evergreen 2023 logo
phage therapy bacterial resistance Escherichia coli lysis dynamics physiological characteristics phage cocktails

PCA-based phage clustering as a powerful tool for effective cocktail preparation

Abstract ID: 65-MC

Tomoyoshi Kaneko 1,3*, Toshifumi Osaka 2, Satoshi Tsuneda 1,3

  1. Department of Life Science and Medical Bioscience, School of Advanced Science and Engineering, Waseda University
  2. Department of Microbiology and Immunology, Tokyo Women’s Medical University
  3. Phage Therapy Institution, Waseda University

Background

In recent years, the effectiveness of phage therapy as a treatment strategy for bacterial infections has been reevaluated. Phage cocktails, which consist of a mixture of multiple phages, are used in phage therapy to delay the emergence of resistant bacteria. One approach to preparing highly effective phage cocktails involves using phages that target different receptors. There are generally two main methods for identifying target receptors: (1) creating phage-resistant strains and investigating the mutations between wild-type and resistant strains to study the functions encoded by those genes, and (2) creating a comprehensive library of knockout mutant strains for all (nonessential) gene and examining the ability of phage to infect these strains. Both methods require significant efforts, with the former necessitating advanced knowledge of genomics, such as comparing the whole genome sequences of resistant strains and identifying mutation-induced gene functions, and the latter requiring the preparation of a comprehensive library of mutant strains of the host bacteria that a particular phage infects. Therefore, in this study, we used Escherichia coli as a target bacterium and investigated the preparation method of a highly bactericidal phage cocktail based on the physiological characteristics of E. coli phage strains, independent of identifying target receptors.

Methods

We isolated phages that lyse E. coli derived from fecal samples of mice with colitis using the double-layer agar method. A total of 29 phage isolates were obtained from the sewage of four wastewater treatment plants. For each phage isolate, the lysis dynamics of E. coli were measured using optical density (OD) curves. Thirteen phages that exhibited clear lytic properties were selected, and the time at which the OD started to decrease was recorded as the lysis onset time. The lysis duration was calculated by subtracting the lysis onset time from the time when the OD started to increase again, indicating the emergence of resistant bacteria. In addition to these values, the physiological characteristics of phage, including the adsorption rate constant, burst size, titer after one-day cultivation, and lysis spectrum, were investigated. These measured values were normalized and subjected to principal component analysis. Furthermore, the Euclidean distance was calculated using the principal component values of each phage, and hierarchical clustering was performed. A total of 78 phage cocktails were created using pairs of phages belonging to the same or different clusters, and their lytic activity was evaluated based on OD measurements. The change in lysis duration due to cocktail formation, referred to as the extension ratio of lysis duration, was calculated by dividing the lysis duration of the cocktail by the longer lysis duration among the constituent phages. The relationship between the cluster membership of the phages composing the cocktail and the extension ratio of lysis duration was investigated. Additionally, whole genome analysis was performed for each phage, including calculation of Average Nucleotide Identity (ANI) to determine genomic similarity, as well as comparison of tail fibers.

Results and Discussions

When clustering based on physiological characteristics into five clusters and grouping based on genomic similarity (ANI > 95%), the results were consistent. When considering the tail fiber, which provides information similar to the target receptor, we were able to divide the phages into five groups, similar to the physiological clustering and genomic similarity. However, the boundaries between these groups were more ambiguous. When the phages in a cocktail belonged to different clusters, there was a tendency for the extension ratio to be higher. Particularly, among the cocktails with an extension ratio of two or more, indicating a lysis duration more than double, 13 of them were cocktails composed of phages from different clusters, and among those, three cocktails had an extension ratio exceeding six. These results suggest the necessary conditions for designing effective phage cocktails.