Recombinant N-terminal nucleotide-binding domain from mouse P-glycoprotein. Overexpression, purification, and role of cysteine 430.

Varying length cDNAs encoding the N-terminal nucleotide-binding domain (NBD1) from mouse mdr1 P-glycoprotein were prepared on the basis of structure predictions. Corresponding recombinant proteins were overexpressed in Escherichia coli, and the shortest one containing amino acids 395-581 exhibited the highest solubility. Insertion of an N-terminal hexahistidine tag allowed domain purification by nickel-chelate affinity chromatography. NBD1 efficiently interacted with nucleotides. Fluorescence methods showed that ATP bound at millimolar concentrations and its 2′,3′-O-(2,4,6-trinitrophenyl) derivative at micromolar concentrations, while the 2′(3′)-N-methylanthraniloyl derivative had intermediate affinity. Photoaffinity labeling was achieved upon irradiation with 8-azido-ATP. The domain exhibited ATPase activity with a K for MgATP in the millimolar range, and ATP hydrolysis was competitively inhibited by micromolar 2′,3′-O-(2,4,6-trinitrophenyl)-ATP. NBD1 contained a single cysteine residue, at position 430, that was derivatized with radiolabeled N-ethylmaleimide. Cysteine modification increased 6-fold the K for 2′(3′)-N-methylanthraniloyl-ATP and prevented 8-azido-ATP photolabeling. ATPase activity was inhibited with a 5-fold increase in the K for MgATP. The results suggest that chemical modification of Cys-430 is involved in the N-ethylmaleimide inhibition of whole P-glycoprotein by altering substrate interaction.

Multidrug resistance of tumor cells is often associated with overexpression of P-glycoprotein, a membrane transporter that extrudes chemotherapeutic drugs using ATP hydrolysis as energy source (1,2). The protein is encoded by the mdr gene family comprising two members in man, mdr1 and mdr2, or three in mouse, mdr1 (or mdr1b), mdr2, and mdr3 (or mdr1a). Only mdr1, and to a lower extent mdr3, was found to convey cellular multidrug resistance; mdr3 appears to be involved in detoxification/protection processes and mdr2 in phospholipid translocation (3). The function of mdr1 P-glycoprotein in normal tissues is still questioned although its relative abundance in mouse pregnant uterus and adrenal glands (4) favors a role in steroid hormone secretion (5).
Structural analysis of the P-glycoprotein sequence, composed of 1276 amino acids in mouse (6), predicts two homologous halves, each containing up to six putative membrane-spanning ␣-helices and one cytoplasmically sided nucleotide-binding domain with characteristic Walker motifs A and B (7). P-glycoprotein structural organization is typical of the ATP-binding cassette (ABC) 1 superfamily including yeast (8) and protozoan parasite (9) drug transporters, and a series of different members from eukaryotic proteins, like the cystic fibrosis gene product CFTR, to bacterial transporters (10). The ATPase activity and related drug transport of P-glycoprotein require both functional nucleotide-binding sites (11,12) and are sensitive to the cysteine-specific modifier N-ethylmaleimide (NEM) (13)(14)(15)(16)(17). The lack of structural data about P-glycoprotein is due to its low abundance, difficult purification and membrane character, and to the lack of a highly overexpressing system (18). A recent approach to circumvent such problems was to overexpress in bacteria recombinant domains predicted to be soluble, in fusion with either the glutathione S-transferase or the maltose-binding protein to allow their purification by affinity chromatography: this was achieved with the C-terminal nucleotidebinding domain (NBD2) from P-glycoprotein (19,20), or with HlyB or CFTR domains (21,22). However, the presence of a relatively high-size fusion protein might be undesirable when studying protein/protein interactions, and its release by specific proteolytic cleavage was only partial or led to unstable nucleotide-binding domains. An alternative was to use a hexahistidine tag for fusion, in order to increase protein solubility and allow its purification by nickel-chelate chromatography.
