Nα-terminal acetylation of proteins by NatA and NatB serves distinct physiological roles in Saccharomyces cerevisiae

N-terminal (Nt)-acetylation is a highly prevalent co-translational protein modification in eukaryotes, catalyzed by at least five Nt-acetyltransferases (Nat) with differing specificities. Nt-acetylation has been implicated in protein quality control but its broad biological significance remains elusive. We investigated the roles of the two major Nats of S. cerevisiae, NatA and NatB, by performing transcriptome, translatome and proteome profiling of natAΔ and natBΔ mutants. Our results do not support a general role of Nt-acetylation in protein degradation but reveal an unexpected range of Nat-specific phenotypes. NatA is implicated in systemic adaptation control, as natAΔ mutants display altered expression of transposons, sub-telomeric genes, pheromone response genes and nuclear genes encoding mitochondrial ribosomal proteins. NatB predominantly affects protein folding, as natBΔ mutants accumulate protein aggregates, induce stress responses and display reduced fitness in absence of the ribosome-associated chaperone Ssb. These phenotypic differences indicate that controlling Nat activities may serve to elicit distinct cellular responses.


INTRODUCTION
Acetylation of the N-terminal α-amino group of polypeptides (Nt-acetylation) is the most prevalent irreversible modification of proteins in eukaryotes, affecting 50-70% of the yeast proteome and up to 90% in higher eukaryotes (van Damme et al., 2011b;Helbig et al., 2010;Aksnes et al., 2016;Aksnes et al., 2019). Nt-acetylation is achieved already early during protein synthesis, as soon as the nascent chains emerge at the ribosomal tunnel exit (Gautschi et al., 2003), and therefore is an intrinsic property of proteins synthesized in eukaryotic cells. Co-translational Nt-acetylation is exerted by a set of structurally related N-terminal acetyltransferases (Nat) which associate with the large ribosomal subunit close to the peptide tunnel exit (Polevoda et al., 2008;Gautschi et al., 2003). Saccharomyces cerevisiae expresses five N-terminal acetyltransferases (NatA to NatE) with distinct specificities for Ntermini (Aksnes et al., 2016), whereby NatA and NatB together are responsible for about 90% of all protein Nt-acetylation. NatA acetylates proteins with S-, A-, V-, G-, T-and C-termini, once the Nterminal methionine has been removed by methionine aminopeptidase (MetAP; Polevoda & Sherman, 2003b). NatB acetylates the N-terminal methionine of proteins starting with MD-, ME-, MQ-, and MN- (Polevoda & Sherman, 2003b). The extent of Nt-acetylation is variable for different proteins, ranging from infrequent acetylation to full acetylation, in particular for NatB substrates (Aksnes et al., 2016).
NatA and NatB form each a dimeric complex consisting of a catalytic subunit and an auxiliary subunit that mediates ribosome binding (Polevoda et al., 2008;Gautschi et al., 2003). In yeast, the ablation of either subunit of NatA and NatB completely abolishes the enzymatic activity of the complex (Park & Szostak, 1992;Polevoda et al., 2003;Polevoda & Sherman, 2003a) and causes a number of Natspecific and general phenotypes, including the NatA-specific mating defects, and a general reduction of growth rate as well as increased stress sensitivity, especially against elevated temperatures (Gautschi et al., 2003;Polevoda et al., 2003;Polevoda et al., 1999;Mullen et al., 1989;Whiteway et al., 1987). In searches for functional roles of N-acetylation, previous studies have implicated Ntacetylation in a range of different cellular processes (Aksnes et al., 2019). First, Nt-acetylation is required for establishing protein-protein interactions, by providing critical contributions to involved interfaces. Examples are the Ubc12-Dcn1 ubiquitin ligase complex (Scott et al., 2011) and interactions of the silencers Sir3/ Orc1 with histone proteins (Whiteway et al., 1987;Geissenhöner et al., 2004;Wang et al., 2004), of the GTPase Arl3 with the Golgi membrane protein Sys1 (Behnia et al., 2004;Setty et al., 2004) and of actin with tropomyosin (Polevoda et al., 2003;Singer & Shaw, 2003). Second, Nt-acetylation contributes to the correct cellular sorting of proteins to the secretory pathway, by preventing the incorrect targeting of cytosolic proteins to the ER translocon (Forte et al., 2011). Third, Nt-acetylation has been connected to protein folding and stability since it reduces the Nterminal charge and thereby can stabilize N-terminal α-helices of proteins such as mitochondrial chaperonin 10 (Cpn10; Jarvis et al., 1995;Ryan et al., 1995), tropomyosin (Greenfield et al., 1994) and α-synuclein (Dikiy & Eliezer, 2014;Bartels et al., 2014). Finally, a number of publications link Ntacetylation to protein degradation in yeast (Oh et al., 2017;Dörfel et al., 2017;Holmes et al., 2014;Zattas et al., 2013;Pezza et al., 2009). There, acetylated N-termini form an N-degron that targets proteins via the Ac/N-end rule pathway for ubiquitination by E3 ubiquitin ligases including Not4 and Doa10, followed by proteasomal degradation (Hwang et al., 2010;Shemorry et al., 2013). It has been proposed that the failure to form protein complexes exposes acetylated N-termini of unassembled subunits for ubiquitination via the Ac/N-end rule pathway, defining Nt-acetylation as mark for cellular control of protein stoichiometries and assembly. In support of this proposal, several different proteins overproduced from plasmids undergo degradation via the Ac/N-end rule pathway under specific conditions (Hwang et al., 2010;Shemorry et al., 2013). However, a recent study systematically analyzing the impact of Nt-acetylation on the stability of proteins in yeast did not provide evidence for a general function as N-degron (Gawron et al., 2016;Kats et al., 2018). The importance of Ntacetylation for protein degradation in vivo is therefore still elusive.
