ppGpp negatively impacts ribosome assembly affecting growth and antimicrobial tolerance in Gram-positive bacteria

Significance

When bacteria encounter stresses such as nutrient deprivation, they react by switching on the stringent response, the effects of which are mediated by two nucleotides collectively referred to as (p)ppGpp. These nucleotides function by binding to target proteins, leading to bacterial cells shutting down active growth and entering a state that promotes survival. In Staphylococcus aureus, relatively little is known about the target proteins with which these nucleotides interact. In this work, a genome-wide nucleotide–protein interaction screen was used to identify protein targets of (p)ppGpp to fully establish the pathways these nucleotides control in Gram-positive bacteria. In doing so, we identify several previously unknown targets with roles in ribosomal assembly, cell growth, and antimicrobial tolerance.

Abstract

The stringent response is a survival mechanism used by bacteria to deal with stress. It is coordinated by the nucleotides guanosine tetraphosphate and pentaphosphate [(p)ppGpp], which interact with target proteins to promote bacterial survival. Although this response has been well characterized in proteobacteria, very little is known about the effectors of this signaling system in Gram-positive species. Here, we report on the identification of seven target proteins for the stringent response nucleotides in the Gram-positive bacterium Staphylococcus aureus. We demonstrate that the GTP synthesis enzymes HprT and Gmk bind with a high affinity, leading to an inhibition of GTP production. In addition, we identified five putative GTPases—RsgA, RbgA, Era, HflX, and ObgE—as (p)ppGpp target proteins. We show that RsgA, RbgA, Era, and HflX are functional GTPases and that their activity is promoted in the presence of ribosomes but strongly inhibited by the stringent response nucleotides. By characterizing the function of RsgA in vivo, we ascertain that this protein is involved in ribosome assembly, with an rsgA deletion strain, or a strain inactivated for GTPase activity, displaying decreased growth, a decrease in the amount of mature 70S ribosomes, and an increased level of tolerance to antimicrobials. We additionally demonstrate that the interaction of ppGpp with cellular GTPases is not unique to the staphylococci, as homologs from Bacillus subtilis and Enterococcus faecalis retain this ability. Taken together, this study reveals ribosome inactivation as a previously unidentified mechanism through which the stringent response functions in Gram-positive bacteria.

The stringent response is a complex mechanism used by all bacteria to deal with cell stresses including amino acid deprivation, carbon source starvation, fatty acid depletion, and osmotic stress (1–3). This response, first characterized over 40 years ago, is coordinated by the rapid synthesis of the nucleotides guanosine tetraphosphate and pentaphosphate, collectively termed (p)ppGpp (2). Once produced, these alarmones are responsible for controlling a cellular switch, resulting in the down-regulation of active growth and an up-regulation of genes involved in the stress response (4). Additionally, these nucleotides have been shown to be vital for controlling the transition of bacteria into stationary phase, biofilm formation, sporulation, virulence, antibiotic tolerance, and persister cell formation (5–9).

In proteobacteria, it has long been established that, after enduring stress, (p)ppGpp is synthesized by both the monofunctional synthetase enzyme RelA and the bifunctional synthetase SpoT, a protein that also contains (p)ppGpp hydrolase activity (4). RelA associates with ribosomes, and synthetase activity is triggered upon an accumulation of uncharged tRNA sensed by the ribosome during amino acid depletion (10–13). The synthetase activity of SpoT, on the other hand, is induced by other stresses such as fatty acid depletion (1). Once synthesized, the major effect of (p)ppGpp production is an alteration in gene transcription, where stable RNAs (rRNA and tRNA), as well as cell proliferation genes, are down-regulated, and genes involved in the stress and starvation response are up-regulated (4). In Gram-negative bacteria, this transcriptional shift is mediated by (p)ppGpp interacting with the RNA polymerase (RNAP), which in combination with the transcription factor DksA, modulates gene expression on a transcriptional level (3, 14). Aside from the RNAP, there are at least 15 other direct (p)ppGpp target proteins in Escherichia coli, such as the translation elongation factors EFG and EF-Tu, the DNA primase DnaG, and a number of amino acid decarboxylases that are involved in the acid stress response (15–17). These nucleotides also play major roles in controlling bacterial persistence in Gram-negative bacteria by activating toxin–antitoxin systems and triggering slow growth (9), leading to cells that persist in the host following antibiotic treatment.

In Staphylococcus aureus, as well as other Gram-positive species, (p)ppGpp is synthesized by RSH, a bifunctional RelA/SpoT homolog that contains both a synthetase and hydrolase domain (18, 19). The genome of S. aureus also encodes two other monofunctional synthetases, RelP and RelQ, and transcription of these genes increases when cells are exposed to cell wall-targeting antimicrobials (20, 21). Recent work on S. aureus has shown that the ability to switch on the stringent response is essential for its virulence and is required for the organism to cause chronic infections (22–25).

In contrast to the situation in proteobacteria, very little is known about the binding targets for (p)ppGpp in Gram-positive species. These nucleotides do not interact with the RNAP (26), and few direct binding proteins have been identified. It has been established that the depletion of cellular GTP, the substrate for (p)ppGpp-synthesizing enzymes, plays a significant role in initiating the stringent response in these organisms (27). Decreased GTP levels lead to a decrease in the transcription of mRNAs with a GTP-initiating nucleotide, which in Gram-positive bacteria includes most rRNA promoters (26). Aside from substrate depletion, (p)ppGpp also actively inhibit GTP synthesis in Bacillus subtilis and Enterococcus faecalis by blocking the functions of the hypoxanthine-guanine phosphoribosyltransferase HprT and the guanylate kinase Gmk, two enzymes involved in the GTP synthesis pathway (28–30). GTP levels are also important in some species for the activation of CodY, a global transcriptional regulator. In a GTP-bound state, CodY binds to DNA and represses the transcription of a number of genes involved in the adaptation to nutrient limitation. However, upon entry of cells in stationary phase, GTP levels decrease, leading to the release of CodY from DNA, de-repression, and transcription of target genes (31). Intracellular GTP levels do therefore play a significant role in modulating the stringent response. However, given the identification of multiple (p)ppGpp-binding proteins in E. coli, it seems unlikely that GTP homeostasis is the sole regulatory function for (p)ppGpp in Gram-positive species.

In this study, we used a genome-wide nucleotide–protein interaction screen to identify previously unidentified targets for the stringent response nucleotides (p)ppGpp in S. aureus. In addition to confirming that both HprT and Gmk from S. aureus can interact with these nucleotides, we demonstrate that (p)ppGpp bind with high affinity and specificity to five putative GTPases—RsgA, RbgA, Era, HflX, and ObgE—implicated in ribosome assembly. Characterization of RsgA, RbgA, Era, and HflX revealed that their GTPase activity is increased in the presence of ribosomes but inhibited by the stringent response nucleotides. With this, we identify an additional mechanism by which the stringent response alarmones can control cell proliferation in Gram-positive bacteria at a posttranscriptional level by actively interfering with ribosome assembly to inhibit cell growth and promote antimicrobial tolerance.

Results

Identification of (p)ppGpp-Binding Proteins Using a Genome-Wide Nucleotide–Protein Interaction Screen.

Our previous work demonstrated the use of a differential radial capillary action of ligand assay (DRaCALA)-based ORFeome screen as a high-throughput platform for identifying interaction partners for c-di-AMP (32). To adapt this screen to identify (p)ppGpp-binding proteins, radiolabeled (p)ppGpp was synthesized (Fig. S1 A–C) and used in combination with an S. aureus protein expression library that contains 2,343 ORFs from the genome of the S. aureus strain COL (85.5% of the total number of ORFs in the genome) fused to a His-MBP-tag and expressed in E. coli. To perform the genome-wide screen, the S. aureus His-MBP protein expression library strains were grown up, protein expression was induced, and crude whole-cell extracts were prepared. These lysates were arrayed in a 96-well format and used in DRaCALAs with a 1:1 mix of radiolabeled pppGpp:ppGpp (Fig. S1D). An average fraction bound for each plate was calculated as described by Roelofs et al. (33), and positive interactions were deemed as being 2.4 times greater than the background. This led to the identification of seven putative (p)ppGpp target proteins.

Fig. S1.

