RelZ-Mediated Stress Response in Mycobacterium smegmatis: pGpp Synthesis and Its Regulation

Bacteria utilize nucleotide messengers to survive the hostile environmental conditions and the onslaught of attacks within the host. The second messengers guanosine tetraphosphate and pentaphosphate [(p)ppGpp] have a profound impact on the long-term survival, biofilm formation, antibiotic tolerance, virulence, and pathogenesis of bacteria. Therefore, understanding the stress response mechanism regulated by (p)ppGpp is essential for discovering inhibitors of stress response and potential drug targets. Mycobacterium smegmatis contains two (p)ppGpp synthetases: Rel_Msm and RelZ. Our study unravels the novel regulatory mechanisms of RelZ activity and its role in mediating antibiotic tolerance. We further reveal its ability to synthesize novel second messenger pGpp, which may have regulatory roles in mycobacteria.

ABSTRACT

Stringent response is a conserved stress response mechanism in which bacteria employ the second messengers guanosine tetraphosphate and guanosine pentaphosphate [collectively termed (p)ppGpp] to reprogram their cellular processes under stress. In mycobacteria, these alarmones govern a multitude of cellular phenotypes, such as cell division, biofilm formation, antibiotic tolerance, and long-term survival. Mycobacterium smegmatis possesses the bifunctional Rel_Msm as a (p)ppGpp synthetase and hydrolase. In addition, it contains a short alarmone synthetase MS_RHII-RSD (renamed RelZ), which contains an RNase H domain in tandem with the (p)ppGpp synthetase domain. The physiological functions of Rel_Msm have been well documented, but there is no clear picture about the cellular functions of RelZ in M. smegmatis. RelZ has been implicated in R-loop induced stress response due to its unique domain architecture. In this study, we elucidate the differential substrate utilization pattern of RelZ compared to that of Rel_Msm. We unveil the ability of RelZ to use GMP as a substrate to synthesize pGpp, thereby expanding the repertoire of second messengers known in mycobacteria. We have demonstrated that the pGpp synthesis activity of RelZ is negatively regulated by RNA and pppGpp. Furthermore, we investigated its role in biofilm formation and antibiotic tolerance. Our findings highlight the complex role played by the RelZ in cellular physiology of M. smegmatis and sheds light upon its functions distinct from those of Rel_Msm.

IMPORTANCE Bacteria utilize nucleotide messengers to survive the hostile environmental conditions and the onslaught of attacks within the host. The second messengers guanosine tetraphosphate and pentaphosphate [(p)ppGpp] have a profound impact on the long-term survival, biofilm formation, antibiotic tolerance, virulence, and pathogenesis of bacteria. Therefore, understanding the stress response mechanism regulated by (p)ppGpp is essential for discovering inhibitors of stress response and potential drug targets. Mycobacterium smegmatis contains two (p)ppGpp synthetases: Rel_Msm and RelZ. Our study unravels the novel regulatory mechanisms of RelZ activity and its role in mediating antibiotic tolerance. We further reveal its ability to synthesize novel second messenger pGpp, which may have regulatory roles in mycobacteria.

INTRODUCTION

The survival of bacteria depends on their ability to sense and adapt to rapidly fluctuating environmental conditions. A key survival mechanism used by bacteria during hostile conditions is the stringent response, which is orchestrated by the nucleotide second messengers guanosine tetraphosphate and pentaphosphate [collectively named (p)ppGpp] (1). Bacteria trigger the stringent response upon encountering any environmental stress or starvation and reprogram their cellular machinery accordingly. (p)ppGpp regulates replication, transcription, and translation (2–4). In addition, (p)ppGpp influences bacterial physiology by binding to several cellular targets such as RNA polymerase and modulating pathways related to metabolism, cell division, and antibiotic resistance, among others. The mechanism and triggers for (p)ppGpp-mediated stress response vary across bacterial species (4). In Gram-positive bacteria, (p)ppGpp has a profound impact on GTP homeostasis and thereby stress survival (5).

The synthesis and hydrolysis of (p)ppGpp are catalyzed by the proteins of the RelA-SpoT homologue (RSH) family (4, 6). Most Gram-negative bacteria such as Escherichia coli possess two RSH proteins: RelA, which can only synthesize (p)ppGpp, and SpoT, which is a (p)ppGpp hydrolase and a weak synthetase (7). In Gram-positive bacteria, including Actinobacteria, the main enzyme responsible for (p)ppGpp turnover is the multidomain Rel, which can both synthesize and hydrolyze (p)ppGpp (8, 9). In addition to these classical RSHs, several single-domain monofunctional RSH enzymes called “short alarmone synthetases” (SASs) have been identified in several bacterial species, such as Bacillus subtilis, Staphylococcus aureus, and Enterococcus faecalis (4). Although the multidomain RSHs have been well studied, little is known about the physiological relevance of SASs. In addition to maintaining basal (p)ppGpp levels, these have been implicated in specific stress responses such as cell wall stress and alkaline stress (10, 11).

In mycobacteria, (p)ppGpp affects long-term survival, virulence, cell division, and antibiotic tolerance (8, 9, 12–14). M. smegmatis contains a bifunctional Rel_Msm for (p)ppGpp synthesis and hydrolysis. Deletion of Rel_Msm leads to changes in cellular morphology, growth, cell wall metabolism, antibiotic tolerance, and biofilm formation (14, 15). The genome of M. smegmatis also encodes a unique SAS, MS_RHII-RSD (here termed RelZ), which contains an RNase H domain (RHII) in tandem with the (p)ppGpp synthesis domain (RSD) (16). RelZ can make but not hydrolyze (p)ppGpp. Similarly, the pathogen Mycobacterium tuberculosis contains a bifunctional Rel (Rel_Mtb) and SAS homolog Rv1366 (17). Interestingly, Rv1366 lacks the RHII domain and is incapable of synthesizing (p)ppGpp in vitro (18). No discernible phenotype could be attributed to the protein, so the presence of SAS homologs such as RelZ and Rv1366 in mycobacterial species remains a mystery (19). In our previous work, we demonstrated the ability of RelZ to cleave RNA/DNA hybrids and hydrolyze triple-stranded structures called R-loops (16, 20). We further revealed the inability of the individual domains to function separately and the necessity of the hexameric oligomerization for its activity. However, it is unclear what other cellular phenotypes are regulated by RelZ.

