In vitro Manganese-Dependent Cross-Talk between Streptococcus mutans VicK and GcrR: Implications for Overlapping Stress Response Pathways

Streptococcus mutans, a major acidogenic component of the dental plaque biofilm, has a key role in caries etiology. Previously, we demonstrated that the VicRK two-component signal transduction system modulates biofilm formation, oxidative stress and acid tolerance responses in S. mutans. Using in vitro phosphorylation assays, here we demonstrate for the first time, that in addition to activating its cognate response regulator protein, the sensor kinase, VicK can transphosphorylate a non-cognate stress regulatory response regulator, GcrR, in the presence of manganese. Manganese is an important micronutrient that has been previously correlated with caries incidence, and which serves as an effector of SloR-mediated metalloregulation in S. mutans. Our findings supporting regulatory effects of manganese on the VicRK, GcrR and SloR, and the cross-regulatory networks formed by these components are more complex than previously appreciated. Using DNaseI footprinting we observed overlapping DNA binding specificities for VicR and GcrR in native promoters, consistent with these proteins being part of the same transcriptional regulon. Our results also support a role for SloR as a positive regulator of the vicRK two component signaling system, since its transcription was drastically reduced in a SloR-deficient mutant. These findings demonstrate the regulatory complexities observed with the S. mutans manganese-dependent response, which involves cross-talk between non-cognate signal transduction systems (VicRK and GcrR) to modulate stress response pathways.


Introduction
Streptococcus mutans, one of the primary etiological agents of dental caries, can metabolize dietary carbohydrates and produce lactic acid as a fermentative endproduct [1]. In addition to its acidogenicity, S. mutans is also aciduric, owing partly to an acid tolerance response [2] that allows it to adapt to conditions of low pH in the plaque environment. Part of this adaptive response is facilitated by differential regulation of genes under acid stress that include those whose products mediate proton extrusion (e.g. atpE/A), alter membrane composition (e.g. fabM, ffh) and assist with DNA repair (e.g. uvrA, recA) [3][4][5][6][7][8].
In previous work, we defined a role for GcrR (also known as CovR), in the S. mutans ATR and noted that gcrR expression was subject to metalloregulatory control by SloR [4]. The SloR metalloregulator in S. mutans is a DtxR homolog that controls the expression of a plethora of genes in response to metal ion availability, particularly manganese [9]. Manganese is an important micronutrient that has been correlated with streptococcal virulence and caries incidence [10]. The SloR regulon controls manganese-responsive genes that encode sucroseindependent and -dependent adherence (spaP, gbpC, gtfB and gtfC, genetic competence (comD/E), and oxidative stress tolerance (sod), all of which were shown to be up-regulated by SloR [4,9,11,12]. More recently, it was demonstrated that manganese limitation increased gcrR expression in a SloRdependent manner [4].
In contrast with what is found in other closely related streptococci, in S. mutans gcrR is not part of a two-component system (TCS) but rather is an orphan response regulator (RR), meaning it is not genetically linked to a cognate histidine kinase [13] [14]. A typical TCS is comprised of a membrane-bound HK, which autophosphorylates when activated by an environmental stimulus and subsequently transphosphorylates a cytosolic RR, which is often co-transcribed as an operon with the HK. The phosphorylation of the RR typically results in its activation, thus facilitating the RRs binding to DNA promoter/operator regions to modulate the expression of genes under the control of the TCS [15]. Unlike in S. mutans, the GcrR ortholog in Streptococcus pyogenes (designated CovR), is genetically linked to its cognate HK, CovS. In fact, CovRS has been extensively studied in group A streptococci (GAS) and more than 15% of the GAS genome, including those genes that mediate growth, virulence, biofilm formation, and stress tolerance are controlled by the CovRS TCS [16][17][18][19][20][21][22].
In S. mutans, GcrR has been implicated in the stress tolerance response; in particular, GcrR was shown to up-regulate expression of atpEA and ffh in the ATR mechanism [4]. These genes encode a proton extruding ATPase and a signal recognition peptide that facilitates ATPase insertion into the bacterial membrane, respectively [4,8]. A clear role for GcrR in the regulation of the bacteria's biofilm phenotype was shown as GcrR modulates S. mutans sucrose-dependent adherence through direct binding to the glucosyltransferase B and C (gtfB and gtfC) promoter regions to repress their transcription [11,12]. Expression of gtfBC is critical for biofilm formation and the pathogenicity of S. mutans [23], and therefore is tightly regulated by multiple signal transduction systems at the transcriptional and/or translational levels [24][25][26][27][28]. One such system is the VicRK TCS, that has a drastic effect on the biofilm phenotype of S. mutans, and was shown to have a positive regulatory impact on gtf expression [29].
