PspC domain-containing protein (PCP) determines Streptococcus mutans biofilm formation through bacterial extracellular DNA release and platelet adhesion in experimental endocarditis

Bacterial extracellular DNA (eDNA) and activated platelets have been found to contribute to biofilm formation by Streptococcus mutans on injured heart valves to induce infective endocarditis (IE), yet the bacterial component directly responsible for biofilm formation or platelet adhesion remains unclear. Using in vivo survival assays coupled with microarray analysis, the present study identified a LiaR-regulated PspC domain-containing protein (PCP) in S. mutans that mediates bacterial biofilm formation in vivo. Reverse transcriptase- and chromatin immunoprecipitation-polymerase chain reaction assays confirmed the regulation of pcp by LiaR, while PCP is well-preserved among streptococcal pathogens. Deficiency of pcp reduced in vitro and in vivo biofilm formation and released the eDNA inside bacteria floe along with reduced bacterial platelet adhesion capacity in a fibrinogen-dependent manner. Therefore, LiaR-regulated PCP alone could determine release of bacterial eDNA and binding to platelets, thus contributing to biofilm formation in S. mutans-induced IE.

Introduction Bacterial biofilm formation determines not only pathogenesis, but also antibiotic resistance upon colonizing host microenvironments to cause infectious diseases. Yet, the adaptation of bacteria or components responsible for forming biofilm in vivo have remained largely unclear. A typical example is infective endocarditis (IE), which is a biofilm-associated disease and most frequently caused by staphylococci or oral streptococci [1,2]. Previously, we demonstrated these IE pathogens can hijack platelets and neutrophil extracellular traps (NETs) to promote biofilm and vegetation formation [3][4][5]. In addition, plasma components can enhance Streptococcus mutans survival in the blood circulation and modulate bacterial biofilm formation through enhancing autolysin maturation for releasing bacterial extracellular DNA (eDNA), which is an important matrix component of biofilm on the damaged heart valve in vivo [6]. These data suggested that host factors can modulate bacterial virulence causing IE, but the bacterial regulatory components responsible for regulating bacterial virulence for causing IE remain unclear.
Bacteria adapt to environmental stimulations and regulate downstream gene expression through two component systems (TCSs) [7]. TCSs typically consist of a sensor kinase and a response regulator, which is generally a DNA-binding protein that modulates the expression of target genes [8,9]. Although sensor kinases and response regulators are usually paired and localized in operons, they also crosstalk with other TCSs for regulating bacterial gene expression in response to a variety of environmental stimuli. Bioinformatic analysis of the S. mutans UA159 genome sequence identified 13 putative TCS (TCS-1 to TCS-13) and one orphan response regulator (GcrR or CovR) in S. mutans [10]. Among these TCSs, the role of several in bacterial biofilm formation has been demonstrated, including TCS-1 (VicKR) [11], TCS-2 (CiaHR) [10], TCS11 (LiaRS) [12], TCS-13 (ComDE) [13], and GcrR [14]. In vitro, these TCSs have been shown to regulate biofilm formation by regulating glucosyltransferase (Gtf) systems, which control the sucrose-mediated glucan production crucial for S. mutans biofilm formation on polystyrene and dental surfaces [11,15]. However, there is no sucrose in plasma, and glucosyltransferase-deficient strains do not reduce their abilities to cause IE [16], suggesting S. mutans exploit other bacterial or host components, instead of glucan, to form biofilms on heart valves in IE. Consistently, our previous study demonstrated that platelets and bacterial eDNA play important roles in S. mutans biofilm formation on the heart valve in IE [6]. The data suggested novel TCS-regulated components in S. mutans may determine biofilm formation in vivo.
In this study, using in vivo survival assay and microarray analysis, we identified LiaR plays an essential role in biofilm formation on damaged heart valves in situ in an experimental IE rat model. The underlying mechanisms mainly involved novel phage shock protein C (PspC) domain-containing protein (PCP)-mediated eDNA release and platelet adhesion capacities. In addition, PCP is well conserved in other streptococcal pathogens, implying their essential roles in pathogenesis through modulating biofilm formation.

