The nucleotide-binding domains of P-glycoprotein Functional symmetry in the isolated domain demonstrated by N -ethylmaleimide labelling Georgina Berridge 1 , Jennifer A. Walker 1 , Richard Callaghan 1 and Ian D. Kerr 1,2 1 Nuffield Department of Clinical Laboratory Sciences, University of Oxford, John Radcliffe Hospital, Oxford, UK; 2 School of Biomedical Sciences, University of Nottingham, Queen’s Medical Centre, Nottingham, UK The two nucleotide-binding domains (NBDs) of a number of ATP-binding cassette (ABC) transporters have been shown to be functionally dissimilar, playing different roles in the transport process. A high degree of co-operativity has been determined for the NBDs of the human multidrug trans- porter, P-glycoprotein. However, the issue of functional symmetry in P-glycoprotein remains contentious. To address this, the NBDs of P-glycoprotein were expressed and purified to 95% homogeneity, as fusions to maltose-binding protein. The NBDs were engineered to contain a single cysteine residue in the Walker-A homology motif. Reactivity of this cysteine residue was demonstrated by specific, time- dependent, covalent labelling with N-ethylmaleimide. No differences in the rates of labelling of the two NBDs were observed. The relative affinity of binding to each NBD was determined for a number of nucleotides by measuring their ability to effect a reduction in N-ethylmaleimide labelling. In general, nucleotides bound identically to the two NBDs, suggesting that there is little asymmetry in the initial step of the transport cycle, namely the recognition and binding of nucleotide. Any observed functional asymmetry in the intact transporter presumably reflects different rates of hydrolysis at the two NBDs or interdomain communications. Keywords: ABC transporter; cysteine; functional symmetry; maleimide; Walker-A. ATP-binding cassette (ABC) transporters are multidomain membrane proteins, responsible for the controlled efflux and influx of substances (allocrites) across cellular mem- branes [1]. They are minimally composed of four domains, with two transmembrane domains (TMDs) responsible for allocrite binding and transport and two nucleotide-binding domains (NBDs) responsible for coupling the energy of ATP hydrolysis to conformational changes in the TMDs [2,3]. A detailed understanding of ABC transporter-medi- ated allocrite flux requires the delineation of the interactions between the four domains. Studies aimed at elucidating aspects of the transport cycle of P-glycoprotein have demonstrated that both NBDs are capable of ATP hydrolysis [4], that inhibition of hydrolysis at one NBD effectively abrogates hydrolysis at the other [5], andthathydrolysisatthetwoNBDsmayoccurinan alternative fashion [6]. However, whether the two NBDs have a functionally identical role in the transport cycle, or if they are functionally nonequivalent remains a contentious issue. The reaction pathway proposed for P-glycoprotein involves ATP binding, hydrolysis, release of phosphate, and release of ADP. As both ATP hydrolysis and phosphate release appear to be rapid events [4], the rate-limiting step in this scheme is proposed to be either ATP association or ADP dissociation. Asymmetry in either of these events would be a critical component of overall functional asymmetry. Approaches to addressing this issue in P-glycoprotein (for example, determining the effects on ATPase activity subsequent to mutagenesis of equivalent residues in the N-terminal and C-terminal NBD) provide data both in favour [7–10] and against [11–13] functional asymmetry. The reasons for the discrepancies are unclear. This is in contrast with many other ABC transporters, for which there is evidence that the two NBDs, although highly similar in sequence, may adopt different functional roles in the transport cycle. Pertinent data have been presented for the multidrug resistance-associated protein (MRP1 [14,15]), the transporter associated with antigen processing (TAP [16,17]), the yeast a-factor transporter (Ste6 [18]), and the cystic fibrosis transmembrane conductance regulator [19,20]. The purpose of this study is to examine the NBDs of human P-glycoprotein in an in vitro system to determine if asymmetry of NBDs can be detected in studies of the isolated domain. We investigate asymmetry at the initial step in the catalytic cycle, namely nucleotide binding. Numerous investigations have demonstrated that nucleotide binding is a key event in the transport cycle. For example, changes in immunoreactivity [21], drug affinity [22], pro- tease sensitivity [23], and tertiary structure [10,24] are all associated with nucleotide binding. Furthermore, binding of nucleotide and the associated conformational change have Correspondence to I. Kerr, School of Biomedical Sciences, University of Nottingham, Queen’s Medical Centre, Nottingham NG7 2UH, UK. Fax: + 44 115 970 9969, Tel.: + 44 115 875 4682, E-mail: ian.kerr@nottingham.ac.uk Abbreviations: ABC, ATP-binding cassette; p[NH]ppA, adenosine 5¢-[b,c-imido]-triphosphate; MBP, maltose-binding protein; MIANS, 2-(4¢-maleimidylanilino)naphthalene-6-sulfonic acid; NBD, nucleotide-binding domain; TAP, transporter associated with antigen processing; TMD, transmembrane domain. (Received 30 October 2002, revised 20 January 2003, accepted 10 February 2003) Eur. J. Biochem. 270, 1483–1492 (2003) Ó FEBS 2003 doi:10.1046/j.1432-1033.2003.03514.x been demonstrated to occur within isolated NBDs, suggest- ing that such domains are a useful model system for analysing nucleotide binding in the intact transporter. Structural investigations of bacterial NBDs have revealed conformational differences in nucleotide-free and nucleo- tide-bound forms [25,26], which can be mirrored by molecular modelling studies (J. D. Campbell, I. D. Kerr & M. S. P. Sansom unpublished results). These studies are in agreement with the hypothesis that nucleotide can bind to a single NBD, as a precursor to NBD dimerization [27,28]. In this study we utilized the ability of thiol-reactive reagents to label endogenous cysteines located within the ATP-binding pocket of the NBDs. Previous data have demonstrated that the Walker-A cysteine residues are both accessible to thiol-specific probes, and that derivitization of these residues prevents ATP hydrolysis [5,8,29]. Recent structural determinations of related ABC transporter NBDs (for example, HisP [30]) confirm that the equivalent residue exposes its side chain to the nucleotide-binding pocket [3]. We describe the engineering, expression and purification of single-cysteine-containing proteins comprising either the C-terminal NBD or