Whole-genome sequencing of Burkholderia pseudomallei from an urban melioidosis hot spot reveals a fine-scale population structure and localised spatial clustering in the environment

Abstract

Melioidosis is a severe disease caused by the environmental bacterium Burkholderia pseudomallei that affects both humans and animals throughout northern Australia, Southeast Asia and increasingly globally. While there is a considerable degree of genetic diversity amongst isolates, B. pseudomallei has a robust global biogeographic structure and genetic populations are spatially clustered in the environment. We examined the distribution and local spread of B. pseudomallei in Darwin, Northern Territory, Australia, which has the highest recorded urban incidence of melioidosis globally. We sampled soil and land runoff throughout the city centre and performed whole-genome sequencing (WGS) on B. pseudomallei isolates. By combining phylogenetic analyses, Bayesian clustering and spatial hot spot analysis our results demonstrate that some sequence types (STs) are widespread in the urban Darwin environment, while others are highly spatially clustered over a small geographic scale. This clustering matches the spatial distribution of clinical cases for one ST. Results also demonstrate a greater overall isolate diversity recovered from drains compared to park soils, further supporting the role drains may play in dispersal of B. pseudomallei STs in the environment. Collectively, knowledge gained from this study will allow for better understanding of B. pseudomallei phylogeography and melioidosis source attribution, particularly on a local level.

Figure 1

The four urban Darwin regions specified by catchment areas and direction of surface runoff at each survey site (a). Levels of seasonal surface runoff and areas of waterlogging are indicated by colour (yellow; nil to low levels of seasonal waterlogging, green stripes; moderate to high levels, blue; severe levels of waterlogging to fully inundated over extended period). B. pseudomallei positive and negative park sites (n = 45, red and green circles, respectively) and positive and negative drain sites (n = 42, red and green squares, respectively) surveyed during the 2017 rainy season (Feb-March). Getis-Ord Gi* Hot Spot Analysis (b–e). Areas with significant “hot” spots are indicated by dark red circles while “cold” spots are shown in green ((b) ST-36, (c) ST-109, (d) ST-327, (e) ST-553). Significant association of Bayes Cluster 1 with drain sites (blue circles – water and soil) as opposed to parks (brown squares - soil) (f). Bayes cluster 1 negative sites are denoted by black circles (ArcGIS v.10.4.1 (ESRI)).

Figure 2

Maximum parsimony phylogeny of 137 B. pseudomallei genomes. MP reconstruction of 134,032 core-genome orthologous SNPs and InDels belonging to environmental survey isolates. MSHR1153 was used as the reference strain and the tree was rooted at MSHR0668. The sample type (a), site type (b), suburb region (c) and MLST type (d) of each isolate are labelled as described. Red circles on branches denote bootstrap values <80.