MINIREVIEW Multidrug efflux pumps: Substrate selection in ATP-binding cassette multidrug efflux pumps – first come, first served? Robert Ernst 1, *, Petra Kueppers 1 , Jan Stindt 1 , Karl Kuchler 2 and Lutz Schmitt 1 1 Institute of Biochemistry, Heinrich-Heine-University Duesseldorf, Germany 2 Max F. Perutz Laboratories, Medical University Vienna, Austria A brief history of multidrug resistance ATP-binding cassette transporters Multidrug resistance (MDR) strongly impacts the treat- ment of infectious diseases and cancer, as well as the development of effective therapeutics. In principle, there are at least four different mechanisms of how a cell copes with xenobiotics. These include drug inactiva- tion, alteration of the drug target, reduced uptake, and active extrusion of the xenobiotic. The latter is medi- ated by dedicated transmembrane proteins belonging either to the family of secondary transporters such as EmrE [1] or AcrB [2] or to the family of ATP-binding cassette (ABC) transporters such as the human P-glyco- protein (P-glycoprotein, MDR1 or ABCB1) [3] or Pdr5 from Saccharomyces cerevisiae [4]. These primary trans- porters are usually composed of four modules, two nucleotide-binding domains (NBD) and two transmem- brane domains (TMD) that are often encoded by a sin- gle gene in eukaryotes [5]. It is now generally accepted that the TMDs recognize and translocate substrate(s), whereas the NBDs fuel the required conformational changes by nucleotide-dependent dimerization and hydrolysis-dependent dissociation. The low sequence conservation of the TMDs reflects this division of Keywords ABCB1; ABC transporter; ATP hydrolysis; kinetic substrate selection; ligand binding; multidrug efflux; multidrug resistance; Pdr5; TAP; transport mechanism Correspondence L. Schmitt, Institute of Biochemistry, Heinrich Heine University Duesseldorf, Universitaetsstr. 1, 40225 Duesseldorf, Germany Fax: +49 211 81 15310 Tel: +49 211 81 10773 E-mail: lutz.schmitt@uni-duesseldorf.de *Present address Whitehead Institute for Biomedical Research, Cambridge, MA, USA (Received 22 July 2009, revised 1 October 2009, accepted 21 October 2009) doi:10.1111/j.1742-4658.2009.07485.x Multidrug resistance is a major challenge in the therapy of cancer and pathogenic fungal infections. More than three decades ago, P-glycoprotein was the first identified multidrug transporter. It has been studied exten- sively at the genetic and biochemical levels ever since. Pdr5, the most abun- dant ATP-binding cassette transporter in Saccharomyces cerevisiae,is highly homologous to azole-resistance-mediating multidrug transporters in fungal pathogens, and a focus of clinical drug resistance research. Despite functional equivalences, P-glycoprotein and Pdr5 exhibit striking differ- ences in their architecture and mechanisms. In this minireview, we discuss the mechanisms of substrate selection and multidrug transport by compar- ing the fraternal twins P-glycoprotein and Pdr5. We propose that substrate selection in eukaryotic multidrug ATP-binding cassette transporters is not solely determined by structural features of the transmembrane domains but also by their dynamic behavior. Abbreviations ABC, ATP-binding cassette; CFTR, cystic fibrosis transmembrane conductance regulator; MDR, multidrug resistance; NBD, nucleotide- binding domain; PDR, pleiotropic drug resistance; TMD, transmembrane domain. 540 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS labor, whereas the relatively high sequence conservation of the NBDs might imply a conserved mechanism. Mammalian P-glycoprotein was the first ABC trans- porter to be identified as a MDR pump [6] and is nor- mally present at varying levels in every human tissue. Analysis of different cancer cell lines demonstrated massive upregulation of MDR1 expression, which was later shown to mediate acquired resistance of cancer cells against a wide variety of chemotherapeutics. In particular, the purification and reconstitution of P-glycoprotein represented a major breakthrough for a better understanding of P-glycoprotein function because it allowed for detailed mechanistic studies under defined conditions. More recently, structural biology has started to contribute to our understanding of ABC multidrug efflux pumps [7,8]. P-glycoprotein is, among others, the best-characterized eukaryotic ABC transporter, but is most certainly not the only MDR pump in mammals. Over the years it has been demonstrated that multidrug resistance-related pro- tein 1 (MRP1; ABCC1) and ABCG2 or breast cancer resistance protein (BCRP) may be of major importance as mediators of MDR phenotypes [9]. The yeast plasma membrane contains an armory of ABC transporters that act together against structur- ally unrelated xenobiotics, thus forming the executive branch of the pleiotropic drug resistance (PDR) network as a first line of defense [4]. Pdr5, the most abundant yeast ABC transporter in exponentially growing cells, was discovered as a gene product con- ferring cycloheximide, mycotoxin and cerulin resis- tance [4]. Like all members of the PDR subfamily, Pdr5 has an inverted architecture (NBD–TMD– NBD–TMD), and exhibits characteristic sequence deviations from otherwise conserved NBD sequence motifs. We are convinced that a molecular