To elucidate the biological functions of Nt-acetylation of proteins we performed a comprehensive multi-level analysis of mutant yeast cells lacking NatA or NatB. Combining transcriptome analysis, ribosome profiling, SILAC-and SWATH-based quantitation of protein stability and aggregation, we find that deletions of NatA and NatB cause distinct phenotypes that do not correlate with, and hence are not explained by quantitative contributions of NatA and NatB to Nt-acetylation of proteins. Our data suggest an unexpected divergence between the two Nat enzymes in the maintenance of cellular physiology. NatA activity is implicated in genetic control of Sir3 and Orc1 mediated gene silencing, transposon activity and mitochondrial activity. NatB activity rather supports protein homeostasis more globally, without major effects on protein stability but potentially by supporting the folding of newly synthesized proteins.

Lack of NatA and NatB activity leads to distinct changes of gene expression profiles
We determined whether the absence of NatA or NatB affects gene expression and protein levels by a series of multi-omics high-throughput screens, comparing yeast wild-type (WT) with mutant cells lacking the catalytic subunits Naa10 or Naa20 of the NatA and NatB complexes (referred to as natAΔ and natBΔ), respectively. mRNA sequencing (RNAseq), ribosome profiling (RP) and quantitative proteomics allowed quantitative assessment of Nat-dependent changes at the transcriptional, translational and protein steady-state level. The deep sequencing-based data sets (RNAseq and RP) comparing natAΔ or natBΔ and WT include more than 4700 quantified genes and are highly reproducible among replicates (r = 0.79-0.95; Figures S1A-S1D and S1F-S1H). Stable isotope labeling by amino acids in cell culture (SILAC; Ong et al., 2002) coupled with mass spectrometry allowed to determine the steady-state levels of 2169 and 2976 proteins in natAΔ and natBΔ mutant cells, respectively (r = 0.68/0.71; Figures S1E and S1I). To complement SILAC-MS analysis, we also employed label-free SWATH-MS (Gillet et al., 2012) and determined the steady-state levels of 2245 (natAΔ) and 2220 (natBΔ) proteins.
For 78 proteins, among them 18 previously unknown Nt-acetylated proteins, we verified the expected, Nat-specific loss of Nt-acetylation in the natAΔ and natBΔ mutants (Figures S2A and S2B). Our data confirm the lack of functional redundancy between NatA and NatB. Supporting previous findings (van Damme et al., 2011a;Aksnes et al., 2016), NatB substrates with MD-, ME-, MN-and MQ-termini are acetylated at efficiencies close to 100%, while some of the A-, G-, S-and T-termini of NatA substrates are only partially Nt-acetylated.
Loss of Nt-acetylation in natAΔ and natBΔ mutants causes substantial, mutant-specific changes in protein synthesis. RP analysis revealed that the deletion of NatA and NatB reproducibly changes the translation levels of approximately 150 and 400 genes more than two-fold, respectively ( Figures 1A   and 1B). The changes are only weakly correlated between the two mutants (r = 0.182; Figures S3B and 1C). Consistent with the more severe growth defects of natBΔ mutant cells on agar plates ( Figure S4), the extent of changes is overall stronger in natBΔ cells as compared to natAΔ cells.
The changes in protein synthesis in natAΔ or natBΔ strongly correlate with changes in mRNA levels as revealed by RNAseq analysis (r = 0.94/0.87; Figures S5A and S5D). This demonstrates that they predominately result from changes at the transcriptional level. We also found good correlations between observed changes in protein synthesis rates and steady-state levels determined by SILAC (r = 0.64/0.49; Figures S5B and S5E). The observed changes of the proteome are therefore mainly due to altered protein synthesis rather than changed protein degradation.