the N-terminal NBD as fusions to Escherichia coli maltose-binding protein (MBP). Having demonstrated accessibility of the unique Walker-A cysteine residue in these fusion proteins, we investigated if nucleotide preincubation could prevent derivitization by maleimide reagents. Our results show that the N-terminal and C-terminal NBDs are functionally similar in their ability to bind ATP and ADP, but show some differences in their ability to bind other nucleotides, including adenosine 5¢-[b,c-imido]-triphosphate (p[NH]ppA). These differences may be attributable to the small but significant sequence differences between the N-terminal and C-terminal NBDs, a hypothesis that we have investigated further by molecular modelling. The relevance of our data to the understanding of P-glycoprotein functional asymmetry is discussed. Experimental Procedures Reagents and chemicals N-Ethylmaleimide was obtained from Sigma (Poole, Dorset, UK). N-[ 3 H]Ethylmaleimide (specific activity 40–60 CiÆmmol )1 ) was from NEN Biochemicals (Zaventem, Belgium). Restriction enzymes were from New England Biolabs (Hitchin, Herts, UK), and PCR primers were from Sigma-Genosys (Cambridge, UK). The following nucleo- tides (Sigma unless stated) were used in N-[ 3 H]ethylmale- imide labelling and ATPase assays: disodium ATP, disodium 2-deoxy-ATP, sodium ADP, tetralithium p[NH]ppA (Calbiochem, Nottingham, UK), disodium CTP, lithium GTP, sodium dTTP, trisodium ITP. 2-(4¢- Maleimidylanilino)naphthalene-6-sulfonic acid (MIANS) was obtained from Molecular Probes (Eugene, OR, USA). All other reagents were of analytical grade or better. Generation of fusion proteins NBDs of human P-glycoprotein were fused to the C-terminus of E. coli MBP in the vector pMal-C2x (New England Biolabs). The N-terminal NBD was amplified by PCR using DNA encoding wild-type human P-glycoprotein as template. The forward and reverse oligonucleotide sequences are 5¢-GAAGAGTGGGCAAC GGATCCGAT AATATTTAAG and 5¢-CATTTCCTGCTGT CTGCAG TCAGACAAGTTTGAAG, respectively. The restriction sites encoded within these primers are underlined. The C-terminal NBD was amplified from DNA encoding cysteine-less P-glycoprotein [31] into which Cys1074 (in the Walker-A motif) had been re-introduced by site-directed mutagenesis (Altered Sites, Promega). The mutagenic primer had the sequence 5¢-GGCAGCAGTGGC TGTGG GAAGAGCACAG, in which the introduced cysteine codon is underlined. After introduction of the Cys1074, the C-terminal NBD was amplified using forward and reverse primers 5¢-CAGCACGGAAGGC GAATTCCCG AACACATTG and 5¢-CTTTGTTCCAGC CTGCAGT CAGACCATTGAAAA, respectively (restriction cloning sites underlined). N-terminal and C-terminal NBDs were cloned into the pMal-C2X vector at the BamHI/PstIand EcoRI/PstI sites, respectively, to generate plasmids pMal- C2x-Nter and pMal-C2x-Cter. Sequences of the NBDs were verified by DNA sequencing (Biochemistry Department, University of Oxford), which also confirmed the fidelity of the mutagenesis reaction. Protein expression pMal-C2x-Nter and pMal-C2x-Cter were transformed into chemically competent E. coli BL21.kDE3 [32]. Single colon- ies were inoculated into 5 mL Luria–Bertani broth supple- mented with 100 lgÆmL )1 ampicillin and grown overnight at 37 °C with shaking at 200 r.p.m., and then diluted 1 : 80 into 400 mL Luria–Bertani broth/ampicillin. Growth was con- tinued at 37 °C until an A 600 of 0.5 was achieved. Cultures were then cooled to 25 °C and induced with 0.2 m M isopropyl thio-b- D -galactoside. After 3–4 h shaking at 25 °C, bacteria were harvested by centrifugation at 3000 g. Purification of fusion proteins Bacterial cell pellets were resuspended in 10 mL lysis buffer (50 m M Tris/HCl, 150 m M NaCl, 20% glycerol, pH 7.4) by vortex-mixing. Bacteria were lysed by sonication (10 · 10 s bursts on ice). Lysis was verified by examination under a microscope. Bacterial lysates were cleared by a low-speed centrifugation (10 000 g for 10 min). Soluble proteins were isolated from this supernatant by centrifugation at 60 000 g for 60 min at 4 °C (Beckman TLA 100 rotor). Soluble protein was diluted to a concentration of 3 mgÆmL )1 and incubated with pre-equilibrated amylose resin (New Eng- land Biolabs) at a protein/resin ratio of 4 : 1 (w/v) for 30 min at room temperature. Resin and protein were loaded on to BioSpin columns (Bio-Rad), and unbound material was discarded. Nonspecifically bound proteins were subse- quently removed by four washes with 2 mL lysis buffer supplemented with 10 l M maltose. Bound fusion proteins were eluted by subsequent washes of 2 mL lysis buffer containing 1 m M maltose. Aliquots of all fractions were analysed by SDS/PAGE, and those containing fusion proteins were pooled and concentrated under nitrogen with an Amicon stirred-cell using a PM10 membrane (10-kDa cut-off). ATPase activities of purified fusion proteins were determined by colorimetric assay [33]. 1484 G. Berridge et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Labelling with N -ethylmaleimide N-[ 3 H]Ethylmaleimide (specific radioactivity 40–60 CiÆ mmol )1 ,1mCiÆmL )1 ) is supplied in pentane to which was added 250 lL dimethyl sulfoxide (1 : 4 the original vol- ume), and the pentane was evaporated under a stream of nitrogen. N-[ 3 H]Ethylmaleimide (4 mCiÆmL )1 )wasthen frozen in aliquots at )20 °C. N-[ 3 H]Ethylmaleimide treated in this way was stable without significant loss in the intensity of protein labelling for 4–6 weeks (data not shown). On longer-term storage, we found that greater exposure times were necessary to achieve the same labelling intensity, suggesting that exchange of the tritium label may occur with the solvent during storage over 2 months. The time dependence of labelling was measured by incubating protein (1.5 lg) with N-[ 3 H]ethylmaleimide (0.6 l M , determined by liquid-scintillation counting) for increasing times at 22 °C. The reaction was stopped by the addition of protein loading sample buffer containing a minimum 10-fold molar excess of a-mercaptoethanol. Control experiments established that this excess prevented further labelling. Proteins were resolved by SDS/PAGE (10% gels), fixed (propan-2-ol/acetic acid/water, 25 : 10 : 65, v/v/v), and soaked in AMPLIFYÒ (Amer- sham, UK). Gels were then dried on to Whatman 3 mm paper and exposed to photographic film (Kodak, MR1 film) at )80 °C. To investigate if nucleotide preincubation can prevent N-[ 3 H]ethylmaleimide labelling, 1.5 lg protein in lysis buffer, supplemented with 5 m M MgCl 2 , was incubated with increasing concentrations of nucleotide at 37 °Cfor 30 min. N-[ 3 H]Ethylmaleimide was then added