Redox Sensing Modulates the Activity of the ComE Response Regulator of Streptococcus mutans

ABSTRACT

Streptococcus mutans, a dental pathogen, encodes the ComDE two-component system comprised of a histidine kinase (ComD) and a response regulator (ComE). This system is necessary for production of bacteriocins and development of genetic competence. ComE interacts with its cognate promoters to activate the transcription of bacteriocin and competence-related genes. Previous transcriptomic studies indicated that expressions of bacteriocin genes were upregulated in the presence of oxygen. To understand the relationship between the aerobic condition and bacteriocin expression, we analyzed the S. mutans ComE sequence and its close homologs. Surprisingly, we noticed the presence of cysteine (Cys) residues located at positions 200 and 229, which are highly conserved among the ComE homologs. Here, we investigated the role of Cys residues of S. mutans ComE in the activation of bacteriocin transcription using the PnlmA promoter that expresses bacteriocin NlmA. We constructed both single mutants and double mutants by replacing the Cys residues with serine and performed complementation assays. We observed that the presence of Cys residues is essential for PnlmA activation. With purified ComE mutant proteins, we found that ComE double mutants displayed a nearly 2-fold lower association rate than wild-type ComE. Furthermore, 1-anilinonaphthalene-8-sulfonic acid (ANS) fluorescence studies indicated that the double mutants displayed wider conformation changes than wild-type ComE. Finally, we demonstrated that close streptococcal ComE homologs successfully activate the PnlmA expression in vivo. This is the first report suggesting that S. mutans ComE and its homologs can sense the oxidation status of the cell, a phenomenon similar to the AgrA system of Staphylococcus aureus but with different outcomes.

IMPORTANCE Streptococci are an important species that prefer to grow under anaerobic or microaerophilic environments. Studies have shown that streptococci growth in an aerobic environment generates oxidative stress responses by activating various defense systems, including production of antimicrobial peptides called bacteriocins. This study highlights the importance of a two-component response regulator (ComE) that senses the aerobic environment and induces bacteriocin production in Streptococcus mutans, a dental pathogen. We believe increased bacteriocin secretion under aerobic conditions is necessary for survival and colonization of S. mutans in the oral cavity by inhibiting other competing organisms. Redox sensing by response regulator might be a widespread phenomenon since two other ComE homologs from pathogenic streptococci that inhabit diverse environmental niches also perform a similar function.

INTRODUCTION

The microaerophilic bacterium Streptococcus mutans is an important constituent of the dental plaque on human teeth and is associated with dental caries formation. The organism has the ability to metabolize various carbohydrates to produce lactic acid and polysaccharides called glucans that help to adhere to the tooth surfaces to form dental plaque (1–3). The dental plaque biofilm contains over 700 different bacterial species where the cell density can be extremely high (4–7). Dental biofilms are constantly exposed to substantial fluxes of environmental conditions, including rapid changes in pH, availability of nutrients, nature of carbohydrates, and variation in redox potentials. These environmental parameters are the most important determining factors for the composition of the microbial population of the plaque biofilm (2, 3, 8). However, S. mutans has the ability to become a predominant species of the community since the organism can adapt, tolerate, and rapidly respond to the hostile environment at the molecular level for changes at the physiological and biochemical levels (9–11). The mechanisms by which S. mutans modulates acid production, adaptation to low pH, or carbohydrate acquisition are relatively well understood, but how S. mutans responds to redox changes is still under intense investigation (2, 3, 12–16).

The oral cavity is a predominantly oxygen-rich environment, and oxygen is utilized by many oral organisms for respiration and energy generation. Oxygen concentration in the oral biofilm fluctuates, and consequently, the biofilm is able to foster a broad range of aerobic, facultative, microaerophilic, and obligately anaerobic organisms (17). Bacteria that initially colonize the tooth or other oral surfaces are exposed to a higher concentration of oxygen, while the bacteria in the mature biofilms are exposed to smaller amounts due to restricted diffusion. The oxygen tension in the oral cavity varies between 5 to 27 mm Hg, but the redox potential of early biofilm is much higher than the mature biofilm (+294 mV versus −141 mV) (12, 17). It is noteworthy that oral streptococci, including S. mutans, do not harbor a complete electron transport chain and thus are unable to perform oxidative phosphorylation. However, these organisms possess an efficient system to metabolize oxygen predominantly with the NADH oxidase enzymes (12, 18, 19).

The effect of oxygen on the pathophysiology of S. mutans has been studied recently. It seems growth under aerobic conditions alters various key attributes of the organism. The transport of various sugars by the sugar-phosphotransferase system is greatly enhanced under aerobic conditions (15). Aeration also causes a decrease in the glycolytic rate and an increase in the production of intracellular storage polysaccharide (15). It has been observed that the ability to form biofilm is drastically reduced when the cells are grown under aerobic conditions (20). It has been demonstrated that synthesis of bacteriocins, ribosomally synthesized antimicrobial peptides, is greatly enhanced when S. mutans cells are grown under aerobic conditions (16). A genome-wide transcriptomic study has demonstrated that nearly 5% of the S. mutans genes are differentially expressed in response to aeration (14). Among the most enhanced genes were those that encode bacteriocin or bacteriocin-like peptides, autolysin-related genes, and some glucosyl transferase-related genes (14). The exact molecular mechanisms by which aeration modulates gene expression in S. mutans are not well understood.

Bacteria employ several transcription regulators to modulate gene expression in response to redox status of the cell. The most studied are the thiol-based redox switches. For example, the transcription factor OxyR of Escherichia coli contains two cysteine residues that are involved in formation of intramolecular disulfide bonds upon oxidation (21, 22). This oxidation makes OxyR activated, which, in turn, involves it in gene transcription. Similarly, Spx is another example where two redox-active cysteine residues convert the protein to an activator (22). S. mutans also encodes Spx regulators; however, it appears that they regulate very specific sets of genes with no overlap of the genes that are modulated during growth under aeration (23, 24).