The aim of the present work was to design the N-terminal nucleotide-binding domain (NBD1) from mouse P-glycoprotein encoded by the mdr1 gene using structure predictions and to overexpress it in E. coli as a hexahistidine-tagged recombinant protein to study nucleotide interactions and the possible role of its single cysteine residue. The results indicate that NBD1 of varying length was highly overexpressed, but only the shortest one exhibited sufficient solubility to be purified as milligram amounts of protein. The domain efficiently bound ATP or ana-logues and exhibited ATPase activity; chemical modification of the single cysteine residue by NEM altered nucleotide interaction and substrate hydrolysis.
Construction of Expression Vectors-Extraction of mRNAs from mouse adrenal glands and synthesis by reverse transcriptase-PCR of the cDNA coding for the longest domain, NBD1-643, were performed as described for NBD2 (19). Here, the two primers specific to mdr1 and corresponding to NBD1 N-terminal Asn-395 and C-terminal Ser-643, were respectively: 5Ј-TATGGATCCAATGTTCACTTCAACTACCC-3Ј and 5Ј-TATAAGCTTCTAGGATCCATAAGCATTATTTCCTGG-3Ј.
The primers allowed the introduction of BamHI and HindIII restriction sites, and the amplified cDNA of 747 base pairs was digested by endonucleases and ligated into the corresponding sites of linearized pQE-30 plasmid (Qiagen). E. coli JM105 cells (supE endA sbcB15 hsdR4 rpsL thi ⌬(lac-proAB) FЈ [traD36 proAB ϩ lacI q lacZ⌬M15]) were transformed with the ligation product and grown on agar plates supplemented with ampicillin (50 g/ml).
The recombinant plasmid was used as a template to produce by PCR amplification the cDNAs to be inserted in the pQE-30 plasmid and coding for NBD1-613 and NBD1-581, by using the same Asn-395related primer together with the following one, 5Ј-TATAAGCTTCT-ACTCATCATGATTTCCTTGCTCCACAA-3Ј or 5Ј-TATAAGCTTCTAG-GTGGTCCGGCCTTCTCTAGCCTTAT-3Ј, corresponding to C-terminal Glu-613 or Thr-581, respectively. In all cases, the PCR conditions were the following: (i) a first denaturation at 92°C for 4 min, (ii) 35 cycles each consisting in denaturation at 92°C for 30 s and elongation at 72°C for 60 s plus a 1-s increase per cycle, and (iii) a final elongation at 72°C for 10 min.
The three recombinant plasmids were restriction-mapped, and the dideoxy sequencing method confirmed the expected sequences for NBD1 cDNAs by reference to the published cDNA sequence of mdr1 P-glycoprotein (6).
Overexpression and Protein Purification-E. coli cells harboring the appropriate recombinant plasmid were grown at 37°C in LB medium (1% (w/v) bacto-tryptone, 0.5% (w/v) yeast extract, 1% (w/v) NaCl) at pH 7.5 containing 50 g of ampicillin/ml, until the absorbance at 600 nm reached 0.7 unit. Expression of the recombinant domains was induced with 0.2 mM IPTG for 1 h at 30°C. Cells were harvested by centrifugation at 5,500 ϫ g for 10 min at 4°C, and resuspended in 10 mM Tris-HCl, pH 8.0, containing 1 mM EDTA, 6 mM MgCl 2 , 1% (v/v) Triton X-100, 1 mM phenylmethylsulfonyl fluoride, and 240 units Benzonase/ ml. The cells were lysed using a SLM-Aminco French Pressure Cell Press at 1,100 psi with a 40-ml capacity cell, and centrifuged at 30,000 ϫ g for 30 min.
The supernatant was applied to a Ni 2ϩ -nitrilotriacetic acid column (24) equilibrated in 50 mM potassium phosphate, 150 mM Na 2 SO 4 , 1% (v/v) Triton X-100, 10% (w/v) glycerol, 40 mM imidazole, at pH 8.5. The column was first extensively washed with the same buffer containing 0.7 M NaCl, and then in the absence of Triton X-100 and NaCl but in the presence of 0.05% (w/v) HECAMEG. The retained proteins were then eluted with 200 mM imidazole, in the presence of 0.01% HECAMEG, and analyzed by SDS-polyacrylamide gel electrophoresis. The fractions were pooled and dialyzed against 50 mM potassium phosphate, 150 mM Na 2 SO 4 , 20% glycerol, 0.01% HECAMEG (dialysis buffer) at pH 8.5. The dialysate was centrifuged to discard possible traces of precipitated material; the supernatant (0.2-0.3 mg of protein/ml) was aliquoted and kept frozen in liquid nitrogen.
