Post-translational Lysine Ac(et)ylation in Bacteria: A Biochemical, Structural, and Synthetic Biological Perspective

Abstract

Ac(et)ylation is a post-translational modification present in all domains of life. First identified in mammals in histones to regulate RNA synthesis, today it is known that is regulates fundamental cellular processes also in bacteria: transcription, translation, metabolism, cell motility. Ac(et)ylation can occur at the ε-amino group of lysine side chains or at the α-amino group of a protein. Furthermore small molecules such as polyamines and antibiotics can be acetylated and deacetylated enzymatically at amino groups. While much research focused on N-(ε)-ac(et)ylation of lysine side chains, much less is known about the occurrence, the regulation and the physiological roles on N-(α)-ac(et)ylation of protein amino termini in bacteria. Lysine ac(et)ylation was shown to affect protein function by various mechanisms ranging from quenching of the positive charge, increasing the lysine side chains' size affecting the protein surface complementarity, increasing the hydrophobicity and by interfering with other post-translational modifications. While N-(ε)-lysine ac(et)ylation was shown to be reversible, dynamically regulated by lysine acetyltransferases and lysine deacetylases, for N-(α)-ac(et)ylation only N-terminal acetyltransferases were identified and so far no deacetylases were discovered neither in bacteria nor in mammals. To this end, N-terminal ac(et)ylation is regarded as being irreversible. Besides enzymatic ac(et)ylation, recent data showed that ac(et)ylation of lysine side chains and of the proteins N-termini can also occur non-enzymatically by the high-energy molecules acetyl-coenzyme A and acetyl-phosphate. Acetyl-phosphate is supposed to be the key molecule that drives non-enzymatic ac(et)ylation in bacteria. Non-enzymatic ac(et)ylation can occur site-specifically with both, the protein primary sequence and the three dimensional structure affecting its efficiency. Ac(et)ylation is tightly controlled by the cellular metabolic state as acetyltransferases use ac(et)yl-CoA as donor molecule for the ac(et)ylation and sirtuin deacetylases use NAD⁺ as co-substrate for the deac(et)ylation. Moreover, the accumulation of ac(et)yl-CoA and acetyl-phosphate is dependent on the cellular metabolic state. This constitutes a feedback control mechanism as activities of many metabolic enzymes were shown to be regulated by lysine ac(et)ylation. Our knowledge on lysine ac(et)ylation significantly increased in the last decade predominantly due to the huge methodological advances that were made in fields such as mass-spectrometry, structural biology and synthetic biology. This also includes the identification of additional acylations occurring on lysine side chains with supposedly different regulatory potential. This review highlights recent advances in the research field. Our knowledge on enzymatic regulation of lysine ac(et)ylation will be summarized with a special focus on structural and mechanistic characterization of the enzymes, the mechanisms underlying non-enzymatic/chemical ac(et)ylation are explained, recent technological progress in the field are presented and selected examples highlighting the important physiological roles of lysine ac(et)ylation are summarized.

Keywords: genetic code expansion; lysine acetylation; lysine acetyltransferases; lysine deacetylases; sirtuin.

