| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
MRC Clinical Sciences Centre, Imperial College London, Hammersmith Hospital Campus, London, UK
1Correspondence: MRC Clinical Sciences Centre, Imperial College, Hammersmith Hospital Campus, Du Cane Rd., London W12 0NN, UK. E-mail: kenneth.linton{at}csc.mrc.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: ATP binding cassette transporter • drug efflux • ABC protein • MDR1 • ABCB1
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. They form one of the largest superfamilies of proteins with multiple paralogs in all extant phyla (and probably all species). Typically they are very specific for the transported ligand, which can include metal ions, sugars and polysaccharides, vitamins, lipids, amino acids, peptides, or proteins. Many are essential for human health, and their dysfunction underlies a number of human genetic diseases (2)
. Some ABC transporters have evolved the ability to recognize a broad spectrum of functionally and structurally unrelated compounds, many of which are cytotoxic and clinically important as therapeutics. Expression of these multispecific ABC transporters in the gastrointestinal tract, and in the liver and kidneys, reduces both the uptake of drugs administered orally and the retention time of drugs in circulation (3)
. They are also important components of permeability barriers, for example, between blood and brain and maternal and fetal circulation, and when expressed in cancer cells they prevent the intracellular accumulation of therapeutic drugs and so cause multidrug resistance (MDR).
A great deal of interest has focused on the of understanding how these ABC transporters work. A considerable body of biochemistry data has been accrued on a wide variety of transporters, ranging from bacterial sugar importers to multispecific drug exporters from humans (4)
. These studies have been complemented more recently by structure determination providing a framework for interpretation of the biochemistry in more mechanistic terms. The resulting general picture of an ABC transporter is of a protein with four domains (or a protein complex if all four domains are not on a single polypeptide). Two nucleotide binding domains (NBDs), which are homologous throughout the entire superfamily, form two ATP-binding pockets at their shared interface; and two transmembrane domains (TMDs) bind the transported ligand, determining the specificity of the transporter, and form the translocation pathway across the biological membrane. The activity of most ABC transporters can be thought of as two coupled cycles, the ATP catalytic cycle of the NBDs and the ligand transport cycle of the TMDs. The binding of ATP at the interface between the two NBDs and the allosteric coupling of the two catalytic sites is well characterized (5
6
7
8)
. Furthermore, several labs have crystallized NBD dimers in multiple conformations with different nucleotides and described the molecular mechanisms for ATP binding, the allosteric interactions between the two sites, and also two conflicting mechanisms for catalysis (9
10
11
12
13)
. The TMDs, on the other hand, are not homologous throughout the ABC transporter superfamily and have been more resistant to analysis. For the MDR class, the drug binding sites of P-glycoprotein are the best characterized. Four sites with distinct affinities for different drugs and their allosteric interactions have been described pharmacologically (14)
. Although these drug binding sites are formed by TMDs (15)
, the molecular basis for drug binding (or ligand binding by the TMDs of any ABC transporter) is not clear.
Many of the drugs that bind to the TMDs of P-glycoprotein stimulate ATP hydrolysis at the NBDs in vitro by 2- to 10-fold (16
17
18)
(the magnitude depending on the drugs used and also on the detergent and lipids used for purification). In ABC transporters, coupling of ligand binding with ATP binding and hydrolysis must require signal and energy transduction across the interface between the TMDs and NBDs. Despite structure determination of the whole transport complex for four bacterial ABC transporters, the vitamin B12 importer, BtuCD, from Escherichia coli (19)
; a metal chelate importer, HI1470/1, from Haemophilus influenzae (20)
; a molybdate importer ModBC from Archaeoglobus fulgidus (21)
; and the drug exporter, Sav1866, from Staphylococcus aureus (two structures have been reported for Sav1866, but they are virtually identical) (22
, 23
; Fig. 1
), the mechanism by which signals and energy are transduced across the NBD:TMD interface remains obscure. This is partly because, at first appraisal, the three structures of the importers appear to be inconsistent with the structure of the drug exporter. The basic ABC transport complex of BtuCD, HI1470/1, and ModBC is a tetramer comprising homodimeric NBDs and homodimeric TMDs (to import, each also requires a ligand-binding protein to deliver ligand to the TMDs). BtuCD and HI1470/1 share homologous TMDs as well as NBDs. The plane of the TMD:TMD interface in both is closely, but not perfectly, aligned with the plane of the NBD:NBD interface, and each TMD makes contact directly with only one NBD (Fig. 1B, C, F
