In Streptomyces lividans, acetyl-CoA synthetase activity is controlled by O-serine and N^ε-lysine acetylation

SUMMARY

Protein acetylation is a rapid mechanism for control of protein function. Acetyl-CoA synthetase (AMP-forming, Acs) is the paradigm for the control of metabolic enzymes by lysine acetylation. In many bacteria, type I or II protein acetyltransferases acetylate Acs, however, in actinomycetes type III protein acetyltransferases control the activity of Acs. We measured changes in the activity of the Streptomyces lividans Acs (SlAcs) enzyme upon acetylation by PatB using in vitro and in vivo analyses. In addition to the acetylation of residue K610, residue S608 within the acetylation motif of SlAcs was also acetylated (PKTRSGK⁶¹⁰). S608 acetylation rendered SlAcs inactive and non-acetylatable by PatB. It is unclear whether acetylation of S608 is enzymatic, but it was clear that this modification occurred in vivo in Streptomyces. In S. lividans, an NAD⁺-dependent sirtuin deacetylase from Streptomyces, SrtA (a homologue of the human SIRT4 protein) was needed to maintain SlAcs function in vivo. We have characterized a sirtuin-dependent reversible lysine acetylation system in Streptomyces lividans that targets and controls the Acs enzyme of this bacterium. These studies raise questions about acetyltransferase specificity, and describe the first Acs enzyme in any organism whose activity is modulated by O-Ser and N^ε-Lys acetylation.

Graphical Abstract

In Streptomyces lividans the activity of acetyl-CoA synthetase (SlAcs) is modulated by the acetylation of residues S608 and K610. The PatB enzyme of this bacterium acetylates K610, but whether S608 is chemically or enzymatically acetylated remains unclear. Deacetylation of SlAcs^Ac is catalyzed by the SlSrtA enzyme, an NAD⁺-dependent SIRT4 sirtuin homologue. To our knowledge, SlAcs is the first example of an AMP-forming acyl-CoA synthetase that is regulated by O-Ser and reversible N^ε-Lys acetylation.

INTRODUCTION

Control of protein function by chemical modifications is widespread in biology. Such modifications are diverse, including phosphorylation, hydroxylation, automodification, methylation, glycosylation, sumoylation, ubiquitination, prenylation, ADP-ribosylation, acylation, etc. (Walsh, 2006). Protein acylation itself is also diverse since acyl chains can vary in length, can have one or more carboxylate functional groups, the carbon skeleton may or may not be unsaturated, and the moieties can be used to generate O-acylated or N-acylated derivatives (Wagner & Payne, 2013, Chen et al., 2007, Garrity et al., 2007, Guan & Fierke, 2011).

Here we investigated the function of GCN5-related N-acetyltransferases (GNATs) (PF00583) from actinomycetes. Some GNATs control the activity of acetyl-CoA (AMP-forming) synthetases (Acs), which are the paradigm of metabolic enzymes whose activities are controlled by sirtuin-dependent, reversible lysine acetylation (sRLA) (Starai et al., 2002, Starai & Escalante-Semerena, 2004b, Chan et al., 2011, Crosby & Escalante-Semerena, 2014). Briefly, AMP-forming Acs enzymes (EC 6.2.1.1) of all domains of life catalyze the reaction shown in reactions 1 and 2. The reactions are reversible as long as pyrophosphate is not hydrolyzed to o-phosphate by pyrophosphatase (reaction 3).

$$\text{Acetate}+\text{ATP}\to \text{acetyl}-\text{AMP}+{\text{PP}}_{\mathrm{i}}(\mathrm{\Delta }{\mathrm{G}}^{\prime }°=10.04\text{kJ}/\text{mol})$$ (1)

(Frey & Arabshahi, 1995)

$$\text{Acetyl}-\text{AMP}+\text{CoA}\to \text{acetyl}-\text{CoA}+\text{AMP}(\mathrm{\Delta }{\mathrm{G}}^{\prime }°=-55.65\text{kJ}/\text{mol})$$ (2)

(Frey & Arabshahi, 1995)

$${\text{PP}}_{\mathrm{i}}+{\mathrm{H}}_2\mathrm{O}\to 2{\mathrm{P}}_{\mathrm{i}}(\mathrm{\Delta }{G}^{\prime }°=-21.92\text{kJ}/\text{mol})$$ (3)

(Thauer et al., 1977)

In a cell lacking sirtuin deacetylase, Acs is acetylated by a protein acetyltransferase (Pat) enzyme effectively abolishing reaction (1) (Starai & Escalante-Semerena, 2004b).

Acs proteins across all domains of life are very similar in length (650-700 residues) and contain 10 conserved regions named A1-A10 (Starai & Escalante-Semerena, 2004a, Gulick, 2009). Region A10 contains the active-site lysine that is modified by GNATs (Starai et al., 2002).

Given the relevance of Acs to central metabolism, it is not surprising that the activity of Acs orthologues in cells of all domains of life is under sRLA control (Castano-Cerezo et al., 2011, Xu et al., 2011, You et al., 2014, Schwer et al., 2006). There are exceptions, however. For example, the Acs enzyme of Streptomyces lividans (SlAcs) is very weakly acetylated by the type II protein acetyltransferase A enzyme (SlPatA) of this bacterium (Tucker & Escalante-Semerena, 2013). This observation is not due to SlPatA being a poor enzyme, because SlPatA can effectively acetylate the Salmonella enterica Acs (SeAcs) enzyme (Tucker & Escalante-Semerena, 2014, Tucker et al., 2014). These results are intriguing, because SlAcs and SeAcs share 52% end-to-end identity and an identical acetylation motif (PKTRSGK).

Recent work with the actinomycete Micromonospora aurantiaca provided insights into why SlPatA may not acetylate SlAcs as efficiently as expected. In the alluded work, Xu and co-workers identified a GNAT that acetylated the Acs enzyme of M. aurantiaca (Xu et al., 2014). This M. aurantiaca GNAT has two domains, a catalytic domain at the C-terminus, and an allosteric ACT-domain (ACT standing for aspartate kinase, chorismate mutase, and TyrA, named after the proteins that contain the domain) at the N-terminus. Members of this type of GNATs are referred to as type III or ACT-Pats and they have only been found in actinomycetes (Hentchel & Escalante-Semerena, 2015). The regulatory domain of type III protein acetyltransferases are substantially smaller (~400 residues) than those in type I and II protein acetyltransferases (~700 residues), which are found in actinomycetes and other microorganisms (Fig. S1) (Hentchel & Escalante-Semerena, 2015). Type I protein acetyltransferases have the large regulatory domain on the N-terminus, with type II acetyltransferases having the large regulatory domain on the C-terminus (Fig S1). Currently, the regulatory function of this domain is unknown but shares homology to NDP-forming CoA ligases (Starai & Escalante-Semerena, 2004b).

Here, we show that the MaACT-Pat enzyme (hereafter referred to as MaPatB) acetylates and deactivates SlAcs, and that the MaPatB homologue from Streptomyces lividans (SlPatB) can also acetylate SlAcs in vivo. Unexpectedly, we found that in S. lividans, SlAcs was acetylated at two sites within the acetylation motif. One modification was on residue S608 located at the −2 position relative to residue K610 of the acetylation motif, which was also acetylated. SlAcs was acetylated at S608 only when purified from S. lividans, but the enzyme had poor activity in vitro and could not be acetylated by MaPatB. These effects were reversed when a variant of SlAcs (SlAcs^S608T) was purified from S. lividans. That is, SlAcs^S608T isolated from S. lividans was active and was acetylated by MaPatB.

SlAcs isolated from E. coli was not acetylated at either Ser or Lys, suggesting that the E. coli PatZ enzyme (Castano-Cerezo et al., 2011) could not acetylate K610 of SlAcs, and that E. coli lacked the enzyme that modifies S608. At present, it is unclear whether or not Ser acetylation is enzymatic.

Lastly, quantification of SlAcs activity in S. lividans strains lacking either of its putative deacetylases (i.e., LdaA, SrtA, CobB) suggested that the SrtA sirtuin of this bacterium was likely responsible for maintaining SlAcs active in vivo via lysine deacetylation. To our knowledge, this is the first example of an acyl-CoA (AMP-forming) synthetase whose function is regulated by O- and N-acetylation.