Protein fractions were analyzed on 12% SDS-polyacrylamide gels as described by Laemmli (25). Protein concentration was routinely determined by the method of Bradford (26) with the Coomassie Blue Plus Protein Assay Reagent kit from Pierce.
ATPase Assay-The ATPase activity was measured by a spectrophotometric method of ADP release using an ATP-regenerating system (27). The medium (0.8 ml final volume) was composed of 30 mM potassium phosphate, 40 mM Tris-HCl, 30 mM KCl, 150 mM Na 2 SO 4 , 20% glycerol, 1 mM MgCl 2 , at pH 8.3, containing 4 mM phosphoenolpyruvate, 0.3 mM NADH, 40 g of pyruvate kinase, and 20 g of lactate dehydrogenase. Increasing equal amounts of ATP and MgCl 2 were added up to 5 mM, and the activity was monitored at 30°C with NBD1-581 aliquots (50 -100 g of protein), by recording NADH disappearance at 340 nm for 5-10 min.
Fluorescence-Experiments were performed at 25.0 Ϯ 0.1°C using a SLM-Aminco 8000C spectrofluorometer with spectral band widths of 2 nm and 4 nm, respectively, for excitation and emission. The measurements were corrected for wavelength dependence on the exciting-light intensity by using rhodamine B in the reference channel. All spectra were corrected for buffer fluorescence and for dilution (less than 5%).
Fluorescence measurements were performed by diluting protein solutions (3-5 M final concentration) in 1 ml of dialysis buffer, at pH 8.5, in the presence of increasing concentrations of fluorescent ATP analogues. When the extrinsic fluorescence of TNP-ATP was studied, the excitation was performed at 408 nm and emission was scanned in the range 540 -570 nm. For MANT-ATP, the excitation wavelength was 350 nm and emission was scanned from 400 to 480 nm. ATP analogue binding was determined from the increase in fluorescence at 545 nm or 432 nm, respectively, for TNP-ATP or MANT-ATP, in the presence as compared to the absence of NBD1-581. Curve-fitting of the concentrationdependent analogue binding was performed with the Grafit program (Erithacus software) as detailed previously (28,29). For ATP-dependent chase of bound TNP-ATP, controls were conducted with the same nucleotide concentrations but in the absence of protein, and curve-fitting was analyzed according to Stinson and Holbrook (30).
Tyrosine-intrinsic fluorescence of 1 M NBD1-581 (5 M tyrosine residues) was measured upon excitation at 275-280 nm by scanning emission in the range 285-360 nm. The binding of MANT-ATP monitored by quenching of emission fluorescence was studied in the presence of increasing analogue concentrations. Correction for nucleotide inner filter effect was determined under the same conditions using N-acetyltyrosinamide. Curve-fitting of nucleotide analogue binding was performed with the Grafit program as described previously for quenching of tryptophan-intrinsic fluorescence (19,28,29).
Photolabeling with 8-Azido-ATP-The NBD1-581 domain (15 g of protein) in 200 l of dialysis buffer containing 5 mM MgCl 2 was placed inside a 48-well tissue culture cluster and first incubated for 10 min in the dark, in ice under continuous stirring, with 8.7 M 8-azido-[␥-32 P]-ATP (200 nCi/nmol). Photoirradiation was then performed at 254 nm for 10 min with a Spectraline UV lamp model EF-280C/F (Spectronics) placed at a 4-cm distance. The samples were transferred into Eppendorf tubes and precipitated by 8% trichloroacetic acid (final concentration). The pellets were resuspended in 0.2 N NaOH and mixed with SDSloading buffer. Proteins were submitted to SDS-polyacrylamide gel electrophoresis, and autoradiography was performed at Ϫ70°C using BioMax MR films (Kodak) with an intensifying screen. Alternatively, the domain was submitted, before 8-azido-[␥-32 P]ATP addition, to pretreatment by 1 mM NEM for 60 min at room temperature (followed by addition of 10 mM dithiothreitol) or incubated with either ATP or TNP-ATP for 10 min on ice.
