Proteomics analyses reveal the evolutionary conservation and divergence of N-terminal acetyltransferases from yeast and humans

Human hARD1-hNAT1 Genes Complement the Yeast ard1-Δ nat1-Δ Phenotypes.

Previously, the ard1-Δ NAT1, ARD1 nat1-Δ and ard1-Δ nat1-Δ yeast mutants were shown to have a variety of phenotypes, including defects in general growth, growth on nonfermentable carbon sources, transition to G_o phase, sporulation, cell growth on salts, caffeine, hydroxyurea and SDS containing media, and in derepression of silent loci (16, 17). These defects are due to the lack of N-acetylation of numerous proteins, including Orc1p and Sir3p (18, 19). The following yeast strains (Exp. 1 in Table 1) were used to demonstrate that the human ARD1-NAT1 genes can complement the yeast defects: the wild type strain yNat; yNatA-Δ containing the deletion yard1-Δ ynat1-Δ; and the y[hNatA] strain containing yard1-Δ ynat1-Δ hARD1 hNAT1. Thus, yNatA-Δ lacks NatA, whereas y[hNatA] lacks yeast NatA, but has human NatA. Expression of both hARD1 and hNAT1 fully complements all observed yNatA-Δ defects, whereas expression of either of the single genes, hARD1 or hNAT1 does not (Fig. 1). Furthermore, heterologous combinations, hARD1 yNAT1 and yARD1 hNAT1, were not functional in yeast, suggesting significant structural subunit differences between the species.

Human hARD1-hNAT1 Genes and the Yeast yARD1-yNAT1 Genes Encode NATs That N-Acetylate the Same Set of Yeast Proteins.

We used a COFRADIC technique to isolate N-terminal peptides (13) and to further determine the N-acetylated and unacetylated proteins from HeLa cells (hNat) (see below), a normal yeast strain (yNat), a mutant of yeast deficient in NatA (yNatA-Δ), and the deficient yeast mutant expressing human NatA (y[hNatA]). Proteome comparison of the yNat and y[hNatA] strains revealed that the yeast NatA and the human NatA act on the same set of proteins. The COFRADIC analysis performed on the 3 yeast strains, yNat, yNatA-Δ, and y[hNatA], is schematically outlined in Fig. 2. As briefly described in the figure, this COFRADIC technique can be used to enrich for N-terminal peptides ending in an arginine residue of α-amino blocked proteins before LC-MS/MS analysis. By mass tagging protein amino groups, including free protein α-amino groups, before the actual COFRADIC step, mass spectrometry can be used to determine the in vivo protein acetylation status and the degree of acetylation. In this study, the following 2 different amino-directed mass tagging strategies were used: trideutero-acetylation [AcD3] and propionylation [N-Prop] (see Fig. 2B). In the former case, acetylated peptides from in vivo blocked proteins and trideutero-acetylated peptides from in vivo free proteins coeluted during RP-HPLC and were used to calculate the degree of acetylation. In the latter case however, propionylated peptides, indicative for in vivo free protein forms, segregated by RP-HPLC from their acetylated counterparts and were used to point to Nat substrates, but not to quantify the degree of acetylation. To further distinguish N-terminal peptides isolated from the 3 yeast strains, Arg-SILAC labeling (20) was performed after introducing the arg4-Δ marker (Exp. 2 in Table 1 and Fig. 2). Such yeast strains are dependent on an external arginine source for growth (Fig. S1), enabling the incorporation of isotopic variants of arginine. Yeast strains were cultured in the presence of 1 of 3 different stable l-arginine isotopic variants (¹²C₆ l-Arg, ¹³C₆ l-Arg, or ¹³C₆¹⁵N₄ l-Arg) (Table 1). Such an approach permits triplexed proteomics; all isolated N-terminal peptides end on arginine and the mass labels provide ample separations, thus allowing distinction of the 3 N-terminal peptide forms by MS (Fig. S2).

In total, 3 different and independent N-terminal COFRADIC analyses were performed: in vitro AcD3-acetylation, N-propionylation (N-Prop-), or no modification (N-noMod) of free protein amino groups (Fig. 2B). The ratio of in vivo acetylated (Ac termini) versus nonacetylated N termini was only inferred from the AcD3-experiment (see above). However, because of the high frequency of partially acetylated N termini, the resulting complex MS-spectra sometimes precluded straightforward analysis. Therefore, we used N-propionylation (N-Prop-) to extend our dataset (Fig. 2B). Finally, in vivo acetylated peptides were also detected in the N-noMod setup. The overall results of these analyses are discussed below.

The AcD3 incorporation data, comprising 262 unique N termini, indicated that 12% of the identified proteins were fully acetylated at their N terminus, whereas 45% were partially acetylated and 43% were nonacetylated (Table 2). The N-propionylation and N-noMod setups confirmed several AcD3 results, which, when combined, comprised 379 unique N termini (Table S1A). Combining these results with previously catalogued yeast N termini (1), a compilation of 614 different endogenous yeast proteins are included, of which at least 58% are partially or fully N-terminally acetylated (Table S1B). Our data further revealed that 163 unique yeast proteins were N-acetylated in the 2 strains yNat or y[hNatA], whereas these proteins were not acetylated in the strain not expressing any NatA (yNatA-Δ) (Table S1C). Furthermore, yNatA and y[hNatA] substrates had a distinct MS-profile as compared with NatB or NatC substrates and to the non-Nat substrates, thus allowing straightforward detection upon examination of MS spectra (Fig. S2 A–C).