). Despite being crystallized in the absence of nucleotide (HI1470/1), or with cyclo-tetravanadate (BtuCD) substituting for nucleotide, the interdomain arrangements of the two importers seem entirely plausible, and the relatively subtle differences between them may represent different conformations in a shared molecular mechanism of transport. The TMD (ModB) of the molybdate importer is not homologous with BtuD and HI1471 and has only 6 transmembrane helices (TMH; Fig. 1D
), compared to 10 in BtuD and HI1471. It shares, however, some structural similarities with the other two importers; TMH2–6 form a TMD:TMD interface, the plane of which mirrors the NBD:NBD interface, and the intracellular loops of the TMDs are short, bringing the NBDs into close proximity with the membrane. Each TMD (ModB), through TMH2–6, also makes contact predominantly with the NBD (ModC) directly below it (Fig. 1G
). [The amino-terminal TMH1 of ModB crosses the membrane in a helical bundle with TMH2–6 of the dimer-associated TMD, and comes into close proximity with two residues of the other NBD (green residues in Fig. 1G
)]. While this is superficially reminiscent of the composite helical bundles formed by the TMDs of Sav1866 (in which TMH1,2 are associated with TMH3',4',5',6' of the dimerically associated TMD), there is no evidence that ModB is homologous with the TMD of Sav1866. So, while the NBD of Sav1866 is homologous with BtuC, HI1470, and ModC, the Sav1866 TMD is nonhomologous with the TMDs of these bacterial importers and is representative of a different class of ABC transporter (the MDR class). In contrast to the interdomain architecture observed in the two importers, the TMD:TMD interface of Sav1866 is dramatically out of alignment with the NBD:NBD interface and each TMD comes into close proximity to both NBDs (Fig. 1A, E
). Sav1866 is a half-transporter (the polypeptide has an amino-terminal TMD and a carboxy-terminal NBD, which homodimerizes to form the transport complex), so it is evident that each TMD makes contact predominantly with the NBD from the other monomer, rather than the one to which it is fused covalently in a two-domain polypeptide. To our knowledge there is no biochemical evidence for this type of arrangement from studies of any ABC transporter. There is, however, contradictory direct evidence from P-glycoprotein, in which all four domains are present on a single polypeptide akin to a head-to-tail fusion of two Sav1866 half transporters. When each domain of P-glycoprotein is engineered as a separate polypeptide and coexpressed in cells in different combinations, NBD1 will stabilize expression of TMD1, and NBD2 will stabilize the expression of TMD2. However, for combinations of NBD1 and TMD2 (or NBD2 and TMD1), corresponding to the major interdomain interfaces proposed by the Sav188 structure, no evidence of association was forthcoming from this cellular trafficking assay (24)
.
|
As Sav1866 and P-glycoprotein are both members of the MDR class of ABC transporter, sharing homologous TMDs as well as homologous NBDs (supplemental Figs. 1 and 2), the domain arrangement of Sav1866 is testable in P-glycoprotein. We have used a cysteine-free P-glycoprotein as a template for the introduction of paired cysteines across the proposed interface between NBD1 and intracellular loop 4 (ICL4) of TMD2. These changes were shown to be minimally invasive using a novel and sensitive assay of P-glycoprotein function, before assessing the proximity of the introduced cysteines by chemical cross-linking. Our data corroborate the structural data reported for Sav1866 and support a Sav1866-like organization of P-glycoprotein at the energy and signal transduction interface between the TMDs and the NBDs. The implications of this conclusion are discussed in relation to the coupling of the ATP-switch mechanism proposed for the NBDs (4
, 25)
to ligand transport by ABC transporters in general.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 27)
. Plasmid Tracer-Cd36 is derived from pTracer-CMV/Bsd (Invitrogen, Carlsbad, CA, USA) and encodes the membrane glycoprotein Cd36 and also the green fluorescent protein (GFP). Novel cysteine codons were introduced into pMDR-cys– by site-directed mutagenesis (QuikChange multisite; Stratagene, La Jolla, CA, USA) using the oligonucleotides L443C, 5'-GTCCAGCTGATGCAGCGCTGCTATGACCCCACAG-3'; S474C, 5'-CTACGGGAAATCATTGGCGTCGTGTGTCAGGAACCTGTATTG-3'; R905C, 5'-GCAATAGAAAACTTCTGTACAGTTGTTTCTTTGAC-3'; and S909C, 5'-CCGAACCGTTGTTTGTCTCACCCAGGAGCAGAAGTTTG-3'. The E556Q mutant was generated by mutation of the wild-type P-glycoprotein using the oligonucleotide E556Q, 5'-CCCCAAGATCCTCCTGCTTGATCAGGCCACGTCAGCCTTGG-3'. Each oligonucleotide also introduced a restriction enzyme site, silent with respect to the protein sequence, to track the mutagenesis. The sequence of the entire gene encoding each mutant protein was verified by DNA sequencing. Human embryonic kidney 293T (HEK293T) cells were transfected with the plasmid DNAs, as described previously (28)
. For cotransfection experiments the volume of the transfection mix was doubled, while the concentration of the individual components (including the total DNA concentration) was kept constant.
