Tandem function of nucleotide binding domains confers competence to sulfonylurea receptor in gating ATP-sensitive K+ channels.

Fundamental to the metabolic sensor function of ATP-sensitive K(+) (K(ATP)) channels is the sulfonylurea receptor. This ATP-binding cassette protein, which contains nucleotide binding domains (NBD1 and NBD2) with conserved Walker motifs, regulates the ATP sensitivity of the pore-forming Kir6.2 subunit. Although NBD2 hydrolyzes ATP, a property essential in K(ATP) channel gating, the role of NBD1, which has limited catalytic activity, if at all, remains less understood. Here, we provide functional evidence that cooperative interaction, rather than the independent contribution of each NBD, is critical for K(ATP) channel regulation. Gating of cardiac K(ATP) channels by distinct conformations in the NBD2 ATPase cycle, induced by gamma-phosphate analogs, was disrupted by point mutation not only of the Walker motif in NBD2 but also in NBD1. Cooling membrane patches to decelerate the intrinsic ATPase activity counteracted ATP-induced K(ATP) channel inhibition, an effect that mimicked stabilization of the MgADP-bound posthydrolytic state at NBD2 by the gamma-phosphate analog orthovanadate. Temperature-induced channel activation was abolished by mutations that either prevent stabilization of MgADP at NBD2 or ATP at NBD1. These findings provide a paradigm of K(ATP) channel gating based on integration of both NBDs into a functional unit within the multimeric channel complex.

. The central role for SUR in defining the ATP sensitivity of K ATP channels is underscored by the abnormal cellular responses, associated with life-threatening disease, that result from malfunction of this regulatory channel module (7)(8)(9)(10). In fact, the ATPase function of the SUR subunit has been proposed to translate intracellular metabolic signals into membrane electrical events (5,11). However, the components of SUR responsible for signal transduction within the K ATP channel complex remain to be established.
As a member of the ATP-binding cassette (ABC) protein family, SUR contains two consensus sequences for nucleotide binding and hydrolysis known as nucleotide binding domains or NBDs (12)(13)(14). Both NBDs are apparently required for optimal performance in ABC proteins (15,16). A deficit in one disrupts the function of the other domain and the ABC protein as a whole, suggesting an interdependence of NBD functions (17)(18)(19). In SUR, mutations in the conserved NBD1 Walker A motif prevent ATP binding at both NBDs (20) 2 and interfere with the stimulatory effect of K ATP channel regulators, which act through NBD2 (21)(22)(23)(24)(25). Conversely, MgADP at NBD2 promotes stabilization of ATP at NBD1 indicating cooperative nucleotide binding at the NBDs of K ATP channels (6,19,26). NBD2 of SUR has been assigned the role of ATP hydrolysis (3,6,19), and discrete conformations driven by this intrinsic ATPase cycle have been identified as essential in channel gating (5). In contrast, NBD1 has limited catalytic activity (3), if at all (6,19), and the role of this domain in K ATP channel gating remains less understood (21,27,28).
Here, we report that an intact NBD1 is mandatory for NBD2 ATPase-dependent K ATP channel gating. Stabilization of ATP at NBD1 depends on, and simultaneously promotes, engagement of NBD2 into a MgADP-bound conformation required to counteract ATP-induced pore closure. Thus, rather than individual components of the regulatory subunit, it is the functional tandem formed by NBD1 and NBD2 that drives SURmediated nucleotide-dependent gating of the K ATP channel complex.

MATERIALS AND METHODS
ATPase Activity in NBD2-The ATPase activity in the second nucleotide binding domain (NBD2) of the cardiac sulfonylurea receptor (SUR2A) was measured as described (3,5). In brief, recombinant NBD2 (Gly 1306 -Thr 1498 ) was purified from Escherichia coli as a fusion to maltose binding protein using affinity chromatography on an amylose resin in (mM) 600 NaCl, 1 EDTA, 20 Tris (pH 7.4), and 10% glycerol. Products of [␥-32 P]ATP hydrolysis, measured in (mM) 34 KCl, 8 MgCl 2 , * This work was supported by National Institutes of Health Grants HL-64822 and HL-07111, the American Heart Association, the Miami Heart Research Institute, the American Physicians Fellowship for Medicine in Israel, the Bruce and Ruth Rappaport Program in Vascular Biology and Gene Delivery, and the Marriott Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Site-directed Mutagenesis-Point mutations in the core consensus sequence of the Walker motifs of NBD1 and NBD2 in the hamster cardiac SUR isoform, SUR2A, were introduced in the pCDNA3.1 plasmid by PCR amplification of both DNA strands with complementary primers containing desired amino acid changes (QuickChange, Stratagene). Primers were designed to contain 30 bases harboring the mutation in the middle, with at least one base at the 3Ј end being C or G. Mutated constructs were sequenced to confirm point mutations and rule out additional changes in the sequence (3).
