Serine-threonine phosphoregulation by PknB and Stp contributes to quiescence and antibiotic tolerance in Staphylococcus aureus

Abstract

Staphylococcus aureus can cause infections that are often chronic and difficult to treat, even when the bacteria are not antibiotic resistant because most antibiotics act only on metabolically active cells. Subpopulations of persister cells are metabolically quiescent, a state associated with delayed growth, reduced protein synthesis, and increased tolerance to antibiotics. Serine-threonine kinases and phosphatases similar to those found in eukaryotes can fine-tune essential bacterial cellular processes, such as metabolism and stress signaling. We found that acid stress–mimicking conditions that S. aureus experiences in host tissues delayed growth, globally altered the serine and threonine phosphoproteome, and increased threonine phosphorylation of the activation loop of the serine-threonine protein kinase B (PknB). The deletion of stp, which encodes the only annotated functional serine-threonine phosphatase in S. aureus, increased the growth delay and phenotypic heterogeneity under different stress challenges, including growth in acidic conditions, the intracellular milieu of human cells, and abscesses in mice. This growth delay was associated with reduced protein translation and intracellular ATP concentrations and increased antibiotic tolerance. Using phosphopeptide enrichment and mass spectrometry–based proteomics, we identified targets of serine-threonine phosphorylation that may regulate bacterial growth and metabolism. Together, our findings highlight the importance of phosphoregulation in mediating bacterial quiescence and antibiotic tolerance and suggest that targeting PknB or Stp might offer a future therapeutic strategy to prevent persister formation during S. aureus infections.

INTRODUCTION

The Gram-positive bacterium Staphylococcus aureus intermittently or permanently colonizes 15 to 50% of the human population (1, 2). Although S. aureus can be part of healthy human microbiota, it is a leading cause of a variety of nosocomial and community-acquired infections (1). Besides skin and soft tissue infections, S. aureus can cause severe and life-threatening, invasive infections such as endocarditis, osteomyelitis, and bacteremia (1). The increasing incidence of resistance against multiple antibiotics, including penicillin, methicillin, and vancomycin (VAN), render these infections difficult to treat and require prolonged treatment and extended hospitalization (3-5).

In addition to antibiotic resistance, S. aureus can also persist and cause chronic and recurrent infections even while remaining susceptible to antibiotics (6-8). This phenomenon of antibiotic persistence is defined as the ability of a subpopulation of bacteria to survive high concentrations of bactericidal drugs to which they remain fully susceptible (9). Subpopulations with different degrees of tolerance can coexist within a bacterial population, facilitating survival in changing environments (9, 10). One mechanism facilitating antibiotic tolerance is bacterial quiescence, which allows the bacteria to survive periods of environmental stress, such as exposure to nutrient limitation or antibiotics, by being in a viable but nondividing state. Because most antibiotics require an active target in the bacterial cell, slowing or arresting growth and central metabolic activities enables the bacteria to survive in the presence of antibiotics for a prolonged period of time (11, 12). The strategy of bacterial bet-hedging through the formation of quiescent persister cells in a clonal bacterial population might therefore be advantageous during infection. The actively growing bacteria in a clonal population are mostly responsible for the colonization of the host and a successful infection, whereas the subpopulation of persisters ensures the survival of the genotype under fluctuating milieus as found in a host and during antibiotic therapy (13).

Bacterial persisters can occur spontaneously in a population or can be triggered by environmental factors such as low pH, starvation, oxidative stress, or drugs, for example, that induce delayed growth, which results in increased bacterial survival, generally referred to as “tolerance by lag” (14-19). Signal transduction through reversible protein phosphorylation is a major regulatory mechanism that facilitates quick responses to environmental changes (20). In bacteria, two-component systems, consisting of a histidine kinase sensor and an associated response regulator (21), are the most common mechanism of signal transduction. In addition, regulatory phosphorylation of serine, threonine, and tyrosine residues has been identified in prokaryotes (22, 23). In S. aureus, the conserved eukaryotic-type serine-threonine (Ser-Thr) kinase STK (alternatively named PknB or Stk1) and the cognate phosphatase Stp affect bacterial cell signaling, central metabolism (24-26), stress response (26, 27), antibiotic resistance (26, 28-30), and virulence (28, 29, 31-33). Stp has been suggested as a potential target for anti-virulence agents because stp mutation or absence decreases S. aureus virulence factor production (34, 35). Ser-Thr phosphorylation is also involved in the regulation of protein synthesis, the most energy-consuming cellular process, and the initiation of translation decreases in response to nutrient limitation (36). For example, in Bacillus subtilis, protein synthesis is reduced by phosphorylation of the elongation factor Tu (EF-Tu), which impairs its essential guanosine triphosphatase (GTPase) activity, during nutrient limitation and subsequent formation of dormant spores (37).

Here, we describe the role of the Ser-Thr kinase-phosphatase pair PknB-Stp in regulating growth and antibiotic tolerance in S. aureus by using pknB and stp deletion mutants. Colony size heterogeneity and growth delays of individual bacterial cells after acidic pH stress and of bacteria derived from murine abscesses were analyzed using single-cell microscopy. We showed that increased Ser-Thr phosphorylation led to the formation of nonstable small colonies (nsSCs) and prolonged lag phases, defined as the time it takes for cells to reach exponential growth upon transfer to different growth conditions or fresh medium. The identified growth delay correlated with increased survival of the bacteria in subsequent antibiotic challenges. Phosphopeptide enrichment by Fe³⁺-immobilized metal affinity chromatography followed by liquid chromatography–tandem mass spectrometry (LC-MS/MS) analysis enabled us to create a detailed map of PknB and Stp targets. In addition, we identified changes in phosphorylation states, a decrease in protein synthesis activity, and increased antibiotic tolerance after acidic pH stress. Together, our identification of phosphorylation targets that led to reduced protein synthesis and cellular quiescence advances the understanding of how S. aureus can survive antibiotic therapy despite being fully susceptible.

RESULTS

Acidic pH exposure prolongs bacterial lag phase and reduces protein synthesis activity

S. aureus encounters a variety of pH conditions during infection, for example, a close to neutral pH of 7.4 in the bloodstream (38) and an acidic pH on the skin (pH 4 to 7) (39, 40), intracellularly in phagolysosomes (pH 4.5 to 5.5) (41-43), or in abscesses (pH 5.5 to 7.3) (44-46). We previously demonstrated that nsSCs, the sizes of which are 5 to 10 times smaller than the most common colony type, result from a bacterial growth delay and correlate with the presence of a persister cell subpopulation (15). To understand whether acidic pH prolongs bacterial lag phase, we exposed the methicillin-sensitive S. aureus strain Cowan I [sequence type (ST) 30, historical strain] and the virulent methicillin-resistant S. aureus strain USA300 JE2 (ST8, community-acquired) to either pH 5.5 or 7.4 for 3 days and quantified the time that single bacterial cells needed to start their first cell division after changing to favorable growth conditions using time-lapse microscopy (Fig. 1A and fig. S1A). Ten percent of the bacterial populations exposed to pH 5.5 showed growth delays of more than 7.8 hours in Cowan I wild type, whereas exposure to pH 7.4 showed growth delays of 4.3 hours (Fig. 1A). A similar behavior was observed for USA300 JE2 wild type, with minimum growth delays of 6.8 hours for pH 5.5 as compared with 4.3 hours for pH 7.4 (fig. S1A). To better understand the molecular state of S. aureus during pH 5.5 and 7.4 exposure and whether the growth delays were linked to reduced cellular energy, we monitored the intracellular adenosine triphosphate (ATP) concentration over time (Fig. 1B and fig. S1B). In addition, we measured the incorporation of the puromycin analog O-propargyl-puromycin (OPP) (47) to monitor protein synthesis activity over time in S. aureus exposed to pH 5.5 or 7.4 (Fig. 1B and fig. S1B). After inoculation of S. aureus in pH 7.4 medium, we observed a rapid increase (sevenfold for Cowan I and threefold for USA300 JE2) in intracellular ATP concentration in the first 6 hours, followed by a gradual decrease over time. In contrast, S. aureus exposed to pH 5.5 showed continuous lower concentrations (3.5-fold increase for Cowan I and no increase for USA300 JE2) of intracellular ATP in the first 6 hours (Fig. 1B and fig. S1B). Furthermore, a steady decrease in protein synthesis activity was observed for S. aureus grown at pH 5.5 as compared with those grown at pH 7.4 (Fig. 1B and fig. S1B).

