The Secondary Multidrug/Proton Antiporter MdfA Tolerates Displacements of an Essential Negatively Charged Side Chain^,

Abstract

The largest family of solute transporters includes ion motive force-driven secondary transporters. Several well characterized solute-specific transport systems in this group have at least one irreplaceable acidic residue that plays a critical role in energy coupling during transport. Previous studies have established the importance of acidic residues in substrate recognition by major facilitator superfamily secondary multidrug transporters, but their role in the transport mechanism remained unknown. We have been investigating the involvement of acidic residues in the mechanism of MdfA, an Escherichia coli secondary multidrug/proton antiporter. We demonstrated that no single negatively charged side chain plays an irreplaceable role in MdfA. Accordingly, we hypothesized that MdfA might be able to utilize at least two acidic residues alternatively. In this study, we present evidence that indeed, unlike solute-specific secondary transporters, MdfA tolerates displacements of an essential negative charge to various locations in the putative drug translocation pathway. The results suggest that MdfA utilizes a proton translocation strategy that is less sensitive to perturbations in the geometry of the proton-binding site, further illustrating the exceptional structural promiscuity of multidrug transporters.

One central mechanism of bacterial drug resistance involves extrusion of drugs from the cell by membrane transporters (1–4). Several transport systems are rather specific and deal with a narrow range of cytotoxic compounds with clear preference toward certain drugs (3, 5–7). Other transporters, which handle a wide spectrum of structurally dissimilar drugs, are called multidrug (Mdr)2 transporters, and their capacity is translated into a multidrug resistance phenotype observed both in eukaryotic and in prokaryotic cells (3, 8, 9). The bacterial Mdr transporters fall into one of five families, of which the major facilitator superfamily (MFS) of secondary transporters is the largest (10). Being abundant in the genomes of many bacterial strains, the MFS-Mdr transporters pose intriguing mechanistic and evolutionarily related questions (e.g. Refs. 11 and 12). Several of these questions are being investigated utilizing MdfA from Escherichia coli (13) as a model for secondary Mdr transporters.

MdfA (14), encoded by the cmr gene (15), is a 410-amino acid residue-long membrane protein with 12 transmembrane helices. Close homologues of MdfA have so far been identified mainly in pathogenic bacteria (for review, see Ref. 13). Cells expressing MdfA from a multicopy plasmid exhibit multidrug resistance against a diverse group of structurally and electrically dissimilar lipophilic and hydrophilic compounds including cationic and electrically neutral drugs (14, 16). Transport experiments have shown that MdfA is driven by the proton electrochemical potential, and in addition to its drug/proton exchange activity, it may also function as a (Na⁺)(K⁺)/proton antiporter (14, 17–19). Clearly, therefore, proton translocation by MdfA is crucial for all of its known functions in the cell.

Of all the open mechanistic questions regarding the function of MFS-Mdr transporters, the question of multidrug recognition has been characterized in detail (9, 20), whereas the mechanism underlying active transport remained the least understood. Specifically, although they are crucial for their drug/proton antiport activity (18), very little is known about proton recognition and translocation by MFS-Mdr transporters. In this regard, studies of several specific secondary transporters have been instrumental in providing clues for determining how protons may be recognized. The best example is the lactose/proton symporter, LacY, where two carboxyl side chains play irreplaceable roles in proton-coupled sugar translocation (21). Similarly, negatively charged residues are mechanistically involved in other antiporters and symporters (22–27). Stimulated by these studies, we have examined the role of acidic residues in MdfA. Surprisingly, we observed that no single acidic residue plays an irreplaceable role in drug resistance and transport by MdfA, although several are important for electrostatic interaction with cationic substrates (28, 29). This observation elicited several hypotheses, regarding the mechanism of proton recognition by MdfA. We tested the possibility that at least two acidic residues might be utilized alternatively. The results described here not only support this idea but also reveal an additional level of promiscuity, suggesting that a functionally critical negative charge still works even when moved to alternative locations inside the putative multidrug translocation pathway of MdfA.

