ABSTRACT
In nearly all bacterial species examined so far, amino acid starvation triggers the rapid accumulation of the nucleotide second messenger (p)ppGpp, the effector of the stringent response. While for years the enzymes involved in (p)ppGpp metabolism and the significance of (p)ppGpp accumulation to stress survival were considered well defined, a recent surge of interest in the field has uncovered an unanticipated level of diversity in how bacteria metabolize and utilize (p)ppGpp to rapidly synchronize a variety of biological processes important for growth and stress survival. In addition to the classic activation of the stringent response, it has become evident that (p)ppGpp exerts differential effects on cell physiology in an incremental manner rather than simply acting as a biphasic switch that controls growth or stasis. Of particular interest is the intimate relationship of (p)ppGpp with persister cell formation and virulence, which has spurred the pursuit of (p)ppGpp inhibitors as a means to control recalcitrant infections. Here, we present an overview of the enzymes responsible for (p)ppGpp metabolism, elaborate on the intricacies that link basal production of (p)ppGpp to bacterial homeostasis, and discuss the implications of targeting (p)ppGpp synthesis as a means to disrupt long-term bacterial survival strategies.
INTRODUCTION
While analyzing nucleotide extracts of Escherichia coli, Cashel and Gallant visualized two spots by thin-layer chromatography that could be implicated in the inhibition of stable RNA accumulation provoked by amino acid starvation, which they dubbed “magic spots” (1). These magic spots were later identified as the hyperphosphorylated guanosine derivatives ppGpp (GDP, 3′-diphosphate) and pppGpp (GTP, 3′-diphosphate), collectively referred to as (p)ppGpp, or alarmones (2, 3). Subsequent studies revealed that (p)ppGpp is responsible for activation of the stringent response (SR), a highly conserved stress response to nutrient starvation (4, 5). Generally speaking, 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 amino acid biosynthesis, nutrient acquisition, and stress survival. In addition to transcriptional control, (p)ppGpp has been shown to directly inhibit the activity of several enzymes, including DNA primase, translation factors, and enzymes involved in GTP biosynthesis (6) (Fig. 1). Ultimately, the SR reallocates cellular resources toward adaptation to a semidormant state, facilitating survival under unfavorable conditions (5, 7). Although initially defined as a response to amino acid and carbon starvation, the term SR has since been expanded to include any regulatory effect exerted by cellular (p)ppGpp accumulation irrespective of the triggering mechanism (4).
Main targets of (p)ppGpp in Firmicutes and Gammaproteobacteria. The regulatory nucleotides (p)ppGpp alter cellular metabolism in response to stress by directly binding to a variety of enzymes. For transcription in Gammaproteobacteria, (p)ppGpp binds directly to RNAP together with the DksA coeffector to modulate promoter selection (8, 9, 11). Repression or activation of target genes is dependent on discriminator sequences encoded in the promoter region (12). Furthermore, (p)ppGpp promotes alternative σ factor usage (64). In Firmicutes, (p)ppGpp is unable to bind to RNAP (8, 9). In this case, transcription is indirectly regulated through depletion of GTP. Low intracellular levels of GTP and BCAA alleviate repression of the global transcriptional regulator CodY, leading to activation of amino acid biosynthesis, nutrient transport, and virulence genes (22, 23, 67). In addition, GTP depletion represses transcription of genes, such as rRNA (rrn) operons, requiring GTP as their initiating nucleotide (+1G). For GTP biosynthesis in Gammaproteobacteria, (p)ppGpp lowers GTP production by inhibiting IMP dehydrogenase (GuaB), the first enzyme in the de novo guanine nucleotide synthesis pathway (107), as well as hypoxanthine phosphoribosyltransferase (HprT) and xanthine-guanine phosphoribosyltransferase (Gpt). In Firmicutes, (p)ppGpp rapidly depletes GTP levels through inhibition of GuaB, HprT, and guanylate kinase (Gmk), halting both de novo and salvage pathways (67). For translation in Gammaproteobacteria, (p)ppGpp directly alters translation initiation and elongation by inactivating initiation factor 2 (IF-2) and elongation factor G (EF-G) (108, 109). Moreover, ribosomal maturation and mRNA translation/stability are affected by inhibition of small GTPases and interaction with the Csr system, respectively (6, 110, 111). In Firmicutes, (p)ppGpp directly inhibits protein translation through inactivation of small GTPases and IF-2 (108, 112). For DNA replication, (p)ppGpp-specific inhibition of DNA primase is observed in both bacterial groups, whereas replication is blocked at the initiation step in Gammaproteobacteria and at the elongation step in Firmicutes (113–115).
