A single intact ATPase site of the ABC transporter BtuCD drives 5% transport activity yet supports full in vivo vitamin B₁₂ utilization

Abstract

In all kingdoms of life, ATP binding cassette (ABC) transporters are essential to many cellular functions. In this large superfamily of proteins, two catalytic sites hydrolyze ATP to power uphill substrate translocation. A central question in the field concerns the relationship between the two ATPase catalytic sites: Are the sites independent of one another? Are both needed for function? Do they function cooperatively? These issues have been resolved for type I ABC transporters but never for a type II ABC transporter. The many mechanistic differences between type I and type II ABC transporters raise the question whether in respect to ATP hydrolysis the two subtypes are similar or different. We have addressed this question by studying the Escherichia coli vitamin B₁₂ type II ABC transporter BtuCD. We have constructed and purified a series of BtuCD variants where both, one, or none of the ATPase sites were rendered inactive by mutation. We find that, in a membrane environment, the ATPase sites of BtuCD are highly cooperative with a Hill coefficient of 2. We also find that, when one of the ATPase sites is inactive, ATP hydrolysis and vitamin B₁₂ transport by BtuCD is reduced by 95%. These exact features are also shared by the archetypical type I maltose ABC transporter. Remarkably, mutants that have lost 95% of their ATPase and transport capabilities still retain the ability to fully use vitamin B₁₂ in vivo. The results demonstrate that, despite the many differences between type I and type II ABC transporters, the fundamental mechanism of ATP hydrolysis remains conserved.

ATP binding cassette (ABC) transporters comprise one of the largest membrane protein superfamilies of any proteome (1–3). From bacteria to humans, they participate in processes such as cancer and bacterial multidrug resistance, antigen presentation, signal transduction, DNA repair, translation, cell division, homeostasis maintenance, detoxification, nutrient import, and antiviral defense (4–13). Hydrolyzing ATP to drive transport, ABC transporters shuttle cargo molecules to and fro the various cellular compartments, through the impermeable barriers of cell membranes. Notable mammalian examples include the multidrug exporters (4) and the transporter associated with antigen presentation (14). In prokaryotes, ABC transporters often function as importers and depend on a high-affinity substrate binding protein (SBP) that delivers the substrate to the cognate transporter. Structural and functional information derived from studies of prokaryotic systems largely shaped our mechanistic view of this superfamily of proteins (15–22). An “alternating access” mechanistic model has been formulated over the years, according to which ATP hydrolysis power a sequence of conformational changes that shift the transporter between intracellular and extracellular accessible conformations (23–26). This model has been most extensively demonstrated for the maltose transporter that is probably the best-characterized ABC transporter. Until quite recently, it was largely accepted that ABC transporters all operate by such a generally conserved mechanism.

However, recent reports on the Escherichia coli vitamin B₁₂ transporter BtuC₂D₂-F contrast with the mechanism suggested for the maltose transporter (27–31). Unlike the maltose transporter, BtuC₂D₂ has high levels of basal ATP hydrolysis rates. These rates are only mildly stimulated by the substrate-free and substrate-loaded binding protein alike (32). BtuC₂D₂ (the transporter) and BtuF (the vitamin B₁₂ binding protein) form an extremely high-affinity, stable complex (K_D ∼ 10⁻¹² M), whereas in the maltose system the complex is transient and is of very low affinity (K_D = 50–100 10⁻⁶ M). In addition, maltose has little (or no) effect on the affinity between the transporter and the binding protein. In contrast, vitamin B₁₂ truly has dramatic effects on complex formation: In the presence of the substrate, the k_on is accelerated ∼25-fold and the k_off by six to seven orders of magnitude, resulting in a ∼10⁵-fold decrease in equilibrium affinity (29). These differences indicate that the ABC transporters for maltose and vitamin B₁₂ operate by very different mechanisms. Currently, the maltose transporter and similar systems are termed type I ABC transporters, whereas BtuC₂D₂ is a type II ABC transporter. The question arises whether type I and type II ABC transporters share some mechanistic features, or will we find differences everywhere? To look for similarities, we have focused on ATP hydrolysis: The nucleotide binding domains (NBDs) of ABC transporters are their most conserved part. Moreover, some features of ATP hydrolysis are maintained also across unrelated protein families (26).

