Genomic instability of TnSMU2 contributes to Streptococcus mutans biofilm development and competence in a cidB mutant

Abstract Streptococcus mutans is a key pathogenic bacterium in the oral cavity and a primary contributor to dental caries. The S. mutans Cid/Lrg system likely contributes to tolerating stresses encountered in this environment as cid and/or lrg mutants exhibit altered oxidative stress sensitivity, genetic competence, and biofilm phenotypes. It was recently noted that the cidB mutant had two stable colony morphologies: a “rough” phenotype (similar to wild type) and a “smooth” phenotype. In our previously published work, the cidB rough mutant exhibited increased sensitivity to oxidative stress, and RNAseq identified widespread transcriptomic changes in central carbon metabolism and oxidative stress response genes. In this current report, we conducted Illumina‐based genome resequencing of wild type, cidB rough, and cidB smooth mutants and compared their resistance to oxidative and acid stress, biofilm formation, and competence phenotypes. Both cidB mutants exhibited comparable aerobic growth inhibition on agar plates, during planktonic growth, and in the presence of 1 mM hydrogen peroxide. The cidB smooth mutant displayed a significant competence defect in BHI, which was rescuable by synthetic CSP. Both cidB mutants also displayed reduced XIP‐mediated competence, although this reduction was more pronounced in the cidB smooth mutant. Anaerobic biofilms of the cidB smooth mutant displayed increased propidium iodide staining, but corresponding biofilm CFU data suggest this phenotype is due to cell damage and not increased cell death. The cidB rough anaerobic biofilms showed altered structure relative to wild type (reduced biomass and average thickness) which correlated with decreased CFU counts. Sequencing data revealed that the cidB smooth mutant has a unique “loss of read coverage” of ~78 kb of DNA, corresponding to the genomic island TnSMU2 and genes flanking its 3′ end. It is therefore likely that the unique biofilm and competence phenotypes of the cidB smooth mutant are related to its genomic changes in this region.

