Understanding the genetic diversity, antimicrobial resistance, and virulence of Klebsiella pneumoniae bacteria

Klebsiella pneumoniae (Kp) is among the top etiological agents of hospital acquired infections behind only Escherichia coli and Pseudomonas aeruginosa. Due to Kp’s large accessory genome, it rapidly acquires antimicrobial resistance (AMR) genes in response to changing antibiotics used in clinical practice. Carbapenemase producing Kp are a particular problem as the resulting infections have very limited treatment options. Kp is usually described as a nosocomial pathogen, but there are hypervirulent strains of Kp (hvKp) endemic in East Asia, which cause community acquired pyogenic liver abscess. While usually susceptible to antibiotics, hvKp infections disseminate rapidly and require aggressive treatment. In Kp, both AMR and hypervirulence are usually the consequence of horizontally acquired genes. Horizontal gene transfer is mediated by plasmids and might be inhibited by bacterial restriction-modification systems that have been suggested as a bacterial immune system. Restriction-modification systems generate methylation patterns that I have identified in clinical Kp isolates (n=8) using data from the PacBio third-generation sequencing platform. Long reads from this platform allowed me to create complete genome assemblies of these isolates. I have also examined the evolving genomic patterns, including of AMR genes and plasmids, in a longitudinal collection of Kp isolates (n=509; years 1980 to 2019) sourced from Portugal that underwent short-read sequencing on an Illumina platform. The analysis revealed the active transmission of strains with AMR genes. A subsequent analysis of global Kp (n=725) characterised hypervirulence biomarkers in the core and accessory genomes using genome wide association study and machine learning methods. This analysis revealed not only known salmochelin and aerobactin loci, but also other genes putatively linked to hypervirulent phenotype. Extending this work, I applied manifold learning and density-based clustering methods to all publicly available Kp assemblies (n=13,176) to investigate the relationship between carbapenemase genes, hypervirulence genes and plasmids. This analysis identified multiple likely outbreaks of carbapenem resistant hvKp and provided insights into the global dynamics of plasmids and genes they carry. In summary, my work has reinforced the importance of genomics and applied statistical methods to understand Kp hypervirulence, epidemiology, AMR and transmission.