The transcription regulator BrsR serves as a network hub of natural competence protein–protein interactions in Streptococcus mutans

Significance

Streptococcus mutans is one of the primary species associated with human dental caries, which affects the vast majority of adults worldwide. Like most streptococci, S. mutans is naturally competent and employs this ability to augment its adaptive evolution, ensuring its persistence within the oral microbiome. To increase the ecological benefit provided by natural competence, S. mutans has also evolved a poorly understood mechanism to coordinately regulate natural competence with its bacteriocin production ability. Here, we reveal the molecular mechanisms supporting this coordination. The current study illustrates how unrecognized posttranslational regulatory abilities may be found among a highly diverse pool of proteins not normally associated with genetic regulation. These hidden abilities are likely critical for the proper control of genetic networks.

Abstract

Genome evolution is an essential and stringently regulated aspect of biological fitness. For bacteria, natural competence is one of the principal mechanisms of genome evolution and is frequently subject to multiple layers of regulation derived from a plethora of environmental and physiological stimuli. Here, we present a regulatory mechanism that illustrates how such disparate stimuli can be integrated into the Streptococcus mutans natural competence phenotype. S. mutans possesses an intriguing, but poorly understood ability to coordinately control its independently regulated natural competence and bacteriocin genetic pathways as a means to acquire DNA released from closely related, bacteriocin-susceptible streptococci. Our results reveal how the bacteriocin-specific transcription activator BrsR directly mediates this coordination by serving as an anti-adaptor protein responsible for antagonizing the proteolysis of the inherently unstable, natural competence-specific alternative sigma factor ComX. This BrsR ability functions entirely independent of its transcription regulator function and directly modulates the timing and severity of the natural competence phenotype. Additionally, many of the DNA uptake proteins produced by the competence system were surprisingly found to possess adaptor abilities, which are employed to terminate the BrsR regulatory circuit via negative feedback. BrsR–competence protein heteromeric complexes directly inhibit nascent brsR transcription as well as stimulate the Clp-dependent proteolysis of extant BrsR proteins. This study illustrates how critical genetic regulatory abilities can evolve in a potentially limitless variety of proteins without disrupting their conserved ancestral functions. These unrecognized regulatory abilities are likely fundamental for transducing information through complex genetic networks.

The term natural competence refers to the prokaryotic ability to actively acquire and internalize extracellular DNA from the environment. Following its first discovery in Streptococcus pneumoniae nearly a century ago (1), natural competence has proven to be a highly tractable model system for the study of molecular genetics. We now know that natural competence is also a critical aspect of microbial ecology and genome evolution, in addition to facilitating the spread of antibiotic resistance worldwide (2–5).

The process of exogenous DNA acquisition requires complex regulatory networks to properly control production of the >20 unique proteins comprising the DNA uptake and recombination machinery (6, 7). Both gram-positive and gram-negative bacteria assemble highly analogous DNA uptake apparatuses, which are encoded by the so-called late competence genes (7, 8). The proximal upstream regulators initiating the competence genetic pathway (i.e., early competence genes) are far less conserved and can diverge greatly between species, presumably due to the disparate stimuli triggering competence development in different organisms. Within the Streptococcus genus, there are at least two distinct classes of intercellular signaling systems that are essential for initiating competence development: the ComCDE two-component signal transduction system and the ComRS Rgg-type regulatory system, a member of the larger RRNPP family (9, 10). The ComCDE system was discovered over two decades ago and controls competence development in the Mitis and Anginosus group streptococci (11–13), whereas, more recently, members of the Salivarius, Mutans, Pyogenic, Bovis, and Suis group streptococci were all found to utilize the ComRS system (10, 14–16). Both systems activate the production of the competence-specific alternative sigma factor ComX (9, 10), which directly stimulates the transcription of the vast majority of late competence genes (17, 18). In Streptococcus mutans, even though the ComRS system is essential for initiating the natural competence genetic pathway, a surprising number of seemingly unrelated genes and environmental stimuli are responsible for controlling the degree to which late competence genes are expressed, ultimately determining the strength of the ensuing natural competence phenotype (18–24). This is one of the more mysterious aspects of the S. mutans natural competence system. For example, through a still unknown mechanism, competence development is tremendously enhanced by the production of certain S. mutans bacteriocins (peptide antibiotics), particularly the bacteriocin mutacin V (25). Given their largely independent genetic pathways, it is intriguing that both bacteriocin production and competence development would exhibit such a strong regulatory linkage. In fact, many other bacterial species have also independently evolved unique mechanisms to coordinate competence development and bacteriocin production, suggesting that a major ecological advantage is provided by connecting these phenotypes to facilitate DNA acquisition from target bacteria (2, 26).

In addition to bacteriocins, a variety of other S. mutans genetic pathways are similarly capable of modulating the natural competence phenotype (9), such as the BrsRM regulatory system (27). BrsRM is a member of a newly identified class of prokaryotic signal transduction system that we refer to as the LytTR regulatory system (LRS) (28). LRS operons are broadly distributed among Gram-positive and Gram-negative bacteria, as well as some archaea (29), and encode a dual regulatory mechanism: an autoregulatory LytTR family transcription activator (30) and a cognate transmembrane protein antagonist of the transcription activator (29, 31, 32). Under laboratory growth conditions, LRSs exist in an inactive state due to the transmembrane protein antagonist, which inhibits its cognate LRS transcription regulator and prevents LRS operon expression (29, 32). One can constitutively activate an LRS simply by mutating the membrane inhibitor protein, which frees its cognate transcription regulator to activate gene expression by binding to direct repeat sequences located in its operon promoter region, as well as the promoters of the other genes within its regulon (29, 31, 32). Using this approach, we previously determined the BrsRM regulon of S. mutans. Deletion of the LRS membrane protein BrsM triggered substantially increased bacteriocin and late competence gene expression (27). All of the bacteriocin gene promoters controlled by the LRS regulator BrsR contain conserved direct repeats homologous to those found in LRS operons, and these repeats are similarly essential for bacteriocin transcription activation by BrsR and other regulators (18, 27, 29, 33, 34). In contrast, no such direct repeats are identifiable in the promoters of genes within the proximal natural competence genetic pathway (9, 15, 18). Thus, competence genes are highly unlikely to be the direct targets of BrsR transcription regulation, suggesting that an unidentified mechanism of competence cross-talk exists with these bacteriocin regulatory pathways. This prompted our interest to examine how the BrsRM LRS is able to stimulate the natural competence system. Here, we show that BrsR is directly responsible for the coordinated regulation of bacteriocin and natural competence genetic pathways by serving as a network hub of protein–protein interactions that mediate posttranslational regulation. Furthermore, newly discovered protein moonlighting functions are essential for both activating this regulatory circuit as well as deactivating it upon its completion (SI Appendix, Fig. S1). These results may be emblematic of an underlying evolutionary strategy employed within genetic networks to connect complementary, but otherwise independently regulated phenotypes.

