ATP Binding Properties of the Nucleotide-binding Folds of SUR1*

Pancreatic beta cell ATP-sensitive potassium (KATP) channels regulate glucose-induced insulin secretion. The activity of the KATP channel, composed of SUR1 and Kir6.2 subunits, is regulated by intracellular ATP and ADP, but the molecular mechanism is not clear. To distinguish the ATP binding properties of the two nucleotide-binding folds (NBFs) of SUR1, we prepared antibodies against NBF1 and NBF2, and the tryptic fragment of SUR1 was immunoprecipitated after photoaffinity labeling with 8-azido-[32P]ATP. The 35-kDa fragment was strongly labeled with 5 μm 8-azido-[32P]ATP even in the absence of Mg2+ and was immunoprecipitated with the antibody against NBF1. The 65-kDa fragment labeled with 100 μm 8-azido-[α-32P]ATP in the presence of Mg2+ was immunoprecipitated with anti-NBF2 and anti-C terminus antibodies. These results indicate that NBF1 of SUR1 binds 8-azido-ATP strongly in a magnesium-independent manner and that NBF2 binds 8-azido-ATP weakly in a magnesium-dependent manner. Furthermore, the 65-kDa tryptic fragment was not photoaffinity-labeled with 8-azido-[γ-32P]ATP at 37 °C, whereas the 35-kDa tryptic fragment was, suggesting that NBF2 of SUR1 may have ATPase activity and that NBF1 has none or little.

Pancreatic beta cell ATP-sensitive potassium (K ATP ) channels regulate glucose-induced insulin secretion. The activity of the K ATP channel, composed of SUR1 and Kir6.2 subunits, is regulated by intracellular ATP and ADP, but the molecular mechanism is not clear. To distinguish the ATP binding properties of the two nucleotide-binding folds (NBFs) of SUR1, we prepared antibodies against NBF1 and NBF2, and the tryptic fragment of SUR1 was immunoprecipitated after photoaffinity labeling with 8-azido-[ 32 P]ATP. The 35-kDa fragment was strongly labeled with 5 M 8-azido-[ 32 P]ATP even in the absence of Mg 2؉ and was immunoprecipitated with the antibody against NBF1. The 65-kDa fragment labeled with 100 M 8-azido-[␣-32 P]ATP in the presence of Mg 2؉ was immunoprecipitated with anti-NBF2 and anti-C terminus antibodies. These results indicate that NBF1 of SUR1 binds 8-azido-ATP strongly in a magnesium-independent manner and that NBF2 binds 8-azido-ATP weakly in a magnesium-dependent manner. Furthermore, the 65-kDa tryptic fragment was not photoaffinity-labeled with 8-azido-[␥-32 P]ATP at 37°C, whereas the 35-kDa tryptic fragment was, suggesting that NBF2 of SUR1 may have ATPase activity and that NBF1 has none or little.
ATP-sensitive potassium (K ATP ) 1 channels of pancreatic beta cells regulate insulin release by altering the beta cell membrane potential (1)(2)(3)(4). Because the channels are inhibited by ATP and activated by MgADP, they act as sensors of intracellular nucleotides, but it is not known how they monitor concentrations of intracellular ATP and ADP. The K ATP channel is a hetero-octamer composed of SUR1 and Kir6.2 subunits (5)(6)(7)(8). SUR1 is a member of the ABC superfamily including P-glycoprotein (MDR1), MRP1, and the cystic fibrosis transmembrane conductance regulator (CFTR) (9, 10), all of which have two nucleotide-binding folds (NBFs) in the molecule; Kir6.2 is an inwardly rectifying potassium channel (11,12).
Electrophysiological studies have shown that K ATP channels are inhibited by ATP through binding to Kir6.2 and are activated by MgADP through binding to SUR1 (13,14). We have reported previously that mutations of either the Walker A or B motifs of NBF1, K719M, and D854N abolish the high-affinity 8-azido-ATP binding of SUR1, whereas the equivalent mutations in NBF2 do not affect ATP binding (15). We have also reported that MgADP and MgATP stabilize binding of prebound 8-azido-ATP to SUR1 and that mutations in the Walker A and B motifs of NBF2 abolish this stabilizing effect of MgADP and MgATP (16). These biochemical results suggest that SUR1 binds 8-azido-ATP strongly at NBF1 and MgADP at NBF2 and that the two NBFs cooperate in nucleotide binding. However, due probably to the high-affinity ATP binding of NBF1, we have not been able to detect nucleotide binding of NBF2 directly.