FIGURE 1

Post-translational lysine ac(et)ylation is dynamically regulated by lysine deac(et)ylases and N-(α)-/N-(ε)-ac(et)yltransferases. (A) The ε-amino group of lysine side chains and the α-amino group at the protein N-termini can be ac(et)ylated by lysine acetyltransferases (KATs) using acetyl-CoA and/or further acyl-CoA donor molecules for the ac(et)ylation. So far, all bacterial protein acetyltransferases belong to the Gcn5-related N-terminal acetyltransferases (GNAT). Next to the enzymatic ac(et)ylation, lysine side chains and protein N-termini can ac(et)ylated non-enzymatically by ac(et)yl-CoA and acetyl-phosphate, the major source for non-enzymatic acetylation in bacteria. Bacteria use Zn²⁺-dependent classical lysine deac(et)ylases and NAD⁺-dependent sirtuin deac(et)ylases to catalyze the deac(et)ylation of lysine side chains. (B) Structures of amino acids used to study lysine acetylation. L-glutamine is often used to mimic lysine acetylation and L-arginine to conserve a non-acetylated state in studies performed in vivo. Trifluoroacetyl-L-lysine and thioacetyl-L-lysine are used as mechanistic inhibitors for sirtuins to stabilize the acetylation at an analyzed site. Notably, these analogs can be potently deacetylated by some classical deacetylases. (C) Diverse acylations identified to occur at lysine side chains in bacteria. Lysine side chains can be modified by various aliphatic or negatively charged acylations in bacteria. Further acylations might be discovered in future. To which extend acetyltransferases are capable to catalyze acylation of the lysine side chains and/or protein N-termini needs further investigation. In general, all acyl-CoA thioesters generated in metabolism can be transferred to the ε-amino group of lysine side chains and/or the α-amino group at the protein N-termini in terms of an non-enzymatic reaction. (D) Polyamines in bacteria shown to be acetylated by acetyltransferases and deacetylated by classical deacetylases. These polyamines might form buffers for acetyl-groups to avoid systemic non-enzymatic protein acetylation. The acetyl-groups can be transferred from acetyl-CoA by the action of polyamine specific acetyltransferases. (E) Increasing the reactivity at the N-(ε)-amino group of lysine side chains or the N-(α)-amino group of the protein N-termini for enzymatic or non-enzymatic ac(et)ylation. Deprotonation of the α- or ε-amino groups by protein acetyltransferases in an important step for acetyl-group transfer from the ac(et)yl-CoA donor molecule during catalysis. A deprotonation can also occur non-enzymatically and is supported by the presence of the lysine side chain in a poly-basic sequence context resulting in the reduction of the substrate lysine side chain’s pK_a value. Moreover, non-enzymatic ac(et)ylation is preferred under alkaline conditions and under high concentrations of the reactive ac(et)yl-CoA thioesters. A deprotonation of the substrate amino group results in an increase in its nucleophilicity for attack of the ac(et)yl-CoA thioesters.

FIGURE 2

Enzymatic protein ac(et)ylation is catalyzed by GNAT ac(et)yltransferases in bacteria. (A) Structure of the E. coli GNAT acetyltransferase RimI in complex with acetyl-CoA (PDB: 2CNS). All bacterial protein acetyltransferases belong to the GNATs. These are characterized by sequence motifs A-D as indicated. Motifs A and B are important for CoA-binding. Motif A contains the characteristic sequence motif Arg/Gln-x-x-Gly-x-Gly/Ala (x: any amino acid), known as P-loop, which contacts the phosphates of the acetyl-CoA/CoA. The acetyl-CoA is shown in ball-and-stick representation [the figure was generated with PyMOL v.2.3.4 (Schrödinger LLC, 2000)]. (B) Domain organization of different bacterial GNAT types. Type I GNATs contain an N-terminal and type II GNATs a C-terminal NDP-forming acyl-CoA synthetase domain. These domains are catalytically inactive, but they bind acetyl-CoA and are important for allosteric regulation of GNAT activity. Type III GNATs encompass an N-terminal ligand binding domain such as a cAMP-binding domain with high similarity for cAMP-binding domains of EPAC, PKA, CAP/CRP, a Rossmann-fold domain specific for NADP⁺-binding, or an ACT domain for binding to amino acids cysteine, arginine and/or asparagine. Binding of these metabolic molecules to the N-terminal domain activates the C-terminal GNAT activity. Type IV GNATs contain only the catalytic GNAT domain and no accessory domain. Type V GNATs are composed of a tandem GNAT and a C-terminal region important for oligomerization and catalytic activity. Only the N-terminal GNAT domain is active, the central GNAT domain is important for structural integrity. (C) Several mechanisms contribute to catalytic activity of bacterial GNATs. Mammalian GNATs are shown to use a general base catalytic mechanism for acetyl-group transfer. In bacteria, not all GNATs contain a catalytic glutamate acting as general base and other mechanisms contribute to catalysis. The electrostatics in the active site might favor substrate amino group deprotonation. Some GNATs use a glutamate as general base for deprotonation of the substrate amino group. Other GNATs were reported to use a remote base, such as an activated serine residue to orient and polarize a catalytic water molecule acting as general base during catalysis. Other GNATs were reported to use an serine residue as catalytic base after activation by an active site water molecule. This catalytic strategy involves the formation of a serine-bound acetyl-enzyme intermediate. (D) Catalytic mechanism exerted by GNATs using a general base catalyst. Many GNATs use a catalytic glutamate as general base that abstracts a proton from the substrate amino group increasing its nucleophilicity for attack of the electrophilic carbonyl carbon of ac(et)yl-CoA. A tetrahedral intermediate is formed, which is resolved to yield the ac(et)ylated substrate amino group and CoA. For some GNATs an active site tyrosine contributes as catalytic acid resolving the tetrahedral intermediate by protonating the sulfhydryl group of the leaving CoA [figure redrawn and modified from Ali et al. (2018) and Blasl et al. (2021)].