Materials and Methods

DNA Manipulation and Strain Construction.

Strains, plasmids, and primers used in this study are described in SI Appendix, Tables S1 and S2. Specific details of strain construction are described in Supplemental Experimental Procedures. The protocol for cloning-independent allelic replacements and markerless mutagenesis of S. mutans have been described previously (35). The S. mutans reference strain UA159 served as the wild-type parent strain for all experiments, except where indicated.

Transformation Assays.

Transformation efficiency assays were performed according to a previously described methodology (36). Transformation reactions utilized marked genomic DNA. Transformation efficiency was determined by calculating the ratio of transformants to total viable cells. Data were reported as the means ± SDs from at least three biological replicates.

Coimmunoprecipitation.

S. mutans cultures were grown to the midlog phase in Todd–Hewitt yeast extract (THYE) medium. Cells were pelleted, washed in phosphate-buffered saline (PBS), and then cross-linked in PBS buffer containing 1% (volume/volume) formaldehyde for 10 to 15 min at 25 °C before quenching the reaction with 0.125 M glycine. Cells were disrupted with sonication, and clarified lysates were incubated with anti-FLAG M2 affinity gel (Sigma-Aldrich) or Pierce anti-DYKDDDDK affinity resin (Thermo Fisher Scientific), as well as monoclonal anti–HA-Agarose (Sigma-Aldrich) or Pierce anti-HA agarose (Thermo Fisher Scientific) for 3 h or overnight with end-over-end rotating at 4 °C. After washing four times, protein was eluted from the matrix with competitor FLAG or HA peptides or with 1% (weight/volume) SDS. The wild-type strain UA159 was included as a no-epitope negative control.

In-Gel Digestion and Liquid Chromatography with Tandem Mass Spectrometry Analysis.

Proteins were analyzed by NuPAGE 10% Bis-Tris Gels (Invitrogen) and stained with Coomassie Blue. In-gel digestion was performed following the ProteaseMAX surfactant digestion protocols (Promega). Mass spectrometric analysis was performed by the Oregon Health and Science University (OHSU) proteomics shared resource.

Western Blot.

S. mutans cultures were lysed using a Precellys evolution (Bertin Technologies) bead disruptor. Equal amounts of clarified total protein lysates were separated by 12 or 15% SDS–polyacrylamide gel electrophoresis (PAGE), followed by transferring onto nitrocellulose membranes (Bio-Rad). The membranes were blocked with 5% (weight/volume) nonfat milk powder in tris-buffered saline with 0.1% (volume/volume) Tween 20 (TBST) and then incubated with primary antibodies (OctA-Probe antibody H-5 anti-FLAG, Santa Cruz Biotechnology; monoclonal anti-HA antibody, Sigma-Aldrich; and anti-MYC Clone 9E10, OriGene) diluted 1:2,000. After incubation with 1:10,000 diluted, HRP-labeled secondary antibodies (Santa Cruz Biotechnology), immunoreactive bands were detected using the ChemiGlow West chemiluminescence substrate kit (Protein Simple). HA or FLAG epitope–labeled lactate dehydrogenase (LDH) was used as a loading control. In some instances, replicate Coomassie-stained SDS gels were imaged and used as loading controls.

RNA Extraction and qPCR.

S. mutans cultures were grown to the midlog phase in THYE medium. Total RNA was extracted from the cells using TRIzol reagent (Invitrogen) according to the manufacturer’s instructions. A total of 1 µg RNA from each sample was reverse transcribed using the SuperScript III first-strand synthesis system for RT-PCR (Invitrogen). qRT-PCR assays were performed using Power SYBR green PCR master mix (Applied Biosystems). 16S ribosomal RNA (rRNA) was used for normalization via the 2^-ΔΔCt method.

Messenger RNA and Protein Stability Assays.

To measure messenger RNA (mRNA) half-life, S. mutans cultures were grown to an optical density (OD₆₀₀) of 0.3 in THYE medium before adding 0.5 mg ⋅ mL⁻¹ rifampicin to halt transcription. Aliquots of the culture were withdrawn over a time course, rapidly chilled to 4 °C by mixing with 20 mL crushed ice, and then collected by centrifugation (6,000 × g, 10 min, and 4 °C). Total RNA was extracted, and comX mRNA abundance at each time point was determined by qRT-PCR. To measure protein half-life, cultures were grown similarly until halting both transcription and translation with an antibiotic mixture consisting of 0.5 mg ⋅ mL⁻¹ rifampicin, 0.8 mg ⋅ mL⁻¹ kanamycin, 0.05 mg ⋅ mL⁻¹ chloramphenicol, 0.025 mg ⋅ mL⁻¹ erythromycin, and 0.02 mg ⋅ mL⁻¹ tetracycline. Aliquots of the culture were withdrawn over a time course and rapidly chilled to 4 °C before performing Western blots on clarified lysates. ImageJ software (https://rsb.info.nih.gov/ij/index.html) was used to quantify band densities on the Western blot images. These data were plotted on a graph to determine protein half-life.

Luciferase Activity Assays.

Luciferase assays were performed using a previously described methodology (37). S. mutans cultures were grown to midlog phase. Luciferase activity was detected after adding 1 µL 0.75 mg ⋅ mL⁻¹ coelenterazine-h (NanoLight Technologies) to 100 µL cell culture. Luciferase activity was measured using a GloMAX Discover 96-well luminometer (Promega). Reporter data were normalized by dividing luciferase values by their corresponding optical density (OD₆₀₀) values. Data are presented as the means ± SDs of at least three biological replicates.

Recombinant Protein Expression and Purification.

Cultures of Escherichia coli BL21 were grown to an optical density (OD₆₀₀) of 0.6 before inducing the production of recombinant proteins with 0.1 mM isopropyl-β–thiogalactoside. Cultures were incubated at 16 °C with aeration for 16 h and lysed by sonication, and then target proteins were isolated using HisPur Ni-NTA resin (Thermo Fisher Scientific) according to the manufacturer’s instructions. All protein purification steps were performed at 4 °C.

Electrophoretic Mobility Shift Assays.

