A Toxin Involved in Salmonella Persistence Regulates Its Activity by Acetylating Its Cognate Antitoxin, a Modification Reversed by CobB Sirtuin Deacetylase

ABSTRACT Bacterial toxin-antitoxin systems trigger the onset of a persister state by inhibiting essential cellular processes. The TacT toxin of Salmonella enterica is known to induce a persister state in macrophages through the acetylation of aminoacyl-tRNAs. Here, we show that the TacT toxin and the TacA antitoxin work as a complex that modulates TacT activity via the acetylation state of TacA. TacT acetylates TacA at residue K44, a modification that is removed by the NAD+-dependent CobB sirtuin deacetylase. TacA acetylation increases the activity of TacT, downregulating protein synthesis. TacA acetylation altered binding to its own promoter, although this did not change tacAT expression levels. These claims are supported by results from in vitro protein synthesis experiments used to monitor TacT activity, in vivo growth analyses, electrophoretic mobility shift assays, and quantitative reverse transcription-PCR (RT-qPCR) analysis. TacT is the first example of a Gcn5-related N-acetyltransferase that modifies nonprotein and protein substrates.

A cetylation of proteins and small molecules is a conserved mechanism that regulates cellular processes in cells from all domains of life. Members of the Gcn5related N-acetyltransferase (GNAT) protein superfamily (PF00583) acetylate proteins and non-protein substrates at the expense of acetyl-coenzyme A (Ac-CoA) (1). In some instances, protein acetylation is reversed by class III NAD ϩ -dependent deacetylases, also known as sirtuins (2).
The genome of the human pathogen Salmonella enterica subsp. enterica serovar Typhimurium LT2 (here S. enterica) encodes 26 putative GNATs, three of which are annotated as toxin acetyltransferases. These putative toxins are part of type II toxinantitoxin (TA) systems (3,4), and the genes encoding these proteins appear to comprise separate operons for each toxin-antitoxin pair.
TA systems have different physiological functions, with some of them contributing to plasmid stabilization as addiction modules (5-9), while others serve as survival management systems under different stress conditions (10,11). TA systems have been shown to downregulate essential functions that can trigger the onset of a persister state that allows cells to survive unfavorable conditions without the need to acquire mutations (12)(13)(14)(15)(16)(17). Importantly, the slowing of essential metabolic processes by TA module expression may allow for tolerance to antimicrobials, which can lead to recalcitrant infections (18,19).
Type II TA systems are in most cases a part of two-gene operons that encode a protein toxin and protein antitoxin (20). Typically, the antitoxin neutralizes the toxin until a signal induces selective degradation of the antitoxin, often by stress-induced proteases (16,(21)(22)(23)(24). Antitoxin degradation creates a stoichiometric imbalance that releases the toxin, thereby increasing its activity and upregulating its transcription (25,26). With some exceptions, antitoxins possess a DNA-binding domain that recognizes the operator site of its own promoter, resulting in modulation of its own synthesis (11,27,28). The S. enterica genome encodes three TA systems that include toxins that are homologous to Gcn5-related N-acetyltransferases (GNATs). All three of these systems contribute to the onset of a S. enterica persister state inside macrophages (29,30). In addition, Helaine and coworkers showed that the TacT protein of a TA type II system comprised of proteins TacT (STM3651) and TacA (STM3652) acetylates the aminoacyl moiety of several charged tRNAs, arresting translation and triggering a persister state (31). Cheverton et al. suggested that exiting TacT-induced persistence was due to replenishment of tRNA pools through the hydrolysis of the acetylated amino group off of the aminoacyl-tRNA by the peptidyl-tRNA hydrolase Pth (31).
Here we show that, in addition to acetylating aminoacyl-tRNAs, TacT acetylates its cognate TacA antitoxin with a concomitant increase in TacT activity without complex dissociation, a unique attribute not seen with other TA modules. We suggest that the acetylation state of the TacA antitoxin, not its degradation, rapidly modulates TacT activity. We also show that the NAD ϩ -dependent CobB sirtuin deacetylase reverses the effect that acetylated TacA has on TacT activity and suggest that reversible TacA acetylation plays a key role in exiting the persister state. In addition, we present evidence that the TacA antitoxin binds to the tacA-tacT promoter when in complex with TacT. The involvement of sirtuin-dependent reversible lysine acetylation (sRLA) frames the persister state of S. enterica within the carbon and energy statuses of the cell.

RESULTS
The absence of TacA antitoxin extends the lag time before the onset of exponential growth in cells with higher levels of tacT ؉ expression. Consistent with published data (31), ectopic expression of tacT ϩ in S. enterica tacA::cat ϩ or tacAT::cat ϩ mutant strains delayed the onset of exponential growth when cells were grown on minimal medium ( Fig. 1) due to increased TacT-mediated aminoacyl-tRNA acetylation. This delay was shortened when cells were grown in rich medium, suggesting that nutrient availability played a role in the observed phenotype. Notably, the extended lag phase did not have an effect on the final cell density or growth rate of the cultures (Fig. 1). The phenotype of tacA::cat ϩ strains in which tacT ϩ was overexpressed (i.e., tacA::cat ϩ /pTacT WT or tacAT::cat ϩ /pTacT WT ) was corrected by in trans expression of tacA ϩ (Fig. 1), indicating that the observed phenotype was dependent on the absence of the TacA antitoxin.
