Abstract
Phenotypic heterogeneity of microbial populations can facilitate survival in dynamic environments by generating sub-populations of cells that may have differential fitness in a future environment. Bacillus subtilis cultures experiencing nutrient limitation contain distinct sub-populations of cells exhibiting either comparatively high or low protein synthesis activity. This heterogeneity requires the production of phosphorylated guanosine nucleotides (pp)ppGpp by three synthases: SasA, SasB, and RelA. Here we show that these enzymes differentially affect this bimodality: RelA and SasB are necessary to generate the sub-population of cells exhibiting low protein synthesis whereas SasA is necessary to generate cells exhibiting comparatively higher protein synthesis. The RelA product (pppGpp) allosterically activates SasB and we find, in contrast, that the SasA product (pGpp) competitively inhibits this activation. Finally, we provide in vivo evidence that this antagonistic interaction mediates the observed heterogeneity in protein synthesis. This work therefore identifies the mechanism underlying phenotypic heterogeneity in the central physiological process of protein synthesis.
Author Summary Upon encountering conditions that are unfavorable to growth, such as nutrient limitation, bacteria enter into a quiescent phenotype that is mediated by group of guanosine nucleotides collectively known as (pp)pGpp. These nucleotides direct the down-regulation of energy intensive processes and are essential for a striking heterogeneity in protein synthesis observed during exit from rapid growth. Here, we show that a network of (pp)pGpp synthases is responsible for this heterogeneity and describe a mechanism that allows for the integration of multiple signals into the decision to down regulate the most energy intensive process in a cell.
Introduction
Nutrient availability is a major environmental cue for bacteria. For example, amino acid starvation results in induction of the stringent response, a conserved mechanism dependent on the synthesis of the nucleotides guanosine penta- and tetra-phosphate ((p)ppGpp). These nucleotides mediate a broad shut down of energy intensive reactions which are required during rapid growth (1, 2). (p)ppGpp directly binds and inhibits key proteins that catalyze processes including transcription (RNA polymerase (3, 4)), translation (GTPase IF2 (5)), GTP biosynthesis (HprT and GmK (6)), DNA replication (DNA primase (7)), and ribosome assembly (ObgE and RsgA (8)).
Gram-positive bacteria typically encode a single, bi-functional RSH enzyme capable of both (p)ppGpp synthesis and hydrolysis as well as two additional small alarmone synthases (SAS) that lack hydrolytic activity. Unlike RSH proteins, which are activated by the binding of deacylated tRNAs to the A-site of the ribosome, SAS enzymes are believed to be transcriptionally regulated (9) and some are also under allosteric control (10). RelA/SpoT and the SAS synthases preferentially produce different molecules in different species. For example, in response to amino acid starvation, E. coli RelA produces approximately equal amounts of the tetra-phosphorylated (ppGpp) and the penta-phosphorylated (pppGpp) guanosines, whereas B. subtilis RelA primarily generates pppGpp using GTP and ATP as substrates (11). B. subtilis SasB preferentially utilizes GDP and ATP to generate the tetra-phosphorylated guanosine (ppGpp) (12) and SasA, the other SAS enzyme in B. subtilis, primarily generates a 5’ monophosphate 3’ di-phosphate guanosine (pGpp) using GMP and ATP as substrates in vivo (12). Together, these three closely related nucleotides are referred to as (pp)pGpp.
Recently, our laboratory demonstrated that accumulation of (pp)pGpp attenuates protein synthesis when populations of B. subtilis cease growing exponentially (5). This attenuation is bimodal and results in a heterogeneity in the protein synthesis activity of individual cells that exhibit either comparatively high or low protein synthesis activity (5). Here we find that a network of interacting (pp)pGpp synthases including a RSH protein (RelA) and two SAS proteins (SasA, SasB) underlies this heterogeneity since the absence of any of these synthases results in the loss of bimodality. The products of SasA and RelA, pGpp and pppGpp respectively, together antagonistically regulate activation of the third synthase (SasB), that is itself responsible for the synthesis of ppGpp, which inhibits Initiation Factor 2 and thereby attenuates protein synthesis (5).