Oral streptococci such as S. mutans can also utilize eDNA for genetic exchange via natural competence. In S. mutans, two quorum-sensing systems have been characterized in regulation of competence. In peptide-rich growth media, the competence-stimulating peptide (CSP, encoded by comC)-based system is used. CSP is a 21AA peptide which is released into the extracellular environment via ComAB. The peptide is then processed by the protease SepM (Hossain & Biswas, 2012) into a final 18AA form which can be recognized by the ComDE two-component system. ComD is a histidine kinase which phosphorylates ComE, a response regulator, which then can activate bacteriocin genes, such as nlmC, and provide negative regulatory feedback for comC (Kreth et al., 2007). Phosphorylated ComE indirectly leads to comX activation, the alternative sigma factor involved in the regulation of competence genes (Li et al., 2002), yet the process by which this occurs remains unclear. Alternative to CSP, a 7AA peptide identified as comX-inducing peptide (XIP) is processed from ComS, which can directly activate comX expression through the regulator ComR. Exported XIP is imported via the Opp permease and forms a complex with ComR, which then binds upstream of the gene encoding the XIP peptide, comS, and comX (Mashburn-Warren, Morrison, & Federle, 2010). Opp-mediated uptake is blocked by peptide-rich media, but ComS has also been demonstrated to participate in competence signaling without leaving the cell (Underhill et al., 2018). While both CSP and XIP induce competence in alternative ways, addition of synthetic forms of either peptide has been shown to stimulate a comX response (Son, Ahn, Guo, Burne, & Hagen, 2012).
The S. mutans cidAB transcriptional unit encodes two proteins (CidA and CidB) which have been shown to affect a multitude of virulence traits including biofilm development and oxidative stress response (Ahn, Rice, Oleas, Bayles, & Burne, 2010). Expression of cidAB has also been shown to be regulated by catabolite control protein A (CcpA; Kim, Waters, Turner, Rice, & Ahn, 2019). While the close homologs of CidA/B, LrgA/B, have recently been characterized in Bacillus subtilis as a transporter of pyruvate (Charbonnier et al., 2017;Esker, Kovács, & Kuipers, 2017), the exact function of the Cid/Lrg proteins in S. mutans remains unknown. Expression data have shown that cidAB RNA levels are upregulated during anaerobic growth (Ahn, Wen, & Burne, 2007) and by growth in the presence of excess glucose (Ahn et al., 2010). Transcriptomic data collected under anaerobic conditions also suggest an interesting effect on the expression of genomic islands (GIs) TnSMU1 and TnSMU2 (Ajdić et al., 2002) of S. mutans in a cidB mutant (Ahn & Rice, 2016). Many of the genes encoded within these regions were found to be significantly upregulated in a cidB mutant as compared to wild type under the same conditions, possibly indicating a link between Cid and these mobile genetic elements (MGEs).
Genomic islands are a major class of MGE commonly acquired via horizontal gene transfer events. There are two identified GIs within the S. mutans UA159 genome designated TnSMU1 and TnSMU2 (Ajdić et al., 2002). TnSMU2 is large, consisting of over 50 kb, and is flanked by transposase remnants. It also has a distinct shift in G + C nucleotide content (~28%) relative to the rest of the S. mutans genome (37.5%; Waterhouse & Russell, 2006). The primary coding regions within this island have been characterized as a cluster containing nonribosomal peptide synthases, polyketide synthases, and accessory proteins involved in pigment synthesis as well as oxidative stress tolerance and biofilm formation (Wu et al., 2010). The region also contains a TetR-like regulator, SMU.1349, which can act as both as an activator for TnSMU2 genes and as a repressor of its own expression (Chattoraj, Mohapatra, & Rao, 2011). It is also notable however that TnSMU2 is not present in all S. mutans strains, with several studies reporting a range of clinical isolates failing to generate PCR products with primers targeted to this GI (Lapirattanakul et al., 2008;Wu et al., 2010).
During routine culturing of the S. mutans cidB mutant from different laboratory frozen glycerol stock cultures, we recently noticed the presence of two stable phenotypes based on colony morphology: a "rough" variant and a "smooth" variant. While the rough variant had been utilized for previous studies characterizing the physiological role of cidB (Ahn & Rice, 2016;Ahn et al., 2010), discovery of the smooth variant was novel. Hi-depth genome resequencing revealed this morphological difference may stem from the loss of TnSMU2, as well as an additional ~20 kb on the island's 3′ end. In this report, we explore the physiological differences between the cidB rough and cidB smooth mutants in relation to the presence or loss of the TnSMU2 region, respectively. Critical S. mutans physiological functions such as genetic competence, oxidative and acid stress resistance, and biofilm formation were also compared between UA159 and both cidB mutant variants. As a whole, this work addresses the physiological role(s) that cidB and TnSMU2 play in S. mutans UA159, and reinforces a previously established regulatory connection (Ahn & Rice, 2016) between these two loci.

| Colony morphology comparisons
To compare general trends in colony morphology between S. mutans UA159 and the cidA, cidB, and cidAB mutants, cells were grown 16-18 hr and then serially diluted in BHI. An equal volume (100 µl) of each 10 -7 dilution was spread-plated onto BHI agar plates and incubated for 48 hr at 37°C in either a regular plate incubator (atmospheric conditions), an incubator supplemented with 5% CO 2 , or in an anaerobic environment (Pouch-Anaero anaerobic Gas Generating System, Mitsubishi Gas Chemical Company, Japan). Representative colony images from n = 3 independent experiments were then captured using a Zeiss Stemi 305 Microscope and corresponding Labscope software (Zeiss, Germany). All images were taken at 10× magnification using darkfield settings.

