Omic AI reveals new autophagy regulators from the Atg1 interactome in Saccharomyces cerevisiae

In <i>Saccharomyces cerevisiae</i>, Atg1 is a core autophagy-related (Atg) protein kinase (PK) in regulating macroautophagy/autophagy, by physically interacting with numerous other proteins, or by phosphorylating various substrates. It is unclear how many Atg1-interacting partners and substrates are also involved in regulating autophagy. Here, we conducted transcriptomic, proteomic and phosphoproteomic profiling of Atg1-dependent molecular landscapes during nitrogen starvation-triggered autophagy, and detected 244, 245 and 217 genes to be affected by <i>ATG1</i> in the autophagic process at mRNA, protein, and phosphorylation levels, respectively. Based on the Atg1 interactome, we developed a novel artificial intelligence (AI) framework, inference of autophagy regulators from multi-omic data (iAMD), and predicted 12 Atg1-interacting partners and 17 substrates to be potentially functional in autophagy. Further experiments validated that Rgd1 and Whi5 are required for bulk autophagy, as well as physical interactions and co-localizations with Atg1 during autophagy. In particular, we demonstrated that 2 phosphorylation sites (p-sites), pS78 and pS149 of Whi5, are phosphorylated by Atg1 to regulate the formation of Atg1 puncta during autophagy initiation. A working model was illustrated to emphasize the importance of the Atg1-centered network in yeast autophagy. In addition, iAMD was extended to accurately predict Atg proteins and autophagy regulators from other PK interactomes, indicating a high transferability of the method. Taken together, we not only revealed new autophagy regulators from the Atg1 interactome, but also provided a useful resource for further analysis of yeast autophagy.