Characteristics of Streptococcus mutans strains lacking the MazEF and RelBE toxin–antitoxin modules


  • José A.C. Lemos,

    1. Department of Oral Biology, University of Florida College of Dentistry, 1600 SW Archer Road, Gainesville, FL 32610-0424, United States
    Search for more papers by this author
  • Thomas A. Brown Jr.,

    1. Department of Oral Biology, University of Florida College of Dentistry, 1600 SW Archer Road, Gainesville, FL 32610-0424, United States
    Search for more papers by this author
  • Jacqueline Abranches,

    1. Department of Oral Biology, University of Florida College of Dentistry, 1600 SW Archer Road, Gainesville, FL 32610-0424, United States
    Search for more papers by this author
  • Robert A. Burne

    Corresponding author
    1. Department of Oral Biology, University of Florida College of Dentistry, 1600 SW Archer Road, Gainesville, FL 32610-0424, United States
    Search for more papers by this author

  • Edited by T. Mitchell

*Corresponding author. Tel.: +1 352 3924370/8462521; fax: +1 352 3927357., E-mail address:


Two pairs of genes were identified in Streptococcus mutans with similarity to relBE and mazEF toxin–antitoxin (TA) modules of Escherichia coli. Transcription of mazEF and relBE was repressed by amino acid starvation, and relBE expression was repressed by low pH. Mutants lacking MazF, RelE, or both toxins (MRT1) grew in broth media and formed biofilms as well as the parent. Biofilm populations of MRT1 were more resistant to acid killing than the parent or single mutants. MRT1 also exhibited a longer diauxie during growth on glucose and inulin and displayed decreased phosphoenolpyruvate:sugar phosphotransferase activity. This is the first report that demonstrates a physiological role for TA modules in Gram-positive bacteria.


Streptococcus mutans thrives in multi-species biofilms on tooth surfaces, where it is subjected to a continuous assault by host defenses and to rapid and dramatic fluctuations in nutrient availability, carbohydrate source and pH. The organism has evolved multiple physiologic and genetic adaptations to optimize growth under these dynamic conditions, gaining a selective advantage over less aciduric species when oral biofilms become acidified.

Toxin–antitoxin (TA) gene pairs, also known as addiction modules, have been implicated in stress management and survival by bacteria. TA modules were first identified in low-copy-number plasmids as preventing plasmid loss through a post-segregational killing effect [1]. TA modules are arranged in operons with the upstream gene coding for an unstable antitoxin and the downstream gene coding for a stable toxin. The antitoxin neutralizes the cognate toxin through protein–protein interaction. When bacteria lose the plasmids, cured cells are selectively killed because unstable antitoxins are degraded faster than the more stable toxin molecules. In this way, the bacterial host becomes addicted to the presence of the plasmid. Chromosomally-borne addiction modules have been identified in Escherichia coli and their mechanism of action and function has been partially elucidated [2–6].

In E. coli, the MazEF module consists of two overlapping genes, mazE and mazF, and is located downstream from the relA gene, which is responsible for the production of the nutritional alarmone, (p)ppGpp [7]. The accumulation of (p)ppGpp, which occurs in response to nutrient starvation, inhibits the transcription of the mazEF operon. Because the antitoxin MazE is a labile protein, its cellular concentration rapidly decreases, leaving MazF free to exert its toxic effect [3]. It has been suggested that the MazEF system serves as an altruistic mechanism for programmed cell death (PCD) during extreme conditions of starvation, in which a sub-population of the starved cells undergoes lysis to enable the survival of the remainder of the population [3]. MazEF-mediated PCD has been shown to be activated by multiple factors, including transcription or translation inhibitors, nutritional stress, thymine starvation, high temperature, DNA damage and oxidative stress [3,8–10].

