Second-site Mutation of Ala-220 to Glu or Asp Suppresses the Mutation of Asp-285 to Asn in the Transposon TnlO-encoded Metal-Tetracycline/H’ Antiporter of Escherichia coli*

A carboxyl group of Asp285 is essential for tetracycline/H+ antiport mediated by the transposon TnlO-encoded metal-tetracycline/H+ antiporter (TetA) of Escherichia coli (Yamaguchi, A., Ono, and T. (1992) J. Biol. Chem. 267,7490-7498). Spontaneous tetracycline resist- ance revertants were isolated from E. coli cells carrying the Asn-285 mutant tetA gene. All of the revertants were due to the second-site mutation at codon 220 of GCG (Ala) to GAG (Glu). The K,,, value of the tetracycline transport mediated by the revertant TetA protein was about 4-fold higher than that of the wild-type, indicating that the revertant is a low affinity mutant. A Glu-220 and Asn-286 double mutant constructed by site-directed mutagenesis showed the same properties as the revertants, confirming that the mutation of Ala-220 is solely responsible for the suppression. The Asp220 mutation of the Asn-285 mutant resulted in a lower level of restoration of the tetracycline resistance and the transport activity than in the case

the Asn-285 mutant tetA gene. All of the revertants were due to the second-site mutation at codon 220 of GCG (Ala) to GAG (Glu). The K,,, value of the tetracycline transport mediated by the revertant TetA protein was about 4-fold higher than that of the wild-type, indicating that the revertant is a low affinity mutant. A Glu-220 and Asn-286 double mutant constructed by site-directed mutagenesis showed the same properties as the revertants, confirming that the mutation of Ala-220 is solely responsible for the suppression. The Asp220 mutation of the Asn-285 mutant resulted in a lower level of restoration of the tetracycline resistance and the transport activity than in the case of the Glu-220 mutation. A single mutation replacing Ala-220 with Glu or Asp caused about a 2-4-fold decrease in the tetracycline resistance, but no crucial change in the transport activity.
It is not likely that Glu-220 is required for a chargeneutralizing salt bridge because an unpaired negative charge in a Glu-220 or Asp-220 single mutant did not cause a serious change in the activity. An alternative explanation is reasonable; Asp-285 directly contributes to the binding of a cationic substrate, metal-tetracycline chelation complex, or proton, and an acidic residue at position 220 can take over the role of Asp-285.
The transposon TnlO-encoded tetracycline resistance protein (TnlO-TetA) of Escherichia coli is a well-studied example of a proton substrate antiport system (1,2). The substrate exported by the antiporter is a monocationic metal-tetracycline chelation complex (3). According to the nucleotide sequence of the TnlO-tetA gene, the TnlO-TetA protein comprises 401 amino acid residues (4,5). Eckert and Beck (6) proposed a secondary structure model comprising 12 transmembrane segments with the amino and carboxyl termini on the cytoplasmic side. We modified the model slightly, in which the charged residues at the boundaries between hydrophobic transmembrane helices and hydrophilic loops are placed in the loop region, on the basis of the results of our site-directed mutagenesis studies on the charged residues (7,8). The general features of the model are * This work was supported by a grant-in-aid for scientific research on priority areas from the Ministry of Education of Japan and a grant-inaid from the Tokyo Biochemical Research Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
supported by the results of protease digestion (6) and antibody binding (9). The same model was also supported, for the pBR322-encoded TetA protein, by the results of analysis of a series of TetA-alkaline phosphatase fusions (10). A 12-membrane-spanning structure is common not only in drug export proteins , but also proton substrate symporters and uniporters (14). The resemblance between antiporters and symporters is not restricted to the secondary structure. They have a commonly conserved sequence motif, GXXXDRXGRR, and a derivative of it in the hydrophilic loop,-, and loop,-, regions, respectively (14,15). However, the distributions of functionally or structurally important residues in the hydrophobic transmembrane regions are quite different from each other. According to the results of site-directed mutagenesis studies on lactose permease (16,17), the important residues are located mainly in the carboxyl-terminal half of the protein, and a truncated mutant deleted from helices 2 to 6 showed downhill transport activity (18). In contrast, the functional residues in the transmembrane regions of the TetA protein show a symmetrical distribution between amino-and carboxyl-terminal halves (19,20). TetA proteins having deletions in their carboxyl-terminal halves show functional complementation with ones having deletions in their amino-terminal halves (211, indicating that both halves contribute to the transport function. The TnlO-TetA protein contains only 4 putative transmembrane charged residues, 3 aspartate and 1 histidine (8,22), and all these residues are conserved in