Constitutive soxR Mutations Contribute to Multiple-Antibiotic Resistance in Clinical Escherichia coli Isolates

Bacterial strains and plasmids.The strains and plasmids used in these studies, along with their relevant properties and source descriptions, are listed in Table 1. Bacteria were cultured overnight in Luria-Bertani (LB) broth (27) at 37°C, unless otherwise stated.

RNA extraction and Northern blotting.Overnight bacterial cultures were diluted 1/100 in fresh LB broth and grown with vigorous shaking (250 rpm) for 90 min at 37°C or for 105 min at 30°C, as indicated. At that point, the cultures were split, and the appropriate samples were treated with 1.3 mM PQ. Shaking at 37°C or 30°C and 250 rpm was continued for another 30 to 40 min, and RNA was harvested from 1.5-ml aliquots of culture using the RNeasy Mini kit from QIAGEN. RNA yield was estimated by measuring absorption at 260 nm and 280 nm. The appropriate amount of RNA was electrophoresed on 1.5% agarose gels containing 2% formaldehyde and 0.07 μg/ml ethidium bromide. The RNA was transferred overnight to a nylon membrane using the TurboBlotter downward transfer system (Schleicher & Schuell). The membranes were hybridized with appropriate ³²P-labeled probes, and gene expression was quantified on a phosphorimager (Bio-Rad). Hybridization with a ³²P-labeled 16S RNA probe was used to control for loading differences.

Sequencing of the soxRS regions from E. coli clinical isolates.A 1-kb fragment corresponding to the soxRS region was amplified by PCR from the genomic DNA of the E. coli clinical isolates. The forward primer was 5′-GGCGAAGCTTCCGCAGGTGTTTATGC-3′ and the reverse primer was 5′-CGTCGGGGGAAGCTTTCCTGTGTACC-3′ (regions corresponding to, respectively, bp 284 to 298 downstream of soxS in the GenBank deposited sequence and bp 1281 to 1271 downstream of soxR are underlined). The amplified fragments were gel purified, and both strands were sequenced at the High-Throughput Sequencing Facility of the Dana-Farber/Harvard Cancer Center.

Cloning of soxR from E. coli clinical isolates.A 600-bp fragment containing the soxR gene and its promoter was amplified by PCR from the genomic DNA of the E. coli clinical isolates. The forward primer was 5′-GCCGCTCGAGCTGAATAATTTTCTGATGGG-3′ (region corresponding to soxS bp 641 to 660 downstream of soxR is underlined) and contained an XhoI site. The reverse primer was the same one used for sequence analysis and contained a HindIII site. After digestion with these enzymes, the PCR fragments were cloned into vector pCR2.1-TOPO using the TOPO TA cloning kit from Invitrogen. Successful clones were confirmed by restriction digestion analysis. Several clones for each mutant were sequenced to verify that no mutations were introduced by the PCR procedure. The cloned soxR alleles were then subcloned under the control of their own promoters into pACYC177 (New England Biolabs) using restriction enzymes XhoI (New England Biolabs) and HindIII (New England Biolabs). The following plasmids were obtained: pAK-WT, which contains the wild-type soxR allele, and pAK-E3, pAK-E17, pAK-E19, pAK-M1, and pAK-I237, which contain, respectively, the soxR alleles from strains E3, E17, E19, M1, and I237 (Table 1). Subcloning was verified by restriction digestion analysis.

Transcriptional activity of cloned soxR alleles.Plasmids pACYC177, pAK-WT, pAK-E3, pAK-E17, pAK-E19, pAK-M1, and pAK-I237 were introduced into strain EH46 (ΔsoxR soxS::lacZ) (Table 2), and the transcriptional activity of the expressed SoxR proteins was assayed by measuring expression of β-galactosidase directed by the reporter gene. The plasmid-bearing strains were grown overnight in LB broth containing 100 μg/ml ampicillin (LB-amp [100 μg/ml]) at 37°C with shaking (235 rpm). The next day, the cultures were subcultured 1/100 into fresh LB-amp (100 μg/ml) medium, grown for 105 min, and split into two cultures, PQ was added at a final concentration of 250 μM to one of each pair, and shaking (235 rpm) continued at 37°C for another 45 min. The cultures were chilled on ice for 15 min, and β-galactosidase activities were assayed in sodium dodecyl sulfate (SDS)-CHCl₃-treated cells as described by Miller (27).