Electrophoretic mobility shift assay (EMSA) was performed using a previously described methodology (29). Double-stranded DNA probes were created by annealing equal molar concentrations of two oligonucleotides. DNA probes were 3′–end labeled with digoxigenin (DIG)-11–2’,3′-dideoxy-uridine-5′-triphosphate (Roche) using terminal transferase (New England Biolabs), according to the manufacturer’s instructions. A total of 1 ng DNA probe was incubated with various concentrations of purified recombinant proteins at 25 °C for 25 min. To measure competitive inhibition by SsbB, a mixture of BrsR and SsbB or SsbB point mutant proteins were incubated at 37 °C for 30 min prior to adding the labeled DNA probe and incubating an additional 25 min at 25 °C. The reaction products were separated by electrophoresis in native 5% Tris/borate/ethylenediaminetetraacetic acid (TBE) polyacrylamide gels at 4 °C and then electroblotted onto Hybond-N membranes (Amersham). The DNA probes were immobilized to the membrane by ultraviolet cross-linking (VWR). Detection of the DIG-labeled probes and visualization with CDP-Star (MilliporeSigma) were performed according to the manufacturer’s instructions. To detect BrsR–SsbB protein complexes following EMSA, the reactions were separated in native 5% TBE polyacrylamide or 1% agarose gels, electroblotted onto nitrocellulose membranes, and then analyzed by Western blot.

Results

The BrsRM LRS Exerts a Hidden Layer of Control over the Natural Competence Phenotype.

Our previous transcriptomic analysis of a brsM mutant revealed the BrsRM regulon to consist primarily of bacteriocin and natural competence genes (27). Since we had not previously determined whether this increase in competence gene expression was associated with changes in the competence phenotype, we compared natural transformation rates after mutating brsM in three separate parent strains of S. mutans: UA159 (serotype C genome reference strain), UA140 (serotype C laboratory strain), and CL1 (serotype K clinical isolate). These strains were chosen based upon their major differences in natural competence and bacteriocin production phenotypes, as well as their serotypes. After deleting brsM in each strain, all exhibited nearly identical, high-competence phenotypes, despite the wide range of transformation rates observed among the parent strains (Fig. 1 A and B). This suggested that competence stimulation by BrsR is likely a conserved function of the protein. Our previous BrsR EMSAs confirmed bacteriocin genes to be direct targets of BrsR regulation due to the presence of critical direct repeat sequences located adjacent to the bacteriocin gene promoters (27). Similar direct repeat sequences were predicted to control BrsR autoregulation of the brsRM operon as well (29). However, no such direct repeats are present in the promoters of the early competence genes comR and comS (15), nor are they present in the promoters of comX or the numerous late competence genes (15, 18). Therefore, we hypothesized that the BrsR-regulated bacteriocin gene nlmC (encoding the bacteriocin mutacin V) may be responsible for stimulating late competence gene expression in the brsM mutant, as mutacin V is a major target of BrsR transcription activation (SI Appendix, Fig. S1), and it is also the bacteriocin principally responsible for activating competence development in the parent strain UA159 (25). Consequently, we compared the transformation efficiency of the wild-type strain UA159 to a mutacin V (nlmC) mutant, a brsM mutant, and a mutacin V/brsM double mutant. As expected, the brsM mutant yielded the highest-transformation efficiency, presumably due to its stimulation of mutacin V production (Fig. 1 C and D) (27). However, the mutacin V/brsM double mutant surprisingly yielded nearly a 10-fold higher transformation efficiency compared to the wild-type and mutacin V single mutant, indicating that BrsR also participates in a mutacin V–independent mechanism of competence stimulation (Fig. 1 C and D). It is worth noting that the wild-type strain UA140 naturally lacks the nlmC gene encoding mutacin V (38), yet it still exhibits enhanced natural transformation in the brsM mutant background (Fig. 1 A and B). This is further evidence that BrsR engages in a mutacin V–independent mechanism of competence stimulation.

Fig. 1.

BrsR stimulates natural competence development independent of the bacteriocin mutacin V. (A) Transformation efficiencies were compared between the following strains: wild-type UA159 (159), wild-type UA140 (140), wild-type CL1 (CL1), UA159 ΔbrsM (Δ59), UA140 ΔbrsM (Δ40), and CL1 ΔbrsM (ΔCL). Transformation efficiency was calculated as the ratio of transformants to total bacteria. Error bars represent the SDs from the means of three biological replicates. (B) Representative images of transformants from the transformation efficiency assay. The numbers along the right side of the image indicate the dilution factors of the transformation reactions. (C) The same transformation efficiency assay was employed using the following strains: wild-type UA159 (WT), UA159 mutacin V mutant (ΔV), UA159 brsM mutant (ΔM), and UA159 and mutacin V/brsM double mutant (ΔV/ΔM). (D) Representative images of transformants from the transformation efficiency assay.

BrsR Forms Heteromeric Protein Complexes with Numerous Late Competence Proteins.

Given the unexpectedly dispensable role of mutacin V during BrsR-dependent competence stimulation, we next focused our attention on alternative, hypothetical regulatory mechanisms. Since S. mutans natural competence genes lack the direct repeat sequences targeted by LRS transcription regulators, it appeared highly unlikely that competence stimulation via BrsR would occur via its transcription activator function. Thus, we were curious whether protein–protein interactions with BrsR might be responsible. To identify the BrsR protein interactome, we engineered a FLAG epitope onto the C terminus of BrsR and expressed the construct as a transcription fusion to the constitutively expressed LDH promoter (Pldh-brsR) to facilitate subsequent coimmunoprecipitation (Co-IP) studies. We performed a variety of control experiments and confirmed that the strain was free of unwanted protein artifacts (SI Appendix, Fig. S2 A–N). Next, cultures were collected as they reached peak competence levels and then were subjected to α-FLAG immunoprecipitation followed by liquid chromatography with tandem mass spectrometry analysis. Over 80 potential BrsR-interacting proteins exhibited ≥twofold enrichment. Notably, most of the particularly strong associations (>20-fold enrichment) were formed with late competence proteins, including DprA (single-stranded DNA [ssDNA]–RecA-loading protein; SMU_RS04605), ComYA (DNA uptake traffic ATPase; SMU_RS09035), SsbB (competence-specific, single-stranded binding protein; SMU_RS08940), ComEA (membrane DNA receptor; SMU_RS02950), ComFA (membrane-associated, DNA-dependent ATPase; SMU_RS02390), and CoiA (DNA transport protein; SMU_RS03050) (SI Appendix, Table S3) (7). To independently verify these interactions, HA epitope tags were engineered onto five of the identified proteins and then tested for BrsR interactions via Co-IP. Since late competence proteins naturally form large protein complexes (7), we also mutated comX to terminate late competence gene expression and then individually ectopically expressed each of the five genes to test for binary interactions with BrsR. Using BrsR as the bait protein, we could readily coimmunoprecipitate each of the selected proteins, which included the well-characterized late competence proteins DprA, ComYA, and SsbB (7), as well as two poorly characterized, competence-related proteins SMU_RS03875 (competence-specific murein hydrolase) and SMU_RS01745 (DNA recombination protein RmuC) (Fig. 2 A and B). In addition, all five proteins could also serve as baits to coimmunoprecipitate BrsR, suggesting highly avid protein–protein interactions (Fig. 2 C and D). The results demonstrate that, in addition to its role as a transcription activator, BrsR directly interacts with a diverse assortment of proteins required for natural competence.

