An N-terminal domain of the tetracycline resistance protein increases susceptibility to aminoglycosides and complements potassium uptake defects in Escherichia coli

Expression of extrachromosomal tet genes increased the susceptibility of gram-negative bacteria to specific aminoglycoside antibiotics. The magnitude of the increase in susceptibility was dependent on the amount and the class of the tet gene product (designated Tet) and the bacterial species in which the tet gene was expressed. Truncated Tet proteins that contained more than the first 33, but not more than the first 97, N-terminal amino acids of Tet also increased the susceptibility to aminoglycosides and complemented the potassium uptake defects in Escherichia coli. The primary structure of this N-terminal Tet fragment has the hydropathic characteristics of a multimeric, transmembrane structure and is highly conserved in three different classes of Tet proteins.

sion also increases the susceptibility of gram-negative bacteria to specific aminoglycoside antibiotics. Defining the physiological or structural differences between tetracyclineresistant (Tcr) and tetracycline-sensitive (Tcs) bacteria that are manifested as a result of these pleiotropic effects could provide new bases for investigating the molecular mechanism of tetracycline efflux and for identifying improved antimicrobial agents.
The physiological basis of the pleiotropic effects of Tet is not known. However, it is reasonable to assume that it is a direct consequence of the structural features of the Tet protein. To study this phenomenon, we have examined the structure-activity relationships of specific fragments of the Tet protein. The results of these studies indicate that truncated Tet proteins containing as little as the first 97 Nterminal amino acids (Tet97), although incapable of conferring Tcr, increased the susceptibility to aminoglycosides and complemented potassium uptake defects in Escherichia coli.
Tet97 has the characteristics of a multimeric transmembrane protein, and its amino acid sequence may be related to those of other membrane proteins, altered forms of which also affect susceptibility to aminoglycosides. MATERIALS  were derived from Tcs strains by calcium chloride-mediated plasmid transformation (21) or transposition after infection with defective lambda bacteriophage (KN561) carrying the transposon TnlO (9).
Construction of plasmids. Plasmid pCC42, a derivative of pBR322 with an inducible tet gene, was constructed by ligating the 1,844-nucleotide PvuII-BamHI fragment of pSC101 that contains the tetR transcription repressor gene (1) to the 3,988-nucleotide EcoRI-BamHI fragment of pBR322 that contains the tet and bla structural genes and the ColEl origin of plasmid DNA replication (32). The regulation of the pCC42 tet structural gene by the tetR repressor was verified by comparing the growth rates of HB101, HB101(pBR322), and HB101(pCC42) in L broth containing 20 ,ug of tetracycline per ml, a concentration that is inhibitory for Tcs HB101, with and without prior derepression (induction) of the tetR repressor with 0.5 ,ug of acid-inactivated tetracycline per ml. Acid-inactivated tetracycline was prepared by refluxing tetracycline in 0.2 N H2S04 for 20 h.
The pH was then adjusted to 6.0 with NaOH, and contaminating 5a,6-anhydrotetracycline was removed by precipitation in the cold. Chromatographic analysis on octadecylsilane columns with 30% methanol-70% 1 mM EDTA indicated that the final product contained less than 1.0% unreacted tetracycline. Accordingly, acid-inactivated tetracycline prepared in this manner had no antibacterial activity at 150 ,ug/ml against E. coli. Acid-inactivated tetracycline was approximately 10-fold more effective than tetracycline at inducing tet gene expression (data not shown).
Plasmid pCC100 was recovered from mutagenized HB101(pBR322) by selecting for Tcr at 75 p.g/ml. Preliminary data indicate that the segment of the pCC100 plasmid that is responsible for the increased Tcr is not located within the tet gene itself, but lies within the PstI-PvuII fragment that contains the origin of replication. This suggests that the mutation may affect the plasmid copy number.