to a final concentration of 0.6 l M and incubated at 22 °C for 15 min. Samples were then subjected to SDS/PAGE and autoradio- graphy as described above. The final concentration of dimethyl sulfoxide in any experiment never exceeded 2.5% (v/v). Parallel controls demonstrated that the addition of dimethyl sulfoxide to 25% (v/v) did not affect N-[ 3 H]ethyl- maleimide labelling. Autoradiographs were analysed using freely available densitometry software (Scion Image, www.scioncorp.com). All images were scanned at a resolution of 300 dpi and analysed without further modification. Exposures of the same gel for different time periods were employed to ensure that saturation of the densitometric signal had not occurred. For the time course of N-[ 3 H]ethylmaleimide labelling experiments, the most intensely labelled band was used as a 100% reference. Plots of time against percentage intensity were obtained, and the single-phase exponential association curve was fitted by nonlinear regression using PRISM (GraphPad, San Diego, CA, USA) to the equation: y ¼ y max Àð1 À exp Àkt Þ where y ¼ percentage saturation, t ¼ time, k ¼ time con- stant for labelling (T 0.5 ¼ 0.69/k). For investigations on nucleotide protection of N-[ 3 H]eth- ylmaleimide labelling, the most intensely labelled band was designated the reference signal. All other data points were quantified as the percentage intensity of the 100% reference signal. Plots of nucleotide concentration against percentage intensity were obtained and the sigmoidal dose–response curve was fitted by nonlinear regression, using Prism, to the equation: y ¼ bottom þ ðtop À bottomÞ 1 þ 10 ðlog IC 50Àlog x Þ n where y ¼ percentage of reference signal, x ¼ nucleotide concentration, bottom ¼ minimum labelling inten- sity, top ¼ saturation of labelling, IC 50 ¼ concentration required to reduce labelling intensity to 50% of its maximum, and n ¼ Hill slope. Data were fitted with either a Hill slope of 1.0 or a variable Hill slope. An F test was used to determine if the data were best fit by an equation containing a fixed or variable slope. Statistical comparison of T 0.5 and IC 50 values was performed using an unpaired Student’s t test. In all cases a value of P < 0.05 was considered significant. Unless indicated otherwise, all data are presented as the mean ± SEM. Molecular modelling of P-glycoprotein NBDs Sequences of P-glycoprotein NBDs and TAP1 were aligned using CLUSTALW [34]. A series of 10 homology models of each NBD was constructed, using the program MODEL- LER _6 V 2 [35], and employing the crystal structure of TAP1 as a structural template [36]. TAP1 is an appropriate choice of template as it shares a high degree of sequence identity to the NBDs of P-glycoprotein, and is a member of the same subfamily of ABC transporters (ABCB). However, the reported crystal structure of TAP1 does not contain ATP. Co-ordinates for ATP were added to TAP1 by least-squares superimposition of TAP1 and HisP, which does contain a bound ATP molecule [30]. Analysis of the individual P-glycoprotein NBD models in terms of stereochemistry and local packing enabled the selection of a preferred N-terminal and C-terminal NBD model. Structural analysis was performed using WHAT - CHECK [37] and structural diagrams were produced using MOLSCRIPT [38]. Results Expression and purification of fusion proteins Previously, we and others have shown that expression in E. coli of the N-terminal and C-terminal NBDs of P-glycoprotein as soluble proteins is very difficult to achieve [39–40, I. D. Kerr, G. Berridge & R. Callaghan, unpub- lished results]. To circumvent this, we expressed NBDs as fusions to the C-terminus of E. coli MBP. Subsequent functional assays utilized covalent attachment of probes to cysteine residues in the NBDs. E. coli MBP is devoid of cysteine residues [42], and the N-terminal NBD of P-glycoprotein contains only a single cysteine in the Walker-A motif [43]. Thus, MBP-NBD-Nter is a 75-kDa fusion protein containing a single cysteine residue located in the ATP-binding pocket. The production of a similar fusion protein containing the C-terminal NDB of P-glycoprotein is hampered by the fact that the C-terminus of P-glycoprotein contains three endogenous cysteine residues. To circumvent this we employed, as a PCR template, DNA encoding cysteine-less P-glycoprotein [31] in which the native cysteine at position 1074 was re-introduced by site-directed Ó FEBS 2003 Functional symmetry of P-glycoprotein NBDs (Eur. J. Biochem. 270) 1485 mutagenesis. Thus, both MBP-NBD-Nter and MBP-NBD- Cter are single-cysteine-containing proteins, in which the cysteine residue is in the Walker-A motif. MBP-NBDs are expressed to high level in a soluble form (Fig. 1, lane 1), % 60% of which binds to the amylose resin. Purification of the fusion proteins is facilely achieved with affinity chromatography (Fig. 1). This one-step purification produced a purity of % 95%. Typically, a 400-mL bacterial culture yielded 1–2 mg fusion protein at a final concentration of 0.5–1.0 mgÆmL )1 . This contrasts with our attempts to purify the N-terminal NBD in isolation as a soluble protein with a hexa-histidine tag, which requires a two-step purifi- cation to remove contaminating proteins (I. D. Kerr, G. Berridge & G. Callaghan, unpublished results). The oligomerization state of the fusion protein was investigated chromatographically. MBP-NBD-Cter or MBP-NBD-Nter was incubated with amylose resin for 30 min at room temperature. A purified, N-terminal NBD (free of MBP; I. D. Kerr, G. Berridge & G. Callaghan, unpublished results) was then added to the immobilized fusion protein. The N-ter- minal NBD passed through the column, and was present entirely in the flow-through fraction, demonstrating that under the conditions used there was no interaction with MBP-NBDs (data not shown). The chromatographic buffer conditions were identical with those used during N-ethylma- leimide labelling studies, suggesting that the fusion proteins are monomeric in solution, under these mild ionic conditions. Although a Factor Xa cleavage site is present in the linker between the two halves of the fusion protein, attempts to cleave NBD from MBP consistently displayed concomitant degradation of the NBD (there are two potential Factor Xa (Gly-Arg) sites one in the N-terminal and one in the C-terminal NBD). Similarly, the pMal-C2 plasmid is available with a subtilisin cleavage site in the linker region. NBDs were cloned into this plasmid, and fusion proteins were expressed. However, attempts at enzymatic hydrolysis of the resultant protein were ineffective, presumably because of