Figures and data
Accessible-volume (AV) simulations of LD fluorophores on TmrAB variants.
AV simulations were performed for LD555 (donor) and LD655 (acceptor) fluorophores attached to the selected TmrAB labeling sites to assess whether donor–acceptor distances are suitable for smFRET measurements30.AV simulations confirm that donor–acceptor distances (RDA) remain within the FRET-sensitive range in both conformations, predicting measurable changes in simulated FRET efficiencies (Esim). Cryo-EM structures of TmrAB reconstituted in lipid nanodiscs11 were used as templates, including apo and ATP-bound (3 mM ATP) states. Outward-facing open (OFopen) and outward-facing occluded (OFoccluded) structures were obtained via orthovanadate trapping (Vi) or the slow-turnover mutant TmrAE523QB (EQ).
Accessible-volume (AV) simulations of LD fluorophores on TmrAB variants.
AV simulations were performed for LD555 (donor) and LD655 (acceptor) fluorophores attached to the selected TmrAB labeling sites to assess whether donor–acceptor distances are suitable for smFRET measurements30. a, TmrABNBD (TmrAC416BL458C) and b, TmrABPG (TmrAC416A, T61CBR56C) in the inward-facing wide (IFwide; PDB: 6RAN, left) and outward-facing open (OFopen; PDB: 6RAH, right) conformations. The approximate membrane position is indicated by the dashed grey line. TmrA is shown in blue with LD655 (orange) and TmrB in yellow with LD555 (green), with labeling positions indicated (416/458 for TmrABNBD; 61/56 for TmrABPG). For illustration, fluorophores are shown in a defined configuration; however, in experiments donor and acceptor dyes are stochastically attached, resulting in random distribution between subunits.
ATP-induced conformational changes of TmrAB analyzed by smFRET.
a, Experimental setup. TmrABNBD (left) and TmrABPG (right) were labeled with LD555/LD655 and immobilized on PEGylated coverslips via a biotinylated conformation-independent, TmrB-specific nanobody (Nb9F10S63C)11,14. b, smFRET imaging was performed using total internal reflection fluorescence (TIRF) microscopy with alternating laser excitation (ALEX; donor: 532 nm; acceptor: 640 nm). Emission was collected in donor (498-620 nm) and acceptor (662-710 nm) channels. Donor emission upon donor excitation 
ATP-dependent shifts in smFRET populations of TmrAB.
a,c, Increasing ATP concentrations (0-3 mM) progressively shifted the population between apo and ATP-bound conformations for (a) TmrABNBD and (c) TmrABPG. FRET efficiency (E) histograms were fitted with two Gaussian populations corresponding to the ATP-free state (blue; defined from apo samples) and the ATP-bound state (orange; determined from saturating ATP conditions). Dotted vertical lines indicate the mean E values of each state, and relative fractions (Gaussian areas) are summarized schematically in each panel. b,d, ATP-binding curves obtained by plotting the fraction of molecules in the ATP-bound states as a function of ATP concentration for (b) TmrABNBD (reporting NBD dimerization) and (d) TmrABPG (reporting PG opening). Data were fitted with a Langmuir isotherm to determine the apparent dissociation constant Kd, ATP of each variant.
Identification of the outward-facing open (OFopen) conformation.
a–c, Three complementary approaches were used to resolve the OFopen state: (a) the slow-turnover variant TmrABPG_EQ, (b) imaging in Mg2+-free buffer supplemented with EDTA, and (c) stabilizing via trans-inhibition using high concentrations of peptide substrate R9L (0.3–2 mM). FRET efficiency (E) histograms were fitted with three Gaussian populations corresponding to the ATP-free state (blue), the ATP-bound state (orange), and the OFopen state (green). All three strategies reveal a distinct OFopen population. Dotted vertical lines indicate the mean E values of each state, and relative fractions (Gaussian areas) are summarized schematically in each panel. d, Comparison of inter-residue distances. Distances (Å) between selected residues on the NBDs and PG of TmrAB (Cβ–Cβ) were determined using smFRET (this study, detergent-solubilized TmrAB), cryo-EM structures of nanodisc-reconstituted TmrAB (PDB 6RAH, 6RAN)11; accessible-volume (AV) simulations (this study), and PELDOR/DEER measurements of detergent-solubilized TmrAB16.
