Microscopy-based Saccharomyces cerevisiae complementation model reveals functional conservation and redundancy of N-terminal acetyltransferases

N-terminal acetylation is a highly abundant protein modification catalyzed by N-terminal acetyltransferases (NATs) NatA-NatG. The Saccharomyces cerevisiae protein Arl3 depends on interaction with Sys1 for its localization to the Golgi and this targeting strictly requires NatC-mediated N-terminal acetylation of Arl3. We utilized the Arl3 acetylation-dependent localization phenotype as a model system for assessing the functional conservation and in vivo redundancy of several human NATs. The catalytic subunit of human NatC, hNaa30 (Mak3), restored Arl3 localization in the absence of yNaa30, but only in the presence of either yeast or human Naa35 subunit (Mak10). In contrast, hNaa35 was not able to replace its yeast orthologue without the co-expression of hNaa30, suggesting co-evolution of the two NatC subunits. The most recently discovered and organellar human NAT, NatF/Naa60, restored the Golgi localization of Arl3 in the absence of yNaa30. Interestingly, this was also true for hNaa60 lacking its membrane-binding domain whereas hNaa50 did not complement NatC function. This in vivo redundancy reflects NatC and NatF´s overlapping in vitro substrate specificities. The yeast model presented here provides a robust and rapid readout of NatC and NatF activity in vivo, and revealed evolutionary conservation of the NatC complex and redundancy between NatC and NatF.


Results and Discussion
Functional conservation of Naa30 revealed by Arl3 localization complementation assay. The dependency of NatC-mediated Nt-acetylation of Arl3 for correct subcellular localization has previously been shown by fluorescence microscopy and mass spectrometry 12,13 . Here, we utilized the Arl3 localization as a model for NatC activity in vivo (Fig. 1). In WT cells, Arl3-GFP localized to the Golgi whereas in naa30Δ and naa35Δ cells the punctate localization was lost (Fig. 1A), as previously reported 12,13 . Yeast Naa38 was shown to be dispensable for Nt-acetylation of Arl3 by mass spectrometry 12 and we accordingly observed the subcellular localization of naa38Δ yeast to be similar to the WT (Fig. 1A).
Next, we used the Arl3 localization model in various rescue assays and found the human orthologue of yNaa30 to be functionally conserved, as it was able to restore the mislocalization of Arl3-GFP in naa30Δ yeast (Fig. 1B). Previously, hNaa30 was reported to restore subcellular localization of Imh1, downstream of Arl3 function 13 . To investigate whether the complementation ability of hNaa30 was due to the catalytic subunit performing Nt-acetylation on its own or in a human-yeast interspecies complex with the other essential NatC subunit, yNaa35, we expressed HA-hNaa30 (hereafter referred to as hNaa30) in naa35Δ cells. In this yeast background, exogenously expressed hNaa30 failed to restore Arl3 localization (Fig. 1B), thus suggesting interspecies NatC functionality of hNaa30-yNaa35.
Human Naa35 is unable to complement yeast Naa35 for Arl3 Nt-acetylation. The functional conservation of hNaa35 was then addressed by expressing hNaa35-FLAG (hereafter referred to as hNaa35) in naa35Δ cells, but this did not restore Arl3 localization. To exclude the possibility of yNaa38 hindering a necessary binding between hNaa35 and yNaa30 we further expressed hNaa35 in naa35Δ naa38Δ yeast. No rescue of Arl3 localization was seen in this condition either (Fig. 1B). Additionally, we performed the complementation assay using hNaa35 without the C-terminal FLAG-tag, which was also negative in terms of restoring Arl3 punctate localization (Fig. 2).
Taken together, these data demonstrate that hNaa30, but not hNaa35, can replace the corresponding yeast orthologue. Whereas hNaa30 functions in the presence of yNaa35, hNaa35 cannot function despite the presence of yNaa30, thus suggesting the formation of a functional hNaa30-yNaa35 interspecies complex, but not yNaa30-hNaa35.