A second chromosomally-encoded TA module in E. coli, RelBE, was shown to promote growth arrest by dramatically reducing the levels of translation [4,5]. During nutrient starvation, the RelB antitoxin is degraded by the Lon protease and the RelE toxin inhibits translation by cleavage of mRNA at the ribosomal A-site [11]. Recently, it was demonstrated that the bacteriostatic effects of the RelE and MazF toxins could be fully reversed by expression of the cognate antitoxin [5]. Presently, there is much debate about whether MazEF and other TA pairs truly govern PCD or whether they are simply involved in reversible cell cycle arrest [5,12–14]. Notwithstanding, it is generally agreed that TA pairs constitute important stress response elements that function as “regulatory switches” under adverse conditions, allowing cells to enter a state that confers protection against severe nutrient limitation and possibly other stresses. Although BLAST searches have identified TA pairs in the majority of prokaryotes for which complete or partial genome sequence data is available [12,15], to our knowledge there have been no studies analyzing the role of TA pairs in Gram-positive bacteria. In this report, we identified and characterized two putative TA pairs in S. mutans, which has dramatically different physiology than E. coli. Mutant strains lacking one or both toxins were used to establish a link between TA pairs, growth arrest and stress responses.

2Materials and methods

2.1Bacterial strains and growth conditions

E. coli DH10B was grown in Luria broth and S. mutans UA159 was grown in brain heart infusion (BHI). If appropriate, ampicillin (100 μg ml−1), kanamycin (50 μg ml−1 for E. coli or 1 mg ml−1 for S. mutans) or erythromycin (300 μg ml−1 for E. coli or 10 μg ml−1 for S. mutans) was added to the media. To assess diauxic growth, cells were grown in tryptone-vitamin (TV) base medium [16] supplemented with 0.05% glucose and 0.5% of sorbitol or inulin. To induce a stringent response, cells were grown in the chemically defined medium FMC [17] to an OD600≅ 0.35, at which point serine hydroxamate (1 mg ml−1) was added [18]. For studies involving acid shock, cells were grown in BHI that was buffered at pH 7.0 to an OD600 of 0.3, washed with 0.1 M glycine buffer, pH 7.0, and then resuspended in fresh BHI that was adjusted to pH 5.0 with HCl. The ability to form stable biofilms was assessed by growing cells in the wells of polystyrene microtiter plates using biofilm medium (BM) [19]. Biofilms used for acid killing experiments were grown in 24-well polystyrene plates as previously described [18].

2.2Nucleic acid methods

Standard recombinant DNA techniques were used as previously described [20,21]. Total RNA was extracted from S. mutans by the hot acid–phenol method as described elsewhere [22]. Levels of mazEF and relBE mRNA were quantified by real-time reverse transcriptase (RT)-PCR. RT-PCR reactions, real time RT-PCR cycling conditions, analysis of data, and quality control were essentially as described elsewhere [22].

2.3Construction of strains

DNA fragments containing the genes of interest were amplified by PCR and cloned onto pGEM5 (Promega) to construct plasmids pJL48 and pJL61. Plasmid pJL48 contained a 1.2kb SphI/PstI fragment carrying the mazEF region. A kanamycin resistance gene flanked by strong transcription and translation terminators (ΩKan) was cloned into a BamHI site located 38 bp from the mazF start codon to give plasmid pJL49. Plasmid pJL61 contained a 0.9 kb SphI/SpeI fragment of the relBE region. An erythromycin marker was inserted at a HindIII site 75 bp from the relE start codon as a blunt-ended cassette to generate plasmid pJL64. Plasmids pJL49 and pJL64 were used to transform S. mutans UA159.

2.4Cell survival assays

To assess long-term survival, strains were grown in FMC containing 10 mM glucose. To investigate death in cells exposed to serine hydroxamate or transcription or translation inhibitors, mid-exponential phase FMC cultures were incubated with inhibitory concentrations of serine hydroxamate (1 mg ml−1), rifampicin (16 μg ml−1) or spectinomycin (50 μg ml−1). Aliquots of the cultures were removed daily, serially diluted and plated on BHI-agar for enumeration of colonies. For acid killing of biofilm-grown cells, 48-h biofilms formed in the wells of tissue culture plates were subjected to acid killing as previously described [18].

2.5Sugar transport assay

To measure sugar transport by the phosphoenolpyruvate:sugar phosphotransferase (PEP-PTS), cells were grown to an OD600 of 0.5 in TV medium supplemented with 0.5% (w/v) glucose. Cells were permeabilized with toluene–acetone (1:9) and glucose-, fructose- and mannose-specific PTS activities were assayed using the method of LeBlanc et al. [23]. PTS activity was expressed as nanomoles of NADH oxidized in a PEP-dependent manner min−1 (milligram of protein)−1. The protein concentration of permeabilized cells was determined by the bicinchoninic acid assay (Sigma).

