A blueprint for broadly effective bacteriophage-antibiotic cocktails against bacterial infections

Abstract

Bacteriophage (phage) therapy is a promising therapeutic modality for multidrug-resistant bacterial infections, but its application is mainly limited to personalized therapy due to the narrow host range of individual phages. While phage cocktails targeting all possible bacterial receptors could theoretically confer broad coverage, the extensive diversity of bacteria and the complexity of phage-phage interactions render this approach challenging. Here, using screening protocols for identifying "complementarity groups" of phages using non-redundant receptors, we generate effective, broad-range phage cocktails that prevent the emergence of bacterial resistance. We also discover characteristic interactions between phage complementarity groups and particular antibiotic classes, facilitating the prediction of phage-antibiotic as well as phage-phage interactions. Using this strategy, we create three phage-antibiotic cocktails, each demonstrating efficacy against ≥96% of 153 Pseudomonas aeruginosa clinical isolates, including biofilm cultures, and demonstrate comparable efficacy in an in vivo wound infection model. We similarly develop effective Staphylococcus aureus phage-antibiotic cocktails and demonstrate their utility of combined cocktails against polymicrobial (mixed P. aeruginosa/S. aureus) cultures, highlighting the broad applicability of this approach. These studies establish a blueprint for the development of effective, broad-spectrum phage-antibiotic cocktails, paving the way for off-the-shelf phage-based therapeutics to combat multidrug-resistant bacterial infections.

Fig. 1. Repeated exposure of bacteria to phages selects for heritable, constitutive resistance.

a To quantify phage-mediated suppression of bacterial growth, we measured growth curves over a 30-hour period and calculated the “Suppression Index,” defined as the area under the curve (AUC) for the non-phage-treated condition minus the phage-treated condition divided by the AUC for the non-treated condition. Growth curves (b) and Suppression Index (c) for P. aeruginosa strain PA14 under different multiplicity of infection (MOI) of DMS3vir phage. d Suppression Index for PA14 exposed to eight different phages, each at an MOI of 100. Representative plaque assays from three independent experiments are shown above each column. e To quantify resistance upon repeat phage exposure, we collected bacteria following an initial phage exposure (1st exposure) and then re-cultured these bacteria in the absence or presence of the identical phage (2nd exposure). Upon this second exposure, we again generated growth curves over 15 h and determined the “Resistance Index,” defined as the AUC of the phage-rechallenged condition divided by the AUC of the non-rechallenged condition. Growth curves (f) for PA14 after initial exposure to DMS3vir at an MOI of 100 (1st exposure) followed by subsequent exposure to either DMS3vir at an MOI of 100 (2nd exposure, dark blue) or Buffer control (light blue) and the resulting Resistance Index (g). h Resistance Index for PA14 challenged initially with one of eight different phages (1st exposure) and subsequently re-challenged with the same phage (2nd exposure). Representative plaque assays from three independent experiments are shown above each column. The bacteria and phage features depicted on the right side of a and e were created in BioRender. Kim, K. (2023) BioRender.com/z47k977. All reported values for b–d, f, g, and h are depicted from three independent experiments with n = 3 and associated error bars represent standard deviation from the mean. Statistical significance in c is derived from One-way analysis of variance (ANOVA) and significance levels include *P < 0.05, **P < 0.01, and ***P < 0.001. Source data are provided as a Source Data file.

Fig. 2. Constitutive phage resistance confers cross-resistance against “complementarity groups” (CG) of genetically diverse phages.

a The Resistance Index of PA14 initially exposed to one of a set of eight phages and then subsequently re-exposed to each of the same set of eight phages, all at an MOI of 100. All reported values are depicted from three independent experiments with n = 3 and associated error bars represent standard deviation from the mean. b Phage exposure matrix representing the average value of the Resistance Index data presented in a. These data highlight the presence of multiple groups of phages wherein constitutive resistance to one confers PA14 cross-resistance against other phages within that group. In this context, CGs are defined as groups of phages that can have cross-constitutive resistance to one another within that group. c Principal component analysis (PCA) of the data presented in the matrix description of b. d Phylogenic dendrogram for the phages in a, b. The scale (A.U.) is represented as relative evolutionary distance (substitution/site), providing a measure of the genetic relatedness among the phages. Source data are provided as a Source Data file.

