Phage-Antibiotic Synergy Is Driven by a Unique Combination of Antibacterial Mechanism of Action and Stoichiometry

Abstract

The continued rise in antibiotic resistance is precipitating a medical crisis. Bacteriophage (phage) has been hailed as one possible therapeutic option to augment the efficacy of antibiotics. However, only a few studies have addressed the synergistic relationship between phage and antibiotics. Here, we report a comprehensive analysis of phage-antibiotic interaction that evaluates synergism, additivism, and antagonism for all classes of antibiotics across clinically achievable stoichiometries. We combined an optically based real-time microtiter plate readout with a matrix-like heat map of treatment potencies to measure phage and antibiotic synergy (PAS), a process we term synography. Phage-antibiotic synography was performed against a pandemic drug-resistant clonal group of extraintestinal pathogenic Escherichia coli (ExPEC) with antibiotic levels blanketing the MIC across seven orders of viral titers. Our results suggest that, under certain conditions, phages provide an adjuvating effect by lowering the MIC for drug-resistant strains. Furthermore, synergistic and antagonistic interactions are highly dependent on the mechanism of bacterial inhibition by the class of antibiotic paired to the phage, and when synergism is observed, it suppresses the emergence of resistant cells. Host conditions that simulate the infection environment, including serum and urine, suppress PAS in a bacterial growth-dependent manner. Lastly, two different related phages that differed in their burst sizes produced drastically different synograms. Collectively, these data suggest lytic phages can resuscitate an ineffective antibiotic for previously resistant bacteria while also synergizing with antibiotics in a class-dependent manner, processes that may be dampened by lower bacterial growth rates found in host environments.IMPORTANCE Bacteriophage (phage) therapy is a promising approach to combat the rise of multidrug-resistant bacteria. Currently, the preferred clinical modality is to pair phage with an antibiotic, a practice thought to improve efficacy. However, antagonism between phage and antibiotics has been reported, the choice of phage and antibiotic is not often empirically determined, and the effect of the host factors on the effectiveness is unknown. Here, we interrogate phage-antibiotic interactions across antibiotics with different mechanisms of action. Our results suggest that phage can lower the working MIC for bacterial strains already resistant to the antibiotic, is dependent on the antibiotic class and stoichiometry of the pairing, and is dramatically influenced by the host microenvironment.

Keywords: Escherichia coli; adjuvant; antibiotic; bacteriophage; clinical isolate; combinatorial treatment; phage; phage therapy; synergy; synogram; synography.

FIG 1

Effect of the bacterial resistance on phage-antibiotic synergy. A 100-fold diluted subculture of JJ2528 was incubated for 4 h, centrifuged, washed, adjusted to an OD₆₀₀ of 1, and inoculated in a 96-well plate to which different treatments had been added to each well: phage alone (ΦHP3), antibiotic alone, phage-antibiotic combined, and untreated control. The OD₆₀₀ was measured every 15 min for a total of 24 h at 37°C with shaking. (A) Synogram showing different treatments. The effects of antibiotic resistance on the gene and allele levels are shown as follow: chloramphenicol-ΦHP3 combined treatment on chloramphenicol sensitive wild-type JJ2528 (B) and chloramphenicol-resistant JJ2528 with CAT (chloramphenicol acetyltransferase) (C); ceftazidime-ΦHP3 combined treatment on wild-type JJ2528 (D), JJ2528 with β-lactamase CTX-M-14 wild-type that confers resistance against ceftazidime (E), and JJ2528 with β-lactamase CTX-M-14 A77V/D240G that confers increased resistance toward ceftazidime (F). Synograms (t = 24 h) represent the mean reduction percentage of each treatment from three biological replicates: Reduction (%) = [(OD_{growthcontrol} − OD_treatment)/OD_{growthcontrol}] × 100. The regions above the dashed lines indicate antibiotic-mediated killing with highly effective doses; the regions between the solid and dashed lines represent the interacting regions of the phage and antibiotic, and the regions below the solid lines indicate phage-mediated killing with ineffective antibiotic concentrations.

