|Home | About | Journals | Submit | Contact Us | Français|
The stringent response (SR), mediated by the alarmone (p)ppGpp, is a conserved bacterial adaptation system controlling broad metabolic alterations necessary for survival under adverse conditions. In Enterococcus faecalis, production of (p)ppGpp is controlled by the bifunctional protein RSH (for “Rel SpoT homologue”; also known as RelA) and by the monofunctional synthetase RelQ. Previous characterization of E. faecalis strains lacking rsh, relQ, or both revealed that RSH is responsible for activation of the SR and that alterations in (p)ppGpp production negatively impact bacterial stress survival and virulence. Despite its well-characterized role as the effector of the SR, the significance of (p)ppGpp during balanced growth remains poorly understood. Microarrays of E. faecalis strains producing different basal amounts of (p)ppGpp identified several genes and pathways regulated by modest changes in (p)ppGpp. Notably, expression of numerous genes involved in energy generation were induced in the rsh relQ [(p)ppGpp0] strain, suggesting that a lack of basal (p)ppGpp places the cell in a “transcriptionally relaxed” state. Alterations in the fermentation profile and increased production of H2O2 in the (p)ppGpp0 strain substantiate the observed transcriptional changes. We confirm that, similar to what is seen in Bacillus subtilis, (p)ppGpp directly inhibits the activity of enzymes involved in GTP biosynthesis, and complete loss of (p)ppGpp leads to dysregulation of GTP homeostasis. Finally, we show that the association of (p)ppGpp with antibiotic survival does not relate to the SR but rather relates to basal (p)ppGpp pools. Collectively, this study highlights the critical but still underappreciated role of basal (p)ppGpp pools under balanced growth conditions.
Drug-resistant bacterial infections continue to pose a significant public health threat by limiting therapeutic options available to care providers. The stringent response (SR), mediated by the accumulation of two modified guanine nucleotides collectively known as (p)ppGpp, is a highly conserved stress response that broadly remodels bacterial physiology to a survival state. Given the strong correlation of the SR with the ability of bacteria to survive antibiotic treatment and the direct association of (p)ppGpp production with bacterial infectivity, understanding how bacteria produce and utilize (p)ppGpp may reveal potential targets for the development of new antimicrobial therapies. Using the multidrug-resistant pathogen Enterococcus faecalis as a model, we show that small alterations to (p)ppGpp levels, well below concentrations needed to trigger the SR, severely affected bacterial metabolism and antibiotic survival. Our findings highlight the often-underappreciated contribution of basal (p)ppGpp levels to metabolic balance and stress tolerance in bacteria.
Despite being a commensal of the gut microbiota, the Gram-positive pathogen Enterococcus faecalis ranks among the leading causative agents of severe and costly nosocomial infections (1). The prevalence of E. faecalis in hospital settings has been attributed to its capacity to survive under adverse conditions, including prolonged starvation and exposure to detergents and commonly used antibiotics, to which E. faecalis has both intrinsic and acquired tolerance (2). This inherent resilience allows E. faecalis to survive under conditions lethal to most other bacteria, pathogenic or nonpathogenic, a trait that seems to be intertwined with its virulence.
The stringent response (SR) is a highly conserved bacterial stress response mediated by the accumulation of the alarmone (p)ppGpp, which refers to two modified guanine nucleotides, pyrophosphorylated GDP or GTP (herein abbreviated as ppGpp and pppGpp, respectively) (3). While initially defined as a response to amino acid starvation, the term SR has since expanded to include any regulatory effect exerted by robust (p)ppGpp accumulation, irrespective of the triggering mechanism. During the SR, the accumulation of (p)ppGpp induces large-scale transcriptional alterations, leading to general repression of genes required for rapid growth, such as rRNA genes, and concomitant activation of genes involved in nutrient synthesis or acquisition and stress survival, a response that shifts cellular resources toward adaptation to a nongrowth state (4–7). In Gram-negative bacteria, (p)ppGpp primarily interacts with RNAP, in synergy with DksA, to directly affect transcription (3). However, direct (p)ppGpp-RNAP interactions do not seem to occur in Gram-positive species (8, 9), and the current model suggests that (p)ppGpp affects transcription of rRNA genes in this bacterial group by reducing the availability of the initiating nucleotide GTP (8). In addition to transcriptional control, (p)ppGpp allosterically inhibits the activity of enzymes other than RNAP, including DNA primase, exopolyphosphatase, lysine decarboxylase, and several enzymes involved in GTP synthesis (10–12).
In Gram-negative proteobacteria, such as Escherichia coli, (p)ppGpp is metabolized by the strong (p)ppGpp synthetase RelA and the bifunctional protein SpoT, which exerts weak synthetase and strong hydrolase activities (13, 14). In Gram-positive organisms, the bifunctional protein RSH (for “Rel SpoT homologue”), also known as RelA or Rel, is a strong (p)ppGpp synthetase, like the E. coli RelA, that also possess (p)ppGpp hydrolase activity characteristic of SpoT (15, 16). In addition to RSH, Gram-positive Firmicutes encode one or two (p)ppGpp synthetases (termed RelP and RelQ) that lack the N-terminal Mn2+-dependent hydrolase domain required for (p)ppGpp hydrolysis and the C-terminal regulatory region of RSH enzymes (17–21). Transcriptomic and phenotypic analysis indicated that the bifunctional RSH is the principal enzyme responsible for the rapid accumulation of (p)ppGpp during nutrient limitation but also controls the magnitude and duration of (p)ppGpp accumulation through regulation of (p)ppGpp synthetase activity, hydrolase activity, or both (5, 15, 17–19). However, the specific roles of RelP and RelQ remain elusive.