Dysregulation in intracellular (p)ppGpp levels impacts important cellular pathways and is detrimental to bacteria (19). Therefore, it is imperative for M. smegmatis to regulate the synthesis activity of both RelZ and Rel_Msm during different growth conditions for maintaining (p)ppGpp homeostasis. The activity of Rel_Msm is regulated at the transcription level and through feedback regulation by (p)ppGpp (21–24). Recent studies have uncovered evidence for the transcriptional and allosteric regulation of SAS in organisms such as B. subtilis and E. faecalis (10, 25, 26). SASs and Rel may also have different substrate specificities (27). Therefore, SASs may have distinct physiological functions from that of Rel, and together their activities are coordinated to generate a cellular response. Our investigation was prompted with the aim to decipher the functions of RelZ in modulating the cellular physiology of M. smegmatis. We have constructed strains devoid of RelZ and Rel_Msm and characterized the role of RelZ in biofilm formation, cell surface-associated phenotypes, and antibiotic tolerance. We have elucidated the substrate utilization preference for RelZ in comparison to Rel_Msm and explored how the activity of RelZ is regulated. We reveal the ability of RelZ to synthesize pGpp using GMP as a substrate and discuss the potential implications of pGpp as a second messenger in mycobacteria.

RESULTS

RelZ but not Rel_Msm can synthesize pGpp.

It has been shown in our earlier work that RelZ can synthesize (p)ppGpp by catalyzing the transfer of a pyrophosphate moiety from the 5′-ribosyl of ATP to the 3′-ribosyl OH of GDP and GTP (16). When resolving the ppGpp synthesis reaction mixtures, we often observed the presence of an additional spot on our chromatograms with a similar migration profile as that of GTP. We hypothesized that GMP is present as a contaminant with GDP and being utilized as a substrate by RelZ. The ability of E. faecalis RelQ_Ef to consume GMP as the substrate further bolstered our hypothesis (27). To examine this possibility, we incubated RelZ and ATP with GMP and observed the appearance of this spot (Fig. 1A). Mass spectrometry revealed the mass of the compound to be 523 Da, which is the same as that of GTP (Fig. 1B). Therefore, there were three possibilities for the identity of this compound: 5′-pppG (GTP), 5′-ppGp, or 5′-pGpp. All of these compounds could be formed by the transfer of the radiolabeled phosphate from ATP to GMP. The migration of this compound differed to that of GTP when 0.75 M KH₂PO₄ is used as a solvent for thin-layer chromatography (TLC), indicating that the compound is not GTP (Fig. 1C) (27). To further distinguish if the compound is pGpp or ppGp, we treated the reaction mixture with NaOH since only pGpp is susceptible to alkali hydrolysis. The degradation of the compound in the presence of alkali validated the identity of the compound to be pGpp (Fig. 1D).

FIG 1.

Synthesis of pGpp by RelZ. (A) TLC of the reaction mixture containing 1 μM RelZ with 1 mM ATP and 1 mM GMP, GDP, or GTP. The reaction was stopped by adding formic acid, and the nucleotides were resolved on PEI-cellulose using 1.5 M KH₂PO₄ as a solvent. (B) Mass spectrometry analysis of the product of the RelZ reaction mixture containing ATP and GMP as the substrates. 522 Da and 544 Da represent the masses of the compound and sodiated compound, respectively. (C) Differential migration of pGpp and GTP. Pure radiolabeled GTP and pGpp were resolved using one-dimensional TLC in 0.75 M KH₂PO₄. (D) TLC of the reaction mixture treated with 0.5 M NaOH for 2 h at 37°C. The control reaction mixture (Ct) was incubated for 2 h in the absence of NaOH. (E) Time course of pGpp synthesis. The ATP spot density on the chromatogram was analyzed by densitometry and plotted for all time points by taking the ATP spot density at beginning to be 100%. Error bars indicate the standard deviations from three independent experiments. (F) TLC showing pGpp synthesis assay with 1 mM ATP, 1 mM GMP containing 1 μM RelZ, or Rel_Msm and sampled at different time intervals.

Subsequently, we probed the ability of Rel_Msm to synthesize pGpp. No significant pGpp formation was observed when Rel_Msm was allowed to react with ATP and GMP for an hour (Fig. 1E and F). During the same time, RelZ could utilize more than 75% of the ATP substrate provided. This suggested that RelZ, unlike Rel_Msm, is an efficient pGpp synthetase in M. smegmatis.

RelZ and Rel have different substrate specificity.

To understand the significance of this pGpp synthesis by RelZ, we further assessed the preference of RelZ for different guanine nucleotides as the substrates. Our previous work demonstrated that RelZ prefers GDP as a substrate over GTP, whereas Rel_Msm prefers GTP as a substrate over GDP (16). We evaluated the kinetic parameters of RelZ using GMP as a substrate and found that though the K_m of RelZ for GMP was similar to that of GTP and GDP, there was a marked difference in the k_cat value (k_cat [GMP] is 2.98 s⁻¹ and k_cat [GDP] is 1.78 s⁻¹) (Table 1). The k_cat/K_m value indicated that RelZ preferred GMP (2.63 mM⁻¹ s⁻¹) as a substrate over GDP (1.81 mM⁻¹ s⁻¹) and GTP (0.35 mM⁻¹ s⁻¹). In a reaction mixture containing equimolar concentrations of GTP, GDP, and GMP, it was evident that RelZ and Rel_Msm have contrasting substrate utilization patterns since pGpp and pppGpp synthesis was highest in the cases of RelZ and Rel_Msm, respectively (Fig. 2A). In conclusion, the substrate hierarchy for RelZ is GMP > GDP > GTP, and for Rel_Msm it is GTP > GDP.

TABLE 1.

Kinetic constants for (pp)pGpp synthesis by RelZ^a

Substrate	kcat (s−1)	Mean Km (mM) ± SEM	kcat/Km (mM−1 s−1)	Efficiency (relative to GMP)
GMP	2.98	1.13 ± 0.03	2.63	1
GDP	1.78	0.98 ± 0.03	1.81	0.69
GTP	0.34	0.96 ± 0.06	0.35	0.13

^aThe kinetic parameters for GMP as the substrate were estimated using reaction conditions described earlier (16). The kinetic constants for GTP and GDP used for comparison were taken from the same study.

FIG 2.

