Identification of Streptococcus cristatus peptides that repress expression of virulence genes in Porphyromonas gingivalis

Dental plaque is a complex multispecies biofilm, and is a direct precursor of periodontal disease. The virulence of periodontal pathogens, such as Porphyromonas gingivalis, is expressed in the context of this polymicrobial community. Previously, we reported an antagonistic relationship between Streptococcus cristatus and P. gingivalis, and identified arginine deiminase (ArcA) of S. cristatus as the signaling molecule to which P. gingivalis responds by repressing the expression and production of FimA protein. Here we demonstrate that direct interaction between P. gingivalis and S. cristatus is necessary for the cell-cell communication. Two surface proteins of P. gingivalis, PGN_0294 and PGN_0806, were found to interact with S. cristatus ArcA. Using a peptide array analysis, we identified several P. gingivalis-binding sites of ArcA, which led to the discovery of an 11-mer peptide with the native sequence of ArcA that repressed expression of fimbriae and of gingipains. These data indicate that a functional motif of ArcA is sufficient to selectively alter virulence gene expression in P. gingivalis, and PGN_0294 and PGN_0806 may serve as receptors for ArcA. Our findings provide a molecular basis for future rational design of agents that interfere with the initiation and formation of a P. gingivalis-induced pathogenic community.

Periodontitis is the 6 th most common infection worldwide 1 and an estimated 5-20% of the population suffers from generalized chronic periodontitis 2, 3 . In the US, around half of the adult population suffers from some form of periodontal disease, which constitutes a significant economic burden 4 . Periodontal diseases and periodontal bacteria are also physically and epidemiologically associated with severe systemic conditions such as coronary artery disease, rheumatoid arthritis, and diabetes [5][6][7] . Chronic periodontitis is the result of a breakdown of periodontal tissue-microbe homeostasis, which then leads to uncontrolled inflammation and tissue destruction 8 . Loss of tissue homeostasis, called dysbiosis, is initiated by communities of organisms colonizing the subgingival area. In most instances these microbial communities are associated with health, and the transition to pathogenesis requires colonization by specific pathogens such as Porphyromonas gingivalis. It has long been recognized that the presence of P. gingivalis alone is insufficient for the disease to occur in a susceptible host, and current models hold that P. gingivalis is a keystone pathogen in that it elevates the virulence of the entire community [9][10][11][12] . Evidence for this comes from murine models in which low levels of P. gingivalis can initiate disease, but only in the context of a microbial community 9 , and from primate studies where a gingipain-based vaccine caused a reduction both in P. gingivalis numbers and in the total subgingival bacterial load, as well as in inhibiting bone loss 13 .
Bacteria within the oral microbial community can exhibit polymicrobial synergy whereby interspecies communication enhances colonization and pathogenic potential. Conversely, oral organisms also engage in antagonistic interactions, whereby one organism inhibits the colonization or growth of another. For example, communities of P. gingivalis and Streptococcus gordonii are synergistically pathogenic in a murine model of alveolar bone loss 14 , whereas Streptococcus cristatus interferes with the colonization and pathogenesis of P. gingivalis in mice 15 . We, and others, have reported previously that arginine deiminase (ArcA) of S. cristatus represses expression of the major fimbrial adhesin of P. gingivalis, FimA 16,17 . ArcA catalyzes the hydrolysis of L-arginine to L-citrulline and ammonia, and the latter is believed to be important for oral biofilm pH homeostasis and the prevention of caries 18,19 . We also found that the expression of arcA was significantly higher in S. cristatus than in S. gordonii 20 , suggesting (1 × 10 7 cells) were initially incubated in the Transwell insert, and the numbers of CC5A migrated to the lower well and 33277 cells in the well were determined by qPCR. Each bar represents the number of bacteria detected in the lower well. Asterisk indicates a statistically significant difference in number of bacteria in the lower wells (n = 3; t test; p < 0.05). (b) Expression of the fimA gene in P. gingivalis in transwell chambers with S. cristatus was measured by real-time qRT-PCR. Each bar represents relative expression level of the fimA, which was normalized to that of 16 S rRNA gene. Standard deviations are indicated. Asterisk indicates a statistically significant difference in expression level of fimA compared to that in P. gingivalis without exposure to S. cristatus (n = 3; t test; p < 0.05).
