Bacterial Protein Acetylation: the Dawning of a New Age

Abstract

Protein acetylation has historically been considered a predominantly eukaryotic phenomenon. Recent evidence, however, supports the hypothesis that acetylation broadly impacts bacterial physiology. To explore more rapidly the impact of protein acetylation in bacteria, microbiologists can benefit from the strong foundation established by investigators of protein acetylation in eukaryotes. To help advance this learning process, we will summarize the current understanding of protein acetylation in eukaryotes, discuss the emerging link between acetylation and metabolism, and highlight the best-studied examples of protein acetylation in bacteria.

Introduction

Since the discovery of post-translational protein acetylation almost fifty years ago (Phillips, 1963, Allfrey et al., 1964), studies of this protein modification have focused primarily on histones and other transcription-associated proteins. Recent reports, however, show that protein acetylation extends far beyond transcription. For example, the mammalian acetylome appears to include more than 2700 acetylated proteins involved in almost every aspect of cellular physiology, including central metabolism, mRNA splicing, protein synthesis and degradation, cell morphology and cell cycle (Choudhary et al., 2009, Zhao et al., 2010, Kim et al., 2006). Thus, as Tony Kouzarides predicted ten years ago, protein modification by acetylation is as common as phosphorylation (Kouzarides, 2000).

During the process of identifying the mammalian acetylome, it became evident that large-scale protein acetylation might also occur in bacterial cells: the mitochondrion, a close relative of α-proteobacteria [reviewed by (Gray et al., 1999)], contains many acetylated proteins (Choudhary et al., 2009, Zhao et al., 2010, Kim et al., 2006). Indeed, recent proteomic studies of two members of the γ-proteobacteria phylum, Escherichia coli (Yu et al., 2008, Zhang et al., 2009) and Salmonella enterica (Wang et al., 2010), have respectively identified 144 and 191 acetylated proteins that contribute to diverse cellular processes. These sets of acetylated proteins only partially overlap, which suggests that these initial studies were not saturating and that many more acetylated E. coli and S. enterica proteins likely exist. Investigation of protein acetylation in members of other bacterial phyla is also warranted.

The Biochemistry of Acetylation

Protein N-acetylation, like the more than 300 reported covalent post-translational modifications (Witze et al., 2007), adds complexity to the fundamental twenty amino acid-encoded proteome by increasing the number of distinct protein isoforms (Polevoda & Sherman, 2002, Walsh, 2006). Protein N-acetylation is the transfer of an acetyl group from an acetyl donor, e.g. acetyl-coenzyme A (acetyl-CoA), to the amino terminus of a protein (N_α) or to the ε-amino (N_ε) group of a lysine residue (Fig. 1). For both N_α- and N_ε-acetylation, an acetyltransferase can establish the proper catalytic microenvironment. The enzyme acts as a scaffold, orienting the acetyl group of acetyl-CoA and the amino group of the acceptor protein into close proximity. The enzyme then promotes deprotonation of the amino group, creating a neutral amino group that serves as a nucleophile that attacks the carbonyl carbon of acetyl-CoA, acetylating the protein and releasing a free CoA molecule (Dyda et al., 2000).

Figure 1. Electron transfer during direct N_ε-acetylation from acetyl-CoA.

At neutral pH, the ε-amino group is positively charged. This residue is deprotonated by a base (not shown in figure). (I) The lysine ε-amino group can now act as a nucleophile to attack the electrophilic carbonyl carbon of acetyl-CoA, (II) forming a tetrahedral intermediate. The electron dense oxygen expels the thiolate group (-SCoA). (III) Deprotonation of the amino group (IV) results in an acetylated lysine side chain and CoASH. Figure adapted from (Walsh, 2006).