under- standing of MDR will greatly benefit from a side-by- side comparison of P-glycoprotein and Pdr5, two pro- totype eukaryotic ABC transporters, which are func- tionally related, but differ significantly in sequence and architecture. A simplistic model of ABC transporters ABC transporters are present in all kingdoms of life and are responsible for translocating an enormous variety of substrates across cellular membranes. Despite their diverse cellular functions, the fundamen- tal mechanism of transport appears conserved. Accord- ing to the alternating access mechanism [10], the TMDs of a membrane transporter can adopt two con- formations, one inward-facing and one outward-facing, allowing for substrate binding at the cellular mem- brane and for substrate release outside the cell, respec- tively. Figure 1A demonstrates this principle for an ABC transporter. The two conformations of the trans- porter exhibit different substrate affinities, which will ultimately result in a transmembrane gradient of the Outward-facing substrate binding site Membrane Cytosol Inward-facing substrate binding site Fig. 1. The alternating access model. The crystal structure of the MDR transporter Sav1866 (PDB 2hyd) from Staphylococcus aureus on the left is representative for an outward-facing conformation of an ABC transporter. As an example for an inward-facing conformation the struc- ture of P-glycoprotein (PDB 2g5u) is depicted on the right. According to the alternating access model, continuous switching between these extreme conformations, which are believed to have different affinities towards the transported substrate, ultimately results in a transmem- brane substrate gradient. Different colours correspond to different polypeptide chains. The NBDs of Sav1866 are fused to the TMDs, while P-glycoprotein is a full-size transporter. Substrates of MDR transporters are believed to enter the substrate-binding region laterally from the cytosolic leaflet of the membrane. R. Ernst et al. Substrate recognition and transport of Pdr5 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS 541 substrate [11,12]. The continuing conformational changes in the TMDs are tightly coupled to the associ- ation and dissociation of NBDs and are powered by nucleotide binding and hydrolysis. When NBDs are open, nucleotides can be exchanged and the substrate- binding region faces the cytosol (Fig. 2). However, binding of ATP can induce association of the NBDs and a simultaneous switch of the TMDs from the inward-facing to the outward-facing conformation [13– 15]. Of course, such coupled behavior of NBDs and TMDs requires a tight physical interaction. Indeed, the crystal structure of the bacterial MDR transporter Sav1866 [8] revealed the existence of a coupling helix that intercalates with a groove in the adjacent NBDs. Genetic and biochemical evidence from studies on human and fungal drug efflux pumps support the importance of this interaction [16,17]. However, it seems inevitable that the cross-talk between TMDs and NBDs must go beyond a rigid physical linkage. Biochemical analysis of P-glycopro- tein ATPase activity disclosed a basal activity, which can be stimulated several-fold when drug substrates are present [18,19]. Thus, substrate binding is somehow signaled to the sites of catalysis. Remarkably, the transmitted signal is not just a binary one, because the degree of stimulation is different for distinct drugs and can vary between a factor of 1 and 10. It is tempting to propose a role for the coupling helices in this more delicate interdomain cross-talk, because they are in close contact with residues of the X-loop in the oppos- ing NBD [8]. Strikingly, the absence of this sequence motif in ABC transporters mediating substrate uptake implies a different coupling mechanism for importers and exporters. Basal ATPase activity – futile hydrolysis? But why does P-glycoprotein exhibit a low, basal ATPase activity in the absence of substrates? Is it an artifact of purification, a partial uncoupling of NBDs and TMDs, or can it be attributed to an elusive co- purifying endogenous substrate? Based on a thermo- dynamic analysis of P-glycoprotein activity, Al-Shawi et al. [18] argued that the basal activity is an intrinsic property of this MDR pump. This apparently futile ATP hydrolysis is crucial to maximize the number of transporters in the inward-facing conformation. After all, ATP hydrolysis results in dissociation of NBDs, returning the TMDs to the inward-facing conforma- tion. Pdr5 exhibits only basal ATPase activity because, thus far, none of the substrates tested is able to stimulate its activity [20,21]. Such lack of substrate stimulation has also been observed for other fungal ABC transporters such as Cdr1 from Candida albicans [22]. Therefore, it becomes increasingly evident that a substrate-independent, basal ATPase activity has a functional relevance in these multidrug ABC trans- porters. It appears that MDR and PDR ABC trans- porters are not optimized for their energy efficiency but rather for their function. The basal ATPase activity is probably an intrinsic property of multidrug ABC transporters rather than just an artifact of the isolation procedure [18]. i) Substrate binding iii) Substrate release ii) ATP-dependent NBD association