ABSTRACT
ATP-binding cassette (ABC) transport systems comprise two transmembrane domains/subunits that form a translocation path and two nucleotide-binding domains/subunits that bind and hydrolyze ATP. Prokaryotic canonical ABC import systems require an extracellular substrate-binding protein for function. Knowledge of substrate-binding sites within the transmembrane subunits is scarce. Recent crystal structures of the ABC importer Art(QN)2 for positively charged amino acids of Thermoanerobacter tengcongensis revealed the presence of one substrate molecule in a defined binding pocket in each of the transmembrane subunits, ArtQ (J. Yu, J. Ge, J. Heuveling, E. Schneider, and M. Yang, Proc Natl Acad Sci U S A 112:5243–5248, 2015, https://doi.org/10.1073/pnas.1415037112). This finding raised the question of whether both sites must be loaded with substrate prior to initiation of the transport cycle. To address this matter, we first explored the role of key residues that form the binding pocket in the closely related Art(MP)2 transporter of Geobacillus stearothermophilus, by monitoring consequences of mutations in ArtM on ATPase and transport activity at the level of purified proteins embedded in liposomes. Our results emphasize that two negatively charged residues (E153 and D160) are crucial for wild-type function. Furthermore, the variant Art[M(L67D)P]2 exhibited strongly impaired activities, which is why it was considered for construction of a hybrid complex containing one intact and one impaired substrate-binding site. Activity assays clearly revealed that one intact binding site was sufficient for function. To our knowledge, our study provides the first biochemical evidence on transmembrane substrate-binding sites of an ABC importer.
IMPORTANCE Canonical prokaryotic ATP-binding cassette importers mediate the uptake of a large variety of chemicals, including nutrients, osmoprotectants, growth factors, and trace elements. Some also play a role in bacterial pathogenesis, which is why full understanding of their mode of action is of the utmost importance. One of the unsolved problems refers to the chemical nature and number of substrate binding sites formed by the transmembrane subunits. Here, we report that a hybrid amino acid transporter of G. stearothermophilus, encompassing one intact and one impaired transmembrane binding site, is fully competent in transport, suggesting that the binding of one substrate molecule is sufficient to trigger the translocation process.
INTRODUCTION
ATP-binding cassette (ABC) systems are found in all three domains of life and form one of the largest protein superfamilies, which includes exporters, importers, and nontransporting ABC proteins that do not possess transmembrane domains (1–3). In general, ABC transporters share a minimal modular organization comprising two transmembrane domains (TMDs), which form the translocation path, and two nucleotide-binding domains (NBDs), which bind and hydrolyze ATP. TMDs display basically no sequence homologies and feature various numbers of transmembrane helices (ranging from 4 to 10) in different ABC transporters (3). Substrates transported by ABC transporters are diverse, including sugars, amino acids, peptides, vitamins, ions, xenobiotics, and even polypeptides, and link ABC transporters to various cellular functions that range from energy supply to osmoregulation, detoxification, and virulence (2–7) Moreover, in mammalian cells, ABC transporters play an important role in the immune system, and dysfunction often causes inheritable diseases, such as cystic fibrosis (8).
ABC transporters are thought to function by an “alternate access” mode, according to which the TMDs switch from an inward-facing to an outward-facing state, depending on the nucleotide status of the NBDs (9).
Canonical prokaryotic ABC importers mediate the uptake of nutrients, osmoprotectants, various growth factors, or trace elements (5). Some also play a role in bacterial pathogenesis (5, 7). While substrate specificity of ABC exporters is accomplished by their TMDs, canonical microbial importers are dependent on high-affinity extracellular substrate-binding proteins (SBP) (10, 11). SBPs bind their substrates at a site located between the N-terminal and C-terminal lobes and switch between open (unliganded) and closed (loaded) conformations. In their closed conformation, SBPs productively interact with the cognate transporter, thereby triggering ATPase activity and initiating the transport cycle (5, 6, 10, 12–19).
Canonical ABC importers may be subdivided into types I and II (20). The first type comprises smaller importers that feature a transmembrane core of 10 to 14 helices. Type II importers, on the other hand, display larger transmembrane domains with up to 20 helices and specificity for metal chelates, heme, and vitamin B12. There is growing evidence for distinct mechanistic differences between subtypes (20–22).
In the case of type I ABC importers, several crystal structures of the maltose/maltodextrin transporter (MalE-FGK2) of Escherichia coli provided a framework for the proposed mode of action (reviewed in reference 14). Accordingly, in the absence of nucleotides (apo state), the transporter resides in an inward-facing conformation. Upon concomitant binding of liganded SBP (MalE; KD [equilibrium dissociation constant], ∼3.5 μM) (14) to MalFG and ATP, which causes the NBDs (MalK2) to fully close (13, 14, 16), the TMDs (MalF and MalG) change to an outward-facing conformation and substrate is released to a low-affinity binding site (estimated KD, 1 to 2 mM) (14) within the transmembrane core (MalF) (23). Subsequent hydrolysis of ATP eventually leads to delivery of the ligand to the cytoplasm and return of the transporter to the apo state.
