Does the ATP-bound EQ mutant reflect the pre- or post-ATP hydrolysis state in the catalytic cycle of human P-glycoprotein (ABCB1)?

Abstract

P-glycoprotein (P-gp, ABCB1) is an ABC transporter associated with the development of multidrug resistance to chemotherapy. During its catalytic cycle, P-gp undergoes significant conformational changes. Recently, atomic structures of some of these conformations have been resolved using cryo-electron microscopy. The ATP hydrolysis-defective mutant of the catalytic glutamate residue of the Walker B motif (E556Q/E1201Q) has been used to determine the structure of the ATP-bound inward-closed conformation of P-gp. Here we show that this mutant does not appear to undergo the same steps as wild-type P-gp. We discuss conformational differences in the EQ mutant that may lead to a better understanding of the catalytic cycle of P-gp and propose that additional structural studies with wild-type P-gp are required.

Introduction

Multidrug resistance in cancer cells has been associated mainly with the overexpression of three ATP-binding cassette (ABC) efflux pumps, P-glycoprotein (P-gp or ABCB1), the breast cancer resistance protein (BCRP or ABCG2) and multidrug resistance-associated protein 1 (MRP1 or ABCC1) [1, 2]. Cancer cells that overexpress these transporters are generally associated with poor patient prognosis, and unfortunately no drug that has been developed to block their transport function has yet reached the clinic. P-gp, one of the most studied full-length ABC transporters, is comprised of two transmembrane domains (TMDs) and two nucleotide-binding domains (NBDs). Each TMD has six homologous transmembrane helices (TMHs) and, together, they form the substrate-binding site and translocation pathway.

The transport function of P-gp is energetically driven by the hydrolysis of ATP and requires conformational changes that begin in the NBD region upon ATP-binding and are then translated across the protein to the transmembrane region, resulting in the efflux of substrates bound to the substrate-binding pocket in the TMDs. The protein transitions from an inward-open (IO) conformation to an inward-closed (IC) conformation. In the IO conformation, the NBDs are separated from each other, and the transmembrane region offers a large cavity able to accommodate a wide variety of amphipathic or hydrophobic ligands. In contrast, in the IC conformation the NBDs dimerize, and the transmembrane region is rearranged, expelling the substrate from the cell [1]. It is clear that this is an oversimplified model and that transporters actually undergo many different conformational changes during the ATP-coupled transport cycle, as recently shown for the bacterial transporters Atm1 and TmrAB, for which at least eight different conformations were solved using X-ray crystallography and cryo-electron microscopy (cryo-EM), respectively [3, 4]. For instance, it has been proposed that the ADP and vanadate (V_i)-trapped conformation of P-gp resembles the transition state during the hydrolysis of ATP [5], yet such a conformation has yet to be solved.

In the case of P-gp, the structures of some of the snapshots taken of conformational states occurring during the transport cycle have been solved either by cryo-EM or X-ray crystallography, using human, mouse and chimeric P-gp. Of note, the structure of the apo conformation (i.e., in the absence of any transport substrate or ATP) has only been solved for mouse P-gp [6–8]. The substrate accesses the drug-binding pocket, located in the transmembrane region, directly from within the membrane. Although biochemical and structural studies have shown that the plasticity of the binding site allows accommodation of ligands in different binding modes [9, 10], it remains unclear what specific conformational changes occur in the transmembrane region to accommodate ligand binding, such as bending or straightening of the helices, rotation or translation. Given the plasticity of the TMD, each ligand might have a unique effect on the binding site.

The structures of the substrates taxol and vincristine as well as those of the inhibitors elacridar, zosuquidar and tariquidar bound to human P-gp have been solved [11, 12]. A major difference between the mouse P-gp structures solved by X-ray crystallography and the human P-gp structures is the bending of TMH4 and TMH10. An ATPase-deficient mutant in which two conserved catalytic glutamate residues in the Walker B domain of the two ATP sites (E556Q and E1201Q) were substituted with glutamine (known as EQ mutant P-gp) has been particularly useful to solve the structures of the IC ATP-bound as well as the IO inhibitor-bound P-gp conformations [11, 13]. Although the ATP-bound structure of EQ mutant P-gp was determined in the presence of the ligand vinblastine, the ligand was not found in the binding site [13]. This was interpreted to mean that the substrate is effluxed from the cell before ATP is hydrolyzed. However, such an interpretation is controversial, as previous findings suggested that the transport occurs after ATP is hydrolyzed, calling into question whether the ATP-bound EQ mutant P-gp structure actually represents the pre-hydrolysis state of wild-type (WT) P-gp. Furthermore, biochemical studies as well as double electron electron resonance (DEER) spectroscopy studies [14, 15] suggest that this mutant might represent an intermediate conformational state during the catalytic cycle compared to WT P-gp.