TacT acetylates residue K44 of TacA. As shown by others, TacA copurified with TacT, forming a complex. TacA and TacT proteins were synthesized from a vector containing the coding sequences for both proteins. Overexpression of both genes allowed for simultaneous synthesis and folding of each protein, leading to a stable TacAT complex. All toxin-antitoxin complexes (wild-type [WT] and variant proteins) reported herein were copurified as described in Materials and Methods. Our assessment of the purity of the proteins used in this study can be found in Fig. S1 in the supplemental material. Results of gel permeation chromatography experiments revealed the TacAT complex was a dimer of dimers (Fig. S2). When purified TacA and TacT proteins were incubated with [1-14 C]Ac-CoA, TacA was acetylated ( Fig. 2A, lane 2), a modification that required TacT (Fig. 2A, lane 2 versus 3). Liquid chromatography-tandem mass spectrometry (LC-MS/MS) peptide fingerprinting analysis of acetylated TacA (TacA Ac ) showed that residues K12, K44, and K83 were acetylated to various degrees (Fig. S3). Figure S3 shows peptide masses with different m/z ratios used search the database using MASCOT software. Analyses of areas under the peaks for peptides corresponding to acetylated K12, K44, and K83 showed a 2.64-fold increase for K12, 18.86-fold increase for K44, and a 3.12-fold increase for K83. 25

Correction of the translation initiation codon of TacA.
In the Salmonella genome, tacAT comprise an operon with tacA being promoter proximal, with the last 13 bp of tacA overlapping the tacT coding sequence. The 5= region of tacA contains the annotated start methionine and an additional methionine at position 8. This information was relevant, because the above-mentioned peptide fingerprinting analysis did not detect the first 7 amino acids (M1 to L7 [gray residues in Fig. S3]) of the annotated primary sequence of TacA; therefore, we considered the possibility that the true start methionine of TacA was residue M8. This misannotation became relevant during the purification of TacAT complexes, because the nucleotides encoding the first 7 amino acids and the S-tag (streptavidin tag) fused to TacA were not translated, resulting in coelution of tagless native TacA with H 6 -TacT. The availability of tagless TacA became useful for the performance of DNA binding experiments, because the DNA-binding domain of TacA was on its N terminus. For the experiments mentioned above and from here on, the translation start codon of TacA was reassigned to the ATG encoding M8. That is, residue M8 became M1, and the numbering for the rest of the residues was modified accordingly. Consequently, residues K19, K51, and K90 became K12, K44, and K83, respectively. The adjusted numbering was used throughout these studies.
Residue K44 is a key acetylation site in TacA, a modification reversed by the NAD ؉ -dependent CobB sirtuin deacetylase. To validate the putative acetylation sites, TacA variants with substitutions at positions K12, K44, and K83 were isolated. As shown in Fig. 2A, transfer of the acetyl moiety of [ 14 C]Ac-CoA was not observed when TacT was in complex with TacA K44A , indicating that either K44 was the only acetylation site in TacA that was modified by TacT or K44 acetylation triggered K12 and K83 acetylation by TacT. The latter scenario was not pursued. Instead we focused on the question of whether or not K44 acetylation was reversible. In Salmonella, the only known protein lysine deacetylase is the NAD ϩ -dependent CobB sirtuin. This result raised the question of whether CobB could deacetylate TacA Ac . To test this idea TacA [14C]Ac was incubated with CobB, NAD ϩ , CobB plus NAD ϩ , or CobB plus NAD ϩ plus nicotinamide (Nm). Under the conditions tested, CobB and NAD ϩ deacetylated~70% of TacA Ac (Fig. 2B, lane 6 versus lanes 2, 3, 4, and 5). As expected, CobB activity was inhibited by Nm (Fig. 2B, lane 5 versus 6), a result consistent with CobB-dependent deacetylation.
TacA acetylation enhances TacT activity in vitro. TacA variants were constructed to determine whether or not acetylation of TacA residue K44 had an effect on TacT-dependent arrest of mRNA translation. TacA variants with substitutions at position K44 that mimicked acetylation (i.e., K44Q) or deacetylation (i.e., K44R) (2) were isolated in complex with wild-type TacT (TacT WT ). No effect on the formation or stability of complexes between TacT WT and TacA variants was detected using gel filtration chromatography (Fig. S2). TacT was purified from homogeneous TacAT complex by denaturation and refolding, as described in Materials and Methods.
In vitro protein synthesis experiments were performed with TacAT WT , TacA K44Q T WT , and TacA K44R T WT complexes or TacT WT alone. Proteins were incubated with or without Ac-CoA in a cell-free protein synthesis system that contained all machinery necessary for transcription and translation from a synthesized DNA product. DNA that is added to the reaction mixtures codes for dihydrofolate reductase (DHFR), and its synthesis is used as a reporter of mRNA translation. DHFR synthesis was monitored using SDS-PAGE (32) (~18 kDa [arrow in Fig. 3]). TacT WT toxin was expected to acetylate aminoacyl-tRNAs, resulting in DHFR protein synthesis inhibition (31). To quantify differences in DHFR synthesis in reaction mixtures containing or lacking Ac-CoA, the intensities of three bands in each lane (asterisks in Fig. 3) were expressed as percentages of the band intensity of DHFR. For example, the level of DHFR synthesis was compared in reaction mixtures containing TacAT WT with or without Ac-CoA (Fig. 3, lanes 4 and 5). Control experiments for these studies included conditions for maximal DHFR synthesis (Fig. 3, lane 2), with the baseline for the production of DHFR being established by a reaction mixture devoid of DNA. As expected, no DHFR was synthesized (Ͻ1%) in the absence of added DHFR-encoding DNA (Fig. 3, lane 2 versus 3). Two reaction mixtures lacking TacA WT , with or without Ac-CoA added, showed TacT WT -and Ac-CoA-dependent arrest of DHFR synthesis (Fig. 3, lanes 10 versus 11), indicating that the TacT WT protein was active.