FIGURE 3

Structural characterization of E. coli GNAT protein acetyltransferases. (A) E. coli GNAT domains of protein acetyltransferases are structurally very similar. For the KATs PhnO, PatZ (aa704-885) and YiaC, Phyre2 models were created. These were superimposed with the structurally characterized KATs RimI (PDB: 2CNS and 2CNT) and YjaB (PDB: 2KCW). The catalytically important tyrosine residue suggested to act as general acid supporting resolving of the tetrahedral intermediate superimposes well. The KATs show a high degree of structural similarity showing root-mean-square-deviations (RMSD) between 0.064 and 2.319 Å toward YiaC [structural models were created with Phyre2 (Kelley et al., 2015); the figure was generated with PyMOL v.2.3.4) (Schrödinger LLC, 2000)]. (B,C) The APBS Electrostatics plugin in PyMOL was used to plot the electrostatic potential on the surfaces of the experimentally determined structures of the KATs RimI (PDB: 2CNS) and YjaB (PDB: 2KWC) (B) or the Phyre2-generated structural models of YiaC, PhnO, and PatZ (C) (Jurrus et al., 2018). All structures were oriented toward the binding site. The electrostatics on the substrate binding area differs considerably suggesting that these KATs use diverse substrates. Moreover, also the electrostatics within the active site differs suggesting that it might support catalysis to different extent. The figure was generated with PyMOL v.2.3.4 (Schrödinger LLC, 2000).

FIGURE 4

Bacterial sirtuins are NAD⁺-dependent lysine-deac(et)ylases. (A) Structure of the E. coli sirtuin deacetylase CobB in complex with an lysine acetylated human histone 4 substrate (PDB: 1S5P). The CobB structure was superimposed to a structure of human SIRT2 in complex with NAD⁺ (PDB: 4RMG) to show localization of NAD⁺. CobB contains a Rossmann-fold domain (salmon) composed of a six-stranded parallel β-sheet flanked by several α-helices containing the NAD⁺-binding site. An NAD⁺-binding region (green) contributing to NAD⁺-binding connecting Rossmann-fold and small helical domain (light blue). The Zn²⁺-binding domain (dark blue) contains a structural Zn²⁺-ion that is coordinated by two pairs of conserved Cys-residues (only two are visible in this structure). The substrate binding loop (red) connects the Zn²⁺-binding domain and the Rossmann-fold domain [the figure was generated with PyMOL v.2.3.4 (Schrödinger LLC, 2000)]. (B) Catalytic mechanism exerted by sirtuins. Sirtuins are NAD⁺-dependent lysine deac(et)ylases enzymes that use NAD⁺ as stoichiometric co-substrate for catalysis. Initially, the carbonyl-oxygen of the lysine’s ac(et)yl-group as a nucleophile performs an attack of the electrophilic C-1′ of the NAD⁺ ribose. This results in fast release of nicotinamide subsequent formation of a C-1′-O-alkylamidate intermediate. Several steps are needed to resolve this intermediate. These include a hydrolysis step, as shown, resulting in formation of the deac(et)ylated lysine and 2′-O-acetyl-ADP-ribose, which exists in a non-enzymatic equilibrium with 3′-O-acetyl-ADP ribose [figure redrawn and modified from Smith and Denu (2006), Ali et al. (2018), Teixeira et al. (2020), and Blasl et al. (2021)].