Scientific RepoRts | 7: 1413 | DOI:10.1038/s41598-017-01551-4 antibody showed that the 47 kDa protein is ArcA of S. cristatus (data not shown). The other two bands were identified by MS analysis as P. gingivalis RagB (PGN_0294) and a MotA/TolQ/ExbB proton channel family protein (PGN_0806), suggesting that these two proteins are receptors for ArcA.
Identification of the key functional motif of ArcA. S. cristatus ArcA is a 47 kDa protein with 409 amino acids 16 . We sought to identify key amino acids and the motif(s) of ArcA responsible for its inhibitory activity toward fimA expression. A peptide microarray was first performed to detect binding sites of ArcA for P. gingivalis. The arrays were incubated with surface extracts of P. gingivalis 33277, or the ΔragB or Δ0806 mutants, and binding was detected with P. gingivalis antibodies. Although the absolute binding capacities (fluorescence intensity) of these strains were significantly varied, likely due to protein degradation of surface extract in some strains, the overall patterns were consistent. Of several peaks observed (Fig. 4), a peptide with the sequence NIFKKNVGFKK (peak 4) and spanning amino acid residues 249-259, was found to have the highest binding affinity to P. gingivalis 33277 proteins, evident as the highest peak. This peak was no longer the highest when the arrays were incubated with surface extracts isolated from the Δ0806 or ΔragB mutants, corroborating the involvement of P. gingivalis proteins PGN_0806 and RagB in recognition of ArcA.
Five peptides (1-5 in Table 1) were synthesized based on ArcA array peaks, and the effect of each peptide on gene expression was determined by inclusion of the peptides in the P. gingivalis growth media. As shown in Table 1, the 11 residue peptide4 from the C-terminal region of ArcA repressed expression of fimA, mfa1, kgp, rgpA/B (encoding catalytic regions of rgpA/B), and rgpA (encoding adhesin domains of RgpA) genes by at least 2 fold, at a concentration of 16 µM. Expression of pgn_0128 encoding immunoreactive 53 kDa antigen was not modulated in response to the presence of peptide4, indicating specificity for a subset of virulence-associated genes. Increased inhibitory activity (60-70%) was observed at a concentration of 64 µM (not shown), suggesting that this region is likely a key active motif of ArcA. These results also establish that the effects of ArcA on P. gingivalis virulence extend beyond repression of fimA and include genes for the gingipains and for the minor Figure 3. Interaction of ArcA and P. gingivalis surface proteins. SDS-PAGE analysis of proteins eluted from Sepharose 4B column. Lane 1, the proteins eluted from untreated Sepharose 4B column exposed to P. gingivalis 33277 extract; Lane 2, the proteins eluted from ArcA antibody-coupled Sepharose 4B column exposed to CC5A extract only; lane 3, the proteins eluted from ArcA antibody-coupled Sepharose 4B column exposed to P. gingivalis and CC5A extracts. Proteins were stained with Coomassie blue. Identification of a binding region of ArcA interacting with P. gingivalis. A peptide array of ArcA was exposed to P. gingivalis 33277 and the ragB and pgn0806 mutants. The intensity plot of the peptide array signals shows as peaks with corresponding regions of ArcA. The five of the highest peaks are numbered. fimbrial subunit, as also shown by others 17 . The half inhibitory concentration (IC50) was determined by constructing a dose-response curve (0, 4, 16, and 64 µM) to measure the effectiveness of peptide4 in repressing expression of these genes. As shown in Fig. 5, the highest efficiency of peptide4 was found in inhibition of rgpA (amplified with primers corresponding to the region encoding the binding domain of RgpA, HGP44), rgpA/B (amplified with primers corresponding to the region encoding catalytic regions of RgpA and B), and fimA with IC50s of 11.2 ± 2.1, 11.7 ± 2.3 or µM, 12.4 ± 3.4, respectively. Peptide4 also showed a significantly lower IC50 for mfa1 (19.2 ± 3.3 µM) compared to that of kgp (64.3 ± 3.8 µM). It should be pointed out that peptide1, 2, 3 and 5 (Table 1) also exhibited some inhibitory activity, although at a lower efficiency. These regions along with peptide4 may be involved in formation of a structural motif that may have a higher binding capacity than peptide4 alone. These findings provide a molecular basis for the future design of inhibitors of P. gingivalis.