LiaR-deficient S. mutans has reduced ability to cause vegetation in vivo
Two component systems sense environmental cues and responses by rapidly regulating gene expression. To identify essential genes of S. mutans in vivo, an in situ biofilm formation was screened on damaged heart valves using a panel of isogenic mutant strains of S. mutans TCS with unknown functions, including SMU.1146, SMU.927, SMU.1815, SMU.659, SMU.1038, SMU.1008, SMU.1964, SMU.576, SMU.487, and SMU.1547 [10] via intravenous infection ( Fig  1A). Briefly, a mixture of 10 transcription regulator isogenic mutant strains of S. mutans [10 8 colony forming unit (CFU)/each] were infected into catheterized rats, and each mutant strain in the mixture could be recognized by specific primers (Table 1; Fig 1B, upper panel). Twentyfour hours after infection, the bacteria colonized inside the vegetation were harvested and identified using specific primers (Fig 1B, bottom panel). Interestingly, SMU.487 (also named liaR)-deficient bacteria were hardly found in the harvested bacterial mixture. To confirm the role of LiaR in the pathogenesis of S. mutans-induced IE, the liaR-deficient strains of S. mutans UA159 and GS5 were tested for their virulence in the experimental rat IE model. LiaR-deficient strains exhibited a significantly reduced ability to colonize on heart valve and form The role of LiaR in the pathogenesis of S. mutans-induced IE was investigated using ΔliaR and comΔliaR strains of GS5 and UA159 in experimental rat endocarditis models. The number of colonized bacteria inside vegetations (C) and vegetation size (D) were measured. Data represent the means ± standard error of the means and were statistically analyzed using the Kruskal-Wallis test with subsequent Dunn's test; � P < 0.05, ns, not significant. IE, infective endocarditis; PCR, polymerase chain reaction; and CLSM, confocal laser scanning microscopy.
https://doi.org/10.1371/journal.ppat.1009289.g001 biofilm-related septic thrombi (vegetation) compared with the parental strain (Fig 1C and 1D, S1 Fig and S2 Fig). Complementation of the liaR gene into liaR-deficient strains restored the phenotype, confirming the role of LiaR (Fig 1C and 1D, S1 Fig and S2 Fig). These data suggest LiaR regulates S. mutans virulence in the pathogenesis of IE.

LiaR regulates PCP-mediated biofilm formation in vivo
To observe bacterial biofilm formation on damaged heart valves in vivo, a green fluorescent protein (GFP)-tagged S. mutans GS5 strain was used in the experimental IE model [3]. Consistently, the liaR-deficient strain showed reduced ability to form biofilms on the damaged valve in situ compared with the parental strain (Fig 2A and S3 Fig). To identify LiaR-regulated downstream genes, microarray analysis was performed (S1 Table). Notably, among the LiaRregulated genes, expression of SMU.753, which encodes a PspC domain-containing protein (named PCP here) [17], was dramatically reduced in the liaR-deficient mutant strain (S1 Table). Data from reverse transcriptase (RT)-and chromatin immunoprecipitation (ChIP)polymerase chain reaction (PCR) assays confirmed LiaR can directly bind the promoter region to regulate expression of pcp (Fig 2B and 2C). The Pcp-deficient mutant also showed a reduced ability to colonize and form a biofilm on the damaged valve (Fig 2A, 2D  PspC is a member of the Psp stress response system, which contains four core components (PspF, -A, -B, and -C) and modulates the bacterial stress response in Gram-negative bacteria, such as Escherichia coli and Yersinia enterocolitica [18]. However, S. mutans and other streptococci only contained a PspC homolog (PCP) and did not have other Psp operon genes compared with Y. enterocolitica ( Fig 3A). A comparison of amino acid sequences with PCPs of the other pathogen streptococci showed good-preservation among these pathogenic streptococci (Fig 3B and 3C). These data suggested that the LiaR-regulated protein, PCP, mediates in vivo

PLOS PATHOGENS
S. mutans biofilm and vegetation formation, and preservation of PCP implied this same role in biofilm formation of other streptococci.