Substrate preference for Rel_Msm and RelZ. (A) TLC of in vitro synthesis reaction mixture containing 5 mM ATP with 1.5 mM (each) GTP, GDP, and GMP. The control reaction (Ct) lacked any enzyme. (B) In vitro pGpp synthesis by RelZ and its RFKD mutant. Representative TLC results of a reaction mixture containing 1 μM RelZ or RelZ-RFKD mutant with 1 mM ATP and 1 mM GTP, GDP, or GMP as the substrates are shown. The control reaction mixture (Ct) lacked an enzyme. (C) The percent activity of RelZ and its RFKD mutant with different substrates is plotted. The activity was estimated by densitometric analysis of the products resolved on TLC and imaged. The RelZ activity with GMP as the substrate was taken to be 100%, and the activities for other substrates with RelZ and the RelZ RFKD mutant were plotted relative to it. Error bars indicate standard errors from three independent experiments. The fold change is represented as “x” times the value of the synthesis activity of the RFKD mutant with respect to RelZ. (D) Sequence alignment of Rel_Msm, E. coli RelA (RelAEc), RelZ, and E. faecalis RelQ_Ef (RelQEf) with the RXKD/EXDD motif marked.

Mutation in DDRD site in RelZ alters its substrate utilization.

We next investigated the underlying reason behind this differential substrate specificity. Differential preference for GTP and GDP as the substrate in RSHs has been attributed to a charged motif that affects interaction with Mg²⁺ ion (28, 29). Monofunctional Rel such as that in E. coli (RelA_Ec) has an acidic EXDD motif that leads to a preference for GDP in comparison to GTP and insensitivity to an increase in Mg²⁺ concentration. On the other hand, bifunctional Rel, such as Rel_Mtb, contains a basic RXKD motif, prefers GTP, and has a lower activity at high Mg²⁺ concentration. Rel_Msm contains an RFKD motif, whereas the RelZ contains an acidic DDRD motif. To elucidate whether the difference in substrate specificity is indeed due to the charged motif, we generated a mutant of RelZ with RFKD as the motif instead of DDRD and analyzed its substrate specificity (Fig. 2B). In comparison to RelZ, the RFKD mutant showed enhanced (p)ppGpp synthesis. This mutant showed a marked increase in its ability to synthesize ppGpp (2.6-fold higher) and pppGpp (1.5-fold higher) (Fig. 2C). The pGpp synthesis by RelZ RKFD mutant was not significantly affected. However, we did not see a reversal of substrate specificity to that of Rel_Msm. When a similar mutation was made in RelA_Ec, the mutant had increased (p)ppGpp synthesis activity and also had a substrate preference opposite to that of RelA_Ec (29). RelZ possesses an acidic motif, but it contains a basic residue at the third site similar to Rel_Msm (RFKD) and RelQ_Ef (EERD) (Fig. 2D). Interestingly, mutating the RXKD motif in Rel_Mtb to EXDD leads to an ability to synthesize pGpp since the mutant can hydrolyze the P_α-O-P_β bond of GTP/GDP (28). This is unlike the case in RelZ, which can directly use GMP as a substrate. The altered activity of the RFKD mutant is not due to a change in the oligomerization and secondary structure as SEC-MALS and circular dichroism analysis did not reveal any deviations from RelZ (data not shown). Together, these results hint at either a difference in catalytic mechanism or the side chain interactions for RelZ.

Rel_Msm can hydrolyze pGpp.

For pGpp-mediated signaling to occur in M. smegmatis, there must be a protein capable of hydrolyzing it once the stress is removed. We assessed the ability of both Rel_Msm and RelZ to hydrolyze pGpp. Rel_Msm showed a remarkable efficiency for pGpp hydrolysis as seen by the almost complete hydrolysis within 15 min (Fig. 3). Rel_Msm hydrolyzes pGpp to GMP and pyrophosphate, as evidenced by the comigration of the radiolabeled product with purified pyrophosphate. In contrast to Rel_Msm, <10% of pGpp was hydrolyzed by RelZ to yield a phosphate moiety, which could be considered a weak hydrolysis. The genome of M. smegmatis also encodes other nucleotidyl phosphohydrolases, but their ability to degrade (pp)pGpp is not yet reported.

FIG 3.

In vitro pGpp hydrolysis assay for Rel_Msm and RelZ. (A) Representative TLC results depicting hydrolysis of pGpp by Rel_Msm and RelZ. First, 1 mM purified pGpp was incubated with 1 μM Rel_Msm or RelZ. The reaction mix was sampled at different time points and resolved on PEI-cellulose using 1.5 M KH₂PO₄. The control reaction mix (Ct) lacked an enzyme and was incubated for 60 min. Pure pyrophosphate (C2) was used as a standard. Rel_Msm shows potent pGpp hydrolase activity, whereas RelZ is a weak hydrolase. (B) The time course of pGpp hydrolysis by Rel_Msm and RelZ is plotted. The radiogram was analyzed by densitometry and plotted by considering the pGpp density at the starting time to be 100%. Error bars indicate standard deviations from three independent experiments.

Single-stranded RNA inhibits pGpp synthesis by RelZ.

RelZ is a bifunctional protein, so the next question we addressed was how the RNase H and (pp)pGpp synthesis activities are regulated. Mutation in the active site of one domain of RelZ does not inactivate the other domain and only the full-length protein is active (20). We first explored whether a regulatory mechanism for RelZ exists. The RHII domain of RelZ hydrolyzes R-loops to release the double-stranded DNA (dsDNA) and single-stranded RNA, so we speculated that the presence of the R-loops and their components modulates the pGpp synthesis activity. We performed the pGpp synthesis assay for RelZ in the presence of R-loops but did not observe any difference in the activity (Fig. 4A and B). Interestingly, we discovered that the presence of 4 μM RNA inhibited >40% of pGpp synthesis. Addition of the same concentration of dsDNA did not affect the pGpp synthesis activity, proving that the inhibition is RNA specific. Single-stranded DNA (ssDNA) did not inhibit RelZ to the same extent as RNA. The reason for the inhibitory effect in the presence of RNA but not in the presence of R-loops is unclear. The E. faecalis RelQ_Ef has a regulatory mechanism in which the (p)ppGpp synthesis activity is inhibited by the presence of single-stranded RNA, but the biological significance remains unknown (26). The evolution of an RHII domain in association with an RSD domain in RelZ hints at a direct link to a potential source of regulatory RNA.