To verify that PGN_0806 and RagB function as receptors in P. gingivalis-S. cristatus communication, we tested gene expression in the Δ0806 and ΔragB strains in the presence or absence of peptide4 and compared these to that in the wild type strain 33277. The results showed that loss of PGN_0806 prevented peptide4-dependent regulation of fimA, mfa1, rgp, and kgp (Fig. 6a). Although the ragB mutation did not completely block pep-tide4 activity, a significantly reduced inhibitory effect was observed toward all of the target genes. Previously, a two component regulatory system (FimS/R) was identified to be activator of the fimA expression 22,23 . We thus tested the role of FimS/R in S. cristatus-P. gingivalis cell-cell communication. Although expression levels of fimA and mfa1 were repressed approximately 20 and 4 fold in the fimS and fimR mutants (data not shown), Peptide4 mediated regulation of FimA expression reminded intact in the absence of FimS and FimR (Fig. 6b), suggesting FimS/R is not involved in this bacterial cell-cell communication. These results provide strong evidence that PGN_0806 and RagB, either separately or in combination, act as receptors in the bacterial cell-cell communication between P. gingivalis and S. cristatus.
Expression of fimbrial proteins and gingipains at the translational level was also determined using Western blot analysis. P. gingivalis 33277 was grown with peptide4 at concentrations of 0, 3, 12, and 48 µM (0×, ¼×,  1×, and 4× IC50 of fimA expression) for 48 h. As shown in Fig. 7a,b, production of FimA, Mfa1, and HGP44 (a binding domain of RgpA) was significantly decreased in the presence of 12 and 48 µM of peptide4. However, production of immunoreactive 53 kDa antigen was not altered, consistent with the expression pattern observed at the transcriptional level. Transmission electron microscopy further showed that there were few fimbriae on the surface of P. gingivalis grown in media supplemented with peptide4 (16 µM), when compared to P. gingivalis cells grown without peptide4 (Fig. 8a,b).

Discussion
P. gingivalis possesses a vast array of effector molecules and systems that enable the organism to colonize and survive in the oral cavity, communicate with other bacteria, and ultimately elevate the virulence of the entire microbial community [24][25][26][27] . Major fimbriae (long fimbriae) composed of FimA, are promiscuous adhesins and contribute to colonization, biofilm formation, cell invasion, bone resorption, and the evasion of host defense systems 25,[28][29][30][31][32][33][34][35][36][37] . With regard to induction of immune dysbiosis, FimA binds the CXC-chemokine receptor 4 (CXCR4) and induces crosstalk with TLR2 that inhibits the MyD88-dependent antimicrobial pathway 24 . Both the major and minor (Mfa1) fimbriae of P. gingivalis mediate coadhesion with S. gordonii and are thus involved in synergistic pathogenicity [38][39][40] . The majority of P. gingivalis clinical isolates are fimbriated, especially those isolated at the base of periodontal pockets [41][42][43] . Other well-known virulence factors are the gingipains which include two arginine-and one lysine-specific cysteine proteinases (RgpA, RgpB, and Kgp) 44 . Thus far, all tested P. gingivalis strains produce gingipains that are both membrane-associated and secreted soluble forms 45 . Besides their role in tissue matrix destruction due to proteolytic activity 46 , gingipains play an important role in biofilm formation of P. gingivalis through the C-terminal adhesive regions of RgpA and Kgp or through processing profimbrillin 47,48 . Gingipains are also involved in modulating immune responses, by cleavage of secreted chemokines and intracellular immune kinases 49,50 . Previously, we reported that S. cristatus ArcA represses fimA expression in P. gingivalis 16,51 . Similar results, reported by others 17,52 , showed downregulation of both fimA and mfa1 fimbriae