Determinants within the C-Terminal Domain of Streptomyces lividans Acetyl-CoA Synthetase that Block Acetylation of Its Active Site Lysine In Vitro by the Protein Acetyltransferase (Pat) Enzyme

Reversible lysine acetylation (RLA) is a widespread regulatory mechanism that modulates the function of proteins involved in diverse cellular processes. A strong case has been made for RLA control exerted by homologues of the Salmonella enterica protein acetyltransferase (SePat) enzyme on the broadly distributed AMP-forming CoA ligase (a.k.a. acyl-CoA synthetases) family of metabolic enzymes, with acetyl-CoA synthetase (Acs) being the paradigm in the field. Here we investigate why the Acs homologue in Streptomyces lividans (SlAcs) is poorly acetylated in vitro by the S. lividans protein acetyltransferase (SlPat) enzyme. Chimeras of S. enterica Acs (SeAcs) and S. lividans Acs (SlAcs) constructed during the course of this work were acetylated by SlPatA in vitro, retained most of their activity, and were under RLA control in a heterologous host. We identified SeAcs residues N- and C-terminal to the target lysine that when introduced into SlAcs, rendered the latter under RLA control. These results lend further support to the idea that Pat enzymes interact with extensive surfaces of their substrates. Finally, we suggest that acetylation of SlAcs depends on factors or conditions other than those present in our in vitro system. We also discuss possible explanations why SlAcs is not controlled by RLA as defined in other bacterial species.


Introduction
Reversible lysine acetylation (RLA) is a post-translational modification that occurs in all domains of life [1] and affects diverse cellular processes and functions. Acetyltransferases transfer the acetyl moiety from acetyl-CoA to the e-amino group of the target lysine. Lysine acetylation can affect enzyme activity [2], protein stability [3], protein-protein interactions, or DNA binding [4]. Yeast Gcn5 protein (yGcn5p)-related N-acetyltransferases (a.k.a., GNATs), classified by amino acid sequence and structure [5], are the only class of acetyltransferases found in all domains of life [6]. GNATs were first identified for their role in modification of histones [7]. Crystal structures and biochemical analyses of the yGcn5p, the founding member of the GNAT family, with representative peptides from histones has provided valuable information about the substrate specificity and substrate recognition by GNATs [8,9].
Members of the GNAT family also acetylate metabolic enzymes. For example, in Salmonella enterica, the enzyme acetyl-CoA synthetase (SeAcs) is acetylated by the protein acetyltransferase (SePat), a two-domain acetyltransferase that contains a large domain of unknown function and a C-terminal GNAT domain [10]. SeAcs is a member of the AMP-forming CoA ligase family of enzymes that converts carboxylic acids to their CoA thioesters via an acyl-AMP intermediate [11]. Acetylation of the active site lysine of AMP-forming CoA ligases prevents the adenylylation of the carboxylic acid. In addition to Pat from S. enterica, GNATs are known to acetylate members of the of AMP-forming CoA ligase family (including Acs) in Rhodopseudomonas palustris [12,13], Bacillus subtilis [14], and Mycobacterium smegmatis [15]. The Acs homologue from Streptomyces coelicolor is acetylated in vivo [16], but the GNAT responsible for acetylation of S. coelicolor Acs is unknown.
Knowledge of the interactions of GNAT with their proteins substrates is limited. R. palustris encodes a single-domain GNAT (RpKatA) and a homologue of the SePat GNAT (RpPat). RpKatA and RpPat discriminate among members of the AMP-forming CoA ligase family produced by R. palustris [13]. In addition to the target lysine, RpPat recognizes a loop greater than 20 Å from the target lysine, suggesting that Pat enzymes interact with a large surface of the acceptor substrate [17]. As a proof of principle, the introduction of this recognition loop into R. palustris methylmalonyl-CoA mutase (RpMatB), an AMP-forming CoA ligase that is not a substrate of RpPat, rendered RpMatB a target of acetylation by RpPat. Thus, synthetic chimeras of AMP-forming CoA ligases have yielded valuable information about how GNATs recognize protein substrates and have produced AMP-forming CoA ligases that are placed under the regulation of lysine acetylation.