In this brief report, we highlight the published findings with this mutant since it was first described in 2002 and present our own additional biochemical studies. We discuss possible implications concerning interpretation of structural studies with the EQ mutant. We propose that additional structural and biochemical studies at the single molecule level with WT P-gp and the EQ mutant are required to conclusively determine whether the conformation of the EQ mutant reflects the pre- or post-hydrolysis step in the catalytic cycle.

Materials and methods

Chemicals

Dodecyl-β-D-Maltoside (DDM) and cholesteryl hemisuccinate (CHS) were obtained from Anatrace (Maumee, OH). Ni-Nitrilotriacetic acid (Ni-NTA) agarose was purchased from Qiagen (Germantown, MD). E. coli polar lipid extract was purchased from Avanti Polar Lipids, Inc. (Alabaster, AL). Bio-Beads SM-2 Resin, size exclusion column protein standards and empty glass columns were purchased from Bio-Rad (Hercules, CA). Other reagents including TPCK trypsin, and trypsin inhibitor were obtained from Sigma-Aldrich (St. Louis, MO).

WT and EQ mutant P-gp expression and purification

WT and the E556Q and E1201Q mutant P-gps were expressed, purified and reconstituted in nanodiscs as previously described [16]. Briefly, High Five insect cells (Thermo-Fisher Scientific, Waltham, MA) were infected with recombinant baculovirus (BV) carrying either the BV-MDR1(H₆) gene (WT P-gp) or the BV-MDR1-EQ(H₆) gene (EQ mutant P-gp) at a multiplicity of infection of 10. Cells were harvested 54-60 h post infection. Total membrane vesicles were prepared by hypotonic homogenization followed by differential centrifugation.

WT or EQ mutant P-gp-expressing membranes were solubilized with a solution containing DDM/CHS detergents. Ni-NTA affinity chromatography followed by size exclusion chromatography were used for purification of the solubilized proteins. Purified P-gp was reconstituted into nanodiscs by combining the purified protein with the MSP1D1 belt protein and E. coli phospholipids in a 1:4:200 ratio. Bio-beads were used to remove the detergent. Size exclusion chromatography was used for purification of nanodiscs containing a single P-gp molecule, using 40 mM Hepes, pH 7.3, 100 mM NaCl and 5 mM DTT as the elution buffer.

Trypsin treatment of WT and EQ mutant P-gp in membrane vesicles

Membrane vesicles (40 μL, 25 μg protein) containing either WT or EQ mutant P-gp were combined with increasing amounts of trypsin (0-2 μg) for 5 min at 37°C. Soy trypsin inhibitor was added to the sample (25 μg) and the samples were incubated for 2 min at 37°C after which 10 μL of 5x loading dye were added and incubated for 20 more min at 37°C. Samples (10 μg total protein/lane) were loaded in a 7% Tris-acetate buffer (Thermo Fisher Scientific, Waltham, MA) and run according to the manufacturer’s instructions. InstantBlue® (Millipore-Sigma, Burlington, MA) was used to stain the corresponding bands. ImageJ (NIH, Bethesda, MD) was used for quantification of the intensity of each band. GraphPad Prism 9.1 (GraphPad Software, San Diego, CA) was used for data plotting and analysis.

Thermal stability of WT P-gp reconstituted in nanodiscs

The thermal stability of WT-Pgp reconstituted in nanodiscs was determined using a protocol similar to one described previously [17]. Briefly, samples of P-gp reconstituted into nanodiscs (0.5 μg total protein) were resuspended in 50 mM MES-Tris pH 6.8 containing 50 mM KCl, 5 mM NaN₃, 1 mM EGTA, 1 mM ouabain and 2 mM DTT in the presence of 5 mM ATP or 10 mM MgCl₂. To determine vanadate sensitivity, another set was treated in the same buffer with 0.3 mM sodium orthovanadate (V_i). Samples were incubated at different temperatures (37-72°C) for 10 min in a C1000 Touch Thermal Cycler (Bio-Rad, Hercules, CA). After this, ATP or Mg²⁺ (5 or 10 mM respectively), whichever was missing during the first step, were added and ATP hydrolysis was allowed to take place for 20 min at 37°C. Subsequently, the reaction was terminated and the amount of inorganic phosphate (P_i) produced was quantified as previously described [17]. The amount of P_i produced by P-gp-mediated ATP hydrolysis was calculated as the difference between the amount in the absence and presence of V_i. The highest value was set as 100%. GraphPad Prism 9.1 was used to process the data.