Mutant pBR322 plasmids lacking specific portions of the tet structural gene were constructed by deleting DNA between the pairs of restriction sites shown in Fig. 2. Briefly, plasmid DNA was digested with the specified pairs of restriction enzymes; if necessary, the staggered ends were nucleotide. The location of each pACYC184 tet point mutation was mapped as described in Materials and Methods. The segment of the pACYC184 Tet amino acid sequence inferred to contain the mutation is indicated. For pBR322 tet mutations, A indicates that DNA has been deleted between the specified restriction sites and + indicates that DNA has been inserted into the specified restriction site. The LD90s of tetracycline for all Tcr and Tcs strains were greater than 60 ,ug/ml and less than 2.5 ,ug/ml, respectively. made flush with Si nuclease, and the DNA was then recircularized with T4 DNA ligase (22). The Tet amino acid sequence encoded by the DNA downstream of the pBR322 BamHI site was also changed by two additional methods. First, plasmid pBR322 DNA. was linearized with BamHI, the staggered ends were made flush with S1 nuclease, and the DNA was then recircularized with T4 DNA ligase (22). Second, random human Sau3A DNA fragments were ligated into the BamHI site (see Fig. 2). All modifications completely abolished Tcr (Table 1). Point mutations in the tet gene were derived and mapped as follows. Plasmid pACYC184 DNA was mutagenized in vitro with hydroxylamine (9) and introduced into calcium chloride-treated E. coli HB101 (21). Transformants harboring mutant tet genes were selected by resistance to chloramphenicol and screening for Tc'. The location of each tet point mutation was then determined by deletion mapping. Briefly, each mutant was introduced into a series of E. coli strains that contained pBR322 plasmids from which the promoter and defined amounts of downstream DNA had been deleted. The presence or absence of Tcr recombinants arising from recombination between the tet mutation on plasmid pACYC184 and each deletion mutation on plasmid pBR322 was then determined.
Measurement of antibiotic resistance. The antibiotic concentration that reduced bacterial plating efficiency by 90% (the LD9J) was determined as follows. Overnight cultures were diluted with fresh L broth (with or without 0.5 jig of acid-inactivated tetracycline per ml) to an A6. of approximately 0.05 and incubated at 370C with vigorous agitation until the A6. reached 0.5. The log-phase cultures were then diluted serially with L broth, and Tcs or Tcr cells (2 x 103 to 5 X 10) were spread on paired L-broth agar plates containing the specified concentrations of the indicated antibiotics (Sigma Chemical Co.) with and without 0.5 jig of acidinactivated tetracycline per ml. In most experiments, susceptibility was measured at increments of antibiotic concentration approximately equal to 10% of the LD90 of the Tcs strain established in preliminary studies. Plates were incubated at 37°C for 18 to 24 h, after which the number of colonies was determined.
Complementation of potassium uptake defects. Plasmids were introduced into calcium chloride-treated E. coli TK2205 cells as described previously (21). Each strain was  grown overnight in minimal medium containing 0.2% glucose and 115 mM potassium ion at 37°C (10). Sufficient cells to produce an A6. in 50 ml were concentrated by centrifugation, and the medium was decanted. The cells were then suspended in 50 ml of prewarmed minimal medium containing 0.2% glucose and either 0, 2.5, or 115 mM potassium ion (Ko, K2.5, and K115, respectively) and shaken for 60 to 90 min at 370C to reequilibrate the cellular potassium pool. The doubling times of the cultures were then determined at 37°C from the subsequent increase in the A6.