Synthesis of ³²P-(p)ppGpp and identification of (p)ppGpp binding proteins. (A) Schematic detailing the synthesis of radiolabeled pppGpp and ppGpp as described in Methods. (B) TLC displaying the conversion of ³²P-GTP to ³²P-pppGpp by Relseq, followed by the hydrolysis of ³²P-pppGpp to ³²P-ppGpp by the phosphatase GppA. (C) DRaCALA binding assay showing the incubation of E. coli whole-cell lysates containing an empty vector (EV) or a strain overexpressing RelA with ³²P-ppGpp and ³²P-pppGpp. (D) Genome-wide nucleotide–protein interaction screen. ³²P-labeled (p)ppGpp was incubated with E. coli lysates and spotted onto nitrocellulose membrane. Well B7 (box) was spiked with pET28b-relA as a positive control. Well F8 (circle) shows a (p)ppGpp interaction with HprT. (E) Coomassie-stained gel of purified recombinant target proteins. Ten micrograms of each protein were run on a 7.5% (vol/vol) PAA-gel, and proteins were visualized by staining with Coomassie brilliant blue. Sizes (in kilodaltons) of protein standards run in parallel are indicated on the Left of the panel.

To interrogate the binding further, the plasmid from each of the seven strains was sequenced to confirm the identity of each gene and retransformed into E. coli cells. Protein expression was once again induced, whole-cell lysates were prepared, and binding to both pppGpp and ppGpp was examined (Fig. 1A). Fraction bound values higher than the empty vector negative control were observed for all seven strains, indicating positive binding interactions. Following this, all seven His-MBP–fused proteins were purified by Ni²⁺-affinity and size exclusion chromatography (Fig. S1E), and the recombinant proteins were used in DRaCALAs with radiolabeled (p)ppGpp (Fig. 1B). Binding assays confirmed positive interactions for six of the target proteins, namely HprT, Gmk, RsgA, RbgA, Era, and HflX. The binding to YqeH, however, was very weak, preventing a determination of binding affinity. For this reason, this protein was not investigated further.

Fig. 1.

Confirmation of the interactions between (p)ppGpp and target proteins. (A) DRaCALA with ³²P-labeled (p)ppGpp and whole-cell lysates prepared from E. coli strains overexpressing the different target proteins. (B) DRaCALA with purified recombinant proteins and ³²P-labeled (p)ppGpp. All experiments were carried out in quadruplicate with the data plotted using the GraphPad Prism software.

(p)ppGpp Bind Specifically to HprT and Gmk from S. aureus to Inhibit Their Function.

Of the six identified (p)ppGpp-binding proteins, two have previously been shown to interact with these nucleotides, namely Gmk and HprT, two proteins involved in GTP synthesis. Gmk is the enzyme responsible for the conversion of GMP to GDP during de novo synthesis of GTP, whereas HprT is involved in the salvage pathway, converting both hypoxanthine to IMP and guanine to GMP. The activities of these enzymes from both B. subtilis and E. faecalis, as well as Gmk from S. aureus (Gmk_SA), have been shown to be inhibited in the presence of (p)ppGpp, thus lowering intracellular GTP levels to a range that supports survival during starvation (28–30).

Using the purified S. aureus Gmk and HprT proteins in DRaCALAs, we show here that both proteins have stronger affinities for ppGpp and pppGpp over GTP (Fig. S2 A and B, and Table S1). Additionally, it was noted that these interactions are specific as only an excess of cold unlabeled ppGpp but not any of the other nucleotides tested could compete for binding with labeled ppGpp or pppGpp (Fig. S2 C and D). To examine whether (p)ppGpp can directly inhibit the function of the staphylococcal HprT enzyme, and to confirm that Gmk_SA can be inhibited, the enzymatic activities of both proteins were monitored in the presence of both ppGpp and pppGpp. Enzymatic assays monitoring the conversion of guanine to GMP by HprT or GMP to GDP by Gmk were set up as previously described (28). This analysis revealed that the HprT and Gmk enzymes from S. aureus are inhibited by both ppGpp and pppGpp (Fig. S2 E and F).

Fig. S2.

Characterization of the interactions between (p)ppGpp and HprT and Gmk by DRaCALA. (A and B) Binding curves and K_d determination for radiolabeled pppGpp, ppGpp, and GTP with purified (A) HprT_SA and (B) Gmk_SA. K_d values were determined from the curves as previously described (33). (C and D) DRaCALA with purified (C) HprT_SA and (D) Gmk_SA and ³²P-labeled (p)ppGpp with an excess of cold competitor nucleotide as indicated. All experiments were carried out in quadruplicate. (E) HprT activity assay. The conversion of GMP to guanine by HprT was monitored over time at an absorbance of 257 nm in the presence of no inhibitor or with 100 μM ppGpp or pppGpp. (F) Gmk activity assay. The conversion of NADH to NAD⁺, in a Gmk-dependent manner, was monitored over time at an absorbance of 340 nm in the presence of no inhibitor or with 100 μM ppGpp or pppGpp. Activity assays were performed four times. Mean values and SDs are shown.

Table S1.

Binding affinities (in micromolar concentration)

	ppGpp		pppGpp		GTP
Protein	Kd	Bmax	Kd	Bmax	Kd	Bmax
HprT	0.37 ± 0.05	0.70 ± 0.02	0.75 ± 0.06	0.75 ± 0.01	6.33 ± 2.10	0.37 ± 0.04
Gmk	6.02 ± 1.50	0.22 ± 0.02	4.01 ± 1.20	0.28 ± 0.02	5.83 ± 3.13	0.08 ± 0.01
RsgA	1.69 ± 0.13	0.52 ± 0.01	10.31 ± 2.59	0.28 ± 0.03	2.73 ± 0.28	0.48 ± 0.01
RbgA	0.49 ± 0.06	0.45 ± 0.01	2.65 ± 0.45	0.25 ± 0.01	5.56 ± 1.88	0.23 ± 0.03
Era	0.95 ± 0.06	0.60 ± 0.01	3.85 ± 0.56	0.37 ± 0.02	2.95 ± 0.41	0.43 ± 0.02
HflX	0.87 ± 0.15	0.67 ± 0.03	13.63 ± 1.57	0.49 ± 0.02	19.39 ± 5.11	0.41 ± 0.05

RsgA, RbgA, Era, and HflX Are Putative GTPases Involved in Ribosomal Biogenesis.

Of the four remaining putative (p)ppGpp-binding proteins identified from the S. aureus strain COL Gateway Clone Set, RsgA is annotated as a hypothetical protein that has 33% identity over 89% of the protein to the E. coli ribosome small-subunit–dependent GTPase A. RbgA and HflX are described as putative GTP-binding proteins. RbgA, although not present in E. coli and other γ-proteobacteria, shows 55% identity over 94% of the protein to the ribosome biogenesis GTPase A from B. subtilis, and HflX has 43% identity over 81% of the protein to the high-frequency lysogenization locus X GTPase from E. coli. Finally, Era is annotated as a GTP-binding protein that has 40% identity over 97% of its length to the E. coli Ras-like protein from E. coli.

Little is known about the functions of these proteins in S. aureus. RsgA from S. aureus, E. coli, and B. subtilis is a nonessential protein that is nonetheless important for normal growth (34–36), whereas both RbgA and Era are essential (37–42). Unlike eukaryotic GTPases that have roles in membrane signaling, members of this family of prokaryotic GTPases appear to have functions linked to ribosome assembly. In E. coli, it has been demonstrated that Era and RsgA bind to the 30S subunit of the ribosome and are critical for 30S ribosomal subunit biogenesis (36, 43, 44). Cryo-electron micrograph images of both proteins in complex with the 30S subunit suggest a chaperoning role, where they may prevent premature association of the 30S with the 50S subunit presumably until the 30S subunit has fully matured (44, 45). Indeed, depletion of these proteins in bacterial cells leads to a decrease in 70S ribosomes with a buildup of 50S and 30S subunits (36, 43, 46). In contrast, both RbgA and HflX have been shown to bind to the 50S subunit and are required for its biogenesis, as cells depleted for RbgA show a reduction in 70S ribosomes, whereas free 50S subunits are completely missing (40, 42, 47–49). HflX has also been implicated as a ribosome-splitting factor, involved in rescuing stalled ribosomes during stress (50).

(p)ppGpp Binds Specifically to the Four Target Proteins, RsgA, RbgA, Era, and HflX.