Two-component signal transduction systems (TCS) are the major mechanisms by which bacteria sense and respond to the changes in environmental parameters such as oxygen tension, pH, ion concentrations, and others (25, 26). The TCS consists of a membrane-bound sensor kinase that facilitates the rapid detection of external signals (for reviews, see references 25–29). Upon detection of an appropriate signal, a conformational change occurs in the sensor kinase, which results in autophosphorylation of the protein. Typically, a conserved histidine residue in the sensor kinase receives a phosphoryl group from ATP, followed by transfer of the phosphoryl group from the kinase to the cognate response regulator. The response regulator is composed of two functional components, a receiver domain with a conserved aspartate residue that accepts the phosphoryl group and an effector domain that is activated upon phosphorylation of the aspartate residue. Phosphorylation of the response regulator alters its ability to interact with the target DNA sequence to activate or repress transcription of one or more target genes. Coordinated gene expression in response to environmental signals is particularly important for many human pathogens (30–32). S. mutans encodes at least 14 TCS that play important roles in bacterial adaptation, bacteriocin production, and biofilm formation (31, 33, 34). Of these, only three systems, CiaRH, LisRS, and VicRK, have been implicated in sensing oxygen tension in the cell (14, 20, 33).

We were interested in understanding the molecular basis behind the upregulation of bacteriocin-related gene expression under aerobic conditions. In S. mutans and related organisms, ComD/E TCS and its homologs BlpH/R are directly involved in the transcription of bacteriocin-related genes (35–38). The response regulator ComE encodes a LytTR DNA binding domain (39). ComE activates the gene expression by directly binding to the promoters of bacteriocin-related genes as well as its own promoter (35, 40–42). Upon comparing the ComE sequences with another well-characterized LytTR family response regulator, AgrA, from Staphylococcus aureus, we noticed that these response regulators contain two highly conserved cysteine residues that are present in the DNA binding domain (Fig. 1 and Fig. S1 in the supplemental material). Recently, Sun and colleagues have demonstrated that AgrA is directly involved in sensing the oxidizing environment (43). These researchers have shown that both conserved cysteine residues (C199 and C228) are involved in an intramolecular disulfide bond formation upon oxidation.

FIG 1.

Importance of cysteine residues of ComE for the activation of the nlmA promoter (PnlmA). (A) Schematic diagram showing the receiver domain and the DNA binding domain of S. mutans ComE (ComE-S. mutans). (B) Putative structure of full-length ComE, generated using I-TASSER, depicting a receiver domain (1 to 133 aa, blue) and a DNA binding domain (153 to 250 aa, green). Three cysteine residues located in the DNA binding domain at 173rd, 200th, and 229th positions are displayed as a stick model (magenta). The distances between cysteine residues are indicated. (C) Effects of ComE and its variants on the induction of PnlmA-gusA in vivo. β-Glucuronidase (Gus) assays of ΔcomE reporter strain (IBS1C15) carrying either wild-type ComE or ComE variants were performed as described in the text. Gus assay of UA159 reporter strain (IBS1C14) and ΔcomE reporter strain (IBS1C15) with empty vector pIB184Em were also performed. Gus activity was measured in Miller units (MU). Error bars indicate means ± standard deviation (SD) from three replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001; ns, not significant.

In this study, we investigated the importance of the conserved cysteine residues of ComE (C200 and C229) for the cognate promoter activation. We found that the presence of any of the two cysteine residues of ComE is essential for the induction of the promoter of the bacteriocin NlmA (PnlmA) in vivo under aerobic conditions. We also demonstrated that ComE homologs from other streptococci such as Streptococcus pneumoniae and Streptococcus gallolyticus are able to activate the PnlmA promoter in S. mutans and respond to aerobic conditions. This is the first report demonstrating that ComE and its homologs directly sense the redox status of the cell and modulate gene expression without involving sensor kinases. The molecular mechanisms by which ComE exerts its action upon oxygen sensing are different from the AgrA system.

RESULTS

Oxidized condition induces nlmA promoter, PnlmA.

Previous studies have demonstrated that the presence of oxygen leads to increased transcription of bacteriocin and bacteriocin-related genes in S. mutans (14, 16). To study the molecular mechanism behind this observation, we selected the two-peptide bacteriocin (mutacin IV) composed of NlmA and NlmB encoded by the nlmAB operon. We used a previously constructed reporter fusion where the entire promoter region of nlmAB, including the first few codons of the nlmA open reading frame (ORF), was fused to a gusA reporter gene to create a transcriptional fusion, PnlmA-gusA. This PnlmA-gusA reporter fusion was then integrated at an ectopic location on the chromosomes of the wild-type UA159 strain and a clean comE-deleted strain (ΔcomE) to generate IBS1C14 and IBS1C15, respectively. We performed β-glucuronidase (Gus) assays to first compare the effect of static and aerobic growth conditions on PnlmA expression. We found that aerobically grown cultures induce PnlmA expression nearly 2-fold compared to statically grown cultures (Fig. 1C, UA159), confirming that oxygen indeed upregulates bacteriocin gene expression. When we used the ΔcomE reporter strain IBS1C15 containing an empty vector, pIB184Em, no measurable Gus activity was noticed (Fig. 1C, vector).

S. mutans ComE and its close homologs contain two highly conserved cysteine (C, Cys) residues in the DNA binding domain (Fig. 1A and B and Fig. S1 in the supplemental material). S. mutans ComE also contains an additional Cys residue at position 173. To study the role of these three Cys residues, we first constructed single mutants where Cys residues were replaced with serine (S) to generate three mutants, C173S, C200S, and C229S. The mutant ComE constructs were cloned under a constitutive promoter P23 in a shuttle plasmid, pIB184Em (44). The mutants were then introduced into the IBS1C15 reporter strain and assayed for Gus activity at the mid-logarithmic growth phase under both static and aerobic conditions. As the positive control, we also introduced wild-type ComE into the ΔcomE reporter strain. We found that the PnlmA expression was nearly two times more in IBS1C15 with ComE expressing from the plasmid than the IBS1C14 strain (with the native comE locus) under static conditions (Fig. 1C). We also found that IBS1C15 carrying the wild-type (WT) ComE plasmid displayed increased Gus activity when grown under aerobic conditions compared to static growth conditions (Fig. 1C, ComE [WT]). We then tested single Cys variants of ComE (C173S, C200S, and C229S) in the IBS1C15 strain. As shown in Fig. 1C, all the single Cys mutants were able to induce PnlmA expression under both static and aerobic growth conditions, suggesting they could interact and activate PnlmA expression. Overall, the Gus activity levels were very similar to the IBS1C14 strain. Importantly, all the single Cys mutants showed increased PnlmA expression when grown under aerobic conditions, indicating they are able to sense oxidation status of the cell similar to the wild-type ComE, but with ∼30 to 60% and ∼22 to 32% reduced capacity under static and aerobic conditions, respectively.