Although it is a key feature in this process, structural information on substrate specificity within the TMDs is thus far available only for the Escherichia coli maltose/maltodextrin transporter (23, 24) and a transporter for positively charged amino acids from the thermophilic bacterium Thermoanaerobacter tengcongensis (synonym Caldanaerobacter subterraneus subsp. tengcongensis), Art(QN)2 (25). In contrast, for type II transporters the existence of a distinct transmembrane substrate binding site(s) is questioned, due to the lack of crystal structures in complex with the substrate, as well as measurable affinities in the transmembrane binding pocket of the E. coli vitamin B12 transporter BtuF-CD (for review, see references 20 and 22).
While in the heterodimeric E. coli MalFGK2 transporter maltose is exclusively in contact with residues of the MalF subunit (23, 24), in the homodimeric Art(QN)2 system, each TMD (ArtQ) was found to bind one arginine molecule (Fig. 1A) (25). This observation implies that two interactions with substrate-loaded binding protein are required to fill both sites before the transport cycle is initiated. If so, this would be inconsistent with the transport mode outlined above. Moreover, the presence of two transmembrane substrate binding sites raises the question of whether both sites actually need to be occupied to trigger ATPase and hence transport activity.
Comparison of arginine-binding sites in the crystal structure of Art(QN)2 and the homology model of ArtM2. (A, left) Overall structure of Art(QN)2 (PDB access code 4YMU). Two transmembrane subunits (ArtQ) are colored in red and purple, and two nucleotide-binding subunits (ArtN) are colored in cyan and light green. Arginine (in ArtQ) and ATP molecules (in ArtN) are shown in ball-and-stick models. (Right) Magnification of one substrate binding site in ArtQ (indicated by a black square). Residues of ArtQ interacting with one arginine molecule and neighboring residues are indicated. Hydrogen bonds are represented as dashed lines. (B, left) Structural alignment of ArtQ2 (colored as in panel A) with the modeled ArtM dimer (colored in green and light yellow). Homology model of ArtM was built by the Swiss-Model homology server, using ArtQ (PDB access code 4YMU) as the template. (Right) Magnification of proposed substrate binding site (indicated by a black square). Residues forming hydrogen bonds with an arginine molecule (represented by dashed lines) and neighboring residues are indicated. The figures were drawn with Biovia Discovery Studio Visualizer v4.5. (C) Sequence alignment of peptide regions from ArtQ and ArtM, encompassing the key residues from the binding pocket interacting with arginine. Residues mutated in this study are highlighted in boldface. Numbers indicate the relative positions of the mutated residues within the polypeptide chains of ArtQ (220 amino acids [aa]) and ArtM (218 aa).
To address this question, we set out to construct and functionally analyze a hybrid transporter containing one intact and one impaired transmembrane substrate binding site. As a prerequisite, knowledge of the role of each of the amino acid residues forming the binding pocket is required. Thus, we first investigated the functional consequences of mutations at the level of purified protein components in a lipid environment. However, since the biochemical properties of the Art(QN)2 system have not yet been studied in detail, e.g., due to the unavailability of sufficient amounts of native lipids from T. tengcongensis, we performed mutational analysis with the closely related and well-characterized Art(MP)2 transporter from the thermophilic bacterium Geobacillus stearothermophilus (26–31).
Here, we demonstrate that only two transporter variants, containing ArtM(L67D) and ArtM(D160L), respectively, displayed strongly impaired ATPase and thus impaired transport activities. Based on these findings, we constructed a hybrid transporter containing one copy of wild-type ArtM and one copy of the ArtM(L67D) variant. The results of ATPase and transport assays clearly revealed that one intact copy of ArtM is sufficient for function.
RESULTS
Structural features and experimental setup.The key feature of the substrate binding sites within each of the two ArtQ subunits of the Art(QN)2 transporter of T. tengcongensis is a negatively charged potential pocket that is mainly formed by five residues, including E152 and E159, which are distributed on three out of five transmembrane helices (2, 3, and 4) (25). While E152 interacts directly with the guanidine group of arginine by forming salt bridges via its δ-carboxyl group, residues L67, N98, M156, and E159 are involved in hydrogen bonds and van der Waals interactions with the substrate. In addition, the benzene ring of Y102 pairs with the guanidine group (Fig. 1A). In the binding pocket of only one ArtQ subunit, the α-carbon atom of P66 forms an additional hydrogen bond to the carboxyl group of the substrate. Together with Y102, the residue was suggested to be generally important for binding amino acid substrates, while the remaining residues from the binding pocket would recognize the specificity of the substrates (25).