Of 14 TCSs present in S. mutans, the VicRK TCS is noteworthy as the only signaling system that is essential for its viability. VicRK has been shown to be essential for survival and virulence in a wide range of bacteria, including the streptococci, bacilli and the staphylococci [29][30][31][32]. Depending on the bacterial species, either both VicK and VicR or just VicR alone are essential (the latter is true in S. mutans). The VicRK homologs (also known as YycGF and WalKR) in Staphylococcus aureus and Bacillus subtilis have been implicated as the master regulatory system for cell wall metabolism by positively regulating autolysin synthesis and biofilm formation [32,33].
Despite these findings, the genetic basis for S. mutans VicRK-modulated stress tolerance is not well-understood and the signal(s) that stimulate VicK activation remain unknown. Therefore, an improved understanding of how VicRK modulates these various stress responses through gene expression could provide insight into how bacterial TCSs might be manipulated, thereby fostering the development of therapeutics against bacterial infections.
In the present study, we report that S. mutans VicK can transphosphorylate not only its cognate RR, VicR, but also the orphan RR, GcrR; the latter is only demonstrable in vitro in the presence of manganese. We also demonstrate that while autophosphorylation and transphosphorylation reactions were enhanced by manganese, the specificity for a given RR appears specific for just VicR and GcrR. In addition, we present evidence to support a role for both VicK and SloR in gcrR transcriptional control. Finally, we show that the DNA binding sites of VicR and GcrR overlap in common downstream gene promoters further integrating the two RRs. A model is proposed for the biological basis for cross-talk between VicK, VicR and GcrR.

Bacterial strains, plasmids and growth conditions
The bacterial strains and plasmids used in this study are described in Table 1. Escherichia coli was grown overnight at 37˚C in Luria-Bertani broth (also known as LB broth, Difco) with gentle aeration, or on Luria-Bertani medium containing 1.5% w/v agar (LB agar). Kanamycin sulfate and ampicillin were added to these media when appropriate, each at a final concentration of 100 mg ml 21 . S. mutans was grown as standing overnight cultures at 37˚C and 5% CO 2 in Todd-Hewitt broth (Becton Dickinson) supplemented with 0.3% w/v yeast extract (THYE).
Kanamycin (700 mg ml 21 ) was added to THYE to maintain selection for the S. mutans GMS905, GMS906 and GMS907 fusion strains. S. mutans GMS584 and SmuvicK, isogenic mutants of the wild-type UA159 strain, were grown in THYE supplemented with erythromycin at a final concentration of 10 mg ml 21 , when needed.

Cloning and purification of VicK, GcrR & ComE
To generate a tagless version of VicK we used the Impact Kit (New England Biolabs) and generated a C-terminal VicK-Intein fusion protein by PCR amplifying the vicK coding sequence from S. mutans UA159 chromosomal DNA with oligonucleotides oSG548 and oSG550 (S1 Table). For the Intein tag to be removed with greater efficiency, the last amino acid of VicK was changed from serine to alanine using primer oSG550. The purified amplicon was digested with NdeI and SapI and ligated into the expression plasmid pTXB1 (New England Biolabs) according to the supplier's protocol. The ligation mixture was transformed into E. coli ER2566 cells, selected for ampicillin resistance, and the construct was confirmed by DNA sequencing. To overexpress VicK-Intein, cells grown to mid-logarithmic phase (OD 600 nm 0.3-0.5) were induced with 1 mM IPTG for 3 hr at 37˚C with aeration before being harvested by centrifugation (4˚C, 5,000 x g, 15 min) and frozen at 220˚C. To purify tagless VicK, cell pellets were thawed and resuspended in Intein column buffer (20 mM Tris-HCl, pH 8.0, 0.5 M NaCl) before lysozyme was added to 1 mg ml 21 . After 30 min on ice, the The coding sequence of S. mutans vicR (smu.1517) was amplified as described above for vicK with oligonucleotides oSG726 and oSG727 (S1 Table). To improve cleavage of the Intein tag by DTT, alanine was substituted for the vicR stop codon in the oSG727 primer. The vicR coding sequence was then cloned into pTXB1, confirmed by sequencing as described above, and transformed into ER2566 cells for protein expression. Freshly transformed cells were used to inoculate an overnight culture for VicR overexpression, which was induced with 1 mM IPTG as described above. The purification was carried out as described for tagless VicK except that the pH of the binding buffer was adjusted to pH 9.0 and the insoluble pellet was incubated with 0.65% sarkosyl for 1 hr to release the fusion protein into the soluble fraction. Pure tagless VicR was visualized by SDS-PAGE, quantified as described above and stored in 25% glycerol at 280˚C.
