Escherichia coli strain DH10B was grown in Luria broth that was supplemented with 300 μg ml^-1 erythromycin and 50 μg ml^-1 of kanamycin or spectinomycin, when needed. S. mutans wild-type (WT) strain UA159 and its derivatives (Table 1) were grown in either brain heart infusion (BHI; Difco) or FMC medium (Terleckyj and Shockman, 1975) that was supplemented with 10 μg ml^-1 erythromycin and 1 mg ml^-1 of kanamycin or spectinomycin, as needed. Synthetic XIP (sXIP, aa sequence = GLDWWSL), corresponding to residues 11–17 of ComS, was synthesized and purified to 96% homogeneity by NeoBioSci (Cambridge, MA, USA). The lyophilized sXIP was reconstituted with 99.7% dimethyl sulfoxide (DMSO) to a final concentration of 2 mM and stored in 40 μl aliquots at -20°C. Unless otherwise noted, cultures were grown overnight in BHI medium with antibiotics, if needed, at 37°C in a 5% CO₂ aerobic atmosphere. The cultures were harvested by centrifugation, washed in 1 mL FMC, and resuspended in FMC to remove all traces of BHI before fresh medium was inoculated. For growth rate comparisons, overnight cultures were diluted 1:50 and grown to mid-exponential phase (OD₆₀₀ = 0.5–0.6). Then, cultures were re-diluted 1:100 into 300 μl of FMC and added to multiwell plates, sXIP was added to the desired final concentrations (0.2, 2, or 20 μM), and cultures were overlaid with 50 μl of sterile mineral oil to reduce the growth inhibitory effects of air on growth of S. mutans. The optical density at 600 nm (OD₆₀₀) was monitored at 30 min intervals for 24 h using a Bioscreen C lab system (Helsinki, Finland) at 37°C with shaking for 10 s before each reading.

Mutant strains of S. mutans were created using a PCR ligation mutagenesis approach (Supplementary Table S1) (Lau et al., 2002). Splice overlap extension was utilized to create single base change mutations within the relA gene using protocols described elsewhere (Ho et al., 1989; Cha et al., 1992; Kaspar et al., 2015). Overexpression of genes was achieved by amplifying the genes of interest from S. mutans UA159 and cloning into the expression vector pIB184 (Biswas et al., 2008; Guo et al., 2014; Kaspar et al., 2015). Transformants were confirmed by PCR and sequencing after selection on BHI agar containing the appropriate antibiotic(s). Plasmid DNA was isolated from E. coli by using QIAGEN (Chatsworth, CA, USA) columns, and restriction and DNA-modifying enzymes were obtained from Invitrogen (Gaithersburg, MD, USA) or New England Biolabs (Beverly, MA, USA). PCRs were carried out with 100 ng of chromosomal DNA using Taq DNA polymerase, and PCR products were purified with the QIAquick kit (QIAGEN). DNA was introduced into S. mutans by natural transformation and into E. coli by the calcium chloride method (Cosloy and Oishi, 1973).

To measure the expression of genes using quantitative real-time PCR (RT-qPCR), three colonies of S. mutans WT (UA159) and the mutant strains of interest (ΔrelAPQ, ΔrelA) were grown overnight and a 1:50 dilution was added to fresh FMC medium the next day. When cells reached an OD₆₀₀ of 0.2, cultures were adjusted to a final concentration of 1% DMSO or 2 μM sXIP and incubated at 37°C in a 5% CO₂ atmosphere for 1 h before being harvested by centrifugation. Cell lysis was achieved through mechanical disruption (bead beating) and RNA was extracted by acidic phenol phase separation. The RNA was further purified with an RNeasy mini kit (QIAGEN) according to the provided protocol, and treatment with DNase 1 (QIAGEN). Purified RNA (1 μg) was used to generate cDNA from gene-specific primers (Supplementary Table S1) using the Superscript III first-strand synthesis (Invitrogen) reverse transcription protocol. Real-time PCRs were carried out using an iCyclerQ real-time PCR detection system (Bio-Rad) and iQ SYBR green supermix (Bio-Rad) according to the supplier’s protocol. 16S rRNA was used as an internal reference. All assays were performed in triplicate with RNA isolated from three biological replicates.