Conformational state distribution and catalytic cycle of TmrAB under active turnover.
Schematic of the TmrAB transport cycle summarizing major conformational states and their estimated population distributions under physiological ATP concentrations (3 mM, 40 °C). a, The inward-facing apo state (IFnarrow and IFwide; blue arc) accounts for ∼20% of molecules and is characterized by separated NBDs and a cytosol-accessible substrate-binding cavity. Substrate binding stabilizes the IFwide conformation11. ATP binding induces NBD dimerization and formation of the ATP-bound ensemble. b,c, Under substrate-bound turnover conditions, TmrAB proceeds via (b) OFoccluded and (c) OFopen states in which substrate release occurs. In steady-state turnover, the ATP-bound ensemble rapidly interconverts between OFoccluded and OFopen, accounting for ∼25% of the ATP-bound population (green circle). These transitions occur faster than the ∼200 ms temporal resolution of smFRET measurements, resulting in an averaged signal under turnover conditions. OFoccluded is proposed as an obligate intermediate between IF and OFopen, preventing substrate backflow by maintaining an occluded binding cavity during rearrangements of the PG and NBDs. Although a substrate-bound OFoccluded state has not been directly observed for TmrAB, its existence is supported by structures of the homodimeric type IV transporter BmrA47. Reduced ATP hydrolysis or substrate trans-inhibition enables trapping of the OFopen. state. d,e, ATP hydrolysis and phosphate (Pi) release generate post-hydrolysis return states (d) URasym and (e) URasym*. Subsequent ADP release restores the apo IF conformation, completing the transport cycle. Overall, the ATP-bound phase (b–e) represents ∼55% occupancy (orange arc) with an estimated dwell time of ∼310 ms, whereas the apo/ATP-rebinding phase (a) lasts ∼90 ms, yielding a total cycle time of ∼400 ms (kcat = 2.57 s−1). TmrA is shown in blue, TmrB in yellow, substrate as a green diamond, and nucleotides as orange symbols. Dotted grey boxes indicate the approximate position of the NBD dimer interface.
Quality of TmrAB purification and fluorophore labeling.
a, SDS-PAGE analysis (10%, reducing gel, Coomassie staining) of successive purification steps: M, molecular weight marker; L, cell lysate; B, Ni-NTA beads after incubation with lysate; FT, flow-through; W, wash; E, eluted TmrAB; BE, buffer-exchanged sample; C, concentrated protein; CFT, concentrator flow-through; 1 and 2, first and second peaks eluted from size-exclusion chromatography (SEC). Only the second peak was used for subsequent FRET experiments. b, SEC (Superdex 200 increase 10/300 GL) showing monodisperse labeled TmrAB and efficient removal of free fluorophores. A representative chromatogram of TmrABPG is shown. c, ATP hydrolysis activity of purified wild-type TmrAB (60 nM TmrABwt) measured at 40 °C for 7 min using the Malachite Green assay. Released inorganic phosphate (Pi) was quantified and data fitted to Michaelis-Menten kinetics, yielding Km = 0.97 ± 0.28 mM and kcat = 2.57 ± 0.38 s-1. d–f, Analytical SEC (Superdex 200 increase 3.2/300) used to determine fluorophore labeling efficiencies of (d) TmrABNBD, (e) TmrABPG, and (f) TmrABPG_EQ. LD555 and LD655 labeling efficiencies were ∼55 and ∼53% (TmrABNBD), ∼43 and ∼52% (TmrABPG), and ∼42% and ∼52% (TmrABPG_EQ), respectively. Total cysteine occupancy exceeds 90% in all variants, reflecting near-complete dual-labeling with equimolar fluorophores.