MATERIALS AND METHODS

Bacterial Strains and Plasmids—E. coli UTmdfA::kan or its derivative UTLmdfA::kan3 with plasmid pT7-5/mdfA-His₆ were used for resistance, expression, transport assays (28) and library selection (see below).

Drug Resistance Assays—Drug resistance was assayed using solid media, as described previously (16). Briefly, transformed cells were grown at 37 °C in LB broth supplemented with the antibiotics ampicillin (200 μg/ml) and kanamycin (30 μg/ml) to 1.0 A₆₀₀ units. A series of 10-fold dilutions was prepared, and 4 μl of each dilution were spotted on plates containing different concentrations of the test drug. Colonies were recorded after 16–24 h of incubation at 37 °C. For selective screening of MdfA mutant libraries, cells were transformed by electroporation with the plasmid library and seeded on plates supplemented by lethal concentrations of the test drug. Resistant colonies were isolated after 30 h at 37 °C.

Expression Analysis by Western Blotting—Overnight cultures were diluted and grown to an A₆₀₀ of 1 unit. Bacteria were harvested, and membranes were prepared as described previously (30). Membrane fractions (5–10 μg of protein) were then subjected to 12.5% SDS-PAGE, electroblotting, and detection using India HisProbe-horseradish peroxidase (Pierce) and enhanced chemiluminescence (ECL).

Transport Assays—EtBr efflux assays were conducted as described (28), with modifications. Overnight cultures were diluted to 0.04 A₆₀₀ units, grown at 37 °C to 1.0–1.2 A₆₀₀ units, and kept on ice. Aliquots of cells (0.3 A₆₀₀ units) were pelleted, resuspended in 2 ml of HEPES buffer (25 mm, pH 6.5), and loaded with EtBr (5 μm) at 37 °C for 5 min in the presence of carbonyl cyanide m-chlorophenyl hydrazone (100 μm). Loaded cells were then centrifuged, resuspended in the same buffer containing EtBr (5 μm) without carbonyl cyanide m-chlorophenyl hydrazone, and subjected to fluorescence measurements. After ∼1 min in the fluorometer, glucose was added (to 0.4%), and EtBr efflux was monitored continuously by measuring the fluorescence, using excitation and emission wavelengths of 480 and 620 nm, respectively. For TPP and chloramphenicol uptake assays, harvested cells (0.6 A₆₀₀) were washed once with potassium phosphate buffer (50 mm, pH 6.5) and 1 mm MgSO₄, resuspended in the same buffer at 10–15 A₄₂₀ units, and divided (50-μl aliquots). Following 2 min of recovery at 37 °C in the presence of 0.2% glucose, transport was initiated by the addition of [³H]chloramphenicol (5 μm) or [³H]TPP (50 μm). Transport was terminated by rapid filtration as described previously (14).

Measurement of Substrate Binding—E. coli UTLmdfA::kan transformed with plasmid pT7-5/araP/mdfA-His₆ encoding the indicated mutants were grown, and membrane vesicles were prepared as described previously (31). For each experiment, 200 μl of membranes (containing 15 mg/ml protein) were quickly thawed in 37 °C and centrifuged for 10 min at 200,000 × g. The membrane pellet was resuspended and homogenized in 1.9 ml of ice-cold 50 mm potassium phosphate buffer, pH 7.3. Next, for each reaction, 40 μl of membranes were mixed with 10 μl of 25 mm TPP or water, and binding was carried out by 30 min of incubation at room temperature. Samples were then stored on ice. Cysteine labeling was initiated by the addition of 10 μl of maleimide-polyethylene glycol (Mal-PEG) (5 kDa) and 3 min of incubation at room temperature. The reaction was terminated by the addition of 10 μl of 60 mm dithiothreitol and 10 min of incubation on ice. The samples were analyzed by Western blotting as described above. The reaction product appeared as a slower migrating MdfA-Mal-PEG band.

