A shared alarmone-GTP switch underlies triggered and spontaneous persistence

Abstract

Persisters describe phenotypically switched cells refractory to antibiotic killing in a genetically susceptible population, while preserving the ability to resume growth when antibiotics are discontinued1,2. Since its proposal 70 years ago, great strides were made to build the framework regarding persistence, including defining triggered, spontaneous and antibiotic-induced persisters. However, challenges remain in characterizing the molecular determinants underlying the phenotypic switch into persistence3. Here we document triggered, spontaneous and antibiotic-induced persistence in a Gram-positive bacterium, all through a common switch involving the alarmone (p)ppGpp and each stemming from a different alarmone synthesis pathway. Starvation-triggered persistence is mediated by Rel synthetase, and spontaneous persistence is through self-amplification via allosteric enzyme activation of alarmone synthetases Rel and SasB, whereas lethal and sublethal concentrations of cell wall antibiotics induce alarmones through an antibiotic-induced alarmone synthetase SasA, consequently enabling adaptive persistence that promotes survival. (p)ppGpp accumulation promotes persistence by depleting intracellular GTP and antagonizing its action. We developed a fluorescent GTP reporter to visualize rare events of persister formation in wild type bacteria, revealing a rapid switch from growth to dormancy in single cells as their GTP levels drop beneath a threshold. While a modest drop of GTP in bulk population slows down growth and promotes antibiotic tolerance, (p)ppGpp drives persistence by allowing the switch-like dynamics to drop GTP beneath the persister threshold in single cells. Persistence through alarmone-GTP antagonism is likely a widespread mechanism to survive antibiotics in Gram positive bacteria and possibly beyond.

Figure 1

(p)ppGpp mediates starvation triggered and spontaneous persistence. (a) Survival curves of Bacillus subtilis wild type (WT) and (p)ppGpp-null mutant ((p)ppGpp0) to lethal concentrations of vancomycin (VAN). MIC: minimal inhibitory concentration. Rekill: cells which survived 20x MIC vancomycin treatments were regrown and treated with vancomycin again. (b) Survival curves of prophage-cured wild type to 20× MIC vancomycin, 20× MIC ciprofloxacin or 8x MIC kanamycin. (c) Starvation-trigggered persistence of WT depends on (p)ppGpp. Biphasic kill curves of vancomycin treatment were performed after several starvation or (p)ppGpp inducing conditions: stationary: stationary phase, RHX: pre-treated with amino acid starvation inducer arginine hydroxamate, +CCCP: pre-treated with ATP synthesis inhibitor CCCP, p-sasA or p-sasASyn: (p)ppGpp0 strains overexpressing the (p)ppGpp synthetase SasA or its synthetase-dead variant SasASyn (SasAD87G). See supplemental data fig 4 for plots in log scale. (d) Nucleotide levels in cells pre-treated with RHX, or ATP synthesis inhibitors CCCP or arsenate (As) for 30 min. Nucleotide levels were measured by thin layer chromatography (TLC) and normalized to their levels before induction. Bottom panel represents persistence to vancomycin treatment of the pre-treated cells. (e-f) Survival of serial-passaged populations to vancomycin for 5 h. Persisters which remained through the passage were defined as spontaneous persisters3. (g) In B. subtilis and many gram-positive bacteria, (p)ppGpp is synthesized by three (p)ppGpp synthetases. Rel is the bifunctional (p)ppGpp synthetase and hydrolase, while SasB (SAS1/RelQ/YjbM) and SasA (SAS2/RelP/YwaC) are monofunctional synthetases. (h) Persister levels in starvation-induced (black) WT and (p)ppGpp mutants with RHX. (i) Levels of spontaneous persisters (green, 2nd passage) in WT and (p)ppGpp mutants. relSyn (RelD264G) is a synthetase inactive variant of Rel which retained its hydrolase activity essential for viability in the presence of sasB and sasA. sasBF42A encodes an allosteric pppGpp binding site mutant of SasB. Values represent mean (n >= 3) and error bars indicate s.d. *: p < 0.05, **: p < 0.01, ns: not significant (Student’s t test).

Figure 2

(p)ppGpp mediates persistence by reducing GTP and is independent of tolerance. (a) (p)ppGpp robustly inhibits de novo GTP biosynthesis by inhibiting the GDP synthetase Gmk and the IMP dehydrogenase GuaB, as well as inhibiting expression of the pur operon through the PurR repressor. (p)ppGpp also inhibits GTP synthesis by purine salvage which is not illustrated here. (b-c) Vancomycin survival curves of wild type (WT), purine biosynthesis mutants ΔpurB, ΔpurL, ΔpurF, GTP synthesis mutants ΔguaA and guaB repressed mutant (guaB↓), slow growth mutants Δpyk (pyruvate kinase) and ΔctaA (heme synthase), as well as (p)ppGpp and (p)ppGpp gmkQ110R. (d) Measurement of low GTP cells in bacterial population. The PlowGTP reporter is activated when cellular GTP level is low. Flow cytometry analysis of wild-type population containing the reporter (~1.5×106 cells). Representative image of cells with low or high PlowGTP fluorescence was shown. (e-j) Correlation between growth rate, tolerance, pesistence, population-averaged PlowGTP signals, and frequency of low GTP cells (high PlowGTP fluorescence cells) in wild-type under different growth conditions. Wild type containing the PlowGTP reporter were grown in minimal media with different carbon sources. Populations were measured for their growth rate, PlowGTP fluorescence, and levels of high PlowGTP fluorescence cells using flow cytometry (~1.5×106 cells), as well as tolerance and persistence from vancomycin survival curves as shown in Supplemental Data Fig 21. Each value represents a single replicate. Three replicates for each condition were shown. r, Pearson correlation coefficient, p, p-value.

