Low Cytoplasmic Magnesium Increases the Specificity of the Lon and ClpAP Proteases

ABSTRACT

Proteolysis is a fundamental property of all living cells. In the bacterium Salmonella enterica serovar Typhimurium, the HspQ protein controls the specificities of the Lon and ClpAP proteases. Upon acetylation, HspQ stops being a Lon substrate and no longer enhances proteolysis of the Lon substrate Hha. The accumulated HspQ protein binds to the protease adaptor ClpS, hindering proteolysis of ClpS-dependent substrates of ClpAP, such as Oat, a promoter of antibiotic persistence. HspQ is acetylated by the protein acetyltransferase Pat from acetyl coenzyme A (acetyl-CoA) bound to the acetyl-CoA binding protein Qad. We now report that low cytoplasmic Mg²⁺ promotes qad expression, which protects substrates of Lon and ClpSAP by increasing HspQ amounts. The qad promoter is activated by PhoP, a regulatory protein highly activated in low cytoplasmic Mg²⁺ that also represses clpS transcription. Both the qad gene and PhoP repression of the clpS promoter are necessary for antibiotic persistence. PhoP also promotes qad transcription in Escherichia coli, which shares a similar PhoP box in the qad promoter region with S. Typhimurium, Salmonella bongori, and Enterobacter cloacae. Our findings identify cytoplasmic Mg²⁺ and the PhoP protein as critical regulators of protease specificity in multiple enteric bacteria.

IMPORTANCE The bacterium Salmonella enterica serovar Typhimurium narrows down the spectrum of substrates degraded by the proteases Lon and ClpAP in response to low cytoplasmic Mg²⁺, a condition that decreases protein synthesis. This control is exerted by PhoP, a transcriptional regulator activated in low cytoplasmic Mg²⁺ that governs proteostasis and is conserved in enteric bacteria. The uncovered mechanism enables bacteria to control the abundance of preexisting proteins.

INTRODUCTION

ClpAP and Lon are two of the five ATP-dependent proteases present in the bacterium Salmonella enterica serovar Typhimurium (1). These proteases play two critical roles: they preserve proteome quality by degrading nonfunctional and potentially toxic proteins, and they implement regulatory decisions by degrading functional proteins (1). The protease ClpAP consists of the chaperone ClpA, which unfolds substrates and shuffles them into a proteolytic chamber of the oligomeric peptidase ClpP (2). A subset of ClpAP substrates is recognized by the protease adaptor ClpS, which delivers them to ClpA (3, 4) and is degraded by ClpP (5, 6). In Lon, the unfoldase and peptidase activities are present in a single polypeptide (7, 8).

Protein acetylation is a major posttranslational modification that regulates protein function in all domains of life (9–12). In bacteria, proteins can be acetylated from acetyl coenzyme A (acetyl-CoA) by the protein acetyltransferase Pat (13). Pat-dependent acetylation is sensitive to the abundance of both acetyl-CoA (13) and acetyl-CoA binding proteins (14). Acetyl-CoA is synthesized from pyruvate after the tricarboxylic acid (TCA) cycle and from acetate through acetyl phosphate (13) (Fig. 1). Qad is an acetyl-CoA binding protein required for the Pat-dependent acetylation of HspQ but not of other investigated proteins (14). In many species of the family Enterobacteriaceae, the qad and hspQ genes are convergently transcribed (14). In addition, bacterial proteins can acetylate nonenzymatically from acetyl phosphate, and protein deacetylases can remove the acetyl group from such proteins (15).

FIG 1.

PhoP increases the specificity of the Lon and ClpSAP proteases by promoting HspQ acetylation. (Top) When bacteria experience low cytoplasmic Mg²⁺, the transcription regulator PhoP activates transcription of the qad gene. The enzyme Pat promotes HspQ acetylation using acetyl-CoA bound to the acetyl-CoA binding protein Qad as the acetyl donor. (Bottom) Acetylated HspQ escapes Lon-dependent degradation and does not stimulate degradation of another Lon substrate (red). Accumulated HspQ binds to ClpS and inhibits ClpS-dependent ClpAP proteolysis. The specificity of the Lon and ClpAP proteases increases upon the PhoP- and Qad-dependent acetylation of HspQ in bacteria experiencing low cytoplasmic Mg²⁺.

HspQ is a Lon substrate that, curiously, enhances proteolysis of a subset of Lon substrates, such as Hha (16). Acetylation of the HspQ protein increases the specificities of both Lon and ClpAP because acetylated HspQ is no longer a Lon substrate or enhancer of Hha proteolysis (14). In addition, acetylated and nonacetylated HspQ bind to ClpS equally well and act as antiadaptors for ClpS but have no effect on proteolysis of the ClpS-independent ClpAP substrate AcnB (14). Nutrients favoring acetyl-CoA accumulation further HspQ acetylation, which decreases proteolysis of both HspQ-enhancing Lon substrates and ClpS-dependent ClpAP substrates (14).

Proteolysis of specific Lon and ClpAP substrates is regulated by PhoP, a DNA binding protein activated by several signals, including low periplasmic Mg²⁺ (17, 18). First, PhoP promotes degradation of the Lon substrate H-NS by displacing the gene silencer H-NS from DNA, a necessary step because Lon does not degrade H-NS when bound to DNA (19). By decreasing H-NS amounts, PhoP promotes expression of horizontally acquired genes silenced by H-NS (19). Second, PhoP represses transcription of the clpS promoter (20) (Fig. 1). By reducing ClpS abundance, PhoP decreases degradation of ClpAP substrates exhibiting low affinity for ClpS but not of those tightly bound by ClpS (20). The PhoP-dependent reduction in ClpS abundance has pleiotropic effects because ClpS is required for degradation of several regulatory proteins (21) and also because the ClpSAP substrate Oat promotes antibiotic tolerance (20).