Results
The SasA and SasB (p)ppGpp synthases contribute to heterogeneity
Cellular heterogeneity in protein synthesis as B. subtilis cultures exit rapid growth is dependent on the presence of the phosphorylated guanosine nucleotides (pp)pGpp (5). We investigated the origins of this heterogeneity by assessing single cell protein synthesis using O-propargyl-puromycin (OPP) incorporation in strains carrying deletion mutations in either of the two B. subtilis (pp)pGpp synthases (SasA and SasB) whose expression increases during exit from rapid growth (12). To quantify these effects we applied a cutoff that specifies the population of cells with low rates of protein synthesis. We set the threshold of this cutoff (850 relative fluorescence units (RFU)) as the magnitude of OPP labeling of wildtype B. subtilis culture in late transition phase (Fig. S1A) that captures 95% of the entire population. We used this threshold to define the fraction of the population with low rates of protein synthesis (“OFF”) (Fig. S1B). By convention, we define the remainder of the population as “ON.”
A strain lacking SasB (ΔsasB) contained fewer “OFF” cells as compared to the wildtype strain (Fig. 1A, B; S2). This result is consistent with our previous observation that the SasB product ppGpp inhibits the function of IF2 and thereby downregulates protein synthesis (5). In contrast, a strain lacking SasA (ΔsasA) does not contain the substantial fraction of “ON” cells seen in the wildtype parent strain (Fig. 1A, C; S2) and most cells in the population are “OFF”. This observation suggests that the SasA product pGpp does not directly inhibit translation, as does ppGpp, but rather acts indirectly.
sasA but not sasB expression is correlated with levels of protein synthesis
sasA and sasB are regulated transcriptionally and expressed post-exponentially (12, 13) when the heterogeneity is observed (Fig 1A). We therefore asked if expression of either sasA or sasB is correlated with protein synthesis using transcriptional fusions of the sasA or the sasB promoters to YFP (PsasA-yfp or PsasB-yfp). Consistent with prior observations (12), expression of both sasA and sasB reporters increased during the exit from exponential growth (Fig 2A, B). We examined the relationship between promoter activity and protein synthesis by measuring both YFP expression and OPP incorporation in single cells. Cells with higher sasA expression (PsasA-yfp) are more likely to have higher levels of protein synthesis than cells with lower sasA expression (Fig 2D). If the population is divided into quartiles of sasA expression, average OPP incorporation in the top two quartiles as compared to the bottom quartile is significantly higher (Fig 2D). In comparison, there was no significant difference in OPP incorporation between any of the quartiles of sasB expression (Fig 2C). Thus, differences in sasA, but not sasB, expression are associated with the observed heterogeneity in protein synthesis.
SasB allosteric activation is necessary for heterogeneity
If changes in sasB transcription are not associated with differences in protein synthesis (Fig 2C), but SasB is necessary for the heterogeneity of protein synthesis (Fig 1B), what mechanism underlies differential SasB activity in single cells? B. subtilis SasB is subject to allosteric activation by pppGpp, the main product of B. subtilis RelA (14). Phe-42 is a key residue in this activation and a SasB mutant protein carrying an F42A substitution (SasBF42) is not allosterically activated by pppGpp in vitro (14). We investigated the importance of this allosteric activation for protein synthesis heterogeneity using a strain expressing SasBF42. Heterogeneity of this strain is significantly attenuated compared to the WT strain, demonstrating the importance of the allosteric activation of SasB by pppGpp for the bimodality of protein synthesis activity (Fig 3A, B; S3 Fig).
This result suggests that the enzyme responsible for pppGpp synthesis could also affect the heterogeneity. RelA is the primary source of pppGpp in B. subtilis (11), so the loss of relA would be predicted to affect SasB activity. We therefore generated a strain expressing a RelA mutant protein (RelAY308A) carrying a single amino acid change at a conserved residue essential for synthase but not hydrolysis activity (15, 16) since RelA hydrolytic activity is essential in a strain that retains functional sasA and sasB genes (17). Labeling of this strain with OPP in late transition phase revealed that the “OFF” population was largely absent (Fig 3C; S3), demonstrating that RelA-mediated pppGpp synthesis is important for the bimodality.