FRET properties of labeled TmrAB variants.
a, Time-correlated single-photon counting histograms of LD555 (left) and LD655 (middle) measured for free dye in buffer (black), LD555/LD655-labeled TmrABNBD (orange), and LD555/LD655-labeled TmrABPG (blue). Amplitude-weighted average fluorescence lifetimes (τ) are summarized in the histogram (right), confirming sufficient fluorophore mobility for reliable FRET measurements. b–d, Donor-excited ensemble emission spectra (550–700 nm, excitation 520 nm) of stochastic LD555/LD655-labeled (b) TmrABNBD, (c) TmrABPG, and (d) the slow-turnover variant TmrABPG_EQ measured at increasing ATP concentrations. Spectra were normalized to donor intensity in the apo state. ATP-dependent donor quenching and acceptor sensitization demonstrate that all variants retain FRET capability. e–g, Fractional fluorescence changes, (F-F0)/F0, plotted as a function of ATP concentration for (e) TmrABNBD, (f) TmrABPG, and (g) TmrABPG_EQ, where F is the acceptor emission intensity and F0 the intensity in the apo state. Data were fitted with a hyperbolic (Langmuir-type) binding model to determine apparent Kd, ATP values, which are consistent with ensemble FRET measurements of ATP binding.
Effect of fluorophore labeling and nanobody binding on TmrAB transport activity.
Single-liposome transport assays were performed by flow cytometry using mean fluorescence intensity at 527 nm (MFI527) as a readout of peptide transport. a, TmrAB (0.6 µM) reconstituted into liposomes (∼50 transporters per liposome) was incubated with C4F peptide (30 µM; RRYCFKSTEL, F, fluorescein) in the presence of 3 mM ATP (green) or 3 mM ADP (orange) at 40 °C for 5 min. After extensive washing, proteoliposomes were analyzed by flow cytometry. b, Flow-cytometry gating strategy. Forward scatter area (FSC-A) versus side scatter area (SSC-A; left) was used to identify homogeneous proteoliposomes, forward scatter height (FSC-H) versus FSC-A (middle) to select singlets, and fluorescence intensity at 527 nm (F527; right) to determine the mean fluorescence intensity (MFI) values of the gated population. c, Transport activity of unlabeled wild-type TmrAB (TmrABWT) compared with LD555/LD655-labeled TmrABPG. d, Effect of nanobody binding on TmrABwt transport activity. Assays were performed in the absence or presence of Nb9F10S63C at a 1:1 molar ratio (0.6 µM each). Fluorophore labeling at the periplasmic gate and nanobody binding did not measurably affect TmrAB transport activity.
Representative smFRET traces of TmrABNBD.
Representative single-molecule FRET (smFRET) traces of TmrABNBD recorded (a) in the absence of ATP and (b, c) in the presence of 3 mM ATP. Hidden Markov modeling (HMM) classified ATP-bound traces as (b) static or (c) dynamic based on the absence or presence of transitions between ATP-free and ATP-bound conformational states. Donor fluorescence upon donor excitation is shown in green, acceptor fluorescence upon donor excitation in orange, FRET efficiency (E) in black, and stoichiometry (S) in grey.
Representative smFRET traces of TmrABPG.
Representative single-molecule FRET (smFRET) traces of TmrABPG recorded (a) in the absence of ATP and (b, c) in the presence of 3 mM ATP. Hidden Markov modeling (HMM) classified ATP-bound traces as (b) static or (c) dynamic based on the absence or presence of transitions between ATP-free and ATP-bound conformational states. Donor fluorescence upon donor excitation is shown in green, acceptor fluorescence upon donor excitation in orange, FRET efficiency (E) in black, and stoichiometry (S) in grey.