PhoP appears to decrease ClpS-dependent proteolysis by a mechanism(s) in addition to repression of clpS transcription because the steady-state levels of the ClpSAP substrates Oat and UvrY were lower in a phoP-null mutant than in a clpS promoter mutant refractory to PhoP repression (20). We now report that this mechanism entails PhoP promoting transcription of the qad gene (Fig. 1). By increasing Qad amounts, PhoP furthers HspQ acetylation, which protects HspQ from Lon, thereby decreasing proteolysis of both HspQ-stimulated substrates of Lon and ClpS-dependent substrates of ClpAP. The identified control is conserved in enteric species and promotes antibiotic tolerance when bacteria experience low cytoplasmic Mg²⁺. Our findings highlight the key role played by the PhoP protein and by cytoplasmic Mg²⁺ concentration in increasing protease specificity.

RESULTS

Deletion of the hspQ gene and prevention of PhoP repression of the clpS promoter have additive effects on ClpSAP substrate abundance.

To identify the PhoP-dependent mechanism(s) responsible for stabilization of ClpSAP substrates (other than PhoP repression of clpS transcription [20]), we examined the steady-state amounts of the ClpSAP substrate Oat-Flag produced from its normal promoter and chromosomal location in a set of isogenic strains. Oat-Flag was detected in wild-type S. Typhimurium but not in a phoP-null mutant when bacteria were grown in a defined medium of low Mg²⁺ for 6 h (Fig. 2A). An hspQ-null mutant and the clpS promoter mutant refractory to repression by PhoP (14, 20) had similarly small Oat-Flag amounts, albeit not as small as the phoP mutant (Fig. 2A). By contrast, the Oat-Flag protein was no longer detected in a double mutant with the clpS promoter mutation and with a deletion of the hspQ gene (Fig. 2A), mimicking the behavior of the phoP mutant (Fig. 2A). Given that PhoP repression of the clpS promoter decreases, but does not eliminate, clpS expression (20), these results suggest that HspQ inhibits the ClpS protein still produced under PhoP-inducing conditions. These results raise the possibility of PhoP stabilizing ClpSAP substrates by increasing HspQ abundance.

FIG 2.

PhoP stabilizes ClpSAP substrate Oat by increasing HspQ abundance. (A) Western blot analysis of crude extracts prepared from oat-Flag (JY655), oat-Flag hspQ (JY686), oat-Flag pclpS-d (JY684), oat-Flag hspQ pclpS-d (JY795), and oat-Flag phoP (JY656) S. Typhimurium following growth in N-minimal medium (pH 7.7) containing 10 μM MgCl₂ for 6 h. Samples were analyzed using antibodies directed to the Flag epitope and AtpB protein. (B) Western blot analysis of crude extracts prepared from hspQ-Flag (JY674), hspQ-Flag clpS (JY696), hspQ-Flag phoP (JY697), and hspQ-Flag clpS phoP (JY698) S. Typhimurium following growth as described for panel A. Samples were analyzed using antibodies directed to the Flag epitope and AtpB protein. (C) Western blot analysis of crude extracts prepared from hspQ-Flag (JY674), hspQ-Flag phoP (JY697), hspQ-Flag lon (JY703), and hspQ-Flag lon phoP (JY706) S. Typhimurium following growth as described for panel A. Samples were analyzed using antibodies directed to the Flag epitope and OmpA proteins. Numbers correspond to the average protein abundance relative to the leftmost lane from independent experiments. Protein abundance is rounded to two decimal places; 0.0 means <0.05. Data are representative of those from three independent experiments, which gave similar results.

Inactivation of the lon gene overcomes the defect in HspQ amounts exhibited by the phoP mutant.

We determined that HspQ-Flag amounts were smaller in a clpS-null mutant than in wild-type S. Typhimurium (Fig. 2B), in agreement with previous results and with the notion that ClpS protects HspQ from Lon (14). Unexpectedly, the phoP mutant also had smaller HspQ-Flag amounts than the wild-type strain (Fig. 2B). This was unexpected because a phoP-null mutant has more ClpS than the wild-type strain (20) and, as stated above, ClpS protects HspQ from Lon (14). HspQ-Flag amounts were similarly small in the phoP clpS double mutant and the phoP single mutant (Fig. 2B). Thus, PhoP increases HspQ amounts in a clpS-independent manner.

PhoP increases HspQ amounts at a posttranscriptional level for the following reasons. First, hspQ mRNA amounts were similar in isogenic wild-type and phoP strains (see Fig. S1A in the supplemental material). By contrast, the mRNA amounts of the PhoP-activated gene pagC, used as control, were markedly smaller in the phoP mutant than in the wild-type strain (Fig. S1B). Second, inactivation of the lon gene rescued the phoP mutant, restoring HspQ-Flag amounts to those of the lon single mutant (Fig. 2C). HspQ-Flag amounts in the lon mutant were larger than those of the wild-type strain (Fig. 2C), in agreement with previous results demonstrating that HspQ is a Lon substrate (14, 16). These results suggest that PhoP increases HspQ amounts by protecting HspQ from Lon.

