INTRODUCTION

The sulfonylurea receptor (SUR)¹ is a member of the ATP-binding cassette (ABC) superfamily (1, 2), which includes cystic fibrosis transmembrane conductance regulator (CFTR), P-glycoprotein (Pgp), and multidrug resistance-associated protein (MRP) (3). Recent studies have shown that a sulfonylurea receptor and Kir6.2, an inward rectifier K⁺ channel member, constitute ATP-sensitive K⁺ (K_ATP) channels (4, 5). K_ATP channels, as the ATP and ADP sensors in pancreatic -cells, regulate glucose-induced insulin secretion by linking the cell's metabolic state to its membrane potential (6). It has generally been accepted that an increase in the ATP/ADP ratio inhibits K_ATP channel activity, while a decrease in the ratio stimulates activity (7-10). ADP, in the absence of Mg²⁺, however, may cause channel inhibition as a weaker analog of ATP. MgADP has been shown to antagonize channel inhibition by ATP (11, 12). ATP inhibition of K_ATP channel activity may require binding at two sites, and a MgADP antagonism might occur through competitive binding at one of the sites (10). The activity of the -cell K_ATP channel thus is regulated by a complex mechanism of intracellular ATP and ADP.

SUR1 has the highly conserved Walker A and B motifs and the SGGQ ABC signature in each putative nucleotide-binding fold (NBF), both of which are thought to be critical for nucleotide regulation of the functional activity of ABC proteins (3, 13). Pgp has been suggested to have no high affinity binding site for MgATP, and the ATP hydrolysis step generates a Pgp-MgADP-P_i conformation of high chemical potential that may be coupled to drug transport (14). Both NBFs of Pgp bind and hydrolyze MgATP. Thus, the roles of the two NBFs of Pgp in drug transport may be equivalent (15). On the other hand, in CFTR, a Cl channel, the roles of the two NBFs are thought to be different: ATP hydrolysis at one site controls channel opening and ATP hydrolysis at the other site regulates channel closing (16, 17). Although our previous studies have suggested that the SUR1 subunit confers the ATP sensitivity of pancreatic -cell K_ATP channels and that the Kir6.2 subunit forms the K⁺ ion-permeable domain of the channel (2, 4), direct evidence of the effects of ATP and ADP on SUR1 has not yet been reported. Nichols et al. (12) have shown that mutations in NBF2 abolish MgATP antagonism of ATP inhibition without affecting ATP inhibition itself.

It has been found recently that the apparent affinity for nucleotides of Pgp is greatly increased in the presence of vanadate (14) probably because a stable and reversible inhibitory complex is generated by trapping a nucleotide at the catalytic site. The vanadate-induced nucleotide trapping method has been used successfully for the analysis of the mechanism of ATP hydrolysis by MRP (18). Utilizing this technique, we show in the present study that unlike other ABC proteins such as Pgp and MRP, SUR1 strongly and stably binds ATP at NBF1 even in the absence of Mg²⁺, and that MgADP, through binding at NBF2, antagonizes the Mg²⁺-independent high affinity ATP binding at NBF1. Accordingly, the mechanism of the interaction of SUR1 with nucleotides is different from that of Pgp and MRP.

EXPERIMENTAL PROCEDURES

The 8-azido-[-³²P]ATP and 8-azido-[-³²P]ATP were purchased from ICN Biomedicals.

African green monkey cells COS-7 were propagated in Dulbecco's modified Eagle's medium with 10% fetal bovine serum under 5% CO₂ at 37 °C. COS-7 cells were transfected with hamster SUR1 expression vectors with LipofectAMINE^TM (Life Technologies, Inc.) according to the manufacturer's directions. Three days after transfection, membrane proteins were prepared by nitrogen cavitation as described previously (19) (without sucrose density gradient centrifugation). Immunoblotting was done with an antibody against the C-terminal 21 amino acids of hamster SUR1.²

Membrane proteins (10 µg) were incubated with 5 µM 8-azido-[-³²P]ATP or 8-azido-[-³²P]ATP, 2 mM ouabain, 0.1 mM EGTA, and 40 mM Tris-Cl (pH 7.5) in a total volume of 5 µl for 10 min at 37 °C. (Variations of these conditions are described in the figure legends.) The reactions were stopped by the addition of 400 µl of ice-cold Tris-EGTA buffer (40 mM Tris-Cl (pH 7.5), 0.1 mM EGTA), and free ATP was removed after centrifugation (15,000 × g, 10 min, 4 °C). Pellets were washed in the same buffer, resuspended in 8 µl of Tris-EGTA buffer, and irradiated for 5 min (at 254 nm, 5.5 milliwatts/cm²) on ice. Samples were electrophoresed on an 7% SDS-polyacrylamide gel, and autoradiographed. The trapped 8-azido-[³²P]ATP in SUR1 was measured by counting the radioactivity of the band of SUR1 excised from an SDS gel or by scanning with a radioimaging analyzer (BAS2000, Fuji Photo Film Co.). Experiments were done in duplicate or triplicate.