SasB allosteric activation is inhibited by pGpp
A strain lacking SasA (ΔsasA) contains more “OFF” cells as compared to the wildtype parent (Fig 1C). The presence of this sub-population of cells depends on a SasB protein that can be allosterically activated (Fig 3B). Integrating these two observations, we hypothesized that the product of SasA (pGpp) inhibits the allosteric activation of SasB by pppGpp. The similarity of pGpp and pppGpp suggests that they could have an antagonistic interaction since they are likely capable of binding to the same site on SasB, but their differing phosphorylation states could affect their ability to allosterically activate SasB.
We tested this model by assaying in vitro whether pGpp inhibits the allosteric activation of SasB. First, we confirmed that SasB generates more ppGpp when reactions are supplemented with pppGpp (14) and observed a ∼2 fold increase in ppGpp production when SasB was incubated with pppGpp (Fig 4A). Using pGpp synthesized in vitro by the recently identified (p)ppGpp hydrolase NahA (18), we observed that pGpp attenuates the allosteric activation of SasB in a dose dependent manner (Fig 4A). Since even the highest concentration of pGpp did not decrease production of ppGpp relative to that generated by SasB without the addition of pppGpp (Fig 4A), the inhibition is likely specific to the allosteric activation. We first tested this directly by assaying the effect of pGpp on SasB activity in the absence of its allosteric activator (pppGpp). Addition of pGpp did not significantly affect SasB activity within the range of pGpp concentrations we used previously (Fig S4). We further confirmed the specificity by assaying a SasBF42 mutant protein that is insensitive to allosteric activation by pppGpp (14). As previously reported SasBF42A has similar activity to a non-allosterically activated WT SasB in the presence of pppGpp (Fig 4B). However, in contrast with wildtype SasB, pGpp does not affect the activity of SasBF42A even when pppGpp is included (Fig 4B).
These in vitro biochemical experiments suggest that the effect of SasA on protein synthesis heterogeneity is dependent on the activity of SasB. If this is true in vivo, then the phenotype of a ΔsasA mutation should be epistatic to that of a ΔsasB mutation. Consistently, the population of “OFF” cells in a ΔsasA strain is absent in a strain lacking both SasA and SasB (sasA sasB) (Fig 5A, B; S5). Thus, the effect of SasA is dependent in vivo on SasB. Finally, since RelA activates SasB, a ΔsasA mutation should be epistatic to a relA mutation with respect to protein synthesis. A strain expressing RelAY308A and carrying a ΔsasA mutation exhibits a loss of heterogeneity similar to the relAY308A strain, demonstrating that the effect of the ΔsasA mutation depends on a functional RelA synthase (Fig 5A, C; S5). This result is consistent with the hypothesis that sasA is epistatic to relA.
While SasA is the only known (pp)pGpp synthase that predominately produces pGpp in vivo in B. subtilis (12), pGpp also accumulates in stationary phase cells as a result of degradation of both ppGpp and pppGpp by the (p)ppGpp hydrolase NahA (18, 19). We therefore asked if NahA contributes to the heterogeneity in protein synthesis by comparing OPP incorporation in wildtype and Δ nahA cells during late transition phase. We observed no difference in heterogeneity (Fig S6) consistent with SasA being the primary source of pGpp.
Discussion
B. subtilis populations experiencing nutrient limitation and entering into quiescence respond bimodally with respect to global protein synthesis activity (5). Here, we find that this bimodality depends on all three (pp)pGpp synthases. We demonstrate that it is dependent on the allosteric activation of SasB by the RelA product pppGpp and that this activation is antagonized by the SasA product pGpp (Fig 6A). Our work therefore provides a mechanism for the phenotypic heterogeneity observed and identifies novel regulatory interactions between (pp)pGpp synthases.