We performed threshold-independent gene ontology (GO) enrichment analyses to identify the gene expression responses to deficiencies in Nt-acetylation activity in the respective mutants ( Figures 1D   and 1E). Most prominently, the lack of NatA activity impacts the expression of pheromone response genes, transposable elements and genes encoding mitochondrial ribosomal proteins ( Table S1). The lack of NatB activity particularly impacts the expression of genes involved in stress-responses, amino acid biosynthesis and translation (Table S2).

NatA regulates gene silencing and mitochondrial protein synthesis
The most strongly down-regulated genes in natAΔ cells are enriched for genes encoding mating typerelated proteins, especially MATa-specific proteins (Figures 2A and 2B). Haploid yeast cells sense mating pheromone as signal for the presence of potential mating partners. This triggers the mating pathway to induce cell cycle arrest at G 1 phase and oriented growth towards the mating partner for cell fusion (Bardwell, 2004;Haber, 2012). Some members of this signaling cascade are expressed in both yeast mating types, MATa and MATα, while others are mating type-specific. In diploid cells, all mating genes are suppressed by the combined, dimeric repressor Mata1-Matα2 derived from both mating type loci, HML (MATα) and HMR (MATa).
We found that MATa-specific genes are almost completely repressed upon NatA deletion, providing a rationale for the strongly impaired mating capacity of MATa natAΔ cells (Whiteway & Szostak, 1985).
We hypothesized that deficient Nt-acetylation causes these alterations in gene expression through effects on the gene silencer Sir3, which confers suppression of gene expression at the silent HML locus of the mating cassette (Rine & Herskowitz, 1987). Nt-acetylation of Sir3 is required for its stable interaction with the nucleosome core particle during gene silencing (Yang et al., 2013). Confirming this assumption, deletion of the HML cassette rescued expression of MATa-specific genes in natAΔ cells (Figures 2B and S6A). Intriguingly, it also restored expression of all five families of transposable elements which are downregulated in natAΔ mutant cells (Figures 2A and S6A). Considering that transposable elements are exclusively expressed in haploidic cells (Elder et al., 1981), we speculate they may be regulated similar to the MATa-specific mating response. Our gene expression analysis in natAΔ cells identified 41 additional, so far unidentified HML-dependent genes with unrelated functions (Table S3).
Of note, the additional HML deletion rescued expression of more than half of the differentially expressed genes in natAΔ mutant cells, i.e. transposable elements, MATa-specific genes and others ( Figure S6A), and this also partially alleviates the NatA-dependent slow growth phenotype ( Figures   2C and S6B). An additional set of NatA-regulated genes is differentially expressed independently of the HML locus and predominantly localized close to sub-telomeric regions ( Figures 2D, 2E, S6D, and S6E).
Intriguingly, sub-telomeric genes are typically repressed by the silencer Orc1, a NatA substrate that is structurally homologous to Sir3, possesses an identical N-terminus and also functions similarly (Geissenhöner et al., 2004). The de-repression of sub-telomeric genes is specific for natAΔ mutants in comparison to mutants affecting other N-terminal acetyltransferases ( Figure S6C and data not shown). These findings strongly suggest that Nt-acetylation of Orc1 is critical for its gene silencing activity.
Based on our RNAseq, RP and SILAC-MS analyses, NatA deletion also elevated the average synthesis of nuclear-encoded mitochondrial ribosomal proteins by more than 40% (Figures 3A and   S7A). This effect was much more pronounced in natAΔ mutants as compared to natBΔ mutants, suggesting specificity of NatA versus NatB activity ( Figure S7B). 94% of all detected genes encoding mitochondrial ribosomal proteins are up-regulated in natAΔ cells (76/81 in the RP data set), leading to similarly elevated ribosomal protein levels ( Figure 3A). The changes in both, ribosomal protein expression and mitochondrial phenotype were not affected by mating-type regulation of the HML cassette ( Figures S7A, S7C, and S7D).
The seemingly elevated protein synthesis capacity in mitochondria however did not coincide with enhanced respiration capacity of natAΔ cells analyzed by MitoTracker. Correcting the measured respiration activity for the larger cell size of NatA-depleted cells even suggests reduced mitochondrial activity ( Figures 3B and 3C). In line with this finding, natAΔ cells displayed a greatly extended lag phase when shifted from the fermentable carbon source glucose to respiration-obligate glycerol ( Figure 3D). Together our data suggest that enhanced synthesis of mitochondrial ribosomal proteins may be induced in response to impaired function of mitochondria or mitochondrial translation in natAΔ cells.