FIG 4.

Regulation of pGpp synthesis and RNase H activity of RelZ. (A) Representative TLC depicting pGpp synthesis by RelZ in the presence of R-loop, dsDNA, ssDNA, and RNA. RelZ was incubated with 1 mM ATP and 1 mM GMP in the presence of 2 and 4 μM R-loops, dsDNA, ssDNA, or RNA. The control reaction (Ct) lacked an enzyme. A RelZ reaction (R) carried out in the absence of the inhibitor served as a positive control. (B) Bar graph illustrating the inhibition of RelZ activity by RNA. The pGpp formation was assessed by densitometry analysis of pGpp spot intensity on the TLC and is plotted as the relative percentage of pGpp synthesis activity in a reaction mixture containing the inhibitor against that in the absence of it. Error bars indicate standard deviations from three independent experiments. (C) In vitro fluorescence assay for RNase H activity. A 12-mer RNA/DNA hybrid consisting of a fluorescein-labeled poly(rA) annealed to a dabsyl-labeled poly(dT) was used as a substrate (AH). The presence of dabsyl leads to quenching of the fluorescence in AH. The RNase H activity of RelZ was assayed by incubating 500 nM AH with 75 nM RelZ in presence or absence of pGpp. The poly(rA) served as the positive control. The fluorescence intensity at 520 nm is plotted, and error bars indicate the standard deviations from three independent experiments. ***, P < 0.001.

pGpp has a moderate effect on the RNase H activity of RelZ.

We subsequently examined whether the RNase H activity of RelZ is affected by the product of the RSD activity, pGpp. We used a poly(rA)⋅poly(dT) hybrid (AH) in which dabsyl quenches the fluorescence of the fluorescein. RelZ cleaves AH, leading to an increase in fluorescence, as a measure of RNase H activity. In the presence of 4 μM pGpp, we observed a minor decrease in fluorescence intensity, indicating a reduction in RNase H activity (Fig. 4C). This suggests that intracellular pGpp could modulate the RNase H activity of RelZ, but the relevance of this mode of regulation is unknown.

A high pppGpp concentration affects the pGpp synthesis activity of RelZ.

Once the environmental conditions become favorable for growth, the cells recover by inhibiting the stringent response. For this recovery, (pp)pGpp synthesis by RelZ has to be halted. We sought to determine whether RelZ possesses a feedback inhibition mechanism in which (pp)pGpp downregulates its own synthesis. Therefore, we carried out pGpp synthesis assay for RelZ in the presence of 1 mM pGpp, ppGpp, or pppGpp (Fig. 5). Our data revealed that pppGpp inhibited the pGpp synthesis activity of RelZ to <80%. However, ppGpp and pGpp did not affect the activity significantly. This suggests that a high concentration of pppGpp is a prerequisite for inhibiting RelZ-mediated synthesis.

FIG 5.

Regulation of RelZ by (pp)pGpp. (A) TLC showing in vitro pGpp synthesis by RelZ and in vitro pppGpp synthesis by Rel_Msm separately. RelZ and Rel_Msm were incubated with 1 mM ATP and 1 mM GMP or GTP, respectively, in the presence of 1 mM pGpp, ppGpp, or pppGpp. The control reaction (Ct) lacked an enzyme. (B) The pGpp synthesis activity of RelZ and pppGpp synthesis activity of Rel_Msm was plotted using densitometry analysis of the pGpp/pppGpp spot intensity on the TLC plate. The pGpp/pppGpp intensity in a reaction devoid of the nucleotide inhibitor was taken to be 100%. Error bars indicate the standard deviations from three independent experiments.

Previously, it has been shown that Rel_Msm has an autoregulatory feedback mechanism (23, 24). We explored the possibility that pGpp could modulate the activity of Rel_Msm. We assessed the pppGpp synthesis assay for Rel_Msm in the presence of 1 mM pGpp, ppGpp, or pppGpp and observed the expected inhibition of the pppGpp synthesis by (p)ppGpp. Furthermore, we observed that pGpp caused a significant inhibition (almost 40%) of Rel_Msm activity, although it was not as potent as pppGpp. Thus, Rel_Msm is a potential cellular target of pGpp.

RelZ affects the antibiotic tolerance of M. smegmatis.

The evidence presented so far demonstrates that RelZ is an efficient pGpp synthetase, which raises an important question as to its regulatory roles in vivo. To gain a deeper insight into how the stress response is modulated by the two RSH proteins with differing substrate specificities and to understand the (pp)pGpp⁰ phenotype of M. smegmatis, we generated strains devoid of RelZ (ΔrelZ). To our surprise, we found that the strain with deletions of Rel_Msm and RelZ (Δrel ΔrelZ) has residual (p)ppGpp, indicating the presence of another synthetase (see Fig. S1 in the supplemental material).

Since antibiotic tolerance is one of the most important phenotypes linked with (p)ppGpp, we analyzed the effect of RelZ deletion on antibiotic susceptibility in M. smegmatis. In contrast to our expectations, we had previously observed the Δrel strain to have a higher tolerance to most antibiotics (15). To examine the role of RelZ in antibiotic tolerance, we carried out phenotype microarray analysis with the strains to test their sensitivity against 240 different antibiotics/chemicals. The growth of the strains is inferred from the dye reduction values plotted as a function of time, and the phenotype microarray (PM) software compares the area under the curve for each strain. Figure 6 depicts the growth of the wild-type, knockout, and complementation strains in the presence of the antimicrobials bleomycin, ofloxacin, and rifampin. Table S3 provides a list of antimicrobials in which the area under the curve (AUC) difference between the strains was >4,000. In contrast to the Δrel strain, the ΔrelZ strain displayed sensitivity to most antibiotics. Several of these chemicals were genotoxic in nature (e.g., ofloxacin and bleomycin). Moreover, a complementation strain (ΔrelZ+pRelZ) containing a copy of the relZ gene under the control of a constitutive promoter led to the reversal of this phenotype, confirming that the antibiotic sensitivity is indeed due to relZ deletion and not due to polar effects. MIC analysis of the strains for the antibiotics using a resazurin microtiter plate assay (REMA), as well as CFU analysis, reflected a similar trend (see Table S4 and Fig. S3 in the supplemental material). The Δrel ΔrelZ strain showed a similar tolerance level or decreased tolerance compared to the Δrel strain. The Δrel+pRelZ strain, in which RelZ is expressed in the Δrel strain, showed reduced survival compared to the Δrel strain though the susceptibility was not identical to the wild-type strain in most cases. This indicated that the (pp)pGpp synthesis by RelZ could reverse the Δrel phenotype to some extent. The antibiotics and chemicals in which RelZ deletion affected growth belong to multiple classes with differing mechanisms of action. It could be hypothesized that the antibiotic influx/efflux is affected in these strains due to the altered (pp)pGpp homeostasis.