UIC2-Fab binding to purified WT and EQ mutant P-gp in nanodiscs

UIC2-Fab was prepared using the Pierce™ Fab preparation kit (Thermo Fisher Scientific, Waltham, MA) according to the manufacturer’s instructions. Nanodiscs containing WT or EQ mutant P-gp (3 μg, 15 nmol) were incubated with UIC2-Fab (1 μg, 20 nmol, 18 μL final volume) for the indicated periods of time ranging from five to forty minutes at 37°C. Next either 5 mM ATP, 10 mM MgCl₂, 5 mM ATP-γ-S, 5 mM AMP-PMP, and/or 0.5 mM V_i were added and the sample (20 μL final volume) was further incubated for 5 to 30 min at 37°C. Experiments in which the ratio between the P-gp and the UIC2-Fab as well as the order of addition was changed were performed. Either ATP, MgCl₂, or other nucleotides and/or V_i were added first and the UIC2-Fab was added second; specific details are presented in the corresponding figures. NativePAGE™ Sample Buffer (Thermo Fisher Scientific, Waltham, MA) was added and samples were loaded in a NativePAGE™ 4-16% Bis-Tris gel, which was run according to the manufacturer’s instructions. Protein bands were stained with InstantBlue® stain, quantified and analyzed as described previously [17]. GraphPad Prism 9.1 was used to process the data. Two-way ANOVA statistical tests were used to show significance among compared data.

Molecular modeling of UIC2 binding to WT and EQ mutant P-gp

Computer molecular modeling and picture creation were done with UCSF Chimera [18]. The structure PDBID: 6QEX [9] was used for the IO conformation, and PDBID: 6C0V [10] was used for the EQ mutant in the ATP-bound IC state. As extracellular loop 1 (ECL1) residues 81-104 are not resolved in the 6C0V structure, this portion was grafted on from 6QEX by superimposing the outer ends of transmembrane helices 1 and 2. The ECL1-bound UIC2-Fab in 6QEX was also transferred to 6C0V to compare the antibody binding sites of these two structures.

Results

We first reported characterization of the EQ mutant in 2002. We have shown that the expression and purification of EQ mutant P-gp is comparable to that of the WT protein [16]. In light of the recent cryo-EM studies on WT P-gp and the EQ mutant, we compared their susceptibility to mild trypsin treatment and their reactivity to the Fab of the conformation-sensitive antibody UIC2 so as to further interrogate the differences between the conformations.

ATP binding favors the dimerization of NBDs

Previously we have shown using membrane vesicles containing P-gp that binding of ATP increases the thermal stability of P-gp [17]. Furthermore, our previous findings indicated that this stabilization occurs in an ATP concentration-dependent manner, and that a concentration of 5 mM is more than sufficient to achieve maximum stability [17]. Fig. S1 shows that the increased thermostability previously shown using WT P-gp in membrane vesicles is also observed when the purified protein is reconstituted into lipidic nanodiscs. We found that the temperature at which 50% of the protein is inactivated (IT₅₀) in the absence of ATP is 42.69 ± 0.76°C, whereas in the presence of ATP it increases to 64.91 ± 0.34°C (~22°C increase). We interpret this change in thermal stability to be caused by the change in conformation that occurs upon ATP binding and dimerization of the NBDs when the molecule transitions from an inward-open (IO) to inward-closed (IC) conformation. Since EQ mutant P-gp is unable to hydrolyze ATP, a similar experiment with EQ mutant P-gp could not be performed.

WT or EQ mutant P-gp in the IO apo-configuration as well as in the IC ATP-bound conformation show a similar sensitivity to trypsin cleavage

Previous studies have shown that silent polymorphisms of P-gp can affect the trypsin digestion rate [19, 20], suggesting a change in the overall conformation of the haplotype with three SNPs as compared to the WT transporter. Interestingly, we found that this was not the case for WT and EQ mutant P-gp. The susceptibility to trypsin of both proteins in inside-out oriented membrane vesicles was comparable in the absence of any nucleotide (Fig. 1A). Furthermore, the trypsin degradation of WT and EQ mutant P-gp was not affected by the presence of ATP, Mg²⁺ and/or V_i. (IC₅₀ values range from 0.2 to 0.5 μg trypsin and are given in Fig. 1B and C). This suggests that differences in the IO or IC conformations of either the WT or EQ mutant do not affect the susceptibility to trypsin digestion. It has been shown previously that treatment with trypsin at low concentrations cleaves P-gp first at residue R680 in the linker region between the first NBD and TMH7 [21]. Interestingly, the structure of this linker region has not been solved for P-gp from any species in any conformation to date. Our results indicate that the trypsin cleavage site in the linker region is equally accessible to trypsin in both WT and EQ mutant P-gp in the presence and absence of ATP.

Fig. 1. WT and EQ mutant P-gp are equally sensitive to trypsin digestion.

The sensitivity of WT and EQ mutant P-gp under various conditions as a function of trypsin concentration was determined by combining membrane vesicles containing the corresponding protein with or without nucleotides, magnesium or vanadate for 5 min at 37°C followed by treatment with the indicated concentration of trypsin for another 5 min at 37°C. The reaction was stopped by addition of excess soybean trypsin inhibitor. Samples were denatured and PAGE was performed (10 μg protein per lane). P-gp bands were quantified as described in the methods section. (A) Representative images showing the WT or EQ mutant P-gp band disappearing as the amount of trypsin increases. The sensitivity of WT and EQ mutant P-gp (B) in the absence of any nucleotide or (C) in the presence of ATP, Mg²⁺ and/or V_i as a function of trypsin concentration. The IC₅₀ values in parentheses represent mean ± SD from three experiments.

