An Unbiased Method for Clustering Bacterial Effectors Using Host Cellular Phenotypes

We present a novel method implementing unbiased high-content morphometric cell analysis to classify bacterial effector phenotypes. This clustering methodology represents a significant advance over more qualitative visual approaches and can also be used to classify, and therefore predict the likely function of, unknown effector genes from any microbial genome. As a proof of concept, we use this approach to investigate 23 genetic regions predicted to encode antimacrophage effectors located across the genome of the insect and human pathogen Photorhabdus asymbiotica. Statistical cluster analysis using multiple cellular measures categorized treated macrophage phenotypes into three major groups relating to their putative functionality: (i) adhesins, (ii) cytolethal toxins, and (iii) cytomodulating toxins. Further investigation into their effects on phagocytosis revealed that several effectors also modulate this function and that the nature of this modulation (increased or decreased phagocytosis) is linked to the phenotype cluster group. Categorizing potential functionalities in this way allows rapid functional follow-up of key candidates for more-directed cell biological or biochemical investigation. Such an unbiased approach to the classification of candidate effectors will be useful for describing virulence-related regions in a wide range of genomes and will be useful in assigning putative functions to the growing number of microbial genes whose function remains unclear from homology searching.