Implementing N-terminomics and machine learning to probe in vivo Nt-arginylation

N-terminal arginylation (Nt-arginylation) is a degradation signal in the ubiquitin-proteasome and autophagy-lysosomal pathways, but its study has been limited by technical challenges. Here, we developed an integrated approach combining N-terminomics with machine learning-based filtering to identify in vivo Nt-arginylation. By using Arg-starting missed cleavage peptides as proxies for ATE1-mediated arginylation, we trained a transfer learning model to predict MS2 spectra and retention times. By applying the prediction models with an additional statistical filter, we identified 134 Nt-arginylation sites in thapsigargin-treated HeLa cells. Arginylation was enriched in proteins from various organelles, especially at caspase cleavage and signal peptide processing sites. Several proteins were further validated for their interaction with p62 ZZ domain. Temporal profiling revealed that ATF4 increased early post-stress, followed by arginylation at caspase-3 substrates and ER signal-cleaved proteins. Our approach enables sensitive detection of rare N-terminal modifications, offering potential for biomarker and drug target discovery.