Membrane topology of the transposon 10-encoded metal-tetracycline/H+ antiporter as studied by site-directed chemical labeling.

The transposon (Tn) 10-encoded metal-tetracycline/H+ antiporter (Tn10-TetA) is predicted to have a membrane topology involving 12 transmembrane domains on the basis of the hydropathy profile of its sequence and the results of limited proteolysis; however, the experimental results of limited proteolysis are not enough to confirm the topology because proteases cannot gain access from the periplasmic side (Eckert, B., and Beck, C. F. (1989) J. Biol. Chem. 264, 11663-11670). One or two cysteine residues were introduced into each predicted hydrophilic loop or the N-terminal segment of Tn10-TetA by site-directed mutagenesis, and then the topology of the protein was determined by examining whether labeling of the introduced Cys residue by membrane-permeant [14C]N-ethylmaleimide ([14C]NEM) was prevented by preincubation of intact cells with the membrane-impermeant maleimide, 4-acetamido-4'-maleimidylstilbene-2,2'-disulfonic acid (AMS). The binding of [14C]NEM to the S36C (loop 1-2), L97C (loop 3-4), S156C (loop 5-6), R238C (loop 7-8), S296C (loop 9-10), Y357C, and D365C (loop 10-11) mutants was completely blocked by pretreatment with AMS, indicating that these residues are located on the periplasmic surface. In contrast, [14C]NEM binding to the S4C (N-terminal segment), S65C (loop 2-3), D120C (loop 4-5), S199C and S201C (loop 6-7), T270C (loop 8-9), and S328C (loop 10-11) mutants was not affected by pretreatment with AMS, indicating that these residues are on the cytoplasmic surface. These results for the first time thoroughly confirm the 12-transmembrane topology of the metal-tetracycline/H+ antiporter.

Tn10-TetA is regarded as a member of the major facilitator family including uniporters and symporters because of the structural similarity and the conserved sequence motif, DRXGRR (9). Tn10-TetA is important for elucidating the common structural and functional features of the major facilitator family because it is the only antiporter in this family of which the molecular mechanism of the transport has been extensively studied in detail. Using inside-out membrane vesicles, it has been shown that Tn10-TetA mediates the 1:1 antiport of a divalent cation-tetracycline complex with a proton (4,10). Sitedirected mutagenesis studies on Tn10-TetA revealed the presence of some functionally essential aspartic acid residues (11,12). One of these aspartic acid residues is located in the sequence motif common to the major facilitator family transporters (13). The functional and/or structural importance of the aspartic acid conserved in this motif was also confirmed in ␣-ketoglutarate permease (14) and lactose permease (15), suggesting a common molecular mechanism for symporters and antiporters.
Most major facilitator family transporters are predicted to contain 12 membrane-spanning domains (9). As to tetracycline/H ϩ antiporters, Eckert and Beck (1), and Henderson and Maiden (16) proposed similar 12-transmembrane segment models for the membrane topologies of Tn10-TetA and pBR322-TetA, respectively, based on the results of hydropathy analysis. The model of Tn10-TetA was partially confirmed by biochemical and immunological studies. (i) 4 of the 5 predicted cytoplasmic loops were cleaved by the proteases used from the inner side of the membrane (1). (ii) The N-terminal methionine is accessible for cyanate modification in inside-out vesicles but not in spheroplasts (1). (iii) The anti-C-terminal peptide antibody could bind to Tn10-TetA only from the inner side of the membrane (17). However, there is no biochemical evidence for the periplasmic location of any of the predicted periplasmic segments of Tn10-TetA, because it is resistant to protease digestion from the periplasmic side of the membrane (1). Allard and Bertrand (18) studied the membrane topology of pBR322-TetA, which is a tetracycline-resistant protein encoded by pBR322, by means of alkaline phosphatase (PhoA) gene fusions. With this technique, they succeeded in determining the periplasmic location of three of the six predicted periplasmic loops, whereas the fusion proteins connected at the predicted periplasmic loop 1-2, loop 3-4, and loop 9 -10 showed unexpectedly low enzyme activity (18). Since the hydropathy plot of Tn10-TetA (1) showed no distinct hydrophilic segment between TM3 and TM4, or TM9 and TM10, the periplasmic location of loop 3-4 and loop 9 -10 was not supported by either the biochemical evidence or the predictive algorithm.