Complementation of a ΔsoxR strain with cloned soxR alleles for increased antibiotic resistance.Plasmids pACYC177, pAK-WT, pAK-E3, pAK-E17, pAK-E19, pAK-M1, and pAK-I237 were introduced into strain TN402 (ΔsoxR soxS⁺) (Table 1), and their contribution to the antibiotic resistance phenotype of the recipient strain was assayed on antibiotic gradient plates. Overnight cultures of TN402 (ΔsoxR) (Table 1) bearing the appropriate plasmids were grown in LB-amp (100 μg/ml) at 37°C with shaking (235 rpm), and samples were applied to an antibiotic gradient plate. The resistance to the antibiotic was determined by measuring growth along the antibiotic gradient after 12 to 24 h of incubation at 37°C. The antibiotics tested were chloramphenicol up to 30 μg/ml, nalidixic acid up to 15 μg/ml, and ciprofloxacin up to 0.125 μg/ml. Antibiotic gradient plates were prepared as described by Cunningham et al. (4), except that each plate contained 60 ml of solid medium.

Multicopy suppression of antibiotic resistance with wild-type soxR.Clinical isolates E17, M1, and I242 and control laboratory strains GC4468, DJ901, JTG936, and JTG1052 were transformed with plasmid pSXR encoding isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible SoxR (Table 1) or the vector (pSE380). Overnight cultures of the transformed strains grown in LB-amp (100 μg/ml) at 37°C with shaking (235 rpm) were applied to antibiotic gradient plates containing 5 μM IPTG. Resistance was assessed as described in the previous section.

Expression of soxS mRNA in E. coli clinical isolates.We sought to determine the molecular basis of the previously reported constitutive soxS expression in antibiotic-resistant clinical E. coli isolates and its possible role in resistance. Strains E3, E17, and E19 (34) (Table 1) were isolated from blood samples from patients at cancer centers in Europe and the Middle East (34). The strains were all highly fluoroquinolone resistant (MIC of 64 μg/ml, compared to ≤0.25 μg/ml for the wild type [34]) and bore mutations in both gyrA and parC (34). Strain M1 (Table 1) was isolated from a urinary tract infection from a nonneutropenic patient (34). Strain M1 had modest fluoroquinolone resistance (MIC, 4 μg/ml [34]) and a gyrA mutation (30). Strains I236, I237, I242, I243, I244, I246, I248, I251, I253, and I254 (40) (Table 1) were obtained from hospitals in Spain. These strains were reported to have intermediate/high fluoroquinolone resistance, resistance to multiple antibiotics and a gyrA mutation (40). All of the above strains had been reported to exhibit high basal soxS mRNA expression (30, 36), but the mechanism(s) underlying this altered expression was unknown.

We first tested whether the strains retained elevated soxS expression under our experimental conditions. In our hands, compared to the wild-type laboratory strain GC4468, strains E17, E19, M1, I237, I242, and I253 all expressed soxS mRNA at elevated levels during normal (“unstressed”) cell growth (Fig. 1). The absolute levels of basal soxS mRNA expression differed among the different strains. Most of the strains tested retained the ability to up-regulate soxS mRNA upon treatment with PQ (Fig. 1A, B, and C and data not shown). In strain E17, basal expression of soxS mRNA was already maximal, as it was not induced further by PQ treatment (Fig. 1A). In strain E19 (Fig. 1A), the high basal level of soxS mRNA was further elevated ∼2-fold by PQ treatment. In strain M1 (Fig. 1B) the basal soxS expression was elevated but to a lesser extent than in strains E17 and E19. Strains I237, I242, and I253 expressed increased basal amounts of soxS mRNA compared to the laboratory strain GC4468 (Fig. 1C), but the absolute level of expression was low compared to the constitutive expression of strain JTG936 (14) (Table 1). None of the other strains expressed detectably increased amounts of soxS mRNA, in contrast to previous reports. However, most of the I-series strains did exhibit elevated basal soxS expression when cultured in Iso-Sensitest broth (M. Webber and L. J. Piddock, personal communication).

