Zinc Acetate Potentiates the Action of Tosufloxacin against Escherichia coli Biofilm Persisters

Formation of bacterial biofilms is a major health threat due to their high levels of tolerance to multiple antibiotics and the presence of persisters responsible for infection relapses. We previously showed that a combination of starvation and induction of SOS response in biofilm led to increased levels of persisters and biofilm tolerance to fluoroquinolones.

ABSTRACT

Formation of bacterial biofilms is a major health threat due to their high levels of tolerance to multiple antibiotics and the presence of persisters responsible for infection relapses. We previously showed that a combination of starvation and induction of SOS response in biofilm led to increased levels of persisters and biofilm tolerance to fluoroquinolones. In this study, we hypothesized that inhibition of the SOS response may be an effective strategy to target biofilms and fluoroquinolone persister cells. We tested the survival of Escherichia coli biofilms to different classes of antibiotics in starved and nonstarved conditions and in the presence of zinc acetate, a SOS response inhibitor. We showed that zinc acetate potentiates, albeit moderately, the activity of fluoroquinolones against E. coli persisters in starved biofilms. The efficacy of zinc acetate to increase fluoroquinolone activity, particularly that of tosufloxacin, suggests that such a combination may be a potential strategy for treating biofilm-related bacterial infections.

INTRODUCTION

Bacterial chronic infections are a serious health care problem that is often associated with the formation of highly antibiotic-tolerant biofilms (1). Although multifactorial, this high level of tolerance is at the origin of most of the treatment failure of biofilm-associated infections and is mainly due to the high frequency of persister bacteria (persisters) in biofilms compared to that of planktonic bacteria (2). Persister cells are phenotypic variants that can revert to wild-type (wt) antibiotic susceptibility and are considered to be the source of relapse of biofilm-related infections and reinfection (3). Therefore, there is a strong need to develop effective treatment strategies against biofilm persisters.

Escherichia coli is a versatile species that comprises commensal strains but also pathogenic strains frequently involved in a broad spectrum of intra- and extraintestinal diseases (4). Uropathogenic E. coli bacteria are the most prominent causative agents of urinary tract infections (UTIs) (5). In the case of UTIs, most relapse events are due to regrowth of E. coli persisters after treatment (6). Recently, the fluoroquinolone antibiotic tosufloxacin and the antimicrobial peptide colistin were reported as effective against uropathogenic E. coli persisters obtained from planktonic stationary-phase populations (6). In addition, tosufloxacin has also been reported to be highly active against planktonic Staphylococcus aureus persisters (7). However, since the levels and nature of persisters may differ greatly between planktonic and biofilm cells, the activity of antibiotics against persisters should also be evaluated under biofilm conditions.

We previously showed that starvation stress experienced by biofilm bacteria increases the level of tolerance to fluoroquinolone ofloxacin in E. coli biofilms (8). This increased tolerance of biofilms to ofloxacin upon starvation is dependent on the presence of a functional bacterial SOS response (8). Tolerance to ofloxacin treatment in the stationary phase was also shown to require activation of the SOS response during the posttreatment recovery phase (9). In addition to tolerance, the SOS response has been shown to be involved in horizontal gene transfer, emergence of antibiotic resistance, and toxin production (10). Several candidate inhibitors of the SOS response have been reported, including zinc, for which the molecular details of activity have been well described (11–15). Zinc inhibits the SOS response by interfering with the RecA ATP-binding site, which is essential for RecA activation (14). The role of SOS response in persister generation therefore raises the possibility of using the SOS response as a target to reduce fluoroquinolone persisters and the emergence of biofilm tolerance to antibiotics in clinical settings (11, 16).

In the current study, we hypothesized that inhibition of the SOS response may be an effective strategy to eliminate planktonic and biofilm fluoroquinolone persisters. After evaluating the effectiveness of various antibiotics against E. coli persisters in starved and nonstarved biofilms, we used zinc acetate as an SOS response inhibitor and showed that zinc acetate can be used as an adjuvant to increase fluoroquinolone activities, including that of tosufloxacin, against persisters. This study supports the possibility to use SOS response inhibitors to improve antibiotic activity against biofilm-related infections and persisters.

