Deciphering the Regulatory Circuitry That Controls Reversible Lysine Acetylation in Salmonella enterica.

UNLABELLED: In Salmonella enterica, the reversible lysine acetylation (RLA) system is comprised of the protein acetyltransferase (Pat) and sirtuin deacetylase (CobB). RLA controls the activities of many proteins, including the acetyl coenzyme A (acetyl-CoA) synthetase (Acs), by modulating the degree of Acs acetylation. We report that IolR, a myo-inositol catabolism repressor, activates the expression of genes encoding components of the RLA system. In vitro evidence shows that the IolR protein directly regulates pat expression. An iolR mutant strain displayed a growth defect in minimal medium containing 10 mM acetate, a condition under which RLA function is critical to control Acs activity. Increased levels of Pat, CobB, or Acs activity reversed the growth defect, suggesting the Pat/CobB ratio in an iolR strain is altered and that such a change affects the level of acetylated, inactive Acs. Results of quantitative reverse transcription-PCR (qRT-PCR) analyses of pat, cobB, and acs expression indicated that expression of the genes alluded to in the IolR-deficient strain was reduced 5-, 3-, and 2.6-fold, respectively, relative to the levels present in the strain carrying the iolR(+) allele. Acs activity in cell-free extracts from an iolR mutant strain was reduced ~25% relative to that of the iolR(+) strain. Glucose differentially regulated expression of pat, cobB, and acs. The catabolite repressor protein (Crp) positively regulated expression of pat while having no effect on cobB. IMPORTANCE: Reversible lysine acylation is used by cells of all domains of life to modulate the function of proteins involved in diverse cellular processes. Work reported herein begins to outline the regulatory circuitry that integrates the expression of genes encoding enzymes that control the activity of a central metabolic enzyme in C2 metabolism. Genetic analyses revealed effects on reversible lysine acylation that greatly impacted the growth behavior of the cell. This work provides the first insights into the complexities of the system responsible for controlling reversible lysine acylation at the transcriptional level in the enteropathogenic bacterium Salmonella enterica.

INTRODUCTION

Reversible lysine acetylation (RLA) is a posttranslational regulatory mechanism present in all domains of life (1). RLA allows an organism to rapidly and reversibly modulate the biological activity of proteins involved in carbon utilization, transcription, translation, and stress responses (2–5) by modulating the acetylation state of the epsilon amino group of lysyl residues critical for function (reviewed in reference 6). In the last decade, studies have provided insights into how the RLA system works in diverse prokaryotes (3, 7–10). As shown in Fig. 1, in Salmonella enterica, the RLA system is comprised of a protein acetyltransferase (Pat) of the Gcn5 N-acetyltransferase (GNAT) family and an NAD+-consuming sirtuin deacetylase (CobB) (2). Relevant to this work is the RLA control of acetyl coenzyme A (acetyl-CoA) synthetase (Acs), an AMP-forming CoA ligase involved in acetate utilization (11). It has been shown that Pat is responsible for the acetylation and inactivation of Acs (2), while removal of the acetyl moiety of AcsAc by the CobB deacetylase reactivates the enzyme (12) (Fig. 1). RLA-dependent regulation of Acs is imperative, as uncontrolled Acs results in growth arrest by depletion of ATP pools (8).

FIG 1 

RLA control of acetyl-CoA synthetase (Acs) in S. enterica. The activity of the AMP-forming acetyl-CoA synthetase is posttranslationally modified by the protein acetyltransferase Pat. This modification is reversible by the activity of the NAD+-consuming class III sirtuin deacetylase CobB. O-AADPR, O-acetyl ADP ribose; Nm, nicotinamide.

In addition to posttranslational regulation, expression of acs is controlled by several transcriptional regulators (13). While the regulatory region of cobB in S. enterica has been examined to some extent (14), the transcriptional regulation of genes encoding the enzymes of the RLA system (pat and cobB) has not been investigated. It has been shown that the catabolite repressor protein (Crp) regulates the Escherichia coli pat homologue (pka) (15), although a role for Crp regulation of pat in S. enterica has not been reported.

In this work, a genetic approach was used to identify S. enterica genes whose products affected the pat promoter (P). We show that inactivation of iolR (stm4417), encoding an RpiR-like transcriptional repressor, decreased pat expression (16). RpiR-like regulators are involved in sugar catabolism and can function as activators and repressors (17, 18). In S. enterica, Bacillus subtilis, Corynebacterium glutamicum, and Sinorhizobium meliloti, IolR negatively regulates expression of the myo-inositol utilization operon (16, 19). myo-Inositol (cyclohexane-1,2,3,4,5,6-hexol) is an abundant cyclic polyol in soil, and its utilization as a sole carbon source depends on the presence of a large number of genes organized as a genomic island (16), which is present in gammaproteobacteria, alphaproteobacteria, and some Gram-positive bacteria (20–26).

Here we present in vivo evidence that IolR activates pat expression in S. enterica and that IolR binds to the pat promoter in vitro. We also report that acs and cobB are transcriptionally activated by IolR, which places the RLA system of S. enterica under IolR control. Significantly, an iolR mutant strain displayed a growth defect in minimal medium containing 10 mM acetate, which we suggest is due to an imbalance of the active (nonacetylated)/inactive (acetylated) Acs ratio caused by changes in pat and cobB expression in the absence of IolR. Finally, we show that Crp, a global regulator of carbon metabolism, regulates pat and acs expression in S. enterica. To our knowledge, this is the first report of global, integrative transcriptional control of genes encoding the enzymes of the RLA system in S. enterica and its effect on carbon metabolism.

