eLife
eLife Sciences Publications, Ltd
A single power stroke by ATP binding drives substrate translocation in a heterodimeric ABC transporter
Volume: 9
DOI 10.7554/eLife.55943
•
•
•
• Altmetric

### Notes

Abstract

ATP-binding cassette (ABC) transporters constitute the largest family of primary active transporters, responsible for many physiological processes and human maladies. However, the mechanism how chemical energy of ATP facilitates translocation of chemically diverse compounds across membranes is poorly understood. Here, we advance the quantitative mechanistic understanding of the heterodimeric ABC transporter TmrAB, a functional homolog of the transporter associated with antigen processing (TAP) by single-turnover analyses at single-liposome resolution. We reveal that a single conformational switch by ATP binding drives unidirectional substrate translocation. After this power stroke, ATP hydrolysis and phosphate release launch the return to the resting state, which facilitates nucleotide exchange and a new round of substrate binding and translocation. In contrast to hitherto existing steady-state assays, our single-turnover approach uncovers the power stroke in substrate translocation and the tight chemomechanical coupling in these molecular machines.

Keywords
Stefan, Hofmann, Tampé, Albers, and Boudker: A single power stroke by ATP binding drives substrate translocation in a heterodimeric ABC transporter

## Introduction

Ubiquitous in all phyla of life, ABC transporters shuttle chemically diverse compounds across cell membranes, ranging from peptides and proteins to lipids, metabolites, antibiotics, and drugs (Dean et al., 2001; Rees et al., 2009; Thomas and Tampé, 2018). Substrate transport is facilitated by large conformational changes, which are chemomechanically coupled to the hydrolysis of ATP (Abele and Tampé, 2004; Higgins and Linton, 2004; Senior and Gadsby, 1997). ABC exporters are composed of two nucleotide-binding domains (NBDs), responsible for ATP binding and hydrolysis, and two transmembrane domains (TMDs) forming the substrate translocation pathway. ABC exporters are involved in multifold physiological pathways and human diseases, ranging from cystic fibrosis to acquired multidrug resistance in cancer chemotherapy (Csanády et al., 2019; Robey et al., 2018). The transporter associated with antigen processing (TAP) plays a critical role in the adaptive immune response against intracellular pathogens and tumors (Blum et al., 2013; Trowitzsch and Tampé, 2020). Notably, more than half of all human ABC transporters have two functionally asymmetric ATP-binding sites (canonical vs. non-canonical). The Thermus thermophilus multidrug resistance transport complex TmrAB represents a structural and functional homolog of the TAP complex with an overlapping peptide specificity and can restore antigen presentation in human TAP-deficient cells (Kim et al., 2015; Nöll et al., 2017; Zutz et al., 2011). In recent years, the molecular understanding of the conformational space of ABC transporters has been expanded. In particular, eight high-resolution cryo-EM structures of TmrAB in different states and conformations were determined, comprising two inward-facing (IF), four outward-facing (OF), and two asymmetric post-hydrolysis states (Hofmann et al., 2019). Based on this detailed insight, two major transitions from IF-to-OF and OF-to-IF conformation can be defined. However, the functional roles of ATP binding and ATP hydrolysis in the substrate translocation cycle are poorly defined. Owing to the fast ATP turnover in ABC transporters paired with the hydrophobicity of most substrates, it has been intrinsically difficult to tackle these indispensable questions.

Notably, exporters to our knowledge ABC have not been analyzed by single-turnover assays, which would require full transport activity of all correctly reconstituted ABC transporters in liposomes paired with sensitive detection of the translocated substrate. Thus, it remained largely elusive until now whether a single-substrate translocation event is mediated by ATP binding or ATP hydrolysis and how substrate and nucleotide binding affect the major conformational changes. Furthermore, some ABC exporters have been described as uncoupled molecular machines that undergo multiple rounds of ATP hydrolysis per translocated substrate (Bock et al., 2019; Eytan et al., 1996; Nöll et al., 2017; Zehnpfennig et al., 2009; Zollmann et al., 2015). The significance of these futile cycles, however, is largely unclear.

Here, we address these questions by establishing single-turnover assays. The translocation complex TmrAB was reconstituted in lipid nanodiscs or liposomes providing an optimal lipid environment (Bechara et al., 2015) for full activity and conformational freedom (Hofmann et al., 2019). We delineate that a single conformational switch from the IF-to-OF conformation, triggered by ATP binding, drives unidirectional substrate translocation across the membrane. After ATP hydrolysis in and phosphate release from the canonical site, the OF-to-IF transition is launched, which allows for nucleotide exchange and a new round of substrate processing. Our single-turnover analyses provide quantitative insights into the inner mechanics and coupling of ATP binding, hydrolysis, and substrate translocation of ABC transporters.

## Results

### The IF-to-OF switch induced by ATP binding precludes substrate binding

ABC transporters pass through very transient pre- and post-hydrolysis states. In order to kinetically separate the events of ATP binding and hydrolysis, we substituted the catalytic glutamate by glutamine at the canonical ATP site (TmrAE523QB, TmrAEQB) to slowdown the ATP turnover. The transporter was reconstituted in lipid nanodiscs (McLean et al., 2018), maintaining its optimal membrane environment for function and conformational dynamics (Hofmann et al., 2019). By scintillation proximity assays (SPA), we confirmed that wildtype TmrAB and TmrAEQB are equivalent in equilibrium peptide- and nucleotide-binding (Figure 1—figure supplement 1). We therefore utilized TmrAEQB with a slowdown in ATP hydrolysis for single-turnover studies to probe the roles of ATP binding, ATP hydrolysis, and both subsequent conformational transitions (IF-to-OF and return) for substrate translocation. When comparing our high-resolution cryo-EM structures of IF and OF conformers (Hofmann et al., 2019), we found that key residues involved in peptide binding are rearranged upon the IF-to-OF transition. Transmembrane helices TM2, TM3, and TM4, framing the substrate-binding site in the IF conformation, get closely packed together in the OF state to form a tightly sealed intracellular gate (Figure 1a). To monitor this conformational switch, we analyzed the equilibrium peptide binding to IF and OF TmrAEQB. After the IF-to-OF transition induced by ATP binding, the association of radiolabeled peptides to TmrAEQB was precluded (Figure 1b). The impairment of substrate binding was also shown by fluorescence anisotropy using the fluorescent C4F peptide (Figure 1c). Loss of peptide binding correlates with the concentration of ATP added to induce the IF-to-OF transition, which is monitored by fluorescence anisotropy using fluorescent peptides (Figure 1d) and by SPA utilizing radiolabeled peptides (Figure 1—figure supplement 2). The EC50 of 30 µM is consistent with the equilibrium dissociation constant of Mg2+-ATP (Figure 1e). It is worth mentioning that we observed a similar loss of substrate binding after nucleotide trapping in the OF conformation by ATPγS, a slowly hydrolyzing ATP analog (Figure 1—figure supplement 2). Using the same experimental conditions, binding of radiolabeled and fluorescent peptides was not affected by Mg2+-ADP (Figure 1d and Figure 1—figure supplement 2), although Mg2+-ADP and Mg2+-ATP display similar affinities for TmrAB in the IF conformation (Figure 1e). In conclusion, only ATP and not ADP can drive the IF-to-OF transition, which in turn abrogates high-affinity substrate binding to the ABC transporter.

Figure 1.
(a) Scheme of the IF-to-OF transition by comparing the IFwide and OFopen conformation of TmrAB. TmrA, TmrB, and ATP are shown in cyan, yellow, and red, respectively. (b) Loss in peptide binding after the ATP-induced IF-to-OF transition. TmrAEQB (0.2 µM) in Nds was incubated for 5 min at 45 °C with the R9LQK peptide (10 µM, supplemented with 3H-R9L) in the presence and absence of Mg2+-ATP (1 mM). Peptide binding was monitored by SPA. Signals were background-corrected and multiplied by the dilution factor of radiolabeled peptides. (c) Peptide binding after the IF-to-OF switch. The fluorescence anisotropy of C4F peptide (50 nM) was determined at λex/em = 485/520 nm in the absence and presence of TmrAEQB in Nds (1 µM). To distinguish between the IF and OF conformation, TmrAEQB was incubated for 5 min at 45 °C with and without Mg2+-ATP (1 mM). (d) Nucleotide-dependent peptide binding. TmrAEQB in Nds (1 µM) was incubated for 5 min at 45 °C with increasing concentrations of nucleotide. (e) Equilibrium binding affinity of ATP or ADP analyzed by scintillation proximity assay. TmrAEQB (0.2 µM) was incubated with increasing concentration of ATP or ADP supplemented with 3H-ATP or 3H-ADP, respectively. The mean of three independent experiments was analyzed by a one-site binding model, KD, ATP = 0.10 ± 0.03 mM, KD, ADP = 0.25 ± 0.10 mM. In (b–e) the mean ± SD (n = 3) is displayed.
The IF-to-OF transition by ATP binding precludes substrate binding.