Real-time BodipyTM-taxol uptake assay
Untransfected HEK293T cells, and cells transiently expressing wild-type or mutant P-glycoprotein (mock transfected cells are cells transfected with a mutant pMDR-wt, which has a frameshift mutation that prevents expression of the protein), were harvested by brief incubation with trypsin in 2 mM EDTA in phosphate-buffered saline. The trypsin was neutralized by the addition of 5.5 volumes of DMEM (Invitrogen) supplemented with 10% FCS (Invitrogen), and the cells recovered by centrifugation at 150 g for 10 min at 4°C (Hettich rotanta 46R; Hettich, Tuttlingen, Germany). The cell pellets were resuspended in FACS medium (phenol red-free DMEM/F-12 supplemented with 1% FCS) to 3 x 107 cells/ml, and incubated with the mouse monoclonal antibody 4E3 (AbD Serotec, Oxford, UK; 22 µg/ml) for 30 min at 4°C, with mixing at 5-min intervals. The cells were recovered by centrifugation at 150 g for 2 min at 4°C (Hettich rotanta 46R), and resuspended in 500 µl FACS medium. This washing step was repeated twice more before finally resuspending the cells at 3 x 107 cells/ml. The cell suspension was then incubated with goat anti-mouse polyclonal antibody conjugated to the fluorescent molecule R-phycoerythrin (R-PE; Dako UK Ltd., Ely, UK; 80 µg/ml) for 30 min at 4°C, with mixing at 5-min intervals. The cells were then pelleted and washed in FACS medium as before, resuspended at 3 x 107 cells/ml, and stored in the dark at 4°C until analysis. For FACS analysis (a detailed description of the assay is given in Supplemental Fig. 3), an aliquot of cells (50 µl; 1.5x106 cells) was transferred to a FACS tube (Becton Dickinson, Franklin Lakes, NJ, USA) and mixed with five volumes of preheated (37°C) FACS medium and Bodipy-taxol (Invitrogen) at 0.01–1 µM final concentration. Fluorescence data were acquired on a FACScan flow cytometer (Becton Dickinson). Cells were gated for normal size and granularity. The R-PE and Bodipy fluorochromes were excited at 488 nm, and emission spectra were measured using the FL-2 and FL-1 channel, respectively. Flow cytometry data were analyzed using the FlowJo software package (Tree Star, Ashland, OR, USA). Transfected cells were gated for equivalent levels of surface P-glycoprotein according to 4E3 binding, as indicated by the FL-2 fluorescence. The initial rates of Bodipy-taxol accumulation in these cell populations were then measured by observing the increase in FL-1 fluorescence over time. Net flux of Bodipy-taxol into cells was linear at all concentrations of Bodipy-taxol used over this initial phase. The initial linear rate of drug uptake into cells was plotted against drug concentration and fitted best by linear regression analysis (Graphpad Prism; Graphpad Software Inc., San Diego, CA, USA). As the rate of transport was found to be directly proportional to the concentration of drug applied, it is described by the first-order rate law:
 |
Thus, subtraction of the rate constant (given by the slope of the line of best fit) for cells expressing wild-type P-glycoprotein from the rate constant of mock-transfected cells provides a measure of the efflux activity of wild-type P-glycoprotein:
 |
The efflux activity of different mutant P-glycoproteins relative to the wild-type protein is calculated by:
 |
Flow cytometric analysis of cotransfection efficiency
HEK293T cells were transfected with pMDR-wt, or pTracer-CD36, or with a 1:1 (w:w) mixture of both plasmids, and subjected to flow cytometry. Efficiency of transfection by pMDR-wt was assessed by antibody binding [4E3 with a red fluorescent secondary (R-PE)], as described above. Efficiency of transfection by pTracer-Cd36 was assessed by GFP fluorescence. Fluorescence data were acquired on a FACScan flow cytometer (Becton Dickinson), as described above. R-PE and GFP were excited at 488 nm and emitted fluoresence measured using the FL-2 and FL-1 channels, respectively. Data were analyzed using the FlowJo software package (Tree Star).
Preparation of crude membrane fractions from HEK293T cells
For chemical cross-linking experiments, membrane fractions were prepared from HEK293T cells transiently expressing the mutant P-glycoproteins. A confluent monolayer of HEK293T cells from a 75 cm2 flask was harvested as described above and resuspended in 1 ml ice-cold lysis buffer (10 mM Tris-HCl, pH 7.4; 250 mM sucrose; 2 mM CaCl2; 40 µM leupeptin (Sigma-Aldrich, Poole, UK), 2 mM benzamidine (Sigma-Aldrich), 2 µM pepstatin A (Merck Biosciences, Nottingham, UK), 10 mM EDTA). Suspensions were homogenized by three 30 s intervals at high speed on ice using an Ultra-Turrax T8 homogenizer (IKA Labortechnik, Stouden, Germany). Undamaged cells, nuclei, and other large organelles were pelleted by centrifugation at 500 g for 10 min. The crude membrane fraction was isolated from the resulting supernatant by ultracentrifugation in a TLA120.2 rotor (Beckman Coulter UK Ltd., High Wycombe, UK) at 100 000 g for 50 min and resuspended in ice-cold resuspension buffer (10 mM Tris-HCl, pH 7.4; 250 mM sucrose, 40 µM leupeptin, 2 mM benzamidine, 2 µM pepstatin A, 10 mM EDTA).
