SOS Response Induction by ß-Lactams and Bacterial Defense Against Antibiotic Lethality

See allHide authors and affiliations

Science  10 Sep 2004:
Vol. 305, Issue 5690, pp. 1629-1631
DOI: 10.1126/science.1101630


The SOS response aids bacterial propagation by inhibiting cell division during repair of DNA damage. We report that inactivation of the ftsI gene product, penicillin binding protein 3, by either β-lactam antibiotics or genetic mutation induces SOS in Escherichia coli through the DpiBA two-component signal transduction system. This event, which requires the SOS-promoting recA and lexA genes as well as dpiA, transiently halts bacterial cell division, enabling survival to otherwise lethal antibiotic exposure. Our findings reveal defective cell wall synthesis as an unexpected initiator of the bacterial SOS response, indicate that β-lactam antibiotics are extracellular stimuli of this response, and demonstrate a novel mechanism for mitigation of antimicrobial lethality.

The ability of bacteria to reduce their susceptibility to antimicrobial drugs importantly affects both bacterial ecology and the treatment of infectious diseases. Previously known mechanisms of bacterial defense against antibiotics include mutation of the drug target, inactivation or destruction of the antimicrobial, and inhibition of antibiotic entry (1). We report a mechanistically novel type of defense mechanism that uses a bacterial two-component signal transduction system to induce the SOS response and temporarily inhibit cell division during exposure to β-lactam antibiotics, consequently limiting the bactericidal effects of these drugs.

Two-component signal transduction systems have a key role in mediating the response of bacteria to environmental stimuli. Normally, receptor-mediated detection of a stimulus at the cell surface leads to autophosphorylation of a sensor kinase component, which then phosphorylates the effector protein component (i.e., the response regulator), enabling the effector to bind to operator/promoter sequences of target genes and either increase or repress transcription (2). DpiA, the effector for the DpiBA two-component system, not only regulates transcription (3) but also regulates DNA replication and segregation by virtue of its uncommon ability to bind to A+T-rich sequences in the replication origins of the E. coli chromosome and certain plasmids (4). Interaction of DpiA with replication origins competes with binding of the replication proteins DnaA and DnaB: When overexpressed, DpiA can interrupt DNA replication and induce the SOS response (4), thereby inhibiting cell division (4, 5).

Previous sequence analysis has suggested that the adjacent E. coli dpiB and dpiA genes (3), like their Klebsiella pneumoniae orthologs citA and citB (6), comprise a polycistronic operon (Fig. 1A). Polymerase chain reaction (PCR) analysis using combinations of primers corresponding to sequences within each of these genes confirmed that dpiB and dpiA are encoded by a common transcript (Fig. 1B). We wished to identify stimuli that activate the dpiBA operon; to monitor such activation, we fused a Hind III–Sma I DNA segment containing the region 5′ to dpiB to a lacZ reporter gene fragment (pHI1508 in Fig. 1A) (3, 7). β-Galactosidase synthesis from this construct was investigated under a variety of growth-limiting conditions known to activate two-component systems and/or the SOS response (including growth in media containing different carbon sources; starvation for O2, PO4, or carbon; heat or cold shock; high salt; exposure to ultraviolet light; culture in stationary phase or in conditioned media; and concentration to a high cell density). Whereas none of the above conditions increased β-galactosidase synthesis, we observed during our experiments that expression of the reporter gene was stimulated by exposure of bacteria to ampicillin and other β-lactam antibiotics (penicillin, cefuroxan, cephalexin, pipericillin) (Fig. 2A) (7, 8). In contrast, none of the non–β-lactam categories of antibiotics we tested activated the dpiBA promoter (Fig. 2A).

Fig. 1.