• Open in new tab
• Download powerpoint

FIG. 1.

Expression of soxS mRNA in clinical E. coli isolates. Overnight cultures were diluted 1/100 in fresh LB broth and grown with vigorous shaking (250 rpm) for 105 min at 30°C (A and C) or for 90 min at 37°C (B). At that point, the cultures were split, the appropriate samples were treated with 1.3 mM PQ, and shaking resumed at 250 rpm and 30°C or 37°C as appropriate for another 30 to 40 min. RNA was then harvested for Northern analysis. (A) Expression of soxS mRNA in E. coli strains GC4468 (soxR⁺), E3, E17, E19, and M1, with or without 1.3 mM paraquat treatment. (B) Basal expression of soxS mRNA in E. coli strains GC4468, JTG936 (soxR^c), I236, I237, I242, I243, I244, I246, I248, I251, I253, and I254.

Mutations in the soxRS region of the E. coli clinical isolates.Theoretically, constitutive soxS expression could be due to mutations in the soxR gene that render the protein constitutively active, mutations in the soxS promoter that turn on transcription of the soxS gene constitutively (16), changes in the soxS gene that prolong the half-life of the soxS mRNA (6), or mutations in other genes that regulate the redox status of SoxR (8, 9, 11, 12, 22). To address some of these possibilities, we first sequenced the entire soxRS region of each of the clinical isolates.

Table 2 shows that strains E17, E19, M1, I237, I242, and I253, which exhibited increased soxS expression, all harbored missense mutations in the soxR gene. An in-frame 3-bp deletion (C₁₁₃₀A₁₁₃₁G₁₁₃₂) in strain E17 resulted in deletion of serine-128 (S128) within the cysteine cluster of SoxR. A deletion of two consecutive thymines in strain E19 (T₁₁₉₁T₁₁₉₂) resulted in a −1 frameshift at leucine-148 and a termination three codons later, which truncated SoxR three residues from the C terminus. A G₈₀₈-to-A transition in M1 changed arginine-20 to histidine within the helix-turn-helix motif of the SoxR DNA binding domain. Strains I237, I242, and I253 all harbored the same two differences in the soxR gene compared to the reference laboratory strain: C₈₆₂ to G changes threonine-38 to serine, and G₉₆₉ to C changes glycine-74 to arginine. Both of these alterations lie within the second helix-turn-helix motif proposed to participate in DNA binding (1, 16).

The sequence analysis (Table 2) also showed that strains E3 and I251 harbor the same mutation (G₁₀₈₀ to A) that changes alanine-111 to threonine. However, the basal expression of soxS mRNA was not elevated in these strains (Fig. 1). The soxS-soxR intergenic region in all the strains was identical to that in the wild-type laboratory strain.

Various polymorphisms were observed in the clinical strains, either silent mutations or amino acid changes without an apparent phenotypic effect. Strains I237, I242, and I253 harbored nine additional silent point mutations in soxR and eight in soxS (Table 2). All the changes, missense and silent, were identical among the three strains, which suggested that I237, I242, and I253 may have arisen from the same clone. Strains I243 and I246 also shared a silent G₁₀₇₉-to-A change in soxR and a silent G₄₉₄-to-A change in soxS. They shared both of these polymorphisms with strain I236, which harbored an additional silent point mutation in soxS: a T₃₄₁-to-C change. Interestingly, strain E17, which we obtained from a different source than strains I236, I243, and I246, also carried the silent G₄₉₄-to-A polymorphism in soxS.