Previous studies from our group identified and characterized the two enzymes responsible for (p)ppGpp production in E. faecalis, the bifunctional RSH and the monofunctional RelQ synthetase; RelP homologues are not found in the genome of enterococci (5, 17). To avoid ambiguity in the nomenclature between the monofunctional RelA of Gram-negative bacteria and the bifunctional enzyme of Gram-positive organisms, we have chosen to adopt the term RSH to refer to the bifunctional enzyme of E. faecalis, called RelA in previous publications (5, 17). Characterization of strains lacking rsh, relQ, or both showed that RSH is responsible for activation of the SR and that (p)ppGpp plays a crucial role in stress survival and virulence (5, 17, 21). A complete lack of (p)ppGpp, as seen in the (p)ppGpp0 (rsh relQ) strain, led to overall decreased tolerance to vancomycin, reduced survival within macrophages, and attenuated virulence in the Caenorhabditis elegans and Galleria mellonella models (5, 17). Interestingly, the rsh strain, which like the (p)ppGpp0 strain is unable to mount the SR, showed enhanced vancomycin tolerance and behaved like the wild-type strain in in vitro macrophage survival and invertebrate virulence models.
Despite the relatively long history of (p)ppGpp in bacterial research, substantially less effort has been made to understand its regulatory effect during exponential (balanced) growth. Here, balanced growth represents any physiological condition in which cellular constituents are produced at constant rates relative to one another, leading to a constant rate of cell division. Previous studies have shown that during steady-state growth, E. coli produces low basal levels of (p)ppGpp (22, 23). In these studies, a general inverse correlation between basal (p)ppGpp levels and growth rate was established whereby lower growth rates are indicative of higher basal (p)ppGpp pools (23–25). It was later postulated that basal (p)ppGpp pools, despite being at concentrations considerably lower than those observed during the SR, may be high enough to have regulatory effects during balanced growth (26). This lead to the idea that (p)ppGpp acts as a gradient over a range of growth rates rather than a bistable switch to control cellular physiology under nutrient replete (“off”) or starvation (“on”) conditions (26).
Given the increasing number of studies that report the association of (p)ppGpp with virulence expression, persister cell formation, and other survival-related phenotypes (3, 27–29), a comprehensive understanding of the pathways affected by (p)ppGpp may reveal potential targets for the development of new antimicrobial therapies. To gain a more thorough understanding of the genetic basis through which (p)ppGpp maintains cell homeostasis during balanced growth, we used microarrays to compare the transcriptome of the wild-type strain E. faecalis OG1RF to a panel of mutant strains producing different basal levels of (p)ppGpp during exponential growth phase. The transcriptome of a bona fide (p)ppGpp0 strain revealed that basal (p)ppGpp pools are essential for a balanced metabolism, providing an additional explanation for the reduced fitness and attenuated virulence of the (p)ppGpp0 mutant strain. As recently shown in Bacillus subtilis (11), we also found that complete lack of (p)ppGpp led to aberrant regulation of guanosine metabolism characterized by accumulation of intracellular GTP and severe inhibition of growth in the presence of exogenous guanosine. Finally, we expanded our previous observations by showing that the (p)ppGpp-mediated antibiotic tolerance, regardless of the drug target, occurs at (p)ppGpp concentrations that are markedly lower than those achieved during the SR. To the best of our knowledge, this report represents the first global transcriptional comparison of strains with different basal (p)ppGpp levels during balanced growth in any microorganism.
Among the many phenotypic differences observed between the different (p)ppGpp-deficient strains, the Δrsh strain showed a slow-growth phenotype in rich medium, whereas the ΔrelQ and (p)ppGpp0 (Δrsh ΔrelQ) strains grew as well as the wild-type strain. Previously, we proposed that the slow growth of the Δrsh strain is a result of the strain’s inability to hydrolyze (p)ppGpp synthesized by RelQ, thereby elevating basal (p)ppGpp above wild-type levels (17). To measure basal levels of (p)ppGpp, cultures were uniformly labeled with 32P from early exponential to mid-log growth phase. The basal (p)ppGpp level is defined here as the amount of (p)ppGpp present in nonstressed, logarithmically growing cells. Quantifications of extracted nucleotides separated by thin-layer chromatography (TLC) (Fig. 1B) indicated that basal levels of ppGpp were approximately 4-fold higher in the Δrsh strain than the wild-type strain. The increased ppGpp level in the Δrsh strain was accompanied by an ~40% reduction in GTP. In contrast to ppGpp, pppGpp levels were similar in the wild-type and Δrsh strains. No significant differences in (p)ppGpp pools were observed between wild-type and ΔrelQ strains, and as expected, no (p)ppGpp was detected in the (p)ppGpp0 strain.