FIGURE 5

Ac(et)ylation is a modification of molecules that is tightly connected to the cellular metabolic state. (A) Left panel: GNATs depend on ac(et)yl-CoA as donor molecule for the transfer of the ac(et)yl group to the substrate amino group. Shown is only acetyl-CoA but dependent on the specificity of the GNATs for different acyl-CoAs, also other acyl groups can be transferred. Acetyl-CoA is generated in the metabolism of all major nutrient classes. Dependent on the metabolic state, the intracellular concentration of ac(et)yl-CoA fluctuates. The catalytic efficiency of the GNATs depend on the intracellular acetyl-CoA/CoA ratio rather than the acetyl-CoA concentration. This is reflected by the similar K_M-values of GNATs for ac(et)yl-CoA and CoA. Moreover, several GNATs are regulated by binding of metabolic molecules such as cAMP, NADP⁺, acetyl-CoA and amino acids to accessory, allosteric sites. Binding of the ligands to the allosteric site modulates GNAT activity constituting another layer for regulation of GNAT activity. Further regulatory systems include post-translational modifications of GNATs, not shown here. Right panel: sirtuin activity is also tightly connected to the cellular metabolic state. Sirtuins use NAD⁺ as stoichiometric co-substrate for catalysis and other endogenous regulators such as c-di-GMP for CobB also exist. As described for GNATs, also sirtuins are regulated by post-translational modifications which are not shown here. (B) Acetyl-phosphate is the major factor for non-enzymatic acetylation in bacteria. Acetyl-CoA is formed during metabolism of all major nutrient classes. Shown is the formation of acetyl-CoA through glycolysis starting from glucose to form pyruvate, which is oxidatively decarboxylated to form acetyl-CoA by the pyruvate dehydrogenase complex (PDH). Acetyl-CoA can subsequently further metabolized either by oxidative phosphorylation under aerobic conditions to drive formation of ATP and regeneration of NAD⁺ or it is converted to acetate by phosphotransacetylase (Pta) catalyzing the formation of acetyl-phosphate and acetate kinase (AK) yielding acetate. Both reactions are reversible, i.e., AK and Pta can also convert acetate to acetyl-CoA under consumption of ATP. Alternatively, the acetate can be converted by acetyl-CoA synthetase (Acs) to form acetyl-CoA. Due to the production of pyrophosphate (PP_i) in the first half reaction yielding acetyl-adenylate (acetyl-AMP), this reaction is almost irreversible. Under conditions of carbon overflow, the intracellular acetate concentration (>5 mM) increases so that also the reaction catalyzed by AK and Pta to produce acetyl-CoA from acetate becomes important. This results in accumulation of acetyl-phosphate under conditions of carbon overflow causing systemic non-enzymatic acetylation.