To determine binding kinetics and interaction specificities between (p)ppGpp and the four putative GTPases, DRaCALAs were performed with the purified proteins. Binding affinities in the low micromolar range were established for all of the proteins and ppGpp (Fig. S3 and Table S1). With the exception of RsgA, the affinities of all four proteins to pppGpp and GTP were 4–16 times weaker, indicating that ppGpp may be a more potent effector in S. aureus than pppGpp (Fig. S3 and Table S1). RsgA, on the other hand, bound ppGpp and GTP with similar affinities, suggesting that the occupancy of the binding site with either ligand is going to depend heavily on the intracellular nucleotide concentration at any given time during the growth cycle. Additionally, it was determined that the interactions between each of these proteins and ppGpp are specific as only an excess of cold unlabeled ppGpp, but not any of the other nucleotides tested, could completely compete for binding with labeled ppGpp (Fig. S4).

Fig. S3.

Characterization of the interactions between (p)ppGpp and RsgA, RbgA, Era, and HflX by DRaCALA. (A–D) Binding curves and K_d determination for radiolabeled pppGpp, ppGpp, and GTP with purified (A) RsgA, (B) RbgA, (C) Era, and (D) HflX. K_d values were determined from the curves as previously described (33). All experiments were carried out in quadruplicate. The data were plotted, and the best-fit lines were determined by nonlinear regression using the GraphPad Prism software.

Fig. S4.

Characterization of the interactions between ppGpp and RsgA, RbgA, Era, and HflX by DRaCALA. (A–D) DRaCALAs with purified (A) RsgA, (B) RbgA, (C) Era, and (D) HflX and ³²P-labeled ppGpp with an excess of cold competitor nucleotide as indicated. All experiments were carried out in quadruplicate. The data were plotted using the GraphPad Prism software.

RsgA, RbgA, Era, and HflX Are GTPases, the Activities of Which Are Inhibited by (p)ppGpp.

To examine whether these four proteins function as GTPases, the proteins were incubated with radiolabeled GTP and the hydrolysis to GDP monitored by TLC. Although the control protein MBP was unable to hydrolyze GTP even after overnight incubation, all four (p)ppGpp-binding proteins hydrolyzed GTP, however to varying degrees (Fig. 2A). As RsgA was able to fully hydrolyze GTP upon overnight incubation, a time course was performed with the enzyme, revealing that full hydrolysis of GTP to GDP occurred within 20 min (Fig. 2B). Previous work on RsgA from E. coli reported that the activity of the protein is increased in the presence of ribosomes (36, 51). To determine the effect of ribosomes on the activity of all four GTPases, 70S ribosomes were purified from the community-acquired methicillin-resistant S. aureus (CA-MRSA) strain LAC* and included in the GTP hydrolysis assays. Although only a slight increase in the enzymatic activity was observed for RsgA, a dramatic increase in activity was noted for RbgA, Era, and HflX (Fig. 2C), indicating that these proteins are indeed all intracellular GTPases, the activities of which are stimulated in the presence of the ribosome.

Fig. 2.

GTPase activity assays in the presence or absence of ribosomes and (p)ppGpp. (A) The GTPase activity of recombinant RsgA, RbgA, Era, and HflX were determined by incubating 10 µM protein with α-³²P-GTP overnight at 37 °C. Hydrolysis was monitored by TLC, and the percentage GDP formed was quantified using ImageJ and values were plotted using GraphPad Prism. (B) The enzymatic activity of RsgA was monitored as above with samples withdrawn over a 1-h period. (C) Quantification of GTPase activity in the presence of 70S ribosomes. GTPase assays were set up as above in the absence or presence of 70S ribosomes. Reactions with RsgA were stopped after 10 min, whereas reactions with RbgA, Era, and HflX were incubated for 60 min. (D) Analysis of GTPase activity of all four target proteins in the presence of (p)ppGpp. GTP hydrolysis in the presence of 70S ribosomes was monitored in the presence of either 1 mM ppGpp or pppGpp. Reactions with RsgA were stopped after 10 min, whereas reactions with RbgA, Era, and HflX were incubated for 60 min. (E) The activity of RsgA was monitored in the presence of increasing concentrations of ppGpp and pppGpp. Reactions were stopped after 10 min and analyzed by TLC. (F) Quantification of the GTPase activity of RsgA in the presence of (p)ppGpp. The enzyme reactions were set up as in E, and the percentage GDP formed was quantified using ImageJ. The data were fitted using a dose–response inhibition algorithm in GraphPad Prism with the corresponding IC₅₀ value given in the text. All experiments were performed in triplicate, and averages and SDs were plotted using GraphPad Prism.

Next, to determine the effect of (p)ppGpp on the enzymatic function, hydrolysis assays were performed in the presence of ribosomes and either 1 mM ppGpp or pppGpp. Interestingly, the hydrolysis activity of all four GTPases was significantly inhibited in the presence of either one of the stringent response nucleotides (Fig. 2D). To examine this in more detail, the activity of RsgA in the presence of increasing amounts of ppGpp or pppGpp was monitored by TLC, revealing an IC₅₀ of 56.8 ± 8.23 μM for ppGpp and 151 ± 18.9 μM for pppGpp (Fig. 2 E and F). During stringent response activation, the levels of (p)ppGpp in the bacterial cell rise to 1–2 mM (28, 52), levels that are more than sufficient to inhibit the functions of these enzymes. Altogether, these data reveal that RsgA, RbgA, Era, and HflX function as GTPases, the activities of which increase upon association with the ribosome and are inhibited upon interaction with (p)ppGpp. These data further suggest that, upon induction of the stringent response, where cellular levels of (p)ppGpp increase to 1–2 mM and levels of GTP fall, the activities of these enzymes are inhibited, which could affect the assembly of functional ribosomes.

The Absence of RsgA, or the Inhibition of Its GTPase Activity, Reduces Intracellular Levels of 70S Ribosomes and Slows the Growth of S. aureus.

We next wanted to examine the contribution of ribosomal GTPases to the growth and viability of S. aureus. Both rbgA and era are essential genes in this organism, and so the construction of deletion mutants was not possible. RsgA and HflX, on the other hand, are encoded by nonessential genes, and S. aureus strains with in-frame deletions in these genes were constructed in the CA-MRSA background strain LAC*. Although no growth defect was observed under the conditions tested for the hflX mutant strain, the rsgA mutant strain grew significantly slower than the wild-type LAC* (Fig. 3A). This growth defect could be complemented fully by the introduction of a plasmid with the rsgA gene expressed under anhydrotetracycline-inducible control (Fig. 3A). As RsgA is reported to be involved in ensuring 30S subunit maturation before binding the 50S subunit (45), we sought to examine the effect of its absence on the ribosomal content in S. aureus. To this end, the ribosomal profiles from extracts of the wild-type strain LAC*, the rsgA mutant, and the complemented strain were evaluated by sucrose density gradient centrifugation (Fig. 3 B–D). These profiles revealed that, in the absence of RsgA, cells contained reduced levels of intact 70S ribosomes, with a concomitant buildup of 50S and 30S subunits, which is in agreement with previous observations that RsgA has a role in ribosomal subunit association.

Fig. 3.

Deletion of rsgA negatively affects the growth and ribosomal composition of S. aureus. (A) Growth of S. aureus strains LAC* iTET, LAC*ΔrsgA iTET, LAC*ΔrsgA iTET-rsgA, and LAC*ΔrsgA iTET-rsgA T199A. Overnight cultures grown in the presence of 100 ng/mL Atet were diluted to an OD₆₀₀ of 0.01 (time = 0 h) and grown in the presence of Atet for 8 h. Growth curves were performed three times, and average OD₆₀₀ readings and SDs were plotted. (B–D) Effect of rsgA deletion on ribosomal profiles. Extracts from wild-type LAC* iTET (B), LAC*ΔrsgA iTET (C), and LAC*ΔrsgA iTET-rsgA (D) grown to exponential phase were fractionated by sucrose density gradient centrifugation. Gradients were fractionated by upward displacement and analyzed for RNA content by measuring the absorbance at 260 nm. Experiments were performed in triplicate with one representative graph shown.