N_α-acetylation

N_α-acetylation is considered to be an irreversible modification of either the deformylated N-terminal methionine or, following cleavage of the N-terminal methionine, the newly exposed amino acid. In eukaryotes, N_α-acetylation of proteins is extremely common (present on more than 80% of mammalian proteins) and is primarily cotranslational. In bacteria, N_α-acetylation is proposed to be less common, occurring posttranslationally on ribosomal proteins. Six putative N_α-acetyltransferases families have been identified in all domains of life, suggesting a common origin for N_α-acetylation [reviewed by (Polevoda & Sherman, 2003)]. A recent study reported that N_α-acetylation might function as a signal for protein degradation (Hwang et al., 2010).

N_ε-acetylation

In contrast to N_α-acetylation, N_ε-acetylation is both dynamic and reversible, a property that permits this posttranslational modification to serve in a regulatory capacity. The impact of N_ε-acetylation varies according to the context. N_ε-acetylation can change the size, charge, and/or conformation of the protein. These changes alter DNA binding affinity, protein stability, protein-protein interactions, protein localization, and protein function. N_ε-acetylation also influences other posttranslational modifications [reviewed by (Glozak et al., 2005, Yang & Seto, 2008a)]. Unless otherwise stated, the remainder of this review will focus on the regulation and impact of N_ε-acetylation.

Acetylation and deacetylation

Proteins with lysine acetyltransferase activity were historically referred to as HATs because of the extensive early studies on histone acetylation. Due to the observation that non-histone eukaryotic proteins become acetylated and the emergence of protein acetylation in bacteria, which possess no true histones, several authors have proposed to rename lysine acetyltransferases as either KATs (Allis et al., 2007) or LATs (Yang, 2004). Up to five groups of proteins with KAT activity have been identified: 1) the Gcn5-related acetyltransferase (GNAT) family, 2) the MYST family, 3) the CBP/p300 co-activators, 4) the SRC family of coactivators, and 5) the TAFII group of transcription factors [reviewed by (Marmorstein & Roth, 2001, Sterner & Berger, 2000)]. Autoacetylation, an intramolecular covalent modification, has also been reported for several eukaryotic KATs (Tip60 (Wang & Chen, 2010) and p300 (Thompson et al., 2004), for example) and for the E. coli proteins ACS and CheY (Barak et al., 2004).

Of these five groups of KATs, the GNATs are the most widely distributed, with over 10,000 members identified across all three domains of life [reviewed by (Vetting et al., 2005)]. Importantly for bacteriologists, GNATs are ubiquitous within the bacterial domain and are capable of carrying out a diverse range of functions. Some GNATs, such as protein acetyltransferase (Pat) from S. enterica, catalyze N_ε-acetylation (Starai & Escalante-Semerena, 2004), while other GNATs perform N_α-acetylation; for example RimI, RimL, and RimJ of E. coli (Tanaka et al., 1989, Yoshikawa et al., 1987). Not all GNATs catalyze protein acetylation. Indeed, the first characterized members of the GNAT family, the aminoglycoside N-acetyltransferases, were discovered as bacterial enzymes that could acetylate and thus inactivate antibiotics [reviewed in (Davies & Wright, 1997, Shaw et al., 1993)].

Two major families of lysine or histone deacetylases (KDACs or HDACs) have been identified. These two families have been grouped into four classes (Gregoretti et al., 2004): the zinc-dependent Rpd3/Hda1 family (classes I, II, and IV) (Yang & Seto, 2008b) and the NAD⁺-dependent sirtuin family (class III) (Blander & Guarente, 2004). For each KDAC class, putative bacterial homologues have been identified [reviewed by (Hildmann et al., 2007)], but only a few have been shown to serve as protein deacetylases and very few of their protein substrates have been identified (Gardner et al., 2006, Gardner & Escalante-Semerena, 2009, Hildmann et al., 2004, Starai et al., 2002, Wang et al., 2010, Li et al., In press).