WT and EQ mutant P-gp show different abilities to bind UIC2 under certain conditions

UIC2 is a conformation-sensitive antibody that binds to the extracellular region of P-gp [22–24]. Specific P-gp conformations can be trapped selectively [25]. For example, ATP-γ-S and APM-PNP are non-hydrolysable ATP analogs that have been proposed to trap P-gp in an IC conformation [26, 27]. The effect of different conformations of P-gp on the binding of UIC2 has been reported previously [28, 29]. Here we compared the ability of WT and EQ mutant P-gp to bind the UIC2-Fab, when trapped in different conformations. The differences in the binding to UIC2 are a consequence of the relative movement of the extracellular loops during the transport-coupled ATP hydrolysis cycle. Fig. 2A shows the binding of UIC2-Fab to WT P-gp reconstituted in nanodiscs under various conditions. In the absence of any nucleotide UIC2-Fab remains bound to the protein (lane 2). Addition of ATP alone in the absence of Mg²⁺ drives the molecule to form an IC configuration (lane 3). We have previously found that binding of ATP under non-hydrolytic conditions (that is, in the absence of Mg²⁺) makes P-gp significantly more thermostable compared to the apo conformation [17]. Here we show that binding of ATP alone in the absence of Mg²⁺ does not result in dissociation of UIC2-Fab from WT P-gp. Furthermore, addition of ATP-γ-S or AMP-PNP alone yields results comparable to those with ATP (Fig. 2A, lanes 4 and 5, respectively). These results taken together with the previously published thermostability data [17] suggest that the pre-hydrolysis conformation of WT P-gp corresponds to an IC conformation that is still able to bind UIC2-Fab in the extracellular region, possibly resembling the previously proposed inward-closed-outward-occluded (IC-OO) conformation [14]. Addition of ATP and Mg²⁺ to WT P-gp results in a protein that is constantly undergoing hydrolysis, and no UIC2-Fab dissociation is observed (Fig. 2A, lane 6). In contrast, addition of ATP-γ-S or AMP-PNP and Mg²⁺ would be expected to result in a pre-hydrolysis conformation since these nucleotide analogs are not hydrolyzed. Interestingly, we see a dramatic difference when comparing the dissociation of the UIC2-Fab from P-gp with ATP-γ-S and AMP-PNP both in the presence of Mg²⁺ (lanes 7 and 8 respectively). We observed partial dissociation of the UIC2-Fab in the presence of AMP-PNP/Mg²⁺. Considering that both molecules should arrest the protein in the pre-hydrolysis conformation, this is an unexpected result. Furthermore, addition of ATP, Mg²⁺ and V_i, which should arrest WT P-gp in a post-hydrolysis conformation, favors the partial dissociation of UIC2-Fab (Fig. 2A, lane 9), in a similar way as AMP-PNP. The dissociation of the UIC2-Fab in the presence of ATP, Mg²⁺ and V_i is in agreement with previous studies done with WT P-gp [28, 29].

Fig. 2. Binding of UIC2-Fab to WT and EQ mutant P-gp during steps of the catalytic cycle.

Nanodiscs containing purified WT (A) or EQ mutant P-gp (B) were incubated with UIC2-Fab for 10 min at 37°C. Next ATP, ATP-γ-S, AMP-PNP, Mg²⁺ and/or V_i were added and incubated for 30 more min at 37°C. Samples were loaded in a native gel and the degree of dissociation of the UIC2-Fab was quantified. (C) Histograms showing the percentage of protein that remains bound to the UIC2-Fab. Please note that no histogram is shown for lane 1 where only the P-gp nanodisc sample was loaded. Experiments were performed at least three times and values are expressed as mean ± SD.

Next, we explored the effect of the same conditions on the dissociation of UIC2-Fab from EQ mutant P-gp, as shown in Fig. 2B. The results were strikingly different for many of the conditions (Fig. 2B). Addition of ATP or ATP-γ-S and no Mg²⁺ results in partial dissociation of the UIC2-Fab from the protein (lanes 2 and 3, respectively). AMP-PNP in the absence of Mg²⁺ shows no UIC2-Fab dissociation for either molecule (lane 4). This suggests that ATP-γ-S and AMP-PNP at 5 mM and in the absence of Mg²⁺ cause WT and EQ mutant P-gp to form different conformations. The presence of ATP/Mg²⁺ also yields partial dissociation, and so do ATP-γ-S/Mg²⁺, AMP-PNP/Mg²⁺ and ATP/Mg²⁺/V_i (lanes 6-9). Hence, we observed that EQ mutant P-gp, except under one condition (the presence of AMP-PNP and no Mg²⁺), directly transitions from the IO conformation to a combination of IC-OO and inward-closed-outward-facing (IC-OF), unlike its WT counterpart, which transitions to a predominantly IC-OO conformation before reaching a state in which both conformations are observed. Figure 2C shows a quantification of WT or EQ mutant P-gp that remains bound to the UIC2-Fab. Clearly, specific conditions, as well as the nature of the protein (WT or mutant) significantly affect the ratio between the unbound and bound.