by Streptococcus intermedius ArcA. In these studies ArcA enzymatic activity is required for an effect of on biofilm formation through arginine depletion, suggesting an additional indirect role of ArcA in P. gingivalis colonization 52 . These observations suggest that ArcA modulates expression of fimbrial proteins in P. gingivalis both directly and indirectly. Collectively, accumulating observations suggest that ArcA modulates expression of fimbrial proteins in P. gingivalis both directly and indirectly. Here, we identified a functional motif of ArcA, located at the C-terminal and spanning amino acids 249-259, and a peptide (peptide4) derived from this region showed inhibitory activity for both mRNA and protein expression of fimbriae (FimA and Mfa1) and gingipains (RgpA/B and Kgp). Hence this peptide is a potential candidate for developing inhibitors against P. gingivalis. Based on our observation that ArcA specifically binds to the surface of P. gingivalis, it is likely that the peptide inhibitors would be specific for this organism and not have a significant inhibitory effect on early biofilm colonizers (streptococci and actinomyces). Targeting P. gingivalis alone would likely be sufficient to impede the development of a dysbiotic biofilm, as P. gingivalis is considered a keystone pathogen 8,53 . Cell surface receptors are important elements in signal transduction, and possess the ability to bind (sense) a specific signal, subsequently eliciting a specific cellular response. A well-known signal transduction process in bacteria involves two-component regulatory systems which involve a sensor histidine kinase and a response  ) in surface extracts of P. gingivalis 33277 exposed to pwptide4 at concentrations 0, 3, 12, and 48 µM were analyzed using a Western blot analysis. (b) Semiquantitation of western blots was conducted with ImageJ software. Each bar represents the intensity of the protein band. An asterisk indicates a significant difference between the relative intensity of the protein bands in P. gingivalis exposed to peptide4 (3, 12, or 48 µM) compared to those seen in P. gingivalis not exposed (0 µM, 1 unit) (P < 0.05 by t test). regulator protein 54 . The FimS/R two-component signal transduction in P. gingivalis predominantly regulates fimA expression 55,56 and some other genes including mfa1 22,23 . However, results from the current study showed that FimS/R was not involved in communication between P. gingivalis and S. cristatus, since expression levels of fimA and mfa1 in the fimS or fimR mutants were also modulated in response to peptide4. However, we identified two P. gingivalis surface proteins, RagB (PGN_0294), a major immunodominant antigen of P. gingivalis, and PGN_0806 annotated as a MotA/TolQ/ExbB proton channel family protein, which interact with ArcA of S. cristatus, especially with the peptide4 region of ArcA. Interestingly, the Pseudomonas aeruginosa TonB-ExbB-ExbD protein complex is reported to be involved in signal transduction 57 . Moreover, RagA, which is thought to associate with RagB on the P. gingivalis surface, is a TonB-dependent receptor [58][59][60] . We also demonstrate that mutation in the ragB gene partially blocks the inhibitory activity of ArcA against fimA, while the P. gingivalis strain carrying mutation in the pgn_0806 gene was completely abrogated in response to peptide4. These data corroborate the role of PGN_0806 and RagB, as receptors, in P. gingivalis-S. cristatus communication. Although we identified a signal peptide and potential receptor(s), the mechanisms of intracellular signal transduction are still unidentified. A previous study showed that expression of rgpA, but not kgp, was decreased in a prtT mutant, which indicated that expression of kgp and rgp is not coordinately regulated 61 . Therefore, we speculate that independent intracellular transmitters are involved in control of fimA, mfa1, kgp, and rgp.