The ArtJ-(MP)2 transporter used as model system to investigate the functional roles of these residues consists of the binding protein ArtJ, which displays 10-fold higher affinity for arginine (KD, 0.04 μM) than for histidine/lysine (27), and a homodimer each of the transmembrane subunit, ArtM, and the nucleotide-binding subunit, ArtP (26). Crystal structures of ArtJ in complex with different ligands (27) and of ArtP2 with bound nucleotides (32) have been solved, whereas structural information on the complete transporter is elusive. However, a homology model of the ArtM dimer was built by the Swiss-Model homology server (http://swissmodel.expasy.org), using ArtQ as the template (PDB access code 4YMU) and guided our study (Fig. 1B).
ArtM displays 44% sequence identity with ArtQ (24), and all but one residue from the binding pocket of ArtQ is conserved in ArtM (Fig. 1C). The only exceptions are E159, which is conservatively replaced by D160, and M156, which is replaced by L157 in ArtM (Fig. 1C).
In order to construct a hybrid transporter containing only one intact transmembrane substrate binding site, we first performed mutational analysis of each of the residues from the binding site in ArtM, except for P66 and L157, to unravel their role in liganding a substrate molecule. (P66 was previously demonstrated not to be crucial for binding [31], and L157 was not considered, since the homology model does not propose interaction with the substrate.) To this end, the consequences of mutations were studied at the level of purified protein components in a lipid environment by monitoring ArtJ/arginine- and ArtJ/histidine-dependent stimulation of ATPase activity and, where appropriate, uptake of radiolabeled arginine. Since only negligible differences in ATPase activity were observed for both substrates, the results are discussed for arginine only.
Mutational analysis of amino acid residues from the putative substrate binding site of ArtM.In the first set of experiments, each residue was individually replaced by alanine (or accidentally by leucine, in the case of D160). The consequences of these mutations on ATPase activity of the variants are shown in Fig. 2A. Consistent with previous findings (27, 28, 31), wild-type Art(MP)2 displayed low basal ATPase activity in the absence of its cognate binding protein, ArtJ, which was stimulated in the presence of ArtJ/Arg or ArtJ/His to yield a specific activity of ∼0.5 μmol Pi · min−1 · mg−1. Replacing N99 and Y103 by alanine resulted in similar ATPase activities, indicating that both residues are dispensable for function.
ATPase and transport activities of Art(MP)2 variants with replacements in the substrate binding site. (A) Purified variants of Art(MP)2 were incorporated into liposomes formed from G. stearothermophilus lipids and assayed for ATPase activity in the absence (−ArtJ) or presence of ArtJ-arginine and ArtJ-histidine, as described in Materials and Methods. The addition of ArtJ in the absence of substrate did not change the basal activities of the transporter variants. Shown are means and standard errors (SE) (n ≥ 3). (B) Proteoliposomes containing the indicated variants were assayed for ArtJ-dependent uptake of l-[14C]arginine, as described in Materials and Methods. Each value is the mean of at least three separate determinations, with standard deviation (SD) shown as error bars. All values are corrected for background counts per min and measured with liposomes lacking complex protein.
Likewise, the conservative mutation L67A did not negatively affect ATPase activity of the variant. To the contrary, the rate of ATP hydrolysis was about doubled compared to that of the wild type. In order to explore the role of L67 further, we next analyzed the effect of a mutation that alters the chemical nature of the side chain, L67D. As shown in Fig. 2A, this variant displayed a strongly impaired rate of ATP hydrolysis. Consistently, the transport activity of the mutant was almost abolished (Fig. 2B). According to the crystal structure of ArtQ and the homology model of ArtM, the nitrogen atom of the α-amino group of L67, which is engaged in the peptide bond, forms a hydrogen bond with the carboxyl group of arginine (Fig. 1A and B). Thus, any residue at this position should actually maintain this interaction. Accordingly, we interpret our finding to mean that the substrate gains access to the binding pocket, but is improperly bound due to electrostatic repulsion between its carboxyl group and the γ-carboxyl group of L67D. Alternatively, electrostatic repulsion might occur between L67D and D160, which according to the homology model are about the same distance (∼5 Å) apart as L67D and the carboxyl group of arginine. Consequently, the triggering of ATPase activity to subsequently initiate the transport cycle is hampered.
Functional analysis of the transporter variant containing a replacement of the ion-pairing E153 by alanine revealed an interesting but not unprecedented phenotype—a substantial up-shift in ATPase activity (Fig. 2A). Kinetic analysis clearly revealed that the mutation affected Vmax (2.67 μmol Pi · min−1 · mg−1, compared to 0.8 μmol Pi · min−1 · mg−1 in the wild type) but not KM (0.7 mM compared to 0.6 mM in the wild type). Furthermore, transport activity of the mutant was also increased by ∼2-fold relative to that of the wild type (e.g., compare data at 60 s; Fig. 2B), suggesting that the increased rate of ATP hydrolysis is partially coupled to transport.