Changing the order of addition (nucleotide or nucleotide derivative, Mg²⁺, and/or V_i first and UIC2-Fab second) did not affect the overall outcome of the results (gels are presented in Fig. S2A and B and quantification of the corresponding protein bands in Fig. S2C and 2D for WT and EQ mutant P-gp, respectively). Similarly, varying the incubation time or the P-gp-UIC2-Fab ratio did not alter the outcome (data not shown). Addition of 50 μM verapamil, a known P-gp substrate only affected the outcome for the EQ mutant P-gp, which showed no UIC2-Fab dissociation in the presence of AMP-PNP and Mg²⁺ (Fig. 3A and 3C for WT P-gp and 3B and 3D for EQ mutant P-gp), suggesting that the presence of a substrate may also affect the transition from the IC-OO to the IC-OF conformation under certain conditions.

Fig. 3. Effect of P-gp substrate verapamil on the binding of UIC2-Fab to WT and EQ mutant P-gp during steps of the catalytic cycle.

Nanodiscs containing purified WT (A) or EQ mutant P-gp (B) were incubated in the presence of 50 μM verapamil and UIC2-Fab for 10 min at 37°C. Next ATP, ATP-γ-S, AMP-PNP, Mg²⁺ and/or V_i were added and incubated for 30 more min at 37°C. Samples were loaded in a native gel and the degree of dissociation of the UIC2-Fab was quantified. Histograms show the percentage of WT protein (C) or EQ mutant P-gp (D) that remains bound to the UIC2-Fab. For comparison, red bars indicate values in the presence of verapamil and black bars values in its absence, as shown in Fig. 2.

To further understand conformational changes resulting in the dissociation of UIC2-Fab from P-gp, we compared the extracellular loops in the currently available structures of human P-gp in the IO and IC conformations: PDBID: 6QEX and 6C0V for WT IO and EQ mutant IC, respectively. That both WT and EQ mutant P-gp bind UIC2 when in the IO state (See Fig. 2) indicates that both proteins are indeed in the IO, UIC2-bound, 6QEX conformation shown in Fig. 4A. To understand the lack of binding in the IC state of the EQ mutant, we had to fill in the unresolved portion of extracellular loop (ECL) 1 of the 6C0V structure with the analogous segment from the WT (see Methods). As shown in Fig. 4B, UIC2 bound to ECL1 of the EQ mutant structure is unable to simultaneously bind ECL3 and ECL4 as it does in the IO conformation (Fig. 4A), suggesting a reduced binding affinity.

Fig. 4. Binding of UIC2-Fab to the extracellular region of WT and EQ mutant P-gp.

(A) Cryo-EM structure (PDBID: 6QEX) of UIC2-Fab bound to the inward-open conformation of WT P-gp. The protein is shown as a gray ribbon, with only ECLs 1, 3 and 4 depicted for clarity. The light and heavy chains of UIC2 are shown as purple and magenta volumes. (B) Comparison model of UIC2 bound to the extracellular loops of EQ mutant P-gp in the inward-closed conformation (based on PDBID: 6C0V, Cryo-EM structure and the UIC2-ECL1 complex in PDBID: 6QEX).

Discussion

During the transport cycle, P-gp transitions through a series of conformations. Although some of the X-ray and cryo-EM structures have provided clues as to how these conformations might look, there are still many unanswered questions.

The ATP-binding site is formed by the A-loop, Walker A, Walker B, D and H loops of one NBD and the LSGGQ signature motif of the other NBD. Both ATP-binding sites in P-gp are able to bind and hydrolyze ATP. However, it has been suggested that there is a structural asymmetry between the two NBDs [30, 31]. The first report to show dimerization of the NBDs with the catalytically inactive EQ mutation was obtained with the MJ0796 bacterial ATP-binding subunits [32]. The analogous mutation was then introduced by our group into human P-gp in 2002. Table 1 presents a summary of reports published since then in which either human or mouse EQ mutant P-gp was used, and some important findings described in each publication. Mutation of the Glu residues to Gln and Ala showed that this mutant binds to 8-azido-ATP more tightly than does WT [33]. Although it was shown that this double mutant exhibits normal ATP binding, the hydrolysis function is significantly impaired. Furthermore, upon binding to ATP the molecule remains in a conformation in which the NBDs are dimerized [13, 33]. Biochemical studies with human and mouse P-gp showed that when only one of these glutamates is mutated, a single hydrolysis cycle occurs, supporting a model in which ATP hydrolysis alternates between the two ATP binding sites [5, 33]. It was shown using purified mouse P-gp in detergent micelles that the EQ mutant could bind ATP, but hydrolyze at a significantly lower rate than the WT protein, likely through a different transition state than the one the WT P-gp undergoes [5]. These earlier studies indicated that the binding of ATP/Mg²⁺ and ADP/Mg²⁺ favored an EQ mutant conformation with the dimerization of two NBDs [5, 33]. It was also suggested that this mutant was unable to form the same transition state as the WT [5, 34], highlighting the role of these residues in the catalysis (cleavage of the gamma phosphate of ATP) and indicating that this, rather than release of ADP is what causes the lower ATP hydrolysis rate.