| Genomic DNA isolation, Illumina Resequencing, and SNP analysis
To analyze and identify single nucleotide polymorphisms (SNPs) and other sequence changes, genomic DNA (gDNA) was isolated from S. mutans UA159, cidB rough, and cidB smooth strains. Cells were grown to late exponential phase (Optical density at 600 nm (OD 600 ) between 0.5-0.8) in BHI, at which time cell pellets were harvested from 7 ml of culture by centrifugation (3,900 × g, 10 min

| Real-time PCR (qPCR)
Relative copy number of genes located within and outside of the low-coverage TnSMU2 region of the cidB smooth mutant were verified using quantitative real-time polymerase chain reaction (qPCR).
Total gDNA was isolated from wild-type, cidB rough, and cidB smooth strains as described above and was used as template with the primer sets indicated in Table 1. qPCRs were performed using iQ SYBR Green Supermix (Bio-Rad, California, USA), 30 ng gDNA per reaction, and with primers at 250 nM per reaction on a Bio-Rad CFX Connect Real-Time System with the following conditions: 95°C for 3 min and 34 cycles of 95°C for 15 s followed by 55°C for 30 s.

TA B L E 1 qPCR primers used in this study
Relative copy number was then compared using the reported C T value for each reaction.

| Aerobic growth assay
Growth of S. mutans UA159 and isogenic cid mutants in aerobic conditions was assayed using a Bioscreen C automated growth system (Growth Curves USA). S. mutans cells were grown 16-18 hr in a chemically defined media (FMC; Terleckyj et al., 1975) and then diluted to an OD 600 of 0.02 per milliliter in fresh FMC. A honeycomb well plate (Growth Curves USA) was then inoculated at a 1:4 well to volume ratio, and cell optical density was recorded over 24 hr with constant shaking at 37°C.

| Acid tolerance assay
Each strain's ability to withstand acid stress was assayed by measuring cell viability over time after exposure to low pH conditions. Cells were first grown 16-18 hr in BHI and then diluted to an OD 600 of 0.02 in 5 ml of fresh BHI. Cultures were then grown for 4 hr before cells were harvested via centrifugation. Supernatant was then removed, and cell pellets were resuspended in 5 ml of 0.1 M glycine buffer at a pH of either 3.5 or 7, and incubated at 37°C, 5% CO 2 .
Samples were then removed at 0, 20, 40, 60, and 90 min incubation and serially diluted before being plated on BHI agar. Total colonyforming units (CFUs) were then enumerated after 48 hr growth at 37°C in a 5% CO 2 incubator.

| Dot drop competition assays
The ability of wild-type and cid mutant strains to compete against S. gordonii DL-1 and its isogenic spxB mutant was determined by a dot drop competition assay as described previously (Kreth, Merritt, Shi, & Qi, 2005). In brief, S. gordonii or spxB mutant cells were grown overnight for 16-18 hr in 0.5x BHI media and then diluted to an OD 600 of 0.5 in fresh 0.5× BHI media. About 10 µl was then dropped onto 0.5× BHI (Difco, BD) agar, and the plate was incubated overnight at 37°C in 5% CO 2 . The following day, 10 µl of each S. mutans competing species was inoculated alongside the S. gordonii drop in a similar manner and allowed to incubate an additional 24 hr at 37°C in 5% CO 2 before being photographed.

| Confocal microscopy and COMSTAT of static biofilms
In order to identify possible differences in biofilm formation between the cidB mutant variants, 5% CO 2 and anaerobic biofilms were cultivated in BM media (Loo et al., 2000)  Microscope using ZEN software (Zeiss, Germany). Z-stacks were generated using 0.5 µm slices at 63 × objective magnification with two random, center fields of view per well. Quantification of biofilm statistics was performed using COMSTAT (Heydorn et al., 2000) running on MATLAB R2010a (MathWorks) with manual thresholding on individual images collected on separate days.

| Cell viability measurement of static anaerobic biofilms
To assay biofilm cell viability, UA159, cidB rough, and cidB smooth biofilms were cultivated in semidefined sucrose biofilm media in anaerobic pouches as described above. After 48 hr growth, growth media was removed, and biofilms were scraped and resuspended in sterile 0.85% NaCl. Serial dilutions were plated on BHI agar, and total CFUs were then enumerated after 48 hr growth at 37°C in a 5% CO 2 incubator.