The ATP binding/hydrolysis properties of the NBF are best studied with MDR1 among the ABC proteins, and its two NBFs are proposed to be equivalent in function (17). The covalent modification of the cysteine residue in the Walker A motif of either NBF has been shown to be sufficient to inactivate the ATPase activity (18 -21). Mutation of the Walker A lysine residue of either NBF, thought to interact with the ␤and ␥-phosphates of ATP, abolishes the ATPase activity of MDR1 and its ability to confer multidrug resistance (22)(23)(24). Both NBFs of MDR1 can hydrolyze nucleotides (25)(26)(27), and mild trypsin digestion showed that two NBFs were labeled in equal proportion with 8-azido-ATP after hydrolysis in the presence of orthovanadate (28). SUR1 photoaffinity-labeled with 8-azido-[ 32 P]ATP was digested mildly with trypsin. Tryptic fragments were immunoprecipitated with antibodies against NBF1, NBF2, and a Cterminal sequence of SUR1. The ATP binding properties of the tryptic fragments indicate that NBF1 of SUR1 binds 8-azido-ATP strongly in a magnesium-independent manner and that NBF2 binds 8-azido-ATP weakly in a magnesium-dependent manner. It is also suggested that, although NBF2 of SUR1 may have ATPase activity, NBF1 has little, if any at all. Photoaffinity Labeling of a High-affinity ATP-binding Site-Membranes from COS-7 cells expressing SUR1, prepared as described (15), were incubated with 5 M 8-azido-[␣-32 P]ATP in 5 l of TEM buffer (40 mM Tris-Cl (pH 7.5), 0.1 mM EGTA, and 1 mM MgSO 4 ) containing 2 mM ouabain for 10 min at 37°C. The reactions were stopped by the addition of 500 l of ice-cold TEM buffer, and free 8-azido-[ 32 P]ATP was removed after centrifugation (15,000 ϫ g, 5 min, 2°C). Pellets were resuspended in 8 l of TEM buffer and irradiated for 5 min (at 254 nm, 5.5 milliwatts/cm 2 ) on ice. Samples were electrophoresed on a 12% SDS-polyacrylamide gel and autoradiographed. When membranes were incubated with 5 M 8-azido-[␣-32 P]ATP in the absence of Mg 2ϩ , the reactions were stopped by the addition of 500 l of ice-cold TE buffer (40 mM Tris-Cl (pH 7.5) and 0.1 mM EGTA). After centrifugation, pellets were resuspended in 8 l of TE buffer and UV-irradiated as described above.
Photoaffinity Labeling of a Low-affinity ATP-binding Site-Membranes were incubated with 100 M 8-azido-[␣-32 P]ATP in 5 l of TEM buffer containing 2 mM ouabain for 10 min on ice. Reactions were done * This work was supported by Grant-in-aid for Scientific Research on Priority Areas "ABC Proteins" 10217205 from the Ministry of Education, Science, Sports, and Culture of Japan and by research fellowships from the Japan Society for the Promotion of Science for Young Scientists. 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.
Preparation of Antibodies against NBF1 and NBF2-NBF1 (amino acids 695-934) or NBF2 (amino acids 1361-1582) of hamster SUR1 was fused with maltose-binding protein in a pMALc2 vector (New England Biolabs Inc.), and fusion proteins were expressed in Escherichia coli strain BL21. The fusion proteins were purified with amylose resin (New England Biolabs Inc.), and rabbit polyclonal antibodies were raised against the purified proteins.
Limited Trypsin Digestion and Immunoprecipitation of Tryptic Fragments of SUR1-Photoaffinity-labeled membranes were resuspended in TE buffer containing 2.5 g/ml trypsin and 250 mM sucrose to 10 g of membrane proteins/l and incubated for 60 min at 37°C. 100 l of radioimmune precipitation assay buffer (20 mM Tris-Cl (pH 7.5), 1% Triton X-100, 0.1% SDS, and 1% sodium deoxycholate) containing 100 g/ml p-amidinophenylmethanesulfonyl fluoride was added to the samples to terminate proteolysis, and membrane proteins were solubilized for 30 min at 4°C. After centrifugation for 15 min at 15,000 ϫ g, tryptic fragments were immunoprecipitated from the supernatant with the antibody raised against NBF1, NBF2, or the C-terminal 21 amino acids of rat SUR1.

