Samples were lysed in 0.5% RapiGest (Waters, Sydney, Australia) in 50 mM NH4HCO3, followed by 3 freeze-thaw cycles in liquid nitrogen.20 (link) Samples were then reduced (60℃, 1 hour; 5 mM dithiothreitol; Chem-Supply), alkylated (room temperature, 1 hour; 15 mM iodoacetamide; Merck, Sydney, Australia) and proteolyzed (37℃, overnight; Trypsin Gold; Promega, Sydney, Australia). Peptides were purified using Oasis HLB solid-phase extraction cartridges (Waters) and resuspended in 0.1% aqueous formic acid (Sigma). Mass spectrometry was performed on a nanoAcquity UPLC connected to a SYNAPT G2-Si (HDMS) spectrometer (Waters). Data was processed using MassLynx software (Waters) and Progenesis QI for Proteomics (version 4.1; Waters). Peptides were searched against the UniProt human protein FASTA database, with variable modifications: carbamidomethyl (C), deamidation (N and Q), oxidation (M), and propionamide (C). Relative protein quantitation was performed with the Hi-N method using the top 3 peptides. Proteins were identified if expressed in at least 2 technical replicates in at least 1 sample. Protein interaction networks with a minimum interaction score of 0.4 were generated using the human STRING database (v11.0).21 (link)