However, when the IBS1C15 strains were complemented with double Cys mutants such as C200S/C229S, C173S/C229S, and C173S/C200S, they induced PnlmA expression at ∼70 to 90% lower level, suggesting that at least any of the two Cys residues are required for efficient PnlmA expression; the presence of a single Cys residue is not sufficient for the efficient activation. Importantly, these double mutants were responsive to oxygen.

Formation of ComE dimers by an intermolecular disulfide bond.

Previous studies have shown that proteins with cysteine residues can form intra- or intermolecular disulfide bonds under oxidizing conditions, and these can be visualized on a nonreducing SDS-PAGE (45, 46). We wanted to investigate the involvement of the cysteine residues of ComE for both intermolecular and intramolecular disulfide bond formation. We purified the ComE and its variants from E. coli and incubated them with or without dithiothreitol (DTT) for 30 min at room temperature before electrophoresis on a 13% nonreducing SDS-PAGE (see Materials and Methods for details). As shown in Fig. 2, we observed various species on the gel. For the wild-type ComE, when DTT was absent, we noticed three species: a predominant band of about 31 kDa that corresponds to reduced monomer, a faster-migrating band that is less than 31 kDa, which we believe to be a monomeric species with intramolecular disulfide bonds, and a minor slow-migrating species of about 62 kDa (Fig. 2, ComE [WT]). The oxidized monomeric form is more compact since it has intramolecular disulfide bonds, and unlike the reduced form, it migrated faster. In the presence of DTT, only the reduced monomeric species was observed. When we used variant C173S, the predominant species was the faster-migrating monomeric species with intramolecular disulfide bonds in the absence of DTT. We also observed two different dimeric species, presumably with different combinations of intermolecular disulfide bond formation. For both C200S and C229S variants, under nonreducing conditions, the faster-migrating monomeric species was absent, and, instead, various dimeric species were observed (Fig. 2). When we used the double cysteine variants (C200S/C229S, C173S/C200S, and C173S/C229S), the results were very similar to the C200S and C229S single mutants. Under the reducing conditions, all the variants formed reduced monomeric species similar to wild-type ComE. We also tested the effect of the addition of oxidizing agents H₂O₂ and 2,2′ dithiodipyridine on wild-type ComE and its variants for disulfide bond formation. The results were similar to the nonreducing conditions with the presence of more monomeric oxidized species for wild-type ComE and C173S and more dimeric and oligomeric species for ComE and all the single and double mutants (data not shown).

FIG 2.

Detection of cysteine-mediated oligomerization of purified ComE and its variants. (A) ComE and its variants were incubated with (+) or without (−) 2 mM DTT for 1 h at room temperature. Samples were loaded on a 13% nonreducing (no DTT) medium-size (17 cm by 15 cm) SDS-PAGE. After electrophoresis, the gel was stained with PageBlue Coomassie R250. The experiment was performed twice, and a representative gel is shown.

Taken together, these results indicate that an intramolecular disulfide bond is formed between the C200 and C229 residues of ComE, and intermolecular bonds are formed involving all the cysteine residues.

Binding of ComE and its variants to the nlmA promoter.

To understand the DNA binding activity of ComE and its variants, we used biolayer interferometry (BLI). We PCR generated the PnlmA fragment that is ∼500 bp long carrying a biotin tag at the 5′ end. This biotinylated-PnlmA fragment was then immobilized on a streptavidin-coated biochip. We first measured the association rate constant for ComE and its variants in the presence (Fig. 3A) or absence of DTT (Fig. 3B). As shown in Fig. 3C, the association rate constant (K_a) value for wild-type ComE was 1.3 ± 0.1 under reducing conditions. Under nonreducing conditions, the value was 1.7 ± 0.1, which is significantly higher (P < 0.05) than the reducing condition.

FIG 3.

Binding of ComE and its variants to biotinylated-PnlmA promoter using biolayer interferometry. A PCR-generated PnlmA fragment that was tagged with biotin at the 5′ end was loaded onto streptavidin biosensor tip. The binding of the biotinylated-PnlmA fragment with ComE and its variants was observed under reduced (2 mM DTT) (A) or oxidized conditions (0 mM DTT) (B). The association reaction was performed for 5 min at room temperature. A representative graph from three independent sets is shown. (C) The association rate constant (K_a) of ComE and its variants to the biotinylated-PnlmA fragment was plotted. Error bars indicate means ± SD. Statistical analysis was performed using the unpaired t test. *, P < 0.05; ***, P < 0.001.

The K_a values for the single variants C173S and C200S were slightly higher than the wild-type ComE protein (Fig. 3C). However, under nonreducing conditions, binding was significantly increased for the C173S and C200S mutants. The K_a values for C229S were similar to the wild type under both reducing and nonreducing conditions. This suggests that intramolecular disulfide bond formation in ComE does not affect binding affinity of ComE to PnlmA. The increased binding affinity could be due to the formation of dimers by an intermolecular disulfide bond. When we used the double mutants (C200S/C229S, C173S/C200S, and C173S/C229S) for the binding assay, we found that the rate constant was 2-fold lower than the wild-type ComE (Fig. 4). The results indicate that the presence of a single cysteine residue in ComE is not sufficient for strong binding with PnlmA.

FIG 4.

(A) Binding of double cysteine variants of ComE to biotinylated PnlmA using biolayer interferometry. PnlmA tagged with biotin at 5′ end was loaded on streptavidin sensor tip. The binding of biotinylated-PnlmA with 1 µM variants was observed under reduced (with 2 mM DTT) conditions. A representative graph of three independent sets is shown. (B) The association rate constant (K_a) of double cysteine variants of ComE to biotinylated PnlmA was plotted. Error bars indicate means ± SD from three independent experiments.

Determination of conformational changes of ComE and its variants.

Previous studies have shown that 1-anilinonaphthalene-8-sulfonic acid (ANS) fluorescence intensity is enhanced upon binding to hydrophobic surfaces of proteins (47–49). ANS fluorescence depends on ion pairing with positively charged side chains, a variation of solvent polarity as well as viscosity. The emission of ANS depends on two distinct excited states. The first step of the excitation event is the nonpolar (NP) state localized on the naphthalene moiety of ANS. The wavelength of fluorescence maximum of ANS from NP state moderately depends on polarity. The NP state relaxes to form the intramolecular charge transfer state (ICT) depending on the solvent that is stabilized through molecular distortion and solvent relaxation. Intermolecular electron transfer (ET), ionization, and subsequent electron salvation are detected in the aqueous solution of ANS. The ET process serves as an efficient mechanism for decay from the CT state, which explains the low-fluorescence quantum yield of ANS.