at least class A, B, and C TetA proteins (23). Site-directed mutagenesis studies revealed that all of these charged residues are important for the tetracycline transport function (8,22, 241, the three carboxyl side chains being especially essential (8). It is characteristic in TetA proteins that the number of transmembrane negatively charged residues is higher than the number of positively charged residues. This feature is very different from the distribution of transmembrane charged residues in lactose permease. In the putative secondary structure model of lactose permease modified by King et a l . (25), the transmembrane regions contain the same numbers of negatively and positively charged residues, most of them forming charge-neutralizing pairs (25,26). Such charge-neutralization is necessary for protein folding in a hydrophobic environment (27). The pairings of the charged residues were found on analysis of second-site suppressor mutants (25,26). Namely, replacement of either charged residue with a neutral residue creates an unpaired charge causing a functional defect, while additional replacement of the unpaired residue with a neutral residue causes functional restoration. In the TetA protein, Asp-285 is located sterically close to His-257 in the putative structure (22); however, the occurrence of charge-pair neutralization between Asp-285 and His-257 is not likely because the replacement of these residues with neutral residues caused different results, that is, the replacement of Asp-285 with a neutral residue, Asn, resulted in the complete loss of the transport activity (8), while on the replacement of His-257 with a neutral residue, Tyr, significant activity remained (20). In addition, there is no other positively charged candidate for pairing with Asp-15, Asp-84, or Asp-285 in the transmembrane region.
To examine the possibility of charge-pair neutralization in the TetA protein, we isolated second-site suppressor mutants of the Asn-285 mutant TetA protein. Unexpectedly, the revertants showed no removal of the positively charged residue. All of them showed the introduction of a new negatively charged residue at a position distant from Asn-285 in the primary sequence. The results indicated that charged residues in the transmembrane region are not always necessary to make charge-neutralizing pairs. It is likely that the carboxyl group of Asp-285 is protonated or neutralized on the binding of a cationic substrate, or a negatively-charged group of Asp-285 may be located in a hydrated transmembrane pathway involved in substrate translocation.
Biochemicals. An oligonucleotide-directed in vitro mutagenesis system was purchased from Amersham. Lambda-Lift Expression Detection Kit for immunoblotting was purchased from Bio-Rad. All other materials were of reagent grade and obtained from commercial sources.
Bacterial Strains and Plasmids-E. coli MV1184 (281, TGl(291, and W3104 (30) were used for single-strand DNA preparation, transformation, and expression of the mutant plasmids, respectively. pUC118 was purchased from TAKARA (Kyoto, Japan). pLGT2 was constructed by cloning of the tetR and tetA genes into a low-copy-number plasmid, pLG339 (31), as described previously (8). pLGD285N is a derivative of pLGT2, in which the Asp-285 codon was changed to an Asn one by site-directed mutagenesis, as described previously (8).
Isolation of Spontaneous Revertants-E. coli W3104lpLGD285N was grown overnight a t 37 "C in 5 ml of 2 x Y T medium 11.6% Bactotryptone, 1% yeast extract, and 0.5% NaCl) containing 50 pg/ml kanamycin. The cells were harvested from 1.5 ml of the medium and then resuspended in 0.1 ml of fresh 2 x YT medium. The concentrated cell suspension was poured on a YT-agar (0.8% Bacto-tryptone, 0.5% yeast extract, 0.5% NaCl, and 1.5% agar) plate containing 50 p g / d tetracycline and then incubated at 37 "C. After a 48-h incubation, colonies were isolated.
Site-directed Mutagenesis-At first, the EcoRI-BamHI fragment (868 bp)' of the D285N mutant tetA gene was subcloned into the EcoRI-BamHI site of pUC118. The fragment contains the 3'-terminal half of the tetA structure gene, which includes codons 220 and 285. The resulting plasmid was named pTBD285N. Site-directed mutagenesis of codon 220 was performed using the mutagenic primers listed in Table I1 and pTBD285N as a template with an oligonucleotide-directed in vitro mutagenesis system (version 2.1, Amersham). The mutation was first detected as the appearance of a new PuuII restriction site and then confirmed by DNA sequencing. The resulting plasmids were pTBA220ED285N and pTBA220DD285N. The double mutant tetA genes a t positions 220 and 285 were reconstmded by replacing the EcoRI-BamHI region of the wild-type pLGT2 plasmid with the corresponding fragment of pTBA220ED285N or pTBA220DD285N. The resulting plasmids were pLGA220ED285N and pLGA220DD285N. respectively. A single mutant at position 220 was constructed by exchange of the EcoRI-BgZI fragment (187 bp), which includes codon 220 but not codon 285, of pLGT2 with the corresponding fragment of pTBA220ED285N or pTBA220DD285N. The resulting plasmids were pLGA220E and pLGA220D, respectively.
