Background

Like many groups working in infection biology, we have gained considerable insights from performing transcriptomics experiments on our pathogen of interest. In our case this has been enterohaemorraghic and enteropathogenic Escherichia coli. A caveat for all these experiments is the growth condition used – is it really relevant to a real infection? Hence, for a long time, we have had a strong desire to try and perform transcriptomic analyses using samples derived from animal-based infections. Sounds simple, but there were a few hurdles. E. coli isn’t a natural pathogen of many small animals, so most models use antibiotics to help reduce the microbiota, thereby allowing an infection. Yet, the microbiota and their metabolites, are known to be important factors in affecting the disease process. We felt the experiments would be most informative if they were performed in a natural infection model. Therefore, we turned to Citrobacter rodentium, a murine pathogen used as a surrogate for E. coli infection. At the University of Glasgow, we collaborate with Dr Gill Douce who has considerable experience in many animal models of infection, including the Citrobacter-murine model. We were also fortunate to be supported for our sequencing costs through a Wellcome Trust funded initiative at our University called the Catalyst grant. 

Findings
The transcriptomic data provided insights into how Citrobacter causes an infection. We could see for the first time the dynamic expression of some of the phage encoded effectors that are secreted via the Type Three Secretion System. We hit a bit of luck here too. We knew the leading group on effector proteins led by Professor Gad Frankel at Imperial College London, had already been looking at some of these effectors. When we contacted Gadi he was immediately keen to collaborate and had some excellent data on the cell biology associated with the effectors. It is also fair to say we faced some challenges. The biggest one was associated with the postdoc who did the majority of the work, Dr James Connolly. James was actually funded to work on an entirely different project, so the vast majority of the work in this paper was carried out on top of a full time and already packed programme of science. But, it was clear this was an exciting story and we were determined to make the most of it.

Our decision to use the Citrobacter-murine model was definitely a good one. It allowed us to find a host-induced metabolic pathway that affected the regulation of virulence. The expression of the Type 3 Secretion System and a suite of associated effectors was found to be fine-tuned co-ordinately through a unique mechanism involving metabolism of 1,2-propanediol, a metabolite derived from commensal species of the microbiota. This was an important discovery for us that revealed novel insights into virulence regulation and also highlights the fact that without the application of a natural infection model we likely would have missed this important link. 

The work was also considerably strengthened by the review process. One question we were asked to address was whether the propanediol pathway was more important in its role as a regulator of virulence or in a purely metabolic manner contributing to the pathogens’ fitness. Addressing this question added a new level of clarity and insight to our study. Having the opportunity to publish the reviews alongside the paper is important to us, as it shows the transparency of the process and role the reviewers played in helping to improve the story as a whole. 

https://www.nature.com/articles/s41467-018-06701-4

Artwork above by Eliza Wolfson, https://lizawolfson.co.uk. The transcriptional landscape of Citrobacter rodentium during infection. The data is illustrated by colored nodes representing gene ontology terms associated with differential expression in vivo.