Fig. 2.

BrsR forms heteromeric complexes with late competence proteins and regulates their abundance. (A–D) The labels α-FLAG or α-HA indicate the antibodies used for Western blot detection. Samples are listed as follows: prestained protein ladder (M), wild-type UA159 (1), BrsR–FLAG + DprA–HA (2), BrsR–FLAG + ComYA–HA (3), BrsR–FLAG + SsbB–HA (4), BrsR–FLAG + SMU_RS03875–HA (5), and BrsR–FLAG + SMU_RS01745–HA (6). The prestained protein ladder molecular weights from top to bottom: 95, 72 (strong reference band), 56, 43, 34, 26, and 17 KDa. (A) The abundance of BrsR–FLAG was compared in each of the input samples used for Co-IP. (B) BrsR–FLAG was used as a bait to coimmunoprecipitate HA epitope–tagged competence proteins using anti-FLAG affinity resin. (C) The abundance of HA epitope–tagged competence proteins were compared in each of the input samples used for Co-IP. (D) HA epitope–tagged competence proteins were used as baits to coimmunoprecipitate BrsR–FLAG using anti-HA affinity resin. (E–I) Cultures were grown to OD₆₀₀ 0.4, and then, total protein was extracted to assess the abundance of various competence proteins. LDH was used as a loading control for each reaction. Strains are listed as follows: mutacin V mutant (ΔV), wild-type UA159 (WT), mutacin V/brsM double mutant (ΔV/ΔM), and brsM mutant (ΔM). The abundances of the following HA epitope–tagged competence proteins were compared: DprA (E), ComYA (F), SsbB (G), SMU_RS03875 (H), and SMU_RS01745 (I). (J) A mixture of antibiotics was added to halt transcription/translation of the WT and ΔM strains after the cultures reached an optical density of OD₆₀₀ 0.4. Aliquots of the cultures were withdrawn after the indicated incubation times and then further analyzed via Western blot.

BrsR Stimulates the Production of Late Competence Proteins.

The surprisingly diverse and avid BrsR interactions with late competence proteins suggested a unique mechanism of competence stimulation. We next compared the abundances of the five confirmed BrsR-interacting competence proteins in the mutacin V mutant, brsM mutant, and mutacin V/brsM double mutant backgrounds. Interestingly, four out of five of these proteins (DprA, ComYA, SsbB, and SMU_RS03875) exhibited expression patterns strongly correlated with the natural competence phenotypes of the parent strains (Fig. 2 E–H). Their abundances in the brsM mutant and mutacin V/brsM double mutant backgrounds were substantially higher than in both the wild-type and mutacin V single mutant strains (Fig. 2 E–H).

In contrast to DprA, ComYA, SsbB, and SMU_RS03875, neither mutacin V nor BrsR altered SMU_RS01745 levels (Fig. 2I). After analyzing the promoter regions of these genes, we noted that SMU_RS01745 is transcribed from a Sigma-70 type promoter, whereas dprA, comYA, ssbB, and SMU_RS03875 all contain “cin-box” (TACGAAT) type promoters utilizing the alternative sigma factor ComX (SI Appendix, Fig. S3) (17, 18). Since ComX is epistatic to dprA, comYA, ssbB, and SMU_RS03875 in the competence genetic pathway (SI Appendix, Fig. S1) (6, 18), we suspected that BrsR stimulation of DprA, ComYA, Ssb2, and SMU_RS03875 production was likely mediated through ComX. In further support of this idea, we also examined SsbB protein stability and found no evidence to suggest a direct BrsR-dependent posttranslational mechanism (Fig. 2J).

BrsR Utilizes a Moonlighting Anti-Adaptor Ability to Modulate the Window of Competence.

Our analysis of protein abundances for DprA, ComYA, SsbB, SMU_RS03875, and SMU_RS01745 all strongly implicated ComX as a key component for BrsR-dependent stimulation of natural competence. ComX was also identified in our previous BrsR interactome screen, albeit with weaker enrichment (11-fold) compared to many late competence proteins in the list (SI Appendix, Table S3). To confirm this interaction, we epitope tagged ComX in addition to its upstream regulator ComR (SI Appendix, Fig. S1), which was not identified in our BrsR interactome screen (SI Appendix, Table S3). For ComX, heteromeric complex formation was detectable with BrsR as the bait protein (Fig. 3A). However, the assay was unsuccessful in the reverse orientation with ComX as the bait (Fig. 3B), suggesting a less avid BrsR–ComX interaction compared to the BrsR–late competence protein interactions. This is also consistent with the weaker ComX enrichment in the BrsR interactome data (SI Appendix, Table S3). As expected, we found no evidence of a BrsR–ComR interaction, further supporting the validity of the interactome results (Fig. 3 C and D and SI Appendix, Table S3).

Fig. 3.

Heteromeric BrsR–ComX complexes antagonize ComX proteolysis. (A–D) The labels α-FLAG or α-HA indicate the antibodies used for Western blot detection. Strains are listed as follows: mutacin V mutant (ΔV), wild-type UA159 (WT), mutacin V/brsM double mutant (ΔV/ΔM), and brsM mutant (ΔM). The prestained protein ladder molecular weights from top to bottom: 95, 72 (strong reference band), 56, 43, 34, 26, and 17 KDa. (A) BrsR–FLAG was used as a bait to coimmunoprecipitate ComX–HA. (B) ComX–HA was used as a bait to coimmunoprecipitate BrsR–FLAG. (C) BrsR–FLAG was used as a bait to coimmunoprecipitate ComR–HA. (D) ComR–HA was used as a bait to coimmunoprecipitate BrsR–FLAG. (E) Total protein was extracted from cultures grown to OD₆₀₀ 0.3. ComX–HA abundance was measured in the following backgrounds: ΔV, WT, ΔV/ΔM, and ΔM. LDH served as a loading control. (F) The same experiment was performed, except the abundance of ComR–HA was detected. (G) Total RNA was extracted from cultures grown to OD₆₀₀ 0.3. The relative abundance of comX mRNA was measured using qRT-PCR in the following strains: ΔV, WT, ΔV/ΔM, and ΔM. Error bars represent the SDs from the mean of three biological replicates. Unpaired Student’s t test *P < 0.05 and **P < 0.01. (H) Rifampicin was added to halt transcription after the ΔV and ΔV/ΔM cultures reached an optical density of OD₆₀₀ 0.3. Aliquots of the cultures were withdrawn at the indicated times, and comX mRNA abundance was determined by qPCR. (I) A mixture of antibiotics was added to halt transcription/translation of the ΔV strain after the cultures reached an optical density of OD₆₀₀ 0.4. Aliquots of the cultures were withdrawn after the indicated incubation times. ComX–HA half-life was determined by comparing its abundance at each time point via Image J analysis of the individual band intensities in the Western blot. (J) The same experiment was performed, except the ΔV/ΔM strain was assayed. (K) BrsR–FLAG, BrsR^K47E–FLAG, and BrsR^K65E–FLAG were immunoprecipitated using anti-FLAG resin. (L) The same output samples were analyzed for the abundance of ComX–HA. The half-life of ComX–HA in the wild-type BrsR background (M) was compared to that of the BrsR^K47E (N) and BrsR^K65E BrsR (O) mutants. A replicate gel of Coomassie-stained protein lysates was included as a loading control (Total).