An increase in ATPase activity was previously observed for the Art(QN)2 variant containing a substitution by alanine of the homologous residue E152 (25). However, since transport assays were not feasible for the T. tengcongensis system, coupling to transport could not be investigated.
Moreover, a similar phenotype was found for mutations affecting ArtM(K159) (31). This residue is thought to be part of a “gate” region controlling access of the substrate to the transmembrane binding site once substrate is released from the binding protein (33). An upshift in ATPase activity was found regardless of the chemical nature of the replacing amino acid, pointing to a specific role of a lysine residue at this position. However, in contrast to ArtM(E153A), mutations affecting K159 caused a significant downshift in transport, thus exhibiting an uncoupled phenotype (see also below) (31).
To further explore the role of E153, we characterized the conservative replacement by aspartate, as well as combinations of E153A with other mutations. In all cases, the variants lacking the glutamate residue at position 153 displayed basically the same phenotype as E153A, albeit to different extents (Fig. 2A). Most interestingly, the negative effect on transporter activity of an aspartate residue replacing leucine at position 67 was compensated for by E153A, resulting in an even greater upshift in ATPase activity (5-fold increase in Vmax compared to that of the wild type) and about the same transport rate as that of the variant containing ArtM(E153A) (Fig. 2B).
However, when combined with a mutation of the above-mentioned K159 (K159I), the specific ATPase activity was highest, reaching the value observed for the single K159I mutant (∼5.5 μmol Pi · min−1 · mg−1 [see Fig. 4 in reference 31]). In contrast, the transport rate of the double mutant was similar to that of the wild type, whereas the single K159I mutant exhibited a rate of only about 23% relative to that of the wild type (see Table 2 in reference 31). Thus, the combination of both mutations somewhat balanced the transport activities of the single mutations.
Noticeable, the E153A/K159I double mutant displayed a significantly higher basal activity (-ArtJ) than that of the wild type (Fig. 2A). However, the molecular basis of intrinsic, uncoupled ATPase activity of ABC transporters is unknown. Possibly, mutations in a peptide region of the transmembrane subunits that is crucial for triggering ATPase activity affect uncoupled activity.
Together, these results clearly demonstrate that a glutamate residue at position 153 is essential for wild-type activity. In the absence of structural information, we can only speculate as to why mutations cause an upshift in ATPase activity. Possibly, fewer interactions of the ligand with the transporter, due to the lack of salt bridges with E153, result in higher structural flexibility. which in turn might more efficiently trigger ATPase and hence transport activity.
In contrast to mutations of E153, exchange of the other negatively charged residue from the binding pocket, D160, to leucine displayed the opposite phenotype, i.e., an almost complete loss of ATPase and transport activity (Fig. 2). This finding corroborates the proposed role of D160 in interacting with the α-amino group of the substrate (Fig. 2B). Thus, we reasoned that the double mutant L67D/D160L might be ideally suited for the construction of a hybrid transporter containing only one inactive substrate binding site. Strikingly, however, in the double mutant, ATPase activity was restored, albeit with higher basal activity, which at early time points was coupled to arginine uptake (Fig. 2). This observation suggests that a positional switch of a negatively charged residue and a hydrophobic residue can at least partially compensate for the negative effects of the single mutations. Moreover, this finding is in line with the above hypothesis that the substrate has access to the binding pocket in the L67D variant.
Based on these results, we concluded that only the single mutations L67D and D160L were suited for the construction of a hybrid transporter. Since L67 is more conserved among homologous transporters (25), the L67D mutant was chosen for the intended study.
Functional analysis of a hybrid Art(MP)2 transporter containing one intact and one impaired substrate binding site.To obtain a hybrid transporter, we first constructed a plasmid (pJH146) bearing two artM alleles, artM and artM(L67D), which translate in ArtM proteins with a C-terminal His6 tag (ArtM-his) and a C-terminal twin Strep-tag, ArtM(L67D)-Strep, respectively (for details, see Materials and Methods). Since coexpression with a plasmid harboring artP turned out to yield only poor amounts of transporter protein, overproduction of ArtM variants and ArtP was performed separately. To begin with, the possible ArtM dimers [ArtM-his–ArtM-his; ArtM(L67D)-Strep–ArtM(L67D)-Strep; and ArtM-his–ArtM(L67D)-Strep] were isolated from membranes of the overproducing strain and subjected to chromatography on a Strep-Tactin Sepharose matrix. Only dimers containing one or two Strep-tags were retained and eluted after removal of proteins with only His-tags (ArtM-his–ArtM-his). In a second step, the eluted proteins [ArtM(L67D)-Strep–ArtM(L67D)-Strep and ArtM-his–ArtM(L67D)-Strep] were incubated with a Talon resin, to which only the desired hybrid ArtM-his–ArtM(L67D)-Strep would bind via its His tag. At this stage, purified ArtP was added to the matrix to assemble Art[M-his-M(L67D)-Strep]P2 complexes, as described previously (30). Complexes were eventually eluted by washing with buffer containing imidazole. See Fig. S1 in the supplemental material for a cartoon illustrating the procedure.