inaccessibility of the cleavage site. Examination of the ATP-binding pockets of bacterial NBDs for which structural data are available (e.g. HisP, Rad50, MalK [30,44,45]) demonstrates that the side chain of the equivalent residue (Ser43 in HisP, Ser34 in Rad50, and Cys40 in MalK) is located within the ATP-binding pocket. Indeed, the mean distance between the equivalent side chain and the a-phosphate of bound nucleotide (or pyrophos- phate in the case of MalK) is only 5.2 A ˚ . This suggests that (a) the single cysteine in MBP-NBD proteins may be susceptible to derivatization, and (b) occupancy of the nucleotide-binding site could alter the accessibility and reactivity of this cysteine. Therefore, the purified single- cysteine-containing fusion proteins were examined for their ability to bind N-ethylmaleimide. The single cysteine residue is accessible to N -ethylmaleimide The initial experiments provided a detailed characterization of the binding of N-[ 3 H]ethylmaleimide to MBP-NBD fusion proteins. Time, temperature and concentration dependence of N-[ 3 H]ethylmaleimide labelling was investi- gated to optimize conditions for nucleotide preincubation experiments described below. It was demonstrated that effective labelling of fusion proteins could be obtained at pH 7.4, 22 °C and approximately equimolar N-ethylmale- imide and protein. Use of these assay conditions avoids nonspecific labelling of noncysteine residues (e.g. histidine) as previously described [46]. In support of this, N-[ 3 H]ethyl- maleimide was shown not to label MBP on its own, or fusion proteins derived from cysteine-less P-glycoprotein (data not shown), consistent with the absence of cysteine residues from these. The representative binding data presented in Fig. 2 for N-[ 3 H]ethylmaleimide investigates the time course of deri- vatization of the cysteine residue in the ATP-binding pocket and demonstrates its accessibility to N-[ 3 H]ethylmaleimide. The data were best fit to a single-phase exponential (again consistent with N-ethylmaleimide reacting with a single thiol residue), analysis of which produced a half-time (T 0.5 )for the association of N-[ 3 H]ethylmaleimide with MBP-NBD- Nter of 30.6 ± 1.5 min (n ¼ 4), and with MBP-NBD-Cter of 38.0 ± 4.3 min (n ¼ 4). Unpaired t test demonstrates no difference in the T 0.5 for N-[ 3 H]ethylmaleimide labelling of the two fusion proteins under these conditions. Preincubation with ATP but not ADP prevents N -[ 3 H]ethylmaleimide labelling The data above show that the cysteine in MBP-NBDs is accessible to modification by N-ethylmaleimide, confirming our initial hypothesis. Our second hypothesis, that nucleo- tide occupancy of the ATP-binding pocket would be manifested as a reduction in N-[ 3 H]ethylmaleimide labelling, was investigated in subsequent experiments. As N-ethyl- maleimide binding is irreversible, whereas nucleotide binding is reversible, we used constant experimental conditions to enable comparison of the effect of different nucleotides on N-[ 3 H]ethylmaleimide labelling of the two fusion proteins. To investigate potential asymmetry in the two NBDs, we preincubated fusion proteins with various concentrations of nucleotides, allowing them to come to equilibrium before Fig. 1. Purification of MBP-NBD fusion protein. MBP-NBD-Nter was purified from E. coli BL21 kDE-3 as described in Experimental pro- cedures. Lanes S (soluble proteins) and F (flow-through) contain 5 lg protein, whereas all other lanes contain 100 lL (1/20th of each frac- tion) precipitated by trichloroacetic acid. W1–W4 indicate four washes with 10 l M maltose. E1–E6 represent elution fractions in 1 m M maltose. Samples were resolved by SDS/PAGE (10% gel), and stained with Coomassie blue. The approximate positions of molecular-mass markers are indicated on the right. 1486 G. Berridge et al.(Eur. J. Biochem. 270) Ó FEBS 2003 adding 0.6 l M N-[ 3 H]ethylmaleimide. This reaction was allowed to proceed for 15 min at 22 °C, over which time any inhibition of N-[ 3 H]ethylmaleimide labelling, caused by the presence of nucleotide in the binding site, would be evident. Characterization of MBP-NBDs by the sensitive Chifflet assay [33] demonstrated that there was no NTPase activity of the fusion proteins (data not shown). Thus, any effects on N-[ 3 H]ethylmaleimide labelling are due to nucleotide bind- ing only. Nucleotide concentrations of up to 3.5 m M were employed. Higher concentrations of nucleotides caused a significant alteration of the pH, which may affect the specificity of maleimide reactivity [46]. Example results of experiments conducted with ATP are shown in Fig. 3A,B, and similar results were obtained in multiple independent experiments with different batches of fusion proteins. The data, fitted by a sigmoid dose–response equation, are plotted as a function of ATP concentration in Fig. 3C. ATP at concentrations higher than 3 m M was able to completely prevent N-ethylmaleimide labelling under the reaction conditions used. The mean data, obtained in at least eight experiments, return IC 50 values for the inhibition of N-[ 3 H]ethylmaleimide labelling for MBP-NBD-Nter of 1.8 ± 0.2 m M (n ¼ 8) and for MBP-NBD-Cter of 2.3 ± 0.2 m M (n ¼ 9). The difference in the potency of reduction in N-[ 3 H]ethylmaleimide labelling between the N-terminal and C-terminal NBDs was not significant. Pretreatment with 2-deoxy-ATP, an effective substitute in ATPase reactions, is also able to confer protection against N-[ 3 H]ethylmaleimide labelling with indistinguishable values of potency and extent of labelling diminution to ATP (Table 1). In contrast, ADP did not offer any protection against N-[ 3 H]ethylmaleimide labelling (Table 1), with negligible displacement at concentrations up to 3.5 m M in either the N-terminal or C-terminal NBD. This result was independ- ently confirmed by fluorescence experiments in which MIANS was used as the thiol-reactive agent. Again, no reduction in MIANS labelling of MBP-NBDs was demon- stratedatconcentrationofupto3.5m M ADP (data not shown). Preincubation with other nucleotides reveals subtle functional differences between N-terminal and C-terminal NBDs Data in the previous section demonstrated that N-[ 3 H]eth- ylmaleimide labelling of MBP-NBD fusion proteins was Fig. 3. ATP protects against N-[ 3 H]ethylmaleimide labelling of fusion proteins. Incubation of proteins with nucleotide and subsequent labelling with N-[ 3 H]ethylmaleimide was carried out as described in the text. (A) MBP-NBD-Nter and (B) MBP-NBD-Cter preincubated with increasing concentrations (0–3.5 m M ) of ATP (displayed beneath each lane). The approximate position of the 80-kDa molecular-mass marker is denoted by a solid line. The sigmoidal dose–response curve fits to the data are shown in (C). (j) MBP-NBD-Nter; (h) MBP-NBD-Cter. Fig. 2. Time-dependence of labelling of fusion proteins by N-[ 3 H]ethyl- maleimide. Labelling of MBP-NBD was carried out as described in the text. (A) MBP-NBD-Nter and (B) MBP-NBC-Cter labelling over time (in minutes displayed beneath each lane). The approximate position of the 80-kDa molecular-mass marker is denoted by a solid line. (C) Percentage saturation plotted as a function of labelling time. Data points (fitted to a single exponential equation) are derived from den- sitometric analysis of the data in (A) and (B). (j) MBP-NBD-Nter; (h) MBP-NBD-Cter. Ó FEBS 2003 Functional symmetry of P-glycoprotein NBDs (Eur. J. Biochem. 