Table 1.

Summary of published biochemical and structural studies performed using the E-Q mutant of P-gp^*

First report of human P-gp EQ showing trapping of Mg-ATP in the absence of Vi in a similar conformation as the WT in the presence of ATP/Vi.	[33]
Mutation of homologous glutamic acid residues in mouse P-gp allows trapping of the ATP-bound conformation even in the absence of Vi.	[45]
Characterization of mouse EQ mutant ATPase activity and reaction intermediates.	[5]
Mouse EQ mutant P-gp was used to characterize the transition states of mouse WT P-gp. Differences in the characteristics of the closed conformation were identified.	[46]
Effect of substrates on the binding of ATP to mouse EQ mutant.	[47]
EQ mutant was used for the characterization of reaction intermediates based on their kinetic properties and the nature of the trapped nucleotide (triphosphate or diphosphate).	[48]
The E·S conformation of P-gp obtained with the EQ mutant can also be obtained with the wild-type protein in the presence of the nonhydrolyzable ATP analog ATP-γ-S.	[26]
Crosslinking studies of Cys residues in a Cysless background with EQ mutant protein was used to determine proximity of specific residues throughout the protein as the conformation changes during catalytic cycle.	[49]
Mouse EQ mutant was used to study the importance of charge and length of residues substituted for glutamate residues for ATPase activity.	[50]
Introduction of a short crosslinker recovers the function in the EQ mutant.	[51]
Differential scanning calorimetry was used to study the stability of the mouse EQ mutant in the presence of Mg-ATP.	[52]
Double electron electron microscopy was used to study changes in distance between specific amino acids in WT and EQ mutant P-gp during the catalytic cycle.	[14, 15]
EQ mutant P-gp was used to determine the P-gp structure in the closed conformation using Cryo-EM.	[13]
Large scale purification of EQ mutant protein using insect cell membranes and reconstitution of purified mutant protein into nanodiscs.	[16]
Thermal stability of the EQ mutant and comparison to WT protein in the absence and presence of ATP.	[17]
Hydrogen-deuterium exchange mass spectrometry was used to assess conformational changes in mouse P-gp. The EQ mutant was used to generate the closed conformation in the presence of ATP.	[53]

^*The function of the catalytic glutamate residue of the Walker B motif in the ATP sites of several ABC transporters has been studied. Due to space limitations and our focus on the catalytic cycle of P-gp, only studies with the double EQ (E556Q/E1201Q) mutant of the human and mouse transporters are summarized in this table.

We have previously demonstrated that the thermostability of P-gp in membrane vesicles depends on its conformation and that the IC pre-hydrolysis ATP-bound conformation of P-gp is significantly more stable than the IO Apo conformation. Furthermore, the post-hydrolysis conformation was found to have a thermal stability comparable to that of the pre-hydrolysis conformation. In addition, we found that the EQ mutant P-gp had a similar thermal stability as the WT, both in the IO and IC states [17]. Here we report that when the purified EQ mutant is reconstituted in nanodiscs this is also the case. Nanodiscs have been proven to be a very useful platform to generate the membrane environment for biochemical and structural studies. Structures of human P-gp at atomic resolution by cryo-EM have been obtained using purified protein in both detergent micelles and nanodiscs. We have shown that the ability of inhibitors to inhibit the ATPase activity of P-gp is affected when the protein is in detergent micelles. In contrast, properties of P-gp in nanodiscs are similar to those in a native membrane environment [16]. Furthermore, we and other groups have shown that the K_m for ATP hydrolysis by P-gp when reconstituted in nanodiscs is comparable to that observed in membranes [17, 28, 35].

The results presented here focus on how conformational changes impact the binding of UIC2. We used nucleotide analogs that presumably trap P-gp in a specific conformation. We found that UIC2 partially dissociates (30-50% remains bound to P-gp) from P-gp under some conditions, for example by addition of ATP/Mg²⁺ and V_i. It has been previously shown that under these conditions the binding of UIC2 to P-gp is altered. Results regarding the effect of V_i on the binding of UIC2 have been mixed. For example, the Atkins group has shown using SPR that both UIC2 and MRK16 do not bind P-gp in nanodiscs in the presence of V_i [28]. Alternative experiments performed in KK-H as well as K562/i-S9 cells showed only a slight decrease in UIC2 reactivity as the concentration of V_i was increased [36]. Clearly, experimental conditions can impact the ratio between the conformations of P-gp that are able or unable to bind UIC2. Although the molecular basis of this difference between UIC2 reactivity with P-gp in intact or permeabilized cells and when the protein is purified and reconstituted in a purified lipid membrane environment is not understood, the glycosylation of the transporter and/or the use of the Fab of UIC2 instead of the full antibody might be the contributing factors. It is also possible that a fraction of the P-gp molecules is unable to go through the hydrolysis cycle (catalytically inactive, unable to transition to the ATP post-hydrolysis step) but yet able to still bind UIC2. It is also possible that the nanodisc has an impact on the conformational changes of P-gp. However, nanodiscs are currently one of the best membrane mimics for biochemical and cryo-EM based structural studies of membrane proteins. Further studies are needed to fully understand the limitations of using nanodiscs in the conformational dynamics of P-gp.