Structure and transcripts of the dpiBA operon. (A) The dpiB and dpiA genes are aligned in the 5′ → 3′ direction; citC is 5′ of the dpiBA promoter/operator region and is transcribed divergently from it (3). The segment of the dpiBA operon included in plasmid pHI1508 (7) is indicated. (B) Agarose gel electrophoresis showing bands generated by reverse transcription PCR amplification of E. coli SC1088 (26) RNA using pairs of oligonucleotide primers corresponding to sequences within dpiB (lane 1, primers a + b), dpiA (lane 2, primers c + d), or both genes (lane 3, primers a + d). Locations of primer sequences are indicated in (A). (C) Induction of dpi transcripts (7) by ampicillin from SC1088 grown at 30°C in the absence (lanes 1, 3, and 5) or presence (lanes 2, 4, and 6) of ampicillin (4 μg/ml for 4 hours), shown with loading controls.

Fig. 2.

Induction of the dpi operon by β-lactam antibiotics. (A) Expression of the dpi operon as measured in lacZ E. coli strain UT481 by a lacZ reporter fusion to the dpiB promoter on pHI1508 [in Miller units (27)]. Bacteria were grown at 30°C with or without (black squares) antibiotics. Values similar to the control without antibiotics were observed after addition at time zero of kanamycin, streptomycin, spectinomycin, mitomycin C, chloramphenicol, tetracycline, nalidixic acid, rifampicin, vancomycin, or phosphomycin: Open symbols indicate lacZ expression after addition at time zero of ampicillin (squares), cephalexin (diamonds), or pipericillin (circles). [See (7) for concentrations.] All points represent the average of at least three separate experiments. (B) Induction of dpiB/lacZ (as measured by β-galactosidase expression calculated in Miller units) in E. coli JOE339, a lacZ mutant strain that carries an ftsIts mutation. Cells were grown at 30°C with (dashed line and open squares) or without (black squares) ampicillin (4 μg/ml), or at 42°C in the absence of ampicillin (gray circles). (C) Same as (B) but in MC4100, the parent of JOE339. For (B) and (C), bars indicate SD.

Increased expression of the dpiB and dpiA operon by β-lactam treatment was confirmed by quantitative PCR (9) analysis (Fig. 1C), which showed a β-lactam–dependent fourfold increase in dpiBA transcripts encoded by the E. coli chromosome. Consistent with these observations was a concurrent threefold increase in expression of the E. coli citC gene (Fig. 1C), which is divergently transcribed from dpiBA (Fig. 1A) and previously was shown to be upregulated by overexpression of the DpiA protein (3). Similarly, lacZ fusions to promoters found in earlier studies to be activated by DpiA (7) showed DpiA-dependent elevation of expression during treatment with ampicillin (Table 1), further establishing the ability of ampicillin to induce the dpiBA operon. Up-regulation of the dpiB/lacZ fusion by ampicillin was also observed in the dpiA null mutant strain (Table 1), indicating that induction of dpiBA expression by ampicillin does not require the DpiA protein.

Table 1.

Induction of lacZ fusions by DpiA or ampicillin in wild-type (WT) and mutant strains. Ampicillin was added at 10 μg/ml and time points were taken after 2 hours of growth at 37°C. DpiA was overproduced at twice the normal amount from a multicopy plasmid pHI1429 (3). pHI1627 carries a lacZ fusion to pabA, which has been identified as a gene up-regulated by DpiA (7). pHI1508 carries a lacZ fusion to the promoter/operator region of dpiB (7). The lacZ E. coli strain 1088 (26) and null mutations of dpiA (3), recA, and lexA (7) were used. Values are averages of at least three experiments. β-Galactosidase production is indicated in Miller units (27).

lacZ- strain and plasmids Miller units Induction ratio
WT + pHI1627 29 ± 4
    + DpiA overproduction 118 ± 20 4
    + ampicillin 185 ± 20 6.4
dpiA null mutant + pHI1627 16 ± 5
    + ampicillin 18 ± 1 1
WT + pHI1508 52 ± 3
    + DpiA overproduction 138 ± 25 2.8
    + ampicillin 345 ± 50 6.6
dpiA null mutant + pHI1508 27 ± 3
    + ampicillin 113 ± 15 4.1
recA null mutant + pHI1508 24 ± 3
    + ampicillin 80 ± 8 3.3
lexA null mutant + pHI1508 22 ± 6
    + ampicillin 110 ± 10 5