### A single ATP turnover and phosphate release trigger the OF-to-IF return

After converting the TmrAEQB complexes into the OF conformation (5 min, 45 °C) and removal of free ATP by rapid gel filtration, we analyzed the occlusion of either [α32P]-ATP or [γ32P]-ATP in TmrAEQB by thin layer chromatography and autoradiography. We then monitored the OF-to-IF transition by (i) the release of the occluded radiolabeled nucleotides, (ii) the restoration of peptide binding, as well as (iii) the ATP hydrolysis and phosphate release (Figure 2a,b). By scintillation proximity assays, we observed a mono-exponential decay of the nucleotides occluded in TmrAEQB and an underlying decline of binding signal over time typical of SPA. The half-life of the nucleotide-occluded state at 20 °C was 24 min, which is in accordance with the ATP turnover of 29 ± 1 min at room temperature (Figure 2c,d). The half-life of the nucleotide-occluded state was slightly affected by temperature. At 4 °C, the half-life was prolonged to 34 min and reduced to 19 min at 45 °C (Figure 2—figure supplement 1). Interestingly, only 50% of bound nucleotides were released at 4 °C suggesting an asymmetric opening of both ATP sites (Figure 2—figure supplement 1). We next sought to trace the OF-to-IF return by a time-dependent recovery of peptide binding. In the absence of Mg-ATP, IF TmrAEQB displayed fast peptide binding, which was linearly increasing over time, typical of SPA-based peptide-binding assays (Figure 2e). As detailed above, substrate binding to OF TmrAEQB was abolished after the IF-to-OF switch. However, we observed a mono-exponential recovery of peptide binding with a half-life of 24 min, reaching the level determined before nucleotide trapping after 60 min (Figure 2e). This demonstrates the full reversibility of the conformational cycle, which we can dissect into two single-turnover events. We also examined the time-dependent release of γ-phosphate upon the OF-to-IF transition. Strikingly, only one of the two occluded ATP molecules is hydrolyzed within the single OF-to-IF transition, approaching a final ATP-to-ADP ratio of 1:1 (Figure 2f). This illustrates the non-equivalence of the two ATP sites in ATP turnover as shown by the two asymmetric unlocked-return states captured by cryo-EM (Hofmann et al., 2019). The rate of phosphate release complies with the overall ATP hydrolysis rate as well as the rates of nucleotide release and peptide re-binding. Taken together, a single ATP turnover and phosphate release at the consensus site launch the OF-to-IF return, restoring the peptide-binding site and allowing nucleotide exchange.

Figure 2.
A single ATP turnover and phosphate release launches the OF-to-IF return.

### The IF-to-OF transition triggered by ATP binding is coupled to substrate translocation

To test whether ATP binding leads to a productive translocation event in the IF-to-OF transition, we reconstituted TmrAEQB in liposomes at a lipid-to-protein ratio of 20:1 (w/w) and performed a single-turnover assay with 50 µM C4F peptide (Figure 3a), which is well above its KD value of 6.8 µM (Nöll et al., 2017). When an IF-to-OF transition was induced by ATP binding (5 min, 45 °C), a substantially higher number of proteoliposome-associated peptides was observed as compared to the background level with ADP. Given that TmrAEQB with occluded Mg2+-ATP remains arrested in the OF conformation with a half-life of 24 min, these data indicate that ATP binding with the ensuring IF-to-OF switch – and not ATP hydrolysis – represent the power stroke driving substrate translocation (Figure 3b). To exclude the possibility that peptides are associated with membranes or protein aggregates in an ATP-dependent manner, we separated liposomes by flotation in sucrose density gradients and quantified the lumenal C4F peptides by fluorescence readout (Figure 3c). For proteoliposomes incubated with Mg2+-ATP, we observed a substantially higher peptide level compared to Mg2+-ADP. Notably, the level of translocated C4F peptides correlated with the amount of TmrAEQB in the floated liposome fraction. In contrast, fractions lacking TmrAB showed only background fluorescence similar to proteoliposomes incubated with ADP, for which the peptide level was independent of the amount of reconstituted TmrAEQB in the corresponding fraction. To determine the coupling ratio of transported peptides per TmrAEQB complex, we characterized proteoliposomes based on (i) TmrAEQB orientation, (ii) incorporation efficiency, and (iii) liposome diameter (Figure 3—figure supplement 1). Firstly, the orientation of TmrAEQB in liposomes was determined by TEV-protease cleavage of the His-tag fused to TmrA. As monitored via anti-His immunoblotting, 50% of the His-tags were cleaved, reflecting TmrAB complex with the right-side-out orientation while equal loading of proteoliposomes was controlled by instant blue staining (Figure 3—figure supplement 1). Secondly, the reconstitution efficiency was determined by carbonate crush and flotation in sucrose density gradients. Hereby, 95% of TmrAEQB complexes were effectively reconstituted in proteoliposomes and were detected in the top fractions as quantified by anti-His immunoblotting (Figure 3—figure supplement 1). Thirdly, the average proteoliposomes diameter (~160 nm) was measured by single-particle tracking (Figure 3—figure supplement 1). Finally, we calculated for each liposome ~30 correctly-oriented, energizable transport complexes. Notably, based on the number of translocated peptides per liposome, a nearly stoichiometric coupling of ~0.5 translocated peptides per TmrAEQB complex was calculated.

Figure 3.
(a) Experimental scheme. TmrAEQB was reconstituted in liposomes and incubated with C4F peptide and Mg2+-ATP or Mg2+-ADP. For single-turnover, transport was performed for 5 min at 45 °C to allow a single IF-to-OF transition. After solubilization of proteoliposomes, transported peptides were quantified by fluorescence detection (λex/em = 485/520 nm). (b) Single-turnover transport studies. Liposomes containing reconstituted TmrAEQB were mixed with C4F peptide (50 µM) and Mg2+-ATP or Mg2+-ADP (3 mM each) for 5 min at 45 °C. Liposomes were extensively washed. After solubilization, peptides were fluorescently quantified. Data are shown as mean ± SD (n = 3). (c) Single-turnover transport and liposome flotation assays. C4F peptide translocation was performed as described in b). Proteoliposomes were floated by sucrose density gradient centrifugation to remove aggregates and unspecifically adsorbed C4F peptides. Transported peptides were fluorescently quantified. TmrAEQB was detected using immunoblotting (α-His).
Single-conformational switch by ATP binding is coupled to an efficient substrate translocation.

### ATP binding drives substrate translocation monitored by single-liposome assays

As demonstrated above, ATP binding induces the IF-to-OF transition, which is tightly coupled to peptide translocation. To reveal insights into this single-transition event, we established a single-liposome transport assay with high signal-to-background sensitivity. For each liposome, the number of translocated peptides was correlated with the number of right-side-out reconstituted TmrAB, which was stoichiometrically labeled with trisNTAAlexaFluor647 (trisNTAAF647) via the C-terminal His-tag of TmrA as previously described (Wieneke and Tampé, 2019). To set up this new approach, we first analyzed substrate translocation by wildtype TmrAB (Figure 4a). After incubation with ATP for 5 min at 45 °C, a 100-fold increase in transported peptides was observed as compared to the background control with ADP (Figure 4b). It is important mentioning that substrate transport was observed for 98% of TmrAB-positive liposomes. To convert the mean fluorescence intensities into the absolute number of translocated peptides, we calibrated the flow cytometry analysis with liposomes loaded with defined numbers of C4F peptide and equal concentrations of dextranAF647 as loading control. We observed a linear increase of the mean fluorescence intensity over the range of 1 to 100 peptides per liposome (Figure 4—figure supplement 1).

Figure 4.
(a) Experimental scheme. TmrAB-containing liposomes were incubated with C4F peptide (30 µM) and Mg2+-ATP or Mg2+-ADP (3 mM each) for 5 min at 45 °C. Proteoliposomes were extensively washed by centrifugation, and His-tagged TmrAB complexes were labeled with trisNTAAF647 (150 nM). The mean fluorescence intensities of correctly oriented TmrAB complexes and translocated peptides were examined by flow cytometry. (b), Multiple rounds of C4F peptide translocation. Proteoliposomes were incubated with C4F peptide and nucleotides as described in a). Proteoliposomes were washed and correctly-oriented TmrAB complexes were labeled with trisNTAAF647 as described in a). The mean fluorescence intensities of 100,000 events were analyzed.Sequential rounds of peptide translocation monitored by flow cytometry.

