The first nucleotide binding fold of the cystic fibrosis transmembrane conductance regulator can function as an active ATPase.

Cystic fibrosis is caused by mutations in the cell membrane protein called CFTR (cystic fibrosis transmembrane conductance regulator) which functions as a regulated Cl channel. Although it is known that CFTR contains two nucleotide domains, both of which exhibit the capacity to bind ATP, it has not been demonstrated directly whether one or both domains can function as an active ATPase. To address this question, we have studied the first CFTR nucleotide binding fold (NBF1) in fusion with the maltose-binding protein (MBP), which both stabilizes NBF1 and enhances its solubility. Three different ATPase assays conducted on MBP-NBF1 clearly demonstrate its capacity to catalyze the hydrolysis of ATP. Significantly, the mutations K464H and K464L in the Walker A consensus motif of NBF1 markedly impair its catalytic capacity. MBP alone exhibits no ATPase activity and MBP-NBF1 fails to catalyze the release of phosphate from AMP or ADP. The Vmax of ATP hydrolysis (30 nmol/min/mg of protein) is significant and is markedly inhibited by azide and by the ATP analogs 2′-(3′)-O-(2,4,6-trinitrophenyl)-adenosine-5′-triphosphate and adenosine 5′-(β,-imido)triphosphate. As inherited mutations within NBF1 account for most cases of cystic fibrosis, results reported here are fundamental to our understanding of the molecular basis of the disease.

Cystic fibrosis is caused by mutations in the cell membrane protein called CFTR (cystic fibrosis transmembrane conductance regulator) which functions as a regulated Cl ؊ channel. Although it is known that CFTR contains two nucleotide domains, both of which exhibit the capacity to bind ATP, it has not been demonstrated directly whether one or both domains can function as an active ATPase. To address this question, we have studied the first CFTR nucleotide binding fold (NBF1) in fusion with the maltose-binding protein (MBP), which both stabilizes NBF1 and enhances its solubility. Three different ATPase assays conducted on MBP-NBF1 clearly demonstrate its capacity to catalyze the hydrolysis of ATP. Significantly, the mutations K464H and K464L in the Walker A consensus motif of NBF1 markedly impair its catalytic capacity. MBP alone exhibits no ATPase activity and MBP-NBF1 fails to catalyze the release of phosphate from AMP or ADP. The V max of ATP hydrolysis (ϳ30 nmol/min/mg of protein) is significant and is markedly inhibited by azide and by the ATP analogs 2-(3)-O-(2,4,6-trinitrophenyl)-adenosine-5triphosphate and adenosine 5-(␤,␥-imido)triphosphate. As inherited mutations within NBF1 account for most cases of cystic fibrosis, results reported here are fundamental to our understanding of the molecular basis of the disease.
Cystic fibrosis is the most common autosomal recessive disease in Caucasians affecting approximately 1 in 2000 people in the United States and Canada (1)(2)(3). The disease is caused by mutations in the CFTR 1 protein which impair its normal function as a regulated phosphorylation-dependent Cl Ϫ channel in epithelial cells (2,3). CFTR, which is comprised within a single polypeptide chain of 1480 amino acids, is predicted to fold into five distinct domains, two nucleotide binding folds (NBF1 and NBF2), a regulatory domain (R), and two transmembrane spanning regions (4). Although inherited mutations throughout the CFTR protein are known to cause cystic fibrosis, there is a high frequency of such mutations within or surrounding NBF1 and NBF2, with those within NBF1, particularly ⌬F508, being responsible for most cases of the disease (1)(2)(3)(4). Therefore, in order to understand the molecular basis of the disease, it is both necessary and fundamental to elucidate the functions of NBF1 and NBF2.
There is considerable evidence that both phosphorylation of the CFTR protein, and its interaction with ATP, independent of phosphorylation events, are required for optimal functions (5)(6)(7)(8). In the latter case, the finding that nonhydrolyzable ATP analogs like AMP-PNP fail to support CFTR function has led to the suggestion that ATP hydrolysis may be required (7,8). Thus, unlike ATP, AMP-PNP fails to open prephosphorylated Cl Ϫ channels in patch clamp experiments using Hela cells or 3T3 fibroblasts expressing CFTR (7,8). Moreover, in single channel analysis of artificial lipid planar bilayers reconstituted with human CFTR, AMP-PNP, unlike ATP, failed to support channel activity (9).
