Tracing the environmental footprint of the Burkholderia pseudomallei lipopolysaccharide genotypes in the tropical “Top End” of the Northern Territory, Australia

The Tier 1 select agent Burkholderia pseudomallei is an environmental bacterium that causes melioidosis, a high mortality disease. Variably present genetic markers used to elucidate strain origin, relatedness and virulence in B. pseudomallei include the Burkholderia intracellular motility factor A (bimA) and filamentous hemagglutinin 3 (fhaB3) gene variants. Three lipopolysaccharide (LPS) O-antigen types in B. pseudomallei have been described, which vary in proportion between Australian and Asian isolates. However, it remains unknown if these LPS types can be used as genetic markers for geospatial analysis within a contiguous melioidosis-endemic region. Using a combination of whole-genome sequencing (WGS), statistical analysis and geographical mapping, we examined if the LPS types can be used as geographical markers in the Northern Territory, Australia. The clinical isolates revealed that LPS A prevalence was highest in the Darwin and surrounds (n = 660; 96% being LPS A and 4% LPS B) and LPS B in the Katherine and Katherine remote and East Arnhem regions (n = 79; 60% being LPS A and 40% LPS B). Bivariate logistics regression of 999 clinical B. pseudomallei isolates revealed that the odds of getting a clinical isolate with LPS B was highest in East Arnhem in comparison to Darwin and surrounds (OR 19.5, 95% CI 9.1–42.0; pbimA genotype bimABm has a similar remote region geographical footprint to that of LPS B. In addition, correlation of LPS type with multi-locus sequence typing (MLST) was strong, and where multiple LPS types were identified within a single sequence type, WGS confirmed homoplasy of the MLST loci. The clinical, sero-diagnostic and vaccine implications of geographically-based B. pseudomallei LPS types, and their relationships to regional and global dispersal of melioidosis, require global collaborations with further analysis of larger clinically and geospatially-linked datasets.