Human Naa30 and Naa35 cooperate to perform NatC activity in yeast. Based on the above results showing that co-expression of hNaa30 and yNaa35 restore NatC function and the inability of yNaa30 and hNaa35 to do so, we proceeded by studying both human NatC subunits in yeast. Human Naa30 and Naa35 were exogenously co-expressed in three different yeast strains: naa35Δ , naa30Δ naa35Δ , and naa30Δ naa35Δ naa38Δ (NatCΔ ) (Fig. 1D). For all deletion strains, a clear rescue of Arl3 localization was observed, indicating that NatC activity was present (Fig. 1C). Thus, the two human subunits Naa30 and Naa35 cooperated and performed NatC activity on a yeast substrate protein in vivo. From this, we suggest a functional conservation of the Naa30-Naa35 NatC complex between S. cerevisiae and humans, and that there has been a co-evolution of these subunits since heterologous expression of yNaa30 and hNaa35 is non-functional. Our observation that human Naa30-Naa35 NatC complex is functionally active in yeast and complements yNatC, is analogous to earlier complementation studies of NatA and NatB 3,21 , and suggests a large degree of conservation of these major NATs from yeast to humans. However, the yeast/human Naa30 complementation observed here, differs from those of NatA and NatB studies in which the human catalytic subunits alone (Naa10 and Naa20, respectively) cannot phenotypically complement a yeast strain lacking its yeast orthologue 3,21 . The punctate localization pattern of Arl3 corresponds to Golgi structures. To verify that the observed localization pattern of Arl3 in the rescue assays did indeed represent restoration of the Arl3 WT localization, the trans-Golgi Sec7-mRFP and cis-Golgi mCherry-Sed5 marker proteins were individually expressed in Arl3-GFP cells. Both Golgi markers partially co-localized with Arl3-GFP in the WT, apparently with equal amount of overlap (Fig. 3). We further examined the restored punctate Arl3-GFP localization in naa30Δ cells expressing hNaa30 and detected similar co-localization with mCherry-Sed5 as for the WT yeast (Fig. 3B). This confirmed that the reappeared punctate structures detected represents restored Arl3 Golgi localization.
Overlapping in vivo activities of human NatC and NatF in yeast. Human NatC, NatE and NatF display overlapping in vitro substrate specificities 4,22,23 . Furthermore, human NatE and NatF exhibit overlapping in vivo substrate profiles when expressed in yeast 4,24 . All three enzymes recognize Met-starting N-termini containing a hydrophobic or amphipathic amino acid residue in the second position. Arl3 has a Met-Phe-starting N-terminus and thus may be a putative target of these enzymes. To challenge the suggested substrate redundancy between human NatC, NatE and NatF in vivo, we again utilized the Arl3 complementation model (Fig. 4). Here, the human catalytic subunits Naa30, Naa50 and Naa60 were individually expressed in Arl3-GFP naa30Δ cells. Human Naa50, which in human cells is cytosolic and partially ribosome-associated 23,28 , was not able to  complement yNaa30 whereas hNaa60 almost completely reinstated the punctate Arl3 localization, similar to hNaa30 (Fig. 4A). The same was true for hNaa60 expressed in naa35Δ cells (Fig. 4C), indicating that hNaa60 may operate independently of any of the yNatC subunits, unlike hNaa30, which requires the assumed ribosome binding subunit Naa35.