v) NBD dissociation vi) Nucleotide exchange iv) ATP hydrolysis Fig. 2. Schematic drawing of the ‘processive clamp model’. In the resting state, the TMDs (rectangles) are in the inward-facing conforma- tion, whereas the NBDs (circles) are open and can exchange nucleotides. Low cellular ADP concentrations (green) favor ATP binding (red) to both NBDs. The association of NBDs is prohibited in the absence of substrate (blue circle). (i) Binding of substrate allows for ATP-dependent association (ii) of the NBDs. Simultaneously, the TMDs switch to the outward-facing conformation and (iii) the substrate dissociates from the low-affinity binding site. Subsequent sequential ATP hydrolysis (iv) results in dissociation of the NBDs (v) and a synchronous resetting of the TMDs. The catalytic cycle closes with the exchange of nucleotides (vi) [15,25]. Substrate recognition and transport of Pdr5 R. Ernst et al. 542 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS A comprehensive comparison of ATPase activities mediated by MDR and PDR transporters will clearly reveal not only conceptual similarities, but also marked differences. Most strikingly, Pdr5 is not a typical ‘canonical’ ABC transporter [21]. By contrast to P-gly- coprotein which contains the consensus sequences of all conserved motifs (Walker-A and -B, C-loop and H- loop) [5], Pdr5 harbors a degeneration within the Walker-A and -B motifs, as well as within the H-loop of the N-terminal NBD and in the C-loop of the C-termi- nal NBD. Based on the generally accepted view of an ATP-bound sandwich dimer of NBDs, this situation would create one active ATPase site and one ‘regula- tory’ site serving as the platform for dimerization. Non- equivalency and asymmetry of the ATPase sites has also been observed to varying degrees for other human ABC transporters such as cystic fibrosis transmembrane con- ductance regulator (CFTR) [23] and TAP [24]. It remains puzzling why some transporters are ‘canonical’, whereas others contain one degenerated ATP-binding site. It has been hypothesized, however, that such an asymmetry provides an energetic and ⁄ or kinetic ‘short- cut’, in which hydrolysis of only one ATP molecule may be sufficient to reset the TMDs to the inward-facing conformation to start a new transport cycle [15,23]. What is the mechanism of substrate transport? The original notion that ATP hydrolysis provides the power stroke for substrate transport has been chal- lenged by Abele & Tampe ´ [13] who first proposed that an ATP-dependent association of the NBDs switches the TMDs from the inward- to the outward-facing conformation. Concurrently, the ATP-switch [14] and the processive clamp model [15,25] suggested that the catalytic cycle proceeds further by substrate release from the low-affinity outward-facing binding site, sequential hydrolysis of two ATP molecules, dissocia- tion of the NBD dimer and simultaneous resetting of the transporter to the inward-facing conformation (Fig. 2). Although the overall scheme appears similar for all models, the details and temporal order of nearly every sub-step is under intense debate. The ATP switch model, for example, predicted ATP hydrolysis at both sites and a completely nucleotide-free intermediate. However, in the case of CFTR, ATP hydrolysis at only one consensus site is sufficient for gating of the chloride channel [23]. Likewise, direct measurements of steady-state ATPase activities of Pdr5 implied that hydrolysis occurs primarily at its consensus site [21]. A strict requirement of ATP hydrolysis at both nucleo- tide-binding sites would render this drug transporter virtually inactive. For ABC importers, an intimate cross-talk between substrate binding and NBDs has been proposed to prevent NBDs from dimerization in the absence of substrates [26]. By contrast, exporters are thought to hydrolyze ATP only after substrate release from the outward-facing binding region [14]. Whether this is indeed true for eukaryotic MDR efflux pumps in general, or whether there are alternative modes of action remains to be established. Binding of highly diverse substrates – a special feature of MDR ABC proteins The most striking feature of MDR pumps is their remarkable spectrum of substrates. How can a single protein such as P-glycoprotein recognize and mediate the vectorial transport of thousands of substrates? What determines the balance of diversity, specificity and selectivity? The principles of key-and-lock and induced fit are probably not applicable here. To date, at least two ligand-binding sites have been identified in P-glycopro- tein [19], referred to as the H-site and the R-site, because certain drugs only compete with Hoechst 33342 trans- port, but not with rhodamine 123 transport and vice versa. Other drugs such as vinblastine do not bind exclu- sively to one of the two sites but instead overlap both. Strikingly, positive cooperativity exists between the R- and H-site as drug binding to either site can stimulate transport from the other. For Pdr5, the fungal equiva- lent of P-glycoprotein, two, or even three independent binding sites have been reported [27,28]. However, the existence of several distinct binding sites does not neces- sarily mean that they form independent structural enti- ties. Structural studies