RpPat and SePat enzymes acetylate their cognate Acs proteins. Although the GNAT responsible for the acetylation of Acs in S. coeolicolor is unknown, the closely related actinomycete Streptomyces lividans encodes SlPatA, a two-domain homologue of SePat and RpPat enzymes. Significantly, SlPatA does not efficiently acetylate the S. lividans Acs (SlAcs) in vitro [18], making this the first Acs enzyme that is not efficiently acetylated by a Pat acetyltransferase. In contrast, SlPatA efficiently acetylates SeAcs. Here we probe the amino acid sequences in SeAcs that rendered it a better substrate for SlPatA than SlAcs is. By replacing amino acids from SeAcs into the C-terminus of SlAcs, we constructed SlAcs-SeAcs chimeras that were efficiently acetylated by SlPatA. One SlAcs-SeAcs chimera contained 41 amino acid differences from SlAcs. As a result of these changes, the SlAcs-SeAcs chimera was subject to regulation by SlPatA. We used a heterologous model system to demonstrate that the SlAcs-SeAcs chimera was subject to RLA regulation in vivo by SlPatA. In sum, we identified regions in SeAcs that were critical for recognition by SlPatA, and transferring of these residues into the poor substrate SlAcs resulted in a SlAcs variant that was efficiently regulated by SlPatA.

Bacterial Strains and Growth Conditions
All strains and plasmids used in this study are listed in Tables 1  and 2, respectively. Escherichia coli and Salmonella enterica strains were grown at 37uC in lysogeny broth (LB, Difco) [19] or no-carbon essential (NCE) minimal medium [20] supplemented with sodium acetate (10 mM), MgSO 4 (1 mM), and ampicillin (100 mg ml 21 ). When necessary, antibiotics were used at the following concentrations: ampicillin, 100 mg ml 21 ; tetracycline, 10 mg ml 21 ; chloramphenicol, 12.5 mg ml 21 , kanamycin, 50 mg ml 21 . L-(+)arabinose was added at varying concentrations (5 or 200 mM) to induce the expression of S. enterica acs, S. lividans acs, and acs chimeras cloned into the expression vector pBAD30 [21]. Isopropyl b-D-1-thiogalactopyranoside (IPTG) was added to a final concentration of IPTG (0-500 mM) to induce expression of S. lividans patA (EFD66247) clones into the expression vector pSRK-Km [22]. Growth experiments were performed at 37uC using a microtiter plate and a microtiter plate reader (Bio-Tek Instruments). All growth data are plotted as the mean of three data points.

Molecular Techniques
DNA manipulations were performed using standard techniques [23]. Restriction endonucleases were purchased from Fermentas. DNA was amplified using Pfu Ultra II Fusion DNA polymerase (Agilent) or Herculase II Fusion DNA polymerase (Agilent). Sitedirected mutagenesis was performed using the Quikchange TM Site Directed Mutagenesis kit (Agilent). Plasmids were isolated using the Wizard Plus SV Miniprep kit (Promega) and PCR products were purified using the Wizard SV Gel and PCR Clean-Up System (Promega). DNA sequencing was performed using BigDyeH (ABI PRISM) protocols, and sequencing reactions were resolved at the University of Georgia Genomics Facility.

Plasmids Used for Protein Overproduction
Chimeric proteins encoded by fusing different regions of S. lividans acs (EFD68454) and S. enterica acs genes were generated by amplifying genomic DNA from S. lividans TK24 genomic DNA from S. enterica strain TR6583, respectively. Fusion plasmids encoding proteins in which the N-terminal domain of SlAcs was fused to the C-terminal domain of SeAcs at residues 520, 550, 560, 566, 582, 617 were generated by overlap-extension PCR [24], followed by standard cloning into plasmid pTEV5 [25]. Fusion plasmids encoding a protein in which an internal sequence of SlAcs was replaced by the corresponding sequence SeAcs were constructed as described below and in Table 2.
Plasmid pSlAcs44 (SlAcs 615-626 SeAcs) -the nucleotides encoding the first 614 residues of SlAcs were amplified from pSlAcs1, the nucleotides encoding the final 40 residues of SeAcs fused to SlAcs amplified from pSlAcs28, and cloned into pTEV5.
The C-terminal domain of SeAcs was amplified from strain TR6583. DNA fragments were cut with NheI and EcoRI and ligated into pTEV5 [25] cut with the same enzymes. The resulting plasmids directed the synthesis of SlAcs-SeAcs chimeras or SeAcs Cterminal domain (pACS38) each with an N-terminal H 6 tag cleavable by recombinant tobacco etch virus (rTEV) protease prepared as described [26].
The C-terminal domain of SlAcs was amplified from S. lividans TK24 genomic DNA. The DNA fragments were cut with KpnI and HinDIII and ligated into pKLD66 [25] cut with the same enzymes. The resulting plasmid pSlAcs7 directed synthesis of the SlAcs C-terminal domain with an N-terminal maltose-binding protein-His 6 tag cleavable by rTEV protease as described above.