RESULTS

Starvation enhanced E. coli biofilm antibiotic tolerance in both SOS-dependent and SOS-independent manners.

We previously demonstrated that starved E. coli biofilms display increased tolerance to ofloxacin (8). We wondered whether starvation also promoted increased tolerance when biofilms were treated with other fluoroquinolones and other classes of antibiotics. We assessed the efficacy of various quinolones/fluoroquinolones (nalidixic acid, ofloxacin, norfloxacin, ciprofloxacin, and tosufloxacin), penicillin (ticarcillin), aminoglycosides (amikacin and gentamicin), chloramphenicol, tetracycline, rifampin, and a polypeptide (polymyxin B). Among the different antibiotics tested, polymyxin B was the most effective antibiotic against E. coli TG1 biofilms (whether starved or not), with an observed 6-log reduction in survival (Fig. 1).

FIG 1.

Survival rates of starved and nonstarved biofilms of E. coli K-12 TG1 and TG1 lexAind3 (SOS response-defective mutant) strains when treated with antibiotics. Biofilms of E. coli TG1 and TG1 lexAind3 (TG1 lexAind3 Tet^r for all antibiotics with the exception of tetracycline and TG1 lexAind3 Cm^r for tetracycline) were grown for 24 h in M63B1Gluc and treated for a period of 24 h with one of several antibiotics in M63B1 Gluc medium (black bars; nonstarved) or M63B1 medium without glucose (gray bars; starved). (A) Nalidixic acid (Nal; 640 mg/liter, 80× MIC), ofloxacin (Oflo; 5 mg/liter, 80× MIC), norfloxacin (Norflo; 5 mg/liter, 80× MIC), ciprofloxacin (Cipro; 2.5 mg/liter, 80× MIC), and tosufloxacin (Tosu; 5 mg/liter, 80× MIC). (B) Ticarcilln (Tica; 100 mg/liter, 100× MIC), amikacin (Amik; 320 mg/liter, 80× MIC), gentamicin (Genta; 40 mg/liter, 80× MIC), chloramphenicol (Chloram; 160 mg/liter, 80× MIC), polymyxin B (Polym; 80 mg/liter, 80× MIC), tetracycline (Tet; 40 mg/liter, 80× MIC), and rifampin (Rifam; 1280 mg/liter, 80× MIC). Surviving cells were quantified by viable cell counts. Percentage (%) of survival represents CFU count of viable cells after 24 h of treatment compared to that of untreated biofilms prior to the addition of the antibiotics. Each percentage (%) value is the mean plus or minus standard error of the mean (SEM) from at least three replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

The fluoroquinolones ofloxacin, norfloxacin, ciprofloxacin, and tosufloxacin were also relatively efficient at reducing E. coli TG1 survival in nonstarved biofilms with tosufloxacin as efficient as ciprofloxacin and norfloxacin. Starvation led to significantly increased tolerance of E. coli TG1 biofilms to some antibiotics, including ofloxacin, ticarcillin, ciprofloxacin, norfloxacin, tosufloxacin, nalidixic acid, amikacin, and gentamicin (Fig. 1). In starved biofilms, tosufloxacin was the most efficient fluoroquinolone at killing E. coli TG1 biofilms. The increased survival of E. coli TG1 in starved biofilms treated with fluoroquinolones was abolished in the SOS response mutant (lexAind3 mutant). In contrast, the SOS response did not appear to be involved in the enhanced tolerance of the starved biofilms treated with the other antibiotics, ticarcillin, nalidixic acid, amikacin, and gentamicin. Time-kill curves of 24-h E. coli TG1 biofilms treated by ofloxacin or tosufloxacin showed a typical persistence biphasic shape with a persister plateau reached after 20 h of treatment and an increased SOS-dependent tolerance in starved biofilms (Fig. S1).

The SOS response is required for increased ofloxacin tolerance of starved biofilms during the posttreatment recovery phase.