Electrophysiological Measurements-Channel behavior was recorded in isolated ventricular myocytes dissociated from guinea pig hearts (29) as well as in COS-1 cells expressing recombinant K ATP channels (5). Pipettes (ϳ7-10 M⍀) were filled with (in mM) KCl 140, CaCl 2 1, MgCl 2 1, HEPES-KOH 5 (pH 7.3). For the inside-out configuration, cells were superfused with "internal solution" (in mM) KCl 140, MgCl 2 1, EGTA 5, HEPES-KOH 5 (pH 7.3). For the open cell-attached patch, internal solution was supplemented with glucose (1 g/liter), malic acid (5 mM), and pyruvic acid (5 mM). Following seal formation, the open cell-attached configuration was obtained by applying digitonin (8 g/ml) through a second pipette (filled with 5 g/ml propidium iodide and 0.5 g/ml rhodamine). Solution flow was visualized by rhodamine under ultraviolet light, and staining of the cell nucleus with propidium iodide served as a criterion for plasmalemmal permeabilization. Data were expressed as means Ϯ S.E.
Nucleotide Occlusion Procedures-In conjunction with current recording, two established approaches (30 -34) were employed to trap nucleotides within the catalytic site of the K ATP channel ATPase. First, ␥-phosphate analogs, orthovanadate and beryllium fluoride, were used to stabilize MgADP in the post-and prehydrolytic states of SUR, respectively (5,(32)(33)(34). To this end, sodium vanadate (100 mM, Sigma) was dissolved in water (pH 10), and orthovanadate was obtained by boiling the vanadate solution (pH 10). Freshly boiled stock was diluted to final concentration (pH 7.3) prior to use. Beryllium fluoride (BeF 2 ), as a 33% stock solution (Alfa), was dissolved in buffer solution containing 50 mM KF to produce sufficient amount of phosphate analogs BeF x (BeF 3 Ϫ and BeF 4 2Ϫ ) (31). For electrophysiological experiments, 50 mM KF replaced 50 mM KCl in the internal solution. Second, nucleotide occlusion was induced in the absence of ␥-phosphate analogs by cooling down the membrane patch after formation of the occluded nucleotide at Ͼ30°C (30). In these experiments we used a heating/cooling bath temperature controller (HCC-100A, Dagan Corp.) equipped with an electronically controlled high precision (Ϯ0.1°C) and broad range Peltier thermocouple set between 4 and 32°C.

NBD1 Mandatory for NBD2-dependent K ATP Channel Gat-
ing-Engagement of the NBD2 ATPase cycle into discrete conformations determines K ATP channel behavior (5). In this regard, ␥-phosphate analogs, orthovanadate and beryllium fluoride, are valuable tools that arrest the ATPase cycle in distinct conformations by stabilizing MgADP at the catalytic site (32)(33)(34). Here, ATPase activity in recombinant NBD2 was 77.6 Ϯ 3.9 nmol P i /min/mg (n ϭ 11) and was suppressed to 17.4 Ϯ 2.8 (n ϭ 11) and 6.4 Ϯ 2.2 (n ϭ 4) nmol P i /min/mg by 1 mM orthovanadate and beryllium fluoride, respectively (Fig.  1A). Vigorous activation of ATP-inhibited wild-type recombinant K ATP channels (Kir6.2/SUR2A, Fig. 1C) produced by orthovanadate through stabilization of NBD2 in a MgADP-bound post-hydrolytic conformation (5) was disrupted not only by mutations in NBD2 but also in NBD1 (Fig. 1, B and E). Specifically, orthovanadate-induced K ATP channel opening was abolished either by replacing aspartate with asparagine in the Walker B motif of the SUR2A NBD2 (D1470N), a mutation that disrupts ATP hydrolysis (3,32), or by exchanging lysine for alanine in the Walker A motif of NBD1 (K708A), which precludes ATP binding to SUR1 (20; Fig. 1, B and E). On average, channel activity in the presence of 0.25 mM ATP and 1 mM orthovanadate was in the wild type 72 Ϯ 18% of maximal channel opening measured in the absence of ATP (n ϭ 4; Fig.  1B). It was reduced to 5 Ϯ 3% in the D1470N mutant and to 3 Ϯ 1% in the K708A mutant (n ϭ 4; Fig. 1B). Similarly, either the D1470N or the K708A mutation reversed the inhibitory effect of beryllium fluoride (Fig. 1, B, D, and E), which traps the NBD2 ATPase in a prehydrolytic ATP-like bound state (5). Accordingly, K ATP channel activity in the presence of 0.1 mM MgADP and 1 mM BeF x was in wild-type, D1470N, and K708A  1 mM), respectively. Channel activity was expressed relative to control activity measured in the absence of ATP or ADP for orthovanadate (n ϭ 5) and BeF x (n ϭ 4), respectively. C, vigorous