A fundamental aspect of ATP hydrolysis by multisubunit enzymes is cooperativity and interdependence. Can one catalytic site function without the other? These questions have been addressed in ABC multidrug exporters and in type I ABC transporters, but never in a type II ABC transporter. We present here such an analysis.

Results

Cooperativity of ATP Hydrolysis by BtuC₂D₂.

We began our investigation of the ATP hydrolytic activity of BtuC₂D₂ by examining the cooperativity of the two ATPase sites. For this, we have used an enzyme-coupled spectroscopic assay in conjunction with an automated 96-well plate reader. Several control experiments were conducted to assess the reliability and robustness of the ATPase assay. As shown in Fig. S1A, the measurements are highly reproducible and three independent repeats superimpose very well. At BtuC₂D₂ concentrations of 0.1–1 μM, the initial rates of ATP hydrolysis are linear for at least 2 min and are a linear function of the concentration of BtuC₂D₂ (Fig. S1 B–D).

The cooperativity of ATP hydrolysis by wild-type BtuC₂D₂ was first studied in detergent solution. Initial rates of ATP hydrolysis (first ∼80 s) were measured and plotted as a function of ATP concentration (Fig. 1A). The experimental data were fit using either the Michaelis–Menten equation or its expanded version that includes also a term for the Hill coefficient (see Materials and Methods for details). Both fittings gave highly comparable results, with very similar rmsd. A Hill coefficient of 1.03 was determined using a nonlinear regression (Fig. 1A). The data were also fit without introducing additional degrees of freedom to the model. A linear regression of log[V/(V_max − V)] as a function of log[ATP] yielded a similar Hill coefficient of 1.04 (determined as the slope of the linear fit; Fig. S2). Three independent experiments yielded kinetic constants V_max = 1.09 ± 0.09 μmol⋅mg⁻¹⋅min⁻¹, K_m = 17 ± 4.6 μM, and a Hill coefficient of 1.03 ± 0.08. These results suggest that in detergent solution the NBDs of BtuC₂D₂ do not hydrolyze ATP cooperatively.

Fig. 1.

Cooperativity of ATP hydrolysis in detergent solution (A) and in proteoliposomes (B). Initial rates of ATP hydrolysis by BtuC₂D₂ were measured over the first 80 s of activity. The experimental data (circles) was fit using the Michaelis–Menten equation (dashed black line) or its expanded version, which includes also a term for the Hill coefficient (solid blue line). Inset in B shows a magnification of the data up to 37.5 μM ATP.

We next tested the cooperativity of ATP hydrolysis by BtuC₂D₂ reconstituted into proteoliposomes. Insertion of BtuC₂D₂ into the liposome’s membrane results in a random orientation of the transporter where a fraction of the molecules adopts an inside-out topology, whereas another fraction inserts in a right-side-out topology. In our assay, Mg-ATP is added to the reaction mixture, thus measuring only the activity of transporter molecules with outward-facing NBDs (i.e., that are “inside-out”).

No ATP hydrolysis was measured in empty liposomes or in liposomes reconstituted with the Walker B mutants E159A or E159Q. In contrast, a robust signal was measured with liposomes reconstituted with wild-type BtuC₂D₂ (Fig. S3A).

Over a range of ATP concentrations, in the first 80 s of hydrolysis the rates were linear and the R² values of the fits (goodness of fit) were between 0.97 and 0.99 (Fig. S3B). Plotting the initial rates as a function of ATP concentration yielded a clearly biphasic curve (Fig. 1B), indicative of positive cooperativity. Unlike what was observed in detergent solution, fitting the data with the Hill equation yielded a much superior fit (fivefold smaller rmsd than the Michaelis–Menten fit) and a random distribution of errors (Fig. S3C). Three independent experiments yielded kinetic constants V_max = 0.36 ± 0.09 μmol⋅mg⁻¹⋅min⁻¹, K_m = 19.9 ± 7.2 μM, and Hill coefficient of 2.09 ± 0.08. Thus, in contrast to what was observed in detergent solution, in proteoliposomes the NBDs of BtuC₂D₂ are highly cooperative.

Construction and Activity of BtuC₂–Tandem BtuD.