270) 1487 differentially affected by the hydrolysable substrate of P-glycoprotein (ATP) and the release product (ADP). However, there are no apparent differences between the N-terminal and C-terminal NBDs to bind nucleotide. To further characterize the molecular properties of each NBD, the potency of a number of different nucleotides to prevent N-[ 3 H]ethylmaleimide labelling was determined. The data for the effects of nonadenine-containing nucleotides is shown in Table 2. Of the compounds examined, the only nucleotide able to confer full protection against the deriva- tization of the cysteine residue in Walker-A was CTP (Table 2). However, the respective potencies to prevent labelling were not significantly different between the N-ter- minal NBD (IC 50 ¼ 2.2 ± 0.2 m M ) and the C-terminal NBD (IC 50 ¼ 2.3 ± 0.3 m M ). Neither dTTP nor ITP was able to fully prevent N-[ 3 H]ethylmaleimide labelling by preincubation at concentrations as high as 3.5 m M nucleo- tide, indicating that these nucleotides have a much reduced affinity for the NBDs of P-glycoprotein, compared with ATP or CTP. The most striking differences in N-[ 3 H]ethylmaleimide labelling between the two NBDs was observed with p[NH]ppA and GTP. GTP preincubation, at concentra- tions up to 3.5 m M did not protect against N-[ 3 H]ethyl- maleimide labelling of the MBP-NBD-Nter protein. In contrast, 44 ± 8% protection was seen for labelling of MBP-NBD-Cter at the highest achievable concentration of 3.5 m M . Thus, the data obtained with GTP demonstrates that the two NBDs contain subtle differences in their nucleotide-binding pocket that are discriminated by the guanine nucleotide. Further differences were highlighted by results obtained using p[NH]ppA protection against N-[ 3 H]ethylmaleimide labelling of the two NBDs (Fig. 4). Whereas N-[ 3 H]ethyl- maleimide labelling of MBP-NBD-Nter is completely prevented by p[NH]ppA over the experimental nucleotide concentration range (IC 50 ¼ 0.9 ± 0.1 m M ), labelling of MBP-NBD-Cter is only 41 ± 4% prevented at the highest nucleotide concentration. This suggests that the C-terminal NBD interacts differently with p[NH]ppA from the N-ter- minal NBD. Table 2. Inhibition of N-[ 3 H]ethylmaleimide labelling by nonadenosine nucleotides. All experimental details are identical with those given in Table 1. Where full protection from N-[ 3 H]ethylmaleimide labelling was observed, an IC 50 is presented, otherwise the mean inhibition of labelling observed at the highest nucleotide concentration (3.5 m M ) is given. N, number of experiments with data presented as the mean ± SEM. An asterisk denotes that the inhibition of N-[ 3 H]ethylmaleimide is significantly different between MBP-NBD-Nter and MBP-NBD-Cter (P <0.05).IC 50 parameter not determined, as it was not possible to fit the data to a dose-response equation, denoted by n/a. GTP ITP dTTP CTP Nter Cter Nter Cter Nter Cter Nter Cter IC 50 n/a n/a n/a n/a n/a n/a 2.2 ± 0.2 2.3 ± 0.3 Inhibition (%) 0 ± 10* 48 ± 3* 44 ± 8 27 ± 6 42 ± 4 41 ± 10 93 ± 7 97 ± 3 N 4 4 6454 6 7 Fig. 4. p[NH]ppA reacts differently with N-terminal and C-terminal MBP-NBD fusion proteins. Incubation of proteins with nucleotide and subsequent labelling with N-[ 3 H]ethylmaleimide was carried out as described in Experimental Procedures. The mean ± SEM is shown. (j) MBP-NBD-Nter; (h) MBP-NBD-Cter. Table 1. Inhibition of N-[ 3 H]ethylmaleimide labelling by adenosine nucleotides. Nucleotide at various concentrations was preincubated with fusion protein, followed by addition of N-[ 3 H]ethylmaleimide. The percentage inhibition of N-ethylmaleimide labelling observed was quantified as described in the text. Nter represents MBP-NBD-Nter; Cter represents MBP-NBD-Cter. Where full protection from N-[ 3 H]ethylmaleimide labelling was observed, an IC 50 is presented. Otherwise the mean protection from labelling observed at the highest nucleotide concentration (3.5 m M ) is given. N, number of experiments. Data presented as the mean ± SEM. An asterisk denotes that the inhibition of N-[ 3 H]ethylmaleimide is significantly different between MBP-NBD-Nter and MBP-NBD-Cter (P < 0.05). IC 50 parameter not determined, as it was not possible to fit the data to a dose-response equation, denoted by n/a. ATP dATP p[NH]ppA ADP Nter Cter Nter Cter Nter Cter Nter Cter IC 50 1.8 ± 0.2 2.3 ± 0.2 1.6 ± 0.2 2.0 ± 0.2 0.9 ± 0.1 n/a n/a n/a Inhibition (%) 94 ± 4 94 + 2 89 ± 9 98 ± 2 100 ± 1* 41 ± 4* 0 ± 12 0 ± 10 N 8956 6499 1488 G. Berridge et al.(Eur. J. Biochem. 270) Ó FEBS 2003 Discussion In this study, we have investigated the similarity between the two NBDs of P-glycoprotein in nucleotide binding. Differ- ences in this initial step of the catalytic cycle would be manifested in asymmetric roles in transport. In this work, we used a fusion protein consisting of either the N-terminal or C-terminal NBD of P-glycoprotein, fused to E. coli MBP. This is necessary because of the inherent insolubility of isolated NBDs of human ABC transporters (I. D. Kerr, G. Berridge & R. Callaghan, unpublished results). A similar fusion protein approach has previously been used to express the NBDs of P-glycoprotein [47,48]. Consistent with this, we were unable to specifically cleave NBD1 from MBP-NBD fusions by digestion with Factor Xa. In contrast with our data, nucleotide hydrolysis was demonstrated in the latter study [48], although at a lower specific activity than observed for full-length P-glycoprotein, potentially because of the quantities of fusion protein employed in ATPase assays. In the light of the inability of our fusion proteins to hydrolyse nucleotide, a novel approach was used to characterize the interaction of nucleotides with NBDs, employing the unique, reactive cysteine residue in the Walker-A sequence. The two NBDs of P-glycoprotein are functionally