Fig. 3. Phages belonging to the same CG utilize the same bacterial receptors.

a Schematic model description of three major surface receptors involved in phage uptake by P. aeruginosa, including Type IV pili (pink), oligosaccharides antigens (OSA) and lipopolysaccharide (LPS, blue), and flagella (green). This schematic model was created in BioRender. Kim, K. (2023) BioRender.com/q97t970. b Representative images of twitching motility assay by PA14, PA14ΔpilA, and PA14 strains exposed to DMS3vir, TIVP-H6, Luz24, and OMKO1 phages. Scale bar = 10 mm. c Suppression Index measured for PA14 (black) and the PA14ΔpilA (pink) by eight different types of phages, each at an MOI of 100. d Colony biofilm phenotypes of PA14, PA14ΔwapR, and the designated phage-exposed strains on Congo red agar medium after 100 hours of growth. Scale bar = 2 mm. e Suppression Index measured for the wild-type PA14 (black) and PA14ΔwapR (blue) by eight different types of phages at an MOI of 100. f Representative images of swimming motility assay by PA14, PA14ΔfliC, and PA14 strains exposed to DMS3vir, TIVP-H6, Luz24, and OMKO1 phages. Scale bars = 10 mm. g Suppression Index measured for PA14 (black) and PA14ΔfliC (green) by eight different types of phages, each at an MOI of 100. Images shown in b, d, and f are based on a minimum of triplicate independent replicates. All reported values for c, e, and g are depicted from three independent experiments with n = 3 and associated error bars represent standard deviation from the mean. Statistical significance in c, e, and g is derived from Two-way analysis of variance (ANOVA) and significance levels include *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Source data are provided as a Source Data file.

Fig. 4. Cocktails containing phages from three or more CG reliably eliminate multi-drug resistant clinical isolates of P. aeruginosa, including biofilm and polyclonal cultures in vivo.

a Growth curves for PA14 treated with phage(s) from DMS3vir, the same CG (DMS3vir + TIVP-H6), two different CGs (DMS3vir + OMKO1), or three different CGs (Luz24 + OMKO1 + PAML-31-1). Notably, the treatment with three phages from different CGs completely prevented growth (highlighted in yellow). All reported values are depicted from three independent experiments (n = 3) and associated error bars represent standard deviation from the mean. b Pooled Suppression Index data for PA14 treated with different phage cocktails; each data point represents as the average Suppression Index value of three independent experiments from a different combination of phages, with the number of phages and number of CGs as indicated. These cocktails are listed in Supplementary Table 2. The error bars represent standard deviation from the mean of pooled Suppression Index data. c Pooled Resistance Index for PA14 treated initially with different phage cocktails and subsequently re-challenged by each component phage that comprises the cocktails, all at an MOI of 100; each data point represents as the average Resistance Index value of three independent experiments. Detailed information on the phages used is in Supplementary Table 2. The error bars represent standard deviation from the mean of pooled Resistance Index data. d Representative three-dimensional renderings of PA14 biofilms after 72 hours in the absence or presence of phage therapy, as indicated. e Pooled data for PA14 cells in biofilms treated with different phage cocktails; each data point represents as the average cell number of three independent experiments. The error bars represent standard deviation from the mean of pooled biofilm data. f Suppression Index for 16 clinical isolates of P. aeruginosa by individual phages Luz24, OMKO1, and PAML-31-1, each at an MOI of 100, or a cocktail of these three phages (KIM-C1) with a combined MOI of 100. g, Suppression Index for these clinical isolates of P. aeruginosa by cocktail KIM-C1 at combined MOIs of 1, 100, or 500. Data in f and g are shown as an average of triplicate independent results. h Suppression Index for monoclonal and polyclonal P. aeruginosa cultures treated individually with KIM-C1 cocktail, its three constituent phages, or KIM-M1 mocktail (a mixture of 3 phages with same CG: DMS3vir, KOR_P1, and KOR_P2) grown in planktonic form or as biofilms. Data are shown as an average of triplicate independent results. The phage dose used here is a combined MOI of 100. i A schematic for wound mice model. This schematic model was created in BioRender. Kim, K. (2023) BioRender.com/s43l916. j Representative wound images of each condition (control, cocktail, and mocktail) over the course of experiments. k CFU of the polyclonal cultures in wounds under the conditions in j. The red line indicates the average value of each condition from at least 15 independent experiments. The black dotted line with “LOD” denotes the limit of detection, which is 100 CFU/g (log(100) = 2). The data under LOD indicates LOD/2 (50 CFU/g or log(50) = 1.7). Statistical significance in b, c, e, and k is derived from One-way analysis of variance (ANOVA) and significance levels include *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Source data are provided as a Source Data file.