The coding sequence of S. mutans gcrR (smu.1924) was amplified by PCR with primers oSG741 and oSG742 (S1 Table). To facilitate cleavage of the Intein tag by DTT, glutamine was substituted for the gcrR stop codon in the oSG741 primer. The gcrR coding sequence was digested with FauI and PvuII and then cloned into the NdeI and SapI sites on pTXB1 before being confirmed by sequencing as described above. Tagless GcrR was purified as described for tagless VicK with one exception; the pH of the column buffer was adjusted to 8.5. Pure tagless GcrR was visualized by SDS-PAGE, quantified as described above and stored in 25% glycerol at 280˚C.
The coding sequence of S. mutans comE (smu.1917) was amplified by PCR with primers oSG691 and oSG692, cloned into pTXB1 and confirmed by sequencing as described above. Tagless ComE was purified as described for VicR except 0.7% sarkosyl was used to solubilize the fusion protein. Pure tagless ComE was visualized by SDS-PAGE, quantified as described above, and stored in 25% glycerol at 280˚C.

Phosphorylation assays
For autophosphorylation experiments, 1 mM VicK was incubated in 100 mM Tris-HCl, pH 7.5 and various metal cations were included at a final concentration of 1 mM. The phosphorylation reactions were initiated by adding 0.10 mM [c-32 P] ATP followed by incubation at room temperature for 15 min [40].
For transphosphorylation reactions, 1 mM VicK was incubated for 15 minutes at room temperature with excess cold ATP (10 mM), to ensure ATP was not limiting, and 0.2 mM [c-32 P] ATP, 100 mM Tris-HCl pH 7.5, 50 mM NH 4 Cl, and 1 mM MgCl 2 . Then GcrR, VicR, or ComE RRs were added to 1 mM to the the VicK autophosphorylation reaction for 1 hr at room temperature. The phosphoryation reactions were diluted 1:1 in 2x SDS loading buffer (125 mM Tris-HCl pH 7.4, 0.005% bromophenol blue, 4% SDS, 20% glycerol). A 4-20% acrylamide Tris-glycine SDS PAGE Novex gel (Invitrogen), the X-Cell SureLock Mini-Cell protein electrophoresis chamber and the Tris-glyine running buffer were pre-chilled at 4˚C for ,3 hrs. The reactions were separated by SDS-PAGE for ,3 hrs at 150 V at 4˚C. The resulting gels were dried, exposed to a phosphor screen [26] and scanned using a Typhoon phosphorimager (GE Healthcare).
For transphosphorylation reactions in the presence of MnCl 2 , 1 mM VicK was incubated in the presence of excess cold ATP (10 mM), 0.2 [c-32 P] ATP mM, 100 mM Tris-HCl pH 7.5, and 1 mM MnCl 2 for 15 min at room temperature. The response regulator (1 mM) was added and the reactions were incubated for 1 hr at room temperature before being diluted 1:1 in 2x SDS loading buffer. As described above, the buffers and gels were pre-chilled at 4˚C. The transphosphorylations reactions were separated on 4-20% acrylamide Tris-glycine SDS PAGE Novex gels for 2.5 hrs at 128 V and then for 1 hr at 150 V at 4˚C. The resulting gels were dried, exposed to a phosphor screen [26] and scanned using a Typhoon phosphorimager (GE Healthcare). The Phos-Tag mobility shift assay was performed as described previously [41], with the exception that a 12% gel was prepared and visualized by silver staining [42].

DNaseI footprinting
DNaseI footprinting analysis was performed as previously described [43]. For protein concentrations used, refer to the figure legend.
Construction of the S. mutans PgcrR:cat fusion strains GMS905, GMS906 and GMS907 The gcrR promoter region was amplified by PCR from the S. mutans UA159 chromosome using primers gcrR_356.FV.F and gcrR_356.FV.R (S1 Table). SacI and BamHI digested amplicons were then cloned into the integration vector pJL84, which contains the S. aureus cat and S. mutans phnA and mtlA genes [44]. Kanamycin-resistant transformants were selected and the presence of the gcrR promoter region fused to cat was confirmed by sequencing. The resulting plasmid, pLM1, was transformed into S. mutans UA159, GMS584 and SmuvicK in the presence of competence stimulating peptide (CSP) according to established protocols, to generate strains GMS905, GMS906 and GMS907, respectively [9,29,45]. Integration of the PgcrR:cat fusion via allelic exchange was mediated by the S. mutans phnA and mtlA genes that are resident on the pJL84 plasmid [44]. PCR amplification and nucleotide sequencing were used to confirm the double allelic crossover event at the desired locus on the S. mutans chromosome.