PhoP is a direct transcriptional activator of the qad gene.

We reasoned that PhoP favors HspQ acetylation because acetylated HspQ is not a Lon substrate (14), and phoP is required for wild-type HspQ abundance (Fig. 2). Given that the protein acetyltransferase Pat acetylates HspQ from acetyl-CoA bound to the Qad protein (14), we investigated the possibility of phoP promoting expression of the qad or pat gene.

We have now determined that PhoP is a direct transcriptional activator of the qad gene for the following reasons. First, qad mRNA amounts were 4-fold larger in wild-type S. Typhimurium than in the phoP mutant following 6 h of incubation in low-Mg²⁺ medium (Fig. 3A). (For reasons explained in the following section, wild-type and phoP S. Typhimurium had similar qad mRNA amounts at 2 h in low-Mg²⁺ medium [Fig. 3A].) Second, a nucleotide sequence resembling a PhoP box (22) is present upstream of the qad coding region (Fig. 3B). Third, the purified phosphorylated-PhoP protein (herein referred to as PhoP-P) shifted DNA fragments corresponding to the qad promoter region (Fig. 3C) and to the PhoP-activated mgtC promoter (Fig. 3C), used as a positive control (22). The observed shifts appear to be specific because they were reversed by cold probes corresponding to the qad and mgtC promoter DNAs, respectively (Fig. 3C), and also because PhoP-P did not shift a DNA fragment harboring the promoter of the hspQ gene (Fig. 3C), which is not regulated by PhoP (Fig. S1A).

FIG 3.

PhoP directly promotes transcription of the qad gene in an mgtA- and mgtC-dependent manner. (A) mRNA abundance of the qad gene produced by wild-type (14028s) and phoP (MS7953s) S. Typhimurium following growth in N-minimal medium (pH 7.7) containing 10 μM MgCl₂ at 37°C for 2 or 6 h. mRNA abundance was normalized to that of the ompA gene. Primers used in qRT-PCR are listed in Table S2. Shown are the means and SD from four independent experiments. (B) Nucleotide sequence of the qad upstream region of the S. Typhimurium (strain 14028s) chromosome showing the nucleotide sequences of the qad start codon (blue box) and putative PhoP binding site (red box). The transcription start sites of the qad gene are indicated by green asterisks according to the S. Typhimurium RNA-seq database (http://bioinf.gen.tcd.ie/cgi-bin/salcom.pl?_HL) (36). (C) Electrophoretic mobility shift assay of DNA fragments harboring the S. Typhimurium qad, mgtC, and hspQ promoter regions with the indicated amounts of purified PhoP-P protein. The gel is representative of that of three independent experiments. Samples were incubated at room temperature for 20 min and then electrophoresed on 6% Tris-borate-EDTA gels at 100 V for 90 min. (D) mRNA abundance of the qad gene produced by wild-type (14028s), mgtC (EL4), mgtA (EG16735), and phoP (MS7953s) S. Typhimurium following growth in N-minimal medium (pH 7.7) containing 10 μM MgCl₂ at 37°C for 6 h. mRNA abundance was normalized to that of the ompA gene. Primers used in qRT-PCR are listed in Table S2. Shown are the means and SD from three independent experiments. Unpaired Student’s t tests for mRNA abundance were performed between the wild-type sample and the mutants. *, P < 0.05; **, P < 0.01.

By contrast, pat mRNA amounts were similar in isogenic wild-type and phoP S. Typhimurium following growth for 2 and 6 h in low-Mg²⁺ medium (Fig. S1C). In addition, wild-type and phoP S. Typhimurium harbor similar amounts of Pat-Flag protein expressed from its normal promoter and chromosomal location (Fig. S2). This was true whether bacteria were grown in low-Mg²⁺ (i.e., 10 μM) or high-Mg²⁺ (i.e., 10 mM) medium (Fig. S2). These results are in agreement with the similar pat mRNA abundance present in the two strains grown at the two Mg²⁺ concentrations (Fig. S1C). Thus, the differences in HspQ abundance exhibited by bacteria experiencing different Mg²⁺ concentrations are due to reasons other than Pat abundance.

Cumulatively, the results in this section suggest that PhoP increases HspQ amounts by promoting expression of qad, which specifies a protein required for HspQ acetylation (14).

Transcriptional activation of qad is mgtA and mgtC dependent.

Transcriptional control of certain PhoP-regulated genes requires amounts of PhoP-P achieved only after the PhoP-activated gene products MgtA and MgtC are made. MgtA enhances PhoP-P abundance by increasing activation of PhoP’s cognate sensor PhoQ in bacteria experiencing low Mg²⁺ (23), and MgtC hinders PhoP proteolysis by ClpSAP (24). The mRNAs corresponding to the mgtA and mgtC genes include unusually long leader sequences that promote transcription termination within the leader unless specific cytoplasmic conditions are met, such as a drop in cytoplasmic Mg²⁺ below a certain threshold, allowing transcription to proceed into the associated coding region (25–27). Thus, during growth in low-Mg²⁺ medium, transcription of mgtA- and mgtC-dependent genes is delayed until the cytoplasmic Mg²⁺ concentration decreases below a threshold that allows RNA polymerase to proceed into the mgtA and mgtC coding regions (25–27).