Constitutive activity of cloned mutant soxR genes.To confirm that the increased basal soxS mRNA expression of the E. coli clinical isolates was actually due to constitutive SoxR activity, we cloned the mutant genes into expression vectors (see Materials and Methods). These plasmids (the empty vector pACYC177 and pAK-WT, pAK-E3, pAK-E17, pAK-E19, pAK-M1, and pAK-I237) were introduced into strain EH46 (ΔsoxR soxS::lacZ) (Table 1). The in vivo activity of the mutant proteins was assayed by measuring β-galactosidase expression in the presence or absence of 250 μM PQ, a redox-cycling agent and inducer of the soxRS regulon (17). Figure 2 shows that the mutant soxR genes from E. coli clinical isolates E17, E19, and M1 expressed proteins that conferred high-level expression of the soxS::lacZ reporter, even in the absence of an activating signal. In contrast, the soxR gene from strain E3 did not appear to encode a constitutively active form of SoxR. These observations are in agreement with the results in Fig. 1A and B, which show expression of soxS mRNA in the clinical isolates. Unexpectedly, the SoxR protein expressed from pAK-I237 did not confer increased basal β-galactosidase expression in strain EH46, despite the elevated soxS expression in strain I237 (Fig. 1C). Treatment with PQ induced β-galactosidase expression greatly and to the same extent (>20-fold) in strains harboring the wild-type (WT), E3, or I237 soxR genes. Paraquat treatment did not increase significantly the expression of β-galactosidase by the E17 and E19 SoxR proteins, while it slightly increased the β-galactosidase activity induced by the M1 protein (Fig. 2). For all three alleles—E17, E19, and M1—uninduced expression was already almost maximal (Fig. 2). PQ-induced expression was highest with pAK-E17, pAK-E19, and pAK-M1 (Fig. 2), while pAK-E3 and pAK-I237 paralleled the levels seen for wild-type SoxR (pAK-WT) (Fig. 2).

• Open in new tab
• Download powerpoint

FIG. 2.

Regulation of soxS::lacZ by mutant SoxR proteins. Plasmids pACYC177, pAK-WT, pAK-E3, pAK-E17, pAK-E19, pAK-M1, and pAK-I237 carrying no soxR, wild-type soxR, or mutant alleles of soxR were introduced into strain EH46 (ΔsoxR soxS::lacZ) (Table 1), and the transcriptional activity of the expressed SoxR proteins was assayed by measuring expression of β-galactosidase in the presence or absence of 250 μM PQ. The data shown are the means and standard errors of two determinations.

Antibiotic resistance conferred by cloned mutant soxR genes.If the SoxR^c proteins contributed to the increased antibiotic resistance phenotype of the E. coli clinical isolates, introduction of these proteins into a laboratory ΔsoxR strain (TN402) (Table 1) would confer antibiotic resistance. Resistance was assayed on antibiotic gradient plates with chloramphenicol, nalidixic acid, or ciprofloxacin (Fig. 3). For all three antibiotics tested, introduction of the E3 (pAK-E3) and I237 (pAK-I237) mutant alleles gave the same basal antibiotic resistance as provided by the wild-type (pAK-WT) allele. The vectors encoding the constitutively active E17 (pAK-E17), E19 (pAK-E19), and M1 (pAK-M1) SoxR proteins conferred significantly higher basal resistance to all three antibiotics tested. For ciprofloxacin, these three mutant alleles increased the basal antibiotic resistance 12-, 8-, and 10-fold, respectively, much more than soxRS-mediated mechanisms alone usually provide (1, 20, 24). Treatment with PQ increased the antibiotic resistance of strains harboring the WT, E3, and I237 soxR alleles significantly and to the same extent for each antibiotic tested (Fig. 3). PQ treatment did not significantly affect the antibiotic resistance conferred by the E17 protein and only moderately increased the antibiotic resistance conferred by the E19 and M1 proteins. These results agree with the reporter gene data shown in Fig. 2.

• Open in new tab
• Download powerpoint

FIG. 3.