Naa60 was recently described in human cells as the first organellar NAT and its catalytic activity is facing towards the cytosol 5 . It might be possible that the ability of human Naa60 to complement the NatC activity on Arl3 relates to Naa60´s association with the yeast Golgi membrane. However, upon expression of truncated Naa60 del-MEM , without the C-terminal end that mediates Naa60 membrane-association in mammalian cells, (ref. 5), the Arl3 localization pattern was again represented as punctate structures that partially co-localized with the cis-Golgi marker Sed5 (Figs 4C and 5A). Human Naa60 and Naa60 del-MEM fused to a C-terminal EGFP tag mainly localized to punctate structures and the cytosol, respectively (Fig. 5B). Taken together these data may indicate a post-translational activity of NatF/Naa60 towards Arl3-GFP at the Golgi or freely in the cytosol. In any case, the data verify Met-Phe as an in vivo N-terminal target sequence of NatF and demonstrate the ability of human Naa30 and Naa60 to act on the same substrate in vivo 4 .
From these results, it is clear that human Naa60 and Naa60 del-MEM , but not hNaa50, are able to rescue the Arl3 localization phenotype, thus enlightening an interesting Nt-acetylation performance of NatF as well as functional redundancy with NatC. Moreover, human Naa50 acts as a negative control in the assay for Arl3 localization, excluding mere NAT overexpression to be sufficient for phenotype rescue.

Conclusions
We applied S. cerevisiae as a model system to study the in vivo function of human NATs. The NatC-dependent subcellular localization of yeast Arl3 proved to be a suitable model for studying functional conservation among NatC orthologues as well as diverging in vivo function among NatC, NatE and NatF harbouring overlapping in vitro substrate specificities (Fig. 6). Naa30 was defined as functionally conserved between yeast and humans through its ability to complement Arl3 mislocalization, which interestingly required the presence of either yeast or human Naa35; accordingly we propose that NatC complex formation occurred (Fig. 6A). Human Naa35 on the other hand, could not complement its yeast orthologue and we suggest this to be due to insufficient binding between yNaa30 and hNaa35 or formation of an impaired complex (Fig. 6B). Taken together, these results indicate that the two subunits Naa30 and Naa35 have co-evolved.
Despite human Naa30, Naa50 and Naa60's shared catalytic activity towards specific N-terminal sequences, we found opposite complementation patterns of NatE/Naa50 and NatF/Naa60 for the NatC-deficient phenotype of Arl3. This clearly hints to unique in vivo functions of the two. Furthermore, in light of the different subcellular localizations of Naa60 and Naa60 del-MEM in human cells 5 and in yeast cells, both being able to rescue the Arl3 localization (Fig. 6C), the difference between Naa50 and Naa60 may reflect differences in the substrate binding site rather than differences in subcellular localization. This assumption is also based on the crystal structures and catalytic efficiencies of human Naa50 29 and Naa60 30 suggesting that Naa60 may accommodate larger and more bulky substrate side chains as compared to Naa50.
The yeast model presented here may act as a useful and rapid in vivo setup to assess the functionality of various NAT subunits, meaning isoforms of human Naa30, Naa35 and Naa60. For a Naa10 mutant causing the Ogden syndrome 31,32 , a yeast model provided useful insights into the impaired functionality of NatA 33 Similarly, the yeast model described here may be used in any future studies of pathological mutants of NatC or NatF.

Methods
Yeast Cultivation. The yeast strains used in this study (see Supplementary Table S1) are derived from one of the two parental strains BY4741 (MATa his3Δ1 leu2Δ0 met15Δ0 ura3Δ0) and BY4742 (MATα his3Δ1 leu2Δ0  lys2Δ0 ura3Δ0). According to their genotype and required selection, the cells were either grown in YPD (Sigma-Aldrich, YPD Broth #Y1375 and YPD Agar #Y1500) or appropriate synthetic defined dropout media (Sunrise  In WT yeast cells Arl3 is N-terminally acetylated by the NatC complex (Naa30 and Naa35, whereas Naa38 is dispensable for Arl3-acetylation) and targeted to the Golgi transmembrane protein Sys1, mediating Arl3 punctate Golgi localization. However, in the absence of a functional NatC complex Arl3 remains unacetylated after protein synthesis and the targeting is abrogated, leading to Arl3 cytosolic mislocalization 12,13 . (A) Human Naa30 is able to function in complex with yNaa35 (and possibly yNaa38) or hNaa35, indicating a functionally conserved catalytic subunit between yeast and humans. WT yNatC is depicted framed in dotted line. Science Products, SD-URA #1703-500, SD-LEU #1707-500, SD-LEU-URA #1721-500, Agar #1910-500). Protein expression of mCherry-Sed5 under the control of the MET25 promoter required growth medium with maximum 10 μ g/ml methionine.