As reported above, binding of (p)ppGpp inhibits the GTPase activity of RsgA, and as shown here a decrease in the number of mature ribosomes is observed in S. aureus in its absence. To determine whether inhibiting the GTPase activity of RsgA alone is sufficient to cause this phenotype, the nucleotides encoding for a threonine residue at position 199 in the switch 1 region of the GTPase domain of RsgA were mutated to encode for an alanine to abolish GTPase activity. This protein variant should have reduced GTPase activity and hence mimic a protein in which the GTPase activity has been inhibited by (p)ppGpp. This variant was expressed and purified from E. coli cells. Nucleotide-binding assays showed that ppGpp could still interact with this protein variant, indicating that it is not essential for ppGpp binding; however, GTP binding was, as expected, severely diminished (Fig. 4A). In agreement with the decrease in GTP binding, the GTPase activity of the protein was drastically reduced, even in the presence of ribosomes (Fig. 4B). This rsgA T199A allele was then also introduced on a complementing plasmid into the rsgA mutant strain LAC*ΔrsgA, creating strain LAC*ΔrsgA iTET-rsgA T199A. Monitoring the growth of this strain revealed a significant defect, similar to that of the rsgA mutant, confirming that inactivation of the GTPase function of this protein results in a slower growth phenotype (Fig. 3A). Next, the ribosomal profile for strain LAC*ΔrsgA iTET-rsgA T199A was determined by sucrose density gradient centrifugation, revealing that there is, similar to the rsgA mutant, a reduction in the amount of mature 70S ribosomes (Fig. 4C). Together, these results suggest that, in the absence of RsgA, or upon inactivation of its GTPase activity, the maturation of ribosomes is severely affected.

Fig. 4.

GTPase activity of RsgA is crucial for its function. (A) DRaCALA with purified recombinant MBP, MBP-RsgA (RsgA), or MPB-RsgA-T199A (T199A) protein and ³²P-labeled GTP and ppGpp. All experiments were carried out in triplicate. The data were plotted using the GraphPad Prism software. (B) The GTPase activity of recombinant RsgA and the T199A variant were analyzed in the absence or presence of 70S ribosomes. Hydrolysis was monitored by TLC, the percentage GDP formed was quantified using ImageJ, and values were plotted using GraphPad Prism. (C) Effect of inactivation of GTPase activity on ribosomal profiles. Strain LAC*ΔrsgA iTET-rsgA T199A was fractionated by sucrose density gradient centrifugation and analyzed for RNA content by measuring the absorbance at 260 nm. (D) Effect of (p)ppGpp production on ribosomal profiles. The production of (p)ppGpp was induced by the addition of 0.05 or 60 µg/mL mupirocin to exponentially grown cultures. Thirty minutes post induction, cells were harvested and extracts were analyzed by sucrose density gradient centrifugation. Peaks corresponding to 70S, 50S, and 30S are highlighted in green, orange, and blue, respectively. Experiments were performed in triplicate with one representative graph shown.

It is known that induction of the stringent response causes a reduction in the overall quantity of ribosomes being produced due to ppGpp-mediated decreases in rRNA transcripts (4, 53). Our previous observations led us to suspect that, upon synthesis of (p)ppGpp, this nucleotide would also bind to the four ribosomal GTPases and inhibit their activity, resulting in a decrease in intact 70S ribosomes. To examine what effect (p)ppGpp synthesis has on the ribosomal profile of wild-type staphylococcal cells, the synthesis of (p)ppGpp was triggered by the addition of a low (0.05 μg/mL) or a high (60 μg/mL) dose of mupirocin for 30 min, conditions known to induce the stringent response, and extracts were analyzed by sucrose density gradient centrifugation. As expected, the overall level of ribosomes in the cell was decreased compared with wild type (Fig. 4D). In addition, the ratio of intact 70S to 50S and 30S subunits altered from 1/0.59/0.29 for the wild type to 1/0.83/0.44 with the addition of 0.05 μg/mL mupirocin and to 1/0.92/0.36 in the presence of high levels of mupirocin, revealing that the levels of 70S ribosomes were indeed decreased after induction of the stringent response compared with untreated cells grown in the absence of mupirocin (Fig. 4D). Altogether, these data lead us to propose a role for (p)ppGpp in binding to intracellular GTPases to inhibit ribosomal assembly and promote slow growth.

Inhibition of GTPase Activity Leads to Increased Tolerance to Antimicrobials.

It has been reported that bacterial cultures naturally contain subpopulations of slower growing cells that are associated with persistence and tolerance to antimicrobials (9, 54, 55). In Gram-negative bacteria such as E. coli, this persistence phenotype has been linked to intracellular (p)ppGpp levels, where high levels of (p)ppGpp activate toxin–antitoxin systems leading to a reduced bacterial growth rate (9). To investigate whether the slower growth phenotype observed in this study as a result of the inactivation of GTPase activity also results in tolerance to antimicrobials, exponentially growing cells of LAC* iTET, LAC*ΔrsgA iTET, LAC*ΔrsgA iTET-rsgA, and LAC*ΔrsgA iTET-rsgA T199A were first exposed to three bactericidal antimicrobials, namely the penicillins penicillin G and oxacillin, and the glycopeptide vancomycin (Fig. 5A). Both the ΔrsgA mutant and the ΔrsgA iTET-rsgA T199A strain expressing the inactive GTPase variant, showed increased survival against all three antimicrobials compared with the wild type, with the introduction of a plasmid expressing the rsgA gene restoring susceptibility to wild-type levels (Fig. 5A). To examine this in more detail, the strains were exposed to both penicillin G and the fluoroquinolone ciprofloxacin and colony-forming unit counts determined over a 24-h period (Fig. 5 B and C). Exposure to penicillin G revealed statistically significant differences between the wild-type and mutant strains at the earlier time points, which became less dramatic over time (Fig. 5B). Incubation of strains with ciprofloxacin showed a highly significant increase in survival for the mutant strains that was still clearly observable after 24 h (Fig. 5C). Together, these data indicate that GTPase inactivation leads to an increase in bacterial survival upon exposure to a number of different types of antimicrobials.

Fig. 5.

Strains lacking RsgA, or producing an inactive GTPase variant, exhibit increased survival upon exposure to antimicrobials. (A) Exponentially growing LAC* iTET, LAC*ΔrsgA iTET, LAC*ΔrsgA iTET-rsgA, and LAC*ΔrsgA iTET-rsgA T199A cells were exposed to 20 times the MIC of penicillin G, oxacillin, and vancomycin. Percentage survival of the mutants and complemented strains after 3-h exposure was compared with that of the wild type. Percentage survival was calculated by dividing the number of colony-forming units per milliliter after antibiotic treatment by the number of colony-forming units per milliliter before addition of the antibiotics. Five independent experiments were performed, with the averages and SDs shown. (B and C) Exponentially growing strains were exposed to 20 times the MIC of penicillin G (B) or ciprofloxacin (C). Percentage survival at the indicated time points was calculated as for A. Four independent experiments were performed, with the averages and SDs shown. For statistical analysis, a two-tailed two-sample equal-variance Student t test was performed between LAC* iTET and LAC*ΔrsgA iTET or LAC*ΔrsgA iTET-rsgA T199A. Asterisks indicate statistically significant differences (*P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001).

GTPases from a Number of Gram-Positive Species Are Also Targets of the Stringent Response Alarmones.

RsgA, RbgA, HflX, and Era are enzymes belonging to the Era/Obg subfamily of GTPases. ObgE from E. coli has been shown to interact with (p)ppGpp (56). To examine whether the homolog of this protein from S. aureus also interacts with these nucleotides, the S. aureus ObgE protein was purified, and binding to (p)ppGpp was determined by DRaCALA (Fig. 6A). This revealed a positive interaction that was somewhat weaker than the binding of ppGpp to RsgA. This weaker affinity is potentially why the protein was not identified as an interacting partner using the whole-cell lysate screen.

Fig. 6.

(p)ppGpp bind GTPases from multiple Gram-positive species. DRaCALAs were performed with purified recombinant (A) MBP-tagged ObgE from S. aureus and the indicated ³²P-labeled nucleotides or (B) His-tagged RsgA, RbgA, Era, and HflX proteins from S. aureus, B. subtilis, and E. faecalis, and ³²P-labeled ppGpp. All experiments were carried out in triplicate. The data were plotted using the GraphPad Prism software.