We then aimed to test whether both ATPase sites are required for full activity, or whether one active site will support full or partial activity. However, such experiments are not straightforward, as BtuC₂D₂ is a homodimer. Hence, any single mutation introduced at the btuD gene will result in two mutated sites in the fully assembled transporter. To circumvent this difficulty, we have constructed a btuCD expression vector (hereafter pBtuC₂–tandem BtuD) where a copy of the btuC gene precedes two fused copies of the btuD gene. For generation of a flexible linker, we introduced four tandem repeats of [glycine₍₄₎serine] (33). Expression of this construct results in an assembled transporter with two copies of BtuC and two NBDs that are fused head to tail, enabling discrimination between the “first” and “second” ATPase sites. Several experiments were conducted to test the integrity of BtuC₂–tandem BtuD. We first examined its in vivo activity using growth complementation assays. For E. coli cells to grow, the amino acid methionine must be either supplied exogenously or synthesized endogenously. Two E. coli methionine synthases are able to catalyze the last step in methionine biosynthesis: MetE is vitamin B₁₂ independent, and MetH fully depends on vitamin B₁₂ as a cofactor (34). Thus, in a strain deleted of metE, under conditions of methionine depletion, high-affinity acquisition of extracellular vitamin B₁₂ is essential for growth, and BtuC₂D₂-F becomes indispensable (35). We deleted btuD from a metE⁻ strain creating a metE⁻/btuD⁻ strain. This strain cannot grow even when supplied with vitamin B₁₂, unless transformed with a BtuD expression plasmid. Fig. 2A shows the time-dependent growth in the presence of 1 nM vitamin B₁₂, and Fig. 2B shows the optical density at 600 nm of cultures grown for 12 h in a range of vitamin B₁₂ concentrations.

Fig. 2.

In vivo activity of BtuC₂–tandem BtuD. (A) Cultures were grown in an automated plate reader in the presence of 1 nM vitamin B₁₂: metE⁻/empty plasmid (squares), metE⁻/btuD⁻/empty plasmid (vertical bars), metE⁻/btuD⁻/pBtuC₂–tandem BtuD (triangles), and metE⁻/btuD⁻/pBtuC₂D₂ (circles). (B) Optical density after 12 h of growth in the absence or presence of the indicated vitamin B₁₂ concentrations: metE⁻/empty plasmid (full squares), metE⁻/btuD⁻/empty plasmid (open circles), metE⁻/btuD⁻/pBtuC₂–tandem BtuD (full triangles), metE⁻/btuD⁻/pBtuC₂D₂ (full circles), metE⁻/btuD⁻/pBtuC₂D₂ E159A (open squares), and metE⁻/btuD⁻/pBtuC₂D₂ E159Q (open triangles).

As shown, the metE⁻ strain, which still harbors its endogenous copy of btuD, grows normally when supplied with 1 nM vitamin B₁₂. In contrast, 1 nM vitamin B₁₂ is insufficient to support growth of the metE⁻/btuD⁻ strain. Transforming the metE⁻/btuD⁻ strain with an empty plasmid or plasmids encoding the E159A/Q mutants does not restore growth. However, full growth of this strain is restored when it is transformed with a plasmid encoding either BtuC₂D₂ or BtuC₂–tandem BtuD. The growth complementation conferred by these two plasmids is almost identical.

We next compared the transport activities of “regular” and BtuC₂–tandem BtuD in spheroplasts using radiolabeled vitamin B₁₂. As previously performed with other ABC importers (36, 37), BtuF and the radiotracer are added to the spheroplast suspension and samples are removed at intervals. Bulk solution is removed by rapid filtration followed by several washes. As shown (Fig. 3A), no uptake of vitamin B₁₂ was detected in spheroplasts prepared from cells expressing the Haemophilus influenzae molybdtae/tungstate transporter MolBC, or those expressing the E159A mutant. In contrast, spheroplasts derived from cells expressing BtuC₂–tandem BtuD rapidly accumulate vitamin B₁₂, generating a ∼1,000-fold substrate concentration gradient (see Tables S1–S3 for details) within 40 min. Judging from the first 2 min of transport, the transport rate by BtuC₂–tandem BtuD is ∼53% of the transport rate of “regular” BtuC₂D₂ (Fig. 3A and Tables S1–S3).

Fig. 3.

Transport and ATP hydrolysis by BtuC₂–tandem BtuD. (A) Uptake of 0.5 μM ⁵⁷Co-vitamin B₁₂ into spheroplasts was measured by the rapid filtration method in the presence of 0.1 μM BtuF. Spheroplasts were prepared from cells expressing the H. influenzae molybdate transporter MolBC (circles), BtuC₂D₂ E159A (triangles), wild-type BtuC₂D₂ (full squares), or BtuC₂–tandem BtuD (open squares). (B) Time-dependent ATP hydrolysis by empty liposomes (triangles), BtuC₂–tandem BtuD liposomes (circles), or BtuC₂D₂ liposomes (squares). At time 0, 1 mM ATP and 5 mM MgCl₂ were added to initiate hydrolysis.

