Conserved and divergent functions of RcrRPQ in Streptococcus gordonii and S. mutans

Abstract

In the dental caries pathogen Streptococcus mutans, an MarR-like transcriptional regulator (RcrR), two ABC efflux pumps (RcrPQ) and two effector peptides encoded in the rcrRPQ operon provide molecular connections between stress tolerance, (p)ppGpp metabolism and genetic competence. Here, we examined the role of RcrRPQ in the oral commensal S. gordonii. Unlike in S. mutans, introduction of polar or non-polar rcrR mutations into S. gordonii elicited no significant changes in transformation efficiency. However, S. gordonii rcrR mutants were markedly impaired in their ability to grow in the presence of hydrogen peroxide, paraquat, low pH or elevated temperature. Sensitivity to paraquat could also be conferred by mutation of cysteine residues that are present in the RcrR protein of S. gordonii, but not in S. mutans RcrR. Thus, stress tolerance is a conserved function of RcrRPQ in a commensal and pathogenic streptococcus, but the study reveals additional differences in regulation of genetic competence development between S. mutans and S. gordonii.

Divergent regulation and different roles in genetic competence of a transcriptional repressor and two ABC efflux pumps are elucidated in a commensal streptococcus when compared with a related oral pathogen.

Graphical Abstract Figure.

Divergent regulation and different roles in genetic competence of a transcriptional repressor and two ABC efflux pumps are elucidated in a commensal streptococcus when compared with a related oral pathogen.

INTRODUCTION

Streptococcus gordonii is an abundant, commensal streptococcus in early oral biofilms that can adhere to multiple surfaces (Jenkinson and Lamont 1997; Nobbs, Lamont and Jenkinson 2009), engage in coadhesion with other oral bacteria, antagonize the growth of pathogens (Wang and Kuramitsu 2005; Kreth, Zhang and Herzberg 2008) and moderate acidification of oral biofilms through arginine metabolism (Nascimento et al. 2009). Because the environments in early and mature supra- and subgingival dental plaque are very different and constantly changing, S. gordonii must adapt to major and substantial variations in pH, nutrient source and availability, oxygen and redox potential, and other inputs that have profound impacts on bacterial gene expression and physiology.

Recently, an MarR-like transcriptional regulator, RcrR (rel-competence related), in S. mutans, a dental caries pathogen, was shown to be an autogenous negative regulator of the expression of genes for two ABC efflux pumps (RcrPQ) encoded in rcrRPQ operon (Seaton et al. 2011). Like other MarR-like regulators, S. mutans RcrR plays an essential role in stress tolerance. However, in concert with RcrPQ, and two effector peptides, it also integrates environmental stimuli and cellular physiology with the decision to become genetically competent (Seaton et al. 2011; Ahn et al. 2014; Seaton, Ahn and Burne 2014).

For all species of streptococci, ComX (sometimes called SigX) is a sigma factor that is required for late competence gene expression and transformation (Lee and Morrison 1999). However, multiple pathways exist for activation of comX transcription. In S. gordonii, comX expression is induced by the response regulator ComE after the ComD sensor kinase is activated by competence stimulating peptide (CSP) (Håvarstein et al. 1996). The transcriptional activator for comX in S. mutans is the ComR-XIP (comX-inducing peptide) complex (Mashburn-Warren, Morrison and Federle 2010). Therefore, the pathways for regulating genetic competence are divergent in these two species of streptococci.

The molecular mechanisms that couple genetic competence with virulence-related traits and stress tolerance in streptococci have been elusive, but the recent discovery of the essential roles of RcrRPQ represents a significant breakthrough in this area. The objective of this study was to characterize the rcrRPQ operon in S. gordonii to determine whether the roles that RcrRPQ play in the pathogen S. mutans were conserved in an organism that has an evolutionarily distinct competence cascade and is more commonly associated with oral health than disease.

