Using machine learning to predict protein-protein interactions between a zombie ant fungus and...


Ian Will, William C. Beckerson, Charissa de Bekker


bioRxiv, 2022



Abstract: Parasitic fungi are known to produce proteins that modulate virulence, alter host physiology, and trigger host responses. Due to their various effects on the host, these proteins are often considered a type of “effector” molecule, many of which act via protein-protein interactions (PPIs) between host and parasite. Cross-species PPIs between Ophiocordyceps camponoti-floridani and Camponotus floridanus likely underlie aspects of infection and parasitic behavioral manipulation in this host-parasite relationship. The fungal parasite Ophiocordyceps (zombie ant fungus) manipulates Camponotus (carpenter ant) behavior to induce a summit disease phenotype: the infected host ascends and attaches to an elevated position that promotes fungal growth and transmission. Machine learning approaches offer high-throughput methods to produce mechanistic hypotheses on how this behavioral manipulation occurs. Using D-SCRIPT to predict host-parasite PPIs, we found ca. 6,000 interactions involving 129 parasite genes that were previously found to be upregulated during manipulated summiting behavior. We analyzed the predicted PPIs to identify overrepresentation of functional annotations among these participating proteins hypothesized to mediate Ophiocordyceps-Camponotus interactions. Additionally, we compared these PPIs to predictions made by testing Camponotus proteins with secretomes from three fungi that differ from Ophiocordyceps by degrees of lifestyle and phylogeny, to find which PPIs were unique for Ophiocordyceps-ant interactions. Intriguingly, Ophiocordyceps-Camponotus specific PPIs carried strong signals for neuromodulatory G-protein coupled receptors in the host and frequent involvement of unannotated small secreted Ophiocordyceps proteins in a variety of PPIs. We also detected less specific overrepresentation of Camponotus oxidation-reduction, structural, and gene-regulatory proteins, in addition to Ophiocordyceps proteases. These hypothesized interactions include potential fungal drivers of ant infection and manipulation, whether by increasing, decreasing, or modulating host protein activity. As such, this work provides a springboard for targeted molecular investigations of the mechanisms underlying fungal manipulation of ant behavior.


Keywords: Effectors, Behavioral Manipulation, Network Analysis


https://doi.org/10.1101/2022.09.09.507359




0 views0 comments