RESULTS

IolR regulates pat expression.

A genetic screen was used to identify genes whose functions affected the expression of pat, the gene encoding the protein acetyltransferase in S. enterica. Changes in pat expression were monitored in strain JE7449, which carried a chromosomal pat::MudJ (lacZ) reporter (hereafter pat-lacZ) (see Table S1 in the supplemental material). This strain was transduced to tetracycline resistance (Tcr) using a P22 lysate grown on a pool of ~100,000 strains, each of which was assumed to contain one Tn10d(tet) element randomly inserted in the genome. Tcr derivatives of the pat-lacZ+ reporter strain (~20,000) were screened for changes in β-galactosidase activity, leading to the identification of two colonies that were less blue than the parental strain. The DNA sequence flanking the Tn10d(tet) elements located both insertions within iolR (stm4417), the gene encoding the repressor of the myo-inositol utilization (iol) genes. No other insertions affecting pat expression were identified. To confirm that the iolR::Tn10d(tet) element was responsible for the reduced expression of the pat-lacZ reporter, phage P22 grown on the original iolR::Tn10d(tet) pat-lacZ strain was used to transduce strain JE7449 (pat-lacZ) to Tcr. The reconstructed iolR::Tn10d(tet) pat-lacZ+ strain (JE10535) displayed the same reduction in pat-lacZ expression measured in the original mutant strain (data not shown).

To independently confirm the effect of IolR on pat expression, an iolR::cat mutation was introduced into strain JE7449 (pat-lacZ). Measurements of β-galactosidase activity of the pat-lacZ iolR::cat+ strain (JE10714) during growth in nutrient broth (NB) showed a reproducible ~2-fold decrease in pat promoter (P) activity relative to that in the pat-lacZ+ iolR+ strain (Fig. 2A). Complementation analysis with a wild-type allele of iolR provided in trans restored pat expression to the wild-type level (Fig. 2B). The effect of IolR on pat expression was confirmed by quantitative reverse transcription-PCR (qRT-PCR) and showed a 5-fold downregulation of the pat transcript in an iolR::cat+ strain compared to the wild type (Fig. 2C). From these data, we concluded that the decrease in pat expression in an iolR strain was due to the absence of IolR.

FIG 2 

IolR activates pat expression in vivo. Activity of a pat-lacZ chromosomal operon fusion was assessed in the presence (JE7449) or absence (JE10714) of iolR to measure pat promoter activity (Miller units [MU]) in vivo (A and B). Cell cultures were grown at 37°C in NB medium. The data presented are the average of two independent experiments from individual cultures performed in triplicate. Error bars represent standard deviations. The unpaired t test gave a P value of 0.0004 (A). (C) qRT-PCR showed a 5-fold downregulation in pat activity in an iolR mutant strain relative to the iolR strain. The wild-type transcript level is set at 1, indicated by the dashed line. Error bars represent standard deviations; pVOC, vector-only control.

The effect of IolR on pat expression was tested on acetate (10 mM) and myo-inositol (55 mM). In the absence of IolR, pat-lacZ expression decreased 1.4-fold on acetate (see Fig. S1A in the supplemental material) and 1.3-fold on myo-inositol (Fig. S1B) compared to the levels of expression in the iolR strain.

Pat does not acetylate IolR and is not required during growth on myo-inositol.

IolR represses the expression of genes encoding myo-inositol-degrading enzymes (16, 27), and it was surprising to find that IolR may also play a role in the activation of genes comprising the RLA system. We considered the possibility of a regulatory system in which Pat would control the DNA binding activity of IolR via acetylation, as reported for Pat and the E. coli transcription factor RcsB (3). However, we did not obtain any experimental evidence of Pat-dependent regulation of IolR function under conditions in which acetyl-CoA synthetase (Acs), a bona fide Pat substrate (12), was acetylated (see Fig. S2 in the supplemental material).

Consistent with the above-mentioned observation, we determined no difference in the growth rate of the pat-lacZ iolR::cat from that of a strain carrying the wild-type pat and iolR alleles when grown on myo-inositol (see Fig. S3A in the supplemental material). However, the pat-lacZ strain consistently showed a slight but reproducible delay in the onset of growth. As reported by others (16), we observed that the onset of growth of the iolR::cat strain occurred substantially earlier than that of a strain carrying the wild-type iolR allele, an observation consistent with the lack of repression of the myo-inositol genomic island in a strain devoid of the IolR repressor (see Fig. S3A).

IolR is a tetramer.

To study the role of IolR regulation of pat, the IolR protein was isolated to 96% homogeneity (see Fig. S4A in the supplemental material). The oligomeric state of IolR was determined using fast protein liquid chromatography (FPLC) gel filtration analysis (see Fig. S4B in the supplemental material). Under the conditions tested, IolR eluted ~24 min after injection, a retention time consistent with the behavior of a protein whose mass was ~134 kDa compared to elution times of molecular mass standards. Since the predicted molecular mass of IolR was 31 kDa, it was concluded that IolR was either a dimer of dimers or a tetramer.

IolR binds to the pat promoter (P) in vitro.