Construction of Untagged SlAcs Complementation Plasmid
The S. lividans acs was amplified from pSlAcs1 with the primers that included an optimized ribosome-binding site. The DNA fragment was cut with EcoRI and HindIII and ligated into pBAD30 [21], cut with the same enzymes. The resulting plasmid pSlAcs6 expresses SlAcs under the control of the P araBAD promoter.

Construction of SeAcs, SlAcs, and SlAcs
Complementation vectors encoding H 6 -tagged SeAcs chimera C3. Genes encoding S. lividans Acs and the S. lividans/S. enterica Acs chimeras were amplified from pSlAcs1 and pSlAcs28, respectively, using primers that included an optimized ribosome-binding site and an N-terminal His 6 -tag. S. enterica acs was amplified from genomic DNA isolated from JE6583 using primers that included an optimized ribosome-binding site and an N-terminal His 6 -tag. The DNA fragments were cut with EcoRI and HindIII and ligated into pBAD30, cut with the same enzymes. The resulting plasmids pSlAcs47, pSlAcs48, and pACS59 produce SlAcs, S. lividans/S. enterica Acs chimera C3, and SeAcs WT , respectively, with His 6 -tags fused N-terminal with a Gly-Ser-Gly linker under the control of at the P araBAD promoter.
Purification of SlAcs-SeAcs chimeras, SlAcs C-terminal domain, and SeAcs C-terminal domain. Plasmids encoding tagged proteins were transformed with pRARE2 (EMD Millipore) into a Dpka derivative of E. coli strain C41l(DE3) [27] (JE9314) to prevent acetylation prior to overproduction. The resulting strains were grown overnight and sub-cultured 1:100 (v/v) into two liters of LB containing ampicillin (100 mg ml 21 ) and chloramphenicol (12.5 mg ml 21 ). The cultures were grown shaking at 25uC to A 600 ,0.7 and protein synthesis was induced with IPTG (0.25 mM). Upon induction, the cultures were grown overnight at 25uC. Cells were harvested at 60006g for 10 min at 4uC in a Avanti J-2 XPI centrifuge fitted with rotor JLA-8.1000 (Beckman Coulter). Cell pellets were re-suspended in 30 ml of cold His-bind  All fractions containing H 6 -SlAcs-SeAcs chimera were combined. rTEV protease was added to H 6 -SlAcs-SeAcs chimera and the SlAcs-SeAcs chimera/rTEV mixture was incubated at room temperature for 3 h. PMSF was added to the protein mixture and incubated 15 min at room temperature. The SlAcs-SeAcs chimera/rTEV mixture was dialyzed at 4uC against buffer D (Tris-HCl (50 mM, pH 8), NaCl (500 mM)) twice for 3 h and again against buffer D containing imidazole (5 mM) for 12 h. After cleavage and dialysis, protein mixtures were passed over 1 ml HisPur TM Ni-NTA Resin (Pierce) using the buffers described above. Cleaved SlAcs-SeAcs chimera passed through the resin and eluted in the flow-through fractions. Purified SlAcs-SeAcs chimera was analyzed by SDS-PAGE. Fractions containing SlAcs-SeAcs chimera were pooled together. SlAcs-SeAcs chimera was stored in Tris-Cl buffer (50 mM, pH 8.0) containing NaCl (100 mM) and glycerol (20%, v/v). SlAcs concentration was determined by measuring absorbance at 280 nm. The molecular weights and molar extinction coefficients used to calculate H 6 -SlAcs-SeAcs chimera concentrations are listed in Table 3. All enzymes were purified to .95% homogeneity.

SeAcs Protein Purification
Plasmid pACS10 was transformed into a Dpka derivative of E. coli strain C41l(DE3) (JE9314). The resulting strain was grown overnight and sub-cultured 1:100 (v/v) into two liters of LB containing ampicillin (100 mg ml 21 ). The culture was grown shaking at 37uC to A 600 ,0.7 and protein synthesis was induced with IPTG (0.25 mM). Upon induction, the cultures were grown overnight at 30uC. SeAcs was purified and stored as described [2]. SlAcs WT and SlPatA WT were purified as described [18].

