Sublethal Concentrations of Ciprofloxacin Induce Bacteriocin Synthesis in Escherichia coli

Abstract

Antibiotics that interfere with DNA replication, as well as cell wall synthesis, induce the SOS response. In this report, we show that ciprofloxacin induces synthesis of colicins, narrow-spectrum antibiotics frequently produced by Escherichia coli strains, in an SOS-dependent manner.

It is well documented that the SOS response is induced by antibiotics that interfere with DNA replication (13, 22), as well as cell wall synthesis (15). Furthermore, the SOS response has been shown to induce Shiga toxin production (9) and the lateral transfer of antibiotic resistances encoded by the Vibrio cholerae integrating conjugative element, SXT (2, 7). In this study, we investigated the influence of subinhibitory concentrations of ciprofloxacin, a fluoroquinolone, on bacteriocin production in Escherichia coli. The bacteriocins of E. coli are designated colicins and are active against cells of the same and closely related species (6). They are found with high frequency among natural isolates (5). Colicin synthesis is characteristically regulated by the LexA protein, the key regulator of the SOS response (4). Recently, colicins have been shown to have an in vivo antagonistic role promoting microbial diversity within E. coli populations in the mammalian colon (10) and the potential to promote microbial genetic diversity (25). The results of our study show that sublethal concentrations of ciprofloxacin induce colicin expression in an SOS-dependent manner and imply that SOS-inducing antibiotics could thus affect microbial strain diversification, as well as promote the acquisition and dissemination of antibiotic resistance.

The assay used in the experiments presented here employed a gene fusion of the colicin K activity gene, cka, and the promoterless lacZ carried by plasmid pIK471 (11). To determine whether ciprofloxacin induces cka expression, the β-galactosidase activity of the cka-lacZ fusion was monitored when cells were grown in the presence of subinhibitory concentrations of the antibiotic (16). The MIC for ciprofloxacin was determined using the broth and agar dilution methods (19). Overnight cultures of strain RW118 [thr-1 araD139 Δ(gpt-proA)62 lacY1 tsx-33 supE44 galK2 hisG4 rpsL31 xyl-5 mtl-1 argE thi-1 sulA211] (3) carrying plasmid pIK471 were diluted 1:500 in LB medium and grown with aeration at 37°C. At an optical density at 600 nm of 0.3, the culture was divided into three parts. Ciprofloxacin was added to two of the parts to obtain concentrations of 1/8 MIC (0.008 μg/ml) and 1/16 MIC (0.004 μg/ml), while the third part served as a control without ciprofloxacin; the cultures were then incubated further. At these doses, the growth rates of cultures with and without the antibiotic were comparable, and induction was highest. Induction was observed in ciprofloxacin-treated cultures, with a significant lag (approximately 3 h) between treatment and induction. Induction was particularly noted during stationary phase, with an approximately threefold increase in β-galactosidase activity (Fig. 1). A lag in induction was previously observed for the colicin E1 activity gene cea upon treatment with UV light. Those authors reasoned that as colicins are released by cell lysis, the delay in induction may enable cells with limited damage to repair DNA and reestablish repression before lethal induction of the colicin lysis gene occurs (23). Alternatively, the delay in induction could be due to an indirect effect of ciprofloxacin on cka expression, or subinhibitory concentrations of ciprofloxacin could exert a general effect on stationary-phase gene expression.

FIG. 1.

Expression of the cka-lacZ fusion with and without ciprofloxacin in RW118 and in the recA-defective mutant RW464. Levels of β-galactosidase activity in Miller units in RW118 (solid symbols) without ciprofloxacin (squares), with ciprofloxacin added at 1/8 MIC (circles), and with ciprofloxacin added at 1/16 MIC (triangles). Levels of β-galactosidase activity in Miller units in RW464 (open symbols) without ciprofloxacin (squares), with ciprofloxacin added at 1/8 MIC (circles), and with ciprofloxacin added at 1/16 MIC (triangles). The dotted lines represent growth of both strains with and without ciprofloxacin. The experiment was performed three times, and the means ± standard errors of the mean (error bars) are shown.

To determine whether induction of cka expression is SOS dependent, the β-galactosidase activity of the cka-lacZ fusion was followed, with and without ciprofloxacin, in the recA-defective strain RW464 (3), which cannot elicit an SOS response. As is evident from Fig. 1, no induction was observed in the presence of the antibiotic.

To further resolve the basis of cka induction, we investigated the effect of ciprofloxacin on expression of another SOS-regulated gene, sulA, and an SOS-independent gene, bolA. SulA is synthesized in large amounts during the SOS response and inhibits cell division by binding to FtsZ, the major component of the cell division machinery (8). BolA is involved in adaptation to stationary-phase growth, as well as heat shock, acid stress, oxidative stress, and sudden carbon starvation (12, 24). Our experiments showed that subinhibitory concentrations of ciprofloxacin induced a rapid, approximately 25-fold increase in expression of a sulA-lacZ fusion in strain ENZ1257 (17) while expression of a bolA-lacZ fusion in strain VIP36 (1) was not affected (Fig. 2). The induction kinetics of the cka-lacZ and the sulA-lacZ fusions were compared, and the results revealed much faster induction of the latter. As has been postulated for the colicin E1 gene, cea, the lag in induction of the cka gene most likely enables cells to repair low levels of damage prior to induction of the lethal colicin lysis gene. Furthermore, the delay in cka induction could be related to its effective suppression, established by binding of LexA to two overlapping SOS boxes present in the cka regulatory region, as well as in ColE operons (14). We can thus conclude that ciprofloxacin induces cka expression due to induction of the SOS response.