PCP modulates bacterial eDNA release, which contributes to bacterial biofilm formation in vitro and in vivo
Our previous data showed that bacterial eDNA release, mediated by autolysin (AtlA), contributes to S. mutans biofilm formation in IE [6]. Therefore, the role of PCP in eDNA-dependent biofilm formation was further investigated. Both of liaR-and pcp-deficient mutant strains showed normal growth rates but increased chain length compared with the wild type parental strain (Fig 4A and 4B). They also showed reduced ability to release DNA to the culture supernatant and reduced the eDNA-dependent biofilm formation capacity in sucrose-free defined M4 medium (Fig 4C-4F) (6). Complementation of bacterial DNA to the culture medium restored the biofilm formation ability of the pcp-deficient strain, confirming the role of eDNA in PCP-mediated biofilm formation ( Fig 4G). Moreover, we further detected eDNA inside the vegetation by staining with propidium iodide (PI) in the rat IE model [6]. The confocal laser scanning microscopy (CLSM) image results showed that eDNA was embedded inside the bacterial aggregates of S. mutans GS5 and the complementation strains ( Fig 5A, yellow areas, white arrows), but liaR-or pcp-deficient mutant strains appeared as single populations or aggregates without embedded eDNA ( Fig 5A, red arrows). Quantification of the fluorescence  (C) eDNA released into the medium of S. mutans cultures, including wild type, ΔliaR, Δpcp, comΔliaR, and comΔpcp was PCR-amplified by bacterial 16S rRNA primers, and the products were resolved on 1% agarose gels. (D) CLSM images represent S. mutans GS5 wild type, ΔliaR, Δpcp, comΔliaR, and comΔpcp biofilms (630× magnification). S. mutans were labeled with GFP (green), and bacterial eDNA was stained with 10 μM propidium iodide. (E) The fluorescence intensity for the eDNA images of biofilm using propidium iodide staining was measured. The results show the quantified data of the extended focus images of biofilms using Image J software (National Institutes of Health). Data are expressed as the means ± standard deviations from three experiments. ��� P < 0.001 by 1-way analysis of variance. ns, not significant. (F) Biofilms of S. mutans GS5 wild type, ΔliaR, Δpcp, comΔliaR, and comΔpcp were stained with 0.1% crystal violet and the absorbance quantified at 550 nm. Data are expressed as the means ± standard intensity of eDNA images using PI staining in yellow areas showed that eDNA was reduced inside the mutant strain biofilms compared with wild-type the complementation strains ( Fig  5B). To further confirm the reduced bacterial eDNA inside the vegetation samples of mutant strains, total DNA was extracted from the harvested vegetations without lysing the bacteria, and bacterial eDNA was semi-quantified by polymerase chain reaction (PCR) with bacterial 16S ribosomal RNA (rRNA) primers [6]. Consistent with the CLSM analysis, the eDNA levels detected in the liaR-or pcp-deficient mutant strains samples were much lower than in the wild-type and complementation strains (Fig 5C and 5D). Together, these data suggest that LiaR and PCP mediate bacteria to release eDNA, which contributes to bacterial biofilm formation in the pathogenesis of IE. The phenotype of the pcp-deficient mutant strain is quite like the atlA-deficient mutant strain, in its increased cell length and reduced ability to release eDNA to biofilm in vitro and in vivo [6]. Therefore, AtlA expression and calcium ion-induced AtlA maturation were investigated [19]. However, neither the liaR-nor the pcp-deficient strains showed reduced ability to express and process AtlA (S4 Fig). These data suggest that PCP modulates S. mutans eDNA release without affecting AtlA expression and maturation.

PCP mediates S. mutans biofilm formation via platelets in plasma in vitro and in vivo
Bacteria bind to platelets and form aggregates to enhance bacterial biofilm formation in IE [3]; therefore, the role of PCP in platelet-dependent bacterial biofilm formation was investigated further. Both liaR-and pcp-deficient mutant strains showed reduced ability to form biofilm with platelets in the plasma in vitro (Fig 6A). Consistently, they also showed reduced ability to form biofilms with platelets on damaged valves in experimental IE rat model (Fig 6B). Bacterial abilities to bind platelets and induce their aggregation mediates bacterial ability to form biofilms with platelets [3]. Interestingly, neither the liaR-nor the pcp-deficient strains showed reduced ability to induce platelet aggregation ( Fig 6C) but rather had a lower ability to bind platelets when fibrinogen (Fg) was added compared with the parental strain and complementation strains (Fig 6D). The liaR-and pcp-deficient mutant strains showed a reduced ability to adhere to immobilized Fg, suggesting that the pcp-deficient strain has reduced ability to bind platelets, which is at least partly due to the reduced Fg binding ability ( Fig 6E). Together, these data suggest that PCP is an essential gene for S. mutans biofilm formation in vivo as it controls eDNA release and bacterial platelet adhesion on damaged heart valves to induce IE.

