Role for Tandem Duplication and Lon Protease in AcrAB-TolC- Dependent Multiple Antibiotic Resistance (Mar) in an Escherichia coli Mutant without Mutations in marRAB or acrRAB

Plasmids, strains, and growth conditions.Bacteria were grown in liquid or on agar of Luria-Bertani (LB) medium at 37°C or 33°C as specified. Ethidium bromide was purchased from Amresco Inc., Solon, Ohio; ampicillin, chloramphenicol, kanamycin, nalidixic acid, tetracycline, and DNP were purchased from Sigma-Aldrich Co., St. Louis, Mo. Tetracycline was used at 12 μg/ml for selection of the Tn10 marker. Kanamycin (30 μg/ml) was used for selection of the Tn10Kan and Kan markers. Plasmid pSP-nfnB1, with an nfnB-promoter::luc fusion, and plasmid pSP-luc+ used to construct pSP-nfnB1 (5), were selected with ampicillin (100 μg/ml).

The E. coli strains used during this work are presented in Table 1. M113 was selected as follows: a single colony of E. coli AG100 was grown at 37°C under vigorous shaking in LB broth up to the late logarithmic phase (A ₆₀₀ of 1.0). One hundred microliters of this culture (representing about 10⁸ bacteria) was spread onto one LB agar plate containing 4 μg/ml of tetracycline. The tetracycline-resistant spontaneous mutant M113 appeared after 3 days of incubation at 33°C. M113 was purified two times on LB plates supplemented with tetracycline (4 μg/ml). One isolated colony was grown in LB broth and stored at −80°C in 20% glycerol. Subsequently, a colony of M113 isolated on LB medium (M113R) that was found to have spontaneously lost its resistance was isolated and purified two more times on LB plates. After growth in LB broth, M113R (shown to bear the lon3::IS186 mutation) was stored frozen in 20% glycerol. The strains constructed were purified on LB plates after transduction, grown in LB broth, and stored frozen. Because of the presence of the ppiD::Tn10 marker near lon and the junction of two amplified units, P1 transductions allowed construction of strains with different genotypes. For example, P1 transduction from the donor strain CAG12017 to the recipient M113 and selection for Tn10 allowed the isolation of (i) a strain which reverted one of the lon-mutated loci, but not the duplication (strain M113HN11); (ii) a strain which reverted the duplication, but not the lon3::IS186 mutation (strain M113HN12); (iii) a strain with no change (strain M113HN2); and (iv) a strain that reverted both the duplication and the lon3::IS186 mutation (strain M113HN1).

Drug susceptibility testing.MICs were determined by two different methods: LB agar plating or Etests (AB Biodisk, Solna, Sweden). Strains freshly grown in LB broth up to mid-logarithmic phase (A ₆₀₀ = 0.5) were diluted in saline solution (0.9% NaCl) to an A ₆₀₀ of 0.05. Alternatively, 5-μl drops of suspensions of isolated colonies, each in 130 μl of saline solution, were put on LB agar plates supplemented with increasing amounts of nalidixic acid or chloramphenicol (0, 2, 3, 4, 6, 8, 10, 12, and 16 μg/ml), and incubated at 37°C for 24 h. The increased susceptibility to DNP was determined on LB agar plates supplemented with 0.6 mM DNP after 24 h of incubation at 37°C. Etests were used as recommended by the manufacturer. When strains with the large tandem duplication dupIS186 were analyzed, bacteria were usually grown in the presence of nalidixic acid (5 μg/ml) to maintain the duplication.

Protein extractions.Whole-cell proteins were analyzed from E. coli cultures freshly grown in LB broth to an A ₆₀₀ of 0.5 as described previously (42). Membranes and cytosol proteins were separated as follows. Ten milliliters of bacteria freshly grown in LB broth up to an A ₆₀₀ of 1.0 was harvested by centrifugation, resuspended in 200 μl of a mixture of cold Tris-HCl, pH 8.0 (20 mM), MgCl₂ (2 mM), EDTA (1 mM), and lysozyme (50 μg/ml), and incubated for 30 min on ice. After addition of 3 μl of protease inhibitor cocktail (Sigma-Aldrich Co., St. Louis, Mo.), bacteria were disrupted by sonication for 1 min (output, 2; duty cycle, 15%; Branson Sonifier 250). Unbroken cells were removed by centrifugation for 5 min at 5,000 × g at 4°C, and the membranes were harvested from the supernatant by centrifugation for 45 min at 100,000 × g at 4°C. Cytosolic proteins from the supernatant were precipitated by addition of deoxycholate at a final concentration of 0.002%, incubation 30 min on ice, addition of trichloroacetic acid at a final concentration of 10%, and incubation overnight at 4°C. Precipitated proteins were harvested by centrifugation for 15 min at 13,000 × g, washed with cold acetone, dried, and finally resuspended at room temperature in 200 μl of 1% sodium dodecyl sufate, 1 mM EDTA, 10 mM Tris-HCl (pH 8.0), 1 mM phenylmethylsulfonyl fluoride. Fifty microliters of the same solution was used to solubilize the membranes for 30 min at room temperature.