Electrophoretic mobility shift assays (EMSAs) were performed to determine if the effect of the IolR regulator on pat expression was the result of direct binding of IolR to P. Experimental promoter analysis data were used to identify the transcription start site (TSS) (http://www.imib-wuerzburg.de/research/salmonella) (28). To probe for the specificity of the interaction between IolR and P, we added a nonspecific competitor DNA probe, P, previously used to study IolR binding (16). Increasing concentrations of IolR shifted the P probe but not the P probe, a result that supported the conclusion that IolR directly and specifically interacted with the P promoter (Fig. 3A). Increasing amounts of IolR protein titrated against a fixed amount of the P DNA probe without the presence of P also yielded increased amounts of IolR/P complex (Fig. 3B). IolR did not shift the mobility of P until a molar excess of 50× protein was reached, indicative of nonspecific binding of IolR to P (Fig. 3C). Previous studies by others showed that IolR negatively regulates the transcription of its own promoter, P (16). Using P (175 nucleotides [nt]) as a positive control with the presence of the competitor probe P, we confirmed the reported specificity of IolR for its promoter (16) (Fig. 3D). The P and P probes each shifted at similar molar excess concentrations of IolR, supporting the conclusion that IolR directly and specifically interacted with P.

FIG 3 

IolR binds to the pat promoter region. Binding of IolR to the 6-FAM 5′-labeled pat promoter (P [150 nt, 51 nM]) was analyzed by electrophoretic mobility shift assays in the presence of increasing concentrations of IolR. (A) P and competitor DNA (P [196 nt]) were incubated together to show binding specificity of IolR to P. (B) The P probe alone was incubated at various concentrations of IolR. (C) Competitor DNA P was incubated with increasing concentrations of IolR to determine at what point nonspecific binding interactions occur. (D) The interaction between IolR and P (175 nt) was performed as a known binding control and incubated in the presence of competitor DNA, P. The protein concentration shown is in molar excess to the probe (picomoles). EMSAs were performed in triplicate.

Region of the pat promoter recognized by IolR.

We performed DNA-footprinting analysis to identify the region within the pat promoter recognized by IolR. A 6-carboxyfluorescein (6-FAM) 5′-labeled 382-nucleotide probe containing the P promoter was incubated with various concentrations of IolR protein or bovine serum albumin (BSA [negative control]). After incubation and subsequent DNase digestion and purification of the DNA, samples were analyzed as described in Materials and Methods. Electropherogram overlays comparing IolR and BSA (negative control) and putative binding sites were analyzed by aligning the sequenced probe data (data not shown). Data presented in Fig. 4 show a region of protection of P from nucleotides −112 to −70 relative to the predicted transcription start site (Fig. 4A). Experimental promoter analysis data from Kroger et al. were used to identify the transcription start site (http://www.imib-wuerzburg.de/research/salmonella) (28) (Fig. 4B). A region of hypersensitivity was seen at position −112, an indicator of DNA bending as the result of the binding of a transcriptional regulator, causing an exposed site susceptible to increased cleavage by DNase. A control was performed in which the amount of IolR was doubled in the reaction (10 µg). With this increase in protein concentration, we expected an increase in signal intensity, as seen in Fig. 4A.

FIG 4 

IolR protein binds pat promoter at positions −112 to −70. (A) DNA-footprinting analysis by capillary electrophoresis was used to define the IolR binding region on the pat promoter (P). On the graph, negative values represent an area of protection, bars indicate the concentration of IolR protein (5 µg, dark gray; 10 µg, light gray), and bar heights represent the area difference between the IolR sample and the negative control (BSA). DNA footprinting was performed and analyzed in two independent experiments. (B) Work by Kröger et al. (28) was used to identify the pat transcriptional start site (TSS).

Previous studies aimed at examining the regulation of the iol genomic island in S. enterica by IolR repression did not identify a conserved binding site (16, 29). The intent of the DNA-footprinting analysis was to compare the binding region of IolR within P to promoter regions regulated by IolR, with the idea of determining a consensus IolR-binding region. While unable to determine a consensus site, the data confirmed the direct interaction between IolR and P.

IolR binds to the region of P identified by DNA footprinting.

A 45-nt probe corresponding to the protected region of P identified as the IolR-binding region was used to validate the DNA-footprinting experiments. A 6-FAM 5′-labeled 45-nt probe, P, was generated by annealing complementary primers, and the binding of IolR to this region was examined. The data presented show that IolR binds to the 45-nt probe, confirming that the IolR-binding site is located within this region (see Fig. S5 in the supplemental material). The reason for the presence of signals of higher-molecular-mass complexes is unclear. Possible explanations include the absence of a ligand sensed by IolR or the formation of higher-order IolR multimers. We speculate that this behavior will be better understood as we learn how IolR interacts with P.

The absence of IolR impairs growth on 10 mM acetate.

Growth of an iolR::cat strain was inhibited on 10 mM acetate, with a growth rate three times slower (doubling time of 36 h) (Fig. 5A, solid triangles) than that of a strain carrying the wild-type iolR allele (doubling time of 11 h) (Fig. 5A, solid squares). Growth of the iolR::cat strain was restored when iolR was expressed ectopically (Fig. 5A, open triangles), indicating that the growth defect was due to the absence of IolR. A similar growth defect was reported for an iolR mutant strain of C. glutamicum, but this observation was not investigated (27). No growth differences were observed for the iolR, iolR::cat, or pat::MudJ strains when grown on 50 mM acetate or glycerol (see Fig. S3C and D in the supplemental material).

