Fig 1.
Klebsiella pneumoniae infections and cell surface phage receptors.
(A) Classical KP strains are typically associated with pneumonia, urinary tract infections (UTIs), and surgical site infections. Hypervirulent KP strains are associated with invasive infections such as meningitis, soft tissue infections, liver and splenic abscesses, and bacteremia. (B) The primary cell surface receptor for KP phage is the capsular polysaccharide (CPS). Other phage receptors include lipopolysaccharide (LPS), outer membrane porins (OMPs), and conjugative plasmid-encoded pili. (C) Phages bind to cell surface receptors using receptor-binding proteins on their tail fibers. These proteins can contain catalytic domains (e.g., depolymerase domains) that aid in targeting cell surface receptors. This figure was created using BioRender.
Fig 2.
K- and O- locus type diversity across different geographic regions.
16,475 KP genomes derived from human blood, urinary tract, or respiratory specimens were accessed from NCBI on October 16, 2023. Genomes were typed using Kaptive to determine K-locus types (A) and O-locus types (B) [28,29]. The most prevalent types are labeled; all other types are grouped as “Other.” K-loci and O-loci flagged as “unknown” by Kaptive are included in the “Other” category.
Fig 3.
KP-targeting phage receptor dynamics.
(A) Because most phages are dependent on CPS presence to adsorb efficiently, pressure from phage predation selects bacteria with reduced, altered, or no CPS production. Bacteria evolve to resist phages but become sensitized to other phages. (B) The arrangement of phage tail fiber and tailspike genes in a cassette-like organization enables rapid adaption to changes in bacterial epitopes. Phage specificity can rapidly evolve by mutating residues in the catalytic pocket (vertical evolution) or through horizontal gene transfer (HGT). HGT events between phages can result in the acquisition of enzymatic domains and the exchange of tailspike modules.
Fig 4.
Models of co-evolution between bacterial and phage populations.
(A) In the arms-race model, continuous adaptation by bacteria and phage leads to frequent selective sweeps and accumulation of new bacterial resistances and phage infectivities. (B) In the fluctuating selection model, phages maintain a narrow host range with infrequent selective sweeps. This enables the co-existence of multiple phage and bacterial genotypes, whose frequencies are driven by negative frequency-dependent selection and where rare genotypes have a fitness advantage.
Table 1.
KP bacteriophage therapy studies in humans.
Table 2.
KP bacteriophage therapy studies in mouse infection models.
Fig 5.
Looking ahead—The future of phage therapy for KP infections.
(A) Phage-derived enzymes like lysins and depolymerases can degrade peptidoglycans and carbohydrates, including those in the bacterial CPS and KP biofilm matrix. (B) Phages could be used as gene transfer agents to deliver predetermined “cargo” to bacterial cells and use CRISPR-Cas systems to kill directly or edit the bacterial genome. (C) Extracellular Contractile Ejection Systems (eCIS) are syringe-like macro-molecular systems that deliver toxins into adjacent cells. eCIS could be reprogrammed to change their specificity and/or express alternative payload molecules to combat bacterial infections. (D) Phage genome editing can be performed through a process called “recombineering”. Recombineering enables modification, reduction, or broadening of phage host range. (E) Computational tools are being developed to predict interactions between phages and potential bacterial hosts. Machine learning and modeling allow rapid identification of candidate phages for a given bacterial infection, enabling the design of highly specific and optimized phage cocktails for use in clinical settings.