Chemical Modification by NEM-NBD1-581 (15 g of protein) was incubated in 150 l of dialysis buffer in the presence of 0.1 to 2 mM [ 14 C]NEM (5, 220 dpm/nmol) for 60 min at 30°C. Excess dithiothreitol (10 mM) was added to stop the reaction, and the protein was precipitated by 10% trichloroacetic acid. After centrifugation, the pellet was washed 2-fold and finally solubilized in 0.2 N NaOH. Aliquots were withdrawn for protein assay by the method of Lowry et al. (31) and for liquid scintillation counting in 5 ml of the Ultima Gold XR mixture from Packard. Corrections for unbound NEM were performed on control assays conducted in the absence of protein.

Design of the N-terminal Nucleotide-binding Domain from
Mouse P-glycoprotein-From P-glycoprotein models based on cDNA sequence and predictions from hydrophobicity profiles (1), both N-terminal (NBD1) and C-terminal (NBD2) nucleotide-binding domains are assumed to be extrinsic with limited membrane interactions (Fig. 1, top scheme). They both contain the Walker motifs A and B of ATP site, the ABC-transporter signature S (32) and a C219 monoclonal antibody epitope (33). NBD1 is adjacent to the phosphorylatable linker region. The present strategy to obtain soluble NBD1 (bottom scheme) was to overexpress in bacteria cDNAs coding for recombinant domains whose limits are determined from structure predictions and which contain a hexahistidine tag to allow purification. Recombinant NBD1 contains a single cysteine residue, located at position 430 inside the Walker motif A. Fig. 2 shows the analysis of the NBD1-containing 350 -708 amino acid sequence from mouse mdr1 P-glycoprotein by using the ANTHEPROT program (34). Four different predictive methods were used concerning hydrophobicity (35), accessibil-ity (36), antigenicity (37), and secondary structures (38). Domain design answered the following criteria: (i) the minimal size had to contain the ABC-transporter signature S and both Walker motifs A and B whose critical roles were shown in other ATP-binding proteins of known three-dimensional structure like adenylate kinase, RecA, or mitochondrial F 1 -ATPase; (ii) the maximal size was limited by membrane proximity on the N-terminal side and by the phosphorylatable linker region starting around position 660 (39) on the C-terminal side; (iii) the limits had to be located inside hydrophilic, accessible, antigenic, and aperiodic regions. Due to the reverse transcriptase-PCR method used for cDNA cloning, an additional constraint was to find mdr1 product sequences sufficiently different from the mdr2 and mdr3 ones, such sequences being rare due to the high similarity in primary structure of the three proteins (40). To fulfill all these conditions, Asn-395 and Ser-643 were chosen, respectively, as N-terminal and C-terminal ends for NBD1 (named NBD1-643); the corresponding cDNA was prepared by reverse transcriptase-PCR from adrenal cells which contain abundant mdr1 mRNA, limited amounts of mdr3 mRNA, and almost no mdr2 mRNA (4). Two other C-terminal limits, Glu-613 and Thr-581 giving the respective shorter domains NBD1-613 and NBD1-581, were determined. Whereas the Thr-581 environment appears hydrophobic inside whole P-glycoprotein, due to following apolar residues, it acquires a hydrophilic value when the residue occupies the C-terminal position. The corresponding cDNAs were prepared by PCR amplification using the cDNA coding for NBD1-643 as a template, and a N-terminal hexahistidine tag was linked to all recombinant domains (cf. Fig. 1).
Overexpression and Purification of NBD1- Fig. 3 shows that the three predicted NBD1 domains were highly overexpressed upon IPTG induction and constituted one main component among total bacterial proteins and lysate. Whereas the NBD1-643 and NBD1-613 domains were almost exclusively recovered as inclusion bodies in the pellet of centrifugation, an apprecia- ble amount of the shortest one, NBD1-581, was obtained in a soluble form in the supernatant. The apparent molecular masses of all domains appeared 1-2.5 kDa higher than the theoretical values, 29.5 kDa, 25.4 kDa, and 22 kDa, respectively, for NBD1-643, NBD1-613, and NBD1-581, as this was also observed for other hexahistidine-tagged proteins (41).
