ABSTRACT
Stress is ubiquitous for bacteria and converts a subpopulation of cells into a dormant state known as persistence, in which cells are tolerant to antimicrobials. These cells revive rapidly when the stress is removed and are likely the cause of many recurring infections such as those associated with tuberculosis, cystic fibrosis, and Lyme disease. However, how persister cells are formed is not understood well. Here we propose the ppGpp ribosome dimerization persister (PRDP) model in which the alarmone guanosine pentaphosphate/tetraphosphate (henceforth ppGpp) generates persister cells directly by inactivating ribosomes via the ribosome modulation factor (RMF) and the hibernation promoting factor (Hpf). We demonstrate both RMF and HpF increase persistence and reduce single-cell persister resuscitation and that ppGpp has no effect on single-cell persister resuscitation. Hence, a direct connection between ppGpp and persistence is shown.
INTRODUCTION
Persisters may are formed from myriad stresses such as nutrient, antibiotic, and oxidative stress (1, 2). In this resting state, persisters survive stresses not by actively combatting the stress but by sleeping through it (3). We have shown that nutrient stress creates persister cells that are equivalent to the cells known as viable but non-culturable (2). Since, nearly all tested cells form persisters (4) and almost all cells will face starvation, the persister state is probably an universal resting state (5). Unfortunately, since persister cells usually exist as a sub-population of less than 1% of cells, they have been difficult to study and little is known about how they form.
By inhibiting the activity of ribosomes, we have developed means to convert E. coli cells into a population that consists solely of persister cells (6) so that insights may be made about persister cell resuscitation. Using this approach, which has been verified by six independent labs to produce bona fide persister cells (7–12), we have shown that persister cells resuscitate rapidly primarily in response to external signals (13), such as fresh nutrients, rather than stochastically (14). Critically, persister cell resuscitation is heterogeneous and depends on the number of active ribosomes; once a threshold of active ribosomes is reached, persister cells elongate and divide (13). We also determined that persister cells sense external nutrients via their chemotaxis (for amino acids) and phosphotransferase membrane proteins (for glucose) (14) rather than utilizing specialized proteins for resuscitation. These external signals are propagated internally from the membrane sensors by reducing cAMP which leads to the rescue of stalled ribosomes and to the dissociation of inactive dimerized 100S ribosomes (14). Furthermore, resuscitating cells undergo chemotaxis toward fresh nutrients since it was the lack of nutrients that caused the persistence in the first place (14).
In contrast to how persister cells revive, many questions remain about how persister cells form. It is clear that inhibiting translation by stopping transcription, corrupting ribosomes, and depriving ribosomes of ATP results in a 10,000-fold increase in persistence (6), so inactivation of ribosome activity is imperative for persistence. This is logical in that cells are roughly 50% protein and over 90% of ATP is used for protein synthesis (15), so inhibiting translation effectively makes cells dormant.
Beyond translation inactivation, there is consensus for a role of ppGpp in the formation of persister cells (16) since ppGpp is required for high persister levels in E. coli (17, 18) and Pseudomonas aeruginosa (19). In E. coli, ppGpp is formed in the stationary phase in response to stress such as nutrient depletion (specifically, due to a lack of charged tRNAs); ppGpp is synthesized by RelA and SpoT (which can also degrade ppGpp) and serves to reduce ribosome synthesis (15) by directly modulating transcription via its interaction with RNA polymerase, via activation of RpoS (sigmaS, the stress response sigma factor for the stationary phase), and via activation of RpoE (sigmaE, the stress response sigma factor for misfolded proteins in the periplasm) (20). Critically, along with re-directing transcription, the alarmone reduces translation by activating the expression of the small (55 aa) ribosome modulation factor (RMF) (21), which converts active 70S ribosomes into inactive 100S ribosomes (22) via an inactive 90S dimer complex (23). This dimerization is essential for survival in the stationary phase (24). Also, ppGpp induces the hibernation promoting factor Hpf (95 aa) which converts 90S ribosomes into 100S ribosomes and is highly conserved as it is found in most bacteria and some plants (25).
