Introduction

ATP-binding cassette (ABC) transporters comprise a large family of primary active transporters found in all domains of life that mediate the ATP-driven translocation of a wide range of substrates into or out of the cell or organelles. They are generally considered to be unidirectional transporters in physiological conditions, acting as either importer or exporter1,2,3,4,5,6,7, with the notable exception of CFTR that functions as an ATP-gated chloride channel8. The cytosolic nucleotide-binding domains (NBDs) of ABC transporters are highly conserved and represent the hallmark of the transporter family, whereas transmembrane domains (TMDs) that form the translocation pathway are more diverse among the members. A recent classification of ABC transporters based on the folding of these more variable TMDs placed all known members into seven distinct groups (type I−VII)2. Type I−III ABC transporters act strictly as importers and are found exclusively in prokaryotes, while the type IV−VII mostly include export systems2. Type IV ABC transporters represent an interesting class in which most members have an export function, but some transporters have recently been shown to be importers, including Rv1819c from Mycobacterium tuberculosis (also referred to as BacAMt)9,10.

A genetic screen using random transposon insertions revealed that BacAMt is the sole transporter responsible for the import of cobalamin (vitamin B12) in M. tuberculosis11. This import may be essential, because unlike other mycobacteria, M. tuberculosis does not appear to have the ability to synthesize the micronutrient de novo12. The protein is important to maintain chronic infections in mice13. Additionally, BacAMt has been implicated in the import of other structurally unrelated hydrophilic compounds, including several antimicrobial compounds, such as the proline-rich peptide Bac7(1-16)13,14, bleomycin9,11,13, evybactin15 and aminoglycosides such as streptomycin16. As it is unlikely that BacAMt has evolved for the specific import of these toxic compounds, it is assumed that they are taken up into the cell as part of a moonlighting activity of the protein. The relevance of the observed polyspecificity or promiscuity is poorly understood. A recently solved cryo-EM structure of BacAMt in the nucleotide-bound occluded state has shed some light on its polyspecificity9. The protein possesses a large internal cavity of ~7700 Å3 that spans the bilayer thickness. The cavity’s surface is lined with negatively charged and polar residues, and appears to lack an obvious high-affinity binding site for its substrates. The cavity instead provides a generally favorable environment for the accommodation of hydrophilic compounds, which may explain how structurally unrelated compounds can be transported.

Although it is generally viewed that ABC transporters act as unidirectional transporters, it has also been speculated that BacAMt not only imports substrates, but may also function as an exporter for other substrates. Several whole-cell-based studies have shown that the treatment of drug-resistant clinical isolates of M. tuberculosis with rifampicin and isoniazid is associated with mutations or increased expression levels of BacAMt17,18,19,20,21,22,23, which is a common observation for drug-efflux pumps24. Accordingly, heterologous expression of BacAMt in M. smegmatis increases its resistance to rifampicin, further supporting its potential role in the export of the drugs from the cell23. BacAMt is also essential for M. tuberculosis to protect itself against the cysteine-rich host peptide human β-defensin (HBD2), as the bacterium becomes more susceptible to the peptide upon the deletion of bacAMt14. At the same time, its heterologous expression increases the resistance of Sinorhizobium meliloti to HBD2 and another cysteine-rich peptide NCR24714, suggesting that BacAMt may also actively transports these peptides out of the cell. It is noteworthy that the characterization of the substrate transport processes mediated by BacAMt has so far been limited to whole-cell-based assays, which are indirect, sometimes difficult to interpret, and complicate mechanistic interpretation of the import and export functions.

In this work, we establish a robust in vitro fluorescence-based transport assay for cobalamin to measure the transport activity of purified and liposome-reconstituted protein. Using BacA from Mycobacterium marinum (BacAMm, or MMAR2696, 88.2% sequence identity with BacAMt, Supplementary Fig. 5), we demonstrate that the transport system indeed operates in the import direction for the substrate cobalamin, as it had been shown previously in whole-cell assays9,11. Unexpectedly, we show that BacAMm also mediates ATP-driven transport of cobalamin in the export direction, demonstrating that a member of the ABC transport family acts as a bidirectional transporter for a single physiological substrate. This work shows that the mechanistic diversity of ABC transporters is larger than previously thought, and that functions typically associated with secondary transporters can also be incorporated in an ABC transporter.

Results

Fluorescence-based transport assay for cobalamin

We used the previously engineered single-cysteine variant D141C of the cobalamin-specific substrate binding protein (SBP) BtuF from Escherichia coli to generate a fluorescent sensor for cobalamin25. We purified the protein and labelled the cysteine with a maleimide version of fluorophore Alexa Fluor 488 yielding the chemically modified protein BtuF488 (Supplementary Fig. 1a). Binding of cobalamin to BtuF488 brings the ligand in close proximity to the conjugated fluorophore, which has previously been shown to quench the fluorescence signal by 76%25. To establish the usefulness of the sensor for high-throughput transport measurements in a 96-well plate format, we titrated cobalamin to the sensor in solution and monitored the fluorescence change. We observed a substrate concentration dependent quenching to a maximum of nearly 75% and calculated a dissociation constant (Kd) of BtuF488 for cobalamin of 11.5 nM (Fig. 1a), which is similar to the value reported previously26,27.