Our results show partial dissociation of UIC2 in the presence of ATP/Mg²⁺ and V_i. We interpret this to mean that there are at least two different IC conformations, and that the relative movement of the extracellular loops in each conformation differs significantly, as one of them is unable to bind to UIC2-Fab. This has been previously proposed to be the case for P-gp [15, 37]. Hence, this suggests at least two different conformations that correspond to dimerization of the NBDs, one which retains the ability to bind to the UIC2-Fab and the other unable to bind to it. The first one is observed in the presence of ATP, ATP-γ-S, AMP-PNP and ATP-γ-S/Mg²⁺ and likely corresponds to a pre-hydrolysis IC-OO conformation. On the other hand, AMP-PMP/Mg²⁺ and ATP/Mg²⁺/V_i seem to favor an equilibrium between this conformation and one in which the extracellular region is changed, resulting in dissociation of UIC2-Fab, likely corresponding to an IC-OF conformation. It is important to note that it is possible that multiple conformations coexist in each condition. Single molecule experiments and cryo-EM studies should facilitate the understanding of these equilibriums. Previous studies using EPR showed that the DEER distance distributions between residues in different domains of NBDs of the EQ mutant P-gp abrogated the A-loop asymmetry and altered the extracellular region, which resembled more an apo-like (IO) conformation than the ADP-V_i (post-hydrolysis) trapped conformation [14]. In addition, the cryo-EM structure of ATP-bound EQ mutant P-gp in the occluded state does not reflect the asymmetry previously suggested for the two ATP-binding sites [13]. This has raised questions as to whether this mutant truly represents the pre-hydrolysis state of P-gp. In particular, a question that remains unanswered is whether the conformation of P-gp in the pre-hydrolysis ATP-bound conformation and the post-hydrolysis conformation before the release of P_i are identical. Although both of these would correspond to IC conformations, there might be significant differences both in the transmembrane region as well as in the extracellular region that are yet to be revealed. DEER spectroscopy experimental data supports the existence of two IC conformations, one with the outward region occluded (inward-closed-outward occluded, IC-OO) and another one open (inward-closed-outward-facing, IC-OF). The cryo-EM structure of the EQ mutant P-gp corresponds to an IC conformation [13]. However, the fact that ATP remains unhydrolyzed and substrate is not present in the binding pocket in the TMD region suggests that ATP hydrolysis is not needed for the transport function but rather to reset the molecule to the IO conformation. Such a mechanism, however, remains controversial. We have shown that the thermostability of P-gp depends on its conformation and that the IC pre-hydrolysis ATP-bound conformation of P-gp is significantly more stable than the IO Apo conformation. Furthermore, the post-hydrolysis conformation was found to have a thermal stability comparable to that of the pre-hydrolysis conformation. In addition, we found that the EQ mutant P-gp has a similar thermal stability as the WT, both in the IO and IC states [17].

We have used here non-hydrolysable nucleotide analogs as well as ATP and V_i to trap the P-gp in the IC conformation. There are a few examples of ABC transporter structures obtained using AMP-PNP, including the bacterial transporters Sav1866 [38], Rv1819c [39] MalK [40, 41] and MsbA [42] as well as the eukaryotic transporter CmABCB1 [43]. ATP-γ-S was used to solve the structure of the bacterial transporter MetNI [44]. Interestingly, not all of them show magnesium bound in the NBDs, indicating that it is not needed to drive the molecule to the IC conformation. However, magnesium might play an important role even for the non-hydrolyzing nucleotide analogs to favor the IO-OO to IO-OF conformation switch. AMP-PMP/Mg²⁺ and ATP/Mg²⁺/V_i seem to favor an equilibrium between this conformation and one in which the extracellular region is changed, resulting in dissociation of UIC2-Fab, likely corresponding to an IC-OF conformation. A similar pattern was observed for the TmrAB transporter, for which two IC conformations were observed for the EQ mutant in the presence of ATP/Mg²⁺ as well as the WT in the presence of ATP/Mg²⁺/V_i [4]. The comparison of the simulated binding of the UIC2-Fab to extracellular loops of WT and EQ mutant suggests a decrease in the affinity in the IC conformation (Fig. 4). This is likely due to changes in the relative positions of the transmembrane helices in the 6QEX and 6C0V structures that affect the relative positions of the ECLs. This supports the idea that the 6C0V structure likely resembles the IC-OF post-hydrolysis configuration of WT P-gp, and not the IC-OO pre-hydrolysis form as previously reported [11]. Consequently, the conformation of the IC-OO state, still able to bind UIC2, remains to be elucidated.