To firmly correlate the level of tolerance of the starved biofilms treated with fluoroquinolones with the level of induction of the SOS response, we used lexA variant strains that display different abilities to induce the SOS response (17). Inactivation of the self-cleavage of LexA (S119A) totally inhibits the SOS response. Mutations in LexA can either decrease (G80A) or increase (E86P) the rate of self-cleavage relative to that of the wild-type strain (MMR102_F′tet) and thus affect the level of SOS induction. S119A, G80A, wild-type, and E86P strains all display successively increased efficacy in SOS response (17). Tolerance of the stationary phase of liquid culture (Fig. S2) and starved biofilms (Fig. 2) of the different strains to ofloxacin directly correlated to the level of functionality of the SOS response. However, the survival rates of the lexA variants in the stationary phase of liquid culture (Fig. S2) and starved biofilms (Fig. 2) treated by ticarcillin were completely independent of the SOS response.

FIG 2.

Survival rate of E. coli lexA variants treated by ofloxacin or ticarcillin in starved biofilms. Biofilms of E. coli were grown 24 h in M63B1Gluc medium and treated for a period of 24 h in M63B1 with ofloxacin (5 mg/liter, 80× MIC; black histograms) or ticarcillin (100 mg/liter, 100× MIC; gray histograms). Surviving cells were quantified by viable cell counts. All strains carried the F′tet conjugative plasmid. lexA variants S119A and G80A, wt MMR102, and lexA variant E86P strains have gradually increased capacity to induce the SOS response. Percentage (%) of survival represents the CFU count of viable cells after 24 h of treatment compared to that of untreated biofilms prior to addition of antibiotics. Each percentage (%) value is the mean plus or minus standard error of the mean (SEM) from at least three replicates. Asterisks indicate values significantly different from that obtained from the wild-type cells (MMR102_F′tet). *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

It was previously shown that upon fluoroquinolone treatment, the survival of planktonic stationary-phase persisters requires SOS-induced DNA repair during the posttreatment recovery period (9). To further clarify SOS response requirements for survival of persisters in starved biofilms upon fluoroquinolone treatment, we used plasmids expressing recA, the main SOS response activator gene, or lexAind3 (or lexA3), a dominant negative lexA allele, to activate or block the SOS response in a timely manner, respectively (9). Ofloxacin persisters were less abundant in the recA-deficient strain compared to those in the wild-type strain (Fig. 3). In this recA-deficient mutant, induction of RecA in trans during the biofilm formation, ofloxacin treatment, and recovery phases was sufficient to restore levels of persisters comparable to those of wt bacteria and recA-deficient mutant bacteria with recA expressed in trans during all phases in starved biofilms (Fig. 3). With impaired SOS induction, which was achieved with lexA3 expression, we observed a decreased survival compared to that in uninduced controls (Fig. 3). This held true with lexA3 induction during biofilm formation only, ofloxacin treatment only, recovery only, and any combination of the three periods of treatment (Fig. 3). Overall, the results indicated that induction of the SOS response before, during, or after fluoroquinolone treatment during recovery was sufficient to increase persister survival.

FIG 3.

RecA and the SOS response are critical to ofloxacin persistence in starved biofilms, at least during the recovery phase. Biofilms of E. coli (TG1, plain bars, and TG1ΔrecA [TG1ΔrecA::FRT], hatched bars) were grown for 24 h in M63B1Gluc medium and treated for a period of 24 h with 80× MIC of ofloxacin in M63B1 medium without glucose (starved). Vector control corresponds to plasmid pUA66. The recA and lexA3 (lexAind3) alleles were expressed from the P_T5-lac promoter. precA corresponds to the pKV012 plasmid (pUA66 PT5lac-recA), and plexAind3 corresponds to the pKV013 plasmid (pUA66 PT5lac-lexA3). The number of ofloxacin persisters in starved biofilms was decreased in the absence of recA and following the induction of lexA3. The stage at which the induction of lexA3 or recA by adding IPTG (1 mM) was performed is indicated above each set of histograms. Percentage (%) of survival represents the CFU count of viable cells after 24 h of treatment compared to that of untreated biofilms prior to addition of antibiotics. Each percentage (%) value is the mean plus or minus standard error of the mean (SEM) from at least three replicates. Asterisks indicate values significantly different from that of the control strain, which harbored the pUA66 plasmid. Biof, during biofilm treatment; Treat, during ofloxacin treatment; Recov, during recovery phase. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

Zinc acetate potentiates the activity of fluoroquinolones against persisters in starved E. coli biofilms.