S. lividans Acetyl-CoA Synthetase (SlAcs) is Functional in vivo in a Heterologous System
The SeAcs homologue from S. lividans converts acetate to acetyl-CoA in vitro [18]. Alignment of the SeAcs and SlAcs amino acid sequences using BLAST revealed 52% sequence identity and 62% sequence similarity in amino acid sequence. To determine whether or not SlAcs functioned in vivo, we expressed S. lividans acs + ectopically in a Dacs Dpta S. enterica strain (JE13238) demanding growth on low concentrations of acetate (10 mM). S. enterica uses two pathways for the conversion of acetate to acetyl-CoA (Fig. 1A) [11,32]. One pathway is comprised of SeAcs, which catalyzes a two-step conversion of acetate to acetyl-CoA via an acetyl-AMP intermediate. RLA controls SeAcs activity [2]. The protein acetyltransferase SePat acetylates and inactivates of SeAcs (discussed further below) [10], and SeAcs is deacetylated and reactivated by the sirtuin type deacetylase SeCobB [2,29]. In the second pathway, acetate kinase (Ack) and phosphotransacetylase (Pta) catalyze the conversion of acetate to acetyl-CoA via an acetyl-phosphate intermediate. Acs activity is used by the cell when the concentration of acetate in the environment is ,10 mM, whilst Pta/Ack is the preferred pathway when S. enterica is growing on concentrations of acetate $25 mM. A S. enterica strain lacking the Acs and Ack/Pta pathways failed to grow on acetate (10 mM, Fig. 1B, squares). When SlAcs was produced ectopically, growth of an S. enterica Dacs Dpta strain was restored (Fig. 1B, circles), demonstrating that SlAcs was active and could substitute for SeAcs function in vivo.
SlPatA Acetylates the C-terminal Domain of SeAcs, but not SlAcs AMP-forming CoA synthetases are two-domain enzymes that activate carboxylic acids to CoA thioesters in a two-step reaction. In the first half-reaction, an invariant lysine in the C-terminal domain (K609 of SeAcs) is buried in the active site cleft located between the Nand C-terminal domains [33]. Upon adenylylation of the carboxylic acid substrate, the C-terminal domain undergoes a ,140u domain rotation to allow for the thioesterification of the fatty acyl-AMP intermediate [34]. The catalytic lysine of AMPforming CoA ligases is surface exposed when the enzyme is in the thioester-forming conformation [33], and this likely represents the conformation that is subject to acetylation by Pat.
Previously, we demonstrated that SlAcs was a poor substrate for the SlPatA enzyme in vitro [18]. That work identified SlAcs as the first example of an acetyl-CoA synthetase that was not recognized by the cognate Pat protein acetyltransferase in vitro [10,12]. However, SlPatA efficiently acetylated and inactivated the acetoacetyl-CoA synthetase SlAacS from S. lividans, and the orthologous SeAcs enzyme [18], indicating that SlPatA was catalytically active, but somehow unable to acetylate SlAcs in vitro.
We considered the possibility that SlAcs favored the adenylyation conformation in vitro, which would likely render the target K610 inaccessible to SlPatA due to its location in the SlAcs active site. To differentiate the inaccessibility of SlAcs K610 from the inability of SlPatA to recognize SlAcs, we isolated the C-terminal domains of SeAcs (a good substrate of SlPatA) and SlAcs. In the absence of the N-terminal domain, the target lysine is no longer protected, thus it is accessible to the acetyltransferase.
Homogeneous C-terminal domains of SlAcs (residues D519-D649, 131 aa) and SeAcs (residues D518-S652, 135 aa) were incubated in the presence of SlPatA and radiolabeled [1-14 C] acetyl-CoA. Differential migration of the C-terminal domains is likely due to differences in hydropathy of (grand average of hydropathy [GRAVY] scores [35] for SlAcs and SeAcs C-terminal domains are +0.023 and 20.160, respectively), which has been shown to affect gel mobility of protein in SDS-PAGE [36]. As shown in figure 2A, the C-terminal domain of SeAcs was acetylated, but the SlAcs C-terminal domain was not. These data showed that the N-terminal domain of SeAcs was not required for acetylation by SlPatA. Additionally, these results strongly suggested that inaccessibility of residue K610 was likely not the reason why SlAcs was poorly acetylated in vitro. We hypothesized that regions within the C-terminal domain of SlAcs enzyme prevented acetylation of SlPatA. As shown in figure 2B, the C-terminal domains of SlAcs and SeAcs share ,50% sequence identity. SlAcs WT can substitute for SeAcs WT in S. enterica during growth on acetate. A. S. enterica encodes a one-enzyme and a twoenzyme pathway for acetate activation. The one-enzyme pathway is composed of acetyl-CoA synthetase (Acs), whose activity is modulated posttranslationally by the protein acetyltransferase (Pat) and sirtuin deacetylase (CobB) enzymes. The two-enzyme pathway is comprised of acetate kinase (Ack) and phosphotransacetylase (Pta). B. Growth behavior of Dacs Dpta S. enterica strain JE13238 as a function of SlAcs WT . Experiments were performed on NCE minimal medium supplemented with acetate (10 mM), at 37uC using a microtiter plate and a plate reader (Bio-Tek Instruments). Synthesis of SlAcs WT was ectopically encoded (plasmid pSlAcs6) and induced using L-(+)-arabinose (5 mM). Cloning vector (pBAD30) lacking S. lividans acs + was used as negative control. All S.D. ,0.01 absorbance units. doi:10.1371/journal.pone.0099817.g001 Chimeras of SlAcs and SeAcs Reveal Regions in the SeAcs C-terminal Domain that are Critical for Acetylation by SlPatA Based on regions of sequence conservation (Fig. 2B), we generated a set of precise fusions between the SlAcs and SeAcs that contained varying amounts of the SeAcs protein. A SlAcs chimera containing the SlAcs N-terminal domain fused to the SeAcs C-terminal (chimera A1) was strongly acetylated by SlPatA, confirming that the C-terminal domain of SlAcs was responsible for the poor acetylation of SlAcs WT (Figs. 3A, B).