Next, we examined whether BrsR affected ComX and ComR protein levels. Similar to DprA, ComYA, SsbB, and SMU_RS03875, ComX abundance strongly correlated with natural competence in each of the mutacin V and brsM mutant backgrounds, whereas the abundance of ComR was completely unaffected (Fig. 3 E and F). From these results, we concluded that BrsR regulation of late competence proteins is likely mediated through the stimulation of ComX abundance but not via ComR. Since ComR is the only known regulator of comX gene expression in S. mutans, it was of particular interest to determine the regulatory mechanism(s) employed by BrsR. Despite being a transcription regulator, BrsR surprisingly only played a minor role in comX gene expression. At the transcriptional level, we observed approximately twofold increase in comX gene expression in the brsM mutant background that was largely mediated by mutacin V (Fig. 3G). We also compared the BrsR effect upon comX mRNA half-life and found a similar twofold increased stability in the mutacin V/brsM double mutant (Fig. 3H). In contrast to these modest effects, BrsR exhibited a major impact upon ComX at the posttranslational level. After antibiotic inhibition of both transcription and translation, we measured the ComX protein half-life, which fell below the detection limit of our protein turnover rate assay (<2 min) (Fig. 3I). However, in the presence of activated BrsR (i.e., ΔbrsM), ComX half-life increased considerably to ∼8 min, demonstrating that BrsR is required to stabilize ComX in vivo (Fig. 3J).

To characterize how this stability is conferred, SWISS-Model (39) and University of California, San Francisco Chimera (40) software were used to create a structural model of BrsR that identified eight candidate surface-exposed charged residues, which were each mutagenized to examine their roles in BrsR–ComX heteromeric complex formation (SI Appendix, Fig. S4A) (41, 42). To avoid potential confounding autoregulatory defects conferred by the BrsR mutations (29), we constitutively expressed brsR using the Pldh-brsR construct. Five out of eight of the BrsR point mutant proteins were stable when expressed (SI Appendix, Fig. S4B). Two of these mutants (BrsR^K47E and BrsR^K65E) were impaired in their abilities to transcriptionally activate nlmC (mutacin V) and bind/stabilize ComX (SI Appendix, Fig. S4 B–D). In addition, substantially less ComX protein was detected in these brsR mutant backgrounds, even when comX was constitutively expressed (i.e., PgyrA-comX), which is consistent with a posttranslational regulatory mechanism (SI Appendix, Fig. S4E). Using these same constitutive comX strains, we also confirmed that the K47E and K65E mutant BrsR proteins were indeed defective in their abilities to form BrsR–ComX complexes (Fig. 3 K and L), and this defect resulted in a pronounced destabilization of ComX (Fig. 3 M–O).

We next determined the impact of these changes in ComX stability upon natural transformation. The development of natural competence in S. mutans and other streptococci typically occurs within a very narrow window of cell densities during planktonic culture (10, 43). This agreed with our observed expression pattern of ComX, which both appeared and disappeared rapidly when the optical density values ranged between OD₆₀₀ 0.3 to 0.4 (Fig. 4A). However, in the presence of activated BrsR (i.e., ΔbrsM), ComX was detectable over a much broader range of optical density values from OD₆₀₀ 0.2 to 0.8 (Fig. 4B). This expanded window of ComX stability correlated with a concomitant increase in the window of natural competence as well. As shown in Fig. 4 C and D, the highest transformation rates in the mutacin V mutant background were detected when DNA was added at the early logarithmic phase, with competence diminishing substantially by OD₆₀₀ 0.8. In contrast, the mutacin V/brsM double mutant retained high levels of competence at OD₆₀₀ 0.8, exhibiting >30-fold higher rates of transformation (Fig. 4 C and D). As expected, the BrsR^K47E and BrsR^K65E mutants failed to stimulate competence development at all time points (Fig. 4 C and D). Taken together, the results demonstrate how BrsR is able to directly antagonize ComX proteolysis as a mechanism to modulate both the timing and strength of natural competence phenotype.

Fig. 4.

BrsR expands the window of competence. (A) ComX–HA protein abundance was examined in the wild type (WT) at an optical density OD₆₀₀ of 0.2 as well as at the indicated optical density values in the mutacin V mutant (ΔV) background. LDH was used as a loading control. (B) The same experiment was performed in the mutacin V/brsM double mutant (ΔV/ΔM) background. (C) Transformation efficiencies were compared between the ΔV, ΔV/ΔM, and brsR K47E/K65E point mutant/brsM double mutant strains (K47E and K65E). Transformation reactions were performed after adding transforming DNA at the indicated optical density values. Transformation efficiency was calculated as the ratio of transformants to total bacteria. Error bars represent the SDs derived from three biological replicates. (D) Representative images of the transformation efficiency assay. The numbers along the right side of the images indicate the dilution factors of the transformation reactions.

Heteromeric Complex Formation with Late Competence Proteins Antagonizes BrsR Autoregulation.

