KIR selectivity filter are particularly susceptible to the "bending" of the inner helices associated with activation (23Jiang Y. Lee A. Chen J. Cadene M. Chait B.T. MacKinnon R. Nature. 2002; 417: 515-522Crossref PubMed Scopus (1222) Google Scholar). The KIR inner helix has two conserved glycines versus one upper glycine in Ca2+-activated MthK channel (23Jiang Y. Lee A. Chen J. Cadene M. Chait B.T. MacKinnon R. Nature. 2002; 417: 515-522Crossref PubMed Scopus (1222) Google Scholar). We speculate that bending at the lower, or both "hinges" (Gly156 and Gly165) is required for KATP opening and that bending stabilizes a "centered" position of the pore helices or a conducting conformation of the selectivity filter, seen in KcsA (24Doyle D.A. Morais Cabral J. Pfuetzner R.A. Kuo A. Gulbis J.M. Cohen S.L. Chait B.T. MacKinnon R. Science. 1998; 280: 69-77Crossref PubMed Scopus (5847) Google Scholar) at quasi-physiologic [K+]. A possibility that the lower glycine specifies gating of KIRs has been recently discussed in a report on closed KirBac1.1, crystallized in 0.1 m Mg2+ (11Kuo A. Gulbis J.M. Antcliff J.F. Rahman T. Lowe E.D. Zimmer J. Cuthbertson J. Ashcroft F.M. Ezaki T. Doyle D.A. Science. 2003; 300: 1922-1926Crossref PubMed Scopus (738) Google Scholar). Our data and interpretation are consistent with this structure, where the long inner helices are in a straight configuration, clashing at their cytoplasmic ends, the pore helices are misaligned, and the conformation of the selectivity filter is altered. Stabilization of the open state in mini-KATP, as in full channels, increases the IC50(ATP) (IC50 value for inhibitory ATP) (Fig. 2B). The Nt196 mini channels lack the NBDs and other cytoplasmic domains illustrating that transmembrane helical interactions are sufficient to antagonize ATP inhibition. The SUR core is needed to restore the low KD for ATP seen in the presence of SUR1. The macroscopic inhibitory curves are all well fit by single component Hill functions consistent with there being essentially homogenous populations of mini-KATP at the cell surface. In support of this assertion IC50(ATP), determined for the most (Nt288) and least (Nt232) inhibited macrocurrents, reduced the PO of equivalent single mini KATP by ∼50%, indicating the lower IC50(ATP) seen in cells expressing the larger Nts does not reflect an increased fraction of KIR6.2ΔC channels. Normally, only tetradimeric KATP complexes reach the cell surface (4Zerangue N. Schwappach B. Jan Y.N. Jan L.Y. Neuron. 1999; 22: 537-548Abstract Full Text Full Text PDF PubMed Scopus (912) Google Scholar, 18Sharma N. Crane A. Clement IV, J.P. Gonzalez G. Babenko A.P. Bryan J. Aguilar-Bryan L. J. Biol. Chem. 1999; 274: 20628-20632Abstract Full Text Full Text PDF PubMed Scopus (132) Google Scholar). We observed no ATP-inhibited K+ channels in >100 macro-patches from 10 independent co-transfections of Nts with KIR6.2 showing TMD0-L0 cannot abolish the KIR retention. On the other hand, Fig. 2C shows that the same Nts enhance surface expression of RKR-free channels, albeit less effectively than SUR1 (6Babenko A.P. Gonzalez G. Aguilar-Bryan L. Bryan J. FEBS Lett. 1999; 445: 131-136Crossref PubMed Scopus (51) Google Scholar). The gating properties of mini-channels, determined from macro and single channel currents, were similar for molar transfections ratios (Nt:KIR) of 2:1 versus 1:2 again indicating essentially homogenous populations of surface channels. The data, consistent with the predominant intracellular distribution of KIR6.2ΔC in mammalian cells (25Ma D. Zerangue N. Lin Y.F. Collins A. Yu M. Jan Y.N. Jan L.Y. Science. 2001; 291: 316-319Crossref PubMed Scopus (323) Google Scholar), show that partially assembled heteromeric channels traffick less effectively to the cell surface. Mini-KATP assemble under physiologic conditions and are stable that TMD0 will co-immunoprecipitate with KIR6.2 (Fig. 3A). We were unable to show a comparable stable association of SUR cores with KIR by co-immunoprecipitation, suggesting that TMD0-L0 is necessary for Cts to decrease the KD for inhibitory ATP (21Babenko A.P. Gonzalez G. Bryan J. J. Biol. Chem. 1999; 274: 11587-11592Abstract Full Text Full Text PDF PubMed Scopus (66) Google Scholar). Nt233 and Nt248 from MRP1, a closely related ABCC protein, had no effect on KATP pores (80 macropatches showed no trapping of KIR6.2ΔC35 inside the cell in eight independent co-transfections). Ct196, Ct232, Ct256, Ct288, and Ct297 cores, and an the core half fragment, Ct1046 (see Fig. 1A), failed to modulate KIR6.2ΔC gating, even after substitution of the SUR RKR retention signal (4Zerangue N. Schwappach B. Jan Y.N. Jan L.Y. Neuron. 