The lethality of β-lactam antibiotics stems from their interaction with transmembrane penicillin binding proteins (PBPs) and the consequent disruption of cell wall integrity (10). Whereas ampicillin binds to all 12 E. coli PBPs (11), pipericillin and cephalexin, which were among the β-lactam drugs we found to increase expression of the dpiBA operon (Fig. 2A), bind only to PBP3 (12, 13), which suggests that PBP3 specifically mediates the β-lactam effect. PBP3 is encoded by ftsI, one of a group of filamentation temperature–sensitive genes implicated in cell division (14), and is a membrane transpeptidase required for peptidoglycan synthesis at the septum generated by cell division (15). Binding of β-lactam antibiotics to PBP3 molecules at the septum inactivates transpeptidase function, leading to lysis of dividing cells in bacterial populations (10).

Inactivation of PBP3 also occurs when cultures of the ftsIts strain, JOE339 ftsI23 (16), are shifted to 42°C (14). We found that shift of JOE339 ftsI23 to 42°C increased expression of the dpi/lacZ reporter gene fusion to a level similar to that observed after addition of ampicillin (4 μg/ml) to the culture medium (Fig. 2B). In contrast, expression from the dpi/lacZ fusion was unchanged at 42°C in the parental strain (Fig. 2C); in a mutant of the rodA gene, which encodes PBP2 (a transpeptidase required for cell wall elongation) [strain S1 (10)]; or in a ts mutant of ftsZ [ftsZ84 in JOE337 (16)], a filamentation temperature–sensitive gene involved in septum ring formation (8). Collectively, these results strongly suggest that inactivation of PBP3 is a stimulus for increased expression of the dpiBA operon.

A biological consequence of DpiA overexpression is induction of the SOS response (4); the extent of such induction can be determined by β-galactosidase synthesis from a lacZ fusion with the SOS-regulated promoter of the sfiA gene [e.g., (17, 18)], which prevents FtsZ polymerization and inhibits cell division when SOS is activated (5). Addition of ampicillin (4 μg/ml) to cell cultures increased lacZ expression from the fusion construct to a level comparable to that observed when DpiA was overproduced from a multicopy plasmid (Fig. 3A) (4). However, we observed no change in sfiA/lacZ expression in bacteria containing a dpiA null mutation (Fig. 3A); this result implies that the increase in sfiA expression by ampicillin requires dpiA function. β-Galactosidase synthesis by the sfiA/lacZ fusion construct was also increased by shifting of the ftsIts strain to 42°C, further establishing the ability of FtsI/PBP3 inactivation to induce SOS (Fig. 3B). This result, which identifies SOS as a response to impaired cell septum synthesis, was also dependent on an intact dpiA gene (Fig. 3B).

Fig. 3.

SOS response induced by PBP3 inactivation. (A) E. coli BR3151, a lacZ mutant strain containing a sfiA/lacZ fusion used to measure the SOS response (5), was grown in the absence (black squares) or presence (open squares) of ampicillin (4 μg/ml). Analogous experiments in the presence of ampicillin (4 μg/ml) used a dpiA (open circles) or recA derivative (open diamonds), which appear as overlapping lines. (B) Expression of the sfiA/lacZ fusion from ftsIts JOE339 (squares) or dpiA JOE339 (diamonds) was followed by measuring β-galactosidase production (in Miller units) in bacteria grown at 30°C in the absence (black solid lines, closed symbols) or in the presence (black dashed lines, open symbols) of ampicillin (4 μg/ml), or at 42°C in the absence of ampicillin (gray lines, closed symbols). Both (A) and (B) are averages of three separate experiments (7).