Construction of MdfA Mutant Library—Plasmid pT7-5/mdfA-His₆ was modified to carry a NotI site before the His₆ tag coding sequence at the 3′ end of mdfA and a NcoI site at the 5′ end, which resulted in a Q2A mutation. The modified gene (termed pT7-5/mdfA-His₆) was examined for MdfA expression and function and found to be indistinguishable from wild type MdfA. Wobble base PCR (20 cycles with Bioline Taq polymerase) was performed using primers: sense 5′-gaaagcgtagctatactcg-3′ and antisense 5′-gacagatcgatgagatagg-3′ with 3 ng of pT7-5/mdfAHis₆(E26T/D34M) in the presence of 0.02 mm 8-oxo dGTP and 1.5 mm MgCl₂. Following DpnI treatment and DNA purification, a second PCR was performed with the same primers. The purified product (150 ng) was overdigested (20 h) with NotI and NcoI and ligated to a NotI/NcoI-digested plasmid, which carries an irrelevant gene instead of mdfA (100 ng). E. coli UTLmdfA::kan was transformed by the ligation products and seeded on ampicillin-supplemented plates. About 10⁵ colonies were collected, and a plasmid library was prepared.

RESULTS AND DISCUSSION

Construction and Characterization of Double Mutants of MdfA—The results of individual neutralizations of acidic residues in MdfA (29) were surprising because they raised the possibility that as an Mdr transporter, MdfA functions in a manner that is different from other known secondary transporters (both symporters and antiporters) driven by the ion electrochemical gradient, where the mechanism involves at least one essential acidic residue. Therefore, we hypothesized that either (i) proton recognition and translocation by MdfA does not require acidic residues, or alternatively, (ii) two or more acidic residues may substitute for each other. To approach the second possibility, we chose to focus on two, putatively membrane-embedded acidic residues, Glu-26 and Asp-34. Previous studies showed that each of these acidic side chains is important for transport of cationic substrates but not neutral ones (28, 29). As controls, we utilized two additional acidic residues, Glu-45 and Asp-52, in the putative periplasmic loop 1. Altogether, six double mutants were constructed by neutralizing combinations of the chosen acidic residues with cysteines (Fig. 1A). The single and double mutants were tested for expression (Fig. 1B) and ability to confer multidrug resistance against positively charged (EtBr, benzalkonium, erythromycin) and neutral (chloramohenicol, thiamphenicol) compounds (Fig. 1C). The results show that replacement of only one specific set of acidic residues (E26C/D34C) abrogated MdfA-mediated resistance to all the test drugs (see also supplemental Table 1S). Next, the EtBr and chloramphenicol transport activity of each double mutant was assayed. As expected from the results of resistance experiments, none of the double mutants catalyzed efflux of the cationic substrate EtBr (not shown), further illustrating the importance of electrostatic interactions in the multidrug recognition pocket of MdfA (28). In contrast, several of the mutants retained chloramphenicol transport activity (Fig. 1D). Interestingly, although all the D34C-harboring double mutants are severely defective in their chloramphenicol resistance and transport activity, only mutant E26C/D34C had lost all of its tested functions, including resistance to thiamphenicol. Of all the other mutants, D34C/E45C was also severely impaired; however, introducing another neutralizing mutation at position 34 (D34M/E45C) did not abolish resistance (supplemental Fig. S1). To further evaluate the suggestion that neutralizing Glu-26 and Asp-34 simultaneously is detrimental, in addition to the cysteine replacement described earlier, we characterized other double E26X/D34Z mutants (Fig. 2), including the two combinations of the most active single substitutions E26Q and E26T with D34M. As shown, all the single mutants retained some resistance activity. In contrast, although the double mutants were expressed at levels comparable with that of wild type MdfA (Fig. 2, A and C), none of them was able to support growth on drug-supplemented plates (Fig. 2, B and D). These results demonstrate that individually, the acidic residues at position 26 and 34 play important roles in cationic substrate recognition but are not essential for transport activity (as shown with neutral substrates). The detrimental effect of neutralizing both sites simultaneously suggests that membrane-embedded acidic residues might play an essential role in MdfA and that this role can be alternatively fulfilled by either Glu-26 or Asp-34. However, the loss of function could also reflect improper folding of the double mutant.

FIGURE 1.