The ATP hydrolytic activity of BtuC₂–tandem BtuD was measured in proteoliposomes (Fig. 3B) and in detergent solution (Fig. S4). In reconstituted liposomes, the maximal rate of ATP hydrolysis by BtuC₂–tandem BtuD was 0.17 ± 0.04 μmol⋅mg⁻¹⋅min⁻¹, whereas in detergent this rate was 0.48 ± 0.07 μmol⋅mg⁻¹⋅min⁻¹ (47% and 44%, respectively, relative to BtuC₂D₂). In the presence of BtuF and vitamin B₁₂, these rates were stimulated by ∼40% in detergent solution and ∼2.2-fold in the reconstituted system (Fig. S4). This modest stimulation is very similar to what was observed in the past with the nontandem BtuC₂D₂ (32). From the initial rates of ATP hydrolysis in liposomes and initial transport rates in spheroplasts, we estimated the ATP/substrate ratio of the transport reaction (Tables S1–S3).

Using these data, a ratio of ∼46 molecules of ATP per molecule of vitamin B₁₂ was determined for BtuC₂D₂, and a ratio of ∼42 for BtuC₂–tandem BtuD. These values are 40–50% lower (i.e., better coupling efficiency) than those determined using transport data from a reconstituted system (32). These differences may indicate that in vivo the coupling is more efficient, or are merely a result of the technical differences between the two experimental systems.

Taken together, the above results demonstrate that BtuC₂–tandem BtuD is functional in vivo and in vitro and is reasonably similar to BtuC₂D₂.

Activity of Single-Site NBD Mutants.

Having concluded that BtuC₂–tandem BtuD is a reasonable mimic of BtuC₂D₂, we mutated the Walker B E159 at one of the NBDs or in both. Hereafter, these will be referred to as single-site mutants or double-site mutants, respectively. In the single-site mutants, only one of the ATPase sites is rendered nonfunctional by a mutation, whereas the other is kept in its original wild-type configuration.

The single-site mutants were purified and their ATPase activity was measured in proteoliposomes (Fig. 4A) and in detergent solution (Fig. S5). In both environments, the single-site mutants appear to have largely lost their ATP hydrolytic activity. However, their comparison with the double-site mutants reveals that the single-site mutants still marginally hydrolyze ATP, at a rate that is ∼5% that of wild-type BtuC₂–tandem BtuD (Fig. 4B and Fig. S5).

Fig. 4.

ATP hydrolysis and binding by the single-site mutants. (A) Wild-type BtuC₂–tandem BtuD (thick solid line), single-site mutants E159A and E159Q (thin solid lines), and the double-site mutant E159A (thin dashed line) were purified and reconstituted into liposomes. One millimolar ATP was added at time 0, and 5 mM MgCl₂ was injected at 15 min to initiate hydrolysis. (B) Same as in A, but shown are duplicates of the single-site mutant E159A (black), single-site mutant E159Q (red), double-site mutant E159A (green), and double-site mutant E159Q (blue). (C) Effect of ATP on the association of the single-site mutants with BtuF. Wild-type BtuC₂–tandem BtuD and the single-site mutants (E159A and E159Q) were immobilized onto Ni-NTA beads and incubated with FLAG-BtuF in the absence or presence of ATP-EDTA. Following incubation, unbound FLAG-BtuF was washed away, and the amount of FLAG-BtuF bound to the immobilized transporters was visualized by SDS/PAGE and immunodetection with an α-FLAG antibody.

We previously observed that ATP binding (rather than hydrolysis) by BtuC₂D₂ reduces its affinity toward BtuF (29). To test whether this effect is maintained in the single-site mutants, we performed pull-down experiments in the absence or presence of nucleotide. In these experiments, the His-tagged BtuC₂–tandem BtuD variants are immobilized onto Ni-NTA beads and incubated with FLAG-tagged BtuF in the absence or presence of ATP-EDTA. Following incubation, unbound FLAG-BtuF is washed away, and the amount of FLAG-BtuF bound to the immobilized transporters is visualized by immunoblotting of SDS/PAGE with an α-FLAG antibody. In the absence of ATP, complex formation between wild-type BtuC₂–tandem BtuD and BtuF was indistinguishable from the levels of complex formation between BtuF and the single-site mutants (Fig. 4C). As expected, addition of ATP inhibited complex formation between wild-type BtuC₂–tandem BtuD and BtuF. A very similar ATP effect was observed with the single-site mutants, resulting in decreased levels of transporter–SBP complex formation.