3Results and discussion

3.1Identification of mazEF and relBE pairs in S. mutans

The amino acid sequences of the E. coli MazEF and RelBE proteins were used to identify two putative TA systems in the S. mutans UA159 annotated genome available at the Oral Pathogens Sequence Database ( [24] (Fig. 1). Smu0154 encodes for a 9.2 kDa protein that showed 31% identity (49% similarity) to the E. coli MazE antitoxin. Immediately downstream, and overlapping by 6 nucleotides with the putative mazE stop codon, was Smu0155, coding for a 12.6-kDa protein that showed 33% identity (51% similarity) with the E. coli MazF toxin. Pfam searches placed Smu0155 as a member of the growth inhibitor PemK-like protein family, which includes E. coli MazF. Smu0154 was placed in the SpoVT-AbrB family from Bacillus subtilis. The abrB gene encodes an ambiactive repressor and activator of transcription during the transition state between vegetative growth and the onset of stationary phase and sporulation [25].

Figure 1.

Diagrams showing the mazEF and relBE regions in S. mutans. (A) The mazEF locus (Smu0154-Smu0155). rplM codes for the 50S ribosomal protein L13. rpsI codes for the 30S ribosomal protein S9. Smu0156 is a conserved hypothetical protein. (B) The relBE locus (Smu0816–Smu0817). hsdM codes for the methylase subunit of the hsd type I restriction–modification system. hsdS codes for the specificity determinant. prrC codes for an anticodon nuclease. hsdR codes for the helicase subunits or restriction subunit of the hsd type I restriction–modification system. Arrows indicate the direction of transcription. The numbers below the schematic indicate the size of the ORFs and the numbers between ORFs indicate the size of the intergenic region in base pairs.

The second pair showed similarity with E. coli relBE. Smu0816 encoded a polypeptide with a predicted molecular mass of 10 kDa that shared 27% identity with E. coli RelB. The Smu0817 gene, which overlaps by 10 nucleotides with the relB stop codon, encodes for a predicted 10.7 kDa protein that shared 17% identity with E. coli RelE. Pfam searches placed Smu0816 in the RelB family while Smu0817 was placed in the DUF32 family, where DUF refers to Domain of Unknown Function. The limited level of homologies of Smu0816 and Smu0817 with RelB and RelE fromE. coli, respectively, is not surprising. The alignment of RelE proteins from both Gram-negative and Gram-positive bacteria indicated the presence of only two conserved residues, and the alignment of RelB proteins did not show any conserved domains [12]. Interestingly, the S. mutans relBE genes are imbedded in the hsd-prr locus, which was originally described in E. coli as coding for a ribonuclease that is activated after phage T4 infection and blocks phage development [26]. The hsd-prrC locus is thought to be a suicide system that may sacrifice individual members for the benefit of the population (altruistic cell death). The tight linkage of hsd-prrC and relBE may indicate that these systems act synergistically to promote growth arrest or altruistic cell death in S. mutans under particular environmental conditions.

3.2Transcriptional analysis of mazF and relE

The transcriptional levels of the mazEF and relBE operons in response to amino acid starvation triggered by the addition of serine hydroxamate, or to low pH, was assessed by real time RT-PCR. Addition of serine hydroxamate (1 mg ml−1) to mid-exponential phase S. mutans UA159 cultures, which induces growth arrest and increases (p)ppGpp pool levels [18], caused significant reductions in the levels of mazEF and relBE mRNA when compared to control cells obtained prior to serine hydroxamate treatment (Table 1). Of note, (p)ppGpp accumulation was also associated with transcriptional inhibition of mazEF in E. coli[3].

Table 1.  Relative quantification of mazEF and relBE mRNA upon exposure to serine hydroxamate
 mRNA (copies/μl)aRatiobp-Valuec
  1. aFollowing reverse transcription from 1 μg of total RNA from S. mutans UA159, the amounts of mazEF and relBE were determined by real time PCR using SYBR green. The data represent mean and standard deviations obtained from three different RNA preparations and RT reactions.

  2. bRatio of mRNA in relation to control (before addition of serine hydroxamate).