We identified regions of the SeAcs C-terminal domain important for acetylation by SlPatA by constructing chimeras that contained decreasing amounts of the SeAcs C-terminal domain relative to chimera A1. To measure the efficiency of acetylation, each chimera was incubated with SlPatA and radiolabeled [1-14 C] acetyl-CoA. SlPatA strongly acetylated chimeras containing at least the final 86 amino acids of SeAcs (chimeras A1, A2, A3, A4; Fig. 3). These chimeras contained at least 43 amino acids Nterminal to the acetylation site, a region previously reported to be important for acetylation of homologous AMP-forming CoA ligase enzymes by the R. palustris Pat homologue [17].
To narrow down the number of SeAcs residues required for acetylation of the SlAcs-SeAcs chimeras, we focused on SlAcs-SeAcs chimera A2, which had the fewest SeAcs-derived residues (Fig. 3B), and the highest level of acetylation (Fig. 3C).
We generated a second set of chimeras in which various stretches of SlAcs-derived residues were substituted into SlAcs-SeAcs chimera A2 (Fig. 3D). SlAcs-SeAcs chimeras B4, B5, and B6 that contained at least 45 residues of SeAcs (including the SeAcs K609 acetylation site) were strongly acetylated (Fig. 3C). Notably, the A10 loop of Acs, which contains the target lysine, is completely conserved between SeAcs and SlAcs (Fig. 2B). However, 17 amino acids C-terminal to the acetylation site of SeAcs were required for acetylation by SlPatA. This revealed a previously unrecognized region of the protein important for acetylation. Of this set of SlAcs-SeAcs chimeras, chimera B4 was the best substrate of SlPatA and contained the fewest SeAcs-derived amino acids.
To determine whether the 77 contiguous SeAcs-derived residues of chimera B4 were critical for acetylation, we identified regions of SlAcs and SeAcs with low amino acid sequence conservation and introduced those sets of SlAcs residues into chimera B4 (Fig. 3E).
Importantly, chimeras containing only the 60 SeAcs-derived residues N-terminal to K610 (chimera B2) or 11 SeAcs-derived residues C-terminal to K610 (chimera C5) were ,30% acetylated relative to SeAcs. Thus, amino acid sequences Nand C-terminal to the target lysine were important for acetylation by SlPatA, and neither set of amino acids rendered SlAcs a strong acetylation target when introduced independently.

Assessment of the Enzymatic Activity of the Chimeras
Chimeras were tested for their AMP-forming acetyl-CoA ligase forming activity. Although the SlAcs WT and SeAcs WT C-terminal domains share a high degree of sequence conservation, not all chimeras were active (Fig. 4A, black bars). To identify active chimeras that were also targets of acetylation, the acetylation of each chimera was measured relative to SeAcs (Fig. 4A, gray bars). SlAcs-SeAcs chimera C3 (hereafter referred to as chimera C3) was identified as the single chimera with the fewest SeAcs residues that was active and efficiently acetylated by SlPatA. As shown in figure 4B, chimera C3 contained 41 amino acid differences from SlAcs. For comparison, we include the analogous sequence from Acs homologues known to acetylated by protein acetyltransferases in other bacteria. Notably, the wildtype SlAcs amino acid sequence replaced by SeAcs sequences shares some sequence homology with these Acs homologues.