MATERIALS AND METHODS

Strains and growth conditions

Strains used in this study are listed in Table S1 (Supporting Information). Streptococcus gordonii and S. mutans were routinely cultured in brain heart infusion (BHI) broth (Difco) at 37°C in a 5% CO₂, aerobic atmosphere. Escherichia coli strains were grown in Luria broth at 37°C with shaking at 250 rpm. Antibiotics were used at the following concentrations: kanamycin (500 μg mL⁻¹ for S. gordonii, 1000 μg mL⁻¹ for S. mutans), spectinomycin (1000 μg mL⁻¹ for S. mutans and S. gordonii, 50 μg mL⁻¹ for E. coli) and erythromycin (5 μg mL⁻¹ for S. gordonii).

Construction of mutant strains

Routine DNA manipulation techniques were used to engineer plasmids and strains (Sambrook and Russell 2001). Streptococcus gordonii rcrR mutants were created using a PCR ligation mutagenesis approach that replaced the majority of the rcrR open reading frame with either polar or non-polar kanamycin cassettes. Primers (Table S2, Supporting Information) were used to amplify flanking regions overlapping the 5^′ and 3^′ ends of the rcrR gene. PCR products and antibiotic resistance cassettes (from plasmids pUC19ΩKm and pALH124) (Ahn and Burne 2006; Seaton et al. 2011) were digested with BamHI and ligated together. The ligation mixture was added to competent S. gordonii and transformants were selected on BHI agar containing Km. Insertion of antibiotic resistance cassettes was confirmed using colony PCR and DNA sequencing to verify that no mutations were introduced into the sequences used for the recombination events.

Gene expression analysis

Quantitative real-time reverse transcriptase PCR (qRT-PCR) was performed to measure mRNA of S. gordonii rcrPQ and manA, the gene directly downstream of rcrQ that encodes for a α-1,2-mannosidase. Streptococcus gordonii DL1 and the mutant strains were harvested in mid-exponential phase (OD₆₀₀ = 0.5), treated with RNAprotect (Qiagen) and total RNA was extracted using phenol (pH 4.3) and the RNeasy mini kit (Qiagen). RNA concentration was determined using a Nanodrop spectrophotometer (Thermo Scientific, Wilmington, DE). Purified RNA (1 μg) was used to synthesize cDNA according to the Superscript III (Invitrogen) supplier's protocol. Primers for each gene were designed using the qPCR settings in the Primer3plus online application (Untergasser et al. 2007) (Table S2, Supporting Information). Standard curves for each gene were generated using eight 10-fold dilutions of PCR products, starting with 10⁸ copies μL⁻¹. Triplicates of standard curve DNAs, samples and cDNA controls were added to wells containing iQSYBR green supermix (Bio-Rad) with primers (0.4 μM). Thermocycling was carried out using an iCycler iQ real-time PCR detection system (Bio-Rad) set to the following protocol: 40 cycles of 95°C for 10 s and 60°C for 45 s, with a starting cycle of 95°C for 30 s. Analysis of the data was performed as described elsewhere (Ahn et al. 2005).

Transformation assays

Streptococcus gordonii strains were diluted 1:100 from overnight cultures into pre-warmed BHI (0.8 mL aliquots) and incubated at 37°C in a 5% CO₂ aerobic atmosphere for 30 min. Then, synthetic CSP (H-DVRSNKIRLWWENIFFNKK-OH) or scrambled CSP (same composition, different sequence: H-DKRFKWWILKVFNSNEINR-OH) was added to the cultures to a final concentration of 1 μM, along with 500 ng of purified pIB184 (Em^r). After a further 4 h of incubation, serial dilutions of cell cultures were made and plated onto BHI agar with or without erythromycin (5 μg mL⁻¹) to enumerate the transformants and total CFU, respectively. Colonies were counted after 48 h and transformation efficiency (i.e. percent transformants of viable cells) was calculated as the (number of transformants × 100)/(total CFU).