Chemical cross-linking
Membrane protein fractions (25 µg) in 20 mM Tris-HCl, pH 7.4; 150 mM NH4Cl; 5 mM MgCl2; 40 µM leupeptin; 2 mM benzamidine; and 2 µM pepstatin A were treated with each of three chemical cross-linkers N,N’-o-phenylenedimaleimide (o-PDM; Sigma-Aldrich); N,N’-p-phenylenedimaleimide (p-PDM; Sigma-Aldrich); or 1,6-bismaleimidohexane (BMH; Pierce, Chester, UK) at 0.2 mM final concentration for 5 min at 26°C. The reaction was quenched by addition of β-mercaptoethanol (BDH, Poole, UK) to a final concentration of 17 mM. Where appropriate the membranes were preincubated with AMP-PNP (5 mM; Sigma-Aldrich) or ADP (5 mM; Sigma-Aldrich) and vanadate (Vi; 5 mM; Sigma-Aldrich) for 15 min at 37°C prior to cross linking. The proteins were separated by SDS-PAGE (29)
and analyzed by Western blot, probing with C219 primary antibody (Dako UK Ltd.) and the goat anti-mouse secondary antibody, conjugated to horseradish peroxidase (HRP; Dako UK Ltd.). HRP activity was detected using a chemiluminescence substrate (GE Healthcare Amersham UK Ltd., Little Chalfont, UK) and autoradiography.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The common origin of the MDR class of ABC transporters with both homologous NBDs and homologous TMDs, therefore, permits the unusual domain arrangement of the Sav1866 structure to be tested by chemical cross-linking in P-glycoprotein in a hypothesis driven experiment. In the Sav1866 structure the TMD of one monomer interacts predominantly with the NBD of the other monomer in the homodimeric transport complex. With six TMHs per TMD, these MDR ABC transporters are arranged topologically with two long intracellular loops (ICLs) in each TMD that interact with the NBDs. In Pgp where all four domains are on one polypeptide, ICL1 and ICL2 separate TMH2 and 3, and TMH4 and 5, respectively, and ICL3 and ICL4 separate TMH8 and 9, and TMH10 and 11, respectively (Supplemental Fig. 2A). One of the most dramatic aspects of the Sav1866 structure predicts that ICL4 within TMD2 of P-glycoprotein will interact exclusively with NBD1 (orange-colored residues in Fig. 1E
). In Sav1866, this brings Ser211 and Ser215 within the equivalent loop of one TMD in close proximity to Gln421 and Phe390 of the NBD of the second half-transporter, respectively (Fig. 2A
). Alignment of the sequences of Sav1866 and P-glycoprotein (Supplemental Figs. 1and 2) identified the most likely corresponding residues within P-glycoprotein and led to the prediction that Arg905 and Ser909 of ICL4 in TMD2 should be sufficiently close to allow chemical cross-linking to Ser474 and Leu443 of NBD1, respectively (Fig. 2A
). Site-directed mutagenesis was used to introduce novel cysteine residues into four positions in the fully functional cysteine-free version of P-glycoprotein [Pgp-cys– (26
, 27)
]. The residues in ICL4 and NBD1 predicted to be in close proximity were replaced individually and also in the following pairwise combinations; Leu443Cys with Ser909Cys (L443C+S909C), and Ser474Cys with Arg905Cys (S474C+R905C).
|
Cysteine mutants of P-glycoprotein retain full drug transport activity
To use disulphide cross-linking data to verify quaternary structural predictions, it was necessary to know that the folding of the mutant proteins reflected the native fold of the wild-type protein. Membrane proteins with impaired folding often fail to reach the plasma membrane (28
, 30
, 31)
. Therefore, detection of the mutant proteins on the surface of transiently transfected cells would suggest that the protein is folded correctly. Monoclonal antibody 4E3 recognizes an extracellular loop of P-glycoprotein allowing detection of P-glycoprotein on the surface of live cells by flow cytometry (Fig. 2B
). These data show that each of the mutant proteins was expressed at the cell surface and, because saturating amounts of primary and secondary antibodies were used (data not presented), that each protein trafficked to the cell surface with similar efficiency as wild-type P-glycoprotein.
If the mutant protein present at the cell surface is also functional, this finding would confirm that the communication between the TMDs and the NBDs remained intact and reflected the native situation. To test the activity of the novel cysteine mutants, a new assay of P-glycoprotein function was developed, which measured the initial rate of transport of a fluorescent derivative of the anticancer drug taxol. Taxol has been shown previously to be a transport substrate for P-glycoprotein both in vitro (32)
and in vivo (33)
, as has its fluorescent derivative Bodipy-taxol (34
, 35)
. Transiently transfected HEK293T cells were labeled with 4E3 as before (which does not inhibit P-glycoprotein function) and incubated with Bodipy-taxol (0.1–1.0 µM). The level of accumulation of the fluorescent drug in individual cells was measured by flow cytometry, adjusting the flow rate to analyze 1.5–3 x 103 cells per second. Cells expressing wild-type P-glycoprotein accumulated drug at a measurable, but low, rate because of the active efflux of drug via the transporter. Untransfected and mock-transfected cells and cells expressing an inactive form of P-glycoprotein [the E556Q mutant (36)
] are indistinguishable from each other and accumulate drug at a high rate, demonstrating that the efflux observed in the cells transfected with pMDR-wt is due to P-gp activity. This allowed the rate of accumulation of drug within the population of cells with equal levels of surface P-glycoprotein to be calculated [the initial rate of accumulation was found to be linear for each concentration of drug over the 10-fold range used). This initial linear phase of drug accumulation proved to be a highly sensitive measure of the activity of P-glycoprotein with a 15-fold difference (Fig. 3A
) between cells expressing wild-type P-glycoprotein and mock-transfected cells, or cells expressing inactive P-glycoprotein in which the Walker B glutamate E556 has been replaced by glutamine (36)
. With this minimally invasive assay it was shown that the activity of Pgp-cys–, the template for the cysteine-scanning mutagenesis, was not significantly different from the wild-type protein when studied in live cells.