RESULTS
Limited Trypsin Digestion of SUR1-It has been suggested that SUR1 binds 8-azido-ATP strongly at NBF1 and binds MgADP at NBF2 and that the two NBFs of SUR1 cooperate in nucleotide binding (16). However, we have not been able to detect nucleotide binding to NBF2 directly due to the highaffinity ATP binding to NBF1. Because mild trypsin digestion of MDR1 or CFTR produces two large polypeptides corresponding to the N-and C-terminal halves (29 -32), we separated the two NBFs of SUR1 by mild digestion with proteases. When SUR1 was photoaffinity-labeled with 5 M 8-azido-[␣-32 P]ATP and digested with 2.5 g/ml trypsin, two photoaffinity-labeled fragments of ϳ100 and 35 kDa were produced (Fig. 1A). The amount of the 100-kDa fragment decreased, and the amount of the 35-kDa fragment concomitantly increased with increasing incubation time with trypsin. When SUR1 was digested with trypsin under the same condition, the anti-C terminus antibody recognized the 100-and 65-kDa fragments, but not the 35-kDa fragment, in immunoblotting (Fig. 1B). These results suggest that the 100-kDa fragment is further digested to the 35-kDa fragment, which contains the high-affinity ATP-binding site, and the 65-kDa fragment, which contains the C terminus, as diagrammed in Fig. 2.
Antibodies against NBF1 and NBF2 of SUR1-To determine which NBF the 35-kDa fragment contained, polyclonal antibodies were prepared against NBF1 or NBF2 fused with maltosebinding protein. These antibodies recognized NBF1 and NBF2 of SUR1 expressed in E. coli, respectively, but scarcely recognized those of MDR1 (Fig. 3). The photoaffinity-labeled 35-kDa fragment was precipitated with the anti-NBF1 antibody, but not with the anti-NBF2 or anti-C terminus antibody (Fig. 4,  lanes 1-3). These results indicate that the 35-kDa tryptic fragment contains NBF1 and support the possibility that the 100-kDa fragment is digested to the 35-kDa fragment containing NBF1 and the 65-kDa fragment containing the C terminus (Fig. 2).
Interaction of the NBFs of SUR1 with 8-Azido-[␣-32 P]ATP-We have suggested that NBF1 of SUR1 binds 8-azido-ATP strongly even in the absence of Mg 2ϩ and that NBF2 binds MgADP (15,16). To investigate the ATP binding properties of the NBFs of SUR1, SUR1 was photoaffinity-labeled with 5 or 100 M 8-azido-[␣-32 P]ATP in the presence or absence of Mg 2ϩ . Proteins were then immunoprecipitated with the anti-NBF1, anti-NBF2, or anti-C terminus antibody after mild trypsin digestion (Fig. 4). The labeled 35-kDa fragment was precipitated with the anti-NBF1 antibody when SUR1 was photoaffinity-labeled with 5 or 100 M 8-azido-[␣-32 P]ATP either in the presence or absence of Mg 2ϩ . In contrast, the labeled 65-kDa fragment was precipitated with the anti-NBF2 and anti-C terminus antibodies only when SUR1 was incubated with 100 M 8-azido-[␣-32 P]ATP in the presence of Mg 2ϩ and UV-irradiated without removing free ligands. When membranes were washed with buffer before UV irradiation, the 65-kDa fragment was not labeled with 100 M 8-azido-[␣-32 P]ATP with Mg 2ϩ (data not shown). These results indicate that NBF1 of SUR1 binds 8-azido-ATP strongly even in the absence of Mg 2ϩ and that NBF2 binds 8-azido-ATP weakly only in the presence of Mg 2ϩ . Less labeling of NBF1 with 100 M 8-azido-[␣-32 P]ATP on ice (lane 4) compared with that with 5 M at 37°C suggests that high-affinity 8-azido-ATP binding to NBF1 is temperature-dependent and that incubation at 37°C is necessary. It also indicates that the 100-kDa tryptic fragment contains both NBF1 and NBF2 because it was precipitated with the anti-NBF1, anti-NBF2, or anti-C terminus antibody.