This result suggests that, despite their reduced ATPase activity, ATP binding is unimpaired in the single-site mutants.

Next, we compared the transport activity (in spheroplasts) of the single-site E159A mutant to that of wild-type BtuC₂–tandem BtuD and the double-site mutants. As shown (Fig. 5), relative to wild-type BtuC₂–tandem BtuD, the single-site mutant has a greatly reduced transport activity. However, this low-level activity was clearly distinguishable from the transport rate observed in control spheroplasts or in ones expressing the double-site mutant. Using the same considerations described above for BtuC₂D₂ and BtuC₂–tandem BtuD, we estimate an ATP/vitamin B₁₂ ratio of ∼37 for the single-site mutant (Tables S1–S3).

Fig. 5.

Vitamin B₁₂ transport by the double-site and single-site mutants. Spheroplasts were prepared from control cells (diamonds), cells expressing the double-site E159A mutant (open circles), single-site E159A mutant (open squares), and wild-type BtuC₂–tandem BtuD (full squares). Uptake of 0.5 μM ⁵⁷Co-vitamin B₁₂ in the presence of 0.1 μM BtuF was measured by the rapid filtration method. The left-hand y axis refers to all samples, and the right-hand one to all samples except the spheroplasts prepared from control cells.

Finally, we tested the ability of the single-site mutants to support growth under conditions that require uptake of vitamin B₁₂. Surprisingly, despite their marginal ATPase and transport activity, cells expressing the single-site E159A/Q mutants grew as well as wild-type BtuC–tandem BtuD (Fig. 6). We hypothesized that perhaps the growth of the single-site mutants is a result of a contaminant expression of wild-type BtuC₂–tandem BtuD, some other contaminant, or an unlikely reversion of the mutation to its wild-type configuration. However, careful repetitions of the experiments under stringent sterility always produced the same results. In addition, at the end of the growth experiment we prepared plasmid DNA from the cells that were grown in the assay and confirmed by sequencing that no reversion has occurred. Importantly, the growth complementation conveyed by the single-site mutants cannot be attributed to overexpression of these variants. The genetic setup of the growth assays (pET vectors in non-DE3 strains) ensures extremely low level of expression, undetectable by Western blot analysis (Fig. S6).

Fig. 6.

In vivo activity of the single-site mutants. metE⁻/btuD⁻ cells expressing the following variants were grown in the absence or presence of the indicated vitamin B₁₂ concentrations: wild-type BtuC₂–tandem BtuD (full triangles), single-site E159A mutant (full squares), single-site E159Q mutant (open squares), double-site E159A mutant (open triangles), and double-site E159Q mutant (full circles).

We thus conclude that, in the single-site mutants, 5% ATP hydrolysis leads to 5% transport activity, which is sufficient for full in vivo growth.

Discussion

Almost in every tested aspect, the maltose transporter and BtuC₂D₂ have been found to fundamentally differ. Differences were observed in coupling efficiency, substrate affinity, stability of the transporter–SBP complex, and effects of substrate and nucleotides on complex stability. The two systems also respond distinctively to nucleotide binding: the maltose transporter is outward facing in its ATP-bound state and inward facing when it is nucleotide-free. BtuC₂D₂ is the exact opposite (28, 29). These and other differences led to the formulation of two different mechanistic models for type I and type II ABC transporters, termed “alternating access” and “peristaltic,” respectively (31).

It was previously demonstrated that the maltose transporter (MalFGK) hydrolyzes ATP cooperatively in proteoliposomes and that the cooperativity is reduced in detergent solution. In addition, if one of the NBDs of MalFGK is inactive, ATPase and transport rates drop by ∼95% (38). We have tested these features in BtuC₂D₂ and observed the exact same behavior: BtuC₂D₂ is also fully cooperative in liposomes but not in detergent solution, and when one of the NBDs is compromised ATPase and transport rates drop by ∼95%. This suggests that, despite the mechanistic divergence of ABC transporters, the fundamental mechanism of ATP hydrolysis remained conserved. Interestingly, despite this conservation, ATP binding and hydrolysis induce the opposite conformational changes in the maltose and vitamin B₁₂ transporters.