Site-directed chemical labeling of cysteine residues with a sulfhydryl reagent is useful for analyzing the topology of a polytopic membrane protein (19). Tn10-TetA contains only one cysteine residue at position 377; however, SH reagents did not bind to Cys-377 due to it being buried in the hydrophobic interior of the membrane (20). There was no difference in [ 14 C]NEM binding between the wild-type and the C377A (Cysless) mutant (30). Therefore, the reactivity of an SH reagent to a mutant TetA protein represents the reactivity of the reagent to an introduced cysteine residue (20). In this study, we constructed 15 site-directed mutants of Tn10-TetA in which cysteine residues were introduced into putative periplasmic or cytoplasmic loop regions, followed by chemical labeling with the membrane-permeant SH reagent, [ 14 C]N-ethylmaleimide, after preincubation with or without a membrane-impermeant SH reagent to determine their sidedness.
Site-directed Mutagenesis-Cysteine mutants were constructed by oligonucleotide-directed site-specific mutagenesis according to the method of Kunkel (21) using pCT1183 as a template, and a synthetic oligonucleotide containing mismatches generating a codon change and silent mismatches to generate a new restriction site in order to check the mutation. Mutations were at first detected as the appearance of a newly introduced restriction site and then verified by DNA sequencing using a Shimadzu DNA sequencer DSQ-1000. Then the mutant tetA gene was transferred to low copy number plasmid pLGT2 by corresponding fragment exchange.
Determination of Tetracycline Resistance-The tetracycline resistance of E. coli W3104 cells harboring pLGT2 or its mutant plasmid was determined by the agar dilution method and expressed as the minimum inhibitory concentration.
Labeling of Mutant TetA Proteins with [ 14 C]N-Ethylmaleimide (NEM) and Its Prevention by 4-Acetamido-4Ј-maleimidylstilbene-2,2Јdisulfonic Acid (AMS)-E. coli W3104 cells harboring mutant plasmids were grown in the minimal medium supplemented with 0.1% casamino acids and 0.2% glucose. At the middle of the logarithmic phase, tetA gene expression was induced with 0.25 g/ml heat-inactivated chlortetracycline (26) for 2 h. The cells were then harvested and washed with 50 mM MOPS-KOH buffer (pH 7.0) containing 0.1 M KCl, followed by suspension in the same buffer and adjustment of the absorbance at 530 nm to 85. The absorbance was determined by dilution of an aliquot. Five l of 100 mM AMS (final concentration 5 mM) or the same volume of distilled water was added to 100 l of the cell suspension, followed by incubation for 30 min at 30°C. Subsequently, 2 l of 25 mM [ 14 C]NEM (final concentration 0.5 mM) was added to the reaction mixture, followed by incubation for 5 min, and then the mixture was diluted with 900 l of the same buffer containing 5 mM unlabeled NEM. Immediately after the dilution, the cells were collected and washed once with the same buffer. Then the cells were disrupted by brief sonication. After removal of unbroken cells, the membrane fraction was collected by ultracentrifugation with a Beckman Ultracentrifuge Optima TL. The resultant precipitate was solubilized with 1% Triton X-100 and 0.1% SDS in phosphate-buffered saline containing unlabeled NEM, and then immunoprecipitated with anti-TetA C-terminal peptide antiserum (17) and Pansorbin Staphylococcus aureus cells (27), as described previously (28). Then the precipitate was subjected to SDS-polyacrylamide gel electrophoresis, followed by Coomassie Brilliant Blue staining. The resulting gel was soaked in Amplify (Amersham Corp.) prior to being dried. The dried gel was exposed to an imaging plate for visualization with a BAS-1000 Bio-Imaging Analyzer (Fuji Film Co., Tokyo, Japan).