The broad physiological alterations induced by (p)ppGpp accumulation rely heavily upon transcriptional alterations. In Gammaproteobacteria, such as the Gram-negative paradigmatic organism E. coli, transcriptional control during the SR is predominately accomplished through the direct interaction of (p)ppGpp with the interface of the β′ and ϖ subunits of RNA polymerase (RNAP) (8–10) and is greatly potentiated by the presence of DksA, a small protein that binds to the RNAP secondary channel (11). As a general rule, the discriminator region between the −10 sequence and the transcriptional start site dictates whether (p)ppGpp will function as a repressor (GC-rich region) or as an activator (AT-rich region) (12, 13). Moreover, (p)ppGpp indirectly regulates transcription by either stabilizing the binding of alternative σ factors or interfering with the activity of transcriptional regulators (14–17).
Despite commonalities in the general outcome, the mechanisms of transcriptional control exerted by (p)ppGpp in low-GC Gram-positive bacteria (Firmicutes) are fundamentally distinct from those described in E. coli. Firmicutes lack a DksA homolog and GC- or AT-rich discriminators and, perhaps most importantly, (p)ppGpp does not physically interact with RNAP (18). Rather, (p)ppGpp indirectly affects transcription in this bacterial group by regulating the concentration of the initiating nucleotide of transcription (iNTP, or position +1). For example, the Bacillus subtilis rrn operons use GTP as the iNTP, while the ilv-leu operon, which is responsible for branch chain amino acid (BCAA) biosynthesis, uses ATP (19, 20). It follows that in B. subtilis, and likely other Firmicutes, SR induction and (p)ppGpp accumulation correspond to a sharp drop in the GTP level that is accompanied by an increase in the ATP pool (21). In addition to changes in iNTP pools, Firmicutes evolved a second regulatory network based on the inverse relationship between (p)ppGpp and GTP. The DNA-binding capacity, and therefore transcriptional regulation, of the global metabolic regulator CodY is modulated by its interaction with GTP and BCAA, particularly isoleucine (22). The link between (p)ppGpp and CodY in Firmicutes has been recently reviewed (23) and will be only briefly discussed in this article. Collectively, these results indicate that (p)ppGpp indirectly controls the action of RNAP in Firmicutes through modulation of intracellular purine concentrations (19, 24, 25).
After its discovery in the late 1960s, a series of contemporaneous investigations defined many of the currently accepted features of (p)ppGpp and the SR. These features include the identification and characterization of the major enzymes responsible for the metabolism of (p)ppGpp, the first glimpses into the mode of action of (p)ppGpp as a regulatory nucleotide, and determination of the pleiotropic effects it exerts on cell physiology. Despite steady progress in (p)ppGpp research since its discovery, the arrival of the genomic era in the late 1990s and the recurrent association of (p)ppGpp with virulence expression and antibiotic persistence reignited the field in recent years. As a result, the number of investigations linking (p)ppGpp to a plethora of disparate processes, such as growth rate control, motility, sporulation, biofilm formation, competence, stress tolerance, persistence, and virulence, have dramatically increased. As part of this new surge in “alarmone” research, new enzymatic sources of (p)ppGpp synthesis and degradation were discovered, and a better appreciation for the biological relevance of (p)ppGpp beyond SR activation has been developed. In the next pages, we focus on recent insights into the diversity of enzymes involved in (p)ppGpp metabolism and elaborate on the intricacies that link incremental production of (p)ppGpp to bacterial homeostasis. Recent developments on antimicrobial therapies that target alarmone production are also discussed. For a complete survey of the field, we direct the reader to other recent reviews (5, 7, 18, 26, 27).
CHANGING PARADIGMS: THE DISCOVERY OF “SHORT” RelA/SpoT HOMOLOGS
Even before the discovery of (p)ppGpp, mutations abolishing stringent control in E. coli had been mapped to the RNA control (RC) locus. The RC locus was later identified as the site of the relA gene and named for the “relaxed” phenotype of these mutants that no longer exhibited an inverse relationship between amino acid availability and stable RNA accumulation (28). RelA is found in Betaproteobacteria and Gammaproteobacteria and is chiefly responsible for the accumulation of (p)ppGpp during amino acid starvation in these organisms (Fig. 2) (5). Initially, the association of RelA with stalled ribosomes containing uncharged tRNAs at the acceptor site was shown to trigger (p)ppGpp synthesis. Some 30 years later, the enzyme was predicted to “hop” from one stalled ribosome to another, such that alarmone production accurately reflected the number of starved ribosomes (29). Recently, single-molecule in vivo investigations have indicated that RelA is tightly bound to the ribosome during active translation but rapidly dissociates when starvation is induced and deacylated tRNA accumulates (30). Once off the ribosome, RelA performs multiple rounds of catalysis in what has been termed the “extended hopping” mechanism (30). Interestingly, heat shock promotes a similar but more transient dissociation, suggesting a mechanism by which other non-starvation-based stresses induce (p)ppGpp synthesis (30). Furthermore, RelA activity is induced by ppGpp, a positive allosteric feedback mechanism that contributes to the rapid activation of the SR (31).