To validate the activity of zinc acetate as an SOS inhibitor in our current study settings, we monitored SOS induction in planktonic bacteria upon ofloxacin treatment through the expression of sulA in the presence and absence of zinc acetate (Fig. S3). Increasing concentrations of zinc acetate were used, ranging from 0 to 1 mM, which was below the 1.5 mM MIC of zinc acetate in E. coli (Fig. S4). While ofloxacin induced the SOS response in E. coli, zinc acetate concentrations of ≥0.4 mM significantly reduced the ofloxacin-induced SOS response in a concentration-dependent manner (Fig. S3). We also evaluated the synergistic effect of fluoroquinolones and zinc acetate against E. coli. The results showed that zinc acetate at concentrations of ≥1.0 mM only slightly decreased the MIC values of the fluoroquinolones ofloxacin and tosufloxacin (Fig. S4). Concentrations of zinc acetate higher than 1.2 mM inhibited the growth rate of the E. coli bacteria tested (Fig. S5).

We then tested the ability of three different concentration of zinc acetate (0.5 mM, 0.75 mM, and 1 mM) to potentiate the activity of the fluoroquinolones ofloxacin and tosufloxacin against biofilm persisters upon starvation. Zinc acetate was added at different times during the protocol to assess whether the inhibition of the SOS response at different stages would equally impact the efficacy of zinc acetate as an adjuvant treatment. To mimic potential clinical treatment options, zinc acetate was added after the biofilm was formed, during the posttreatment recovery phase, or at both times. We chose not to assess the adjuvant effect of zinc acetate during biofilm formation, since it would not be clinically applicable for use as an inhibitor prior to biofilm formation being detected. We observed that while zinc acetate alone did not impact the survival of biofilm cells (Fig. S6), its addition to ofloxacin and tosufloxacin treatments significantly but moderately reduced the levels of persisters in starved biofilms when it was added during only the treatment phase, during only the recovery phase, and during both the fluoroquinolone treatment phase and the posttreatment recovery phase (Fig. 4). Regarding potentiation of fluoroquinolone activity against stationary-phase bacteria, we determined that while zinc acetate alone did not impact the survival of stationary cells (Fig. S7), its addition significantly reduced the levels of persisters treated with ofloxacin and tosufloxacin when it was added during only the treatment phase, during only the recovery phase, and during both the fluoroquinolone-treatment phase and the posttreatment recovery phase (Fig. 5). Treatment with 80-fold the MIC of tosufloxacin in combination with 1.0 mM zinc acetate during both the treatment phase and recovery phase actually eradicated the E. coli persisters.

FIG 4.

Zinc acetate decreases the survival rate of fluoroquinolone (ofloxacin and tosufloxacin)-treated starved biofilms. Biofilms of E. coli TG1 were grown for 24 h in M63B1Gluc medium and treated for a period of 24 h with 80× MIC of ofloxacin (5 mg/liter) or tosufloxacin (5 mg/liter) in M63B1 medium without glucose (starved). Zinc acetate was added during fluoroquinolone treatment only, during posttreatment recovery only, or during both fluoroquinolone treatment and posttreatment recovery. Each percentages of survival (%) value is the mean plus or minus standard error of the mean (SEM) from at least nine replicates. Asterisks indicate values significantly different from that of the fluoroquinolone-only treatment group. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

FIG 5.

Zinc acetate decreases the survival rates of fluoroquinolone (ofloxacin and tosufloxacin)-treated E. coli in the stationary phase. E. coli TG1 cells were grown for 24 h in LB broth and treated for a period of 24 h with 80× MIC of ofloxacin (5 mg/liter) or tosufloxacin (5 mg/liter). Zinc acetate was added during fluoroquinolone treatment only, during posttreatment recovery only, or during both fluoroquinolone treatment and posttreatment recovery. Each percentage of survival (%) value is the mean plus or minus standard error of the mean (SEM) from at least three replicates. Asterisks indicate values significantly different from those of the fluoroquinolone-only treatment group. *, P < 0.05; **, P < 0.01; ***, P < 0.001; and ****, P < 0.0001.