similar The similar time course of binding of N-[ 3 H]ethylmaleimide to the Walker-A cysteine residue in the N-terminal and C-terminal NBDs suggests that local steric effects and accessibility of the cysteine are identical in the two halves of P-glycoprotein. We therefore used nucleotide protection of N-[ 3 H]ethylmaleimide labelling to investigate the inter- actions of diverse nucleotides with P-glycoprotein NBDs. Protection against derivatization is afforded when nucleo- tide is in the binding pocket, and therefore the potency of protection is a measure of binding affinity. Of course, as nucleotide binding is a reversible process and N-ethylmale- imide labelling is irreversible, we are unable to determine absolute binding affinities of nucleotides from such data. However, under constant experimental conditions, we are able to compare the relative affinities of nucleotides. As an alternative approach we investigated the possibility of using 8-azido-[ 32 P]ATP labelling to determine relative affinities of nucleotides. However, the maximum concentration of commercially available 8-azido-[ 32 P]ATP is only 100 l M in methanol. Avoiding excess solvent in labelling experiments would impose an upper limit on the achievable concentra- tion of 8-azido-[ 32 P]ATP of % 5 l M . This is two orders of magnitude below the K m for 8-azido-ATP [29], suggesting that only 2–4% of fusion protein would be labelled, thus preventing meaningful competition binding studies. The data we present show that the hydrolysable substrate (ATP) and the hydrolysis product (ADP) are nonequivalent in this system. Whereas ATP can fully inhibit maleimide labelling (within the 15 min time course of the experiment), ADP is incapable of preventing cysteine derivatization over the same time period. This suggests that either ADP has a lower binding affinity or a more rapid dissociation from the NBD than ATP. Rapid dissociation of hydrolysis product may be expected as part of the kinetics of the transport cycle of the intact transporter. Furthermore, the data demon- strate that the two NBDs interact in a similar manner with 2-deoxy-ATP and CTP. Lastly, the data provide evidence that ITP and dTTP bind weakly, and that their respective interaction with the N-terminal and C-terminal NBDs is identical. Our results are comparable to data obtained on the inhibition of ATPase activity of full-length P-glycoprotein by N-ethylmaleimide. Although P-glycoprotein contains additional cysteine residues which might constitute male- imide-binding sites, it has been demonstrated that both Walker-A cysteines are accessible to covalent modification, and that inhibition of ATPase activity is achieved by derivatization of either cysteine [5,8,29]. It has been observed that ATP incubation offered protection against this N-ethylmaleimide-mediated inhibition, with between 2 and 10 m M nucleotide being necessary to restore ATPase activity [8,49]. The protection of N-[ 3 H]ethylmaleimide labelling by preincubation with diverse nucleotides suggests that ATP, 2-deoxy-ATP, p[NH]ppA and CTP can all bind effectively at the NBDs of P-glycoprotein. This is consistent with ATP and dATP being hydrolysis substrates with K m values approaching 1 m M [5,29]. It is also consistent with p[NH]ppA being effective as an inhibitor of this hydrolysis with an EC 50 of 0.4 m M [8,49]. Although we have demon- strated that CTP is effective at preventing N-[ 3 H]ethyl- maleimide labelling, this nucleotide is not an effective substrate for continuous hydrolysis [8,49], suggesting that its binding does not induce the conformational changes that accompany nucleotide hydrolysis. Sequence and structural considerations of P-glycoprotein NBDs Our data do show some differences between the NBDs of P-glycoprotein with respect to two nucleotides. The non- hydrolysable analogue p[NH]ppA and the purine GTP interact differentially. GTP fails to inhibit N-[ 3 H]ethyl- maleimide labelling of the N-terminal NBD, whereas p[NH]ppA appears to have a lower affinity for the C-ter- minal NBD. Can these differences be related to sequence and structural properties of the two NBDs? To address this, we have generated homology models for the N-terminal and C-terminal NBDs of P-glycoprotein based on the crystal structure of TAP1 [36] (Fig. 5). A representative model of the N-terminal NBD is shown in Fig. 5A (the C-terminal NBD model has a similar structure and therefore is not shown). Figure 5B displays in detail the vicinity of the ATP molecule demonstrating the exposure of the cysteine residue (Cys431) to the ATP-binding pocket. Two sequence differ- ences between the N-terminal and C-terminal NBD in this region are highlighted in ball-and-stick fashion in Fig. 5B. The first is the presence of an asparagine (Asn428) in the Walker-A motif of the N-terminal NBD, which is replaced by a serine (Ser1071) in the C-terminal NBD. The side chain of this asparagine is less than 6 A ˚ from the Pb–Ob phospho- anhydride bond of ATP. This is the bond that is altered in p[NH]ppA, suggesting that replacement of Asn428 by Ser1071 could confer subtle alterations on p[NH]ppA binding, in agreement with the results obtained here. The nonidentical interaction of GTP may be related to another variation in sequence in the vicinity of the ATP-binding pocket, specifically in the ABC-specific b-sheet subdomain Ó FEBS 2003 Functional symmetry of P-glycoprotein NBDs (Eur. J. Biochem. 270) 1489 (Fig. 5A). Our homology models of P-glycoprotein NBDs suggest that replacement of residue Ser400 in the N-terminal NBD by the bulkier Asn1043 in the C-terminal NBD may be sufficient to impart the observed effects on interaction with GTP (possibly mediated via water molecules). Functional and structural dissimilarity in the NBDs of ABC transporters For many eukaryotic ABC transporters, there are many data to support functional asymmetry between the NBDs in mediating transport. This includes demonstrations that this functional asymmetry is a result of the inherent properties of the NBDs, rather than of their interactions with the TMDs (e.g [50] and see introduction for further references). Recent structural determinations of both intact ABC transporters and NBD proteins also lend weight to asymmetry in these proteins. The crystal dimer of the NBD of the maltose- uptake system of Thermococcus littoralis (MalK) consists of two monomers with deviation from perfect twofold sym- metry [44], although an analysis of how this might be related to nucleotide binding is precluded by the presence of pyrophosphate in the binding pocket, rather than ATP [44]. In addition, the cryo-electron