Fig. 5. Phage CGs exhibit predictable, class-dependent interactions with conventional antibiotics.

a Exposure matrix for PA14 exposed to combinations of phages and antibiotics simultaneously. Blue areas represent synergy, while red areas represent antagonism. The antibiotic concentrations were below their MIC values (sub-MIC) as detailed in the Supplementary Table 3. CTZ ceftazadime, AZT aztreonam, MER meropenem, GEN gentamicin, TOB tobramycin, AMI amikacin, DOX doxycycline, ERY erythromycin, CIP ciprofloxacin, TMP trimethoprim, COL colistin, and RIF rifampin. Data are shown in the matrix as an average of triplicate independent results. b PCA data for responses to phages and antibiotics. c Effectiveness Ratio (%) of 153 clinical P. aeruginosa isolates treated with the KIM-C1 cocktail (Luz24, OMKO1, and PAML-31-1), AZT treatment alone, or the KIM-C1 cocktail plus aztreonam (KIM-C1 + AZT) grown in planktonic form or as a biofilm. Effectiveness was defined as >60% growth suppression over 30 hours. d Effectiveness Ratio (%) of 153 clinical P. aeruginosa isolates treated with either one of three different phage cocktails (KIM-C1, -C2, and -C3), each containing three or four phages from all four different CGs, or a cocktail plus aztreonam. e Suppression Index for 153 clinical P. aeruginosa isolates treated with the KIM-C1 cocktail, AZT alone, or the KIM-C1 cocktail plus aztreonam (KIM-C1 + AZT) grown in planktonic form or as biofilms. Data are shown in the matrix as an average of triplicate independent results. Source data are provided as a Source Data file.

Fig. 6. Cocktails containing phages from multiple CGs reliably kill multi-drug-resistant clinical isolates of S. aureus and also are effective for polymicrobial infections.

a Schematic model description of two major surface receptors involved in phage uptake by S. aureus. This schematic model was created in BioRender. Kim, K. (2023) BioRender.com/p71c092. b Growth curves for S. aureus strain 1203 following treatment with phage K. c Suppression Index for strain 1203 exposed to five different phages, each at an MOI of 100. All reported values in b and c are depicted from three independent experiments (n = 3) and associated error bars represent standard deviation from the mean. d Phage exposure matrix for strain 1203 initially exposed to one of five phages and then subsequently “re-exposed” to each of the same five phages, all at an MOI of 100. e PCA analysis, and f phylogenic dendrogram for these phages. g Synergic interaction matrix for 1203 exposed to combinations of phages and antibiotics simultaneously. VAN vancomycin 1.0 μg/ml, DOX doxycycline 0.05 μg/ml, ERY erythromycin 1.0 μg/ml, AMP ampicillin 0.1 μg/ml, TMP trimethoprim 1.0 μg/ml, RIF rifampin 0.03 μg/ml. h PCA data for the data in g. i Suppression Index for the individual phages vFB433, vFB009 (100 MOI each), a cocktail (KIM-C4) containing both phages (100 of combined MOIs), vancomycin (VAN), or and phage-antibiotic cocktail (KIM-C4 + VAN) grown in planktonic form or as biofilms. j Effectiveness Ratio (%) of 15 clinical S. aureus isolates to the treatments in i. k Suppression Index for monoclonal versus polyclonal S. aureus isolates cultures (CSA007, CSA009, CSA010) treated individually with a cocktail (KIM-C4) or its two constituent phages (vFB433, vFB009). l Suppression Index for polyclonal S. aureus (CSA007, CSA009, CSA010) or P. aeruginosa (CPA077, CPA086, CPA149) and m Suppression Index for polymicrobial S. aureus + P. aeruginosa isolates (all six strains) cultures treated individually with cocktails of phages with the different CGs or “mocktails” of phages sharing the same CG as controls. P. aeruginosa mocktail (KIM-M1) includes DMS3vir, KOR_P1, and KOR_P2. S. aureus mocktail (KIM-M2) includes vFB433 and vFB468. P. aeruginosa cocktail (KIM-C1) includes Luz24, PAML-31-1 and OMKO1. S. aureus cocktail (KIM-C4) includes vFB433 and vFB009. While the polyclonal S. aureus/P. aeruginosa isolates cultures include three clinical isolates each, the polymicrobial S. aureus and P. aeruginosa cultures include all six isolates. All data shown here are from triplicate independent results. The Suppression Index data in i, k, l, and m are shown as an average of triplicate independent results. Source data are provided as a Source Data file.