Measurement of Bacterial Resistance to Zktracycline-Bacterial resistance to tetracycline was measured by the agar dilution method (32) and expressed as the minimum inhibitory concentration.
Preparation of Inverted Membrane Vesicles-Cells were grown in 1 liter of the minimal medium supplemented with 0.2% glucose and 0.1% casamino acids. At the middle of the logarithmic phase, tetA gene expression was induced for 2 h by incubation with 0.25 pg/ml heat-inactivated chlortetracycline. Inverted vesicles were prepared by disruption lino)propanesulfonic acid. Immunoblot Analysis-SDS-polyacrylamide gel electrophoresis of the inverted vesicles was followed by electroblotting of the proteins. The TetA protein was detected by means of an enzyme-linked immunosorbent assay using an anti-carboxyl-terminal peptide antibody and an Express blot assay kit (Bio-Rad), as described in the previous paper (9).
Tetracycline Bansport Assaying of Inverted Vesicles-Amixture of 10 p1 of the vesicle suspension (3.5 mg of protein/ml) and 0.5 pl of 250 m~ NADH was preincubated a t 30 "C for 1 min. The tetracycline uptake was initiated by the addition of 40 pl of MOPS-KOH buffer (pH 7.0) containing 0.1 M KCl, CoCl, (final concentration, 50 p d , and PHItetracycline (final concentration, 10 p d , unless otherwise stated. After incubation a t 30 "C for the indicated periods, 2 ml of 5 m~ MOPS-KOH (pH 7.0) containing 0.15 M LiCl was added, the mixture was immediately filtered through a Millipore filter (pore size, 0.45 pm) and washed twice, and then the radioactivity of the filter was measured.

RESULTS
Isolation of Second-site Revertants from E. coli Strains Carrying the D285N Mutant Plasmid-When concentrated cell suspension ofE. coli W3104/pLGD285N was spread on nutrient agar plates and incubated at 37 "C for 48 h, 10 colonies were formed. Plasmids were prepared from 6 of these 10 revertant strains and transferred into E. coli W3104. Table I shows the minimum inhibitory concentrations of tetracycline when E. coli W3104 cells carrying these plasmids were grown on agar plates. The tetracycline resistance level (0.8 pg/ml) of the cells carrying pLGD285N was the same as that of the host cells. In contrast, the cells carrying the revertant plasmids showed the same resistance level (200 pg/ml) as the cells carrying the wildtype pLGT2 plasmid.
The nucleotide sequence, 913GATAGT918 (the numbers indicate the base positions from the HincII site located at 60 bp upstream from the start codon of tetA), of the wild-type tetA gene is changed to 913AACTCGs1s in the D285N mutant tetA gene. The former encodes Asp-285-Ser-286 and the latter Asn-285-Ser-286. The latter sequence contains a n additional XhoI site (CTCGAG). All of the 6 revertant plasmids (pLGD285NB-1 to pLGD285NB-6) retained this XhoI site at the same position as in pLGD285N. Then, the nucleotide sequence of about 150 bases around codon 285 was determined. All of the 6 revertant plasmids showed the same sequence as pLGD285N in the region from 850 to 900. The sequence from 913 to 918 of the revertant genes was confirmed to be AACTCG. Thus, the restoration of the tetracycline resistance was not due to the backmutation from Asn-285 to Asp. Determination of the Position of Second-site Mutation-In order to limit the range of the location of the second-site mutation, the revertant tetA gene of pLGD285NB-1 was cut at the middle with EcoRI and at 262 bp downstream from the stop codon with BamHI. The resultant EcoRI-BamHI fragment (868 The EcoRI-BamHI fragments of pLGT2, pLGD285N, and pLGD285NB-1 were individually subcloned into the EcoRI-BamHI site of pUC118 in order to determine their entire sequences. The sequence around codon 285 was GCAGATAGT (Ala-Asp-Ser) in the wild-type tetA, and GCAAACTCG (Ala-Asn-Ser) in D285N and the revertant. The only difference between the nucleotide sequences of D285N and the revertant was at nucleotide position 719 (Fig. 1). Namely, C719 in D285N and the wild-type was changed to Ain the revertant. This single mutation results in a codon change from GCG (encoding Ala-220) to GAG (encoding Glu). The results showed that the removal of a carboxyl side chain at position 285 was compensated for by the introduction of a new carboxyl side chain at position 220. As shown in Fig. 2, Ala-220 is located in putative transmembrane helix 7, which is distant from Asp-285 in helix 9 in the secondary structure. The results indicated that the two positions may be sterically close to each other, although a n alternative possibility such as a n independent new pathway opened by Glu-220 cannot be ruled out.