We used the fluorescence of ANS to monitor the conformational changes of ComE induced by the cysteine substitutions under both reducing and nonreducing conditions. We found that free ANS showed a weak fluorescence, and ComE without ANS did not exhibit any fluorescence (Fig. 5 A and B; Fig. S2A). Interestingly, under reducing conditions, a 2-fold increase in the fluorescence intensity was observed upon binding of ANS to the hydrophobic surface of ComE (Fig. 5A). We also observed that the emission maxima of ANS shifted from 550 nm (ANS only) to 500 nm (ANS-ComE complex), suggesting strong binding of ANS to the ComE protein. Furthermore, the subtracted fluorescence intensities (ΔF) at 500 nm for ComE and its variants were compared (Fig. 5C). The ΔF value of C173S was slightly higher than ComE, indicating that the mutant has slightly wider binding pockets for ANS access. In contrast, the ΔF values of ComE C200S and ComE C229S were 3 to 6 times higher than the wild-type ComE protein, indicating much wider hydrophobic pockets in these mutants. When the fluorescence measurement was done under nonreducing conditions, the wild-type ComE displayed a ΔF value that is 4-fold higher than reducing conditions (Fig. 5C, ComE/WT). Similarly, the single Cys variants also displayed increased ΔF values under nonreducing conditions compared to reducing conditions. Among these Cys variants, C200S displayed the greatest change, with a difference of nearly 4-fold, and C229S displayed the lowest change, with a difference of only 1.4-fold between reducing and nonreducing conditions (Fig. 5C). The C173S variant displayed intermediate changes (2.0-fold) and fell between the C200S and C229S variants. Taken together, these results suggest that ComE and its Cys mutants showed more hydrophobic surfaces under nonreducing conditions than under the reducing conditions.

FIG 5.

Determination of conformational changes of purified ComE and its variants. ComE and its variants were incubated with 2 mM DTT (A) or without DTT (B) for 1 h at room temperature. ANS was added into each reaction mixture and incubated for 10 min at room temperature. Fluorescence spectra of the ComE-ANS complexes were monitored using the excitation wavelength of 360 nm and emission range of 400 to 600 nm. Fluorescence spectra of free ANS were also monitored. The experiment was performed in triplicate. A representative graph is shown. (C) Fluorescence intensity of free ANS at 500 nm was subtracted from the fluorescence intensities of ANS complexed with ComE and its variants. The subtracted fluorescence intensities (ΔF) of ComE and its variants are plotted. Error bars indicate means ± SD. Statistical analysis was performed using the unpaired t test using GraphPad (***, P < 0.001).

When we analyzed the double Cys mutants, the ΔF values for C200S/229S, C173S/C200S, and C173S/C229S were nearly 10 times higher than ComE under reducing conditions (Fig. 5C; Fig. S2A and B). These data suggest that these mutants expose a wide area of the hydrophobic surfaces where ANS could bind. Overall, these results also indicated that the presence of a single cysteine residue is not enough for proper folding and requires at least two cysteine residues, especially C200 and C229, for stabilization of its tertiary structure.

Activation of the PnlmA promoter by ComE homologs.

The S. mutans ComE is a highly conserved response regulator, and its homologs (called BlpR) are present in several streptococcal and other species (Fig. 6). Most of these homologs contain the key cysteine residues in the DNA binding domains similar to S. mutans ComE and the Asp residue in the receiver domain. We selected Streptococcus sobrinus, which belongs to mutans streptococci, S. gallolyticus, S. pneumoniae, and S. aureus AgrA. These homologs show between 55% (AgrA) to over 70% (S. pneumoniae) sequence similarities. We cloned the genes encoding the respective ComE homologs under the constitutive P23 promoter and assayed for PnlmA activation in the IBS1C15 (ΔcomE) strain to study whether these homologs could induce PnlmA activation under static and aeration conditions. As shown in Fig. 7, we observed that only two homologs from S. pneumoniae and S. gallolyticus could efficiently induce PnlmA activation. Both the S. pneumoniae and S. gallolyticus homologs responded positively to aeration since the PnlmA activation was increased when oxygen was present. We found that the homolog from S. pneumoniae activates the PnlmA promoter ∼1.5-fold higher than the S. mutans ComE (Miller units [MU], 50 versus MU 71; S. mutans versus S. pneumoniae, respectively). On the other hand, the overall activity of the homolog from S. gallolyticus was lower than ComE (MU 50 versus MU 34; S. mutans versus S. gallolyticus, respectively). In contrast, AgrA and the homolog from S. sobrinus, despite belonging to a mutans group with high sequence similarity to ComE, did not activate PnlmA (Fig. 7). To better understand the reason for the failure to complement, we reanalyzed the amino acid sequences and identified several conserved residues that were absent in both S. sobrinus BlpR and AgrA. Among them, the most important one is at position 198, where the cysteine residue was replaced with valine. The other notable differences were serine at 37, glutamic acid at 53, tyrosine at 54, glycine at 174, and tyrosine at 178 positions in S. sobrinus BlpR. These residues are somewhat conserved among the homologs. We replaced these residues with appropriate changes to construct S. sobrinus BlpR variants with S37F, Q53H/T54Q, G174T/Y178R, and V198C and tested their ability to activate the PnlmA promoter. We found that none of these S. sobrinus BlpR variants were able to activate the promoter (the Gus values were similar to the WT S. sobrinus; data not shown). Taken together, our data suggest that some streptococci ComE homologs can efficiently induce PnlmA expression and the cysteine residues function similar to S. mutans ComE, whiles other are unable to activate.

FIG 6.

Multiple-sequence alignment of ComE-S. mutans and its homologs. (A) Multiple-sequence alignment of ComE-S. mutans and its homologs from various streptococcal species and S. aureus was performed using Clustal Omega. Conserved residues were highlighted using the BoxShade server. Black and gray boxes represent identical and similar amino acid residues, respectively. Cysteine residues of ComE-S. mutans C173, C200, and C229 are marked by C1, C2, and C*, respectively. Protein sequences were obtained from GenPept. Accession numbers of BlpR are as follows: S. aureus, AKB00121.1; S. sobrinus, AWN61517.1; S. mutans, AAN59528.1; S. gallolyticus subsp. gallolyticus, EFM28803.1; S. pneumoniae, AAK99267.1; Streptococcus mitis, CBJ21637.1; and S. constellatus, AGU74171.1. Accession numbers for ComE are S. pneumoniae, WP_000866065.1; S. mitis, CBJ23317.1; and S. constellatus, AGU75413.1.