FIG 5 

An iolR strain has a growth defect on 10 mM acetate. Growth of S. enterica strains was examined in NCE minimal medium containing 10 mM acetate. Expression of iolR was induced using 100 µM l-(+)-arabinose, cobB expression was induced with 10 µM l-(+)-arabinose, and pat expression was induced with various concentrations of inducer, as indicated. Growth curves were performed using a Powerwave XS2 microplate reader (Bio-Tek Instruments) at 37°C with shaking in triplicate in three independent experiments. The strains analyzed had the following genotypes: iolR (JE6583), iolR::cat (JE10713), iolR::cat/pVOC (JE16934), iolR::cat/P (JE16935), pat::MudJ (JE7449), pat::MudJ iolR::cat (JE10714), cobB::MudJ (JE2845), cobB::MudJ iolR::cat (JE14972), iolR::cat/P (JE18927), and iolR::cat/P (JE18891). Error bars represent standard deviations. pVOC, vector-only control.

Because the observed phenotype on 10 mM acetate correlated with lower levels of pat expression, we hypothesized that increases in the expression of pat under the control of an IolR-independent promoter would restore growth of the iolR::cat strain on 10 mM acetate. Indeed, the iolR::cat strain grew almost as well as the iolR strain when pat was expressed in trans (Fig. 5B, inverted triangles). An increase in the level of inducer (10 to 100 µM) compromised growth of the iolR::cat/P strain. The negative effect of higher pat expression was not surprising since increased Pat levels are known to increase the level of acetylated, inactive Acs (8).

The phenotype of the iolR strain is caused by an imbalance in the Pat/CobB ratio, which affects Acs activity.

In S. enterica, Pat and CobB control Acs activity (6). Given that pat expression decreased in the iolR::cat strain, we hypothesized that the absence of IolR created an imbalance in the Pat/CobB ratio that favored Pat activity and thus a decrease in Acs activity due to Acs acetylation. We reasoned that such loss of Acs activity could be counteracted in several ways. First, inactivation of pat in the iolR strain would block Acs acetylation and should restore growth. Indeed, the poor growth of the iolR strain on 10 mM acetate (Fig. 5C, solid triangles) was reversed when pat was inactivated (Fig. 5C, open triangles). As expected, a cobB strain failed to grow on 10 mM acetate because acetylated Acs could not be reactivated by deacetylation (Fig. 5C, solid circles). Although inactivation of iolR presumably reduced Pat levels in the cobB strain (by lowering the expression of pat), the reduced level was apparently sufficient to keep Acs acetylated (i.e., inactive); thus, growth was not restored (Fig. 5C, open squares). Additionally, inactivation of pat in an otherwise wild-type background had minimal effect on growth likely caused by an excess of Acs activity due to CobB deacetylation (2).

Second, if the net result of the change in the Pat/CobB ratio in the iolR::cat strain was an increase in acetylated, inactive Acs, an increase in the level of CobB sirtuin deacetylase in the iolR::cat strain would restore Acs to its active, deacetylated state, and consequently growth on 10-mM acetate would occur. This prediction was confirmed, as shown in Fig. 5D.

Third, if the growth defect of the iolR::cat strain on 10 mM acetate was caused by a change in the level of Acs activity, it followed that overexpression of acs in the Δacs iolR::cat strain would restore growth on 10 mM acetate. Results obtained using control strains are shown, and as expected, the Δacs strain failed to grow on 10 mM acetate (Fig. 6A, inverted triangles), and growth was restored by expression of acs in trans (Fig. 6A, circles). Shown in Fig. 6B is the effect of ectopic synthesis of wild-type Acs in the Δacs iolR::cat strain. Wild-type growth of the Δacs iolR::cat strain on 10 mM acetate was observed upon induction of acs expression [20 µM l-(+)-arabinose] (Fig. 6B, circles). Unsurprisingly, excessive levels of Acs [100 µM l-(+)-arabinose] (Fig. 3, open squares) had a deleterious effect on growth, as reported elsewhere (8).

FIG 6 

Induction of acs expression restores growth of an iolR strain on 10 mM acetate. Growth of a iolR::cat strain containing acs expressed ectopically under the control of an l-(+)-arabinose-inducible promoter was examined in minimal medium containing acetate (10 mM). Control strains are shown in panel A. The effects of acs induction are shown in panel B. Growth curves were performed using a Powerwave XS2 microplate reader (Bio-Tek Instruments) at 37°C with shaking in triplicate in three independent experiments. The strains analyzed had the following genotypes: iolR (JE6583), iolR::cat (JE10713), Δacs (JE7758), Δacs/P (JE9912), and Δacs iolR::cat/P (JE16596). Expression of acs was induced with 0 µM (open triangles), 20 µM (open circles), or 100 µM (black circles in panel A and open squares in panel B). Error bars represent standard deviations.

If the growth phenotype of the iolR::cat strain on acetate was due to lower Acs activity, we should be able to detect differences in Acs activity in cell-free extracts. Indeed, a reproducible and statistically significant reduction (~25%) in Acs activity was found in cell-free extracts of the iolR::cat strain relative to extracts of the iolR+ strain (Fig. 7).