opening of wild-type (wt) Kir6.2/SUR2A channels visualized in inside-out patches as a downward deflection relative to the zero current level (dotted line). Recombinant channel activity was readily inhibited by ATP (0.25 mM). ATPinduced channel inhibition was reversed by the ␥-phosphate analog orthovanadate, an effect that required 14.2 Ϯ 0.3 min (n ϭ 5) to reach maximal effect. During the 3-min-long break, no change in channel activity was observed. D, in contrast to orthovanadate, beryllium fluoride (BeF x ϭ BeF 2 ϩ KF 50 mM) did not antagonize ATP-induced channel inhibition. Rather, in the presence of Mg 2ϩ and ADP, BeF x inhibited Kir6.2/SUR2A channel opening. E, mutation (K708A) in the NBD1 Walker A motif of SUR2A prevented the effect of both orthovanadate and BeF x on K ATP channels. During the 12-min-long break, no change in channel activity was observed. mutants, respectively, 8 Ϯ 2%, 63 Ϯ 14%, and 79 Ϯ 10% of the maximal activity measured in the absence of nucleotide (n ϭ 3; Fig. 1B). Thus, an intact NBD1 is required for the NBD2 ATPase dependent K ATP channel gating.
ATP Stabilized at NBD1 Is Necessary for NBD2-dependent K ATP Channel Gating-MgADP at NBD2 promotes K ATP channel opening (5) and concomitantly stabilizes ATP at NBD1 even in the absence of ␥-phosphate analogs (6,19,26). Whether such stabilization of ATP at NBD1 is required for NBD2-dependent channel gating has not been resolved thus far, as the lifetime of the MgADP-bound conformation is limited because of dissociation from NBD2 (5,19,26). To slow ATPase activity and promote the lifetime of MgADP at the catalytic site, we cooled membrane patches and monitored K ATP channel behavior online. ATP inhibition of K ATP channels (at Ͼ30°C) was reversed by transient cooling (to 5°C) of membrane patches ( Fig. 2A). On average, in inside-out patches (Fig. 2B, upper panel) ATPinduced channel inhibition was markedly reduced from an initial IC 50 (Fig. 2B, lower panel), a distinct patch condition, the sensitivity of K ATP channels toward ATP was reduced from an initial IC 50 of 270 Ϯ 21 M (n ϭ 6 -14) to 1.6 Ϯ 0.2 mM (n ϭ 2-7) after cooling. In contrast, cooling-induced K ATP channel activation was not achieved in the presence of poorly hydrolyzable ATP analogs, AMP-PNP (n ϭ 5; Fig. 2C) or Ap 4 A (n ϭ 3; Fig.  2D). Thus, cooling does not disrupt hydrolysis-independent K ATP channel gating but rather promotes MgADP stabilization at the catalytic site (30), which in turn is essential for channel opening. Although MgADP can be stabilized at NBD2 (5), excluding ATP, by application of the ADP-regenerating system hexokinase plus glucose prevented cooling-induced reduction of K ATP channel sensitivity to ATP (Fig. 2E). This suggests that the presence of ATP at NBD1 is a prerequisite for MgADP to serve as a K ATP channel regulator at NBD2. Indeed, in the same patch, cooling performed in the presence of ATP induced K ATP channel opening, which was abolished by activation of the ADP-scavenging creatine phosphate/creatine kinase system The solid lines represent Hill plots reconstructed based on parameters obtained by fitting experimental points. Cooling failed to activate K ATP channels inhibited by the nonhydrolyzable ATP analogs AMP-PNP (C) or Ap 4 A (D). In the same patches, replacing AMP-PNP or Ap 4 A with ATP reversed K ATP channel inhibition following cooling. E, K ATP channel inhibition by 0.5 mM ATP was partially antagonized by MgADP (38 Ϯ 7% of control activity in the absence of nucleotides, n ϭ 3). Then, 0.5 mM ADP was clamped in the presence of the ATP-scavenging hexokinase/glucose system. Following the cooling interval, the sensitivity of K ATP channel toward ATP plus MgADP (44 Ϯ 9% of control, n ϭ 3) as well as the sensitivity of channels to ATP alone were not significantly changed. In the same patch, cooling in the presence of MgATP antagonized channel inhibition (84 Ϯ 6% of control activity, n ϭ 3), an effect reversed by activation of the ADP-scavenging creatine kinase/creatine phosphate (CrP) system. All experiments were performed in open cell-attached patches. The temperature gradient is shown by a color bar on top of records. (Fig. 2E). Therefore, cooperative stabilization of ATP and MgADP at NBDs translates into K ATP channel opening at inhibitory levels of ATP.