CAT assay
Overnight cultures of S. mutans GMS905, GMS906 and GMS907 were diluted 1:10 in pre-warmed THYE and grown to an OD 600 nm 0.6-0.7. Cells were harvested by centrifugation, resuspended in 1 ml of 10 mM Tris-HCl at pH 7.8 and lysed by mechanical disruption in a BIO101 Savant FastPrep (Thermo Savant) for 1.5 min with intermittent cooling on ice. Unlysed cells and debris were removed by centrifugation (9,300 x g, 4 min, 4˚C) and the resulting cell lysates were stored at 220˚C for subsequent protein determination with a BCA protein assay kit (Pierce) and chloramphenicol acetyltransferase [35] assays. CAT assays were performed according to the method described by Shaw [46]. Briefly, a 360 ml reaction mixture consisting of 100 mM Tris-HCl pH 7.8, 0.4 mg ml 21 5,59dithiobis(2-nitrobenzoic acid) (DTNB, (e 412 nm 513.6 mM 21 cm 21 ), and 0.1 mM acetyl-CoA was mixed with 40 ml of whole cell lysate. To monitor CAT specific activity, 0.1 mM chloramphenicol (CM) was added and absorbance readings at OD 412 were obtained every 10 sec over a five min interval in a Synergy HT Microtiter Plate Reader (BioTek). To assess background activity, absorbance readings for wells that contained only the reaction mixture and the whole cell lysate were obtained in parallel. The rate of change due to addition of CM was determined by subtracting the background activity from the rate of change after the addition of CM. This value was divided by 0.0136 (extinction coefficient of DTNB) to yield CAT activity and then subsequently divided by the total protein concentration to express the CAT specific activity result in nM min 21 mg 21 . All CAT assays were performed as three independent experiments each in triplicate.

Quantitative real time PCR (qRT-PCR)
Overnight cultures of S. mutans UA159 and SmuvicK were grown to midexponential phase (OD 600 nm ,0.4) in Tryptone (Bioshop) Yeast Extract Glucose medium (TYEG; 10% w/v tryptone, 5% w/v yeast extract, 17.2 mM K 2 HPO 4 , 0.5% w/v glucose, pH 7.5). Following incubation, the cells were pelleted, resuspended in TYEG at pH 7.5 or pH 5.5 and incubated for 1 h at 37˚C and 5% CO 2 . The cells were harvested by centrifugation, snap frozen in liquid nitrogen and stored at 280˚C. RNA was isolated from the pellets and reverse transcribed as previously described [36]. The resulting cDNAs were used as a template in qRT-PCR reactions with primers listed in S1 Table according to established protocols [36]. Gene expression was normalized to that of S. mutans 16SrRNA which was invariable under the experimental test conditions (data not shown). Relative expression of the target genes was calculated using results from 3 independent experiments, according to the method of Pfaffl et al [47].
To determine vicR expression in a sloR-deficient (GMS584) mutant vs. wildtype UA159, overnight cultures of both strains were grown in a semi-defined medium (SDM) [48] to OD 600 nm ,0.6 before the cells were pelleted and snap frozen as described above. RNA was isolated from the pellets and reverse transcribed as previously described [4]. The expression of vicR was normalized to that of gyrA, which did not change under the experimental conditions tested (data not shown). These qRT-PCR experiments were performed in triplicate in each of three independent experiments.

Autophosphorylation of VicK is facilitated by manganese and inhibited by iron
To demonstrate that VicK can undergo autophosphorylation, VicK was purified to.90% homogeneity as determined by Coomassie blue staining of SDS-PAGE gels (data not shown) and subsequently used in phosphorylation assays. Initial phosphorylation assay conditions were based on results described previously by Clausen et al [40]. VicK was readily phosphorylated when incubated at room temperature for 15 min in the presence of NH 4 Cl, MgCl 2 and [c-32 P] ATP (data not shown). We next tested a variety of different divalent metal cations (1 mM MgCl 2 , CdCl 2 , CaCl 2 , MnCl 2 , ZnCl 2 , NiCl 2 , CoCl 2 , CuSO 4 , FeCl 3 , or FeSO 4 ) in lieu of NH 4 Cl and MgCl 2 to assess their potential impact on VicK phosphorylation. As shown in Fig. 1, VicK was able to autophosphorylate at relatively low levels in the presence of MgCl 2 , CaCl 2 , CoCl 2 , or FeCl 3 , and at significantly increased levels in the presence of MnCl 2 (an over 3-fold increase compared to the next highest phosphorylation condition that was observed with MgCl 2 ).