FIG. 2.

Expression of the sulA-lacZ and bolA-lacZ fusions with and without ciprofloxacin in strains VIP36 and ENZ1257, respectively. Levels of β-galactosidase activity in Miller units from the sulA-lacZ fusion (solid symbols) without ciprofloxacin (squares), with ciprofloxacin added at 1/8 MIC (circles), and with ciprofloxacin added at 1/16 MIC (triangles). Levels of β-galactosidase activity in Miller units from bolA-lacZ (open symbols) without ciprofloxacin (squares), with ciprofloxacin added at 1/8 MIC (circles), and with ciprofloxacin added at 1/16 MIC (triangles). The dotted lines represent growth of both strains without and with ciprofloxacin. The experiment was performed three times, and the means ± standard errors of the mean (error bars) are shown.

To verify that ciprofloxacin induces colicin K synthesis, an in vivo biological assay was employed. Colicin synthesis was monitored with and without ciprofloxacin by a colicin production assay. Strain RW118 harboring the colicin K-encoding plasmid pKCT1 (18), as well as RW464 recA with pKCT1, was cultivated as described above to the stationary-phase optical density at 600 nm of approximately 5, when expression is highest. One-milliliter samples were collected and treated by sonication for 30 seconds. Cell debris was removed by centrifugation at 15,000 × g for 3 min. The supernatants, crude colicin extracts, were then used for colicin assays. A twofold dilution series of colicin extracts was prepared. Subsequently, 10 μl of the diluted extracts was placed in wells in an LB plate and overlaid with 0.7% soft agar with 100 μl of the indicator strain AB1133. As is evident from Fig. 3, ciprofloxacin induces a significant increase in colicin K levels in the wild-type strain, while in the recA-defective strain, no induction was observed. To determine whether ciprofloxacin induces synthesis of other colicins, strains BZB2101, BZB2104, and BZB2123, producing three pore-forming colicins, A, E1, and N, respectively, as well as strain BZB2120, producing the DNase colicin E7, were tested for induction. Indeed, for all of the strains, subinhibitory concentrations of ciprofloxacin were shown to induce synthesis (Fig. 3). To evaluate the levels of induction of the individual colicins, the sizes of the lysis zones were compared and dilution of the crude colicin extracts was taken into account. On the basis of the results of three experiments, we could conclude that at both of the tested concentrations, ciprofloxacin induces a ≥16-fold increase in colicin K and E1 levels. For colicin A, a ≥30-fold increase at 1/8 MIC and a ≥16-fold increase at 1/16 MIC were determined. Colicin E7 exhibited a ≥30-fold increase at both concentrations, while colicin N showed an approximately 16-fold increase at 1/8 MIC and an approximately 2-fold increase in colicin levels at 1/16 MIC. The results of all three experiments were comparable, except for colicins A and E7, for which a ≥30-fold increase at 1/8 MIC was determined in two of the three experiments, while in one, a ≥16-fold increase of colicin levels was determined. With regard to induction of colicin K, a discrepancy between results obtained using the cka-lacZ fusion, where only a threefold induction was observed, and results using the colicin extract was evident. A similar inconsistency was observed in a previous study (18) and was found to be due to loss of plasmid pIK471 upon profound upregulation of cka-lacZ expression.

FIG. 3.

Colicin production assay of cells harboring colicin-encoding plasmids with and without ciprofloxacin. The indicated concentrations of ciprofloxacin added to the colicin-producing cultures are 0 (none), 1/8, and 1/16 MIC. Aliquots (10 μl) of twofold serial dilutions of colicin extracts were placed into wells in an LB plate and overlaid with soft agar harboring the indicator strain, as described in the text. (a) colicin N, (b) colicin K, (c) colicin E1, (d) colicin K expression in recA-defective strain RW464, (e) colicin E7, and (f) colicin A.

To investigate the significance of the SOS response in the induction of colicin synthesis by ciprofloxacin, all of the plasmids encoding the studied colicins were transferred to the wild-type RW118 and recA-defective RW464 strains by transformation. Immunity to the colicin produced allowed selection of transformants. Therefore, prior to plating, 1 ml of the appropriate crude colicin extracts was added to the transformation mixtures and incubated for 1 h. Using this procedure, only wild-type RW118 harboring the individual plasmids was isolated. To allow more efficient selection, transposition of Tn3, encoding resistance to ampicillin, from the conjugative plasmid pHly152-T8 (20) to the colicin-encoding plasmids was performed (16). The recA-defective RW464 transformants carrying the colicin-encoding plasmids were then isolated by selection for ampicillin resistance. Subsequently, as for colicin K, no induction of colicins A, E1, N, and E7 by ciprofloxacin was observed in the recA-defective strain RW464 (data not presented).

Recently, in vivo experiments with strains producing a pore-forming colicin, E1, and colicin E2 with endonucleolytic activity provided evidence that colicins are antagonistic agents within E. coli populations (10). Furthermore, monitoring the transcriptional response of E. coli to colicins E9, an endonuclease, and E3, an RNase, has shown that the former induces the SOS response while the latter upregulates expression of DNA integrases, invertases, and recombinases. These results indicated that colicins also have the potential to promote microbial diversity through the induction of error-prone DNA polymerases, gene transfer, and DNA rearrangements (25). Thus, our results, even though anticipated, reinforce the need for great caution in the use of SOS-inducing antibiotics. Such antibiotics not only promote the dissemination of antibiotic resistance genes and the production of toxins regulated by repressors sensitive to RecA cleavage, but also promote colicin synthesis, which is frequently encoded by pathogenic and nonpathogenic E. coli strains of human and animal origin (5, 21).