A preparative purification of NBD1-581 is illustrated in Fig.  4. Similarly as in Fig. 3C, the overexpressed domain appeared abundant in both the total protein fraction (lane 2) and the soluble one (lane 3). Affinity chromatography using a nickelchelate resin was particularly efficient to discard bacterial proteins as a pass-through fraction (lane 4) and to selectively bind and purify the NBD1-581 domain which was then eluted with 200 mM imidazole (lane 5). After dialysis in 50 mM potassium phosphate, 150 mM Na 2 SO 4 , 20% glycerol, 0.01% HECAMEG, at pH 8.5, about 4 mg of protein of approximately 95% pure NBD1-581, at a 10 -15 M concentration, could be obtained from a 3-liter culture. Purified NBD1-581 reacted strongly with C219 monoclonal antibody, which is specific for the VQXALD sequence (42), when assayed by immunoblotting (not shown here).
Nucleotide Binding and Cysteine Chemical Modification-The binding of TNP-ATP, a fluorescent ATP analogue, to NBD1-581 was monitored by changes of extrinsic fluorescence upon excitation at 408 nm: the emission spectrum was enhanced and blue-shifted from 558 to 545 nm. The enhancement at 545 nm depended on the probe concentration in a saturable manner with a K d value for TNP-ATP of 2.2 Ϯ 0.6 M. Bound TNP-ATP was efficiently chased by ATP addition (K d ATP ϭ 6.3 Ϯ 0.9 mM); reciprocally, TNP-ATP binding was specifically prevented by preincubation with ATP (data not shown).
The binding of MANT-ATP, another fluorescent derivative, was also monitored by the enhancement of extrinsic fluorescence as illustrated in Fig. 5A. When MANT-ATP bound to NBD1-581, its fluorescence intensity increased while the wavelength of maximal emission was blue-shifted from 444 to 432 nm. At fixed protein concentrations, the increase at 432 nm was dependent on MANT-ATP concentration in a saturable manner and allowed the determination of a K d value for MANT-ATP of 25.2 Ϯ 4 M (Fig. 5B). The binding of MANT-ATP could also be monitored by changes in the intrinsic fluorescence of NBD1-581 which contained five tyrosine residues. Fig. 5C illustrates the tyrosine-characteristic fluorescence of 1 M NBD1-581: the excitation spectrum exhibited a maximum at 275 nm, with a shoulder around 280 nm, whereas the emission was maximal at 299 nm. Addition of MANT-ATP produced a significant quenching of the domain fluorescence emission. The quenching was dependent on the analogue concentration (Fig. 6, empty symbols) giving a K d value for MANT-ATP of 25 Ϯ 8 M. Incubation of NBD1-581 for 1 h with increasing concentrations of radioactive NEM led to increasing incorporation of the reagent: NBD1-581 which contains a single cysteine residue, at position 430 inside the Walker motif A, maximally incorporated 1 mol of NEM/mol when using a NEM concentration in the range 0.8 -2 mM. The inset to Fig. 6 indicates that NEM altered MANT-ATP interaction by increasing its K d value up to 6-fold. When assayed at a high protein concentration in the presence of millimolar MgATP, NBD1-581 exhibited a low but significant ATPase activity (Fig. 8A); linear double-reciprocal plots allowed graphical estimation of a V max value of 25 nmol of ATP hydrolyzed/min ϫ mg of protein, corresponding to a turnover number of 0.5 min Ϫ1 , and a K m value for MgATP of 2.1 mM (B, empty circles). Addition to the assay medium of TNP-ATP at 5 M (empty squares) or 10 M (closed squares) produced a competitive inhibition of ATP hydrolysis. Chemical modification by NEM inhibited ATPase activity (A, closed circles): the inhibition was more pronounced at low substrate concentration (74% at 0.12 mM MgATP) as compared to high concentration (45% at 4.8 mM MgATP), and double-reciprocal plots indicated that the K m MgATP was increased about 5-fold (not shown here). DISCUSSION The original aspects of this paper concern the first preparation of a soluble N-terminal nucleotide-binding domain from P-glycoprotein, the measurement of its interaction parameters with substrate ATP and analogues, and the characterization of the cysteine residue located inside the Walker motif A.