To determine whether any of the metal cations tested above might inhibit VicK autophosphorylation in the presence of Mn 2+ , we repeated the autophosphorylation assays with reaction mixtures containing VicK, MnCl 2 and equimolar amounts of MgCl 2 , CdCl 2 , CaCl 2 , ZnCl 2 , NiCl 2 , CoCl 2 , CuSO 4 , FeCl 3 , or FeSO 4 . CaCl 2 , NiCl 2 , and ZnCl 2 inhibited VicK autophosphorylation by 46%, 77% and 43%, respectively, relative to the 100% phosphorylation of VicK that we observed in the presence of Mn 2+ alone (Fig. 2). Interestingly, CoCl 2 , FeCl 3 and FeSO 4 had an even more pronounced effect on VicK phosphorylation in the presence of Mn 2+ , diminishing activity to 9%, 18% and 7%, respectively (Fig. 2). Importantly, the addition of either FeCl 3 or FeSO 4 (freshly prepared before use to limit oxidation) did not alter the overall pH of the phosphorylation reaction, consistent with a direct inhibitory effect for iron on VicK phosphorylation. Collectively, these autophosphorylation experiments demonstrate that while standard conditions (NH 4 Cl/MgCl 2 ) and Mn 2+ can readily stimulate autophosphorylation of VicK, not all divalent cations can effectively do so.
VicK differentially transphosphorylates VicR and GcrR in the presence of manganese S. mutans VicK's ability to transphosphorylate VicR has recently been demonstrated [41,49,50]. To further explore this in the presence of different divalent metal cations, we phosphorylated VicK in the presence of excess ATP, and buffer containing MgCl 2 and NH 4 Cl and then supplemented the reaction mixture with various combinations of the VicR, ComE or GcrR responder proteins. ComE and GcrR were selected since both have overlapping pathways with the VicRK TCS based on transcriptome analysis and regulation of common phenotypes that include genetic competence, acid tolerance and biofilm formation [29,36,39,[51][52][53]. Despite this knowledge cross-talk with VicK has not been demonstrated. Each RR was allowed to incubate individually with VicK. As seen in Fig. 3A (lanes 5-7), VicR was the only RR that was phosphorylated under these conditions. As an additional control, each RR was added to phosphorylation reactions in the absence of VicK and no phosphorylation was observed (Fig. 3A,  lanes 2-4). ComE was also added to the transphosphorylation reaction containing either VicR or GcrR. Even with the addition of ComE, only VicR was efficiently phosphorylated by VicK (Fig. 3A, lane 8). Likewise no detectable transphosphorylation was noted when GcrR and ComE were combined (Fig. 3A, lane 9). Previous studies have shown that the GcrR regulon is manganese-responsive [4], and here we demonstrate that VicK autophosphorylation is stimulated by manganese. To date, no cross-talk has ever been observed between any of the S. mutans HKs tested (VicK, CiaH and LiaS) and their noncognate RRs (CiaR and LiaR) [49]; albeit these experiments were performed in the absence of divalent cations. We explored whether VicR or the GcrR orphan RR could be phosphorylated by VicK-P in the presence of manganese. The autophosphorylation and transphosphorylation reactions were allowed to proceed in the presence of excess ATP, and buffer containing MnCl 2 . As seen in Fig. 3B lane 4, VicR was phosphorylated under these conditions, although at slightly lower levels than those observed previously in transphosphorylation reactions lacking manganese. In contrast to Fig. 3A and in the presence of manganese, VicK was able to phosphorylate the non-cognate S. mutans response regulator, GcrR (Fig. 3B, lane  5). Importantly, neither VicR nor GcrR were capable of autophosphorylation under these conditions (Fig. 3B, lanes 2-3).
To determine whether VicK has a preference for the phosphorylation of VicR and/or GcrR in the presence of manganese, the reaction containing 1 mM GcrR was repeated in the presence of 0.01-1 mM VicR (Fig. 3B, lanes 6-9). For these experiments VicR was added to the reaction mixture after GcrR was added. The addition of even 0.02 mM VicR outcompeted GcrR as a substrate for VicK transphosphorylation (Fig. 3B, lane 7) although measurable GcrR phosphoryla- tion persists under these conditions through VicR concentrations as high as 0.04 mM (Fig. 3B, lane 8). To demonstrate that this competition was specific for VicR and GcrR, we transphosphorylated VicR and GcrR in the presence of manganese and 1 mM ComE. As shown in S1 Fig. as much as 1 mM ComE had no effect on the phosphorylation state of either VicR or GcrR. To confirm that the bands shown in Fig. 3b are the proteins indicated, we performed transphosphorylation assays in the presence of MnCl 2 followed by Phos-Tag mobility shift analysis [41] and silver staining with purified VicK (51.7 kDa), VicR (26.9 kDa), and GcrR (26.7 kDa) under the same conditions described above (S2 Fig.). The protein migration patterns shown in S2 Fig. correspond to those seen in Fig. 3b, demonstrating that these protein bands are the specified proteins.