FIG 7.

Activation of PnlmA by ComE homologs. The ΔcomE reporter strain (IBS1C15) was complemented with pIB184Em plasmids carrying ComE or its homologs from S. mutans (ComE-SMU), S. pneumoniae (BlpR-SPN), S. gallolyticus (BlpR-SGA), S. sobrinus (BlpR-SOB), or S. aureus AgrA (AgrA-SAU). The cultures were grown in THY medium at 37°C under static or aerobic conditions, and Gus activities were measured as described in Materials and Methods. Error bars indicate means ± standard deviation (SD) from four replicates. Statistical analysis was performed using the unpaired t test. **, P < 0.01; ***, P < 0.001; ns, not significant.

DISCUSSION

Bacteria employ several cysteine-based systems to sense and regulate gene expression in response to alteration in oxygen tension and reactive oxygen species (ROS). Most of the well-studied transcription factors contain cysteine residues that are involved in intramolecular disulfide bond formations such as OxyR of E. coli (21). These transcription factors are mostly standalone regulators (22). Sensing redox status by two-component systems (TCS) is not very common. However, a few studies have shown that the sensor kinases containing redox active cysteine residues are involved in either intermolecular or intramolecular disulfide bond formation to modulate the activity. For example, the RegB sensor kinase, which is present in both Gram-negative and Gram-positive bacteria, is involved in modulating gene expression in response to changes in oxygen tension in the cell (50). Similarly, the SrrB sensor kinase of S. aureus contains two cysteine residues in the catalytic domain that form intramolecular disulfide bonds. The kinase activity of SrrB is highly sensitive to the redox status of the cysteine residues (51). As mentioned previously, in S. aureus, the well-studied agr TCS also responds to redox status, but it involves the AgrA response regulator that contains two cysteine residues (43). Recently, in S. pneumoniae, another orphan response regulator, RitR, which contains a single cysteine residue, responds to redox status by creating an intermolecular disulfide bond to form oligomers (52). In this study, we discovered a new group of well-conserved response regulators, ComE, which contain two or more cysteine residues that respond to redox status by forming intermolecular disulfide bonds to generate oligomers. This ComE system appears to be functionally different from both AgrA and RitR that sense oxygen tension in the cell.

Based on our oligomerization study (Fig. 2), we believe both intermolecular and intramolecular disulfide bonds are formed in ComE. We found that when C173 is absent, both C200 and C229 were engaged in forming both intermolecular and intramolecular disulfide bonds. When C200 is absent, C229 was more active than C173, and we found more C229 dimers (Fig. 2, C200S). When C229 is absent, it appears that C173 was more active than C200, and we observed more formation of C173 dimers than C200 dimers. The biological relevance of the C173 residue is not entirely clear since our sequence analysis indicated that the C173 residue is only present in S. mutans and not in other streptococci. However, when we searched the NCBI database containing nearly 200 strains, we found that the C173 residue is highly conserved among S. mutans ComE. Our modeling analysis (Fig. 1B) indicated that C173 resides about 13 Å apart from the C200 and 18 Å from the C229 residues. Thus, it is unlikely that C173 is involved in intramolecular disulfide bond formation. This is supported by our observation that both the wild-type and C173 mutant displayed oxidized monomeric species, but not the C200 and C229 mutants (Fig. 2).

AgrA functions both as an activator and a repressor for virulence genes (53, 54). The sensing of redox status in AgrA is obtained by two highly conserved cysteine residues at positions C199 and C228 (corresponding to C200 and C229 of ComE) (Fig. 6) that create an intramolecular disulfide bond formation upon oxidation. This oxidized AgrA dissociates from the promoters, leading to downregulation of RNAIII and upregulation of a glutathione peroxide gene. AgrA also contains additional cysteine residues at positions 6, 55, and 123, but these residues are not involved in redox sensing. It is important to point out that the DNA binding studies of AgrA were performed using only the effector domain and not the full-length protein. In contrast to AgrA, our data suggest and indicate that intramolecular disulfide bond formation may not be as important as intermolecular bond formation. This is because all the single ComE mutants were biologically functional and were able to complement the ΔcomE strain (Fig. 1C). Since ComE functions as an activator, we speculate that under oxidative conditions, enhanced or stable dimer (or oligomer) formation by disulfide bonding led to increased gene transcription.

Both AgrA and ComE contain the LytTR DNA binding domain instead of helix-turn-helix or other motifs that are more common (55, 56). Only less than 3% of all prokaryotic response regulators contain the LytTR domain (57). The structure of the LytTR DNA binding domain has been solved for AgrA (58). The DNA binding domain is comprised of a 10-stranded β fold with three short loops. The DNA binding is accomplished by residues present in the loops between the β sheets (H169, N201, and R233). From the structure, it appears that the two cysteine residues are flanked by N201 and R233 residues that make critical contact with the DNA. Intramolecular bond formation between these two cysteine residues, which are present in the β6 and β10 sheets, presumably disrupts the contact with the DNA. However, the structure does not predict how a dimer (or oligomer) interacts with the DNA since the structure of only the LytTR domain and not the entire AgrA was determined. We speculate that intermolecular disulfide bond formation between two ComE monomers does not interfere with the DNA binding but, rather, that it stabilizes the interaction. The mode of action of ComE is fundamentally different from AgrA; one is activated and the other one is deactivated upon oxidation; thus, new structural determination of ComE is necessary to understand the exact molecular mechanism. Based on our results, we propose a model for how ComE and its homologs function in response to aerobic conditions (Fig. 8). The main difference between the AgrA and ComE homologs is that in ComE under oxygen-rich conditions, intermolecular disulfide bond formation leads to stable dimer formation that activates the target promoter.

FIG 8.

Model for the regulation of ComE under reducing and oxidizing conditions. Under reducing conditions, ComE (WT) and its Cys variants are available in monomeric form, shown as oval and circle, respectively. Cysteine residues are shown as blue dots. Although the single Cys variants have slightly wider conformation than ComE (WT), they could interact and activate cognate promoters. This activation leads to production of bacteriocins and competence-related proteins, shown as purple and light blue circles, respectively. The double Cys variants (shown as heptagonal) have a wider conformation, so they bind weakly to the cognate promoters, resulting in less protein production. Under oxidizing conditions, ComE (WT) and single Cys variants form dimers using intermolecular disulfide bonds. These dimers bind strongly to the cognate promoters and thus increase the gene transcription. In contrast, the dimers formed by double Cys mutants are unable to enhance transcription compared to ComE (WT) and single Cys mutants due to the presence of wider hydrophobic pockets in their monomeric form.

