An unbiased proteomic platform for ATE1-based arginylation profiling

Abstract

Protein arginylation is an essential post-translational modification catalyzed by arginyl-tRNA-protein transferase 1 (ATE1) in mammalian systems. Arginylation features a post-translational conjugation of an arginyl to a protein, making it extremely challenging to differentiate from translational arginine residues with the same mass. Here we present a general ATE1-based arginylation profiling platform for the unbiased discovery of arginylation substrates and their precise modification sites. This method integrates isotopic arginine labeling into an ATE1 assay utilizing biological lysates (ex vivo) rather than live cells, thus eliminating ribosomal bias and enabling bona fide arginylation identification. The method has been successfully applied to peptide, protein, cell, patient and mouse samples, with 235 unique arginylation sites revealed from human proteomes using 20 µg of input. Representative sites were validated and followed up for their biological functions. This global platform, applicable to various sample types, paves the way for functional studies of this difficult-to-characterize protein modification.

Fig. 1. Isotopic arginine labeling and detection strategy for ATE1 substrates and arginylation sites.

a, Scheme of arginyl installation onto proteins by ribosomal synthesis (translational) and ATE1 (post-translational). b, Arginylation profiling platform for arginylation site and substrate discovery from biological samples. Lysate is labeled by isotopic arginine molecules (Arg¹⁰ and Arg⁰), respectively, using ATE1 assay, mixed and digested. Peptides are fractionated and analyzed by mass spectrometry in data-dependent acquisition mode. Proteomics data are searched to produce peptide identifications (IDs), among which peptide pairs modified by H and L Arg are further evaluated for MS1 isotopic features. c, Isotopic arginylation of a peptide ATE1 substrate. EICs were extracted using monoisotopic peaks based on calculated m/z values. EIC in black indicates chromatography of the unmodified peptide. d, Isotopic feature in MS1 spectra and their summary (upper, center and lower: 75, 50 and 25 percentiles, respectively). The first monoisotopic peaks in technical MS1 scans (n = 81) from a representative LC–MS run is set to 1 for normalization. Relative intensities of other isotopic peaks (n = 81, 80, 61, 28, 8, 81, 81, 79, 53, 25 and 3 technical scans) are displayed. A 10-ppm error is set for all MS1 isotopic peaks. e, Ratio summary of MS1 pairs in four replicates (n = 4) using doublet, quartet and sextet peaks. Detailed ratios from the replicates are provided. The numbers of pairs in quartet and sextet are normalized to the numbers of pairs in doublet from respective replicate. Nle, norleucine. Source data

Fig. 2. Arginylation analysis of whole-proteome peptides using in-house software.

a, Software flowchart of customized ArginylomePlot. Input data are mzXML and peptide identification files. Output data are MS1, MS2 and the summary of unbiased arginylation sites. b, Experimental workflow for arginylation of tryptic peptides from HEK293T (ATE1 KO) cells. c, Numbers of identified MS1 pairs, arginylation sites, unique peptides and unique proteins. d, H/L ratio distribution of all MS1 pairs and their corresponding numbers of MS1 scans (doublets, error ≤10 ppm). The retention times of peptide IDs belonging to a pair were averaged. The software will extract matching MS1 scans within 1.25 min (±1.25 min) of the averaged retention times. e, Analysis of the N-terminal residues of all unique arginylated peptides. Q is arginylated after deamidation (Q → E). f, Sequence logo calculated from unmodified forms of all unique arginylated peptides. The frequency plot is generated by WebLogo. The peptide sequences were aligned and extended to include 14 positions downstream of the arginylation sites as the P1′ position. g, Comparison of tryptic and nontryptic N-terminal of all arginylated peptides. A nontryptic N-terminal may indicate endogenously exposed N-termini of proteins.