Constitutive stress response in natBΔ mutant cells
Our differential gene expression analysis revealed that natBΔ cells strongly up-regulate genes encoding machinery for protein refolding, including the stress-inducible chaperones Hsp26 and Hsp70 (Ssa4) as well as the cytosolic and mitochondrial disaggregases Hsp104 and Hsp78 ( Figure 4A).
These genes were only poorly induced in natAΔ cells, indicating that the lack of NatB activity, but not of NatA activity, causes significant protein folding stress eliciting cell responses ( Figure 4B).
Consistent with the elevated expression of chaperone genes, natBΔ cells more efficiently refolded heat-denatured luciferase after thermal stress treatment of the cells as compared to natAΔ cells ( Figure 4C). Interestingly, genes more than two-fold induced in natBΔ cells were enriched for targets of the transcription factors Msn2/Msn4 which control expression of the general stress response of yeast (Martínez-Pastor et al., 1996;Schmitt & McEntee, 1996) (Figure 4D). Furthermore, in natBΔ cells the burden of unfolded proteins on the cellular protein folding machinery is counteracted by reduced synthesis of proteins involved in cytosolic translation ( Figure 4E). The reduction of translation activity was also observed by polysome profiling that showed a higher monosome to polysome ratio in natBΔ cells as compared to WT and natAΔ cells ( Figure 4F). In agreement with the reduction in growth rate, ribosome synthesis and translation activity, natBΔ mutant cells also showed an about 2-fold lower rate of incorporation of radiolabeled methionine into newly synthesized proteins as compared to WT cells ( Figure 4G).

Lack of Nt-acetylation has no global impact on substrate stability
The exposure of acetylated N-termini was recently described as the targeting signal for degradation of surplus or unassembled protein complex subunits via the Ac/N-end rule pathway (Hwang et al., 2010;Shemorry et al., 2013). Accordingly, the constitutively upregulated stress response observed in natBΔ cells, and to a smaller extent in natAΔ cells, may be triggered by the accumulation of non-acetylated Nat substrates that fail to be degraded. However, a series of independent experiments do not support this idea. First, the average steady-state levels of both, verified NatA and NatB substrates and all cellular proteins did not differ when comparing WT and natAΔ or natBΔ cells grown at physiological conditions at 30°C (Figures 5A and 5B, first panels). Second, as revealed by two independent approaches, protein stabilities are similar for Nat substrates versus all proteins. In one approach, we calculated protein stabilities by normalizing protein levels (determined by quantitative proteomics) to synthesis rates (determined by RP). The results showed that global protein stabilities are not affected by deletion of NatA or NatB (Figures 5A and 5B, second panels). In a second approach, focusing on natBΔ mutant cells, we measured protein stabilities using the recently developed tandem fluorescent protein timer approach (tFT; Khmelinskii et al., 2012) and compared natBΔ mutant to WT cells. The tFT comprises two single-color fluorescent proteins (sfGFP and mCherry) that mature with different kinetics and are fused to a protein of interest. Determining the ratio of mCherry to sfGFP allows measuring the relative protein stability, independent of protein abundance. Analyzing a genome wide library of chromosomally encoded tFT fusion genes did not show any significant difference of tFTtagged verified NatB substrates as compared to all measured proteins, confirming previous results ( Figure 5C, first panel). Third, the Ac/N-end rule pathway does not play a detectable role as quality control mechanism when protein homeostasis is perturbed by environmental stress. RP, SILAC and tFT analyses of cells transiently exposed to heat stress (at least two doublings at 37°C) did not detect global effects of NatA and NatB deletion on protein levels or stability (Figures 5A-5C, right panels).
Fourth, we did not detect a role of Nt-acetylation on protein subunit degradation when subunit stoichiometries of specific protein complexes are disturbed. This was experimentally investigated by duplicating genes encoding individual subunits of complexes, fusing one of the alleles to the tFTencoding sequence, and measuring the stability of the tagged subunits. The 2-fold molar excess of the subunits encoded by the gene duplications over the other subunits of the corresponding protein complex should lead to an experimentally detectable degradation of up to 50% of the tagged subunits.
The chosen candidates are Nt-acetylated members of oligomeric complexes, with two of them (Hsp104 and Ubp6) being positively tested in a previous study for stability changes under strong overexpression (Hwang et al., 2010). Deficient Nt-acetylation in natAΔ and natBΔ cells did not increase the stability of the excess subunits. Ubp6 and Sup45 were destabilized in both mutant cells, indicating that stability was independent of Nt-acetylation. The only exception is Hsp104, which is slightly stabilized in natBΔ mutants ( Figure 5D).