DHFR synthesis was reduced by~30% in reaction mixtures containing TacAT WT in the presence of Ac-CoA (Fig. 3, lane 4 versus 5). This result was unexpected because, to our knowledge, this is the first report of a toxin suggested to be active in complex with its cognate antitoxin. Reaction mixtures containing TacA K44R T WT (deacetylation mimic variant) and acetyl-CoA showed a 40% reduction in DHFR synthesis (Fig. 3, lane 6 versus 7). Although it appears there are differences in aminoacyl-tRNA acetylation by the TacAT WT and TacA K44R T WT proteins, these differences are not statistically significant (see percentages as standard deviations in Fig. 3). In sharp contrast, DHFR synthesis was reduced by 92% in reaction mixtures containing the acetylation mimic variant TacA K44Q T WT complex, and Ac-CoA (Fig. 3, lane 8 versus 9), suggesting that the TacA K44Q variant enhanced the acetylation of aminoacyl-tRNA by TacT WT (Fig. 3, lane 8 versus 9). These results strongly suggested that TacA WT acetylation by TacT WT upregulated TacT WT -dependent acetylation of aminoacyl-tRNAs. A summary of the above results is presented as percentages underneath the SDS-PAGE gel (Fig. 3).
TacA acetylation enhances TacT activity in vivo. To validate in vitro results in vivo, tacA alleles encoding TacA variants with single-amino-acid substitutions at position K44 were introduced by site-directed mutagenesis into a plasmid carrying tacAT ϩ (pTacAT-1). One TacA variant within the operon had a K44Q substitution (TacA K44Q ) for the purpose of mimicking acetylation (pTacAT-10), another variant had a K44R substitution (TacA K44R ) to mimic deacetylation (pTacAT-9). The resulting plasmids, which also carried the tacT ϩ allele, were individually introduced into tacAT::cat ϩ strains. As shown in Fig. 4, tacAT::cat ϩ strains that synthesized TacA K44Q T WT complex had a striking growth phe- . Lane 2 represents a positive control for DHFR production, where the reaction mixture contained DHFR DNA (100 ng) and acetyl-CoA to ensure acetyl-CoA stocks did not interfere with reaction components. The reaction mixture in lane 3 lacked DNA and contained TacAT WT complex to control for any added bands that might interfere with interpretations. Lanes 4 to 9, reaction mixtures that contained TacAT WT or variant complexes (indicated by labels above each lane) with or without acetyl-CoA (with DNA added to lanes 4 to 11). Lanes 10 and 11, positive control for aminoacyl-tRNA acetylation. These reaction mixtures contained TacT with or without acetyl-CoA. All reaction mixtures were incubated for 2 h at 37°C. Details of the procedure can be found in Materials and Methods. Asterisks represent bands used to normalize the intensity of the DHFR protein within each lane (ImageQuant v5.2 software). Comparison of DHFR band intensity to bands with asterisks was averaged, and the standard deviation of these three numbers was calculated. The percentage of decrease with SD for each sample is indicated below the corresponding lanes.

RLA Modulates Salmonella Persistence
® notype compared to strains making TacA K44R T WT (Fig. 4, gray circles and black diamonds). This result was consistent with an increase in aminoacyl-tRNA acetylation by TacT WT that slowed down mRNA translation, causing the observed phenotype. We note that the growth arrest of strains making TacA K44Q T WT was more severe than the growth delay of strains expressing tacT ϩ alone and is most likely due to lower induction of plasmids coding for TacA K44Q T WT causing this phenotype (i.e., overexpression of tacT alone at 25 M arabinose is not a high enough induction to cause a growth delay). This is most likely due to stability of the toxin when it is in or not in complex with its antitoxin, and these results are further discussed below.
Acetylation of residue K44 of TacA alters the DNA binding of the protein. Electrophoretic mobility shift assays (EMSAs) were performed to determine whether modifications of K44 would affect the DNA-binding activity of TacA. A 157-bp 6-carboxyfluorescein (6-FAM) 5=-labeled probe (positions Ϫ163 to Ϫ6 [probe 1]) upstream of the corrected ATG transcription initiation codon for tacAT was chosen for probe design. TacAT WT complex bound to this probe, and this binding was specific because TacAT did not bind to the promoter for argS, a negative control (see Fig. S4 in the supplemental material). To narrow down the minimal sequence necessary for TacAT binding, two probes within the 157-bp region upstream of the putative ATG codon for tacAT were designed. One probe was 71 bp long (positions Ϫ163 to Ϫ92 [probe 2]), and the other was 75 bp long (positions Ϫ81 to Ϫ6 [probe 3]) (probes 2 and 3 in Fig. S5). There was no shift seen with probe 2, but the electrophoretic mobility of the 75-bp probe (probe 3) changed in the presence of the TacAT WT complex (Fig. S5).