  3. cPaired student's t-test. p < 0.01 indicates statistical significance. NS indicates not significant (p > 0.01). NA indicates that t-test is not applicable.

0 min4.06E + 07 ± 2.22E + 071NA
30 min5.68E + 06 ± 4.42E + 050.102 ± 0.068<0.01
60 min3.26E + 05 ± 2.34E + 050.006 ± .0047<0.01
0 min3.52E + 05 ± 1.98E + 051NA
30 min1.24E + 05 ± 8.6E + 040.29 ± 0.2NS
60 min3.79E04+± 2.21E + 040.11 ± 0.02<0.01
0 min3.63E + 05 ± 2.54E + 051NA
30 min2.81E + 03 ± 1.95E + 030.06 ± 0.034<0.01
60 min2.13E + 04 ± 1.94E + 040.123 ± 0.093<0.01
0 min1.97E + 04 ± 1.43E + 041NA
30 min6.2E + 03 ± 3.87E + 030.32 ± 0.077<0.01
60 min2.1E + 03 ± 5.06E + 020.23 ± 0.22<0.01

Interestingly, a shift in the culture pH from 7.0 to 5.0 resulted in the inhibition of relBE transcription, but transcription of mazEF was not affected (Table 2). Rapid and sustained acidification of S. mutans occurs frequently in the oral cavity and the organism is believed to gain a considerable ecological advantage at the low pH values attained in dental biofilms. The finding that relBE is regulated in response to low pH is thus significant. Further, a lack of induction of mazEF in S. mutans in response to acid, compared with down-regulation of both operons during a stringent response, highlights that there are important differences in the regulation of expression of these TA modules and implies that the TA pairs have distinct contributions to cellular physiology depending upon the stress encountered.

Table 2.  Relative quantification of mazEF and relBE mRNA in response to low pH
 mRNA (copies/μl)ap-Valueb
  1. aFollowing reverse transcription from 1 μg of total RNA from S. mutans UA159, the amounts of mazEF and relBE were determined by real time PCR using SYBR green. The data represent mean and standard deviations obtained from three different RNA preparations and RT reactions.

  2. bPaired student's t-test. p < 0.01 indicates statistical significance. NS indicates not significant (p > 0.01). NA indicates that t-test is not applicable.

pH 7.03.13E + 08 ± 1.03E + 08NA
pH 5/30 min3.1E + 08 ± 1.8E + 08NS
pH 5/60 min4.21E + 08 ± 9.43E + 07NS
pH 5/120 min3.57E08 ± E + 07NS
pH 7.03.05E + 05 ± 1.31E + 05NA
pH 5/30 min1.24E + 05 ± 6.32E + 04NS
pH 5/60 min2.16E + 05 ± 5.37E + 04NS
pH 5/120 min2.4E + 05 ± 1.29E + 05NS
pH 7.02.24E + 06 ± 1.03E + 06NA
pH 5/30 min9.25E + 05 ± 5.48E + 05NS
pH 5/60 min9.22E + 05 ± 4.08E + 05NS
pH 5/120 min3.96E + 04 ± 1.67E + 04<0.01
Control1.25E + 06 ± 4.26E + 05NA
pH 5/30 min4.97E + 05 ± 3.24E + 05NS
pH 5/60 min4.02E + 05 ± 2.38E + 05NS
pH 5/120 min8.98E + 03 ± 3.57E + 03<0.01

3.3Influence of the MazF and RelE toxins on biofilm formation and acid tolerance of S. mutans

The formation of stable biofilms and the ability to grow and metabolize carbohydrates in an acidic environment are central to the virulence of S. mutans. Not surprisingly, stress tolerance pathways, which are intimately involved in growth and homeostasis, have a direct impact on the capacity of S. mutans to form stable biofilms (for a review, see [27]). Thus, we evaluated the impact of the loss of the toxins MazF (TJ-mazF), RelE (JL-relE) or both (MRT1) on biofilm formation and acid tolerance. The mutant strains showed no significant differences in biofilm formation when grown on microtiter plates with either glucose or sucrose as the primary carbohydrate source (data not shown). Acid killing at pH 2.8 of mid-exponential or stationary phase planktonic cells grown in either FMC or BHI revealed no differences between the mutants and parental strains (data not shown). In contrast, acid killing of biofilm cells showed that MRT1 was approximately 10 times more resistant than the parent strain (Fig. 2). Both single mutants showed no significant differences in biofilm acid resistance (data not shown). To rule out that the phenotype of biofilm-grown MRT1 cells was related to biofilm architecture or diffusion limitation, acid killing was also carried out on dispersed biofilm cells. The data obtained was comparable to what was found in intact biofilms with an approximately 10-fold increase in the MRT1 acid resistance suggesting that changes in cellular physiology were responsible for the increased acid resistance phenotype.