Fig. 1: Fluorescence-based transport assay for cobalamin.
figure 1

a Fluorescence quenching of 25 nM BtuF488 in solution in the presence of various concentrations of cobalamin. The Kd value is 11.5 nM. b Release of cobalamin from proteoliposomes reconstituted with the cobalamin transporter ECF-CbrT or the folate transporter ECF-FolT2. Proteoliposomes were loaded with 13 µM cobalamin and 25 nM of the sensor BtuF488 was added to the external buffer. The release of cobalamin was monitored for liposomes reconstituted with the cobalamin-specific transporter ECF-CbrT in the presence of 10 mM external Mg-ATP (yellow circles), 10 mM external Mg-ADP (white triangles) or 10 mM luminal Mg-ATP (blue circles). The folate-specific transporter ECF-FolT2, which is closely related to ECF-CbrT, was used as negative control in the presence of 10 mM external Mg-ATP (yellow triangles) or 10 mM external Mg-ADP (grey triangles). Cartoons on the right side of the graph explain the experimental setup. The liposome membrane is shown as grey circle, the sensor in the external buffer in yellow, cobalamin as red dots, the active ECF-CbrT population in colours, and the inactive population in grey. Export and import directions are indicated by blue and yellow arrows, respectively. The black cross indicates that export is not catalyzed by ECF-CbrT. c Accumulation of cobalamin into liposomes reconstituted with ECF-CbrT. Proteoliposomes were loaded with 2 µM BtuF488 inside the lumen and 65 µM cobalamin was added externally. Transport activity was measured in the presence of 10 mM internal Mg-ATP (yellow circles), 10 mM external Mg-ATP (blue circles) or 10 mM external Mg-ADP (white triangles). Cartoons on the right side of the graph explain the experimental setup, as in (b). a−c The data were obtained from biological duplicates each containing technical triplicates. The data is presented as the mean with error bars indicating the standard deviation calculated from all individual data points (n = 6).

Full size image

Next, we envisaged that BtuF488 could be used as sensor to monitor transport of cobalamin across liposomal membranes (Supplementary Fig. 4). To develop a transport assay, we purified and reconstituted the energy-coupling factor (ECF) transporter ECF-CbrT from Lactobacillus delbrueckii into liposomes (Supplementary Fig. 1b). ECF-CbrT is a Type III ABC transporter that mediates ATP-dependent import of cobalamin in bacteria28,29. We define the side of the membrane where the NBDs are located as cis-side, and the opposite side the trans-side. ECF-CbrT functions as an importer and therefore transports cobalamin from the trans to the cis side. The reconstitution procedure of ECF transporters has been shown to result in both inside-out and right-side-out oriented protein complexes in the liposomal membrane30, where the transporters expose their cytosolic NBDs to the external buffer or to the aqueous solution within the liposomal lumen, respectively. To monitor the transport activity of only one of these two populations of ECF-CbrT, we added Mg-ATP and cobalamin to the buffers on opposite sides of the membrane, and the sensor to the same compartment as Mg-ATP (Fig. 1b). To probe the activity of the population exposing their NBDs (cis-side) to the external buffer, we encapsulated cobalamin in the lumen of the proteoliposomes (trans-side), and added the sensor to the external buffer. The transport reaction was subsequently started by adding Mg-ATP to the external buffer and we followed the fluorescence over time in a 96-well plate. While trans-to-cis transport physiologically leads to import of cobalamin, in our liposome set-up it leads to release of the substrate from the lumen (physiological import direction). A time-dependent quenching of the sensor fluorescence was observed in the presence of Mg-ATP, indicating that cobalamin is actively transported out of the proteoliposomes (yellow circles, Fig. 1b). The fluorescence remained unaffected when Mg-ATP was replaced by Mg-ADP (white triangles, Fig. 1b), or when liposomes were reconstituted with the folate-specific ECF transporter ECF-FolT2 (yellow and grey triangles, Fig. 1b, Supplementary Fig. 1c), confirming that the observed quenching is derived from ATP-dependent transport of cobalamin across the membrane by ECF-CbrT. Next, we co-encapsulated Mg-ATP together with cobalamin in the lumen of the proteoliposomes, while keeping the sensor on the outside. In this configuration, potential transport from the cis to the trans side can be detected (physiological export direction) by the population of ECF-CbrT exposing their NBDs within the lumen of the liposomes. As expected for an importer, we did not observe quenching of the sensor (blue circles, Fig. 1b), confirming that ECF-CbrT lacks the ability to transport cobalamin in the export direction. To confirm that ECF-CbrT only transports cobalamin in the import direction, we reversed the transport setup and examined the transport activity by monitoring the accumulation of cobalamin into the lumen of proteoliposomes (Supplementary Fig. 4). To do so, we encapsulated the sensor within the lumen of the proteoliposomes, added cobalamin to the external buffer, and had Mg-ATP present either in the lumen, or in the external solution. Although this set-up leads to slightly noisier traces (since the amount of sensor molecules that can be incorporated in the lumen is limited), the results are clear (Fig. 1c). Quenching of the sensor was observed only when Mg-ATP was co-encapsulated in the lumen of the liposomes (yellow circles, Fig. 1c). In contrast, when Mg-ATP was added to the external buffer (blue circles, Fig. 1c), no quenching was observed confirming that ECF-CbrT transports cobalamin only in the import direction.