TacA WT or TacT WT alone did not bind to probe 3 (see Fig. S6A in the supplemental material). This result suggested that TacA needed TacT to bind to DNA. An electrophoretic mobility shift assay (EMSA) was conducted in which TacA and TacT proteins were present to test whether or not they could reform a complex capable of binding to probe 1. No binding was detected at 5-fold excess TacA and 50-fold excess TacT, nor was DNA binding observed when TacA protein was incubated with probe 1 at 50-fold excess (Fig. S6B). We concluded that refolded TacT was active based on the results presented in Fig. 3. However, we cannot rule out the possibility that TacA did not refold correctly, thus preventing DNA binding. It is also possible that the complex could not reform after TacA and TacT were separated.
We observed two different probe shifts (Fig. 5A, labeled with one-and two-site binding), which were not due to a mixed population of complex (i.e., acetylated versus  non-acetylated) since a TacA K44R T WT complex (i.e., not the acetylatable variant, but a positively charged substitution of K44) produced the same two shifts (Fig. 5B). To ascertain whether the observed two bands reflected the presence of two binding sites within probe 3, the probe was split into two pieces: probes 4 and 5 (sequence representation in Fig. S5). Probe 4 (Ϫ81 to Ϫ46) or probe 5 (Ϫ46 to Ϫ6) was mixed with TacAT WT complex, and interactions were assessed by EMSAs (Fig. S5). The two bands observed when probes 1 and 3 were used ( Fig. 5; Fig. S5) were not observed when probes 4 and 5 were used (Fig. S5), indicating that only one binding site was present in each probe. Work on the MqsRA TA system in Escherichia coli showed that two band shifts occur when two binding sites are present (46). Such a scenario is consistent with our observations (Fig. 5A) in that the lower band assigned to TA complex bound to either site and the upper band reflecting interactions of complexes bound to two sites (Fig. 5A, cartoon representation).
In contrast to the TacA K44R T WT complex, the TacA K44Q T WT complex (i.e., mimicking acetylated K44) did not display the second band, in support of the idea that TacA K44Q T WT binds to both sites (Fig. 5). These data suggested that residue K44 did not directly affect the DNA-binding activity of TacA, but that acetylation altered TacA recognition of or binding to the tacAT promoter. It was further validated that K44 was not directly involved in DNA binding because the TacA K12A T WT complex had DNA binding activity that was abrogated (Fig. S7), suggesting the N terminus included the DNA-binding domain of TacA. In contrast to this finding, the binding levels of TacA K83A T WT and TacAT WT complexes to probe 3 were very similar, suggesting that residue K83 and therefore the C terminus was not involved in TacA binding to DNA. A predicted structural representation of these residues can be seen in Fig. S8.
To determine whether the altered binding to DNA by TacA variants in complex with TacT WT had an effect on repression of tacAT transcription, chromosomal mutations coding for K44R or K44Q TacA variants were constructed as described in Materials and Methods. Cells coding for TacA WT T WT , TacA K44R T WT , or TacA K44Q T WT or cells lacking TacT RLA Modulates Salmonella Persistence ® were grown to mid-log phase on minimal medium supplemented with 22 mM glycerol, and total RNA was extracted as described previously (33). We performed RT-qPCR with total RNA to measure the differences in level of mRNA transcript of the tacA gene in the mutants compared to the tacAT ϩ strain. We observed a large increase in tacA transcript (28.6-fold) in the strain in which the tacT gene was deleted and only tacA remained (Fig. 6). This suggested that the native TacAT complex repressed the operon under this condition and that the presence of TacT was needed for TacA-mediated repression. When measuring tacA transcript in the chromosomal variants, we did not detect significant transcript differences between the TacA K44R T WT or TacA K44Q T WT strain and the parent strain. These results suggested either that the altered DNA binding activity of TacA K44Q T variants seen in vitro might be an artifact of the assay or that an additional factor for tacAT expression might be needed and may only be induced under different conditions.

DISCUSSION
Here we report insights into the functionality of one of the three type II toxinantitoxin systems encoded by the genome of Salmonella enterica, a human pathogen. The S. enterica TacAT system (encoded by tacA and tacT [formerly STM3651 and STM3652, respectively]) is the focus of the studies reported herein. The TacAT system is the first example of a toxin-antitoxin system whose function is not regulated by dynamic association/dissociation of its components and the first example of a system in which the toxin is a GNAT that recognizes a protein and non-protein substrate. In the case of TacAT, TacT acetylates the ␣-amino group of aminoacyl-tRNAs (29) and also acetylates a lysine residue of TacA (this work). Furthermore, we have shown that sirtuin-dependent reversible lysine acetylation (sRLA) deacetylates TacA Ac and may play a role in controlling the activities of TacA and TacT in vivo.
A new role for sRLA in Salmonella pathogenesis. We hypothesize that the S. enterica TacAT system is different from other type II toxin-antitoxin systems in that the activity of TacT is not upregulated as a result of the cleavage or cellular degradation of the cognate antitoxin (i.e., TacA) (22)(23)(24). Our data support the conclusion that TacAT activities are modulated posttranslationally through sRLA ( Fig. 3 and 4). The involvement of sRLA in the modulation of the entrance or egress into or out of a persister state suggests a metabolic link between the latter and the carbon (Ac-CoA) and energy (NAD ϩ ) statuses of the cell (see below).  Although it is clear that acetylation of TacA alters binding of TacA to the two sites present inside the tacAT promoter in vitro (Fig. 5), more work is needed to understand why and how TacA Ac recognition of two sites is favored over recognition of either one of the two sites and whether or not there are additional factors involved in tacAT expression.