Interaction of the NBFs of SUR1 with 8-Azido-[␥-32 P]ATP-Among the eukaryotic ABC superfamily proteins, MDR1, MRP1, CFTR, and ABCR (the rod photoreceptor-specific ABC transporter responsible for Stargardt's disease) have been reported to have ATPase activity (23,25,(33)(34)(35)(36)(37). We have proposed that NBF2 of SUR1 has ATPase activity and that MgADP, either by direct binding to NBF2 or by hydrolysis of bound ATP at NBF2, stabilizes prebound 8-azido-ATP binding at NBF1 (16). Because it was possible to distinguish the two FIG. 1. Limited digestion of SUR1 with trypsin. A, autoradiogram of membrane proteins (25 g) from COS-7 cells expressing SUR1 photoaffinity-labeled with 5 M 8-azido-[␥-32 P]ATP as described under "Experimental Procedures." Proteins were digested with 2.5 g/ml trypsin at 37°C for the indicated periods and separated by 12% SDSpolyacrylamide gel electrophoresis. B, immunoblot of membrane proteins (20 g) from COS-7 cells expressing SUR1 with the anti-C terminus antibody. Proteins were digested and separated as described for A. The band indicated by the asterisk is a nonspecific one because membranes from untransfected COS-7 cells produce the same band (data not shown).

FIG. 2. Predicted diagram of limited trypsin digestion of SUR1.
Mild trypsin digestion of SUR1 (180 kDa) first produces a 100-kDa fragment containing NBF1, NBF2, and the C terminus then produces a 35-kDa fragment containing NBF1 and a 65-kDa fragment containing NBF2 and the C terminus.
NBFs by their interaction with ATP by limited trypsin digestion, we tried to determine whether SUR1 has ATPase activity. Membranes expressing SUR1 were incubated with 50 M 8-azido-[␣-32 P]ATP or 8-azido-[␥-32 P]ATP in the presence or absence of Mg 2ϩ for 10 min at 37°C and UV-irradiated without removing free ligands. SUR1 was mildly digested with trypsin, and the tryptic fragments were immunoprecipitated with the anti-NBF1, anti-NBF2, or anti-C terminus antibody (Fig. 5). The 35-kDa fragment containing NBF1 was photoaffinity-labeled with both 8-azido-[␣- 32 5 and 6). These results suggest that ␥-phosphate dissociates from 8-azido-ATP bound at NBF2 and that NBF2 of SUR1 may have ATPase activity. In contrast to NBF2, NBF1 appears to have little or no ATPase activity under the conditions examined.
To demonstrate loss of the terminal phosphate from 8-azido-[␥-32 P]ATP bound at NBF2, we tried to show photoaffinity labeling of NBF2 with 8-azido-[␥-32 P]ATP at time 0. Membranes were incubated with 50 M 8-azido-[␥-32 P]ATP in the presence or absence of Mg 2ϩ on ice and photoaffinity-labeled without removing free ligands. The 65-kDa tryptic fragment containing NBF2 was not, however, photoaffinity-labeled with 8-azido-[␥-32 P]ATP even when the reaction with trypsin was done on ice, although the same fragment was photoaffinitylabeled with 8-azido-[␣-32 P]ATP (data not shown).

DISCUSSION
In this study, we have determined that NBF1 of SUR1 binds 8-azido-ATP strongly in a magnesium-independent manner and that NBF2 binds 8-azido-ATP weakly in a magnesium-dependent manner. We have also demonstrated that NBF1 has no or very little ATPase activity, whereas NBF2 of SUR1 may have ATPase activity. These characteristics of the two NBFs of SUR1 are quite different from those of MDR1: MDR1 has no high-affinity binding site for ATP; both NBFs of MDR1 hydrolyze ATP; and the roles of the two NBFs in drug transport are assumed to be equivalent in MDR1 (28). After hydrolysis, MDR1 is thought to form the metastable state MDR1⅐MgADP⅐P i complex, and the phosphate ion is released before MgADP. When ATP is hydrolyzed in the presence of orthovanadate, orthovanadate binds to this intermediate in place of the released P i to form a stable inhibitory complex, MDR1⅐MgADP⅐V i (17). Therefore, when MDR1 is incubated with 8-azido-[␣-32 P]ATP in the presence of orthovanadate, MDR1 traps 8-azido-[␣-32 P]ADP and is specifically and strongly photoaffinity-labeled (15,17,21).