RESULTS

The Micromonospora aurantiaca MaPatB enzyme acetylates Streptomyces lividans acetyl-CoA synthetase (SlAcs) in vitro.

Previous work from our laboratory showed that SlPatA (a type II Pat, Fig. S1) did not recognize SlAcs as substrate, but it did acetylate the acetoacetyl-CoA synthetase (SlAacS) from the same organism (Tucker & Escalante-Semerena, 2013). Recent work with Micromonospora aurantiaca identified a type III Pat enzyme containing an ACT-domain (PF01842, MaPatB, Fig. 1, Fig. S1) that acetylated its cognate Acs (Micau_0428, hereafter MaAcs) (Xu et al., 2014). Analysis of the genome of S. lividans identified locus EFD70633 as one encoding a homologue of MaPatB (41% identity, Fig. 1; hereafter SlPatB). Results in figure 2A show that SlPatA did not recognize MaAcs or SlAcs as substrates, but it did recognize SlAacS, a bona fide substrate of SlPatA (Tucker & Escalante-Semerena, 2013), indicating that SlPatA was active. In contrast, MaPatB acetylated SlAcs at residue K610 (Fig. 2B, lanes 2, 3, 4). These results were similar to those obtained with control experiments where MaPatB was incubated with its known substrate MaAcs (Fig. 2B, lanes 5, 6, 7). Despite the level of identity between MaPatB and SlPatB (Fig. 1), SlPatB failed to acetylate SlAcs or MaAcs under the conditions tested (Fig. 2B, lanes 9, 10).

Figure 1.

Protein alignment of MaPatB and SlPatB. Protein sequences of type III Pat enzymes containing ACT-domains (PF01842), MaPatB (Micau_0428) and SlPatB (EFD70633), were aligned using ESPrit 3.0 (Robert & Gouet, 2014) with default parameters. MaPatB and SlPatB share 41% identity (BLASTp) and the active site glutamate of these proteins is indicated by the red asterisk.

Figure 2.

SlAcs is acetylated in vitro by varying degrees with various protein acetyltransferases. Target proteins of interest (Acs) were incubated with acetyltransferase and [1-¹⁴C]-acetyl-CoA. Proteins were separated by SDS-PAGE and stained with Coomassie Brilliant Blue R to visualize proteins. Precision Plus Protein™ (BioRad) standard was used as a molecular marker (MM) with relevant molecular masses indicated in kDa. Acetylation was visualized by phosphor imaging (separated by horizontal black line). A. SlAcs and MaAcs incubated with [1-¹⁴C]-acetyl-CoA and ± SlPatA. SlAacS and SlPatA used as a positive control. SlAcs and MaAcs are bands corresponding to 71 kDa, SlAacS at 72 kDa, and SlPatA at 107 kDa. B. SlAcs^WT, SlAcs^K610A, MaAcs^WT, or MaAcs^K620A, were incubated with [1-¹⁴C]-acetyl-CoA and ± MaPatB or SlPatB (~35 kDa). MaAcs incubated with MaPatB was used as a positive control.

Effects of acetylation on SlAcs activity

Since SlPatB did not acetylate SlAcs in vitro but MaPatB did (Fig 2B), we used MaPatB to assess the effect of acetylation on SlAcs activity. The SlAcs enzyme used in the experiments described below was isolated from E. coli. Under the conditions tested, SlAcs activity decreased proportionally to the amount of MaPatB used in the experiment (0–3 μM), but leveled off after an ~50-60% decrease even in the presence of excess MaPatB (Fig. 3). In these experiments, a specific activity of ~30 μmol AMP min⁻¹ mg⁻¹ was considered to be 100%. The observed decrease in activity was consistent with MaPatB acetylation on MaAcs (Xu et al., 2014). The limited reduction in SlAcs activity was unusual in that homologues from other bacteria are >80% deactivated upon acetylation (Gardner & Escalante-Semerena, 2008, Gardner et al., 2006, Starai & Escalante-Semerena, 2004b, Crosby et al., 2010, Crosby et al., 2012a, Crosby et al., 2012b, Tucker & Escalante-Semerena, 2013). We investigated possible reasons for the limited deactivation of SlAcs.

Figure 3.

SlAcs activity decreases after incubation with acetyl-CoA and MaPatB. SlAcs (2 μM) isolated from E. coli (i.e., SlAcs in which residue S608 was not acetylated) was incubated with MaPatB (concentrations indicated on x-axis) and acetyl-CoA and specific activity (μmol AMP min⁻¹ mg⁻¹) was calculated using a continuous spectrophotometry assay as described in Experimental procedures. The activity of SlAcs measured after incubation in the absence of MaPatB was set as a value of 100% activity and SlAcs activity was compared with increasing amounts of MaPatB as indicated by the y-axis. Error bars represent standard deviation. Experiment was repeated three times in technical triplicate.

The acetylation state of SlAcs purified from S. enterica

To determine whether the loss of SlAcs activity was due to the acetylation of residue K610, we analyzed the acetylation state of SlAcs proteins synthesized by S. enterica using rabbit polyclonal anti-acetyllysine (α-AcK) antibodies. Results of control experiments with non-acetylated SlAcs and in-vitro acetylated SlAcs (acetylated by MaPatB) showed that AcK was readily detected under the conditions tested (Fig. 4, lanes 17, 18). A signal for AcK was detected in SlAcs purified from S. enterica cells that synthesized SlPatB or MaPatB but not from cells harboring vectors coding the inactive variants of both PatB enzymes (Fig. 4, lanes 13-16). These results raised two points. First, that SlPatB and MaPatB recognized SlAcs as substrate, and second, that residues E366 (SlPatB) and E269 (MaPatB) were important for enzyme activity.

Figure 4.

SlAcs is acetylated in S. enterica strains that synthesize active PatB acetyltransferases. Cell lysates were obtained from Δacs/pSlAcs strains expressing the acetyltransferase as indicated above each lane. pSlPatB^E366Q and pMaPatB^E269Q refer to plasmids carrying alleles that encode inactive forms of SlPatB or MaPatB, respectively. Acetylation state of SlAcs from lysates (left panel, SDS-PAGE) was analyzed by western blot analysis using anti-acetyllysine (right panel, α-K^Ac). In vitro acetylated and non-acetylated SlAcs were used as positive and negative controls respectively. SlAcs is the band at ~72 kDa. CFEs, cell-free extracts.

Discovery of O-Ser acetylation in SlAcs

To determine acetylation state of SlAcs when it is synthesized in S. lividans, H₆-SlAcs^WT protein was purified from S. lividans. Unexpectedly, the specific activity of SlAcs^WT enzyme isolated from S. lividans (~2 μmol AMP min⁻¹ mg⁻¹) was 20- to 30-fold lower than that of SlAcs^WT purified from E. coli (Fig. 5A). In fact, the activity of SlAcs purified from S. lividans was similar to that of an inactive SlAcs variant (SlAcs^K610A; Fig. 5A). Results from size-exclusion chromatography experiments (column volume 25 mL with a flow rate of 0.5 ml min⁻¹) showed that the retention time and peak distribution of SlAcs purified from S. lividans (24 min) was unchanged from that of SlAcs purified from E. coli (24 min, Fig. S3), suggesting that the proteins were probably not misfolded.

Figure 5.

SlAcs^WT purified from Streptomyces strains does not have activity in vitro and is not acetylated by MaPatB. A. His₆-SlAcs^WT was purified from E. coli or Streptomyces as indicated on x-axis, and activity of SlAcs was tested using a spectrophotometry coupled assay. SlAcs purified from E. coli was used as a positive control and an inactive variant (SlAcs^K610A) was used as a negative control. Specific activity is reported as μmol AMP min⁻¹ mg⁻¹ of SlAcs. SlAcs^WT was purified from S. lividans in biological triplicate and each biological replicate was tested in technical triplicate. The technical triplicates of one biological replicate shown as a representative with error bars indicating standard deviation. Asterisks represent significance of SlAcs^WT purified from E. coli compared to SlAcs^WT purified from S. lividans strains (indicated on x-axis) by calculating P value using an unpaired t-test, ^**** represents a P value below 0.00005. B. SlAcs^WT purified from Streptomyces (lane 2) or E. coli (lane 3) was incubated with MaPatB and [1-¹⁴C]-acetyl-CoA. Last lane (lane 3) was a negative control of SlAcs purified from E. coli incubated only with [1-¹⁴C]-acetyl-CoA. Proteins were separated by SDS-PAGE and stained with Coomassie Brilliant Blue R to visualize proteins. Precision Plus Protein™ (BioRad) standard was used as a molecular marker (MM) with relevant molecular masses indicated in kDa. Acetylation was visualized by phosphor imaging (separated by horizontal black line). His₆-SlAcs at ~72 kDa and MaPatB at ~35 kDa.