RESULTS

We investigated whether SUR1 trapped a nucleotide at the catalytic site in the presence of vanadate, as has been reported of Pgp. Membrane proteins from COS-7 cells transiently expressing SUR1 were first preincubated with vanadate, Mg²⁺, and 5 µM 8-azido-[-³²P]ATP for 10 min at 37 °C, and then unbound ligands were removed. The proteins were then irradiated with UV light, and a 140- to 150-kDa protein was found to be specifically photoaffinity-labeled (Fig. 1, lane 1). No specific photoaffinity-labeled protein was observed in membrane proteins from untransfected COS-7 cells (lanes 10 and 11). The mobility in SDS-polyacrylamide gel electrophoresis of the photoaffinity-labeled membrane protein was identical to that of SUR1 in Western blotting (data not shown), indicating that SUR1 was efficiently photolabeled with 8-azido-[-³²P] ATP in the presence of Mg²⁺ and vanadate.

Photoaffinity labeling of P-glycoprotein and SUR1 with 8-azido-[-³²P]ATP.

When membrane proteins from COS-7 cells transiently expressing human Pgp were incubated with vanadate, Mg²⁺, and 8-azido-[-³²P]ATP, a 170-kDa protein was specifically but weakly photoaffinity-labeled (lane 5). Vanadate-induced trapping of nucleotide was increased in the presence of verapamil, a substrate for Pgp (lane 6). Both vanadate and Mg²⁺ were required for the photoaffinity labeling (lanes 7-9). These results suggest that the apparent affinity for MgATP of Pgp was low and was increased in the presence of vanadate, as reported previously.

We then ascertained whether or not vanadate or Mg²⁺ were required for the photoaffinity labeling of SUR1. Surprisingly, neither vanadate nor Mg²⁺ was required (lanes 2-4).

Plasma membrane proteins were incubated with different concentrations of 8-azido-[-³²P]ATP in the presence of Mg²⁺ and vanadate (Fig. 2A) or with different concentrations of 8-azido-[-³²P]ATP in the absence of Mg²⁺ or vanadate (Fig. 2B). In both cases, nucleotide binding was apparently saturated at around 10 µM and then increased with increasing concentrations of 8-azido-ATP. These results indicate that the bound nucleotide in SUR1 is ATP, and that ATP binding is not dependent on Mg²⁺. These results also suggest that SUR1 may have two ATP binding sites, one high affinity and the other low affinity.

We examined the effects of mutations in NBF1 or NBF2 on Mg²⁺-independent ATP binding in SUR1. All the mutant and wild-type SUR1s expressed in transfected cells were detected as the 140- to 150-kDa mature form (Fig. 3A). Substitutions of the conserved lysine in Walker A, K719R and K719M (lanes 2 and 3), or the aspartate in Walker B, D854N (lane 4), abolished the binding of 5 µM 8-azido-[-³²P]ATP, although substitutions at equivalent sites in NBF2, K1385R, K1385M, or D1506N (lanes 5, 6, and 7) did not affect it. SUR1 with mutations in NBF1 binds ATP only slightly even when incubated with 40 µM 8-azido-[-³²P]ATP.

Immunoblot analysis (A) and photoaffinity labeling with 8-azido-[-³²P]ATP (B and C) of wild-type SUR1 and the SUR1 mutants.

ADP, in the absence of Mg²⁺, weakly antagonized Mg²⁺-independent high affinity ATP binding with a half-maximal inhibitory concentration (IC₅₀) of approximately 200 µM (Fig. 4A, lanes 1-5). ADP, in the presence of Mg²⁺, strongly antagonized the ATP binding (IC₅₀ < 10 µM) (Fig. 4A, lanes 6-10). Because Mg²⁺ did not affect the apparent affinity for ATP (Fig. 2), we assume that Mg²⁺ increases the affinity for ADP of SUR1. The K1385M mutation reduced MgADP antagonism (IC₅₀, approximately 80 µM) (B, lanes 6-10), although it did not affect ADP antagonism in the absence of Mg²⁺ (B, lanes 1-5).