While our results provided a clear explanation for the regulatory role of BrsR–ComX protein complexes, they still did not explain the functional role of BrsR interactions with late competence proteins. Our initial hypothesis was that BrsR may be regulating the function or stability of the late competence proteins within its interactome, similar to its effect upon ComX. However, we found no evidence to support this (Fig. 2J). Thus, we later suspected that the function of these protein interactions may be to regulate BrsR, possibly by antagonizing its transcription regulatory function. Since BrsR is a positive regulator of the brsRM operon (29), we created a brsRM luciferase transcription fusion reporter to examine the impact of late competence proteins upon BrsR autoregulation. To stimulate the production of late competence proteins, we supplemented the cultures with “SigX-inducing peptide” (XIP), which is the peptide signal molecule encoded by comS that directly initiates the proximal competence genetic pathway (SI Appendix, Fig. S1) (9, 15). Interestingly, exogenous XIP supplementation was indeed found to inhibit the luciferase activity of the brsRM operon reporter both in the wild-type and brsM mutant backgrounds, whereas this effect was fully abrogated in a comX mutant background, suggesting that late competence proteins may be responsible (Fig. 5 A–D). To examine this more directly, we first confirmed via EMSA that BrsR binds to the direct repeat sequences located directly upstream of the brsRM operon promoter, as previously predicted (SI Appendix, Fig. S5 A–C) (29). Next, we selected SsbB from our group of confirmed BrsR interactors to determine whether it could serve as a competitive inhibitor in the same EMSA reaction. In addition, we developed a structural model of the BrsR–SsbB heteromeric complex to identify potential SsbB residues critical for BrsR interactions and then individually mutagenized each (SI Appendix, Fig. S6 A–C). Two of these SsbB mutant proteins (D47A and R78A) exhibited severely reduced affinities toward BrsR (SI Appendix, Fig. S6 D and E) and were therefore heterologously expressed and purified along with the wild-type SsbB (SI Appendix, Fig. S7) to include in the BrsR EMSA of the brsRM operon promoter. As shown in Fig. 5 E–G, wild-type SsbB functions as a robust competitive inhibitor of BrsR DNA binding, whereas this inhibitory activity is severely compromised with the D47A and R78A mutant SsbB proteins. We also noticed that the EMSA reactions containing SsbB each yielded an additional, slower migrating band shift. Upon further analysis, we determined that this was due to SsbB binding nonspecifically to double-stranded DNA (dsDNA) in vitro (SI Appendix, Fig. S8 A–D), but with obviously lower affinity compared to ssDNA. As shown in SI Appendix, Fig. S8D, substantially less SsbB is required to bind all of the ssDNA probe compared to the dsDNA probe. The EMSA results further indicated that the D47A and R78A mutant SsbB proteins retained their DNA-binding abilities (Fig. 5 E–G). Next, it was of interest to determine whether the inhibitor function of SsbB in the EMSA reactions was dependent upon heteromeric complex formation with BrsR. While wild-type SsbB could clearly titrate BrsR away from the brsRM promoter DNA in the EMSA reaction, the BrsR–SsbB heteromeric complexes responsible for this effect were apparently too large to resolve via 5% native PAGE, as they mostly remained near the wells of the gel, even after extended run times (Fig. 5H). Presumably, this was a consequence of BrsR interactions with large SsbB multimers (44–46). Fortunately, we were able to circumvent this limitation by electrophoresing in nondenaturing 1% agarose gels prior to performing Western blot (45). Using this approach, we could readily detect the dose-dependent formation of BrsR–SsbB complexes in vitro, whereas these complexes were not detectable using the D47A and R78A mutant SsbB proteins (Fig. 5 I–J). Because of the substantial disparities in the electrophoretic migration rates of BrsR–SsbB complexes versus free BrsR, it was not possible to simultaneously resolve both the complexed and free BrsR in the same agarose gels (Fig. 5 H–J). Lastly, we compared BrsR protein abundance in a comX mutant background ± ectopic expression of either dprA, comYA, ssbB, or point mutant ssbB genes. The ectopic expression of each of the wild-type late competence genes substantially reduced the abundance of BrsR, whereas this effect was compromised with the D47A and R78A point mutant ssbB genes and completely abolished with the D47A/R78A double point mutant ssbB (Fig. 5K). This result confirmed that multiple late competence proteins do antagonize BrsR levels in the cell, which we conclude is at least partially due to the inhibition of BrsR autoregulation of the brsRM operon promoter.

Fig. 5.

Competence proteins antagonize BrsR autoregulation. (A–D) brsRM luciferase transcription fusion reporter strains were grown to an optical density of OD₆₀₀ 0.1 before adding 1 µM XIP to stimulate late competence gene expression. Aliquots of the cultures were withdrawn at 20-min intervals, and normalized luciferase values were measured in the following strain backgrounds: wild-type UA159 (WT) (A), brsM mutant (ΔM) (B), comX mutant (ΔcomX) (C), and comX/brsM double mutant (ΔcomX/ΔM) (D). (E–G) EMSAs were performed with recombinant BrsR–FLAG (0.05 µM) ± various concentrations of SsbB–HA or point mutant SsbB–HA proteins, as well as ±1 ng digoxigenin-labeled DNA probe, encompassing the direct repeat (DR) region upstream of the brsRM operon promoter. Red arrows indicate mobility-shifted BrsR–DNA complexes. Wild-type SsbB–HA (E), SsbB^D47A–HA (F), or SsbB^R78A–HA (G) were added to the reactions at 0.05, 0.1, 0.2, 0.3, 0.4, 0.6, 0.8, and 1 µM at 37 °C for 30 min prior to performing EMSA. (H) BrsR–FLAG (0.05 µM) was coincubated ±0.1, 0.8, or 1.6 µM SsbB–HA at 37 °C for 30 min prior to performing EMSA and Western blot. (I) BrsR–FLAG (0.05 µM) was coincubated ±0.8 or 1.6 µM wild-type or mutant SsbB proteins at 37 °C for 30 min prior to performing EMSA and Western blot. Samples were electrophoresed in 1% native agarose gels to resolve BrsR–SsbB heteromeric complexes. (J) The same experiment was performed without DNA in the reactions. (K) A brsM/comX double mutant strain was transformed with plasmids encoding constitutively expressed late competence proteins to determine their effect upon BrsR–FLAG abundance. The following expression plasmids were compared: no plasmid (1), empty vector (2), dprA (3), comYA (4), ssbB (5), ssbB D47A point mutant (6), ssbB R78A point mutant (7), and ssbB D47A/R78A double point mutant (8).

Late Competence Proteins Possess Hidden Adaptor Functions Targeting BrsR for Proteolysis.