To determine whether the binding of (p)ppGpp to the GTPases identified in this work is specific for Staphylococcus or is a more general regulatory mechanism used by a number of Gram-positive species, the rsgA, rbgA, era, and hflX genes from B. subtilis and E. faecalis, as well as S. aureus as a control, were amplified and cloned in a vector allowing the expression of the respective proteins as N-terminal His-tagged fusion proteins. All proteins were subsequently expressed in E. coli and purified by Ni²⁺-affinity chromatography. DRaCALA binding assays with radiolabeled ppGpp revealed a positive interaction between the nucleotide and all S. aureus and E. faecalis His-tagged proteins (Fig. 6B). Of the B. subtilis homologs, all with the exception of RsgA_BS showed strong binding (Fig. 6B). RsgA from S. aureus and B. subtilis share 45% identity but must have amino acid differences at the binding site for ppGpp that prevent binding. Altogether, these data suggest that the binding and inhibition of bacterial GTPases upon induction of the stringent response is likely a ubiquitous process in Gram-positive bacteria.

Discussion

Upon detection of an environmental stress, bacteria use the nucleotides (p)ppGpp to mediate a complex and multipronged approach leading to cells rapidly shutting down growth and entering a persistent state that promotes drug tolerance. The work presented here demonstrates the use of a genome-wide nucleotide–protein interaction screen to systematically identify target proteins for (p)ppGpp to unravel the mechanisms behind this process. As expected, this screen identified two previously known target proteins for these nucleotides, HprT and Gmk, providing validation for the screening technique. In addition, the screen identified four previously uncharacterized S. aureus GTPases as binding partners.

GTPases are a superfamily of ubiquitous enzymes with roles in signal transduction, cell division, and protein translation. This superfamily consists of several subfamilies, grouped based on identity and function, which include the translation elongation factor subfamily, the FtsY/Ffh subfamily, the Era subfamily, and the Obg subfamily. The translation elongation factor group contains a number of proteins, the functions of which have been demonstrated to be inhibited by (p)ppGpp. For instance, (p)ppGpp can inhibit the activities of the elongation factors EFG and EF-Tu (57, 58), as well as the initiation factor IF2 (59), which power the translocation of the ribosome during protein synthesis, the binding of new aminoacyl tRNAs to the ribosome, and the formation of the initiation complex, respectively. Several bacterial GTPases of the Era/Obg subfamilies, into which RsgA, RbgA, HflX, and Era group, are known to function in ribosomal assembly, more specifically in the maturation of the individual 50S and 30S ribosomal subunits before mature 70S formation. Only one of these proteins, ObgE from E. coli, is known to interact with (p)ppGpp (56). ObgE has been implicated in DNA replication (60) and has also been shown to bind to the Gram-negative (p)ppGpp synthetase/hydrolase enzyme SpoT from E. coli (61). Similar to RsgA, RbgA, HflX, and Era, it has recently been shown that ObgE also has a role in 50S and 30S ribosomal subunit association, and that (p)ppGpp binding can enhance the association of ObgE to the 50S subunit of the ribosome (59).

In the present work, we identify putative GTPases in S. aureus that have the ability to bind specifically and with high affinity to both ppGpp and pppGpp. Enzymatic analysis reveals that these enzymes are all active GTPases, the activities of which are enhanced in the presence of ribosomes but are inhibited when they are bound to (p)ppGpp. The identification of these previously unidentified target proteins allows us to propose an additional mechanism by which cells undergoing stress can use (p)ppGpp to rapidly shut down growth, namely by preventing the assembly of 70S ribosomes (Fig. 7).

Fig. 7.

Model depicting the functions of (p)ppGpp. Upon exposure to nutrient deprivation, the bacteria respond by activating the stringent response. This response is controlled by two nucleotide messengers, ppGpp and pppGpp, which function to shut down active growth and promote survival. [1] Once synthesized, these nucleotides can bind to the RNAP in Gram-negative bacteria, leading to altered transcription and decreased growth. In Gram-positive bacteria, these nucleotides instead bind to HprT and Gmk, two enzymes involved in the GTP synthesis pathway. Here, they inactivate the functions of these enzymes, resulting in decreased intracellular levels of GTP. This in turn results in altered transcription of a number of genes, mediated in part by the GTP-regulated control of the transcriptional repressor CodY and also by a decrease in the availability of GTP as an initiating nucleotide for transcription. [2] (p)ppGpp can bind to bacterial GTPases. In unstressed cells, these proteins associate with the ribosome and are thought to control the ribosome maturation processes leading to the formation of 70S ribosomes. In stressed cells, the synthesis and binding of (p)ppGpp to these enzymes inhibits their GTPase activity, resulting in decreased 50S and 30S association and a reduction in the number of mature 70S ribosomes. This in turn slows growth, a consequence of which is an increase in the tolerance of bacterial cells to antimicrobials. [3] In addition to factors controlling transcription and ribosomal assembly, (p)ppGpp can also bind to proteins involved in translation, such as the elongation factors EFG and EF-Tu, replication, such as DnaG or lipid metabolism. Binding of (p)ppGpp to these proteins inhibits their function, again promoting a slower growth state.

The stringent response alarmones help bacteria to adjust their growth to stress conditions in a number of different ways (Fig. 7): (method 1) in the α-, β-, and γ-proteobacteria, (p)ppGpp bind to the RNAP and in conjunction with the transcription factor DksA, alter the transcription of approximately one-third of the genome (3, 14, 62). Due to alterations in amino acid sequences that render (p)ppGpp unable to bind, the RNAP is not a target for (p)ppGpp in the Firmicutes, Actinobacteria, or Deinococcus-Thermus genera (26, 63). Instead, (p)ppGpp regulate transcription by binding to HprT and Gmk, enzymes involved in the GTP synthesis pathway (28, 30). These nucleotides are able to bind with high affinity and specificity to both of these enzymes, resulting in an inhibition of enzymatic function (Fig. S2 and Table S1) (28). This inhibition results in a decrease in cellular GTP levels triggering a de-repression of the transcriptional regulator CodY, as well as inhibiting the transcription of many rRNA genes due to the lack of availability of GTP as an initiating nucleotide (27, 28). Of note is that, although Gmk is present in Gram-negative species, this protein is not able to bind (p)ppGpp due to conformational changes in the nucleotide-binding pocket, suggesting that the regulation of GTP levels in this way may be unique to Gram-positive organisms (30); (method 2) ppGpp can interact with GTPases involved in ribosomal assembly to inhibit the association of the 50S and 30S subunits, as now shown in this study. In normal unstressed cells, RbgA and HflX bind to the 50S subunit (40, 42, 48, 49), whereas both RsgA and Era bind to the 30S subunit and interact with the 16S rRNA (35, 36, 43, 44). There is evidence to suggest that this occurs while the proteins are in the GTP-bound state, as for RsgA, RbgA, and ObgE, the inhibition of GTPase activity by the binding of nonhydrolysable analog of GTP causes increased association of the protein to ribosomal subunits (36, 45, 64). Here, the proteins are thought to have a caretaking or checkpoint role where they could function to facilitate proper RNA folding or processing or could promote correct protein–protein or protein–RNA interactions. Support for this conclusion comes from the observation that a B. subtilis strain depleted for RbgA shows an increase in immature 50S subunits, caused by the incorrect incorporation of the ribosomal protein L6 before the binding of other late assembly proteins (42). Additionally, it has been shown that deletions of both era and rsgA results in an accumulation of immature 17S RNA, a precursor of 16S RNA (36, 44), with cryo-EM images suggest a chaperoning role in processing the 3′ end of rRNA (44, 45). Furthermore, the position at which the RsgA and Era proteins bind to the 30S subunit, as revealed in cryo-EM studies, likely prevents the formation of a complex with the 50S subunit while they are bound (44, 45). Upon the sensing of an as-yet-unknown signal, these proteins are then released from the 50S and 30S subunits by GTP hydrolysis, allowing the now mature subunits to interact and form 70S ribosomes. In this way, these GTPases control ribosome assembly and so protein synthesis. In strains where these proteins are absent, it is likely that the subunits fail to successfully mature, seriously affecting association and mature 70S formation (36, 43).