Acetylation links metabolism and transcription

In eukaryotic cells, the flow of carbon, and thus cellular energy status, plays a key role in regulating histone acetylation status (Fig. 2). For example, a substantial fraction of the central metabolic enzymes ATP-citrate lyase (ACL, purple lines) and/or acetyl-CoA synthetase (ACS, green lines) can localize to the nucleus, where each can provide acetyl-CoA for histone acetylation: ACL performs this role in mammalian cells that have been exposed to excess glucose, while ACS performs this function in yeast cells exposed to acetate (Wellen et al., 2009, Takahashi et al., 2006). The nuclear localization presumably permits direct delivery of acetyl-CoA to KATs [reviewed by (Ladurner, 2009, Rathmell & Newgard, 2009)]. Similarly, carbon flux dictates the function of sirtuin-dependent histone deacetylation. Since glycolysis consumes NAD⁺ (blue lines), NAD⁺-dependent deacetylation by sirtuins is not favoured during rapid metabolism of excess glucose. Instead, sirtuin-dependent deacetylation must await consumption of that glucose and the subsequent recovery of NAD⁺. For this reason, sirtuins are proposed to communicate the metabolic status of the cell [reviewed by (Buck et al., 2004, Starai et al., 2004, Schwer & Verdin, 2008, Denu, 2005)].

Figure 2. Relevant central metabolic pathways.

Glycolysis (blue) metabolizes glucose to acetyl coenzyme A (acetyl-CoA) in an NAD⁺-dependent manner. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PDH, pyruvate dehydrogenase. Acetate dissimilation requires the Pta-AckA pathway (red). The enzyme PTA (phosphotransacetylase) converts acetyl-CoA and inorganic phosphate (P_i) into coenzyme A (CoA) and the high-energy pathway intermediate acetyl phosphate (acetyl-P). The enzyme ACKA (acetate kinase) converts acetyl-P and ADP to acetate and ATP. The acetate freely diffuses across the cell envelope into the environment. Acetate assimilation (green) requires the high-affinity enzyme ACS (acetyl-CoA synthetase). In a two-step process that involves an enzyme-bound intermediate (acetyl-AMP), Acs converts acetate, ATP, and CoA into AMP, pyrophosphate (PP_i), and acetyl-CoA. The acetyl-CoA replenishes the NAD⁺-dependent tricarboxylic acid (TCA) cycle (orange). Mammalian ATP-citrate lyase (ACL) uses ATP and CoA to convert the TCA cycle intermediate citrate into oxaloacetate (OAA), ADP, P_i, and acetyl-CoA (purple). Figure adapted from (Wolfe, 2008).

Acetylation controls central metabolic flux

Growing evidence argues that protein acetylation can regulate the flux of carbon through central metabolism. For example, histone acetylation appears to impact expression of central metabolic enzymes. Furthermore, reduced expression of ACL, which results in decreased histone acetylation, is reported to reduce transcription of the genes that encode the glucose transporter Glut4, the glycolytic enzymes hexokinase 2 and phosphofructokinase 1, and the fermentation enzyme lactate dehydrogenase A (Wellen et al., 2009).

The impact of acetylation on central metabolism may not be restricted to transcription. Lysine acetylation of central metabolic enzymes is quite common in both bacteria and eukaryotes (Wang et al., 2010, Zhao et al., 2010, Yu et al., 2008, Zhang et al., 2009, Kim et al., 2006). Acetylation occurs on proteins involved in many central metabolic processes including the synthesis of acetyl-CoA itself [via ACS and pyruvate dehydrogenase (PDH)], glycolysis, gluconeogenesis, the TCA cycle and the glyoxylate bypass, glycogen biosynthesis, amino acid biosynthesis, fatty acid metabolism, and urea detoxification (Kim et al., 2006, Wang et al., 2010, Yu & Auwerx, 2009, Zhang et al., 2009).

