N-terminal (Nt) acetylation is a highly abundant protein modification, occurring on approximately 80% of the human proteome (Arnesen et al., 2009; Aksnes et al., 2015). The process is catalyzed by Nt acetyltransferases (NATs), transferring an acetyl group from Acetyl CoA to the amino group of the first amino acid in the protein sequence. To date seven NATs have been found in human cells, NatA-NatF, and NatH, having distinctive features in terms of subunit composition, subcellular localization and substrate specificity (Aksnes et al., 2019). Some NATs, NatA, NatB and NatC, have broad substrate pools and act co-translationally. Other NATs have a very specific substrate pool, such as NatD co-translationally acetylating histones H2A and H4 (Songkyu et al., 2003; Hole et al., 2011), NatF post-translationally acetylating transmembrane proteins (Van Damme et al., 2011; Aksnes et al., 2015), or NatH/NAA80 post-translationally acetylating actins (Drazic et al., 2018).

The first catalytic mechanism of a human NAT (Evjenth et al., 2012), as well as the first structures of a NAT and NAT-complex bound to its peptide substrate, have been presented (Liszczak et al., 2011; Liszczak et al., 2013; Støve et al., 2016). Taking advantage of this information, we and others have developed peptidic NAT inhibitors that are based on covalently linking mimics of the two substrates of the biochemical reaction, i.e., coenzyme A (CoA) and a short peptide carrying an N-terminal bromoacetyl group (Figure 1) giving a thioether-linked acetyl moiety (Foyn et al., 2013; Støve et al., 2016; Goris et al., 2018; Deng et al., 2021). In these studies it has been found that it is sufficient to use the four amino acids from the N-terminus of the protein substrates to obtain potent bisubstrate inhibitors.

Such bisubstrate inhibitors demonstrate specificity and significant inhibitory potential with IC₅₀ and K_i in the low μM/nM range. For NatD, increasing the length of the acetyl linker to propionyl has proven beneficial for the design of highly potent inhibitors (Deng et al., 2021). Further, for most NATs, access to bisubstrate inhibitors has led to significant advances in the field by enabling the determination of protein crystal structures (Liszczak et al., 2013; Støve et al., 2016; Hong et al., 2017; Goris et al., 2018; Deng et al., 2020; Deng et al., 2023).

Actin is a major component of the cytoskeleton and cytoskeletal dynamics are important for several cellular activities such as cell motility, division, and intracellular trafficking (Pollard and Cooper, 2009). The cellular activity of actin is connected to its dynamic transition between monomeric (G-actin) and filamentous (F-actin) forms. This is regulated in cells by actin-binding proteins as well as by post-translational modifications. It was established 4 decades ago that animal actins undergo a unique Nt-maturation process (Redman and Rubenstein, 1981; Rubenstein and Martin, 1983). Now we know that this involves the following steps (Arnesen and Aksnes, 2023): a) a general Nt-processing step for all eukaryotic proteins of co-translational N-terminal acetylation (Aksnes et al., 2019) (preceded by methionine excision for muscle actins), b) a unique animal actin-specific post-translational cleavage of the Nt-acetylated residue by ACTMAP (Haahr et al., 2022) and finally c) an animal actin-specific post-translational Nt-acetylation by NAA80/NatH to generate acidic actin Nt-termini (Drazic et al., 2018; Goris et al., 2018; Wiame et al., 2018). NAA80 binding to profilin, specifically PFN2, primes the interaction with actin monomers and the acetylation of actin’s N-terminus which sticks out from the folded actin monomer (Rebowski et al., 2020; Ree et al., 2020). In human cells lacking NAA80, actin is not Nt-acetylated (near 0% Nt-acetylation) while in the presence of NAA80, actin is acetylated to a near 100% stoichiometry (Drazic et al., 2018; Drazic et al., 2022). This suggests that actin Nt-acetylation is essential for optimal actin functionality in humans. Indeed, human NAA80-KO cells where actin is unacetylated display fragmentation of the Golgi apparatus, altered cytoskeletal organization including decreased G-actin/F-actin ratio, increased F-actin, increased cell size, increased filopodia and lamellopodia and increased cell migration (Aksnes et al., 2018; Drazic et al., 2018; Beigl et al., 2020). The physiological impact of actin Nt-acetylation is not fully understood due to the lack of NAA80 KO animal models. However, two brothers carrying a homozygous NAA80 variant resulting in a partial reduction in cellular actin Nt-acetylation were recently presented (Muffels et al., 2021). These individuals showed hearing loss, mild muscle weakness and developmental delay.