Other unique features of the S. enterica TacAT system. As stated above, a unique feature of TacT is that it can acetylate non-protein and protein substrates. Also unique is the fact that TacT activity is enhanced as a result of the acetylation of TacA by TacT while the proteins are in complex. At present, there is no indication that TacA and TacT ever dissociate from each other. In fact, the aminoacyl-tRNA-acetylating activity of TacT is enhanced when TacT is in a complex with TacA K44Q (an acetylation mimic) (Fig. 3). In contrast, TacT activity decreases in complexes containing TacA or TacA K44R (a deacetylated mimic).
In vivo evidence supports the idea that TacA acetylation enhances TacT activity. Results from in vivo experiments are consistent with in vitro results. For example, the phenotype associated with the synthesis of TacA K44Q T WT occurred even at low levels of induction (25 M arabinose), whereas the phenotype generated by a high level of expression of TacT alone was erratic, and when it was observed, it occurred only at high levels of induction (150 to 1,000 M arabinose). These differences could be due to an increased turnover rate or to poor solubility of TacT in the absence of TacA. Therefore, when TacT is in complex with TacA K44Q , TacT is not only stabilized, but its activity is increased, leading to growth arrest (Fig. 4). This result suggests that TacA Ac T WT blocks protein synthesis very efficiently. We posit that interaction of Salmonella with the macrophage leads to increased Ac-CoA levels as a result of the inhibition of Ac-CoA-consuming processes. Such an increase in Ac-CoA could trigger TacA acetylation by TacT, increased stability of the complex, and increased TacT activity, ultimately resulting in the arrest of protein synthesis. These ideas await further experimentation.
Additional perspective. One central, unanswered question is how S. enterica monitors and modulates the levels of TacA WT T WT and TacA Ac T WT complexes. Our data suggest that the NAD ϩ -dependent CobB protein deacetylase prevents the accumulation of TacA Ac T WT complex that would drive the cell into a prolonged persister state. Given that CobB can deactylate TacA Ac and that NAD ϩ is needed for CobB activity, a plausible answer to the above question may lie on the intracellular NAD ϩ level. A decrease in the Ac-CoA level could be due to its consumption by anabolic processes, which would require a robust energy charge and would correlate with increased NAD ϩ levels in the cell. NAD ϩ would then be used by CobB to reduce TacT activity as a result of TacA Ac deacetylation.
Importance. The TacAT toxin-antitoxin system of S. enterica appears to have evolved to use sirtuin-dependent reversible lysine acetylation (sRLA) as a mechanism to rapidly enter and leave a persister state. There are several methods of other type II TA systems that use transcriptional regulation to either replenish pools of antitoxin to neutralize toxin activity, as with HicAB (34), or more commonly use the toxin as a corepressor until a stoichiometric imbalance occurs to derepress the operon (known as conditional cooperativity) (35,36). With the TacAT system, the use of sRLA could quickly and efficiently modulate toxin and antitoxin activities without the need for threshold derepression of transcription. By maintaining a stable TacAT complex, the cell may enter or egress the persister state as a function of Ac-CoA, NAD ϩ , and probably other as yet unknown signals. Additional evidence for the involvement of sirtuins in the control of TacAT activities inside the macrophage is needed to advance our understanding of the persister state and how S. enterica gets into and out of it. Regardless, the new knowledge reported here suggests a new use of sirtuin activators to help reduce the probability of S. enterica maintaining a persister state inside a host.

MATERIALS AND METHODS
Bacterial strains. All strains constructed were derivatives of Salmonella enterica subsp. enterica serovar Typhimurium LT2 (here S. enterica) using the Wanner in-frame gene deletion method (37). Bacteria were grown shaking at 37°C, and the media used for growth are described under "Culture media and chemicals." Culture media and chemicals. All of the bacterial strains and plasmids used are listed in Table S1 in the supplemental material. Strains were grown in lysogenic broth (LB [Difco]), nutrient broth (NB [Difco]), or no-carbon essential (NCE) minimal medium (38). Growth studies in which glycerol was the sole carbon and energy source were performed in NCE medium supplemented with MgSO 4 (1 mM), Wolfe's trace minerals (1ϫ) (39), and glycerol (22 mM). When used, antibiotics were added at the following concentrations: ampicillin, 100 g/ml; chloramphenicol, 20 g/ml; and kanamycin, 50 g/ml. Cultures used as an inoculum were grown overnight at 37°C in NB, and small samples (1% [vol/vol]) were used to inoculate 198 l of fresh medium placed in each well of a 96-well microtiter plate. L-(ϩ)-Arabinose was used as an inducer wherever indicated. Microtiter plates were incubated at 37°C inside the temperaturecontrolled chamber of a PowerWave microtiter plate reader (Bio-Tek Instruments), and plates were continuously shaken using the medium setting of the instrument. Cell density was monitored at 630 nm, and data were analyzed using Prism 6 software package (GraphPad).