Here, microarrays were used to obtain the transcriptional profiles of OG1RF and the Δrsh, ΔrelQ, and Δrsh ΔrelQ strains during exponential growth. Given that the E. faecalis V583 genome annotation (30) has been widely used in the literature, we adopted the same gene designation in tables depicting the microarray data. However, genes present in the OG1RF genome but absent in V583 are presented with the original OG1RF designations (31).
At an assigned P value of ≤0.001 and a 2-fold cutoff, 51 genes in the Δrsh strain, 17 genes in the ΔrelQ strain, and 246 genes in the Δrsh ΔrelQ [(p)ppGpp0] strain were differentially expressed compared to the wild-type strain. The complete list of these genes is shown in Table S1 in the supplemental material. A subset of these genes (n = 7) was selected for real-time quantitative RT-PCR validation, and all results were consistent with the expression trends observed in the microarrays (see Fig. S1 and Table S2 in the supplemental material). The small number of genes found to be differentially expressed in the ΔrelQ strain is in agreement with the nearly identical basal levels of (p)ppGpp in this strain and the wild type (Fig. 1). Conversely, a much larger number of genes were differentially expressed in the Δrsh and (p)ppGpp0 strains, which have 4-fold-higher basal (p)ppGpp levels and no (p)ppGpp, respectively. Among the genes that were differentially transcribed in the Δrsh strain, 60.8% were upregulated and 39.2% were downregulated. Most notable among the downregulated genes were those involved in nucleotide metabolism, including those encoding nucleoside diphosphate kinase (ndk) and GMP reductase (guaC). Among the genes activated in the Δrsh strain, a subset correspond to de novo pyrimidine biosynthesis (nrdG, nrdD, pyrDII, pyrB, pyrDB, pyrE, purA, pyrC, and carB), suggesting that elevated basal levels of (p)ppGpp interfere with pyrimidine metabolism.
The most prominent differences were observed in the transcriptome of the (p)ppGpp0 strain. Remarkably, the overwhelming majority of the differentially expressed genes in this strain were activated (92.6%). A closer evaluation of these genes revealed a dramatic upregulation of genes associated with energy generation, in particular, genes involved in pyruvate production from alterative carbon sources (Fig. 2; also, see Table S1 in the supplemental material). This is surprising, as the cells were grown in a defined but complete medium containing 10 mM glucose, which is sufficient to repress alternate carbon metabolism pathways (32). For example, many genes (n = 35) of the various phosphoenolpyruvate:sugar phosphotransferase systems (PTS), the major sugar transport systems at low carbohydrate concentrations, were activated. Likewise, all 13 genes encompassing the citCL locus, which is involved in citrate transport and metabolism, were highly induced (ranging from 27- to 134-fold induction) as well as the genes encoding glycerol-3-phosphate oxidase (glpO) and glycerol uptake protein (glpF), involved in the GlpK pathway of glycerol metabolism (33). Three genes involved in serine degradation (serS1, sdhA-1 and sdhB-1), which convert serine into pyruvate, were also highly induced (≥66-fold induction). Finally, the gene encoding the catabolite control protein CcpA, which is responsible for carbon catabolite repression (CCR), was induced 3.2-fold. Among the few repressed genes in the (p)ppGpp0 strain (7.4%), a subset are involved in purine metabolism and salvage pathways, including GMP reductase (guaC), guanine deaminase (guaD), a putative xanthine/uracil permease, IMP cyclohydrolase (purH), and adenylosuccinate synthase (purA).
As a typical lactic acid bacterium, E. faecalis usually exhibits homolactic fermentation. However, environmental changes, including changes in oxygen availability, pH, and carbon source, can rapidly trigger a switch from homolactic to mixed-acid (heterolactic) fermentation (34). The major end products from heterolactic fermentation include lactate, acetate, formate, acetoin, and ethanol. To investigate whether the transcriptional activation of alternate carbon and energy metabolism genes in the (p)ppGpp0 strain is manifested at the physiological level, we compared the intracellular concentration of pyruvate as well as the production of lactate, ethanol, formate, and acetoin secreted into the growth medium between wild-type and (p)ppGpp0 strains. We also attempted to measure acetate production, but even with the removal of exogenous acetate, the chemically defined FMC medium was incompatible with the enzymatic assay, leading to high background readings. Despite the transcriptional induction of genes involved in different pathways linked to pyruvate production, the intracellular pyruvate levels of the (p)ppGpp0 strain were comparable to those observed in OG1RF (Fig. 3A). However, the (p)ppGpp0 strain produced less lactate than OG1RF (e.g., an 11% decrease at an optical density at 600 nm [OD600] of 0.7; P ≤ 0.0001) and produced more acetoin, formate, and ethanol (Fig. 3B to E). The relatively small total amounts of formate produced [e.g., 2.4% of the total lactate produced at an OD600 of 0.7 for the (p)ppGpp0 strain] and the marginal difference in ethanol production between OG1RF and the (p)ppGpp0 strain (e.g., 4.6% at an OD600 of 0.7) is likely to have little, if any, impact on cell physiology. However, the increased production of acetoin (105% increase at an OD600 of 0.7; P ≤ 0.0001) coupled with the reduced production of lactate observed in the (p)ppGpp0 strain is more likely to have biological implications, as acetoin is a nonacidic fermentative end product. To assess this possibility, the pH of culture supernatants from wild-type and (p)ppGpp0 strains was measured over the different growth phases. In agreement with our metabolite analysis, the wild-type strain was able to reduce the external pH to values significantly lower than those achieved by the (p)ppGpp0 strain (Fig. 3F). Collectively, these results serve to validate the microarray results and suggest that, to maintain pyruvate homeostasis, the (p)ppGpp0 strain switches from a predominant homolactic fermentation to heterolactic fermentation.