FIGURE 6

Bacterial Zn²⁺-dependent classical deacetylases. (A) Structural analyses of bacterial classical deacetylases. Left panel: several bacterial classical deacetylases were structurally characterized. The enzyme AphA from M. ramosa encodes for a acetyl-polyamine amidohydrolase (AphA) that was categorized into class II. A. aeolicus histone deacetylase-like protein (HDLP) was classified to class I and Bordetella/Alcaligenes strain FB188 histone deacetylase-like amidohydrolase (HdaH) to class IIb. Both enzymes are active as protein deacetylases. The class IIb enzyme P. aeruginosa APAH (PA3774) was shown to be inactive toward polyamines and suggested to act as protein deacetylase. For comparison the structures were superimposed to the structure of class I KDAC8 from human. Notably, the L1 loop in bacterial classical deacetylases is a structural feature needed for oligomerization of and catalytic activity and the polyamine specificity determination loop (PSL) conveys substrate specificity toward polyamines. Right panel: The active site is totally conserved in the enzymes compared. The only difference is the residue N186 in P. aeruginosa APAH PA3774 (and M. ramosa AphA, Bordetella/Alcaligenes strain FB188 HdaH), which is a aspartate in KDAC8 (D183), and A. aeolicus HDLP (D173). Functionally, both asparagine and aspartate contact, orient and polarize the general base histidine (P. aeruginosa APAH PA3774 H144) [the figure was generated with PyMOL v.2.3.4 (Schrödinger LLC, 2000)]. (B) Catalytic mechanism exerted by bacterial classical deacetylases. As the active site is totally conserved between mammalian and bacterial classical deacetylases, a similar catalytic mechanisms can be assumed. Classical deacetylases use a catalytic Zn²⁺-ion that is coordinated by an two Asp, one His, one Tyr and the ac(et)yl-moiety of the substrate amino group, i.e., derived from a polyamine of a protein lysine side chain. The Zn²⁺-ion contacts the carbonyl oxygen thereby polarizing the carbonyl group of the acetyl-group. This results in an increase in the electrophilicity at the carbonyl C-atom enabling nucleophilic attack of an active site water molecule. This water molecule is activated by the histidine residue (human KDAC8: H143; P. aeruginosa APAH PA3774: H144). This histidine acts as general base abstracting a proton from the water thereby increasing its nucleophilicity. A second histidine (human KDAC8: H142; P. aeruginosa APAH PA3774: H143) fulfills the role of electrostatic catalyst. Both histidine residues are integral parts of Asp-His charge-relay systems. A tetrahedral oxyanion intermediate is build that is stabilized by a tyrosine (human KDAC8: Y306; P. aeruginosa APAH PA3774: Y313). This is resolved by the catalytic histidine (human KDAC8: H143; P. aeruginosa APAH PA3774: H144) acting as proton donor (catalytic acid) to the substrate amino group finally resulting in release of acetate and the deacetylated amino group [figure is redrawn and modified from Ali et al. (2018) and Blasl et al. (2021)].

FIGURE 7

Non-enzymatic N-(ε)-lysine and/or N-(α)-N-terminal ac(et)ylation. (A) Several mechanisms contribute to the efficiency of non-enzymatic acetylation. The high energy ac(et)yl-CoA thioesters and the mixed anhydride acetyl-phosphate are very reactive molecules. Upon increase of the intracellular concentrations the level of systemic lysine and/or N-terminal ac(et)ylation is increased. Furthermore, an alkaline pH-value supports non-enzymatic ac(et)ylation as it affects the protonation state of the ε- and α-amino groups. Under more alkaline conditions the amino groups are in a more deprotonated state increasing their nucleophilicity prone for non-enzymatic ac(et)ylation. Moreover, the sequence context of the lysine side chain affects its reactivity. The presence of the substrate lysine side chain in poly-basic patches containing several basic residues such as arginine and lysine lowers the lysine side chains pK_a value resulting in a more reactive, deprotonated state under physiological pH. Finally, spatial vicinity and proper orientation of the substrate amino group to an acidic residue such as glutamate can favor its deprotonation increasing its nucleophilicity. (B) Reaction mechanism for non-enzymatic ac(et)ylation of substrate amino groups. In bacteria most of non-enzymatic acetylation is due to acetyl-phosphate. However, also ac(et)yl-CoA contributes to non-enzymatic ac(et)ylation in bacteria. Shown is the reaction of non-enzymatic lysine acetylation by acetyl-phosphate. A tetrahedral intermediate is formed following the nucleophilic attack of the deprotonated lysine side chain to the electrophilic carbonyl carbon of acetyl-phosphate. This intermediate decomposes to form ac(et)yl-lysine and CoA (figure redrawn and modified from Ali et al. (2018) and Blasl et al. (2021)].

