Improvement of a therapeutic phage cocktail against multidrug-resistant Pseudomonas aeruginosa using directed evolution
Kevin Burke 1*, Olga Kirillina 1, Yonas Alamneh 1, Caitlin Urick 1, Nino Mzhavia 1, Lillian Musila 2, Paphavee Lertsethtakarn 3, Wanwen Su 1, Rania Abu-Taleb 1, Alexander Bobrov 1, Mikeljon Nikolich 1, Andrey Filippov 1
- Wound Infections Department, Bacterial Diseases Branch, Walter Reed Army Institute of Research, Silver Spring, MD
- Department of Emerging Infectious Diseases, United States Army Medical Research Directorate-Africa, Nairobi, Kenya
- Armed Forces Research Institute of Medical Sciences, 315/6 Rajvithi Rd., Bangkok, Thailand
Antibiotic resistance represents a growing and serious global challenge, with clinicians struggling to provide effective antibacterial treatments. Multidrug-resistant (MDR) strains of P. aeruginosa are spreading throughout the world and cause especially difficult-to-treat infections, with antibiotics losing efficacy. Phage therapy represents a powerful tool to supplement or possibly replace failing antibiotics for the treatment of these MDR infections. Phages offer attractive characteristics for therapeutics, including replication at the site of infection and minimal off-target killing because of the high strain specificity of most phages. However, such tight specificity makes development of durable, off-the-shelf phage cocktail therapeutics with broad killing spectra difficult to generate. In this work, we employed an in vitro directed evolution approach (the Appelmans protocol) to develop phages with broader host ranges against a panel of global, diverse MDR clinical strains. As a result we generated, isolated and characterized a phage clone with expanded host range and incorporated it into a phage cocktail that provided therapeutic efficacy in a mouse model of wound infection.
We employed the phage training approach known as the Appelmans protocol, in which we mixed three genetically similar phages with complimentary lytic activity and passaged against a 12-strain panel of P. aeruginosa phage-resistant clinical isolates and a phage-permissive host. Following multiple rounds of phage training, we were able to isolate phage clones with altered and/or expanded host ranges when tested against a 100-strain global diversity panel of MDR P. aeruginosa clinical isolates. Altogether, we isolated 10 phage variants that were plated on a previously pan-phage-resistant clinical strain of P. aeruginosa, MRSN 20176. Of these 10 phages, six had expanded host range, while the remaining four showed host ranges either roughly equal to or reduced compared with the parental phages. We sequenced the phage clones and characterized recombination and SNP events. Upon deeper analysis, we observed four major sites of recombination and the presence of numerous SNPs when compared with the most closely similar parental phage genome. We identified point mutations in genes predicted to encode tail fibers. We additionally identified mutations and recombination events in genes predicted to encode DNA polymerase and RNA polymerase and other phage structural proteins. We tested the stability of phages with expanded host ranges to determine whether the observed spectral expansion would be retained after repeated propagation on a single host. Phage clones 20176-4, 20176-5, 20176-6, and 20176-7 were passaged against phage-permissive strain PAO1 serially over five rounds. Lysates were then plated, and three clones were collected from each, purified, and re-plated against the 100-strain global diversity set to assess host range. Of these, phage subclones collected from 20176-4 showed an increase in host range, while the remaining clones showed either retention of host range or a reduction. We selected phage subclone 20176 4-2, a subclone of 20176-4, for use in phage cocktail development. Phage 20176 4-2 was assessed for compatibility and stability in mixes with other top performing phages from our library, assessed for in vitro killing efficacy, and then, endotoxin was removed from the preparation for incorporation into a phage cocktail. The phage had stable titer over two months of storage, was compatible with other cocktail phages, and rapidly killed susceptible P. aeruginosa strains. A new phage cocktail containing the trained phage was designated PAM3T. PAM3T was derived from therapeutic phage cocktail WRAIR_PAM3, which consists of six P. aeruginosa phages and has a combined host range of 83% of the global diversity panel. We removed two components of WRAIR_PAM3 and replaced them with 20176 4-2. PAM3T consists of only five phages but has expanded activity, with a combined host range of 85% of the diversity panel. PAM3T was then tested in a mouse dorsal wound model of infection with P. aeruginosa strain PAO1::lux, with four phage doses provided daily both topically and intraperitoneally. While 7/8 (87.5%) of control mice treated with saline died, the phage treatment protected 100% of mice from lethal septicemic infection. PAM3T reduced the burden of P. aeruginosa PAO1::lux in infected wounds to undetectable levels within two days, and phage-treated wounds closed as rapidly as by Day 14, versus by Day 21 for positive control mice treated with ceftazidime.
In conclusion, we demonstrated that a directed in vitro evolution approach can yield phages with expanded or altered host ranges, and that these expanded host range clones can provide powerful additions to therapeutic phage cocktail design by reducing the number of phages required for a broad host range cocktail and by providing activity directed at phage-resistant strains or clonal groups. Our trained phage 20176 4-2 had a stably expanded host range and compatibility with top performing phages in our library, and phage cocktail PAM3T that included this phage showed high therapeutic efficacy in a mouse wound infection model.