Overexpression of Soluble N-terminal Nucleotide-binding Domain from P-glycoprotein-The present preparation of a soluble N-terminal nucleotide-binding domain from P-glycoprotein constitutes the first one ever reported. Up to now, NBD1 was obtained as a part of the membrane-bound N-terminal half of P-glycoprotein (12,15). Only the C-terminal domain, NBD2, could be obtained in a soluble form, in fusion with either the glutathione S-transferase (19) or the maltose-binding protein (20). An additional problem for NBD1 was to define an appropriate C-terminal end which is shown here to be critical for solubility. The ANTHEPROT program allowed us to predict several possible limits; one of them, Thr-581, indeed gave a domain of sufficient solubility. Another important parameter for solubility was the presence of the N-terminal hexahistidine tag since its replacement by the glutathione S-transferase, by using the pGEX-KT construct (43) instead of the pQE-30 plasmid, gave a considerably less soluble NBD1-581 domain. 2 Also in the case of NBD1 from CFTR, the absence of any fusion at the N-terminal end provided an insoluble and unstable domain (44) whereas fusion with the maltose-binding protein considerably increased its solubility (22,45). Finally, the presence of the uncharged detergent HECAMEG, at a minimal 0.01% concentration, during purification also improved the P-glycoprotein NBD1 solubility. This correlates to the positive effect of Tween 20 previously reported on the solubility of NBD2 after thrombin cleavage (19), but HECAMEG which exhibits a negligible ultraviolet absorbance does not interfere with subsequent fluorescence measurements. These beneficial detergent effects suggest that some portions of NBD1-581 might hydro-2 G. Dayan and A. Di Pietro, unpublished data. phobically interact with membrane components. The fact that, among hexahistidine-tagged N-terminal domains, NBD1-643 and NBD1-613 were much less soluble than NBD1-581 suggests that at least part of the amino acid sequence 582-643 might interact with, or be part of, another domain of P-glycoprotein such as the adjacent linker region (39); it might alternatively interact with a membrane component since mutation of Thr-578 in the mdr3 product, corresponding to Thr-581 in the mdr1 one, was recently reported to alter signal transduction (46).
Interactions with Nucleotides-This work describes the first direct interaction of ATP or analogues with P-glycoprotein NBD1. A very high affinity, 3 orders of magnitude higher than ATP, was observed for TNP-ATP by direct binding experiments monitoring the increase of extrinsic fluorescence, which gave a K d value in the micromolar range similar to that observed for NBD2 fused to the glutathione S-transferase (19) or to CFTR NBD1 fused to the maltose-binding protein (22). Micromolar concentrations of TNP-ATP also produced a competitive inhibition of ATPase activity. In the case of MANT-ATP binding, a 10-fold lower affinity, but still 2 orders of magnitude higher than ATP, was obtained by both an increase of extrinsic fluorescence and quenching of the domain tyrosine-intrinsic fluorescence. A large increase in affinity for the MANT analogues as compared to unmodified nucleotides was also observed for other proteins (27-29, 47, 48). Among the five tyrosine residues present in NBD1-581, Tyr-464 is equivalent to Trp-1106 of NBD2 whose fluorescence emission was shown to be quenched by fluorescent nucleotide binding (19). It seems therefore probable that the quenching of NBD1-581 tyrosine-intrinsic fluorescence at least partly concerns Tyr-464. A quenching of tyrosine-intrinsic fluorescence was also observed in RecA upon interaction with DNA (49). The fact that NBD1-581 covalently bound 8-azido-[␥-32 P]ATP at a 8.7 M concentration in the presence of magnesium ions is consistent with a very low turnover number: assuming that 8-azido-ATP is hydrolyzed similarly to ATP, as observed for whole P-glycoprotein (50), a value of 0.002 min Ϫ1 could be estimated meaning that a single turnover would require more than 8 h at such a low concentration.