Bidirectional transport of cobalamin by BacAMm

We then turned to the mycobacterial type IV ABC transporter BacA. Previously, it was shown that the cobalamin auxotrophic strain E. coli ΔFEC requires the expression of a functional cobalamin transporter for growth28,31. By monitoring its growth, it was concluded that BacAMt mediates the import of cobalamin9. We used the same assay to show that expression of the closely-related homologue BacA from M. marinum (BacAMm, 88.2% sequence identity with BacAMt) also rescues growth of the E. coli ΔFEC strain (Supplementary Fig. 2a), showing that BacAMm is also a functional cobalamin importer. For experiments with purified transporters, reconstituted in liposomes, we chose to continue with the protein from M. marinum, because the wild-type protein is less prone to aggregation upon purification in detergent solution than its M. tuberculosis counterpart. To conclusively demonstrate that the BacAMm is indeed responsible for the import of cobalamin, we purified and reconstituted BacAMm into liposomes (Supplementary Fig. 1d). We then assessed its ability to release cobalamin from the liposomes from the trans-to-the cis side of the protein (physiological import direction) using the fluorescence-based transport assay presented above. For experiments with BacAMm we initially encapsulated a 5-fold higher concentration (65 µM) of cobalamin inside the lumen of the proteoliposomes than used with ECF-CbrT, as the BacA proteins seem to lack a high-affinity binding site for the substrate. In later experiments (see below) we used concentrations ranging from 0.5 µM to 100 µM cobalamin. Similar to ECF-CbrT, we also observed an ATP-dependent quenching of the fluorescence signal over time in the presence of externally added nucleotides for BacAMm (yellow circles, Fig. 2a). We did not observe quenching of the sensor in the presence of external Mg-ADP (white triangles, Fig. 2a) or the non-hydrolysable ATP analogue AMP-PNP (grey circles, Fig. 2a), demonstrating that the transport of cobalamin is strictly coupled to the hydrolysis of ATP. This experiment shows that a purified member of the BacA family alone is sufficient for transport of cobalamin in the physiological import direction.

Fig. 2: Bidirectional transport of cobalamin by BacAMm.
figure 2

a Release of cobalamin from proteoliposomes reconstituted with BacAMm. Proteoliposomes were loaded with 65 µM cobalamin and the sensor BtuF488 was added to the external buffer with a concentration of 25 nM. Transport was monitored in the presence of 10 mM external Mg-ATP (yellow circles), 10 mM internal Mg-ATP (blue circles), 10 mM Mg-ATP on both sides (purple circles), 10 mM Mg-AMP-PNP (grey circles) or 10 mM external Mg-ADP (white triangles). Cartoons on the right side of the graph explain the experimental setup. The liposome membrane is shown as grey circle, the sensor in the external buffer in yellow, cobalamin as red dots, the active BacAMm population in green-orange, and the inactive population in grey. Export and import directions are indicated by blue and yellow arrows, respectively. b Accumulation of cobalamin into liposomes reconstituted with BacAMm. Proteoliposomes were loaded with 2 µM BtuF488 inside the lumen and 65 µM cobalamin was added externally. Transport activity was measured in the presence of 10 mM internal Mg-ATP (yellow circles), 10 mM external Mg-ATP (blue circles), 10 mM Mg-ATP on both sides (purple circles) or 10 mM internal Mg-ADP (white triangles). Cartoons on the right side of the graph explain the experimental setup, as in (a). a, b The data were obtained from biological duplicates each containing technical triplicates. The data is presented as the mean with error bars indicating the standard deviation calculated from all individual data points (n = 6).