| CSP competence assays
To assay the ability of the cid mutants to take up externally added plasmid DNA, a quantitative competence assay was performed using a previously published protocol (Seaton, Ahn, Sagstetter, & Burne, 2011)

| XIP competence assays
Transformation efficiencies in chemically defined media were determined as described above, except that cultures were grown in defined FMC media (Terleckyj et al., 1975) and were supplemented with 1 µg synthetic SigX-inducing peptide (sXIP, Sigma-Aldrich) and 500 ng unmethylated pORI23 when cultures reached an OD 600 of 0.135-0.15.

| Statistical analysis
All statistical analyses were performed using SigmaPlot 13.0 (Systat Software). Data were tested for normality and equal variance prior to selection of appropriate parametric or nonparametric tests as indicated in each figure legend. Number of biological replicates analyzed in each experiment is specified in each figure legend.

| The cidB mutant yields two stable colony phenotypes
During routine culturing of cidB mutant frozen glycerol stocks, it was recently noted that two variant colony morphologies formed when struck out onto a BHI plate and grown in 5% CO 2 supplemented conditions. One cidB variant was rough and granular matching the colony phenotype of the parental wild-type strain S. mutans UA159, as well with isogenic cidA and cidAB mutants ( Figure 1). The other cidB variant was observed to be smooth and round, lacking the granularity of both wild-type and the cidB "rough" mutant ( Figure 1).
When grown anaerobically, this morphology repeated itself with the cidB "smooth" variant remaining round and mucoid while each of the other strains maintained their rough shape and granularity.
When cultured in conditions with atmospheric levels of oxygen, both cidB colony variants, as well as the cidAB mutant, failed to grow ( Figure 1), as was observed in our previous publication (Ahn et al., 2010). However, aerobic growth of the wild-type and the cidA F I G U R E 1 Influence of growth environment on S. mutans colony morphology. S. mutans wild-type (UA159) and isogenic cid mutants were grown on BHI plates at 37°C for 48 hr in either normal atmospheric conditions ("aerobic"), 5% CO 2 , or in an anaerobic pouch. All images were taken at 10× magnification and are representative of n = 3 independent experiments mutant displayed consistent colony morphology to that observed in anaerobic and CO 2 supplemented growth conditions. Thus, the two cidB mutant variants were renamed: cidB rough for the granular variant and cidB smooth for the round, mucoid variant. Both cidB variant colony morphologies were deemed stable, as repeated subculturing from frozen glycerol stocks and colonies of cidB rough and cidB smooth always yield all rough or smooth colonies, respectively.

| The cidB smooth mutant genome has lost TnSMU2 and neighboring genes
In order to determine what genetic variations may have led to altered colony morphologies of the cidB rough and cidB smooth variants, high depth whole genome resequencing was performed. Total gDNA was extracted from each of wild-type UA159, cidB rough, and cidB smooth, which were then sequenced using a MiSeq Illumina Platform to a depth of 1,000 reads. Deletion of the cidB gene was confirmed in both cidB variants ( Figure A1 in Appendix), with a small degree of aligned read noise attributed to cidB's sequence homology to the lrgB gene (SMU.575c). Single nucleotide polymorphisms (SNPs) were identified based on differences from the S. mutans UA159 reference genome (NC_004350.2) and showed two SNPs within the wild-type genomic resequencing that were not found in either cidB mutant (Table 2). SNP analysis of both cidB variants displayed many of the same SNPs as the isogenic wild-type strain (Tables 2-4 in Appendix), but also presented four variations found in both cidB variants that were not present in the reference genome or our wildtype strain (Table 1) Given that coverage of the TnSMU2 region by sequence reads was not completely absent in the cidB smooth mutant, qPCR was also performed on gDNA isolated from wild type, cidB rough, and cidB smooth, using primers specific for genes both within and outside the "low-read coverage" TnSMU2 region of the cidB smooth strain.
Primers were generated for two genes inside of the low-coverage region, bacA2 and gbpC, as well as two genes outside of the region, pdhB and comE (Table 1). Relative copy numbers were determined as a function of the C T values generated by each qPCR, with higher C T values indicating lower levels of initial gDNA template copy number available for amplification. Analysis of C T values for those genes outside of the TnSMU2 region indicated similar abundance for each of pdhB and comE, with little difference in C T values noted between each of the three strains ( Figure 3). C T values for bacA2 and gbpC products were also comparable between wild type and cidB rough, with mean C T values of ranging from 12 to 13. In the cidB smooth variant however, these C T values were significantly higher, ranging from 30 to 31 for gbpC and bacA2 products ( Figure 3). Although these qPCR data correlate with the sequencing read coverage patterns observed in wild type, cidB rough, and cidB smooth (Figure 2), it is not clear whether the TnSMU2 genomic region is in very low abundance or completely absent in the cidB smooth variant.