GcrR and VicR binding sites overlap at co-regulated genes
VicRK and GcrR regulate a number of overlapping genes including gtfC [2,11,12,29,52,54]. We recently further characterized theconsensus sequence of VicR [43] and sought to gain a better understanding of the direct regulation of VicR and GcrR. Specifically, we wanted to examine the binding site boundaries of VicR and GcrR particularly since a GcrR binding consensus has proven elusive. To explore this, we used DNaseI footprinting with VicR and GcrR individually or both RRs present at an equimolar amount, with the gtfC promoter region as DNA substrate. As shown in Fig. 4A, VicR and GcrR independently displayed protection that overlapped the VicR consensus sequence (indicated by the solid and dashed lines respectively). Unlike VicR, the GcrR footprint exhibited enhanced cleavage (indicated by the arrows) at two specific locations. These sites correspond to the bases 59-TGTG and 59-GTGT that flank the VicR consensus sequence. Such ''hypersensitive sites'' reflect an increased accessibility of specific phosphodiester linkages to DNase I cleavage, often indicative of DNA bending. The same enhanced cleavage was also observed when VicR and GcrR were both present in the assay at an equimolar amount, whereas VicR alone displayed protection at this site (Fig. 4A, solid line). This indicates that not only do VicR and GcrR have overlapping binding sites but that GcrR under equimolar conditions exhibits greater binding affinity for gtfC compared to VicR.
To determine if GcrR demonstrated DNA binding dominance at another gene locus (in addition to gtfC), we also investigated the binding affinities of VicR and GcrR to the gcrR promoter region. To examine this we used DNaseI footprinting by amplifying a 160 bp fragment upstream of the gcrR start codon [43]. As shown previously, VicR protected a region that overlapped its consensus sequence [43] clearly seen with the highest concentration of VicR (0.5 mM) (Fig. 4B, solid line). In contrast, GcrR displayed protection that overlapped that of VicR, but the binding affinity was much stronger with protection observed at 0.25 mM (dashed line). When VicR and GcrR were present simultaneously at equimolar amounts, protected regions again overlapped but resembled that of GcrR compared to VicR, suggesting that GcrR demonstrates stronger binding affinity for the gcrR substrate (Fig. 4B). These results provide evidence that further integrate the functions of VicR and GcrR at select promoter regions.

SloR positively regulates vicRKX expression
We previously reported the results of microarray experiments that support involvement of S. mutans SloR in regulating gcrR expression [4]. Further review of these transcriptome data revealed that expression of the vicRKX tricistronic operon was drastically reduced, relative to wild type, in the isogenic sloR-deficient GMS584 strain, consistent with a role for SloR in the transcriptional regulation of this locus. To corroborate these findings, qRT-PCR experiments were performed with cDNAs from S. mutans UA159 and GMS584 with vicR-specific primers (S1 Table). Consistent with the microarray data, vicR expression was over 7-fold greater in the UA159 wild-type strain as compared to vicR expression in the GMS584 SloR-deficient mutant (data not shown), thereby supporting a role for SloR as a positive regulator of the vicRKX tricistronic operon.

Transcription of gcrR is subject to VicK and SloR control
To further elucidate the impact of VicK and SloR on gcrR expression, PgcrR:cat fusions were constructed in S. mutans wild-type (GMS905) and mutant (GMS906 and GMS907 which harbor sloR and vicK insertion-deletion mutations, respectively) backgrounds (Fig. 5). The resulting fusion strains were confirmed by nucleotide sequencing with Smu_1179c-and cat-specific primers (S1 Table). For cat assays, whole cell lysates were prepared from mid-exponential phase cultures of each strain so that CAT-specific activity could be monitored according to the spectrophotometric assay of Shaw [46]. Importantly, we observed CAT-specific activity for the wild-type GMS905 strain (5.84¡1.24 nM min 21 mg 21 ), that was 3-fold greater than that of the GMS906 SloR-deficient derivative (1.78¡0.74 nM min 21 mg 21 ), and 2.5-fold greater than that of the VicK-deficient strain, GMS907 (2.30¡0.69 nM min 21 mg 21 ). These findings support a role for both SloR and VicK in modulation of the S. mutans gcrR gene in vivo. VicK positively regulates gcrR and ATR related gene transcription under low pH VicK has previously been implicated in facilitating S. mutans ability to respond and adapt to low pH [39]. To further examine the role of VicK in regulating the S. mutans ATR, we compared the expression of known ATR genes, atpE/A, ffh, radA and gcrR, in the UA159 wild-type strain and a vicK-knockout derivative, SmuvicK [29]. Exposure of wild-type UA159 cells to a sub-lethal acid challenge (pH 5.5) resulted in over a 2-fold increase in expression at these loci compared to cells grown at pH 7.5 (data not shown). The greatest induction of gene expression in the UA159 wild-type strain was observed for atpA and atpE (.3-fold), which encode the alpha and c subunits of the F o F 1 membrane-bound protontranslocating ATPases, respectively. Not surprisingly, the involvement of these ATPase subunits in the S. mutans ATR is well established in the literature [4,6,7,55,56].