RelA-SpoT homolog (RSH) family enzymes controlling (pp)pGpp metabolism. In Betaproteobacteria and Gammaproteobacteria, the synthesis and hydrolysis of (p)ppGpp is catalyzed by RelA and SpoT. RelA functions as the primary (p)ppGpp synthetase responsible for induction of the SR. SpoT is bifunctional, acting as the primary (p)ppGpp hydrolase but also capable of weak (p)ppGpp synthetase activity in response to nutritional signals not sensed by RelA (5, 116, 117). Outside of Betaproteobacteria and Gammaproteobacteria, the bifunctional Rel is the primary enzyme responsible for (p)ppGpp metabolism in most bacterial species (39). Like RelA, Rel is responsible for the induction of the SR during amino acid starvation and, like SpoT, responsible for the hydrolysis of (p)ppGpp (118). In addition, many bacteria can have one or more stand-alone SAS enzymes that appear to have weak but constitutive activity (39, 43, 44, 46, 47). Small alarmone hydrolases (SAHs) have so far only been characterized in higher eukaryotes, but they appear to have potent hydrolase activity comparable to SpoT (56).
The second enzyme regulating (p)ppGpp accumulation in Betaproteobacteria and Gammaproteobacteria is the RelA homolog SpoT (Fig. 2). SpoT possesses both synthetic and hydrolytic activities, albeit the synthetase activity is weak compared to RelA. Interestingly, (p)ppGpp synthesis by SpoT is induced by unique signals not sensed by RelA, which include carbon, fatty acid, and iron starvation (32–34). In addition, SpoT interacts with the GTP-binding protein CgtA (35, 36), which has been proposed to promote hydrolase activity, thereby maintaining low (p)ppGpp levels under nutrient-rich conditions (37). However, the effects of CgtA on basal (p)ppGpp pools are quantitatively minor and deserve further verification.
For some time, RelA and SpoT were considered the enzymatic paradigms for (p)ppGpp metabolism. However, over the past 2 decades, our understanding of the enzymes controlling the synthesis and degradation of (p)ppGpp has undergone several adjustments. The first modification came with the characterization of a RelA/SpoT homolog (RSH) protein from Streptococcus dysgalactiae subsp. equisimilis, named RelSeq, containing functional characteristics of both RelA and SpoT, i.e., strong synthetase activity like RelA and hydrolase activity like SpoT (38) (Fig. 2). Recently, phylogenetic studies indicated that the combined strong synthetase and hydrolase activities characteristic of Rel enzymes are ancestral to RelA and SpoT and more widely distributed in prokaryotes, including Actinobacteria, Firmicutes, and Alpha-, Delta- and Epsilonproteobacteria among many others (39). In vitro studies with the Mycobacterium tuberculosis Rel (RelMtb) revealed that, like RelA, synthetase activity is stimulated by a complex of ribosomes, uncharged tRNA, and mRNA (40). Finally, biochemical and structural studies revealed that synthetase and hydrolase activities of RelSeq are reciprocally regulated through two mutually exclusive conformations adapted by the catalytic N-terminal half of the full-length protein (41, 42). When Rel is in a hydrolase-ON state, the active configuration of the hydrolase domain is structurally different from that of the inactive synthetase domain. When the synthetase domain is ligand bound, it becomes active, placing the enzyme in a synthetase-ON and hydrolase-OFF state. It should be noted that a certain level of ambiguity has and still exists around the nomenclature of this enzyme. Rel has been commonly referred to as both RelA, because of its essentiality to SR activation, and Rsh, due to its RelA/SpoT hybrid characteristic. Recently, Atkinson and colleagues proposed that the enzymes involved in (p)ppGpp metabolism be divided into two categories: “long,” multidomain RSHs, which include the bifunctional Rel and SpoT and the monofunctional RelA, and the more recently discovered “short” single-domain RSHs described below (Fig. 2) (39). In the interest of a unified nomenclature, we will follow this naming convention here.
More recently, the genomes of different Gram-positive organisms were found to encode single-domain (p)ppGpp synthetases, known as small alarmone synthetases (SASs), which lack both the C-terminal regulatory domain and the Mn2+-dependent hydrolase domain present in long RSHs (43). These enzymes are ubiquitous in Firmicutes and have also been found in Actinobacteria and Vibrio species (39, 44–47).