FIG 7 

Activity of acetyl-CoA synthetase (Acs). The activity of Acs from whole-cell extracts of an iolR or iolR::cat+ strain grown on acetate (10 mM) was measured using a coupled NADH-consuming spectrophotometric assay (41). The strains analyzed were iolR (JE6583) and iolR::cat (JE10713). Samples were analyzed in triplicate. Error bars represent standard deviations. The unpaired t test gave a P value of 0.002.

IolR controls expression of acs and cobB.

Since ectopic expression of acs restored growth of the iolR::cat mutant, we surmised that acs expression was lower in the mutant than in the wild-type strain. To address this possibility we used an acs-lacZ reporter fusion to determine whether IolR was also involved in the regulation of acs in S. enterica. Since alterations in pat expression were likely to affect CobB levels, we also used a cobB-lacZ fusion to assess the effect of the absence of IolR on cobB expression. Data obtained from experiments with the above-mentioned transcriptional reporters support the idea that IolR somehow activated expression of both genes (Fig. 8). In the absence of IolR, expression of acs (Fig. 8A) and cobB (Fig. 8B) was reduced on average by 40% (acs) or 22% (cobB), respectively. The effect of IolR on acs and cobB expression was confirmed by qRT-PCR and showed a >2-fold and >3-fold downregulation, respectively, of the transcripts in an iolR::cat+ strain compared to the wild type (Fig. 8C). At present, it is unclear whether the effect of IolR on Acs and CobB levels is direct or indirect.

FIG 8 

IolR controls expression of acs and cobB. The activity of acs-lacZ and cobB-lacZ reporters was assessed in backgrounds ± iolR to measure P and P activity (Miller units [MU]). Cultures were grown at 37°C in NB medium (A and B). Optical density (650 nm) and β-galactosidase activity (420 nm) were measured hourly. The data were obtained from individual cultures performed in triplicate. Error bars represent standard deviations. The strains analyzed had the following genotypes: cobB::MudJ (JE2845), cobB::MudJ iolR::cat (JE14972), pACS3 P (JE4637), and iolR::cat/pACS3 P (JE14962). (C) qRT-PCR showed a 2.6-fold downregulation in acs activity and a 3-fold decrease in cobB activity in an iolR strain relative to the iolR+ strain. The wild-type transcript level is set at 1, indicated by the dashed line. Error bars represent standard deviations.

Glucose differentially affects pat, cobB, and acs expression.

Due to the previously established role of the catabolite repressor protein (Crp) in the regulation of the E. coli pat homologue (pka) (15) we examined the effect of catabolite repression on genes encoding the RLA system ± iolR. Expression of pat in cells grown in NB plus glucose was reduced by a factor of 2 relative to the expression of pat in cells grown in NB lacking glucose (Fig. 9, compare black bars). This suggested that pat expression was subjected to catabolite repression, an idea that was further explored. Regardless of the presence of glucose in the medium, the absence of IolR reduced pat-lacZ expression 30 to 40% (Fig. 9A). The absence of IolR had a small but reproducible negative effect (~20%) on cobB expression in the absence of glucose, an effect that was magnified to ~40% when glucose was added (Fig. 9B). Significantly, in contrast to pat expression, the expression of cobB increased ~30% when glucose was added, suggesting that unlike pat, expression of cobB was not subject to catabolite repression (Fig. 9B, compare black bars).

FIG 9 

Glucose differentially controls expression of pat, cobB, and acs. Cultures were grown at 37°C in NB medium ± glucose (10 mM). Optical density (650 nm) and β-galactosidase activity (420 nm) were measured at mid-log phase (OD650 of ~0.7) to assay for pat and cobB promoter activity (Miller units [MU]) in a pat-lacZ or cobB-lacZ strain background ± iolR. The data are the average of two independent experiments from individual cultures performed in triplicate. The strains analyzed had the following genotypes: pat::MudJ (JE7449), pat::MudJ iolR::cat (JE10714), cobB::MudJ (JE2845), cobB::MudJ iolR::cat (JE14972), pACS3 P (JE4637), and iolR::cat/pACS3 P (JE14962). Error bars represent standard deviations.

In E. coli, Crp controls the expression of acs (30). Consistent with the idea that in S. enterica acs expression is controlled by catabolite repression, transcription of acs was reduced 80% when glucose was present in the medium (Fig. 9C, compare black bars). IolR function also appeared to be important for the activation of acs in medium devoid of glucose, with ~50% reduction in acs expression in the iolR::cat strain relative to the wild type (Fig. 9C, NB medium). In the presence of glucose, expression of acs ± iolR was very similar (Fig. 9C, NB plus glucose).

Crp activates pat expression.

The effect of Crp on pat expression was examined in cultures grown in NB plus ribose (10 mM). P activity in the ΔiolR crp::cat strain was 2.5-fold lower during the mid-log phase than the P activity measured in a strain containing wild-type iolR and crp alleles (Fig. 10). This decrease in P activity was similar to the one measured in a strain lacking iolR and was restored when crp was provided in trans. P activity was slightly lower in the ΔiolR crp::cat strain than that in strains lacking only crp or iolR (Fig. 10). Collectively, the data indicated that Crp was required for wild-type levels of pat expression in S. enterica. Pat-dependent acetylation of Crp was tested; however, the data indicate that Pat does not acetylate Crp (see Fig. S2 in the supplemental material).