Stress response analysis

Growth of S. gordonii and S. mutans strains was monitored using a Bioscreen C (Growth Curves USA, Piscataway, NJ). Overnight cultures were diluted 1:50 into pre-warmed BHI, grown to mid-exponential phase (OD₆₀₀ = 0.5), diluted 1:100 into 350 μL of fresh BHI broth and transferred to Bioscreen micro-well plates. The OD₆₀₀ of cultures was then monitored by the Bioscreen C system. A sterile mineral overlay (50 μL) was placed on top of cultures, when desired, to avoid inhibitory effects on growth of exposure to oxygen. Hydrogen peroxide (H₂O₂, 1 mM), 2 mM sodium nitrite (NaNO₂) or paraquat (25 mM) was added to cultures to assess growth during exposure to oxidative stress. Growth in BHI that had been acidified to pH 5.5 using HCl was used as a measure of acid tolerance. Bioscreen plates were also incubated at 43°C to assess the effects of particular mutations on thermal stress.

Complementation of mutant strains

Vectors containing wild-type SGO_1752 (S. gordonii rcrR) or SMu.921 (S. mutans rcrR) were created by amplifying the genes from the parental strains, including sequences 100 bp upstream and 50 bp downstream of the structural gene. Amplified DNA was cloned into the E. coli–Streptococcus shuttle vector pDL278 (LeBlanc, Lee and Abu-Al-Jaibat 1992) using HindIII and SacI for SGO_1752, or BamHI and SacI for SMu.921. Ligation reactions were mixed with competent E. coli DH10-beta (New England Biolabs, Ipswich, MA) and transformants were selected on LB agar supplemented with spectinomycin. Cloning was confirmed by Sanger sequencing and purified plasmid was then transferred to S. gordonii or S. mutans strains. Transformants of streptococcal strains were verified again by Sanger sequencing. The two cysteine residues at positions 26 and 118 in the S. gordonii RcrR protein were changed to serine residues by mutating the TGT codons to TCT on the complementing plasmids. In these cases, site-specific mutagenesis was carried out using the Q5^® site-directed mutagenesis kit (NEB) according to the suppliers's protocol. Substitution primers were designed using the NEBaseChanger online application. Primers used during complementation are listed in Table S2 (Supporting Information).

RESULTS AND DISCUSSION

Bioinformatic analysis of S. gordonii rcrRPQ

Multiple sequence alignment revealed that the protein sequence of RcrR is well conserved in the bovis, pyogenic, salivarius, mutans, mitis and anginosus groups of streptococci (Fig. 1A). The rcrRPQ operon is flanked by different genes in S. gordonii and S. mutans (Fig. 1B). In S. gordonii DL1, rcrRPQ is downstream of glmS (glucosamine-6-phosphate activated ribozyme) and scrK (fructokinase), and upstream of manA (α-1,2-mannosidase); gene encoding products that are involved in metabolism of carbohydrates and that may interact, according to the STRING 9.1 database (S1) (Franceschini et al. 2013). By contrast, the rcrRPQ operon of S. mutans UA159 is upstream of a gene for a thiol peroxidase (tpx) that is involved in oxidative stress resistance, the cipI gene encoding an endogenous bacteriocin resistance protein and the relPRS operon; RelP being the primary enzyme catalyzing (p)ppGpp accumulation during exponential growth and RelRS is a two-component system that influences relPRS expression (Lemos et al. 2007; Seaton et al. 2011). Importantly, the peptides encoded in the very 3^′ end of the S. mutans rcrQ gene that affect competence and CSP sensitivity could not be found in the 3^′ end of S. gordonii rcrQ. Although a putative peptide (56-aa) encoded in a different reading frame than rcrQ was found, it is larger than, and shares no significant sequence homology with, the S. mutans rcrQ peptides (BLASTP analysis) (Fig. 1C).

Figure 1.