Since exposure to different carbon sources seems to alter the profile of acetylated central metabolic enzymes, acetylation might regulate metabolic flux, directing carbon down one set of pathways under one condition (e.g. growth on excess glucose) and down a different set under a completely different condition (e.g. growth on a TCA cycle intermediate) (Zhao et al., 2010, Wang et al., 2010). Given that some enzymes that synthesize acetyl-CoA regulate histone acetylation (i.e. ACS and ACL) and/or are themselves acetylated (i.e. ACS and PDH), and given that most central metabolic enzymes can become acetylated, it seems likely that fluctuations in the acetyl-CoA pool(s), and thus acetylation, regulate the expression of central metabolic enzymes and/or the stability of those enzymes and/or their activity, which, in turn, could adjust the acetyl-CoA pools themselves.

Drilling down: two bacterial examples

Now that researchers have begun to compile lists of N_ε-acetylated bacterial proteins, the next step is to answer the following questions: Under what condition(s) does a particular protein become N_ε-acetylated? How does it become N_ε-acetylated? How does N_ε-acetylation affect its expression, stability, and/or function? Currently, we possess precious few examples. In fact, virtually everything we know about N_ε-acetylation in bacteria comes from the study of two proteins: the central metabolic enzyme ACS and the signalling protein CheY. For ACS, reversible acetylation of a single lysine residue serves as a simple on-off switch that controls the activity of this stand-alone enzyme. For CheY, however, reversible acetylation of multiple lysines coupled with reversible phosphorylation of a single aspartate controls the ability of this signalling protein to form complexes with three distinct protein targets. The physiological consequences of these CheY acetylations remain unclear.

Acetyl-CoA synthetase

In a two-step process, the ubiquitous high-affinity enzyme ACS converts acetate, ATP and CoA into acetyl-CoA, pyrophosphate, and AMP (Fig. 2, green lines). The resultant acetyl-CoA can then be used to generate ATP, produce biosynthetic subunits (Wolfe, 2005, Starai & Escalante-Semerena, 2004), or serve as a source of acetyl groups for modification of proteins (Takahashi et al., 2006, Yang & Seto, 2008a).

Reversible acetylation regulates ACS activity (Fig. 3). In bacteria, a member of the GNAT family (PAT in S. enterica and AcuA in Bacillus subtilis) inhibits the synthesis of acetyl-AMP by acetylating a lysine residue located within the ACS active site (Gardner et al., 2006, Starai et al., 2002). How acetylation inhibits activity remains unknown; however, reactivation requires removal of the acetyl group (Starai et al., 2002). In S. enterica (Starai et al., 2002, Starai et al., 2003) and mammalian mitochondria (Schwer et al., 2006), this deacetylation reaction is catalyzed by a sirtuin (CobB and SIRT3, respectively). In B. subtilis, reactivation involves both a sirtuin (SrtN) (Gardner & Escalante-Semerena, 2009) and a class I KDAC (AcuC) (Gardner et al., 2006).

Figure 3. Reversible acetylation of ACS and CheY.

The activity of ACS is regulated by reversible acetylation. In a form of feedback inhibition, a GNAT (either Pat or AcuA depending on the organism) uses acetyl-CoA as its acetyl donor to acetylate ACS (^AcACS), which is inactive. When the glycolytic carbon source is depleted, NAD⁺ is recycled (not shown). An NAD⁺-dependent sirtuin (CobB or SrtN, depending on the organism) can now deacetylate ACS, generating the by-product 2'-O-acetyl-ADP-ribose (2'-OAADPr). In B.subtilis, the class III KDAC AcuC also can catalyze deacetylation in an NAD⁺-independent manner (not shown). Deacetylated, and thus reactivated, ACS can now synthesize acetyl-CoA. The activity of CheY is also regulated by reversible acetylation. In this case, acetylation occurs by autoacetylation (auto), ACS, or some unknown acetyltransferase (AT). CobB deacetylates acetyl-CheY (^AcCheY).