We note surprisingly, that the SlAcs^WT protein purified from S. lividans strains was not acetylated by MaPatB in vitro compared to the positive control (Fig. 5B, lane 2 vs 3.). To determine whether this observation was due to unknown posttranslational modifications on SlAcs, LC/MS/MS peptide fingerprinting was performed on SlAcs purified from different sources, namely i) SlAcs purified from E. coli and incubated with MaPatB + AcCoA, ii) SlAcs purified from E. coli and incubated with AcCoA but no MaPatB, and iii) SlAcs purified from S. lividans. SlAcs^WT protein purified from S. lividans was acetylated at residue serine (S608) and residue K610 (Fig. 6A). In contrast, residue S608 was not acetylated in SlAcs^WT protein purified from E. coli, and this protein could be acetylated by MaPatB at residue K610 as shown by mass spectrometry (Fig. 6B). The SGKIMR peptide of untreated SlAcs protein purified from E. coli was not detected in the LC/MS/MS most likely because trypsin cleaved this peptide at K and R, rendering it too small for detection (data not shown).

Figure 6.

SlAcs purified from Streptomyces is acetylated at S608 and L610. A. S^AcGK^AcIMR peptide identified by LC/MS/MS analysis of SlAcs purified from S. lividans. B. SGKIMR peptide identified by LC/MS/MS of SlAcs acetylated by MaPatB in vitro. No peptide was detected for non-acetylated control. b ions are the series of fragments that extend from the N-terminus; y ions are the series of ions that extend from the C-terminus. MASCOT software (http://www.matrixscience.com) was the online search engine used to identify peptides on the basis of their masses.

In the past, our group reported that single substitutions in the acetylation motif can render an acyl-CoA synthetase non-acetylatable (Crosby & Escalante-Semerena, 2014). For this reason, the S. lividans acs⁺ allele was mutated to encode a S608T variant (SlAcs^S608T). Threonine was chosen because Thr is a residue that is conserved in other S. lividans CoA synthetases (Tucker & Escalante-Semerena, 2013). The specific activity of variant H₆-SlAcs^S608T purified from S. lividans was ~25% lower than that of SlAcs^WT purified from E. coli (Fig. 7A, black vs blue bar), but more importantly, the H₆-SlAcs^S608T protein purified from S. lividans became a substrate for MaPatB in vitro (Fig. 7B, lanes 2–7). Notably, the level of activity of SlAcs^WT isolated from S. lividans was not different than the level of the activity of the inactive variant SlAcs^K610A also isolated from E. coli (Fig. 7, red vs gray bar). From an enzyme functionality standpoint, this result showed that although Ser and Thr have a hydroxyl group in their side chains that could be modified, the Ser side chain was acetylated, inactivating the enzyme, while the SlAcs^S608T protein was not, resulting in active enzyme.

Figure 7.

Variant SlAcs^S608T has higher activity compared to SlAcs^WT when purified from Streptomyces, and SlAcs^S608T is acetylated in vitro by MaPatB. Panel A. His₆-SlAcs^S608T and His₆-SlAcs^WT was purified from Streptomyces and SlAcs activity was calculated using a continuous spectrophotometric assay and reported in μmol AMP min⁻¹ mg⁻¹. SlAcs was purified from S. lividans in biological triplicate and each biological replicate was tested in technical triplicate. The technical triplicates of one biological replicate shown as a representative with error bars indicating standard deviation. SlAcs purified from E. coli was used as positive control and an inactive variant (SlAcs^K610A) was used as negative control. Asterisks represent significance of SlAcs^WT compared to SlAcs^S608T by calculating P value using an unpaired t-test, ^*** represents a P value below 0.0005. Panel B. His₆-SlAcs^S608T purified from Streptomyces strains (as indicated above lanes) or E. coli was incubated with MaPatB and [1-¹⁴C]-acetyl-CoA. Last lane was a negative control of His₆-SlAcs purified from E. coli incubated with only [1-¹⁴C]-acetyl-CoA.

The SlSrtA enzyme (a SIRT4-like sirtuin) deacetylates SlAcs^Ac in S. lividans

To determine whether acetylated SlAcs^S608T could be enzymatically deacetylated, we looked for genes in the S. lividans genome putatively encoding protein deacetylases; we found three such genes (ldaA, srtA, and cobB). One of them encoded a homologue of the NAD⁺-dependent SIRT5 deacetylase, (a homologue of SeCobB), another one encoded a homologue of the NAD⁺-dependent SIRT4 sirtuin, and a third one encoded a homologue of the Zn-dependent, acetate-forming RpLdaA deacetylase (Crosby et al., 2012a, Tucker & Escalante-Semerena, 2013).

H₆-SlAcs^S608T protein was purified from five S. lividans strains, namely i) patB⁺ ldaA⁺ cobB⁺ srtA⁺, ii) ΔpatB ldaA⁺ cobB⁺ srtA⁺, iii) patB⁺ΔldaA cobB⁺ srtA⁺, iv) patB⁺ ΔldaA ΔcobB srtA⁺, and v) patB⁺ ΔldaA ΔcobB ΔsrtA. The H₆-SlAcs^S608T variant was purified from each strain and its activity was quantified. The activity of H₆-SlAcs^S608T was not significantly different among proteins purified from patB⁺ ldaA⁺ cobB⁺ srtA⁺, patB⁺ΔldaA cobB⁺ srtA⁺, or patB⁺ ΔldaA ΔcobB srtA⁺ strains (Fig. 8A). However, the activity of SlAcs^S608T isolated from the patB⁺ ΔldaA ΔcobB ΔsrtA strain was 50% lower than that of the H₆-SlAcs^S608T variant purified from the control strain patB⁺ ldaA⁺ srtA⁺ cobB⁺ (Fig. 8A, black vs yellow bar)). This reduction of activity strongly suggested that in S. lividans, SlAcs was under reversible lysine acetylation control exerted by the SlPatB and SrtA enzymes. In S. coelicolor, SrtA is equivalent to CobB1, which was shown to deacetylate ScAcs^Ac (Mikulik et al., 2012). Unfortunately, S. lividans SrtA and CobB deacetylases were not active when purified from E. coli (Tucker & Escalante-Semerena, 2013). Based on the in vivo data reported in figure 8 we propose that SrtA is likely the deacetylase that used SlAcs^Ac as substrate in S. lividans.

Figure 8.

SlAcs is deacetylated by SrtA in S. lividans. A. SlAcs^S608T was purified from S. lividans strains indicated on the x-axis. Specific activity is reported as μmol AMP min⁻¹ mg⁻¹ of SlAcs. SlAcs^S608T was purified from S. lividans in biological triplicate and each biological replicate was tested in technical triplicates. Genes deleted in strains from which SlAcs^S608T was isolated are shown in red type. The technical triplicates of one biological replicate shown as a representative with error bars indicating standard deviation. Asterisks represent significance compared to patB⁺ ldaA⁺ cobB⁺ srtA⁺ by calculating P value using an unpaired t-test, ^** represents a P value below 0.005. B. SlAcs^WT or SlAcs^S608T (5 μM) both isolated from E. coli, was incubated with Acetyl-CoA with or without MaPatB (1 μM). Activity of SlAcs was monitored using a continuous spectrophotometry assay and activity reports as μmol AMP min⁻¹ mg⁻¹ of SlAcs. ^**** represents a P value of < 0.0001. Experiment was repeated in triplicate and error bars represent standard deviation of triplicates from one replicate.