on soluble multidrug receptors and more recently on P-glycoprotein have demonstrated that different ligands can be accommodated by a large, hollow binding pocket that can be divided into sub-sites with preferences for certain different drugs [7,29]. Also the crystal structure of the bacterial MDR pump Sav1866 [8], obtained in the absence of drugs, revealed a large, outward-facing putative substrate-binding region. Hence, a large, hydrophobic drug-binding site is consis- tent with available biochemical and structural data, but we are far from an understanding of how these pumps select their substrates when confronted with a variety of highly diverse substrates. Learning from random and site-directed mutagenesis Because of the lack of high-resolution structures of Pdr5 or related MDR pumps, and the low sequence R. Ernst et al. Substrate recognition and transport of Pdr5 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS 543 conservation of the TMDs, it has been extremely difficult to devise a strategy for site-directed muta- genesis experiments aimed at identifying amino acid residues crucial for substrate binding and transport. Hence, random mutagenesis approaches have turned out to be the method of choice. For a summary of known P-glycoprotein mutants, we refer to http://www. nottingham.ac.uk/  mbzidk/P-gp%20Mutations.htm. A random mutagenesis approach on PDR5 yielded 80 distinct Pdr5 variants, which were functionally clas- sified based on transport phenotypes. Residue changes in Pdr5 can affect folding, trafficking and stability, substrate specificity and inhibitor susceptibility of the transporter [30]. Most remarkable were the observa- tions for S1360 in the hydrophilic face of the predicted TMS11 [21,30]. A S1360F substitution resulted in altered drug transport for only a subset of substrates, such as azole derivatives and cycloheximide. Simulta- neously, the inhibitory effect of the immunosuppres- sant compound FK506 on Pdr5-mediated antifungal resistance was greatly reduced by the S1360F muta- tion. These data imply that substrate binding and transporter inhibition might depend on similar struc- tural features. Other approaches focused on the recognition princi- ple of the substrate-binding region [27,31,32]. Remark- ably, with only one exception [32], these studies did not manipulate Pdr5 itself but instead systematically permutated its potential substrates. In essence, the data suggest that Pdr5-type pumps may harbor two or more substrate-binding sites, as proposed earlier by Kolacz- kowski et al. [28]. Nevertheless, it remains enigmatic how these isolated and ⁄ or overlapping binding sites determine the substrate specificity and selectivity of Pdr5. The diversity of substrates and the low sequence conservation of the TMDs make understanding sub- strate selection extremely challenging, but strikingly, the mechanisms of ATP hydrolysis, performed by highly conserved NBDs, are also controversial. An explicit difference between P-glycoprotein and Pdr5 is that ATP hydrolysis by P-glycoprotein is proposed to alternate between the two ATP-binding sites [18], whereas deviation from consensus sequences in Pdr5 implies that only one ATP-binding site has significant catalytic activity. Such functional asymmetry of the NBDs of Pdr5 has been verified by site-directed muta- genesis [21]. But which residues are catalytically rele- vant and what is the mechanism of ATP hydrolysis? Several studies have highlighted the crucial importance of the carboxylic residues in the Walker-B for coordi- nation of the Mg 2+ ion and catalytic activity [33,34]. Subsequently, two different models for the mechanism of ATP hydrolysis have been proposed: general base catalysis and the substrate-assisted catalysis model [33,35]. Expanding on the latter, the catalytic dyad model was formulated based on functional character- izations of the H-loop in isolated NBDs of human TAP1 and bacterial HlyB [35–37] and, for example, in the assembled, bacterial maltose import system [12]. Because both residues of the putative catalytic dyad are only present in NBD2 of Pdr5, the opposing pre- dictions of the two models could be tested in a fully assembled ABC transporter [21]. These data disproved an essential role of the ‘linchpin’ histidine (H1068) for the steady-state ATPase activity of Pdr5. Very surpris- ingly, however, the H1068A mutant exhibited an altered substrate selection – a result that apparently contradicts the generally accepted picture that the TMDs ‘take care’ of substrate recognition, whereas the NBDs serve ‘only’ as fueling devices. Hence, the data suggest intensive cross-talk between energy consump- tion and substrate selection. Multidrug ABC transporters – ‘normal’ efflux pumps? The above-mentioned simplistic model of an ABC transporter with an inward-facing and an outward- facing conformation is helpful for describing dedicated membrane transporters [10,14,15]. However, it does not explain how MDR ABC transporters select between substrates of diverse structures. This selection process becomes even more fascinating when consider- ing that some drugs may bind to the transporter and interfere with or promote the binding of other drugs [19,38]. At this point, we want to remind the reader of two important facts regarding