Because PhoP repression of clpS transcription is mgtA and mgtC dependent (20), we wondered whether this is also the case for PhoP activation of qad transcription. We established that qad mRNA amounts were smaller in both mgtA and mgtC single mutants than in wild-type S. Typhimurium (Fig. 3D), albeit not as small as in the phoP mutant (Fig. 3D), following 6 h in low-Mg²⁺ medium. Wild-type and phoP strains had similarly low qad mRNA amounts at 2 h in low-Mg²⁺ medium (Fig. 3A), a time before the MgtA and MgtC proteins are made because the cytoplasmic Mg²⁺ has not dropped below the critical threshold (23, 24). (In agreement with previous results [23, 24], the mRNA amounts of the control gene pagC were markedly smaller in the mgtA, mgtC, and phoP mutants than in the wild-type strain following 6 h in low-Mg²⁺ medium [Fig. S1D], and the mgtA mutation had a stronger effect than the mgtC mutation [Fig. S1D].) These data indicate that qad expression responds to the cytoplasmic Mg²⁺ signals governing synthesis of the MgtA and MgtC proteins.

Growth in low-Mg²⁺ medium enhances HspQ abundance in a qad-dependent manner.

The data presented above suggest the following model for changes in HspQ abundance that take place in wild-type S. Typhimurium experiencing low Mg²⁺ for an extended period: the MgtA and MgtC proteins further accumulation of PhoP-P to levels that promote qad transcription, resulting in increased qad mRNA amounts. The resulting Qad protein amounts enhance HspQ acetylation by Pat, which protects HspQ from proteolysis by Lon. Four independent sets of data provide experimental support for this model.

First, HspQ-Flag amounts were eight times larger in wild-type S. Typhimurium grown for 6 h in low-Mg²⁺ (i.e., 10 μM) than in high-Mg²⁺ (i.e., 10 mM) medium (Fig. 4A). This difference is qad dependent because it was not observed in a qad mutant (Fig. 4A). Second, when S. Typhimurium strains were grown in high-Mg²⁺ medium, HspQ-Flag amounts were 10-fold larger in wild-type S. Typhimurium harboring a plasmid in which the qad gene was transcribed from a heterologous promoter than in the isogenic strain with the plasmid vector control (Fig. 4B). By contrast, the two strains exhibited similarly high HspQ-Flag abundance when grown in low-Mg²⁺ medium (Fig. 4B), a condition in which the chromosomal qad gene is highly expressed from its PhoP-activated promoter (Fig. 3). Third, the qad-expressing plasmid restored HspQ-Flag abundance upon the phoP mutant reaching wild-type amounts, whereas the plasmid vector control did not (Fig. 4C). Fourth, inactivation of the lon gene overcame the defective HspQ-Flag amounts of the qad mutant (Fig. 4D), supporting the notion that qad protects HspQ from Lon (14). The lon qad double mutant had HspQ-Flag amounts similar to those of the lon single mutant and larger than those present in the wild-type strain (Fig. 4D). In sum, by promoting qad transcription in low-Mg²⁺ medium, PhoP protects HspQ from Lon.

FIG 4.

Qad protects HspQ from the Lon protease under low-Mg²⁺ conditions. (A) Western blot analysis of crude extracts prepared from hspQ-Flag (JY674) and hspQ-Flag qad (JY740) S. Typhimurium following growth in N-minimal medium (pH 7.7) containing 10 μM or 10 mM MgCl₂ for 6 h. Samples were analyzed using antibodies directed to the Flag epitope and AtpB proteins. (B) Western blot analysis of crude extracts prepared from hspQ-Flag (JY674) S. Typhimurium harboring the plasmid vector (pUHE21-2laci^q) or a plasmid expressing the qad gene following growth in N-minimal medium (pH 7.7) containing 10 μM or 10 mM MgCl₂ with 0.05 mM isopropyl-β-d-thiogalactopyranoside (IPTG) for 6 h. Samples were analyzed using antibodies directed to the Flag epitope and AtpB proteins. (C) Western blot analysis of crude extracts prepared from hspQ-Flag (JY674) and hspQ-Flag qad (JY740) S. Typhimurium harboring the plasmid vector (pUHE21-2laci^q) or a plasmid expressing the qad gene following growth in N-minimal medium (pH 7.7) containing 10 μM MgCl₂ and 0.2 mM IPTG for 6 h. Samples were analyzed using antibodies directed to the Flag epitope and GroEL proteins. Numbers correspond to the average protein abundances relative to the leftmost lane from independent experiments. Data are representative of those from three independent experiments, which gave similar results. (D) Western blot analysis of crude extracts prepared from hspQ-Flag (JY674), hspQ-Flag qad (JY740), hspQ-Flag lon (JY703), and hspQ-Flag qad lon (JY754) S. Typhimurium following growth in N-minimal medium (pH 7.7) containing 10 μM MgCl₂ for 6 h. Samples were analyzed using antibodies directed to the Flag epitope and AtpB proteins. Numbers correspond to the average protein abundance relative to the leftmost lane from independent experiments. Protein abundance rounded to two decimal places; 0.0 means <0.05. Data are representative of those from three independent experiments, which gave similar results.

Growth in low-Mg²⁺ medium increases the abundance of Lon and ClpSAP substrates.

We hypothesized that the increased qad expression taking place in low-Mg²⁺ medium inhibits proteolysis by Lon and ClpSAP because Qad is required for HspQ acetylation and acetylated HspQ no longer enhances degradation of the Lon substrate Hha (14) and also because the accumulated HspQ inhibits ClpS-dependent proteolysis by ClpAP (14).

