Evidence for interactions between helices 5 and 8 and a role for the interdomain loop in tetracycline resistance mediated by hybrid Tet proteins.

An interdomain hybrid Tet protein consisting of a class C alpha domain and a class B beta domain (Tet(C/B)) lacks detectable efflux ability and provides only minimal levels of resistance to tetracycline (Tc) (3 microg/ml) compared with intact class B (256 microg/ml) and class C (64 microg/ml). Twenty-one independently isolated mutants of the Tet(C/B) protein with increased Tc resistance were generated by random chemical mutagenesis. Nine mutants with a Glu substitution for Gly-152 in helix 5 of the class C alpha domain produced a resistance of 48 microg/ml, whereas another 9 with an Asp replacement of Gly-247 in helix 8 of the class B beta domain mediated resistance at 32 microg/ml. The third type of mutation, found in 3 mutants expressing 24 microg/ml resistance, was a S202F replacement in the putative interdomain cytoplasmic loop of Tet(C/B). The latter underscores a previously unappreciated function of the interdomain cytoplasmic loop. All three types of Tet(C/B) mutant proteins were expressed in amounts comparable with that of the original protein and demonstrated restored energy-dependent efflux of tetracycline. Site-directed mutational analysis demonstrated that a Gly-247 to Asn mutation could also facilitate Tc resistance by the Tet(C/B) hybrid, and a negatively charged side chain at position 152 was required for Tet(C/B) activity. These mutations appear to promote the necessary functional interactions between the interclass domains that do not occur in the Tet(C/B) hybrid protein and suggest a direct association between helix 5 and helix 8 in the function of Tet efflux proteins.

Tetracycline (Tc) 1 resistance is mediated by many different but related determinants in Gram-negative bacteria, designated as classes A through E, G, H, and J (1). Each determinant encodes a cytoplasmic membrane protein, Tet, that catalyzes the active efflux of a tetracycline-divalent cation complex from the cell in exchange for a proton (2). Tet proteins are members of the major facilitator superfamily that includes transporters of various drugs, antibiotics, sugars, and the wellstudied lactose permease (3). Members of this group share a common topology as well as regions of amino acid sequence identity.
The Tet proteins of classes B and C are predicted to contain 12 transmembrane (TM) regions divided in half into two domains, ␣ and ␤, by a large putative cytoplasmic loop designated the interdomain region (4 -7). Extensive genetic and mutagenic analyses of the class B Tet protein has shown that the two domains, both of which are required for Tet function (8), contribute differently to the efflux of Tc. In the ␣ domain, amino acid residues in TM2 and TM3 line a putative substrate translocation pathway (9,10). The cytoplasmic loop connecting TM2 and TM3 could act as a gate that undergoes a conformational change during Tc transport (11,12). In the ␤ domain, His-257 in TM8 appears to be involved in proton translocation (13), whereas Asp-285 in TM9 is important for Tc binding (14).
Tc resistance proteins from classes A and C share 78% amino acid sequence identity, whereas the class B determinant is more distantly related to the other two, exhibiting 45% identity (15,16). A hybrid Tet protein consisting of a class A ␣ domain and class C ␤ domain (Tet(A/C)) specified levels of Tc resistance not unexpected for the combination of the two functional proteins (17). In contrast, a Tet(C/B) protein (class C ␣ domain and class B ␤ domain) provided only minimal resistance to Tc, and a Tet(B/C) protein (class B ␣ domain and class C ␤ domain) did not confer resistance (17). However, an easily detectable level of Tc resistance (10 -15% that of the level of wild-type Tet(C)) was observed when both the Tet(C/B) and the Tet(B/C) proteins were expressed together in the same cell. This result indicated that the individual domains of each hybrid protein were active and capable of interacting with the corresponding opposite domain of the same class (18). Presumably, differences between the amino acid sequences of the class C and class B prevented the domains of the Tet(C/B) hybrid from functioning.
In the present study, we have isolated and characterized active Tet(C/B) mutants. The locations of the specific mutations identify areas that probably mediate functional interactions between the ␣ and ␤ domains of Tet proteins and also implicate the interdomain region in the function of the Tet protein.

MATERIALS AND METHODS
Reagents-Restriction enzymes were obtained from New England Biolabs (Beverly, MA). Taq polymerase and T4 DNA ligase were purchased from Life Technologies, Inc. Antibiotics were obtained from Sigma, except that AHTc was prepared by Mark Nelson of this laboratory. [ 3 H]Tc was purchased from NEN Life Science Products. The remaining reagents were analytical grade.