the nature of MDR substrates. First, many substrates of MDR transport- ers have been shown to cross biological membranes passively at considerable rates, such that even effi- ciently transported substrates may not exhibit steep concentration gradients across the membrane [39]. Sec- ond, the (partial) hydrophobic nature of many sub- strates results in membrane partitioning and it has been shown that multidrug transporters can use this feature by taking up substrates directly from the cyto- solic leaflet of the membrane [40]. What are the factors that determine the substrate selection in MDR ABC transporters? Numerous studies have disclosed some potential key factors. Whereas the charge, the ability to form hydro- gen bonds and the hydrophobicity are crucial factors Substrate recognition and transport of Pdr5 R. Ernst et al. 544 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS for efficient recognition by P-glycoprotein [41,42], the substrates of Pdr5 seem to be selected primarily by their volume (the optimum being about 200–225 A ˚ 3 ) and their ability to form hydrogen bonds [27,43]. It is stunning that these functional homologues and protag- onists of MDR obviously use very different ways to select their substrates. Towards a simplistic model of Pdr5 that includes substrate selection First, one may assume that substrate binding and selection occur at the same time or at least in very close temporal order. Hence, selection occurs in anal- ogy to more specific efflux pumps, during the inward- facing conformation (Fig. 2). However, the possibility remains that substrate release also affects the overall selectivity. Second, Pdr5 exhibits only basal ATPase activity and is not stimulated by substrates. Therefore, substrate binding cannot be rate limiting for the cata- lytic cycle of Pdr5 [20,21]. Third, because high sub- strate concentrations inhibit the ATPase activity of Pdr5, it is fair to assume that substrate release becomes the apparent rate-limiting step under these conditions. A plausible explanation is that non-released substrates may lock the NBDs in a conformation inca- pable of exchanging nucleotides with the surrounding medium (essentially as proposed for P-glycoprotein) [44]. Such inhibition of a transporter by its own sub- strate can be seen to be analogous to product inhibi- tion of enzymes [45] and has been implemented in kinetic models for P-glycoprotein [18,46]. Fourth, the kinetics of substrate–protein interactions, as well as the nucleotide–protein interactions, are interdependent [21]. For a better understanding of substrate selection in multidrug transport systems one has to consider that the hydrophobic drug-binding pocket is not a static structure. Even in the absence of substrates, MDR transporters hydrolyze ATP and the resulting confor- mational changes also drive continuous rebuilding of the substrate-binding site [18,21]. Hence, it is hard to envision that all sorts of substrates can fully equilibrate with the drug-binding site. Given that transport sub- strates can interact with the binding pocket, it seems feasible that the selection of substrates is not realized through a key–lock phenomenon, but instead is a highly dynamic process. It depends on the motion of the ABC transporter, on the order of drug binding events, on dis- tinct affinities towards the binding pocket, on the kinetics of drug binding and release, and so forth. In fact, several studies and observations with Pdr5 have challenged the dogma that solely structural features of the TMDs determine substrate selection. In 1998, Egner et al. [30] identified mutations within the NBDs that affected the drug substrate spectrum. This raises the question of how an amino acid substitution in the highly conserved NBD changes substrate bind- ing in the TMDs. Similar observations by others strengthen this concern [32,47]. Recently, we reported a mutation within the H-loop of NBD2 (H1068A) that had no significant effect on the steady-state ATPase activity, but greatly diminished the transport of rhoda- mine 6G in vitro [21]. Importantly, crystal structures of the isolated NBDs of bacterial HlyB and human TAP1, in their wild-type and mutant forms, showed that a mutation of the H-loop histidine to alanine does not grossly affect the overall NBD fold [35,48]. How can a mutation in close proximity to the c-phosphate of a bound ATP molecule, buried deep within the NBD, affect substrate selection in the TMDs? Interest- ingly, this kind of cross-talk between nucleotide bind- ing and the substrate-binding sites, appears to be bilateral, because some drugs inhibit the ATPase activ- ity of Pdr5 differently from its UTPase ⁄ GTPase activ- ity [43,49]. Most strikingly, ATP hydrolysis powers Pdr5 to transport rhodamine 6G, whereas UTP was unable to do so, even though it is efficiently hydro- lyzed [28]. How does the NBD signal to the TMD, whether ATP, GTP or UTP is bound and hydrolyzed [43]? The kinetic substrate selection model Based on the available literature, we argue that there is no ‘signaling relay’ required to explain these various observations. Instead, we propose that both the kinet- ics of transporter–substrate and transporter–nucleotide interactions affect the substrate selectivity of the MDR transporter Pdr5, and