In wild-type S. Typhimurium, the amounts of the Lon substrate His-Hha were nine times larger following growth for 6 h in low-Mg²⁺ (i.e., 10 μM) than in high-Mg²⁺ (i.e., 10 mM) medium (Fig. 5A). This difference is phoP and qad dependent because it was no longer observed in phoP and qad single mutants or a phoP qad double mutant (Fig. 5A). That the phoP mutation has a slightly stronger effect than the qad mutation may reflect that the phoP mutant accumulates large amounts of ATP, thereby stimulating proteolysis by ATP-dependent proteases (28). By contrast, the four strains exhibited similarly low His-Hha abundance when grown in high-Mg²⁺ medium (Fig. 5A), a condition that hinders qad expression because it does not activate the PhoP protein (29). The increased His-Hha abundance observed in low-Mg²⁺ medium (Fig. 5A) parallels the qad-dependent increase in HspQ-Flag amounts (Fig. 4A).

FIG 5.

Accumulation of HspQ in low-Mg²⁺ medium increases the abundance of Lon and ClpSAP substrates and promotes antibiotic persistence. (A) Western blot analysis of crude extracts prepared from oat-Flag (JY655), oat-Flag phoP (JY656), oat-Flag qad (JY774), and oat-Flag phoP qad (JYDN12) S. Typhimurium harboring a plasmid expressing the His-hha gene following growth in N-minimal medium (pH 7.7) containing 10 μM or 10 mM MgCl₂ for 6 h. Transcription of His-hha was induced with 0.1 mM IPTG. Samples were analyzed using antibodies directed to the Flag epitope, the His epitope, and AtpB protein. (B) Western blot analysis of crude extracts prepared from hspQ-Flag (JY674), hspQ-Flag lon (JY703), hspQ-Flag qad (JY740), and hspQ-Flag lon qad (JY754) S. Typhimurium harboring a plasmid expressing the His-hha gene following growth in N-minimal medium (pH 7.7) containing 10 μM or 10 mM MgCl₂ for 6 h. His-hha transcription was induced with 0.1 mM IPTG. Samples were analyzed using antibodies directed to the Flag epitope, the His epitope, and AtpB protein. (C) Western blot analysis of crude extracts prepared from oat-Flag hspQ-Flag (JY865), oat-Flag hspQ-Flag clpS (JY867), oat-Flag hspQ- Flag qad (JYDN10), and oat-Flag hspQ-Flag qad clpS (JYDN11) S. Typhimurium following growth in N-minimal medium (pH 7.7) containing 10 μM or 10 mM MgCl₂ for 6 h. Samples were analyzed using antibodies directed to the Flag epitope and AtpB protein. (D) Persister frequency by wild-type (14028s), qad (JYS7), pclpS-d (JY665), and qad pclpS-d (JYS8) S. Typhimurium was determined by the number of colonies on LB agar plates after treatment with kanamycin (100 μg/ml) for 2 h. Shown are the means and SD from three independent experiments. (E) Persister cell formation by wild-type (14028s), qad (JYS7), pclpS-d (JY665), and qad pclpS-d (JYS8) S. Typhimurium following growth in N-minimal medium (pH 7.7) containing 10 μM MgCl₂ at 37°C for 6 h and incubation with kanamycin (100 μg/ml) for 2 h. Shown are 10-fold serial dilutions plated on LB agar plates. In all Western blot data, numbers correspond to the average protein abundance relative to the leftmost lane from independent experiments and rounded to two decimal places. Data are representative of those from three independent experiments, which gave similar results.

Wild-type and qad strains harbored similar amounts of both His-Hha (Fig. 5A) and HspQ-Flag (Fig. 4B) when grown in high-Mg²⁺ medium, which makes sense given that qad is a PhoP-activated gene (Fig. 3) and that PhoP is not active in a high-Mg²⁺ environment (29). In addition, lon inactivation increased the abundance of both His-Hha (Fig. 5B) and HspQ-Flag (Fig. 5B) in the qad mutant, resulting in the lon qad double mutant having His-Hha amounts similar to those of the lon single mutant (Fig. 5B). His-Hha amounts were larger in the lon and lon qad mutants than those of the wild-type strain (Fig. 5B). Therefore, the Qad protein is required to inhibit Lon proteolysis of Hha and HspQ.

We determined that the amounts of the ClpSAP substrate Oat-Flag are nine times larger following growth for 6 h in low-Mg²⁺ (i.e., 10 μM) than in high-Mg²⁺ (i.e., 10 mM) medium (Fig. 5A). Oat-Flag amounts were smaller in the qad mutant than in the wild-type strain in low-Mg²⁺ medium (Fig. 5A), in agreement with the amounts of the ClpS inhibitor HspQ being smaller in a qad mutant than in the wild-type strain (Fig. 4A). Oat-Flag amounts were more reduced in the phoP and phoP qad mutants than in the qad mutant (Fig. 5A), in agreement with PhoP also repressing clpS expression (20) and having larger ATP amounts (28).

Following growth in high-Mg²⁺ medium, the four strains discussed in the previous paragraph exhibited similarly small Oat-Flag amounts (Fig. 5A). The reduced Oat-Flag amounts resulting from qad inactivation are due to increased amounts of free ClpS (i.e., not bound by HspQ) because Oat-Flag amounts were similarly large in clpS single and clpS qad double mutants (Fig. 5C). By contrast, the amounts of the Lon substrate HspQ-Flag present in the clpS qad double mutant were 10-fold smaller than those in the clpS⁺ qad⁺ strain (Fig. 5C). That HspQ-Flag amounts in the clpS and qad single mutants were 3- to 4-fold smaller than in the clpS⁺ qad⁺ strain (Fig. 5C) argues that both ClpS and Qad contribute to HspQ-Flag steady-state levels. Taken together, these findings show that Qad-mediated protection of the HspQ protein taking place in low-Mg²⁺ medium reduces proteolysis of Lon and ClpSAP substrates.