By the same protocol, complex variants containing two wild-type copies of ArtM or two copies of ArtM(L67D) were purified as controls. The composition of the resulting transporter complexes was verified by sodium dodecyl sulfate (SDS) gel electrophoreses and immunoblotting using anti-His and anti-Strep-tag antibodies, respectively. As shown in Fig. 3A, ArtM variants with His tags and Strep-tags migrate differently and can clearly be distinguished by their reaction with the respective antibodies.
Analysis of hybrid Art(MP)2 complexes containing no, one, or two L67D replacements in ArtM2. (A) Coomassie-stained SDS gel (left) and immunoblots (center and right) of Art(M-his-M-Strep)P2 (lane 1), Art[M-his-M(L67D)-Strep]P2 (lane 2), Art[M(L67D)-his-M(L67D)-Strep]P2 (lane 3), and Art(MP-his)2 (lane 4). Immunoblots were developed with anti-His and anti-Strep-tag antibodies, respectively, to verify the presence of ArtM copies with His tags and Strep-tags in each complex. For details, see Materials and Methods and Fig. S1. (B) ATPase activities of proteoliposomes containing the considered hybrid complexes were monitored as described in Materials and Methods and in the legend to Fig. 2A. SE mean (n ≥ 3 biological replicates). (C) Uptake of l-[14C]arginine into proteoliposomes containing the considered hybrid complexes. SE mean (n ≥ 3 biological replicates). For clarity, only the residues located at position 67 of the ArtM dimers are indicated. “Wild type” represents the activities of Art(MP)2 complexes prepared by the standard protocol.
In the case of hybrid and double mutant (lanes 2 and 3, respectively) staining intensities are similar, suggesting a 1:1 stoichiometry of ArtM with His tags and Strep-tags. For the Art(M-his-M-Strep)P2 construct (lane 1), we repeatedly observed a stronger staining intensity of ArtM-Strep-tag compared to ArtM-his, the reason for which remained unclear. Comparing the subunit compositions of hybrid complexes (lanes 1 to 3) with an Art(MP)2 complex prepared under standard conditions by taking advantage of a His tag C-terminally fused to ArtP (lane 4) clearly revealed that the hybrid complex preparations were not contaminated with tagless ArtM.
After incorporation into proteoliposomes, all variants, together with a wild-type transporter, Art(MP-his)2, purified by the standard procedure, were assayed for ATPase and transport activity. Compared to the wild type the Art(M-his-M-Strep)P2 variant displayed about 60% of ArtJ/Arg-stimulated ATPase activity (Fig. 3B, column for L67/L67). This difference is probably due to a less efficient complex assembly under in vitro conditions. A similar specific activity was observed for the hybrid Art[M-his-M(L67D)-Strep]P2 (Fig. 3B, column for L67/L67D), whereas the double mutant (Fig. 3B, column for L67D/L67D) displayed only 20% of activity, in line with data shown in Fig. 2A. Thus, the presence of one intact substrate binding site is sufficient to trigger ATP hydrolysis—but is it coupled to transport? Fig. 3C demonstrates that this is the case. Uptake of radiolabeled arginine by the hybrid (L67/L67D) was similar to that of the positive control (L67/L67) and was thus consistent with the ATPase data. In contrast, uptake was clearly impaired, albeit not fully abolished (see also Fig. 2B), when proteoliposomes containing the double mutant (L67D/L67D) were assayed.
DISCUSSION
Crystal structures of the homodimeric type I ABC transporter, Art(QN)2, for positively charged amino acids of T. tengcongensis revealed the presence of one substrate-binding pocket in each of the transmembrane ArtQ subunits (Fig. 1A) (25). The homology model of the closely related Art(MP)2 transporter of G. stearothermophilus proposes strong conservation of residues interacting with substrate in the binding sites (Fig. 1B). Results of the mutational analysis of ArtM presented in this communication confirm the proposed crucial role of two negatively charged residues—E153 and D160—in complexing substrates, such as arginine and histidine. In contrast, N99 and Y103 were found to be dispensable.