To date, SASs have been best characterized in Firmicutes that encode the RelP and RelQ enzymes, albeit RelP is absent in certain species (39, 43). Several lines of evidence indicate that SAS activity is important for maintaining low basal levels of (p)ppGpp under balanced growth conditions, as their inactivation reverted the slow growth phenotype of Δrel strains, a finding that was attributed to high basal levels of (p)ppGpp (43–45). Moreover, both relP and relQ were shown to contribute to the essentiality of hydrolase activity of the Rel enzyme in Staphylococcus aureus, and spontaneously generated mutations on both relP (yjbM) and relQ (ywaC) were found to suppress the growth defect caused by rel inactivation in B. subtilis (47, 48). In M. tuberculosis, inactivation of the SAS, Rv1366, did not provide a noticeable phenotype, even when introduced in the relMtb background, suggesting Rv1366 lacks appreciable synthetase activity in vivo (49). This is consistent with recent studies by Bag and colleagues, who demonstrated that Rv1366 is unable to synthesize ppGpp in vitro (50). In contrast, the Mycobacterium smegmatis SAS, MSM_5849, exhibited weak (p)ppGpp synthetase activity as well as Mn(II)-dependent RNase HII activity (51). Of note, this SAS-RNase HII fusion appears to occur with some frequency among species of mycobacteria. Finally, one SAS, termed RelV, has been characterized in V. cholerae and found to be highly conserved among Vibrio species (46). The presence of RelV was unexpected, as Vibrio species encode the canonical relA/spoT genes of other Gammaproteobacteria and this is the only known example of a SAS-containing enzyme in this bacterial group (46). RelV was shown to efficiently synthesize (p)ppGpp both in vitro and in vivo (46, 52).
The seemingly stable coexistence of long RSHs with SASs suggests some evolutionary advantage to this apparent enzymatic redundancy. The contribution of a SAS to stress tolerance and cell homeostasis, particularly in the presence of a fully active Rel enzyme, is not entirely clear, albeit recent studies have shed some light onto their biological roles. In S. aureus, simultaneous inactivation of relP and relQ significantly enhanced susceptibility to cell wall-targeting antibiotics (47). Interestingly, competence and basal (p)ppGpp production by RelP appears to be interconnected in Streptococcus mutans. The transcription of the S. mutans relP is under the control of and cotranscribed with the RelRS two-component system. A gene cluster located upstream of relP encoding two ABC transporters (RcrPQ) and a DNA-binding transcriptional regulator (RcrR) were found to be critical for competence development and maintenance of basal (p)ppGpp through activation of relRSP (53). It has been hypothesized that the Rcr (Rel competence-related) operon secretes a quorum-sensing molecule sensed by RelRS that modulates cell growth and competence in response to specific signals (53). As detailed below, low levels of (p)ppGpp produced by RelQ, the sole SAS found in enterococci, are sufficient for persistence and full virulence in Enterococcus faecalis (45, 54). However, this linkage of low basal (p)ppGpp levels, rather than the SR, to virulence is not entirely unexpected. In several Gram-negative pathogens, virulence is only abolished or attenuated in relA spoT double-knockout strains and not in relA single mutants, which produce low levels of (p)ppGpp due to weak SpoT synthetase activity (7, 26). Based on these pioneering studies, it appears that bacteria could utilize SASs to fine-tune cell physiology in a species-specific manner.
During the biochemical characterization of E. faecalis RelQ, we recently found that this enzyme was able to efficiently utilize GMP as a pyrophosphate acceptor to synthesize pGpp (GMP, 3′-diphosphate). We followed this observation by showing that pGpp has specific regulatory effects on (p)ppGpp-controlled processes, including strong inhibition of enzymes involved in GTP biosynthesis (A. O. Gaca, C. Colomer-Winter, P. Kudrin, J. Beljantseva, K. Liu, B. Anderson, J. D. Wang, D. Rejman, K. Potrykus, M. Cashel, V. Hauryliuk, and J. A. Lemos, submitted for publication). Interestingly, pGpp was previously shown to accumulate in B. subtilis during amino acid limitation, but its source, either direct synthesis from GMP or nonspecific degradation of (p)ppGpp, was not determined (55).
The most recent major discovery in (p)ppGpp metabolism was the identification of stand-alone small alarmone hydrolases (SAHs) (Fig. 2) (56). These enzymes were first discovered in metazoans and were named Mesh1. Deletion of Mesh1 in Drosophila melanogaster resulted in slow body growth and impaired starvation resistance, features reminiscent of the bacterial SR. In fact, heterologous expression of Mesh1 in E. coli can substitute for the absence of SpoT-mediated hydrolase activity and abolish RelA-induced cell growth delays (56). The fact that (p)ppGpp has not been detected in eukaryotes besides plant chloroplasts suggests that Mesh1 might function to degrade related polyphosphorylated nucleotides with a similar function in metazoans. Putative SAH homologs have since been identified in a diverse range of organisms, including prokaryotes, but functional studies are still missing (39). In addition to SAHs, nonspecific hydrolases, including Nudix and the phosphohydrolase MazG, also appear to be important for maintaining intracellular (p)ppGpp at levels compatible with balanced growth (57, 58). The presence of nonspecific hydrolases may explain why deletion of the primary (p)ppGpp hydrolase (i.e., Rel) is not lethal in some SAS-encoding bacterial species.