Bacterial Strains and Plasmids-Escherichia coli DH5␣ cells (recA1; laboratory collection) were used for the propagation of plasmid DNA, determining Tc resistance, and for assessing protein expression. Tc efflux assays were performed with E. coli AG100A cells (⌬acrAB (19)). Cell cultures were routinely grown in LB media supplemented with chloramphenicol (20 g/ml) or Tc (10 g/ml), as needed. AHTc (0.025 g/ml) was used as a gratuitous inducer of Tet protein where applicable. * This work was supported by National Institutes of Health Grant GM55430. 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 (20). A XhoI/BglII fragment of pRAR1013, carrying the tet(C) gene (17), was used to replace the tet(C/B) hybrid gene in pRAR1033. The resulting construct, pCR2, was used as a source of Tet(C/C) (class C ␣ domain and class C ␤ domain) in this study.
Random Mutagenesis of Plasmid DNA and Isolation of Tc-resistant Mutants-Plasmid DNA was purified using a Qiaprep spin miniprep kit (Qiagen, Valencia, CA). 3 g of plasmid pRAR1033 DNA was incubated in 400 mM hydroxylamine, 50 mM potassium phosphate (pH 6.6), 100 mM EDTA in a 100-l volume for 1 h at 65°C followed by overnight incubation at 37°C. The entire volume was drop-dialyzed against sterile distilled water for 2 h at room temperature on a type VS filter (0.025 M pore size; Millipore, Bedford, MA). After ethanol precipitation and a 70% ethanol wash, the DNA was suspended in sterile distilled water.
E. coli DH5␣ cells were transformed with 50 ng of mutagenized DNA using the CaCl 2 heat shock method (21). Upon recovery, a total of 5 ϫ 10 5 transformants were plated on LB plates containing 20 g/ml chloramphenicol (for maintenance of the plasmid) and 10 g/ml Tc. Apparent Tc-resistant mutants were visible after 24 -48 h of incubation at 37°C. Only one colony per transformation was chosen for analysis to ensure that the mutants isolated were generated in separate mutagenic events. Tc resistance was verified by overnight growth of liquid LB cultures containing chloramphenicol and Tc. Plasmid DNA was isolated from Tc-resistant cultures and retransformed, and transformants were grown on Tc to verify that resistance was plasmid-mediated.
Specific mutations within the tet(C/B) gene were identified by DNA sequencing. The entire coding region of hybrid tet genes containing mutations in either domain was subcloned on a Xmn I/BglII fragment into the original unmutagenized plasmid (pRAR1033) to ensure that any Tc resistance observed was due solely to the identified mutation.
The mutations that produced Tc resistance by Tet(C/B) were introduced into the wild-type Tet protein class from which they were derived. The G247D mutation was introduced into the class B tet gene on pLR1068 by replacement of an EcoRI/PvuI fragment. A BamHI/SalI fragment of pRAR1033 encoding the G152E mutation was exchanged into the tet(C) gene on pCR2. Using the same restriction sites, the G152E mutation was also added to a tet(C/B) hybrid gene already containing the G247D mutation to create a double mutant. The exchanges were verified by sequencing.
Site-directed Mutagenesis-Site-directed mutagenesis of the (C/B) hybrid tet gene on plasmid pRAR1033 or the class C tet gene was performed by a PCR overlap method (22). Mutagenic PCR primers were designed to incorporate a restriction endonuclease site along with the desired mutation where possible ( Table I). The final PCR product potentially carrying the specific mutation was digested with PvuI and BamHI and exchanged into the unmutated parental tet gene. Plasmids were screened for the specific mutation by restriction enzyme analysis. The integrity of the mutated tet genes was verified by DNA sequencing.
DNA Sequencing-Sequencing of plasmid DNA was performed at the Tufts University Core Facility using a 373 stretch DNA sequencer (Applied Biosystems).
Tc Susceptibility-Cells harboring various wild-type and hybrid tet genes were grown in the presence of chloramphenicol and AHTc to an A 530 of 0.8. Cells were swabbed for confluent growth onto an LB plate containing AHTc. E-test strips (a gift of AB Biodisk, Solna, Sweden) containing Tc were applied to the inoculated plate. After overnight incubation at 37°C, growth inhibition around the test strip corresponding to the MIC of Tc was examined.
Membrane Isolation and Western Blot Analysis-DH5␣ cells expressing various Tet proteins were grown in the presence of AHTc, and membranes were prepared from them as described previously (23).