Previous evidences for nucleotide interaction were more indirect, based on site-directed mutagenesis altering overall ATP hydrolysis and related drug transport (11,51), on the ATPase activity of membrane-bound N-terminal half (12,15) and on chemical modifications of whole P-glycoprotein (50). Using the NBD1-581 domain, we find here an apparent K d value for ATP in the absence of magnesium ions, to avoid hydrolysis, about 3-fold higher than the K m MgATP for ATPase activity. The latter value, 2.1 mM, is not very different from those obtained with membrane-bound or purified whole P-glycoprotein (14 -17, 52, 53) but contrasts with the unexpectedly high value, 20 mM, reported for NBD2 fused to the maltose-binding protein (20). The ability of purified NBD1 to hydrolyze ATP appears to be an intrinsic activity of the domain since: (i) the maximal rate, 25 nmol of ATP hydolyzed/min ϫ mg of protein, is comparable to that reported for recently characterized NBD1 from CFTR (45) and for NBD2 fused to the maltose-binding protein (20); (ii) the activity is competitively inhibited by TNP-ATP, at micromolar concentrations related to the analogue binding as determined by fluorescence; (iii) it is inhibited by NEM, as whole P-glycoprotein (13,16,17,53) and its N-terminal half (15), which increases the K m MgATP similarly as it increases the K d MANT-ATP. The maximal rate is, however, much lower as compared to domains from the bacterial ABC transporters HlyB (21) and MalK (54). It is also much lower than whole purified P-glycoprotein (17,55,56) which suggests that full activity would require critical interactions with NBD2 and/or transmembrane domains, and possibly with the phosphorylatable linker region as proposed for the CFTR regulatory domain (57). The fact that the same activity was obtained upon renaturation of NBD1-581 inclusion bodies favors that the domain overproduced and purified as a soluble protein retained a fully native conformation.
Reactivity and Role of Cysteine 430 -The single cysteine residue present in purified NBD1-581 is shown here to be reactive to NEM and to be specifically and completely derivatized by the reagent. A number of arguments might favor a critical role for Cys-430: (i) its chemical modification is shown here to decrease the affinity for ATP and analogues and thereby to inhibit MgATP hydrolysis; (ii) a cysteine residue is conserved in as many as 127 Walker motif A sequences from ATP/GTP-binding proteins among which 66 belong to the ABC transporter superfamily, when the Swiss-Prot data base is scanned with the GX 2 GCGK(S/T) sequence; (iii) the mutation of the immediately following Gly and Lys residues in human P-glycoprotein NBD1 or NBD2 abolishes the ATPase activity (51). However, the absence of any cysteine in other ABC transporters and many ATP/GTP-binding proteins indicates that the residue seems not directly involved in nucleotide binding and/or hydrolysis; indeed, mutant cells where all P-glycoprotein cysteines were substituted were still multidrug-resistant (58). It is therefore probable that introduction of the bulky group of NEM produces an important steric hindrance which is responsible for inhibition. The NEM concentrations required to inhibit the domain activity are higher than those reported for whole P-glycoprotein ranging from 5 to 200 M (14, 17, 55), but are comparable to those inhibiting the N-terminal half of the transporter (15).
In Chinese hamster membrane P-glycoprotein, two cysteine residues, among a total amount of seven, were derivatized by radioactive NEM when complete inhibition of the drug-stimulated ATPase activity was achieved, the radioactivity being equally distributed in each half of the molecule (50). The results reported here with purified mouse NBD1 strongly support that Cys-430 inside the Walker motif A is the essential residue derivatized by NEM in the N-terminal half of whole P-glycoprotein. This finding corroborates the very recent report showing that reintroduction of a single cysteine residue in the Walker motif A of the N-terminal half of a Cys-less P-glycoprotein mutant restores the sensitivity of the drug-stimulated ATPase to NEM inhibition (58). Accordingly, the ATPase activity of purified HlyB nucleotide-binding domain, which does not contain any cysteine in its Walker motif A, is not inhibited by NEM (21). In addition, we show here that the inhibition of the domain-intrinsic ATPase activity by NEM, in the absence of drugs, is due to alteration of substrate ATP binding.
Work is in progress concerning NBD1 structural and biochemical properties and its in vitro interactions with other recombinant domains of P-glycoprotein.