RcrR alignment and rcrRPQ genome placement among selected Streptococcus species. (A) Sequence alignment of streptococcal RcrR amino acid sequences performed using ClustalW2, where *, : and . indicate identical, similar and weakly similar residues respectively. The helix-turn-helix DNA-binding domain is highlighted by a gray box. For S. gordonii, cysteine residues (at positions 26 and 118) are highlighted in red and circled. (B) Genome context of the rcrRPQ operons of S. mutans UA159 and S. gordonii DL1 rcrRPQ. (C) Peptide identified in the 3^′ end of rcrQ. The TAA stop codon of rcrQ and the ATG start codon of manA are shown in bold plus italics.

Mutations in rcrR do not affect competence development by S. gordonii

Polar and non-polar mutants of S. gordonii rcrR were generated by allelic exchange mutagenesis as detailed in the ‘Methods’ section. In S. mutans, the expression levels of the rcrPQ genes are strongly associated with competence behaviors in polar and non-polar mutants, and even in strains expressing an intact RcrR protein (Seaton et al. 2011; Ahn et al. 2014; Seaton, Ahn and Burne 2014). Quantitative real-time PCR was used to measure the transcript levels in S. gordonii of rcrP and rcrQ, and the gene immediately downstream of the rcrRPQ operon, manA, which encodes a product required for mannose utilization. Expression of rcrPQ was increased roughly 20-fold in the non-polar S. gordonii rcrR mutant (Fig. S1, Supporting Information). For the polar mutant, constructed with a kanamycin marker that contains transcription terminators, expression of rcrP and rcrQ mRNA was 5-fold and 16-fold lower, respectively, than in wild-type S. gordonii (Fig. S1, Supporting Information). No significant differences in manA mRNA levels were found between wild-type S. gordonii and the rcrR mutants in cells growing exponentially with glucose as the primary carbohydrate source (Fig. S2, Supporting Information). Interestingly, although no growth defect was observed for the rcrR mutants growing with glucose as the carbohydrate source, both rcrR mutant strains showed slower growth on mannose (Fig. S2; Table S3, Supporting Information). This finding will require further investigation, but we cannot exclude that manA expression may be altered in rcrR mutants at other phases of growth or in growth conditions different than those used to measure manA levels.

Interestingly, neither of the rcrR mutants of S. gordonii differed from the parental strain in their ability to be transformed regardless of whether cells were treated with synthetic CSP, which greatly enhanced the percentage of cells that could be transformed, or with a peptide of the same size and composition as CSP with a different primary sequence (scrambled CSP, see Methods section) (Fig. 2). The two major differences between S. mutans and S. gordonii rcrRPQ are the lack of peptides encoded at the 3^′ end of rcrQ and the complete absence of any impact of changes in rcrR or rcrPQ expression levels on competence. These findings may be explained, at least in part, by the fact that XIP-ComR, not ComE, is the proximal regulator of comX expression in S. mutans and that the peptides and RcrPQ have been proposed to exert their influence on the ComRS pathway.

Figure 2.

Transformation efficiencies of S. gordonii DL1, S. gordonii rcrR-P and S. gordonii rcrR-NP strains. The assay was performed with 500 ng plasmid DNA plus 1 μM synthetic CSP, or 1 μM synthetic peptide of identical composition, but scrambled sequence (Scrmb). Independent data are shown as a scatterplot, where open circles show average measurements from each replicate (three total) and black lines show group medians.

rcrRPQ mutants of S. gordonii have major defects in stress tolerance

A strong connection of RcrRPQ with the tolerance of stressors that are common in oral biofilms was demonstrated in S. mutans by showing that mutants lacking the RcrPQ transporters were impaired in growth at low pH, in the presence of air or when exposed to the superoxide anion-generator paraquat (Seaton et al. 2011). The ability of the S. gordonii rcrR polar or non-polar mutants to tolerate a spectrum of environmental stressors was evaluated in planktonic cultures (Fig. 3; Table S3, Supporting Information). Although growth in the presence of air or of nitrite was not altered in mutants, compared to the wild-type strain, clear differences were observed for the mutants growing in the presence of hydrogen peroxide, paraquat, low pH and increased temperature. These phenotypes were reversed by complementation with wild-type rcrR (data not shown).