Chimera C3 Activity is Modulated by Acetylation and Deacetylation
As shown in figure 5A, the catalytic residue K610 residue is the only residue of chimera C3 that was acetylated. To test whether the activity of chimera C3 was under the control of acetylation, the protein was incubated with SlPatA acetyltransferase in the presence and absence of the acetyl donor, acetyl-CoA. Upon acetylation, chimera C3 activity decreased .98%, similar to the regulation of SeAcs activity (Fig. 5B, gray bar). The SlAcs enzyme retains .75% activity upon incubation with SlPatA and Ac-CoA (Fig. 5B, gray bar). As mentioned above, acetylation of SeAcs WT is reversed by the NAD + -dependent sirtuin deacetylase CobB in S. enterica, and deacetylation reactivates the SeAcs WT enzyme [2]. We tested whether chimera C3 could be deacetylated by incubating acetylated chimera C3 with SeCobB, the co-substrate NAD + , or both. When SeCobB and NAD + were present in the reaction mixture, chimera C3 Ac was completely deacetylated (Fig. 5C), demonstrating that the reversibility of the acetylation process was not affected by the substitutions in chimera C3.

SlAcs-SeAcs Chimera C3 is Acetylated in vivo in S. enterica by SlpatA
To determine the efficiency of SlPatA acetylation of chimera C3 in vivo, we used S. enterica acetate utilization (Fig. 1A) as a heterologous model to demonstrate the effects of SlPatA acetylation on activity of the Acs homologues. In this system, His-tagged chimera C3, SlAcs, and SeAcs (H 6 -chimera C3, H 6 -SlAcs, H 6 -SeAcs, respectively) were produced from plasmids in S. enterica acs pat cobB and S. enterica acs pat cobB + strains JE9152 and JE9894, respectively. All the experiments were conducted in S. enterica pat strains to prevent acetylation by SePat. We characterized the effect of SlPatA acetylation on the H 6 -Acs homologues by measuring growth of each strain harboring a plasmid with an inducible SlPatA allele or an empty cloning vector. Additionally, we isolated H 6 -SeAcs WT , H 6 -SlAcs WT , and H 6 -chimera C3 from cells grown in the presence or absence of SlPatA to quantify the effects of SlPatA acetylation on each Acs protein. As shown in figure 6A, production of H 6 -chimera C3, H 6 -SlAcs, or H 6 -SeAcs supported growth of S. enterica acs pat cobB strain (open symbols). This was the expected result, since the strain lacked Pat activity, thus the cell could not acetylate (i.e., inactivate) any of the Acs enzymes. We attributed the lag in the strain producing H 6 -chimera C3 to the decreased activity of this chimera (Fig. 5B). When production of SlPatA was induced in each strain (25 mM inducer), growth of S. enterica acs pat cobB strains producing H 6 -SeAcs WT or H 6 -chimera C3 was significantly reduced, while growth of the S. enterica acs pat cobB strain producing H 6 -SlAcs WT was unaffected. Importantly, inhibition of an S. enterica acs cobB strain producing H 6 -SlAcs WT required high levels of SlPatA WT induction (500 mM inducer, Fig. 6B). No growth inhibition occurred when SlPatA WT was induced at low levels (#5 mM inducer, Fig. 6C).
As expected, the presence of SeCobB WT in a S. enterica acs pat cobB + strain resulted in no significant growth defects upon SlPatA WT induction in strains expressing H 6 -SlAcs WT or H 6 -SeAcs WT (Fig. 6D). However, we did note a slight inhibition of growth of a S. enterica acs pat cobB + strain producing H 6 -chimera C3. We surmised that such an effect was likely due to a decreased ability of SeCobB WT to deacetylate and reactivate H 6 -chimera C3 and restore growth. This idea was supported by the observation that increased induction of SlPatA WT inhibited a S. enterica acs pat cobB + strain producing H 6 -chimera C3 (Fig. 6E), but not those producing H 6 -SlAcs WT nor H 6 -SeAcs WT (Fig. 6F).
Since high levels of SlPatA induction were required to inhibit growth of an S. enterica acs cobB strain producing H 6 -SlAcs WT , we expected that H 6 -SlAcs WT to be poorly acetylated by SlPatA WT and thus more active in vivo. We also expected higher proportions of acetylated to non-acetylated H 6 -SeAcs WT and H 6 -chimera C3 in vivo. To measure the effect of SlPatA WT acetylation on the activity of H 6 -SlAcs of, H 6 -SeAcs of, and H 6 -chimera C3, we grew S. enterica acs cobB strains expressing the corresponding acs alleles while inducing SlPatA WT at low levels (5 mM) to allow for growth and biomass accumulation for all strains (Fig. 6C). H 6 -SlAcs WT , H 6 -SeAcs WT and H 6 -chimera C3 enzymes were isolated from strains harboring plasmid-borne SlPatA WT or empty vector.