Despite the above findings, it remains unresolved whether one or both of the two nucleotide binding folds of CFTR exhibit the capacity to hydrolyze ATP. In earlier reports from this laboratory (10 -12), we have demonstrated that peptide segments of both NBF1 and NBF2 containing the Walker A consensus motif (13) can bind TNP-ATP and that this analog can be displaced by ATP. However, no ATP hydrolytic capacity for either the NBF1 or NBF2 segment could be demonstrated (10 -12). Purified preparations of the complete NBF1, and of NBF1 in fusion with the maltose-binding protein (MBP-NBF1), have been obtained also (14,15). Similar to the peptide segments, NBF1 and MBP-NBF1 bind ATP and/or TNP-ATP (14,15). However, these preparations have not been examined in detail for their capacity to hydrolyze ATP.
Studies described in this report have focused on the MBP-NBF1 fusion protein which was originally overexpressed in Escherichia coli and purified in this laboratory (15). Below, compelling evidence is provided that NBF1 within the MBP-NBF1 complex is an active ATPase.

EXPERIMENTAL PROCEDURES
Overexpression and Purification of the MBP-NBF1 Fusion Protein-Methodology is essentially as described previously (15), with the modifications that 1) the lysis buffer included also per 25 ml, 1000 units protein kinase A, and 5 mM ATP, and 2) the final protein product was stored as an ammonium sulfate pellet. The expression cassette polymerase chain reaction (16,17) was used to synthesize the NBF1 of CFTR, which included base 1429 (Phe-433) to base 1899 (Ser-589). A pMal-cR1 expression plasmid containing the cDNA for NBF1 in frame with MBP, and under control of the tac promoter, was used to transform TB1 E. coli cells. After induction with isopropyl-1-thio-␤-D-galactopyranoside, incubation, lysis, subfractionation, and precipitation with ammonium sulfate, the final pellet was suspended in column buffer (10 mM sodium phosphate, 0.5 M NaCl, and 1.0 mM EGTA, pH 7.2) and loaded onto an amylose column. Elution was carried out with 20 mM maltose in column buffer. The purified MBP-NBF1 fusion protein was stored at Ϫ80°C as a 55% ammonium sulfate pellet.
Preparation of Mutant Proteins-Mutations in the lysine residue at position 464 in the primary sequence of NBF1 within the MBP-NBF1 fusion protein were introduced by use of the Mutagene oligonucleotide * This work was supported by grants from the National Institutes of Health (NIDDK) and the Cystic Fibrosis 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.
SDS-PAGE-This was carried out by the method of Laemmli (21). Gels were stained with Coomassie dye to determine the degree of overexpression of recombinant wild type and mutant proteins and to estimate their purity.
Assay for ATP Hydrolytic Activity-Three different methods were used to monitor the ATP hydrolytic capacity of wild type and mutant MBP-NBF1 fusion proteins. In the first method, the disappearance of ATP was followed by monitoring the loss of the chemiluminescence signal in an assay system containing luciferin and luciferase. The reaction was performed at 25°C at pH 6.7 in a 1.2-ml system containing 10 mM KP i , 50 mM KCl, 1 mM MgSO 4 , 10 l of luciferin/luciferase (Bio-orbit kit, Turku, Finland), and the indicated amounts of ATP, MBP-NBF1, MBP, or mitochondrial F 0 F 1 -ATPase in membrane bound form with a specific activity of 3.0 mol/min/mg of protein and prepared as described previously (22). In the second assay method the formation of ADP coupled to the enzyme pair pyruvate kinase and lactic dehydrogenase was monitored spectrophotometrically at 25°C by following the decrease in absorbance of NADH at 340 nm. The assay system contained in a volume of 1 ml, 50 mM Tris-Cl, 5 mM MgCl 2 , 0.6 mM phosphoenolpyruvate, 0.3 mM NADH, 3.9 units of pyruvate kinase, 1.2 units of lactic dehydrogenase, and the indicated concentrations of ATP and MBP-NBF1. In the third assay, the release of inorganic phosphate was monitored at 25°C in a 0.5-ml system containing 50 mM Tris-Cl, 5 mM ATP or 5 mM AMP, 5 mM MgCl 2 , pH 7.4, and the indicated amount of MBP-NBF1. After 75 min, 3.5 ml of water was added to 0.4 ml of the reaction mixture followed by immediate addition of acid molybdate and Elon reagent as described previously (23). After 15 min the optical density at 660 nm was measured.
Determination of Protein-Protein was estimated using the Coomassie plus dye binding assay kit from Pierce.