Full size image

It has been speculated that in addition to its import function, BacAMt may also mediate the export of some antibiotics and antimicrobial peptides from the cell14,17,18,19,20,21,22,23. This suggests that the protein may use different transport directions for different substrates. We then looked into the possibility that BacAMm might be able to transport a single substrate (cobalamin) not only in the import direction (trans-to-cis side), but also in the export direction (cis-to-trans side). To test this, we monitored the release of cobalamin into the external buffer in the presence of lumenal cobalamin and Mg-ATP. Strikingly, we observed a time-dependent quenching of the fluorescence signal in the presence of lumenal Mg-ATP (but not in the presence of Mg-ADP), indicating that BacAMm is also able to transport cobalamin in the export direction (blue circles, Fig. 2a), albeit with a 2-3-fold lower transport rate than observed for the import direction. This is in stark contrast to the strict importer ECF-CbrT, for which export activity could not be demonstrated even at higher concentrations of lumenal cobalamin (blue circles, Fig. 1b and Supplementary Fig. 3). Extensive washing during sample preparation makes it unlikely that sufficient external ATP was available for the inside-out oriented BacA population to transport cobalamin in the trans-to-cis direction. The lack of transport observed for ECF-CbrT using the same washing procedure (Fig. 1b and Supplementary Fig. 3) confirms that external ATP is sufficiently removed during sample preparation. This confirms that the release of cobalamin from the proteoliposomes resulted from actual transport activity of BacAMm. Finally, we prepared proteoliposomes in the presence of Mg-ATP on both sides of the membrane, loaded them with cobalamin in the lumen and monitored the flux of cobalamin to the external solution where the sensor was present. For BacAMm, a faster quenching of the sensor was observed compared to the quenching derived from the import or export activity alone (purple circles, Fig. 2a). This shows that both oriented protein complexes within the liposome simultaneously contribute to the release of cobalamin, which is consistent with the notion that BacAMm can transport in the import and export directions. We compared this result to that of the same experiment using ECF-CbrT. Since ECF-CbrT can only transport in the import direction, we expected that the presence of ATP on both sides of the membrane would not increase the flux of cobalamin out of the liposomes, in contrast to what we observed for BacAMm. Indeed, there was no additive flux of cobalamin from the liposome lumen in this case, but surprisingly we found that additional lumenal Mg-ATP had the opposite effect for ECF-CbrT, as we observed a decrease in the degree of quenching of the fluorescence signal (purple circles, Supplementary Fig. 3). This effect is attributed to the re-transport of cobalamin into the proteoliposomes by import activity of the other orientation, which is possible due to the low-nanomolar binding affinity of CbrT28. In this case, ATP-dependent cycling of cobalamin between the compartments may lead to a steady state flux in which the level of sensor quenching is reduced, compared to unidirectional efflux only. Altogether, we show that a wild-type ABC transport system drives bidirectional transport of a single substrate.

The apparent transport rates suggest that BacAMm transports cobalamin 2-3-fold more efficiently in the import direction than the export direction, but this conclusion only holds if the populations of BacAMm reconstituted in the right-side-out and inside-out orientations are equal. Any deviation from the 1:1 orientation will affect the observed rates of transport. In addition, the membrane curvature could also affect the proteins oriented for transport in the import and export directions differently. To test the effect of the protein orientation on the observed activity, we reversed the transport setup (similar to what we had done for ECF-CbrT, Fig. 1c) and examined the transport activity of BacAMm by monitoring the accumulation of cobalamin into the lumen of proteoliposomes. We encapsulated the sensor within the lumen of the proteoliposomes, added cobalamin to the external buffer, and had Mg-ATP present either in the lumen, or in the external solution, or on both sides. Quenching of the sensor was observed both when Mg-ATP was co-encapsulated in the lumen of the liposomes (yellow circles, Fig. 2b), and when Mg-ATP was added to the external buffer (blue circles, Fig. 2b) confirming that BacAMm transports cobalamin both in the import and export direction. Again, a 2-3-fold higher transport rate was found in the import direction compared to the export direction, indicating that transport in the trans-to-cis (import) direction is somewhat more efficient than in the trans-to-cis (export) direction, and that the transporter had been reconstituted in equal proportions in the right-side-out and inside-out orientation. When Mg-ATP was added to both the internal and external solutions, there was again an additive effect on the transport rate (purple circles, Fig. 2b).

Transport kinetics of cobalamin in the export and import direction

To gain more insight into the unusual bidirectionality of BacAMm, we assessed the transport kinetics of cobalamin in both transport directions. We monitored the import of cobalamin into proteoliposomes and triggered the transport activity in the presence of increasing concentrations of cobalamin in the external buffer. A fixed concentration of Mg-ATP was present either in the lumen (trans-to-cis or import reaction) or in the external buffer (cis-to-trans or export reaction). The initial transport rates depended on the substrate concentrations for both transport directions (Fig. 3a). Remarkably, the apparent KM for cobalamin for transport in the import direction of 3.9 µM was lower than the apparent KM for transport in the export direction (40 µM, but it is noteworthy that this value may be less accurate than the one for the import direction, because we could not saturate the transport rates). The absolute Vmax values are difficult to determine from these experiments, because only a fraction of the sensors encapsulated within the liposomes is accessible to transported cobalamin due to the expected heterogeneity in the proteoliposome population (including multilamellar liposomes, and liposomes containing sensors without reconstituted active transporters). Nonetheless, it is possible to compare the relative differences in Vmax between import and export activity. Although there is some variation between experiments, this comparison shows that the Vmax values in the two transport directions are very similar and differ by less than two-fold. Altogether, this data demonstrates that BacAMm operates somewhat more efficiently as a cobalamin importer at physiological concentrations, which is consistent with the observation that the protein mediates the transport of cobalamin in the import direction in vivo (Supplementary Fig. 2). Finally, we tested the transport dependency on the Mg-ATP concentration for both transport directions, and found that the apparent KM for ATP was nearly identical for both the import and export activity (Fig. 3b).

