Genomics informed sero-epidemiology of Klebsiella pneumoniae

Klebsiella pneumoniae (Kp) is a Gram-negative bacterium and priority antimicrobial resistant (AMR) pathogen, ranked #1 in the World Health Organization's AMR pathogen list. 1 Kp is a major cause of opportunistic healthcare-associated infections globally, and among the top three causes of neonatal sepsis in low resource countries. 2Strains with resistance to the 3rd generation cephalosporins and the carbapenems are increasing in frequency worldwide and cause infections that are extremely difficult to treat.Therefore, there is an urgent need for novel control strategies, and growing interest in the development of vaccines, monoclonal antibody and phage therapies.The capsule (K) and outer-lipopolysaccharide (O) are the two major Kp surface polysaccharides.Both are pathogenicity factors and major targets for vaccines and monoclonal antibody therapies; however, there is some debate over the potential efficacy of anti-O strategies due to the possibility that the capsule will block anti-O antibodies. 3he capsule also serves as a major binding receptor for anti-Klebsiella bacteriophage, wherein most can bind only one or a subset of capsule types. 4ighty-two K phenotypes are recognised by serological typing techniques, 5 but more than 160 have been predicted from genome data based on distinct gene content within the capsule synthesis (K) locus. 6,7Similarly, nine O phenotypes were recognised by traditional typing approaches, but genomics data indicate >20 distinct O polysaccharides can be produced through combinations of distinct O loci and genes elsewhere in the genome that are known to play a role in O polysaccharide modification. 8Given that protein-polysaccharide conjugate vaccines and monoclonal antibody therapies have limited valency (up to 20-25 antigens in a single vaccine 3 ), and phage cocktails generally cannot exceed 5-10 phage, there is a clear need to understand the epidemiology of K and O types in order to prioritise them as targets.K and O loci can be rapidly identified from Kp whole-genome sequences using the bioinformatics tool, Kaptive, 6,7 which is available as a stand-alone command line tool, and integrated into the Kp-specific genomic typing tool, Kleborate. 9For broader accessibility, Kaptive was also made available as the graphical web tool, Kaptive Web (kaptiveweb.erc.monash.edu)and has been integrated into the Pathogenwatch genomic surveillance platform (pathogen.watch).Kaptive can identify K and O loci from draft genome assemblies with very high accuracy (98%), 7 and a recent comparison of >700 matched genomes with serological K typing data indicated high concordance (85%), but highlighted that there remain gaps in the knowledge-base about the K genotype-phenotype relationship. 5enomic evolutionary analyses have revealed that K and O diversity differs among common Kp sub-lineages, or 'clones' (referenced here by the dominant 7-gene multi-locus sequence type, ST).Specifically, clones associated with so-called 'hypervirulent' communityacquired Kp infections have carried KL1 (encoding K1, found in ST23) or KL2 (encoding K2, ST86, ST65) throughout their evolutionary histories. 10In contrast, the common globally distributed multidrug resistant (MDR) healthcare-associated clones display diverse K and O loci, with acquisition of novel loci driven through chromosomal recombination. 10However, despite this diversity, each of the major MDR clones is dominated by just one or two K loci e.g.KL107 among ST258, 9,10 KL51 and KL64 among ST231. 11 While it is well documented that hypervirulent infections are dominated by K1 and K2-producing Kp, healthcare-associated infections have been associated with a diverse array of K types.In a 12-month study at the Alfred Health Network (Melbourne, Vic, Australia), 91 distinct K loci were represented among 294 unique Kp infections, reflecting the high diversity of clones capable of causing disease (n = 179 STs detected). 12The top 16 K loci accounted for just 50% of all infections.In contrast, only 12 predicted O types were detected, and the top four accounted for >50% of infections, reflecting a comparatively lower level of diversity among O antigens that is replicated across broad geographies. 9,11,13,