Acetylation of the SlAcs^S608T variant protein by MaPatB decreases its activity

To assess the effect of acetylation on SlAcs^S608T activity, SlAcs^S608T and SlAcs^WT were purified from E. coli and were acetylated (5 μM) in vitro by MaPatB (1 μM). When SlAcs^WT was acetylated, it lost ~20% activity compared to the non-acetylated control (Fig. 8B). These results are comparable to the data in Figure 3, where a 20% decrease in SlAcs^WT activity was seen with similar ratios of SlAcs to MaPatB (i.e ~1:5) In contrast, when SlAcs^S608T was acetylated under the same conditions it lost ~80% activity (Fig. 8B). These results raised important questions about the control of CoA synthetases in S. lividans and the potential sequential control of CoA synthetases through posttranslational modifications of multiple residues. These data also raised questions about how this protein may be deacetylated, and if the deacetylation of serine or lysine is dependent on deacetylation of the subsequent residue. These results are discussed below.

Initial in vitro studies of protein substrate specificity among Pat enzymes

The differences in SlAcs acetylation by SlPatA and MaPatB (Fig. 2A, B) raised questions about the specificity of Pat enzymes for Acs substrates. To further examine this specificity, SlAcs was incubated with other Pat enzymes. We found that Rhodopseudomonas palustris Pat (RpPat, a type I enzyme) efficiently acetylated SlAcs (Fig. S4A, lanes 2, 3, 4) compared to the positive control of RpPat with its known substrate RpBadA (Fig. S4A, lane 8) (Crosby et al., 2010). In contrast, incubation of SlAcs with Salmonella enterica Pat (SePat) yielded a signal that was only ~16% as strong as the signal obtained with SeAcs, a bona fide substrate of SePat (Fig. S4, lane 8). Although MaAcs and SlAcs are 76% identical (Fig. S2), SePat acetylated MaAcs more efficiently (~5-fold stronger signal than SlAcs) (Fig. S4B, lanes 2 vs 5). With the exception of SePat, all other Pat enzymes tested acetylated MaAcs and SlAcs very similarly (Fig. 2).

In vivo evidence supports the observed difference in substrate specificity among Pat enzymes

We performed in vivo experiments to determine whether the differences in protein substrate specificity of Pat enzymes observed in vitro, could be replicated under conditions where Acs activity was required for growth. To do this, we grew S. enterica strains with low concentrations of acetate (10 mM), a condition where Acs activity is needed (Starai & Escalante-Semerena, 2004a). In wild-type S. enterica, the absence of CobB results in a substantial decrease in cell growth with 10 mM acetate (Starai et al., 2003). Such a phenotype is due to the acetylation of Acs by Pat (Starai & Escalante-Semerena, 2004b). Surprisingly, a S. enterica Δacs ΔcobB pat⁺ strain carrying a plasmid encoding SlAcs grew better than the acs⁺ cobB⁺ pat⁺ strain with 10 mM acetate (Fig. S5, compare gray squares, black circles, white diamonds). These results were consistent with the idea that SePat, which was present in all strains tested, did not acetylate SlAcs in vivo to the point of affecting the ability of the cell to grow with low acetate concentrations. In contrast, when the Δacs ΔcobB pat⁺ strain carrying a plasmid encoding SlAcs also carried the pSlPatB plasmid growth was delayed for ~20 h, after which the cultures grew at a rate comparable to that of the wild-type strain and reached full density (Fig S5, black vs white circles).

Exploring the specificity of Pat enzymes for SlAcs using an in vivo heterologous system

To monitor SlAcs activity and its control by different Pat enzymes in vivo, we used Salmonella enterica strains in which the posttranslational control of Acs and the effects of the absence of control elements (i.e., Pat, CobB) on growth have been extensively characterized (Starai et al., 2002, Starai & Escalante-Semerena, 2004b, Starai et al., 2003, Chan et al., 2011). Several genes were cloned, introduced into a S. enterica Δacs strain. The growth behavior with 10 mM acetate as the sole source of carbon and energy was monitored as a function of time. The presence of the pat⁺ allele in all strains was not a concern since SePat did not acetylate SlAcs extensively in vitro (Fig. S4B). As discussed below, results from the experiments confirmed this observation.

A S. lividans acs⁺ allele coding for the wild type protein with a hexahistidine N-terminal fusion was cloned into an arabinose inducible complementation vector encoding resistance to ampicillin (VanDrisse & Escalante-Semerena, 2016). The genes encoding different types of Pat enzymes included S. lividans patA⁺, S. lividans patB⁺, and M. auriantica patB⁺, which were cloned into a compatible arabinose inducible vector encoding chloramphenicol resistance and the resulting plasmids were transformed into strain JE21742 (Δacs/pSlAcs). The resulting strains were grown in minimal medium containing acetate (10 mM). Under those conditions cell growth required Acs function. We predicted that acetylation of Acs would lower its activity reducing the growth rate and final cell density of the cultures.

In figure 9A, the behavior of positive- and negative-control strains growing in minimal medium with low acetate (10 mM) as the sole source of carbon and energy is shown by black circles (acs⁺) and grey squares (Δacs), respectively. When the Δacs strain synthesized SlAcs, growth occurred without delay (Fig. 9A, white diamonds), indicating that SlAcs was functional. Notably, growth of the Δacs strain that synthesized SlPatA and SlAcs was not different than that of the Δacs strain that only synthesized SlAcs (Fig. 9A, white diamonds vs grey triangles). These results suggested that neither SePat nor SlPatA inactivated SlAcs. In contrast, cultures of Δacs strains that synthesized SlAcs and either SlPatB or MaPatB displayed an ~30-h long lag phase before the onset of exponential growth (Fig. 9A, white diamonds vs grey diamonds or white circles). We hypothesized that the observed lag phase was associated with an increase in SlAcs acetylation, which would result in lower SlAcs activity. To test this idea, the same strains were grown with acetate, cells were collected at mid-log (OD₆₃₀ ~0.15), and lysed and SlAcs activity was quantified. SlAcs activity decreased by 10% in Δacs/pSlAcs cells harboring a plasmid encoding SlPatA compared to the vector-only control (Fig. 9B). In contrast, a ~60-70% decrease in SlAcs activity was measured when the enzyme was purified from strain Δacs/pSlAcs/pSlPatB or from strain Δacs/pSlAcs/pMaPatB. These percent activities were relative to that of SlAcs purified from cells that did not express an acetyltransferase (Fig. 11B, black bar). These results were consistent with the idea that that the above-mentioned lag phase was likely due to a decrease in SlAcs activity due to acetylation. These results also suggested that although SlPatB was not active in vitro, it did acetylate SlAcs in vivo, thereby reducing its activity.

Figure 9.

Overexpression of PatB acetyltransferases in S. enterica lowers SlAcs activity in cell lysates. A. The gene coding for SlAcs was cloned into pCV1 and introduced into a S. enterica Δacs strain carrying genes coding for acetyltransferases of interest cloned into pCV3. Overnight cultures were grown in NB + antibiotic. Cells were sub-cultured (1% v/v) and grown on NCE minimal medium supplemented with acetate (10 mM). Growth curves were obtained in technical triplicate and was repeated three times using a microplate reader (BioTek instruments). Error bars represent standard deviation of technical triplicates. B. Cells grown in 1 L of same medium conditions from A were harvested and lysed and SlAcs specific activity (μmol AMP min⁻¹ mg⁻¹) was quantified using a continuous spectrophotometry assay. SlAcs was isolated from strains containing plasmids listed below bars. Error bars represent standard deviation of technical triplicates. Asterisks represent significance compared to vector control by calculating P value using an unpaired t-test, ^*** represents a P value below 0.0005, ^**** represents a P value below 0.00005, and ns represents no significance.

DISCUSSION

The fact that the AMP-forming acetyl-CoA synthetase from Streptomyces lividans (SlAcs) was poorly acetylated by SlPatA (Tucker & Escalante-Semerena, 2013) was the driving force for the work presented in this paper. Here we show that SlAcs activity is controlled by sirtuin-dependent reversible lysine acetylation (sRLA), and that the enzyme responsible for this modification is a type III protein acetyltransferase thus far found only in actinomycetes. The sRLA control of SlAcs activity is novel because of the unprecedented involvement of an O-acetylation modification of the side chain of a serine residue close (position -2) to the lysyl chain that is typically acetylated in AMP-forming acyltransferases. Serine and threonine acetylation in bacteria was only recently reported, with the paradigm acetyltransferase being YopJ from Yersinia pestis (Mukherjee et al., 2007). YopJ acetylates lysine, threonine, and serine residues of host kinases (Mittal et al., 2010). While YopJ is a prokaryotic acetyltransferase that O-acetylates a eukaryotic protein, SlAcs is the first example of a prokaryotic AMP-forming acyl-CoA synthetase that is controlled by O-Ser and N^ε-Lys acetylation. Additionally, a recent acetylome study in Mycobacterium tuberculosis identified many acetylated serine residues on a myriad of proteins (Birhanu et al., 2017). Although the enzyme responsible for acetylating serine residues has not yet been identified in M. tuberculosis, these studies highlight the prevalence of O-serine acetylation in prokaryotes, an area that needs further investigation.