FIG 6.

Phenotype microarray analysis in M. smegmatis. The strains were grown in 96-well plates in the presence of different antibiotics with tetrazolium violet as an indicator for growth. The change in the color of the dye due to respiration was measured, and the absorbance was plotted as a function of time. Representative growth profiles of strains in the presence of three antibiotics on PM plates are shown. The y axis represents the dye reduction values generated due to bacterial respiration, and the x axis represents the incubation time of the PM plates. The graphs were plotted using the ggplot2 of the R statistical package.

The Δrel ΔrelZ strain is defective in biofilm formation and has altered cell surface properties.

Biofilm formation is a major stress-associated phenotype governed by (p)ppGpp. To understand the effect of RelZ deletion, we assessed the biofilm formation of all the strains. The Δrel strain is defective in biofilm formation (15) and the Δrel ΔrelZ strain mirrored the phenotype (Fig. 7A). The ΔrelZ strain did not show robust biofilm formation, unlike the wild-type strain, but the effects were not as drastic as that of the Δrel strain. The biofilm did not have a thick and wrinkled appearance like that of the wild-type strain. This suggests that RelZ does not impact biofilm formation significantly and that the biofilm formation is more substantially affected by Rel_Msm. Biofilm quantification using the crystal violet method estimated a significant reduction (almost 50%) in the biofilm formation ability of the Δrel ΔrelZ strain, thus supporting our conclusion that it is impaired in biofilm formation (Fig. 7B).

FIG 7.

Biofilm formation and aggregation in M. smegmatis. (A) Biofilm formation. The cultures were allowed to form biofilms in Sauton’s medium for 5 days and then imaged. (B) Biofilm quantification by the crystal violet method. The absorbance at 570 nm has been plotted as a bar graph with the standard deviations for three sets of experiments. (C) Aggregation percentages of the wild-type, knockout, and complementation strains. The percentage of aggregated cells in the stationary-phase cultures has been plotted as a bar graph with the standard deviations for three sets of experiments. The graph was plotted using GraphPad Prism 5. ***, P < 0.001; **, P < 0.01; *, P < 0.05.

The deletion of Rel_Msm leads to altered surface phenotypes with respect to colony morphology, sliding motility, aggregation, and cell wall hydrophobicity (15). We examined these cell surface properties in cells devoid of RelZ. We noticed that the stationary-phase cultures of the ΔrelZ strain often settle faster than the wild-type cultures. Moreover, we found that the promoter of RelZ is constitutive in nature, but its activity is elevated during stationary phase, further emphasizing the significance of RelZ under normal growth conditions, as well as during starvation (Fig. S2). The Δrel ΔrelZ strain cultures have a high tendency to form clumps and settle down quickly. We quantified this by calculating the percentage of aggregated cells in stationary-phase cultures of these strains (Fig. 7C). As expected, the ΔrelZ (>40% aggregation) and Δrel ΔrelZ (>80% aggregation) strains showed significantly higher aggregation compared to the wild type. RelZ complementation could rescue the phenotype in the ΔrelZ+pRelZ strain validating that the change in phenotype is due to the deletion of relZ. Both Δrel and Δrel ΔrelZ strains aggregate more than the ΔrelZ strains and complementation of RelZ in a Δrel strain did not recover the phenotype to that of the wild-type strain. This suggested that the deletion of Rel_Msm has a more profound effect on these phenotypes. A similar trend was reflected in the case of Congo red binding assay (used as a measure of cell wall hydrophobicity) and sliding motility (30) (Fig. S4).

Taken together, our findings indicate that RelZ has a role in cell wall metabolism, as seen by the altered cell surface properties of the ΔrelZ strain. The Δrel ΔrelZ strain is similar to the Δrel strain in the macroscopic properties. It has increased cell wall hydrophobicity, which is most likely due to the change in the lipid composition of the cell wall. We can conclude that lowering the alarmone levels severely affects the cell envelope in M. smegmatis.

DISCUSSION

In our study, we have attempted to decode the role of the RelZ in M. smegmatis stress response. In contrast to the SAS known from other bacteria, RelZ also contains an RNase H domain, which led us to decipher its role in R-loop induced stress response (16, 20). In our current work, we reveal yet another function of RelZ: the synthesis of pGpp. Importantly, GMP is the most preferred substrate for RelZ. On the other hand, Rel_Msm prefers GTP as the substrate over GDP (16, 28). To our knowledge, this is the first reported mycobacterial SAS with the ability to synthesize pGpp, and it raises an interesting question about its significance in vivo.

Recently, the SAS from E. faecalis and Corynebacterium glutamicum were shown to be capable of synthesizing pGpp (27, 31). The authors of those studies speculated the involvement of pGpp in prolonging the stringent response after the depletion of the cellular pools of GTP and GDP. The triggering of stringent response involves utilization of GTP and GDP for synthesis of (p)ppGpp, which further inhibit purine biosynthesis (5). We hypothesize that an SAS, like RelZ, capable of utilizing GMP would be beneficial to M. smegmatis under such circumstances. The cellular concentration of GTP is known to be much higher than GMP in bacteria, so it remains a question whether the intracellular levels of GMP increase to such levels that pGpp synthesis by RelZ is likely. Previous studies on this subject lend credence to our hypothesis that pGpp synthesis by RelZ could have biological relevance under specific conditions. A substantial increase in GMP levels during amino acid deprivation has been documented in B. subtilis (32). In addition, amino acid-starved cultures of B. subtilis have been previously shown to contain pGpp, albeit at a much lower concentration than (p)ppGpp (33). We have been unsuccessful in our attempts to detect the presence of pGpp in nucleotide extracts of M. smegmatis using two-dimensional TLC, high-pressure liquid chromatography (HPLC), and mass spectrometry due to the identical masses and migration patterns of pGpp and GTP. The presence of pGpp-binding proteins in E. faecalis (Gmk and Hpt) and in M. smegmatis (Rel_Msm) hints at putative regulatory roles for pGpp as a second messenger (27).