In conclusion, we have identified a functional domain of S. cristatus ArcA that has high affinity to the surface of P. gingivalis and is able to repress expression of several well-known virulence genes involved in production of fimbriae and gingipains. We also uncovered two surface proteins of P. gingivalis, RagB and PGN_0806, which interact with ArcA and are required for bacterial cell-cell communication between P. gingivalis and S. cristatus. These results functionally characterize and molecularly dissect the antagonistic relationship between these oral bacteria that we reported earlier 16,62 . Application of these findings should provide the basis for therapeutic strategies designed to reduce colonization of P. gingivalis in the oral cavity and suppress the pathogenicity of periodontitis-associated dental plaque.
Transwell co-culture assay. P. gingivalis cells (10 5 ) were inoculated in each well of a six well plate, and S. cristatus CC5A or its arcA mutant (10 7 ) was added into the transwell inserts with a polycarbonate porous membrane of pore size 0.4 µm or 8 µm. After 16 h growth, the bacterial cells in each well of the plate were collected by centrifugation. To determine the number of CC5A migrated to the lower wells, the bacteria in the lower wells were harvested by centrifugation, and DNA was released by boiling the samples for 20 mins. Numbers of bacteria were determined by qPCR using specific primers for CC5A arcA and 33277 16s-rRNA (Table S1). P. gingivalis RNA was purified using an RNeasy mini spin column which selectively lyses P. gingivalis cells and not S. cristatus cells. Expression of the fimA gene in P. gingivalis was measured using real time qRT-PCR (see below). RNA isolation and qPCR. P. gingivalis were homogenized in Trizol Reagent (Invitrogen, Carlsbad, CA) and RNA was purified using an RNeasy mini spin column (Qiagen, Valencia, CA). RNA samples were digested on-column with RNase-free DNase, and total RNA was tested using an Agilent 2100 Bioanalyzer to ensure the quality of the samples. Primers are listed in Table S1. Amplification reactions consisted of a reverse transcription using a Bio-Rad iScript Reverse Transcription Supermix on a TC-3000 thermal cycler (Techne, Staffordshire, ST15 OSA, UK) and a real-time qRT-PCR analysis using a QuantiTect SYBR Green RT-PCR Kit (Qiagen) on an iCycler MyiQ TM Real-Time PCR detection system (Bio-Rad Laboratories, Inc, Hercules, CA) according to the manufacturer's instructions. P. gingivalis 16 S rRNA gene was used as a normalizing gene. The melting curve profile was analyzed to verify a single peak for each sample, which indicated primer specificity. The expression levels of the investigated genes for the test sample were determined relative to the untreated calibrator sample by using the comparative cycle threshold (ΔC T ) method. ΔC T was calculated by subtracting the average C T value of the test sample from the average C T value of the calibrator sample, and the value used to calculate the ratio between the two by assuming 100% amplification efficiency. By loading the same amount of total RNA for any comparable samples, the ΔC T represents the difference in gene expression between the samples. Confocal microscopy. ArcA (50 µg) was purified from S. cristatus CC5A as described 22 , mixed with P.
gingivalis cells (10 8 ) in PBS and incubated at room temperature for 1 h. After washing three times with PBS, bacteria-protein complexes were blocked in PBS with 5% BSA for 1 h. ArcA bound to P. gingivalis was detected by staining with rabbit anti-ArcA polyclonal antibody (1:400) and tetramethyl rhodamine isothiocyanate (TRITC, 1:500)-conjugated AffiniPure Goat Anti-Rabbit IgG antibody. Visualization was with a LSM 510 inverted confocal microscope with selected filters (543 nm excitation and 560 nm emission).