Detection of MarA, AcrA, and TolC by Western blot hybridization.Proteins were loaded on a 15% (MarA) or 10% (AcrA and TolC) sodium dodecyl sulfate-polyacrylamide gel and separated by electrophoresis (17). The proteins from the gel were transferred electrophoretically for 1 h at 25 V (Trans-Blot SD semidry transfer cell; Bio-Rad, Richmond, Calif.) to a polyvinylidene difluoride membrane (Polyscreen PVDF transfer membrane; NEN Life Science Products, Inc., Boston, Mass.). The membrane was blocked overnight at room temperature in Tris-buffered saline (TBS; 17.1 mM NaCl, 5.1 mM KCl, 2.5 mM Tris-base, pH 7.4) containing 1% bovine serum albumin, washed three times in wash buffer (0.045% Tween 80 in TBS), and hybridized for 1 h at room temperature with polyclonal antibodies diluted 1:10,000 (anti-AcrA), 1:3,000 (anti-MarA) (22), or 1:5,000 (anti-TolC) in buffer A (0.5% Triton X-100, 0.2% sodium dodecyl sulfate, 0.1% bovine serum albumin in TBS). After three washes in wash buffer, the membrane was hybridized at room temperature with horseradish peroxidase conjugated to anti-rabbit antibodies (Cell Signaling Technology, Inc., Beverly, Mass.) diluted 1:2,000 in buffer A. After three washes in wash buffer, the blots were developed with the Western Lighting Chemiluminescence Reagent Plus (Perkin-Elmer, Boston, Mass.). Quantification of the band intensities was done with ImageQuant software (Amersham Biosciences Inc., Piscataway, N.J.).

Luciferase assays.Luciferase was quantified with the Steady-Glow luciferase assay system (Promega, Madison, Wis.) and a Wallac Victor³ V microplate reader (Perkin-Elmer, Boston, Mass.) as described previously (5).

P1 transduction.Fifty microliters of an overnight culture of the donor strain was diluted into 5 ml of LB broth supplemented with 5 mM CaCl₂. After 1.5 h of incubation at 37°C under vigorous shaking, the culture was inoculated with 100 μl of bacteriophage P1vir lysate of AG100. After complete lysis of the donor cells (about 3 h of incubation at 37°C), 200 μl of chloroform was added and the culture was vortexed and centrifuged (15 min at 1,500 × g). Chloroform was added to the supernatant, which was vortexed and stored at 4°C. Two milliliters of an overnight culture of the recipient strain was centrifuged for 10 min at 9,000 × g. Pellets were resuspended in 1 ml of a solution of 10 mM MgSO₄ and 5 mM CaCl₂ and incubated for 20 min at 37°C. P1vir lysate (1, 10, or 100 μl) was added to 100 μl of recipient cells (controls of cells or lysate only were included). After 25 min of incubation at 37°C without shaking, 200 μl of LB broth supplemented with 50 mM sodium citrate was added. After 1 h of incubation at 37°C under vigorous shaking, tubes were centrifuged (10 min at 9,000 × g), and the pellets were resuspended and plated on LB selective medium supplemented with 10 mM sodium citrate and the appropriate antibiotic and incubated overnight at 37°C.

Nucleic acid extraction.For high-quality DNA, extraction from an overnight culture of bacteria in LB broth was performed with the DNeasy tissue kit (QIAGEN, Inc.). For noncomparative PCR-grade DNA template, one colony was suspended in 130 μl of sterile water. Alternatively, 100 μl of bacterial culture was centrifuged for 10 min at 9,000 × g, and the pellet was suspended in 50 μl of sterile water. Twenty-five microliters of those suspensions was heated for 10 min at 98°C and then centrifuged for 5 min at 9,000 × g, and a maximum of 3 μl of the supernatant was used as template DNA for PCR.

Total RNA was extracted from cultures in exponential phase (A ₆₀₀ of 0.6) with the RNeasy minikit (QIAGEN, Inc.), and treated twice with DNase (QIAGEN, Inc.). RNA quality was verified after electrophoresis in 1% agarose gel containing 0.25 μg/ml of ethidium bromide. Comparison of RNA concentrations in each extraction was assessed by quantification of the rRNA bands with the Gel doc 1000 digital camera system and the Molecular Analyst software (both by Bio-Rad).

Reverse transcription.About 200 ng of total RNA was used for cDNA synthesis by reverse transcription using the Superscript III First-Strand synthesis system for reverse transcription (RT)-PCR (Invitrogen, Carlsbad, Calif.) and the random hexamer primers, following the conditions recommended by the manufacturer. Control reactions without reverse transcriptase were also performed.

Primers and routine PCR.Primers used for PCR amplifications (Table 2) were synthesized by the Tufts University Core Facility. PCRs in a final volume of 25 μl were performed in the Gene Amp PCR Systems 2700 and 9700 apparatus (Applied Biosystem, Foster City, Calif.) using Taq polymerase enzyme (Invitrogen) and the conditions recommended by the manufacturer. Amplified products were separated by electrophoresis with TAE buffer (40 mM Tris-acetate, 1 mM EDTA, 0.114% glacial acetic acid) in 1% agarose gels containing ethidium bromide at a final concentration of 0.25 μg/ml and visualized under UV light, and their size was determined by comparison with the 1-kb DNA ladder (Invitrogen).