1999; 22: 537-548Abstract Full Text Full Text PDF PubMed Scopus (912) Google Scholar) with AAA to force surface expression (>200 macro-patches from six series of independent co-transfections). Fig. 3B confirms that Ct196 and Ct288 are expressed and are able to bind and photolabel with [γ-32P]8-azidoATP. A three-plasmid experiment provides a positive control for the Ct1046 fragment, whose misfolding cannot be tested using nucleotide binding; while Ct1046 has no effect on KIR6.2 pores, co-expression with Nt1046 rescues generation of "fragmented" KATP indistinguishable from wild type (Fig. 3C). The results reveal strong interdomain interactions within the SUR core and show that TMD0-L0 is indispensable for assembly of KATP channels. TMD0-L0 is a combination of unique gatekeeper and coupling module anchoring the SUR core to KIR6.0 and modulating their trafficking and activity. The identification of TMD0-L0 as the principal coupling module positions it in contact with KIR in a composite model (Fig. 4). Contact is through the outer helix and involves T76MS critical for assembly of chimeric KIRs with SUR (15Schwappach B. Zerangue N. Jan Y.N. Jan L.Y. Neuron. 2000; 26: 155-167Abstract Full Text Full Text PDF PubMed Scopus (149) Google Scholar), illustrated here in a homology model of the closed channel built using recent crystal structures (10Nishida M. MacKinnon R. Cell. 2002; 111: 957-965Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar, 11Kuo A. Gulbis J.M. Antcliff J.F. Rahman T. Lowe E.D. Zimmer J. Cuthbertson J. Ashcroft F.M. Ezaki T. Doyle D.A. Science. 2003; 300: 1922-1926Crossref PubMed Scopus (738) Google Scholar, 12Chang G. J. Mol. Biol. 2003; 330: 419-430Crossref PubMed Scopus (248) Google Scholar) (for details, see Supplement Material). Although the lack of a structural template for TMD0-L0 prohibits its explicit docking with KIR and the VC-MsbA dimer-like ABC core of SUR, activation of KIR by TMD0 argues that intramembrane intersubunit interactions are sufficient to reposition the KIR outer helices. Consistent with coupled movement of the inner and outer helices during activation of other K+ channels (26Perozo E. Cortes D.M. Cuello L.G. Science. 1999; 285: 73-78Crossref PubMed Scopus (498) Google Scholar, 27Jiang Y. Lee A. Chen J. Ruta V. Cadene M. Chait B.T. MacKinnon R. Nature. 2003; 423: 33-41Crossref PubMed Scopus (1656) Google Scholar), repositioning would give the KIR inner helices freedom to bend and initiate ligand-independent bursting. Nt232 mimics the effect of SUR saturated with nucleotide diphosphates, revealing intersubunit interactions involving a submembrane part of L0 (red), which open KATP channels in the presence of inhibitory concentrations of intracellular ATP. Consistent with the MsbA model (12Chang G. J. Mol. Biol. 2003; 330: 419-430Crossref PubMed Scopus (248) Google Scholar) ATP-driven dimerization of the SUR NBDs is expected to reposition TMD1 and 2, in a manner sensitive to MgADP and drug binding to SUR core, and by inference move L0. We propose that the submembrane amphipathic helix of L0 (19Bakos E. Evers R. Calenda G. Tusnady G.E. Szakacs G. Varadi A. Sarkadi B. J. Cell Sci. 2000; 113: 4451-4461Crossref PubMed Google Scholar) interacts with the KIR α-helix (17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar), named "slide helix" in KirBac1.1 (11Kuo A. Gulbis J.M. Antcliff J.F. Rahman T. Lowe E.D. Zimmer J. Cuthbertson J. Ashcroft F.M. Ezaki T. Doyle D.A. Science. 2003; 300: 1922-1926Crossref PubMed Scopus (738) Google Scholar), which is anchored at the membrane-water interface in KcsA (28Cortes D.M. Cuello L.G. Perozo E. J. Gen. Physiol. 2001; 117: 165-180Crossref PubMed Scopus (218) Google Scholar). Activation of chimeric channels containing the KcsA pore (29Lu Z. Klem A.M. Ramu Y. Nature. 2001; 413: 809-813Crossref PubMed Scopus (278) Google Scholar) required the proximal N-domain preceding the outer helix of KIR2.1 or the S4-S5 linker preceding S5 (analogous to the outer helix) from the Shaker KV channel. Part of S4 and the S4-S5 linker appear to be submembranous in the closed KvAP and their movement is proposed to pull the outer helix and the coupled inner helix away from the central pore axis during opening (27Jiang Y. Lee A. Chen J. Ruta V. Cadene M. Chait B.T. MacKinnon R. Nature. 