The sequences around codon 220 of the five other revertant plasmids were also determined. In all of them, Ala-220 was changed to Glu. Therefore, position 220 is a unique position compensating for the mutation at position 285.
Site-directed Mutagenesis of Ala-22O"In order to determine whether or not the mutation of Ala-220 is the only cause of suppression of the D285N mutation, site-directed mutagenesis of Ala-220 to Glu or Asp was performed using the mutagenic primers shown in Table 11. As shown in Table 111, the A220E,D285N (Glu-220-Asn-285) double mutant strain showed as high tetracycline resistance as the wild-type and revertant strains (minimum inhibitory concentration, 200 pg/ml). Therefore, it is clear that the change from Ala-220 to Glu is solely responsible for suppression of the D285N mutation.
The A220DD285N (Asp-220-Asn-285) double mutant strain showed a lower resistance level (50 pg/ml) than the A220E/ D285N mutant. A longer spacer of the carboxyl side chain a t position 220 seems more favorable for suppression of the D285N mutation than a shorter one. This is in contrast with the results as to position 285. Asp-285 is far more favorable than Glu-285, as described previously (8).
A single mutation of Ala-220 to Glu or Asp in the wild-type TetA caused a decrease in the resistance level (Table 1111, but the mutant strains retained significant tetracycline resistance. Thus, the introduction of an extra negative charge at position 220 did not cause crucial damage t o the TetA function. The resistance level of the A220E mutant (100 pg/ml) was some-657 what higher than that of the A220D mutant (50 pg/ml). This is very different from the result of the removal of a negative charge from position 285. The latter causes the complete loss of the function (8). If the second-site suppression is based on the charge neutralization of an unpaired positive charge in the D285N mutant, the insertion of an extra negative charge at position 220 in the wild-type TetA should cause crucial damage in the function through the generation of a n unpaired charge, similar to that in the case of the negative-charge removal from position 285. This is clearly not the case.
Determination of TetA Protein Production and Kinetics of Tetracycline Dunsport in Inverted Membrane Vesicles-SDSpolyacrylamide gel electrophoresis of the inverted vesicles (10 pg of membrane protein) was performed, followed by electroblotting onto a nitrocellulose filter. TetA proteins were detected by enzyme-linked immunosorbent assay using anti-carboxylterminal specific antiserum and alkaline phosphatase-linked goat anti-rabbit IgG. The amounts of D285N, the revertant (B-l), and the A220ED285N, A220D/D285N, A220E, and A220D mutant TetA proteins were the same as that of the wild-type (Fig. 3). Thus, the differences in the tetracycline resistance between the wild-type and these mutant strains are not due to differences in the amount of the TetA protein.
Tetracycline uptake by inverted membrane vesicles was measured in the presence of 10 p~ [3H]tetracycline and 50 VM CoC12 (Fig. 4, A and B ) . The initial rate of tetracycline uptake by wild-type vesicles was 2.1 nmoVmg of proteid30 s. D285N mutant vesicles showed no tetracycline transport; as reported in the previous paper (8). Single mutants A220E and A220D showed rates of about 4.0 (190% as to wild-type) and 2.1 (loo%), respectively. On the other hand, the double mutants showed unexpectedly low tetracycline transport activity. The initial rates of the revertant (B-l), A220E/D285N, and A220D/ D285N mutant vesicles were about 0.9,0.9, and 0.2 nmoVmg of proteid30 s, respectively, which correspond to only 43%, 43%, and 10% as to the wild-type, respectively.
