Single-molecule visualization of ATP-induced dynamics of the subunit composition of an ECF transporter complex under turnover conditions

Abstract

Energy-Coupling Factor (ECF) transporters are ATP-binding cassette (ABC) transporters essential for uptake of vitamins and cofactors in prokaryotes. They have been linked to pathogen virulence and are potential targets for antimicrobials. ECF transporters have been proposed to use a unique transport mechanism where a substrate-translocating subunit (S-component) dynamically associates with and dissociates from an ATP-hydrolyzing motor (ECF module). This model is contentious, because it is based largely on experimental conditions without compartments or continuous bilayers. Here, we used single-molecule spectroscopy to investigate the conformational dynamics of the vitamin B12 transporter ECF-CbrT in membranes under vectorial transport conditions. We observed ATP hydrolysis-dependent dissociation of the S-component CbrT from, and re-association with the ECF module, in absence and presence of vitamin B12 consistent with futile ATP hydrolysis activity. The single-molecule spectroscopy experiments suggest that S-component expulsion from and re-association with the ECF module are an integral part of the translocation mechanism.

Fig. 1. Confocal and TIRF smFRET of ECF-CbrT sensors in apo condition.

a Structure of Ecf-CbrT (pdb 6fnp) with the four subunits represented in different colors. b Schematic representation of confocal smFRET, an ECF-CbrT complex in detergent micelle diffuses through the confocal volume. c Distribution of bursts measured by confocal microscopy showing each one main FRET population for the H- and C-sensors (total number of bursts cumulated from technical triplicate recordings are indicated). d FRET efficiency distribution of donor and acceptor containing bursts (stoichiometry between 0.25 and 0.85) showing a maximum at 0.68 and 0.56 for the H- and C-sensors, respectively (number of bursts indicated). e Schematic representation of TIRF smFRET, with a proteoliposome immobilized on a PEG-coated, neutravidin-treated surface via biotinylated DOPE lipids. Biotin and neutravidin are shown as grey diamonds and a purple cross, respectively. f Representative TIRF traces for H and C-sensors in apo condition, top panels display the donor (green) and acceptor (red) intensities, and the bottom panels display the FRET efficiency. The colored and black lines are the signal before and after Chung-Kennedy filtering, respectively (see “Methods” section). The first and last 5 s (pink shaded areas) correspond to direct acceptor excitation. g Distribution of FRET efficiency for H and C-sensors (calculated from signal after CK filtering) showing a maximum at 0.72 and 0.60, respectively (total number of traces is indicated, see Supplementary Table 3 for replicates information).

Fig. 2. Dynamics of the ECF-CbrT complex in liposomes using smTIRF.

a Representative dynamic FRET traces showing dissociation (top panels), association (middle panels), and multiple transitions (bottom panels). The blue and red bars on top indicate the conditions without or with vitamin B12 in the lumen, respectively. In both cases, ATP was present on the outside (ATP_out). Donor and acceptor intensities and FRET efficiency are shown before (green, red, and blue, respectively) and after (black) CK filtering. The first and last 5 s (pink shaded areas) correspond to direct acceptor excitation. b Distribution of measured FRET efficiencies in the indicated conditions (number of traces is indicated in Fig. 2c, see Supplementary Table 3 for replicates information). c Percentage of dynamic traces in different conditions, the weighted average and standard deviation are shown, with the black circles representing the values for individual experiments. After a Welch’s ANOVA test (right-tailed) using weighted average and standard deviation followed by a Tukey-HSD test (two-tailed), the apo condition was found to be significatively different from the ATP_out and the ATP_out + B12_in conditions (p = 0.000378 and p = 0.000021 respectively), while differences between other conditions were not significant (p = 0.063662; 0.162573; 0.222560 and 0.318457 for Apo vs. ATP_{in & out}; ATP_{in & out} vs. ATP_out; ATP_{in & out} vs. ATP_out + B12_in and ATP_out vs. ATP_out + B12_in respectively). The total number of traces for each condition is indicated in the table (See Supplementary Table 3 for replicates information).

Fig. 3. Co-reconstitution of single-labeled ECF-CbrT complexes.

a Representative traces in apo (top trace) condition or in the presence of ATP (middle and bottom traces), showing no FRET, static and dynamic FRET, respectively. Donor and acceptor intensities and FRET efficiency are shown before (green, red, and blue, respectively) and after (black) CK filtering. The first and last 5 s (pink shaded areas) correspond to direct acceptor excitation. b Schematic representation of the experimental conditions. Ecf-CbrT single-cysteine mutants expressed, purified, and labeled separately are co-reconstituted in the same liposomes with the indicated protein to lipid ratios (EcfA_K122C_Alexa Fluor 647 shown with the red ATPase and CbrT_A182C_Alexa Fluor 555 shown with the green S-component). c Distribution of FRET efficiencies in the indicated conditions. In the apo and ATP_out + B12_in conditions, reconstitution with EcfA_K122C_Alexa Fluor 647 at a 1:1.000 ratio was used, the 1:2.000 was used for the ATP only and B12 only conditions. d Fraction of static high-FRET (solid colors), low-FRET (no color), and dynamic traces (dashed) in different conditions. Numbers of analyzed traces are shown in the table below the bars (# FRET corresponds to High-FRET + Dynamic traces), see Supplementary Table 3 for replicates information.

Fig. 4. Effect of vitamin B12 and ATP on the ECF-CbrT complex in liposomes and rate of dissociation.

Conditions in the presence or absence of 10 μM vitamin B12 in the proteoliposomes lumen are shown in red and blue, respectively. a Peak-to-peak ratio of the dissociated (FRET efficiency of 0.08) and associated (0.60) populations as measured in the FRET Efficiency distributions. Weighted average and standard deviation are shown, individual experiment values in black circles (See Supplementary Table 3 for replicates information). b ATP dependence of the dissociated/associated ratio. c ATP dependence of the percentage of dynamic traces. Weighted average and standard deviation are shown (See Supplementary Table 3 for replicates information). d Time of first dissociation event upon ATP addition. The fraction of traces for which a transition in FRET efficiency has not yet occurred after the addition of ATP in the outside buffer is shown as a function of the recording time. As a comparison, binding of maltose to maltose-binding protein (MBP) was recorded (grey). At time t = 0, the pump starts. The activation solution (ATP or maltose) reaches the channel in a few seconds. Within traces classified as dynamic, dissociation was observed in 36 and 43 cases upon ATP addition to proteoliposomes with or without loaded vitamin B12, respectively (measured over 6 and 9 recording sessions, respectively). In the MBP control, a double mutant T36C/S352C labeled with Alexa Fluor 555 and 647 was immobilized, and recording was done with the addition of 200 μM maltose. A transition from 0.6 to 0.8 FRET efficiency was recorded in 86 traces corresponding to maltose binding to MBP (3 recording sessions). See Supplementary Fig. 4 for further details. A one-exponential phase decay fitting gave half-lives of 2.5 s for MBP, and 7 and 10 s for the conditions ATP_out + B12_in and ATP_out, respectively. Two-tailed unpaired t tests showed that the difference between decay rates exponential fits in the conditions ATP_out + B12_in and ATP_out are non-significant (P = 0.95), while they are both significantly slower than maltose binding to MBP (P < 0.0001 in both cases).