Once (p)ppGpp is present in the cell, we show that these nucleotides can interact with high affinity with RsgA, RbgA, Era, and HflX (Figs. S3 and S4, and Table S1) and efficiently inhibit their GTPase activity (Fig. 2). In an rsgA mutant strain, the lack of GTPase activity results in a decrease in ribosomal subunit association, resulting in fewer mature 70S ribosomes (Figs. 3 and 4) (36). The decrease in 70S ribosomes would lead to a stall in protein production, which could explain the observed slower growth phenotype seen for the rsgA mutants in the absence of GTPase activity (Fig. 3A), as well as the increase in antimicrobial tolerance (Fig. 5). Further investigation is needed to conclusively say if rbgA, era, or hflX mutant strains behave in a similar fashion; (method 3) in addition to transcription and ribosomal assembly, these nucleotides can also bind to a number of other targets such as the E. coli proteins PlsB and PgsA to shut down lipid metabolism (65), to DnaG from both Gram-negative and Gram-positive bacteria to inhibit DNA replication (66), or the elongation factor GTPases from E. coli to inhibit protein translation (57, 58).

Together, these modes of growth inhibition combine to ensure a rapid shut down in bacterial growth. Although the exact biochemical mechanism by which (p)ppGpp can inhibit GTPase activity has not yet been fully elucidated, the data presented here clearly point to the control of ribosomal assembly as a potent contributor to bacterial stress survival.

Methods

Bacterial Strains and Culture Conditions.

E. coli strains were grown in LB or LB-M9 (67) and S. aureus strains in TSB at 37 °C with aeration. Strains and primers used are listed in Tables S2 and S3. The S. aureus (MRSA), Strain COL Gateway Clone Set, Recombinant in Escherichia coli, Plates 1–25, NR-19277, was obtained through BEI Resources, National Institute of Allergy and Infectious Diseases (NIAID), NIH. Information on strain construction is provided in SI Methods.

Table S2.

Bacterial strains used in this study

Strain	Relevant features	Source
Escherichia coli strains
XL1-Blue	Cloning strain: TetR	Stratagene
DH5α	Cloning strain: ANG397	Ref. 70
BL21(DE3)	Strain used for protein expression	Novagen
T7IQ	Strain used for protein expression: CamR	NEB
MG1655	F− λ− rph-1	Ref. 71
ANG292	pCL55iTETr862 in XL1-Blue: AmpR	Ref. 32
ANG474	pMUTIN-HA in E. coli; source for Erm (ermAM) marker: AmpR	Bacillus Genetic Stock Center
ANG1824	pET28b in XL1-Blue: KanR	Novagen
ANG1867	pET28b in BL21(DE3): KanR	Novagen
ANG2154	pIMAY in DH10B: CamR	Ref. 72
ANG2815	pVL791 in XL1-Blue: AmpR	Ref. 73
ANG2999	pVL847-GW in DB3.1: GnR	Ref. 73
ANG3032	pET28b-gppA in XL1-Blue: KanR	This study
ANG3033	pET28b-gppA in BL21(DE3): KanR	This study
ANG3030	pET28b-relA in XL1-Blue: KanR	This study
ANG3031	pET28b-relA in BL21 (DE3): KanR	This study
ANG3374	pET21a-relseq in BL21(DE3): AmpR	Ref. 74
ANG3959	pVL847-obg in T7IQ: GnR CamR	This study
RMC313	pIMAY-rsgA in DH5α: CamR	This study
RMC315	pIMAY-hflX in DH5α: CamR	This study
RMC303	pCL55iTETr862-rsgA in DH5α: AmpR	This study
RMC361	pCL55iTETr862-rsgA T199A in XL1-Blue: AmpR	This study
RMC362	pVL847-GW-rsgA T199A in XL1-Blue: GnR	This study
RMC363	pVL847-GW-rsgA T199A in BL21(DE3): GnR	This study
RMC387	pVL791-rsgA-Bs in XL1-Blue: CarbR	This study
RMC379	pVL791-rsgA-Bs in BL21 (DE3): CarbR	This study
RMC391	pVL791-rsgA-Ef in XL1-Blue: CarbR	This study
RMC383	pVL791-rsgA-Ef in BL21 (DE3): CarbR	This study
RMC380	pVL791-rbgA-Bs in XL1-Blue: CarbR	This study
RMC388	pVL791-rbgA-Bs in BL21 (DE3): CarbR	This study
RMC392	pVL791-rbgA-Ef in XL1-Blue: CarbR	This study
RMC384	pVL791-rbgA-Ef in BL21 (DE3): CarbR	This study
RMC389	pVL791-era-Bs in XL1-Blue: CarbR	This study
RMC381	pVL791-era-Bs in BL21 (DE3): CarbR	This study
RMC393	pVL791-era-Ef in XL1-Blue: CarbR	This study
RMC385	pVL791-era-Ef in BL21 (DE3): CarbR	This study
RMC390	pVL791-hflX-Bs in XL1-Blue: CarbR	This study
RMC382	pVL791-hflX-Bs in BL21 (DE3): CarbR	This study
RMC394	pVL791-hflX-Ef in XL1-Blue: CarbR	This study
RMC386	pVL791-hflX-Ef in BL21 (DE3): CarbR	This study
RMC395	pET28b-rsgA-Sa in XL1-Blue: KanR	This study
RMC399	pET28b-rsgA-Sa in BL21 (DE3): KanR	This study
RMC396	pET28b-rbgA-Sa in XL1-Blue: KanR	This study
RMC401	pET28b-rbgA-Sa in BL21 (DE3): KanR	This study
RMC397	pET28b-era-Sa in XL1-Blue: KanR	This study
RMC402	pET28b-era-Sa in BL21 (DE3): KanR	This study
RMC398	pET28b-hflX-Sa in XL1-Blue: KanR	This study
RMC403	pET28b-hflX-Sa in BL21 (DE3): KanR	This study
Staphylococcus aureus strains
SEJ1	RN4220 spa; protein A negative derivative of RN4220; ANG314	Ref. 67
LAC*	LAC*: Erm sensitive CA-MRSA LAC strain (AH1263)	Ref. 75
RMC355	LAC* pCL55iTETr862 (iTET): CamR	This study
RMC358	LAC*rsgA::erm: ErmR	This study
RMC368	LAC*rsgA::erm pCL55iTETr862 (iTET): ErmR, CamR	This study
RMC369	LAC*rsgA::erm pCL55iTETr862-rsgA (iTET-rsgA): ErmR, CamR	This study
RMC371	LAC*rsgA::erm pCL55iTETr862-rsgA (iTET-rsgA) T199A: ErmR, CamR	This study
RMC377	LAC*hflX	This study
Other strains
ANG196	Bacillus subtilis 168—transformable laboratory strain, trpC2	Ref. 76
ANG254	Enterococcus faecalis FA2-2—common laboratory strain	Ref. 77

Antibiotics were used at the following concentrations: for E. coli cultures: kanamycin (KanR), 30 μg/mL; ampicillin (AmpR), 100 μg/mL; carbenicillin (CarbR), 50 μg/mL; gentamicin (GnR), 20 μg/mL; for S. aureus cultures: erythromycin (ErmR), 10 μg/mL; chloramphenicol (CamR), 7.5 μg/mL; IPTG at 1 mM.

Table S3.