Strain construction. All strains constructed were ara-9 derivatives of S. enterica using the Wanner in-frame gene deletion method (37). The primers used in this study were synthesized by Integrated DNA Technologies, Inc. (IDT, Coralville, IA), and are listed in Table S2 in the supplemental material. tacA::cat ϩ and ΔtacA strains were engineered as follows. Using Pfu Ultra II Fusion DNA polymerase (Stratagene), flanking regions of plasmid pKD3 (37) were amplified with primers designed with 36 to 39 bp of overlapping region at the beginning of either the tacA or tacT gene and with 50 bp of overlapping region at the end of the tacA or tacT gene. PCR fragments were analyzed by agarose gel electrophoresis on 1% (wt/vol) agarose gels poststained with 0.5 g/ml ethidium bromide for 15 min. PCR fragments were PCR cleaned up using the Wizard SV gel and PCR cleanup system (Promega), and 5 to 10 l of product was electroporated into S. enterica strain JE10813 Δara-9 harboring plasmid pKD46 (37). After electroporation, cells were grown up to an optical density at 630 nm (OD 630 ) of~0.6 at 30°C, followed by three washes with glycerol (10% [vol/vol]). Electroporation was performed in 0.2-cm electroporation cuvettes (MidSci) in a Bio-Rad MicroPulser electroporator on the Ec2 setting. Cells were then recovered in 0.5 ml of LB for 1 h at 37°C, plated on LB plus agar plus antibiotic, and incubated overnight at 37°C. Drug-resistant transformants were streaked repeatedly on antibiotic plates at 42°C to cure the strains of plasmid pKD46. Strains were then reconstructed by P22-mediated transduction of the drug marker into strain JE10079. Strains containing chloramphenicol insertions were transformed with plasmid pCP20 to resolve out the chloramphenicol insertion and make a scarred deletion.
Plasmid construction for complementation and overexpression. All plasmids used in this work are listed in Table S1. Primers used in this study were synthesized by Integrated DNA Technologies, Inc. (IDT [Coralville, IA]), and are listed in Table S2. We used the high-efficiency cloning method described elsewhere (40) to clone the tacA, tacT, tacAT, and cobB genes into pCV1 and pCV3 vectors. Plasmid pCV1 is a modified plasmid of pBAD24 (41) with BspQI sites added and confers ampicillin resistance, and expression of genes cloned into it can be induced with L-(ϩ)-arabinose. Plasmid pCV3 is a modified plasmid of pBAD33 (41) with added BspQI sites and confers chloramphenicol resistance and arabinose induction. Genes tacA, tacT, tacAT, and cobB were amplified from the S. enterica chromosome using Pfu Ultra II fusion DNA polymerase (Stratagene). PCR fragments were analyzed by agarose gel electrophoresis on 1% (wt/vol) agarose gels stained with ethidium bromide. PCR fragments were PCR cleaned up using the Wizard SV gel and PCR cleanup system (Promega) and digested with the restriction enzyme BspQI (NEB) at 50°C for 1 h; products were ligated with T4 DNA ligase (Fisher).
The overexpression vector pACYCDuet (EMD Millipore Biosciences) possesses two multiple-cloning sites (MCSs) for which two separate genes may be cloned into and overexpressed simultaneously from the same vector. This vector was used for the overexpression of TacT and TacA from the same plasmid. pACYCDuet was first digested with FastDigest (Thermo, Fisher Scientific) BamHI and EcoRI for 1 h at 37°C. tacA and tacT were amplified from the S. enterica chromosome using Pfu Ultra II fusion DNA polymerase (Stratagene). PCR fragments were analyzed by agarose gel electrophoresis on 1% (wt/vol) agarose gels stained with ethidium bromide. PCR fragments were PCR cleaned up using the Wizard SV gel and PCR cleanup system (Promega). The tacT PCR product was digested with FastDigest (Thermo, Fisher Scientific) BamHI and EcoRI for 1 h at 37°C. The digested PCR product and pACYCduet plasmid were PCR cleaned up with the same cleanup kit stated above, and the digested tacT product was ligated into MCS1 of digested pACYCDuet with T4 DNA ligase (Fisher) at room temperature for 30 min. Once transformants were verified, the tacA PCR product was digested with EcoRI and XhoI for 1 h at 37°C, PCR cleaned, and ligated as described above into the second MCS of pACYCDuet, which had the tacT gene cloned into MCS1.
Plasmids were isolated using the Wizard Plus SV miniprep kit (Promega). To confirm the correct sequences were cloned without mutations, DNA sequencing reactions were analyzed at the Georgia Genomics Facility, University of Georgia-Athens. Site-directed mutagenesis (Stratagene) was performed on pTacAT-2 or pTacAT-1 to change the mentioned residues K12, K44, and K83 to A, Q, or R. PCR was performed using Pfu Ultra II DNA polymerase with the primers listed in Table S2. Modifications included an annealing time of 60 s, an extension temperature of 68°C, and an extension time of 2.5 min kb Ϫ1 . DNA changes were confirmed by sequencing.