Taken together, our analyses do not indicate a prominent role of Nt-acetylation for global protein stability at physiological and heat stress conditions and upon perturbance of protein stoichiometries as occurring upon gene duplication.

Figure 5. Loss of Nt-acetylation has no global impact on substrate stability
Protein levels of natAΔ (A) and natBΔ (B) mutant relative to WT cells based on quantitative proteomics (SILAC) and protein stabilities as normalized protein levels (SILAC relative to protein synthesis determined by RP). (C) Stability of tFT-tagged proteins in natBΔ mutant relative to WT cells measured using the mCherry/GFP ratio as read-out. Cells were grown under physiological (30°C) and heat stress (37°C for two doubling times (SILAC) or 1 day (tFT)) conditions. Verified Nt-acetylation substrates are compared versus all quantified proteins. The number of proteins in each group is indicated between brackets. (D) Comparison of the stability of tFT-tagged proteins (mCherry/GFP as read-out) expressed in WT and natAΔ and natBΔ mutant cells.

Global protein aggregation upon lack of Nt-acetylation
Another possible reason for the induction of the protein refolding machinery in natBΔ cells ( Figure   4A) would be that the loss of Nt-acetylation compromises the structural integrity of NatB substrates, necessitating the refolding and sequestration activity of chaperones and ultimately resulting in increased protein aggregation. To address this possibility, we isolated protein aggregates from natAΔ and natBΔ mutant and WT cells grown at 30°C. Absence of NatB, and to a smaller extent also of NatA, led to enhanced aggregation of a diverse set of endogenous proteins ( Figure 6A). Employing SILAC-based quantitative proteomics analysis of the fraction of aggregated proteins, we identified 613 proteins that are more than 2-fold enriched in the aggregate fraction in natAΔ relative to WT, and 794 proteins in natBΔ cells ( Figure 6B). The majority of the aggregating proteins in the mutants had no or only modest changes in their total levels in comparison to WT, indicating that protein aggregation is not driven by altered total protein levels in the mutants (Figures S8A and S8B). Unexpectedly, the two sets of aggregated proteins in natAΔ and natBΔ cells largely overlap (54-70%; Figure 6B). This implies a general folding deficiency conferred by deficiency in Nt-acetylation that is not specific to NatA or NatB substrates. Consistently, protein aggregates in natAΔ and natBΔ cells were not enriched for the specific substrates of the missing Nat enzyme (Figure S9A), and in the natΔ mutant cells the aggregation propensity of all cellular proteins was not significantly different from that of Nat substrates ( Figure S9B).
We surmised that protein misfolding and aggregation could also be caused by the inactivation of chaperones, due to impaired Nt-acetylation of chaperones themselves or to elimination of a so far unknown chaperone activity of the Nat enzymes. Arguing against the first possibility, the large majority of chaperones are substrates of NatA (e.g. Hsp26, Ssa1/2, Ssb1/2, Ssz1, Sse1/2, Egd2, Hsc82, Hsp82). The sole relevant exception is Hsp104, which is a substrate of NatB. However, the chaperone activity of Hsp104 is restricted to the solubilization of aggregated proteins (Mogk et al., 2018), and this activity (as measured for aggregated luciferase) was even about 2-fold elevated in natBΔ cells ( Figure 4C). These arguments strongly suggest that protein aggregation in natBΔ cells cannot be explained by chaperone inactivation. To investigate the hypothesis that Nat complexes may function as chaperones, we tested the capacity of a previously described catalytically inactive NatA-Naa10 mutant (Liszczak et al., 2013) to suppress protein aggregation in natAΔ cells. However, the expression of plasmid-encoded Naa10 E26A mutant protein did neither prevent protein aggregation nor revert the natAΔ growth phenotypes, in contrast to the expression of NatA WT as control ( Figures   S10A and S10B). This result argues against this possibility and suggests that the enzymatic activity of Nt-acetylation is critical for NatA functionality in protein homeostasis in vivo.
We next searched for enriched biological processes involving the fraction of aggregated proteins in natAΔ and natBΔ cells, by performing a threshold-dependent GO analysis. The most enriched group among the aggregated proteins in natAΔ and natBΔ mutants are cytoplasmic translation components including cytosolic small and large ribosomal subunits, translation factors, and tRNA-aminoacylation machinery (Figures 6C and 6D). Although a number of aggregating translation factors are known stress granule markers, we did not detect any stress granule formation in natΔ mutant cells. This suggests that impaired Nt-acetylation reduces synthesis and selectively inactivates and sequesters components of the protein synthesis machinery, which further reduces the flux of newly synthesized proteins into the network of chaperones and factors involved in nascent polypeptide maturation. As expected, chaperones implicated in protein folding and disaggregation were also enriched in protein aggregates in natAΔ and natBΔ cell (Figures 6C and 6E).