Pull-down assays. To isolate and identify P. gingivalis surface molecule(s) that interacts with ArcA, we performed a pull-down assay. Surface extracts of P. gingivalis were prepared by sonication with a Sonic Dismembrator (Fisher Scientific; output control 8, 20 s × 3), and the cell debris were removed by centrifugation followed by filtration (0.2-μm pore size). Surface extracts of 33277 were mixed with purified ArcA on a rotator at room temperature for 1 h, and then added to an ArcA antibody coupled Sepharose 4B column (Sigma-Aldrich).
After incubation at room temperature for 1 h, the column was washed three times with PBS containing 0.01% Tween, and proteins were eluted by 0.1 M Glycine pH 2.4. After SDS-PAGE, bands were excised and identified by liquid chromatography-mass spectrometry.
Protein Interaction Screening. Surface extracts of P. gingivalis were collected by sonication (1 min for three times) and centrifugation (13 000 g for 30 min) followed by filtration (0.2 μm pore size). Peptide microarray analysis was performed by PEPperPRINT (Heidelberg, Germany). An ArcA microarray was generated with 409 different peptides of ArcA, and each peptide contained 15 amino acids with a peptide-peptide overlap of 14 amino acids. After blocking and washing, the array was incubated with surface extracts (100 µg/ml) isolated from P. gingivalis 33277, the 0806 mutant, or the ragB mutant. P. gingivalis surface proteins bound on the ArcA array were detected using anti-P. gingivalis antibodies and sheep anti-rabbit IgG (H + L) DyLight680. The arrays were analyzed with a LI-COR Odyssey Imaging System. Peptide synthesis and activity. Peptides were synthesized by PEPperPRINT and Biomatik (Delaware) and purified with High Performance Liquid Chromatography (HPLC) at ≥90% purity. Peptides were resuspended in nuclease/proteinase-free PBS, aliquoted, and stored at −20 °C. Inhibitory activity of peptides were determined as described 16 with slight modification. Peptides were mixed with 5 × 10 5 cells of P. gingivalis, spotted onto a TSB blood agar plate, and cultured for 60 h anaerobically. The expression levels of fimbrial and gingipain mRNA or proteins were measured using qRT-PCR or western blot analyses.
Western Blot Analysis. P. gingivalis cells were lysed using BugBuster ® Protein Extraction Reagent (EMD Millipore, Darmstadt, Germany), and the protein concentration determined using a Bio-Rad protein assay (Bio-Rad). Lysates (0.5 µg) were separated by 12% sodium SDS-PAGE, and transferred to nitrocellulose membranes (Invitrogen, Carlsbad, CA) with a Mini Transblot Electrophoretic transfer cell (Bio-Rad) at 100 V for 1 h. The membrane was blocked with 3% BSA in PBS for 1 h and incubated with polyclonal anti-FimA, anti-Mfa1, anti-HGP44 (a C-terminal adhesin domain of gingipains), or anti-PGN-0128 antibodies diluted 1:1,000 for 1 h. After washing with PBS, the membrane was incubated with anti-rabbit horseradish peroxidase-conjugated secondary antibodies for 1 h. The proteins were visualized using enhanced chemiluminescence (GE Healthcare Bio-Sciences Corp, Pittsburgh, PA) and measured using a semi-quantitative Western blot technique with ImageJ software (NIH, Bethesda, Maryland).
Transmission electron microscopy. P. gingivalis cells were grown on TSB blood plates for 48 h with or without peptide4. Bacterial cells were collected and resuspended in PBS. 20 µl of bacterial suspension were applied to a Formvar-coated copper grid (200 mesh, Electron Microscopy Sciences, PA) and air dried. The bacterial cells were then negatively stained with 0.5% ammonium molybdate for 4 min and observed with a transmission electron microscope (Philips CM-12, Portland, OR) operated at 80 kV.

Statistical analyses.
A student's t-test was used to determine the statistical significance of differences in gene expression profiles and growth rates of P. gingivalis strains. A p < 0.05 was considered significant.