DISCUSSION

This study showed that zinc acetate inhibited the SOS response and enhanced fluoroquinolone activity, especially that of tosufloxacin, against E. coli biofilms and persisters. This combination (fluoroquinolone with zinc) is a potential strategy for treating biofilm-related bacterial infections.

Our study showed that, like the last-resort antimicrobial but highly toxic agent polymyxin B, fluoroquinolones, including tosufloxacin, were effective against E. coli persisters in biofilm, although most of the antibiotics tested in this study could not efficiently kill E. coli, especially in starved biofilms. Unlike in E. coli, Kobayashi et al. showed that tosufloxacin was relatively inefficient to kill Pseudomonas aeruginosa biofilm bacteria (18). The toxicity of tosufloxacin on articular cartilage is milder than that of the well-used fluoroquinolones norfloxacin and ciprofloxacin (19), and no other specific toxicity or enhanced toxicity was observed. Tosufloxacin is reported to display high clinical efficacy, especially for the treatment of chronic urinary tract infections (20, 21). To control chronic infections, including UTIs, that can be related to persisters in biofilms, tosufloxacin may be an effective choice.

We showed that starved biofilm tolerance was dependent on a functional SOS response when treated by fluoroquinolones but not when treated by other classes of antibiotics, including beta-lactams, that are known to induce SOS response in E. coli. This points to other mechanisms of tolerance developed by E. coli to sustain these high antibiotic concentrations, such as possible growth arrest causing tolerance to beta-lactams (8). Unlike fluoroquinolones, biofilm tolerance to the quinolone nalidixic acid was relatively high and was not dependent on a functional SOS response. In addition to a potential lower capacity of nalidixic acid compared to that of fluoroquinolones to penetrate into biofilm matrix and biofilm cells, this difference could be explained mechanistically. While nalidixic acid treatment induces the SOS response, drug damages caused by this drug are repaired by the recombination repair system rather than by the SOS DNA repair system (22). In contrast, we and others have shown that tolerance to different fluoroquinolones is absolutely dependent on the SOS response. A study by the group of Kim Lewis has notably shown that SOS-dependent DNA repair genes, such as dinG, ruvAB, and uvrD, were directly necessary for E. coli tolerance to fluoroquinolones (23). Therefore, tolerance to nalidixic acid may be due to SOS-independent recombination repair, while tolerance to fluoroquinolones instead depends on SOS-dependent DNA repair.

Our previous results showed that starvation in biofilm bacteria induced an SOS-dependent ofloxacin tolerance, but that it was independent of the toxin-antitoxin (TA) loci induced by the SOS response (8). Here, using different variants of LexA with gradually increased capacity to induce the SOS response, we demonstrated that the observed levels of tolerance were directly correlated with the levels of SOS induction. Additionally, we showed using recA- and lexAind3-expressing variants, in accordance with Völzing and Brynildsen (9), that SOS induction during the recovery phase was sufficient to increase the survival of biofilm bacteria treated by fluoroquinolone. In addition, we also showed that induction of the SOS response during biofilm formation or during treatment was also sufficient to promote increased survival of biofilm bacteria when exposed to fluoroquinolones.

Accordingly, we showed that the use of an SOS inhibitor, zinc acetate, in combination with tosufloxacin during the antibiotic treatment phase or during the posttreatment recovery phase efficiently killed persisters in starved biofilms and in the stationary phase. We did not evaluate the activity of zinc acetate combined with tosufloxacin during biofilm formation because of the absence of clinical relevance of this situation. While for biofilm treatment there was only a very slight increased gain of efficacy from adding zinc acetate both during treatment and recovery phase compared to only during recovery phase, this gain was much stronger when treating planktonic population and actually enabled eradication of E. coli persisters (80-fold MIC of tosufloxacin combined with 1.0 mM zinc acetate). A similar eradication of another E. coli strain was obtained in a different study using a higher concentration (333-fold MIC) of tosufloxacin (6). These results suggest that the combination of tosufloxacin and zinc acetate may reduce the required dose of tosufloxacin, the duration of antibiotic administration, and the toxicity of tosufloxacin. These results may also have some implications in term of treatment schedule.