We next followed peptide translocation by TmrAEQB under single-turnover conditions and quantified the chemomechanical coupling (Figure 5a). Proteoliposomes incubated with ATP showed a substantial increase in fluorescent peptide compared to the background level with ADP (Figure 5b). Peptide translocation by TmrAEQB showed a biphasic time dependency characterized by a fast burst phase (pre-steady state) and a much slower linear steady-state rate (Figure 5c). Within the first minutes, the transported peptides increased until all TmrAEQB complexes had been switched and arrested in the pre-hydrolytic state (burst phase). With longer incubation periods (steady-state condition), a very low translocation rate reflecting multiple ATP turnover events was observed. In the burst phase (0–5 min), however,~30 peptides are transported per liposome (y-axis intercept of the linear term), reflecting the number of transport-competent TmrAEQB complexes per liposome (Figure 3—figure supplement 1).

Figure 5.
(a) Experimental setups. Liposomes containing reconstituted TmrAEQB were incubated with C4F peptide (30 µM) and Mg2+-ATP or Mg2+-ADP (3 mM each). To guarantee a single-turnover, transport was performed for 5 min at 45 °C (switch 1). For two consecutive cycles, proteoliposomes were incubated for 1 hr at 4 °C (return) followed by 5 min at 45 °C (switch 2). Proteoliposomes were extensively washed, and His-tagged TmrAEQB complexes were labeled with trisNTAAF647 (150 nM). Proteoliposomes were washed by centrifugation. The mean fluorescence intensities of correctly oriented TmrAEQB complexes and translocated peptides were examined by flow cytometry. (b) Single-translocation events monitored by flow cytometry. TmrAEQB proteoliposomes were incubated with C4F peptide and nucleotide as described in a). The mean fluorescence intensities of 100,000 events were analyzed by flow cytometry. (c) Time-dependent translocation of peptides induced by ATP binding and IF-to-OF switching. TmrAEQB proteoliposomes were incubated with C4F peptide and nucleotides as described in a) for increasing time at 45 °C. For ATP, signals were subtracted by mean fluorescence intensities of untreated proteoliposomes and fitted by a burst equation (Equation 6), yielding ~35 peptides per liposome at the Y-intercept (Spanfast) and kslow ≈ 0.5 peptides liposome−1 min−1. Data are shown as mean ± SD (n = 3). (d), Two consecutive rounds of C4F peptide translocation driven by ATP binding and IF-to-OF switching. TmrAEQB proteoliposomes were incubated with C4F peptide and nucleotides as described in a). The mean fluorescence intensities of 105 events were analyzed by flow cytometry. Data are shown as mean ± SD (n = 3).
A single IF-to-OF transition driving substrate translocation.

### Single translocation events are unidirectional and stoichiometrically coupled

We next investigated sequential cycles of substrate translocation induced by ATP binding (Figure 5a). After the first IF-to-OF switch (switch 1, 5 min at 45 °C), we gave the transport complexes sufficient time to return to the IF conformation (60 min at 4 °C, see Figure 2). We then performed a second IF-to-OF transition under single-turnover conditions (switch 2). Proteoliposomes displayed an additive accumulation of peptides along each IF-to-OF switch, while the number of translocated peptides per liposome did not change during each OF-to-IF return, indicating that peptides are not retro-translocated during the OF-to-IF transition (Figure 5d). We calculated that ~1.3 peptides are translocated per correctly oriented TmrAEQB upon each IF-to-OF transition. To follow the fate of the substrate after its first translocation, we performed a first single-turnover with an Alexa Fluor 647-labeled peptide (C4AF647), followed by a second single-turnover experiment with the C4F peptide (Figure 5—figure supplement 1). In the presence of ATP, proteoliposomes displayed elevated signals for both peptides compared to proteoliposomes incubated with ADP. Notably, during the second IF-to-OF switch, the fluorescence signal of C4AF647 remained constant, demonstrating that substrate transport occurs unidirectionally. Taken together, sequential rounds of ATP binding drive a unidirectional and stoichiometrically coupled substrate translocation via a single IF-to-OF switch. In turn, ATP hydrolysis and phosphate release in the canonical site is used to master the directionality of the ABC transporter.

## Discussion

In this work, we advance the quantitative mechanistic understanding of ABC exporters. We reveal that ATP binding, which initiates the IF-to-OF transition, is stoichiometrically coupled to substrate translocation (Figure 6). Along this power stroke by occlusion of two ATP molecules, the substrate-binding site is remodeled, priming substrate release on the extracellular site. In contrast, ATP hydrolysis and subsequent phosphate release at the consensus site trigger the OF-to-IF return, which restores substrate binding and nucleotide exchange to start a new translocation cycle. By using equilibrium ligand-binding assays, we demonstrate that the IF-to-OF transition initiated by ATP binding abolishes peptide binding.

Figure 6.
IFwide conformation of TmrAB allows binding of nucleotides and peptide. ATP binding triggers the IF-to-OF conformational transition, which drives substrate translocation. ATP-occluded OF conformations represent pre-hydrolysis states, in which peptide binding and nucleotide exchange is abolished. After ATP hydrolysis at the consensus site, phosphate (Pi) release triggers the OF-to-IF transition via asymmetric unlocked-return (URasym and URasym*) conformers, revealed by cryo-EM (Hofmann et al., 2019), to reactivate peptide and nucleotide binding.Translocation cycle.

The use of a catalytically less active variant in combination with Mg-ATP has been described as an effective strategy for colonizing ABC exporters in the pre-hydrolytic OF state (Hofmann et al., 2019; Kim and Chen, 2018). Based on this data, we employed the TmrAEQB variant with a slowdown in ATP turnover to kinetically outline the major conformational transitions (IF-to-OF and return) and to correlate these rearrangements with substrate translocation. This approach enabled us to resolve substrate transport in single-turnover mode. Regarding wildtype TmrAB, single-turnover transport studies would require conformation-selective inhibitors which arrest the transport complex at different stages of the translocation cycle and which to our knowledge are not available.

A comparison of IF and OF TmrAB points to a drastic reorganization of residues aligning the substrate-binding site (Hofmann et al., 2019), similar to P-glycoprotein (Martin et al., 2000; Martin et al., 2001), ABCG2 (McDevitt et al., 2008), and MRP1 (Payen et al., 2003). In addition, we delineate that the OF-to-IF return directly correlates with (i) the ATP turnover, (ii) the release of inorganic phosphate, (iii) the restoration of peptide binding, and (iv) nucleotide exchange. We observed a full recovery of peptide binding, demonstrating reversibility of conformational transitions to start subsequent processing cycles.

The processive-clamp/ATP-switch model (Abele and Tampé, 2004; Higgins and Linton, 2004; Janas et al., 2003) has gained increasing acceptance in the light of new structures of ABC transporters solved in the nucleotide-bound state (Johnson and Chen, 2018; Manolaridis et al., 2018). However, because of major experimental demands, it is an open debate whether the substrate translocation event is driven by ATP binding or ATP hydrolysis. We addressed this pressing question by establishing a sensitive single-turnover translocation assay. We uncovered that ATP binding drives the single-translocation event of different substrates along the IF-to-OF transition. By analyzing sequential steps in the translocation cycle, we revealed that substrate transport by TmrAEQB is unidirectional, which was previously shown for human TAP (Grossmann et al., 2014).

In ABC transporters, the coupling ratio between conformational transitions and substrate translocation remain largely elusive. The coupling between ATP hydrolysis and the transported substrate can be quite ineffective, reaching up to 100 ATP-turnover events per translocated substrate (Bock et al., 2019; Manolaridis et al., 2018; Nöll et al., 2017; Zehnpfennig et al., 2009). Experimental parameters such as detergents, lipid environment, sample preparation, protein activity, and sensitivity of applied methods can substantially affect the energetic coupling. Here, our newly established single-turnover approaches reveal a coupling ratio of 0.5 to 1.3 peptides per TmrAEQB. Taking the experimental error and sensitivity of these approaches into account, we revealed a stoichiometric coupling between a single conformational transition and a translocation event. Sensitive translocation assays with single-liposome or single-transporter resolution will be indispensable to elucidate the main factors that control and affect the chemomechanical coupling.