Type III protein acetyltransferases acetylate Acs in Actinomycetes

In several Alpha- and Gamma-proteobacteria, AMP-forming Acs enzymes are regulated by type I protein acetyltransferases [for a review see (Hentchel & Escalante-Semerena, 2015)]. Here we identified the SlPatB enzyme (a type III protein acetyltransferase) of Streptomyces lividans as the enzyme that controls the activity of SlAcs at the expense of AcCoA (Fig. 2). SlPatB is a homologue of MaPatB from Micromonospora aurantiaca, which was shown to be allosterically regulated by cysteine and arginine (Xu et al., 2014). It is likely that Acs control in S. lividans is further affected by physiological conditions that modulate cysteine and arginine metabolism, given that fact that MaPatB activity is allosterically regulated by these two amino acids. At present, it is unclear whether or not SlPatB is allosterically controlled because this enzyme was not active under the in vitro conditions tested. Bioinformatics analyses show that only genomes of actinomycetes encode type III protein acetyltransferases, raising important questions about the possible link between amino acid metabolism and protein acetylation in these bacteria.

O-Serine and N^ε-lysine acetylation controls the activity of SlAcs in Streptomyces lividans

The regulation of Acs in S. lividans appears to be more complex than that of Acs homologues found in other bacteria. To our knowledge, SlAcs is the first AMP-forming AcCoA synthetase whose activity is posttranslationally controlled by O-Ser and reversible N^ε-Lys acetylation.

SIAcs synthesized by S. lividans was acetylated at the expected Lys residue within the acetylation motif (site 0), and at residue S608 located two positions upstream the Lys acetylation site (site -2; Fig. 6A). At this point, it is not clear what the order of acetylation may be, however, we propose that two different acetyltransferases modify SlAcs in S. lividans, since synthesis of SlAcs in E. coli did not yield enzyme acetylated at S608, suggesting that E. coli lacks the O-acetyltransferase that modifies residue S608 (Fig. 6B). Although chemical (i.e. non-enzymatic) serine acetylation is rare, at present we cannot rule out this possibility. It appears unlikely that this would be the case, since SlAcs isolated from E. coli S608 was not acetylated. Further studies on S608 acetylation in Streptomyces are needed.

We suggest that in S. lividans full downregulation of SlAcs activity only occurs when S608 is acetylated (Fig. 8B). We suggest that this is the case because when we attempted to acetylate SlAcs isolated from E. coli, SlAcs activity was only decreased by 20-40%, even after extended incubation with the acetyltransferase. Since E. coli does not acetylate S608, the acetyltransferase can modify K610. In contrast, SlAcs isolated from S. lividans was modified at residue S608 (Fig. 6A) and lacked activity (Fig. 5). The absence of activity correlated with the presence of acetylated serine, because when serine variant SlAcs^S608T was isolated from S. lividans strong SlAcs activity was measured (Fig. 7A). The results with SlAcs^S608T also suggest that the putative serine acetyltransferase cannot recognize threonine as its substrate.

Acetylation of S608 is of interest given the location of this residue within the acetylation motif found in CoA synthetases [PX₄GK, (Crosby & Escalante-Semerena, 2014)]. In SlAcs, the acetylation motif is PKTRSGK⁶¹⁰, with the modified serine and lysine residues shown in bold typeface. Based on the data reported here, it appears like there are several layers of control of SlAcs in S. lividans. At present is not known whether S608 acetylation is reversible, but if it were, the identification of the S608^Ac deacetylase would be an important step towards understanding the integration of two reversible acetylation/deacetylation events that control S608 and K610 modification. Why would such a mechanism of Acs posttranslational control be needed in Streptomyces? Is this control mechanism used to modulate the activity of other enzymes in prokaryotes or eukaryotes? These and other interesting questions warrant further investigation in this area of Streptomyces physiology.

An opportunity to identify determinants that play a critical role in acyltransferase specificity

Questions regarding how acyltransferases recognize their protein targets are of interest. We have published structural studies aimed at identifying residues contributing affecting the recognition and specificity of SlPatA for SeAcs (Tucker et al., 2014). One limitation of the alluded work was that it was performed with proteins from different sources. The studies reported here open the door for structural work involving enzymes partners from the same organism. A better understanding of the specificity of acyltransferases for their protein targets may help identify such targets.

Deacetylation of SlAcs in Streptomyces

The S. lividans genome codes for three deacetylases, two of which are NAD⁺-dependent sirtuin deacylases (Sauve, 2010). The two sirtuins in this bacterium are different from each other. One of them, SlCobB (formerly CobB2), is homologous to human SIRT5 a known demalonylase, desuccinylase (Matsushita et al., 2011). The other, SlSrtA (formerly CobB1) is homologous to human SIRT4, a known ADP-ribosyltransferase (Haigis et al., 2006). SIRT5-like enzymes are the best characterized family of sirtuins in bacteria (e.g., SeCobB, RpSrtN, EcCobB).

Results presented in figure 8A suggest that SlSrtA is responsible for maintaining SlAcs deacetylated (thus active) in vivo. Unfortunately, the function of putative deacetylases was not studied in vitro due to difficulties in obtaining active enzymes. At present, it is unclear why SlCobB and SlSrtA proteins were inactive in isolation. However, results of in vivo experiments show that SlSrtA is is an active participant in the regulation of SlAcs activity, hence offering avenues for future research. Prior to this publication there were no known substrates for S. lividans protein deacylases.

EXPERIMENTAL PROCEDURES

Bacterial strains, culture media, chemicals, and sequencing methods

All strains used in this study are listed in Table 1. Escherichia coli C41 (λDE3) (Miroux & Walker, 1996) and DH5α (New England Biolabs) strains were grown in lysogeny broth (LB, Difco) at 37°C. Salmonella strains used for growth analysis were derivatives of Salmonella enterica subsp. enterica sv Typhimurium strain LT2 (hereafter S. enterica) and grown at 37°C in nutrient broth (NB, Difco) with NaCl (85 mM), or no-carbon essential (NCE) minimal medium (Berkowitz et al., 1968) supplemented with sodium acetate (10 mM), L-methionine (0.5 mM), MgSO₄ (1 mM), and trace minerals (Balch & Wolfe, 1976). When in the medium antibiotics were present at the following concentrations: ampicillin, 100 μg mL⁻¹; chloramphenicol, 20 μg mL⁻¹.

Table 1.