We found that two heterologous ComE homologs (BlpRH) from S. gallolyticus and S. pneumoniae were able to complement the S. mutans ΔcomE strain. Furthermore, these two ComE homologs also responded in a similar way under aerobic conditions. However, the degree of induction, although statistically significant, was much less than the S. mutans ComE protein. The comED TCS system is under the positive feedback loop where the promoter of comE is regulated by ComE in response to cell density and competence stimulating peptide (CSP) concentration. In our assay, we used a plasmid-based constitutive P23 promoter for the expression of ComE to maintain constant amounts of proteins (ComE and its variants) in the cell. When we used the wild-type strain (IBS1C14), the difference of PnlmA expression in the presence or absence of oxygen was much higher (Fig. 1C). We believe that the BlpRH pathway in S. gallolyticus and S. pneumoniae responds to oxygen in a similar manner to the S. mutans ComED pathway.

The DNA binding residues are poorly conserved among the LytTR family transcriptional regulators (39). When we compared the key residues of AgrA with the ComE and its homologs, we found that two of the three residues (H169 and N201) are highly conserved among streptococci but not the R229 (Fig. 6) Yet the ComE homolog BlpR from S. sobrinus was unable to complement the S. mutans ΔcomE strain. We identified a few highly conserved residues that are absent in S. sobrinus BlpR. However, when we changed the residues to conserved residues, it did not restore the activity. The ComE binds to two direct repeats of sequence containing the TCBTAAAYSGT motif, which is present in PnlmA and other ComE-regulated promoters (41). It is possible that BlpR of S. sobrinus binds to very different DNA consensus sequences and is unable to recognize the PnlmA promoter. The same is probably true for AgrA, which also recognizes a different conserved binding sequence (54).

The LytTR domain-containing regulators are typically found in only a few proteins per genome (59). However, the S. mutans genome encodes seven putative LytTR domains containing proteins, including ComE (60). None of the LytTR family regulators, except ComE, encode the double cysteine residues that we studied here. In that sense, ComE is the only LytTR protein in S. mutans that responds to the oxygen status of the cell. Whether these other LytTR regulators respond to oxygen tension by some unknown mechanism remains to be evaluated. However, S. pneumoniae and other mitis groups of streptococci encode at least two response regulators (called BlpR and ComE) containing both a LytTR domain and the conserved double cysteine residues (Fig. 6). The BlpR regulates bacteriocin production, whereas ComE is involved in competence development in S. pneumoniae (61, 62). The role of these conserved cysteine residues in the pathophysiology of S. pneumoniae during changes in oxygen status remains to be evaluated.

Streptococci are microaerophilic but catalase-negative organisms. While they can tolerate the presence of oxygen, their level of tolerance is much lower than the aerophilic organisms such as S. aureus. Streptococci are abundantly present in various human niches, including the oral cavity, where nearly 30% of the 700 different species are streptococci (7). Some commensal oral streptococci, such as S. gordonii and S. sanguinis, produce hydrogen peroxide to inhibit competing bacteria, including S. mutans (16). We speculate that S. mutans employs a different strategy to compete with the niche-specific organisms. It utilizes the ComE/D pathway to sense the oxygen status and increase the secretion of numerous bacteriocins to inhibit the competing bacteria. We analyzed the available genomes of S. mutans and the sanguinis group of streptococci (S. cristatus, S. gordonii, and S. sanguinis) for the presence of bacteriocin genes by BAGEL4 (63). It appears that the S. sanguinis group generally lacks bacteriocin-encoding genes, whereas S. mutans genomes are highly enriched with the bacteriocin-encoding genes. The other streptococci probably utilize the ComE/D or similar LytTR system for other physiological functions, such as increased competence development, to take up heterologous DNA for adaption purposes.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

Bacterial strains and plasmids used in this study are listed in Table 1. E. coli DH5α or E. coli BL21(DE3) cells were grown in Luria-Bertani (LB) medium with appropriate antibiotics such as 100 μg/ml ampicillin (Ap) and 500 μg/ml erythromycin (Em). E. coli cells were grown at 37°C in shaking conditions. Streptococcal strains were grown in Todd-Hewitt medium with 0.2% yeast extract (THY medium) containing 300 μg/ml kanamycin (Km) and 10 μg/ml Em whenever required. For transformation, S. mutans cells were grown until the optical density at 600 nm (OD₆₀₀) reached 0.2. Then, CSP21 peptide (400 nM) was added into the culture, followed by 10 min incubation. Afterward, 500 ng of DNA was added to the culture and incubated for 2 h. Cultures were spread on THY agar plates with appropriate antibiotics and incubated at 37°C in a candle jar. For assaying promoter activity, cells were incubated at 37°C in a candle jar for microaerophilic (static) conditions or with shaking at 140 rpm for aerobic growth conditions.

TABLE 1.