2003; 423: 33-41Crossref PubMed Scopus (1656) Google Scholar). KATP channels are a novel variation on this theme. The model provides a structural interpretation for the decrease in sensitivity of the KIR to inhibitory ATP upon stimulation by SUR. Models of the KIR6.2 and KIR6.1 cytoplasmic pores (see Supplement Material) suggest that C-terminal residues critical for ATP inhibition (e.g. Ile182, Lys185, and Gly334, predicted to be co-localized (17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar)) lie in proximity (within the dimensions of an ATP molecule) to Arg50 adjacent to the slide helix of the neighboring KIR6.2 subunit. A R50Q/K185Q double mutation disrupts the ATP inhibitory machinery (20Babenko A.P. Gonzalez G. Bryan J. Biochem. Biophys. Res. Commun. 1999; 255: 231-238Crossref PubMed Scopus (70) Google Scholar). We hypothesize that ATP binding at intersubunit N-/C-terminal interfaces constrict the ring of KIR cytoplasmic domains repositioning the slide helices to lock channels in the interburst state with the lowest KD for ATP (8Babenko A.P. Bryan J. J. Biol. Chem. 2001; 276: 49083-49092Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). Conversely, pulling the slide helices away from the KIR pore, via their interactions with L0 helices, moves Arg50 and opens the ATP-binding pocket, thus increasing the KD. This mechanism can explain the marked increase in IC50(ATP) for Nt232-containing channels (∼0.7 mm; Fig. 2B). Based on the data from longer Nts, we propose that L0 parts, separable from the amphipathic helix, partner with the peripheral N terminus of KIR6.2 (17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar) to attenuate the PO(max). This idea is supported by the finding that deletion of the peripheral KIR N terminus (8Babenko A.P. Bryan J. J. Biol. Chem. 2001; 276: 49083-49092Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar, 20Babenko A.P. Gonzalez G. Bryan J. Biochem. Biophys. Res. Commun. 1999; 255: 231-238Crossref PubMed Scopus (70) Google Scholar), or application of a hydrophilic N-terminal synthetic peptide competing with the endogenous N terminus (17Babenko A.P. Bryan J. J. Biol. Chem. 2002; 277: 43997-44004Abstract Full Text Full Text PDF PubMed Scopus (58) Google Scholar), locks heteromeric, but not homomeric, channels in the burst state. Similarly, perturbing the mobility or positioning of the KIR N terminus by concatemerization increases the PO(max) (Fig. 2A). Deletion of the mobile peripheral N terminus has been needed to crystallize KIR pores (10Nishida M. MacKinnon R. Cell. 2002; 111: 957-965Abstract Full Text Full Text PDF PubMed Scopus (344) Google Scholar, 11Kuo A. Gulbis J.M. Antcliff J.F. Rahman T. Lowe E.D. Zimmer J. Cuthbertson J. Ashcroft F.M. Ezaki T. Doyle D.A. Science. 2003; 300: 1922-1926Crossref PubMed Scopus (738) Google Scholar), and beyond the amphipathic helix, L0 is predicted to be flexible. Transient interactions between these domains, rather than tight binding, is expected to increase the frequency of burst termination but not close the channel permanently. The model suggests how SUR1 mutations in the 5′ portion of human ABCC8, e.g. A116P, cause familial hyperinsulinemia (30Aguilar-Bryan L. Bryan J. Endocr. Rev. 1999; 20: 101-135Crossref PubMed Scopus (631) Google Scholar) and how a polymorphism, E23K, in the distal N terminus of KIR6.2 can increase the PO of KATP channels in insulin secreting β-cells of pancreatic islets, increasing the risk of type 2 diabetes (31Schwanstecher C. Meyer U. Schwanstecher M. Diabetes. 2002; 51: 875-879Crossref PubMed Scopus (215) Google Scholar). The Pro should kink TM3 of TMD0 that can cause misfolding and prevent normal SUR/KIR coupling or trafficking. Indeed, co-expression of SUR1A116P with KIR6.2 resulted in no ATP-inhibited K+ currents in >100 macro-patches (eight independent transfections). Lys23 could reduce the frequency of collisions between the KIR N terminus and L0. Similar alterations of SUR2 and KIR6.1 are predicted to contribute to certain forms of angina and vascular distonia, respectively. We thank Drs. L. Aguilar-Bryan, A. Crane, and W. Vila-Carriles for sharing biochemical data and G. Zhao for preparation of plasmids. Download .pdf (.11 MB) Help with pdf files