Primers used in this study

Number	Name	Sequence
ANG168	R-pCL55seq	CACGTTTCCATTTATCTGTATACGGATC
ANG169	F-pCL55seq	AATTCCTCCTTTTTGTTGACACTCTATC
ANG1704	F-NdeI-RelA	GGGCATATGGTTGCGGTAAGAAGTGCACATATC
ANG1705	R-EcoRI-RelA	AAAGAATTCCTAACTCCCGTGCAACCGACGC
ANG1721	F-NcoI-GppA	GGGCCATGGGTTCCACCTCGTCGCTG
ANG1722	R-HindIII-GppA	GGGAAGCTTATGCACTTCCAGCGGCCAG
ANG1749	F-pIMAYseq	TGCTTTATCGGCCGTATGTG
ANG1750	R-pIMAYseq	AATACCTGTGACGGAAGATC
RMC001	F-pVL847seq	CACGTATTGCCGCCACTATGG
RMC002	R-pVL847seq	CGCAGTCAGGCACCGTGTATG
RMC008	F-RsgA-check	CGAAGAGACACCTGATGTAATC
RMC009	R-RsgA-check	GTTGATCGATAACTTCTATATG
RMC014	F-RsgA-T199A	CATTAAATCGAGGAAAGCATACTGCAAGACATGTCGAACTATTCG
RMC015	R-RsgA-T199A	CGAATAGTTCGACATGTCTTGCAGTATGCTTTCCTCGATTTAATG
RMC018	F-EcoRV-HflXup	GGGGATATCCAAGCATCTACTGAAAGTGAAG
RMC019	R-HflXup-ErmAM	ATTTTTGTTCATCTCCTTAATAAAATCCTACTCAAA
RMC020	F-ErmAM-HflXup	TTTATTAAGGAGATGAACAAAAATATAAAATATTCT
RMC021	R-ErmAM-HflXdown	TTAAATCCTTTTTTATTTCCTCCCGTTAAATAATAG
RMC022	F-HflXdown-ErmAM	GGGAGGAAATAAAAAAGGATTTAAAAAATAATAAAA
RMC023	R-KpnI-HflXdown	GGGGGTACCCATTTCAAGCAATGCATTTAATGATG
RMC027	F-EcoRV-RsgAup	GGGGATATCGACAACTCCTAATACTGGTGAACG
RMC028	R-RsgAup-ErmAM	ATTTTTGTTCATAATGGCACCTCTCGATTAATTTTA
RMC029	F-ErmAM-RsgAup	AGAGGTGCCATTATGAACAAAAATATAAAATATTCT
RMC030	R-ErmAM-RsgAdown	CCTTTCTATTTGTTATTTCCTCCCGTTAAATAATAG
RMC031	F-RsgAdown-rErmAM	GGGAGGAAATAACAAATAGAAAGGTTAGATATTAAA
RMC032	R-KpnI-RsgAdown	GGGGGTACCCTATCAAATCCAAGTGCAACAGC
RMC033	F-AvrII-RsgA	GGGCCTAGGGCAGTAAAATTAATCGAGAGGTGCCATT
RMC035	R-SacII-RsgA-His	CCCCCGCGGTTAGTGATGGTGATGGTGATGACCATATCTAACCTTTCTATTTG
RMC046	F-NdeI-RsgA-Bs	GGGCATATGCCTGAGGGCAAAATTATTAAG
RMC047	R-BamHI-RsgA-Bs	GGGGGATCCCTAATACCTCGGCTTTCTGTC
RMC048	F-NdeI-RsgA-Ef	CCCCATATGGTTTATCTGAAAGGTCAAATC
RMC049	R-BamHI-RsgA-Ef	CCCGGATCCCTATGATTTTTTCTTATAAACAG
RMC072	F-NheI-RbgA-Bs	CCCGCTAGCACAATTCAATGGTTCCCGG
RMC051	R-XhoI-RbgA-Bs	CCCCTCGAGTTACATCGTCGGCTGTTCAAATGAC
RMC073	F-NheI-RbgA-Ef	CCCGCTAGCACCATTCAATGGTTTCCCG
RMC053	R-XhoI-RbgA-Ef	CCCCTCGAGTTATTCGTCTCCTAGTTCTTCC
RMC054	F-NdeI-Era-Bs	CCCCATATGACGAACGAAAGCTTTAAATCAG
RMC055	R-BamHI-Era-Bs	CCCGGATCCTTAATATTCGTCCTCTTTAAAGC
RMC056	F-NdeI-Era-Ef	CCCCATATGACAACTGAACATAAATCTG
RMC057	R-BamHI-Era-Ef	CCCGGATCCTTAATATTCTTCTTTGCGATAACC
RMC058	F-NdeI-HflX-Bs	GGGCATATGAACGAACAAGAAACGATTCAGG
RMC059	R-BamHI-HflX-Bs	CCCGGATCCCTACATATACTTCTTTAATTCACC
RMC060	F-NdeI-HflX-Ef	GGGCATATGACCACTCATGAAAAAGTTATTCTAG
RMC061	R-BamHI-HflX-Ef	CCCGGATCCTTATTCTTTTTCGGATTCAGC
RMC064	F-NdeI-RsgA-Sa	CCCCATATGAAGACAGGTCGAATAGTGAAAT
RMC065	R-BamHI-RsgA-Sa	CCCGGATCCTTAATATCTAACCTTTCTATTTGA
RMC066	F-NheI-RbgA-Sa	CCCGCTAGCGTTATTCAATGGTATCCAGGAC
RMC067	R-BamHI-RbgA-Sa	CCCGGATCCTTAATTGTTAGCGTCATTTGCTA
RMC068	F-NdeI-Era-Sa	CCCCATATGACAGAACATAAATCAGGATTTGT
RMC069	R-BamHI-Era-Sa	CCCGGATCCTTAATCTTGGTCTTCAACATAACC
RMC070	F-NdeI-HflX-Sa	CCCCATATGGCTCAGCAACAAATTCATGA
RMC071	R-BamHI-HflX-Sa	CCCGGATCCTTATTTTTTAAATCCTTTTATACGA

Restriction sites in primer sequences are underlined.

Protein Purifications.

Proteins were purified from 1- to 2-L E. coli cultures. Cultures were grown to an OD₆₀₀ of 0.5–0.7, protein expression induced with 1 mM IPTG, and incubated overnight at 16 °C. Protein purifications were performed by nickel-affinity and size exclusion chromatography as previously described (68, 69). Protein concentrations were determined by A₂₈₀ readings.

Construction of the S. aureus ORFeome Expression Library.

The 2,343 E. coli strains containing pDONR221 vectors with S. aureus strain COL ORFs (BEI Resources, NIAID, NIH) were grown in 1.5 mL of LB-M9 in 2-mL 96-well deep dishes selecting for kanamycin resistance. The cultures were centrifuged, and the plasmids were extracted using 96-well MultiScreen_HTS PLASMID plates (Millipore). The S. aureus gateway ORFeome library was shuttled from the pDONR221 entry plasmids into the protein overexpression destination vector pVL847-GW using LR clonase enzyme II as per manufacturer’s guidelines (Invitrogen). Subsequently, the destination plasmid library was introduced into E. coli strain T7IQ, selecting for gentamicin resistance.

Preparation of E. coli Whole-Cell Lysates.

Protein expression strains were grown in LB-M9 medium overnight at 30 °C, and protein induction was subsequently induced for 6 h with 1 mM IPTG. Bacteria were collected by centrifugation and suspended in 1/10th of their original volume in 40 mM Tris, pH 7.5, 100 mM NaCl, 10 mM MgCl₂ binding buffer containing 2 mM PMSF, 20 μg/mL DNase, and 0.5 mg/mL lysozyme. Cells were lysed by three freeze/thaw cycles. Lysates were directly used in binding assays or stored at −20 °C.

DRaCALA.

This assay was performed as described previously with slight modifications as outlined in SI Methods (32, 33).

Synthesis of (p)ppGpp.

³²P-labeled pppGpp was synthesized from α-³²P-GTP (Perkin-Elmer) by incubating 55.5 nM α-³²P-GTP with 2 μM Relseq protein in 25 mM Bis-Tris propane, pH 9, 100 mM NaCl, 15 mM MgCl₂ binding buffer, using 8 mM ATP as the phosphate donor, at 37 °C for 1 h. The Relseq protein was separated from the radiolabeled pppGpp by filtration on 3-kDa cutoff spin column. To synthesize ³²P-ppGpp, the ³²P-pppGpp was incubated with 1 μM of the phosphatase GppA for 15 min at 37 °C. The GppA protein was separated from the radiolabeled ppGpp by filtration on 3-kDa cutoff spin column. Reaction products were visualized by spotting 1 μL on PEI-cellulose F TLC plates (Merck Millipore) and separation in 1.5 M KH₂PO₄, pH 3.6. The radioactive spots were visualized using an LA 7000 Typhoon PhosphorImager. Unlabeled (p)ppGpp was synthesized in the same way but with the addition of 6 mM GTP instead of the 55.5 nM α-³²P-GTP. Spiking a duplicate reaction with radiolabeled GTP confirmed complete conversion of GTP to (p)ppGpp.

GTP Hydrolysis Assays.

The ability of proteins to hydrolyze GTP to GDP was determined by incubating 10 μM recombinant protein with 2.78 nM α-³²P-GTP in 40 mM Tris, pH 7.5, 100 mM NaCl, 10 mM MgCl₂ at 37 °C for the indicated times. Ribosomes at a final concentration of 118 nM and increasing concentrations of ppGpp or pppGpp were added to the initial mixture where indicated. The reactions were inactivated with the addition of formic acid to a final concentration of 1.2 M. Precipitated proteins were pelleted by centrifugation at 17,000 × g for 10 min. Reaction products were then visualized by spotting 1 μL on PEI-cellulose F TLC plates (Merck Millipore) followed by separation in 1 M KH₂PO₄, pH 3.6, buffer. The radioactive spots were visualized using an LA 7000 Typhoon PhosphorImager, and images were quantified using ImageJ.