Characterization of double mutants of MdfA. A, ribbon representations of MdfA (32) show the mutated acidic residues. B, expression of the mutants is shown by Western blotting. WT, wild type; vec, vector. C, multidrug resistance activity of MdfA and mutants spotted on LB agar plates. D, uptake of [³H]chloramphenicol by cells expressing the indicated mutants was assayed by rapid filtration. The experiments were performed as triplicates and repeated three times. Bars indicate standard errors of mean.

FIGURE 2.

Characterization of 26/34 double mutants of MdfA. A and C, Western blot analysis of expression. WT, wild type; vec, vector. B and D, multidrug resistance activity of MdfA and mutants spotted on LB agar plates.

MdfA E26T/D34M Is Properly Folded in the Membrane—Because the double mutant has no transport activity, we examined whether replacement of Glu-26 and Asp-34 has a detrimental structural effect in MdfA, utilizing drug binding assays with membranes harboring MdfA mutants. To this end, we utilized a mutant of MdfA that is devoid of any native cysteines and contains a single cysteine at position 54 (single Cys-54) (31). In control experiments, we observed that binding of TPP to Cys-less MdfA Cys-54 stimulated the reactivity of residue Cys-54 with the cysteine-reactive label, Mal-PEG (Fig. 3). This increase in cysteine reactivity indicates that substrate binding induces a conformational change in MdfA that is accompanied by exposure of Cys-54 to the aqueous solvent (31). As observed previously (31), chloramphenicol did not have a detectable effect of labeling. In contrast to Cys-less MdfA Cys-54, TPP did not stimulate labeling of residue Cys-54 in the inactive MdfA R112M mutant, suggesting that TPP-induced stimulation requires a functional transporter (Fig. 3). Notably, a faster migrating band is more pronounced in the sample of inactive R112M mutant, possibly representing a degradation product, an indication that this mutant might indeed be structurally impaired. Studying the reactivity of single Cys-54 also carrying simultaneously E26T and D34M mutations showed that TPP does stimulate the reactivity of Cys-54 in this mutant, suggesting that it is correctly folded and conformationally responds to the substrate (Fig. 3). These results suggest that mutating Glu-26 and Asp-34 does not have a detrimental structural effect on MdfA. Therefore, considering the mechanisms of other secondary transporters, we propose that proton recognition and translocation by MdfA may be mediated through either Glu-26 or Asp-34.

FIGURE 3.

Effect of TPP on single cysteine labeling in various MdfA mutants. A, membranes from cells overexpressing the indicated single cysteine mutant were incubated with or without TPP (T) or chloramphenicol (C) and then with maleimide-PEG. The labeled product appeared as a slower migrating band in SDS-PAGE as probed by Western blotting. The results shown are representative, B, TPP binding is indirectly deduced from its quantified stimulatory effect on labeling. Averages and standard errors of the mean (bars) from three independent experiments are shown for the experiments ± TPP.

Functional Displacement of an Essential Acidic Residue by Site-directed Mutagenesis—If our proposal that Glu-26 and Asp-34 are able to act alternatively as mediators of proton-coupled antiport by MdfA is true, then it must mean that unlike in solute specific secondary transporters, the exact location of the negative charge in the transport pathway of MdfA may not be critical. To test this, we first examined whether an acidic residue can be functionally placed in between and instead of the acidic residues at positions 26 and 34. Accordingly, several mutants of MdfA(E26Q/D34M) were characterized. All the mutants were expressed in the membrane fraction at levels similar to that of wild type MdfA (Fig. 4A), but only two, E26Q/T29E/D34M and E26Q/G32E/D34M, reproducibly showed a low resistance activity against thiamphenicol (Fig. 4B). The orientation of the newly inserted acidic residues (T29E and G32E) with respect to the putative translocation pathway is different from that of Glu-26 or Asp-34 (Fig. 4C), thus offering a reasonable explanation for why the mutants exhibit such a low resistance activity.

FIGURE 4.