FIGURE 8

CE-clan related bacterial virulence factors with dual deubiquitinase (DUB)- and acetyltransferase (AcT)- activity and technological advances to study lysine ac(et)ylation. (A) Structure of the catalytic domain of Chlamydia trachomatis ChlaDUB1 (PDB: 6GZT; PDB: 6GZS). ChlaDUB1 was shown to act as DUB with specificity for K63-linked ubiquitin chains and as acetyltransferase for lysine, threonine and serine side chains. To discriminate this activity from lysine acetyltransferase activity, the abbreviation AcT is used. Notably, both activities, namely a deubiquitination, i.e., a hydrolysis reaction, and the transfer of an acetyl-group, i.e., a condensation reaction, is catalyzed by the same active site. The enzyme is a CE-clan related protease using a catalytic triad (order: His-Asp-Cys) for catalysis. Mutational studies showed that mutation of the catalytic cysteine to alanine (ChlaDUB1: C345A) abolished both activities. In contrast to other CE-clan related virulence factors that only are active as DUB or AcT, ChlaDUB1 contains an α-helix (VR-3: variable region-3) that mediates binding toward ubiquitin or acetyl-CoA/CoA using different surfaces. The structure with ubiquitin was obtained by using a ubiquitin activity-based probe ubiquitin-propargylamide (Ub-PA), resulting in covalent linkage of the Ub-PA to the active site Cys. Coenzyme A (CoA) was also covalently bound by formation of a disulfide bond between the CoA cysteamine and the active site Cys [the figure was generated with PyMOL v.2.3.4 (Schrödinger LLC, 2000)]. (B) Improved mass-spectrometry workflows to obtain information of the lysine ac(ety)lation stoichiometry on a systemic scale. Initially, mass-spectrometry was used to identify lysine acetylation sites in proteins without obtaining information on the dynamics or stoichiometry. Workflows that were developed, such as SILAC, allow to systemically compare alterations in lysine ac(et)ylation in diverse cellular states. Finally, chemical labeling of all lysine side chains (and N-terminal amino groups) in proteins (before enzymatic digestion) using isotopically labeled molecules such as N-acetoxy-succinimide or acetic-anhydride allows to perform absolute quantification of ac(et)ylation on a global scale. This workflow can be combined with SILAC and internal standards to generate powerful workflows to assess the kinetics and stoichiometry of ac(et)ylation. (C) Genetically encoding N-(ε)-ac(et)yl-lysine in proteins using the genetic code expansion concept (GCEC). By applying an orthogonal and synthetically evolved ac(et)yl-lysyl-tRNA (AcKRS)/tRNA_CUA pair based on the pyrrolysyl-tRNA_CUA (PylRS)/PylT pair from Methanosarcina barkeri or M. mazei acetyl-L-lysine and other lysine acylations, such as, propionyl-, butyryl-, crotonyl-lysine can be site-specifically incorporated into any protein as response to an amber stop codon. The system is orthogonal in all used model organisms. (1) Cells are fed with acetyl-L-lysine, deacetylases can be inhibited by addition of selected deacetylase inhibitors, (2) cells express the evolved AcKRS/tRNA_CUA pair, (3) AcKRS charges the cognate amber suppressor tRNA_CUA with the ac(et)yl-lysine, (4) Ac(et)yl-lysine is co-translationally incorporated into proteins as response to an amber stop codon using the host cells translation machinery. Using this technology enables to obtain natively-folded and quantitatively lysine ac(et)ylated proteins for structural and functional studies [figure modified from Lammers et al. (2019) and Blasl et al. (2021)].