(p)ppGpp BEYOND THE SR: BASAL (p)ppGpp POOLS AND CELL HOMEOSTASIS
Most research regarding the protective and regulatory aspects of (p)ppGpp has focused on SR induction. Here, we have defined the SR as the rapid and dramatic accumulation of (p)ppGpp that occurs during stress, resulting in the strong repression of macromolecular biosynthesis and activation of stress survival pathways. This is not to say that the contributions of subtle fluctuations in (p)ppGpp pools to the control of core cellular processes and growth rate have been completely overlooked (59–62). However, more recently, the notion that (p)ppGpp exerts important regulatory effects at concentrations below those needed to activate the SR has been more thoroughly revisited (10, 54, 63, 64). In an elegant study using cells that were progressively starved, the Conway lab identified two distinct regulatory cascades activated by (p)ppGpp in E. coli (64). First, at slightly elevated (p)ppGpp concentrations, the leucine-responsive protein (Lrp) regulon, a global regulator of genes involved in amino acid biosynthesis and transport, is activated (65). At higher (p)ppGpp concentrations, the RpoS regulon, controlled by the general stress sigma factor (σS), is induced. Thus, E. coli cells have discretely calibrated responses to a gradient of (p)ppGpp (64). In this biphasic stress response, activation of Lrp represents an initial attempt to restore intracellular amino acid pools at the onset of starvation. However, if this homeostatic mechanism fails to meet cellular amino acid demands, the RpoS regulon is activated to ensure survival (Fig. 3A). Of note, Lrp is restricted to Betaproteobaceria and Gammaproteobacteria and, although more widely distributed than Lrp, σS is absent in several bacteria, including some major pathogenic species. Thus, the concentration-dependent regulatory networks of (p)ppGpp in other bacteria have yet to be investigated.
Simplified view of the concentration-dependent effects of (p)ppGpp on growth and survival in the most studied bacterial groups. In Gammaproteobacteria, Lrp and RpoS (σS) function as discretely calibrated regulatory units triggered by increasing concentrations of (p)ppGpp (64). Moderate accumulation of (p)ppGpp due to mild nutrient limitation activates expression of the Lrp regulon, which is responsible for inducing pathways needed to restore metabolic homeostasis. Severe nutrient limitation causes a more dramatic increase in (p)ppGpp pools responsible for activation of the RpoS (σS) regulon, inducing the expression of stress survival genes. In Firmicutes, an inverse relationship between (p)ppGpp and GTP dictates whether cells maintain a physiological program compatible with growth or switch to a state best suited for survival under slow growth or stasis (23, 71). Under nutrient-rich conditions compatible with rapid growth, CodY is active, controlling gene expression to direct pyruvate into fermentation and secretion of by-products like lactate, acetate, and ethanol. Nutrient starvation triggers the accumulation of (p)ppGpp and a subsequent drop in GTP. Alleviation of CodY regulation due to BCAA and GTP depletion activates transcription of genes involved in nutrient transport, amino acid biosynthesis, and virulence. The inability to maintain GTP and (p)ppGpp homeostasis leads to extreme fluctuations in GTP which can be highly inhibitory and, possibly, lethal (67). This cartoon is not inclusive of all pathways affected by alteration in (p)ppGpp or GTP levels but highlights the unique pathways used by several well-studied bacterial groups to propagate incremental alterations in these nucleotide pools.
We have explored in some detail the importance of (p)ppGpp outside the SR in the opportunistic pathogen E. faecalis. During phenotypic characterization of Δrel and Δrel ΔrelQ [(p)ppGpp0] strains, both lacking the ability to mount the SR, we observed important differences between the two strains that could be clearly attributed to either the presence or absence of RelQ. In the Δrel strain, basal levels of (p)ppGpp synthesized by RelQ supported wild-type levels of antibiotic persistence and virulence (45, 54, 66). Only the deletion of both rel and relQ, completely abolishing (p)ppGpp synthesis, resulted in virulence attenuation and increased sensitivity to antibiotics (45, 54, 66). These phenotypes can be explained, in part, by the metabolic dysregulation observed in the Δrel ΔrelQ strain (54). In the absence of (p)ppGpp, E. faecalis undergoes large-scale transcriptional alterations in secondary carbon metabolism, which result in a shift from a homolactic to a heterofermentative metabolism with a concomitant increase in H2O2 production (54). These results suggest that E. faecalis, and likely other Firmicutes, depends on small amounts of (p)ppGpp to control the pace and direction of carbon flow and to accurately respond to external and internal metabolic cues. In addition, complete loss of (p)ppGpp also leads to dysregulation of GTP homeostasis (see below), which can severely impair cell fitness (54, 67).