Plasmid construction.
A record of all plasmids and primers used is found in Supplementary Table S2 and   Table S3, respectively. Human NAA30 (Gene ID 122830) was sub-cloned from pRS315-HA-hMAK3, which was a kind gift from Professor Sean Munro, MRC, Cambridge 13 , into the pBEVY-U yeast vector 34 downstream of the ADH1 promoter using restriction enzyme sites XmaI (5´) and EcoRI (3´). The N-terminal HA-tag was retained from the original gene construct.
Human NAA35 (Gene ID 60560) was amplified from pCMV6-AC-NAA35-GFP (OriGene, #RG213022) with the insertion of XbaI (5´) and PstI (3´) restriction enzyme sites as well as a C-terminal FLAG-tag and ligated into pBEVY-L after the GPD promoter. Additionally, pBEVY-L-hNAA35 was generated by introducing a stop codon downstream of the NAA35 ORF by mutagenesis (Agilent Technologies, #210514).

Yeast transformation. Yeast transformation was based on a method developed by Gietz and Schiestl 37 .
Exponentially growing yeast cells (10 ml culture, OD 600 1.0) were harvested, washed twice with water and resuspended in 240 μ l 50% PEG-3350 before addition of 36 μ l 1 M lithium acetate, 100 μ g salmon sperm ssDNA, DNA template (10 μ g gene deletion cassette or 5 μ g plasmid), and water up to 360 μ l. After a 40 min heat-shock at 42 °C the cells were washed four times with water, before plating or incubation in YPD, depending on selection.
Yeast gene deletions were performed by homologous recombination and a complete overview of the primers used to amplify the respective deletion cassettes as well as the PCR-based screen for positive mutants are listed in Supplementary Table S3. Protein extraction and Western blot analysis. Protein extraction was done according to Kushnirov 38 with modifications. Exponentially growing yeast cells (4 ml yeast culture, OD 600 1.0) were harvested for 10 min at 5,000 × g. The pellet was washed twice with water, resuspended in 200 μ l water before adding 200 μ l 0.2 M NaOH for 5 min. Cells were harvested for 1 min at 12,000 × g, resuspended and boiled for 3 min in 100 μ l 1x Sample buffer (10% glycerol, 2% SDS, 50 mM Tris-HCl, pH 6.8, 0.0025% bromophenol blue, 50 mM dithiothreitol), before pelleting cell debris. Typically, 6 μ l supernatant was loaded for SDS-PAGE on 4-20% Mini-PROTEAN ® gels (Bio-Rad, #456-8093).
Nt-acetylation activity of hNaa30 is observed in the absence of yeast or human Naa35, which is thought to be the ribosome-binding subunit. Human Naa35 cannot replace its yeast orthologue and function together with yNaa30, neither in the presence nor absence of yNaa38. (C) Naa30 and Naa60 exhibit in vivo redundancy. The Arl3 localization phenotype is complemented by human Naa60 and Naa60 del-MEM in both naa30Δ and naa35Δ yeast. Based on the subcellular localizations of human Naa60 and Naa60 del-MEM , a possible post-translational NAT activity at the cytosolic side of Golgi or freely in the cytosol is suggested. Abbreviations: y30, yNaa30; y35, yNaa35; y38, yNaa38; h30, hNaa30; h35, hNaa35; h50, hNaa50; h60, hNaa60; h60del-MEM, hNaa60 without membrane binding region; Ac-CoA, acetyl-coenzyme A.