Strains of S. mutans were grown overnight and then diluted 1:50 in 200 μl of FMC medium. When the cultures reached an OD₆₀₀ = 0.2, either a final concentration of 1% DMSO or 2 μM sXIP was added to the growing culture. After 10 min of incubation, 500 ng of purified plasmid pDL278, which harbors a spectinomycin resistance (Sp^R) gene, was added. Following 3 h of further incubation, dilutions of the cultures were plated on BHI agar plates with or without 1 mg ml^-1 spectinomycin. Colony forming units (CFUs) were counted after 48 h. Transformation efficiency was determined by dividing the number of Sp^R transformants by the total number of CFU recovered on non-selective medium.

Overnight cultures of S. mutans were diluted 1:50 into 35 mL of fresh FMC medium and harvested by centrifugation when the cultures reached an OD₆₀₀ = 0.5. When desired, 1% DMSO or 2 μM sXIP (final concentrations) was added when the cultures reached an OD₆₀₀ value of 0.2. Cell pellets were washed once with buffer A (0.5 M sucrose; 10 mM Tris–HCl, pH 6.8; 10 mM MgSO₄) containing 10 μg ml^-1 of phenylmethanesulfonyl fluoride (PMSF; ICN Biomedicals Inc.), collected by centrifugation, and resuspended in 0.5 mL Tris-buffered saline (50 mM Tris–HCl, pH 7.5; 150 mM NaCl). Cells were lysed using a Mini Bead Beater (Biospec Products) in the presence of 1 volume of glass beads (average diameter 0.1 mm) for 30 s intervals, three times, with incubation on ice between homogenizations. Lysates were then centrifuged at 3,000 ×g for 10 min at 4°C. Protein concentrations of the resulting supernates were determined using the bicinchoninic acid assay (BCA; Thermo Scientific) with purified bovine serum albumin as the standard. ComX protein was detected by immunoblotting using a protocol detailed elsewhere (Kaspar et al., 2015). Protein preparations (10 μg) were loaded on a 15% polyacrylamide gel with a 6% stacking gel and separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred to Immobilon-P polyvinylidene difluoride membranes (Millipore), then treated with primary polyclonal anti-ComX antisera at a 1:1,000 dilution and a secondary peroxidase-labeled, goat anti-rabbit IgG antibody (1:5,000 dilution; Kirkegaard & Perry Laboratories, USA). Detection was performed using a SuperSignal West Pico Chemiluminescent Substrate kit (Thermo Scientific) and visualized with a FluorChem 8900 imaging system (Alpha Innotech, USA).

For measurements of LacZ produced from promoter:reporter gene fusions, strains were grown overnight, washed in FMC, resuspended, and diluted 1:40 into fresh FMC medium. At OD₆₀₀ = 0.2, sXIP was added to a final concentration of 2 μM, cells were incubated for an additional 90 min, then cells were harvested by centrifugation. β-galactosidase activity was measured by using a modification of the Miller protocol (Zubay et al., 1972). Briefly, cells were washed once with Z buffer [Na-phosphate buffer (pH 7.0), 10 mM KCl, 1 mM MgSO₄, 5 mM β-mercaptoethanol), and resuspended in 1.3 ml Z buffer. Part of the sample (500 μl) was vortexed with 25 μl of a toluene–acetone mix (1:9) for 2 min and kept at 37°C, the remaining 0.8 ml of the suspension was used to determine OD₆₀₀. The LacZ reaction was initiated by the addition of 100 μl of o-nitrophenyl-β-D-galactopyranoside (4 mg ml^-1) and was terminated by the addition of 500 μl of 1 M Na₂CO₃. Samples were centrifuged for 1 min using a tabletop centrifuge and the OD of the supernatant fluid was measured at 420 and 550 nm. Activity was expressed in Miller units (Zubay et al., 1972).