RESULTS
Increased susceptibility of Tcr bacteria to aminoglycosides. The presence of plasmid pBR322 in each of three E. coli strains and a Salmonella typhimurium strain significantly increased the susceptibilities of these strains to kanamycin ( Table 2). The mean LDg's of kanamycin for the three Tcr E. coli strains varied between 52 and 68% of that for the isogenic Tcs controls. Chromosomal mutations conferring RecAand streptomycin-resistant (Smr) phenotypes did not significantly affect the plasmid-dependent increase in Kms. The average effect of plasmid on Kms was nearly twofold higher in the S. typhimurium strain than in the E. coli strains. Expression of the pBR322 tet gene also reduced comparably the LD90s of three closely related aminoglycosides: gentamicin, amikacin, and tobramycin. However, no measurable effect 'was noted on the susceptibility to several other inhibitors; including another aminoglycoside, kasugamycin, and the aminocyclitol spectinomycin (data not shown). tet expression required for increased anminoglycoside susceptibility. The plasmid-dependent infcrease in Kms was not suppressed by deletion of DNA containing the plasmid pBR322 bla gene promoter (plasmid pCC99), but was completely suppressed by deletion of DNA containing the pBR322 tet gene promoter (plasmid pCC47) ( Table 3). To confirm that the increased susceptibility to aminoglycosides resulted from the expression of the tet gene, we measured Kms in strains in which tet expression is inducible. Plasmid pCC42 expressed the tet gene only when derepressed with tetracycline or a tetracycline analog such as acid-inactivated tetracycline (see Materials and Methods), but was otherwise identical to pBR322 which expresses the tet gene constitutively (22). The effect of inducers of tet gene expression on Kms was compared in HB101, HB1O1(pBR322), and HB1O1(pCC42). The LDgos of kanamycin for HB101 and HB1O1(pCC42) were identical when assayed without acidinactivated tetracyline (Table 3). However, in the presence of acid-inactivated tetracycline, the LD90 of kanamycin for HB101(pCC42) fell to a value that was indistinguishable from that of kanamycin for HB101(pBR322). Acid-inactivated tetracycline did not significantly affect the Kms of either HB101 or HB1Ol(pBR322) ( Table 3).
The effect of tet expression on Kms was also reproduced with two unrelated plasmids that contain the same class C tet gene as pBR322. The tet gene of plasmid pACYC184, like that of pBR322, is expressed constitutively (6). HB101 (pACYC184) was more Kms than plasmid-free HB101 was in the absence of acid-inactivated tetracycline ( Table 3). The tet gene of plasmid pSC101, like that of pCC42, is normally repressed (7). Accordingly, HB1O1(pSC101) and plasmidfree HB101 were equally Kms in the absence of acidinactivated tetracycline, but HB101(pSC101) became more Kms in its presence (Table 3). Together, these results indicate that the increased susceptibility to aminoglycosides requires tet gene expressiofi.
Correlation of increased susceptibility to aninoglycosides with the level of tet gene expression. Although plasmids pSC101, pACYC184, and pBR322 contain the same tet gene (3, 6, 7), they differ in copy number (22), resulting in different tet gene dosages. pSC101 has a lower copy number than either pACYC184 or pBR322. The copy number of the mutant pBR322 plasmid pCC100 has not been determined. However, the fact that the mutation is not in the tet gene is consistent with the possibility that the mutation increases plasmid copy number (see Materials and Methods). HB101(pBR322) and HB1O1(pACYC184) both expressed greater Tcr and greater Kms than acid-inactivated tetracycline-induced HB1O1(pSC101) ( Table 3). HBlO1(pCC100) expressed greater Tcr and greater Kms than HB101 (pBR322), HB1O1(pACYC184), or acid-inactivated tetracycline-induced HB1O1(pSC101). Linear regression analysis indicated a highly significant correlation between the level of  The data in Table 3 were used to calculate the fractional reduction in the mean LD90 of kanamycin that results from tet expression in each strain (LD90 Km Tcr/LD90 Km Tc1). These values were then plotted versus the mean LD90 of tetracycline for the strain. The data for strains containing class C tet determinants (0) were fitted to a linear plot by the least-squares equation. The linear regression coefficient (-0.952) indicates that there is a very strong linear correlation between these two parameters. The data for the strains containing the class A, B, and D tet genes (0) are shown for comparison.