FIG 2

Growth characteristics and interaction plots for phage-antibiotic synergy. Bacterial growth over time was assessed for 24 h in the presence or absence of phage and antibiotic (top), and synergy was assessed via interaction plots (bottom). (A) Combination of phage and antibiotic resulted in additive, synergism, antagonism, and/or no effect. Representative interactions between ΦHP3 and antibiotics on wild-type JJ2528 (antibiotic dose plus phage titer): (B) trimethoprim, 0.5 μg/ml plus 10⁵ PFU/ml and 64 μg/ml plus 10⁹ PFU/ml; (C) colistin, 4 μg/ml plus 10⁸ PFU/ml and 4 μg/ml plus 10⁹ PFU/ml; (D) kanamycin, 16 μg/ml plus 10⁴ PFU/ml and 32 μg/ml plus 10⁹ PFU/ml; (E) ciprofloxacin, 16 μg/ml plus 10⁸ PFU/ml and 16 μg/ml plus 10⁹ PFU/ml; (F) ceftazidime, 2 μg/ml plus 10⁴ PFU/ml and 16 μg/ml plus 10⁹ PFU/ml; (G) chloramphenicol, 4 μg/ml plus 10⁵ PFU/ml and 4 μg/ml plus 10⁹ PFU/ml. Two-way ANOVA was employed for statistical significance testing. *, P < 0.05; **, P < 0.01; ***, P < 0.001; n.s., not significant. Growth curves show means ± standard deviations (SDs).

FIG 3

Effect of antibiotic class on phage-antibiotic synergy. A 100-fold diluted subculture of wild-type JJ2528 was incubated for 4 h, centrifuged, washed, adjusted too an OD₆₀₀ of 1, and inoculated in a 96-well plate coated with ΦHP3 and antibiotics, and the OD₆₀₀ was measured every 15 min for a total of 24 h with shaking. Effect of different antibiotics was studied with: (A) trimethoprim; (B) colistin; (C) kanamycin; (D) ciprofloxacin; (E) ceftazidime; (F) chloramphenicol. Synograms (t = 24 h) represent the mean reduction percentage of each treatment from three biological replicates: Reduction (%) = [(OD_{growthcontrol} − OD_treatment)/OD_{growthcontrol}] × 100. The regions above the dashed lines indicate antibiotic-mediated killing with highly effective doses; the regions between the solid and dashed lines represent the interacting regions of the phage and antibiotic, and the regions below the solid lines indicate phage-mediated killing with ineffective antibiotic concentrations.

FIG 4

Effect of combinatorial treatment on preventing the rise of resistance. ExPEC strain JJ2528 wild-type cells were treated with different titers of ΦHP3-alone (A) and in combination with low (B), intermediate (C), and high (D) doses of antibiotics. Growth curves and bar graphs (t = 24 h) represent means ± SDs. Kruskal-Wallis test was performed, followed by Dunn’s test for multiple comparisons. *, P < 0.05; **, P < 0.01.

FIG 5

Effects of human urine and serum on phage-antibiotic synergy. A 100-fold diluted subculture of wild-type JJ2528 was incubated for 4 h, centrifuged, washed, adjusted to an OD₆₀₀ of 1, and inoculated in a 96-well plate coated with ΦHP3 and ceftazidime. The OD₆₀₀ was measured every 15 min for a total of 24 h for LB and urine and 8 h for serum with shaking. Bacterial cells were cultured in LB (A), pooled human urine (B), pooled human urine plus 10% LB (C), human serum (D), and human serum plus 10% LB (E). Synograms represent the mean reduction percentage for each treatment in urine (N = 3) and serum (N = 2): Reduction (%) = [(OD_{growthcontrol} − OD_treatment)/OD_{growthcontrol}] × 100. The regions above the dashed lines indicate antibiotic-mediated killing with highly effective doses; the regions between the solid and dashed lines represent the interacting regions of the phage and antibiotic, and the regions below the solid lines indicate phage-mediated killing with ineffective antibiotic concentrations.

FIG 6

Effect of genetically similar phages on phage-antibiotic combined therapy. A 100-fold diluted subculture of wild-type JJ2528 was incubated for 4 h, centrifuged, washed, adjusted to an OD₆₀₀ of 1, and inoculated in a 96-well plate coated with phages (ΦHP3 and ΦES12) and antibiotics in LB medium. OD₆₀₀ was measured every 15 min for a total of 24 h with shaking in between. Synograms (t = 24 h) show wild-type ExPEC treated with ΦHP3 (A) and ΦES12 (B) with antibiotics (ceftazidime and ciprofloxacin). Synograms represent the average reduction percentage for each treatment from three biological replicates: Reduction (%) = [(OD_{growthcontrol} − OD_treatment)/OD_{growthcontrol}] × 100. The regions above the dashed lines indicate antibiotic-mediated killing with highly effective doses; the regions between the solid and dashed lines represent the interacting regions of the phage and antibiotic, and the regions below the solid lines indicate phage-mediated killing with ineffective antibiotic concentrations.