Figure 2.

Acid killing of S. mutans strains UA159 (triangles) and MRT1 (squares). Cells from 48-h biofilms grown on the surface of polystyrene plates were subjected to acid killing in 0.1 M glycine (pH 2.8). Cell viability at each time point is expressed as the percentage of viable cells [CFU (ml of culture−1)] at time zero.

The fact that planktonic and biofilm populations of S. mutans MRT1 strain display different behaviors in terms of acid resistance is itself notable and confirms that there are important physiological differences between sessile and planktonic populations of microorganisms. Interestingly, we previously reported that biofilm populations of relA strains of S. mutans became more resistant to acid killing, yet there were no differences in acid sensitivity of planktonic cultures [18]. The molecular basis for these findings has yet to be uncovered, but it is notable that there are significant differences in the growth rates of biofilm and planktonic populations. A significant proportion of the cells in mature biofilms are not actively growing and exhibit metabolic activity resembling that of stationary phase cells. Both our previous investigation [18] and the present study have demonstrated that stationary phase planktonic cells still behave differently than cells grown in biofilms. Thus, it appears that S. mutans does indeed possess a “biofilm phenotype”[27]. Mass transport limitations encountered in biofilms impose stresses on the cells, such as nutrient limitation and pH gradients, that may not be encountered during growth in planktonic cultures. The aberrant acid tolerance properties of the RelA-deficient and MRT strains only in biofilms may reflect that the stresses and corresponding environmental signals imposed in the biofilm environment are distinct from those attained by planktonic populations, under the conditions tested here.

3.4Influence of the toxins in long-term cell survival and antibiotic-triggered cell death

Because TA pairs have been linked to altruistic cell death and to the control of macromolecular biosynthesis during starvation, the long-term survival rates of the mutants and wild-type strains were investigated. Strains were incubated for up to 15 days, but no differences in the number of survivors were observed between strains, even when serine hydroxamate was added to the media to induce a stringent response (data not shown).

In E. coli, exposure to antibiotics that inhibit transcription or translation was shown to trigger MazF-mediated cell death [9]. Thus, we explored the effect of inhibitory concentrations of spectinomycin, an inhibitor of translation, on the survival of the wild-type strain and its derivatives in FMC media. The results indicated that, even after several days, the viability of the wild-type strain and the TJ-mazF, JL-relE and MRT1 derivatives was equivalent (data not shown). Moreover, contrary to data obtained with E. coli that showed dramatic decreases in the number of survivors after 1 h incubation with spectinomycin and other inhibitors of transcription and translation [9], the decrease in the number of viable cells in FMC containing spectinomycin did not differ from the control group incubated in FMC alone (data not shown). Similar results were obtained when the transcription inhibitor, rifampicin, was used (data not shown). These data indicate that, at least under the conditions tested, general inhibition of transcription or translation does not trigger cell death in S. mutans.

The concept of PCD in bacteria is relatively recent [28] and there is some controversy surrounding whether TA modules, in particular MazEF, mediate cell death or simply induce a reversible state of bacteriostasis [5,12–14,29,30]. The theory of PCD or “altruistic cell death” proposes that under nutrient starvation conditions, a large part of a bacterial population living in complex communities, such as biofilms, undergoes lysis to allow surviving cells to scavenge nutrients from dead siblings [3,13,31]. In the theory of bacteriostasis, TA modules simply down-regulate macromolecular synthesis during nutritional stress, and when environmental conditions become favorable again, cells resume normal growth [5,12,29,30]. The data presented here do not support that MazF mediates cell death in S. mutans. Still, it is important to note that there are fundamental differences in the physiology of S. mutans and E. coli, including that S. mutans allows major fluctuations in adenylate energy charge and intracellular pH [32,33]. Moreover, the environments occupied by these bacteria are dramatically different. S. mutans does not have a free-living lifestyle and exists overwhelmingly in densely populated oral biofilms. Although there are sustained periods where saliva is the primary nutrient for oral organisms, bacteria in the mouth do not undergo prolonged periods of starvation in severely oligotrophic conditions, such as those that can be encountered in the environment. For S. mutans, the need for altruistic PCD may have been obviated by evolution in an environment where sufficient nutrients are always present to afford survival of the population. Thus, it would not be surprising if the role of TA modules in these physiologically dissimilar organisms had diverged substantially through evolution. It also seems likely that, if bacterial TA pairs do mediate PCD, the systems could work in concert with other effectors that are activated through environmental signals that are not mimicked in specific in vitro systems, which may explain in part the differences in survival of the MRT1 mutants in biofilms versus planktonic culture.