Visualization of (p)ppGpp accumulation was conducted as described elsewhere (Lemos et al., 2007). Briefly, overnight cultures from strains of interest were diluted 1:50 into a modified FMC medium containing a lower concentration of inorganic phosphate (Lemos et al., 2007). When cells reached an OD₆₀₀ of 0.2, [³²P]-orthophosphate, along with the desired amount of sXIP, was added. Cells were then incubated at 37°C for the specified duration, then harvested by centrifugation. (p)ppGpp was extracted by addition of an equal volume of ice-cold 13 M formic acid, followed by two freeze–thaw cycles in a dry ice/ethanol bath. After extraction, the CPM μl^-1 of the supernatant fraction was measured in a scintillation counter and 2.0 × 10⁵ CPM of each sample was spotted onto a polyethyleneimine (PEI) cellulose plate (Selecto Scientific) for thin-layer chromatography (TLC). The TLC plates were chromatographed in one dimension with 1.5 M KH₂PO₄ (adjusted to pH 3.4 with phosphoric acid), air-dried and exposed to X-ray film at -70°C. The signal density of spots corresponding to ppGpp (GP4) and pppGpp (GP5) was analyzed using ImageJ (v1.47) software¹.

While previous communications indicated that a deletion of the relA gene of S. mutans does not appear to have effects on transformation efficiency (Lemos et al., 2004), those experiments were performed in the complex medium BHI in the presence of CSP. Since that time, it has come to light that the ComRS system is the proximal regulator of alternative sigma factor comX expression, not ComDE, in S. mutans. Importantly, the ComRS system has been shown to function optimally in a chemically defined medium (Son et al., 2012), and requires ComS-derived XIP for ComR-dependent activation of gene expression (Mashburn-Warren et al., 2010). To determine if (p)ppGpp influenced the development of competence by ComRS in a chemically defined medium (FMC), we compared the growth kinetics between the S. mutans WT strain UA159 and a (p)ppGpp⁰ mutant (ΔrelAPQ) in the presence of various concentrations of sXIP, which was dissolved in DMSO. Cells treated with the same final v/v concentration (1%) of DMSO served as controls. The WT strain UA159 displayed apparent growth inhibition, as assessed by monitoring doubling time and final optical density, as the concentration of sXIP was increased (Wenderska et al., 2012; Guo et al., 2014; Figure 1A). We specify apparent growth inhibition because cell lysis may also influence the total optical density at any given time point. The exponential doubling time of UA159 increased from 66 ± 1 to 204 ± 5 min as the concentration of sXIP was increased from 0.2 to 2 μM. A doubling time of 229 ± 18 min was observed in the presence of 20 μM sXIP. A reduction in the final yield (final OD₆₀₀ at the 36-h time point) was seen with increasing concentrations of sXIP; 1.08 ± 0.01 with no XIP and 0.45 ± 0.08 at the highest concentration of XIP. Interestingly, the growth inhibitory effects of sXIP were not observed in the (p)ppGpp⁰ strain (Figure 1B). As sXIP concentrations were increased from 0.2 μM to 2 or 20 μM, there were only minimal increases in the doubling times: 98 ± 2, 111 ± 1, and 109 ± 2 min, respectively. Similarly, there were only marginal reductions in the final yields in the strain lacking (p)ppGpp.

To confirm that changes in growth inhibition were due to a loss of (p)ppGpp, we added back the relP gene, encoding a SAS that lacks hydrolase activity, under the control of the constitutive P23 promoter on a shuttle plasmid (Biswas et al., 2008) and compared the strains growth kinetics to one containing only the empty pIB184 vector. While the (p)ppGpp⁰ strain containing the empty vector showed limited growth inhibition in the presence of XIP (Figure 1C), addition of relP led to increased doubling times of 208 ± 13 and 273 ± 24 min with addition of 2 or 20 μM sXIP, respectively, compared to 86 ± 3 min with the 1% DMSO control (Figure 1D). Additionally, final yields dropped from 0.88 ± 0.02 at 24 h after growth without sXIP to 0.68 ± 0.03 and 0.60 ± 0.03 with 2 or 20 μM sXIP. These data suggest that modulation of (p)ppGpp levels, as seen with the complementation of the (p)ppGpp⁰ strain, can lead to phenotypic changes in response to the competence inducing peptide XIP, altering the response level to the signaling peptide.