Intriguingly, the aggregation-prone proteins that we identified in Nat-deficient mutants are highly similar to those isolated from cells lacking the ribosome associated Hsp70s, Ssb1 and Ssb2 (Koplin et al., 2010) (Figure 6F). This overlap includes many ribosomal proteins, which do not aggregate under general stress conditions ( Figure 6G). Furthermore, deletion of SSB1/2 in natBΔ cells strongly reduced cellular fitness (Figure 6H), suggesting Ssb1/2 and NatB act in parallel pathways. Given that Nat enzymes and Ssb1/2 both act co-translationally, we tested whether the loss of Nt-acetylation has a particularly strong impact on the de novo folding, and hence the structural integrity of nascent proteins, by radioactive pulse-labeling of newly synthesized proteins followed by aggregate isolation.
Indeed, aggregates isolated from natBΔ cells contained elevated amounts of 35 S-methionine labeled proteins as compared to WT, suggesting that non-acetylated nascent proteins have an increased tendency to misfold and aggregate ( Figure S11).

DISCUSSION
Our study provides new insights into the functional importance of Nt-acetylation of proteins. Multi-level high-throughput analysis of gene expression and proteome homeostasis of natAΔ and natBΔ mutant cells revealed that the major N-terminal acetyltransferases NatA and NatB have distinct roles in physiology and proteostasis of S. cerevisiae. Lack of NatB activity causes much stronger perturbation of protein homeostasis than the lack of NatA activity, resulting in misfolding and aggregation of several hundred proteins and induction of stress-induced genes controlled by the Msn2/4 transcription factors. Lack of NatA activity instead affects primarily gene expression, genome integrity and metabolic regulation. The disparate phenotypes cannot be explained by differences in substrate pool sizes of the two enzymes since the substrate spectrum of NatB is considerably smaller than that of NatA (3-fold less protein species, 13-fold lower number of substrate molecules, less complete acetylation of a given protein species; Aksnes et al., 2016). Rather, they appear to rely at least in part on a set of enzyme-specific targets, such as transcription silencers Sir3 and Orc1 acetylated by NatA.
Our data furthermore suggest that for both NatA and NatB targets, the Ac/N-end rule degradation pathway is a less prominent cellular response to misfolding stress than proposed earlier (Hwang et al., 2010;Shemorry et al., 2013).
One possible explanation for the prominent protein aggregation in natBΔ mutants is that the loss of Nt-acetylation impairs the folding or destabilizes the folded state of non-acetylated proteins as previously shown for some endogenous proteins, e.g. chaperonin 10 Jarvis et al., 1995) and tropomyosin (Greenfield et al., 1994). However, we find that Nat-specific substrates are not enriched in the aggregate fraction of the respective mutant cells. The aggregate fraction is also not enriched in subunits of protein complexes, suggesting that the lack of Nt-acetylation of complex subunits is also not a major reason for the spectrum of aggregating proteins. Nat substrates that misfold and aggregate in natΔ mutants due to the lack of Nt-acetylation may instead promote the coaggregation of other proteins, thereby causing global disbalance of the proteome.
A second plausible mechanism explaining the aggregation phenotype is that natBΔ cells have reduced chaperone activity. This may result from (i) lack of Nt-acetylation of the chaperones themselves, or because (ii) Nat complexes have chaperone-like activities or (iii) Nat complexes affect the activity of chaperones (Ssb and NAC) that act co-translationally in the vicinity of Nat complexes at the ribosomal tunnel exit. The first possibility is unlikely since the vast majority of chaperones is Ntacetylated by NatA, and hence retains functionality in natBΔ mutants. We performed tests of the second possibility for NatA but did not obtain evidence for a chaperone function. The most compelling evidence points towards the third possibility. Deletion of SSB1/2 in natBΔ cells strongly diminished cellular growth, implying a functional overlap between co-translational folding assisted by Ssb and Ntacetylation by NatB. Consistent with this scenario, aggregates isolated from pulse-labelled natBΔ cells include newly synthesized proteins, indicating that a fraction of them misfolds and becomes sequestered into aggregates. A functional overlap was previously suggested also for NatA and Ssb, based on the partial rescue of heat sensitivity and other natAΔ phenotypes by SSB1/2 overexpression (Gautschi et al., 2003). Importantly however, natAΔ mutants lacking SSB1/2 did not display a reduced cellular fitness that is comparable to natBΔ cells lacking SSB1/2. We speculate that Nat enzymes coordinate ribosome association and function of Ssb, thereby coordinating co-translational protein folding and enzymatic processing. Consequently, the lack of Nt-acetylation may impose challenges for the co-translational folding of proteins, which in turn enhances the need for co-translational folding assistance, explaining the negative effect of simultaneous deletion of NatB and Ssb on growth.