Figure 3.

Growth curves of S. gordonii DL1 and rcrR mutant strains in the presence of 1 mM H₂O₂, 25 mM paraquat, oxygen, 2 mM NaNO₂, pH 5.5 or 43°C. Red lines indicate the wild-type strain, blue lines indicate the rcrR-P strain and green lines indicate the rcrR-NP strain. Data are representative of three independent experiments performed at least in triplicate. Final optical densities and doubling times are available in Table S3 (Supporting Information).

The oral biofilm is a dynamic environment with substantial changes in oxygen, its metabolites and redox, pH, temperature and carbohydrate availability. Most likely, RcrRPQ allows S. gordonii to cope with a constantly changing environment and certain stressors via modulation of gene expression by RcrR and removal of deleterious substances by the two ABC efflux pumps (RcrPQ). With the tight intertwining of competence with RcrRPQ levels in S. mutans and the intimate association of competence regulation with stress tolerance in this organism, it is difficult to ascribe particular phenotypes directly to RcrR or RcrPQ, since changes in these proteins translate to changes in expression of the competence regulon. However, absent an influence of RcrRPQ on competence in S. gordonii, one would predict a direct involvement of the RcrR transcriptional regulator and RcrPQ efflux pumps in coping with stress in this organism. In light of these new findings, contrasting the RcrR and competence regulons of these two organisms could lead to new strategies to interfere with growth of S. mutans without perturbing the ability of S. gordonii to persist.

Redox sensing by S. gordonii RcrR

One striking difference between the RcrR proteins of S. gordonii and S. mutans is the presence of two cysteine residues in the S. gordonii protein, whereas RcrR of S. mutans is devoid of cysteine (Fig. 1A). To determine whether these cysteine residues were functionally significant, the rcrR mutation in the rcrR-NP strain was complemented, either by using a wild-type copy of rcrR or one that was altered so that either, or both, cysteine residues were changed to serines. Growth assays confirmed that wild-type rcrR complemented the rcrR-NP strain, restoring resistance to stressors to which the rcrR-NP strain was sensitive. For example (Fig. 4), the wild-type and complemented strains exited the lag phase significantly faster and attained higher final optical densities when grown in the presence of paraquat. Importantly, all cysteine to serine mutant strains (C26S, C118S and the C26S/C118S double mutant) exhibited increased lag phases in the presence of 25 mM paraquat and lower final optical densities than S. gordonii DL1 or the rcrR-NP strain carrying a wild-type copy of rcrR. There are instances where RcrR proteins of other Streptococcus spp., including S. mitis and S. pneumoniae (Table 1), contain cysteine. In this case, it appears that S. gordonii RcrR operates as a 2-Cys redox sensor similar to the Enterococcus faecium AsrR protein (Cys-11, Cys-61), which contains cysteine residues that sense superoxide anion (Lebreton et al. 2012).

Figure 4.

Contribution of cysteine residues to superoxide stress sensing of S. gordonii RcrR. Growth curves in the presence of 25 mM paraquat of S. gordonii rcrR-NP complemented strains. Data are representative of three independent experiments performed at least in triplicate.

Table 1.

Distribution of cysteine residues in RcrR among Streptococcus spp.