Notably, under the experimental conditions applied, ATPase activities of the mutants were equally affected by arginine and histidine, respectively, although the KD values of the high-affinity binding site in ArtJ differ 10-fold (27). However, these results are consistent with similar positioning of both substrates in the crystal structures of Art(QN)2 (see Fig. 4A in reference 25), although no binding affinities for either ArtQ or ArtM are known. This might be different in transporters that interact with more than one substrate-binding protein differing in substrate specificity and affinities. The GlnPQ transporter of Lactococcus lactis contains two copies each of substrate-binding domains (SBD1 and SBD2) fused to the TMDs. While SBD1 exhibits preference for asparagine over glutamine, SBD2 displays a high affinity for glutamine only. In their elegant study using single-molecule spectroscopy, Gouridis et al. provided evidence that late steps in the translocation cycle might be dependent on the amino acid (34). Whether these findings are of general importance for type I ABC transporter or are confined to systems interacting with two or multiple substrate-binding proteins/domains remains to be elucidated, because to date, the number and chemical nature of the putative transmembrane substrate binding site(s) in GlnPQ are unknown.
Furthermore, we demonstrate by ATPase and transport assays of a hybrid Art[M-M(L67D)P]2 transporter that, mechanistically, one intact substrate binding site in the transmembrane subunits is sufficient for function. Thus, the question arises of whether these findings are physiologically meaningful. First of all, the crystal structures of Art(QN)2, complexed with one molecule of arginine/histidine per ArtQ subunit, show the transporter in an inward-facing conformation, with the ArtN dimer residing in a semiopen state (25). This is in contrast to an X-ray structure obtained for a catalytic intermediate of the E. coli maltose/maltodextrin transporter, which was captured in the outward-facing conformation with a closed NBD dimer (MalK2) and complexed with discharged SBP (maltose binding protein) (23). This difference was attributed to the failure to crystallize the Art(QN)2 transporter in complex with its cognate binding protein, ArtI (25), as both ATP and liganded SBP are required for full closure of the NBDs (13, 15, 16). Thus, it cannot be ruled out that under in vivo conditions only one binding site is loaded (by chance?) with substrate at any given time point, which would be in line with the in vitro analysis of Art(MP)2 variants shown here.
We must also take into account that the experimental setup to monitor ATPase and transport activities of the Art(MP)2 mutants does not exactly represent the conditions in intact cells. In Gram-negative bacteria such as E. coli, SBPs freely diffuse in the periplasm, the space between the cytoplasmic and the outer membrane (5, 35). Although unknown for the majority of ABC importers, in the case of the maltose transporter of E. coli and the histidine transporter of Salmonella enterica, a 30- to 50-fold excess of SBP over transport complexes has been determined (36, 37). As shown for the maltose transporter in vitro, mechanistically, a 1:1 ratio of SBP and transport complex is sufficient for function, but ATPase and transport activities are substantially increased by increasing concentrations of liganded maltose-binding protein (15).
In contrast, T. tengcongensis and G. stearothermophilus are Gram-positive bacteria, lacking an outer membrane. Here, SBPs are extracellularly tethered to the cytoplasmic membrane via fatty acids that are covalently bound to the amino-terminal cysteine residue of the proteins (5, 35, 38). However, to our knowledge, the number of SBPs associated with one transport complex is not known for any of the studied systems. Moreover, due to the toxic effects of lipoproteins when overproduced in E. coli host strains (38), activity measurements in vitro, as in this study, are usually performed with excess soluble SBP that lacks the N-terminal cysteine. Thus, we cannot fully exclude the possibility that under in vivo conditions, perhaps with just one or only a few copies of SBP per complex, the mechanistic situation differs from that of the in vitro data.
On the other hand, homodimeric amino acid ABC importers are not confined to Gram-positive bacteria. Actually, crystal structures of the E. coli methionine transporter, Met(NI)2, in the inward-facing conformation but lacking a substrate(s), show high overall structural similarity in the transmembrane subunit MetI to that of ArtQ (25, 39, 40). This observation therefore renders rather unlikely the possibility mentioned above that the use of soluble rather than membrane-associated ArtJ affected the results.
Assuming that both binding sites are actually occupied with substrate under in vivo conditions, this would imply as already pointed out in the Introduction that two consecutive interactions with substrate-loaded SBP are required. However, such a scenario would be inconsistent with the alternate access model of type I importers based on crystal structures and with biochemical evidence of the heterodimeric E. coli maltose transporter. Here, once substrate is delivered to the transmembrane binding site in the MalF subunit in the ATP-bound state of the transporter, maltose-binding protein in its now open conformation forms a tight complex with the transporter, prior to ATP hydrolysis (14, 23, 41, 42). It is thought that back diffusion of substrate from the low-affinity transmembrane binding site is thereby prevented.