microscopy structure of YvcC supports structural asymmetry in ABC transporters [51]. YvcC is a homodimeric protein (each monomer consisting of a single TMD and NBD). The structures identified as the NBDs are of different dimensions and thus, presumably, different conformations [51]. Whether this is attributable to NBD–NBD interactions or NBD–TMD interactions awaits a higher-resolution structure. Summary We have shown that the two NBDs of P-glycoprotein are substantially functionally symmetrical in terms of their binding to diverse nucleotides. Any functional asymmetry observed in the intact transporter is probably not entirely due to inherent properties of the NBD, and presumably reflects either differences in the rate of hydrolysis or the effects of interdomain interactions. In particular, our data demonstrate that, in isolation, both NBDs interact identi- cally with ATP, in agreement with a recent spectroscopic study of P-glycoprotein, which observed secondary struc- tural asymmetry as a result of nucleotide hydrolysis, but not nucleotide binding [10]. Our resultant hypothesis that either NBD may be recruited to hydrolyse ATP during the transport cycle will be tested by further experiments involving single-cysteine isoforms of full-length P-glycoprotein. Acknowledgements This work was funded by a Wellcome Trust Career Development Fellowship to I.D.K., a Wellcome Trust Vacation Scholarship to J.A.W., and Cancer Research UK Grant to R.C. We thank Natalie Gabriel for preliminary purifications of MBP-NBD fusion proteins, and Georgios Samoilis for assistance with site-directed mutagenesis. We thank Catherine Martin, Mark Gabriel, Alice Rothnie, Janet Storm, andDrB.Nardfordiscussions. References 1. Higgins, C.F. (1992) ABC transporters: from microorganisms to man. Annu.Rev.CellBiol.8, 67–113. Fig. 5. Homology modelling of P-glycoprotein NBDs. Structural models for the N-terminal and C-terminal NBD of P-glycoprotein were constructed as described in the text. (A) N-terminal NBD model presented in ribbon format, with a-helices represented as spirals, and a-strands as arrows. The ATP molecule is shown in ball-and-stick format. The scale bar represents 5 A ˚ .(B)Aclose-upviewofthe nucleotide-binding pocket in the N-terminal homology model. ATP is shown as before, as are the side chains of Cys431, Asn428 and Ser400. The scale bar represents 2 A ˚ . 1490 G. Berridge et al.(Eur. J. Biochem. 270) Ó FEBS 2003 2. Holland, I.B. & Blight, M.A. (1999) ABC-ATPases, adaptable energy generators fuelling transmembrane movement of a variety of molecules in organisms from bacteria to humans. J. Mol. Biol. 293, 381–399. 3. Kerr, I.D. (2002) Structure and association of ATP binding cas- sette transporter nucleotide-binding domains. Biochim. Biophys. Acta 1561, 47–64. 4. Urbatsch, I.L., Sankaran, B., Bhagat, S. & Senior, A.E. (1995) Both P-glycoprotein nucleotide-binding sites are catalytically active. J. Biol. Chem. 270, 26956–26961. 5. al Shawi, M.K. & Senior, A.E. (1993) Characterization of the adenosine triphosphatase activity of Chinese hamster P-glyco- protein. J. Biol. Chem. 268, 4197–4206. 6. Senior, A.E., al Shawi, M.K. & Urbatsch, I.L. (1995) The catalytic cycle of P-glycoprotein. FEBS Lett. 377, 285–289. 7. Takada, Y., Yamada, K., Taguchi, Y., Kino, K., Matsuo, M., Tucker, S.J., Komano, T., Amachi, T. & Ueda, K. (1998) Non- equivalent cooperation between the two nucleotide-binding folds of P-glycoprotein. Biochim. Biophys. Acta 1373, 131–136. 8. Loo, T.W. & Clarke, D.M. (1995) Covalent modification of human P-glycoprotein mutants containing a single cysteine in either nucleotide-binding fold abolishes drug- stimulated ATPase activity. J. Biol. Chem. 270, 22957–22961. 9. Hrycyna, C.A., Ramachandra, M., Germann, U.A., Cheng, P.W., Pastan, I. & Gottesman, M.M. (1999) Both ATP sites of human P-glycoprotein and essential but not symmetric. Biochemistry 38, 13887–13899. 10. Vigano,C.,Julien,M.,Carrier,I.,Gros,P.&Ruysschaert,J.M. (2002) Structural and functional asymmetry of the nucleotide- binding domains of P-glycoprotein investigated by attenuated total reflection Fourier transform infrared spectroscopy. J. Biol. Chem. 277, 5008–5016. 11. Szakacs, G., Ozvegy, C., Bakos, E., Sarkadi, B. & Varadi, A. (2001) Role of glycine-534 and glycine-1179 of human multidrug resistance protein (MDR1) in drug-mediated control of ATP hydrolysis. Biochem. J. 356, 71–75. 12. Urbatsch,I.L.,Gimi,K.,Wilke-Mounts,S.&Senior,A.E.(2000) Investigation of the role of glutamine-471 and glutamine-1114 in the two catalytic sites of P-glycoprotein. Biochemistry 39, 11921– 11927. 13. Urbatsch,I.L.,Gimi,K.,Wilke-Mounts,S.&Senior,A.E.(2000) Conserved Walker A Ser residues in the catalytic sites of P-gly- coprotein are critical for catalysis and involved primarily at the transition state step. J. Biol. Chem. 275, 25031–25038. 14. Gao,M.,Cui,H.R.,Loe,D.W.,Grant,C.E.,Almquist,K.C., Cole, S.P. & Deeley, R.G. (2000) Comparison of the functional characteristics of the nucleotide binding domains of multidrug resistance protein 1. J. Biol. Chem. 275, 13098–13108. 15. Hou, Y., Cui, L., Riordan, J.R. & Chang, X. (2000) Allosteric interactions between the two non-equivalent nucleotide binding domains of multidrug resistance protein MRP1. J. Biol. Chem. 275, 20280–20287. 16. Lapinski, P.E., Neubig, R.R. & Raghavan, M. (2001) Walker A lysine mutations of TAP1 and TAP2 interfere with peptide translocation but not peptide binding. J. Biol. Chem. 276, 7526– 7533. 17. Saveanu, L., Daniel, S. & van Endert, P.M. (2001) Distinct functions of the ATP binding cassettes of transporters associated with antigen processing: a mutational analysis of Walker A and B sequences. J. Biol. Chem. 276, 22107–22113. 18. Proff, C. & Kolling, R. (2001) Functional asymmetry of the two nucleotide binding domains in the ABC transporter Ste6. Mol. Gen. Genet. 264, 883–893. 19. Anderson, M.P. & Welsh, M.J. (1992) Regulation by ATP and ADP of CFTR chloride channels that contain mutant nucleotide- binding domains. Science 257, 1701–1704. 20. Carson, M.R., Travis, S.M. & Welsh, M.J. (1995) The two nucleotide-binding domains of cystic fibrosis transmembrane conductance regulator (CFTR) have distinct functions in con- trolling channel activity. J. Biol. Chem. 270, 1711–1717. 21. Druley, T.E., Stein, W.D. & Roninson, I.B. (2001) Analysis of MDR1 P-glycoprotein conformational changes in permeabilized cells using differential immunoreactivity. Biochemistry 40, 4312– 4322. 22. Martin, C., Berridge, G., Mistry, P., Higgins, C.F., Charlton, P. & Callaghan, R. (2000) Drug binding sites on P-glycoprotein are altered by ATP binding prior to nucleotide hydrolysis. Biochem- istry 39, 11901–11906. 