Perturbation of purine nucleotide levels by (p)ppGpp affects bacterial fitness and antibiotic sensitivity (34). Tipping the balance of (p)ppGpp toward synthesis or hydrolysis affects mycobacteria, and the importance of this homeostasis is highlighted by the effect of a hydrolysis-defective Rel mutant on the growth and pathogenesis of M. tuberculosis (19). Our data indicate that the pGpp synthesis by RelZ is elegantly modulated via two mechanisms: RNA and pppGpp. The inhibition of the SAS RelQ_Ef by RNA is sequence specific and counteracted by (p)ppGpp (26). RelZ is involved in hydrolyzing RNA/DNA hybrids and R-loops, so once they have been degraded by the RHII domain, the alarmone synthesis would no longer be required. Hence, it could be envisaged that the hydrolyzed RNA would inhibit the pGpp synthesis by RelZ. However, direct evidence for this conjecture is lacking. Alternatively, the RNA arises from the cellular pool of mRNA, and any sequence specificity could imply potential cellular targets. More structure-function insights are needed to understand the catalytic and regulatory mechanism of the SAS RelZ.

We speculate that the inhibition of RelZ by pppGpp could help modulate the overall alarmone levels within the cell, and it could be beneficial when the cells no longer require any alarmone synthesis.

To elucidate the (p)ppGpp⁰ phenotype in M. smegmatis, we characterized the strain devoid of both its alarmone synthetases. We found that the Δrel ΔrelZ strain is capable of synthesizing ppGpp indicative of the presence of another (p)ppGpp synthetase. A recent study by Njire et al. demonstrated the weak (p)ppGpp synthesis activity of M. tuberculosis Rv2783 encoding a PNPase (35). We suspect that the M. smegmatis PNPase homolog (MSMEG_2656) could have a weak (p)ppGpp synthesis activity, but the presence of any other protein with such an activity cannot be ruled out.

Decreased (p)ppGpp levels can lead to decreased fitness, especially under nutrient-limiting conditions, and have an effect on GTP homeostasis (4, 5, 7, 36–39). M. smegmatis Δrel cells are longer, multiseptate, and multinucleate and show an altered lipid profile and defective biofilm formation (14, 15). Unlike the case with Rel_Msm, the deletion of RelZ increases sensitivity to many antibiotics. The reason for the antibiotic sensitivity is difficult to comprehend since it could either be a pleiotropic effect of (p)ppGpp or due to the RNase H activity, and it is difficult to delineate the role of each domain. It is tempting to speculate that the RNase H activity of RelZ affects the tolerance of DNA-damaging agents. The role of SASs in antibiotic resistance tends to be species specific. In S. aureus, the SASs RelP and RelQ are upregulated during cell wall stress and provide tolerance against cell wall active antibiotics (11). In E. faecalis, the order of vancomycin sensitivity for the various strains was found to be as follows: ΔrelA ΔrelQ_Ef strain < ΔrelQ_Ef strain < wild type < ΔrelA strain (40). Higher basal levels correlated with higher antibiotic tolerance in E. faecalis. The constitutive (pp)pGpp synthesis by RelZ is therefore significant in this scenario. The deletion of Rel_Msm leads to elevated levels of polar lipids, making the cell wall more hydrophobic and hindering the uptake of several antibiotics (15). The important consequence of lowering (p)ppGpp levels is reflected in the profound alterations in the cell surface properties of the Δrel ΔrelZ strain. The complementation by RelZ rescued the phenotypes of Δrel strain to a certain extent, thus underscoring the importance of alarmone synthesis by this SAS despite the prominent role played by Rel_Msm. Deletion of Rel_Msm also causes an upregulation of several genes linked with antibiotic resistance (14). We argue that the Δrel ΔrelZ strain also would have an altered transcriptome that impacts this phenotype. We believe that the (pp)pGpp levels are fine-tuned during the different growth conditions in M. smegmatis, and the SAS RelZ plays a significant role in this regard in association with Rel_Msm. Knowledge of the (pp)pGpp levels is essential to conceive a clear picture about the significance of (pp)pGpp on antimicrobial susceptibility.

Realization of the importance of SASs has led to exciting discoveries, and we expect there will be more interesting findings related to their cellular functions and downstream effectors. More extensive research is needed in the future with the hope of developing inhibitors of stress response in bacteria.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

E. coli strains DH5α and BL21(DE3) were grown in Luria-Bertani broth at 37°C with shaking at 180 rpm. The cultures were supplemented with ampicillin at 100 μg ml⁻¹. M. smegmatis cultures were grown in Middlebrook 7H9 broth (MB7H9; Difco) containing 2% glucose and 0.05% Tween 80 at 37°C. Hygromycin and apramycin were used at a concentration of 50 μg ml⁻¹. The list of strains and plasmids is in Table S1 in the supplemental material.

Plasmid pRelZ was generated by cloning the relZ gene in pMV261 vector (containing constitutive promoter) using the primers listed in Table S2 in the supplemental material. Plasmid pRelZ was transformed into the M. smegmatis ΔrelZ and Δrel strains to generate the ΔrelZ+pRelZ and Δrel+pRelZ strains, respectively. The clones were confirmed by sequencing and the RelZ expression in the strains were analyzed by SDS-PAGE. The cultures were grown in cultures supplemented with kanamycin at 25 μg ml⁻¹.

Site-directed mutagenesis.

The RFKD mutant of RelZ was generated using pET-MS_RHII-RSD as the template. The primers incorporating the desired mutations are listed in Table S2 in the supplemental material. The products were phosphorylated, ligated, and transformed into E. coli DH5α, and the mutation was confirmed by sequencing.

The sequences of the proteins were retrieved from UniProt, and the multiple sequence alignment was done using Clustal Omega with default parameters.

Protein expression and purification.