Comparative PCR protocols.The comparative PCR protocols were derived essentially from references 13 and 9 and were done in thin-walled 0.2-ml PCR tubes (USA Scientific, Inc.). DNA at a concentration of 1 ng/μl was diluted 1:40 in sterile water. From this solution, eight twofold dilutions were performed and used as template DNA (10 μl) in different PCR amplifications of 30 cycles using the same mixture of water, enzyme, deoxynucleoside triphosphates (dNTPs), and primers hybridizing to marA (Table 2). Amplified products were separated by electrophoresis in a 1.2% agarose gel in TAE buffer. The gel was stained for 30 min in TAE buffer supplemented with 0.5 μg/ml of ethidium bromide, followed by 30 min of destaining in TAE buffer solution. Amplified bands were visualized and quantified under UV light (see RNA extraction protocol). PCR amplifications stopped in their log phase (faint bands with proportionality between the quantity of DNA amplified and the quantity of DNA used as template) were used to compare the quality and concentration of DNA in each of the extractions. Amplifications of ybaO were done following a similar protocol and with the same dilutions of template DNA, but the intensities of the amplified bands were normalized according to the results obtained with the standard marA. The final result is a comparison of the ratio of the quantity of gene ybaO detected per gene marA present in the extracted DNA.

Comparative RT-PCR was achieved using a similar approach using template cDNA obtained from 200 ng of total RNA (see above). The sequence of reference (16S rRNA) was amplified from 10 μl of template cDNA diluted 130,000 to 4,160,000 times. acrA, marA, and lon were amplified from 10 μl of cDNA diluted 8 to 2,048 times. Absence of DNA in the RNA extractions used for cDNA synthesis was checked by 16S rRNA gene PCR amplification using the reverse transcription controls as template DNA.

DNA sequencing.PCR-amplified DNA fragments purified with the QIAquick PCR purification kit (QIAGEN) were quantified by measure of their A ₂₆₀ and then sequenced at the Tufts University Core Facility.

Isolation and characterization of a spontaneous multidrug-resistant E. coli mutant.A spontaneous tetracycline-resistant mutant of E. coli AG100 was isolated after 3 days of incubation at 33°C on an LB agar plate containing 4 μg/ml of tetracycline (see Materials and Methods). This mutant, M113, took only 1 day to form visible colonies when replated on similar medium. When compared with the parental strain AG100 and a marR mutant, AG112, M113 presented a typical multiple antibiotic resistance (Mar) phenotype, as seen by its increased resistance to compounds belonging to different families of antibiotics (Table 3). M113 was also serendipitously found to have an unexpected increase in DNP sensitivity (DNP^s) (MIC, 0.6 mM) compared to AG100 and AG112 (MIC, 1.2 mM). When M113 from the −80°C stock was spread on an LB plate and the phenotype of five isolated colonies was checked, two had lost their Mar phenotype. Therefore, M113 inoculations, made from the frozen stock, were grown overnight in LB broth in the presence of nalidixic acid (5 μg/ml) to maintain the resistance phenotype.

A Western blot analysis with anti-MarA antibodies showed at least 9 times more MarA in M113 than in the parental strain, AG100 (Fig. 1A). M113 produced 55% ± 12% of the amount of MarA detected in AG112 (mutant marR). Comparative RT-PCR showed no difference in marA transcription between AG100 and M113, while an increase of ∼15-fold was seen for AG112 (Fig. 2A). Therefore, unlike in previously described Mar mutants, MarA overproduction in M113 was not linked to increased marA transcription. In accord with the RT-PCR findings, no mutation predicted to affect marA transcription was found by sequencing marR and the marRAB promoter (data not shown).

• Open in new tab
• Download powerpoint

FIG. 1.

Analysis of M113. (A) Western blot showing MarA production in AG100, M113, and AG112. A portion of the gel stained with Coomassie blue to show protein loading is presented above the MarA protein identified by Western blot hybridization with anti-MarA antibodies. The quantities of total proteins loaded are 30 μg (lanes a), 10 μg (lanes b), and 3.3 μg (lanes c). (B) Detection of the duplications by PCR. The recipient strain M113 grown in the presence of nalidixic acid (5 μg/ml) was P1 transduced with lysate from the donor strain CAG12017 (ppiD::Tn10). Random tetracycline-resistant transductants were tested by PCR for the presence of the wild-type ppiD and the ppiD::Tn10 loci. Arrows indicate transductants in which both loci were detected, indicating the presence of a duplication. Wild-type E. coli AG100 was used as a control. (C) Relationship between MarA overproduction, Mar phenotype, DNP sensitivity, and the presence of the duplication. M113 was transduced with a P1 lysate from the donor strain CAG12017 (ppiD::Tn10). One example of each of the four different types of transductants obtained (lanes 1 to 4) is presented. a, MarA overproduction measured by Western blot hybridization with anti-MarA; b, persistence of the Mar phenotype detected by an increased resistance to nalidixic acid and chloramphenicol compared to AG100; c, sensitivity to 0.6 mM 2,4-dinitrophenol; d, duplications detected by PCR (as in panel B).