|
Each of the single cysteine substitution mutants L443C, S474C, R905C, and S909C retain significant levels of drug transport activity, as do the L443C+S909C and S474C+R905C double mutants (Fig. 3B, C
). Statistical analysis of the results by Student’s t test (paired, two-tailed) showed that none of the mutant proteins with introduced cysteines were significantly different to the cysteine-free protein (Table 1
). The efficient trafficking and drug transport activity of all of the cysteine-substitution mutants indicated that their tertiary and quaternary folds reflect that of the wild-type P-glycoprotein and that they are therefore suitable for cross-linking analyses. The activity of the mutants also implies that the physicochemical character of the side chains of these amino acids in the wild-type protein is not critical for function.
|
Chemical cross-linking of introduced cysteines confirms the close proximity of NBD1 and TMD2-ICL4 in P-glycoprotein
Cross-linking across the NBD1:TMD2 interface of P-glycoprotein would constrain the shape of the protein when treated with the ionic detergent SDS. The lariat structure formed by the cross-linked protein would likely retard its electrophoretic mobility through a polyacrylamide gel. Both of the double-cysteine mutants generated (L443C+S909C and S474C+R905C) exhibited distinct new forms following separation of the cross-linked protein species within the samples by SDS-PAGE and detection by Western analyses (Fig. 4
, right-hand panels). The new forms, which migrated more slowly through the gel, were formed with each of the chemical cross linkers of different length, suggesting a degree of flexibility at this interface (we preferred to use chemical cross-linkers rather than induction of disulphide bridging because oxidation of the sample is difficult to control, which leads to a lack of reproducibility). The chemically cross-linked gel retardation was reproducible and dependent on the presence of both cysteines (the single cysteine mutants, shown in the left-hand and central panels of Fig. 4
, fail to form the same structures). However, to rule out the possibility that the gel retardation is the product of intermolecular cross linking between, for example, L443C on one P-glycoprotein with S909C on a second molecule, we cotransfected HEK293T cells with plasmids encoding both the single cysteine forms of the double mutants. Membrane protein samples from the cotransfected cells were subjected to chemical cross-linking, as before, but found not to form the slow migrating protein species (Fig. 5
). This control for intermolecular cross-linking is, of course, valid only if significant numbers of individual cells in the population receive both plasmids during cotransfection. To demonstrate the efficiency of cotransfection, HEK293T cells were transfected with pTracer-Cd36, which expresses GFP, or pMDR-wt, or with both plasmids, and subjected to flow cytometry. Cells transfected with pMDR-wt were detected by surface staining with the anti-P-glycoprotein antibody 4E3 in combination with a red-fluorescent secondary antibody (Fig. 6
A, top left quadrant), and cells receiving pTracer-Cd36 were detected by the green fluorescence of GFP (Fig. 6B
, bottom right quadrant). Following successful cotransfection, cells receiving both plasmids would fluoresce in both the green and red wavelengths and would therefore appear in the top right-hand quadrant of the dot plot (Fig. 6C
). In this control experiment, 70% of the total population of cells received both plasmids, which clearly demonstrates that the cotransfection protocol is very efficient. Taken together with the results of the cross-linking, the simplest interpretation of these data is that L443C is in close proximity to S909C and that S474C is in close proximity to R905C in the tertiary structure of monomeric P-glycoprotein, entirely consistent with the hypothesis based on the crystal structure of Sav1866. NBD1 of P-glycoprotein therefore makes contact with ICL4 of TMD2.
|
|
|
Trapping of P-glycoprotein isoforms with AMP-PNP or ADP.Vi suggests energy is transduced across the NBD1:TMD2-ICL4 interface
To test for conformational changes at this TMD:NBD interface during the ATP catalytic cycle, membrane samples were preincubated with the nonhydrolysable trinucleotide analog AMP-PNP, or ADP and vanadate (ADP.Vi), before addition of cross-linker. Both treatments caused a reproducible reduction in the level of cross-linking between L443C and S909C (fewer slow migrating P-glycoprotein species and more noncross-linked species) with each of the three chemical cross-linkers, although the effect seems more pronounced with BMH (Fig. 7
A). Repetition with the mutant P-glycoprotein with a cysteine pair at positions S474C and R905C showed no evidence of conformational change (Fig. 7B
), providing a convenient negative control for non-specific effects of the added nucleotides.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
. To derive a structure for any membrane protein requires a great deal of manipulation, beginning with perhaps the most potentially deleterious step, solubilization of the protein from its native environment of the lipid bilayer using detergent. During the numerous steps in the process between extraction of the protein from the membrane and structure determination, a dimeric protein like Sav1866 would need to maintain both its tertiary fold and quaternary interactions. This could potentially lead to crystallization of the protein in a nonphysiological conformation. The structure reported for Sav1866 had to be viewed with initial caution because it has some unusual features at both the TMD:TMD and the NBD:TMD interfaces, which are not common to the other available structures of ABC transporters, BtuCD (19)
, HI1470/1 (20)
, and ModBC (21)
. Nor was there supporting biochemical evidence for any of these features. Here, we have tested one aspect of the structure of Sav1866 in the homologous P-glycoprotein, using the reactivity of the sulfhydryl side-chains of cysteines introduced across the unusual interface between TMD2 and NBD1. These changes introduced at the interface between the drug binding and ATP-binding domains might be expected to alter signal and energy transduction within the protein, and therefore it was essential to confirm the functionality of the mutant proteins generated. Function was assessed using a novel assay performed in whole cells. Comparison of the transport kinetics of different P-glycoproteins can be difficult to measure in transfected cells for several reasons. First, the cellular accumulation of drug will be dependent on the amount of P-glycoprotein present in the plasma membrane. The mutants described herein are all expressed and trafficked to the plasma membrane with similar efficiency making this less of an issue. However, having first stained the transfected cells with saturating amounts of anti-P-glycoprotein antibody, the use of a two-color flow cytometric assay permits gating on cells that express equivalent amounts of P-glycoprotein. Differences in drug flux in different populations of cells thus depend on protein activity rather than the number