Temperate phage-antibiotic synergy across antibiotic classes reveals new mechanism for preventing lysogeny

Abstract

A recent demonstration of synergy between a temperate phage and the antibiotic ciprofloxacin suggested a scalable approach to exploiting temperate phages in therapy, termed temperate phage-antibiotic synergy, which specifically interacted with the lysis-lysogeny decision. To determine whether this would hold true across antibiotics, we challenged Escherichia coli with the phage HK97 and a set of 13 antibiotics spanning seven classes. As expected, given the conserved induction pathway, we observed synergy with classes of drugs known to induce an SOS response: a sulfa drug, other quinolones, and mitomycin C. While some β-lactams exhibited synergy, this appeared to be traditional phage-antibiotic synergy, with no effect on the lysis-lysogeny decision. Curiously, we observed a potent synergy with antibiotics not known to induce the SOS response: protein synthesis inhibitors gentamicin, kanamycin, tetracycline, and azithromycin. The synergy results in an eightfold reduction in the effective minimum inhibitory concentration of gentamicin, complete eradication of the bacteria, and, when administered at sub-optimal doses, drastically decreases the frequency of lysogens emerging from the combined challenge. However, lysogens exhibit no increased sensitivity to the antibiotic; synergy was maintained in the absence of RecA; and the antibiotic reduced the initial frequency of lysogeny rather than selecting against formed lysogens. Our results confirm that SOS-inducing antibiotics broadly result in temperate-phage-specific synergy, but that other antibiotics can interact with temperate phages specifically and result in synergy. This is the first report of a means of chemically blocking entry into lysogeny, providing a new means for manipulating the key lysis-lysogeny decision.IMPORTANCEThe lysis-lysogeny decision is made by most bacterial viruses (bacteriophages, phages), determining whether to kill their host or go dormant within it. With over half of the bacteria containing phages waiting to wake, this is one of the most important behaviors in all of biology. These phages are also considered unusable for therapy because of this behavior. In this paper, we show that many antibiotics bias this behavior to "wake" the dormant phages, forcing them to kill their host, but some also prevent dormancy in the first place. These will be important tools to study this critical decision point and may enable the therapeutic use of these phages.

Keywords: antimicrobial resistance; bacteriophage; lysis-lysogeny; phage-antibiotic synergy; temperate phage.

Fig 1

Temperate phage-antibiotic synergy across quinolones. Checkerboard assay of HK97 with quinolones: (A) nalidixic acid, (B) oxolinic acid, and (C) levofloxacin. Endpoint growth relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap. Readings at 9 and 12 h. Checkerboard assay of HK97 and oxolinic acid after 9 h (D) or 12 h (E). Growth relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap. Area under the curve readings. Checkerboard assay of HK97 and (F) nalidixic acid, (G) oxolinic acid, (H) levofloxacin, and (I) ciprofloxacin. Area under the curve relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap. (J) Checkerboard assay of HK97 and ciprofloxacin in recA mutant. Area under the curve relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap.

Fig 2

Temperate phage-antibiotic synergy across SOS-inducing antibiotics. Checkerboard assay of HK97 (A) mitomycin C, (B) trimethoprim, (C) ceftazidime, (D) ampicillin, (E) cefotaxime, and (F) cefixime. Color is kept consistent across drugs of the same class. Area under the curve relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap. Checkerboard assay of HK97 and (G) mitomycin C, (H) trimethoprim, and (I) ceftazidime in recA mutant. Area under the curve relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap.

Fig 3

HK97 lysogen sensitivity to β-lactams and lysogeny frequency. MIC in liquid culture for wild-type E. coli K-12, and lysogen control was tracked after challenging with serial dilutions of (A) ampicillin, (B) ceftazidime, (C) cefotaxime, and (D) cefixime (means ± SD, n = 3 biological replicates, each with three technical replicates). (E) Percentage of lysogen and non-lysogen survivors from 20 colonies after overnight HK97 and β-lactams challenges averaged. Concentrations were selected where we have seen the highest difference in AUC between phage challenge and phage + antibiotic treatment. (F) Phage quantification after 18 h when unchallenged and when challenged with MIC and 1/2 MIC concentrations of ampicillin, cefixime, ceftazidime, and cefotaxime, (means ± SD, n = 3 biological replicates). Significance for panel F was calculated using two-way ANOVA with no significant difference. Shapes represent different biological replicates.

Fig 4

Protein synthesis inhibitors result in temperate phage-antibiotic synergy. Checkerboard assay of HK97 and (A) gentamicin, (B) kanamycin, (C) tetracycline, and (D) azithromycin. AUC relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap. (E) Bars show the average number of survivors relative to untreated cultures in three biological replicates, each of three technical replicates. Each biological replicate is represented by its own shape: circle, square, or triangle. Limit of detection (10 CFU/mL) is represented at all points of MIC and 1/2 MIC tPAS data, as no counts were obtained, except “square” at 1/2 MIC. Error bars depict the SD, while ***P from 0.001 to 0.0001 from a one-way ANOVA and Tukey’s post hoc test. (F) Bars show the observed effect (green) versus the expected (gray) effect determined by multiplying the effect of the phage and antibiotic alone. Average from the three biological replicates for each observed tPAS data from Fig. 4E was compared to the calculated expected effect at the corresponding antibiotic concentration using a paired t test, *P ≤ 0.05. Checkerboard assay of HK97 and (G) gentamicin and (H) tetracycline in a recA mutant. Area under the curve relative to untreated bacterial control, averaged among three biological replicates, plotted as a heatmap.

Fig 5

Mechanism of gentamicin HK97 synergy. (A) Percentage of lysogen and non-lysogen survivors after overnight HK97 and gentamicin challenges averaged. PCR was performed in duplicate for confirmation. (B) Phage adsorption of survivors. Bars showing the percentage of adsorbed phage in four survivors from 1/4 MIC gentamicin phage challenge; non-lysogens are in gray and orange is the single lysogen survivor. Error bars represent SD. Percent was compared using two-way ANOVA and Tukey’s multiple comparison tests, *P ≤ 0.05. (C) MIC in liquid culture for wild-type E. coli K-12, and lysogen control was tracked after challenging with serial dilutions of ciprofloxacin (means ± SD, n = 3). MIC is ~1.024 µg/mL for wild type and lysogen. (D) Growth curves in liquid culture of wild-type E. coli K-12 and lysogen control was tracked in the absence and presence of gentamicin at 1/2 MIC, averaged among three biological replicates ± SD. (E) Phage quantification at time 0 for lysogen control and after 2 and 6 h when challenged with serial dilutions of gentamicin, averaged among three biological replicates, each of three technical replicates. Error bars represent SD. Each biological replicate is represented by its own shape: circle, square, or triangle. (F) Percentage of lysogens and a representation of bacterial growth tracked over time for the HK97 or the HK97 and gentamicin challenge using qPCR at five time points (n = 3 performed in biological triplicates) for 1/2 MIC antibiotic. Significance of the results was studied using two-way ANOVA and Tukey’s multiple comparison post hoc test, ****P ≤ 0.0001.

Fig 6

Antibiotics synergizing with temperate phages. Antibiotics that activate the SOS response act as prophage inducers and synergize with temperate phages (top right). Protein synthesis inhibitors also synergize with temperate phages but do so by blocking entry to lysogeny (bottom, center). While some β-lactams do show synergy with temperate phages, mechanistically this appears to be independent of the lysis-lysogeny decision (L) Created with BioRender.com.