• Open in new tab
• Download powerpoint

FIG. 2.

Amounts of AcrA, MarA, and TolC proteins and induction of genes of the mar regulon in AG100 and mutants grown in LB broth. All results are presented in arbitrary units based on the value obtained with AG100 = 1. marA and acrA transcription was analyzed by comparative RT-PCR. MarA overproduction was quantified by Western blot hybridizations with anti-MarA antibodies using whole-cell extracts of bacteria. The quantifications were done in duplicate. Similar amounts of MarA were detected in M113R and M113 (data not shown). nfnB transcription was analyzed by luciferase dosage using bacteria transformed with the plasmid pSP-nfnB1 carrying an nfnB-promoter::luc fusion. The amounts of AcrA and TolC proteins were quantified by Western blot hybridizations with anti-AcrA or anti-TolC antibodies using whole-cell or membrane protein extracts of bacteria. Experiments were done at least in duplicate.

Mar phenotype linked to chromosomal duplication in M113.In the course of evaluating (by P1 transduction) the cause of the multidrug resistance, we noted that a mutation affecting the Mar phenotype seemed to map near, but not in, the acrAB locus, although the instability of the phenotype impaired precise mapping of the mutation. No mutation was found by sequencing the acrAB promoter or acrR coding for the transcriptional repressor of the acrAB operon. We then transferred by P1 transduction a ppiD::Tn10 marker (also called zba-3054::Tn10 (27) and located 18 kb to the left of acrB) from the donor strain, CAG12017, to M113. PCR amplification with primers that detected either the wild-type or the mutated ppiD locus (Table 2) unexpectedly showed both the wild-type and the mutant loci in some single transductants (Fig. 1B), indicating a duplication of the ppiD locus in M113. When the same P1vir lysate was used with the recipient strain AG100, only the ppiD::Tn10 locus was detected in the selected tetracycline-resistant transductants (data not shown), indicating that the duplication detected in M113 was neither carried by the donor strain CAG12017 nor present in the parental strain AG100.

To analyze the linkage between the MarA overproduction, the Mar resistance phenotype, the DNP sensitivity and the presence of the duplication in M113, additional P1 transductions of M113 with P1 grown on the donor strain CAG12017 were performed. Twenty-three tetracycline-resistant transductants (bearing ppiD::Tn10) were examined, and four different types of transductants were encountered (Fig. 1C). We concluded from this analysis that (i) the MarA overproduction and the Mar phenotype were separable and resulted from different mutations (mutant types 2 and 4 in Fig. 1C); (ii) the MarA overproduction and the DNP^s phenotype were always found together (mutant types 3 and 4) and thus possibly resulted from the same mutation; reversion of this DNP^s mutation was independent of the presence of the duplication (mutant types 1 and 2); and (iii) the Mar resistance phenotype was tightly linked to the presence of the duplication (mutant types 2 and 3), but not the MarA overproduction (mutant types 3 and 4).

Presence of a lon::IS186 mutation in M113.Using P1 transduction from the donor strain CAG12017 and DNP^s as a possible marker for MarA overproduction, we mapped the mutation responsible for the apparent MarA overproduction. The DNP^s of M113 was reverted in 90 out of 102 ppiD::Tn10 transductants analyzed. This 88.2% frequency of cotransduction with ppiD::Tn10 revealed that the mutation causing DNP^s was ∼3.7 kb away from ppiD. No mutation was found in the 10 kb to the right of ppiD::Tn10 (primers used for the PCR and sequencing not shown). PCR amplification of the lon promoter region located 3.6 kb to the left of ppiD::Tn10 showed an additional 1.3-kb-long sequence in M113 compared to its parental strain, AG100. This additional sequence was found in all DNP^s transductants examined, but was absent from the DNP^r ones (data not shown). Sequencing revealed a 1,306-bp IS186 inserted in the promoter of the lon gene of M113 (Fig. 3). We named this mutation lon3::IS186.

• Open in new tab
• Download powerpoint

FIG. 3.

Mutations in M113. (A) Sequence of the mutational events. The acrAB locus is shown as a black dot and the IS186 insertion sequence as black arrow on the wild-type E. coli K-12 chromosome. The letters a, b, c, and d are used as visual references along the chromosome. Main phenotypes of M113R and M113 are presented in italic. The sequence of events is as follows. Event 1 is stable spontaneous IS186 insertion in the lon promoter of AG100. Event 2 is spontaneous tandem duplication of the 149-kb-long unit. The event probably occurred by recombination between the IS186 present on sister chromosomes (41) and allowed growth in the presence of the tetracycline used for M113 selection. Primers hokE1 and lonR1 were designed to specifically detect this dupIS186 event. In event 3, frequent spontaneous loss of the duplication in the absence of selective pressure for the Mar phenotype allowed the selection of M113R. (B) Detail of the lon3::IS186 mutation. Nucleotide coordinates are indicated according to E. coli K-12 genome sequence (GenBank accession no. U00096). The IS186 sequence is partially presented and boxed. Its orientation (noted “I”) is given according to reference 35. The 7-nucleotide-long duplication of the target site is underlined, and the −10 box of the σ³² promoter of lon is shown in boldface and italic.