Interestingly, however, deviations from this scheme seem to exist. By applying fluorescence correlation spectroscopy, Doeven et al. observed release of the cognate binding protein OppA from the oligopeptide ABC transporter OppBCDF of Lactococcus lactis after delivery of the substrate (43). Therefore, it is tempting to speculate that under these conditions, a second interaction of a substrate-loaded binding protein with the transporter and hence delivery of the substrate to the second transmembrane binding pocket might occur prior to ATP hydrolysis. Consequently, the question of whether one or both substrate molecules are released to the cytoplasmic side of the membrane per the transport cycle remains to be elucidated. However, our finding that ATPase and transport activities of the hybrid transporter are similar to those of the wild-type variant argue in favor of the thus-far assumed stoichiometry of two molecules of ATP consumed per molecule of substrate translocated (44). Clearly, further investigations will be required to clarify this matter.
To our knowledge, the results presented in this communication provide together the first biochemical insight into the chemical nature of a transmembrane substrate-binding site of a type I ABC importer. Moreover, our data do at least not exclude the possibility that homodimeric and heterodimeric importers might differ in details of the alternate access mode.
MATERIALS AND METHODS
Bacterial strains, plasmids, and oligonucleotides.Bacterial strains and plasmids used in this study are listed in Table S1 in the supplemental material. Strain XL10-Gold was used for mutagenesis, strain BL21(DE3) T1R was applied for protein overproduction, and strain NEB 5α was used for plasmid storage and production. Oligonucleotides used for plasmid constructions are listed in Table S2 in the supplemental material.
Cloning and site-directed mutagenesis.ArtM[P-Strep] with a C-terminal Strep-tag was constructed by using plasmid pRcCbiM(H2D)N(S), which was kindly provided by the Eitinger group (Humboldt Universitaet zu Berlin). The gene cbiMN was cut out by using the flanking restriction sites for NcoI and BglII. Plasmid pRF2 containing artMP was digested with the same enzymes and the resulting artMP fragment that was subsequently ligated into the restricted vector, obtaining pJH62.
Cloning of the hybrid Art[M-his][M-Strep] was performed in two steps, making use of a HindIII restriction site of artM-his on pVE15, downstream of the stop codon. A fragment containing artM with a sequence encoding Strep-tag and a Shine-Dalgarno sequence flanked by HindIII restriction sites was amplified using the oligonucleotides artMHindIIIStrep_up and artMHindIII_down (Table S2) and pVE15 as a template. The resulting PCR product and pVE15 were restricted with HindIII, ligated, and transformed, obtaining the plasmid pJH136 with artM-his, followed by artM-Strep which is cotranscribed but separately translated. Because a single Strep-tag proved to be insufficient for binding of Art[M-his][M-Strep] to a Strep-tag affinity matrix, a twin Strep-tag was introduced using the oligonucleotides LinkerStrep_down and Strep_up (Table S2), with the help of the Q5 site-directed mutagenesis kit (NEB), resulting in the peptide sequence AWSHPQFEKGGGSGGGSGGSAWSHPQFEK of two Strep-tags (underlined), separated by a linker sequence fused C-terminally to ArtM. The mutation L67D in one or both artM alleles was introduced by repeating the cloning procedure with plasmid pJH133 carrying artM(L67D), which was derived by site-directed mutagenesis of pVE15 using the QuikChange Lightning site-directed mutagenesis kit. All constructs described above were verified by commercial sequencing (LGC, Berlin, Germany).
Protein preparations.Art(MP)2 (wild-type and variants) was purified by making use of the Strep-tag engineered to the carboxy terminus of ArtP. E. coli strain BL21(DE3)T1R was transformed with placI and pJH62 or its derivatives and grown in Luria broth (LB)-ampicillin-chloramphenicol to an optical density at 650 nm (OD650) of 0.5 prior to the addition of 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for induction of gene expression. For purification of Art(MP)2, the membrane fraction obtained after disintegration of cells by ultrasonication and ultracentrifugation (1 h at 200,000 × g) was resuspended in 50 mM 3-(N-morpholino)propanesulfonic acid (MOPS)–KOH (pH 7.5), 5% (vol/vol) glycerol, 0.3 M NaCl, and 0.1 mM phenylmethylsulfonyl fluoride (PMSF) and solubilized with 1.1% n-dodecyl-β-d-maltopyranoside (DDM). After ultracentrifugation (30 min at 200,000 × g) the supernatant was incubated with Strep-Tactin Sepharose (IBA Lifescience, Germany) for 1 h, after which the matrix was washed with 15 column volumes of 50 mM MOPS-KOH (pH 7.5), 5% (vol/vol) glycerol, 0.3 M NaCl, and 0.05% DDM. Subsequently, transport complex protein was eluted with buffer supplemented with 5 mM desthiobiotin. Pooled protein fractions were concentrated by a centrifugal device (Amicon, Millipore) and buffered in 50 mM MOPS-KOH (pH 7.5), 5% (vol/vol) glycerol, 0.1 M NaCl, and 0.02% DDM by passage through a PD10 desalting column (GE Healthcare). Art(MP)2 (wild type only) carrying a His tag fused to the carboxy terminus of ArtP was purified as described elsewhere (31).