Catabolic control, or the ability to regulate the rate and direction of carbon flow, is important for balancing proliferative capacity with the damaging effects of metabolic by-products, such as reactive oxygen species (ROS) (35). Enterococcus faecalis is a potent producer of ROS, including H2O2 and superoxide, which are by-products of aerobic glycerol metabolism and the interaction of respiratory chain semiquinone radicals with oxygen (36, 37). At neutral or low pH, like those in our culture media, H2O2 can also rapidly form (8 × 104 M−1 s−1) from the spontaneous dismutation of superoxide resulting from the random interaction of two superoxide molecules (38). We compared the production of H2O2 between the wild-type OG1RF and (p)ppGpp0 strains. The (p)ppGpp0 cells produced at least 6.8-fold more H2O2 than the wild type at early log, mid-log, and late log growth phases (Fig. 4). This increased H2O2 production further suggests that the (p)ppGpp0 strain is unable to maintain a balanced metabolism.
Recently, (p)ppGpp was shown to directly inhibit multiple enzymes involved in GTP synthesis, and disruption of GTP homeostasis in a B. subtilis (p)ppGpp0 strain resulted in metabolic changes that negatively affected cell viability, even in the absence of starvation (11). Addition of exogenous guanosine to the B. subtilis (p)ppGpp0 strain, which is converted to GTP via the salvage pathway, dramatically increases GTP levels, suggesting that (p)ppGpp maintains GTP homeostasis via negative feedback control (11). As demonstrated in B. subtilis, the hypoxanthine-guanine phosphoribosyltransferase (HprT) homologue from E. faecalis was also specifically inhibited by pppGpp in a dosage-dependent manner with a 50% inhibitory concentration (IC50) of 89.8 ± 10.49 µM (Fig. 5A, top). Although guanylate kinase (Gmk) was inhibited by pppGpp with an IC50 of 461.9 ± 106.4 µM (data not shown), this was nonspecific, as GTP was able to inhibit Gmk with an efficiency similar to that of pppGpp (Fig. 5A, bottom). To assess whether (p)ppGpp also exerts posttranscriptional control over GTP synthesis in vivo, we investigated the effects of exogenous guanosine on the wild-type and (p)ppGpp-defective strains. Similar to what was seen with B. subtilis, addition of guanosine resulted in a sharp increase in GTP levels in the (p)ppGpp0 strain, whereas GTP levels remained constant in the wild-type, Δrsh, and ΔrelQ strains (Fig. 5B). As an apparent result of GTP accumulation, growth of the (p)ppGpp0 strain was completely inhibited by as little as 1 mM guanosine (Fig. 5C). In contrast with what was seen with B. subtilis (11), no significant effect on cell viability was observed in the (p)ppGpp0 strain following exposure to exogenous guanosine (data not shown). These results serve to highlight another crucial function of (p)ppGpp, that seems to be separate from the SR activation, in balancing guanosine metabolism with cellular demands for macromolecular biosynthesis and energy generation.
Although the association of (p)ppGpp metabolism with antibiotic tolerance has been known for several years (39, 40), recent studies suggested that activation of the SR is central to the formation of persister cells in Gram-negative bacteria (28, 41). Previously, we showed that the E. faecalis (p)ppGpp0 strain (Δrsh ΔrelQ) was significantly more sensitive to vancomycin, whereas the Δrsh strain showed increased survival rates (17). Here, we expanded this initial finding by conducting time-kill kinetics studies with two additional bactericidal drugs: ampicillin (β-lactam) and norfloxacin (DNA gyrase inhibitor) (Fig. 6). In both cases, the survival rates of the ΔrelQ strain were not statistically different from those of the wild type. However, the (p)ppGpp0 strain was killed more rapidly (P ≤ 0.001). Despite the fact that the rsh strain has only RelQ as a source of (p)ppGpp synthesis and is unable to activate the SR (5), this strain showed rates of survival against ampicillin or norfloxacin similar to those of the wild-type strain. To verify that the survival differences among strains were not due to differences in long-term viability, cells were exposed to the bacteriostatic antibiotic chloramphenicol. Despite a small reduction in the viability of the (p)ppGpp0 strain after 48 h, the differences observed were not statistically significant (Fig. 6C). These results indicate that, in E. faecalis, (p)ppGpp-mediated protection occurs at levels that are well below those needed to trigger the SR.