Using comparative RT-PCR, we found a decrease of about 85% in lon transcription in M113 compared to AG100 after growth in LB broth up to mid-logarithmic phase. The stability of DNP^s was examined in the strain M113R (bearing only a lon3::IS186 mutation). After seven sequential overnight growths on LB plates at 37°C, no loss of the DNP^s was found among 10 isolated colonies from the seventh plate.

lon mutations caused DNP sensitivity. lon mutants of strains AG100 and AB1157 presented an increased sensitivity to DNP (Table 3). As shown above, the increased sensitivity for M113 compared to AG100 was twofold. Although no DNP^s was seen for the lon mutant GJ1912 (Table 3), when its lon-103::IS186 mutation was transduced into AG100, it did confer the DNP^s (strain AG100HN55 in Table 3). Therefore, the DNP^s phenotype associated with a lon mutation appears to be strain-dependent.

MarA overproduction linked to lon3::IS186 and other lon mutations in additional E. coli strains.We examined MarA overproduction by Western blot hybridization in the constructed strains M113R, M113HN12, AG100HN17 (all harboring a lon3::IS186 mutation), and AG100HN55 (lon-103::IS186) (see Materials and Methods) and in the spontaneous mutants AB1899 (lon1::IS186) isogenic to AB1157 (14) and GJ1912 (lon-103::IS186) isogenic to GJ1913 (36). The different lon mutations in different E. coli strains all resulted in increased amounts of MarA as compared to the wild-type parent strains (Table 3).

Relationship of the lon3::IS186 mutation to the Mar phenotype.Although the lon mutation increased the amounts of MarA (Fig. 2A), it was, surprisingly, not responsible for the multidrug resistance phenotype of M113 (see M113R in Table 3). Introduction of the lon3::IS186 mutation into a marR mutant caused a fivefold increase in MarA (Fig. 2A) but had only a small effect on drug resistance (compare strains AG112HN36 and AG112HN48 in Table 3) (Fig. 2A). The results clearly showed that MarA overproduced in a lon mutant failed to confer Mar resistance. Still, the lon mutation did not seem to affect the MarA-dependent Mar phenotype in the marR lon double mutant (compare AG112HN36 and AG112HN48 in Table 3).

A small increased sensitivity to nalidixic acid was observed in M113R and other lon mutants of AG100, while no apparent difference was observed with norfloxacin, another quinolone (Table 3). Because of their sensitivity to SOS system-inducing conditions (24), lon mutants usually present an increased sensitivity to quinolones (23, 38).

The spontaneous lon1::IS186 mutant AB1899 has been reported to present increased resistance to multiple antibiotics compared to its parental strain, AB1157 (Table 3) (32). However, further analysis showed that AB1899 was not isogenic to its parent, AB1157, since reversion of the lon mutation of AB1899 to wild type by P1 transduction with the donor strain CAG12017 only minimally decreased the chloramphenicol resistance of AB1899 (compare strains AB1899HN67 and AB1899HN68 in Table 3). Evidently, at least one other mutation which causes most of the Mar phenotype is present in this strain. Similarly, P1 transduction of the lon3::IS186 mutation from the strain M113HN12 to AB1157, producing the strain AB1157HN70, did not confer a Mar phenotype (Table 3). These findings confirm that the multidrug resistance phenotype of AB1899 results from another “unknown” mutation, not from lon1::IS186.

Endpoints of the large tandem duplication dupIS186 responsible for multidrug resistance. clpX is the gene directly upstream of lon in the E. coli K-12 chromosome, and the clpX-IS186-lon locus is found in lon3::IS186 mutants. When we transduced M113 with a P1vir lysate from the donor strain CAG12017, transductants that had a duplication of the ppiD locus (with one copy carrying the ppiD::Tn10 marker) and a wild-type lon locus (and thus reverted the lon3::IS186 mutation; verified by PCR amplification) also still had another IS186-lon locus, but no clpX-IS186 locus. Therefore, the lon gene was part of the duplication, but clpX was not. We hypothesized that the IS186 inserted in the lon promoter could be one of the two borders of the amplified unit, with the second border being another IS186. We searched the sequenced E. coli K-12 genome for the nearest IS186 present to the right of lon and inserted in the same orientation as in lon. One such IS was present 149 kb to the right of lon at positions 607231 to 608537 according to GenBank sequence accession no. U00096. A forward primer, hokE1, that hybridized to the left of this IS186 was designed and used along with the reverse primer lonR1 to detect by PCR the potential linking region of two amplified units (see Table 2 and Fig. 3A for primer sequences and hybridization sites). Sequencing of the PCR product amplified from M113 confirmed the order hokE-IS186-lon at the junction of two amplified units; therefore, the amplified unit was indeed bordered by the two IS186s. We called this duplication dupIS186. PCR amplification confirmed that the tandem duplication studied was present in M113 but not AG100 and was not induced by P1 transduction per se. Comparative PCR amplification of the marA and ybaO genes present outside and inside the 149-kb amplified unit, respectively, revealed an average of 2.5 copies of the amplified unit after overnight growth of M113 in the presence of nalidixic acid (5 μg/ml) (Fig. 4A). The duplication was maintained during growth of M113 in nalidixic acid, but it was rapidly lost along with the Mar phenotype when M113 was grown in its absence (Fig. 4B).