In contrast, loss of VicK failed to induce any of these genes at low pH, suggesting a requirement for VicK in modulating their transcription under conditions of acid stress (Fig. 6). In addition, all of the genes were significantly down-regulated in the VicK mutant relative to that in wild type (p,0.001, with the exception of gcrR at pH 7.5), further supporting a positive regulatory role for VicK in their transcription (S2 Table).

Discussion
In this report, we demonstrate that in vitro autophosphorylation of the S. mutans VicK HK is enhanced by manganese and inhibited by ferrous iron, suggesting two potential roles for VicRK in sensing manganese availability and conditions of redox. We further demonstrate that VicK, in addition to phosphorylating its cognate RR, VicR, facilitates in vitro cross-talk by phosphorylating the orphan RR GcrR, in the presence of manganese albeit only when VicR is present at comparatively low relative concentration. Our findings are similar to those of Guckes et al who report similar cross talk in the form of one HK being able to transphosphorylate specific non-cognate RRs, but only under specific conditions [57]. While VicK-facilitated phosphorylation of VicR and GcrR appears to be specific in vitro, it remains to be seen to what extent this phenomenon occurs in vivo. Interestingly, Stipp et al recently demonstrated that VicRK and GcrR work in concert to form structurally stable biofilms by coordinating surface biogenesis and cell division in S. mutans [54]. However their model did not show any direct interaction between VicK and the orphan RR GcrR. Liang et al identified a CovS mutant in S. pyogenes that retained CovR-mediated virulence gene regulation, via an unidentified alternate pathway for CovR phosphorylation [58].
Here we provide insight into the communication pathways of the S. mutans SloR, VicRK and GcrR regulators. Cross-regulation of GcrR by the VicK sensor kinase, may explain the overlap between the VicRK-regulon and that of GcrR, in modulating genes whose products contribute to the oxidative stress and acid tolerance responses of S. mutans. Work conducted previously by Dunning et al showed that SloR interacts directly with the gcrR promoter region to facilitate its expression [4]. Here we demonstrate that SloR also positively regulates expression of the vicRKX operon, but whether its impact on vicRKX transcription is direct or indirect remains to be determined. Moreover, CAT-specific activity observed in the S. mutans cat fusion strains support gcrR expression that is subject to both SloR and VicRK control.
Results of phosphorylation assays highlight a role for manganese in integrating the VicRK, GcrR and SloR regulatory pathways. Manganese is an essential micronutrient that affects S. mutans genes whose products are conducive to its virulence, which include those that mediate adherence and biofilm formation [59][60][61][62]. In fact, manganese functions as a cofactor for a S. mutans superoxide dismutase (SOD), which converts damaging reactive oxygen species, into less toxic substances [63,64]. Reports in the literature describe a correlation between intracellular manganese and iron concentrations with the sensitivity of bacteria to oxidative stress [65,66]. It has been suggested that bacteria may utilize manganese instead of ferrous iron to avoid redox conditions that are associated with Fenton chemistry [67]. This could be especially important in a biofilm environment where extracellular manganese can reach mM concentrations in the oral cavity [68].
A novel aspect of this study is our demonstration of VicK's ability to transphosphorylate GcrR under low VicR concentrations in the presence of manganese (Fig. 3B). Further, the presence of VicR causes the turnover of phosphorylated VicK, something that is not seen with other tested RRs. This effect may be important biologically as VicR-VicK interactions may activate VicK phosphatase activity reducing the net steady state levels of phosphorylated VicK; this effect seems enhanced with increasing concentrations of manganese. In contrast, we also show that iron inhibits VicK phosphorylation especially in the presence of manganese (Figs. 1 and 2). These findings are consistent with GcrR regulation by the VicRK system as well as by the metal ion-dependent SloR metalloregulator that binds manganese preferentially over iron. It is worth pointing out that we have used tagless but not necessarily native proteins for these phosphotransfer reactions. As yet to be determined, post translational modifications to each protein may be critical in each of the aforementioned reactions.
We previously described modulation of the S. mutans ATR by SloR with GcrR as an essential intermediary [4]. We also confirmed sloABC expression that is responsive to SloR and manganese concentrations that are physiological conditions likely representative of feast (.10 mM) or famine (,0.1 mM) [4]. Our expression analysis to determine whether sloABC transcription was dependent on VicK revealed that loss of VicK did not significantly affect expression of sloABC (data not shown); suggesting that VicRK does not modulate metal ion uptake via the sloABC transport system. Given the in vitro effect of manganese on steadystate levels of GcrR and VicR phosphorylation, it is possible that VicK, like SloR, may use GcrR as an intermediary to modulate acid tolerance in S. mutans by responding to manganese.