Discussion
Using an in vivo survival assay, the present study identified an important S. mutans response regulator, LiaR, which regulates bacterial biofilm formation on the damaged heart valve in experimental rat IE model. Distinct from other identified biofilm-regulating TCSs of S. mutans, which regulate Gtf-mediated glucan formation, LiaR regulates bacterial eDNA release and platelet adhesion in vivo. In addition to the response regulators, we also tried to identify the sensor kinases involved in pcp regulation and bacterial biofilm formation on the damaged heart valve, and we found none of the tested sensor kinases (including SMU.1128, SMU.1145, deviations of triplicate experiments; �� P < 0.01 by 1-way analysis of variance. ns, not significant. (G) Biofilms of S. mutans GS5 wild type and ΔliaR grown with or without series concentrations of purified bacterial eDNA were stained with 0.1% crystal violet. Staining was quantified by measuring the absorbance at 550 nm. Data are expressed as the means ± standard deviations of triplicate experiments; ��� P < 0.001 by 1-way analysis of variance. These. experiments were repeated three times, with a representative experiment shown. GFP, green fluorescent protein; eDNA, extracellular deoxyribonucleic acid; PI, propidium iodide; and CLSM, confocal laser scanning microscopy.  (A) GFP-tagged S. mutans GS5 wild type, ΔliaR, Δpcp, comΔliaR, or comΔpcp strains were intravenously injected into experimental IE rats, and the biofilm inside the vegetations harvested from injured heart valves were observed by CLSM (630× magnification). The data represent the CLSM images of biofilm formation inside the vegetation. S. mutans were labeled with GFP (green), and bacterial eDNA was stained with 10 μM propidium iodide. Yellow areas represent the presence of both S. mutans and eDNA (white arrows) and red arrows indicate single populations or aggregates without embedded eDNA. These experiments were replicated at least three times, with a representative experiment shown. (B) The fluorescence intensity of eDNA images using propidium iodide staining in the yellow aera was measured. The results show the quantified data of the extended focus images of biofilms using Image J software (National Institutes of Health). Scatter plots show values from individual rats, and each rat represents an independent experiment. Data are expressed as the means ± standard deviations. ��� P < 0.001 by 1-way analysis of variance. (C and D) Total DNA isolated from vegetation (without lysing bacteria) was PCR-amplified with bacterial 16S rRNA primers and products were resolved on 1% agarose gels (C) and quantified with Image J software (National Institutes of Health) (D). Scatter plots show values from individual rats, and each rat represents an independent experiment. Data are expressed as the means ± standard deviations; ��� P < 0.001 by 1-way analysis of variance. GFP, green fluorescent protein; IE, infective endocarditis; eDNA, extracellular deoxyribonucleic acid; and CLSM, confocal laser scanning microscopy; PI, propidium iodide.  [20]. Therefore, the sensor kinase responsible for regulating PCP expression and biofilm formation on heart valves in IE needs further examination.
In addition to LiaR, the current in vivo survival assay results also revealed the SMU.1146deficient strain had reduced ability to colonize damaged heart valves in vivo, compared with other mutant strains (Fig 1B). Our preliminary data showed SMU.1146 deficiency did not  [21]. Previously, we demonstrated the role of VicK in regulation of AtlA maturation, which plays important role in bacterial eDNA-dependent biofilm formation in IE [6]. However, because VicR is essential for bacterial viability, vicR-deficient strains could not be generated to investigate the role of VicR in AtlA maturation and S. mutans biofilm formation in IE. Therefore, the roles of VicR, GcrR, CiaR, and ComE in the pathogenesis of IE needs further investigation.
In addition to regulate S. mutans biofilm formation on damaged valves, LiaR has also been shown to play a role in the formation of biofilm under oxidative stress, development of genetic competence, and resistance to acidic pH [12,22]. It has also been reported to regulate expression of numerous genes, including pcp (SMU.753) [23]. However, no study has yet characterized the function of PCP in the modulation of S. mutans virulence in bacterial pathogenesis. PCP is a small protein composed of only 99 amino acids that contains a domain of PspC which is a member of the Psp stress response system [24]. The main function of the Psp system is to protect bacteria from several extreme stresses, which has been well-characterized in Gram-negative bacteria, including E. coli and Y. enterocolitica [25]. The Psp system in Gramnegative bacteria contains four core components: PspF, -A, -B, and -C [24,26]. PspA forms a complex with PspF to inhibit its activity and can also interact with the PspB-PspC complex in the inner membrane, where it may assist bacterial stress relief and maintains the proton motive force [27,28]. PspB and