Fig. 1. Exposure to acidic pH prolongs lag phase, reduces translational activity, and elicits distinct phosphorylation changes in wild-type S. aureus.

(A) Lag phase determination by single-cell microscopy in S. aureus wild-type (WT) Cowan I transferred to nutrient-rich medium after 3 days of pH 5.5 or 7.4 exposure. N = 3 biological replicates. Shaded areas indicate the SD. Cowan I WT (pH 5.5), n = 232 bacterial cells; Cowan I WT (pH 7.4), n = 627 cells. The gray zone marks the period at the beginning of the experiment when cell divisions could occur without being observed. Differences between the two groups were statistically significant for the period during which 90% of the bacteria had started to divide (red arrow). Asterisks indicate significant difference at **P < 0.01. (B) Intracellular ATP concentration and protein synthesis activity in S. aureus WT Cowan I grown in DMEM (pH 5.5 or 7.4) for 3 days. Protein synthesis activity is indicated as relative fluorescence units (RFUs), representing total fluorescence of OPP-labeled S. aureus normalized to colony-forming units (CFUs). N = 3 biological replicates. (C) Heatmap of hierarchical cluster analysis of proteins involved in metabolic processes showing phosphopeptide abundance in the phosphoproteomes of S. aureus WT Cowan I grown at pH 7.4 or 5.5. N = 4 biological replicates per condition. Blue indicates less abundant phosphopeptides; red indicates more abundant phosphopeptides. Black indicates nondetected phosphopeptides. (D) Pie charts and volcano plots comparing the total proteome and phosphoproteome changes between Cowan I WT grown at pH 5.5 and 7.4, identifying differentially abundant proteins or phosphopeptides [cutoff adjusted P < 0.05 and a fold change (FC) of ∣ log₂FC∣ > 1]. Underrepresented proteins and phosphopeptides are depicted by blue symbols, and overrepresented proteins and phosphopeptides are depicted with red symbols. Symbol type indicates Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway. (E) Changes in Ser-Thr phosphorylation of ribosomal proteins in Cowan I WT grown at pH 5.5 compared with WT grown at pH 7.4. Green color indicates increased Ser-Thr phosphorylation (at least one phosphosite log₂FC ≧ 1, adjusted P < 0.05). Blue color indicates decreased Ser-Thr phosphorylation (at least one phosphosite log₂FC ≦ −1, adjusted P < 0.05). If an increased and decreased abundance of phosphopeptides for the same protein were found, then only the one with the greatest change is shown. (F) Phosphorylation (P) sites to protein plot for the selected proteins FtsZ, enolase (Eno), PknB, and EF-G. Phosphorylation sites and exact amino acid positions are indicated. Y axis indicates log₂ FCs of phosphopeptides in Cowan I WT grown at pH 5.5 as compared with WT grown at pH 7.4. Dashed lines indicate pseudo-FCs (phosphopeptides that were only detected in one condition). Statistical significance in phosphopeptides is shown as *. P values in (D) to (F) were determined by a moderated t test, as implemented in the R package limma, and were adjusted for multiple testing using the Benjamini-Hochberg method. Significance threshold was set to 0.05.

Phosphoproteomic analysis shows increased Ser-Thr phosphorylation patterns during acid stress

To further investigate the impact of acid stress on protein phosphorylation and signal transduction, major strategies used in response to marked environmental changes, we performed total proteomic and phosphoproteomic analysis of S. aureus Cowan I exposed to pH 5.5 or 7.4 for 3 days. A heatmap overview of all the proteins with identified phosphopeptides (fig. S1C) and one including certain proteins that are involved in metabolic processes both showed a clear sample clustering of pH 5.5 and 7.4 and increased phosphorylation events in S. aureus exposed to pH 5.5 as compared with cells exposed to pH 7.4 (Fig. 1C and fig. S1C). Consistent with the prolonged lag phase and decreased protein synthesis, acid-stressed S. aureus exhibited a greater number of proteins with decreased abundance (430 of 1671 identified proteins) as compared with pH 7.4 (Fig. 1D and data files S1 to S3). Following this pattern, we also identified more phosphosites with reduced abundance in pH 5.5–stressed bacteria (144 of 567 identified phosphosites), with an almost identical proportion of the proteins (25.7%) or phosphopeptides (25.9%) changed. However, proteins involved in cell division (such as FtsZ), glycolysis (such as Eno), signaling (such as PknB), and translation, such as ribosomal proteins (such as RplB, RplC, and RplS) and elongation factors (such as EF-G), displayed increased Ser-Thr phosphorylation in acid-stressed bacteria as compared with bacteria at pH 7.4 (Fig. 1, D and E). Specifically, we found a strong Thr phosphorylation in the activation loop (residues 160 to 171) of the Ser-Thr kinase PknB (pThr¹⁶⁴, pppThr¹⁶⁴, and ppThr¹⁶⁶), indicating a potential role of Ser-Thr phosphoregulation in acid stress and bacterial quiescence (Fig. 1F and data file S2).

Lack of Ser-Thr dephosphorylation increases colony growth heterogeneity during acid stress exposure

We examined the role of the Ser-Thr kinase PknB and the Ser-Thr phosphatase Stp in regulating the stress response, colony growth, and lag phase in S. aureus by generating isogenic mutants for stp and pknB in the two distinct wild-type strains, Cowan I and USA300 JE2. These were exposed to defined stress in pH-buffered medium (pH 7.4, 5.5, or 4.0) for 3 and 7 days. Whereas we did not identify any significant differences in the proportion of nsSCs between the wild-type strains and their corresponding isogenic mutants at pH 7.4, we found a significantly increased proportion of nsSCs in the Δstp mutants (13 to 22%) as compared with the wild type (5 to 14%) and ΔpknB (1 to 15%) after exposure to pH 5.5 for 3 days (Fig. 2A and fig. S2C). This effect was even more pronounced when bacteria were exposed to pH 4.0, with nsSC levels of up to 37% in the Δstp mutants (Fig. 2A and fig. S2C). In addition, there was a trend toward reduced nsSC formation in both Cowan I and USA300 JE2 ΔpknB after pH 4.0 exposure; however, this was not statistically significant (Fig. 2A and fig. S2C). Extending the pH stress to 7 days showed a similar pattern and doubled the proportions of nsSCs in wild type and Δstp mutant in both Cowan I and USA300 JE2, whereas a significant reduction of nsSCs in ΔpknB was observed (Fig. 2B and fig. S2D). Furthermore, the deletion mutants Δstp and ΔpknB did not exhibit any significant differences in their ability to grow under nutrient-rich and under various pH conditions when compared with wild-type parent (figs. S2, A and B, and S3, A to D).

Fig. 2. Deletion of stp leads to increased formation of nsSCs in S. aureus after acid stress.