In the case of labeling using everted membrane vesicles, the vesicles were prepared from E. coli W3104 cells harboring mutant plasmids after tetA gene induction as described previously (28). Then the vesicles (0.5 mg of protein) were used in place of intact cells as described above. Transport Assay-Tetracycline uptake by everted membrane vesicles was assayed as described in our previous paper (4) in the presence of 10 M [ 3 H]tetracycline and 50 M CoCl 2 .

[ 14 C]N-Ethylmaleimide Binding to the S65C and L97C Mutant TetA Proteins and Prevention by Preincubation with 4-Ac
we examined the binding of [ 14 C]NEM to the S65C and L97C mutants of Tn10-TetA in either intact cells or everted membrane vesicles after preincubation in the presence or absence of AMS. Ser-65 and Leu-97 are located in predicted cytoplasmic loop 2-3 and periplasmic loop 3-4, respectively. Ser-65 is not important for the transport function, but position 65 is a unique hot spot in loop 2-3 for inactivation by an SH reagent (28). Leu-97 is a hot spot in loop 3-4 similar to position 65. 2 It is certain that the S65C and L97C mutants retain the wild-type conformation of TetA, because they showed wild-type drug resistance. A binding experiment was performed as described under "Experimental Procedures." Since NEM is a membrane-permeant reagent, both the S65C and L97C mutants were equally labeled by [ 14 C]NEM in either intact cells or everted membrane vesicles in the absence of AMS (Fig. 1), indicating that these positions are exposed to the aqueous phase. Under these conditions containing 0.5 mM NEM, about 90% of the tetracycline transport activity was abolished (data not shown), indicating that the labeling reaction was almost saturated. With 1.0 mM NEM, the degree of inactivation no longer increased, whereas 0.2 mM NEM abolished only 65% activity. The degree of binding of NEM to the S65C mutant was not affected by preincubation with a membrane-impermeant reagent, AMS, in intact cells, whereas the binding of NEM in everted membrane vesicles was prevented by pretreatment with 5 mM AMS (Fig. 1), indicating that position 65 is located inside of intact cells and exposed to the outside medium in everted membrane vesicles. This confirms the cytoplasmic location of position 65. In contrast, the binding of [ 14 C]NEM to the L97C mutant was completely prevented by preincubation with 5 mM AMS for 30 min in intact cells (Fig. 1), indicating that position 97 is located on the outside surface of the cyto- FIG. 1. [ 14 C]NEM binding to the S65C and L97C mutant Tn10-TetA proteins and its prevention by preincubation with a membrane-impermeant sulfhydryl reagent, AMS, in intact cells and everted membrane vesicles. Suspensions (100 l) of intact cells or everted membrane vesicles expressing S65C or L97C mutant TetA were preincubated in the presence or absence of 5 mM AMS for 30 min at 30°C, followed by incubation with 0.5 mM [ 14 C]NEM for 5 min. The reaction was stopped by dilution with the same buffer containing excess unlabeled NEM. The cells were disrupted by brief sonication. The disrupted cells or everted vesicles were solubilized, and the TetA proteins were precipitated using anti-C-terminal peptide antiserum and Pansorbin S. aureus cells (28). The radioactive bands on SDS electrophoresis gels of these precipitates were visualized with a BAS-1000 Bio-Imaging Analyzer (Fuji Film Co., Tokyo, Japan). The symbols, ϩ and Ϫ, under the lanes indicate the presence or absence of AMS during the preincubation, respectively. plasmic membrane. With 2 mM AMS, the degree of protection decreased to about half, while various preincubation times, from 5 to 30 min, did not have a significant effect on the degree of protection. The binding of NEM to the L97C mutant was partially blocked in everted membrane vesicles (Fig. 1), probably due to contamination by right-side-out or unsealed membrane vesicles. These results clearly showed that an experiment on the binding of [ 14 C]NEM to a cysteine mutant and the protection by AMS is useful for determining whether a position is located on the cytoplasmic or periplasmic surface.