Cell pellets were resuspended in 50 ml of buffer A containing 4-(2-hydroxymethyl)-1-piperazineethanesulfonic acid (50 mM HEPES [pH 7.0 at 4°C]), NaCl (500 mM), imidazole (20 mM), glycerol (20% [vol/vol]), lysozyme (1 mg/ml), DNase (1 g/ml), and protease inhibitor phenylmethylsulfonyl fluoride (PMSF [1 mM]). Cells were sonicated for 60 s using a Qsonica sonicator at 60% duty with 2-s pulses. Lysates were centrifuged using a Beckman Coulter, Inc., Avanti J-251 centrifuge equipped with a JA-25.50 rotor at 40,000 ϫ g for 30 min. Clarified lysates were filtered through a 0.45-m-pore filter and applied to a 2-ml HisTrap FF column (GE Healthcare Sciences) using an Äkta fast protein liquid chromatography (FPLC) system (GE Healthcare Sciences). The column was washed with 10 column volumes of bind buffer, 7 column volumes of 8% elution buffer (50 mM HEPES [pH 7.0 at 4°C], 500 mM NaCl, 500 mM imidazole, 20% [vol/vol] glycerol), and a 20-column-volume gradient to 100% elution buffer. When separation of TacT and TacA complexes was necessary, proteins were denatured on the column as described previously (31). Fractions were run on an SDS-PAGE gel, and fractions containing the desired protein were . Proteins were flash frozen in liquid N 2 and stored at Ϫ80°C. Proteins were quantified using a NanoDrop 1000 spectrophotometer (Thermo Scientific) using the molecular mass (28.67 kDa) and extinction coefficient (18,700 M Ϫ1 cm Ϫ1 [Expasy ProtParam]) of the complex (assuming a 1:1 ratio of TacT to TacA, as determined via size exclusion chromatography [described below]). Percentage of purity was calculated using ImageQuant v5.2 software. CobB protein was purified as described elsewhere (43). TacT was isolated from TacAT complex using a purification protocol reported elsewhere (31). Briefly, TacAT WT complex was purified as described above and was separated into its components by denaturation with guanidine-HCl (5 M) followed by overnight dialysis. Denatured proteins were resolved by Ni-affinity chromatography and refolded by dialysis of the denaturant. As refolding occurred, the bulk of TacT became insoluble, while TacA remained stable in solution. This procedure yielded 6-fold larger amounts of TacA than TacT (e.g., 0.5 mg of TacA versus 0.08 mg TacT per liter of culture).
Size exclusion chromatography. A Superose 12 10/300 GL gel filtration column (GE Healthcare Life Sciences) was equilibrated as per the manufacturer's protocol using water and elution buffer (50 mM HEPES [pH 7.0 at 4°C], 150 mM NaCl, 20% [vol/vol] glycerol). Samples were applied to the column using a 100-l superloop. Gel filtration standards (Bio-Rad) were applied to the column first, and a standard curve was calculated using the log molecular weight (MW log ) of each standard against the retention time of each protein. Purified TacAT complexes were eluted from the column in the same manner, retention times were recorded, and molecular weights were determined using an equation calculated from the standard curve.
In vitro DHFR protein synthesis assay. The PureExpress in vitro protein synthesis kit (New England Biolabs) reactions were set up per the manufacturer's protocol in RNase-free tubes with the following modifications. Reaction mixtures were adjusted to a total volume of 15 l (5 l of solution A, 3.75 l solution B, 100 ng DHFR DNA) and, when noted, supplemented with acetyl-CoA (2 mM), TacAT complex (2 M), or TacT (2 M). All samples contained 0.5 l of SUPERase In RNase inhibitor (Thermo, Fisher Scientific). Reaction mixtures were brought up to equal volumes with the addition of RNase-free water, per the manufacturer's protocol. Reaction mixtures were incubated in a 37°C sand bath for 2 h, after which tubes were placed on ice, and 2.5 l of each reaction mixture was added to 12 l of 1ϫ loading dye (60% [vol/vol] glycerol, 0.3 M Tris-HCl [pH 6.8], 12 mM EDTA, 12% SDS, 0.87 mM 2-mercaptoethanol, 0.05% [wt/vol] bromophenol blue) and heated at 100°C for 10 min. Five microliters of each denatured sample was resolved by SDS-PAGE on a 15-lane 15% (wt/vol) polyacrylamide gel with Tris-HCl buffer at pH 8.8 for the resolving gel or Tris-HCl at pH 6.8 for the stacking gel. Gels were run for 45 min at 220 V and visualized via Coomassie blue staining and acetic acid destaining. Gels were imaged and analyzed for DHFR protein production using ImageQuant v5.2 software. The intensity of DHFR from each lane on the polyacrylamide gel was normalized to the asterisk-indicated bands in Fig. 3 using the ImageQuant v5.2 software. These values were used to calculate the percentage of decrease in DHFR per reaction. The mean percentages were plotted using Prism6 software to obtain the standard deviation of each percentage of decrease as shown at the bottom of the figure.
Construction of a strain carrying a chromosomal tacA allele encoding TacA K44Q or TacA K44R . A region 792 bp upstream of tacAT in frame with the coding region of tacAT was cloned into pCV1 (pTacAT-17). Site-directed mutagenesis was performed to mutate tacA to encode either TacA K44R or TacA K44Q , as described above (pTacAT-22 and pTacAT-23). This DNA fragment (792 bp upstream of tacAT in frame with tacAT coding for variants) was designated fragment 1 (Fig. S8). Gene splicing by overlap extension quantitative PCR (SOE-qPCR) was utilized to fuse this fragment to a cat ϩ gene (fragment 2) and the downstream region of tacAT (fragment 3). The cat ϩ gene from pKD3 (37) was used as a template for fragment 2. Fragments 1, 2, and 3 were amplified with Pfu Ultra II fusion DNA polymerase (Stratagene) with primers listed in Table S2 and an annealing temperature of 61.8°C. PCR products were sized on a 1% (wt/vol) agarose gel and were purified using a Wizard SV PCR cleanup system (Promega). Fragments 1 and 2 (50 ng each per 50-l PCR mixture) were annealed using SOE-PCR primers (5= SOE-PCR fragment 1 and 3= SOE-PCR fragment 2 [Table S2]) with an annealing temperature of 61.8°C and an extension time of 30 s/kb. PCR products were sized on a 1% (wt/vol) agarose gel and were purified using a Wizard SV PCR cleanup system (Promega). Fragments 1 and 2 were annealed to fragment 3 using the same protocol. The linear PCR fragments 1 to 3 were transformed into a ΔtacAT strain harboring the helper plasmid pKD46 using the protocol described above under "Strain construction" (37). The linear PCR fragment recombined with the upstream and downstream regions of tacAT, which inserted the tacAT operon coding for K44Q or K44R in place of its absence. Cells were plated on LB agar plus chloramphenicol (10 g/ml), and individual colonies were screened for acquisition of tacAT compared to ΔtacAT. Chromosomal mutations were confirmed by sequencing.