• Open in new tab
• Download powerpoint

FIG. 4.

Copy number and instability of the amplification. (A) Number of copies of the amplified unit. Three overnight growths of AG100 in LB broth and M113 in LB broth supplemented with (+Nal) or without (−Nal) nalidixic acid at 5 μg/ml were done. DNA was extracted, and a comparative PCR was performed as described in Materials and Methods to compare the number of copies of ybaO (present on the amplified unit) detected per copy of marA (present outside of the amplified unit) in each of the DNA extractions. (B) Instability of the Mar phenotype and the dupIS186 mutation. Strain M113HN11 harboring dupIS186 was passaged four times sequentially on agar plates in two independent experiments (a and b). In experiment a, the plate was supplemented with 5 μg/ml nalidixic acid at the second passage (+). For each experiment (a and b), after each passage on the plate, four isolated colonies were studied to determine the proportion of bacteria harboring a Mar phenotype. The detection of the dupIS186 mutation in the isolated colonies was done by PCR as described in Fig. 1B. Only the bacteria harboring a Mar phenotype had the dupIS186 mutation.

Mar phenotype resulting from the dupIS186 mutation.The strains M113HN11 and AG100HN3, each harboring dupIS186 and a wild-type lon gene, were constructed by P1 transduction and analyzed (see Materials and Methods and Table 3). They did not overproduce MarA and were DNP^r (Table 3). dupIS186 conferred increased resistance to chloramphenicol and nalidixic acid, as well as resistance to ampicillin and norfloxacin at a lower level (Table 3). The resistance to tetracycline could not be tested in these Tn10-bearing strains, but the original M113 strain presented increased resistance to tetracycline that was not caused by the lon3::IS186 mutation and was hence presumably caused by the dupIS186 mutation. When it was P1 transduced into the marR mutant AG112, the dupIS186 mutation conferred high multidrug resistance (mutant AG112HN2-74 in Table 3).

Segregation of duplicated copies.M113 was transduced with a P1vir lysate from strain CAG12017, and one colony with a ppiD::Tn10-marked duplication was detected and purified twice on LB plates to give single colonies. One of these colonies that still harbored the duplication was then plated one more time on LB to give single colonies. Nine isolated colonies were then tested for the presence or loss of the duplication and for their Mar phenotype. Five colonies still had the duplication and the Mar phenotype, and four colonies had lost the duplication and the Mar phenotype. Of these four, three had lost the copy carrying the ppiD::Tn10 locus and one had lost the copy with the wild-type ppiD gene. So, the Mar phenotype of M113 was not caused by a mutation carried by one of the amplified units, but resulted from the presence of several amplified units.

Role of acrAB in the multidrug resistance associated with dupIS186.The 149-kb-long amplified fragment harbored the multidrug resistance efflux locus acrAB. AcrAB is largely responsible for mar-mediated multiple antibiotic resistance in E. coli (31). Duplication of this locus would explain the Mar phenotype. To test this possibility, two approaches were investigated.

First, we deleted acrAB in M113 by P1 transduction from the donor strain AG100A (ΔacrAB::Kan). Of six M113 ΔacrAB transductants analyzed, one had lost the duplication, retained only the ΔacrAB locus, and had the phenotype of AG100A. Of the five others with a duplication, one was isolated on LB and lost the Mar phenotype but retained at least one copy of both wild-type acrAB and ΔacrAB. This transductant lost the Mar phenotype presumably because of the lower number of copies of the wild-type acrAB locus. When similar experiments were performed with a donor strain carrying a marker ppiD::Tn10 and a wild-type acrAB locus, no colony with a detectable duplication ever lost its Mar phenotype. So, decreasing the number of copies of intact acrAB reverted the Mar phenotype in bacteria that still carried a dupIS186. This finding suggested a critical role for the acrAB locus as opposed to other loci in the duplication.

In a second approach, the dupIS186 was transferred by P1 transduction from the donor strain M113HN2 to the recipient strain AG100A (ΔacrAB). Transductants were selected for their resistance to tetracycline (marker ppiD::Tn10). A transductant carrying a dupIS186 but no wild-type acrAB locus (AG100AHN2-21) was detected by PCR. No increased resistance to the different antibiotics tested was found, revealing again that the presence of amplified acrAB was needed for dupIS186-dependent multidrug resistance (Table 3).