Regulation of protein synthesis during nutrient limitation
The downregulation of protein synthesis in B. subtilis cells experiencing nutrient limitation occurs as a result of ppGpp binding and thereby inhibiting IF2 (5). SasB is the main source of ppGpp and this work identifies how ppGpp synthesis by SasB and the subsequent downregulation of protein synthesis is coupled to changes in environmental conditions. First, SasB allosteric activation by the RelA product pppGpp is required for the downregulation of protein synthesis in a subpopulation of cells (Figs 3, 5). RelA activity reflects tRNA charging levels (11), thereby coupling SasB-dependent regulation of protein synthesis to amino acid availability. Second, the SasA product pGpp inhibits the allosteric activation of SasB (Fig 4). Although SasA is constitutively active, sasA expression, at least in part, reflects availability of the Lipid II peptidoglycan precursor (20-22), thereby coupling SasB-dependent regulation of protein synthesis to cell wall metabolism. Thus, the roles of RelA and SasA in regulating SasB activation provides a mechanism to integrate multiple environmental signals in the decision to attenuate protein synthesis.
Physiological sources of variability in SasB activity
Phenotypic heterogeneity such as that observed here in the context of protein synthesis can arise from stochastic differences in gene expression (23). Although sasB expression exhibits substantial variability in expression cell to cell (Fig 2A), it does not correspond with the level of protein synthesis in individual cells (Fig 2C). Thus, variability of SasB activity in single cells is likely relevant. What could be responsible? Our observations link heterogeneity to the convergent regulation of SasB allosteric activation by the products of the RelA and SasA synthases, pppGpp and pGpp, respectively (Fig 4A). Thus, both enzymes are potential sources of variability and, consistently, strains carrying either relAY308A or ΔsasA mutations exhibit a loss in heterogeneity as compared to the wildtype (Figs 1C, 3C). Since RelA is a cellular sensor of tRNA charging, levels of which are highly sensitive to growth conditions (24), variations in this parameter could contribute to variability in protein synthesis via modulation of RelA activity. Noise in sasA transcription is dependent on the activity of PrkC, a membrane Ser/Thr kinase that regulates sasA via the essential WalRK two component system (20). Since both WalRK (25) and PrkC (26) activities reflect cell wall metabolism, variation in this process could also impact sasA variability. Thus, differences in the protein synthesis activity of individual cells may reflect cellular variations in amino acid and cell wall metabolism.
Allosteric activation of (pp)pGpp synthases
Many genes encoding SAS proteins such as sasB are transcriptionally regulated (9). In addition, we observe here there that allosteric activation of SasB by pppGpp (14) is required for the attenuation of protein synthesis (Fig 3) demonstrating sasB transcription is necessary but not sufficient, at least in the physiological context of nutrient limitation. We also find that this allosteric activation is antagonized by the SasA product pGpp (Fig 6A), consistent with the epistatic relationship between sasB and sasA (Fig 5A). Antagonistic regulatory mechanisms are likely widespread in this family of synthases. For example, the SasB homolog Enterococcus faecalis RelQ is attenuated by RNA that competes with pppGpp for binding to the allosteric site (27). Given the very recently observed allosteric activation of B. subtilis RelA by (p)ppGpp (28), an important question for future study is to determine whether this activation is also subject to antagonism by pGpp and, if so, to characterize the physiological consequences of this regulation.
(pp)pGpp synthases
The different protein synthesis activity of strains carrying a mutation in one of the genes encoding a (pp)pGpp synthase (Figs 1, 3C) is consistent with previous reports that SAS enzymes differ between themselves and also with RelA in the guanosine nucleotide that they preferentially produce (18, 29-31). Our observations demonstrate that each particular product differs in its in vivo function, thereby extending previous observations that ppGpp and pppGpp can differ in their effect on gene transcription in E. coli (32). The biochemical experiments demonstrating that pGpp antagonizes pppGpp allosteric activation of SasB, but itself is not capable of activation (Fig 4A; S4) are consistent with our physiological experiments. The biochemical activity of these nucleotides have been reported to differ, including observations that pppGpp is much more potent than ppGpp in stimulating SasB (14), that pGpp is a significantly more potent inhibitor of purine salvage enzyme XPRT than ppGpp (33), and that ppGpp, but not pppGpp, inhibits the function of IF2 in stimulating subunit joining (34). Thus, these three closely related nucleotides have distinct biochemical and, as we show here, physiological activities.