This lack of growth inhibition phenotype in cells treated with sXIP is commonly associated with strains that have mutations in either comR or comX, indicating a severe reduction in early com gene expression that would lead to the (p)ppGpp⁰ failing to respond to the sXIP signal (Perry et al., 2009; Wenderska et al., 2012). This same phenotype is also observed in the ΔrcrR-NP strain, which fails to express a full-length comX mRNA (Kaspar et al., 2015). To test for com gene down-regulation, expression was measured by qRT-PCR in the WT and (p)ppGpp⁰ strains 1 h after addition of 2 μM sXIP to cells at OD_{600 nm} = 0.2. Interestingly the comX gene, encoding the alternative sigma factor required for late competence gene expression (Supplementary Figure S1A), and the late competence gene comYA (Supplementary Figure S1B) were both induced to the same level in the mutant and WT strains treated with XIP. The qRT-PCR results were consistent with the finding that similar levels of ComX protein (Kaspar et al., 2015) were present in lysates of mutant and WT cells that had been treated with XIP (Supplementary Figure S1C). Similar gene expression profiles between the WT and (p)ppGpp⁰ strains following addition of sXIP were also observed for a direct target of ComE (comD), for the comRS genes, and for rcrR, the first gene in the rcrRPQ operon that has been shown to influence competence, stress tolerance, and (p)ppGpp accumulation (Supplementary Figures S1D–G). Thus, the ComRS signaling pathway was intact and functional in both the WT and (p)ppGpp⁰ strains, so the specific impact of complete loss of (p)ppGpp on the growth and competence phenotypes apparently does not occur at the level of comRS expression or formation of an active ComR–XIP complex.

Clearly, (p)ppGpp levels can alter the growth phenotype of cells exposed to sXIP, so we reasoned that addition of sXIP may lead to an increase in internal (p)ppGpp pools. To determine how (p)ppGpp accumulation was affected by the addition of exogenous XIP, the WT strain UA159 was grown in FMC medium and labeled with ³²P-orthophosphate in the presence of either 1% DMSO (control) or increasing concentrations of sXIP. More specifically, cultures of S. mutans were grown to OD₆₀₀ = 0.15 and then ³²P-orthophosphate and sXIP were added together. Cultures were incubated for 60 min, cells were harvested, and formic acid extracts of the phosphorylated guanosine nucleotides were prepared and separated by thin layer chromatography (Figure 2A). Addition of mupirocin, which leads to RelA-dependent accumulation of (p)ppGpp, served as a positive control. Only a very modest increase in (p)ppGpp accumulation was evident with addition of 0.2 μM sXIP, compared to the 1% DMSO control. However, there were substantial increases in (p)ppGpp pools in the samples treated with 2 or 20 μM sXIP (Table 2). The latter concentrations of sXIP correspond to those that induce the greatest degree of growth inhibition in UA159, providing further evidence of an essential role for (p)ppGpp in XIP-dependent inhibition of growth. Induction of (p)ppGpp after sXIP exposure is also time-dependent, as cells harvested at 10, 20, 40, and 60 min after addition of ³²P-orthophosphate and sXIP showed the greatest accumulation of (p)ppGpp at the 40 and 60 min time points (Supplementary Figure S2). To determine how addition of sXIP to the growth medium might influence expression of individual rel genes, we measured the expression of relP, relQ, and relA mRNA levels in UA159 after addition of 2 μM sXIP (Figure 2B). Expression of the relQ gene was increased 1.9-fold after sXIP addition, compared to the 1% DMSO control, whereas relA gene expression was decreased 2.8-fold with XIP treatment. There were no increases noted in relP mRNA associated with sXIP treatment. However, it was recently reported that there is a ComE binding site in front of the cipI gene (SMU.925, immB; Khan et al., 2016), so we cannot rule out that ComE-dependent activation of cipI following XIP treatment might influence downstream relP (SMU.926) mRNA levels under other conditions that lead to activation of competence. Collectively, though, the results indicate increased production of RelQ and possible allosteric modulation of RelA synthetase/hydrolase activity can impact the quantity of (p)ppGpp in XIP-treated cells.