In both B. subtilis and E. faecalis, (p)ppGpp directly and strongly inhibits GTP biosynthesis by targeting HprT (hypoxanthine-guanine phosphoribosyltransferase) and Gmk (GMP kinase). The 50% inhibitory concentrations of (p)ppGpp for HprT and Gmk are extremely low, ranging from 11 to 80 μM, indicating that basal (p)ppGpp levels are important for GTP homeostasis (54, 68). This is evidenced by the observation that the addition of exogenous guanosine to (p)ppGpp0 strains, which is converted to GTP via the salvage pathway, significantly increased GTP levels, whereas GTP levels remained constant in (p)ppGpp+ strains (54, 68). The accumulation of GTP in (p)ppGpp0 strains was highly inhibitory and could even induce cell death (“death by GTP”) in B. subtilis (67). In contrast, in E. coli uptake of purine as well as pyrimidines is blocked by ppGpp, and yet the presence of both bases or nucleosides in the medium does not give lethal elevations of the corresponding nucleotides in (p)ppGpp0 strains (69). In addition to the association with rRNA gene transcription and CodY activation (see below), GTP activates a wide variety of anabolic processes, including nucleic acid synthesis and all three steps of translation (70). Based on the central role of GTP in cell homeostasis, it is not surprising that a tight inverse correlation between GTP pools and bacterial survival has been observed (67). The reduction of intracellular GTP pools in B. subtilis, either through mutation of enzymes in the de novo GTP biosynthesis pathway or by treatment with decoyinine, a selective inhibitor of GMP synthetase, enhanced resistance to amino acid limitation (67). Collectively, these studies indicate that one of the primary protective mechanisms of (p)ppGpp is to regulate intracellular GTP pools (Fig. 3B). As discussed below, the inverse relationship of GTP and (p)ppGpp may function as a metabolic switch that controls growth and survival.
(p)ppGpp AND CodY
As stated before, Firmicutes also integrate the inverse relationship between (p)ppGpp and GTP levels to control activity of the transcriptional regulator CodY (23). Specifically, reduction in cellular pools of the coeffector GTP by (p)ppGpp results in less stable CodY-DNA interactions, thereby alleviating CodY regulation (71). Importantly, the (p)ppGpp/CodY association is critical for nutrient stress tolerance and virulence in several bacterial pathogens (72–77). Yet, the relative contributions of GTP- and CodY-dependent mechanisms for stress tolerance and virulence controlled by (p)ppGpp are not entirely clear, as GTP depletion can also affect stress survival in a CodY-independent manner (67, 68). In addition, CodY in both Lactococcus and Streptococcus species appears to be insensitive to GTP (73, 78, 79). Parsing the contributions of GTP pool fluctuations and CodY to (p)ppGpp-controlled phenomena in Firmicutes will be an important step in fully understanding the protective effects of (p)ppGpp. Additionally, it will be important to determine if more subtle changes in GTP levels due to maintenance of basal (p)ppGpp pools are sufficient to influence CodY activity.
THE CENTRAL ROLE OF (p)ppGpp IN BACTERIAL PERSISTENCE
Bacterial persisters are phenotypic variants that enter a slow-growing or dormant state and transiently become multidrug tolerant (80). Because persisters are able to grow when the antibiotic is removed, bacterial persistence has been implicated in chronic and recurrent infections, particularly those of biofilm origin. Although (p)ppGpp accumulation has been linked to antibiotic tolerance for a while (81–84), only recently have the underlying mechanisms by which (p)ppGpp mediates bacterial persistence begun to be elucidated (Fig. 4). In particular, different laboratories have systematically shown that increased (p)ppGpp levels parallel observed increases in persistence (85–89). By coupling microfluidics with fluorescent reporter gene fusions in E. coli, the Gerdes group demonstrated that (p)ppGpp levels vary stochastically in exponentially growing cultures, and these investigators confirmed that persistence and (p)ppGpp levels are positively correlated at the single-cell level (89). In the current E. coli-based model, (p)ppGpp triggers persistence by activation of toxin-antitoxin (TA) loci through a regulatory cascade that involves (i) allosteric inhibition by (p)ppGpp of the exopolyphosphatase (Ppx) enzyme responsible for inorganic polyphosphate (PolyP) degradation, (ii) activation of the Lon protease by poly(P), (iii) Lon-dependent degradation of antitoxins from different TA modules, and (iv) inhibition of transcription and translation by the free toxins of the TA modules (86). Interestingly, the Hip (high-persistence) TA system, the first genetic factor implicated in persister cell formation (90), was shown to impair glutamyl tRNA synthase (GltX) activity via direct phosphorylation of GltX by the HipA toxin (91, 92). This phosphorylation event was shown to induce accumulation of uncharged tRNAs, thereby triggering substantial amounts of (p)ppGpp production through RelA. This persister-activation cascade is further supported by studies with Salmonella enterica serovar Typhimurium that have shown that (p)ppGpp, Lon, and TA modules are also required for Salmonella persistence, either within macrophages or during antibiotic treatment (93–95). In addition to stochastic activation of TA modules by (p)ppGpp, Nguyen and colleagues proposed that the SR protected E. coli and Pseudomonas aeruginosa from the lethal effects of antibiotics through an active mechanism, which included the induction of antioxidant defenses (96, 97).