perhaps also affect the basal mode of P-glycoprotein and other ABC transporters. For an easy understanding, one can imagine an over- simplified situation (Fig. 3). A MDR transporter like Pdr5 with only basal ATPase activity switches between its inward-facing and outward-facing conformation, paralleling the rate of ATP binding and hydrolysis. Clearly, a prerequisite for such behavior is a strict mechanical coupling between NBDs and TMDs [8,16]. Let us assume further that two different drugs have identical affinities for the inward-facing substrate- binding site, but do not compete with each other for binding (Fig. 3A,B). One drug, called FAST, displays fast on kinetics and fast off kinetics, whereas the other drug, called SLOW, has slow on and off kinetics. At time zero, the transporter switches to its drug- accepting, inward-facing conformation and both drugs R. Ernst et al. Substrate recognition and transport of Pdr5 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS 545 start to equilibrate with the substrate-binding site (Fig. 3C,D). Thus, it becomes obvious that the rate at which the transporter switches back to the outward- facing conformation determines which of the two sub- strates is transported more efficiently, even though both compounds have the same affinity. If the trans- porter remains in the inward-facing conformation only for a short period (Fig. 3C), the FAST substrate is transported several-fold more efficiently than the SLOW substrate. By contrast, if the transporter switches more slowly (Fig. 3D), both drugs will be transported with virtually identical efficiencies. This simplistic model does not yet include aspects of multi- drug binding and transport, such as the binding of multiple competing drugs, drug release and so forth, which may have further impact on the selection of sub- strates. However, these missing aspects are capable of enhancing even the effects on substrate selection. In summary, the kinetic substrate selection model explains how two distinct substrates with identical or similar affinities can be transported with different effi- ciencies. More importantly, it provides a theoretical framework of how physically coupled NBDs and TMDs can cross-talk by their dynamic behavior: any impact affecting the duration of drug equilibration with the substrate-binding site, whether it is a mutation (e.g. H1068A in Pdr5) or the use of another nucleotide (e.g. UTP instead of ATP), will affect the substrate selection. We would like to emphasize that many substrates may not be dramatically affected by manipulations such as those described here. This could be the case if they equilibrate with their binding site orders of magnitude faster than the multidrug transporter can switch its conformation. However, the concept of kinetic substrate selection could potentially even be extended to other MDR ABC transporters that exhi- bit substrate-stimulated ATPase activities. A testable prediction of our model is that Pdr5 should transport a different set of substrates at low ATP concen- trations ([ATP] <K M ) as opposed to saturating con- centrations ([ATP] >> K M ). It will be exciting to find out whether and how MDR transporters use their dynamic behavior to tune or fine-tune drug transport. Fast conformational changes short drug equilibration time Slow conformational changes long drug equilibration time Time Binding (%) Binding (%) Time AB CD Fig. 3. Kinetic substrate selection in multidrug ABC-transporters. (A) Cartoon depicting a scenario in which a MDR ABC transporter in its inward-facing conformation equilibrates with two transport substrates, namely the drugs FAST (red) and SLOW (blue). Binding of FAST does not promote or interfere with the binding of SLOW, and vice versa. The transporter is not stimulated ⁄ activated by the presence of drugs. (B) Kinetic and thermodynamic parameters of FAST and SLOW. The affinities of FAST and SLOW to the drug-binding site are identical. Both the on and off rates of FAST are higher than for SLOW. (C) Scenario of MDR transporter switching rapidly between the inward- and out- ward-facing conformations because of high basal activity. Association curves for FAST and SLOW are shown in the lower left panel. At the indicated time-point, the transporter switches to its outward-facing conformation and transports whatever is bound. Under given conditions, FAST is transported more efficiently than SLOW. (D) Scenario of MDR transporter switching slowly between its inward- and outward-facing conformations. FAST and SLOW can equilibrate for much longer with inward-facing binding site (time-point as indicated). The transport effi- ciencies for FAST and SLOW are almost identical. Throughout the figure we depicted P-glycoprotein (PDB 3g5u) as representative of the inward-facing conformation. The crystal structure of Sav1866 (PDB 2hyd) represents an outward-facing transporter. Substrate recognition and transport of Pdr5 R. Ernst et al. 546 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS Conclusions In this minireview, we have discussed possible trans- port mechanisms of P-glycoprotein and Pdr5, two pro- totypical eukaryotic multidrug transporters. Although they serve similar functions, their mode of action appears to be quite different. Moreover, we