PspC form an integral inner membrane complex involved in activation of the Psp response and prevents toxicity of mislocalized secretins in Y. enterocolitica strains [29,30]. Most review manuscripts indicate the crucial role of PspA in the Psp stress response system [24,31]. However, pspC-null, but not pspA-null, strain of Y. enterocolitica shows severely attenuated virulence in a mouse infection model [32], suggesting the independent role of PspC in bacterial virulence modulation. In addition, PspA homologs are only found in some Gram-positive bacteria, including Bacillus subtilis [33] and Mycobacterium tuberculosis [34], but PCP can be found in most streptococci (https://www.ncbi.nlm.nih.gov/gene). Although PspC has been reported to have a role in preventing the toxicity of mislocalized secretins in Y. enterocolitica strains [29,30], little is known about the actions of PspC and the function of the PspC domain. PCP is a protein family with relatively low molecular weight and lacks any other identified functional domains, suggesting it may influence bacterial eDNA release and platelet adhesion by modulating the expression or function of other bacterial surface proteins in S. mutans. Therefore, our current working hypothesis is that PspC or PCP may affect the function and/or stability of bacterial surface-located proteins or outer membrane instability in some Gram-negative bacteria. The function of PCP will be further investigated, which may also provide important information on the function of PspC in Gramnegative bacteria.
Our previous study demonstrated that bacterial eDNA contributes to S. mutans biofilm formation on damaged heart valves of endocarditis rats [6]. The same study also showed that the autolysis protein AtlA in S. mutans contributes to bacterial eDNA release to biofilm in vitro and in vivo [6]. The phenotype of the pcp-deficient strain is quite similar to that of the atlAdeficient strain in its increased chain length and reduced ability to release eDNA (Fig 4), suggesting that PCP may play a role in AtlA-mediated autolysis. However, AtlA expression and calcium ion-induced AtlA maturation remain unchanged in pcp-deficient strain compared with the wild type strain (S1 Fig). Similarly, a previous study also showed inactivation of S. mutans chaperones RopA and PrsA increased bacterial chain length but did not influence AtlA expression [35]. These data imply that PCP may modulate the function of AtlA or influence S. mutans autolysis systems. In addition to AtlA-mediated autolysis, an S. mutans eDNA release system involving protein secretion machinery-mediated vesicle formation has also been proposed [36], but its role in the pathogenesis of IE and the underlying mechanism remain unclear. Therefore, the role of PCP in protein secretion machinery-mediated eDNA release will be also further investigated. A recent study showed that eDNA crosslinks glucan to stabilize the S. mutans biofilm on dental surface [37]. However, our previous study showed that glucosyltransferase-deficient strains do not have a reduction in the ability to cause IE [16], suggesting that glucan may not play a role in S. mutans biofilm formation on damaged heart valves. Moreover, the pcp-deficient mutant strain had no defect in the expression of glucosyltransferases or in glucan-dependent biofilm formation (S6 Fig). These data further confirm that glucan does not play a major role for bacterial biofilm formation during the pathogenesis of IE.
In addition to mediating eDNA release, PCP also mediates bacterial ability to bind platelets and form biofilm with them in an Fg-dependent manner. Fg-binding proteins, such as ClfA in Staphylococcus aureus, have been identified in several IE pathogens and play important roles in bacterial binding to platelets and IE pathogenesis [38,39]. However, no Fg-binding protein was identified in S. mutans. Here, addition of Fg enhanced S. mutans binding to platelets, suggesting an unidentified Fg-binding protein plays important roles in S. mutans biofilm formation with platelets. It is posited that other S. mutans Fg-binding protein(s) will be identified by investigating the function of PCP further. In addition, our previous study showed bacterial biofilm with platelets on damaged valves recruits neutrophils and induces NET formation, which contributes to vegetation expansion. Therefore, the liaR-or pcp-deficient strains, which have a reduced ability to form biofilms with platelets on damaged valves, also show a reduced ability to induce NETs (Fig 6B), which partly contributes to the reduced vegetation formation in the IE rat models (Figs 1D and 2E).
Taken together, the present study identified an S. mutans response regulator (LiaR) and downstream gene (PCP) which play roles in S. mutans biofilm formation on damaged heart valves in the pathogenesis of IE. These data also confirm the crucial roles of bacterial eDNA release and platelet adhesion in the pathogenesis of IE. In addition, conservation of PCP among streptococci further suggest the putative role of the PCP in biofilm formation of other streptococci.