Physiological implications of heterogeneity in protein synthesis
(p)ppGpp has long been thought to mediate entry into bacterial quiescence (35, 36). This transition facilitates survival in nutrient limited environments and its regulation depends upon the integration of a multitude of rapidly changing environmental signals that themselves may impair decision-making. One way bacteria deal with such uncertainty is to generate subpopulations, with distinct, often bimodal phenotypes from a population of genetically identical cells (23). Examples of phenotypic variation in B. subtilis include heterogeneity in specific metabolic activities such as acetate production (37) or in developmental transitions such as sporulation (38) and competence (39). The phenotypic variation in protein synthesis activity we observe here has potentially broad functional implications given its central role in cellular physiology. A global reduction in protein synthesis activity, if accompanied by a constant rate of protein degradation, would have the effect of reducing overall metabolic capacity, especially by affecting processes like ribosome assembly. Global effects also could have specific regulatory consequences. For example, the alternative sigma factor B. subtilis SigD drives expression of genes controlling daughter cell separation and motility that exhibit well characterized phenotypic variation. RelA affects both this variability as well as absolute levels of SigD (40), suggesting that differences in protein synthesis between cells contribute to SigD variability.
In summary, this work demonstrates that three differentially phosphorylated nucleotides and their respective synthases comprise a signaling network responsible for the heterogenous regulation of protein synthesis as B. subtilis cultures enter quiescence. We find that this heterogeneity is dependent on the RelA product pppGpp, which allosterically activates SasB, and the SasA product pGpp, which antagonizes this activation. Since the synthesis of pppGpp and pGpp reflects amino acid and peptidoglycan precursor availability, respectively, these parameters are thereby coupled to protein synthesis activity and facilitate cell decision making during the entry into quiescence.
Materials and Methods
Strains and media
Strains were derived from B. subtilis 168 trpC2. sasA (ywaC) and sasB (yjbM) gene knockouts were from transformed into B. subtilis 168 trpC2 using genomic DNA from BD5467 (41). The sasB transcriptional reporter strain was constructed similarly as described (20). Briefly, a 107 bp region encompassing the sasB operon promoter (PsasB) was amplified and inserted into AEC 127 using EcoRI and BamHI sites. The resulting AEC 127 PsasB was integrated into B. subtilis 168 trpC2 at sacA. sasBF42A and relAY308A strains were generated using integration of pMINIMAD2 derivatives (pMINIMAD2 sasBF42A and pMINIMAD2 relAY308A, respectively). Briefly, sasB was amplified excluding start and stop codons and F42A mutation was introduced using overlap extension PCR. sasBF42A was inserted into pMINIMAD2 vector using EcoRI and SalI sites. pMINIMAD2 sasBF42A vector was transformed into B. subtilis 168 trpC2 using a standard transformation protocol. Transformants were selected for erythromycin resistance at 45 °C overnight and grown for 8 hours at RT in LB. Cultures were diluted 1:10 in LB and grown overnight. Cultures were plated for single colonies and grown overnight at 45 °C. Single colonies were checked for erythromycin sensitivity and sensitive clones were checked for sasBF42A allele by Sanger sequencing of sasB amplified genomic region. The relAY308A strain was generated in a similar way but EcoRI and BamHI sites were used to insert the relAY308A gene into pMINIMAD2.