The SOS response is involved in bacterial adaptive responses and horizontal gene transfer, potentially leading to the onset of antibiotic resistance in a broad range of bacterial species. Recent studies have shown that zinc is efficient at inhibiting both resistance mutation induced by quinolones in E. coli and Klebsiella pneumoniae (14) and SOS-induced development of chloramphenicol resistance in Enterobacter cloacae (12). Additionally, the same authors also showed that zinc was able to inhibit the transfer of an extended spectrum beta-lactamase (ESBL) gene from Enterobacter cloacae to E. coli (12). At much higher concentrations than the one used in our study, zinc has a direct antibacterial and antibiofilm activity (24). However, zinc toxicity in humans needs to be considered. Zinc acetate is commonly used in human for treating the common cold and zinc deficiency, and it is considered relatively nontoxic for humans. In these situations, it is delivered at the concentration of 150 mg/day with no detected side effect (25). The use of zinc acetate at 1 mM corresponds to 183 mg/liter and 65.3 mg/liter of zinc. This is comparable to zinc concentrations described in the literature to be toxic for humans and that can, for example, cause anemia or neutropenia, corresponding to 100 to 300 mg of zinc per day (25). Thus, it is possible to consider the use of a relatively low volume (5 to 100 ml) of 1 mM zinc acetate to locally help the treatment of biofilm-related infections. While further assays using different in vitro and in vivo models as well as clinical studies are necessary to evaluate efficacy and toxicity of zinc acetate in these situations, local adjuvant strategy of antibiotic treatment of infected wounds, recurrent UTI treatment by bladder instillation together with systemic antibiotic treatment, and the use of zinc acetate during or right after using conservative catheter-lock strategy could be considered in the future. Based on our findings, more specifically, in these clinical situations, a treatment of zinc acetate given concomitantly with the fluoroquinolone treatment or subsequently alone after the antibiotic treatment could reduce the formation of persisters during a potential regrowth of the biofilms. As we showed, using zinc acetate both during the fluoroquinolone treatment and after it, during recovery, may also improve treatment efficacy in targeting both biofilms and planktonic bacteria.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Bacterial strains and plasmids used in this study are described in Table S1. All experiments were performed at 37°C using M63B1 minimal medium containing 0.4% glucose (M63B1Gluc) or Miller’s LB medium (Thermo Scientific, Rochester, NY) unless otherwise specified. Antibiotics were added, when required, at the following concentrations: kanamycin, 50 mg/liter; tetracycline, 15 mg/liter; chloramphenicol, 25 mg/liter; and streptomycin, 100 mg/liter. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

E. coli survival against several antibiotics in starved and nonstarved biofilms.

Test concentrations of the antibiotics used in survival assays were based on MIC values determined using microbroth dilution methods in M63B1Gluc, as previously described (8). The antibiotics tested were ofloxacin, ciprofloxacin, norfloxacin, nalidixic acid, amikacin, gentamicin, chloramphenicol, polymyxin B, rifampin, and tetracycline (all obtained from Sigma-Aldrich), ticarcillin (Ticarpen; GlaxoSmithKline, Marly le-Roi, France), and tosufloxacin (Wako Chemical, Tokyo, Japan). The MIC values for the strains of bacteria tested are shown in Table S2.

In the survival assays, in vitro biofilms were grown in triplicate for 24 h on UV-sterilized 96-well polyvinyl chloride (PVC) plates (Thermo Scientific) as previously described (8). Briefly, stationary-phase cultures were diluted to an optical density at 600 nm (OD₆₀₀) of 0.05 in fresh M63B1Gluc, and 100 μl of this inoculum was used in each well. Biofilms were grown at 37°C in static conditions. After 24 h, each well was washed twice with 1× phosphate-buffered saline (PBS) to remove the unattached and planktonic bacteria. The remaining biofilms were treated for 24 h using 100 μl of M63B1Gluc (nonstarved) or M63B1 without glucose (starved), with or without antibiotics. The specific concentration used for each antibiotic was 80-fold its respective MIC against E. coli strain TG1, except for ticarcillin, for which a 100-fold MIC was used. After 24 h, each well was washed twice with 1× PBS to remove the antibiotic. Surviving bacteria were quantified after vigorous pipetting to detach all biofilm bacteria from the surface of the wells. After serial dilution, 10 μl per dilution of sample was spotted onto LB agar plates and incubated for 24 h at 37°C, and colonies were counted per sample. The total biofilm area cell density prior to antibiotic treatment was 4.9 ± 2.7 × 10⁵ CFU/mm². The CFU values were compared with those from untreated 24-h biofilms and expressed as the percentage of survival.