As part of a separate study, we serendipitously discovered another key moonlighting function of the late competence proteins. As shown in SI Appendix, Table S4, in a constitutive brsR expression background, the protein interactome of the housekeeping ClpXP protease is surprisingly similar to the BrsR interactome (SI Appendix, Table S3). Strong enrichment of multiple late competence proteins, ComX, and BrsR were all observed, suggesting they could be substrates for ClpXP proteolytic degradation. To first test whether BrsR is subject to an additional layer of posttranslational regulation, we repeated the previous dprA, comYA, and ssbB ectopic expression experiment and incorporated the constitutive Pldh-brsR construct to eliminate the influence of autoregulation upon BrsR abundance. As shown in Fig. 6A, a transcription activation-defective ComX (i.e., no sigma factor activity because of a C-terminal 3× HA epitope tag) was produced together with a constitutively expressed brsR. In this genetic background, the subsequent ectopic expression of wild-type dprA, comYA, and ssbB still triggered a substantial reduction of BrsR protein abundance in the cell, whereas the point mutant ssbB genes were impaired in this ability. As expected, ComX abundance in each of these strains mirrored that of BrsR, which is further evidence that BrsR protects ComX from degradation (Fig. 6A). However, unlike the BrsR results, the ectopic expression of the mutant ssbB genes only partially restored ComX levels, indicating that late competence proteins likely employ both BrsR-dependent and BrsR-independent mechanisms to modulate ComX abundance (Fig. 6A). Next, we were interested to determine whether the depletion of BrsR by late competence proteins is mediated by the Clp protease, as suggested by the ClpX interactome results (SI Appendix, Table S4). In the constitutive brsR expression background, wild-type or mutant ssbB genes were ectopically expressed after deleting the Clp ATPases ClpC, ClpE, and ClpX, as well as the ClpP protease. Each of the individual Clp ATPase deletions failed to suppress the large disparities in BrsR levels in the wild-type ssbB and D47A/R78A double mutant ssbB ectopic expression strains (Fig. 6B). However, in a clpP mutant background, wild-type SsbB lost its ability to trigger BrsR depletion, indicating an impairment of BrsR proteolytic degradation (Fig. 6B). These results further indicated that BrsR–SsbB heteromeric complexes likely stimulate BrsR proteolytic degradation by any of the AAA+ Clp protease complexes (i.e., ClpCP, ClpEP, or ClpXP), possibly as a consequence of the redundancy between these proteases (47, 48). As further evidence, we utilized ClpP as a bait to coimmunoprecipitate BrsR–SsbB heteromeric complexes. Based upon the results shown in Fig. 6B, we reasoned that more BrsR would be coimmunoprecipitated in the presence of wild-type SsbB versus the D47A/R78A double mutant SsbB. Indeed, this is precisely what was observed. While ClpP pulldowns yielded identical amounts of coimmunoprecipitated wild-type and mutant SsbB proteins (Fig. 6C), considerably more BrsR was coimmunoprecipitated together with the wild-type SsbB, compared to its double mutant version (Fig. 6D). In fact, the mutant SsbB reaction yielded nearly identical BrsR Co-IP results to the negative control reaction that completely lacked SsbB (Fig. 6D). The ability of SsbB to stimulate BrsR degradation by the Clp protease is highly analogous to the role of adaptor proteins like MecA, which enhance the Clp-dependent, proteolytic degradation of proteins that would otherwise serve as poor substrates for proteolysis (49). Based upon these results, we conclude that BrsR interactions with various late competence proteins prevent new BrsR synthesis by antagonizing BrsR autoregulation, while concurrently stimulating the proteolytic degradation of extant BrsR proteins via the Clp protease. The combined action of both inhibitory mechanisms results in the potent depletion of BrsR levels in the cell that can be observed soon after late competence genes are expressed.

Fig. 6.

Late competence proteins function as adaptor proteins targeting BrsR for proteolysis. (A) The constitutively expressed Pldh-brsR strain was transformed with plasmids encoding constitutively expressed late competence proteins to determine their effect upon BrsR–FLAG abundance in a comX mutant background. The following expression plasmids were compared: no plasmid (1), empty vector (2), dprA (3), comYA (4), ssbB (5), ssbB D47A point mutant (6), ssbB R78A point mutant (7), and ssbB D47A/R78A double point mutant (8). (B) The Pldh-brsR strain was transformed with either the wild-type ssbB expression plasmid (1, 3, 5, 7 and, 9) or the double point mutant ssbB expression plasmid (2, 4, 6, 8, and 10) prior to mutating comX. BrsR–FLAG abundance was compared after mutating the following: clpC (3 and 4), clpE (5 and 6), clpX (7 and 8), and clpP (9 and 10). (C) ClpP–HA was used as a bait protein to coimmunoprecipitate SsbB–Myc from Pldh-brsR strains transformed with the following expression plasmids: empty vector (Vector), ssbB (SsbB), and double point mutant ssbB (SsbB mutant). The Western blot was probed with anti-Myc antibodies. (D) The same experiment was performed, except the Western blot was probed with anti-FLAG antibodies to measure BrsR–FLAG Co-IP.

Discussion

The current study focuses upon one of the more puzzling aspects of the S. mutans natural competence system: its propensity for regulation from outside of the proximal, natural competence genetic pathway (21). The S. mutans competence system is an excellent model to explore the mechanisms supporting such linkages between genetic pathways, as natural competence defects are exceptionally common mutant phenotypes in this organism, yet the underlying mechanisms have remained largely enigmatic. The mechanisms utilized to link the BrsRM LRS to the natural competence system might also exemplify a more fundamental regulatory strategy used by cells to connect complementary phenotypes encoded by otherwise independent genetic pathways.

End Product Inhibition Model of Competence Regulation by the BrsRM LRS.

Based upon the current results, we propose the following model to describe how a noncanonical competence regulatory system like the BrsRM LRS is able to achieve potent control over the natural competence phenotype (Fig. 7). Firstly, the proximal ComRS competence regulatory system is required to initiate the transcription of comX, which encodes a competence-specific alternative sigma factor that is essential for competence development. ComX is exceptionally unstable in S. mutans (Fig. 3I), likely as a consequence of its highly efficient, proteolytic degradation by the Clp protease (50–52). To progress to a productive state of natural competence, the cell must either produce ComX at a rate faster than its efficient proteolysis or the cell must reduce the rate at which ComX is degraded. The canonical ComRS competence regulatory system depends heavily upon the former strategy, such that the strength of the upstream ComS-dependent, intercellular-signaling stimulus directly affects the rate of comX transcription, and ultimately, the strength of the ensuing natural competence phenotype (SI Appendix, Fig. S1) (9, 10). Irrespective of comX transcription rate, once the ComX protein is produced, its inherent instability can be suppressed by noncanonical regulators like BrsR, which moonlights as an anti-adaptor protein antagonist of ComX proteolysis (Fig. 3 A, E, and J). Thus, the net effect upon competence development is actually the sum of both the upstream competence-inducing, extracellular-signaling cascade, as well as the secondary action(s) of noncanonical competence-regulatory pathways that are controlled via independent stimuli. This allows the cell to integrate a potentially wide array of environmental and/or physiological stimuli into the circuitry controlling the competence phenotype. After the late competence proteins have been produced because of the function of ComX, the cell must engage a competence escape mechanism to remain viable, as cells cease to divide while in the competent state (7). For this reason, it is necessary to alter the BrsR–ComX-binding equilibrium to disfavor these heteromeric complexes and revert ComX back to its basal state of low intrinsic stability, thus ending the production of nascent late competence proteins. To achieve this, the cell terminates the BrsR regulatory effect by initiating a genetic mechanism analogous to classic end product inhibition. This negative feedback mechanism depends upon the prior production of the many different BrsR-interacting late competence proteins (SI Appendix, Table S3). Together, these interactions shift the BrsR-binding equilibrium to favor BrsR–late competence protein heteromeric complexes in lieu of BrsR–ComX complexes, thereby freeing ComX for Clp-dependent, proteolytic degradation. To finally inactivate the circuit, these newly formed BrsR–late competence protein heteromeric complexes deplete BrsR by inhibiting BrsR autoregulation and concurrently serving as adaptor proteins to enhance BrsR degradation by the Clp proteases (Figs. 5 A–K and 6 A–D). In this manner, the BrsR activation of competence is terminated only after the competence genetic pathway has reached its endpoint (i.e., the production of late competence proteins), thus tethering BrsR depletion directly to the successful activation of the competence phenotype. Conceptually, this is reminiscent of the numerous biosynthetic and catabolic pathways that respond to the production of reaction intermediates and end products with negative feedback regulation upon key genes within the pathways (53–57). A fundamental distinction here is that BrsR originates from outside of the proximal genetic pathway it is regulating. Furthermore, since the primary products of the competence pathway are proteins rather than metabolites, this allows for an additional layer of regulatory control over protein turnover rates via the moonlighting functions of BrsR and numerous late competence proteins.