Fig. 3. Summary of arginylation sites and ATE1 substrates in human proteins and proteomes.

a, Experimental workflow for CALR arginylation. b, Numbers of identified MS1 pairs, arginylation sites, unique peptides and unique proteins in human cells and patient tissues. c, Overview of arginylation sites in human cells and patient tissues. d, Ratio plot of all MS1 pairs detected in all sample fractions. e, Sequence logo calculated from unmodified forms of all unique arginylated peptides. Frequency plots are generated by WebLogo. The arrow indicates the cleavage site before arginylation. f, Arginylation type comparison of all unique sites. g, Biological function analysis of ATE1–protein substrates using PANTHER. iPS, induced pluripotent stem. CF, cardiac fibroblasts. CM, cardiomyocytes. R, arginylation. R_deami, N/Q arginylation after deamidation. PD, Parkinson’s disease. PDD, Parkinson’s disease with dementia. AD, Alzheimer’s disease. Source data

Fig. 4. Validation of representative arginylation sites.

a, Synthetic peptide validation of seven arginylation sites. b, Representative MS2 spectra indicating RM12 arginylation on Cys45 with tri-oxidation (C45 RO3) and its MS1 summary (upper, center and lower: 75, 50 and 25 percentiles, respectively). Its first monoisotopic peak in MS1 scans (n = 44 technical scans) from a representative run is set at 1 for normalization. Relative intensities of other isotopic peaks (n = 43, 38, 30, 18, 2, 44, 44, 40, 30, 9 and 0 technical scans) are displayed. c, Open E24 in ERO1A and E17 in SSBP are arginylated in an in-bacteria arginylation system. Protein and ATE1 are coexpressed in E. coli for in-bacteria arginylation. The protein was purified and digested for proteomics analysis. d, Arginylation dependency of CALR, ERO1A and SSBP sites on endogenous ATE1 (n = 1 biological replicate). Anti-ATE1 was used to detect the expression of endogenous ATE1. β-tubulin was used as a loading control. Anti-Flag antibody was used to detect expressions of CALR, ERO1A and SSBP proteins. EICs of arginylated peptide in each protein in WT and KO cells after pull-down and proteomics are provided. e, Relative arginylation levels of CALR, ERO1A and SSBP sites after cooverexpression of ATE1. Protein was purified by antibody pull-down experiment followed by proteomics analysis. Peak areas of arginylated peptides were normalized to the sample with the highest signal and relative ratios are displayed. Different amounts (0, 0.5 and 1 µg) of ATE1 plasmids were used for transfection. β-tubulin was used as a loading control. Anti-Flag antibody was used to detect expressions of CALR, ERO1A and SSBP proteins. KO, ATE1 KO; Std, standard peptide. Source data

Fig. 5. Biological functions of representative arginylation sites.

a, Imaging of ERO1A and 24R-ERO1A compared with endoplasmic reticulum marker and cooverexpressed PDI. b, Expression levels of cytosolic ERO1A species in the CHX chase experiment. c, Imaging of SSBP and 17R-SSBP compared with mitochondria protein COX4. d, Bioenergetic profiles of HEK293T cells after SSBP and 17R-SSBP transfection measured by Seahorse XF24. Data are presented as mean ± s.d. (n = 5 biological replicates). Mitochondria (mito.) respiration was analyzed with basal respiration, ATP production, proton leak, maximal respiration, spare capacity and nonmitochondria (Nonmito.) respiration. The average values of three Seahorse measurements were used for comparison. DAPI, 4,6-diamidino-2-phenylindole; ER, endoplasmic reticulum; FCCP, carbonyl cyanide-4-(trifluoromethoxy)phenylhydrazone; Rot/A, rotenone/antimycin A; OCR, oxygen consumption rate. Source data

Extended Data Fig. 1. Design and flowchart of the arginylation database and its website.

Arginylation data are from peptides, whole-proteome peptides, proteins, human proteomes, and mouse proteomes. Visualization includes isotopic ratios and annotated isotopic MS1 and MS2 spectra. All indexed MS1 and MS2 scans are accessible to the public to download. A dashed arrow means a one-to-multiple inclusion relationship.