Our new single-liposome approaches revealed a fast burst phase followed by a slower steady-state phase, indicating that a step after substrate translocation is rate-limiting. In our cryo-EM structures of TmrAB, we captured two asymmetric ADP/ATP-bound states under steady-state conditions, leading to the assumption that the separation of the NBD dimer represents the rate-limiting step (Hofmann et al., 2019). Here, we demonstrate that the rates of phosphate, ATP, and ADP release correlate with the OF-to-IF transition. In the future, single-molecule fluorescence resonance energy transfer (Catipovic et al., 2019), time-resolved small-angle X-ray scattering (Josts et al., 2020), and stopped-flow analyses (Robson et al., 2009) could provide further insights into turnover kinetics and rate-limiting steps.

In conclusion, we delineate the mechanistic framework of major conformational transitions and the fundamental energetic principles in heterodimeric ABC transporters. Our single-turnover analyses significantly advance our quantitative mechanistic understanding how transitional rearrangements in ABC transport complexes master substrate binding and translocation. We uncovered that the IF-to-OF transition by ATP binding drives unidirectional substrate translocation across the membrane. In the future, single-turnover approaches will be an indispensable method to unfold hidden facets of conformational coupling, energy transduction, and mechanistic details of substrate translocation by ABC transporters but also other membrane translocation machineries.

## Materials and methods

Key resources table
Reagent type
(species) or
resource
information
AntibodyMonoclonal α-HisSigma-AldrichSAB1305538Mouse origin; 1:2000 dilution
AntibodyPolyclonal α-mouse-HRP conjugateSigma-AldrichAP130PGoat origin; 1:20,000 dilution
Chemical compound, drugβ-n-dodecyl β-D-maltoside (DDM)Carl RothCN26.5
Chemical compound, drugBovine brain lipid extractSigma-AldrichB1502
Chemical compound, drugE. coli polar lipidsAvanti100600
Chemical compound, drug1,2-Dioleoyl-sn-Glycero-3-Phosphocholine (DOPC)AnatraceD518
Chemical compound, drug[2,5’,8-3H(N)]-ATP (3H-ATP)PerkinElmerNET118900
BiotrendART-0386
Chemical compound, drug32P]-ATPHartmann AnalyticFP-207
Chemical compound, drug32P]-ATPHartmann AnalyticFP-201
Chemical compound, drugCopper-chelated PVT SPA beadsPerkinElmerRPNQ0095
Gene (Thermus thermophilus)TmrAQ72J05TTC0976
Gene (Thermus thermophilus)TmrBQ72J04TTC0977
Peptide, recombinant proteinRRY-[3H]L-KSTELHartmann AnalyticCustom synthesis
Peptide, recombinant proteinRRY-C*-KSTELThis study (Peptide labeling and purification)C* denotes fluorescein labeled Cys
Peptide, recombinant proteinRRY-C*-KSTELThis study (Peptide labeling and purification)C* denotes Alexa Fluor 647 labeled Cys
Recombinant DNA reagentpET-22bMerck Millipore69744Vector for protein expression in E. coli
Software, AlgorithmPyMOLSchrödinger
Strain, strain background (Escherichia coli)BL21(DE3)Thermo FisherC600003Chemically competent cells

### Expression and purification of TmrAB

Expression and purification of TmrAB was performed as described (Hofmann et al., 2019; Zutz et al., 2011). TmrAB was expressed in E. coli BL21(DE3) grown in LB high salt media at 37 °C. At an OD600 of 0.6, protein expression was induced with 0.5 mM isopropyl β-D-thiogalactopyranoside (IPTG, Carbolution). Cells were grown for 3 hr at 37 °C. Harvested cells were resuspended in lysis buffer (20 mM HEPES-NaOH pH 7.5, 300 mM NaCl, 50 μg/ml lysozyme, 0.2 mM PMSF) and disrupted by sonication. Membranes were pelleted at 100,000 g for 30 min at 4 °C. Membranes were solubilized for 1 hr at 4 °C with 20 mM β-n-dodecyl β-D-maltoside (β-DDM, Carl Roth) in purification buffer (20 mM HEPES-NaOH pH 7.5, 300 mM NaCl). After centrifugation at 100,000 g for 30 min at 4 °C, solubilized TmrAB was applied to Ni-NTA agarose (Qiagen) for 1 hr at 4 °C. Resin was washed with 20 column volumes of purification buffer containing 50 mM imidazole and 1 mM β-DDM. Bound protein was eluted with purification buffer including 300 mM imidazole and 1 mM β-DDM. The eluate was buffer exchanged to TSK buffer (20 mM HEPES-NaOH pH 7.5, 150 mM NaCl, 1 mM β-DDM) using PD-10 desalting column (GE Healthcare).

### Expression and purification of membrane scaffolds

MSP1D1 was expressed in E. coli BL21(DE3) grown in LB high salt media at 37 °C (Hofmann et al., 2019). Protein expression was induced by addition of 0.5 mM IPTG at an OD600 of 0.6. Cells were grown for 3 hr at 37 °C. Purification was performed as described (Ritchie et al., 2009). Briefly, harvested cells were resuspended in 40 mM Tris-HCl pH 8.0, 300 mM NaCl, 1 mM PMSF and 1% Triton X-100. Cells were disrupted using sonication, and lysate was cleared at 30,000 g for 30 min at 4 °C. MSP1D1 was applied to Ni-NTA agarose (Qiagen) for 1 hr at 4 °C. Resin was washed consecutively with 20 column volumes of Tris-HCl buffer (40 mM Tris-HCl pH 8.0, 300 mM NaCl), containing (i) 1% Triton X-100, (ii) 50 mM Na-cholate, 20 mM imidazole, and (iii) 50 mM imidazole. Bound MSP1D1 was eluted in Tris-HCl buffer containing 400 mM imidazole. Eluate was buffer exchanged to 20 mM Tris-HCl pH 7.4, 100 mM NaCl and 0.5 mM EDTA using PD-10 desalting column (GE Healthcare).

### Reconstitution in lipid nanodiscs

Reconstitution of TmrAB in lipid nanodiscs was performed as described (Hofmann et al., 2019). Bovine brain lipids (Sigma-Aldrich) were solubilized in 20 mM of β-DDM. Purified TmrAB and MSP1D1 were mixed with DDM-solubilized bovine brain lipids in a TmrAB/MSP1D1/lipid molar ratio of 1/7.5/100 in TSK buffer without detergent. Samples were incubated at 20 °C for 30 min. SM-2 Bio-beads (Bio-Rad) were added in two steps at 4 °C (1 hr and overnight) to remove detergent. Samples were concentrated using Amicon Ultra-0.5 ml centrifugal filters with 50 kDa cut-off (Merck Millipore). TmrAB reconstituted in lipid nanodiscs were separated from empty nanodiscs by size exclusion chromatography (SEC) using Superdex 200 Increase 3.2/300 (GE Healthcare).

### Liposome reconstitution

Reconstitution of TmrAB in liposomes was performed as described (Nöll et al., 2017). Briefly, liposomes were composed of E. coli polar lipids/DOPC (Anatrace) in a 7/3 (w/w) ratio. Purified TmrAB was added to Triton X-100 destabilized liposomes in a 1/20 (w/w) ratio. SM-2 Bio-beads (Bio-Rad) were added in four steps at 4 °C (1 hr, overnight, 2 hr, and 2 hr) to remove the detergent. Proteoliposomes were harvested at 270,000 g for 30 min at 4 °C and resuspended in transport buffer (20 mM HEPES-NaOH pH 7.5, 150 mM NaCl, 5% (v/v) glycerol) to a final lipid concentration of 5 mg/ml. TmrAB orientation in liposomes was examined using TEV protease specific cleavage of the C-terminal His-tag. Briefly, liposomes were mixed with TEV protease in a TmrAB/TEV protease ratio of 16/1 (w/w). Digestion was performed for 16 hr at 4 °C including 3 mM DTT. As positive control, proteoliposomes were solubilized during digestion with 1% (w/v) Triton X-100. His-tag cleavage of TmrAB was quantified by immunoblotting using α-His detection (Sigma-Aldrich). An equal load of TmrAB was confirmed by SDS-PAGE and InstantBlue staining (Sigma-Aldrich). The reconstitution of TmrAB in liposomes was analyzed by carbonate crush and sucrose density gradient flotation assays. Proteoliposomes were harvested at 270,000 g for 30 min at 4 °C and resuspended in 100 mM Na2CO3 pH 11.5, 30% (w/v) sucrose. Samples were incubated for 30 min on ice. Liposomes were layered with HEPES-NaOH buffer (20 mM HEPES-NaOH pH 7.5, 150 mM NaCl) containing (i) 30% sucrose, (ii) 20% sucrose and (iii) 0% sucrose. Samples were centrifuged at 270,000 g for 30 min at 4 °C. The gradient was fractionated, and TmrAB was quantified by immunoblotting using α-His antibody (Sigma-Aldrich). The size of the proteoliposomes were determined by the nanoparticle tracking system NanoSight LM10 (Malvern Panalytical). Reconstituted proteoliposomes were diluted to a lipid concentration of 1 mg/l with transport buffer, and vesicle tracking was performed at 25 °C.