| Both cidB mutant variants are more sensitive to oxidative stress but display comparable acid tolerance relative to wild type
As a defect in oxidative stress tolerance had been previously ob-    (Figures 4, 5c), the cidB rough mutant may be better able to tolerate H 2 O 2 stress generated by S. gordonii on BHI agar plates relative to cidB smooth, presumably due to its intact TnSMU2 locus.
The ability for each cidB mutant variant to tolerate acid stress was also assessed by challenging wild type and each strain to pH 3.5 ( Figure A3 in Appendix). In this experiment, all strains displayed comparable decreases in cell viability over time in the pH 3.5 treatment condition (~99% loss of viability by 90 min) and demonstrated similar levels of survival in the pH 7.0 control condition. These results suggest that neither CidB nor TnSMU2 is required for acid tolerance under the conditions tested in this study.

| Both cidB and the TnSMU2 genomic region influence S. mutans biofilm structure
Biofilm formation is another key physiological process of S. mutans which has been previously described as impacted by the Cid/Lrg system (Ahn et al., 2010). Both CO 2 and anaerobic growth conditions were chosen in order to study the alterations in biofilm  (Figure 7b-d). These analyses revealed a modest trend in increased biofilm biomass and thickness for wild type F I G U R E 3 qPCR gene copy measurements of TnSMU2-related genes. Relative copy numbers were determined by qPCR using wildtype (UA159), cidB rough, and cidB smooth genomic DNA. The C T value represents the qPCR cycle at which detectable amplification occurred for each gene product. Data represent the average of n = 3 (wild type, cidB rough) or n = 4 (cidB smooth) independent experiments; error bars display standard deviation with * denoting statistical significance relative to wild type (Dunn's test (bacA2)  during anaerobic growth compared with CO 2 growth, but these results were not statistically significant (p > .05, Mann-Whitney rank sum test, Figure 7b, c). The opposite trend was noted however with the roughness coefficient (Figure 7d), whereby wild-type anaerobic biofilm values were significantly reduced relative to wild-type CO 2 biofilms (p < .001, Mann-Whitney rank sum test), suggesting that anaerobic wild-type biofilms are more homogenous than CO 2 grown samples. The cidB rough mutant biofilms were closely matched to wild type in terms of biomass, thickness, and roughness coefficient in the CO 2 growth condition, but anaerobic cidB rough mutant biomass and thickness were reduced almost twofold compared with wild-type under this same condition (Figure 7b, c), which correlates with the decreased CFU counts observed in this strain (Figure 7a).
An almost threefold increase in roughness coefficient was also observed in the anaerobic cidB rough biofilm relative to wild-type properties that are observed in opposite growth conditions. Given that the increased cell PI staining phenotype was unique to the cidB smooth mutant, it is likely that this is associated with genomic loss of TnSMU2 and/or neighboring genes.