Tcr and the magnitude of the increase in Km' in E. coli and S. typhimurium strains carrying class C tet genes (Fig. 1). Increased KmS conferred by all classes of tet genes. The finding that the class C tet gene conferred increased Kms is of greater significance if the increased susceptibility is also conferred by the other classes of tet gene determinants commonly encountered in gram-negative organisms (18,23,25,26). To address this question, we compared Kms with and without acid-inactivated tetracycline in HB101 containing inducible members of the other three classes (i.e., A, B, and D) of tet determinants. The mean LD90s of kanamycin for strains containing class A, B, and D determinants in the presence of acid-inactivated tetracycline were 72%, 70% and 80%, respectively, of the LD90 of the matched controls in the absence of acid-inactivated tetracycline ( Table 3). The quantitative relationship between the level of Tcr and the magnitude of the increase in Km' was different in these strains than in strains containing the class C tet gene (Fig. 1). The LD90 of kanamycin in the absence of acid-inactivated tetracycline was greater in the strain containing TnlO than in plasmid-free strains. The basis of this difference is not known.
Mapping the Tet domain that confers the pleiotropic effects of Tet. To determine whether the increased Km1 of Tcr bacteria required the activity of a functional Tet protein, we first assessed the effect on Km1 of defined structural tet gene mutations that abolished Tcr. None of six regionally mapped tet point mutations that abolished Tcr suppressed the tetdependent increase in Km1 (not shown). Therefore, the tet-dependent increase in Km1 does not require a functional Tet protein.
These results suggested that plasmids lacking specific portions of the let gene might be used to map the minimum Tet domain that is necessary to express the pleiotropic effects of Tet. Mutant pBR322 plasmids were constructed by deleting the portions of the tet gene diagrammed in Fig. 2. Although every deletion abolished Tcr (Table 1), only deletions that removed the tet gene DNA sequence upstream of the BamHI restriction site suppressed the increased Km1 (Fig. 2). All   shown. The LD90 of kanamycin for each plasmid-containing strain and plasmid-free HB101 was compared in three independent experiments as described in Materials and Methods. Data from a single experiment are shown. Two independent isolates of most deletion plasmids were tested and yielded similiar results. The LD90s for HB101 containing 10 different human DNA insertion plasmids were examined. Of the 10, 9 were more Kms than plasmid-free HB101 was. Abbreviations: Km, kanamycin; NT, nucleotide. nucleotide Sau3A fragments of human DNA into the tet gene BamHI site (e.g., pCCl-1) abolished Tcr (Table 1), but did not suppress Km1 (Fig. 2). Changes in the reading frame of mRNA encoded by DNA downstream of the BamHI site (e.g., pCC45) also abolished Tcr (Table 1), but did not suppress the increased Km1 (Fig. 2). Thus, the increase in Km1 that accompanies tet expression requires neither the activity of a functional Tet protein nor a structurally intact Tet protein.
Mutations (kdpABC, trkA, and trkD) in E. coli TK2205 significantly impair potassium uptake (10,13). Accordingly, TK2205 cannot grow without a high external concentration of potassium ions. This defect is partially complemented by expression of the pBR322 tet gene (10). The doubling time of  Table 4). Complementation of the potassium uptake defect requires tet gene expression; TK2205 containing plasmid pCC42 will not grow in K2.5 unless tet gene expression is induced by tetracycline or acid-inactivated tetracycline. We next determined whether the N-terminal portion of the Tet protein also was sufficient to complement the potassium uptake defect. A summary of these experiments is presented in Table 4. All strains failed to grow in potassium-free medium (KO) and had doubling times of 55 to 60 min in K115.
There was complete concordance between the ability of the mutant plasmids to increase Kms (Fig. 2) and to grow in K2.5 ( Table 4). The Tet protein initiation codon, the EcoRV site, and the BamHI site are located at nucleotides 86, 185, and 375, respectively (Fig. 2). Thus, a fragment containing more than the first 33 amino acid residues, but no more than the first 97 amino acid residues, of the class C Tet protein (Tet97) was sufficient to confer at least two of the pleiotropic effects of native TET, but not to confer Tcr. The fact that these pleiotropic effects were not suppressed by several different tet mutations that altered the amino acid sequence encoded by DNA downstream from the BamHI site indicates that the expression of these pleiotropic effects is independent of the C-terminal amino acid sequence.