Bacterial Strains and plasmids used in this study

Strain	Relative Genotype	Source1
E. coli strains
E. coli DH5α	Φ80dlacZ∆M15 recA1 endA1 gyrA96 thi-1 hsdR17 (rk-, mk+) supE44 relA1 deoR ∆(lacZYA-argF) U169 phoA	NEB
E. coli C41 (DE3)	pka12::kan+ ompT hsdS (rBmB) gal λ (DE3)	Laboratory collection
S. enterica strains
JE6583	metE205 ara-9	K. Sanderson via J. Roth
Derivatives of JE6583
JE22070	pCV1/pCV3
JE7758	Δacs2	Laboratory collection
JE21853	Δacs2/pCV1/pCV3
JE21760	Δacs2/pSlAcs50/pCV3
JE21763	Δacs2/pSlAcs50/pSlPatA57
JE21762	Δacs2/pSlAcs50/pSlPatB4
JE21761	Δacs2/pSlAcs50/pMaPatB1
JE23482	Δacs2 Δpat3/pSlAcs50/pCV3
JE23483	Δacs2 Δpat3/pSlAcs50/pSlPatA57
JE23484	Δacs2 Δpat3/pSlAcs50/pSlPatB4
JE23486	Δacs2 Δpat3/pSlAcs50/pSlPatB7
JE23485	Δacs2 Δpat3/pSlAcs50/pMaPatB1
JE23487	Δacs2 Δpat3/pSlAcs50/pMaPatB5
S. lividans strains
TK24	Wild type
Derivatives of S. lividans TK24
JE23392	TK24/pSlAcs31
JE23583	ΔEFD68590/pSlAcs31
JE23581	ΔEFD68590 ΔEFD71509/pSlAcs31
JE23398	ΔEFD68590 ΔEFD71509 ∆EFD65580/pSlAcs31
JE23639	ΔEFD70633/pSlAcs31
JE23900	TK24/pSlAcs52
JE23952	ΔEFD68590/pSlAcs52
JE23954	ΔEFD68590 ΔEFD71509/pSlAcs52
JE23956	ΔEFD68590 ΔEFD71509 ΔEFD65580/pSlAc52
JE23948	ΔEFD70633/pSlAcs52

Streptomyces strains were all derivatives of Streptomyces lividans strain TK24. ISP-2 medium (Shirling & Gottlieb, 1966) or R2YE medium (Kieser et al., 2000c, Kieser et al., 2000a) was used to culture S. lividans on solid medium. Liquid cultures were grown on yeast extract-malt extract (YEME) rich medium in baffled flasks (cultures under 200 mL) or with marine-grade stainless steel springs (NMMP) (cultures over 500 mL) to reduce cell clumping. Strains were grown 2-5 days at 30°C at 220-250 rpm. Antibiotics were used at the following concentrations: thiostrepton, 10 μg mL⁻¹, ampicillin, 100 μg mL⁻¹, chloramphenicol, 20 μg mL⁻¹. All chemicals were purchased from Fisher Chemical Co. unless noted otherwise; chloramphenicol, L(+)-arabinose (Sigma-Aldrich); and isopropyl β-D-1-thiogalactopyranoside (IPTG, IBI Scientific). All restriction enzymes were purchased from Thermo Scientific™ with the exception of BspQI (New England Biolabs).

Molecular techniques

DNA manipulations were carried out using standard molecular techniques (Elion et al., 2007). DNA was amplified using Pfu Ultra II Fusion DNA polymerase (Agilent) or Phusion High-Fidelity DNA Polymerase (New England Biolabs). Site-directed mutagenesis was performed using the Quikchange™ Site Directed Mutagenesis protocol (Agilent). Plasmids were purified 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 at the Georgia Genomics Facility. Primers were synthesized from Integrated DNA Technologies and are listed in Table 2.

Table 2.

Plasmids used in this study

Plasmid	Genotype	Source1
pTEV5	lacI+ bla+	(Rocco et al., 2008)
pTEV6	lacI+ malE+ bla+	(Rocco et al., 2008)
pTEV18	lacI+ bla+	(VanDrisse & Escalante-Semerena, 2016)
pTYB1	lacI+ bla+
pCV1	araC+ bla+	(VanDrisse & Escalante-Semerena, 2016)
pCV3	araC+ cat+	(VanDrisse & Escalante-Semerena, 2016)
pKC1139	Temperature sensitive shuttle vector, apr+	(Bierman et al., 1992)
pRK2013	Self-transmissible helper plasmid, mob+, tra+, kan+	(Figurski & Helinski, 1979)
pSE34	PermE constitutive expression vector, bla+, tsr+	(Yoon et al., 2002)
Overexpression plasmids
pSlAcs1	S. lividans acs+ (EFD68454) in pTEV5, bla+	(Tucker & Escalante-Semerena, 2014)
pSlAcs2	S. lividans acs+ (AcsK610A) in pTEV5, bla+
pMaAcs3	M. aurantiaca acs + (Micau_0428) in pTEV18, bla+
pACS67	S. enterica acs+ in pTEV18, bla+	(VanDrisse & Escalante-Semerena, 2016)
pACS28	S. enterica acs+ (AcsK609A) in pTYB1, bla+	(Chan et al., 2011)
pPAT8	S. enterica pat+ in pTEV6, bla+	(Thao & Escalante-Semerena, 2011)
pSlPatA1	S. lividans patA+ (EFD66247) in pTEV5, bla+	(Tucker & Escalante-Semerena, 2013)
pMaPatB3	M. aurantiaca patB+ (Micau_1670) in pTEV18, bla+
pSlPatB1	S. lividans patB+ (EFD70633) in pTEV5, bla+
pCOBB71	S. enterica cobBs+ in pTEV6, bla+	(Tucker & Escalante-Semerena, 2010)
pRpPat13	R. palustris pat in pTEV5	(Crosby et al., 2010)
pBadA3	R. palustris badA in pTEV5	(Crosby et al., 2010)
Complementation plasmids
pSlAcs50	S. lividans acs+ in pCV1, bla+
pMaPatB1	M. aurantiaca patB+ in pCV3, cat+
pSlPatB4	S. lividans patB+ in pCV3, cat+
pSlPatA57	S. lividans patA+ in pCV3, cat+
pSlPatB7	S. lividans patB coding for PatBE366Q in pCV3, cat+
pMaPatB5	M. aurantiaca patB coding for PatBE269Q in pCV3, cat+
Streptomyces deletion constructs
pKC1139EFD68590	Upstream and downstream regions of EFD68590 (acuC) in pKC1139
pKC1139EFD71509	Upstream and downstream regions of EFD71509 (cobB) in pKC1139
pKC1139EFD65580	Upstream and downstream regions of EFD65580 (srtA) in pKC1139
pKC1139EFD70633	Upstream and downstream regions of EFD70633 (patB) in pKC1139
Streptomyces complementation
pSlAcs31	S. lividans acs+ (His6, EFD68454) in pSE34, tsr+, bla+
pSlAcs52	S. lividans acs coding for His6-AcsS608T (EFD68454) in pSE34, tsr+, bla+

¹Unless noted, all plasmids and strains were constructed in this study.

onstruction of plasmids for protein production and complementation

The following genes were amplified from purified genomic DNA using primers listed in Table 3: Streptomyces lividans TK24 acs (EFD68454), patA (EFD66247), patB (EFD70633), aacS (EFD70521), Micromonospora aurantiaca ATCC 27029 acs (Micau_0428), patB (Micau_1670), Salmonella enterica LT2 pat (STM2651), acs (STM4275), Rhodopseudomonas palustris CGA009 badA (RPA0661), and pat (RPA4240). For Salmonella complementation studies, the start codons of S. lividans acs and patA genes (GTG, TTG, respectively) were changed to ATG. The DNA fragments and corresponding plasmids were digested with BspQI and cloned using protocols described elsewhere described (VanDrisse & Escalante-Semerena, 2016, Galloway et al., 2013).

Table 3.