Figure 3

Direct observation of persistence switch in cells through (p)ppGpp-GTP antagonism. (a) Schematic of FACS sorting and antibiotic survival assay using the PlowGTP reporter. Bacterial population containing the PlowGTP reporter were sorted based on PlowGTP fluorescence and immediately treated with antibiotic. Survival was measured by serial dilution and plating followed by colony counting. (b) Wild type containing PlowGTP-GFP reporter was FACS-sorted into low or high PlowGTP fluorescence. Approximately 1 × 106 cells were sorted per sample per replicate. Fractions were treated with vancomycin and measured for survival over time. Values represent mean ± s.d., n = 3. See also supplemental data figure 22. (c) Time lapse micrograph of wild type cells containing the PlowGTP reporter in response to antibiotic treatment. Exponentially growing wild type containing the PlowGTP reporter were patched on agarose pads made with growth media. High PlowGTP fluorescence cells were identified (arrow) and monitored for antibiotic survival using carbenicillin and Sytox-blue staining (cyan) for viability. Values in yellow indicate time in min. (d) Survival of low GTP cells in biofilm. Wild type containing both PlowGTP (for low GTP, green) and PrrnBP1 (for high GTP, red) reporters were grown into colony biofilm followed by vancomycin treatment for 6 h. Bar graphs indicate the changes in high GTP and low GTP populations before and after treatment. Four random regions in the biofilm were measured. Values represent mean (n >= 3) and error bars indicate s.d. *: p < 0.05, **: p < 0.01, ns: not significant (Student’s t test). (e) Representative single cell dynamic trace of persister formation. Cells were patched onto agarose pad made with growth media at roughly one cell per field of view, and the monitored for growth till before saturation (see also supplemental data figure 24a-b). A total of roughly 30,000 cell growth and division events were recorded (~150 events per microcolony) to reveal ~10 persistence entrance events. Numbers indicate time in min. Arrows indicate a cell with increased PlowGTP fluorescence followed by growth attenuation. (f) Quantitated changes in cell length vs PlowGTP fluorescence (left) vs PlowGTP fluorescence (right) from the single trace in (e). Dotted lines indicate persistence entrance. (g) Summary data of specific growth rate and reporter fluorescence from ~100 independent single cell traces similar to e (ten entrance traces). Solid lines indicate non-linear regression (variable slope) model fitted to the data.

Figure 4

Antibiotic-induced persistence through alarmone accumulation contributes to adaptive survival. (a) Cell wall antibiotics induce PlowGTP reporter fluorescence. WT containing the PlowGTP reporter were treated with 20× MIC ciprofloxacin (CIP), 4× MIC kanamycin (KAN), 200× MIC carbenicillin (CARB), 20x MIC vancomycin (VAN) or 3× MIC bacitracin (BAC) for 2 h. Fluorescence was measured using microscopy (> 200 cells each, n = 3) UT: before treatment. (b to d) Levels of ppGpp, ppApp and GTP in wild type (WT) treated with sublethal or lethal concentrations of bacitracin for 0.5 h. UT: before treatment, SL: 0.5× MIC bacitracin, L: 3x MIC bacitracin. (e-f) Levels of cell wall antibiotic-induced persisters in WT and (p)ppGpp mutants. (e) Strains containing the PlowGTP reporter before and after 0.5× MIC bacitracin treatment for 0.5 h were measured for generation of low GTP cells using flow cytometry (~1×106 cells per sample per replicate, n = 3, see also Supplemental data figure 28). Levels of induced low GTP cells were determined by subtracting the levels in uninduced populations. (f) Survival of WT or (p)ppGpp mutants pretreated with 0.5× MIC for 0.5 h, followed by lethal (3 × MIC) bacitracin treatment for up to 5 h. Levels of induced persistence were determined by subtracting survival of uninduced populations at 5 h. See also Supplemental data figure 29. Values represent mean and error bars represent s.d. Statistical tests were done between WT and mutant pairs. *: p < 0.05, **: p < 0.01, ns: not significant (Student’s t test). (g) Survival curve of WT or ΔsasA pretreated with sublethal (SL, 0.5× MIC) bacitracin for 0.5 h, followed by lethal (L, 3 × MIC) bacitracin treatment for up to 5 h. Statistical tests were done between WT and mutant pairs. *: p < 0.05, **: p < 0.01, ns: not significant (Student’s t test). (h) Model for persistence pathways mediated by (p)ppGpp-GTP antagonism. Starvation-triggered persistence (purple) is mediated primarily through the (p)ppGpp synthetase Rel. Spontaneous persistence is mediated through activities of two (p)ppGpp synthetases Rel and SasB (green). Persisters can also be induced by cell wall antibiotics through SasA (orange). Thus, (p)ppGpp enables integration of different signals to trigger persistence through GTP depletion.