(A and B) Proportion of nsSCs in S. aureus Cowan I WT, Δstp, and strains after 3 (A) or 7 (B) days of pH 7.4, 5.5, or 4.0 exposure. N = 6 or 7 biological replicates (A); N = 3 biological replicates (B). (C) Proportion of nsSCs in S. aureus Cowan I WT, Δstp, and ΔpknB strains after exposure to the intracellular environment of human lung epithelial A549 cells for 3, 5, or 7 days. Statistical significance was determined by the Kruskal-Wallis test with Dunn’s post hoc test. Error bars indicate SEMs. Asterisks indicate significant difference at *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Exposure to conditions representative of the intracellular milieu, such as acidic pH, can induce the formation of nsSCs, and this increases over time (43, 48). To evaluate whether the mutant lacking Stp displays an increased colony size heterogeneity under more complex stress conditions, we infected human lung epithelial cells (A549) with Cowan I and its isogenic mutant strains. The Cowan I strain was used for the following intracellular stress assay because of its lower cytotoxicity and hemolysis activity as compared with USA300 JE2 (fig. S4, A to C). We monitored the formation of nsSCs over time postinfection (p.i.) in these nonprofessional phagocytes and found that the Δstp mutant strain formed a significantly higher proportion of nsSCs after 5 and 7 days p.i. as compared with its wild-type or ΔpknB counterparts (Fig. 2C). In contrast, the kinase deletion mutant ΔpknB showed a reduction in nsSC formation over time as compared with both wild type and Δstp (Fig. 2C). Neither of the mutants displayed a significantly increased or decreased ability to survive intracellularly, but we observed a trend toward a reduced capacity to survive intracellularly for ΔpknB when compared with the wild type and Δstp at 5 and 7 days p.i. (fig. S3E). Together, our data suggest that both the phosphatase Stp and the kinase PknB play a role in the S. aureus acid stress response and influence bacterial growth during acid stress conditions in vitro and in cells.

Acid stress extends lag phase and reduces intracellular ATP and protein synthesis in Δstp

To verify that the proportion of nsSCs reflects a growth delay of individual bacterial cells, we used time-lapse microscopy to measure the time single bacterial cells needed to start their first cell division after changing to favorable growth conditions. This lag phase was assessed for pH 5.5–exposed bacteria and for exponential phase bacteria grown in nutrient-rich medium at pH 7.4 that served as a control (Fig. 3A and figs. S5, A to B, and S6A). Whereas exponentially growing bacteria did not exhibit differences in growth upon transfer to fresh medium (fig. S5, A and B), the lag phase of mutants after acid stress changed significantly, as expected from our persister measurements on nsSCs (Fig. 3A and fig. S6A). We observed growth delays of up to 12 hours for Cowan I Δstp and 10 hours for USA300 JE2 Δstp. The wild-type strains showed lag phases of up to 8 hours for Cowan I and 4 hours for USA300 JE2. Although there was no significant difference in the lag phases between the wild type and ΔpknB in Cowan I (90% of the bacteria started to grow after 5.5 hours), USA300 JE2 ΔpknB had a significantly reduced lag phase as compared with its wild-type counterpart after pH 5.5 exposure (90% of the bacteria started to grow after 2.8 hours for ΔpknB as compared with 3.8 hours in wild type Fig. 3A and fig. S6A).

Fig. 3. Acid stress increases lag phase and reduces ATP concentration in S. aureus Δstp.

(A) Lag phase determination by single-cell microscopy in S. aureus Cowan I WT and the isogenic mutants Δstp and ΔpknB transferred to nutrient-rich medium after 3 days of pH 5.5 exposure. N = 3 biological replicates. Shaded areas indicate the SD. Cowan I WT (pH 5.5), n = 475 bacterial cells; Cowan I Δstp (pH 5.5), n = 246; Cowan I ΔpknB (pH 5.5), n = 554. The gray zone marks the period at the beginning of the experiment when cell divisions could occur without being observed. Statistical significance was determined for the time at which 90% of the bacteria started to divide (red arrow). (B) Intracellular ATP concentration in S. aureus Cowan I WT, Δstp, and ΔpknB grown in DMEM (pH 5.5) for 3 days. N = 6 biological replicates. Statistical significance was determined by one-way ANOVA with Tukey’s post hoc test after passing the Shapiro-Wilk normality test. Error bars indicate SEMs. (C to E) S. aureus strains Cowan I WT, Δstp, and ΔpknB were exposed to pH 5.5 for 3 days and analyzed at the indicated time points (0, 2, 4, 6, 24, 48, and 72 hours) for the proportion of nsSCs (C), intracellular ATP concentration (D), and protein synthesis activity (E). N = 4 biological replicates. Statistically significant difference at 72 hours was determined by Mann-Whitney U test. (F to H) Proportion of nsSCs (F), intracellular ATP concentration (G), and protein synthesis activity (H) in S. aureus strains Cowan I WT, Δstp, ΔpknB, PknB^K39M, and ΔpknB/Δstp exposed to pH 5.5 for 3 days. N = 4 biological replicates. Statistical significance was determined by one-way ANOVA with Tukey’s post hoc test after passing the Shapiro-Wilk normality test. Error bars indicate SEMs. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

To determine whether the lag phase changes were linked to reduced cellular energy status, we measured intracellular ATP concentrations of the bacteria after exposure to acidic pH (Fig. 3B an fig. S6B). We found that the Δstp mutant, which showed significantly prolonged lag phases after pH 5.5 exposure (Fig. 3A and fig. S6A), displayed significantly lower ATP concentrations (50% reduction) as compared with the wild type and ΔpknB mutant in both Cowan I and USA300 JE2 (Fig. 3B and fig. S6B). Slightly, but significantly, reduced ATP concentration was found for Cowan ΔpknB as compared with the wild type, whereas there was no difference between USA300 JE2 wild type and ΔpknB. These findings are consistent with previously published metabolomics data showing significantly reduced ATP concentration in stationary S. aureus strain 8325 Δstp but not in ΔpknB (49).

To delve into the dynamics of nsSC formation, intracellular ATP concentration, and protein synthesis activity in the wild type and its isogenic mutant strains, we performed kinetic studies at pH 5.5 and 7.4 (Fig. 3, C to E, and figs. S6, C to E, S7, A to C, and S7, G to I). The Δstp mutant showed significantly higher proportions of nsSCs over time as compared with wild type and ΔpknB mutant at pH 5.5 (Fig. 3C and fig. S6C). In contrast, the ΔpknB mutant exhibited continuously lower proportions of nsSCs as compared with wild type and the Δstp mutant (Fig. 3C and fig. S6C). Intracellular ATP concentration and protein synthesis activity were reduced in the Δstp mutant and showed a continuous decrease over time (Fig. 3, D to E, and fig. S6, D to E). In contrast, the ΔpknB mutant displayed a significant increase in protein synthesis activity over time (Fig. 3E and fig. S6E). No significant differences in nsSCs, intracellular ATP concentration, or protein synthesis activity were observed between the wild type and corresponding mutant strains over time while growing in pH 7.4 medium (fig. S7, A to C and G to I). Additionally confirming these findings, cells producing only a catalytically inactive mutant form of PknB (PknB^K39M) and a double knockout (ΔpknB/Δstp) showed similar results as the ΔpknB deletion mutant (Fig. 3, F to H, and figs. S6, F to H, and S7, D to F and J to L). Overall, our data show that, in the absence of Stp, the single-cell growth delay and nsSC formation after acid stress are associated with reduced intracellular ATP concentration and reduced protein synthesis activity.