### Peptide binding assays

Equilibrium peptide binding to TmrAB was examined using scintillation proximity assays (SPA). TmrAB (0.2 µM) reconstituted in lipid nanodiscs was mixed with the peptide RRYQKSTEL (9.75 µM of R9LQK) traced with tritylated RRY-[3H]L-KSTEL (0.25 µM of 3H-R9L, Hartmann Analytic). To induce the IF-to-OF transition, TmrAEQB was incubated with 1 mM ATP (Sigma-Aldrich) and 3 mM MgCl2 for 5 min at 45 °C. If indicated, the excess of free ATP and peptide was removed by rapid gel filtration (Bio-Spin columns P-30, Bio-Rad). Copper-chelated SPA beads (PerkinElmer) were added to a final concentration of 5 mg/ml. The scintillation proximity assay was performed at 20 °C in cpm mode (Wallac MicroBeta). Background was determined in the presence of 200 mM imidazole. Data represent mean ± SD from three experiments. The total binding signal was background-corrected. The specific binding was multiplied by the dilution factor of radiolabeled 3H-R9L. The recovery of peptide binding during the OF-to-IF transition was followed by mono-exponential fit

$Y={Y}_{0}+\left(Plateau-{Y}_{0}\right)\bullet \left(1-{e}^{kt}\right)$
with Y: binding, Y0: binding at X = 0, k: rate constant. The half-life was calculated as
${t}_{1/2}=\frac{ln2}{k}$

Peptide binding was analyzed by a one-site Langmuir-type binding model

$Y=\frac{{B}_{max}\bullet \left[P\right]}{{K}_{D}+\left[P\right]}$
with Y: binding, Bmax: maximum binding, and KD: equilibrium dissociation constant.

### Nucleotide binding assays

Nucleotide binding to TmrAB was examined using SPA. 200 nM TmrAB reconstituted in lipid nanodiscs were incubated with 2.8 µM ATP supplemented with 0.2 µM of [2,5’,8-3H(N)]-ATP (3H-ATP, PerkinElmer) or 2.8 µM ADP supplemented with 0.2 µM of [2,8-3H]-ADP (3H-ADP, Biotrend), and 3 mM MgCl2. If indicated, the excess of free nucleotides was removed by rapid gel filtration (Bio-Spin columns P-30, Bio-Rad). In some cases, rebinding of occluded nucleotides was blocked by adding an excess of ATP (1 mM). Samples were kept on ice to prevent ATP turnover and ATP-induced conformational changes. Nucleotide binding was analyzed by a one-site binding model (analog to Equation 3). Due to excessive background signals, the nucleotide concentration could not be increased above 0.3 mM. Data represent mean ± SD from three experiments.

### Fluorescence anisotropy

Binding of fluorescent peptide RRYCfluoresceinKSTEL (C4F) was analyzed as described (Nöll et al., 2017). Briefly, TmrAB (1 µM) reconstituted in lipid nanodiscs was incubated for 5 min at 45 °C in the presence or absence of 1 mM ATP and 3 mM MgCl2 in 20 mM HEPES-NaOH pH 7.5, 150 mM NaCl. In some cases, ATP and ADP were titrated. The fluorescein anisotropy of the C4F peptide (50 nM) was analyzed at λex/em of 485/520 nm using a microplate reader (CLARIOstar, BMG LABTECH). The fluorescence anisotropy was calculated using

$r=\frac{{I}_{||}-{I}_{\mathrm{\perp }}}{{I}_{||}+2\cdot {I}_{\mathrm{\perp }}}$

Data represent mean ± SD from three experiments. To determine nucleotide effects on peptide binding, fluorescence anisotropy was normalized to free and 100%-bound C4F peptide and fitted by dose-response function

$Y=Bottom+\frac{Top-Bottom}{1+{10}^{\left(X-log{EC}_{50}\right)}}$
with Y: normalized binding, Bottom and Top: plateau values of regression function, EC50: effective concentration 50%, X: log[nucleotide].

### ATPase activity assays

TmrAEQB (1 µM) reconstituted in lipid nanodiscs was incubated with 1 mM ouabain, 5 mM NaN3, 50 µM EGTA, 3 mM MgCl2, and 1 mM ATP traced with [γ32P]-ATP (Hartmann Analytic). Background hydrolysis was recorded in the presence of 10 mM EDTA and in the absence of TmrAB. Samples were incubated for 5 min at 45 °C and 15 min at 20 °C to mimic the trapping procedure of TmrAEQB. Samples were spotted onto polyethyleneimine cellulose plates (Merck Millipore). Thin layer chromatography was performed with 0.8 M LiCl-acetic acid pH 3.2. Plates were developed overnight on Exposure Cassette-K (Bio-Rad) and evaluated on Personal Molecular Imager System (Bio-Rad). Due to time delays between nucleotide trapping, thin layer chromatography and autoradiography, background levels of ADP resulting from autohydrolysis could not be avoided. Data represent mean ± SD from three experiments.

### Nucleotide occlusion

After nucleotide trapping of TmrAEQB, occluded nucleotides were analyzed as described (Zutz et al., 2011). In brief, TmrAEQB (2 µM) reconstituted in lipid nanodiscs was incubated with 3 mM MgCl2 and 1 mM ATP traced with [γ32P]-ATP or [α32P]-ATP (Hartmann Analytic) for 5 min at 45°C. After arresting TmrAEQB in the OF conformation, 2 mM of cold ATP were added and free radiolabeled nucleotides were removed by rapid gel filtration (Bio-Spin columns P-30, Bio-Rad). The nucleotide-trapped TmrAEQB was directly transferred into 2 mM ATP at 20 °C to follow a single-turnover. At different time points, samples were spotted onto polyethyleneimine cellulose plates (Merck Millipore). Thin layer chromatography was performed with 0.75 M KH2PO4 pH 3.4. Plates were developed overnight on Exposure Cassette-K (Bio-Rad) and evaluated on Personal Molecular Imager System (Bio-Rad). Data represent mean ± SD from three experiments.

### Peptide transport assays

TmrAEQB (1.5 µM) reconstituted in liposomes was incubated with 50 µM C4F peptide, 3 mM ATP or ADP, and 5 mM MgCl2. Transport was performed for 5 min at 45 °C and stopped with 10 mM EDTA. Proteoliposomes were washed and centrifuged three times for 30 min at 270,000 g, 4 °C in 20 mM HEPES-NaOH pH 7.5, 150 mM NaCl, 5% (v/v) glycerol, and 10 mM EDTA. Afterwards, proteoliposomes were either (i) solubilized with 1% (w/v) SDS, and peptides were quantified at λex/em of 485/520 nm using microplate reader, or (ii) applied to a sucrose density gradient flotation assay to remove any peptide bound to liposomes or aggregated TmrAEQB. For this, liposomes were resuspended in 20 mM HEPES-NaOH pH 7.5, 150 mM NaCl, and 55% (w/v) sucrose. The liposome fraction was covered with layers of buffer with decreasing concentrations of sucrose (55% to 0%). Liposomes were separated at 100,000 g for 16 hr at 4 °C. The gradient was harvested in fractions. TmrAB was quantified by immunoblotting using an α-His antibody (Sigma-Aldrich). Fractions were solubilized with 1% (w/v) SDS, and peptides were quantified at λex/em of 485/520 nm using a microplate reader. Data represent mean ± SD from three experiments.

### Single liposome flow cytometry

$Y={Span}_{fast}\bullet \left(1-{e}^{-{k}_{fast}\bullet t}\right)+{k}_{slow}\bullet t$
with Spanfast: intercept of y-axis, kfast: rate constant of burst phase, kslow: slope of linear steady-state phase. Data were recorded in triplicates and mean ± SD are displayed.