Fig. 3: Transport kinetics of BacAMm for transport of cobalamin.
figure 3

a Initial cobalamin transport rates determined for BacAMm reconstituted into liposomes in the presence of various concentrations of cobalamin. The transport rates were tested in the presence of 10 mM Mg-ATP for the import direction (yellow) and the export direction (blue). Cartoons on the right side of the graph explain the experimental setup. The liposome membrane is shown as grey circle, the sensor in the lumen in yellow, cobalamin as red dots, the active BacAMm population in green-orange, and the inactive population in grey. Export and import directions are indicated by blue and yellow arrows, respectively. The KM values for cobalamin are 3.9 µM and 40 µM for transport in the import and export direction, respectively. b Initial transport activity of BacAMm reconstituted into liposomes in the presence of various concentrations of Mg-ATP. The transport rates were tested in the presence of 65 µM cobalamin for the import direction (yellow) and in the presence of 100 µM cobalamin for the export direction (blue). The KM values for Mg-ATP are 0.25 mM and 0.15 mM for transport in the import and export direction, respectively. Cartoons on the right side of the graph explain the experimental setup, as in (a). a, b The data were obtained from biological duplicates each containing technical triplicates. Linear regression was used to determine the initial transport rates of the averaged independent experiments. The activity was then normalised to the highest activity within the data set. The mean of the two averaged biological duplicates was plotted with error bars indicating the standard deviation calculated between the two biological repeats (n = 2). The apparent KM and Vmax values were derived using the Michaelis-Menten equation.

Full size image

Discussion

Conventionally, assays to detect transport across liposomal membranes make use of radioactive substrates. While these assays have proven to be sensitive, robust and capable of accurate quantification, the throughput is often very low. In addition, radioactive substrates are increasingly expensive, generate radioactive waste, and their use is subject to stringent regulation. In the particular case of the substrate cobalamin, the use of 57Co-cyanocobalamin requires elaborate safety precautions28. Hence, development of alternative transport assays that support high-throughput experiments are desirable.

Methods established in the past to detect cobalamin in solution or in cellular or liposomal compartments without the use of radioactive substrates32,33,34,35,36,37,38,39,40,41 often do not allow for high-throughput measurements, or are incompatible with assays to measure transport in both import and export direction. We chose to use a protein-based sensor derived from the SBP BtuF25. This sensor exploits the fluorescence quenching of a conjugated fluorophore by cobalamin upon binding of the substrate to the protein25,42 and has been used to monitor the presence of cobalamin in liposomes at the single-molecule level25. Similar sensors have been used to measure binding of cobalamin to BtuF in bulk solution26,43 and at single molecule level44,45. The sensor is membrane impermeable, binds cobalamin with low nanomolar Kd, and allows for sensitive detection of substrate imported into, or exported from the liposomal lumen (Supplementary Fig. 4). While the high-affinity binding property of BtuF488 on one hand provides sensitivity to the transport assay, it may on the other hand also act as a sink, making it difficult to demonstrate transport of cobalamin against its concentration gradient. We established a fluorescence-based transport assay using the well-characterized cobalamin transporter ECF-CbrT, and then used it to characterize the mycobacterial ABC transporter BacA.

Our study reveals that the mycobacterial type IV ABC transporter BacAMm supports the ATP-driven transport of cobalamin in both the import and export direction, which is exceptional among ABC transporters, as these proteins generally use the free energy from the hydrolysis of ATP to ensure unidirectionality1,2,3,4,5,6,7. Physiologically relevant bidirectional substrate movement in the ABC transporter family has only been observed in CFTR8, which is mechanistically distinct from all ABC transporters and acts as an ATP-gated channel rather than a transporter. It is noteworthy that a few other type IV ABC transporters have been genetically engineered to change their transport mode by introducing different type of mutations. For instance, mutations in the D-loop and the outer gate of the human ABC transporters TAP46 and ABCB147, respectively, converted the exporters into facilitators that translocate their substrates along the concentration gradient. In contrast to BacAMm, facilitated diffusion of peptides by the mutated TAP transporter was no longer coupled to the hydrolysis of ATP. In the case of the multidrug efflux pumps ABCB148 and ErfCD49, mutations in the substrate binding site have been demonstrated to reverse the direction of transport for some of their substrates, while the transport directionality is maintained for others. However, to the best of our knowledge, bidirectional transport has not been presviouly observed for a native ABC transporter. It is possible that bidirectionality may be found in more ABC transporters, but has so far remained undetected because the assay conditions prevented detection.