Ethics statement
Human blood samples were collected from healthy volunteers, and the study was approved by the National Taiwan University Hospital Committee for Regulation of Human Specimens and Volunteers (Taipei, Taiwan; No. 201901005RINC). All human volunteers provided informed consent.

ΔliaR, Δpcp, and comΔpcp strain construction
All references to genomic loci are based on the S. mutans UA159 genome database (https:// www.ncbi.nlm.nih.gov/genome/). To construct ΔliaR and Δpcp strains, these genes in S. mutans GS5 were disrupted by insertion of a promoterless kanamycin cassette using a ligation-PCR mutagenesis strategy [40]. Briefly, the promoterless kanamycin resistance gene fragment was isolated from the pALH124 plasmid by EcoRI digestion [41]. Approximately 500-bp fragments before and/or after liaR and pcp were PCR-amplified using the forward and reverse primers listed in Table 1. PCR reaction products were digested with EcoRI, ligated with the kanamycin resistance gene fragment, and transformed directly into S. mutans GS5. Correct allelic replacement in the derived isogenic mutant strains was confirmed by PCR amplification and RT-PCR analysis was performed to verify the loss of liaR and pcp expression and confirm expression of genes downstream of liaR and pcp. For construction of the comΔliaR and comΔpcp strains, gene fragments of the liaR promoter, liaR, pcp promoter, and pcp were PCRamplified using the primers listed in Table 1. PCR products were digested with XbaI or EcoRI and cloned into the pMC340B_Cm plasmid that contained the chloramphenicol resistance gene [19]. The resulting plasmid was transformed into the ΔliaR or Δpcp strains to generate the comΔliaR or comΔpcp strains. Expression of liaR or pcp was further confirmed by RT-PCR.

S. mutans-induced IE rat model and in vivo survival assay
Approval of animal use was obtained from the National Taiwan University Institutional Animal Care and Use Committee (Taipei, Taiwan, Republic of China) prior to initiation of experiments. A modified rat model of experimental streptococcal endocarditis was performed as previously described [16]. Briefly, a polyethylene tube with a stainless-steel wire embedded inside was inserted into the left carotid artery above the clavicles and advanced toward the chest to create injury at the aortic valves. Twenty-four hours after catheter placement, rats were intravenously infected with the bacteria at 1 × 10 9 CFU. For in vivo survival assay, a mix of 10 insertion mutants of transcription regulators was injected (10 8 for each). At 24 h postinfection, the vegetation was harvested and homogenized by ultrasonication. The bacteria colonized inside the vegetation was detected by PCR using specific primers listed in Table 1. For confocal laser scanning microscopy (CLSM) analysis, GFP-tagged S. mutans GS5 wild type, ΔliaR, Δpcp, or comΔpcp strains (1 × 10 9 CFU) were intravenously injected into the tail veins of Wistar rats. For the detection of bacterial eDNA, the vegetation was stained with propidium iodide. For platelets, the vegetation was stained with an anti-CD42d antibody (1:50 dilution; eBioscience, San Diego, CA, USA), followed by Texas red-tagged anti-rabbit immunoglobulin G antibody (1:500 dilution; Jackson ImmunoResearch Labs, West Grove, PA, USA), and DNA was stained with Hoechst 33258 (Sigma-Aldrich, St. Louis, MO, USA). Bacterial biofilms were then observed on a confocal microscope (Leica TCS SP5). For histologic analyses and gram staining, heart tissue was fixed with 2% paraformaldehyde, embedded in paraffin, cut into 3μm-thick slices, and stained with hematoxylin and eosin or modified gram staining [42].