As shown in figure 7, activity of the H 6 -SlAcs WT enzyme isolated from a strain producing SlPatA WT was not significantly reduced compared to H 6 -SlAcs WT isolated from a strain with no SlPatA WT . However, activities of the H 6 -SeAcs WT and H 6 -chimera C3 enzymes were significantly lower when isolated from strains expressing SlPatA WT compared to those with no SlPatA WT . Activities of the SeAcs and H 6 -chimera C3 were restored upon incubation with SeCobB deacetylase. These data suggested that SlPatA WT more efficiently acetylated H 6 -SeAcs WT and H 6 -chimera C3 than it did H 6 -SlAcs in a heterologous in vivo model.

Discussion
Herein we report the first Acs enzyme that is not a substrate of Pat homologues in vitro. This finding is important, since Acs is the paradigm for the analysis RLA in all metabolic systems reported thus far. Our results begin to shed some light onto why the SlAcs is not efficiently acetylated by the SlPatA WT enzyme of S. lividans. By constructing chimeras of SlAcs that are acetylated by SlPat WT and retain biological activity we gained insights into structural, physiological and possibly evolutionary questions raised by this work.

Is Acs Activity under RLA Control in Streptomycetes?
At present, the answer to this question is unclear. It is not known whether SlAcs WT is a bona fide substrate of SlPatA WT in vivo in S. lividans. The literature adds to the challenge of determining whether or not in streptomycetes Acs is under RLA control. Work performed by others in Streptomyces coelicolor suggested that the Acs enzyme of this actinomycete may be under RLA control, because results of in vitro experiments showed that acetylated ScAcs was a substrate of a sirtuin deacetylase present in that bacterium [16]. The same authors also reported the isolation of acetylated ScAcs from S. coelicolor. Since the S. coelicolor genome contains a gene SlAcs-SeAcs Chimera C3 is active and efficiently acetylated. A. Acetyl-CoA synthetase activity of each chimera and SlAcs WT relative to SeAcs WT (gray bars). Amount of acetylation in figure 3C and 3F was quantified and normalized to the total acetylation of SeAcs (black bars). SlAcs-SeAcs chimera C3, the most efficiently acetylated, active chimera with the fewest SeAcs WT -derived residues, is noted with a star. Values are reported as the mean 6 S.D. of three experiments. B. Sequence alignment of SlAcs WT , SeAcs WT , chimera C3, Rhodopseudomonas palustris CGA009 Acs (RpAcs), and Mycobacterium smegmatis mc 2 155 Acs (MsAcs). Residues in chimera C3 that are derived from the SeAcs WT amino acid sequence are highlighted in black. SlAcs residues conserved in the MsAcs homologue are shown in bold typeface in the sequence of the latter. Black box indicates the target lysine. doi:10.1371/journal.pone.0099817.g004 encoding a SlPatA homologue, they concluded that ScAcs was under RLA control.
Our initial work with the S. lividans SlPatA WT and SlAcs WT enzymes paints a complex picture for the regulation of SlAcs WT function in this organism, and by extrapolation, maybe in S. coelicolor. Because the specific activity of SlAcs WT is similar to that of SeAcs WT in vitro (Fig. 5B), we hypothesize that SlAcs WT activity is also tightly controlled by S. lividans. To account for the inability of SlPatA WT to acetylate SlAcs WT , we propose that SlPatA WT has evolved unique strategies for substrate recognition, or SlPatA WT is not the primary modifier of SlAcs WT . We discuss each possibility further below.
In vitro, SlPatA WT does not Recognize SlAcs WT Pat homologues acetylate Acs in R. palustris and S. enterica [10,12]. Clearly, acetylation of SlAcs WT by SlPatA WT does not occur efficiently in vitro or in a heterologous model system (Figs. 4B, 5A, 6A, 7) [18]. The following possibilities should be taken into consideration when thinking about the potential regulation of SlAcs WT by RLA. First, it is possible that SlAcs WT may have evolved to evade acetylation by SlPatA WT . Secondly, since S. lividans encodes ,72 predicted GNAT-type acetyltransferases (Pfam00583) it is possible that one of these GNATs, not SlPatA WT , acetylates SlAcs WT . If a GNAT other than SlPatA acetylated SlAcs, it begs the questions of what selective pressure drove the conformational change SlAcs to avoid recognition by SlPatA, and what the physiological benefits of such a change are. And thirdly, the reversed domain organization of SlPatA, relative to RpPat and SePat, may prevent recognition of SlAcs WT by SlPatA WT .