propose that MDR and possibly other multispecific transport- ers exploit ‘kinetic substrate selection’ to distinguish between two substrates of identical affinities. Clearly, this hypothesis needs to be tested and calls for further analysis with high temporal and spatial resolution. The final goal of all these efforts is obvious – a detailed molecular understanding of processes underlying MDR development will open up new ways to over- come a major obstacle in the fight against infectious diseases and cancer. Acknowledgements We apologize to all our colleagues whose work could not be properly cited due to space restrictions. We would like to thank Sander Smits, Carola Hengsten- berg and Wiebke Weinbrenner for critically reading the manuscript. Robert Ernst is a recipient of an EMBO Long Term Fellowship (ALTF 379-2008). Work in our laboratories was supported by the DFG (grants Schm1279 ⁄ 2-3 and 5-3 and SFB 575 project A9) to LS, and by a grant from the Austrian Science Foundation (‘Fonds zur Fo ¨ rderung der wissenschaftli- chen Forschung’ project SFB035-04) to KK. References 1 Tate CG, Kunji ER, Lebendiker M & Schuldiner S (2001) The projection structure of EmrE, a proton- linked multidrug transporter from Escherichia coli,at 7A ˚ resolution. EMBO J 20, 77–81. 2 Seeger MA, Schiefner A, Eicher T, Verrey F, Diederichs K & Pos KM (2006) Structural asymmetry of AcrB trimer suggests a peristaltic pump mechanism. Science 313, 1295–1298. 3 Higgins CF (2007) Multiple molecular mechanisms for multidrug resistance transporters. Nature 446, 749–757. 4 Ernst R, Klemm R, Schmitt L & Kuchler K (2005) Yeast ABC-transporters – cellular cleaning pumps. Methods Enzymol 400, 460–484. 5 Oswald C, Holland IB & Schmitt L (2006) The motor domains of ABC-transporters. What can structures tell us?. Naunyn Schmiedebergs Arch Pharmacol 372, 385–399. 6 Juliano RL & Ling V (1976) A surface glycoprotein modulating drug permeability in Chinese hamster ovary cell mutants. Biochim Biophys Acta 455, 152–162. 7 Aller SG, Yu J, Ward A, Weng Y, Chittaboina S, Zhuo R, Harrell PM, Trinh YT, Zhang Q, Urbatsch IL et al. (2009) Structure of P-glycoprotein reveals a molecular basis for poly-specific drug binding. Science 323, 1718– 1722. 8 Dawson RJ & Locher KP (2006) Structure of a bacte- rial multidrug ABC transporter. Nature 443, 180–185. 9 Haimeur A, Conseil G, Deeley RG & Cole SP (2004) The MRP-related and BCRP ⁄ ABCG2 multidrug resis- tance proteins: biology, substrate specificity and regula- tion. Curr Drug Metab 5, 21–53. 10 Jardetzky O (1966) Simple allosteric model for mem- brane pumps. Nature 211, 969–970. 11 Al-Shawi MK & Omote H (2005) The remarkable transport mechanism of P-glycoprotein: a multidrug transporter. J Bioenerg Biomembr 37, 489–496. 12 Davidson AL, Dassa E, Orelle C & Chen J (2008) Structure, function, and evolution of bacterial ATP- binding cassette systems. Microbiol Mol Biol Rev 72, 317–364. 13 Abele R & Tampe ´ R (2004) The ABCs of immunology: structure and function of TAP, the transporter associ- ated with antigen processing. Physiology (Bethesda) 19, 216–224. 14 Higgins CF & Linton KJ (2004) The ATP switch model for ABC transporters. Nat Struct Mol Biol 11, 918–926. 15 van der Does C & Tampe R (2004) How do ABC trans- porters drive transport? Biol Chem 385, 927–933. 16 Sauna ZE, Bohn SS, Rutledge R, Dougherty MP, Cronin S, May L, Xia D, Ambudkar SV & Golin J (2008) Mutations define cross-talk between the N-termi- nal 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 283, 35010– 35022. 17 Zolnerciks JK, Wooding C & Linton KJ (2007) Evi- dence for a Sav1866-like architecture for the human multidrug transporter P-glycoprotein. FASEB J 21, 3937–3948. 18 Al-Shawi MK, Polar MK, Omote H & Figler RA (2003) Transition state analysis of the coupling of drug transport to ATP hydrolysis by P-glycoprotein. J Biol Chem 278, 52629–52640. 19 Shapiro AB & Ling V (1997) Positively cooperative sites for drug transport by P-glycoprotein with distinct drug specificities. Eur J Biochem 250, 130–137. 20 Decottignies A, Kolaczkowski M, Balzi E & Goffeau A (1994) Solubilization and characterization of the overex- pressed PDR5 multidrug resistance nucleotide triphos- phatase of yeast. J Biol Chem 269, 12797–12803. 21 Ernst R, Kueppers P, Klein CM, Schwarzmueller T, Kuchler K & Schmitt L (2008) A mutation of the H-loop selectively affects rhodamine transport by the R. Ernst et al. Substrate recognition and transport of Pdr5 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS 547 yeast multidrug ABC transporter Pdr5. Proc Natl Acad Sci USA 105, 5069–5074. 22 Shukla S, Rai V, Banerjee D & Prasad R (2006) Char- acterization of Cdr1p, a major multidrug efflux protein of Candida albicans: purified protein is amenable to intrinsic fluorescence analysis. Biochemistry 45, 2425– 2435. 23 Gadsby DC, Vergani P & Csanady L (2006) The ABC protein turned chloride channel whose failure causes cystic fibrosis. Nature 440, 477–483. 24 Chen M, Abele R & Tampe R (2004) Functional non- equivalence of ATP-binding cassette signature motifs in the transporter associated with antigen processing (TAP). J Biol Chem 279, 46073–46081. 25 Janas E, Hofacker M, Chen M, Gompf S, van der Does C & Tampe R (2003) The ATP hydrolysis cycle of the nucleotide-binding domain of the mitochondrial ATP- binding cassette transporter Mdl1p. J Biol Chem 278, 26862–26869. 