Profiling of N-acetylated protein termini provides in-depth insights into the N-terminal nature of the proteome

Abstract

N-terminal processing of proteins is a process affecting a large part of the eukaryotic proteome. Although N-terminal processing is an essential process, not many large inventories are available, in particular not for human proteins. Here we show that by using dedicated mass spectrometry-based proteomics techniques it is possible to unravel N-terminal processing in a semicomprehensive way. Our multiprotease approach led to the identification of 1391 acetylated human protein N termini in HEK293 cells and revealed that the role of the penultimate position on the cleavage efficiency by the methionine aminopeptidases is essentially conserved from Escherichia coli to human. Sequence analysis and comparisons of amino acid frequencies in the data sets of experimentally derived N-acetylated peptides from Drosophila melanogaster, Saccharomyces cerevisiae, and Halobacterium salinarum showed an exceptionally higher frequency of alanine residues at the penultimate position of human proteins, whereas the penultimate position in S. cerevisiae and H. salinarum is predominantly a serine. Genome-wide comparisons revealed that this effect is not related to protein N-terminal processing but can be traced back to characteristics of the genome.

Fig. 1.

Scheme of experimental approach for analysis of N-acetylated protein N termini in human HEK293 cells. Four individual sets of peptides, generated by trypsin (twice), Lys-C, and Lys-N, each originating from 1 mg of HEK293 lysate, were subjected to SCX separation, LC-MS/MS, and database searches.

Fig. 2.

Schematic of SCX fractionation of different classes of proteolytic peptides. The bottom graph displays the observed elution patterns of unmodified doubly charged peptides, typically very abundant in (tryptic) digests, in dark gray, whereas the elution of doubly and singly phosphorylated peptides is displayed in light grey. In the top diagram, the experimentally derived elution profiles of the number of N-acetylated peptides are displayed in a cumulative manner for the four experiments using trypsin, Lys-C, and Lys-N. Characteristic patterns are observed for N-acetylated peptides containing zero, one, or two basic residues whereby the first two categories can be nearly base-line resolved from the other classes of peptides.

Fig. 3.

Overview of number of non-redundant N-acetylated peptides observed per SCX fraction. The graphs display the observed elution patterns of N-acetylated peptides for the four experiments (trypsin (twice), Lys-C, and Lys-N). Although there is some variation, they all possess similar elution patterns and number of identified N-acetylated peptides. Shown in black are the number of peptides originating from N termini predicted by the annotated genome (starting at either position 1 or 2 in the protein), whereas displayed in gray are peptides with a start residue at a different position in the proteins.

Fig. 4.

Redundancy, reproducibility, and complementarity of multienzyme approach and non-redundant data set of in vivo N-acetylated protein termini in human HEK293 cells. Each of the Venn diagrams displays the overlap in identified acetylated protein N termini between a single trypsin data set and each of the other data sets. As expected, the overlap between the trypsin replicates is larger than between trypsin and Lys-C or Lys-N. After filtering for redundant protein N termini, we obtained experimental identification of a total of 1391 in vivo N-acetylated protein termini. FDR, false discovery rate.

Fig. 5.

Characteristics of experimentally measured N-terminal proteome. A, three-way comparison of N-acetylated peptides identified by studies performed on human teratocarcinoma Nt2/d1 cells (14), HeLa cells (11), and HEK293 cells (our present data) revealing a small overlap of 14% between the HeLa and HEK293 study. Combining the results of these three studies provides a list of 1861 non-redundant acetylated N-terminal peptides (supplemental Table 6). B, frequency distribution of acetylated N-terminal amino acid residues. Results are summarized for data on HEK293 cells (this work), HeLa cells (11), D. melanogaster (20), H. salinarum (19), and S. cerevisiae (11). The data on the two human cell lines are very similar. N-terminal acetylation on serine residues is frequently observed in organisms such as H. salinarum and S. cerevisiae, whereas acetylation on alanine is more abundant in human HEK293 and HeLa cells. C, bar chart illustrating the effect of the penultimate amino acid residue on the efficiency of methionine cleavage. If, for example, an alanine is in the second position, the N-terminal methionine is cleaved off in nearly 100% of the cases; however, when a valine is in this position, only 50% of the N-terminal peptides undergo methionine cleavage.

Fig. 6.

Sequence logos illustrating frequency of amino acid residue occurrence in primary N-terminal stretch of proteins. The amino acid position (1 marking the ultimate N-terminal residue) is indicated below each sequence logo. The top row shows the logos obtained from our data set of 1391 N-acetylated peptides detected in HEK293 cells and those generated from the in vivo N-acetylated peptide data sets from HeLa cells, S. cerevisiae (11), D. melanogaster (20), and H. salinarum (19). These sequence logos reveal that the relative frequency of the terminal amino acid residues is very comparable between the HEK293 and HeLa cells, whereas different relative abundances are observed for the N-terminal residues of S. cerevisiae. The second and third rows reveal subsets of the experimentally measured HEK293 N-acetylated peptides, dividing them in classes of peptides starting with an alanine, serine, threonine, methionine, glycine, and valine. Similarly, the fourth row contains sequence logos for peptides starting with an Ala-Ala, Ala-(Asp/Glu), or Ala-Ser stretch. The fifth row displays for comparison the sequence logos for all proteins in the human genome with a (Met)-Ala-Ala sequence (from the Swiss-Prot database) and shows amino acid frequency plots of N-terminally acetylated peptides from D. melanogaster and H. salinarum.

Fig. 7.

Proteome-wide amino acid frequency distributions. The frequency of occurrence for the penultimate amino acid (i.e. Met-X) residue of protein N termini is given in black solid bars, and for comparison, the frequency of occurrence over all intact proteins present in the proteome is give by the striped bars (data were taken from the Swiss-Prot v56.2 database for H. sapiens (n = 18,821), D. melanogaster (n = 2789), S. cerevisiae (n = 6551), and H. salinarum (n = 443)).