The three (p)ppGpp synthetases in S. mutans appear to contribute to accumulation of alarmone under different circumstances (Lemos et al., 2007), so we explored the behavior of strains carrying individual or selected combinations of RSH and SAS enzymes on competence pathway signaling through ComRS. First, the growth kinetics of strains harboring individual deletion:replacement mutations of relA (Figure 3A), relP, and relQ were evaluated. The ΔrelA strain showed the lowest sensitivity to sXIP, with a doubling time of 99 ± 5 to 101 ± 4 min in the presence of 0.2 and 2 μM sXIP, respectively. A longer doubling time (147 ± 3 min) was observed with 20 μM sXIP. Although doubling times increased with increased concentrations of sXIP, the final yields did not differ (0.70 ± 0.06 to 0.77 ± 0.03 min). In contrast, a strain carrying mutations in both relP and relQ (ΔrelPQ) showed growth inhibitory effects to XIP that were similar to those seen with the WT strain (Figure 3B), as was the case for the strains carrying only relP or relQ mutations (data not shown). The observed doubling times for the ΔrelPQ mutant strain were substantially greater than for the parental strain in the presence of 2 μM (229 ± 39 min) or 20 μM XIP (324 ± 68 min) and final yields were also lower (0.47 ± 0.14, 0.39 ± 0.05). Of note, a lag time of approximately 18 h was seen with 2 μM XIP in the ΔrelPQ strain. When mutants lacking the relA gene were combined with individual relP or relQ mutations (ΔrelAP, ΔrelAQ), the strains grew like the ΔrelA single mutant, providing further evidence that the loss of the Rel (RSH) enzyme alone could alter sensitivity to sXIP.

Fluctuations in sensitivity of strains to sXIP can be a potential indicator for changes in the ability to be transformed by exogenous DNA (Ahn et al., 2014; Kaspar et al., 2015). Transformation efficiency was measured in the ΔrelA, ΔrelP, and ΔrelQ strains grown in FMC and treated with 2 μM XIP to induce competence. The ΔrelA mutant strain showed a five-log reduction in transformation efficiency (2.0 × 10^-7), compared to the WT strain (3.8 × 10^-2; Figure 3C). In contrast to the ΔrelA mutant, the ΔrelP and ΔrelQ strains showed no reduction in transformation efficiency. To confirm the results of the transformation assay, ComX levels were measured by western blot in the ΔrelA and ΔrelPQ strains. A clear decrease in ComX protein was seen in the ΔrelA strain when treated with sXIP, but ComX protein in the ΔrelPQ and WT strains were similar (Figure 3D). Interestingly, the ΔrelA strain showed a decrease in transformation efficiency of less than one log compared to the WT strain when cells were grown in BHI medium and synthetic competence stimulating peptide (sCSP) was used to induce competence (3.7 × 10^-3 to 1.2 × 10^-2), comparable to the behavior noted for the ΔrelP strain (3.4 × 10^-3; Supplementary Figure S3). It is of interest that the effects of loss of the RSH enzyme were more profound in chemically defined medium. In particular, XIP is known to be ineffective at inducing com gene expression or enhancing transformation in complex medium (BHI), whereas sCSP only induces comX and competence in complex medium. Thus, it is likely that the effects of loss of the RSH enzyme are exerted through the ComRS pathway in a WT genetic background. Importantly, the behavior of the WT strain contrast dramatically with that of the (p)ppGpp⁰ strain, in which com gene expression was not similarly impacted. This difference in behavior is likely attributable to the fact that the (p)ppGpp⁰ strain completely lacks (p)ppGpp, whereas the WT and relA mutant have higher basal levels of (p)ppGpp due to a lack of (p)ppGpp hydrolase activity. Thus, com gene expression may only be influenced once (p)ppGpp production reaches a critical threshold and is not impacted when only low levels of (p)ppGpp are present, as is the case for the (p)ppGpp⁰ strain; highlighting further the critical role of (p)ppGpp in manifestation of these important phenotypes.