List of strains and plasmids used in this study

Strain or plasmid	Description	Reference or source
Strains
UA159	Wild type	26
IBS1C14	UA159::PnlmA-gusA reporter, Kmr	This study
IBSN39	UA159::ΔcomE (markerless deletion of comE)	This study
IBS1C15	ΔcomE::PnlmA-gusA reporter, Kmr	This study
IBS1C18	IBS1C15/pIB1B68, complemented with comE, Emr	This study
IBS1C20	IBS1C15/pIB184Em, ΔcomE reporter strain bearing plasmid pIB184Em, Emr	This study
IBS1C21	IBS1C15/pIB1B80, complemented with comE C173S, Emr	This study
IBS1C22	IBS1C15/pIB1B81, complemented with comE C200S, Emr	This study
IBS1C23	IBS1C15/pIB1B82, complemented with comE C229S, Emr	This study
IBS1C24	IBS1C15/pIB1B83, complemented with comE C200S/C229S, Emr	This study
IBS1C26	IBS1C15/pIB1B94, complemented with comE C173S/C229S, Emr	This study
IBS1C27	IBS1C15/pIB1B95, complemented with comE C173S/C200S, Emr	This study
S. gallolyticus TX20005	Wild type	ATCC
S. sobrinus ATCC 27352	Wild type	ATCC
S. pneumoniae TCH8431	Wild type	BEI
IBS1C19	IBS1C15/pIB1B37, complemented with S. pneumoniae blpR, Emr	This study
IBS1C33	IBS1C15/pIB1E1, complemented with S. gallolyticus blpR, Emr	This study
IBS1C34	IBS1C15/pIB1E2, complemented with S. sobrinus blpR, Emr	This study
IBS1C35	IBS1C15/pIB1B99, complemented with S. aureus agrA, Emr	This study
E. coli DH5α	Cloning strain	Lab stock
E. coli BL21(DE3)	Protein expression with T7 promoter	Lab stock
Plasmids
pIB184Em	E. coli streptococcal shuttle plasmid, Emr	27
pET15b	Expression plasmid, Apr	Novagen
PIB1B2	pET15b with His-ComE, Apr	This study
PIB1B61	pET15b with His-ComE C173S, Apr	This study
PIB1B62	pET15b with His-ComE C200S, Apr	This study
PIB1B63	pET15b with His-ComE C229S, Apr	This study
PIB1B65	pET15b with His-ComE C200S/C229S, Apr	This study
PIB1B85	pET15b with His-ComE C173S/C200S, Apr	This study
PIB1B86	pET15b with His-ComE C173S/C229S, Apr	This study
pIB1B68	pIB184Em with ComE, Emr	This study
pIB1B80	pIB184Em with ComE C173S, Emr	This study
pIB1B81	pIB184Em with ComE C200S, Emr	This study
pIB1B82	pIB184Em with ComE C229S, Emr	This study
pIB1B83	pIB184Em with ComE C200S/C229S, Emr	This study
pIB1B94	pIB184Em with ComE C173S/C229S, Emr	This study
pIB1B95	pIB184Em with ComE C173S/C200S, Emr	This study
PIB1B37	pIB184Em with S. pneumoniae BlpR, Emr	This study
pIB1B99	pIB184Em with S. aureus AgrA, Emr	This study
PIB1E1	pIB184Em with S. gallolyticus BlpR, Emr	This study
PIB1E2	pIB184Em with S. sobrinus BlpR, Emr	This study

Construction of comE variants for purification and complementation.

For purification purposes, the wild-type comE gene was PCR amplified using primers NdeI-comE-F and XhoI-comE-R (for all the primer sequences, please see Table 2). The amplified PCR product was cloned into NdeI-XhoI-digested vector pET15b to generate plasmid pIB1B2. The sequence of comE in pIB1B2 plasmid was confirmed by DNA sequencing. To generate single cysteine variants of comE by site-directed mutagenesis, appropriate complementary forward and reverse primers were designed for each comE variant as listed in Table 2. Plasmid pIB1B2 was used as the template for PCR amplification. The amplified products were then digested with DpnI and transformed into E. coli DH5a cells. The isolated plasmids were confirmed by DNA sequencing and named pIB1B61 (C173S), pIB1B62 (C200S), and pIB1B63 (C229S). A similar method was used for generating double cysteine variants of comE but with various templates carrying appropriate single cysteine variants. The double cysteine variants were also confirmed by DNA sequencing and named pIB1B65 (C200S/C229S), pIB1B85 (C173S/C200S), and pIB1B86 (C173S/C229S).

TABLE 2.

List of primers used in this study

Primer name	Sequence (5′–3′)
BamHI-rbsComE-F	CGCGGATCCAAGAAGGAGGATATACAAATGATTTCTATTTTTGTATTGG
XhoI-ComE-R	CCGCTCGAGTCATTTTGCTCTCCTTTGATCAGCAATC
NdeI-ComE-F	CCGCATATGATTTCTATTTTTGTATTGGAAGATGAT
ComE-C173S-F	CAACAGCCCATAAGCTCAGCCTTTATACTTATGATG
ComE-C173S-R	CATCATAAGTATAAAGGCTGAGCTTATGGGCTGTTG
ComE-C200S-F	TGGATAAGAGACTTTTTCAGAGCCATCGCTCTTTTATTGTC
ComE-C200S-R	GACAATAAAAGAGCGATGGCTCTGAAAAAGTCTCTTATCCA
ComE-C229S-F	CGAAATAATAAGTCTAGTCTTATTTCACGAAC
ComE-C229S-R	GTTCGTGAAATAAGACTAGACTTATTATTTCG
BamHI-Sobrinus BlpR-F	CGCGGATCCAAGAAGGAGGATATACAAATGTTATATATCTATGCC
XhoI-Sobrinus BlpR-R	CCGCTCGAGTTACCTCCTATGTAATTTCTTCAAAGCATTGGCTAAGGG
BamHI-Gallolyticus BlpR-F	CGCGGATCCAAGAAGGAGGATATACAAATGTTAGATATTTATGTACTAGAGG
XhoI-Gallolyticus BlpR-R	CCGCTCGAGTCACCTCATCTCATTTAATAGAGG
BamHI-Aureus AgrA-F	CGCGGATCCAAGAAGGAGGATATACAAATGAAAATTTTCATTTGCG
XhoI-Aureus AgrA-R	CCGCTCGAGTTATATTTTTTTAACGTTTCTCACCG
pnlmA-F	TACAAATATGGCAATCGAAG
Biotin-pnlmA-F	TACAAATATGGCAATCGAAG
pnlmA-R	TCAAATGCCTGTGTATCCAT

For complementation studies, wild-type comE and its variants were PCR amplified using primers BamHI-rbscomE-F and XhoI-comE-R (Table 2) and appropriate pET15b-derivative plasmids as the templates. The amplified products were digested with BamHI and XhoI and ligated into BamHI-XhoI-digested pIB184Em plasmid to generate plasmids pIB1B68 (ComE), pIB1B80 (C173S), pIB1B81 (C200S), pIB1B82 (C229S), pIB1B83 (C200S/C229S), pIB1B94 (C173S/C229S), and pIB1B95 (C173S/C200S), which were transformed into a ΔcomE reporter strain (IBS1C15) and named IBS1C18, IBS1C21, IBS1C22, IBS1C23, IBS1C24, IBS1C26, and IBS1C27, respectively.

β-Glucuronidase assay.