RNA isolation. Strains JE10079 (tacAT ϩ ), JE23754 (tacA1 encoding TacA K44Q T WT ), JE23755 (tacA2 encoding TacA K44R T WT ), and JE23438 (ΔtacT) were grown overnight in triplicate in nutrient broth (NB [2 ml]; Difco) with shaking at 37°C. After incubation, strains were diluted 1:100 into 5 ml of fresh NCE minimal medium supplemented with MgSO 4 (1 mM), Wolfe's trace minerals (1ϫ), and glycerol (22 mM). Cultures were grown with shaking at 37°C to an optical density at 600 nm of 0.5, and then 5 ml of each sample were quickly centrifuged in 1.5-ml Eppendorf tubes at 6,000 ϫ g, supernatant was removed, and pellets were flash-frozen in liquid nitrogen and kept on dry ice. RNA was isolated following the RNAsnap protocol (33). Pellets were resuspended in 150 l of boil solution (18 mM EDTA, 0.025% [wt/vol] SDS, 95% [vol/vol] RNA-grade formamide, 1% [vol/vol] 2-mercaptoethanol in RNase-free water) and were mixed vigorously to break up the cell pellet. Pellets were incubated at 95°C for 7 min and centrifuged at 16,000 ϫ g for 5 min at room temperature; 100 l of supernatant was transferred to a fresh tube. A sodium acetate-ethanol RNA precipitation was then conducted by the addition of 300 l of RNase-free water, 40 l of sodium acetate (3 M [pH 5.2]; final concentration, 0.3 M), and finally 400 l of ice-cold absolute ethanol (100%), with mixing briefly before the addition of the next reagent. The mixture was incubated on ice for 15 min and centrifuged at 16,000 ϫ g for 15 min at 4°C, and ethanol was decanted off. Ethanol (400 l, cold, 70% [vol/vol]) was added, and pellets were centrifuged at 12,000 ϫ g for 10 min at 4°C in an Eppendorf 5415D centrifuge. Ethanol was removed, and pellets were allowed to dry. RNA pellets were resuspended in RNase-free water at 4°C on ice overnight. Subsequent RNase-free DNase I treatment was conducted using the Ambion Turbo DNA-free kit according to the manufacturer's instructions (Thermo, Fisher Scientific). After DNA cleavage, a final sodium acetate-ethanol precipitation was performed as described above. RNA was allowed to resuspend at 4°C on ice for 4 h, then was flash-frozen in liquid nitrogen and stored at Ϫ80°C until used. A small aliquot of each sample was sent for quality control analysis using the RNA 600 Nano kit of the Agilent 2100 bioanalyzer through the Georgia Genomics Facility. Primers for qPCR were designed using primer 3 software and were evaluated for specificity and melting curve prior to running the qPCR.
cDNA synthesis and quantitative reverse transcription PCR. Total RNA (620 ng) from each sample was used for the synthesis of cDNA using the iScript cDNA synthesis kit from Bio-Rad Laboratories according to the manufacturer's protocol. Each cDNA reaction mixture was then diluted to 7.5 ng/l and used as the template for PCR. For real-time PCR, 20-l reaction mixtures were prepared with 10 l of 2ϫ FastSYBR green master mix (Applied Biosystems), 500 nM each gene-specific primer (1 l of 10 M primer stock), and 15 ng of cDNA (2 l of 7.5-ng/l cDNA). The real-time PCR was performed using a 7500 Fast real-time PCR system (Applied Biosystems). The threshold cycle values of rpoB and gyrB were checked first to ensure that both genes were optimal for use as reference genes for these strains under the conditions chosen for RT-qPCR. Cycle threshold (C T ) data were normalized to the rpoB gene (44). These normalized values (ΔC T ) were transformed using 2(eϪΔC T )/10 Ϫ6 (45) and were reported as arbitrary gene expression units (EU), or the gene expression ratio of the mutant strains to the parent strain (JE10079 tacAT ϩ ). Mean EU values were used to calculate the standard error of the mean (SEM) using Prism6 from three biological replicates that were each tested in technical triplicates. Differences in EU between mutant strains and JE10079 and between JE23754 (tacA1 encoding TacA K44Q T) and JE23755 (tacA2 encoding TacA K44R T) were compared using Welch's t test with the GraphPad Prism6 software as shown in Fig. 6.
Reagent and resource sharing. Further information and requests for reagents may be directed to and will be fulfilled by the corresponding author, Jorge C. Escalante-Semerena.