Rel_Msm, RelZ, and its RFKD mutant were expressed from pET 21b and purified using nickel affinity chromatography as described earlier (16). Briefly, the cultures were induced at an optical density at 600 nm (OD₆₀₀) of 0.6 with 1 mM IPTG (isopropyl-β-d-thiogalactopyranoside) and grown for 3 h at 37˚C. The cells were harvested and lysed by sonication in a lysis buffer (50 mM Tris [pH 7.9] at 4°C, 500 mM NaCl, 1 mM phenylmethylsulfonyl fluoride). The lysate was centrifuged, and the supernatant was loaded onto a Ni-nitrilotriacetic acid (Ni-NTA)–agarose column preequilibrated with the same buffer. The protein was eluted using 500 mM imidazole and dialyzed in buffer containing 25 mM Tris (pH 7.9) at 4°C, 125 mM NaCl, 25 mM imidazole, and 5% glycerol.

In vitro (pp)pGpp synthesis assay.

The (pp)pGpp synthesis activity of the proteins was assayed as described earlier using 10-μl reaction mixtures containing 1 mM ATP, 10 μCi of [γ-³²P]ATP ml⁻¹ (3,000 Ci mmol⁻¹; BRIT, India), 1 mM GMP/GDP/GTP, 50 mM HEPES (pH 7.5), 100 mM NaCl, 1 mM dithiothreitol (DTT), and 1 mM MgCl₂ in the presence of 1 μM protein (16). The reaction mixture was kept at 37°C for 30 min, and the reaction was stopped by adding 1 N formic acid. Then, 2 μl of the reaction mixture was spotted onto a PEI-cellulose sheet (Merck) and resolved by one-dimensional TLC in 1.5 M KH₂PO₄ (pH 3.4). The nucleotides were visualized by phosphorimaging (Molecular Imager; Bio-Rad) and quantified by densitometric analysis using ImageJ software.

Kinetic constants were determined for RelZ with GMP as the substrate. We used 10 mM ATP and 1 μM enzyme in all reactions. The GMP concentration was varied from 10 μM to 5 mM, and the Mg²⁺ concentration was kept at 5 mM. The reaction was analyzed by TLC as mentioned above.

To analyze the substrate preference for Rel_Msm and RelZ, the reaction mixture was maintained at an equimolar concentration (1.5 mM) of GMP, GDP, and GTP, along with 5 mM ATP, 5 mM Mg²⁺, and 1 μM protein and analyzed as described above.

Alkali hydrolysis.

Alkali hydrolysis of pGpp was carried out as described previously (27, 41). Briefly, 0.5 M NaOH and 10 mM MgCl₂ was added to the pGpp reaction mixture, followed by incubation at 37°C for 2 h. Formic acid was added to a final concentration of 1 N, and the reaction products were analyzed by TLC as described above. The control mixture lacked protein.

Preparation of (pp)pGpp.

To make pure pGpp, 2 μM RelZ was incubated with 1 mM ATP and 2 mM GMP in the same reaction buffer used to assay pGpp synthesis. Phenol-chloroform extraction was performed to remove the protein. The sample was diluted 1:3 in 25 mM Tris (pH 7.9) and subjected to anion-exchange chromatography (MonoQ 10/100 GL; GE Healthcare). The nucleotides were resolved by using a 0 to 1 M LiCl gradient. The peak fraction was collected and precipitated by the addition of 1 M LiCl and 4 volumes of ethanol. After incubation on ice for 30 min, the fraction was centrifuged at 5,000 × g for 20 min at 4°C. The pellets were washed twice with absolute ethanol, air dried, and stored at –20°C. The quality was controlled by HPLC and electron spray ionization-mass spectrometry (ESI-MS).

³²P-labeled pGpp (³²P-pGpp) was purified as described above, but the reaction mixture contained radiolabeled ATP.

ppGpp and pppGpp were purchased from Jena Biosciences, Germany.

pGpp hydrolysis assay.

The reaction mixture contained 1 mM ³²P-pGpp, 50 mM HEPES (pH 7.5), 100 mM NaCl, 1 mM DTT, 2 mM MnCl₂, and 1 μM protein. The reaction mixture was kept at 37°C, and aliquots were spotted onto TLC plates at specified time intervals. The reaction products were visualized and quantified as described above. Pure pyrophosphate was cored out from TLC of a Rel_Msm pppGpp hydrolysis reaction and extracted in 10 mM Tris-HCl.

Activity inhibition assays.

R-loops were synthesized as described earlier by annealing the oligonucleotides D1, D2, and R3 mentioned in Table S2 (20). In vitro pGpp synthesis reaction mixture containing RelZ was incubated with either 2 or 4 μM single-stranded DNA (80-nucleotide [nt] D1), single-stranded RNA (20-nt R3), double-stranded DNA (80-bp D1.D2), or R-loop. The reaction mixture was kept at 37°C for 30 min, and then the reaction was stopped, visualized, and quantified as described earlier.

To analyze the effect of (pp)pGpp on RelZ pGpp synthesis, the pGpp synthesis reaction mixture was incubated with either 1 mM pGpp, ppGpp, or pppGpp at 37°C for 30 min and analyzed as described above. For Rel_Msm, the inhibition was tested by using GTP as a substrate instead of GMP. All assays were performed with multiple replicates.

In vitro RNase H fluorescence assay.

An RNA/DNA hybrid (AH) was synthesized by annealing the 12-mers 5′-fluorescein-poly(rA) and 3′-dabsyl-poly(dT) as described earlier (16). The hydrolysis reaction mixture consisted of 75 nM RelZ and 500 nM AH in a reaction buffer (50 mM Tris-Cl [pH 7.5], 50 mM KCl, and 5 mM MnCl₂) and was kept at 37°C for 15 min. A control mixture lacking protein was used as the negative control. A concentration of 4 μM pGpp was added to assess its effect on RNase H activity. Fluorescence was recorded using an excitation wavelength of 490 nm. Then, 4 μM pGpp was added to assess its effect on RelZ activity.

M. smegmatis strain construction.

The Δrel ΔrelZ strain was generated using phage-mediated specialized transduction method as described earlier (20, 42). The allele exchange substrate plasmid (phAE159-ms_rhII-rsd) used to generate the ΔrelZ strain previously was transduced into the M. smegmatis Δrel strain and screened for apramycin resistance (Table S1). The deletion of the gene was confirmed by sequencing.

Biofilm formation and quantification.

Stationary-phase cultures grown in MB7H9 medium were harvested, washed, and suspended in Sauton’s medium containing 2% glucose. For biofilm formation, 2 ml of Sauton’s medium was poured into a 24-well plate, and each well was inoculated with 2 μl of the culture. Each sample was sealed, kept in a humidified incubator at 30°C for 5 days, and imaged.

