Target site specificity and in vivo complexity of the mammalian arginylome

Abstract

Protein arginylation mediated by arginyltransferase ATE1 is a key regulatory process essential for mammalian embryogenesis, cell migration, and protein regulation. Despite decades of studies, very little is known about the specificity of ATE1-mediated target site recognition. Here, we used in vitro assays and computational analysis to dissect target site specificity of mouse arginyltransferases and gain insights into the complexity of the mammalian arginylome. We found that the four ATE1 isoforms have different, only partially overlapping target site specificity that includes more variability in the target residues than previously believed. Based on all the available data, we generated an algorithm for identifying potential arginylation consensus motif and used this algorithm for global prediction of proteins arginylated in vivo on the N-terminal D and E. Our analysis reveals multiple proteins with potential ATE1 target sites and expand our understanding of the biological complexity of the intracellular arginylome.

Figure 1

Four mouse ATE1 isoforms have different, partially overlapping specificity toward their N-terminal target sites. Cropped image of the autoradiogram showing Arg incorporation as the ¹⁴C signal (gray or black after the film exposure). ATE1 isoforms are indicated on the left. In each ATE1 isoform group, two rows show the two repeats of the same experiment, performed using the arrays synthesized in duplicates on the same membrane and analyzed in duplicates in the arginylation assays (see Fig. S1 for the original uncropped image). The peptide sequences are shown underneath, with the target residue underlined. The peptide backbone was selected based on the N-terminal beta actin peptide, as explained in the text.

Figure 2

Four mouse ATE1 isoforms have different specificity toward midchain target sites. Cropped image of the autoradiogram showing Arg incorporation as the ¹⁴C signal (gray or black after the film exposure). ATE1 isoforms are indicated on the right, two rows show the two repeats of the same experiment, performed using the arrays synthesized in duplicates on the same membrane and analyzed in duplicates in the arginylation assays (see Fig. S1 for the original uncropped image). The peptide sequences are shown underneath, with the target residue underlined.

Figure 3

Four mouse ATE1 isoforms show different reactivity with natural peptides. Cropped image of the autoradiogram showing Arg incorporation as the ¹⁴C signal (gray or black after the film exposure). ATE1 isoforms are indicated on the right, two rows show the two repeats of the same experiment, performed using the arrays synthesized in duplicates on the same membrane and analyzed in duplicates in the arginylation assays (see Fig. S1 for the original uncropped image). The peptide sequences are shown underneath, with the target residue underlined.

Figure 4

Arginylation of “non-canonical” N-terminal residues. (A) Arginylation assay with synthesized peptides in solution with varied residues in the N-terminal position (sequences are indicated on the x axis). Y axis shows [³H-R] incorporation signal, normalized to that of the D-containing peptide. Error bars represent SEM, n = 3 independent repeats. See Dataset 1 for the mass spectra of the D- and C-peptides. (B) Arginylation assay with the peptide containing N-terminal D (as a positive control) as well as unoxidized and oxidized C (sequences are indicated on the x axis; oxidation is denoted as (O₃). Error bars represent SEM, n = 3 independent repeats.

Figure 5

Mouse ATE1-1 exhibits complex specificity toward randomly selected peptide targets. (A) Two-sample logo of the arginylated peptides in the literature set against the (restricted) unlabeled peptides. Two Sample Logo was run with default parameters and color settings (no Bonferroni correction was applied). (B) Cropped image of the autoradiogram showing R incorporation as the ¹⁴C signal (gray or black after the film exposure), using peptides with potential side chain D/E target sites (top row), N-terminal D/E target sites with different context in the adjoining residues (middle row), and randomly designed peptides with variable target sites immobilize on the same membrane (bottom row) performed in two independent repeats (see Fig. S3 for the original uncropped image). The peptide sequences are shown underneath. The last three positions in the array were left empty for the negative control.

Figure 6

Arginylation of N-terminal D or E occurs independently of the specific order of amnino acid residues around the target site. Cropped image of the autoradiogram showing R incorporation as the ¹⁴C signal (gray or black after the film exposure), using pairs of peptides with the same amino acid sequence, synthesized in the original (top row) and shuffled (bottom row) order of the residues. In all cases except two, shuffling did not affect the arginylation signal on the peptide. Two repeats are shown for each reaction.

Figure 7

(A) ROC curves for the motif-based and logistic regression ensemble (LRE) predictors described in this study. The dashed lines represent the models’ performance when trained on the literature-derived peptides and peptides from the first array. The solid lines represent the models’ performance when trained on the final combined data set. (B) Two-sample logo of all the arginylated peptides in the final combined set against the (restricted) unlabeled peptides. Two Sample Logo was run with default parameters and color settings (no Bonferroni correction was applied).