### Peptide labeling and purification

Peptides were synthesized on solid phase by standard Fmoc chemistry. For cysteine labeling, the C4 peptide (RRYCKSTEL) was incubated with 5-iodoacetamide fluorescein (Sigma-Aldrich) or Alexa Fluor 647 maleimide (Thermo Fisher) in PBS DMF buffer (8.1 mM Na2HPO4 pH 6.5, 137 mM NaCl, 2.7 mM KCl, 1.8 mM KH2PO4, 20% (v/v) DMF) for 1 hr at 20 °C using a molar fluorophore excess of 1.2. Samples were purified by reversed-phase HPLC (Jasco; PerfectSil 300 ODS C18) applying a linear acetonitrile gradient from 5–80% supplemented with 0.1% (v/v) TFA. Purified peptides were snap frozen in liquid nitrogen and lyophilized (Lyovac GT2, Heraeus).

### Data presentation and statistics

All measurements were performed in triplicates (n = 3). In all column and XY diagrams, mean values ± SD were presented. Diagrams were prepared in GraphPad Prism5. Figures of protein structures were prepared using PyMOL (Schrödinger, 2015).

## Funding Information

This paper was supported by the following grants:

http://dx.doi.org/10.13039/501100001659Deutsche ForschungsgemeinschaftTa157/12-1 to Robert Tampé.
http://dx.doi.org/10.13039/501100000781European Research CouncilERC_Ad 789121 to Robert Tampé.
http://dx.doi.org/10.13039/501100001659Deutsche ForschungsgemeinschaftSFB 807 to Robert Tampé.

## Acknowledgements

We thank Dr. Rupert Abele, Dr. Christoph Thomas, Andrea Pott, Inga Nold, and all members of the Institute for Biochemistry (Goethe University Frankfurt) for discussion and comments. This research was supported by the German Research Foundation (SFB 807 – Membrane Transport and Communication as well as Ta157/12-1 to RT). The support by the European Research Council (ERC Advanced Grant 789121 to RT) is gratefully acknowledged.

### Competing interests

No competing interests declared.

### Author contributions

Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing.
Data curation, Investigation.
Conceptualization, Data curation, Formal analysis, Supervision, Funding acquisition, Validation, Visualization, Writing - review and editing.

## Data availability

All data generated or analysed during this study are included in the manuscript and supporting files. Source data files have been provided for Figures 1, 2, 3, 4, and 5.

## References

1

Abele R, Tampé R 2004. . The ABCs of immunology: structure and function of TAP, the transporter associated with antigen processing. Physiology 19: , pp.216-224, doi: 10.1152/physiol.00002.2004

2

Bechara C, Nöll A, Morgner N, Degiacomi MT, Tampé R, Robinson CV 2015. . A subset of annular lipids is linked to the flippase activity of an ABC transporter. Nature Chemistry 7: , pp.255-262, doi: 10.1038/nchem.2172

3

4

Bock C, Zollmann T, Lindt KA, Tampé R, Abele R 2019. . Peptide translocation by the lysosomal ABC transporter TAPL is regulated by coupling efficiency and activation energy. Scientific Reports 9: 11884, doi: 10.1038/s41598-019-48343-6

5

Catipovic MA, Bauer BW, Loparo JJ, Rapoport TA 2019. . Protein translocation by the SecA ATPase occurs by a power-stroke mechanism. The EMBO Journal 38: e101140, doi: 10.15252/embj.2018101140

6

Csanády L, Vergani P, Gadsby DC 2019. . Structure, gating, and regulation of the CFTR anion channel. Physiological Reviews 99: , pp.707-738, doi: 10.1152/physrev.00007.2018

7

Dean M, Rzhetsky A, Allikmets R 2001. . The human ATP-binding cassette (ABC) transporter superfamily. Genome Research 11: , pp.1156-1166, doi: 10.1101/gr.GR-1649R

8

Eytan GD, Regev R, Assaraf YG 1996. . Functional reconstitution of P-glycoprotein reveals an apparent near stoichiometric drug transport to ATP hydrolysis. Journal of Biological Chemistry 271: , pp.3172-3178, doi: 10.1074/jbc.271.6.3172

9

Grossmann N, Vakkasoglu AS, Hulpke S, Abele R, Gaudet R, Tampé R 2014. . Mechanistic determinants of the directionality and energetics of active export by a heterodimeric ABC transporter. Nature Communications 5: 5419, doi: 10.1038/ncomms6419

10

Higgins CF, Linton KJ 2004. . The ATP switch model for ABC transporters. Nature Structural & Molecular Biology 11: , pp.918-926, doi: 10.1038/nsmb836

11

Hofmann S, Januliene D, Mehdipour AR, Thomas C, Stefan E, Brüchert S, Kuhn BT, Geertsma ER, Hummer G, Tampé R, Moeller A 2019. . Conformation space of a heterodimeric ABC exporter under turnover conditions. Nature 571: , pp.580-583, doi: 10.1038/s41586-019-1391-0

12

Janas E, Hofacker M, Chen M, Gompf S, van der Does C, Tampé R 2003. . The ATP hydrolysis cycle of the nucleotide-binding domain of the mitochondrial ATP-binding cassette transporter Mdl1p. Journal of Biological Chemistry 278: , pp.26862-26869, doi: 10.1074/jbc.M301227200

13

Johnson ZL, Chen J 2018. . ATP binding enables substrate release from multidrug resistance protein 1. Cell 172: , pp.81-89, doi: 10.1016/j.cell.2017.12.005

14

Josts I, Gao Y, Monteiro DCF, Niebling S, Nitsche J, Veith K, Gräwert TW, Blanchet CE, Schroer MA, Huse N, Pearson AR, Svergun DI, Tidow H 2020. . Structural kinetics of MsbA investigated by Stopped-Flow Time-Resolved Small-Angle X-Ray scattering. Structure 28: , pp.348-354, doi: 10.1016/j.str.2019.12.001

15

Kim J, Wu S, Tomasiak TM, Mergel C, Winter MB, Stiller SB, Robles-Colmanares Y, Stroud RM, Tampé R, Craik CS, Cheng Y 2015. . Subnanometre-resolution electron cryomicroscopy structure of a heterodimeric ABC exporter. Nature 517: , pp.396-400, doi: 10.1038/nature13872

16

Kim Y, Chen J 2018. . Molecular structure of human P-glycoprotein in the ATP-bound, outward-facing conformation. Science 359: , pp.915-919, doi: 10.1126/science.aar7389

17

Manolaridis I, Jackson SM, Taylor NMI, Kowal J, Stahlberg H, Locher KP 2018. . Cryo-EM structures of a human ABCG2 mutant trapped in ATP-bound and substrate-bound states. Nature 563: , pp.426-430, doi: 10.1038/s41586-018-0680-3

18

Martin C, Berridge G, Mistry P, Higgins C, Charlton P, Callaghan R 2000. . Drug binding sites on P-glycoprotein are altered by ATP binding prior to nucleotide hydrolysis. Biochemistry 39: , pp.11901-11906, doi: 10.1021/bi000559b

19

Martin C, Higgins CF, Callaghan R 2001. . The vinblastine binding site adopts high- and low-affinity conformations during a transport cycle of P-glycoprotein. Biochemistry 40: , pp.15733-15742, doi: 10.1021/bi011211z

20

McDevitt CA, Crowley E, Hobbs G, Starr KJ, Kerr ID, Callaghan R 2008. . Is ATP binding responsible for initiating drug translocation by the multidrug transporter ABCG2?. FEBS Journal 275: , pp.4354-4362, doi: 10.1111/j.1742-4658.2008.06578.x

21

McLean MA, Gregory MC, Sligar SG 2018. . Nanodiscs: a controlled bilayer surface for the study of membrane proteins. Annual Review of Biophysics , pp.107-124, doi: 10.1146/annurev-biophys-070816-033620

22

Nöll A, Thomas C, Herbring V, Zollmann T, Barth K, Mehdipour AR, Tomasiak TM, Brüchert S, Joseph B, Abele R, Oliéric V, Wang M, Diederichs K, Hummer G, Stroud RM, Pos KM, Tampé R 2017. . Crystal structure and mechanistic basis of a functional homolog of the antigen transporter TAP. PNAS 114: , pp.E438-E447, doi: 10.1073/pnas.1620009114

23

Payen LF, Gao M, Westlake CJ, Cole SP, Deeley RG 2003. . Role of carboxylate residues adjacent to the conserved core walker B motifs in the catalytic cycle of multidrug resistance protein 1 (ABCC1). Journal of Biological Chemistry 278: , pp.38537-38547, doi: 10.1074/jbc.M305786200

24

Rees DC, Johnson E, Lewinson O 2009. . ABC transporters: the power to change. Nature Reviews Molecular Cell Biology 10: , pp.218-227, doi: 10.1038/nrm2646