Mutations in recA or lexA that preclude induction of the SOS response (19) prevented the effects of either ampicillin treatment or temperature shift of the ftsI mutant strain on expression of the sfiA/lacZ fusion protein (Fig. 3A) (8), confirming that sfiA induction by the β-lactam–PBP3–DpiA pathway is SOS dependent. Still further confirmation that the observed activation of sfiA expression by this pathway is due to induction of the SOS response was provided by Western blot data showing that the RecA protein also was elevated by DpiA overproduction and by inactivation of PBP3 through ftsI temperature inactivation or by ampicillin, and that the effect of PBP3 inactivation was dependent on an intact dpiA gene (7). The dependence of β-lactam/PBP3–mediated SOS induction on dpiA contrasted with the lack of effect of the dpiA mutation on RecA expression induced by the DNA damaging agent mitomycin C (7), indicating the distinctive nature of the cell wall–mediated and DNA damage–mediated paths to SOS induction.

Because β-lactam antibiotics kill only bacteria that are dividing (20), we hypothesized that the ability of these drugs to induce the SOS response, and consequently delay cell division by increasing the expression of sfiA (5), may provide temporary protection from β-lactam lethality. We therefore tested wild-type E. coli cells, dpiA null mutant bacteria, bacteria known to be unable to generate an SOS response [i.e., recA mutant cells (19)], and sfiA mutant bacteria (21) for their relative ability to withstand exposure to ampicillin, as measured by survival in cultures containing different concentrations of this antibiotic. Mutation of dpiA, recA, or sfiA increased bacterial susceptibility to killing by ampicillin (Fig. 4): 90% of cells mutated in these genes were unable to form colonies after 1 hour of exposure to ampicillin (3 μg/ml), whereas the same extent of killing of wild-type cells required 1 hour of exposure to ampicillin at 9 μg/ml or 4 hours of exposure at 3 μg/ml. During overnight exposure to pipericillin (2 μg/ml), a PBP3-specific β-lactam, 10 times as many wild-type bacteria as dpiA mutant bacteria survived (0.01% versus 0.001% of cells relative to the number before addition of antibiotic). However, the minimum inhibitory concentration of ampicillin required to permanently inhibit cell growth (1.5 μg/ml) was unchanged by mutation of dpiB, dpiA, or both. Thus, although dpiBA-mediated induction of the SOS response delayed β-lactam antibiotic lethality, it did not reverse the effects of extended exposure to these drugs.

Fig. 4.

Effect of SOS response induction on survival of bacterial cells expressing DpiA during β-lactam exposure. SC1088 wild-type (squares) cultures were exposed to ampicillin at time zero, and percent survival was determined. Survival data are also shown for strains containing null mutations in recA (open diamonds), dpiA (3) (open circles), or sfiA (open triangles). Ampicillin was added at 3 μg/ml (black lines) or 9 μg/ml (gray lines) (7).

Our results indicate a hitherto unsuspected role for the SOS response in temporarily halting cell division when the transpeptidase encoded by the ftsI gene at the cell septum is functionally impaired, and additionally demonstrate a novel role for both the DpiBA two-component system and the sfiA gene in this process. The consequence of dpiBA operon-dependent induction of SOS by β-lactam antibiotics is to mitigate the lethal effects of these drugs on bacteria. Recent evidence indicates that even subinhibitory concentrations of a variety of antibiotics can modulate transcription in bacteria (22), and microarray data suggest that altered expression of SOS and other stress response genes are among the many global changes that can result from exposure to antibiotics (23, 24). Additionally, induction of the SOS response also can affect the interbacterial transfer of genetic material, increasing dissemination of antibiotic resistance among microbial populations (25). The further ability of the SOS response to enhance the survival of bacteria exposed to β-lactams identifies the SOS response as a potential target for drugs aimed at enhancing the efficacy of β-lactam antimicrobials.

Supporting Online Material

Materials and Methods


References and Notes

View Abstract

Stay Connected to Science

Navigate This Article