Species	N-terminus	C-terminus
Anginosus and Mitis Groups
Streptococcus intermedius B196	C48	C118
Streptococcus anginosus SK52	C48	C118
Streptococcus gordonii DL1	C26	C118
Streptococcus oralis SK313	C26, C44
Streptococcus mitis B6	C26
Streptococcus pneumoniae R6	C26
Mutans, Salivarius, Bovis and Pyogenes Groups
Streptococcus mutans UA159
Streptococcus salivarius 57.1	C42
Streptococcus thermophilus LMD-9
Streptococcus gallolyticus UCN34
Streptococcus agalactiae NEM316
Streptococcus pyogenes M1		C108
Others
Lactococcus lactis KF147
Streptococcus suis D9	C34

To begin to probe in more detail the roles of S. gordonii and S. mutans RcrR proteins in sensing stress, non-polar rcrR strains of S. mutans and S. gordonii were complemented with rcrR from the other species. Whereas S. gordonii RcrR reversed the paraquat sensitivity of S. mutans rcrR-NP, S. gordonii rcrR-NP complemented with S. mutans rcrR was unable to grow in 25 mM paraquat (Fig. 5), although this strain grew well in non-stressed conditions (S3). We posit that the reason that S. mutans RcrR did not restore oxidative stress tolerance to the S. gordonii rcrR mutant is that derepression of rcrRPQ is critical to tolerance of ROS, perhaps superoxide anion in particular. Because S. mutans RcrR lacks the cysteine residues found in the S. gordonii protein, there may be no capacity for sensing this stressor and relieving repression of the operon. In terms of cis-acting elements, the RS1 and RS2 binding sites identified in S. mutans are both recognized in vitro by purified RcrR, but RS1 appears to function as the primary binding site (Seaton, Ahn and Burne 2014). The RS1 site identified in S. gordonii (Fig. 5) shares 75% sequence homology with that of S. mutans, whereas the RS2 sites were only 35% identical. Therefore, we cannot exclude that differing affinities for the predicted binding sites that are present in the S. gordonii rcrRPQ promoter region (Fig. 5), as well as gene dosage effects of rcrR, contribute to the behaviors of strains in the cross-species complementation experiments.

Figure 5.

Cross-complementation of S. gordonii and S. mutans rcrR-NP with the rcrR gene of the other species. (A) Consensus RcrR binding site sequence generated by RegPrecise (Novichkov et al. 2010). The binding site was derived from 21 sequences. (B) Predicted RcrR promoter binding regions in S. gordonii and S. mutans rcrR promoter sequences. Two putative binding sites exist in both promoter regions, RS1 (in red) and RS2 (in blue). −35 and −10 transcriptional start regions are boxed and were predicted using the BPROM software for identification of bacterial promoters (Softberry). The predicted −35 (TTGACT) and −10 (TAGAAT) each differ from the consensus sigma A-type promoter by 1 nt. Growth curves of S. gordonii (C) and S. mutans (D) rcrR-NP strains complemented with the rcrR gene of the other species (purple) in comparison to wild-type (red) and rcrR-NP (green) strains in the presence of 25 mM paraquat. Data are representative of three independent experiments performed at least in triplicate. Lag times (standard error) for strains were as follows: DL1, 12.3 h (±0.21 h); Sg rcrR-NP, 17.1 h (±0.67 h); rcrR:SMu.921, not applicable; UA159, 9.6 h (±0.34 h); Sm rcrR-NP, 12.9 h (±0.34 h); rcrR:SGO_1752, 7.7 h (0.61).

CONCLUSION

Divergence exists in the roles of RcrRPQ in genetic competence, but contribution to stress tolerance is conserved in S. gordonii and S. mutans. RcrR of S. gordonii appears to have evolved to play a more direct role in redox sensing, which is of interest because commensal streptococci often produce large amounts of hydrogen peroxide, which is a potent antagonist of many of the late colonizers of oral biofilms, including S. mutans. The enhanced redox-sensing capacity of RcrR in S. gordonii may confer a competitive advantage in mixed-species oral biofilm by tightly controlling the expression of the RcrPQ transporters. Future studies to explore this hypothesis in mixed-species in vitro models or competition studies in a suitable in vivo model would be of value since therapies may be designed that exploit these fundamental differences in rcrRPQ operon regulation.

SUPPLEMENTARY DATA

Supplementary data are available at FEMSLE online.

FUNDING

This study was supported by NIDCR R01 DE13239 and DE23339.

Conflict of interest. None declared.