Enzymatic Assays.

Gmk and HprT activity assays were performed as previously described and are outlined in SI Methods (28).

70S Ribosome Purification.

70S ribosomes were purified as detailed by Daigle and Brown (51) with the following exceptions: ribosomes were purified from 4 L of the S. aureus strain LAC* grown in TSB medium. The S. aureus culture was grown to an OD₆₀₀ of 0.8 before the addition of 100 μg/mL chloramphenicol. Following a 3-min incubation at 37 °C, cultures were allowed to cool to 4 °C before centrifugation. Cells were suspended in buffer A (20 mM Tris⋅HCl, pH 7.5, 10.5 mM magnesium acetate, 100 mM NH₄Cl, 0.5 mM EDTA, 3 mM 2-mercaptoethanol) and lysed with 0.2 μg/mL lysostaphin and 75 ng/mL DNase for 30 min at 37 °C. Lysates were centrifuged at 30,000 × g for 1 h, and the protocol continued as per Daigle and Brown (51).

Ribosomal Profiles from S. aureus Cell Extracts.

Crude isolations of ribosomes from S. aureus cell extracts were achieved as described by Uicker et al. (42) with some modifications. Briefly, 150-mL cultures of the different S. aureus strains were grown to an OD₆₀₀ of 0.6 in TSB 100 ng/mL Atet. For induction of the stringent response, mupirocin was added to cultures 30 min before harvesting. Cultures were allowed to cool to 4 °C before centrifugation. The cells were suspended in lysis buffer (80 mM Tris⋅HCl, pH 7.8, 7 mM magnesium acetate, 150 mM NH₄Cl, and 2.5 mM DTT), normalized to an OD₆₀₀ of 25, lysed by the addition of 0.2 μg/mL lysostaphin and 75 ng/mL DNase, and incubated for 30 min at 37 °C. The extracts were centrifuged at 17,000 × g for 5 min, and subsequently 500 μL was layered onto 10–25% (wt/vol) sucrose density gradients in 10 mM Tris⋅HCl, pH 7.8, 10 mM MgCl₂, and 300 mM KCl. Gradients were centrifuged for 3.5 h at 210,000 × g. Gradients were fractionated by upward displacement of 250-μL aliquots, which were analyzed for RNA content at an absorbance of 260 nm.

Antimicrobial Tolerance Assay.

Overnight cultures of S. aureus strains in TSB containing 100 ng/mL Atet were diluted to an OD₆₀₀ of 0.05 and grown until an OD₆₀₀ of 0.4 was reached. The 1.5-mL aliquots were then incubated with 20 times the minimum inhibitory concentration (MIC) value for vancomycin (40 μg/mL), oxacillin (1.28 mg/mL), penicillin G (20 μg/mL), or ciprofloxacin (320 μg/mL), as previously determined by e-test strips. Aliquots were further incubated at 37 °C for the times indicated. Colony-forming unit counts were determined by removing 500-μL samples, centrifuging, and suspending cells in fresh medium. The cells were subsequently serially diluted and plated. Percentage survival was calculated by dividing the number of colony-forming units per milliliter after antibiotic treatment by the number of colony-forming units per milliliter before addition of the antibiotics.

SI Methods

Plasmid and Strain Construction.

Strains used in this study are listed in Table S2. Plasmids pET28b-gppA and pET28b-relA were constructed by amplifying the gppA and relA genes from MG1655 genomic DNA using primer pairs ANG1721/ANG1722 and ANG1704/ANG1705, respectively, and inserting them into appropriately digested pET28b. pVL847-GW-rsgA T199A was created using overlap primer PCR with plasmids pVL847-GW-rsgA as a template and primers RMC014/RMC015. The rsgA, rbgA, era, and hflX genes were amplified from S. aureus strain LAC*, B. subtilis 168, and E. faecalis FA2-2 genomic DNA using the primers listed in Table S3. The PCR products were digested with the appropriate enzymes and cloned into pET28b or pVL791. The plasmid pCL55iTETr862-rsgA was created by amplifying the rsgA gene from LAC* chromosomal DNA with primer pair RMC033/RMC035. The resulting fragment was cloned into pCL55iTETr862 that had been cut with AvrII/SacII. Overlap primer PCR with primers RMC014/RMC015 and pCL55iTETr862-rsgA as a template was used to introduce a T199A mutation. All plasmids were initially transformed into E. coli strain XL1-Blue, and sequences of all inserts were verified by fluorescence automated sequencing at GATC Biotech. For protein expression and purification, all plasmids were transformed into E. coli strain BL21(DE3).

For the deletion of the rsgA and hflX genes in S. aureus, 1-kb fragments upstream and downstream of both genes were amplified from LAC* genomic DNA using primer pairs RMC027/RMC028 and RMC031/RMC032 for rsgA and RMC018/RMC019 and RMC022/RMC023 for hflX, which incorporate 5′-EcoRV and 3′-KpnI sites, respectively. The erythromycin cassette ErmAM was amplified from pMUTIN-HA plasmid DNA using primers RMC029/RMC030 and RMC020/RMC021. Purified PCR products were then fused by SOE PCR, digested with EcoRV and KpnI, and cloned into the allelic exchange vector pIMAY, yielding plasmids pIMAY-rsgA and pIMAY-hflX. These plasmid were then electroporated into SEJ1 and stably maintained at 28 °C in the presence of 10 μg/mL Cam before phage transduction into LAC*. Shifting the temperature to 37 °C resulted in insertion of the plasmids into the chromosome. Downshift of the temperature to 28 °C in the absence of Cam resulted in excision of the pIMAY plasmids and created strain RMC358, with an in-frame deletion of the chromosomal copy of the rsgA gene and strain RMC377 with a deletion of hflX. For complementation analysis of RsgA, the plasmids pCL55iTETr862, pCL55iTETr862-rsgA, and pCL55iTETr862-rsgA T199A were initially electroporated into S. aureus strain SEJ1. The integrated plasmids were then transduced with Φ85 into LAC* ΔrsgA, yielding strains RMC368, RMC369, and RMC371 (Table S2).

DRaCALA.

E. coli whole-cell lysates or 10 µM purified protein in binding buffer were mixed with ∼2.78 nM ³²P-labeled pppGpp, ppGpp, or GTP, synthesized as described in the main text, and incubated at room temperature for 5 min before spotting 2.5 μL on a nitrocellulose membrane. For the whole-genome screen, the ³²P-labeled pppGpp and ppGpp were mixed at a 1:1 ratio, dispensed into lysate-containing 96-well plates, and the mixture spotted onto nitrocellulose membrane using a 96-well pin tool (V&P Scientific). For competition assays, 100 μM cold nucleotides (ATP, GTP, cAMP, cGMP; Sigma; c-di-AMP, c-di-GMP; BioLog; ppGpp; tebu-bio) were added to the initial mixture, and 2.5 μL of reactions were spotted onto nitrocellulose membranes (Amersham Hybond-ECL; GE Healthcare). All spots were then air-dried, and radioactivity signals were detected using a LA 7000 Typhoon PhosphorImager. The fraction of ligand bound and K_d values were calculated as previously described (33).

Enzymatic Assays.

The Gmk activity assay contained 100 mM Tris⋅HCl, pH 7.5, 100 mM KCl, 10 mM MgCl₂, 4 mM ATP, 1.5 mM phosphoenolpyruvate, 2 U of pyruvate kinase, 2.64 U of lactate dehydrogenase, 150 μM NADH, 10 nM Gmk, and 100 μM of either pppGpp or ppGpp. Reactions were initiated by the addition of 50 μM GMP, and the absorbance at 340 nm was monitored over time. The HprT assay contained 100 mM Tris⋅HCl, pH 7.5, 1.2 mM MgCl₂, 50 μM guanine, 20 nM HprT, and 100 μM of either pppGpp or ppGpp. Reactions were initiated by the addition of 1 mM phosphoribosyl pyrophosphate, and the absorbance at 257 nm was monitored over time.