Analysis of selection methods to develop novel phage therapy cocktails against antimicrobial resistant clinical isolates of bacteria

Antimicrobial resistance (AMR) is a major problem globally. The main bacterial organisms associated with urinary tract infection (UTI) associated sepsis are E. coli and Klebsiella along with Enterobacter species. These all have AMR strains known as ESBL (Extended Spectrum Beta-Lactamase), which are featured on the WHO priority pathogens list as critical for research. Bacteriophages (phages) as viruses that can infect and kill bacteria, could provide an effective tool to tackle these AMR strains. There is currently no gold standard for developing a phage cocktail. Here we describe a novel approach to develop an effective phage cocktail against a set of ESBL-producing E. coli and Klebsiella largely isolated from patients in UK hospitals. By comparing different measures of phage efficacy, we show which are the most robust, and suggest an efficient screening cascade that could be used to develop phage cocktails to target other AMR bacterial species. A target panel of 38 ESBL-producing clinical strains isolated from urine samples was collated and used to test phage efficacy. After an initial screening of 68 phages, six were identified and tested against these 38 strains to determine their clinical coverage and killing efficiency. To achieve this, we assessed four different methods to assess phage virulence across these bacterial isolates. These were the Direct Spot Test (DST), the Efficiency of Plating (EOP) assay, the planktonic killing assay and the biofilm assay. The final ESBL cocktail of six phages could effectively kill 23/38 strains (61%) for Klebsiella 13/19 (68%) and for E. coli 10/19 (53%) based on the planktonic killing assay data. The ESBL E. coli collection had six isolates from the prevalent UTI-associated ST131 sequence type, five of which were targeted effectively by the final cocktail. Of the four methods used to assess phage virulence, the data suggests that planktonic killing assays are as effective as the much more time-consuming EOPs and data for the two assays correlates well. This suggests that planktonic killing is a good proxy to determine which phages should be used in a cocktail. This assay when combined with the virulence index also allows phage synergy to inform cocktail design.