25

Ritchie TK, Grinkova YV, Bayburt TH, Denisov IG, Zolnerciks JK, Atkins WM, Sligar SG 2009. . Chapter 11 - Reconstitution of membrane proteins in phospholipid bilayer nanodiscs. Methods in Enzymology 464: , pp.211-231, doi: 10.1016/S0076-6879(09)64011-8

26

Robey RW, Pluchino KM, Hall MD, Fojo AT, Bates SE, Gottesman MM 2018. . Revisiting the role of ABC transporters in multidrug-resistant Cancer. Nature Reviews Cancer 18: , pp.452-464, doi: 10.1038/s41568-018-0005-8

27

Robson A, Gold VA, Hodson S, Clarke AR, Collinson I 2009. . Energy transduction in protein transport and the ATP hydrolytic cycle of SecA. PNAS 106: , pp.5111-5116, doi: 10.1073/pnas.0809592106

28

Schrödinger L 2015. The PyMOL Molecular Graphics System

29

Senior AE, Gadsby DC 1997. . ATP hydrolysis cycles and mechanism in P-glycoprotein and CFTR. Seminars in Cancer Biology 8: , pp.143-150, doi: 10.1006/scbi.1997.0065

30

Thomas C, Tampé R 2018. . Multifaceted structures and mechanisms of ABC transport systems in health and disease. Current Opinion in Structural Biology 51: , pp.116-128, doi: 10.1016/j.sbi.2018.03.016

31

Trowitzsch S, Tampé R 2020. . Multifunctional chaperone and quality control complexes in adaptive immunity. Annual Review of Biophysics 49: 081643, doi: 10.1146/annurev-biophys-121219-081643

32

Wieneke R, Tampé R 2019. . Multivalent chelators for in Vivo Protein Labeling. Angewandte Chemie International Edition 58: , pp.8278-8290, doi: 10.1002/anie.201811293

33

Zehnpfennig B, Urbatsch IL, Galla HJ 2009. . Functional reconstitution of human ABCC3 into proteoliposomes reveals a transport mechanism with positive cooperativity. Biochemistry 48: , pp.4423-4430, doi: 10.1021/bi9001908

34

Zollmann T, Moiset G, Tumulka F, Tampé R, Poolman B, Abele R 2015. . Single liposome analysis of peptide translocation by the ABC transporter TAPL. PNAS 112: , pp.2046-2051, doi: 10.1073/pnas.1418100112

35

Zutz A, Hoffmann J, Hellmich UA, Glaubitz C, Ludwig B, Brutschy B, Tampé R 2011. . Asymmetric ATP hydrolysis cycle of the heterodimeric multidrug ABC transport complex TmrAB from Thermus thermophilus. Journal of Biological Chemistry 286: , pp.7104-7115, doi: 10.1074/jbc.M110.201178

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

ATP-binding cassette transporters form one of the most important classes of transporters and the mechanisms of substrate binding and translocation have been studied for several decades. This report uses single-turnover analyses on a system reconstituted in liposomes to demonstrate that ATP binding and not ATP hydrolysis is the step which drives the transport of the substrate across the membrane. Such an approach might be highly valuable for other transport systems.

Decision letter after peer review:

Thank you for submitting your article "A single power stroke by ATP binding drives substrate translocation in a heterodimeric ABC transporter" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Olga Boudker as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Konstantinos Beis (Reviewer #2); Dirk Slotboom (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

All three reviewers like the approach and thoroughness of the work and agree that the results provide solid evidence on the previous assumptions that ATP binding and not hydrolysis drives the export of substrates by ABC transporters. They agree that the authors have setup an elegant system to study substrate transport by ABC transporters in detail and that your work is potentially applicable to any ABC transporter assuming that fluorescent labelled or radioligands are available.

No extra experiments are required but the reviewers asked for several clarifications.

Essential revisions:

None, but I would suggest that you address the points below.

Minor points:

1) The reviewers agree that the use of the EQ mutant is essential for this manuscript, however they would like you to discuss shortly to use the EQ mutant might affect their results.

2) In subsection “The IF-to-OF transition triggered by ATP binding is coupled to substrate translocation” the authors suggests that ATP binding and not hydrolysis drives transport, based on their observation that after 5 minutes incubation they see “a substantially higher number of peptides” transported and their observation that ATP binding is much faster than ATP hydrolysis. As long as the amount is not quantified or placed in relation to the total amount of peptides transported, no difference can be made between full binding or only a partial hydrolysis of ATP. The authors set-out to do this quantification. I would suggest that the authors make their suggestion that the powerstroke is driven by binding and not by hydrolysis after this quantification.

3) In Figure 2F the error bars seem to be missing.

4)The authors state that “each liposome contained ~30 correctly-oriented, energizable transport complexes.” Can they clarify what evidence they have apart from the transport of peptides, about the number of molecules? Have they quantified from their fluorescent labelled protein?

5) Figure 2B could benefit by having a darker contrast.

6) The Results section is extremely condensed and hard to read. Since eLife does not have a firm word limit to manuscripts, the authors really should do better in explaining the (nice) experiments. In addition, I find the discussion a bit shallow, and disappointing considering the beautiful data.

7) Subsection “A single ATP turnover and phosphate release trigger the OF-to-IF return”: "as described above" Removal of ATP was not described above.

8) Figure 2D and related supplement figure: The decay does not look monoexponential. I suspect that there are 2 different koff values: One for the dissociation of nucleotide from the canonical site (after hydrolysis) and the other for dissociation from the degenerate site. In that respect, it is surprising that in Figure 2F the nucleotide from the degenerate site does not seem to come off. Why does more than 80% of bound nucleotide dissociate in the experiment shown in 2D (this number indicates that dissociation from both sites takes place), but only 50% in 2F (this number suggests that ATP in the degenerate site does not come off).

9) Figure 2E: In the "No nucleotide" control there is a clear increase in bound peptide over the time course of the experiment. This must be explained. My guess is that ATP co-purifies with the protein.

10) Figure 2—figure supplement 1 panel B: At 4 degrees, only 50% of bound nucleotide is released. Explain. It may be that the dissociation from the degenerate site is much slower at 4 degrees than at higher temp, or that the dissociation from the canonical site is much slower. To tell these two events apart would really give new insight.

11) Subsection “A single ATP turnover and phosphate release trigger the OF-to-IF return”: The difference in half-life at the different temperatures is very small. Given the large conformational change needed for the dissociation to occur, I find the marginal temperature dependence surprising. Explain.

12) Figure 1D and Figure 1—figure supplement 2 panel B: It is unclear whether these experiments are identical or not. If yes: why show it twice? If not: explain the difference better.

13) Figure 1B and Figure 1—figure supplement 1 panel C: why are the absolute values (cpm) so different?

14) Subsection “ATP binding drives substrate translocation monitored by single-liposome assays”: it is not clear what "untreated" refers to.

15) Figure 3—figure supplement 1 panel A. It is unclear what the panels show, and there is no explanation in the text. Why are the instant blue data different from the α-His data? The samples were incubated for the same amount of time, right? Why is there no quantification shown? From the qualitative images, the authors seem to draw quantitative conclusion. Please show proper quantification.

16) Figure 5C. I do not see biphasic behavior, and do not see the offset at 35 per liposome. I guess the wrong figure is shown?

17) In general, the authors should not oversell their nice work by using words like "ultrasensitive", unscientific qualifications like "almost unlimited". Also, it does not help the reader if it is unclear what "this" and "these" refer to. For instance, in paragraph two of the Introduction "this question": there was no question asked. Paragraph three: “these questions”: again, I cannot find questions.

Essential revisions:

None, but I would suggest that you address the points below.

Minor points:

1) The reviewers agree that the use of the EQ mutant is essential for this manuscript, however they would like you to discuss shortly to use the EQ mutant might affect their results.

Thank you for your comment. Using a catalytically reduced variant in combination with Mg-ATP appeared as an effective strategy to populate the OF conformation as reported for TmrAB but also for other ABC exporters (Hofmann et al., 2019; Manolaridis et al., 2018; Kim et al., 2018; Dastvan et al., 2019; Barth et al., 2018). In order to investigate the events of ATP binding and ATP hydrolysis with their respective effects on substrate translocation, we utilized the TmrAEQB variant with a drastic slowdown in ATP turnover. This allowed us to kinetically outline the IF-to-OF switch and the returning OF-to-IF transition to correlate these conversions with substrate translocation. Considering wildtype TmrAB, single-turnover transport studies are technically challenging and would require conformation-selective inhibitors that arrest TmrAB at various stages of the translocation cycle which to our knowledge are presently not available. We observed an effective coupling ratio of consumed ATP per translocated substrate for TmrAEQB in contrast to TmrAB which acts largely uncoupled (Bock et al., 2019). As the speed of conformational transitions is slowed down for TmrAEQB compared to TmrAB, the allosteric coupling might be favored. For explanation, we extended the main text accordingly.