RESULTS
All studies reported here were carried out with the first nucleotide binding fold (NBF1) of CFTR in fusion with the MBP. The MBP-NBF1 fusion protein was prepared by an earlier method developed in this laboratory (15) and modified as described under "Experimental Procedures." Fig. 1 summarizes results of experiments in which this procedure was used to obtain highly purified MBP-NBF1 for those studies described in this report. The overexpression of the fusion protein in E. coli under control of the tac promoter after induction with isopropyl-1-thio-␤-D-galactopyranoside, its elution from an amylose column, and its electrophoretic profile on SDS-PAGE are illustrated in Fig. 1, A-C, respectively. To assure maximal purity of the product only a single peak fraction was selected from the amylose column, which from SDS-PAGE is estimated to be Ͼ95% pure. Purified MBP-NBF1 has the molecular mass expected of ϳ60 kDa (42 kDa for MBP ϩ 18 kDa for NBF1) and remains stable for several months when stored as a 55% ammonium sulfate pellet at Ϫ80°C.
To determine directly whether MBP-NBF1 has the capacity to catalyze the hydrolysis of ATP, two completely different assays, one based on the disappearance of the substrate ATP and the other on the formation of the product ADP were used initially. The assay based on the disappearance of ATP, which includes the luciferin/luciferase reaction to monitor ATP levels, was employed first. This is one of the most sensitive ATPase assays known, detecting ATP concentrations in the nanomolar to micromolar range via chemiluminescence. The assay is illustrated in Fig. 2A in a control experiment with a known ATPase, the mitochondrial F 0 F 1 -ATPase. Here, it can be seen that upon addition of 3.3 M ATP to a buffered system containing only luciferin and luciferase a light flash (chemiluminesence response) is detected which peaks and levels off to a near zero decay rate. Upon addition of F 0 F 1 -ATPase, the chemiluminescence signal rapidly decreases as ATP disappears upon its conversion to ADP and P i . In Fig. 2B, it can be seen that when  this experiment is repeated with MBP-NBF1 a slower but significant rate of disappearance of ATP is observed. Moreover, this rate is ATP-dependent (Fig. 2B, inset) and remains linear at the very low concentrations of ATP employed. Finally, in Fig.  2C it can be seen that MBP alone at concentrations higher than MBP-NBF1 has no capacity to catalyze the hydrolysis of ATP, implicating NBF1 as the catalytic unit. (It should be noted that MBP has no nucleotide binding domains and neither binds nor hydrolyzes ATP (24).) Although the chemiluminescence assay clearly demonstrates that MBP-NBF1 and not MBP alone exhibits the capacity to hydrolyze ATP, the assay is limited in the amount of ATP that can be added. Thus, V max values must be extrapolated rather than determined directly. For this reason, the second assay noted above, which monitors the formation of ADP coupled to the pyruvate kinase and lactic dehydrogenase reactions, was employed. In this coupled system, the ADP formed in the ATPase reaction is monitored by the decrease in absorption of NADH when pyruvate is converted to lactate. Fig. 3A shows that when increasing amounts of MBP-NBF1 are added to the assay the rate of disappearance of NADH increases. Fig. 3B shows that this increased rate of hydrolysis is a linear function of the amount of MBP-NBF1 added assuring that the increased rate is not due to a non-protein contaminant. In data not presented, MBP alone had no capacity to catalyze ATP hydrolysis as indicated above for the luciferin-luciferase based assay. Results presented in Fig. 3C show that a typical Michaelis-Menten plot is obtained when ATPase activity is plotted versus ATP concentration resulting in K m and V max values, respectively, of 0.11 mM and ϳ30 nmol/min/mg of protein (first order rate constant ϭ 0.5 min Ϫ1 /NBF1).