Primers used in this study

Primer Name	Primer Sequence2 5′ → 3′
Overexpression and complementation primers
5′ SlAcs ATG-His6 pCV1	NNGCTCTTCNTTCATG CATCATCACCATCACCACAGCAAC GAATCCTTGGCCAAC
3′ SlAcs pCV1	NNGCTCTTCNTTATCAGTCCTCGCTGGGGGCGGCCG
5′ MaAcs pCV1 pTEV18	NNGCTCTTCNTTCATGAGCGAGGCATTGGCCAACTTGCTG
3′ MaAcs pCV1 pTEV18	NNGCTCTTCNTTATCAGTCCTCGTCGGACTTCCCGCC
5′ SlPatA ATG pCV3	NNGCTCTTCNTTCATGTCGTACGCGAGCCGTACTCTG
3′ SlPatA pCV3	NNGCTCTTCNTTATCAGCCGCCTGGGCGGTGGTGCCCG
5′ SlPatB pCV3	NNGCTCTTCNTTCATGAGTCGAGACATGTCTGATGTG ACGG
3′ SlPatB pCV3	NNGCTCTTCNTTATCACTCCCGGCCGATCCGCTCCGCCT
5′ SlPatB pTEV5
3′ SlPatB pTEV5
5′ MaPatB pCV3 pTEV18	NNGCTCTTCNTTCATGGCGCTCTGGCGGATCCGAGCCA
3′ MaPatB pCV3 pTEV18	NNGCTCTTCNTTATCAACTCGACCGCTGTGCCGGCACCGT
5′ SlAcs pSE34
3′ SlAcs pSE34
Site directed mutagenesis primers
5′ SlPatB E366Q	CAGCGCGACCTGCGTCTCGTCCC
3′ SlPatB E366Q	GGGACGAGACGCAGGTCGCGCTG
5′ MaPatB E269Q	CAGCGCCGCCTGGGCCTCGTCGC
3′ MaPatB E269Q	GCGACGAGGCCCAGGCGGCGCTG
5′ SlAcs S608T	ATGATCTTGCCGGTGCGGGTCTTCGGC
3′ SlAcs S608T	ATGATCTTGCCGGTGCGGGTCTTCGGC
S. lividans primers
EFD70633 5′ EcoRV del	CTATGACATGATTACGAATTCGATATCGCCCAGCTCGTGCTGCTCGGC
EFD70633 R linker XbaI	GTGACGGATGCCGTGGCGCGCTCTAGAGAGCGGATCGGCCGGGAGTGA
EFD70633 F linker XbaI	TCACTCCCGGCCGATCCGCTCTCTAGAGCGCGCCACGGCATCCGTCAC
EFD70633 3′ HindIII del	TGTAAAACGACGGCCAGTGCCAAGCTTGGCCGCACTCCCACTGCTGG
EFD65580 5′ EcoRV del	GGC GAT ATC CGG CAA CGC CAC GTG CGG CTT C
EFD65580 R linker XbaI	TTC TCT AGA CTT GCC GGT CAT GCG GTC GAG
EFD65580 F linker XbaI	GTC TCT AGA CTGCTGCGCGGGCTGGGCTGA
EFD65580 3′ HindIII del	AAGAAGCTTGAACGTCCCTTCGCCGAGACC
EFD68590 5′ EcoRV del	GCTGATATCTCGGACTTCGACGACGTCTTCC
EFD68590 R linker XbaI	GACTCTAGACA GGG CGA GCC GGA CCG GGTC
EFD68590 F linker XbaI	TTGTCTAGAGTTGCGGGGGTTGTTGGCGTAG
EFD68590 3′ HindIII del	CTCAAGCTTCTCGATCACGTCGTCCTCGTGC
EFD71509 5′ EcoRV del	GCGGATATCGCGAAGTACCTGTCCCAGTCGGAG
EFD71509 R linker XbaI	TACTCTAGACATGCGGCAAGGGTACGGAAC
EFD71509 F linker XbaI	AGGTCTAGAGGACGCGGCGACGGCCTGAC
EFD71509 3′ HindIII del	GCCAAGCTTCCGAGCTTCATCAGTGCTGCTC

²Nucleotides shown in bold typeface indicate restriction sites

The resulting plasmids were: pSlAcs1, pSlPatA1, pSlAacS1, pSlPatB1, pMaAcs3, pMaPatB3, pRpPat2, pRpBadA3, pSeAcs65, and pSePat8. All plasmids directed the synthesis of proteins fused to an N-terminal H₆ tag (with the exception of pRpPat2, which was fused to an N-terminal H₆-MBP tag). All tags were cleavable by recombinant tobacco etch virus (rTEV) protease purified as described (Blommel et al., 2007). Plasmids that directed the synthesis of SlAcs^K610A (pSlAcs2), and MaAcs^K620A (pMaAcs5) variants were generated from plasmids pSlAcs1 and pMaAcs3. The resulting plasmids (i.e., pSlAcs50, pSlPatB4, pSlPatA57, pMaAcs1 and pMaPatB) were used for complementation in Salmonella. For purification from Streptomyces, pSlAcs31 or pSlAcs52 directed the synthesis of a protein fused to an N-terminal H₆-tag from plasmid pSE34. For identification of plasmid backbones and antibiotic markers for the above-mentioned constructs, refer to Table 1.

Construction of gene deletion in Salmonella enterica

An in-frame deletion of acs in Salmonella enterica was constructed using the phage lambda Red recombinase system as described elsewhere (Datsenko & Wanner, 2000). Plasmids were transformed into strains as described (VanDrisse et al., 2017). In cells that carried two plasmids, pCV1 plasmids were transformed into competent S. enterica cells and selected for on ampicillin. Competent cells harboring pCV1 plasmids were then transformed with pCV3 plasmids, whose inheritance was selected by chloramphenicol resistance. Transformants were streaked onto LB + ampicillin and chloramphenicol to select for cells containing both plasmids.

Construction of gene deletions in Streptomyces lividans

In-frame deletions of ΔldaA, ΔsrtA, and ΔcobB were generated as described (Martinez et al., 2004). DNA fragments of 1.5-kb upstream and downstream of the gene of interest were amplified from S. lividans TK24 purified genomic DNA. The resulting fragments were cloned into plasmid pKC119 (Bierman et al., 1992) using the In-fusion ND Cloning Kit (Clontech) and transformed into E. coli Stellar competent cells (Clontech). Resulting plasmids were transformed into the E. coli helper strain HB101 harboring pRK2013 (Figurski & Helinski, 1979). Plasmids were conjugated into S. lividans and plated on mannitol soya (MS) agar (Kieser et al., 2000c). After 20 h of growth, plates were flooded with apramycin (50 μg ml⁻¹ final concentration for 25 ml of plate in 3 mL of soft nutrient agar) to select for the acquisition of deletion construct plasmids. Apramycin-resistant strains were inoculated into 25 ml YEME + apramycin grown in 500 mL baffled flasks at 30°C for 4 days. Cells were pelleted, re-suspended and plated on ISP-2 medium + apramycin at 42°C for 3 d to select for cells that had integrated the plasmids into the chromosome. Apramycin-resistant cells were grown in 25 mL of YEME medium in 500 mL baffled flasks at 30°C for 4 d and plated on ISP-2 at 30°C to promote loss of plasmid. Isolated colonies were screened on ISP-2 and ISP-2 medium containing with apramycin to screen. Apramycin-sensitive strains were screened by PCR for deletion of ldaA, cobB, or srtA.

Protein purification from Escherichia coli or Salmonella enterica

Proteins purified from cells expressing pSlAcs1, pSlAcs2 pSlPatA1, pSlAacS1, pRpBadA3, were obtained from laboratory stocks. Plasmids pSlACT-Pat1, pMaAcs3, pMaAcs5, pMaPatB3, pRpPat2, pSePat8, pSeAcs65 were transformed into E. coli C41λ(DE3) pat::kan⁺. The resulting strains were grown overnight in 50-mL cultures in medium containing ampicillin (100 μg ml⁻¹), and were sub-cultured 1:100 into 2 L of lysogeny broth containing ampicillin (100 μg ml⁻¹). Cultures were grown shaking at 30°C to an optical density at 600 nm (OD₆₀₀) ~0.6, and protein synthesis was induced by the addition of IPTG (0.5 mM). Upon induction, cultures were grown at 25°C overnight and harvested at 6000 X g for 15 min at 4°C in an Avanti J-2 XPI centrifuge with rotor JLA-8.1000 (Beckman Coulter). Cell pellets were stored at −80°C until used. Pellets were thawed and re-suspended in 50 mL of buffer A [(4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer [(50 mM, pH 7.0 at 4°C) containing NaCl (500 mM], imidazole (20 mM), lysozyme (1 mg/mL), DNase (1μg/mL), and protease inhibitor phenylmethanesulfonyl fluoride (PMSF, 0.5 mM)]. Cells were lysed on ice by two rounds of sonication (1 min (2s, 50% duty, ea.) using a 550 Sonic Dismembrator (Fischer Scientific) at setting 5. Clarified cell lysates were obtained after centrifugation for 30 min at 4°C at 40,000 X g in a Beckman Coulter Avanti J-25I with JA-25.50 rotor followed by filtration of the supernatant through a 0.45 μm filter (Millipore). Samples were applied at 4°C to a pre-equilibrated 1-ml HisPur nickel-nitrilotriacetic acid (Ni-NTA) resin (Thermo Scientific). The column was washed with 10 column volumes (CV) of buffer A, followed by 7 CV of buffer B (HEPES (50 mM, pH 7.0 at 4°C] containing NaCl (500 mM], and imidazole (60 mM). Proteins were eluted with 5 CV of buffer C (HEPES (50 mM, pH 7.0 at 4°C] containing NaCl (500 mM], and imidazole [500 mM]). Proteins were dialyzed at 4°C in decreasing concentrations of NaCl (400 mM, 200 mM, and 150 mM) with a final buffer composition of HEPES (50 mM, pH 7.0 at 4°C), NaCl (150 mM), and glycerol (20% v/v). Proteins concentration was quantified using a NanoDrop with molecular weights and extinction coefficients of each protein as calculated by inputting protein sequences into ExPASy (ProtParam). Proteins were drop-frozen in liquid nitrogen and stored at −80°C.

