Kingdom-wide Discovery of Type VI Secretion Effectors via AI Uncovers Three New Bacterial Toxins
Akerman-Arad, A.; Danov, A.; Chausho, Y.; Friedman, A.; Benedict, S.; Shalom, M.; Fisher, T.; Fraenkel, R.; Mamou, G.; Oppenheimer-Shaanan, Y.; Tzarum, N.; Levy, A.
Show abstract
The type VI secretion system (T6SS) is a molecular harpoon used by 19% of bacteria to inject toxic proteins into microbial or host cells. Despite its widespread distribution, many T6SS effector (T6Es) genes remain unidentified. Here, we developed an XGBoost classifier integrating 42 genetic and biochemical features to identify novel T6Es. Applying our model to 2,559 T6SS-encoding bacterial genomes revealed 19,958 T6E candidates. We validated our computational predictions and confirmed that three effector candidates, having new 3D structures, demonstrated cytotoxicity to bacteria and yeast, mostly in the bacterial periplasm. For two of these, cognate immunity proteins neutralizing toxicity have been validated. T6EC11, a new peptidoglycan degrading effector, quickly leads to spherical E. coli, with inner membrane separation, and gradual peptidoglycan degradation in a unipolar manner, demonstrating an intriguing moon-like structural phenotype. T6EC12, an effector toxic to bacteria and yeast, carries a conserved zinc-binding motif. Mutations the motif markedly reduced toxicity, suggesting that T6EC12 represents a novel metalloprotease. Our study expands the known T6E repertoire, and provides thousands of T6E predictions in all T6SS-encoding bacteria.
Matching journals
The top 6 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- FliO is an evolutionarily conserved yet diversified core component of the bacterial flagellar type III secretion system 95%
- Structure of a Tc holotoxin pore provides insights into the translocation mechanism 95%
- Panacea: a hyperpromiscuous antitoxin protein domain for the neutralisation of diverse toxin domains 94%
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.