PhoP control of the qad gene is conserved in enteric bacteria.

We analyzed the DNA sequence upstream of the qad open reading frame in the genomes of several enteric species, looking for the presence of predicted PhoP boxes, which may be an indication that the PhoP control of qad transcription is conserved. We determined that the nucleotide sequences in the qad promoter region are highly conserved in S. enterica, Salmonella bongori, Enterobacter cloacae, and Escherichia coli (Fig. S3A). By contrast, the PhoP box is only partially conserved in the region upstream of the Yersinia pestis qad coding region (Fig. S3A).

As in S. Typhimurium (Fig. 3A), the mRNA amounts of the qad gene were smaller in an E. coli phoP mutant than in the wild-type strain following growth in low-Mg²⁺ medium (Fig. S3B); conversely, hspQ mRNA amounts were similar in isogenic wild-type and phoP E. coli (Fig. S3B), as in S. Typhimurium (Fig. S1A). The mRNA amounts of the control PhoP-activated gene mgtA were smaller in the phoP mutant than in E. coli in low-Mg²⁺ medium (Fig. S3B). These results indicate that the PhoP-mediated qad regulation mechanism is conserved in E. coli.

Qad promotes antibiotic persistence by stabilizing the ClpSAP substrate.

We established that the ClpSAP substrate Oat protein heightens the persistence to aminoglycoside antibiotics (20). In agreement with this notion and the results discussed above (Fig. 5C), we reasoned that Qad may increase antibiotic persistence in low-Mg²⁺ medium.

We determined that the clpS promoter mutant refractory to PhoP repression exhibits lower persistence to kanamycin than wild-type S. Typhimurium in low-Mg²⁺ (10 μM) medium (Fig. 5D and E). Likewise, qad inactivation decreased persistence (Fig. 5D and E). Persister cell formation in a qad clpS promoter double mutant was lower than in qad and clpS promoter single mutants (Fig. 5D and E), reflecting that Qad-mediated acetylation and ClpS reduction independently contribute to Oat abundance (Fig. 2A).

DISCUSSION

We have now identified low cytoplasmic Mg²⁺ as a condition that hinders the degradation of subsets of Lon and ClpAP substrates. Low cytoplasmic Mg²⁺ decreases proteolysis of the Lon substrate Hha (Fig. 5A) and of the ClpS-dependent substrate Oat (Fig. 5A) by preventing HspQ degradation by Lon (Fig. 5B). Low cytoplasmic Mg²⁺ promotes expression of the MgtA and MgtC proteins, which act as posttranslational activators of the transcriptional regulator PhoP (23, 24). The resulting enhanced abundance of active PhoP protein enables transcriptional activation of qad by PhoP-P (Fig. 3). The qad gene specifies an acetyl-CoA binding protein required for the Pat-mediated acetylation of HspQ (14). This chemical modification renders HspQ resistant to Lon proteolysis and unable to stimulate Hha proteolysis by Lon (14). In addition, the accumulated HspQ is an antiadaptor for ClpS (14), which hinders degradation of the ClpSAP substrate Oat (Fig. 5A and C), thereby increasing antibiotic persistence (Fig. 5E). Thus, by altering the abundance of acetylated HspQ, bacteria can modify the stability of specific protease substrates.

PhoP controls proteolysis in general and particular ways.

The transcriptional regulator PhoP is emerging as a key regulator of protein homeostasis by regulating both protein synthesis (30) and degradation (28). PhoP exerts systemic effects on ATP-dependent proteolysis by reducing the amounts of cytoplasmic ATP under slow growth conditions (28), such as those resulting when cytoplasmic Mg²⁺ drops below a certain threshold (30). Given that ATP molecules are required for proteases to unfold their substrates (1), the reduction in ATP concentration decreases the activity of multiple ATP-dependent proteases, allowing microbes to preserve functional proteins, thereby enabling a speedy return to a normal growth state once Mg²⁺ becomes available (28).

The S. Typhimurium PhoP protein also regulates proteolysis in specific ways. First, it promotes transcription of the adaptor protein MgtR, which is necessary for degradation of the MgtA (31), MgtB (32), and MgtC (33) proteins by the FtsH protease. Second, PhoP promotes expression of MgtU, a small protein that inhibits FtsH-mediated degradation of MgtB (32). Third, PhoP promotes transcription of IraP (34), an antagonist of MviA (referred to as RssB and SprE in E. coli), the adaptor that delivers the RpoS protein to the ClpXP protease for degradation (34, 35). (Transcription of the E. coli-specific iraM gene, which specifies an antiadaptor for RssB, is also PhoP dependent [35].) Fourth, PhoP represses expression of the clpS gene in both S. Typhimurium and E. coli, thereby reducing degradation of ClpS-dependent substrates of ClpAP (20). Last, we have now established that PhoP promotes expression of the qad gene, which reduces proteolysis by both ClpSAP and Lon by increasing HspQ acetylation (14) (Fig. 1).

Cytoplasmic conditions control protein acetylation from acetyl-CoA.