of transporter molecules per cell and the assay would, therefore, be suitable for analyses of different P-glycoproteins expressed at different levels. Second, the uptake of cytotoxic drugs is, by definition, deleterious to the cell and will ultimately have an effect on the function of the active transport protein at the cell surface. By measuring only the initial linear rate of drug accumulation into the cell, the cytotoxic effect of the drug is minimized. This also eliminates the third problem associated with end point assays of drug transporter function, which is that most cytotoxic drugs bind to intracellular targets that act as sinks for drug thereby driving passive influx and rendering the drug unavailable to the transporter for efflux from the cell. End point assays used previously by others and ourselves are, therefore, limited in their sensitivity and unlikely to detect subtle changes in activity. Using our novel assay, linear rates of uptake were recorded over a 10-fold range of drug concentration, generating many data points for statistical analyses and providing an unprecedented sensitivity with a 15-fold difference between P-gp-expressing and nonexpressing cells. Each of the cysteine mutant proteins generated proved to be fully functional for the transport of Bodipy-taxol, strongly suggesting that they share the protein fold of the wild-type protein and validating their use in testing structural predictions in their native environment of the lipid bilayer.
The cross-linking data show that Arg905 and Ser909 in ICL4 of TMD2 are, respectively, in close proximity to Ser474 and Leu443 in NBD1 of P-glycoprotein. This finding is entirely consistent with the most surprising feature of the interdomain architecture suggested by the crystal structure of Sav1866. Previously, we have used cysteine-scanning mutagenesis to provide evidence for modeling P-glycoprotein (37)
on the structure of the bacterial, lipid A exporter, MsbA [since retracted (38)
]. In that study cysteines were introduced into the extracellular end of TMH2 of P-glycoprotein and shown to be close enough to cross-link with cysteines toward the extracellular end of TMH11. Likewise, TMH5 was shown to be close to TMH8. Although the premise of the earlier study (the physiological veracity of the MsbA structure) was flawed, the cross-linking data remains correct because the equivalent TMH in the new Sav1866 structure (TMH2-TMH5' and TMH5-TMH2') form an interface at the extracellular face of the membrane. When TMD1 and TMD2 of P-glycoprotein are aligned with the TMD of Sav1866 to identify the residues equivalent to those replaced with cysteines in the 2003 study (Supplemental Figs. 2B, C), it is apparent that their side chains come within 12Å for TMH2-TMH5' and 4Å for TMH5-TMH2'.
While the previous data cross-linking TMD1 to TMD2 of P-glycoprotein would be unable to distinguish between models based on the retracted MsbA structure and Sav1866, the cross-linking of ICL4 of TMD2 with NBD1 of P-glycoprotein is consistent only with the Sav1866 structure and would not be predictable from the MsbA-based model of P-glycoprotein, or any of the four bona fide structures available for the bacterial ABC importers. The likely veracity, therefore, of the Sav1866 structure has some important implications for the mechanism of action of the MDR class of ABC transporters. First, the proximity of the intracellular loops of the TMDs to the ATP binding pockets of the NBDs suggests that the TMDs could influence directly the structure and movement of NBD motifs involved in ATP binding and hydrolysis and vice versa. With respect to P-glycoprotein, ICL1 and ICL3 contact both NBDs, spanning the nucleotide binding pockets. ICL2 and ICL4 contact only NBD2 and NBD1, respectively, but they come into close proximity to motifs known to be involved in ATP binding and hydrolysis. For example, Ser474, one of the residues mutated to cysteine in this study, is adjacent to the highly conserved Q-loop glutamine. The Q-loop has long been implicated in the allosteric coupling of the two ATP binding pockets because it links the core subdomain containing the Walker A, Walker B, stacking aromatic and H-loop, with the 
-helical subdomain, which contains the ABC signature and D-loop motifs (each ATP binding pocket is a composite site composed of the core subdomain from one NBD and the -helical subdomain of the second [9
, 11
)]. Leu443 is eight amino acids carboxy-terminal from the Walker A motif that makes multiple contacts with the bound nucleotide and has been observed in an inappropriate conformation in some NBDs crystallized in the absence of their TMDs (12
, 39)
. Both Ser474 and Leu443 are, therefore, in close proximity to NBD1 motifs that are likely involved in the allosteric coupling between the two ATP binding pockets and also between the drug transport cycle of the TMDs and the ATP catalytic cycle of the NBDs. Evidence in support of the interaction between ICL4 and NBD1 being a site of energy transduction through which the ATP and drug binding sites are coupled allosterically is provided by the modulation of the cross-linking efficiency between L443C and S909C by the nonhydrolysable ATP analog AMP-PNP, and also by ADP in the presence of vanadate (ADP.Vi) (Fig. 7A
). (It should be noted that the side chains of L443 and S909 are not likely to be involved directly in this energy transduction because they can be replaced with cysteines without reduction in transport activity of the protein.) The allosteric coupling of the two ATP catalytic sites of P-glycoprotein was first demonstrated in 1994 (40)
. The likely mechanism involved, positive cooperativity in both ATP binding (41)
and ATP hydrolysis (7
, 8
, 41
, 42)
, has only become apparent from studies of the behavior of isolated bacterial NBDs in solution. Urbatsch et al., (40)
also demonstrated that the protein is not phosphorylated during catalysis allowing the catalytic cycle to be inhibited by vanadate by replacing the liberated phosphate in the post catalytic conformation. AMP-PNP and ADP.Vi, therefore, trap the protein in different states (the ATP-bound and post hydrolytic states, respectively) that have been shown to be distinct by electron microscopy (43)
. These nucleotides can also alter the conformation of P-glycoprotein in membrane fractions as shown their ability to reduce binding of the conformation-sensitive monoclonal antibody UIC2 (43)
. While it is unlikely that all the P-glycoprotein molecules in a membrane fraction would be accessible to nucleotide (the NBDs of some in the fraction are likely to be intravesicular), the data suggest that in the presence of AMP-PNP or ADP.Vi the L443C and S909C cysteines are less available for cross-linking. The effect is more pronounced with BMH, suggesting that residues may be brought closer together in a less flexible conformation, preventing bifunctional interaction with the longest cross-linker. The differential response of the two cysteine pairs (cross-linking in the S474C and R905C double mutant appears insensitive to added nucleotide), suggests that S474 and R905 may move in tandem during the ATP catalytic cycle, if they move at all.