No yeast protein was N-acetylated in yNat that was not N-acetylated in y[hNatA], and vice versa, although there was variation in the degree of N-acetylation by yNat versus y[hNatA], which was determined for 81 substrates identified in the AcD3 setup (Table S1C). Up to 85% of the partially acetylated NatA-type substrates were acetylated to a higher degree in the y[hNatA] strain as compared with the yNat strain, as exemplified by the La protein homolog (Table S1C and Fig. S2D). However, some substrates were more prone to acetylation in the yNat strain compared with the y[hNatA] strain, as for example, the 60S ribosomal protein L1 (Table S1C and Fig. S2E).

Determination of Protein N-Acetylation in HeLa Cells and Analysis of hARD1-hNAT1 Knockdown Effects.

To investigate the level of N-acetylation in humans and elaborate on the substrate specificity of hNatA in its physiological context, we determined the N-terminal sequences of acetylated and nonacetylated soluble proteins from the HeLa cell lines listed in Table 1 (Expts. 3 and 4). hNat denotes the normal HeLa cell line and hNatA-si denotes a cell line in which the expression of hARD1 and hNAT1 was reduced by sihARD1 and sihNAT1 knockdown (7). N-terminal COFRADIC analyses were carried out using trideutero-acetylation to block free protein N termini (similar to Fig. 2). Here, again, AcD3- peptides and their in vivo acetylated counterparts could be distinguished after MS analysis by their 3-Da mass difference. As illustrated in Fig. S3, the N terminus of the ADP-ribosylation factor GTPase-activating protein 3 was in vivo acetylated (A), the pirin N terminus was partially acetylated (B), whereas the developmentally-regulated GTP-binding protein 2 carries an in vivo free N terminus (C). In this experiment, we identified the N termini and the proportion of acetylation of 742 different proteins from the normal hNat HeLa strain (Tables 2 and S2A-B). These results constitute the largest dataset of N-acetylation of human proteins and reveal that ≈76% of HeLa proteins are fully acetylated at their N terminus, whereas 8% are partially acetylated and 16% are unacetylated. A comparison of N-acetylated and unacetylated proteins from yeast and humans reveal that the distribution N termini of naturally occurring substrates differs significantly (Table 2). Most N-acetylated proteins from yeast begin with serine, whereas most N-acetylated proteins from humans begin with alanine. A more striking difference is the N-acetylation of 8 of 19 Met-Lys- proteins from humans and none from yeast (Table 2). The enzyme acetylating Met-Lys- proteins may represent a NAT found in humans but not in yeast.

The overall protein N-acetylation patterns from the hNat and hNatA-si cell lines (Exps. 3 and 4 in Table 1) were surprisingly similar. Reduced expression of hARD1 and hNAT1 in the hNatA-si strain was verified by Western blot analysis of hArd1p and hNat1p, using aliquots from the samples analyzed by COFRADIC (Fig. S4A). The hNatA-si strain contained ≈30% of the normal amount of Nat1p, and ≈5% of the normal amount of Ard1p. Unexpectedly this low amount of hNatA is still sufficient to acetylate most of the NatA substrates, and the acetylation of only a few of the NatA substrates appeared to be reduced. We here considered a hNatA substrate to be affected by knockdown of the hNatA complex when the difference in percentage of N-acetylation was >5% between Exp. 3 and Exp. 4 (Table 1). Only 16 hNatA substrates were susceptible to the diminished level of hNatA out of a total pool of 242 acetylated NatA type substrates (i.e., 6.6%) identified in both setups (Table S2C and Fig. S4 B–R). Interestingly, 13 of the 16 affected N termini having diminished levels of acetylation in Exp. 4 (hNatA knock-down) were also found to be only partially acetylated in Exp. 3 (control setup). In contrast, nearly all NatA type substrates with completely acetylated termini from Exp. 3 were also found fully acetylated in Exp. 4. It therefore appears that N termini can be assigned to different classes with various degrees of efficacy for hNatA acetylation: (i) completely N-acetylated and nearly unaffected by hNatA-si; (ii) partially acetylated and N termini of which the degree of acetylation is not affected by hNatA-si; (iii) partially acetylated and N termini of which acetylation is diminished by NatA-si, and (iv) those that are not acetylated.

In Vitro Acetylation by hArd1p + hNat1p.

It is essential to determine whether the different classes of hNatA-type N termini identified in our HeLa experiments, from fully acetylated to nonacetylated, are actually differentially preferred substrates of the hNatA complex per se. If a positive correlation is observed, this would establish that the observed differences in N-acetylation of NatA type N termini in human cells are directly linked to the substrate preferences of the hNatA complex. We addressed this question by carrying out in vitro N-acetylation (Exp. 5 in Table 1) of representatives of the following 4 groups of synthetic 24-mer peptides with an immunoprecipitated hArd1p and hNat1p complex (6): group A, a previously described in vitro hNatA substrate (SYSM-); group B (class i above), hNatA substrates found to be fully acetylated and not affected in Exp. 4 (SESS-, STPD-, ANSA-); group C (class iii above), hNatA substrates found to be partially acetylated and affected in Exp. 4 (TSAL-, AVFA-, ADGK-); and group D (class iv above), a peptide (SPTP-) found to be unacetylated in vivo. The results demonstrate that the hArd1p-hNat1p complex acetylates in vitro all 3 class i N termini (Fig. S5). Two of the in vivo partially acetylated class iii substrates, TSAL and ADGK, were also found to be acetylated in vitro, but generally at a lower level compared with their fully in vivo acetylated counterparts, whereas AVFA was not significantly acetylated. The class iv peptide starting with serine (SPTP-), found to be unacetylated in vivo (Table S2A), was also not acetylated in vitro by hNatA.