Growth curves
Growth curves were performed in a Tecan Infinite m200 plate reader at 37 °C with continuous shaking and OD600 measurements were made every five min. Cultures were grown from single colonies from fresh LB plates grown overnight at 37 °C. Exponential phase starter cultures (OD600∼ 0.5 - 1.5) were diluted to OD600 = 0.01 and grown in 96-well Nunclon Delta surface clear plates (Thermo Scientific) with 150 μL per well. All growth curves were done in triplicate and media-only wells were used to subtract background absorbance.
OPP labeling
OPP labeling of cells was as described (5). Exposure times were 30 msec for phase contrast, and 20 msec for mCherry. Fluorescence intensity of ∼1570 single cells per experiment was determined using ImageJ. Cells were binned based on fluorescence intensity using 50 a.u. wide bins in all experiments and number of cells in each bin presented as a histogram.
Protein expression and purification
Wildtype and F42A SasB proteins were expressed and purified essentially as described (14). Wildtype sasB was amplified from B. subtilis 168 trpC2. The F42A mutation was introduced using overlap extension PCR. WT and sasBF42A PCR products were inserted into pETPHOS expression vector using EcoRI and BamHI sites. pETPHOS WT sasB and pETPHOS sasBF42A were transformed into E.coli BL21 and proteins were induced with 1 mM IPTG for 2h at OD600 ∼0.5.
Cells were harvested at 4 °C and lysed using a Fastprep (MP biomedicals) in 50 mM Tris (pH 8.0), 250 mM NaCl, 5 mM MgCl2, 2 mM BME, 0.2 mM PMSF, and 10mM imidazole. Lysates were clarified and bound to a Ni-NTA column (Qiagen) for 1h. Columns were washed using 20 mM imidazole. Protein was eluted using 500 mM imidazole, dialyzed into 20mM Tris, 500 mM NaCl, 5mM MgCl2, 2 mM BME, and 10% glycerol and stored at -20 °C. NahA protein was purified in a similar way except that NahA was induced for 1h at 30 °C and NahA expressing cells (JDE3138) were lysed, washed, and eluted in 250 mM NaCl instead of 500 mM.
pGpp synthesis
pGpp was synthesized in vitro by purified NahA enzyme as described (18). Briefly, 10 nM purified B. subtilis NahA was incubated with 30 nM pppGpp (Trilink Biotechnologies) in 40 mM Tris-HCl (pH 7.5), 100 mM NaCl, 10 mM MgCl2 at 37 °C for 1 hour. Reactions were monitored for conversion of pppGpp to pGpp using thin layer chromatography on PEI-cellulose plates in 1.5 M KH2PO (pH 3.6). Nucleotides were visualized using short wave UV light. NahA enzyme was precipitated using ice cold acetone and nucleotides were stored at -20 °C.
SasB activity assays and TLC
SasB activity was assayed by measuring the amount of ppGpp generated similar to (5). Briefly, 0.8 μM purified B. subtilis WT or F42A SasB was incubated with 0.5 μCi of [γ-32P]-ATP (PerkinElmer) and 50 μM GDP in 20 mM Tris (pH 7.5), 500 mM NaCl, 5 mM MgCl2, 2mM BME. SasB was allosterically activated using 12.5 μM pppGpp (Trilink Biotechnologies) and pGpp was added as noted. Reactions were performed in a total volume of 10 μL, and each reaction was incubated at 37 °C for 1 min before being stopped using 5 μL of ice cold acetone. Conversion of ATP to ppGpp was visualized using thin layer chromatography on PEI-cellulose plates in 1.5 M KH2PO4 (pH 3.6). Plates were dried completely at RT and exposed for 5 min on a phosphor storage screen and visualized (GE Typhoon). ATP and ppGpp spot intensities were quantified using ImageJ.
Supporting information
Acknowledgements
SD was supported in part by the Columbia University Graduate Training Program in Microbiology, Immunology and Infection (R01 AI106711, Program Directors D. Fidock and L. Symington). JD was supported by NIH R01GM141953, R35GM141953, R21AI156397, and is a Burroughs-Welcome Investigator in the Pathogenesis of Infectious Disease.