The unusual bidirectionality is consistent with previous studies in whole cells showing that on the one hand BacAMt facilitates the import of several hydrophilic compounds9,11,13,14,15,16, including cobalamin, whilst on the other hand there are also indications that the protein may be involved in the export of antimicrobial compounds14,17,18,19,20,21,22,23. From an evolutionary perspective, BacA may be a generalist transport system with the ability to transport a wide range of hydrophilic substrates not necessarily related to physiologically relevant processes, which over time may evolve into a specialized transporter50. Although BacAMm transports cobalamin in both directions, the apparent KM for cobalamin is approximately 10-fold lower for transport in the import direction (Fig. 3a), while Vmax values are similar. Therefore, in the presence of physiological concentrations of cobalamin, operation of BacAMm could lead to a low level of accumulation, albeit much less what the hydrolysis of ATP could achieve if all free energy was used for accumulation. The capacity to export the micronutrient may be of physiological relevance, because M. marinum (in contrast to M. tuberculosis) is a bacterium that can synthesise cobalamin51,52. Cobalamin is often shared between cobalamin producers and consumers within microbial communities53,54. While multiple membrane transport systems have been identified that are used to import cobalamin55, it has remained elusive how synthesised cobalamin is released into the environment54. It is likely that cobalamin producing microbes live in symbiosis with cobalamin auxotrophic organisms and therefore share the vitamin via the export through membrane transporters54,56. Other B-type vitamins are hypothesised to be shared through the active export via secondary active bidirectional transporters54. In contrast, BacA is a primary active ATP-driven transporter. We speculate that transporters belonging to the BacA family57 may contribute to the release of cobalamin, and possibly other shared metabolites, into the surroundings.

It remains elusive how, unlike other well-characterised ABC transporters, BacAMm can mechanistically transport substrates in both directions. The structural information of the BacA family is scarce, as only a single structure of BacAMt has been solved, reflecting the nucleotide-bound occluded state with an unidentified substrate trapped in its cavity9. The protein possesses a large cavity of ~7700 Å3 lined with polar and negative residues, which can explain its ability to transport multiple structurally unrelated hydrophilic compounds. However, structures under turnover conditions may be needed to gain insights into the bidirectional transport mechanism of BacAMm.

Methods

Cloning

bacAMm was amplified from the genomic DNA of M. marinum and cloned into the E. coli expression vector pBXC3H from the FX cloning kit58. Site-directed mutagenesis was performed using partially overlapping primers to remove all cysteines from ECF-CbrT, as we noted that the wild-type protein had the tendency to be purified in the absence of the S-component CbrT. All primers are listed in Supplementary Table 2.

Overexpression and membrane vesicle preparation

The membrane proteins ECF-CbrT28, ECF-FolT59 and BacAMm9 were expressed as described previously. Briefly, E. coli MC1061 cells were grown in 5 L baffled flasks filled with 2 L LB miller broth containing 100 µg ml−1 ampicillin at 37 °C while continuously shaking. Expression of the protein was induced at an OD600 of 0.8 with 0.01% (w/v) and 0.02% (w/v) L-arabinose (Sigma Aldrich) for both ECF transporters and BacAMm, respectively. Cells were harvested 3 h after induction by centrifugation (20 min, 6000 x g, 4 °C) and were washed with ice-cold 50 mM potassium phosphate (KPi), pH 7.5. Cells were resuspended in breaking buffer (50 mM KPi, pH 7.5, 10% (v/v) glycerol, 2 mM MgSO4, 0.5 mM PMSF (Roth) and 4 µg mL−1 DNase I (Sigma-Aldrich)) and were lysed by three passages through a high-pressure homogeniser HPL6 system (Maximator) at 25 kpsi. Cell debris was removed by high-speed centrifugation (40 min, 31,000 x g, 4 °C), after which the crude membranes were collected by ultracentrifugation (150 min, 186,000, 4 °C). Membranes were homogenised in storage buffer (50 mM KPi pH 7.5, 10% (v/v) glycerol) and stored in aliquots at −70 °C until further use after flash freezing in liquid nitrogen. The total protein concentration was determined using the bicinchoninic acid assay kit (Thermo Fisher Scientific). For the preparation of membrane vesicles containing BacAMm, the breaking buffer and storage buffer was supplemented with 1 mM DTT, and a complete EDTA-free protease inhibitor tablet (Roche) was added to the breaking buffer.