| Mutation of cidB and loss of TnSMU2 correlate with altered competence
We had previously observed that mutation of lrgA leads to CSPrelated competence deficiency (Ahn, Qu, Roberts, Burne, & Rice, 2012). Therefore, the influence of cid genes on the ability of S. mutans to uptake foreign DNA was tested in this study using quantitative competence assays in both complex and defined media. Both media were tested in order to probe both the CSP-and XIP-mediated competence systems found within S. mutans. With addition of synthetic CSP (sCSP), each of the wild type and cid mutants had similar transformation efficiencies (Figure 8 and Figure A4 in Appendix) with cidB smooth being slightly lower than wild type and the cidB rough variant (Figure 8a). As expected, transformation efficiencies without the addition of sCSP were dramatically lower for all strains.
F I G U R E 5 Dot drop competition between S. mutans cidB mutants and S. gordonii, and quantitative H 2 O 2 challenge assay. The ability of S. gordonii wild-type (Sg) (a) and isogenic spxB mutant (b) to inhibit S. mutans wild-type (UA159) and isogenic cidB mutant growth was assessed using a dot drop competition assay as described in Materials and Methods. Data represent n = 3 independent experiments. The ability to grow planktonically in FMC containing 1 mM H 2 O 2 was also quantified in all three strains by OD 600 measurements over a 24-hr period (C). Data represent n = 3 independent experiments, error bars = SEM However, the cidB smooth transformation efficiency was significantly lower (~1 log reduction, p < .05, Student-Newman-Keuls test) compared with the wild-type strain and cidB rough (Figure 8a).
Mutation of either cidA or cidAB did not alter transformation efficiency in BHI media in the presence or absence of sCSP ( Figure A4 in Appendix).
When grown in defined media and supplemented with synthetic XIP (sXIP), distinct phenotypes were observed with each cidB variant. The transformation efficiency of the cidB rough variant was modestly decreased compared with wild type (Figure 8b; ~0.5 log decrease, p = .014, Holm-Sidak Test), but the cidB smooth mutant transformation efficiency was significantly lower with a F I G U R E 6 Representative biofilm images of wild-type, cidB rough, and cidB smooth mutants. Confocal microscopy images of wild-type, cidB rough, and cidB smooth biofilms grown for 48 hr in semidefined media with 20 mM sucrose. Biofilms were stained for viable (green) or dead/damaged (red) using Syto-9 and propidium iodide, respectively. Images show the orthogonal view of the biofilm centered on a Z-stack in the first third and are representative of n = 18-27 random fields of view taken over n = 3-5 independent experiments F I G U R E 7 CFU and COMSTAT analysis of wild-type, cidB rough, and cidB smooth biofilms. Quantification of viable biofilm cells by CFU serial dilution plating (a) was performed on 48 hr UA159 (wild-type), cidB rough, and cidB smooth anaerobic biofilms. Data represent n = 3 biological replicates, error bars = SEM. * represents significant difference compared to wild type (p < .01, Holm-Sidak). Total biomass (b), average biofilm thickness (c), and the roughness coefficient (d) were quantified using pixel density and calculated through the COMSTAT algorithm (Heydorn et al., 2000) in MATLAB. The roughness coefficient represents the heterogeneity of the biofilm surface, with higher values indicative of a less uniform surface. Light bars represent biofilms generated during CO 2 growth, and dark bars represent biofilms generated during anaerobic conditions. Data represent the average of n = 18-27 random fields of view acquired over n = 3-5 independent experiments. Error bars = standard error of the mean (SEM), * represents significant difference between CO 2 growth conditions compared to wild type (p < .05, Holm-Sidak for biomass and thickness, Dunn's test for Roughness coefficient), ** represents significant difference between anaerobic growth conditions compared to wild type (p < .05, Dunn's test), # represents significant difference between anaerobic wild type and CO 2 wild type (p < .001, Mann-Whitney rank sum test)