Construction and Activity of Cysteine Mutants-We constructed 15 different cysteine mutants to determine the sidedness of the hydrophilic loop regions of Tn10-TetA. In addition to positions 65 and 97, cysteine residues were introduced into positions 4, 36, 120, 156, 199, 201, 238, 270, 296, 328, 357, 360, and 365 using the mutagenic oligonucleotide primers listed in Table I. The locations of the mutations are depicted in Fig. 2. These positions covered all of the 11 putative hydrophilic loop regions and the N-terminal segment. The cytoplasmic location of the C-terminal is evident because the anti-C-terminal peptide antibody is only accessible from the inner side of the membrane (17). Mutagenesis was performed using pCT1183 as a template, and the sequences were determined by DNA sequencing, followed by transfer of the mutant tetA gene to a low copy number plasmid, pLGT2, by fragment exchange. No additional unexpected mutations occurred within the fragment transferred to pLGT2. The expression of the mutant TetA proteins was detected by 1) immunoblotting using anti-C-terminal peptide antiserum, and 2) Coomassie Brilliant Blue staining of gels after SDS electrophoresis of the immunoprecipitated TetA proteins. The degrees of expression of the TetA proteins were not significantly affected by these mutations (data not shown).
At first, the tetracycline resistance levels of E. coli W3104 cells harboring these mutant plasmids were measured on agar plates. As shown in Table II, 10 of the 15 mutants showed tetracycline resistance comparable with that of the wild type. S156C, S201C, and R238C showed moderate resistance, indicating that these mutations did not cause any significant alteration in the protein conformation. In contrast, the D120C and Y357C mutants had almost completely lost the drug resistance (Table II). As to the D120C mutant, it retained comparable tetracycline transport activity to that of the wild type when the activity was measured in everted membrane vesicles (Fig. 3). A similar discrepancy between transport activity and the resistance level was also observed for the D120N mutant (12), of which minimum inhibitory concentration was 19 g/ml but the transport activity was comparable with that in the wild type. This discrepancy may be based on alteration of the kinetic properties of the transporter (12). Thus, some of the conformational change must be caused by the Asp-120 mutation. However, as judged from the high level transport activity of the D120C mutant, the conformational change should not be large enough to destroy the structure of the transporter. The Y357C mutant also showed a similar discrepancy. The mutant retained about 10% of the transport activity (Fig. 3), indicating that it also retains the structure of the transporter, whereas the conformational damage in Y357C may be larger than that in the D120C mutant. So, an additional mutant, D365C, was constructed as to the same predicted loop.

Reactivity of Cysteine Residues Introduced into the N-terminal Half of Tn10-TetA with [ 14 C]NEM and Its Prevention by AMS-
The binding of [ 14 C]NEM to the cysteine residues introduced into the N-terminal half of Tn10-TetA was examined. As shown in Fig. 4, the reactivity of the S4C, S36C, S65C, L97C, and D120C mutants to [ 14 C]NEM was high in the absence of AMS, indicating that these positions are located in regions exposed to the aqueous phase. On the other hand, the reactivity  of the S156C mutant was somewhat lower than those of the other five mutants; however, binding was still detectable in the absence of AMS. After preincubation with excess AMS, the [ 14 C]NEM binding to the S36C, L97C, and S156C mutants was completely blocked, whereas the binding to the S4C, S65C, and D120C mutants was not affected at all by AMS preincubation (Fig. 4). These results were highly reproducible, and it should be noted that the protective effect of AMS was almost all or none under these conditions, as shown on quantitative measurement (Table III). These results clearly indicated that positions 36, 97, and 156 are located on the periplasmic side of the membrane, whereas positions 4, 65, and 120 are located on the cytoplasmic surface of the membrane. Positions 36 and 97 are located in predicted periplasmic loop 1-2 and loop 3-4, respectively. These results are the first direct biochemical evidence for the periplasmic location of these two loops. In particular, it is interesting that loop 3-4 is highly exposed to the aqueous medium, as judged from the reactivity of the L97C mutant to [ 14 C]NEM, although the hydrophilicity of this loop is not distinct and neither protease cleavage nor PhoA fusion data provide direct support for a periplasmic location (1,18).