Like BtuC₂D₂ and MalFGK, the histidine transporter (HisPQM) and MJ0796 also hydrolyze ATP cooperatively (39–42). In all these systems, a single gene encodes the ATPases, resulting in the formation of identical NBDs in the assembled transporter. Perhaps cooperativity is inherent to this symmetry between the NBDs. In contrast, when the NBDs are products of two different genes, or are translated as a single polypeptide, they are not cooperative and often have distinct functions (43–47).

However, symmetry alone cannot account for cooperative catalysis, as there are many examples of symmetric enzymes that are noncooperative. In a sense, MalFGK is also nonsymmetric within the context of the fully assembled transporter. One explanation is that in ABC importers like MalFGK and BtuCD not only the ATPase sites are symmetric but they are also formed by a head-to-tail organization of the symmetric units, creating a physical conformational link between the two catalytic sites. However, perhaps there is also a mechanistic reason for the persistence of cooperativity, which may be connected to the mechanism of transport. It has been proposed that the ATPase sites of ABC transporters hydrolyze ATP consecutively rather than simultaneously. Cooperativity fits well within this framework and argues against simultaneous hydrolysis. If this is indeed the case, it would imply that the transport stoichiometry is at least two molecules of ATP per transport cycle. Clearly, this remains an open question.

One perplexing result presented here concerns the in vivo activity of the single-site mutants. Despite having lost ∼95% of their ATPase and transport activities, these mutants still supported full growth under conditions of limiting vitamin B₁₂ concentrations. The explanation may lie with the physiological context of the substrate: Like other vitamins and trace elements (substrates of type II systems), vitamin B₁₂ is needed in very small amounts. It is estimated that as little as ∼10–20 molecules per cell are needed for normal growth (48). The transport activity of the single-site mutant generates a ∼60-fold substrate concentration gradient (Fig. 5 and Tables S1–S3). Thus, even at 1 nM of externally added vitamin B₁₂, cells expressing the single-site mutants will be able to acquire ∼100 molecules of vitamin B₁₂, and 100% growth will be maintained. This highlights another important difference between type I and type II systems: Substrates of type I systems serve as energy sources or metabolic building blocks (sugars, amino acids, or polypeptides) and their beneficial effect is proportional to their intracellular concentration. Thus, for a type I system it is unlikely that 5% transport activity will support 100% in vivo utilization.

Materials and Methods

Bacterial Strains and Plasmids.

DH5α (Invitrogen) was used for cloning procedures, and BL21-Gold(DE3) (Stratagene) was the host for protein expression. For vitamin B₁₂ utilization assays, we used the wild-type strain BW25113 (49) and its isogenic deletion mutants, ∆metE::kan from the Keio collection (50) and ∆metE ∆btuD::kan. The latter was constructed as follows: pCP20 (51) was used as described (49) to eliminate the kanamycin-resistance gene from ∆metE::kan, and the resulting strain was introduced with the ∆btuD::kan mutation (50) by P1 transduction. The correct structure of all deletion mutants was confirmed by PCR.

pBtuC₂D₂ is a pET-21b(+) (Novagen) derivative that encodes BtuC (with an N-terminal 10×His tag) and BtuD under the control of a single T7 RNA polymerase promoter (each coding sequence has its own ribosome-binding site). pBtuC₂–tandem BtuD is a pET-19b (Novagen)-derived plasmid and its structure is essentially as that of pBtuC₂D₂ except that it encodes a genetically fused BtuD dimer.

Protein Purifications.

BtuC₂D₂, BtuC₂–tandem BtuD, double-site, and single-site mutants were purified as previously described (15). N-terminal FLAG-tagged BtuF was purified from osmotic shock extracts by size exclusion chromatography. Protein samples were snap frozen as small aliquots in liquid nitrogen and stored in –80 °C for up to 3 mo.

Reconstitution of BtuCD.

BtuC₂D₂, BtuC₂–tandem BtuD, double-site, and single-site mutants were reconstituted essentially as previously described (29, 32) only that 0.1% (wt/vol) n-dodecyl-N,N-dimethylamine-N-oxide (LDAO) was used throughout the purification and reconstitution process. The lipid:protein ratio used was 30:1 (wt/wt). Proteoliposomes were resuspended to a final concentration of 10 mg⋅ml⁻¹ lipids, snap frozen as small aliquots in liquid nitrogen, and stored in –80 °C for up to 3 mo.