Previously, we demonstrated that inactivation of the bifunctional RSH resulted in a slow-growth phenotype, a characteristic that could be rescued with the simultaneous inactivation of RelQ (17). Identical observations have been made in Streptococcus mutans and B. subtilis, which encode two small (p)ppGpp synthetases, indicating that the slow growth of rsh mutants is associated with reduced intracellular GTP due to constitutive production of (p)ppGpp by one (in the case of E. faecalis) or two monofunctional (p)ppGpp-synthetases (18, 19, 42, 43). In the present study, by extending the duration of isotopic labeling under nonstringent conditions, we acquired unequivocal evidence that the Δrsh strain accumulates higher basal levels of (p)ppGpp at the expense of intracellular GTP. Interestingly, the differences in (p)ppGpp levels were restricted to ppGpp, not pppGpp. This observation may have important biological implications, as studies conducted with E. coli revealed that ppGpp is a more effective regulator (~10 times more potent than pppGpp) for several SR-controlled phenomena (44). Previous studies have shown that RelQ paralogs can efficiently utilize GDP or GTP to synthesize, respectively, ppGpp and pppGpp, although no substrate preference experiments have been conducted (19, 45). The accumulation of ppGpp in the Δrsh strain may be an indication that RelQ preferentially utilizes GDP over GTP. Another possible explanation for the ppGpp accumulation in Δrsh is that pppGpp produced from GTP is unstable and rapidly degraded into ppGpp (46). In Gram-negative bacteria, guanosine pentaphosphatases (Gpp) are responsible for the conversion of pppGpp to ppGpp in vivo, although alternate pathways for (p)ppGpp degradation have been shown in vitro (47, 48). However, no homologues of gpp have been identified in E. faecalis and closely related Gram-positive bacteria. At the present time, the reasons for the accumulation of ppGpp, but not of pppGpp, in the rsh strain remain undetermined. Work is under way to investigate the possibility that the E. faecalis RelQ preferentially utilizes GDP over GTP.
The key finding from the microarray analysis was the observation that a complete lack of (p)ppGpp caused large-scale transcriptional alterations during balanced growth. A closer examination of these transcriptional alterations showed that genes involved in the production of pyruvate via multiple pathways, including glycolysis, secondary-carbon-source metabolism (citrate, glycerol, and malate), and serine degradation, were highly induced in the (p)ppGpp0 strain. Although we did not observe increased production of pyruvate in the (p)ppGpp0 strain, the ability of this strain to maintain pyruvate homeostasis is likely linked to enhanced acetoin production, as two molecules of pyruvate are consumed for every molecule of acetoin generated. Additionally, the acetolactate synthase (ALS) enzyme has a lower Km for pyruvate than other competing fermentative enzymes. Hence, acetoin production is favored when intracellular pyruvate is abundant.
Among the energy generation pathways transcriptionally induced in the (p)ppGpp0 strain were multiple PTS genes, two divergent citrate metabolism operons, and glycerol uptake and metabolism genes. Notably, these metabolic pathways are known to be under carbon catabolite regulation (CCR), mediated by CcpA, in different Gram-positive bacteria (49–51). CcpA is a global regulator of carbon metabolism, either positive or negative, and its activity is stimulated through interactions with PTS enzymes and by the availability of the glycolytic intermediates glucose-6-phosphate and fructose-1,6-bisphosphate (FBP) (52). Thus, metabolic alterations observed in the (p)ppGpp0 strain may serve as an explanation for CCR alleviation. Moreover, when compared to the parent strain, transcription of the ccpA gene was also induced. While the concomitant upregulation of ccpA with CcpA-regulated genes may appear counterintuitive, this likely represents an attempt by the cell to constrain alternate carbon metabolism, diverting it back to the utilization of preferential carbon sources such as glucose. The upregulation of alternate carbon catabolism pathways suggests that the (p)ppGpp0 strain is unable to accurately sense external and internal metabolic cues.
The linkage of enhanced ROS generation with the (p)ppGpp0 strain phenotypes is of particular interest due to the strain’s enhanced susceptibility to bactericidal antibiotics, reduced survival within murine macrophages, and attenuated virulence in two invertebrate models (5, 17). In every case, a negative correlation between ROS generation and bacterial viability appears to exist, although the model proposing endogenous ROS generation as a common killing mechanism of bactericidal antibiotics was recently challenged (53–55). In line with the transcriptome analysis that suggested enhanced metabolic flux, an indicator of increased ROS generation (35), we detected a >5-fold increase in H2O2 production by the (p)ppGpp0 strain during exponential growth. As a potential protective mechanism to compensate for increased ROS production, the transcription of several oxidative stress genes was induced in the (p)ppGpp0 strain, including NADH peroxidase, peroxiredoxin, thioredoxin disulfide reductase, and ferrodoxin reductase (see Table S1 in the supplemental material). Thus, while the (p)ppGpp0 strain appears to be able to cope with intrinsically higher rates of intracellular ROS under favorable growth conditions, this strain may be at the limit of its antioxidant capacity and is, therefore, more susceptible to exogenous ROS exposure.