The sequence of events in the transmembrane region resulting in the efflux of the substrate remains to be clearly elucidated. Structural studies of the bacterial transporters Atm1 and TmrAB have shown that the conformational landscape of these transporters is rather complex [3, 4]. Fig. 5A is a schematic representation of a possible mechanism showing conformational changes that occur in WT and EQ mutant P-gp during the ATP hydrolysis cycle and the Table in Fig. 5B summarizes proposed predominant conformations of WT and EQ mutant P-gp based on published and present results. Both conformations are similar in the absence of ATP (IO), and able to bind to UIC2-Fab in the extracellular region. As ATP binds, WT P-gp transitions to an IC-OO pre-hydrolysis conformation in which the two NBDs are dimerized, but UIC2-Fab is still able to bind to the extracellular region. After ATP is hydrolyzed, but before release of P_i or ADP, the WT P-gp loses the ability to bind to UIC2-Fab. This corresponds to an IC-OF conformation which is likely a very short-lived species that can be trapped by addition of V_i. We speculate that in intact cells the dissociation of UIC2 might occur for a very short period of time and the molecules bind again before diffusing, as the UIC2 seems to remain bound under hydrolysis conditions. Based on our results, we suggest that EQ mutant P-gp transitions through various conformations in a manner different from the WT protein. In addition, we propose that careful consideration should be taken when interpreting P-gp structural data in the presence of ATP or its non-hydrolysable analogs or when using the EQ mutant and we believe that further structural studies are needed to fully understand differences in the conformational landscape of WT and EQ mutant P-gp.

Fig. 5. Comparison of the conformations of WT and EQ mutant P-gp during the ATP hydrolysis cycle.

(A) Binding to UIC2 in both WT and EQ mutant P-gp is similar in the absence of ATP, in the inward-open (IO) configuration. ATP binding induces conformational changes in P-gp (1). These changes include dimerization of the NBDs as well as reconfiguration of the transmembrane region, forming an inward-closed (IC) configuration. While the NBDs are dimerized, the molecule transitions from an inward-closed-outward-occluded (IC-OO) pre-hydrolysis conformation with ATP bound to an inward-closed-outward-facing (IC-OF) post-hydrolysis conformation with ADP and phosphate bound (2), followed by rapid release of phosphate, and subsequent release of ADP, followed by reset of the transporter to the IO conformation (3). (B) The table summarizes proposed predominant conformations of WT and EQ mutant P-gp based on published and present results. The data presented in this report suggest that EQ-mutant P-gp does not appear to follow the same steps as the WT, possibly transitioning to a mixture of IC-OO/IC-OF conformations under certain conditions, such as in the presence of ATP/Mg²⁺ when both conformations are observed.

We found that purified WT and EQ mutant P-gps in nanodiscs exhibit differences in binding of UIC2-Fab under certain conditions. Given that EQ mutant P-gp has been widely used to study P-gp, we would like to raise some important questions: Is the EQ mutant really a good structural mimic of WT P-gp? Does the structure of P-gp (PDBID 6C0V) truly represent the ATP-bound pre-hydrolysis state of P-gp or does it reflect the post-hydrolysis conformation? What is the true structure of P-gp in the pre-hydrolysis conformation? The data presented here as well as the DEER spectroscopy studies [14, 15] and the thermal stability results [17] all suggest that the EQ mutant conformation might include elements from both published cryo-EM structures [11, 13]. The extracellular region seems to resemble the IO conformation reported by the Locher group [11], whereas the NBDs are dimerized, similar to the structure of the IC conformation solved by the Chen group [13].

Taken together, the past and present findings suggest that WT P-gp and the EQ mutant P-gp do not appear to transition through the same conformational changes, which can have major implications for our understanding of the ATP hydrolysis cycle. New structural studies of WT P-gp under non-hydrolysable conditions (i.e., in the presence of ATP alone or using its non-hydrolysable analogs/Mg²⁺) might thus reveal an inward-closed conformation that corresponds to the pre-hydrolysis state. Such studies with P-gp as well as other ABC transporters are critical in order to understand whether the EQ mutant conformation reflects either the pre- or post-ATP hydrolysis state in the catalytic cycle.

Abbreviations

P-gp

P-glycoprotein

BCRP

breast cancer resistance protein

MRP1

the multidrug resistance-associated protein 1

TMD

transmembrane domain

TMH

transmembrane helix

DEER

double electron electron resonance

EPR

electron paramagnetic resonance

cryo-EM

cryo-electron microscopy

ECL

extracellular loop

AMP-PNP

adenylyl-imidodiphosphate

WT

wild type

IO

inward-open

IC

inward-closed

OO

outward-occluded

IF

inward-facing

OF

outward-facing