Since overproduction of toxins of toxin/antitoxin (TA) systems and production of any toxin increases persistence (18) and since ppGpp is necessary for full activity of toxins MazF (26) and HipA7 (17), the prevailing model is that ppGpp activates TA system toxins (27). In a now defunct model (28), the Gerdes group proposed that persisters cells were formed through ppGpp based on its activation of Lon protease (via polyphosphate); Lon would then degrade preferentially antitoxins so that toxins would be activated. Here we propose a simpler model: ppGpp induces persistence directly through ribosome dimerization, and persister cell resuscitation is through reactivation of ribosomes. Hence, we provide the long-sought connection between ppGpp and persistence as well as show RMF is important not only for the stationary phase but also for persistence.
RESULTS
We discovered that the protein that activates translation by dissociating hibernating 100S ribosomes, the GTPase HflX (29, 30), significantly increases the rate persister cells resuscitate and that the hflX mutation prevented resuscitation (14). Therefore, we hypothesized that dimerization of ribosomes by RMF causes persistence and reactivation of 100S ribosomes resuscitates persister cells. It is well-established that (i) ppGpp is formed during myriad stresses such as depletion of nutrients, phosphate, fatty acids, and iron as well as due to osmotic and acid stress (20) so ppGpp is utilized by stressed cells, (ii) ppGpp induces rmf (21), (iii) RMF dimerizes and inactivates ribosomes (22), and (iv) ppGpp increases persistence (16–19). To complete our model, we investigated the role of RMF in forming persister cells and their resuscitation. We also tested the impact of the 100S dimerization factor Hpf, that converts 90S ribosome into 100S ribosomes (25) and ppGpp on persister cell resuscitation.
RMF and Hpf increase persistence
If RMF increases persistence, then deleting rmf should reduce the number of persister cells with antibiotics. Using ampicillin at 10 MIC (100 µg/mL), we found there was a 1,480-fold reduction in the number of persister cells (Fig. 1A). Similarly, inactivating Hpf resulted in a 12-fold reduction in persistence with ampicillin. Therefore, RMF is more important than Hpf for forming persister cells. In addition, as expected, overproducing both RMF and Hpf reduced single-cell resuscitation, but only moderately (∼20%) (Fig. 1B).
ppGpp doe s not affect waking
Although ppGpp is the single-most important factor for forming persisters, its role in resuscitation in not clear. To investigate this at the single cell level in a manner that allows us to capture initial waking events (13, 14), we converted the complete population into persister cells (6), a process that has been confirmed eight ways (13), and investigated single-cell resuscitation in the absence of ppGpp via the relA spoT mutant. We utilized rich medium since the relA spoT mutant does not grow on minimal medium and found there was little effect of eliminating ppGpp on single-cell persister resuscitation (Fig. 1B) although reviving relA spoT cells were clearly elongated (Fig. S1). These results agree with those found for resuscitating starved P. aeruginosa on agar plates where relA spoT had no effect (33).
RMF and Hfp reduce single-cell waking
We hypothesized that inactivating dimerization factors RMF and Hpf would lead to faster persister cell resuscitation since ribosomes could not be dimerized during persister cell formation; hence, we quantified single cell resuscitation for the isogenic rmf mutant. In agreement with the hypothesis, we found cell resuscitation with minimal glucose medium was increased dramatically (17-fold) (Fig. 1B). Corroborating our model, the isogenic hpf mutant also increased waking dramatically (12-fold) (Fig. 1B). Hence, RMF and Hpf reduce persister waking by maintaining ribosomes in their inactive state.
HflX increases waking only in the presence of RMF
Since inactivating HflX, which dissociates hibernating 100S ribosomes to resuscitate persister cells, prevents cell resuscitation (14), we hypothesized that HflX would only increase waking in a background where RMF was used to induce dormancy (by forming inactive 100S ribosomes). Therefore, we compared producing HflX in the rmf mutant vs. producing HflX in a hflX mutant (so there was no background HflX from the chromosome but there was RMF to dimerize ribosomes) and found there was a four-fold increase when RMF was present (Fig. 1B). Therefore, HflX increases waking when ribosomes are dimerized by RMF.