26 Oldham ML, Khare D, Quiocho FA, Davidson AL & Chen J (2007) Crystal structure of a catalytic inter- mediate of the maltose transporter. Nature 450, 515– 521. 27 Golin J, Ambudkar SV, Gottesman MM, Habib AD, Sczepanski J, Ziccardi W & May L (2003) Studies with novel Pdr5p substrates demonstrate a strong size depen- dence for xenobiotic efflux. J Biol Chem 278, 5963– 5969. 28 Kolaczkowski M, van der Rest M, Cybularz-Kol- aczkowska A, Soumillion JP, Konings WN & Goffeau A (1996) Anticancer drugs, ionophoric peptides, and steroids as substrates of the yeast multidrug transporter Pdr5p. J Biol Chem 271, 31543–31548. 29 Schumacher MA, Miller MC & Brennan RG (2004) Structural mechanism of the simultaneous binding of two drugs to a multidrug-binding protein. EMBO J 23, 2923–2930. 30 Egner R, Rosenthal FE, Kralli A, Sanglard D & Kuch- ler K (1998) Genetic separation of FK506 susceptibility and drug transport in the yeast Pdr5 ATP-binding cas- sette multidrug resistance transporter. Mol Biol Cell 9, 523–543. 31 Conseil G, Perez-Victoria JM, Renoir JM, Goffeau A & Di Pietro A (2003) Potent competitive inhibition of drug binding to the Saccharomyces cerevisiae ABC exporter Pdr5p by the hydrophobic estradiol-derivative RU49953. Biochim Biophys Acta 1614, 131–134. 32 Tutulan-Cunita AC, Mikoshi M, Mizunuma M, irata D & Miyakawa T (2005) Mutational analysis of the yeast multidrug resistance ABC transporter Pdr5p with altered drug specificity. Genes Cell 10, 409–420. 33 Moody JE, Millen L, Binns D, Hunt JF & Thomas PJ (2002) Cooperative, ATP-dependent association of the nucleotide binding cassettes during the catalytic cycle of ATP-binding cassette transporters. J Biol Chem 277, 21111–21114. 34 Urbatsch IL, Julien M, Carrier I, Rousseau ME, Cayrol R & Gros P (2000) Mutational analysis of conserved carboxylate residues in the nucleotide bind- ing sites of P-glycoprotein. Biochemistry 39, 14138– 14149. 35 Zaitseva J, Jenewein S, Jumpertz T, Holland IB & Schmitt L (2005) H662 is the linchpin of ATP hydroly- sis in the nucleotide-binding domain of the ABC trans- porter HlyB. EMBO J 24, 1901–1910. 36 Ernst R, Koch J, Horn C, Tampe R & Schmitt L (2006) Engineering ATPase activity in the isolated ABC cassette of human TAP1. J Biol Chem 281, 27471– 27480. 37 Zaitseva J, Jenewein S, Wiedenmann A, Benabdelhak H, Holland IB & Schmitt L (2005) Functional charac- terization and ATP induced dimerization of the isolated ABC-domain of the haemolysin B transporter. Biochem- istry 44, 9680–9690. 38 Borst P, Zelcer N, van de Wetering K & Poolman B (2006) On the putative co-transport of drugs by multidrug resistance proteins. FEBS Lett 580, 1085– 1093. 39 Seelig A (2007) The role of size and charge for blood– brain barrier permeation of drugs and fatty acids. J Mol Neurosci 33, 32–41. 40 Bolhuis H, van Veen HW, Molenaar D, Poolman B, Driessen AJ & Konings WN (1996) Multidrug resis- tance in Lactococcus lactis: evidence for ATP-dependent drug extrusion from the inner leaflet of the cytoplasmic membrane. EMBO J 15, 4239–4245. 41 Chiba P, Burghofer S, Richter E, Tell B, Moser A & Ecker G (1995) Synthesis, pharmacologic activity, and structure–activity relationships of a series of propafe- none-related modulators of multidrug resistance. J Med Chem 38, 2789–2793. 42 Seelig A (1998) A general pattern for substrate recognition by P-glycoprotein. Eur J Biochem 251, 252– 261. 43 Golin J, Ambudkar SV & May L (2007) The yeast Pdr5p multidrug transporter: how does it recognize so many substrates? Biochem Biophys Res Commun 356, 1–5. 44 Aanismaa P & Seelig A (2007) P-Glycoprotein kinetics measured in plasma membrane vesicles and living cells. Biochemistry 46, 3394–3404. 45 Pall ML & Kelly KA (1971) Specificity of transinhibi- tion of amino acid transport in neurospora. Biochem Biophys Res Commun 42, 940–947. 46 Gatlik-Landwojtowicz E, Aanismaa P & Seelig A (2006) Quantification and characterization of P-glycoprotein–substrate interactions. Biochemistry 45, 3020–3032. Substrate recognition and transport of Pdr5 R. Ernst et al. 548 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS 47 Jha S, Dabas N, Karnani N, Saini P & Prasad R (2004) ABC multidrug transporter Cdr1p of Can- dida albicans has divergent nucleotide-binding domains which display functional asymmetry. FEMS Yeast Res 5, 63–72. 48 Procko E, Ferrin-O’Connell I, Ng SL & Gaudet R (2006) Distinct structural and functional properties of the ATPase sites in an asymmetric ABC transporter. Mol Cell 24, 51–62. 49 Conseil G, Decottignies A, Jault JM, Comte G, Barron D, Goffeau A & Di Pietro A (2000) Prenyl-flavonoids as potent inhibitors of the Pdr5p multidrug ABC trans- porter from Saccharomyces cerevisiae. Biochemistry 39, 6910–6917. R. Ernst et al. Substrate recognition and transport of Pdr5 FEBS Journal 277 (2010) 540–549 ª 2009 The Authors Journal compilation ª 2009 FEBS 549 . MINIREVIEW Multidrug efflux pumps: Substrate selection in ATP-binding cassette multidrug efflux pumps – first come, first served? Robert Ernst 1, *,. within the NBD, affect substrate selection in the TMDs? Interest- ingly, this kind of cross-talk between nucleotide bind- ing and the substrate- binding