DISCUSSION
Factors affecting Tet-dependent aminoglycoside susceptibility. Each of the strains examined became more Kms when the tet genes were expressed. A direct relationship existed between the levels of Tcr and the increase in Kms in strains containing the same class C tet gene (Fig. 1). The magnitude of the increase in Kms was also affected by the bacterial host in which the tet gene was expressed. For example, pBR322 reduced the LD90 of kanamycin an average of 40% in three different E. coli strains, but 69% in the S. typhimurium strain (Table 2). Nonetheless, the quantitative relationship between the levels of Tcr and Kms in these strains was similar ( Fig. 1). Thus, the difference in Kms in these species may reflect strain-specific factors that affect the level of tet gene expression. Increasing the strength of the pBR322 tet gene promoter also increases Tcr and susceptibility to fusaric acid (15), whereas decreasing the strength of the pBR322 tet gene promoter decreases Tcr and the ability to complement potassium uptake defects (10). Thus, the magnitude of at least three of the known pleiotropic effects of tet is directly related to the level of tet gene expression, as defined by the level of Tcr.
Another variable that affects Kms is the class of Tet protein that is expressed. Members of all four Tet classes increased Kms in HB101 (Table 3). However, the quantitative relationship between Tcr and Kms expressed by the class A, B, and D tet genes was different from that expressed by the class C tet gene (Fig. 1). Heterogeneity in the effects of the individual tet classes on resistance to tetracycline and its analogs and complementation of potassium uptake defects also have been reported (10,18,26).
Correlating the functional differences between the individual classes of Tet proteins with the known differences in their amino acid sequences could help to define the structural basis of the pleiotropic effects of Tet. Increased Kms is conferred by class A to D Tet proteins and by fragments of the class C Tet protein containing Tet97. We compared the amino acid sequences of Tet97 among the three classes of Tet proteins for which sequence data are available (16,29,32,34). Two regions of TET97 differ significantly in the conservation of their amino acid sequences. Between pBR322 Tet residues 1 and 15, only 2 of the 15 residues (13%) are identical in the three Tet proteins. A more highly conserved region, however, is located between pBR322 Tet residues 16 and 97, in which 49 of the 82 residues (60%) are identical in the three Tet proteins. Hydropathic analyses performed by using the algorithims of Kyte and Doolittle (17) and Eisenberg et al. (11) also identified three potential multimeric, membrane-spanning structures in Tet97 such as those found in multiprotein complexes and membrane channel-forming proteins (11). These characteristics were conserved among the three classes of Tet proteins. It is reasonable to assume that the amino acid residues and/or the structural motifs which they dictate and that are essential for the expression of the pleiotropic effects of Tet are located within the more highly conserved region.
Basis of the pleiotropy of tet expression. We have shown that two of the pleiotropic effects of Tet, increasing susceptibility to aminoglycosides and complementing potassium uptake defects, are conferred by Tet97. Other data suggest that Tet97 also increases susceptibility to fusaric acid and cadmium (our unpublished studies). Similiarly, Moyed et al. have reported that all mutations that suppress the effect of Tet overexpression on viability map to a restriction fragment encoding the first 100 N-terminal Tet amino acids (27,28). Since many of these mutations did not affect Tcr, it is more likely that they affected the Tet protein than the tet promoter or the tet mRNA ribosome-binding site (27,28). Together, these data suggest that all of the pleiotropic effects of Tet are mediated by Tet97 and therefore may have a common mechanistic basis.
Expression of either native class C Tet or Tet97 increases potassium uptake (10). This suggests that a common basis for the increased susceptibilities of Tcr bacteria to aminoglycosides, cadmium, and fusaric acid might also be increased uptake of these compounds. One possibility that could account for the increased uptake of these compounds is that integration of Tet into the membrane results in a nonspecific increase in permeability. However, as described above, most antibiotics that we and others have examined have equal potency against Tcs and Tcr strains (2; our unpublished results). Moreover, tet expression increases susceptibility only to specific aminoglycosides (e.g., kanamycin but not kasugamycin) and to specific organic acids (2). Therefore, the effect of tet expression is not nonspecific.