Expression, purification and labelling of BtuF

Expression of BtuF and cell lysis was accomplished with the same procedure as described above with minor modifications. The temperature during cell growth was reduced to 25 °C when the cells reached an OD600 of 0.5 and protein was expressed at an OD600 of 0.8 with 0.01% (w/v) L-arabinose. Cell lysis was done in the presence of 1 mM DTT, and the supernatant obtained after removal of the cell debris was directly used for subsequent purification steps. The supernatant was incubated for 1 h with 2 mL bed volume of Nickel-Sepharose 6 Fast flow beads (Cytiva) that was pre-equilibrated with 20 column volumes (CV) of 50 mM KPi, pH 7.5, 300 mM NaCl, 10% (v/v) glycerol and 1 mM DTT. The resin was then packed into an Econo-Pac column (BioRad) and the unbound proteins were allowed to flow through. Non-specifically bound proteins were removed by 40 column volume (CV) of wash buffer (50 mM KPi, pH 7.5, 300 mM NaCl, 50 mM imidazole, pH 7.5, 10% (v/v) glycerol, 1 mM DTT). BtuF was eventually eluted from the column with elution buffer (50 mM KPi, pH 7.5, 300 mM NaCl, 350 mM imidazole, pH 7.5, 10% (v/v) glycerol, 1 mM DTT). Fractions containing most protein were further purified by size-exclusion chromatography (SEC) using a Superdex 200 increase 10/300 GL column (Cytiva) equilibrated with SEC buffer (50 mM KPi, pH 7.5, 200 mM NaCl, 10% (v/v) glycerol, 1 mM DTT). 21 nmol BtuF was rebound to 0.3 mL bed volume of Nickel-Sepharose 6 Fast flow beads, that was pre-equilibrated with 20 CV SEC buffer. DTT and glycerol were removed with 9 CV labelling buffer (50 mM KPi, pH 7.5, 200 mM NaCl), after which the protein was immediately incubated at 4 °C for 2 h with a 2-fold molar excess of AF488 C5 maleimide (Jena Science) or XDF-488 C5 maleimide (AAT Bioquest) dissolved in labelling buffer. Excess fluorophore was removed with 20 CV washing buffer followed by the elution of the labelled protein (50 mM KPi, 7.5, 200 mM NaCl, 350 mM imidazole, pH 7.5, 5% (v/v) glycerol). Labelled protein was further purified by SEC using a Superdex 200 increase 10/300 GL column equilibrated with SEC buffer (50 mM KPi, pH 7.5, 150 mM NaCl, 5% (v/v) glycerol). Aliquots were flash frozen in liquid nitrogen and stored at −70 °C until further use.

Binding of cobalamin to BtuF488

25 nM of BtuF488 in 50 mM KPi, pH 7.5 and 150 mM NaCl solution was mixed with various concentrations of Cyano-cobalamin (Acros Organics), ranging from 1 nM to 3 µM, to determine the dissociation constant. The mixture was incubated for 5 min at 30 °C in a black polystyrene 96-well plate (Greiner) coated with bovine serum albumin (BSA, Roth), to reduce interactions of BtuF488 with the 96-well plate. The fluorescence intensity was then measured on a Spark 10 M microplate reader (TECAN) using an excitation wavelength of 485 nm (bandwidth 5) and an emission wavelength of 520 nm (bandwidth 10). The normalised decrease in fluorescence was plotted against the substrate concentration. The measurements were performed as biological duplicates each containing technical triplicates. The KD was derived by fitting the formula \({{{{{\rm{Y}}}}}}=1-\left(\frac{{{{{{\rm{A}}}}}}\left(\left({{{{{\rm{Et}}}}}}+{{{{{\rm{X}}}}}}+{{{{{\rm{Kd}}}}}}\right)-\sqrt{{\left({{{{{\rm{Et}}}}}}+{{{{{\rm{X}}}}}}+{{{{{\rm{Kd}}}}}}\right)}^{2}-4*{{{{{\rm{Et}}}}}}*{{{{{\rm{X}}}}}}}\right)}{2}\right)\) in GraphPad Prism (version 10.0.2).

Purification of ECF-CbrT and ECF-FolT2

Membrane vesicles containing 40–60 mg of total protein were solubilized in solubilisation buffer (50 mM KPi, pH 7.5, 300 mM NaCl, 10% (v/v) glycerol and 1% (w/v) dodecyl-β-maltoside (DDM, Glycon)) for 1 h while gently agitating. Insoluble material was subsequently removed by ultracentrifugation (25 min, 444,000 x g, at 4 °C). The supernatant was incubated for 1 h with 0.5 mL bed volume of Nickel-Sepharose 6 Fast flow beads that was pre-equilibrated with 20 CV of wash buffer (50 mM KPi, pH 7.5, 300 mM NaCl, 0.05% (w/v) DDM, 50 mM imidazole, pH 7.5) while gently agitating. The resin was then packed into a polypropylene column (BioRad) and the unbound proteins were allowed to flow through. Non-specifically bound protein was removed with 30 CV wash buffer. Bound proteins were eluted with elution buffer (50 mM KPi, pH 7.5, 300 mM NaCl, 0.05% (w/v) DDM and 300 mM imidazole, pH 7.5) in fractions of 450 µl−800 µl−800 µl. The elution fraction containing most protein was further purified by SEC using a Superdex 200 increase 10/300 GL column equilibrated with SEC buffer (50 mM KPi, pH 7.5, 200 mM NaCl, 0.05% (w/v) DDM). The peak fractions corresponding to the purified protein were pooled and used for reconstitutions into liposomes.

Purification BacAMm

Purification was performed similar to ECF-CbrT and ECF-FolT2 with minor modifications. 50 mM KPi, pH 7.5 was replaced for Tris, pH 8.0 and the concentration was lowered to a concentration of 25 mM except for the solubilisation. DDM was replaced for DDM/CHS (10:1) to stabilise the protein. The elution buffer was supplemented with 350 mM imidazole, pH 7.5. All buffers were supplemented with 1 mM DTT.