To determine where in the competence signaling cascade the effects of deletion of relA were occurring, the expression of selected com genes was measured by RT-qPCR in cells growing in FMC and treated with 2 μM sXIP. Expression levels of comR were one-log higher in UA159 than in the strain lacking the RSH enzyme (ΔrelA), both in control (1% DMSO) samples and in cells treated with sXIP (Figure 4A). The levels of comR expression were not altered by exposure to XIP in either strain. A different pattern of expression was noted for comX (Figure 4B), comYA (Supplementary Figure S4A) and comS (Supplementary Figure S4B). In particular, expression of comX in the relA-deficient strain was modestly lower than in WT cells in the absence of sXIP treatment and about 1-log lower after treatment with XIP; a similar pattern was noted for comYA. It is significant, though, that induction of comX and comYA still occurs in response to sXIP treatment in the relA-deficient strain, albeit less so than in the WT genetic background. In contrast, the expression of comS, while induced by XIP in both UA159 and the ΔrelA mutant, did not differ between strains (Supplementary Figure S4B). Additionally, expression of comD and cipB were monitored and no differences were detected between UA159 and the ΔrelA strains in cells treated only with DMSO. However, when the ΔrelA strain was treated with sXIP there was no induction of comD and a lower level of induction of cipB, compared to that seen in the WT strain (Supplementary Figures S4C,D). Thus, the data are most consistent with the influences of Rel on competence arising from alterations in the expression of comR, although other inputs outside the ComRS–XIP system may modulate gene expression, sensitivity to XIP, and competence development.

To examine if the changes in comR mRNA levels were correlated with increased comR promoter activity, PcomR, was fused to a Streptococcus salivarius lacZ gene and integrated in single copy into the chromosome of S. mutans at the mtlA-phn locus (Son et al., 2012; Guo et al., 2014). A significant decrease in LacZ activity was measured in the ΔrelA background, compared to the WT strain (Figure 4C), with or without addition of sXIP. A similar reduction in promoter activity in the relA-deficient strain was also observed using a PcomX-lacZ fusion (Figure 4D). Thus, deletion of relA can influence PcomR activity, with the down-regulation of comR likely accounting for diminished activation of PcomX.

Previously, a complemented ΔrelA strain was constructed by providing an intact copy of relA gene in trans on plasmid pMSP3535, which allows for induction of the gene with subinhibitory concentrations of nisin (Lemos et al., 2004). To confirm that reductions in transformation efficiency and com gene expression were due to the loss of the Rel enzyme, the relevant phenotypes of the complemented strain (ΔrelA/relA⁺) were compared with those of the relA mutant. The ΔrelA/relA⁺ strain displayed similar transformation efficiency as the WT strain (1.4 × 10^-3 versus 1.5 × 10^-3, respectively), whereas a strain in which the relP gene was provided on a plasmid in the relA mutant background (ΔrelA/relP⁺) displayed transformation efficiencies equivalent to the ΔrelA strain (2.8 × 10^-7 to 2.4 × 10^-7, respectively; Figure 5A). qRT-PCR of selected com genes in the complemented strains revealed that the ΔrelA/relA⁺ strain had similar quantities of mRNA as the WT strain for comR (Figure 5B) and comX (Figure 5C), whereas com gene expression in the ΔrelA/relP⁺ strain remained comparable to that in the ΔrelA mutant. Thus, it was not sufficient to provide a plasmid-borne copy of a gene encoding only (p)ppGpp synthase activity (relP) in the relA mutant to achieve full complementation of RelA deficiency. Rather complementation of the RelA deficit was only achieved with an RSH enzyme containing both (p)ppGpp synthase and hydrolase activities (relA).