Recent advances defining the direct regulation of GTP homeostasis by (p)ppGpp in B. subtilis and, more importantly, showing that GTP depletion enhances stress survival independently of (p)ppGpp have provided new mechanistic insight onto the underlying causes of the often divergent phenotypes in our two SR-defective mutants (the rsh and rsh relQ strains). In accordance with results for B. subtilis (11), we observed that (p)ppGpp directly controls GTP homeostasis in E. faecalis. Specifically, HprT and Gmk were inhibited by pppGpp (IC50 ~ 100 µM for HprT and 400 µM for Gmk) at physiologically relevant levels, as (p)ppGpp levels can reach low-millimolar levels during a developed stringent response in other bacteria (56). At an concentration equivalent to that of pppGpp, GTP had no effects on HprT activity but similar effects on Gmk activity, indicating that pppGpp inhibition is specific to HprT but nonspecific for Gmk. Based on our results, pppGpp appears to be a more potent and specific inhibitor for HprT than Gmk, suggesting that regulation of HprT by pppGpp is probably the major mechanism for maintaining GTP homeostasis in E. faecalis. In addition, we observed GTP accumulation and severe growth inhibition when the (p)ppGpp0 strain, but not the rsh strain, was cultured in the presence of excess guanosine. Thus, it is very likely that the different, often opposing, phenotypes of rsh and (p)ppGpp0 strains result from differences in GTP levels (low and high GTP, respectively) in these strains. Changes in GTP pools will directly affect the activity of the metabolic regulator CodY, which senses both carbon and nitrogen availability by responding to branched-chain amino acids and GTP levels, creating a complex and integrated system to control nutrient acquisition and utilization in Gram-positive bacteria (57). The opposing 4-fold-elevated and no basal (p)ppGpp levels in rsh and (p)ppGpp0 strains may then correlate to reduced and high CodY activity, respectively. In addition to interference with CodY activity, the guanine energy charge (GTP/GDP ratio) could also play a more direct role in the phenotypes associated with the (p)ppGpp0 strain. GTP is a key component of anabolic cellular processes, including synthesis of stable RNAs, polyamine synthesis, and all three stages of translation (8, 58).
In Gram-negative bacteria, recent studies proposed that (p)ppGpp participates in antibiotic tolerance by (i) controlling ROS metabolism and activating detoxification enzymes (41, 59), (ii) functioning as an unconventional metabolic toxin-antitoxin (TA) module (28), and (iii) potentiating the action of the HipAB TA module (60). Although we have a relatively good understanding of the underlying mechanisms by which (p)ppGpp modulates antimicrobial tolerance in Gram-negative bacteria, much less is known about the process in Gram-positive species. Interestingly, two independent whole-genome sequencing studies identified point mutations in the rsh gene of Staphylococcus aureus isolates, which resulted in increased accumulation of (p)ppGpp, as being directly responsible for the emergence of antibiotic-tolerant (persister) cells (61, 62). Previously, we showed that a complete lack of (p)ppGpp, as seen in the (p)ppGpp0 strain, resulted in increased sensitivity to vancomycin (17). However, the rsh strain showed enhanced tolerance to vancomycin (17). By expanding our previous observation to other bactericidal antibiotics, we showed here that only the (p)ppGpp0 strain, not the Δrsh strain, displayed enhanced sensitivity to bactericidal antibiotics. Collectively, the differences in antibiotic tolerance observed between Δrsh and (p)ppGpp0 strains clearly indicate that (p)ppGpp is central for antibiotic tolerance but that this phenomenon occurs at (p)ppGpp concentrations that are much lower than those necessary to activate the SR. While the SR was abolished in both Δrsh and (p)ppGpp0 strains (5), the differences in basal (p)ppGpp and GTP pools between these two strains appear to be directly associated with antibiotic tolerance. Work is under way to understand how small changes in intracellular guanine nucleotide pools, mediated by RelQ, enhance E. faecalis tolerance to bactericidal antibiotics.
Collectively, our microarray data and supporting physiological assays indicate that the (p)ppGpp0 strain cannot control the pace and direction of carbon flow and, as a result, is unable to sense and respond to environment changes accordingly. As a consequence, a strain unable to synthesize (p)ppGpp may have uncontrolled consumption of energy stores, unbalanced NAD+/NADH ratios, increased ROS generation due to increased metabolic activity, and unbalanced GTP homeostasis. All of these factors combined likely reduce the long-term fitness of the cells.
Since the identification of (p)ppGpp more than four decades ago, mounting evidence has shown that activation of the SR is central for controlling responses that promote cell survival during adverse conditions. In the past few years, stringent control and (p)ppGpp metabolism have been, once again, the subject of intense research. The picture emerging from these most recent investigations is that bacteria utilize (p)ppGpp in many different ways, and that the paradigm for Gram-negative organisms does not apply fully to Gram-positive bacteria. In this study, we provided new evidence that the role of (p)ppGpp in the physiology of Gram-positive bacteria goes well beyond activation of the SR.