Substantial Changes in the C-terminal Domain of SlAcs WT Lead to its Recognition by SlPatA WT Forty-one amino acid changes in the C-terminal domain of SlAcs WT were needed to allow SlPat WT to recognize and acetylate SlAcs (Fig. 3). If we assumed that the domain organization of SlPat WT was not a factor in SlAcs WT recognition, such a large number of substitutions would suggest that the protein underwent dramatic evolutionary changes to prevent modification by SlPat WT . Importantly, we note that some SlAcs sequences that were replaced in the C3 chimera exhibit homology to Acs homologues that are acetylated by GNAT enzymes in other bacteria (Fig. 4B). This suggests acetylation of Acs and other AMPforming acyl-CoA synthetases cannot be predicted by amino acid sequence [17]. Our results indicate, however, that SlAcs recognition by SlPatA WT is reversible by mutation, and that the resulting SlAcs variant can be reversibly acetylated.
How do Changes in the C-terminal Domain of SlAcs Affect its Acetylation and Activity?
Studies of R. palustris Pat (RpPat) substrate specificity indicate that this enzyme recognizes a loop .20 residues N-terminal to the target lysine in the substrate protein, suggesting that the RpPat interacts with a relatively large surface of substrate proteins [17]. Here, we demonstrate that the identities of residues ranging from 8-52 amino acids N-terminal to the target lysine of SeAcs WT , in combination with 5-17 amino acids C-terminal to the target lysine of SeAcs WT are critical for recognition of this substrate by SlPatA WT in isolation. This indicates that SlPatA WT recognizes several regions of the SeAcs C-terminal domain including the target lysine, residues N-terminal to the target lysine, and residues Cterminal to the target lysine. It is possible that these regions of the SeAcs C-terminal domain are necessary for direct interactions with the SlPatA protein. Alternatively, these regions may be necessary to position the target lysine for entry into the SlPatA active site. The crystal structure of SlPatA WT substrate SeAcs WT is known (PDB 1PG3, 1PG4) [33]. Comparison of this structure with the structure of SlAcs (structure not known) may distinguish these possibilities. Efforts to obtain the crystal structure of SlAcs are ongoing.
Is SlAcs WT Regulated by One or More Protein Acetyltransferases?
As mentioned above, SlAcs WT may have evolved to evade acetylation specifically by SlPatA WT . However SlAcs WT may be acetylated in vivo by one of the additional 72 predicted GNAT-type acetyltransferases (Pfam00583) encoded by the genome of this bacterium or by an enzyme independent pathway. The possibility that an alternative GNAT acetylates SlAcs WT more efficiently than SlPatA WT does is not unprecedented. It is known that the genome of R. palustris encodes a Pat homologue and a single-domain   . Activities of Chimera C3 and SeAcs WT are reduced in strains expressing SlPatA WT . H 6 -Chimera C3, H 6 -SlAcs, and H 6 -SeAcs were produced in S. enterica Dacs pat DcobB strain JE9152 harboring either a plasmid producing SlPatA WT or an empty vector. Strains were grown in NCE minimal medium supplemented with acetate (10 mM). Acs proteins were incubated in the presence or absence of SeCobB deacetylase and its cosubstrate NAD + . Acs activity was measured in an NADH-consumption assay. Values are reported as the mean 6 S.D. of three activity measurements. doi:10.1371/journal.pone.0099817.g007 GNAT protein acetyltransferase that share overlapping protein acetyltransferase substrates, and that both enzymes acetylate these substrates with different affinities [13]. Alternatively, SlAcs WT may be acetylated directly and non-enzymatic by the reactive metabolite acetyl-phosphate. This phenomenon has been characterized in E. coli and has been shown to affect the activity of the target enzymes [37,38]. Therefore, the possibility of SlPatA WT not being the sole regulator of SlAcs WT activity in S. lividans needs to be further investigated.
Does the Unique Domain Organization of SlPatA WT Affect Substrate Specificity?
Pat acetyltransferases are two-domain enzymes composed of a GNAT (acetyltransferase) domain and a large domain whose function is likely regulatory. In SlPatA WT , the GNAT domain is located at the N-terminus of the protein [18]. In contrast, in R. palustris and S. enterica, the domain order is reversed (i.e., GNAT domain is at the C-terminus of protein). SlPatA WT also has a collagen-like Gly-Pro-Ser motif in the large domain [18]. S. enterica and R. palustris Pat homologues efficiently acetylate their cognate Acs enzymes in vitro [10,12]. The alternate domain organization of SlPatA WT may account for the poor acetylation of SlAcs WT compared to SeAcs WT and SlAacS WT in vitro [18]. If SlPatA WT has evolved strategies for recognition of protein substrates differently from SePat and RpPat, our in vitro assay may be missing a factor that promotes efficient SlPatA WT recognition of SlAcs WT such as a small molecule, macromolecule (e.g. protein), or an as-yetunidentified intracellular condition. If this were the case, the amino acid changes introduced into SlAcs WT to generate the SlAcs-SeAcs chimera C3 obviate the need for additional factors or conditions for SlPatA WT recognition.