FIG 10 

Crp activates pat expression. Cultures were grown at 37°C in NB medium with ribose (10 mM). Optical density (650 nm) and β-galactosidase activity (420 nm) were measured at the peak of pat expression (OD650 of ~0.7) to assay for pat promoter activity (Miller units [MU]) in a pat-lacZ strain background. Plasmids were induced with l-(+)-arabinose (100 µM). The data presented are the average of two independent experiments from individual cultures performed in triplicate. The strains analyzed had the following genotypes: pat::MudJ (JE7449), pat::MudJ iolR::cat (JE10714), pat::MudJ crp::cat (JE16743), pat::MudJ ΔiolR crp::cat (JE16744), pat::MudJ iolR::cat/pVOC (JE10727), pat::MudJ iolR::cat/pIOLR1 (JE10728), pat::MudJ crp::cat/pVOC (JE16771), and pat::MudJ crp::cat/P (JE17322). Error bars represent standard deviations. pVOC, vector-only control.

DISCUSSION

In S. enterica, IolR regulates the RLA system.

The chief finding from the studies reported herein is that the IolR protein controls and integrates the expression of pat, cobB, and acs (Fig. 2 and 8). Although we do not yet understand the molecular details of how IolR integrates the expression of the above-mentioned genes, collectively our in vivo genetic evidence supporting this claim is compelling (Fig. 5 and 6). Furthermore, in vitro data obtained support the conclusion that IolR directly interacts with the pat promoter (Fig. 3 and 4). Whether or not the effect of IolR on cobB and acs expression is direct remains to be determined.

IolR function is needed for growth on 10 mM acetate, which requires RLA and Acs.

IolR function is necessary for optimal growth on 10 mM acetate (Fig. 5), and the growth defect of an iolR strain suggests that IolR regulation of pat and cobB impacts the levels of Acs activity in the cell (Fig. 8). This conclusion is supported by data showing that the ectopic expression of acs complements growth of an iolR strain (Fig. 6). The subtle effects of the absence of IolR on pat, cobB, and acs expression make it difficult to determine the precise magnitude of the changes in Pat, CobB, and Acs protein levels, and in the case of Acs, there is also a need to distinguish between acetylated versus nonacetylated protein. Our attempts to gain insights into these changes using Western blot analysis were unsuccessful due to the lack of required sensitivity to define the magnitude of the predicted changes (data not shown). However, the lower levels of Acs activity present in cell extracts of the iolR strain (Fig. 7) support our conclusions.

The differentially repressive effect of glucose on pat and cobB expression is needed to ensure sufficient activation of acetate by Acs.

The differential effect that glucose has on the expression of pat and cobB (Fig. 9) can be explained by considering the acetogenic nature of glucose. During glucose catabolism, excess acetyl-CoA is diverted through the acetate kinase (AckA)/phosphotransacetylase (Pta) pathway, yielding ATP via substrate-level phosphorylation and releasing CoA, which is needed by the pyruvate dehydrogenase to make more acetyl-CoA. In S. enterica, the assimilation of glucose-derived acetate is known to require the functions of AckA/Pta and Acs. Notably, during growth on glucose, less Acs is made because Crp cannot fully activate acs expression when cyclic AMP (cAMP) levels are low (31). One plausible way the cell can ensure that the limited amount of Acs made by the cell in the presence of glucose supplies enough acetyl-CoA to support growth is to lower the expression of pat while increasing the expression of cobB (Fig. 9). By so doing, less Acs becomes acetylated, and whatever Acs is acetylated is reactivated by the higher levels of CobB deacetylase made under these conditions. Given that pat expression is reduced whenever IolR or Crp is not made, it was surprising to find out that the absence of both regulators did not have an additive effect on pat expression (Fig. 10). This result could simply reflect the basal level of pat expression or possibly represent a more complex regulatory network in which Crp affects regulation of iolR. Such an idea would not be unprecedented, since in B. subtilis, the iol operon, including iolR, is under the control of catabolite repression (32).

Why does IolR activate expression of the RLA system?

The dual regulatory role of IolR as a repressor of iol genes and an activator of pat, cobB, and acs is intriguing. The ability of IolR to function as an activator is not unprecedented. A report on C. glutamicum demonstrated that IolR activates expression of pck (encodes phosphoenolpyruvate carboxykinase) (27). IolR also regulates srfJ, coding for a type III secretion effector protein, in Salmonella (33), a result that is not surprising as srfJ lies within the iol genomic island, which is regulated by IolR.

Under conditions where acetate and myo-inositol are simultaneously present in the environment, the cell needs to integrate their metabolism. In thinking about this issue, one must consider that acetate and myo-inositol are catabolized at substantially different rates, as acetate enters the central metabolism as soon as it is converted into acetyl-CoA. In contrast, myo-inositol degradation requires the generation of a signal that upon binding to IolR lifts repression of the iol genes, and synthesis of myo-inositol-degrading enzymes can occur. Generation of the signal needed to transcribe the iol genes is apparently an extensive process (see Fig. S3A in the supplemental material for comparison of the differences in lag phases between the iolR and iolR strains). The use of IolR to integrate myo-inositol and acetate metabolism would be an efficient way to generate as much acetyl-CoA for anabolic purposes as possible while maintaining the capability of modulating the activity of the RLA system for the purpose of controlling the level of Acs activity.

Is myo-inositol utilization regulated by RLA?