(p)ppGpp-triggered systems controlling persister cell formation. (A) In E. coli, (p)ppGpp sits atop a sequential biochemical signaling network for the induction of persisters. The accumulation of (p)ppGpp inhibits the activity of exopolyphosphatase (PPX), allowing poly(P) to accumulate. Poly(P) then activates Lon protease to target and degrade antitoxins of type II toxin-antitoxin (TA) pairs. Free toxins go on to inhibit processes essential for active cell growth (89). (B) The nucleotides (p)ppGpp and GTP may act as a metabolic switch in Firmicutes. The accumulation of (p)ppGpp leads to a dramatic reduction in GTP levels by directly consuming GTP and GDP but also inhibits the activity of Gmk and HprT by blocking both de novo and salvage pathways of GTP biosynthesis (67). GTP is an essential cofactor in numerous anabolic process needed for cell growth. This reduction in GTP slows bacterial growth and leads to alleviation of CodY regulation (see Fig. 3). An inverse correlation between GTP and cell fitness has been observed, with a reduction in GTP pools having a general protective effect (67, 119).
Although reasonable progress has been made in understanding the underlying mechanisms by which (p)ppGpp mediates persistence in E. coli and related Gammaproteobacteria, homologous molecular pathways leading to persistence in other bacterial groups remain poorly understood. Yet, recent studies support the involvement of (p)ppGpp in persister cell formation in members of the Firmicutes phylum. For example, whole-genome sequencing analysis of bacterial isolates from a patient with recurrent S. aureus infections identified a single-nucleotide substitution in the rel gene that affected hydrolase activity and caused accumulation of (p)ppGpp (98). Similar observations have been made with laboratory strains, as increased (p)ppGpp production due to spontaneous point mutations in rel were observed in S. aureus populations that survived lethal doses of methicillin (87, 88).
In agreement with the generalized role of (p)ppGpp in persister activation, complete lack of (p)ppGpp in E. faecalis (Δrel ΔrelQ strain) dramatically reduced the number of persisters (45, 54, 66). However, loss of Rel did not lower persistence rates and, depending on the drug target, the Δrel strain, which has ∼4-fold-higher basal levels of ppGpp due to constitutive alarmone synthesis by RelQ (54), produced a significantly higher number of persisters than the parent strain (45). This finding is a departure from the general concept that the SR mediates bacterial persistence, as activation of the SR in E. faecalis and Firmicutes in general is dependent on the Rel enzyme (43, 45, 66, 74, 99–101). Resolution of this notion will require systematic studies on basal (p)ppGpp elevation in other related species. Taking into consideration that RelP and RelQ mediate tolerance against cell wall-active antibiotics in S. aureus, it becomes clear that the association of (p)ppGpp with persistence in Firmicutes does not relate directly to the SR but rather to basal (p)ppGpp pools. Based on the critical role of (p)ppGpp in GTP homeostasis (54, 67), it is tempting to speculate that (p)ppGpp mediates persistence in Firmicutes via GTP regulation. Thus, while (p)ppGpp may indeed function as a universal mediator for persister formation, the underlying molecular mechanisms acting upstream and downstream of (p)ppGpp signaling may vary among bacteria. Clearly, additional investigations are necessary to address this possibility, particularly studies that draw a direct comparison between the mechanisms of persistence in phylogenetically diverse bacteria.
(p)ppGpp AS A TARGET FOR ANTIMICROBIAL DRUG DEVELOPMENT
The intimate association of (p)ppGpp regulation in bacterial persistence and virulence makes (p)ppGpp signaling interference a promising target for drug development. In fact, two recent studies have confirmed the potential usefulness of antimicrobial approaches that interfere with (p)ppGpp accumulation. In the first study, Relacin, a (p)ppGpp analog, was shown to specifically inhibit synthesis activity of RelA and Rel enzymes (102). Relacin was also shown to reduce in vitro survival of B. subtilis and Streptococcus pyogenes and hinder biofilm formation and sporulation processes in B. subtilis (102). In a more recent study, a broad-spectrum peptide (peptide 1018) was found to specifically bind and promote (p)ppGpp degradation (103). Interestingly, peptide 1018 showed stronger activity toward biofilms causing cell death at concentrations that did not affect planktonic cells. In a follow-up report, those authors demonstrated that low doses of peptide 1018 act synergistically with conventional antibiotics to kill a variety of drug-resistant pathogens (104). Based on these two promising examples, it seems reasonable to envision the usefulness of specific (p)ppGpp inhibitors as antipersister drugs in combination therapies.
CONCLUDING REMARKS
About 45 years ago, the discovery of two hyperphosphorylated guanosine nucleotides that were capable of reprogramming cell physiology provided one of the first clues into the complexity of how bacteria utilize alarmone production to regulate a multilayered network controlling bacterial growth and survival. Although much has been learned in the interceding years, several outstanding questions remain. For example, we have just started to understand the biological significance of basal (p)ppGpp under balanced growth conditions. Despite decades of research, many aspects of the mechanism by which RSH enzymes catalyze the conversion of ATP and GTP/GDP to pppGpp/ppGpp are still unknown. Moreover, basic questions as to how and when (pp)pGpp is produced by SASs and how (p)ppGpp mediates persistence in organisms with very different lifestyles are also poorly understood.