Protein acetylation is a major posttranslational modification that regulates protein function in all organisms (9–12). The acetyltransferase Pat promotes protein acetylation in the presence of acetyl-CoA as an acetyl donor. Given that S. Typhimurium Pat abundance is very low and constant under various conditions (36) (Fig. S3), acetyl-CoA accessibility would be a rate-limiting step for protein acetylation. Our study uncovered that expression of the acetyl-CoA binding protein Qad furthers HspQ acetylation under low-cytoplasmic-Mg²⁺ conditions (Fig. 4) by providing the acetyl group from acetyl-CoA bound to Qad to Pat (14).

Bacteria reduce general and specific ATP-dependent proteolysis when experiencing Mg²⁺ starvation.

When the cytoplasmic Mg²⁺ concentration drops below the threshold that compromises ribosome assembly, bacteria execute a gene expression program designed to reduce rRNA synthesis so that functional ribosomes can assemble and protein synthesis can continue, albeit at a much reduced rate (30). Under these conditions, bacteria preserve preexisting pools of functional proteins by reducing proteolysis by ATP-dependent proteases (28); By contrast, unfolded, misfolded, or truncated proteins are still degraded (28), thereby preserving the quality of the proteome. This gene expression program is directly controlled by PhoP, the master regulator of Mg²⁺ homeostasis (17, 18).

Specific changes in ATP-dependent proteolysis take place once the PhoP-activated MgtA and MgtC proteins are made and bacteria experience an increase in the amount of active PhoP protein (23, 24). First, transcription of the protease adaptor clpS gene is reduced, thereby decreasing proteolysis of a subset of ClpSAP substrates (20). Second, a transcriptional increase in the acetyl-CoA binding protein Qad furthers acetylation of HspQ (14), which reduces proteolysis of subsets of ClpSAP and Lon substrates (Fig. 1). These changes are required for long-term survival and virulence (24) as well as for a speedy return to a growth state when Mg²⁺ is no longer limited (28).

Regulation of protease specificity in enteric bacteria can give rise to phenotypic differences.

Divergence in the deduced amino acid sequences of the hspQ and gad genes or their expression may give rise to phenotypic differences among closely related bacteria (37). For instance, the lysine residue acetylated in the S. Typhimurium HspQ and critical for HspQ to regulate protease specificity is largely conserved in enteric bacteria but not in Yersinia pestis (14). In S. Typhimurium, PhoP binds to the qad (Fig. 3C) and clpS (20) promoters, regulating transcription of the corresponding genes in an MgtA- and MgtC-dependent manner (Fig. 3D). PhoP also controls the qad and clpS genes in E. coli (Fig. S3B) (20), which lacks MgtC (38). By contrast, the PhoP box upstream of the Y. pestis qad coding region shows only partial conservation (Fig. S3A), which may explain why qad mRNA amounts are similar in isogenic wild-type and phoP Y. pestis (39). Given that proteases control a wide variety of cellular processes (1), differences in the regulation of protease specificity may provide a source of phenotypic variation among closely related bacteria.

Proteases can interact to increase specificity.

One protease can alter the abundance of a different protease in response to specific stress conditions, thereby changing the activity of particular cellular pathways. For instance, Lon and ClpAP activate Lon by degrading the C terminus of this protease in Caulobacter crescentus (40). We have now established that low cytoplasmic Mg²⁺ increases the specificity of the very same two proteases in S. Typhimurium and E. coli by promoting both a decrease in the abundance of the adaptor ClpS (20) and an increase in the abundance of the anti-ClpS and Lon stimulator HspQ (Fig. 4A). The interaction between the HspQ and ClpS proteins hinders proteolysis of ClpS-dependent substrates of ClpAP (14) and protects HspQ from Lon (Fig. 2B and C). These effects are tuned, in turn, by metabolic conditions that determine the Qad- and Pat-dependent acetylation of HspQ because acetylated-HspQ and nonacetylated HspQ bind ClpS equally well, but only the nonacetylated form of HspQ is a Lon substrate and Lon-enhancing factor (14). Thus, by controlling degradation by ClpAP and Lon under stress conditions, bacteria modify pathways to maintain proteome quality and to adjust the amounts of specific proteins (40) (Fig. 1). The ability to control protease activity is likely important in clinical settings because Lon deficiency increases E. coli tolerance to trimethoprim concentrations below the MIC but increases susceptibility to high concentrations of this drug (41).

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Bacterial strains and plasmids used in this study are presented in Table S1 in the supplemental material. All S. Typhimurium strains are derived from strain 14028s (42) and were constructed by phage P22-mediated transductions as described previously (43). DNA oligonucleotides used in this study are presented in Table S2. Bacteria were grown at 37°C in Luria-Bertani broth (LB) and N-minimal media (pH 7.7) (44) supplemented with 0.1% Casamino Acids, 38 mM glycerol, and the desired concentrations of MgCl₂. E. coli DH5α was used as the host for preparation of plasmid DNA. Ampicillin was used at 50 μg/ml, kanamycin at 50 μg/ml, chloramphenicol at 25 μg/ml, and tetracycline at 12.5 μg/ml.

Construction of chromosomal mutants and plasmids.

Chromosomal mutants were constructed with the one-step disruption method (45), with minor modifications. To construct the pat-Flag (JYDN13) strain, a cat cassette was introduced into the pat gene as follows: a cat gene fragment was amplified from plasmid pKD3 using primer pair B3D5/B3D6 and then introduced into wild-type S. Typhimurium strain 14028s harboring plasmid pKD46.