Functional Tandem of NBD1 and NBD2 Secures K ATP Channel Gating-The cooperative binding of nucleotides at the NBDs of K ATP channels (6,19,26) suggests a joint action of NBD1 and NBD2 on channel gating. Disruption of either NBD1 or NBD2 through mutation precluded the reduction in ATPsensitivity observed after cooling in wild-type Kir6.2/SUR2A K ATP channels (Fig. 3A). Specifically, mutations in the Walker motifs of NBD2 (K1349A and/or D1470N), which diminish the intrinsic ATPase activity (3), attenuated K ATP channel activation following cooling (Fig. 3B). The time course of K ATP channel activation was fitted by the Boltzmann's function, I max ⅐[1 ϩ exp((T 0.5 Ϫ t)/k)] Ϫ1 , where I max is the maximal channel activity expressed relative to the activity in the absence of ATP, T 0.5 the time of half-activation, t the relative time of reheating, and k the slope of the time course. In wild-type Kir6.2/SUR2A channels, parameters defining the time course were as follows: I max ϭ 1.12 Ϯ 0.04, T 0.5 ϭ 0.60 Ϯ 0.03, and k ϭ 0.24 Ϯ 0.03 (n ϭ 3; Fig. 3D). The K1349A/D1470N mutations significantly decreased and delayed activation of K ATP channels (I max ϭ 0.42 Ϯ 0.04, T 0.5 ϭ 0. 92 Ϯ 0.05, k ϭ 0.16 Ϯ 0.04, n ϭ 3; Fig. 3D). Furthermore, the K708A mutation that prevents ATP binding to NBD1 (20) also abolished K ATP channel activation (n ϭ 3; Fig. 3, C and D). Thus, disrupting either NBD1 or NBD2 impedes K ATP channel opening upon cooling, indicating that NBDs act as a functional unit rather than as independent determinants of channel gating. DISCUSSION In the hetero-octameric K ATP channel complex (35), the sulfonylurea receptor confers fine nucleotide modulation of K ϩ permeation through the channel pore (7)(8)(9)(10)(11)(12)(13)(14). In fact, the powerful metabolic sensor role of K ATP channels may stem from the nonequivalent properties of NBD1 and NBD2 recognized within SUR (6, 19 -22, 26). NBD1 has been demonstrated to bind nucleotides, whereas NBD2 hydrolyzes ATP (6,19,20,26); yet the individual and/or collective contribution of NBDs in K ATP channel gating is not fully understood. In this study, we provide functional evidence that cooperative interaction rather than the independent contribution of each NBD is critical for K ATP channel regulation. These findings provide a paradigm for K ATP channel gating based on the integration of both NBDs into a functional unit within the multimeric channel complex.
Specifically, although gating of cardiac K ATP channels was related to discrete conformations in the ATPase cycle at NBD2 (5), it is now shown that the intactness of NBD1 is critical for this function. In agreement with cooperative binding of nucleotides to SUR, where MgADP at NBD2 promotes stabilization of ATP at NBD1 (6,19), here ATP, but not ADP, at NBD1 was required to promote MgADP-induced opening of K ATP channels in the presence of normally inhibitory concentrations of ATP.
The requirement for "cross-talk" between NBDs appears to be a common feature of several members of the ABC family. In fact, mutations of p-glycoprotein that preclude nucleotide binding or arrest ATP hydrolysis at one NBD prevent normal function at the other NBD (18, 34, 36 -38). Moreover, ATP hydrolysis at both NBDs is necessary for transport of a single molecule in the p-glycoprotein transport cycle (16,38). In the cystic fibrosis transmembrane regulator, the close proximity of NBDs (39) permits nucleotide hydrolysis at one NBD to influence nucleotide binding at the other NBD site, thereby regulating chloride conductance (40). Crystal structures of ABCrelated proteins (such as the HisP of histidine permease, MalK of the trehalose/maltose transporter or the Mre11/Rad50-ATPase DNA repair complex) suggest dimerization of the two FIG. 4. K ATP channels gated through interaction between nucleotide binding domains of the SUR subunit. A "bird's eye view" of the hetero-octameric K ATP channel complex composed of four poreforming Kir6.2 and four regulatory SUR subunits, which possess two nucleotide-binding domains, NBD1 and NBD2. Under basal metabolic state (in the presence of high levels of intracellular ATP), ATP is bound to NBD1 and hydrolyzed at NBD2 by the intrinsic ATPase. Under this condition, the ADP-scavenging creatine kinase (CK) system prevents accumulation of ADP at the channel site and perpetuates the ATPase cycle impeding ATP stabilization at NBD1. Under metabolic stress, a drop in ATP regeneration increases MgADP at the channel site (5,11) and prolongs the lifetime of the MgADP-bound conformation at NBD2 leading to entrapment of ATP at NBD1 and channel opening. Cooling and the ␥-phosphate analog orthovanadate (ortho-V) also induce channel activation by promoting cooperative nucleotide interactions at NBDs.