Bisubstrate inhibitors of NAT enzymes could potentially be used as tools to study the effects of inhibiting Nt-acetylation to shed more light on the roles of Nt-acetylation in the cell. NAA80 inhibitors could specifically be applied to manipulate cytoskeletal dynamics and to increase cell migration. Herein we describe the optimization of a bisubstrate inhibitor for NAA80.

The NATs constitute an important class of enzymes and potent inhibitors can prove to be useful as tool compounds to study their roles both in vitro and in vivo. To identify such a tool compound for NAA80, we have in this work optimized the earlier identified bisubstrate inhibitor CoA-Ac-DDDI-NH₂ (see Figure 1 for general structure). In our earlier work we tested the in vitro activity of NAA80 toward a broad library of potential substrates such as amino acids, nucleosides, coenzymes, various amines (e.g., serotonin and spermidine), vitamins and a number of 24-mer peptides representing the N-terminal part of proteins (Goris et al., 2018). In these substartes, the four N-terminal amino acids were varied while the remaining peptide sequence was kept constant. We found that for all of these potential substrates, NAA80 only acetylated three peptides, MDEL₂₄, DDDI₂₄ and EEEI_24. DDDI₂₄ and EEEI₂₄ correspond to the N-terminal part of β- and γ-actin in their cytosolic processed forms, respectively, while MDEL₂₄ represents the unprocessed N-terminal part of the protein p65. The level of product formation for actylation of MDEL₂₄, DDDI₂₄ and EEEI₂₄ was determined to be 119 ± 5.61 μM, 50.5 ± 1.0 μM, and 42.4 ± 0.95 μM, respectively (Goris et al., 2018). In our inhibitor design, we decided to continue using a tetrapeptide for the protein-mimicking part as previous studies have shown that these are the most important residues for inhibitory activity (Liszczak et al., 2011; Foyn et al., 2013; Liszczak et al., 2013; Støve et al., 2016; Hong et al., 2017; Goris et al., 2018; Deng et al., 2020; Deng et al., 2021; Deng et al., 2023). Co-crystal structures of NAT-bisubstrate inhibitor complexes typically show that these four amino acids are most important for protein-inhibitor binding interactions. All inhibitors were prepared with the C-terminal carboxylic acid capped as amides to avoid the negative charge of a C-terminal carboxylic acid and better mimic a protein N-terminal.

The peptide part of all inhibitors was prepared using Fmoc-based solid phase peptide synthesis where the C-terminal amino acid was loaded onto a Rink amide resin (Scheme 1). After removal of the Fmoc-group of the N-terminal amino acid, coupling with bromoacetic acid introduced the acetyl-linker. The bromoacetyl peptide was then deprotected and cleaved from the resin and purified by RP-HPLC. Next, triethylammonium bicarbonate buffer (pH 8.5) was used to facilitate the conjugation of CoA to the bromoacetyl peptide. The final bisubstrate inhibitors were purified by RP-HPLC and characterized by MS and NMR to confirm their structures (see Table 1; Supplementary Table S1).

We have already reported that bisubstrate inhibitors based on the N-terminus of processed β- and γ-actin (DDDI and EEEI) display IC₅₀ values of 0.38 and 1.16 μM, respectively (Table 2). The difference in inhibitory potency was somewhat higher than was expected based on the small difference in how well NAA80 acetylated the 24-mer peptides DDDI₂₄ and EEEI₂₄ as their four N-terminal amino acids. Surprisingly, while the MDEL₂₄ peptide showed a higher degree of acetylation than the DDDI₂₄ and EEEI₂₄ peptides when we screened for potential substrates, we found that the inhibitor based on the MDEL sequence was less potent and displayed an IC₅₀ value of 1.26 μM.