On the other hand, the revertant (B-1) vesicles showed high tetracycline transport activity, comparable to that of the wildtype, in the presence of 100 p~ [3H]tetracycline (Fig. 4C). It is likely that the revertant and the double mutants may be low affinity mutants. Therefore, the kinetics of tetracycline transport in inverted membrane vesicles were measured in the presence of a high CoC12 concentration (1 mM). As shown in Table  IV, the K,,, values for tetracycline transport by the revertant and the A220ED285N mutant were 158 p~ and 145 PM, respectively, which are the same within experimental deviation. These K , values were about 4.3-fold higher than the value in the case of the wild-type (36 p~), indicating that the A220E/ D285N mutant has significantly lower substrate affinity than the wild-type. Similarly, the K,,, value of the A220DD285N mutant was also higher than that of the wild-type. The K, values of the A220E/D285N and A220DD285N mutants were very close to each other. This suggests that the decrease in the substrate affinity is mainly due to the change of the location of an essential negative charge from position 285 to 220. On the other hand, the V , , values of the revertant and the A220El   D285N mutant are just the same, being about 2-fold higher than that of the wild-type. On the other hand, the V, , , value of the A220DD285N mutant was about one-half that of the wildtype. Therefore, the length of the spacer a t position 220 affects the turnover rate of the transporter rather than the substrate binding affinity. The introduction of an extra negative charge, a t position 220, into the wild-type TetA did not cause crucial change to the protein function. Instead, the K,,, values are decreased to about 2-fold that of the wild-type by this extra negative charge. The results may indicate that dual negative charges in the active site probably strengthen the binding of the cationic substrate. The K,,, values of the A220E and A220D mutants were similar to each other. In contrast, the V, , , value of the A220D mutant was lower than that of the A220E mutant. The latter was about the same as that of the wild-type. That is, the length of the side chain a t position 220 has no effect on the affinity, but it affects the turnover rate. A longer chain is better than a shorter one. This result is the same as that in the case of the Asp-285-Ala-220 double mutants.

The kinetic constants for tetracycline uptake by inverted membrane vesicles
The rates of [3H]tetracycline uptake for the initial 30 s were measured in the presence of 1 l l l~ CoC12. The net rate of the active uptake was calculated by subtracting the rate in the absence of NADH from the rate in the presence of 2.5 II~M NADH. Kinetic constants were calculated from Lineweaver-Burk plots.  (34). transmembrane charged residues other than the 2 pairs described above exhibited no such charge-neutralizing suppression. These chargeneutralizing salt bridges contribute to the protein folding (27) andor may act as a gate openedclosed through protonation/ deprotonation (26).
In the putative secondary structure of the TnlO-TetA protein (81, there are 3 aspartates and 1 histidine in the deep transmembrane locus. The numbers of transmembrane negative and positive charges are not equal, but a modification like that made by King et al. (25) is not possible in the central loop region of TnlO-TetA. The current topology of TetA has a large central loop common to the other 12-membrane-spanning transporters (14) and the modified structure of lac permease according to King et al. (25). In the present study, a charge-neutralizingtype revertant was not obtained. It is certain that the presence of a n unpaired charge in the hydrophobic environment requires high free energy (25). However, the location of unpaired charges in the hydrated transmembrane channel does not require high free energy. Thus, charge-pair neutralization is not essential in such a hydrated channel. The transmembrane functional residues of TnlO-TetA including Asp-285 are distributed on the hydrophilic side of each amphiphilic transmembrane helix (20). This indicates that these helices form a hydrated transmembrane channel.
The predominance of transmembrane negative charges in the TetA protein may be related to the positive charge of the transported substrate, the metal-tetracycline chelation complex (3). Unpaired negative charges in the transmembrane locus of the TetA protein probably contribute to the binding of a positively charged substrate in the hydrated substrate-translocation channel. This prediction is supported by the fact that an extra negative charge created by a single mutation of Ala-220 increases the substrate binding affinity.
There is significant symmetry in the distribution of functional residues of TetA (19,201. Two sets of quartets of residues symmetrically conserved in the amino-and carboxyl-terminal transmembrane loci are important for the transport function (20). Symmetrically conserved sequence motifs in the hydrophilic loop region also play a functional role (19). The origin of the symmetry may be a tandem duplication of the tetA gene (36). As to the transmembrane negative charges, Asp-285 is symmetrical to Asp-84. However, Asp-15 is an exception; there is no corresponding acidic residue in the carboxyl-terminal half. The position symmetrical t o h p -1 5 is occupied by Ala-220. This is the position suppressing the mutation of Asp-285. It seems not incidental that a 1-base-change of codon 220 gives a Glu codon. Position 220 might have been originally occupied by Glu in the ancestral TetA protein.
The revertant reported in this study is based on charge compensation. A similar type of revertant was recently reported in a bacterial photosynthetic reaction center of Rhodobacter sphaeroides (35). The removal ofhp-213 in the L-subunit was suppressed by the mutation of Asn-44 to Asp in the "subunit (35). These Asp residues are predicted to act as a proton donor in the proton transfer reaction. Previously, we pointed out the possibility that Asp-285 of TetA plays dual roles in substrate binding and proton transfer (8). According to the fact that the double mutants of TetA showed low substrate affinity but high Second-site Revertants of the Asp-285 + Asn Mutant Tetracycline Carrier 26995 -V,,, values, a n acidic residue at position 220 may be fully able "*, ,-Io to act as a proton donor/acceptor but less effective as a substrate binding group.