Ethidium bromide resistance conferred by other transporter gene(s) present on the amplified unit.The genes cusAB, emrE, fsr, and mdlAB code for transporters and are present, along with acrAB, on the amplified unit of dupIS186. When overproduced, emrE (positions 567538 to 567870 on the chromosome of E. coli K-12; GenBank accession no. U00096) confers resistance to ethidium bromide (28). Overproduction of acrAB, but not of cusAB, fsr, and mdlAB also confers an ethidium bromide resistance (28). The strain AG100AHN2-21 carrying a dupIS186, but no wild-type acrAB locus, showed an increased resistance to ethidium bromide (MIC of 18 μg/ml) compared to AG100A (MIC of 6 μg/ml). Therefore, the genetic amplification of other transporter genes—such as emrE—can also result in increased resistance to specific drugs.

Activity of MarA stabilized in a lon mutant.The finding that MarA overproduced in lon mutants did not confer the Mar phenotype suggested that the MarA stabilized in lon mutants might not be active. We therefore examined transcription of the MarA-regulated genes acrAB, nfnB, and tolC (6) and/or amounts of their products in the lon mutant M113R (Fig. 2) as an assay for MarA activity. The transcription of acrA and nfnB was increased in M113R (Fig. 2B). The amounts of AcrA found in a lon or marR mutant were both increased similarly compared to the wild-type strain (Fig. 2C). The marR lon double mutant showed increased amounts of TolC protein (Fig. 2C), suggesting that the stabilized MarA could also induce tolC transcription. Therefore, MarA stabilized in a lon mutant appeared to be active.

Effect of a lon mutation on the AcrAB-TolC pump function.Overexpression of AcrAB alone can confer the Mar phenotype (31). Therefore, the absence of a Mar phenotype in a lon mutant which overproduced AcrA (Fig. 2) was unexpected. One possibility was that a general increased sensitivity to antibiotics was caused by the lon mutation. If the pump was functional in a lon mutant but the lon mutation also increased the sensitivity of bacteria, a ΔacrAB lon double mutant (AG100AHN60) would show increased sensitivity to antibiotics such as chloramphenicol and ampicillin compared to the ΔacrAB mutant carrying a wild-type lon gene (AG100AHN61). This was not observed (Table 3), revealing that the absence of Mar phenotype in a lon mutant was not caused by a general increase in drug susceptibility.

The remaining possibility was that the AcrAB-TolC pump itself was present at a reduced amount or had a reduced activity in the absence of Lon protease. The AcrAB-TolC pump consists of two transmembrane proteins (AcrB in the inner membrane, TolC in the outer membrane) linked by periplasmic AcrA (40). Similar amounts of TolC were detected in the membrane protein extract (Fig. 2C), but TolC was not found in the cytosol fraction (data not shown) of AG100, AG112, and M113R. Therefore, the decreased activity of the AcrAB-TolC pump in a lon mutant was not the result of a decreased amount of the TolC protein present in the membrane fraction. Similarly, increased amounts of AcrA were found in whole-cell extracts of AG112 and M113R (Fig. 2C). AcrA was found in both the cytosol and the membrane fractions in AG100, AG112, and M113R (data not shown), presumably because of its interactions with AcrB and TolC. Thus, the lon mutation did not seem to affect the MarA-induced overproduction of AcrA or its compartmentalization. The effect of Lon protease on AcrAB-TolC function appears to occur via a mechanism other than an effect on production and compartmentalization of AcrA and TolC.

A lon mutation stabilized MarA.The lon::IS186 mutations resulted in a ≥9-fold increase in MarA protein (Fig. 1A and 2A), without any increase in marA transcription. During the time of our investigation, Griffith and coworkers showed that the Lon protease, along with at least one other still unknown protease, was the only E. coli ATP-dependent protease (of those presently known) involved in a rapid turnover of MarA, and a lon-510 mutation resulted in a ≥24-fold decrease in the breakdown rate of MarA (12). Our data agree with their result; the apparent overproduction of MarA observed in lon mutants presumably resulted from the stabilization of the basal quantities of MarA produced. In contrast to these investigators, we were able to detect increased quantities of MarA expressed from the wild-type chromosomal marRAB locus in lon mutants using our anti-MarA antiserum.

Lon indirectly regulates the mar regulon via MarA.M113R (lon3::IS186) overproduced about half of the amount of MarA found in the marR mutant AG112 (Fig. 2A), yet the levels of induction of the MarA-regulated acrAB and nfnB transcription and the production of AcrA were apparently similar in both strains (Fig. 2). This lack of proportionality was not caused by a saturation effect of MarA, because nfnB transcription and AcrA and TolC production were further increased in a marR lon double mutant in which increased amounts of MarA were found (Fig. 2A and B). A possibility would be that SoxS, which is also partially stabilized by a lon mutation (12) and which also induces the transcription of those genes (33), could contribute to the induction of the genes of the mar regulon in a lon mutant. Unlike MarA, overproduced SoxS was described as inducing marRAB transcription in the presence of a wild-type MarR repressor (25). Since we did not observe any induction of marRAB transcription in a lon mutant (Fig. 2A), the contribution, if any, of SoxS to the increased transcription of the mar regulon in a lon mutant seems minimal.