Sequence conservation within the NBDs suggests that ABC transporters share a similar molecular mechanism for ATP binding and catalysis. The NBDs have been suggested to alternate between a nucleotide-free open conformation and a nucleotide-bound closed conformation with ATP binding and ATP hydrolysis driving the transition between the two states (25
, 44)
. While biochemical differences between different transporters are likely to affect the kinetics of the ATP catalytic cycle, and so the kinetics and efficiency of conformational change at the NBDs (for example the ABCC class of transporter may only require one hydrolysis event to destabilize the closed conformation of the NBDs (45
46
47)
, while P-glycoprotein likely requires both ATPs to be hydrolyzed), the conformational changes at the NBD:NBD interface to which the TMDs respond are likely to be similar. It is less clear whether the proposed common conformational changes at the NBD:NBD interface are coupled to the ligand binding sites of the TMDs in the same way throughout the family. This uncertainty is due mainly to the lack of homology between different classes of TMD but also to the differences apparent at the NBD:TMD interface of the drug exporters and solute importers highlighted in Fig. 1
. However, when the surface of the pink NBD of the NBD:NBD dimer of Sav1866 shown in Fig. 1H
is colored by proximity to any Sav1866 ICL (within 4Å they are colored gold and blue) and compared to the residues of BtuD, which are similarly proximal to the ICLs of BtuC (these are mapped onto this NBD in green and blue), it becomes apparent that there is significant overlap (the blue region). These 14 residues are not highly conserved, but if there is a common conduit within the NBDs for signal transduction, it may well be here. This group is not a linear sequence within the NBD but includes residues close to the Walker A and B motifs and, predominantly, residues in and around the Q-loop (the group also includes the residues corresponding to Ser474 and Leu443 of P-glycoprotein).
In summary, we report that analysis of P-glycoprotein within its native environment of the lipid bilayer by chemical cross-linking of introduced sulfhydryls suggests that ICL4 within TMD2 forms an interface with NBD1. The interface appears to be relevant to the function of the transporter because the level of cross-linking observed is sensitive to added nucleotide. The data are consistent with the veracity of the unusual TMD:NBD interface observed in Sav1866, and it is proposed that the evolutionarily distinct TMDs (at least of bacterial importers and drug exporters) have evolved different mechanisms to energetically couple to common conformational changes at their NBDs.