2) In subsection “The IF-to-OF transition triggered by ATP binding is coupled to substrate translocation” the authors suggests that ATP binding and not hydrolysis drives transport, based on their observation that after 5 minutes incubation they see “a substantially higher number of peptides” transported and their observation that ATP binding is much faster than ATP hydrolysis. As long as the amount is not quantified or placed in relation to the total amount of peptides transported, no difference can be made between full binding or only a partial hydrolysis of ATP. The authors set-out to do this quantification. I would suggest that the authors make their suggestion that the powerstroke is driven by binding and not by hydrolysis after this quantification.

We like to thank the reviewers for this comment. We have changed the main text accordingly.

3) In Figure 2F the error bars seem to be missing.

The error bars were already displayed in the diagram but hard to read. We have changed the figure accordingly.

4)The authors state that “each liposome contained ~30 correctly-oriented, energizable transport complexes.” Can they clarify what evidence they have apart from the transport of peptides, about the number of molecules? Have they quantified from their fluorescent labelled protein?

Thank you for this helpful comment. We would like to refer to Figure3—figure supplement1. Proteoliposomes were characterized by nanoparticle tracking to calculate the average diameter. TmrAEQB was randomly oriented (50/50%) in proteoliposomes determined by TEV-protease cleavage of the His-tag fused to TmrA. Incorporation efficiency of TmrAEQB in proteoliposomes was analyzed by carbonate crush and sucrose density gradient floatation. Here, 95% of TmrAEQB were reconstituted. These three parameters were used to calculate the average number of correctly-inserted TmrAEQB complexes per proteoliposome. We added further information in the main text for clarification.

5) Figure 2B could benefit by having a darker contrast.

Thanks for the comment. We originally kept the illustration conservatively, however changed the panel accordingly.

6) The Results section is extremely condensed and hard to read. Since eLife does not have a firm word limit to manuscripts, the authors really should do better in explaining the (nice) experiments. In addition, I find the discussion a bit shallow, and disappointing considering the beautiful data.

Thanks for the comment. We have changed the main text accordingly.

7) Subsection “A single ATP turnover and phosphate release trigger the OF-to-IF return”: "as described above" Removal of ATP was not described above.

Thank you for your comment. We have corrected the main text.

8) Figure 2D and related supplement figure: The decay does not look monoexponential. I suspect that there are 2 different koff values: One for the dissociation of nucleotide from the canonical site (after hydrolysis) and the other for dissociation from the degenerate site. In that respect, it is surprising that in Figure 2F the nucleotide from the degenerate site does not seem to come off. Why does more than 80% of bound nucleotide dissociate in the experiment shown in 2D (this number indicates that dissociation from both sites takes place), but only 50% in 2F (this number suggests that ATP in the degenerate site does not come off).

Thank you for your comment. In scintillation proximity assays (SPA), ATP binding signals are slightly declining over time. This linear decrease is underlying the exponential decay of nucleotide dissociation. In principle, the signals could be corrected by the linear decrease. However, we decided to present the original raw data without further background subtraction. Regarding Figure2F, we analyzed the identity of nucleotides that were occluded by TmrAEQB along the IF-to-OF switch after the removal of free ATP. After the hydrolysis event, hydrolysis products (ATP, ADP, and Pi) are released from TmrAEQB. Therefore, over time the ATP to ADP ratio is approaching 1/1. In the main text, we have clarified these differences.

9) Figure 2E: In the "No nucleotide" control there is a clear increase in bound peptide over the time course of the experiment. This must be explained. My guess is that ATP co-purifies with the protein.

Thank you for your comment. When measuring peptide binding by SPA, we observe a slight increase of binding over longer time periods. Since TmrAEQB is immobilized on SPA beads, we suggest that peptide equilibration takes longer. We have changed the main text accordingly.

10) Figure 2—figure supplement 1 panel B: At 4 degrees, only 50% of bound nucleotide is released. Explain. It may be that the dissociation from the degenerate site is much slower at 4 degrees than at higher temp, or that the dissociation from the canonical site is much slower. To tell these two events apart would really give new insight.

Thank you for your comment. It is an interesting observation that the koff rate is decreased at 4°C while the amount of released nucleotide is reduced to 50%. We assume an asymmetric opening of both nucleotide-binding sites (NBSs) which is in line with previous reports (Barth et al., 2018, J Am Chem Soc). However, we can only speculate about the asymmetric contributions of both NBSs along the OF-to-IF transition.

11) Subsection “A single ATP turnover and phosphate release trigger the OF-to-IF return”: The difference in half-life at the different temperatures is very small. Given the large conformational change needed for the dissociation to occur, I find the marginal temperature dependence surprising. Explain.

Thank you for your comment. Since the OF-to-IF transition is initiated by a single turnover of ATP, we expected a large temperature dependency between 4°C, 20°C and 45°C. However, the physiological temperature of TmrAB is 68°C (Zutz et al., 2011), making it impossible to monitor nucleotide dissociation by SPA. For lower temperatures, the OF-to-IF return appears to show marginal temperature dependency.

12) Figure 1D and Figure 1—figure supplement 2 panel B: It is unclear whether these experiments are identical or not. If yes: why show it twice? If not: explain the difference better.

Thanks for your comment. We utilized two different substrate peptides and two distinct methods to confirm the ATP-induced abrogation of substrate binding. We have changed the main text accordingly.

13) Figure 1B and Figure 1—figure supplement 1 panel C: why are the absolute values (cpm) so different?

Thanks for your comment. In SPA, radiolabeled ligands are added in trace amounts to reduce background signals. In Figure1—figure supplement1C, the raw data is shown, while in Figure1B the signals are multiplied by the dilution factor of radiolabeled compound, resulting in much larger values. This is now explained more clearly in the figure legends.

14) Subsection “ATP binding drives substrate translocation monitored by single-liposome assays”: it is not clear what "untreated" refers to.

Thank you for your comment. We have changed the main text accordingly.

15) Figure 3—figure supplement 1 panel A. It is unclear what the panels show, and there is no explanation in the text. Why are the instant blue data different from the α-His data? The samples were incubated for the same amount of time, right? Why is there no quantification shown? From the qualitative images, the authors seem to draw quantitative conclusion. Please show proper quantification.

Thank you for your comment. The α-His immunoblot was used to determine the fraction of cleaved TmrA calculating the orientation of TmrAB in proteoliposomes. Equal loading of proteoliposomes was controlled by instant blue staining. We have explained this approach more clearly in the main text.

16) Figure 5C. I do not see biphasic behavior, and do not see the offset at 35 per liposome. I guess the wrong figure is shown?

Thanks for your comment. We observed an initial burst phase followed by a linear increase of the number of transported peptides per liposome for longer time periods. The offset of ~30 transported peptides per liposome resembles the y-intercept of the linear term of the regression function. We have explained this evaluation more clearly in the main text.

17) In general, the authors should not oversell their nice work by using words like "ultrasensitive", unscientific qualifications like "almost unlimited". Also, it does not help the reader if it is unclear what "this" and "these" refer to. For instance, in paragraph two of the Introduction "this question": there was no question asked. Paragraph three: “these questions”: again, I cannot find questions.

We have changed the main text accordingly.

Supplementary materials
• elife-55943-transrepform.pdf info     save_alt
• elife-55943-fig5-data1.docx info     save_alt
• elife-55943-fig4-figsupp1-data1.docx info     save_alt
• elife-55943-fig2-data1.docx info     save_alt
• elife-55943-fig3-data1.docx info     save_alt
• elife-55943-fig1-data1.docx info     save_alt

Citing articles via
https://www.researchpad.co/tools/openurl?pubtype=article&doi=10.7554/eLife.55943&title=A single power stroke by ATP binding drives substrate translocation in a heterodimeric ABC transporter&author=Erich Stefan,Susanne Hofmann,Robert Tampé,Sonja V Albers,Olga Boudker,Sonja V Albers,Konstantinos Beis,Dirk Slotboom,&keyword=conformational transition,membrane proteins,molecular machines,peptide transport,transporter associated with antigen processing,TAP,single-turnover analysis,ABC transporter,multidrug resistence,E. coli,Human,&subject=Research Article,Biochemistry and Chemical Biology,Structural Biology and Molecular Biophysics,