Microarray analysis
The microarray analysis of GS5 wild type and ΔliaR strains was performed as previously described [19]. S. mutans oligonucleotide microarrays were provided by The Institute for Genomic Research and supported by the National Institute of Dental and Craniofacial Research (NIH, USA). RNA of the bacteria grown in the exponential phase was extracted using an RNeasy Mini Kit (Qiagen, Germantown, MD, USA) following the manufacturer's protocol. After purification of RNA using an RNA cleanup kit (Qiagen), cDNA synthesis, Cy dye coupling, and array hybridization was performed as The Institute for Genomic Research. Briefly, cDNA synthesis was carried out with random hexamers and RT, and the mixture was incubated at 42˚C for 16 h. Following completion of cDNA synthesis, RNA was hydrolyzed with NaOH and the aminoallyl-labeled cDNA samples were purified using a QIAquick purification kit (Qiagen) followed by drying in a speed vacuum. The samples were then resuspended in 1 M NaCO 3 and incubated with Cy dye (resuspended as recommended by the manufacturer) for 2 h in the dark. Each Cy3-labeled experimental sample was mixed with a Cy5-labeled reference sample and the mixtures were dried using a speed vacuum. For hybridization, the samples were resuspended in hybridization buffer and heated at 90˚C for 10 min. Following a quick centrifugation, the Cy dye-coupled probes were applied to microarray slides and incubated at 42˚C water bath for 17 h. Slides were then washed and scanned using an Axon Gene-Pix 4000B scanner (Axon Instruments, Foster City, CA, USA). The data were analyzed by GeneSpring software and Significance Analysis of Oral Pathogen Microarray Data [43]. Microarray analysis data were further confirmed by RT-PCR, which was performed as previously described [44]. The primers for RT-PCR are listed in Table 1.

Recombinant LiaR construction and anti-LiaR antibody preparation
Recombinant LiaR expression plasmids were constructed by cloning of relevant PCR amplification products into the BamHI site on the pQE30 vector (Qiagen) containing a sequence encoding an N-terminal in-frame 6XHis-tag. Th PCR primers for generation of LiaR are listed in Table 1. For expression, resultant LiaR expression plasmids were transformed into E. coli M15, and protein expression was induced during logarithmic growth by addition of 1 mM isopropyl-β-D-thiogalactopyranoside. After induction, the proteins were purified via nickel chelating affinity chromatography (Qiagen) as previously described [45]. The purity of the proteins was analyzed by Coomassie blue staining following sodium dodecyl sulfate (SDS)polyacrylamide gel electrophoresis analysis. Antisera against LiaR were prepared as previously described [46]. Briefly, the purified rLiaR was suspended in 3 mL of phosphate-buffered saline (PBS) and emulsified in Freund's complete adjuvant before being used to immunize rats by intracutaneous injection. Two subsequent booster injections were intravenously given at 2-week intervals using incomplete Freund's adjuvant, and sera was collected 2 d after the last boost. The specificity of the anti-LaR rat serum was confirmed by Western blot analysis using total protein extracts from wild type S. mutans or the liaR-deficient mutant strain.