Reconstitution into liposomes

Freshly purified ECF-CbrT, ECF-FolT2 and BacAMm were reconstituted into Triton X-100 destabilised liposomes consisting of E. coli polar lipid extract (Avanti Polar Lipids) and egg phosphatidylcholine (Avanti Polar Lipids) at a 3:1 ratio dissolved in 50 mM KPi, pH 7.5, as described previously60. Briefly, E. coli polar lipid extract, prepared from E. coli total lipid extract (Avanti Polar Lipids) and egg phosphatidylcholine (Avanti Polar Lipids) dissolved in chloroform were mixed at a 3:1 ratio (w/w) and dried in a rotavopor. The obtained lipid film was dissolved to a concentration of 20 mg ml−1 in 50 mM KPi, pH 7.5 buffer, and then extruded eleven times through a 400 nm polycarbonate filter (Avestin). The liposomes were destabilised by adding Triton X-100, until half of the initial absorbance at 540 nm was reached. ECF transporters and BacAMm were added with a protein: lipid ratio of 1:400 (w/w) and 1:250 (w/w), respectively. The mixture was then incubated at 4 °C for 30 min while gently rotating, after which the detergent was removed by the addition of BioBeads SM-2 (Bio-Rad) in four steps. Proteoliposomes were collected by ultracentrifugation (30 min, 444,000 x  g, at 4 °C), resuspended in 50 mM KPi, pH 7.5 to a concentration of 40 mg ml−1 lipids, flash-frozen in aliquots and stored in liquid nitrogen until further use. All buffers were supplemented with 1 mM DTT when liposomes were reconstituted with BacAMm.

Fluorescence-based release assay for cobalamin

Proteoliposomes dissolved in assay buffer (50 mM KPi, pH 7.5 and 150 mM NaCl) were loaded with either 13 µM or 65 µM Cyano-cobalamin and optionally 10 mM Mg-ATP (Roche) through five freeze-thaw cycles, followed by extrusion of eleven passes through a 400 nm polycarbonate filter (Cytiva). Excess of cobalamin was removed by ultracentrifugation (30 min, 444,000 x g, at 4 °C) and the proteoliposomes were once more washed in assay buffer. Transport reactions were performed in black polystyrene 96-well plate (Greiner) using proteoliposomes resuspended in assay buffer with a lipid concentration of 2.5 mg ml−1 in a final volume of 200 µl. The sensor BtuF488 was supplemented to the external buffer with a concentration of 25 nM. Mg-ATP or Mg-ADP (Sigma Aldrich) was added to a final concentration of 10 mM (unless stated otherwise) to trigger transport in the import direction. The transport activity was immediately monitored on a Spark 10 M microplate reader (TECAN) by measuring the fluorescence signal every 20 s at an excitation wavelength of 485 nm (bandwidth 5) and an emission wavelength of 520 nm (bandwidth 10) using the instruments’ SparksControl software version 2.3 (TECAN). All steps were performed in the presence of 1 mM DTT for liposomes reconstituted with BacAMm. The measurements were performed at 30 °C as biological duplicates each containing technical triplicates. Signals were normalised to the first reading after the addition of nucleotides.

Fluorescence-based accumulation assay for cobalamin

The fluorescence-based accumulation assay was performed as described for the fluorescence-based transport assay with minor modifications. Proteoliposomes were loaded with 2 µM BtuF488 and optionally 10 mM Mg-ATP or Mg-ADP through two freeze-thaw cycles, followed by twenty-three passes through a 400 nm polycarbonate filter. Transport reactions were performed using proteoliposomes with a lipid concentration of 1.25 mg ml−1 or 2.5 mg ml−1. Mg-ATP or Mg-ADP was optionally added to the external buffer to a final concentration of 10 mM to trigger transport in the export direction. Cyano-cobalamin was added to a final concentration of 65 µM (unless stated otherwise) to trigger transport.

Cobalamin-dependent growth assays

Growth assays were performed as previously described31. Briefly, the cobalamin-deficient E. coli ΔFEC strain was transformed with various expression vectors to express BacAMt, ECF-CbrT or ECF-FolT2. Single colonies were then innoculated into M9 medium supplemented with 0.00001% (w/v) arabinose (Sigma-Aldrich) and 50 µg ml−1 methionine (Sigma-Aldrich). Cells were grown at 37 °C for 24 h and were then diluted with a ratio of 1:250 into 200 µl fresh M9 medium supplemented with 0.00001% (w/w) arabinose and either 1 nM cyano-cobalamin or 50 µg ml−1 methionine (Sigma-Aldrich) in 96-well plates (Greiner). Plates were sealed with sterile and gas-permeable foil (Breathe-Easy, Diversified Biotech) and incubated at 37 °C for 24 h. The OD600 was measured every 10 min on a SpectraMax ABS plus Microplate reader (Molecular Devices). The measurements were performed as biological triplicates each containing technical triplicates and collected using the instruments’ SoftMax Pro version 7.1.2 (Molecular Devices).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.