Before electrophoresis, membrane fractions were incubated in reducing sample buffer (21) for 1 h at 37°C. Proteins were separated on a 10% polyacrylamide gel, then electroblotted to an Immobilon-P membrane (Millipore) as per the manufacturer's directions. The blot was probed with a polyclonal antibody specific for the 14 carboxyl-terminal amino acid residues of Tet(B) (anti-Ct antibody, kindly provided by A. Yamaguchi (24)) followed by a secondary antibody conjugated to horseradish peroxidase (Phototope-HRP; New England Biolabs). Blots were developed with a Renaissance Western blot chemiluminescence kit (NEN Life Science Products).
Detection of Tet(C) Proteins-To detect Tet(C) proteins with the anti-Ct antibody, the 6 carboxyl-terminal amino acids were removed and replaced with the 14 terminal amino acids of Tet(B). This was achieved by inserting a HpaI site between base pairs 1169 and 1170 of tet(C) by PCR amplification. A BamHI/HpaI fragment encompassing all but the last 22 base pairs of the class B gene was then excised from plasmid pRAR1033 and replaced with the corresponding fragment of tet(C). The resulting construct encoded Tet(C), with the anti-Ct epitope replacing the carboxyl-terminal cytoplasmic tail. This procedure was also used to alter the Tet(C) S202F and G152E mutants.
Tc Efflux Assays-Assays were performed with E. coli AG100A cells as the chromosomal deletion in the acr locus eliminated host background Acr-mediated efflux of Tc. Energy-dependent Tc efflux was measured as the amount of

RESULTS
Generation of Tc-resistant Mutants-A total of 21 Tc-resistant mutants were independently isolated from separate mutagenic and transformation events. Sequence analysis of the hybrid tet gene from each mutant revealed three types of mutations, each consisting of a single base change (Table II).
A GGA 3 GAA transition within codon 152 resulted in a Gly to Glu mutation in nine different mutants (Table II). Gly-152 is located in putative helix 5 of the ␣ domain of Tet(C) in the Tet(C/B) hybrid (Fig. 1). The Tc MIC of cells expressing this protein was 48 g/ml, approximately 16-fold greater than that of the original Tet(C/B) protein (3 g/ml). This Gly residue (position 152 in Tet(C)) is conserved among all Tet proteins in Gram-negative bacteria.
In nine additional mutants, a GGC 3 GAC transition resulted in the substitution of Gly-247 of the class B ␤ domain in Tet(C/B) with an Asp residue (Table II). Based on the current

5Ј-CCGTGGCCGGGGCCCTGTTGGGCG-3Ј ApaI
5Ј-GCATGATGGTTAACTTTTCATTAG-3Ј HincII Interactions between ␣ and ␤ Domains in Tc Resistance Finally, three of the Tet(C/B) mutants isolated contained a UCC to UUC codon change that resulted in a Ser-202 3 Phe mutation. A host bearing this mutation exhibited a Tc MIC of 32 g/ml. Ser-202 is within the Tet(C) ␣ domain of the Tet(C/B) hybrid and resides in the interdomain cytoplasmic loop connecting the ␣ and ␤ domains (Fig. 1).
Each type of mutation was subsequently introduced into the Tet protein from which the corresponding region of the Tet(C/B) protein was derived. Introduction of the G152E mutation into Tet(C/C) (consisting of both an ␣ and a ␤ domain from the class C protein) or the G247D mutation into Tet(B/B) (comprised of class B ␣ and ␤ domains) resulted in inactive proteins, as demonstrated by a Tc MIC of 1.5 g/ml. A S202F mutation in Tet(C) also led to a severe reduction in Tc resistance to 4 g/ml from 64 g/ml mediated by the wild-type protein.
Expression of Parental Versus Mutant Hybrid Tet Proteins-Western blot analysis of the proteins was performed using the anti-Ct antibody directed against the 14 carboxyl-terminal amino acids of Tet(B). The ␤ domain of Tet(C) proteins was altered so that they would be detected by anti-Ct (see "Materials and Methods"). This replacement resulted in a small increase in the Tc MIC provided by these proteins (Table III).
The parental hybrid Tet(C/B) protein was abundantly present in the membranes of the cells in which it was expressed ( Fig. 2A, lane 3). The near background level of Tc susceptibility of Tet(C/B) has been attributed to inefficient functional interactions between ␣ and ␤ domains from the two different classes (17).