GcrR exists as an orphan RR on the S. mutans chromosome and has 65.7% similarity to the orphan RR RitR from Streptococcus pneumoniae, which has a role in oxidative stress tolerance by regulating the expression of iron transport systems [69]. However, in other bacteria (such as group A streptococci where GcrR is known as CovR) GcrR is co-transcribed along with a cognate HK, CovS. Interestingly, extracellular Mg 2+ stimulates covRS expression in group A streptococci (GAS) and increasing concentrations of exogenous Mg 2+ have been shown to increase gcrR expression [70,71]. Moreover, CovS can dephosphorylate (and thus inactivate) CovR in GAS under stress-inducing conditions including high temperature, low pH, high salt and iron starvation [17,72]. It would be interesting to investigate a potential physical interaction between VicR and GcrR, and characterize the VicR-GcrR-and SloR-binding sites to validate potential cross-regulation between their respective regulons in S. mutans. In fact, reports in the literature supported the binding of VicR and GcrR to overlapping sequences upstream of S. mutans gtfB/C, encoding sucrose-dependent glucosyltransferases that are critical determinants of colonization and subsequent virulence [11,29]. Indeed, we confirmed this hypothesis using DNaseI footprinting analysis. We demonstrated that both GcrR and VicR bind to the same regions upstream of gtfC and gcrR. When incubated together at equimolar amounts, GcrR displayed higher DNA binding affinity than VicR suggesting that in vitro, GcrR predominates under these conditions. Although these studies were performed using unphosphorylated forms of the RRs in vitro, the phosphorylation states of these RRs likely play an important role in their regulation in vivo. Additional studies and genomic analyses need to be performed in vivo to support our results.
Based on our results, we propose that it is highly likely that manganese is the common denominator for cross-communication between the VicK, GcrR, and SloR regulatory networks (Fig. 7). Both VicK and SloR activation are manganesedependent, whereas VicK has been shown to respond to pH, oxidative and cell wall stresses [32,33,39,73]. Thus, we hypothesize that VicK autophosphorylation and the subsequent transphosphorylation of VicR and GcrR (the latter in the presence of manganese) leads to modulation of numerous S. mutans virulence genes that facilitates bacterial survival in the presence of reactive oxygen species.
To ensure that the cross-talk between VicK and GcrR is tightly regulated, we propose that S. mutans regulates expression of VicR so that it may compete for phosphorylation by VicK. In the presence of manganese SloR represses gcrR expression and enhances VicR expression while manganese also makes GcrR a substrate for VicK transphosphorylation. In addition, when VicK, SloR or manganese are limiting there is increased expression of GcrR. It is possible that in the absence of manganese there is sufficiently more GcrR than VicR in S. mutans such that GcrR becomes the favored phosphoryl acceptor of VicK. Intriguingly in vitro, even when GcrR and VicR are present at equimolar amounts on overlapping binding sites there is evidence that GcrR binding predominates over VicR, at least in the unphosphorylated state (Fig. 4). According to our model (Fig. 7), when manganese is encountered, GcrR is phosphorylated (and thus activated) by VicK to respond to the new metal cation and possibly conditions of oxidative stress. This activation of GcrR is likely short lived due to the fact that VicR expression will increase and GcrR expression will decrease in response to manganese binding by SloR. As a result, VicR may rapidly outcompete GcrR for VicK phosphorylation, allowing for the tight regulation of crosstalk between VicK and GcrR. It is during this brief period when GcrR is phosphorylated that it alters S. mutans gene expression to accommodate the transition to manganese rich conditions.
Further studies are needed to determine if the phosphorylated forms of these RRs are the active or inactive forms. These additional studies are necessary to dissect the complex interactions that occur between VicK, VicR, GcrR and SloR and the larger role these regulators likely have in governing the cellular physiology of S. mutans.   Fig. 7. In vitro model of manganese-independent (A) and -dependent cross-regulation involving S. mutans SloR, VicRK and GcrR. A) In the absence of Mn 2+ (approximating conditions of free-floating planktonic cells) S. mutans gcrR expression is de-repressed (VicR expression is not induced in the absence of Mn 2+ ). Even though GcrR is the more abundant substrate, VicR is the favored species for VicK phosphorylation under these conditions. B) During this so-called ''transition stage'' (approximating conditions of an early biofilm) a ''spike'' in Mn 2+ renders GcrR (still the predominant species) the favored substrate for phosphorylation by VicK, but only transiently. C) As the biofilm matures and Mn 2+ concentrations increase, SloR is activated to repress gcrR expression, thereby reducing the availability of GcrR as a substrate. The activated SloR-Mn 2+ complex encourages vicR expression, and hence VicR becomes the favored substrate for VicK phosphorylation once again.