| D ISCUSS I ON
In order to successfully persist and cause disease within the oral cavity, S. mutans must be able to survive numerous environmental stresses. The cidB gene has been shown to be an important con- Change in colony morphology due to loss of the TnSMU2 region may not be from the loss of any one of these ORFs, but the collective loss of all related genes may cause larger physiological changes that alter the appearance of colonies on an agar plate. It is also important to note that TnSMU2 is not found in all S. mutans strains (Lapirattanakul et al., 2008;Wu et al., 2010) and that many S. mutans backgrounds have differing colony morphologies (Emilson, 1983).
Previous work in other serotype c S. mutans isolates has identified large, mucoid colonies similar in description to the cidB smooth strain, and characterized this alteration as a result of altered fructosyltransferase (FTase) activity (Okahashi, Asakawa, Koga, Masuda, & Hamada, 1984). Although we have not identified any genes within this region predicted to take part in fructan synthesis, the possibility remains that one of the many hypothetical proteins may indeed have a role in this process.
Due to the presence of low-read coverage within the TnSMU2 region of the cidB smooth mutant genome, we are not able to conclusively state this region has been completely lost. In order to better understand if genes within this region were present, as well as to quantify the gap in coverage, qPCR primers were targeted to genes within the region. These results confirmed that copy numbers of comE and pdhB, two genes outside of the low-coverage area, are consistent throughout our wild-type S. mutans and both cidB mutant variants. This was expected, as neither gene showed gaps in coverage during our sequence analysis even though pdhB resides on the edge of the 3′ end of the low-coverage region. However, relative quantities for bacA2 and gbpC differed in this qPCR analysis: Both genes were present in comparable relative copy number between wild-type and cidB rough genomic DNA (as indicated by near-identical C T values to each other, which were also in line with C T values observed for pdhB and comE for all three strains). However, bacA2 and gbpC C T values in the cidB smooth strain were very high (~30), indicating a low copy number relative to wild type and cidB rough.
Although qPCR amplification of these genes was detectable in cidB smooth, their C T values approach the qPCR cutoff which is normally considered background (C T ≥ 35). This result alone still does not clarify whether or not this region is completely absent in the cidB smooth genome, or whether the TnSMU2 region is somehow being maintained in a small subpopulation of this strain. Experiments are currently in progress to distinguish between these two scenarios, as well as to probe the frequency of loss of TnSMU2 in cidB mutants in both previously and newly constructed mutants.
Two vital physiological processes for S. mutans persistence within the oral cavity are biofilm formation and the ability to withstand oxidative stress (Kreth et al., 2005;Tsumori & Kuramitsu, 1997). While acid tolerance was not shown to be affected in either cidB mutant variant ( Figure A3 in Appendix), mutation of cidB has previously been shown to have a significant effect on oxidative stress tolerance in S. mutans grown in BHI (Ahn & Rice, 2016) and on agar plates in atmospheric levels of O 2 (Ahn et al., 2010). Congruent with these results, neither cidB variant nor the cidAB mutant grew successfully in atmospheric aerobic conditions planktonically (Figure 4), or on agar plates (Figure 1). The cidA mutant however was able to grow comparatively well in defined media during aerobic growth. Competing each cidB variant against S. gordonii did reveal a modest separation of these two mutants in terms of H 2 O 2 resistance when cultured on BHI agar plates (Figure 5a). The cidB smooth mutant appeared to display greater inhibition by wild-type S. gordonii, with a larger area of obstructed growth as directly compared with either the cidB rough variant or wild-type (Figure 5a). This phenotype was lost when competed against a S. gordonii spx mutant which is defective in H 2 O 2 production, confirming this inhibition was a function of H 2 O 2 generated by S. gordonii rather than production of an excreted secondary metabolite or peptide. Increased inhibition of cidB rough as compared with wild type was also observed, suggesting that cidB itself is important for competitive fitness within the oral cavity, but loss of the TnSMU2 region in cidB smooth may contribute to its increased growth inhibition by H 2 O 2 -producing S. gordonii on BHI agar plates. Deletion of the bac/mub operon (encoding a nonribosomal peptide synthetase-polyketide synthase gene cluster responsible for pigment biosynthesis) within TnSMU2 has been previously shown to increase sensitivity to oxidative stress in three different S. mutans strain backgrounds (Wu et al., 2010). Therefore, it is likely that loss of these genes in the cidB smooth mutant contributes to its increased growth inhibition by H 2 O 2 -producing S. gordonii. Defining a specific stress-related effect of elimination of this region is made even more complex due to the variability of this region seen in S.