Lack of Stp-mediated dephosphorylation contributes to antibiotic tolerance

To determine whether high proportions of late-growing bacteria with reduced ATP concentrations and protein synthesis activity were associated with increased antibiotic tolerance, we challenged acidic pH–exposed wild-type strains of S. aureus and their deletion mutants Δstp and ΔpknB with high concentrations of antibiotics [40× the minimum inhibitory concentration (MIC); table S1]. To test the survival capacity of Cowan I, we used the cell wall–targeting antibiotic flucloxacillin (FLU), the DNA replication inhibitor fluoroquinolone ciprofloxacin (CIP), and the transcription inhibitor rifampicin (RIF). For USA300 JE2, we used the cell wall–targeting antibiotics FLU and VAN, as well as RIF. Consistent with the increased lag phase (Fig. 3, A, C, and F, and fig. S6, A, C, and F), the reduced intracellular ATP concentration (Fig. 3, B, D, and G, and fig. S6, B, D, and G), and protein synthesis activity (Fig. 3, E and H, and fig. S6, E and H) in the Δstp mutants of both strains after pH 5.5 exposure, we observed increased tolerance toward several different antibiotics (Fig. 4, A and B, and fig. S8, A and B). Calculating the minimum duration to kill 90% of the initial bacterial population (MDK₉₀) for FLU and CIP, the MDK₉₉ for the rapidly acting antibiotic RIF for Cowan I (Fig. 4B), and the MDK90 for FLU, VAN, and RIF for USA300 JE2 (fig. S8B) allowed us to better compare between strains and treatments. We found that 90% of the initial population of pH 5.5–exposed Cowan Δstp was killed by FLU after 19 hours, whereas it took only 6 hours for the wild type and 5 hours for ΔpknB. A similar pattern was observed for CIP and RIF (Fig. 4B) and for USA300 JE2 challenged with FLU, VAN, or RIF (fig. S8B). To exclude resistance development as a confounding factor for RIF tolerance, we additionally plated on RIF plates and excluded experiments where resistance development occurred from this analysis (fig. S9, A to C). Furthermore, acid-stressed S. aureus wild type, Δstp, ΔpknB, catalytically inactive PknB^K39M, and ΔpknB/Δstp were exposed to FLU and RIF (40× MIC) for 24 hours, and survival was assessed. Overall, the Δstp mutant exhibited a higher tolerance against both FLU and RIF as compared with wild type, ΔpknB, PknB^K39M, and ΔpknB/Δstp. In particular, the Δstp mutant showed significantly higher survival as compared with wild type, ΔpknB, and PknB^K39M in FLU and against wild type and ΔpknB in RIF (Fig. 4C and fig. S8C), further corroborating our findings. During logarithmic growth, only a small proportion of cells were able to survive the high concentration of antibiotics, and no significant differences between the wild-type and mutant strains were observed (fig. S8D).

Fig. 4. Increased antibiotic tolerance in Δstp after acidic pH exposure.

S. aureus strains Cowan I and its isogenic mutants were grown in DMEM (pH 5.5) for 3 days before inoculation in pH 7.4 medium supplemented with the indicated antibiotics (40× MIC). (A) Time kill curves (0, 1.5, 3, 6, and 24 hours) for different antibiotics and strains. Survival was calculated relative to the initial inoculum of ~10⁵ CFU/ml. (B) Minimum duration to kill (MDK) 90 or 99% of the initial population for different strains and antibiotics. Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparison testing after passing the Shapiro-Wilk normality test. Error bars indicate SEMs. (C) Survival after exposure to FLU or RIF for 24 hours. Survival was calculated relative to the initial inoculum. Statistical significance was determined by one-way ANOVA with Tukey’s post hoc test. Error bars indicate SEMs. Asterisks above bars indicate significant difference to Cowan I Δstp. *P < 0.05, **P < 0.01, and ***P < 0.001.

Loss of stp increases lag phase and persister formation in mice

To test whether the stp deletion leads to increases in lag phases and persister subpopulations in vivo, we infected mice with either S. aureus wild type or the corresponding isogenic mutants Δstp or ΔpknB (Fig. 5A). Here, we used the less cytotoxic Cowan I strain to induce abscess formation and pus production in the mice as described previously (15, 50). Mice were euthanized 5 days p.i; the abscess size was measured, and the pus was harvested for further analysis. The Δstp strain caused significantly larger abscesses compared with ΔpknB, but there was no significant difference in abscess formation between the wild type and Δstp and wild type and ΔpknB (fig. S10A). Analysis of the colony growth heterogeneity showed that Δstp formed a higher proportion of nsSCs (up to 13%) as compared with both wild type (up to 5.9%) and ΔpknB (up to 4.8%), indicating the presence of a larger quiescent subpopulation in the absence of Stp in in vivo abscess settings (Fig. 5B and fig. S10, B and C). Single-cell microscopy confirmed prolonged individual lag phases in Δstp as reflected by the increased proportion of nsSCs (Fig. 5C). To test whether S. aureus lacking the phosphatase Stp formed not only larger abscesses and more nsSCs but also a larger subpopulation of persister cells, we exposed bacteria isolated from murine abscesses to high concentrations of FLU and CIP (40× MIC) for 24 hours and analyzed their survival capacity (Fig. 5D). We found that the Δstp mutant showed significantly higher survival levels (up to 16% in FLU) as compared with the wild type (up to 6% in FLU), correlating with the increased proportion of nsSCs (Fig. 5E and fig. S10, D and E). In addition, the kinase mutant ΔpknB showed reduced antibiotic tolerance; we found significantly lower survival in FLU (up to 3%) as compared with both wild type (up to 6%) and Δstp (up to 16%) and a trend to lower survival compared with the wild type in CIP (Fig. 5D). Together, our data demonstrate that the kinase-phosphatase pair PknB-Stp modulates antibiotic tolerance in S. aureus in mice and that a lack of Ser-Thr dephosphorylation can lead to an increased subpopulation of persisters.

Fig. 5. Deletion of stp leads to increased bacterial lag phase and persister formation in mice.

(A) Scheme of the murine infection model. (B) Colony size heterogeneity in S. aureus Cowan I WT, Δstp, and ΔpknB isolated from murine abscess pus. Proportion of nsSCs was calculated relative to total CFU (fig. S10, B and C). Statistical significance was determined by one-way ANOVA with Tukey’s multiple comparison testing. (C) Lag phase determination by single-cell microscopy in cells isolated from murine abscess pus. Cowan I WT, n = 727 cells; Cowan I Δstp, n = 1038 cells; Cowan I ΔpknB, n = 776 cells. The gray zone marks the period at the beginning of the experiment when cell divisions were not monitored. (D) Survival assay of bacteria harvested from murine abscesses and exposed to FLU or CIP for 24 hours. Survival was calculated relative to the initial inoculum. Statistical significance was determined by a two-way ANOVA with Tukey’s multiple comparison testing after passing the Shapiro-Wilk normality test. (E) Relationship between the proportion of nsSCs and the proportion of persisters. Each symbol represents bacteria from one murine abscess. All data represent six abscesses from three mice. Error bars indicate SEMs. Asterisks above bars indicate significant difference. *P < 0.05, **P < 0.01, and ****P < 0.0001.

Phosphoproteomic analysis reveals protein synthesis as a target of Stp during acid stress

To elucidate the role of Ser-Thr phosphorylation in S. aureus growth regulation and to further identify potential downstream targets of PknB and Stp, we performed total proteomics and phosphoproteomic analyses on the pH 5.5–exposed strain Cowan I wild type and its respective pknB and stp deletion mutants (Fig. 6A). A heatmap overview of all identified phosphopeptides (Fig. 6B) showed a comparable pattern in wild type and ΔpknB, with only 33 significantly underrepresented and seven overrepresented phosphopeptides in ΔpknB as compared with wild type (fig. S11, A to C). As previously reported (51), S. aureus favors single phosphorylated peptides, with more than 80% of identified phosphopeptides phosphorylated only at one position. However, deletion of stp led to an overall increase of phosphopeptides, with a higher percentage of peptides phosphorylated at either one or two positions, whereas deletion of pknB showed no major differences as compared with the wild type (Fig. 6C), suggesting the presence of additional Ser or Thr kinases in S. aureus. A prominent difference in phosphopeptides was revealed between Δstp and wild-type, with 240 phosphopeptides significantly overrepresented (90.0% on Thr) and 137 significantly underrepresented (55.5% on Ser; Fig. 6D and fig. S11C). At the level of the entire proteome, 1550 proteins (coverage of 59.5%) were identified, of which 2.8% (43 proteins) were differentially expressed. Specifically, 1.5% (23 proteins) were found to have significantly lower abundance, and 1.3% (20 proteins) had significantly increased abundance in Δstp as compared with the wild type (Fig. 6D). Consequently, most of the changes occurred exclusively on the phosphoproteome and not on the total proteome (fig. S11C), which is in line with previously published data (51). Phosphorylation of ribosomal proteins such as RplS (pThr¹¹), RpsC (pThr¹⁷⁶), and RpsP (pThr⁷⁰); translation elongation factors EF-Tu (pThr³⁴) and EF-G (pThr³⁴, pThr⁴³, and pThr⁴²⁴); and cell division proteins such as FtsZ (pThr³³³, pSer³³⁸, and pThr³⁶⁷) were prominently overrepresented in the stp deletion mutant as compared with the wild type (Fig. 6, D to F). Deletion of the kinase PknB, on the other hand, showed a similar phosphorylation pattern as the wild type for both ribosomal proteins and the translation elongation factor EF-G (fig. S11, D and E). Phosphorylation of translation elongation factors has been linked to reduced translational activity and bacterial dormancy in B. subtilis (37), and our data show that phosphorylation of translation elongation factors plays an important role in S. aureus as well (Fig. 6, F and G). The increased phosphorylation on several Thr residues in EF-G, for example, the 16-fold increase in pThr⁴³, was confirmed by immunoblotting, wherein only in Δstp the antibody signal from EF-G and pThr overlapped but did not align in the wild type or in ΔpknB (Fig. 6G and fig. S11F). Similar results were found for S. aureus USA300 JE2 (fig. S11G).