E. faecalis OG1RF and its derivatives JAL1 (Δrsh, formerly ΔrelA), JAL2 (ΔrelQ), and JAL3 (Δrsh ΔrelQ, formerly ΔrelA ΔrelQ) strains have been previously described (17). For microarray analysis, cells were grown at 37°C in the chemically defined medium FMC (63) containing all amino acids, eight vitamins, three nucleobases (adenine, guanine, and uracil), and salts and supplemented with 10 mM glucose (FMCG) to an optical density at 600 nm (OD600) of 0.3. To assess the effects of excess guanosine on growth, cells were cultured in complete FMCG until the OD600 was 0.3 and diluted 1:100 into FMCG lacking nucleobases but supplemented with increasing concentrations of guanosine. For detection of (p)ppGpp, overnight cultures were diluted in fresh FMCG containing reduced phosphate (8.6 mM) (17), grown to an OD600 of ~0.25, and labeled with 150 µCi ml−1 of carrier-free [32P]orthophosphate (PerkinElmer, Waltham, MA) until each culture reached an OD600 of 0.4 (~50 min for the OG1RF and relQ strains, ~55 min for the Δrsh relQ strain, and ~70 min for the Δrsh strain). For GTP accumulation experiments, cells were grown in low-phosphate FMCG lacking nucleotide, nucleoside, or nucleobase supplementation to an OD600 of 0.3 and labeled with 150 µCi ml−1 of 32P along with the simultaneous addition of 2 mM guanosine to the growth medium for 15 min. Control samples received no exogenous guanosine.
For (p)ppGpp and GTP detection, nucleotides were extracted by adding an equal volume of 13 M formic acid followed by two freeze-thaw cycles in a dry-ice--ethanol bath. Acid extracts were centrifuged, and the supernatants were spotted onto polyethyleneimine cellulose plates (J. T. Backer/Avantor Performance Materials, Center Valley, PA). For two-dimensional (2D) TLC separation, samples were first separated in 3.3 M ammonium formate, 4.2% boric acid (pH 7.0 with NH4OH), followed by a second-dimension separation in 1.5 M KH2PO4 (pH 3.4). TLC plates were desalted by soaking for 5 min in methanol before and after each run. (p)ppGpp and GTP pools were quantified using a phosphorimager (molecular imager FX; Bio-Rad, Hercules, CA).
To isolate RNA from E. faecalis, cells were harvested by centrifugation at 4°C and then treated with the RNA Protect reagent (Qiagen, Inc., Chatsworth, CA). Total RNA was isolated from homogenized E. faecalis cells by the hot acid-phenol method as described previously (64). RNA pellets were resuspended in nuclease-free H2O and treated with DNase I (Ambion/Life Technologies) at 37°C for 30 min. The RNA was purified again using the RNeasy minikit (Qiagen), including a second on-column DNase treatment that was performed as recommended by the supplier. RNA concentrations were determined using a NanoDrop ND-1000 spectrophotometer (Thermo, Fisher Scientific, Waltham, MA).
Transcriptome analysis was performed using the E. faecalis microarrays provided by the J. Craig Venter Institute Pathogen Functional Genomics Resource Center (PFGRC). Additional details regarding the arrays can be found at http://pfgrc.jcvi.org/index.php/microarray/array_description/enterococcus_faecalis/version1.html. Microarray experiments were carried out as previously described (5).
A subset of genes was selected to validate the microarray analysis by real-time quantitative reverse transcription (qRT) PCR. Gene-specific primers (see Table S3 in the supplemental material) were designed using Beacon Designer 2.0 software (Premier Biosoft International). Reverse transcription and real-time reverse transcriptase PCR were carried out on a StepOnePlus real-time PCR system (Life Technologies, Grand Island, NY) according to protocols described elsewhere (64). Student’s t test was performed to verify significance of the real-time PCR quantifications.
Intracellular pyruvate levels were determined as previously described (65). Briefly, cells were grown in FMCG to early log phase (OD600 = 0.3), mid-log phase (OD600 = 0.5), and late log phase (OD600 = 0.7), harvested by centrifugation, and washed twice in 50 mM sodium phosphate buffer (pH 6.7). Cell pellets were lysed using a bead beater and centrifuged to remove insoluble matter. In opaque 96-well microtiter plates, 50 µl of soluble lysate was added to 50 µl of reaction buffer containing 100 mM NaPO4 (pH 6.7), 0.2 mM MgSO4, 10 µM FAD, 0.2 mM thiamine pyrophosphate (TPP), 0.2 U ml−1 pyruvate oxidase, 0.5 U ml−1 horseradish peroxidase (HRP), and 50 µM Amplex UltraRed (Life Technologies). Reaction mixtures were incubated at 37°C for 30 min, and fluorescence was measured by excitation at 470 nm and emission at 590 nm. Pyruvate concentrations were normalized to total cell protein content, determined by bicinchoninic acid (BCA) assay.
Lactic acid, formate, and ethanol were measured using Megazyme enzymatic kits (Megazyme International, Wicklow, Ireland). Assays were conducted according to the manufacturer’s protocol using a 96-well plate format. Acetoin production was measured using a Voges-Proskauer test adapted for use in a 96-well plate as described previously (66). For all analyses, cells were grown to early log, mid-log, and late log phase in FMCG as described above. At the respective time points, aliquots were centrifuged for 5 min, and the cell-free culture supernatants were stored on ice, except for aliquots used for pH determination, which were assessed immediately after centrifugation. Before use in fermentative assays, chilled supernatants were removed from ice and allowed to equilibrate to room temperature.