Preparation of Art[M-his][M-Strep]P2 hybrid complexes was performed in two steps. First, his-ArtP was purified as described in reference 30, and the His tag was subsequently cleaved off by thrombin treatment. Second, the ArtM hybrid was purified as described above for Art(MP)2. The eluate from the Strep-Tactin Sepharose was directly loaded onto a cobalt-loaded Talon resin and washed with 10 column volumes of buffer. Then, ArtP was added to approximately twice the amount of the ArtM hybrid and incubated with the matrix on ice for 15 min. After washing the matrix with 10 column volumes of buffer, the Art[M-his][M-Strep]P2 complex was eluted with 250 mM imidazole, concentrated, and rebuffered as described above (see Fig. S1 for a schematic presentation of the procedure).
ArtJ was purified as described in reference 31. Denaturation and renaturation of ArtJ in order to remove prebound substrate for transport assays was carried out as described in reference 25.
Preparation of proteoliposomes.Art(MP)2 or Art[M-his][M-Strep]P2 were incorporated into liposomes for ATPase and transport assays, essentially as described in Weidlich et al. (31) Briefly, lipids prepared from G. stearothermophilus (20 mg) were dried in a rotating vacuum evaporator, redissolved in 1 ml of 50 mM MOPS-KOH (pH 7.5) containing 1% octyl-β-d-glucopyranoside (OG), and sonicated for 15 min. Subsequently, the arginine transporter variants (50 μg) were added to 125 μl of the lipid-detergent mixture, resulting in a final volume of 300 μl. For ATPase activity assays, proteoliposomes were formed by removal of detergent by adsorption to 100 mg of Bio-Beads (Bio-Rad) at 4°C overnight. After replacing the beads with a new batch, incubation continued for 1 h.
For transport assays, proteoliposomes were formed by fast dilution of the transporter-lipid mixture in 15 ml of 50 mM Tris-HCl (pH 7.5) containing 7.5 mM ATP. In both cases, the mixture was centrifuged for 1 min at 10,000 × g to remove aggregates, and proteoliposomes were subsequently recovered by ultracentrifugation for 30 min at 220,000 × g, and resuspended in 50 mM MOPS-KOH (pH 7.5) or 50 mM Tris-HCl (pH 7.5), respectively.
ATPase assay.ATPase activity was measured essentially as described in references 25 and 31. Assay temperature was 60°C. Reactions were started by adding 60 μg ArtJ with either 1 mM l-arginine or 1 mM l-histidine, 3 mM MgCl2, and 2 mM ATP to the preheated reconstituted complex. Aliquots were taken in 1 min-intervals and placed into wells of a microtiter plate containing 25 μl of a 12% SDS solution. The amount of liberated phosphate was determined colorimetrically with ammonium molybdate complexes, using Na2HPO4 as the standard.
Transport assay.The transport assays were carried out essentially as described in reference 30. The assay temperature was 55°C. Briefly, proteoliposomes were mixed with ArtJ and l-[14C]arginine (50 mCi/mmol; Hartmann Analytics) to yield final concentrations of 6 and 10 μM, respectively, in 50 mM Tris-HCl (pH 7.5) and 1 mM MgCl2. After 15, 30, 60, 120, and 240 s, 25-μl aliquots were diluted in 1 ml of 50 mM Tris-HCl (pH 7.5) and 1 mM l-arginine, followed by rapid filtration through a 0.22-μm nitrocellulose filter (Millipore) that had been equilibrated in the same buffer. Subsequently, the filters were washed once with 5 ml of 50 mM Tris-HCl (pH 7.5) and dried for 1 h at room temperature, and retained radioactivity was determined by liquid scintillation counting. For background correction, liposomes formed in the absence of protein were used.
Analytical procedures.Protein determination, SDS-polyacrylamide gel electrophoresis (SDS-PAGE), and immunoblotting were carried out as described in reference 29.
ACKNOWLEDGMENT
This work was supported by the Deutsche Forschungsgemeinschaft (SCHN 274/14-2 and SCHN 274/16-1).
FOOTNOTES
- Received 13 February 2018.
- Accepted 20 March 2018.
- Accepted manuscript posted online 26 March 2018.
- Address correspondence to Erwin Schneider, erwin.schneider{at}rz.hu-berlin.de.
Citation Heuveling J, Landmesser H, Schneider E. 2018. One intact transmembrane substrate binding site is sufficient for the function of the homodimeric type I ATP-binding cassette importer for positively charged amino acids Art(MP)2 of Geobacillus stearothermophilus. J Bacteriol 200:e00092-18. https://doi.org/10.1128/JB.00092-18.
Supplemental material for this article may be found at https://doi.org/10.1128/JB.00092-18.
REFERENCES
- Copyright © 2018 American Society for Microbiology.