Fig. 6. Protein abundance and phosphorylation alterations in acid-stressed S. aureus.

(A) Experimental workflow for proteomic and phosphoproteomic analysis of S. aureus Cowan I WT, Δstp, and ΔpknB grown at pH 5.5 for 3 days by label-free LC-MS/MS. The scheme was created with BioRender. (B) Heatmap of hierarchical cluster analysis of phosphopeptide abundance in phosphoproteomes (N = 4). Blue indicates less abundant phosphopeptides; red indicates more abundant phosphopeptides. Black indicates nondetected phosphopeptides. (C) Number of identified phosphorylation sites and their multiplicity. (D) Pie charts and volcano plots comparing protein changes in the total proteome and phosphopeptide changes between Cowan I Δstp and WT, identifying differentially expressed proteins or phosphopeptides (cutoff adjusted P < 0.05 and a FC of ∣log₂FC∣ > 1). Underrepresented proteins and phosphopeptides are depicted by blue symbols, and overrepresented proteins and phosphopeptides are depicted with red symbols. Symbol type indicates KEGG pathway. (E) Changes in Ser-Thr phosphorylation of ribosomal proteins in Cowan I Δstp as compared with WT. Green color indicates increased Ser-Thr phosphorylation (at least one phospho-site log₂FC ≧ 1, adjusted P < 0.05). Blue color indicates decreased Ser-Thr phosphorylation (at least one phosphosite log₂FC ≦ −1, adjusted P < 0.05). If an increased and decreased abundance of phosphopeptides for the same protein were found, then only the one with the greatest change is shown. (F) Phosphorylation (P) sites to protein plot for the elongation factor EF-G. Phosphorylation sites and exact amino acid positions are indicated. Y axis indicates log₂ FCs of phosphopeptides in Cowan I Δstp as compared with WT. Dashed lines indicate pseudo-FCs (phosphopeptides were only detected in one condition). Statistical significance in phosphopeptides is shown as *. P values in (D) to (F) were determined by a moderated t test as implemented in the R package limma and were adjusted for multiple testing using the Benjamini-Hochberg method. Significance threshold was set to 0.05. (G) Representative immunoblots of bacterial cell lysates with an antibody specific for phospho-Thr and rabbit serum raised against B. subtilis EF-G that also recognizes S. aureus EF-G. Cowan I WT, Δstp, and ΔpknB were grown at pH 5.5 for 3 days; B. subtilis was grown to stationary phase for 16 hours. (H) Representative images of OPP-labeled Cowan I WT, Δstp, and Δpknb. Fluorescence indicates protein synthesis. Scale bar, 20 μm. (I) Total fluorescence of OPP-labeled S. aureus normalized to CFUs. (J) Total isolated protein from S. aureus Cowan I WT, Δstp, and ΔpknB grown at pH 5.5 for 3 days. Statistical significance was determined by Mann-Whitney U test. Error bars indicate SEMs. Asterisks above bars indicate significant difference as indicated. *P < 0.05, **P < 0.01, and ***P < 0.001.

To assess whether the overrepresented phosphorylation of ribosomal and translation elongation factors is associated with differences in protein synthesis activity, we again measured incorporation of OPP to monitor protein synthesis in pH 5.5–exposed S. aureus Cowan I. We found that the stp deletion mutant showed significantly reduced protein synthesis activity as compared with the wild type, whereas the pknB deletion mutant displayed increased protein synthesis (Fig. 6, H and I). Similarly, Δstp showed significantly reduced total isolated protein concentration as compared with wild type and ΔpknB (Fig. 6J). Together, these data suggest that phosphoregulation of proteins involved in translation leads to reduced protein synthesis in S. aureus during acid stress exposure.

In summary, the results presented here demonstrate that the absence of Stp and, therefore, increased Ser-Thr phosphorylation promote features associated with S. aureus quiescence and persistence, including increased heterogeneity within the population, reduced protein synthesis and intracellular ATP, and enhanced antibiotic tolerance (Fig. 7A and fig. S12). The loss of Ser-Thr phosphorylation mediated by PknB had opposite but milder effects on these phenotypes, suggesting that additional Ser-Thr kinases likely contribute to quiescence and persistence.

Fig. 7. Phosphorylation changes linked to cellular quiescence and antibiotic tolerance in S. aureus.

(A) Summary of the individual assay results for S. aureus Cowan I Δstp and ΔpknB relative to WT. Dots represent individual biological replicates. Statistically significant deviation from WT was calculated by Wilcoxon matched pairs signed-rank test. (B) Scheme summarizing the results of this study. Sensing of extracellular stressors or metabolic precursors by the eukaryote-like Ser-Thr kinase PknB leads to phosphorylation of proteins involved in protein synthesis, cell division, and growth in S. aureus, resulting in reduced protein synthesis, cell division, and ATP concentration. This metabolic quiescence leads to prolonged lag phases and thus to a delayed onset of growth after switching to favorable conditions. This growth delay manifests in the formation of nsSCs and eventually in colony size heterogeneity on agar plates. Ultimately, acid-stressed S. aureus can survive antibiotic challenges and shows increased antibiotic tolerance as compared with non–acid-stressed cells. Yellow circles indicate phosphorylated residues. Stp, Ser-Thr phosphatase; FtsZ, core divisome component protein; GTP, guanosine triphosphate; ROS, reactive oxygen species. The diagram was created with BioRender.

DISCUSSION

Here, we showed that the Ser-Thr kinase PknB and its cognate Ser-Thr phosphatase Stp regulate antibiotic tolerance in S. aureus (Fig. 7B). Measurement of protein synthesis activity showed significantly reduced protein synthesis consistent with increased Ser-Thr phosphorylation after acid stress in the absence of the phosphatase Stp. Similarly, deletion of stp increased the subpopulation of quiescent bacteria upon stress exposure, which was reflected by increased amounts of nsSCs and by lag phase analysis of individual bacterial cells using single-cell microscopy. This was also true for wild-type bacteria that showed an increased lag phase after pH 5.5 exposure as compared with those exposed to pH 7.4. Additional analysis revealed a decreased concentration of intracellular ATP in the Δstp mutant, further indicating a metabolically reduced state of the bacteria. Both in vitro acid stress and in vivo stress exposure in murine abscesses led to increased numbers of bacteria, showing substantial growth delays. Our data showed that the growth delay and the reduced ATP concentration of a Δstp mutant related to its increased ability to survive high concentrations of antibiotics from different classes (Fig. 7A and fig. S12A). Last, phosphoproteomic analysis of S. aureus Cowan I revealed increased phosphorylation of proteins involved in translation such as ribosomal proteins and translation factors in Δstp as compared with wild type (Fig. 6, D to G).