The G152E and S202F C/B proteins were detected in the membrane in quantities comparable with the original inactive Tet(C/B) hybrid ( Fig. 2A, lanes 3, 4, and 11). The G247D mutant was detected in slightly lower quantities in the membrane compared with the original Tet(C/B) and the other mutant proteins (Fig. 2A, lane 9). Although combining the G152E and G247D mutations in Tet(C/B) did not produce resistance to Tc, the protein was observed in higher quantities in the membrane than the original hybrid or any of the single mutants (Fig. 2B,  lane 5).
A strain expressing Tet(B) protein containing the G247D mutation was inactive, which could be attributed to a decreased amount of this mutant protein in the membrane compared with the wild type (Fig. 2B, lane 3). The similar lack of Tc resistance by the Tet(C) G152E could also be due, at least in part, to lower protein expression (Fig. 2C, lane 3). This protein also had a higher mobility in the gel relative to the wild type, which may reflect either a conformational change or the charge difference between the two proteins. It is unlikely, however, to be due to truncation of the G152E protein, as recognition by the anti-Ct antibody requires an intact carboxyl terminus. In addition, a Tet(C) protein containing the G152E mutation complements a deficiency in potassium transport (26), an activity that requires the amino terminus of Tet(C) to be associated with the membrane (27). The presence of the S202F mutation in Tet(C) (generated by site-directed mutagenesis) did not affect the amount of protein in the membrane (Fig. 2C, lane 4). In general, although the Tet(C) proteins analyzed by Western blot were engineered to express the anti-Ct epitope at their carboxyl termini, even the wild-type protein was not as reactive with the antibody as Tet(B) (compare Fig. 2C, lanes 2 and 5). In addition, although Tet(C) with the anti-Ct tag contained only two more amino acid residues, it had a noticeably lower mobility than Tet(B) (Fig. 2C).
Site-directed Mutants-It was of interest to determine whether the negatively charged side chain at either 152 or 247 was a requisite for Tc resistance by the hybrid protein. To this end, G152Q and G247N mutants were generated by site-directed mutagenesis. The Tet(C/B) G247N mutant provided a moderate level of Tc resistance (24 g/ml; Table III). This protein was observed in the membrane in a similar quantity to the wild-type Tet(C/B) hybrid ( Fig. 2A, lane 10) and more than the G247D mutant ( Fig. 2A, lane 9). Cells expressing the G152Q mutant, however, failed to show the Tet(C/B) protein in the membrane (Fig. 2A, lane 7) and were susceptible to Tc (Table III).
To further investigate the side chain requirements at position 152, additional replacements were constructed. Ala, Asp, and Lys were each substituted for Gly-152; only G152D resulted in a protein capable of mediating a significant level of Tc resistance (10 g/ml; Table III). Each mutant protein was detectable in the membrane ( Fig. 2A), although the G152K protein was present in relatively lower amounts.
These results indicated that an acidic residue was required at position 152 to produce active Tet(C/B). Furthermore, introduction of a neutral or positively charged side chain at position 152 decreased the stability of the Tet(C/B) protein. In contrast, replacing Gly-247 with either Asp or Asn was structurally tolerated and also rendered Tet(C/B) capable of mediating Tc resistance.
Tc Efflux Assays-Energy-dependent Tc efflux was measured as the relative uptake of [ 3 H]Tc before and after deenergization of the cells with the protonophore CCCP. As shown in

Interdomain region 24
a Determined via E-test. The original wild-type Tet(C/B) protein provided a Tc MIC of 3 g/ml.

DISCUSSION
Three different single amino acid changes out of a total of 21 independently isolated mutants restored Tc resistance to an otherwise inactive hybrid Tet(C/B) protein. Each mutant demonstrated a MIC for Tc above the 10 g/ml on which they were selected. Of the 5 ϫ 10 5 transformants screened, nine contained a G152E mutation in helix 5 of the class C ␣ domain, and nine more contained the G247D replacement in helix 8 of the class B ␤ domain. This finding suggests that critical interactions between the ␣ and the ␤ domains that are required for the function of Tet(C/B) occur between helices 5 and 8.
Both Gly-152 and Gly-247 are conserved among all classes of Gram-negative Tet proteins and are predicted to be relatively close to the periplasmic side of the membrane (Fig. 1). As conserved amino acid residues are usually of functional or structural significance, it was expected that substitutions leading to an active Tet(C/B) would have been at nonconserved positions. Substitution of Gly-152 with Glu in Tet(C) or Gly-247 with Asp or Asn in Tet(B) had a detrimental effect on Tc resistance, indicating the importance of Gly at these two positions in the wild-type proteins.