Antibiotic resistance conferred by mutant soxR genes. Plasmids pACYC177, pAK-WT, pAK-E3, pAK-E17, pAK-E19, pAK-M1, and pAK-I237 encoding no soxR, wild-type soxR, or a mutant allele of soxR were introduced into strain TN402 (ΔsoxR) (Table 2), and their contribution to the antibiotic resistance phenotype of the recipient strain was assayed on antibiotic gradient plates in the presence or absence of 50 μM PQ. (A) Chloramphenicol (CHL), 0 to 30 μg/ml; (B) nalidixic acid (NAL), 0 to 15 μg/ml; (C) ciprofloxacin (CIP), 0 to 0.125 μg/ml. The data shown are the means and standard errors of three determinations.

It is worth noting that, even in the absence of PQ, introduction of the wild-type soxR vector pAK-WT into strain TN402 (ΔsoxR) (Table 1) increased its resistance to chloramphenicol about twofold compared to the vector-only control (Fig. 3A). The pAK-E3 vector had a similar effect, and neither pAK-WT nor pAK-E3 affected basal resistance to nalidixic acid or ciprofloxacin (Fig. 3). The soxR-dependent increase in chloramphenicol resistance might reflect activation of SoxR by this drug; however, such activation has not been reported. Chloramphenicol can activate MarA expression, but it was not tested whether this effect was soxR dependent (15).

Conversely, PQ treatment increased the antibiotic resistance of the vector-only control to ciprofloxacin about twofold (Fig. 3). This effect was specific to ciprofloxacin. PQ has been shown to be a weak activator of marA expression (35). This could account for the increased ciprofloxacin resistance in the absence of SoxR, particularly if ciprofloxacin is more sensitive than nalidixic acid and chloramphenicol to the marRAB-mediated resistance mechanisms.

Multicopy suppression of antibiotic resistance with wild-type soxR.Nonactivated wild-type SoxR competes with mutant activated protein (20), which allowed us to assess the contribution of mutant soxR alleles to the antibiotic resistance profile of the clinical strains. We introduced a multicopy vector with or without wild-type soxR into the clinical strains and assayed the basal antibiotic resistance. The soxR vector pSXR significantly reduced the basal antibiotic resistance of strains E17 and M1 to all the antibiotics tested (Fig. 4). The antibiotic resistance of the laboratory soxR constitutive strains JTG936 and JTG1052 was also diminished by the introduction of pSXR, while GC4468 (soxR⁺) and DJ901 (ΔsoxRS) were unaffected (Fig. 4). In contrast, the antibiotic resistance of strain I242 was not affected by pSXR (Fig. 4), consistent with the nonconstitutive nature of the soxR allele in this strain. Strain E19 was not included in this experiment because of its high resistance to all the available antibiotics. Strains I237 and I253 were not included because attempts to transform them with pSXR or pSE380 were unsuccessful.

• Open in new tab
• Download powerpoint

FIG. 4.

Multicopy suppression of antibiotic resistance with wild-type soxR. Clinical isolates E17, M1, and I242, as well as control laboratory strains GC4468 (soxR⁺), DJ901 (ΔsoxRS), JTG936 (soxR^c), and JTG1052 (soxR^c), were transformed with the IPTG-inducible plasmid pSXR (Table 1) or the vector alone. The antibiotic resistance of the control and pSXR-containing strains was assayed on antibiotic gradient plates in the presence of 5 μM IPTG. The maximum antibiotic concentrations per plate were as follows: for strain E17, chloramphenicol (CHL) at 1.5 mg/ml, ciprofloxacin (CIP) at 250 μg/ml, and tetracycline (TET) at 750 μg/ml; for strain M1, chloramphenicol at 40 μg/ml, ciprofloxacin at 5 μg/ml, and tetracycline at 19 μg/ml; for strain I242, chloramphenicol at 20 μg/ml, ciprofloxacin at 1 μg/ml, and tetracycline at 300 μg/ml; for strains GC4468, DJ901, JTG936, and JTG1052, chloramphenicol at 20 μg/ml, ciprofloxacin at 0.05 μg/ml, and tetracycline at 15 μg/ml. The data shown are the means and standard deviations of three determinations.