Fig. 7.

Genetic end product inhibition model of BrsR-dependent, natural competence stimulation. The proximal ComRS early competence regulatory system triggers the production of ComX proteins, which are exceptionally labile because of their efficient degradation by the Clp protease. When the BrsRM LRS is activated, positive feedback autoregulation results in a rapid increase in BrsR abundance, triggering the moonlighting anti-adaptor ability of BrsR to form BrsR–ComX heteromeric complexes that antagonize the Clp-dependent proteolysis of ComX. This substantially increases the half-life of ComX, resulting in ComX accumulation, increased ComX-dependent transcription of late competence genes, and ultimately, higher levels of natural competence. A functionally diverse set of late competence enzymes and structural proteins, such as DprA, ComYA, SsbB, etc., all have moonlighting adaptor abilities mediated by heteromeric complex formation with BrsR. Consequently, after the late competence proteins have been produced, the large number of newly synthesized BrsR-binding partners shifts the BrsR-binding equilibrium to disfavor ComX interactions and instead promote the creation of various BrsR–late competence protein heteromeric complexes. The resulting unbound ComX then reverts back to its basal state of low intrinsic stability, ending the production of nascent late competence proteins. In addition, BrsR–late competence protein heteromeric complexes deplete BrsR levels in the cell by antagonizing BrsR-positive feedback autoregulation of the brsRM operon, as well as by serving as adaptor proteins to stimulate the Clp-dependent degradation of extant BrsR proteins. The depletion of BrsR triggered by the terminal products of the competence genetic pathway resets the regulatory circuit, allowing for subsequent rounds of BrsR-dependent competence stimulation. This figure was partially created using BioRender.com.

Implications for Competence Regulation in S. mutans.

The activation of proximal natural competence genetic pathways has been characterized in detail for multiple Streptococcus species (9, 12), but substantially less is known about the mechanisms terminating streptococcal competence. Competence escape is best characterized in S. pneumoniae, in which the late competence protein DprA plays a central role by antagonizing the transcription regulatory function of the early competence protein ComE (58, 59). This is analogous to SsbB inhibition of BrsR autoregulation (Fig. 5 E–K). However, unlike ComE in S. pneumoniae, multiple late competence proteins have the ability to form heteromeric complexes with BrsR to regulate its activity (Fig. 5K and SI Appendix, Table S3) (58). Another distinction from S. pneumoniae is that ComX protein stability is a major determinant of the S. mutans window of competence (Figs. 3 I and J and (4 A–D). In S. pneumoniae, ComX is labile, but its natural turnover rate was not found to be regulated (58). Furthermore, there is reason to suspect that S. mutans late competence proteins may also participate in an end product inhibitory mechanism targeting ComX as well as BrsR. As shown in Fig. 6A, the ectopic expression of dprA, comYA, and ssbB all triggered the potent destabilization of ComX, which was anticipated because of their roles in BrsR depletion. However, we were surprised to discover that the D47A/R78A mutant SsbB retained much of its ability to antagonize ComX levels in the cell, even though this protein is incapable of complexing with BrsR (Fig. 6A and SI Appendix, S6 D and E). This indicates that SsbB likely utilizes both BrsR-dependent and BrsR-independent mechanisms to modulate ComX stability. A separate BrsR-independent inhibitory mechanism would be an interesting candidate for the still unknown, canonical competence escape mechanism of ComRS-encoding streptococci like S. mutans.

Connecting Complementary Phenotypes via Protein Moonlighting Functions.

It was initially surprising to discover BrsR regulating competence independent of its bacteriocin regulatory role because we had anticipated the activation of mutacin V production to be the critical regulatory step. While mutacin V could enhance the magnitude of the BrsR effect, it is not essential, as neither mutacin V nor the transcription regulatory function of BrsR are required (Figs. 1 C and D and 3 M–O and SI Appendix, Fig. S4 B–D). The key level of regulation occurs directly between BrsR and numerous proteins within the proximal competence pathway. Given the aforementioned ecological advantage afforded by coordinating bacteriocin production and competence development (2), we suspect that the BrsR competence regulatory mechanism evolved independently to link these two complementary phenotypes. Considering the diversity of competence proteins that avidly bind to BrsR (SI Appendix, Table S3), we anticipate that analogous interactions would be utilized with other genetic pathways that influence competence development, particularly those involved in bacteriocin production (SI Appendix, Fig. S1). At a more fundamental level, the moonlighting activities revealed in the current study may be just one example of a much broader underlying strategy employed to link independent genetic pathways. Such moonlighting activities may even be a source of the inexplicable transcriptional changes frequently observed in mutant transcriptomic datasets or the plethora of unexpected protein interactions often identified in interactome studies. While the function of a given protein is commonly defined by its primary enzymatic or structural role, this likely underestimates the true breadth of its functionality. For example, S. mutans late competence proteins are a functionally diverse collection of proteins that are only transiently produced in a minority subpopulation of cells (60, 61), yet they have still managed to evolve posttranslational regulatory abilities that are entirely independent of their enzymatic and structural roles in DNA uptake. As such, it would be quite surprising if pervasive protein moonlighting abilities were the exception rather than the rule.

Data Availability

All study data are included in the article and/or SI Appendix.