While internal accumulation of CipB induced by CSP via the ComDE system is thought to be involved in altruistic cell death and fratricide in a sub-population of cells (Perry et al., 2009), XIP-mediated cell death in chemically defined medium requires the ComRS system (Wenderska et al., 2012). This is apparently attributable to a recently discovered feedback circuit in which ComX binds to the comE promoter and leads to enhanced activity of the ComE regulon (Reck et al., 2015; Son et al., 2015b). Since expression of comR was decreased in the ΔrelA strain, deficiencies in activation of comX by the ComRS system could be responsible for the observed changes in apparent growth inhibition by sXIP and transformation efficiencies. To explore this possible mechanistic linkage, we created strains that overexpressed comS, comR, and comX under the control of the P₂₃ promoter on the shuttle plasmid pIB184 (Biswas et al., 2008) in the ΔrelA genetic background. Growth kinetic studies in response to sXIP were conducted and compared to the ΔrelA strain carrying only the pIB184 vector. Overexpression of comS elicited no changes in growth inhibition by sXIP (Figure 6A). Overexpression of comR in the relA mutant strain (184ComR/ΔrelA) lead to modest growth inhibition in the presence of 2 μM sXIP, with an increase in the doubling time to 109 ± 4 min and reduction in final OD to 0.83 ± 0.05 (Figure 6B). Greater inhibition was observed in the presence of 20 μM sXIP, with an increased lag time (3 h), a doubling time of 194 ± 8 min, and a decrease in final yield to 0.71 ± 0.06. The 184ComX/ΔrelA strain (overexpressing comX) showed the greatest growth inhibition in the presence of sXIP, with lag times of 24 and 29 h in the presence of 2 or 20 μM XIP, respectively (Figure 6C). Growth of the 184ComX/ΔrelA strain in 20 μM sXIP resulted in doubling times of 240 ± 4 min and a final yield of 0.70 ± 0.04 after 48 h, very close to that observed for the WT strain. Thus, the loss of sensitivity of the strain lacking the RelA (RSH) enzyme to sXIP is clearly due to the aberrant decrease in comX mRNA levels and ComX production and can be compensated by overexpression of comR and comX, but not of comS. These results also suggest that the competence deficiencies of ΔrelA stem from a lack of ComR, with enough ComR–XIP complexes present to fully saturate expression from the PcomS promoter, but not the PcomX promoter under similar conditions leading to deviations in transformation efficiency and cell lysis through ComX-dependent mechanisms.

The RelA enzyme can synthesize and degrade (p)ppGpp, with the catalytic domains for these activities located in the N-terminal portion of the enzyme. Synthetase-ON/hydrolase-OFF and synthetase-OFF/hydrolase-ON states are controlled by allosteric regulation mediated by the C-terminal domain (Mechold et al., 1996, 2002). Analysis of the crystal structure of the N-terminal fragment (residues 1–385) of the Rel enzyme of Streptococcus dysgalactiae subsp. equisimilis (Rel_Seq) revealed amino acid residues critical for the activities of both domains, with mutation of specific amino acids leading to loss of activity in one domain without affecting the function of the other (Hogg et al., 2004). RelA of S. mutans shares 85% identity at the amino acid level with Rel_Seq, and residues found to play critical roles in the catalytic activity of Rel_Seq are conserved in S. mutans (data not shown). Amino acid substitution mutations in the Rel_Seq synthetase and hydrolase domains were recapitulated in the S. mutans RelA enzyme to determine if one or both enzymatic activities were critical for normal com gene expression. Inhibition of the synthetase function was achieved by mutating the aspartate residue at position 264 to glycine (D264G, relA^{ΔSY N}; Hogg et al., 2004). When grown with increasing concentrations of sXIP, this strain displayed growth inhibition similar to the WT strain (Figure 7A). Mutation of the catalytic site of the hydrolase domain was achieved by substituting a threonine residue with proline at position 151 (T151P; Hogg et al., 2004). Interestingly, viable colonies containing this single substitution could not be recovered, and DNA sequence analysis of transformants that did contain the desired mutation were found to have secondary mutations in the relA gene; either an additional mutation in the synthetase domain or a frameshift mutation that led to production of a truncated enzyme that was predicted to be non-functional. Thus, we were unable to obtain a relA^{ΔHY D} strain. These results suggest that strains of S. mutans in which the loss of function of the RelA (p)ppGpp hydrolase domain has occurred without concurrent loss of the synthetase domain are non-viable. Notably, the relA^{ΔHY D} mutation could also not be obtained in a strain lacking relPQ, so synthesis of (p)ppGpp by RelA and the associated over accumulation of alarmone in the absence of the hydrolase domain appears sufficient to account for lack of growth of an otherwise-WT relA^{ΔHY D} mutant. We were, however, able to incorporate the T151P mutation into the relA^{ΔSY N} strain to create a viable strain expressing a mutant derivative of RelA with no (p)ppGpp synthase or hydrolase activity (relA^{ΔSY NΔHY D}). When relA^{ΔSY NΔHY D} strain was evaluated for its sensitivity to XIP (Figure 7B), the strain displayed a level of resistance to XIP similar to the ΔrelA mutant, adding further support that loss of the hydrolase domain is responsible for the changes in XIP sensitivity. These observations are consistent with the previous findings (Figure 5) that addition of a synthetase only (relP) back to the ΔrelA strain is unable to complement the reduction in competence activity phenotypes. To further confirm the importance of the hydrolase domain in responses of S. mutans to XIP, we examined a strain (relA³⁸⁵) that includes an insertion of an antibiotic resistance cassette beginning at amino acid 386, resulting in a 385-aa peptide containing only the N-terminal domain of RelA. This strain produces high basal levels of (p)ppGpp (data not shown), with the produced peptide containing no detectable (p)ppGpp degrading activity (Mechold et al., 2002). The relA³⁸⁵ strain showed growth kinetics in the presence of increasing concentrations of sXIP similar to the ΔrelA and relA^{ΔSY NΔHY D} strains (Figure 7C). Transformation efficiency (Figure 7D) and ComX levels in each of these strains was consistent with the growth characteristics in the presence of XIP (Figure 7E). Taken together, these results show that the hydrolase domain of RelA plays an essential role in optimizing (p)ppGpp levels to allow for normal responses to XIP and expression of com genes.