Evidences for the pivotal role of NatB in folding of nascent proteins are (i) the aggregation of 35 S-methionine labeled proteins in natBΔ cells after a short pulse of 35 S-methionine, (ii) the overlap between protein aggregates in natBΔ and ssb1,2Δ cells, (iii) the impaired fitness of natBΔ ssb1,2Δ double deletion mutants, and (iv) the co-translational action of NatB, like Ssb1/2, via association with the ribosomal exit tunnel. Misfolding and aggregation of nascent chains in natB∆ cells may drive coaggregation of ribosomes and other misfolding-prone proteins through unspecific protein-protein interactions. This could account for the observed proteostasis disbalance including induction of the heat shock response and overall reduced protein synthesis.
Nt-acetylation was proposed to confer a quality control pathway for unassembled orphan subunits, by creating a degron, a signal for ubiquitination and proteasomal degradation (Shemorry et al., 2013).
Accordingly, the lack of Nt-acetylation should lead to accumulation of unassembled subunits, eventually triggering protein aggregation and stress responses, consistent with what we observed in natBΔ and to a smaller extent in natAΔ cells. However, our analyses of the effects of Nt-acetylation on protein levels and stability using quantitative proteomics, RP and tFT approaches do not provide further support for this model. We do not detect a significant increase of steady-state levels or halflives of experimentally verified Nat substrates upon loss of Nt-acetylation, neither under physiological growth conditions nor after perturbation of protein folding and complex assembly by heat stress and gene duplication. Our results do not exclude that N-acetyl mediated degradation may control the stability of specific proteins, or control protein stability under specific conditions such as strong overproduction of Nt-acetylated orphan subunits. We also do not exclude the existence of redundant degradation pathways that target Nat substrates for degradation in the absence of Nt-acetylation, as shown previously for a subset of unacetylated NatC substrates .
Although the number and diversity of proteins acetylated by NatA is much higher than that of NatB (Aksnes et al., 2016;van Damme et al., 2011a), protein aggregation is less prominent in natAΔ cells, indicating non-random distribution of aggregation-prone substrates among the two enzymes. The most prominent phenotypic changes related to NatA ablation instead affect gene expression, genome integrity and metabolic regulation. NatA is specifically involved in gene silencing to control sexual differentiation and mating, and expression of mitochondrial genes. natAΔ cells, due to their incapability to Nt-acetylate Sir3, fail to suppress the HML locus and artificially express not only the MATa-specific genes MATa1 and MATa2 from the HMR locus but also the MATα-specific genes MAT1 and MAT2 from the HML locus, resulting in a pseudodiploid, sterile phenotype. The functional impairment of Sir3 in natAΔ cells most likely also causes the repression of transposable elements (Elder et al., 1981;Bilanchone et al., 1993), while defective Nt-acetylation of Orc1, a component of the DNA-binding Origin Recognition Complex (ORC), confers the enhanced expression of sub-telomeric genes (Geissenhöner et al., 2004). The increased expression of mitochondrial genes may represent the cellular response to reduced mitochondrial activity. The identification of the underlying mitochondrial defects and involved NatA substrates requires further investigation.
We speculate that NatA is particularly responsive to environmental cues and confers full acetyltransferase activity only when acetyl-CoA levels are high, i.e. upon growth in glucose rich media. A recent study confirmed changed Nt-acetylation levels of Nat substrates in a protein-specific and not global manner under growth conditions conferring low acetyl-CoA levels (Varland et al., 2018). Overall, the selective effect on fewer protein targets (as compared to NatB) qualifies NatA to play a regulatory role in specific cell responses. As consequence, exponentially growing WT cells may express transposon genes, suppress expression of sub-telomeric genes and allow mating to efficiently protect and reshape the genome, with impact on the genetic diversity of the population, while this is prevented under conditions of perturbed proteostasis and metabolism, with signals transmitted via NatA.
In summary, the two major N-terminal acetyltransferases of yeast, NatA and NatB, have a broad range of functions in maintaining proteome integrity and stress signaling, executed by a clear division of labor between the two enzymes. It is an intriguing possibility that the evolution of NatA and NatB is not only driven by the need for distinct catalytic activities to modify the entire spectrum of N-termini of proteins, but also to elicit enzyme-specific cellular responses.