The Lon protease is required for an efficient AcrAB-TolC-dependent Mar phenotype.The amount of MarA in M113R was at least 9 times that of the wild-type strain AG100. As seen in uropathogenic E. coli strains, such MarA overproduction should confer a Mar phenotype (4). Furthermore, quantities of AcrA and TolC proteins found in M113R were not less than that found in AG112 (Fig. 2C), where overproduction of the AcrAB-TolC pump causes a Mar phenotype (31) (Table 3). The lack of multidrug resistance in lon mutants appears linked to an inactive AcrAB-TolC pump. This dependence on Lon was only partial, since the marR lon mutant AG112HN48 and the dupIS186 lon mutant M113, which respectively overproduced 2.3 and 1.6 to 2.4 times the amount of AcrA found in AG112 (Fig. 2C) (data not shown for M113), did have a multidrug resistance, with almost the same phenotype as AG112 (Table 3). An increased multidrug resistance was also observed with a mutant carrying a deletion of the lon gene and a duplication including acrAB (data not shown). Therefore, the partial activity of AcrAB-TolC observed in lon3::IS186 mutants was not explained by the remaining 15% of lon transcription detected, but rather by an intrinsic partial activity of the pump in the absence of Lon. Using these data, we estimated that the absence of Lon protease reduced by ∼60% the ability of AcrAB-TolC to confer a Mar phenotype.

The reason for inactivity in the absence of Lon protease did not appear to be altered compartmentalization of AcrA and TolC proteins. A possible clue to the effect of Lon on AcrAB-TolC comes from the observation that the acrR mutant AG100B, which overproduced larger amounts of AcrA than a marR mutant (∼2.5 times the amounts found in AG100 versus ∼1.7 times the amount for AG112; Fig. 2C) (data not shown; see also references 42 and 43), did not have a higher multidrug resistance than AG112 (Table 3). This finding was unexpected because increased amounts of AcrAB in a marR dupIS186 double mutant (∼5.5 times the amount of AcrA found in AG100; data not shown) caused a much higher multidrug resistance than that of AG112 (Table 3). Therefore, in marR mutants carrying a wild-type lon gene, efficient up- or down-regulation of other genes by MarA might somehow increase the AcrAB-TolC-dependent multidrug resistance, while in a lon mutant, or in an acrR mutant which does not overproduce MarA, those genes might not be efficiently regulated. Current efforts are focused on identifying the effect of Lon protease on AcrAB-TolC function.

Unstable multidrug resistance phenotypes caused by the dupIS186 mutation.Despite the lack of effect on the Mar phenotype of the overproduced AcrAB-TolC pump in the lon mutant, multidrug resistance was observed in lon mutants with an amount of AcrAB further increased. The latter was achieved by the amplification of the acrAB genes present on the large genetic duplication dupIS186 (mutant M113 overproducing 2.7 to 3.8 times more AcrA than AG100; data not shown) or by a marR mutation in AG112HN48 (Fig. 2C; Table 3). We found that 2 to 3 copies of the amplified unit of dupIS186 were maintained per chromosome after an overnight growth of M113 in the presence of 5 μg/ml of nalidixic acid (Fig. 4A). Because of the numerous chemically unrelated substrates that AcrAB expels (28), this transporter likely confers most of the resistance phenotype associated with dupIS186.

Mutations conferring low multidrug resistance seem to serve as intermediate steps to reach higher antibiotic resistances (see reference 11, for example), and mutations in marR, soxR, or acrR were found in numerous highly antibiotic-resistant isolates of E. coli (21, 30, 43). We found that, combined with a marR mutation, the dupIS186 mutation caused high multidrug resistance (resistance to 128 and 192 μg/ml of chloramphenicol and nalidixic acid respectively; mutant AG112HN2-74 in Table 3). Up to 5.5 times more AcrA proteins were found in a marR dupIS186 mutant compared to the wild-type strain (data not shown). Such high quantities of AcrA proteins were also found in highly antibiotic-resistant E. coli mutants in which the mutations causing the increased amounts of AcrAB were not characterized (21).

Because it required the presence of a preexisting lon3::IS186 mutation, the dupIS186 mutation studied here might not be a frequent spontaneous mutation. However, if other genetic amplifications of acrAB that do not require the presence of a first mutation were found, multidrug resistance caused by genetic amplification of acrAB could be an important mechanism of resistance among antibiotic-resistant isolates of E. coli. Unstable duplications could represent the mechanism for loss of resistance in some clinical strains upon subculturing in the laboratory, a phenomenon previously attributed to plasmid loss.

Other implications of tandem amplifications.Because of their instability, tandem genetic amplifications might represent an efficient mechanism for prokaryotes to temporarily respond to external toxic threats, such as antibiotics, without alteration of their genome with a stable mutation. As observed by Matthews and Stewart in some spontaneous methicillin-resistant Staphylococcus aureus mutants having an amplification causing the resistance, loss of the amplification but not loss of the resistance can also be observed after prolonged storage on high concentration of antibiotics (20). So, persistent selection for the resistance phenotype would ultimately favor the growth of mutants in which a new stable mutation conferring resistance allowed the loss of the amplification. Genetic amplification could then represent an early mechanism for selection of stable mutations conferring resistance to antibiotics. Our work also shows that transporter genes can be important determinants for the selection of large tandem amplifications under specific environmental conditions.