Another possibility is that the uptake of these compounds is directly mediated by Tet. For example, the uptake of these compounds could be directly coupled to the efflux of tetracycline. This explanation can be excluded because the pleiotropic effects of Tet are conferred in the absence of tetracycline by constitutively expressed tet genes (Tables 2  and 3) (10,14,15,20). Similarly, Dosch et al. have suggested the Tet protein could have a low substrate specificity and a Km for influx that favors the uptake of these specific molecules (10). In other words, the increased uptake of compounds such as kanamycin could occur by the reverse of the same mechanism which mediates tetracycline efflux. This possibility seems unlikely, since the pleiotropic effects of Tet are conferred by severely truncated Tet proteins which are unable to confer Tcr (Fig. 2).
The mechanism which we propose is that Tet increases the transmembrane proton gradient (Ap), possibly via an effect on membrane respiratory proteins or ATP synthetase. This proposal is based on the following considerations. First, the uptake of most, if not all, of the compounds known to be affected by tet expression is also affected by Ap. The uptake of aminoglycosides in gram-positive and gram-negative bacteria (5,8,12,24), cadmium in Staphylococcus aureus (33), and potassium in the E. coli TK2205 mutant (13) and the Tet-dependent efflux of tetracycline in E. coli (18) are all dependent on the membrane potential. The mode of fusaric acid uptake is not known. However, lipophilic acids similiar to fusaric acid have been used to measure the transmembrane pH gradient in E. coli, and the accumulation of these organic acids is proportional to the magnitude of the pH gradient (30). Therefore, Dosch et al. have interpreted the fact that the antibacterial activity of fusaric acid is greatest at low pH (2) to indicate that the uptake of fusaric acid may be dependent on the transmembrane pH gradient (10).
The second line of evidence supporting our hypothesis is that mutations that reduce the levels of functional cytochromes and coenzyme Q also reduce the membrane potential, the uptake of aminoglycosides, and bacterial susceptibility to aminoglycosides (5,8). Likewise, inhibitors and mutations that mimic the proposed action of the Tet protein, i.e., that affect the ATP synthetase complex so as to reduce proton influx, also increase the membrane potential, the uptake of aminoglycosides, and bacterial susceptibility to aminoglycosides (5,8).
The third line of evidence is that the amino acid sequence of Tet97 is similiar to the sequences of membrane proteins involved in respiration, altered forms of which can also affect susceptibility to aminoglycosides (5,8). Using the FASTP and RDF algorithims (19), we compared the amino acid sequence of the N-terminal 97 residues of the class C Tet protein with the sequences of the 4,253 other proteins in the National Biomedical Research Foundation database. Similiarities to the subunits of multimeric respiratory proteins, including cytochrome b (Z score of 4.5 standard deviations) were detected. One possibility suggested by these similiarities to respiratory proteins is that Tetg7 possesses structural features that enable it to increase Ap, perhaps by directly affecting the activities of protein complexes involved in energy transduction or proton transport or both.
An observation that is apparently inconsistent with the hypothesis that the pleiotropic effects of Tet are caused by an increase in Ap is that expression of the class B TnJO tet gene does not measurably increase the growth rate of TK2205 in potassium-poor medium (10). However, the effect of the class B Tet protein on potassium uptake may simply be smaller than that of class C Tet and may therefore not be detected by observing changes in growth rate.
Our hypothesis makes several testable predictions. Calculating from the data of Damper and Epstein, we would expect the membrane potentials of HB101 and HB101 (pBR322) to differ by about 10 mV. The hypothesis also predicts that tet expression will increase the pH gradient and tip-dependent aminoglycoside uptake. We would also expect that mutations affecting ATP synthetase or other respiratory proteins might suppress the pleiotropic effects of Tet. Likewise, it may also be possible to isolate chromosomal mutations in other genes that suppress specifically the pleiotropic effects of Tet. These studies may define genes whose products interact with Tet in the membrane to cause its pleiotropic effects. Together, such approaches will further elucidate the physiological and structural basis of the pleiotropic effects of Tet.