Gly-152 is the last residue of an amino acid sequence motif (GX 8 GX 3 GPX 2 GG) found among the antiporter members of the major facilitator superfamily (28). Our finding that a G152E mutation inactivates Tet(C) agrees with the findings of McNicholas et al. (26). Substitutions at other positions within this conserved motif in Tet(C), specifically Gly-147 (replaced with any other amino acid) (28) or Gly-143 (replaced with Asp or Asn) (26), also reduced or eliminated Tc resistance. Based on molecular modeling of helix 5 of Tet(C), Gly residues within this sequence were concluded to participate in the formation of a substrate binding pocket devoid of side chains (28). If this model holds true, then our result implies that within the Tet(C/B) hybrid, the putative Tc binding site is ineffective, and a G152E mutation suppresses this defect. The fact that only a negatively charged side chain at position 152 could render Tet(C/B) active suggests that an ionic interaction is formed with another region of the protein or that the side chain contributes in some manner to Tc binding. Residues within helix 5 are important for substrate recognition by other major facilitator superfamily members, particularly the multidrug transporter Bmr (29) and the lactose permease (30).
With respect to Gly-247, a helical wheel projection predicts that it is located on the same hydrophilic side of amphipathic helix 8 as Gln-261 and His-257. Gln-261 appears to contribute to substrate recognition by Tet(B) (31), and His-257 has been proposed to play an essential role in proton translocation (13). Introduction of the G247D mutation into Tet(B) apparently destabilized the protein, as this protein was present in reduced quantities in membrane vesicles relative to the wild type. In Tet(C/B), substitution of Gly-247 with a larger, hydrophilic (Asp or Asn) residue may also perturb the structure of the helix, at least enough to render the protein active without severely displacing the key amino acid side chains at positions 257 and 261. If Gly-247 does indeed face, or is proximal to, the Tc/proton transport pathway rather than the membrane, intro-duction of the hydrophilic side chains of Asp or Asn at that position would not necessarily be thermodynamically unfavorable.
Although the presence of either G152E or G247D is sufficient to activate Tet(C/B), the addition of the second mutation inactivates the protein without decreasing its stability (Fig. 2B, lane 5). Considering that the class C and class B proteins share a relatively low sequence similarity (45%) (15,16), it is surprising that a single amino acid substitution could render the previously inactive Tet(C/B) hybrid functional. In this sense, the individual mutations act as suppressors that compensate for the large differences in the amino acid sequences between the respective domains of the two classes, whereas the presence of the second mutation "desuppresses" the compensatory action of the first. This effect supports the hypothesis that a specific functional interaction exists between regions of helices 5 and 8. Alternatively, the mutations could reflect two entirely independent alterations, which individually promote functional interactions between ␣ and ␤ domains but together, do not. Our results are consistent with the general model for the arrangement of the TM regions of major facilitator superfamily members proposed by Goswitz and Brooker (32), in which helices 5 and 8 are predicted to be adjacent to one another and line the substrate translocation channel.
Not much attention has been given to the interdomain loop of Tet proteins owing to their poor amino acid sequence conservation. However, the random isolation of the active Tet(C/B) S202F mutant, without any additional mutations in the membrane-spanning regions, demonstrates that the interdomain loop is functionally important. Ser-202 is a nonconserved residue, found only in Tet classes C and H. The fact that the S202F replacement inactivates Tet(C) indicates that Ser-202 somehow contributes to Tc resistance. This finding confirms and extends other studies, which suggest that other residues within the interdomain region are important for Tet function (4,33,34).
Overall, our results strongly indicate that helices 5 and 8 interact and provide further evidence that the interdomain loop has a role in the function of Tet proteins. Proteins were quantitated using the NIH Image 1.6 program available on the Internet. Quantities of mutant Tet(B), Tet(C), and Tet(C/B) proteins are represented as a percentage of the respective wild-type proteins within the same blot. Wild-type proteins of different classes within the same blot were each assigned a value of 100%.
c The two letters indicate the class of Tet protein from which the respective ␣ and ␤ domains were derived. d Tc resistance levels for Tet(C) proteins reflect the MIC before the addition of the anti-Ct epitope to the carboxy terminus. The addition of the tag increased the Tc MIC for the wild-type to 76 g/ml, 6 g/ml for the S202F mutant, and 2 g/ml for the G152E mutant.