Strain JY795 was made by transducing an hspQ::cat insertion into strain JY684 (20) using a P22 lysate generated in strain JY683.

Strains JY697, JY698, JY706, JYDN12, and JYDN14 were made by transducing the phoP::Tn10 (46) insertion into strains JY674, JY696, JY703, JY774, and JYDN13 using a P22 lysate generated in strain MS7953s, respectively.

Strain JY867 was made by transducing a clpS::cat insertion into strain JY865 (14) using a P22 lysate generated in strain JY570 (24).

Strains JYDN10 and JYDN11 were made by transducing a qad::kan insertion into strains JY865 and JY867 using a P22 lysate generated in strain JY889, respectively.

Strains JYS7 and JYS8 were made by transducing a qad::kan insertion into strains 14028s and JY665 (20) using a P22 lysate generated in strain JY889 (14), respectively. The resulting strains were transformed with pCP20 to remove the kan cassette.

Purification of the HspQ and PhoP proteins.

Purification of the PhoP protein was performed as described previously (20). Purification of HspQ was done as described previously (14).

Western blot assay.

Cells were grown in N-minimal medium containing 10 μM or 10 mM MgCl₂. Crude extracts were prepared in B-PER reagent (Pierce) with 100 μg/ml of lysozyme and EDTA-free protease inhibitor (Roche). Samples were loaded onto 4 to 12% NuPAGE gels (Life Technologies) or 4 to 15% TGX gels (Bio-Rad) and then transferred to a nitrocellulose membrane using the iBot machine (Life Technologies) or the Trans-Blot Turbo machine (Bio-Rad). Membranes were blocked with 3% skim milk solution at room temperature for 2 h. Then samples were analyzed using antihemagglutinin (anti-HA), anti-Flag, anti-HNS, anti-OmpA, anti-His, anti-GroEL, or anti-AtpB antibodies. Rabbit anti-HA (Sigma-Aldrich; H6908) and anti-Flag (Sigma-Aldrich; F7425) antibodies were used at a 1:2,000 dilution. Mouse anti-His6 (Abcam; ab18184) antibody was used at a 1:2,000 dilution. Rabbit anti-OmpA (LSbio; LS-C369146) was used as control at a 1:5,000 dilution. Mouse anti-AtpB (Abcam; ab110280) and anti-GroEL (Abcam; ab82592) were used as controls at a 1:5,000 dilution. Secondary horseradish peroxidase-conjugated anti-rabbit or anti-mouse antiserum (GE Healthcare) was used at a 1:5,000 dilution. The blots were developed with the Amersham ECL Western blotting detection reagents (GE Healthcare) or SuperSignal West Femto chemiluminescent system (Pierce). Numbers at the bottom of blot images correspond to average protein abundances relative to the leftmost lane from independent experiments.

Electrophoretic mobility shift assay.

DNA fragments corresponding to the qad, mgtC, and hspQ promoter regions were generated by PCR using S. Typhimurium strain 14028s DNA as the template and primer pairs 16930 and 16931, 15939 and 15940, and 16932 and 16933, respectively. PCR products were separated on a 1.0% agarose gel and purified with a QIAquick gel extraction kit (Qiagen). A total of 100 ng of DNA was labeled with T4 polynucleotide kinase (New England BioLabs) and [γ-³²P]ATP (Perkin Elmer). Unincorporated [γ-³²P]ATP was removed by using G-50 microcolumns (GE Healthcare). A total of 10 nmol of purified His-tagged PhoP (47) was phosphorylated in PB buffer (20 mM Tris HCl [pH 8.0], 2 mM MgCl₂, 180 mM KCl₂, 0.1 mM dithiothreitol [DTT], and 40 mM acetyl-P_i) for 1 h at 37°C. A total of 10 fmol of labeled DNA probes was mixed with various amounts of purified PhoP protein in binding buffer [20 mM Tris HCl (pH 8.0), 2 mM MgCl₂, 10 mM KCl₂, 10% (vol/vol) glycerol, 0.1 mM DTT, 60 μg/ml of bovine serum albumin (BSA), and 10 μg/ml of poly(dI-dC)] in a total volume of 20 μl. Following 20 min of incubation at room temperature, samples were electrophoresed on 6% Tris-borate-EDTA gels (Life Technologies) at 100 V for 90 min.

Quantitative RT-PCR.

To measure mRNA abundance, cells were grown in N-minimal medium containing 10 μM MgCl₂ at 37°C for the desired times. Total RNA was purified by using RNeasy kit (Qiagen) with on-column DNase treatment, and cDNA was synthesized by using VILO Super Mix (Life Technologies). Quantification of transcripts was carried out by quantitative real-time-PCR (qRT-PCR) using SYBR green PCR master mix (Applied Biosystems) in a QuantStudio 6 Flex real-time PCR system (Applied Biosystems). The mRNA abundance was determined by using a standard curve obtained from PCR products generated with serially diluted genomic DNA, and results were normalized to the levels of ompA. Data shown are averages from at least three independent experiments. Primers used in quantitative RT-PCR assay are presented in Table S2.

Antibiotic persistence assay.

Salmonella cells were grown in N-minimal medium containing 10 μM or 10 mM MgCl₂ for 6 h at 37°C. Then kanamycin (100 μg/ml) was added and incubation was continued for 2 h. Cells were washed with fresh N-minimal medium containing 10 μM or 10 mM MgCl₂. Persister cell formation was determined by spotting 10-fold serial dilutions onto LB agar plates. Plates were then incubated at 37°C for 15 h.