As discussed in this review, the ubiquity and critical role of RSH enzymes in bacteria, combined with their absence in mammalian cells, mark the enzymes involved in (p)ppGpp metabolism as potential targets for the development of new antimicrobial strategies. However, an important aspect that should not be overlooked in the development of such antimicrobials is the prevalence of SASs in important bacterial pathogens. Given the essential role of basal (p)ppGpp pools in bacterial virulence and persistence, it will be critical to identify compounds that target both “long” RSHs and SASs, thereby effectively eliminating basal (p)ppGpp production. Another interesting possibility is a reverse approach, i.e., identifying compounds that target (p)ppGpp hydrolase activity, thus locking the bifunctional enzymes (SpoT or Rel) in a synthetic mode that, conceivably, would lead to toxic accumulation of (p)ppGpp. A similar approach that relied on target activation rather than inactivation was recently shown to aid in the elimination of bacterial persisters. Specifically, the acyldepsipeptide (ADEP4) antibiotic has been shown to activate the ClpP protease, leading to uncontrolled proteolysis and ultimately cell death (105). Although clpP null mutants arose at a high frequency, combinatorial treatment with ADEP4 and rifampin eradicated biofilm persisters in both in vitro and in vivo models (106). Regardless of the direction taken to interfere with (p)ppGpp metabolism, a better structural and biochemical understanding of the mechanisms of action of the enzymes responsible for synthesis and degradation of (p)ppGpp can be an invaluable resource in the development of clinically effective antibiotics.
ACKNOWLEDGMENTS
We thank Mike Cashel for helpful discussions and critical reading of the manuscript. We also thank James H. Miller for comments and suggestions to improve the text.
A.O.G. was supported by an NIDCR Training Grant in Oral Science (T90 DE021985).
- Copyright © 2015, American Society for Microbiology. All Rights Reserved.
REFERENCES
- 1.↵
- 2.↵
- 3.↵
- 4.↵
- 5.↵
- 6.↵
- 7.↵
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.↵
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.↵
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.↵
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.↵
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.↵
- 72.↵
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.↵
- 90.↵
- 91.↵
- 92.↵
- 93.↵
- 94.↵
- 95.↵
- 96.↵
- 97.↵
- 98.↵
- 99.↵
- 100.↵
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.↵
- 107.↵
- 108.↵
- 109.↵
- 110.↵
- 111.↵
- 112.↵
- 113.↵
- 114.↵
- 115.↵
- 116.↵
- 117.↵
- 118.↵
- 119.↵
Author Bios

Anthony O. Gaca was raised in Grand Rapids, MI, and completed his B.S. in Microbiology at Michigan State University in 2008. While at Michigan State, he contributed to research characterizing metabolic pathways and nutritional requirements for the bioremediation of phenolic compounds and heavy metals. In 2013, Anthony completed his Ph.D. at the University of Rochester under the guidance of José Lemos. His doctoral research focused on defining the physiological contributions and transcriptional hierarchy under the control of (p)ppGpp in the opportunistic pathogen Enterococcus faecalis. Currently, he is characterizing the enzymatic function of small (p)ppGpp synthetases in E. faecalis and other Firmicutes. He began working in the area of bacterial stress tolerance to blend his bioremediation experience with his interest in bacterial pathogenesis.

Cristina Colomer-Winter is a Ph.D. student at the University of Rochester. She was raised in Barcelona, Spain. In 2009, she received a B.S. in Biology at the Universitat Autònoma de Barcelona, where she completed several internships at Boehringer Ingelheim (Germany) and at the Universität Würzburg. Later on, she completed her M.Sc. degree in Microbiology at the Universitat de Barcelona under the mentorship of Jaume Jané at B. Braun Medical (Rubí, Spain), characterizing the heat resistance of Bacillus spp. spores present in the bioburden. After graduating, she continued working for 3 years at the R&D Department of B. Braun, developing antibacterial medical devices and novel applications for tissue regeneration. She joined the lab of José Lemos in the summer of 2014 to start work on the interaction of (p)ppGpp, GTP, and CodY in Enterococcus faecalis pathophysiology.

José A. Lemos was raised in Rio de Janeiro, Brazil, and obtained his Ph.D. in Microbiology and Immunology from the Federal University of Rio de Janeiro in 2000, working with Angela Castro. He did postdoctoral research with Robert Burne at the University of Rochester and later at the University of Florida, where he initiated his studies on the stress response mechanisms of Firmicutes. Around this time, he became interested in the (p)ppGpp-mediated bacterial stringent response while investigating the regulation of toxin-antitoxin modules in Streptococcus mutans. In 2002, he became a research assistant professor in the Department of Oral Biology at the University of Florida. In 2007, he joined the faculty of the University of Rochester in the Center for Oral Biology, where he is currently an associate professor of microbiology and immunology. His laboratory investigates the stress survival mechanisms of streptococci and enterococci.