Reactivity of Cysteine Residues Introduced into the C-terminal Half of Tn10-TetA with [ 14 C]NEM and Its Prevention by
AMS-At first, cysteines were introduced into positions 201, 238, 270, 296, 328, and 360. Positions 201, 270, and 328 are located in predicted cytoplasmic loops, whereas positions 238, 296, and 360 are located in predicted periplasmic loops (Fig. 2). The reactivity of [ 14 C]NEM to four of these six mutants was high (Fig. 5), confirming that these positions are located in loops exposed to the aqueous phase. However, the reactivity of [ 14 C]NEM to the S201C mutant was very low, and [ 14 C]NEM did not bind to the S360C mutant, indicating that these positions are hardly exposed to the aqueous phase. There are two alternative possibilities: 1) predicted loop 6 -7 and loop 11-12 containing positions 201 and 360, respectively, might not be exposed to the aqueous phase. If this is the case, the topology of Tn10-TetA should be greatly changed. 2) Positions 201 and 360 may be unique points in the water-exposed loops at which the side chains are oriented into the protein interior. Our previous work on cysteine-scanning mutants as to loop 2-3 (29) indicated the presence of such a unique cryptic point in this waterexposed loop. In order to determine whether the segments containing these positions are exposed or not, we introduced cysteine residues into several positions close to positions 201 and 360. The S199C and D365C mutants showed high reactivity to [ 14 C]NEM in the absence of AMS (Fig. 6 and Table III). The Y357C mutant showed low but detectable reactivity to NEM ( Fig. 6 and Table III). When the residues are located in the transmembrane region in TetA proteins, Cys mutants of these residues show no or very low binding (less than 0.1 mol of bound [ 14 C]NEM per mol of TetA) (30). Thus, it was confirmed that loop 6 -7 and loop 11-12 are exposed to the aqueous phase.
The [ 14 C]NEM binding to the R238C, S296C, Y357C, and D365C mutants was completely prevented by preincubation with excess AMS, whereas the binding to the S199C, T270C, and S328C mutants was not affected by AMS at all (Figs. 5 and 6 and Table III). Binding of [ 14 C]NEM to the S201C mutant was also observed after preincubation with AMS ( Fig. 5 and Table III). These results clearly indicated that large central loop 6 -7, loop 8 -9, and loop 10 -11 are located on the cytoplasmic surface, whereas loop 7-8, loop 9 -10, and loop 11-12 are located on the outside surface of the cytoplasmic membrane. It may be surprising that position 296 is highly exposed to the outside medium, as judged from the reactivity of the S296C mutant to [ 14 C]NEM (Fig. 5B), in spite of the low hydrophilicity of the segment around this position and the lack of direct biochemical evidence of the periplasmic location of this loop (1, 18).

FIG. 4. [ 14 C]NEM binding to the cysteine residues introduced into the predicted loop regions of the N-terminal half of Tn10-TetA and the prevention by preincubation of intact cells with AMS.
Intact cells expressing the cysteine mutants were preincubated in the presence or absence of AMS, followed by labeling with [ 14 C]NEM as described in Fig. 1. The radiolabeled TetA proteins were visualized with a BAS-1000 Bio-Imaging Analyzer after immunoprecipitation and SDS gel electrophoresis. ϩ and Ϫ indicate the presence and absence of AMS during the preincubation, respectively.

