The emergence of multidrug-resistant strains of Klebsiella pneumoniae poses a serious threat to public health. This forces us to search for alternative therapies including phage-borne capsule depolymerases. These enzymes are capable of degrading bacterial capsules, making cells accessible for antimicrobials and the immune system. Wide distribution and high diversity of prophage sequences in K. pneumoniae genomes could be explored for tailspikes/tail fibers with potential capsule-degrading activity.

In silico analyses of 99 K. pneumoniae genomes allowed us to identify ORFs encoding prophage proteins with putative depolymerase activity and with typical β-helical structure present in lytic phage depolymerases (structures predicted by AlphaFold2 software). Out of 31  ORFs selected, 5 depolymerases were successfully produced and defined in terms of substrate specificity. In addition, by site-directed mutagenesis, we identified crucial amino acids responsible for catalysis. We made assumption that substrate specificity could be predicted based on the host strain serotype, which was correct for most cases. However, two proteins showed activity against serotypes other than expected. Depolymerases 248gp38 and 914gp77 were active against K46 serotype instead of K55 and K32 serotype instead of K62, respectively.

Our results suggest that K. pneumoniae prophages may be a rich source of depolymerases, but the substrate specificities cannot be directly predicted based on host strain serotype, but it should be verified experimentally.