Purification of SlAcs from S. enterica cells was carried out in the same manner, but the column volume of the nickel resin was 100 μL and 2 L of cells were grown in NCE minimal medium supplemented with acetate (10 mM) as carbon and energy source and L-(+)-arabinose (100 μM) as inducer. Purification of SlAcs from S. enterica was performed in biological triplicates and the experiment was carried out twice.

In vitro protein acetylation assays

Protein acetylation assays were performed as described (Starai & Escalante-Semerena, 2004b). Briefly, reaction mixtures contained HEPES buffer (50 mM, pH 7.0 at 25°C), tris(2-carboxyethyl)phosphine hydrochloride (TCEP; 1 mM), [1-¹⁴C]-acetyl-CoA (20 μM), protein substrate (3 μM), and acetyltransferase (1 μM). Reactions (25 μL total volume) were incubated at 37°C for 60 min. Reactions were stopped by the addition of 5 μL of 6× loading dye (60% (v/v) glycerol, Tris-HCl pH 6.8 (0.3 M), EDTA (12 mM), 12% SDS (w/v), 2-mercaptoethanol (0.87 mM), bromophenol blue (0.05%, w/v)) and heated at 100 °C for 10 min. Samples (3 μL) were resolved using SDS-PAGE (Laemmli, 1970), and visualized by Coomassie Blue R staining (Sasse, 1991). Gels were dried, exposed overnight to a MultiPurpose phosphor screen (Packard), and imaged using a Typhoon (GE Healthcare). In vitro protein acetylation assays using [1-¹⁴C]-acetyl-CoA were performed in triplicate.

In vitro acyl-CoA synthetase assays

Acyl-CoA synthetases (concentrations indicated in figure legends) were incubated with acetyltransferase (concentration indicated in figure legends) with or without acetyl-CoA (60 μM) for 1 h at 37°C using reaction conditions described above. Activity of acyl-CoA synthetases was quantified using an NADH consumption coupled assay (Reger et al., 2007). Reactions contained HEPES buffer (50 mM, pH 7.0 at 25 °C), TCEP (1 mM), ATP (2.5 mM), coenzyme A (CoASH, 0.5 mM), MgCl₂ (5 mM), phosphoenol pyruvate (3 mM), NADH (0.1 mM), pyruvate kinase (1 U), myokinase (5 U), lactate dehydrogenase (1.5 U), and organic acid substrate (0.2 mM). All reactions were started by addition of 2 μL of above acetylation reactions (final concentration of acyl-CoA synthetase, 60 nM). Changes in NADH absorbance was monitored at 340 nm for 8 min in a 96-well plate format using a Spectramax Plus UV-visible spectrophotometer (Molecular Devices). Enzyme activities were calculated as described (Garrity et al., 2007). Data presented here were obtained from technical triplicates of one biological replicate, which was repeated in triplicate.

Western blot analysis of His₆-SlAcs purified from S. enterica

Plasmids directing the synthesis of His₆-SlAcs were transformed into a metE205 ara-9 Δacs Δpat strain (JE11993). Total cell lysates were prepared from 100 ml of S. enterica cells grown on minimal medium supplemented with acetate (10 mM), ampicillin, and arabinose (0.1 mM). Cells were pelleted at 6,000 × g for 10 min and re-suspended in 10 ml of buffer A (above) with 1 mg ml⁻¹ of lysozyme and 1 μg ml⁻¹ DNase. Total cell lysates were quantified using a Bradford assay kit (BioRad). From each total cell lysate 7 μg of protein was loaded onto an SDS-PAGE gel. SlAcs purified from E. coli was acetylated in vitro with MaPatB as described above and loaded on the same SDS-PAGE gel. Three gels were loaded, one for Coomasie Blue R staining and two for technical replicates for transfer. Proteins were resolved by SDS-PAGE at 220 V for 45 min and subsequently transferred to a polyvinylidene fluoride (PVDF) membrane (Millipore) using a Trans-Blot Turbo System (BioRad) using BioRad StandardSD mini gel setting (1.0 A, 25V, 30 min) and transfer buffer (Tris•HCl (25 mM, pH8), glycine (192 mM) and methanol (10% v/v). Membranes were incubated with blocking buffer consisting of condensed milk (12.5% w/v) in 25 ml of PBST (10 mM NaH₂PO₄ (pH 7.2 at 25 °C), NaCl (0.9 % w/v), and Tween-20 (0.1% (v/v). Membranes were probed with polyclonal rabbit α-acetylated lysine antibodies (1:866 dilution into blocking buffer) (Calbiochem) overnight at 4 °C. Membranes were washed thrice with 20 ml of phosphate buffered saline with tween 20 (PBST) and goat α-rabbit immunoglobuilin G conjugated to calf intestinal alkaline phosphatase (Pierce) were diluted (1:5,000) into blocking buffer and applied to membrane for 1 h at 25 °C to detect SlAcs^Ac. Membranes were washed thrice with 20 mL PBST and signal was detected using nitro-blue tetrazolium chloride [~19 mg/1 ml of 67% dimethylsulfoxide (DMSO)] and 5-bromo-4-chloro-3′-indolylphosphate p-toluidine salt (~9 mg/1 mL of 67% DMSO) (NBT-BCIP).

Protein purification from Streptomyces lividans

Plasmids pSlAcs31 (encoding His₆-SlAcs) or pSlAcs52 (encoding His₆-SlAcs^S608T) were introduced into S. lividans strains (wild-type, ΔldaA, ΔldaA ΔcobB, and ΔldaA ΔcobB ΔsrtA) by polyethyleneglycol (PEG)-assisted protoplast transformation as described (Kieser et al., 2000c, Kieser et al., 2000b). Transformed cells were plated on R2YE medium and grown at 30 °C for 20 h. Plates were then flooded with thiostrepton (10 μg ml⁻¹, final concentration for 25 ml plate volume in 3 ml soft agar nutrient broth) and incubated for 1-2 d at 30 °C to select for strains harboring plasmids. Cells containing pSlAcs31 or pSlAcs52 were grown in 25 mL YEME medium + thiostrepton in 500-ml baffled flasks for 4-5 d at 30 °C. Cells were harvested by diluting 1:2 into water and centrifuged at 2000 × g for 10 min. Cell pellets were re-suspended in 1-2 ml of YEME medium and 300 μl were plated on multiple MS agar plates. Spores were harvested as described (Kieser et al., 2000a) and counted by serial dilutions plated on ISP-2 medium + thiostrepton. Spores (1 × 10⁹/L) were inoculated into 100 mL YEME medium + thiostrepton and grown for 2-3 days at 30 °C. Cells were harvested by diluting cultures 1:2 into water and centrifuging at 2000 × g for 15 min. Cells were re-suspended in 30 ml of buffer A (above) with 1 mg ml⁻¹ of lysozyme, 1 μg ml⁻¹ DNase, and protease cocktail inhibitor (100 μL, Sigma). Cells were lysed by sonication on ice (46 s total with 2 s on, 2 s off) and cell debris was removed by centrifugation at 45,000 × g for 30 min. His₆-SlAcs or His₆-SlAcs^S608T protein was purified as described above using a 200 μl of a Ni-NTA resin (Pierce). SlAcs-containing fractions were combined and dialyzed at 4 °C for 3 h into HEPES (50 mM, pH 7 at 4°C), NaCl (250 mM), followed by a final dialysis in HEPES (50 mM, pH 7 at 4°C), NaCl (150 mM), glycerol (20%, v/v). Proteins were drop frozen into liquid-N₂ until use. His₆-SlAcs or His₆-SlAcs^S608T activity was assessed using a coupled spectrophotometry assay as described above. The above procedure was performed in triplicate with error bars of activity assays representing technical triplicates of a biological replicate.