DISCUSSION

The apparent affinity for MgATP of Pgp and MRP is low, and it increases in the presence of vanadate (Fig. 1) (14, 18). The nucleotide trapped in Pgp and MRP is ADP, and the metastable state, E-MgADP-P_i, where E is Pgp or MRP, is formed after hydrolysis (14, 18). As are other nucleotide-binding and -hydrolyzing proteins, nucleotide and Mg²⁺ coordinately bind to Pgp. We show here that SUR1 strongly binds ATP even in the absence of Mg²⁺, suggesting that the interaction of SUR1 with nucleotides is different from that of Pgp and MRP.

NBF1 of SUR1 may be a Mg²⁺-independent high affinity ATP binding site because mutations in NBF1 impaired ATP binding, and mutations at equivalent sites in NBF2 did not (Fig. 3) (12). The K1385M mutation reduced MgADP antagonism, although it did not affect ADP antagonism in the absence of Mg²⁺ (Fig. 4). Because the K1385M mutation did not affect Mg²⁺-independent high affinity ATP binding (Fig. 3), MgADP may antagonize the ATP binding at NBF1 through binding at NBF2, and ADP probably antagonizes the ATP binding as a weaker analog of ATP. SUR1 may have two ATP binding sites, one high affinity and the other low affinity, which may work cooperatively because SUR1 with mutations in NBF1 binds ATP only slightly even when incubated with 40 µM ATP (Fig. 2).

Mutations of SUR1 in humans cause persistent hyperinsulinemic hypoglycemia of infancy (PHHI) (20), and K_ATP channel function in pancreatic -cells isolated from patients with PHHI is lost or severely impaired (21). Nichols et al. (12) have reported that mutations in NBF2 of SUR1 found in patients with PHHI did not affect the ATP sensitivity of the -cell K_ATP channels in the absence or presence of Mg²⁺, but impaired MgADP antagonism of ATP inhibition. Their conclusions regarding the essential role of NBF2 of SUR1 in MgADP antagonism are supported by the ATP binding and ADP antagonism in vitro shown in this report.

Walker A and B motifs form a portion of a nucleotide-binding pocket (13). The lysine residue in Walker A may interact with the phosphoryl moiety of the bound nucleotide (22) and the aspartic acid in Walker B to coordinate with Mg²⁺ in the MgATP complex (23). Both NBFs are necessary for ATPase and transport activities of Pgp, because mutations of the lysine in Walker A or the aspartic acid in Walker B of either one of the two NBFs impaired those activities^3,4 (24, 25). The ATP binding affinity of Pgp was decreased when lysine residues in both NBFs were replaced by methionine at the same times³ (26). Since substitution of the lysine in Walker A of SUR1 impaired Mg²⁺-independent high affinity ATP binding, the lysine may interact with ATP, as do other ATP-hydrolyzing proteins. The aspartate in Walker B in NBF1 of SUR1, different from other ATP-hydrolyzing proteins which probably act via Mg²⁺, may interact directly with ATP, because replacement of this residue with asparagine-impaired Mg²⁺-independent high affinity ATP binding of SUR1.

It is postulated that drug transport of Pgp is coupled to relaxation of a high energy catalytic site conformation generated by ATP hydrolysis (15) and that ATP hydrolysis provides the free energy necessary to effect a conformational change that leads to the conducting state of CFTR (27). The Mg²⁺-independent high affinity ATP binding in SUR1 could support the ATP-dependent inhibition of K_ATP channel activity which is antagonized by MgADP (8, 11). Mg²⁺-independent high affinity ATP binding was less when the reaction was done at 0 °C than when it was done at 37 °C (data not shown). MgADP antagonism was not observed when MgADP (a final concentration of 100 µM) was added after SUR1 was incubated with 5 µM 8-azido-[-³²P]ATP for 10 min at 37 °C (data not shown), but MgADP completely blocked ATP binding when added at the same times (Fig. 4). This suggests that Mg²⁺-independent ATP binding at NBF1 causes a conformational change of SUR1 that makes the binding to ATP tighter and more stable.

Gribble et al. (28) reported recently that both NBFs in SUR1 are not essential for K_ATP channel inhibition by ATP, but that they are essential for channel activation by MgADP, using K719A and K1385M SUR1 mutants. However, we have observed that while K719M and K719R mutants severely impair functional expression of K_ATP channels, K1385M and K1385R mutants do not.² Although whether or not SUR1 has ATP hydrolysis activity is unknown, ATP binding to NBF1 of SUR1 might be important in maintaining K_ATP channels in the operative state. It has recently been shown that a truncation of Kir6.2 itself also functions as a K_ATP channel, with lower ATP sensitivity (29). Since SUR1 endows the truncated Kir6.2 with higher ATP sensitivity than that of the truncated Kir6.2 alone, both SUR1 and Kir6.2 subunits are probably involved in the regulation of channel activity by ATP and MgADP.