Strain and Plasmid Construction
The Saccharomyces cerevisiae S288C strain BY4741 served as background for all generated strains in this study. Gene deletion was performed via homologous recombination of the complete indicated ORF by one of the selection markers kanMX4, natMX6, hphNT1, K.l.URA3 as described previously (Janke et al., 2004). Knockouts were confirmed by growth resistance to indicated antibiotics/ auxotrophies and colony PCR. For complementation with WT or mutant genes, the plasmids pRS416 and p413CYC1 were used. The required genes were generated by PCR using yeast gDNA and respective primer pairs. Double digestion with SpeI-NotI (for pRS416) or SmaI-SalI (for p413CYC1) was performed for the plasmids as well as the PCR products. Corresponding plasmid backbone and insert were purified by agarose gel electrophoresis, ligated with T4 ligase and transformed into chemically competent E. coli cells.
After selection on ampicillin plates, plasmids were extracted and sequenced to confirm correct gene insertion. Verified plasmids were used for transformation into yeast knockout strains. Mutants of NAA10 genes were generated by amplification of the generated plasmid using mismatch primers, followed by digestion of the template DNA using Dpn1, transformation, plasmid purification, and sequencing as described earlier.
All yeast strains were grown on full media (YPD) or synthetic defined media (SCD/ SCG) complemented by 2% glucose (SCD) or 2% glycerol (SCG). Heat treated cells were shifted from 30°C to 37°C two doublings prior to harvest. Otherwise cells were grown at 30°C and 130 rpm in liquid culture or at 30°C on agar plates. Used primers, generated plasmids and strains are given in Tables S4-S6, respectively.

Gene Duplication and Tagging
Genes were duplicated and tagged with a single step PCR protocol following the strategy developed by Huber and coworkers . Using primers 17-26 (Supplementary Table 5) and plasmid pMaM61 (Khmelinskii et al., 2012), a PCR construct was generated that comprises a tFT-tag, a marker (natMX4) to select for successful cloning, and a 3' and 5' overhang of 55 nt aligning downstream and upstream to the gene of interest, respectively. This construct was introduced into WT yeast cells by transformation according to standard protocols (Gietz & Schiestl, 2007) and followed by deletion of NatA and NatB, respectively. Verified strains were grown to an OD 600nm of 0.5 and harvested by centrifugation at 8,000xg for 1 min at RT. Thorough washing with 1 ml 1x PBS removed residual medium. 100,000 cells of each sample were measured by FACS and analyzed as described previously (Khmelinskii et al., 2016). The median of 2 replicates are shown and statistically analyzed as indicated.

Growth Assay
The growth behavior of all strains used in this study was analyzed in liquid culture and on solid agar plates. For both methods, a pre-culture was grown to mid-log or early stationary phase and washed three times with dH 2 O (5000xg, 3 min, RT) if medium or carbon source was switched. Spot assays were prepared by stamping 1:10 serial dilutions on respective agar plates. Main liquid cultures were prepared as 200 µl cultures with a start OD 600nm of 0.1 and incubated in clear 12-well or 96-well plates, at 30°C and under continuous shaking (600 rpm). The optical density was measure every 5-20 min in a plate reader (FLUOstar Omega or SPECTROstar Nano, BMG). OD calibration and doubling time calculation was performed as described by Fernandez-Ricaud and coworkers (Fernandez-Ricaud et al., 2007). Each strain was measured in at least three technical and three biological replicates.
spectral library was generated as described in Schubert et al., 2015. The SWATH MS data was extracted with openSWATH (Röst et al., 2014) and aligned with TRIC (Röst et al., 2016). The final protein matrix and statistical analysis was generated using mapDIA (Teo et al., 2015). The raw data and analysis results have been deposited to the ProteomeXchange Consortium via PRIDE (Perez-Riverol et al., 2019) with the dataset identifier PXD015217.

Tandem Fluorescent Timer Assay
Tandem fluorescent timer analysis was performed as described (Khmelinskii et al., 2014), with the following specifications: The screen was done in 1536 format using pinning robotics (Singer instruments) with three biological replicates per clone. Haploids that have both tFT gene fusion and natBΔ mutation were grown on synthetic complete media without histidine (for haploid selection) for 1 day followed by high throughput fluorescence measurements of colonies for sfGFP and mCherry intensities. To test the effect of heat stress, plates were incubated at 30°C for 1 day then shifted to 37°C for 1 day followed by high throughput fluorescence measurement of colonies using Infinite M1000 or Infinite M1000 Pro plate readers equipped with stackers for automated plate loading (Tecan) and custom temperature control chambers. Data analysis for the measured fluorescence intensities was performed as described (Khmelinskii et al., 2014).