The chemotaxis response regulator CheY

As a response regulator of the family of two-component signal transduction systems, CheY autophosphorylates on a conserved aspartyl residue, using its phosphorylated cognate sensor kinase (CheA) as the phosphoryl donor. The resultant phospho-CheY binds with high affinity to the switch component FliM, increasing the probability that the flagellar motor rotates clockwise. Phospho-CheY is unstable, a condition accentuated by the auxiliary phosphatase CheZ [reviewed by (Vladimirov & Sourjik, 2009)].

Intriguingly, CheY has been reported to be N-acetylated in vivo (Yan et al., 2008, Li et al., In press). This acetylation can occur on six lysines located on the carboxy-terminal surface of CheY that interacts with CheA, CheZ, and FliM [reviewed by (Eisenbach, 2007)]. Two CheY acetylation mechanisms have been reported (Fig. 3): autoacetylation using acetyl-CoA as the acetyl donor (Barak et al., 2006) and Acs-catalyzed acetylation using acetate as the acetyl donor (Barak et al., 2004, Barak et al., 1992). A third mechanism, acetylation by a currently unknown acetyltransferase, has been proposed (Yan et al., 2008, Li et al., In press). Two deacetylation mechanisms have been reported: one that depends on Acs, which mediates reversible CheY acetylation (Barak et al., 2004, Barak et al., 1992) and a second that depends on the sirtuin CobB. It is likely, however, that the CobB-dependent mechanism predominates in vivo (Yan et al., 2008, Li et al., In press).

So, what role does CheY acetylation play? Five of the six acetylated lysines participate in the interaction with one or more of the CheY interaction partners. Thus, acetylation could simply inhibit those interactions, essentially protecting some proportion of the total CheY from being acted upon by components of the phosphorylation-dependent signal pathway (Liarzi et al., in press). Surprisingly, a cheA cheZ cobB mutant, which cannot reversibly phosphorylate CheY rapidly nor effectively deacetylate CheY, has been reported to respond robustly to chemotactic repellents. The suggestion is that a phosphorylation-independent repellent response exists and that it involves a protein that can become acetylated - perhaps CheY (Li et al., In press). Other evidence appears to support this conclusion, particularly the observation that replacement of one of the acetylatable lysines with a non-acetylatable arginine results in a defective repellent response (Barak & Eisenbach, 2001). These reports conflict with the well-established dogma that the bacterial chemotaxis absolutely depends on a phosphorylation cascade and that the cell uses the same machinery to respond to both attractants and repellents [reviewed by (Vladimirov & Sourjik, 2009)]. Yet, clearly CheY becomes acetylated and clearly that acetylation can affect function.

What’s next for bacterial protein acetylation?

Whole genome studies are powerful hypothesis generators. Like the early DNA array studies, the pioneering bacterial acetylome studies (Yu et al., 2008, Zhang et al., 2009, Wang et al., 2010) have provided the microbiology community with a wonderful opportunity to develop and test a large number of hypotheses. For example, does acetylation actually regulate central metabolic flux as proposed (Wang et al., 2010)? If so, how does acetylation alter the function of these central metabolic enzymes? Which acetyltransferases and deacetylases control each reversible acetylation? How are those acetylation-associated enzymes themselves regulated? How do these enzymes recognize their targets?