Gus assays were performed to measure the effect of the aerobic environment on the induction of PnlmA. Briefly, overnight grown cells were diluted to 1:20 and grown until the OD₆₀₀ reached ≈0.6. Cells were grown at 37°C in two different conditions. The first one is a partial anaerobic environment where a 15-ml culture tube with 12 ml culture medium was incubated under tightly capped conditions, and the second one is an aerobic environment where 250-ml flask with 50 ml medium was incubated under shaking at 140 rpm. Once the OD₆₀₀ of cultures in partial anaerobic and aerobic environments reached ≈0.6 (after 4 to 5 h and 6 to 8 h, respectively), the cultures were harvested and washed three times with phosphate-buffered saline (PBS). Samples were stored at −20°C. Pellets were resuspended in 760 μl of Z buffer (60 mM Na₂HPO₄, 40 mM NaH₂PO₄, 10 mM KCl, 1 mM MgSO₄, and 20 mM DTT) with 40 μl of freshly prepared 10 mg/ml lysozyme and 8 μl of 10% Triton X-100. The reaction mixture was incubated at 37°C for 30 min followed by the addition of 200 μl of p-nitrophenyl-d-glucoside (PNPG) (4 mg/ml in Z buffer) and further incubated at 37°C. After developing a yellow color, 400 μl of 1 M Na₂CO₃ was added to the reaction mixture to arrest the reaction, and the time was noted. The reaction mixture was centrifuged, and the absorbance of the supernatant at 420 nm was measured. Gus activity was calculated as (1,000 × OD₄₂₀)/(time in min × OD₆₀₀) in Miller units (21, 22). The experiment was performed in triplicate.

Purification of ComE protein.

Plasmid pET15b derivatives containing wild-type ComE and its variants with histidine tag were transformed into E. coli Bl21 cells. For protein expression, cells were grown in 2× yeast extract with tryptone (2×YT) medium containing 100 μg/ml ampicillin at 37°C with shaking at 200 rpm until the OD₆₀₀ reached ≈0.6. The cultures were then shifted to 18°C with shaking at 150 rpm for 30 min before the addition of IPTG (isopropyl-β-d-thiogalactopyranoside; 0.5 mM). After 16 h incubation, cells were harvested, and pellets were dissolved in lysis buffer (20 mM Tris, pH 7.8, 200 mM NaCl, and 10% glycerol) containing 0.5 mg/ml lysozyme and 2 mM phenylmethylsulfonyl fluoride (PMSF) and kept at 4°C for 1 h. The resuspended culture was lysed by sonication, and the supernatant was collected by centrifugation. Imidazole (10 mM) was added to the supernatant and then passed through Ni-nitrilotriacetic acid (Ni-NTA) resin that was preequilibrated with buffer containing 20 mM Tris (pH 7.8), 200 mM NaCl, 10% glycerol, and 10 mM imidazole. The column was extensively washed with washing buffer (20 mM Tris [pH 7.4], 200 mM KCl, and 10% glycerol) containing 50 mM imidazole. Bound ComE protein was eluted with the washing buffer containing 250 mM imidazole. Eluted protein was passed through Sephadex G-25 spin column for desalting. Protein samples were aliquoted in 20 mM Tris, pH 7.4, 200 mM KCl, and 10% glycerol and stored at −80°C. The protein was prespun at 14,000 rpm for 10 min before each experiment. The protein concentration was measured by Bradford’s reagent (24) using bovine serum albumin (BSA) as standard. The percentages of purity were ∼95%, calculated using ImageJ (data not shown).

Assay for oligomerization of ComE and its variants by SDS-PAGE.

Cys-mediated monomeric and dimeric forms of wild-type ComE and its variants were observed on SDS-PAGE. Briefly, 8 μg of ComE proteins were incubated with or without 2 mM DTT for 30 min at room temperature in buffer containing 20 mM Tris (pH 7.4), 200 mM KCl, and 10% glycerol. Samples were then electrophoresed on a 13% nonreducing (no DTT) SDS-PAGE with constant voltage (100 V) for 7 h, and the gel was stained with Coomassie after electrophoresis.

Biolayer interferometry for DNA-protein interaction.

BLI was performed to observe the DNA-protein interaction of ComE and its variants to the PnlmA using a similar method as described previously with some modifications (23). Briefly, PnlmA (≈500 bp) was PCR amplified with biotin tag at the 5′ end using biotinylated forward primer (Table 2). The biotinylated PnlmA fragment was diluted to 40 ng/μl in binding buffer [20 mM Tris, 50 mM KCl, 0.01 mM DTT, 5% glycerol (vol/vol), 1 mM EDTA, 0.01 mg/ml BSA, 5 mM MgCl₂, and 10 μg/ml poly(dI-dC) at pH 7.5] and loaded onto a hydrated streptavidin biosensor tip for 5 min. The biosensor tip was washed with binding buffer for 2 min to remove the excess amount of unbound biotinylated-PnlmA fragment. Further, association reaction was performed for 5 min by exposing the DNA-bound biosensor tip with 1 μM protein incubated in the binding buffer. The binding of ComE and its variants to biotinylated PnlmA was also observed without DTT in the binding buffer. Experiments were performed three times.

Conformational change assay by ANS fluorescence.

Conformational changes of ComE and its variants were measured using ANS, a hydrophobic fluorescence dye commonly used for probing conformational changes (18–20). About 4 μM ComE or its variants were incubated with or without 2 mM DTT in buffer containing 20 mM Tris (pH 7.4), 200 mM KCl, and 10% glycerol for 1 h at room temperature. Then, 50 μM ANS was added to each reaction mixture and further incubated for 10 min. Fluorescence spectra of the ComE-ANS complex were monitored using the excitation wavelength of 360 nm and the emission wavelength range of 400 to 600 nm. The fluorescence spectrum of free ANS was also observed. The fluorescence intensity of free ANS at 500 nm was subtracted from the fluorescence intensity of ANS with ComE and its variants. The subtracted fluorescence intensities (ΔF) of ComE and its variants were plotted. The experiment was performed in triplicate.

Homology modeling of ComE using I-TASSER.

A putative structure of the full-length ComE (1 to 250 amino acids [aa]; GenPept accession number AAN59528.1) was predicted using I-TASSER (25), keeping the crystal structure of the AgrA protein (PDB ID 3BS1) as the template. Since the crystal structure of AgrA contains only the DNA binding domain (136 to 238 aa), the DNA binding domain of ComE-S. mutans (153 to 250 aa) was superimposed on the crystal structure using PyMOL (64). The root mean square deviation (RMSD) was estimated to be 0.98 Å, indicating the predicted structure’s reliability. Furthermore, the distance between cysteine residues of ComE was measured to be 6 Å (between C200 and C229), 13 Å (between C173 and C200), and 18 Å (between C173 and C229).