Recently, the total population of acetylated proteins (the “acetylome”) of the myo-inositol-utilizing bacteria B. subtilis and Erwinia amylovora were reported (9, 10). Notably, two enzymes involved in the degradation of myo-inositol, malonate semialdehyde dehydrogenase (IolA) and carbohydrate kinase (IolC), were among the acetylated proteins identified. It is unclear whether the activity of IolA and/or IolC is controlled by RLA in either organism or in S. enterica. If any Iol proteins are under RLA control, this could provide a link between IolR regulation of RLA and RLA involvement in myo-inositol utilization.

MATERIALS AND METHODS

Culture media and chemicals.

Nutrient broth (NB) (Difco) containing NaCl (85 mM) was used as rich medium. The minimal medium used was no-carbon essential (NCE) minimal medium (34) containing MgSO4 (1 mM), Wolfe’s trace minerals (1×) (35), and a carbon source (acetate [10 or 50 mM], glycerol [22 mM], or myo-inositol [50 mM]). Antibiotics were added at the following concentrations: tetracycline, 20 µg·ml−1; chloramphenicol, 20 µg·ml−1; kanamycin, 50 µg·ml−1; and ampicillin, 100 µg·ml−1. When added, 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside (X-Gal) was present at 40 µg·ml−1, and the calcium chelator ethyleneglycol tetraacetic acid (EGTA) was present at 10 mM. Chemicals were purchased from Sigma-Aldrich.

Bacterial strains.

All strains studied were derivatives of S. enterica serovar Typhimurium strain LT2 (unless otherwise noted in Table S1 in the supplemental material). All primers used in this study are listed in Table S2 in the supplemental material (IDT, Coralville, IA). Strains carrying a deletion of iolR or crp were constructed following described protocols (36) (see Text S1 in the supplemental material).

Plasmid construction.

Plasmids are listed in Table S1 in the supplemental material. For details on plasmid construction, see Text S1 in the supplemental material.

Isolation of Tn10d(tet) insertion in iolR.

A pool of ~100,000 S. enterica strains, each assumed to contain one Tn10d(tet) element randomly inserted in the chromosome, was generated as described previously (37). A P22 lysate grown on this pool of strains was used to transduce recipient strain JE7449 (metE ara pat::MudJ) to tetracycline resistance (Tcr) on NB agar plates containing X-Gal and EGTA. Colonies displaying altered coloration were freed of phage, and P22 phage lysates were generated to use as donors in crosses with the parental JE7449 strain. The location of the insertion was determined in the reconstructed strains by sequencing the DNA flanking the Tn10d(tet) element using a PCR-based protocol with degenerate primers (38). DNA sequencing was performed using BigDye Terminator v3.1 protocols (Applied Biosystems).

Growth studies.

Cultures were grown overnight at 37°C in NB and used to inoculate medium with (5% [vol/vol]) in a volume of 200 µl per well of a 96-well plate. NCE minimal medium containing MgSO4 (1 mM), Wolfe’s trace minerals (1×), and a carbon source (acetate [10 or 50 mM], glycerol [22 mM], or myo-inositol [50 mM]) was used. Plasmids were induced with l-(+)-arabinose, as described. Plates were incubated at 37°C in a Powerwave microplate reader (Bio-Tek Instruments). Data were analyzed using Prism v6 (GraphPad) software.

β-Galactosidase assays.

β-Galactosidase activities were determined as described previously (39). Three independent overnight cultures were grown in NB plus ampicillin, subcultured (1:100 [vol/vol]) into 200 ml of medium plus ampicillin, and induced with 100 µM l-(+)-arabinose. Cultures were grown at 37°C in NB, NB plus ribose (10 mM), NB plus glucose (10 mM), acetate (10 mM), or myo-inositol (55 mM) medium. Acetate cultures were inoculated with 2.5% (vol/vol) of overnight culture.

qRT-PCR.

Cultures of strains JE10713 (iolR::cat+) and JE6583 (iolR+) were grown in NB to an optical density at 600 nm (OD600) of 0.6. RNA extraction was performed as described previously (40). cDNA synthesis was performed using iScript cDNA synthesis kit (Bio-Rad). qRT-PCRs were performed using Fast SYBR green master mix (Thermo Fischer) and a 7500 Fast real-time PCR system (Applied Biosystems).

Purification of IolR and Crp.

IolR was purified to 96% homogeneity by a 2-step nickel affinity chromatography purification using an ÄKTA FPLC system (GE Healthcare). Crp was purified using a 2-step nickel affinity chromatography by gravity column. For protocol details, see Text S1 in the supplemental material.

Analytical gel filtration.

Experiments were performed using a Superdex 200 HR 10/30 gel filtration column (GE Healthcare) attached to an ÄKTA FPLC system (GE Healthcare). For protocol details, see Text S1 in the supplemental material.

DNA-binding assays.