PknB was initially identified in a genetic screen designed to detect factors that regulate methicillin resistance in S. aureus (52). It is highly conserved throughout all sequenced S. aureus genomes (24, 32, 53) and has a domain architecture consisting of extracellular penicillin-binding protein– and STK-associated domains, a transmembrane domain, and an intracellular catalytic kinase domain that is similar to B. subtilis PrkC and to Mycobacterium tuberculosis PknB (33). Previous phosphoproteomic studies identified PknB target proteins involved in metabolic processes such as glycolysis, citrate cycle, cell wall synthesis, and protein synthesis (25, 51, 54). Both PknB and Stp play an important role in cell wall synthesis, and it has been shown that deletion of stp leads to a thickened and more robust cell wall in S. aureus, resulting in decreased susceptibility to lysostaphin and VAN (29, 53-55). Similarly, we observed a twofold increase in the MICs for cell wall–targeting antibiotics such as FLU and VAN (table S1). However, because we used very high antibiotic concentrations for our killing assays, doubled MIC values for these antibiotics do not alone explain the increased survival of the stp deletion mutants during challenge with them.

We previously found increased amounts of proteins involved in protein, nucleotide, and amino acid biosynthesis and signal transduction, including PknB, in a persister-enriched population of a clinical isolate of S. aureus that was isolated from a difficult-to-treat infection (50). Here, we identified strong threonine phosphorylation of PknB (pThr¹⁶⁴) in Cowan I wild type after pH 5.5 exposure, indicating autoactivation of PknB as previously described for M. tuberculosis PknB (56, 57). Higher amounts of PknB might lead to autophosphorylation and increased target phosphorylation that, in turn, can reduce translation, alter cell wall synthesis, and lead to decreased intracellular ATP concentrations. Inhibition of protein synthesis by phosphorylation of the EF-Tu in B. subtilis has been shown to occur during nutrient limitation and ultimately leads to the formation of dormant spores, an extreme form of bacterial persistence (37). In Escherichia coli, phosphorylation of EF-Tu leads to its inactivation and to the inhibition of protein synthesis during environmental stress such as starvation (58). In general, inhibition of protein synthesis has been shown to increase antibiotic tolerance in E. coli (59, 60).

Here, we show that deletion of stp leads to increased threonine phosphorylation (pThr³⁴, pThr⁴³, and pThr⁴²⁴) of EF-G and extensive phosphorylation of ribosomal proteins in acid stress–exposed S. aureus. Together with reduced protein synthesis, low ATP concentration, and a delayed onset of growth after stress removal, this correlated with increased antibiotic tolerance in S. aureus. It has been shown that E. coli exposed to cyclic antibiotic treatment evolves high antibiotic tolerance and displays reduced expression of EF-G and decreased abundance of most proteins when compared with its parental strain (61). Because EF-G catalyzes the translocation of tRNA and mRNA through the ribosome, phosphorylation of EF-G can be linked to reduced protein synthesis. In addition, proteins involved in cell division and bacterial growth were found to be phosphorylated by PknB. For example, the core divisome component FtsZ showed increased phosphorylation in the absence of Stp after acid stress exposure. Phosphorylation of FtsZ might inhibit its GTPase activity, leading to arrested cell division (62, 63). Furthermore, it has been shown that Ser-Thr phosphorylation of FtsZ is important for formation of a complex with the cell division protein FtsQ in M. tuberculosis during oxidative stress, highlighting the importance of Ser-Thr phosphorylation in maintaining bacteria in a division-competent state depending on the growth conditions (64).

Bacterial growth delays and cellular quiescence have been linked to antibiotic tolerance, and the presence of nsSCs can be indicative of a subpopulation of persister cells (15, 65, 66). The increased proportion of nsSCs in S. aureus lacking Stp and the growth delay observed on the single-cell level are associated with reduced energy levels and increased bacterial survival during antibiotic exposure. Reduced intracellular ATP concentrations have been linked to antibiotic tolerance because most antibiotics require an active target and bacterial metabolism to eradicate bacteria (67-70). Growth delays, nsSCs, and antibiotic persistence were detected not only after in vitro stress exposure but also in vivo after harvesting bacteria from murine abscesses.

Because deletion of the kinase PknB did not eliminate Ser-Thr phosphorylation in S. aureus, other phosphodonors, such as acetyl phosphate, ammonium hydrogen, and carbamoyl phosphate, and other currently unidentified Ser-Thr kinases (51, 71-73) may participate in this response. S. aureus harbors two other known serine kinases, RsbW and HprK, which might explain the identified residual Ser phosphorylation in cells lacking PknB, but their targets are not well characterized (74). Our data confirm recent findings by Prust et al. (51) that deletion of pknB does not abolish Thr phosphorylation, suggesting that there are other Ser-Thr kinases present.

In this study, we generated a comprehensive map of the phosphoproteome of S. aureus Cowan I and its isogenic Δstp and Δpknb mutant strains exposed to acidic pH, linking Ser-Thr phosphorylation to reduced protein synthesis and intracellular ATP concentrations and thereby to increased bacterial survival during antibiotic exposure. The kinase inhibitor staurosporine increases methicillin-resistant S. aureus sensitivity to the β-lactam antibiotic nafcillin (75), suggesting that, together with the data presented here, inhibition of PknB during infection might be beneficial for increasing susceptibility to certain antibiotics and reducing the risk of persister cell formation. On the other hand, with rising interest in anti-virulence therapy, inhibition of the phosphatase Stp should be followed with caution and needs further evaluation. The combination of Stp inhibitors with environmental stressors typically found in patients could favor antibiotic tolerance by inducing the formation of quiescent bacteria. Hence, the kinase-phosphatase pair PknB and Stp might offer a future target to prevent persister formation during chronic and difficult-to-treat infections that require prolonged antibiotic therapies.

MATERIALS AND METHODS

Bacterial culture and growth conditions

Bacterial strains used in this study are listed in table S2. S. aureus was grown on Columbia agar plates supplemented with 5% sheep blood (bioMérieux) overnight before use or on tryptic soy broth (TSB; Becton Dickinson) containing 1.5% agar (Becton Dickinson) for bacterial enumeration. For exponential growth conditions, bacteria were grown in TSB overnight (16 hours), shaking (220 rpm) at 37°C, diluted, and regrown in TSB, shaking at 37°C, for 2 hours. For pH stress assays, S. aureus was inoculated into defined pH media [Dulbecco’s modified Eagle medium (DMEM), pH 4.0, 5.5, and 7.4] at a starting optical density at 600 nm (OD₆₀₀) of 0.2. The media consisted of DMEM [glucose (4.5 g ml⁻¹), without phenol red, pyruvate, and l-glutamine; Life Technologies] supplemented with 10% fetal bovine serum (FBS), 4 mM of l-glutamine (Life Technologies), and buffers prepared with Na₂HPO₄ and citric acid to ensure a stable pH of 5.5 and 4. pH 7.4 was maintained by 50 mM Hepes buffer. After incubation at 37°C and 5% CO₂ for 3 days and daily mixing, cultures were analyzed by plating and counting nsSCs or by determining antibiotic persistence in a survival assay. nsSC enumeration was performed as described previously (15). For growth curves, bacteria were grown to stationary phase in TSB shaking at 37°C for 16 hours and diluted to OD₆₀₀ of 0.005 in fresh, prewarmed TSB in a 96-well flat-bottom plate (total volume, 200 μl). Bacteria were incubated at 37°C and shaking at 220 rpm for 12 hours, and OD₆₀₀ was measured every 30 min on a Tecan m200 nano plate reader. Specific growth conditions are described in the corresponding method sections.

Genetic manipulations

For both strains, Cowan I and USA300 JE2, unmarked deletions for either pknB or stp were created using the pIMAY vector as previously described (76). Briefly, S. aureus strains were transformed with pIMAY containing the modified genomic sequences missing either pknB (pSP11; fig. S13) or stp (pSP13; fig. S14) by electroporation. Primers and plasmids used to create pSP11 and pSP13 are described in table S3. Colonies grown on TSB agar plates supplemented with chloramphenicol (10 μg/ml), containing the plasmid, were used for chromosomal integration by homologous recombination. Deletion of pknB and stp was confirmed by polymerase chain reaction (PCR) and sequencing. To rule out inherent growth defects of the deletion mutants, we performed growth curves and did not identify any differences in growth of the mutants compared with their wild types (fig. S2, A and B).