Characterization of the E26Q/D34M triple mutants of MdfA with engineered acidic residue in transmembrane helix 1. A, Western blot analysis of expression. WT, wild type; vec, vector. B, multidrug resistance activity of MdfA and mutants spotted on LB agar plates. C, ribbon representation of transmembrane helix 1 (Ref. 32) with residues Glu-26, Thr-29, Gly-32 and Asp-34 shown as sticks.

Functional Displacement of the Essential Acidic Residue of MdfA by Random Mutagenesis—Encouraged by the results that show partial functional displacements of the essential negative charge, we designed a genetic screen as a means to identify better sites that may functionally accommodate the essential negative charge. This was accomplished by utilizing the inactive double MdfA mutant E26T/D34M as a template for construction of a library of MdfA mutants by error-prone PCR (see “Materials and Methods”). Cells transformed with the library were plated on LB agar plates supplemented with thiamphenicol (10 μg/ml) or chloramphenicol (3 μg/ml). Two resistant colonies out of ∼10,000 were isolated, and the purified plasmids were characterized by retransformation and sequenced along the entire mdfA gene. In addition to the original E26T/D34M replacements, both mutants had two mutations: A150E/I327L (isolated from thiamphenicol plates) and A150E/S309P (isolated from chloramphenicol plates). Next, each of the new mutations was individually subcloned into the double mutant E26T/D34M, and the resulting triple mutants were tested as above. The results show that all the mutants were expressed (Fig. 5A) and that only mutants containing the A150E substitution conferred significant resistance against several of the test drugs (Fig. 5B). Notably, although single mutants neutralized at positions 26 or 34 are unable to transport cationic substrates (29), the new mutants with a single negative charge at position 150 also regained function against positively charged substrates, such as TPP. Therefore, introduction of a negative charge at position 150 (instead of Glu-26 and Asp-34) restores both the transport activity and the proposed electrostatic interaction with cationic substrates (28). The three-dimensional structural model of MdfA (32) shows that residue 150 points into the putative Mdr recognition pocket of MdfA, on the same region as Glu-26 and Asp-34 (Fig. 5C). To evaluate the activity of the new mutants further, transport assays were conducted with the respective cationic and neutral substrates TPP and chloramphenicol. Fig. 6 shows that, as with wild type MdfA, TPP and chloramphenicol uptake was significantly decreased in cells expressing the triple mutant E26T/D34M/A150E, whereas cells harboring empty vector or the inactive double mutant E26T/D34M accumulated the radiolabeled substrate. These results are in accordance with the behavior of the mutants in drug resistance assays.

FIGURE 5.

Characterization of the MdfA constructs harboring combinations of A150E, S309P, and I327L. A, Western blot analysis of expression. WT, wild type; vec, vector. B, multidrug resistance activity of MdfA and mutants spotted on LB agar plates. C, ribbon representations of MdfA (Ref. 32) show acidic residues at positions 26, 34, and 150.

FIGURE 6.

Transport activity of selected MdfA mutants. Uptake of [³H]TPP (A) or [³H]chloramphenicol (B) was assayed by rapid filtration. The experiments were performed as triplicates and repeated three times. Bars indicate standard errors of mean. WT, wild type; vec, vector.

Hypothesis—What is the role of the essential negative charge in secondary multidrug transport, and why is it spatially rather insensitive? We favor the possibility that the negative charge is involved in both substrate and proton recognition as shown previously for the small Mdr transporter EmrE (26), although protonation-deprotonation remains to be shown directly with MdfA. A tentative explanation for the observation that the negative charge can be displaced functionally would be that active transport is driven by competition between protons and substrate molecules on binding to an acidic residue. As discussed previously, drugs interact with MdfA and other multidrug transporters by spatially promiscuous interactions (20, 33). Therefore, if the drug interaction site would also contain a negative charge, on which protons and substrates could compete, a transport mechanism might be established. Such a mechanism is entirely different from that proposed for solute-specific secondary transporters, which require precise location and geometry of catalytic residues (21, 34). Therefore, we speculate that secondary Mdr transporters have a unique structural design that allows a remarkable degree of freedom. Acidic residues can be functionally implanted at various faces of the translocation pathway, provided that they form a binding site shared by substrates and protons in a mutually exclusive manner as a prerequisite for active antiport.