A few things are certain. First, the list of acetylated bacterial proteins will continue to expand. The initial studies were performed on two closely related bacterial species, each grown under a set of extremely limited conditions. Second, the current list of bacterial protein acetyltransferases and deacetylases will grow. To the best of our knowledge, no bacterial members of the non-GNAT families of acetyltransferases have been reported. Furthermore, the number of known and putative GNATs greatly surpasses the number of known or putative deacetylases. For example, the sirtuin CobB is the only known protein deacetylase in S. enterica or E. coli, yet compelling evidence argues that there must be more (Wang et al., 2010) (L. I. Hu, B. P. Lima and A. J. Wolfe, unpublished data). Third, the impact and regulation of bacterial protein acetylation will be highly nuanced. Many bacterial proteins are acetylated on multiple lysine residues (Barak et al., 2004, Barak et al., 1992, Yu et al., 2008, Zhang et al., 2009, Wang et al., 2010). Some proteins could be modified by both acetylation and a second modification. For example, CheY is both acetylated and phosphorylated (Eisenbach, 2007). Furthermore, individual subunits of large protein complexes could each be modified. It will be exciting to learn how multiple acetylation events interact and how multiple modifications influence each other. With the recognition that bacterial proteins can be multiply modified, the question arises: could there be a bacterial version of the proposed eukaryotic protein modification code, in which the interactions of multiple posttranslational modifications are read differently by the cell “to define” the function of a given protein (Takahashi et al., 2006, Yang & Seto, 2008a)?

Recent studies in our laboratory have begun to answer some of these questions. For example, we have identified growth conditions that dramatically alter the protein acetylation profile (A. J. Walker and A. J. Wolfe, unpublished data); thus, supporting the hypothesis that bacterial acetylation is indeed regulated (Wang et al., 2010). Furthermore, we possess evidence supporting the nuance and complexity of the system that controls acetylation. For example, we have learned that a putative GNAT and a lysine on the surface of the α subunit of RNA polymerase both inhibit glucose-induced transcription from an extracytoplasmic stress response promoter, while glucose-activated transcription requires a different GNAT, a different lysine on the surface of α, and a phosphorylated response regulator (B. P. Lima and A. J. Wolfe, unpublished data).

As new, highly sensitive technologies for detecting protein modifications develop, we are likely to find that acetylation is not the only bacterial protein acylation. For example, evidence exists that propionyl-CoA and butyryl-CoA both can serve as acyl donors for protein modification. Propionylated and/or butyrylated lysine residues have been identified in histones (Kim et al., 2006, Leemhuis et al., 2008, Zhang et al., 2009, Chen et al., 2007), CBP/p300 acetyltransferases (Chen et al., 2007, Cheng et al., 2009), and bacterial propionyl-CoA synthetase (PrpE) (Garrity et al., 2007), albeit at lower stoichiometries than acetyllysine residues (Cheng et al., 2009). Like acetylation, both propionylation and butyrylation are likely to be reversible. Members of the CBP/p300 (Cheng et al., 2009, Chen et al., 2007, Leemhuis et al., 2008, Liu et al., 2009), GNAT (Garrity et al., 2007), and MYST (Berndsen et al., 2007) acetyltransferase families can catalyze proprionylation and/or butyrylation, while both modifications can be reversed by sirtuins (Garrity et al., 2007, Smith & Denu, 2007, Cheng et al., 2009, Liu et al., 2009)

In addition to acetyl-CoA, propionyl-CoA, and butyryl-CoA, the list of metabolically relevant acyl-CoA derivatives is quite long. In theory, each could function as an acyl donor (Cheng et al., 2009, Liu et al., 2009). The result of these modifications could be CoA homeostasis via end-product inhibition, as exemplified by acetylation of the acetyl-CoA-producing enzyme ACS and propionylation of the propionyl-CoA-producing enzyme PrpE (Garrity et al., 2007).

Concluding Remarks

Bacteria have long been considered simple relatives of eukaryotes. Obviously, this misperception must be modified. From the presence of a cytoskeleton to the packaging of DNA to the existence of multiple posttranslational modifications, bacteria clearly implement highly sophisticated mechanisms to regulate diverse cellular processes precisely. From even a quick scan of the recently generated lists of acetylated proteins, it is clear that many of these processes are likely regulated by acetylation in bacteria and eukaryotes and, almost certainly, archaea as well. If this is the case, then our deep knowledge of bacterial physiology and the power of bacterial genetics and biochemistry should rapidly advance our understanding of protein acetylation and the diverse cellular processes that it regulates.