3.5Role of MazF and RelE in diauxic growth

Diauxic growth is characterized by a transient growth arrest associated with the exhaustion of the preferred substrate and derepression of new enzymes needed to metabolize a non-preferred substrate. In BHI or TV glucose (0.5% w/v) broth, the mutant strains grew as well as the parent (data not shown). Growth curves of the wild-type and mutant strains in the presence of glucose and the non-preferred carbohydrate, inulin, revealed that the MRT1 mutant displayed a diauxie that was longer than the wild-type strain (Fig. 3). The TJ-mazF strain did not exhibit a longer diauxie, whereas JL-relE showed only a slightly longer diauxie than UA159 (data not shown). MRT1 also displayed a slightly longer diauxie in sorbitol (data not shown). The basis for the differences in the diauxic lag is not readily apparent at this time, although clearly TA modules do impact growth cessation or resumption of growth in this organism. Notably, when the activity of the bacterial sugar:phosphotransferase activity was measured in the MRT1 double mutant, the glucose, mannose and fructose PTS was markedly lower than in the wild-type (Fig. 4). Since the PTS is the major route for internalization of carbohydrate by S. mutans, this finding reveals a central role of TA modules in regulating, directly or indirectly, the major energy acquisition pathway of the organisms. Notably, the PTS also regulates the expression of non-preferred carbohydrate sources, including inulin [34,35], which may provide a molecular link between TA pairs and diauxic growth.

Figure 3.

Diauxic Growth of S. mutans UA159 and MRT1 strains. Symbols: UA159 wild-type (squares) and MRT1 strains (triangles). (A) Growth in TV containing 0.05% glucose and 0.5% inulin. (B). Glucose–inulin diauxie phase from A. The graphs are representative of multiple experiments.

Figure 4.

Glucose, fructose and mannose PTS activity of S. mutans UA159 and MRT1 strains grown in TV glucose. PTS activity was expressed as nanomoles of NADH oxidized in a PEP-dependent manner min−1 (milligram of protein)−1. The values are the mean ± standard deviations from three individual experiments.

4Concluding remarks

In summary, this is the first report that demonstrates physiological roles of the MazEF and RelBE pairs in Gram-positive bacteria. From diauxic growth and acid killing experiments, it appears that loss of a single toxin, particularly MazF, has little effect on S. mutans. However, in the absence of both toxins, acid tolerance and growth properties are clearly impacted. The effects of single and double mutations had similar effects on the PTS, but importantly, a direct linkage between the function of TA modules in the cell and regulation of the primary energy acquisition system of the organism was observed. These results may indicate that MazEF and RelBE are, at least in part, redundant systems for control of growth, although the RelE-deficient strain showed properties intermediate to UA159 and MRT1, and relBE transcription was acid-responsive. Thus, these TA modules may have both overlapping and unique regulatory circuits and roles in stress tolerance. During the preparation of this article, an extensive search for TA loci in completely sequenced bacterial genomes identified a putative solitary relE gene in the S. mutans UA159 chromosome [15]. Work is underway to investigate the function of this gene in S. mutans. It is possible that inactivation of both RelE copies will be necessary to thoroughly study pathways that are modulated by RelBE in S. mutans. In the future, identification of gene products or pathways that are affected by TA modules may provide important information regarding the ability of S. mutans to respond and adapt to fluctuations in nutrient availability and pH.


This work was supported by grant DE13239 from the NIDCR.