Wyres lab

Klebsiella pneumoniae is a global priority antimicrobial resistant pathogen associated with 750,000 deaths annually. As rates of resistance to the ‘drugs of last resort’ continue to increase, infections are increasingly difficult to treat, and there is a growing need for new prevention strategies and therapeutics.

Many bacterial treatment and prevention strategies target or exploit metabolic processes. Over the past decade it has become clear that K. pneumoniae are very diverse, including substantial diversity of genes associated with metabolic processes, but until now very little was known about the extent of phenotypic metabolic diversity within this species.

In our recent PLoS Biology paper we use a combination of comparative genomics and genome scale metabolic modelling to evaluate metabolic diversity among K. pneumoniae and closely related Klebsiella species. Leveraging our high-throughput metabolic modelling pipeline, Bactabolize, and validated pan-reference model, we predicted growth outcomes in >1000 different conditions for each of >7000 individual bacterial strains.

Our data showed that there is considerable diversity of substrate usage- particularly in the sources of carbon that can be used for growth- and that this diversity is structured in the population. While there was variation within and between sub-lineages, each of 48 sub-lineages we explored in detail was associated with its own metabolic profile (Figure). These differences allowed strains from different sub-lineages to ‘cross-feed’ in laboratory experiments, which we propose may contribute to their coexistence in natural populations.

Importantly, the carbon sources that could not be ubiquitously used by all sub-lineages, included those implicated in mammalian gut colonisation- an essential pre-requisite for infection, and target for novel infection prevention strategies.

Figure caption: Sub-lineage specific growth phenotypes. Heatmap showing frequency of variable substrate usage across 48 global K. pneumoniae sub-lineages (SL). Sub-lineage labels are coloured to indicate those that are globally-distributed causes of disease. Substrate shown along X-axis. The element source of each substrate is semi-colon separated and abbreviated for brevity: C = Carbon, N = Nitrogen and S = Sulfur. O2 indicates aerobic conditions. Frequency of substrate usage within sub-lineages is indicated by shading. Substrates are grouped to indicate those that can be used ubiquitously among all or almost all SLs (majority SL-specific core), those that are core among many SLs (common SL-specific core), those that are core among only a minority of SLs (rare SL-specific core) and those  are not core to any SL (SL-specific accessory). Where available, validated substrate-specific growth prediction accuracies are indicated in coloured circles below the X-axis.

So what does this mean?

This work has implications for microbial ecology studies, which generally assume that bacterial species form metabolically homogeneous groups- an assumption that our data clearly shows is erroneous. What’s more, the data highlight that studies seeking to target or exploit K. pneumoniae metabolism must consider and test a broad diversity of clinically relevant strains. The metabolism of one or a small number of strains is not representative of the entire population and should not be used to estimate population impacts.

This work was led by former post-doc in the lab, Dr Ben Vezina, and builds on the tremendous efforts of former PhD student, Dr Helena Cooper, who curated our K. pneumoniae pan-reference metabolic model. The project also included our first foray into the world of metabolomics, to confirm the presence of metabolites that were predicted to support cross-feeding between sub-lineages. Huge thanks to Chris Barlow at the Monash Proteomics and Metabolomics Platform for helping us with this.