To construct the double deletion of stp/pknB and introduce the noncatalytic pknB (K39M) mutation, we followed the protocol of Monk and Stinear (77). Genomic DNA from either USA300 or Cowan I was used as a template for the PCR amplification (primers are described in table S3) of the respective constructs by splicing by overhang extension (SOE)-PCR. Amplimers were introduced into pIMAY-Z by seamless ligation cloning extract (SLiCE) cloning and then transformed into either E. coli IM08B for USA300 or E. coli IM30B for Cowan I. Electroporation into S. aureus, allelic exchange, and conformation by whole-genome sequencing (Illumina, NextSeq) were conducted as described previously (77). To screen for the PknB^K39M point mutation, we performed colony PCR with primers IM1535/IM255 (annealing at 66°C; table S3). This yielded a product (~600 base pairs) in the mutant but not the wild type.

Eukaryotic cell infections

The human lung epithelial carcinoma cell line A549 [the American Type Culture Collection (ATCC) CCL-185] was grown in DMEM and glucose (4.5 g ml⁻¹; Life Technologies) with 10% FBS (GE Healthcare) and 2 mM l-glutamine (Life Technologies) and infected with S. aureus strain Cowan I at a multiplicity of infection of 1. Three hours after infection, the cells were washed with phosphate-buffered saline (PBS; Life Technologies), and FLU (1 mg ml⁻¹ in DMEM; Actavis) and lysostaphin (25 μg ml⁻¹ in DMEM; Sigma-Aldrich) were added to kill any extracellular bacteria. To evaluate the initial number of intracellular bacteria and the proportion of nsSCs, cells were washed three times with PBS, lysed [0.08% Triton X-100 (Fluka) in PBS], and plated on Columbia agar plates supplemented with 5% sheep blood for colony-forming units (CFUs), with nsSC enumeration 30 min after the addition of FLU and lysostaphin.

Every day, the washing step was repeated, and fresh medium [containing FLU (1 mg ml⁻¹) and lysostaphin (25 μg ml⁻¹)] was added to the host cells. Supernatants were monitored for the absence of any extracellular bacteria by plating. Three, 5, and 7 days p.i., host cells were washed three times with PBS, lysed, and plated on Columbia agar plates supplemented with 5% sheep blood for CFUs and nsSC enumeration.

Single-cell time-lapse microscopy

Agar pads were prepared using a custom-made device made of polyvinyl chloride. Bacterial dilutions were added onto solidified 2% agar (bacteriological grade, BD) with Columbia media (BD) and 5% sheep blood (Life Technologies) and then covered with a cover glass. The setup was then placed under a microscope at 37°C at the latest 30 min after inoculation. Bright-field images were acquired using a fully automated Olympus IX81 inverted microscope at 100× resolution (U-FLN-Oil lens) and cellSens software. Up to 3000 positions were monitored per experiment, and the lag phase of 134 to 2000 bacteria per strain was analyzed (15). Statistical significance was determined for time points where 90% of bacteria started to divide using unpaired t tests.

Murine infection

The protocols (ZH251/14 and ZH050/18) were approved by the Institutional Animal Care and Use Committee of the University of Zurich, and all experiments were conducted in approval of the Cantonal Veterinary Office Zurich.

S. aureus Cowan I wild type and the isogenic mutants ΔpknB and Δstp were grown to logarithmic phase, and 10⁸ CFUs were mixed 1:1 with Cytodex beads (Sigma-Aldrich) (78). Bacteria were injected into the flanks of 7- to 9-week-old female wild-type C57BL/6 mice (Janvier Labs, France). Mice were euthanized 5 days p.i. Abscess area was measured, and abscess volume was calculated as previously described (79). Abscess pus was harvested, homogenized, lysed, and directly analyzed by single-cell time-lapse microscopy and plating serial dilutions for bacterial enumeration and nsSC quantification. In addition, bacteria from abscess pus were used for subsequent persister assays.

ATP assay

Intracellular ATP concentrations were measured using a BacTiter-Glo kit (Promega) according to the manufacturer’s instructions. We washed S. aureus grown in DMEM supplemented with 10% FBS and 1% l-glutamine at pH 5.5 for up to 3 days with PBS at indicated time points before measuring luminescence on a SpectraMax i3 MiniMax 300 plate reader (Molecular Devices) in white 96-well flat-bottom plates (Greiner Bio-One). ATP concentrations were normalized to CFU counts. N = 3 to 6 biological replicates.

Persister assay and time kill curves

About 10⁵ bacteria from different stress conditions (pH 5.5 or murine abscess pus) were inoculated in DMEM (pH 7.4) supplemented with antibiotics (40× MIC) as indicated. For time kill curves, bacterial survival was determined by CFU enumeration after 1.5, 3, 6, and 24 hours of incubation at 37°C and 5% CO₂ and calculated relative to the inoculum. For a single–time point evaluation, bacterial survival was determined after 24 hours of incubation with antibiotics. MICs for FLU, VAN, RIF, and CIP were assessed by microbroth dilution method (table S3), following the European Committee on Antimicrobial Susceptibility Testing recommendations. Minimum duration of killing 90 or 99% of the initial population (MDK90/99) was calculated by linear interpolation for each replicate.

OPP staining and translational activity measurement

Click-iT Plus OPP Alexa Fluor 488 Protein Synthesis Kit (Invitrogen) was used to label S. aureus with OPP following the manufacturer’s instructions. Briefly, 450 μl of bacteria exposed to pH 5.5 or 7.4 for up to 3 days were washed with PBS at indicated time points and resuspended in prewarmed TSB supplemented with 20 μM OPP. Labeling was performed at 37°C and 220-rpm shaking for 20 min. All subsequent steps were performed at room temperature. After incubation, bacteria were washed with PBS and fixed in 3.7% formalin for 15 min. To permeabilize the bacteria, fixative was removed, and bacteria were incubated in 100 μl of 0.5% Triton X-100 in PBS for 15 min. After washing twice with PBS, bacteria were labeled using 1× Click-iT cocktail for 30 min in the dark. Bacteria were harvested and washed once with Click-iT rinse buffer and resuspended in 40 μl of PBS for imaging or 150 μl of PBS for fluorescence measurement (excitation, 488 nm and emission, 530 nm) on a SpectraMax i3 MiniMax 300 plate reader in black 96-well flat-bottom plates.

Proteomics sample preparation

A total of 12 samples (four biological replicates for each strain) for dataset 1 (comparison of pH 5.5 grown strains) and eight samples (four biological replicates for Cowan I wild type pH 5.5 and 7.4) for dataset 2 (comparing Cowan I wild type pH 5.5 versus 7.4) were processed for MS-based proteomics. To extract the proteins, 30 μl of SDS lysis buffer [4% (w/v) SDS, 100 mM tris-HCl (pH 8.2), and 0.1 M dithiothreitol complemented with Halt Protease and Phosphatase Inhibitor (Thermo Fisher Scientific)] were added to the frozen cell pellets. Samples were mechanically lysed with glass beads in a tissue lyser (QIAGEN) for four cycles of 2 min, running at 30 Hz. Between the cycles, samples were cooled down on ice. After we spun down the tubes, samples were boiled for 5 min at 95°C and placed 1 min in high-intensity focused ultrasound sonication at 100% amplitude, 80% cycle (UP200St with VialTweeter, Hielscher Ultrasonics Technology). Samples were then diluted with 200 μl of 8 M urea and 100 mM tris-HCl (pH 8.2) and centrifuged for 10 min at 8000g at 4°C. The clear supernatant was transferred into a new tube. Protein concentration was estimated with Qubit (Invitrogen, CA, USA) or Lunatic (Unchained Labs, CA, USA) measurement for runs 1 and 2, respectively. Before filter-aided sample preparation (FASP) (80), protein extracts were incubated for 45 min with 1% (v/v) Benzonase Nuclease (Sigma-Aldrich, E1014). For the FASP digestion, the clear protein lysate was deposited in a Microcon-30kDa filter unit (Millipore, MRCF0R030) and centrifuged at 14,000g for 20 min to pass the totality of the sample through the filter. After a wash step with 200 μl of buffered urea [8 M urea and 100 mM tris-HCl (pH 8.2)] with the same centrifugation parameters as above, samples were incubated for 10 min with 50 mM iodoacetamide. The filter units were subsequently washed with 3× 100 μl of urea and 2× 100 μl of 50 mM triethylammonium bicarbonate (TEAB). For overnight digestion at room temperature, sample filters were incubated in a wet cell, adding 1:50 (w/w) trypsin:protein ratio (Promega, V5111), and solubilized in 120 μl of 50 mM TEAB. The next day, peptides were recovered by centrifugation, and a small aliquot of 3 μl was taken for proteome analysis and diluted with 3% acetonitrile and 0.1% formic acid to 0.3 raw absorbance at 280 nm (NanoDrop, DeNovix).