The production of H2O2 was measured using an H2O2-peroxidase assay kit (Life Technologies). Briefly, cells were grown to early log, mid-log, or late log phase in FMCG as described above, harvested by centrifugation, washed once in assay buffer (50 mM Tris-HCl at pH 7.4), and diluted 1:2 in the assay buffer. Reactions were initiated by mixing 50 µl of cell suspension with 50 µl of reaction mix (50 mM Tris-HCl [pH 7.4], 20 µM Amplex UltraRed, and 0.2 U ml−1 horseradish peroxidase) in opaque 96-well microtiter plates at 37°C for 30 min. Fluorescence was read in a SpectraMax MX5 reader (Molecular Devices LLC, Sunnyvale, CA) with excitation at 490 nm and emission at 590 nm. To normalize fluorescence by CFU, cells aliquots from each growth phase were serially diluted and plated onto BHI agar for colony enumeration.
The genes encoding Gmk and HprT from E. faecalis OG1RF were cloned into pLICTrPC-HA (67) using a ligation-independent cloning technique, and the recombinant plasmids were transformed into E. coli BL21(DE3). Cells were grown from a single colony at 37°C in LB supplemented with 100 µg ml−1 carbenicillin to an OD600 of ~0.6, and isopropyl β-d-1-thiogalactopyranoside was added to a final concentration of 1 mM. Cells were grown for another 3 h before harvest. Proteins were purified using Ni-NTA spin columns (Qiagen) following the manufacturer’s instructions. Protein concentration was determined either by Bradford protein assay (Bio-Rad) or by measuring absorbance at 280 nm using theoretical molar extinction coefficients. Standard protocols for Gmk and HprT enzymatic assays were followed, with minor modifications (11). HprT reactions were performed at 25°C in a 100-µl reaction mix containing 100 mM Tris-HCl (pH 7.4), 1.2 mM MgCl2, 1 mM 5-phosphoribosyl 1-pyrophosphate (PRPP), 50 µM guanine, 20 nM purified HprT enzyme, and various pppGpp concentrations. Reactions were initiated by adding the enzyme and monitored for 10 min by measuring change of absorbance at 257 nm. Gmk reactions were performed at 25°C in a 100-µl reaction mix containing 100 mM Tris-HCl (pH 7.5), 100 mM KCl, 10 mM MgCl2, 4 mM ATP, 1.5 mM phospho(enol)pyruvic acid, 250 µM NADH, 2 U pyruvate kinase (from rabbit muscle; Sigma), 2.64 U l-lactic dehydrogenase (from bovine heart; Sigma), 50 µM GMP, 10 nM purified Gmk, and various pppGpp concentrations. Reactions were initiated by adding GMP and monitored for 10 min by measuring the change in absorbance at 340 nm. Data were fitted into the equation y = 100%/[1 + (x/IC50)S], where y is the relative enzyme activity, x is the inhibitor concentration, and S is the slope factor (GraphPad Prism version 5.02 for Windows; GraphPad Software, San Diego, CA). To confirm the specificity of HprT and Gmk inhibition by pppGpp, control reactions were performed using the conditions described above except that GTP was substituted for pppGpp at concentrations approximately equivalent to the IC50s of pppGpp (100 µM for HprT and 400 µM for Gmk).
Cultures were grown in FMCG to exponential phase and diluted in fresh FMCG to 5 × 106 to 1 × 107 CFU ml−1. Time-kill studies were initiated by adding ampicillin (8 µg ml−1), norfloxacin (64 µg ml−1), or chloramphenicol (16 µg ml−1), which represents 5 to 10 times the MIC of these antibiotics for the wild-type OG1RF strain. Viable counts were determined by plating cultures on tryptic soy agar (TSA) plates at time zero, immediately before the addition of antibiotic, and then every 24 h following antibiotic exposure.
Microarray data have been deposited in NCBIs Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo) and are accessible through GEO Series accession number GSE34561.
Quantitative real-time PCR data used for microarray validation. Error bars represent the standard deviations derived from at least three independent replicates assayed in triplicate. P values of qRT-PCR data were determined using a Student’s t test. P values below 0.05 were considered statistically significant and are indicated by an asterisk. Download
List of genes differentially expressed in strains with altered basal (p)ppGpp levels. Column titles indicate comparisons between each mutant and the parent strain OG1RF during logarithmic growth. Listed values represent fold change of expression and are color coded following the scheme indicated at the bottom. All genes were subjected to a 2-fold differential expression cutoff at a P value of ≤0.001.
Quantitative real-time PCR validation of microarray trends. All microarray genes listed were subjected to a 2-fold differential expression cutoff at a P value of ≤0.001. Quantitative real-time PCR (qPCR) values represent the average fold change derived from at least three independent replicates assayed in triplicate. P values of qPCR data were determined using Student’s t test. P values below 0.05 were considered statistically significant. NC, no change (below the 2-fold cutoff).
Quantitative real-time PCR primers used to validate microarray results.
We thank the J. Craig Venter Institute for supplying the E. faecalis microarrays and Mike Cashel (NIH) for helpful discussions.
A.O.G. was supported by the NIDCR training program in oral science, grant T90 DE021985. J.D.W. was supported by NIH grant GM084003.
Citation Gaca AO, Kajfasz JK, Miller JH, Liu K, Wang JD, Abranches J, Lemos JA. 2013. Basal levels of (p)ppGpp in Enterococcus faecalis: the magic beyond the stringent response. mBio 4(5):e00646-13. doi:10.1128/mBio.00646-13.