Time-kill curve of E. coli starved and nonstarved biofilms.

In vitro biofilms were grown in triplicate for 24 h on UV-sterilized 96-well PVC plates (Thermo Scientific). Stationary-phase cultures were diluted to an optical density at 600 nm (OD₆₀₀) of 0.05 in fresh M63B1Gluc, and 100 μl of this inoculum was used in each well. Biofilms were grown at 37°C in static conditions. After 24 h, each well was washed twice with 1× PBS to remove the unattached and planktonic bacteria. Twenty-four-hour biofilms were treated using 80-fold MIC of ofloxacin or tosufloxacin at 1, 2, 4, 8, 20, 24, and 48 h, after which each well was washed twice with 1× PBS to remove the antibiotics. Surviving bacteria were quantified after vigorous pipetting to detach all biofilm bacteria from the surface of the wells. The CFU values were compared with those from untreated 24-h biofilms and expressed as the percentage of survival.

E. coli survival against ofloxacin in planktonic stationary and starved biofilms with or without the SOS response.

To clarify the role of the SOS response in ofloxacin tolerance in the stationary phase of liquid culture and starved biofilms, we determined the survival rate of different strains of E. coli expressing different variants of lexA. Several lexA variants (MMR102-lexA wt, S119A, G80A, and E86P; see Table S1) previously described were kindly provided by Rahul M. Kohli (17). The wt and lexA variant-carrying strains received the F′tet plasmid through conjugation to favor biofilm formation (26, 27) (Table S1).

For testing the ofloxacin and ticarcillin tolerance in the stationary phase of liquid culture, the survival rates of these strains treated with ofloxacin/ticarcillin in the stationary phase of growth were determined as previously described (9), with some modifications. Briefly, these strain cultures were grown in 0.5 ml of Miller’s LB medium overnight at 37°C with shaking (200 rpm). A 30-μl aliquot of this starter culture was used to inoculate 3 ml of Miller’s LB medium and was subsequently grown at 37°C with shaking for another 16 h. To prepare cultures for ofloxacin/tosufloxacin treatment, 25 ml of Miller’s LB medium were inoculated with overnight cultures and grown to an OD₆₀₀ of 0.01. Cultures were then propagated for 20 h at 37°C with shaking. After 20 h, 500 μl of sample was collected, washed with PBS, and the CFU determined. The remaining samples were centrifuged, washed, and treated with 80-fold MIC of ofloxacin or 100-fold MIC of ticarcillin for 24 h in M63B1 medium. After the 24-h treatment, 500-μl samples were collected, washed by PBS, and the CFU determined. The CFU values were compared with those from untreated 24-h samples and the data expressed as the percentage of survival.

For testing the ofloxacin and ticarcillin tolerance in starved biofilm, the lexA variant-carrying strains received the F′tet plasmid through conjugation to favor biofilm formation (26, 27) (Table S1). The survival rates of the conjugated variants against ofloxacin and ticarcillin in starved biofilms were determined as described above.

The survival rates of E. coli TG1 carrying plasmids expressing recA or lexAind3 (Table S1) in the presence of ofloxacin were determined as described above. Induction of recA or lexAind3 was initiated using 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) during several growth phases to determine when starved biofilm persisters in the presence of ofloxacin required the SOS response for survival. The phases considered included the biofilm formation phase, ofloxacin treatment phase, and posttreatment recovery phase. For the biofilm formation phase and ofloxacin treatment phase, 1 mM IPTG was added to the medium. For the posttreatment recovery phase, 10 μl per dilution of detached biofilm was spotted onto LB agar plates containing 1 mM IPTG and incubated for 24 h at 37°C, and colonies were counted per sample.

Evaluation of zinc acetate as an SOS response inhibitor by β-galactosidase assay.