Assessment of biofilm formation between the cidB variants also
revealed an interesting effect of this gene on the structure and amount of biofilm generated by S. mutans. During growth in CO 2 supplemented conditions, biofilms produced by cidB rough closely resembled those of wild type, whereas biofilms produced by cidB smooth were significantly diminished with lower biomass and average thickness under this same growth condition. This cidB smooth mutant biofilm phenotype may be driven in part by the loss of the glucan binding protein gbpC, which is not a part of TnSMU2 but was part of the 3′ end missing from cidB smooth (Figure 2). Biofilm growth of cidB smooth under anaerobic conditions reversed the phenotypic effects observed in CO 2 growth. In comparison, biofilms formed by the cidB rough mutant were altered during anaerobic growth (decreased biomass and thickness) but not during CO 2 growth, suggesting that loss of cidB alone leads to a reduction of anaerobic biofilm formation, while concurrent loss of the TnSMU2 region may compensate this biofilm reduction in some manner. Viable cell counts within each anaerobic biofilm reflected these findings, as the cidB rough mutant displayed a significantly lower cell count than either the wild type or cidB smooth. Indeed, the smooth variant anaerobic biofilm displayed the highest viable cell count of all three tested strains despite the increase in PI staining observed in the confocal images ( Figure 6). The function of cidB may thus be tied to S. mutans anaerobiosis and persistence. Loss of the large coding sequences within the TnSMU2 region may also reduce the overall carbon requirements for S. mutans in the cidB smooth mutant during anaerobic growth, possibly allowing it to maintain wild-type levels biofilm formation despite increased presence of cell damage.
Reductions in genetic competence can prevent horizontal gene transfer within the oral microbiome, disabling transfer of antibiotic resistance genes (Hannan et al., 2010;Li, Lau, Lee, Ellen, & Cvitkovitch, 2001), disrupting acquisition of MGEs (Ciric, Mullany, & Roberts, 2011), and/or decreasing subpopulation heterogeneity and overall fitness of the S. mutans population. In this study, a significant decrease in the ability for both cidB variants to uptake plasmid DNA was shown in defined media where the XIP-medi-  (Khan et al., 2016). Some of these genes, such as the hypothetical 3-isopropylmalate dehydrogenase SMU.1383, displayed increased expression in the wild-type background at both timepoints, but decreased expression in a comS mutant.
Others, such as gbpC, displayed consistent expression profiles in all three treatments with increased expression throughout.
While several were stated to be statistically significant, many of these alterations failed to clear a mean fold change cutoff of 2.0.
Regulation of competence is incredibly complex; thus, alterations of expression for many of these genes may be downstream effects from more significant changes in other loci. Further experiments beyond the scope of this current study are necessary to determine what the causal agent of the observed competence defects is in both cidB variants or if these competence defects derive directly from the loss of TnSMU2 and its downstream region.
While the precise function of CidAB remains a mystery, this study has reinforced a link between these genes and the genomic island TnSMU2. Our data also provide novel insight into the physiological role CidB plays within the cell, in addition to how the presence or absence TnSMU2 affects S. mutans physiology.
Determining the importance of cidB and TnSMU2 in S. mutans oxidative stress tolerance, biofilm production, and competence helps provide new insights as to how this bacterium is able to persist within the oral cavity and cause disease, while also providing new evidence in regard to the function of these relatively uncharacterized areas of S. mutans physiology.

CO N FLI C T O F I NTE R E S T
None declared.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analyzed during this study are included in this published article and its corresponding appendices.
Additionally, all DNA raw sequencing data files are accessible via the NCBI Sequence Read Archive (SRA) under the accession number PRJNA560077.

E TH I C S S TATEM ENT
None required. Grey SNPs common to WT, cidB rough, and cidB smooth relative to published S. mutans UA159 genome (NC_004350.2). Green: SNPs only found in cidB rough variant.