Electrophoretic mobility shift DNA-binding assays were performed using probes with the fluorophore 6-carboxyfluorescein (6-FAM) attached at the 5′ end. Probes were generated by PCR amplification from the S. enterica chromosome (for details, see Text S1 in the supplemental material). Probes were purified with the Wizard SV gel and PCR cleanup system (Promega). Reaction mixtures (10 µl) contained 6-FAM 5′-labeled probe (50 ng), Tris-HCl buffer (50 mM [pH 7.5]), KCl (50 mM), MgCl2 (10 mM), EDTA (0.5 mM), glycerol (10% [vol/vol]), and IolR protein (shown in molar excess to probe [2.5 to 10 pmol]). Reaction mixtures were incubated at 22°C for 45 min and resolved on a Criterion 10% native polyacrylamide gel (BioRad) in 0.5 × Tris-borate-EDTA (TBE) buffer (Tris-HCl [45 mM], boric acid [45 mM], EDTA [1 mM] [pH 8.3]). The signal was detected using a Typhoon Trio+ variable mode imager (GE Healthcare) with ImageQuant v5.2 software.

DNA-footprinting analysis.

The pat promoter (P) was amplified from S. enterica genomic DNA using 6-FAM 5′-labeled primer (P 382 FAM) and 3′ primer (P 382 Rev). Reaction mixtures contained various amounts of IolR or bovine serum albumin (BSA [negative control]), Tris-HCl (50 mM [pH 7.5]), KCl (50 mM), MgCl2 (10 mM), EDTA (0.5 mM), and glycerol (10% [vol/vol]) and were incubated for 10 min at 25°C. The labeled DNA probe (120 ng) was added to the reaction mixture, and the mixture was incubated for 20 min at 25°C, followed by addition of DNase for 5 min at 25°C, and then heat inactivated for 10 min at 78°C. Reaction mixtures were purified using the MinElute PCR purification kit (Qiagen). Samples were analyzed with an Applied Biosystems 3730 DNA analyzer (Plant-Microbe Genomics Facility, Ohio State University) set to the default run module for LIZ600 dye, with 0.1 µl of size standard (LIZ600), 0.5 to 1.0 µl of sample, and 9 µl of HiDi per well. For protocol details, see Text S1  in the supplemental material.

In vitro acetylation assay.

Protein acetylation assays were performed as described, using radiolabeled [1-14C]acetyl-CoA (2, 7, 14). Briefly, reaction mixtures contained S. enterica IolR (SeIolR) or SeCrp (5 µM) with or without SePat (3 µM). SeAcs was used as a positive control. Laboratory stocks of homogeneous SePat and SeAcs proteins were used in these studies. Reactions were resolved and visualized by SDS-PAGE. Radiolabeled proteins were visualized using a Typhoon Trio+ variable mode imager (GE Healthcare) with ImageQuant v5.2 software.

Acetyl-CoA synthetase assay.

Acs activity was measured from treated whole-cell lysates of iolR (JE6583) and iolR::cat (JE10713) strains using an NADH-consuming assay as described previously (41). For protocol details see Text S1 in the supplemental material.

Supplemental information. Download

Text S1, PDF file, 0.2 MB

Expression of pat on acetate and myo-inositol. Cultures were grown in NCE minimal medium with acetate (10 mM) (A) or myo-inositol (55 mM) (B). β-Galactosidase activity was measured at mid-log phase. Assays were determined in duplicate from three biological replicates. Download

Figure S1, PDF file, 0.4 MB

Pat does not acetylate IolR or Crp. An in vitro acetylation assay using [1-14C]acetyl-CoA was used to test Pat-dependent acetylation of IolR or Crp. Acs was used as a positive control. Reactions were analyzed SDS-PAGE and stained with Coomassie blue to visualize proteins (A). Transfer of the acetyl group was visualized by phosphorimaging (B). +Ctrl, positive control. Download

Figure S2, PDF file, 1.1 MB

Growth study controls. Growth of S. enterica strains in NCE minimal medium with myo-inositol (50 mM) (A), acetate (50 mM) (B), and glycerol (22 mM) (C). Growth curves were performed using a Powerwave XS2 microplate reader (Bio-Tek Instruments) at 37°C with shaking in triplicate in three independent experiments. The strains analyzed had the following genotypes: iolR (JE6583), iolR::cat (JE10713), iolR::cat/pVOC (JE16934), iolR::cat/P (JE16935), and pat::MudJ (JE7449). Plasmids were induced with 100 µM l-(+)-arabinose. Error bars represent standard deviations. pVOC, vector-only control. Symbols shown in panel C apply to all panels. Download

Figure S3, PDF file, 0.5 MB

IolR is a tetramer. (A) S. enterica IolR protein (30.8 kDa) was purified by nickel affinity purification. An SDS-PAGE gel shows molecular size standards (kilodaltons) (lane 1) and IolR protein (lane 2) purified to 96% homogeneity. (B) The molecular mass of IolR (solid circle) in solution was estimated by gel filtration. The molecular mass standards (open circles) are bovine thyroglobulin (670 kDa), bovine gamma globulin (158 kDa), chicken ovalbumin (44 kDa), horse myoglobin (17 kDa), and vitamin B12 (1.35 kDa). Download

Figure S4, PDF file, 0.4 MB

IolR protein binds to the P 45-nt probe. Electrophoretic mobility shift assays were used to validate the binding region identified by footprinting experiments. A 6-FAM 5′-labeled 45-nt probe of the pat promoter region, P (166 nM), corresponding to the identified binding region was incubated with increasing concentrations of IolR protein. The protein concentration shown is in molar excess to probe. EMSAs were performed in triplicate. Download

Figure S5, PDF file, 0.7 MB

Strains and plasmids used in this study.

Table S1, PDF file, 0.1 MB

Primers used in this study.

Table S2, PDF file, 0.1 MB