23. Wang, G., Pincheira, R., Zhang, M. & Zhang, J T. (1997) Con- formational changes of P-glycoprotein by nucleotide binding. Biochem. J. 328, 897–904. 24. Rosenberg, M.F., Velarde, G., Ford, R.C., Martin, C., Berridge, G., Kerr, I.D., Callaghan, R., Schmidlin, A., Wooding, C., Lin- ton, K.J. & Higgins, C.F. (2001) Repacking of the transmembrane domains of P-glycoprotein during the transport ATPase cycle. EMBO J. 20, 5615–5625. 25. Karpowich, N., Martsinkevich, O., Millen, L., Yuan, Y.R., Dai, P.L., MacVey, K., Thomas, P.J. & Hunt, J.F. (2001) Crystal structures of the MJ1267 ATP-binding cassette reveal an induced- fiteffectattheATPaseactivesiteofanABCtransporter.Structure 9, 571–586. 26. Yuan, Y.R., Blecker, S., Martsinkevich, O., Millen, L., Thomas, P.J. & Hunt, J.F. (2001) The crystal structure of the MJ0796 ATP- binding cassette: Implications for the structural consequences of ATP hydrolysis in the active site of an ABC-transporter. J. Biol. Chem. 276, 32313–32321. 27. Locher, K.P., Lee, A.T. & Rees, D.C. (2002) The E. coli BtuCD structure: a framework for ABC transporter architecture and mechanism. Science 296, 1091–1098. 28. Smith, P.C., Karpowich, N., Millen, L., Moody, J.E., Rosen, J., Thomas, P.J. & Hunt, J.F. (2002) ATP binding to the motor domain from an ABC transporter drives formation of a nucleotide sandwich dimer. Mol. Cell. 10, 139–149. 29. Urbatsch, I.L., al Shawi, M.K. & Senior, A.E. (1994) Character- ization of the ATPase activity of purified Chinese hamster P-gly- coprotein. Biochemistry 33, 7069–7076. 30. Hung, L.W., Wang, I.X., Nikaido, K., Liu, P.Q., Ames, G.F. & Kim, S.H. (1998) Crystal structure of the ATP-binding subunit of an ABC transporter. Nature (London) 396, 703–707. 31. Taylor, A.M., Storm, J., Linton, K.J., Gabriel, M., Blott, E.J.,Martin,C.,Higgins,C.F.,Mistry,P.,Charlton,P.& Callaghan, R. (2001) Subtle pharmacological differences between a Cys-less variant of the human multidrug resistance protein, P-glycoprotein, and wild-type. Br.J.Pharmacol.134, 1609–1618. 32. Hanahan, D. (1983) Studies on transformation of Escherichia coli with plasmids. J. Mol. Biol. 166, 557–580. 33. Chifflet, S., Torriglia, A., Chiesa, R. & Tolosa, S. (1988) A method for the determination of inorganic phosphate in the presence of labile organic phosphate and high concentrations of protein: application to lens ATPases. Anal. Biochem. 168,1–4. 34. Thompson, J.D., Higgins, D.G. & Gibson, T.J. (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22, 4673– 4680. 35. Sali, A. & Blundell, T.L. (1993) Comparative protein modelling by satisfaction of spatial restraints. J. Mol. Biol. 234, 779–815. 36. Gaudet, R. & Wiley, D.C. (2001) Structure of the ABC ATPase domain of human TAP1, the transporter associated with antigen processing. EMBO J. 20, 4964–4972. Ó FEBS 2003 Functional symmetry of P-glycoprotein NBDs (Eur. J. Biochem. 270) 1491 37. Rodriguez, R., Chinea, G., Lopez, N., Pons, T. & Vriend, G. (1998) Homology modeling, model and software evaluation: three related resources. Bioinformatics 14, 523–528. 38. Kraulis, P.J. (1991) MOLSCRIPT: a program to produce both detailed and schematic plots of protein structure. J. Appl. Crys- tallogr. 24, 946–950. 39. Dayan, G., Jault, J.M., Baubichon Cortay, H., Baggetto, L.G., Renoir, J.M., Baulieu, E.E., Gros, P. & Di Pietro, A. (1997) Binding of steroid modulators to recombinant cytosolic domain from mouse P-glycoprotein in close proximity to the ATP site. Biochemistry 36, 15208–15215. 40. de Wet, H., McIntosh, D.B., Conseil, G., Baubichon-Cortay, H., Krell, T., Jault, J.M., Daskiewicz, J.B., Barron, D. & Di Pietro, A. (2001) Sequence requirements of the ATP-binding site within the C-terminal nucleotide-binding domain of mouse P-glycoprotein: structure-activity relationships for flavonoid binding. Biochemistry 40, 10382–10391. 41. Reference withdrawn. 42. Duplay, P., Bedouelle, H., Fowler, A., Zabin, I., Saurin, W. & Hofnung, M. (1984) Sequences of the malE gene and of its pro- duct, the maltose-binding protein of Escherichia coli K12. J. Biol. Chem. 259, 10606–10613. 43. Chen,C.J.,Clark,D.,Ueda,K.,Pastan,I.,Gottesman,M.M.& Roninson, I.B. (1990) Genomic organization of the human multi- drug resistance (MDR1) gene and origin of P-glycoproteins. J. Biol. Chem. 265, 506–514. 44. Diederichs, K., Diez, J., Greller, G., Muller, C., Breed, J., Schnell, C.,Vonrhein,C.,Boos,W.&Welte,W.(2000)Crystalstructure of MalK, the ATPase subunit of the trehalose/maltose ABC transporter of the archaeon Thermococcus litoralis. EMBO J. 19, 5951–5961. 45. Hopfner, K.P., Karcher, A., Shin, D.S., Craig, L., Arthur, L.M., Carney, J.P. & Tainer, J.A. (2000) Structural biology of Rad50 ATPase: ATP-driven conformational control in DNA double- strand break repair and the ABC-ATPase superfamily. Cell 101, 789–800. 46. Brewer, C. & Riehm, J. (1967) Evidence for possible non-specific reactions between N-ethylmaleimide and proteins. Anal. Biochem. 18, 248–255. 47. Wang, C., Castro, A.F., Wilkes, D.M. & Altenberg, G.A. (1999) Expression and purification of the first nucleotide-binding domain and linker region of human multidrug resistance gene product: comparison of fusions to glutathione S-transferase, thioredoxin and maltose-binding protein. Biochem. J. 338, 77–81. 48. Wilkes, D.M., Wang, C., Aristimuno, P.C., Castro, A.F. & Altenberg, G.A. (2002) Nucleotide triphosphatase activity of the N-terminal nucleotide-binding domains of the multidrug resistance proteins P-glycoprotein and MRP1. Biochem. Biophys. Res. Commun. 296, 388–394. 49. al Shawi, M.K., Urbatsch, I.L. & Senior, A.E. (1994) Covalent inhibitors of P-glycoprotein ATPase activity. J. Biol. Chem. 269, 8986–8992. 50. Daumke, O. & Knittler, M.R. (2001) Functional asymmetry of the ATP-binding-cassettes of the ABC transporter TAP is determined by intrinsic properties of the nucleotide binding domains. Eur. J. Biochem. 268, 4776–4786. 51. Chami, M., Steinfels, E., Orelle, C., Jault, J.M., Di Pietro, A., Rigaud, J.L. & Marco, S. (2002) Three-dimensional structure by cryo-electron microscopy of YvcC, an homodimeric ATP-binding cassette transporter from Bacillus subtilis. J. Mol. Biol. 315, 1075– 1085. 1492 G. Berridge et al.(Eur. J. Biochem. 270) Ó FEBS 2003 . The nucleotide-binding domains of P-glycoprotein Functional symmetry in the isolated domain demonstrated by N -ethylmaleimide labelling Georgina Berridge 1 ,. 75-kDa fusion protein containing a single cysteine residue located in the ATP-binding pocket. The production of a similar fusion protein containing the C-terminal NDB