Phosphopeptide enrichment

For ferric nitrilotriacetate (Fe-NTA) phosphoenrichment, tryptic peptides were first dried in an ultra-vacuum centrifugation. Then, equal amounts for each sample (600 μg based on protein estimation) were enriched by Fe-NTA spin columns (Thermo Fisher Scientific, A32992) following the provided instructions. Dried phosphopeptides were redissolved in 10 μl of 3% acetonitrile and 0.1% formic acid for MS injections.

Mass spectrometry

MS analyses were performed on an ACQUITY UPLC M-Class System (Waters) coupled to an Orbitrap Fusion Lumos (Thermo Fisher Scientific). For LC-MS/MS injections of proteome and phosphoproteome samples, 2 and 4 μl, respectively, were loaded on a trap column [nanoEase M/Z Symmetry C18 100 A, 5 μm, 1/PK 180 μm by 20 mm (Waters)] and separated on the analytical column [nano-Ease M/Z HSS C18 T3 Col 100 A, 1.8 μm, 1 P/K 75 μm by 250 mm (Waters)] using a linear gradient of 5 to 24% to 24 to 36% acetonitrile (0.1% formic acid) over 90 min (80- to 10-min step) followed by a 20-min reequilibration time. The analytical column was temperature-controlled at 50°C, and the flow rate of 300 nl/min was constant over the chromatographic separation.

MS survey scans from mass/charge ratio (m/z) 300 to 1500 were acquired at a resolution of 120,000 at m/z 200 in profile mode, whereas tandem mass spectra were acquired in the ion trap in centroid mode. For higher-energy collisional dissociation (HCD) fragmentation, a normalized collision energy of 35 was applied, and isolation windows of m/z 0.8 (proteome) or 1.2 (phosphoproteome) were used for MS2 acquisitions. The automatic gain control (AGC) target was set to 10 × 10⁵ for both types of samples, whereas the maximum injection time was either 50 (proteome) or 120 ms (phosphoproteome). The MS proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE (81) partner repository with the dataset identifiers PXD025124 (dataset 1) and PXD035015 (dataset 2).

Protein identification and quantification

Peptide and protein identification/quantification were performed using label-free quantification and match between run option in MaxQuant (version 1.6.2.3). Spectra were searched against the UniProt S. aureus reference database (downloaded 27 November 2020), searching for the variable modifications acetylation (protein N-term), oxidation (M), and, if applicable, phosphorylation (STY) at 1% false discovery rate (FDR). For quantification, only protein groups with at least two peptides were kept. For the phosphoproteome, the MaxQuant output report was used. Intensities were normalized to systematically remove variability in abundance due to different sample loads on the LC column. The log₂-transformed intensities were z-scored and updated by the average of the SD in all samples. After normalization, all samples had a similar intensity distribution for computing a moderated t test (82), using the R package limma (83). All these functions are implemented in the R package SRMService (84). The resulting P values were adjusted for multiple testing using the Benjamini-Hochberg method to control the FDR. Significance threshold was set to 5%. Pathway information of proteins was retrieved from the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (85) for visualization purposes.

Immunoblotting

Bacterial total protein was isolated by bacterial lysis using glass beads. Briefly, bacteria were washed twice with ice-cold PBS and resuspended in ice-cold PBS supplemented with 1× Halt Protease and Phosphatase Inhibitor (Thermo Fisher Scientific). Bacteria were mechanically lysed with glass beads in a tissue lyser (QIAGEN) for three cycles of 1.5 min, running at 30 Hz. Between the cycles, samples were cooled down on ice. Cell debris and intact cells were removed by centrifugation (1200g for 10 min at 4°C).

Protein concentration was determined by Bradford assay (86). Ten micrograms of total protein were separated by SDS–polyacrylamide gel electrophoresis (10% acrylamide) and transferred onto a nitrocellulose membrane. Membranes were stained for total protein using the Novex Reversible Membrane Protein Stain Kit (Invitrogen) following the manufacturer’s instructions. After destaining, membranes were blocked using 4% bovine serum albumin (Sigma-Aldrich) in tris-buffered saline for 2 hours at room temperature. An antibody recognizing phospho-Thr [phospho-Thr (42H4) mouse monoclonal antibody no. 9386; Cell Signaling Technology] was used to detect Thr phosphorylation, and a polyclonal rabbit serum was used to detect the translational EF-G, both from S. aureus and B. subtilis. Visualizing was performed using IRDye secondary antibodies (LI-COR) and the Fusion FX6 EDGE Imaging System (Vilber Lourmat S.A.).

Quantitative hemolysis assay

Overnight cultures of USA300 JE2 wild type and its corresponding mutants were grown in Todd-Hewitt medium as previously described (87, 88). OD was adjusted to an OD₆₀₀ of 2, and bacteria were centrifuged; supernatant was sterile-filtered, and 100 μl was added to 100 μl of washed 5% sheep blood (Thermo Fisher Scientific). After incubation at 37°C for 30 min and 4°C for 30 min, the suspension was centrifuged at 1600 rpm for 15 min, and hemoglobin absorbance was measured in the supernatant at 415 nm with the VersaMax tunable microplate reader (Molecular Devices).

Cell viability assay

To assess bacterial cytotoxicity, an MTT (3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyltetrazolium bromide) assay was performed. As previously described (89), epithelial cells (A549, ATCC CCL-185) were infected at a multiplicity of infection of 1. After 3.5 hours, extracellular bacteria were inactivated by the addition of FLU (1 mg/ml; Actavis) and lysostaphin (25 μg/ml; Sigma-Aldrich) (invasion). Twenty-four hours after invasion, the tetrazolium dye MTT (Sigma-Aldrich) was added to the medium to a final concentration of 1 mg/ml and incubated for 2 hours at 37°C and 5% CO₂. After removing the medium, the formed formazan crystals were dissolved in 0.04 M HCl in isopropanol. Absorbance was measured at 570 nm with the VersaMax-tunable microplate reader (Molecular Devices). Percentage of survival of the eukaryotic cells was calculated relative to an uninfected control.

Statistical analysis

All experiments, except proteomics data, were analyzed for statistical significance by the Kruskal-Wallis test with Dunn’s post hoc test, Mann-Whitney U test by one-way or two-way analysis of variance (ANOVA) with Tukey’s post hoc test after normality testing by the Shapiro-Wilk normality test, or Wilcoxon matched pairs signed-rank test using GraphPad Prism 8.0.0 (GraphPad Software Inc., La Jolla, CA). Analysis of proteomics data is explained in the “Protein identification and quantification” section. P < 0.05 was considered significant (*P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001).

Supplementary Material

This PDF file includes:

Figs. S1 to S14

Tables S1 to S3

References (76, 90)

Other Supplementary Material for this manuscript includes the following:

Data files S1 to S3

MDAR Reproducibility Checklist