DISCUSSION
Since we have shown previously that the protein that reactivates translation by dissociating 100S hybridized ribosomes, HflX, is important for single-cell persister resuscitation (14), it is logical that hybridization of ribosomes by RMF should be required for the formation of persister cells. Hence, our ppGpp ribosome dimerization persister (PRDP) model for persister cell formation and resuscitation focuses on ribosome inactivation and activation, respectively (Fig. 2). Critically, 100S ribosome dissociation and recovery is extremely rapid (less than one minute) (25) which allows for rapid persister cell resuscitation; this matches our results where we found the resuscitation of single persister cells is rapid and the speed depends on the number of active ribosomes (13). Furthermore, since RMF is conserved in bacteria (25), our results suggest the PRDP model be general for the formation of the persister state. Also, since our results showing RMF increases persistence with the antibiotic ampicillin are consistent with previous results showing that RMF increases tolerance to netilmicin (34), gentamicin (35), acid (32), osmotic stress (36) and nutrient limitation in the stationary phase (37, 38), the PRDP model has general applicability for explaining how myriad stresses lead to persister cell formation as well as for explaining how persister cells resuscitate from these myriad resting states.
Along with RMF, we found converting 90S ribosomes into 100S ribosomes via Hpf is important for increasing persistence and inactivating Hpf increases waking. Our results differ from those of P. aeruginosa since, in E. coli, inactivating Hpf increases waking dramatically (Fig. 1B) whereas in P. aeruginosa, Hpf is important for both entering dormancy and resuscitation (33). However, both our work and that with P. aeruginosa demonstrate the link of ribosomes with persistence, but our work discovers the direct link between the stringent response (i.e., ppGpp), persistence, and resuscitation.
With this direct link between ppGpp and persistence, our PRDP model bypasses the need for ppGpp to activate toxins of TA systems for the cell to become persistent. Unfortunately, the main links between TA systems and persistence are based on use of the HipA7 variant that lacks toxicity (17, 39). Instead, the roles of TA systems appear now to be primarily phage inhibition (40), plasmid maintenance, stress response, and biofilm formation (5). This fits well with recent results that diminish the role for TA systems in forming persister cells (31) and instead emphasize the role of ppGpp (16). Therefore, RMF and Hpf catalyzing ribosome inactivation through the formation of 100S ribosomes is the elusive connection between the stringent response and persistence.
MATERIALS AND METHODS
Bacteria and growth conditions
The strains (Table S3) were grown routinely in lysogeny broth (41) at 37°C. For overnight cultures, chloramphenicol (30 μg/mL) was used to retain pCA24N-based plasmids (42) and kanamycin (50 μg/mL) was used for deletion mutants.
Persister cells
Persister cells were generated (6, 13) from exponentially-growing cells (turbidity of 0.8 at 600 nm) by adding rifampicin (100 µg/mL) for 30 min to stop transcription, centrifuging, and adding LB with ampicillin (100 µg/mL) for 3 h to lyse any non-persister cells. Cell pellets were washed twice with 0.85% NaCl then re-suspended in 0.85% NaCl.
Persister cell assays
To determine the number of persister cells after antibiotic treatment, exponentially-growing cells (turbidity of 0.8 at 600 nm) were treated with ampicillin (100 µg/mL) for 3 h, washed twice with 0.85% NaCl to remove the ampicillin, diluted in 0.85% NaCl, and the number of persister cells was determined via a drop assay (43).
Single-cell persister resuscitation
Briefly, cell populations (5 µL) of 100% persister cells were added to 1.5% agarose gel pads containing either LB or M9 glucose (0.4 wt%) medium (44) as indicated, and resuscitation was monitored at 37°C via a light microscope (Zeiss Axio Scope.A1, bl_ph channel at 1000 ms exposure). At least two independent cultures were used at with two different positions per culture (ca. 150 to 300 individual cells were monitored).
ACKNOWLEDGEMENTS
This work was supported by funds derived from the Biotechnology Endowed Professorship at the Pennsylvania State University. The authors have no conflicts of interest.