Photoaffinity labeling of CFTR with 8-azido-[␣-32 P]ATP has also been recently reported (32). NBF1 of CFTR was preferentially labeled with 5 M 8-azido-[␣-32 P]ATP in the absence of orthovanadate. Because this high-affinity labeling of NBF1 is magnesium-and temperature-dependent, Szabó et al. (32) predicted that the nucleotide occlusion occurs in the ATP hydrolysis cycle of CFTR. The high-affinity labeling of NBF1 of SUR1 is also temperature-dependent (Fig. 4), but does not require Mg 2ϩ (15). Accordingly, a structural change (a nucleotide oc- Membrane proteins were digested with 2.5 g/ml trypsin for 60 min at 37°C and solubilized with radioimmune precipitation assay buffer. The tryptic fragments were immunoprecipitated with the anti-NBF1 ( lanes  1 and 4), anti-NBF2 (lanes 2 and 5), and anti-C terminus (lanes 3 and 6) antibodies and separated by 12% polyacrylamide gel electrophoresis. Undigested SUR1, a 100-kDa tryptic fragment containing both NBF1 and NBF2, a 65-kDa fragment containing NBF2, and a 35-kDa fragment containing NBF1 are indicated. Experiments were done in triplicate.  4 -6) in the presence of Mg 2ϩ for 10 min at 37°C, followed by UV irradiation without removing free ligands. Membrane proteins were mildly digested with trypsin, and tryptic fragments were immunoprecipitated with the anti-NBF1 (lanes 1 and  4), anti-NBF2 (lanes 2 and 5), or anti-C terminus (lanes 3 and 6) antibody. Experiments were done in triplicate. clusion) might also occur in NBF1 of SUR1, but it would not be a step in the ATP hydrolysis cycle in the case of SUR1 because Mg 2ϩ is not required. The addition of orthovanadate increases photoaffinity labeling and results in the labeling of both NBFs of CFTR (32), but has no effects on photoaffinity labeling of SUR1 (15). These results suggest that the catalytic mechanism of the NBFs of SUR1 may be different from that of the NBFs of MDR1 and CFTR. This is in agreement with experiments demonstrating that orthovanadate does not influence K ATP channel activity (38,39).
To examine if NBF2 of SUR1 has ATPase activity, photoaffinity labeling of SUR1 with 8-azido-[␥-32 P]ATP was investigated. However, we could not confirm that NBF2 of SUR1 has ATPase activity because the 65-kDa fragment containing NBF2 was not photoaffinity-labeled with 8-azido-[␥-32 P]ATP even at time 0. This is probably because 8-azido-[␥-32 P]ATP bound at NBF2 is hydrolyzed, and the terminal phosphate is released during trypsin digestion. Recently, we found that NBF2 of SUR2A and SUR2B binds ATP in a magnesiumindependent manner, and we could observe photoaffinity labeling of the 65-kDa fragment containing NBF2 with both 8-azido-[␣-32 P]ATP and 8-azido-[␥-32 P]ATP in the absence of Mg 2ϩ , but could not observe photoaffinity labeling of the fragment with 8-azido-[␥-32 P]ATP in the presence of Mg 2ϩ . 2 These results support the idea that NBF2 of SUR1 and SUR2 has ATPase activity.
In CFTR, ATP hydrolysis is important in channel regulation. Electrophysiological studies have suggested that both NBFs of CFTR hydrolyze ATP. ATP hydrolysis at NBF1 is predicted to be involved in channel opening, whereas that at NBF2 is predicted to be involved in channel closing (40,41). Orthovanadate, which is assumed to trap ADP at an NBF, causes extreme stabilization of the open conformation of the CFTR channel (42,43). MgADP, produced by hydrolysis of bound MgATP at NBF2 of SUR1, is thought to stabilize ATP binding at NBF1, and K ATP channel activation may be induced primarily by the cooperative interaction of ATP binding at NBF1 and MgADP binding at NBF2 (16). Accordingly, ATP hydrolysis at NBF2 could modulate K ATP channel regulation. Because NBF1 was photoaffinity-labeled with 8-azido-[␣-32 P]ATP and 8-azido-[␥-32 P]ATP, ATP is not hydrolyzed at NBF1 under the conditions examined in this study. If NBF1 of SUR1 showed ATPase activity under some other conditions such as by binding pharmacological agents or unknown endogenous ligands, the activity of the K ATP channel would also be affected.
In summary, we have analyzed properties of the two NBFs of SUR1 and determined that NBF1 binds ATP strongly even in the absence of Mg 2ϩ and that NBF2 binds ATP weakly in a magnesium-dependent manner. It is now possible to examine the interaction with ATP at each NBF of SUR1. Because ATP binding and hydrolysis are important in the regulation of the K ATP channel, this study provides evidence of a critical part of the molecular mechanism of pancreatic beta cell K ATP channels.