Biofilm quantification was done using the crystal violet method described by Gupta et al. (15) and O’Toole (43). The stationary-phase cultures were resuspended in Sauton’s medium (containing 2% glucose) to a final OD₆₀₀ of 0.05. Subsequently, 200 μl of the culture was poured per well of the 96-well plates, sealed, and kept in a humidified incubator at 30°C for 7 days. Subsequently, the cells were removed from the wells, and the wells were washed twice with water. The adherent biofilms attached to the wells were stained with 0.1% crystal violet for 45 min. The residual dye was removed, and the wells were washed again with water and allowed to dry. The bound dye was solubilized in 200 μl of 80% ethanol, and the absorbance at 570 nm was recorded using a microtiter plate reader (SpectraMax 340PC³⁸⁴; Molecular Devices). The experiment was repeated three times with seven technical replicates, and the resulting graph was analyzed using GraphPad software.

Phenotype microarray analysis.

PM analysis was performed using the protocol of Gupta et al. (15). In brief, transmittance of the cultures was adjusted to 81%, and tetrazolium violet was added to a final concentration of 0.01%. Portions (100 μl) of these cultures were inoculated into the wells of PM plates 11 to 20 coated with antibiotics/chemicals and then kept in a Biolog incubator at 37°C for 60 h. The dye reduction values were converted to areas under the curve (AUCs) by using the parametric software module of the Biolog Omnilog system. The AUCs for the strains were compared and overlaid as a test strain versus reference strain AUC. A cutoff of 4,000 in the AUCs was used to estimate the number of antibiotics in which the growth difference was significant. Dye reduction curve values were plotted using ggplot2 in R statistical software (http://cran.r-project.org).

REMA.

MIC values were estimated using REMA adapted from an earlier protocol (44). In brief, the transmittance of the cultures was adjusted to a McFarland turbidity standard of 1 and then diluted 1:10. Next, 100-μl portions of the diluted culture were inoculated into 96-well microtiter plates containing a 2-fold dilution series of the antibiotics rifampin, ofloxacin, and bleomycin. The plates were sealed and incubated in a humidified incubator at 37°C. After 36 h, 30 μl of 0.01% resazurin was added to the wells of the microtiter plates, and the plates were further incubated for 4 h. The color of the resazurin dye changes from blue to pink due to bacterial growth. The MIC is defined as the minimum antibiotic concentration at which the resazurin dye did not change color.

Analysis of cell surface properties.

M. smegmatis wild-type, knockout, and complementation strains were grown until reaching the stationary phase in MB7H9 medium supplemented with 2% glucose and 0.05% Tween 80. The protocol for the aggregation assay was adapted from that of Deshayes et al. (45). The aggregates were removed by centrifuging the cells at 10 × g for 1 min. The OD₆₀₀ values of the supernatants were recorded. The OD₆₀₀ values of the cultures in which the aggregates were broken by vortexing with glass beads were measured. The percent aggregation was calculated by using the OD₆₀₀ of the supernatant and the OD₆₀₀ of the vortexed culture.

The motility assay was adapted from Martinez et al. (46). MB7H9 plates were solidified with 0.3% agarose and inoculated in their centers with 5 μl of culture (OD₆₀₀ normalized to 1). The plates were incubated at 37°C for 4 days, and the motility was evaluated by measuring the diameter of the halo of growth formed by the cells.

The Congo red accumulation assay was adapted as follows (30). The cultures were cultivated to stationary phase in MB7H9 broth containing Congo red at 100 μg ml⁻¹ and 0.05% Tween 80. The cells were then washed extensively with water until the supernatants were colorless. The Congo red dye associated with the cells was extracted with 400 μl of acetone for 1 h with gentle shaking. The absorbance of the extract was measured at 488 nm. The Congo red binding index was defined as the A₄₈₈ of the acetone extracts divided by the OD₆₀₀ of the culture. All assays were performed in replicates and repeated at least three times.

The statistical analysis was performed using unpaired t tests (two-tailed) and plotted by using GraphPad Prism 5.0 software.

Measurement of intracellular nucleotides by TLC.

M. smegmatis cultures were grown to mid-log phase in MB7H9 media containing glucose. Secondary cultures were grown in morpholinepropanesulfonic acid medium supplemented with 2% and 0.02% glucose, 0.08 mg/ml Casamino Acids, and 0.05% Tween 80 as described earlier (16). The cultures were labeled at an OD₆₀₀ of 0.3 by adding ³²P-o-phosphoric acid (specific activity, >3,000 mCi/mmol; BRIT, Hyderabad, India) to a final concentration of 100 μCi/ml. The cells were harvested after 36 h of growth, washed, and lysed using formic acid. The nucleotide extracts, normalized to an OD₂₆₀ of 2, were spotted onto TLC plates and visualized as described above.

Mass spectrometry analysis of nucleotides (p)ppGpp was carried out according to an earlier protocol (19). The cultures were lysed by the addition of 50% ethanol and homogenized in a Beadbeater (Biospec), and the dried extracts analyzed by ESI-MS.

Estimation of RelZ promoter activity.

The upstream sequences (200 bp for p200 and 1,000 bp for p1000) of the relZ gene were cloned in a pSD5B plasmid upstream of the lacZ gene (21). The M. smegmatis wild-type strain was transformed with promoter-lacZ fusion constructs and grown in MB7H9 medium containing 2% glucose. LacZ expression levels were estimated by performing β-galactosidase activity assays using ONPG (o-nitrophenyl-β-d-galactopyranoside) as a substrate. The activity was calculated as described earlier using the following formula: activity (Miller units) = 1,000 × [{A₄₂₀ – 1.75(OD₅₅₀)}/{time × culture volume × OD₆₀₀}].

CFU analysis for antibiotic sensitivity.

The M. smegmatis strains were grown until mid-log phase, and the OD₆₀₀ was normalized to 0.6. The cultures were subsequently exposed to 4 μg/ml bleomycin, 20 μg/ml rifampin, 4 μg/ml ofloxacin, and 30 μg/ml erythromycin for 24 h. The bacteria were diluted by 10-fold, and the serial dilutions were plated into MB7H9 agar medium. The bacterial CFU were estimated after 3 to 4 days of growth at 37°C. The survival percentage was calculated as the percentage of CFU remaining after exposure to antibiotics compared to the CFU before antibiotic treatment.