|
ACKNOWLEDGMENTS |
|---|
Received for publication May 15, 2007. Accepted for publication June 14, 2007.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. A. Aleksandrov, L. Cui, and J. R. Riordan Relationship between nucleotide binding and ion channel gating in cystic fibrosis transmembrane conductance regulator J. Physiol., June 15, 2009; 587(12): 2875 - 2886. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Oancea, M. L. O'Mara, W. F. D. Bennett, D. P. Tieleman, R. Abele, and R. Tampe Structural arrangement of the transmission interface in the antigen ABC transport complex TAP PNAS, April 7, 2009; 106(14): 5551 - 5556. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P H Dixon, S W C van Mil, J Chambers, S Strautnieks, R J Thompson, F Lammert, R Kubitz, V Keitel, A Glantz, L-A Mattsson, et al. Contribution of variant alleles of ABCB11 to susceptibility to intrahepatic cholestasis of pregnancy Gut, April 1, 2009; 58(4): 537 - 544. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. P Locher Structure and mechanism of ATP-binding cassette transporters Phil Trans R Soc B, January 27, 2009; 364(1514): 239 - 245. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. Dietrich, H. Schmuths, C. D. M. Lousa, J. M. Baldwin, S. A. Baldwin, A. Baker, F. L. Theodoulou, and M. J. Holdsworth Mutations in the Arabidopsis Peroxisomal ABC Transporter COMATOSE Allow Differentiation between Multiple Functions In Planta: Insights from an Allelic Series Mol. Biol. Cell, January 1, 2009; 20(1): 530 - 543. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Z. E. Sauna, S. S. Bohn, R. Rutledge, M. P. Dougherty, S. Cronin, L. May, D. Xia, S. V. Ambudkar, and J. Golin Mutations Define Cross-talk between the N-terminal Nucleotide-binding Domain and Transmembrane Helix-2 of the Yeast Multidrug Transporter Pdr5: POSSIBLE CONSERVATION OF A SIGNALING INTERFACE FOR COUPLING ATP HYDROLYSIS TO DRUG TRANSPORT J. Biol. Chem., December 12, 2008; 283(50): 35010 - 35022. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. W. Loo, M. C. Bartlett, and D. M. Clarke Processing Mutations Disrupt Interactions between the Nucleotide Binding and Transmembrane Domains of P-glycoprotein and the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) J. Biol. Chem., October 17, 2008; 283(42): 28190 - 28197. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Pagant, E. Y. Brovman, J. J. Halliday, and E. A. Miller Mapping of Interdomain Interfaces Required for the Functional Architecture of Yor1p, a Eukaryotic ATP-binding Cassette (ABC) Transporter J. Biol. Chem., September 26, 2008; 283(39): 26444 - 26451. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. He, A. A. Aleksandrov, A. W. R. Serohijos, T. Hegedus, L. A. Aleksandrov, L. Cui, N. V. Dokholyan, and J. R. Riordan Multiple Membrane-Cytoplasmic Domain Contacts in the Cystic Fibrosis Transmembrane Conductance Regulator (CFTR) Mediate Regulation of Channel Gating J. Biol. Chem., September 26, 2008; 283(39): 26383 - 26390. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. W. Loo, M. C. Bartlett, and D. M. Clarke Arginines in the First Transmembrane Segment Promote Maturation of a P-glycoprotein Processing Mutant by Hydrogen Bond Interactions with Tyrosines in Transmembrane Segment 11 J. Biol. Chem., September 5, 2008; 283(36): 24860 - 24870. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Shi, X. Chen, Z. Wu, W. Shi, Y. Yang, N. Cui, C. Jiang, and R. W. Harrison cAMP-dependent Protein Kinase Phosphorylation Produces Interdomain Movement in SUR2B Leading to Activation of the Vascular KATP Channel J. Biol. Chem., March 21, 2008; 283(12): 7523 - 7530. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
90° anticlockwise, illustrates that the interdomain contacts of the TMDs are predominantly with the NBD to which they are not attached covalently. This is most easily seen with the gold TMD, which clearly contacts the teal NBD as well as making fewer contacts with the pink NBD to which it is fused. B) Cartoon representation of the vitamin B12 importer BtuCD (pdb 1L7V). The TMDs are shown in red and blue. The NBDs are colored teal and pink. The bound tetra-vanadate is shown in stick format. Each domain in BtuCD is on a separate polypeptide. The view shown, looking down the domain interface of the NBD:NBD dimer similar to that for Sav1866 in (A), illustrates that the symmetry plane of the TMD:TMD interface is more closely aligned with the NBD:NBD interface and that the interdomain contacts of each TMD is solely with one NBD. C) Cartoon representation of the metal-chelate importer HI1470/1 crystallized in the absence of nucleotide or nucleotide analog (pdb 2NQ2). Both the TMDs and NBDs of HI1470/1 are paralogous with BtuCD and so are rendered similarly. D) Cartoon representation of the molybdate importer ModBC (pdb 2ONK). The TMDs, which are nonhomologous with those shown in (A, B, or C), are shown in pale blue and pale green. The NBDs are colored teal and pink. The bound phosphate is shown in stick format. Each domain in ModBC is on a separate polypeptide. E) Surface representation of the Sav1866 NBD:NBD dimer viewed from the perspective of the hidden TMDs. The NBDs are color-coded as before, except for residues within 4Å of ICL1 of the gold TMD (colored yellow) and within 4Å of ICL2 of the gold TMD (colored gold). The bound ADP is shown as spheres (F). Surface representation of the BtuCD NBD:NBD dimer viewed from the perspective of the hidden TMDs. NBDs are colored, predominantly, in teal and pink and the bound tetra-vanadate is shown as sheres. All residues in close proximity to the red TMD are colored as follows: within 4Å of, the amino-terminal tail (green), ICL3 (dark blue), ICL4 (cyan). His92 is close to both the amino terminus and ICL4 of the red TMD and is colored gold. G) Surface representation of the ModBC NBD:NBD dimer viewed from the perspective of the hidden TMDs. NBDs are colored, predominantly, in teal and pink and the bound phosphate is shown in stick form. All residues in close proximity to the pale green TMD are colored green if within 4Å of the amino terminus and yellow if within 4Å of ICL2. H) Surface representation of the Sav1866 NBD:NBD dimer as per (E). The residues on the pink NBD that are within 4Å of the ICLs of the Sav1866 TMDs are colored gold. Following alignment of the NBD of Sav1866 and BtuD, the close contacts made by the BtuC TMD with the BtuD NBD have been mapped onto the Sav1866 NBD:NBD dimer and colored green. Residues common to both data sets of TMD contacts are colored blue. Elemental coloring of bound nucleotides or nucleotide analogues throughout: C, green; N, blue; O, red; P, orange; V, gray.