Two genes (tpx, cipI) separate the rcrRPQ and relPRS operon, all transcribed in the same direction. The products of rcrRPQ and relP profoundly and concurrently impact stress tolerance, (p)ppGpp production, and genetic competence in S. mutans (Seaton et al., 2011). Moreover, the MarR-type regulator RcrR was found to interact with the relP promoter in vitro (Seaton et al., 2015) and the levels of the ABC transporters RcrP and RcrQ and two small peptides encoded at the 3′ end of rcrQ exert dominant control over competence development (Ahn et al., 2014). Of particular relevance here, a ΔrcrR-P (polar) mutant is hyper-transformable, whereas a ΔrcrR-NP (non-polar) mutant is completely non-transformable, in part due to lack of expression of full-length comX mRNA and production of very low levels of ComX (Kaspar et al., 2015). As the phenotypes displayed by the RcrRPQ operon and (p)ppGpp are intertwined, we reasoned that RcrRPQ and its effectors could be a source for modulation of (p)ppGpp accumulation leading to changes in the transformable state of cells based on environmental signals/conditions. The extremely disparate competence phenotypes displayed by the ΔrcrR polar and non-polar strains provided us the opportunity to explore mechanisms related to RelA activity and its role in com gene expression, which would have been more difficult to analyze in the WT background. Hence, we examined the effects of deleting relA (ΔrelA) in the rcrR mutants. The transformation efficiency of the ΔrcrR-PΔrelA double mutant (Figure 8A) had reduced transformation compared with the ΔrcrR-P mutant (6.8 × 10^-4 versus 1.2 × 10^-2, respectively). A decrease in ComX protein abundance was also observed in this double mutant (Figure 8B). Interestingly, a complete reversal of the non-transformable phenotype of the ΔrcrR-NP strain was evident when the ΔrelA mutation was present in the same strain. In particular, the ΔrcrR-NP strain alone could not be transformed, but the ΔrcrR-NPΔrelA double mutant displayed transformation efficiency comparable to the hyper-transformable ΔrcrR-P strain (1.3 × 10^-2), and ComX levels were also restored to a level similar to those expressed in the ΔrcrR-P strain. Collectively, these data provide further confirmation of the critically important roles of the RelA enzyme and (p)ppGpp in modulation of genetic competence and peptide signaling behaviors in S. mutans.