To elucidate inhibitor binding we have previously solved the crystal structure of DmNAA80 bound to a DDDI bisubstrate inhibitor (Goris et al., 2018). This structure revealed that mainly the α1-α2 region of NAA80 contributed to peptide binding in the inhibitor, while the β6-β7 loop which is important for peptide binding in many of the other NATs, is slightly shifted away from the peptide without contributing to inhibitor binding (Figure 2A). More specifically, D1 of the inhibitor was stabilised by a hydrogen bond from the backbone oxygen of D1 to the backbone nitrogen in S88, while the D1 side chain was stabilized by hydrogen bonds to the side chain of S124, and through a water-mediated hydrogen bond to T125 and I126. D2 of the inhibitor formed a hydrogen bond with W36 and a salt bridge with R43. Further, D3 made a water-mediated contact with R43 and hydrogen bonds with S46 and S48 and finally, the backbone oxygen of I4 formed a hydrogen bonds with R38. As can be seen from the multiple sequence alignments of DmNAA80 and hNAA80 (Figure 2B), most of the key contributing residues to binding in the DmNAA80-DDDI structure and hNAA80-profilin-actin ternary complex (W36, R43, R38 in DmNAA80 iso2 (Goris et al., 2018) corresponding to W105, R107 and R112 in hNAA80 iso 2 (Rebowski et al., 2020) is conserved. However, a key difference between the two is the extended region around the β6-β7 loop in hNAA80, a region that has shown importance for substrate binding in other NATs (Liszczak et al., 2011; Liszczak et al., 2013; Støve et al., 2016; Hong et al., 2017; Goris et al., 2018), and likely further contributes to optimimal binding of the peptide part of the bisubstrate analogue and increased potency of these inhibitors.

Since the DDDI-inhibitor was found to be significantly more potent that the EEEI-inhibitor, we decided to investigate whether the introduction of Asp into the EEEI sequence would increase inhibitory activity. It turned out that the introduction of Asp in position three was the most beneficial and the inhibitor based on the EEDI sequence showed almost ten-fold higher inhibitor activity compared to EEEI (Table 2). The EEDI sequence also proved to outperform DDDI and was found to inhibit NAA80 with an IC₅₀ of 0.17 μM. Inhibitors based on the DEEI and EDEI sequences gave comparable or slightly improved IC₅₀ values respectively compared to the EEEI inhibitor.

We also tested replacing the Ile residue with a Leu and found that the EEEI and EEEL sequences gave inhibitors with equal potency. On the other hand, the DEEL and EEDL sequences gave however a loss of activity compared to DEEI and EEDI while the inhibitor based on EDEL showed a four-to five-fold increase in potency compared to EDEI. Replacing Glu in position three of the EDEL sequence with a Gln residue led to an increase in IC₅₀ from 0.15 μM to 0.85 μM and the replacement of the Asp residue with Ser led to a dramatic loss of potency. Interestingly, a PDEL sequence proved to give only slightly lower inhibitory potency compared to the ESEL sequence.

The introduction of two Asp residues into the EEEI sequence gave inhibitors with a large difference in inhibitory potency as the DDEI sequence gave an IC₅₀ of 2.67 μM, where CoA-Ac-EDDI-NH₂ turned out to be the most potent inhibitor with an IC₅₀ of 0.12 μM.

The NAA80/NatH is a particular case among the NAT enzymes in having actin as its sole substrate. The various cytoplasmic and muscle actin substrates all have different combinations of Asp and Glu in the first N-terminal positions (Aksnes et al., 2019). Here, we show that using specific combinations of Asp and Glu in the peptide sequence makes us able to identify a highly potent NAA80 inhibitor. In conclusion, we have found that for NAA80, the choice of acidic residue at different positions of the peptide part of bisubstrate inhibitors is important and varying these has led to an optimized inhibitor with up to 3 fold higher inhibitory activity compared to those previously described. The CoA-Ac-EDDI-NH₂ was the most potent NAA80 bisubstrate inhibitor with an IC₅₀ of 0.12 μM. This inhibitor should be useful in further in vivo studies by micro injection in Danio rerio and for cell-based studies by micro injection into human cells to elucidate NAA80 function and the effects upon Nt-acetylation of actin.