ATP Hydrolysis Assays and Data Fitting.

ATP hydrolysis was measured using Molecular Probes EnzCheck kit, at 37 °C, in a 96-well format, according to the manufacturer’s specifications. To initiate hydrolysis, 5 mM MgCl₂ was injected to a solution containing 0.1–1 μM BtuCD (as indicated) in 25 mM Tris⋅HCl, pH 7.5, 0.15 M NaCl, 0.1% LDAO, 50 μM EDTA, and the indicated ATP concentration. Where applicable, proteoliposomes were subjected to two cycles of freeze–thaw and reextruded through 400-nm polycarbonate filters. The final lipid concentration in the assay was 30 μM, and the approximate BtuCD concentration was ∼0.1–0.2 μM.

Data were fitted using either the Michaelis–Menten equation or its expanded version, which includes also a term for the Hill coefficient:

V is the observed hydrolysis rate, V_max is the maximal hydrolysis rate, K_m is the Michalis–Menten constant, [S] is the concentration of ATP, and n is the Hill coefficient.

Pull-Down Experiments.

Purified His-tagged BtuCD variants in 25 mM Tris⋅HCl, pH 7.5, 0.15 M NaCl, 0.1% LDAO were immobilized onto Ni-NTA beads (Qiagen; 10 μL per sample), followed by a wash step to remove unbound protein. A volume of 250 μL of purified FLAG-tagged BtuF was added at the indicated concentrations. Following a 30-min incubation, unbound material was removed by three cycles of pelleting/washing with 0.4 mL of buffer. Where appropriate, nucleotides and/or substrate were included in the washing buffer. Bound material was then eluted in a single step with buffer (100 μL) containing 1 M imidazole. The amount of retained FLAG-tagged protein in the sample was visualized by standard immunoblotting procedures, using an α-FLAG M2-HRP antibody (Sigma).

Vitamin B₁₂ Utilization Assays.

Cells were grown in LB media supplemented with 50 μg/mL kanamycin and 100 μg/mL ampicillin to mid log phase. Cells were then harvested, washed with water, and resuspended in Davis minimal media (52) depleted of methionine to OD₆₀₀ of 0.05. The 0.2 mL cultures were grown in the absence or presence of the indicated vitamin B₁₂ concentration in an automated plate reader (Infinite M200 Pro; Tecan). The optical density of the cultures was measured every 5 min for 12 h.

Transport of Vitamin B₁₂ in Spheroplasts.

Spheroplasts were prepared essentially as described previously (53). BL21-Gold(DE3) cells (Stratagene), transformed with the indicated plasmids, were grown in LB-ampicillin medium at 37 °C with shaking to midexponential phase. Protein overexpression was induced by isopropyl β-D-1-thiogalactopyranoside (0.5 mM), and cultures were grown for another hour. Cells were harvested and resuspended in 10 mM Tris⋅HCl, pH 7.5, and 0.75 M sucrose. After the addition of lysozyme (100 µg/mL) and 2 vol of 1.5 mM EDTA, the cell suspension was incubated on ice for 20 min. Following the addition of MgCl₂ (25 mM) and DNase (100 µg/mL), spheroplasts were pelleted and resuspended in 100 mM Tris⋅HCl, pH 7.5, 150 mM NaCl, and 5 mM MgCl₂ to an OD₆₀₀ of about 10, and kept on ice until use.

Spheroplasts were supplemented with 0.2% glucose and incubated at 37 °C for 5 min, and the transport reaction was initiated by the addition of 0.5 μM ⁵⁷Co-vitamin B₁₂ (10 µCi/mL; MP Biomedicals), immediately followed by the addition of 0.1 µM BtuF. At the indicated times, 200-µL aliquots were removed, diluted with 2 mL of ice-cold stop buffer (100 mM Tris⋅HCl, pH 7.5, 150 mM NaCl, and 100 µM vitamin B₁₂), and vacuum-filtered through 0.22-µm PVDF membrane filters (Millipore). The filters were washed twice more with ice-cold stop buffer. The amount of ⁵⁷Co-vitamin B₁₂ retained by the spheroplasts was determined by gamma radiation counting.