DISCUSSION
The membrane topologies of Tn10-TetA and its homolog, pBR322-TetA, have been predicted to include 12-transmembrane segments, and the N and C termini are located on the cytoplasmic side, as judged from the hydropathy profile (1,16). However, the predicted 12-transmembrane structure is not obvious because there are no distinct hydrophilic segments between predicted transmembrane segment (TM) 3 and TM4, and TM9 and TM10 (1). Hydropathy analysis of Tn10-TetA indicated the presence of 10 hydrophobic clusters of different sizes (1). Neither the presence of water-extruding loops nor the periplasmic location of segments between TM3 and TM4 and TM9 and TM10 has been directly supported by any biochemical data (1,18). The results presented in this article for the first time clearly confirmed the validity of the 12-transmembrane topology of the TetA protein.
We performed site-directed chemical labeling of cysteine mutants with sulfhydryl reagents to determine the membrane topology of Tn10-TetA. This method is applicable to Tn10-TetA because wild-type Tn10-TetA has only one cysteine residue at position 377, but maleimide derivatives are not reactive to this cysteine (20) due to it being buried in the hydrophobic interior of the membrane. The criteria used for determining the location of an introduced cysteine are as follows. 1) When cysteine residues are not labeled with [ 14 C]NEM, they are embedded in the hydrophobic interior of the membrane or the folded loops, because the active species of a sulfhydryl group for reaction with maleimide is the deprotonated form. 2) When the labeling of cysteine residues with [ 14 C]NEM is not blocked by membrane-impermeant AMS, the residues are located on the cytoplasmic surface. 3) When the labeling with [ 14 C]NEM is blocked by AMS, the residues are located on the periplasmic surface.
We constructed 15 site-directed mutants in which cysteine residues were introduced into predicted periplasmic or cytoplasmic loops. Except for the S201C and S360C mutants, all mutants were highly labeled by [ 14 C]NEM in the absence of AMS. Since cysteine residues introduced into positions nearby 201 and 360 also showed good reactivity to [ 14 C]NEM, positions 201 and 360 are likely to be local cryptic points in the waterextruding loop regions. The binding of [ 14 C]NEM to the cys-teine residues introduced into the predicted periplasmic loops was completely blocked by preincubation of intact cells with a membrane-impermeant SH reagent, AMS, whereas the binding to the cysteine residues introduced into the N-terminal segment and the predicted cytoplasmic loops was not affected by AMS, clearly indicating that the former cysteine residues are located on the periplasmic surface and the latter ones on the cytoplasmic surface. These results obviously support the 12-transmembrane structure of Tn10-TetA. Since the intervals between introduced cysteine residues are enough for the protein to traverse the membrane once but shorter than required to traverse it more than twice, the model of Tn10-TetA composed of more than 13 transmembrane segments is impossible.
The predicted periplasmic loops tend to be shorter than the cytoplasmic loops (Fig. 2). It has been suspected that TetA proteins are hardly exposed to the periplasmic surface of the membrane because no proteases can gain access from the periplasmic side (1). However, as judged from our results as to [ 14 C]NEM binding to the cysteines introduced into the periplasmic loops, they were certainly exposed to the aqueous phase, similar to the loops located on the cytoplasmic surface. The inaccessibility for proteases from the periplasmic side may be due to the protein conformation. FIG. 6. [ 14 C]NEM binding to the S199C, Y357C, and D365C mutants of Tn10-TetA and the prevention by preincubation of intact cells with AMS. Intact cells expressing these mutants were preincubated in the presence or absence of AMS, followed by labeling with [ 14 C]NEM as described in Fig. 1. The radiolabeled TetA proteins were visualized with a BAS-1000 Bio-Imaging Analyzer after immunoprecipitation and SDS gel electrophoresis. (ϩ) AMS and (Ϫ) AMS indicate the presence and absence of AMS during the preincubation, respectively.