SOS induction was measured using an E. coli strain carrying the sulA SOS promoter fused to lacZ (psulA::lacZ, JJC610) (28). The strain was cultured to an OD₆₀₀ of 0.6 in LB medium. Ofloxacin (0.125 mg/liter) and several concentrations of zinc acetate (Sigma-Aldrich) ranging from 0 to 1 mM were added, and the cultures were incubated for an additional 1 h at 37°C. To determine the expression level of sulA, β-galactosidase activity was directly measured as previously described (29).

Susceptibility and survival of E. coli against ofloxacin/tosufloxacin with or without zinc acetate.

The combination of the fluoroquinolone antibiotics ofloxacin and tosufloxacin with zinc acetate was evaluated using a checkerboard assay. The broth microdilution method used to determine the MIC values in 96-well microtiter plates was modified for the checkerboard assay. Briefly, 50 μl of M63B1Gluc was distributed into each well of the microtiter plates. The ofloxacin/tosufloxacin mixture was 2-fold serially diluted along the abscissa, while the zinc acetate was diluted along the ordinate. An inoculum equal to a 0.5 McFarland turbidity standard was prepared using E. coli TG1 in M63B1Gluc. Each microtiter well was inoculated with 50 μl of bacteria (5 × 10⁴ CFU/well), and the plates were incubated at 37°C for 24 h.

The effect of zinc acetate on bacterial growth was determined by following the optical density of cultures (OD₆₀₀) using a Sunrise Rainbow Thermo microplate reader (Tecan, Kawasaki, Japan). E. coli TG1 cultures were grown in 0.5 ml of Miller’s LB medium overnight at 37°C, and 1 × 10⁵ CFU/ml bacteria were subsequently cultured in 0.1 ml of Miller’s LB medium in 96-well plates at 37°C. The OD₆₀₀ values were measured every 1 h for 12 h.

For the survival assays, the survival rates of E. coli TG1 against ofloxacin/tosufloxacin in starved biofilms were determined as described above. To clarify the effect of zinc acetate on the survival of persisters in starved biofilms treated with ofloxacin/tosufloxacin, several concentrations of zinc acetate were added to the cultures during the ofloxacin/tosufloxacin 24-h treatment phase and/or the posttreatment recovery phase. As a control we also assessed the effect of different concentrations of zinc acetate when added alone for 24 h after biofilm growth and during posttreatment recovery. For the posttreatment recovery phase, 10 μl per dilution of detached biofilm was spotted onto LB agar plates with or without several concentration of zinc acetate, and incubated for 24 h at 37°C, and colonies were counted per sample.

In addition, the survival rates of E. coli TG1 treated with ofloxacin/tosufloxacin in the stationary phase of growth were determined as previously described (9), with some modifications. Briefly, E. coli TG1 cultures were grown in 0.5 ml of Miller’s LB medium overnight at 37°C with shaking (200 rpm). A 30-μl aliquot of this starter culture was used to inoculate 3 ml of Miller’s LB medium and subsequently grown at 37°C with shaking for another 16 h. To prepare cultures for ofloxacin/tosufloxacin treatment, 25 ml of Miller’s LB medium were inoculated with overnight cultures and grown to an OD₆₀₀ of 0.01. Cultures were then propagated for 20 h at 37°C with shaking. After 20 h, 500-μl samples were collected, washed with PBS, and the CFU determined. The remaining samples were treated with 80-fold the MIC of ofloxacin/tosufloxacin for 24 h. After the 24-h treatment, 500-μl samples were collected, washed by PBS, and the CFU determined. The CFU values were compared with those from untreated 24-h samples and the data expressed as the percentage of survival. To determine the effect of zinc acetate on the survival of persisters treated with ofloxacin/tosufloxacin during the stationary phase, several concentrations of zinc acetate were added during the ofloxacin/tosufloxacin treatment phase and/or the posttreatment recovery phase. As a control we also assessed the effect of different concentrations of zinc acetate when added alone for 24 h on the stationary-phase cultures and during posttreatment recovery. For the posttreatment recovery phase, 10 μl per dilution of planktonic bacteria were spotted onto LB agar plates with or without several concentration of zinc acetate and incubated for 24 h at 37°C, and colonies were counted per sample.

Statistical analysis.

Each experiment was performed at least three times. Statistical analysis was performed using a two-tailed unpaired t test. Differences were considered statistically significant for P values of <0.05.