Results from the two different assay procedures described  4. Evidence that the ATP hydrolytic capacity of MBP-NBF1 is not the result of a contaminating phosphatase and is a property of NBF1. A, relative capacity of MBP-NBF1 to hydrolyze the release of P i from ATP and AMP. Phosphate release was monitored colorimetrically exactly as described under "Experimental Procedures" in an assay system containing 5 mM ATP or 5 mM AMP, 5 mM MgCl 2 , and 220 g of MBP-NBF1. (In data not presented, ADP when substituted for ATP, resulted in no phosphate release.) All assays were carried out in duplicate. B, comparison of the purity of wild type and mutant MBP-NBF1 proteins (K464H and K464L). SDS-PAGE by the method of Laemmli (21) was carried out on 4 g of wild-type MBP-NBF1 and on 5 g of the indicated mutant proteins. C and D, the effect of the mutations K464H and K464L within NBF1 on the ATP hydrolytic activity of MBP-NBF1. ATP hydrolytic activity was assayed using the coupled spectrophotometric method exactly as described under "Experimental Procedures." In C 550 g of wild type and 670 g of mutant protein were included in the assay, whereas in D 270 g of wild type and 170 g of mutant protein were included. (Note: specific activity is plotted so that differences in protein concentrations are normalized.) All experiments were repeated, and similar results were obtained. above demonstrate that the highly purified MBP-NBF1 fraction exhibits the capacity to catalyze ATP hydrolysis. However, they do not rule out the possibility that a contaminating protein, either another cell ATPase or a nonspecific phosphatase, is responsible for the results obtained. For this reason two different experiments were performed. In the first, ATP hydrolysis was monitored by a third assay in which phosphate release is measured. Results presented in Fig. 4A show that in the presence of MBP-NBF1, phosphate is released from ATP as expected, but no phosphate is released from AMP, and in data not presented from ADP; phosphate release from AMP or ADP would be expected if a nonspecific phosphatase were contaminating the preparation. Even more convincing are results presented in Fig. 4, C and D, where it is seen that the mutations, K464H and K464L, in the Walker A nucleotide binding motif (GX 4 GKT) of NBF1 reduce the ATP hydrolytic capacity of purified mutant MBP-NBF1 proteins by over 80%. Taken together, the two different types of experiments described here leave little doubt that it is the NBF1 portion of MBP-NBF1 that is responsible for its ATP hydrolytic capacity.
Finally, results presented in Fig. 5 summarize those agents which were found to inhibit the ATPase activity of MBP-NBF1 in the coupled spectrophotometric assay (see "Experimental Procedures"). Here it is seen that, as in the case of the F 1 moiety of F 0 F 1 -ATPase, the ATPase activity of MBP-NBF1 is inhibited by the ATP analogs TNP-ATP (25) and AMP-PNP (26) as well as by sodium azide. However, in contrast to mitochondrial F 1 , where sulfite activates ATP hydrolysis at least 2-fold (28), the ATPase activity of MBP-NBF1 is unaffected. In results not presented here, AMP-PNP was shown to be a more effective inhibitor in the P i release assay. DISCUSSION This is the first report to provide direct evidence in vitro that the first nucleotide binding fold (NBF1) of the CFTR protein can function as an active ATPase. Assays conducted on the MBP-NBF1 fusion protein monitored either disappearance of the substrate ATP (Fig. 2B) or appearance of the products ADP (Fig. 3) and P i (Fig. 4A) leaving no doubt that the purified protein catalyzes the hydrolysis of ATP. The additional experimental results demonstrating that MBP alone has no catalytic capacity (Fig. 2C) and that mutations (K464H and K464L) within the Walker nucleotide binding motif GX 4 GKT markedly inhibit ATPase activity (Fig. 4, C and D) localize the catalytic site to NBF1. These results do not exclude the possibility that within intact CFTR the NBF1 domain may require stabilizing interactions with another domain. Thus, it should be noted that in the F 1 moiety of the F 0 F 1 -ATPase, the ␤ subunit, although containing all residues essential for catalysis, is itself not catalytic (29,30). Rather, interaction with the noncatalytic ␣ subunit is required to elicit ATP hydrolysis (31).
Relative to other proteins that catalyze the hydrolysis of ATP, the V max rate of MBP-NBF1 of ϳ30 nmol/min/mg is slightly higher than that observed for chaperones (32) but considerably lower than that observed for the CFTR related MDR-protein (multidrug resistance protein) (33,34). The much lower rate characteristic of NBF1 of CFTR may reflect a role in using ATP hydrolysis to only open Cl Ϫ channels, as opposed to transporting a drug, as in the case of MDR. However, it remains possible that the activity observed here for CFTR NBF1 may be a minimal value and that in vivo hitherto unknown substrates are transported by CFTR, in which case a higher ATP hydrolytic rate may be required. Significantly, it has been reported recently that in addition to Cl Ϫ , CFTR can also translocate ATP (27,35). FIG. 5. The effect of known ATPase modulators on the ATPase activity of MBP-NBF1. ATPase activity was assayed by coupling the release of ADP to the pyruvate kinase, lactic dehydrogenase reactions, and measuring the decrease in absorbance at 340 nm exactly as described under "Experimental Procedures." In all cases controls were run using ADP instead of MBP-NBF1 and ATP to assure that the modulators tested had no effect on the coupled enzymes. The indicated concentrations of modulators are shown. MBP-NBF1 (1.3 mg) was present in all assays, and all assays were carried out in duplicate. Experiments were repeated at least once with all modulators.