FIG. 3. Intact NBD1 and NBD2 of SUR2A necessary for cooling induced K ATP channel opening. A, in an inside-out patch, ATPinduced inhibition of wild-type (w.t.) Kir6.2/SUR2A channels, was reversed by a cooling period. B, mutations in Walker A (1349 lysine to alanine) and B (1470 aspartate to asparagine) motifs of NBD2 in the SUR2A subunit (K1349A/D1470N) significantly reduced ATP-induced inhibition of recombinant K ATP channels following cooling. C, mutation of the Walker A (708 lysine to alanine) domain of NBD1 in SUR2A (K708A) abolished the antagonism of ATP-induced K ATP channel inhibition following cooling. In A-C, the temperature gradient is shown by a color bar above the records. D, average activation time course of wild-type (WT) or mutated recombinant K ATP channels expressed relative to control activity measured in the absence of ATP. Time of activation was normalized to the time required for heating a particular membrane patch from 7 to 30°C. Solid lines represent Boltzmann's curves constructed using parameters described in the text.
NBDs with transfer of mechanical energy from one NBD to the other within the protein architecture (41)(42)(43). In this process, ATP binding has been found critical in engaging the two NBD domainsintoacompactdimeressentialforsupportinghydrolysisdependent protein function (42). Specifically, ATP binding has been proposed to bring the ATPase domain into a position that ultimately contributes to the "signaling-competent state" of the protein complex (43,44). Thus, by analogy, it is conceivable that ATP binding to NBD1 of SUR is a necessary step in securing the proper structural arrangement of NBD2 required to translate conformational transitions during the ATPase cycle into K ATP channel gating. This is in accord with the recent suggestion that the SUR NBD1 in its ATP-bound state directly interacts with the pore-forming Kir6.2 subunit of the K ATP channel, counteracting ATP-induced channel inhibition (45).
In a cell in the basal metabolic state, ATP exceeds ADP concentration such that ATP should always be bound to NBD1, whereas hydrolysis of ATP would produce MgADP at NBD2 (Fig. 4). Although this nucleotide combination should in principle be associated with channel opening (6,19), in a cardiac cell K ATP channels are normally closed (Fig. 4). Indeed, despite continuous ATPase activity at NBD2, the product of ATP hydrolysis is rapidly removed by cellular ADP-scavenging systems, such as that catalyzed by creatine kinase, limiting the lifetime of the MgADP-bound conformation and preventing channel opening (5,11,46,47) (Fig. 4). However, under metabolic stress, which suppresses creatine kinase activity (48), ADP will increase at the channel site (5). A prolonged lifetime of the MgADP-bound conformation promotes ATP stabilization at NBD1 and thereby channel opening (5,6,19) (Fig. 4).
Although here we employed temperature to cooperatively stabilize the MgADP-bound conformation of NBD2 and ATP at NBD1 (19,30), such a phenomenon may actually be relevant in nature as well. Indeed, K ATP channel behavior has been studied in cardiac myocytes from goldfish that had been acclimated to low temperatures (7°C) as used in this study (49). K ATP channels from these animals were nearly insensitive to concentrations of ATP that were completely inhibitory in non-coldacclimated animals (49). This observation is given a mechanistic basis by the present data, which suggest that membrane cooling promotes cooperative nucleotide stabilization at NBDs resulting in reduction of the channel's ATP sensitivity. The resulting alteration in channel activity is proposed to promote survival at low temperature by membrane potential clamping, as well as maintenance of ionic and energetic homeostasis (49). Moreover, in mammals, a cardioprotective effect of K ATP channels is also present at low temperature, and this effect has been exploited in cardioplegia procedures (50). Thus, the results of the current study provide a working model of K ATP channel function and point to potential avenues in addressing the biology of cold tolerance (51).