ChIP assay
ChIP assays were performed with the anti-LiaR antibody as previously described [47] with some modifications. Briefly, S. mutans GS5 were cultured to exponential phase (OD 550 = 0.4), then formaldehyde at a final concentration of 1% was added directly to the culture. The crosslinking reaction mixture was incubated at room temperature for 15 min with continuous shaking. Then, the reaction was stopped by addition of glycine (to 125 mM) for 5 min at room temperature. After washing with PBS three times by centrifugation, the chilled cells were resuspended in lysis buffer [50 mM Tris-HCl (pH 8.0), 500 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 0.1% SDS, 0.1% sodium deoxycholate, and 100 mM PMSF] containing glass beads (BioSpec, Bartlesville, OK, USA). The bacterial cells were sonicated (FastPrep FP120 homogenizer, Qbiogene Inc., Montreal, QC, Canada) to fragment chromosomal DNA to a majority size range of 0.5-1.0 kb. After centrifugation, the supernatant was harvested and used as the "input" fraction for the ChIP assay. The anti-LiaR or preimmune rat serum (5 μL/mL) was added to the input fraction for immunoprecipitation. After incubation overnight at 4 o C on a rotating platform, protein A Sepharose beads (Amersham Pharmacia Biotech, Piscataway, NJ, USA) were added to the mixtures and incubated for 1 h. After washing with lysis buffer twice, the beads were sequentially washed twice by ChIP wash buffer [10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 0.25 M LiCl, and 0.5% sodium deoxycholate] and twice with TE buffer [10 mM Tris-HCl (pH 8.0) and 1 mM EDTA]. The antibody-protein-DNA complexes were further eluted with elution buffer [1% SDS, TE (pH 8.0), and 10 mM dithiothreitol] for 30 min at 37 o C. After protein digestion by proteinase K (20 mg/mL; Invitrogen, Carlsbad, CA, USA), the DNA were further purified by addition of equal volumes of a phenol-chloroform-isoamyl alcohol mix (25:24:1). After vortexing, the mixture was centrifuged at 4˚C for 5 min at 16,000 × g. The aqueous phase was harvested, and the DNA was further precipitated by addition of 0.1 volumes of 3 M sodium acetate and 2.5 volumes of ice-cold 100% ethanol and incubated at -80˚C. After centrifugation, the precipitated sample was air dried and suspended in 30 μL of TE buffer, which was used as the "output" fraction for PCR analyses.

Semiquantification of eDNA
Bacterial eDNA was detected and purified as previously described [6]. Briefly, the biofilms adhered to the culture tubes were disrupted and suspended by violent agitation. The culture medium was harvested by centrifugation. 1 μL of supernatant was used as a template for semiquantitative PCR (25 cycles) using specific bacterial 16S rRNA primers (Table 1). M4 medium (1 μL) without seeding bacteria was used as a negative control, and medium containing S. mutans GS5 genomic DNA was used as a positive PCR control. For the detection of bacterial eDNA inside the vegetation, total DNA was extracted without lysing bacteria using the Gentra Puregene Kit (Qiagen) following the manufacturer's protocol. The DNA pellet was hydrated in 100 μL DNA hydration solution, and 1 μL was used as a template for semiquantitative PCR (35 cycles) using specific bacterial 16S rRNA primers ( Table 1).

Biofilm formation assay
The eDNA-dependent biofilm assay was performed as previously described using sucrose-free defined M4 medium [6]. Briefly, bacterial biofilm growth was initiated by inoculating individual wells of a 96-well polystyrene microtiter plate with approximately 10 7 CFU in 200 μL M4 medium. Each assay was performed in triplicate, and wells without biofilms were used as blank controls. After a 16-h incubation at 37˚C, the biofilm were stained with 0.1% crystal violet and quantified by measuring the absorbance at 550 nm using a MicroELISA reader (Dynatech Corp., Alexandria, VA, USA). For CLSM analysis, GFPuv-tagged strains were transformed with pPDGFPuv, and the bacterial biofilms were cultured in a 24-well plate with a round glass coverslip in individual wells. For eDNA-dependent biofilm, the biofilms were cultured in M4 medium. For bacteria-platelet aggregate biofilm, the biofilms were grown in the PRP [3]. After a 16-h incubation, biofilms that had formed on the glass coverslip were gently washed with PBS three times and then fixed with 2% paraformadehyde for 15 min. After fixation, bacterial eDNA embedded in the eDNA-dependent biofilm was stained with 10 μM propidium iodide, and the platelets in the bacteria-platelet aggregate biofilm were stained with rhodamine-conjugated phalloidin (1:200 dilution; Invitrogen). After washing with PBS three times, the coverslips were transferred to a slide and observed by CLSM (Leica TCS SP5).

Platelet adhesion assay
The platelet adhesion assay was performed as previously described with some modifications [50]. Microtiter plate wells were coated with 100 μL of bacteria (OD 550 = 1.0); fibrinogen (20 μg/mL) and BSA (1%) coated wells served as positive and negative controls, respectively. After blocking with 1% BSA at 37˚C for 1 h, the plate was washed with Tyrode solution three times, 50 μL of platelet suspension (3-5 × 10 8 platelets/mL) with or without Fg (100 μg/mL) was added and incubated at 37˚C for 30 min. Pure Fg was purchased from Sigma-Aldrich. Nonadherent platelets were removed by three washes with Tyrode solution, and bound platelets were lysed at 37˚C for 2 h in acid phosphatase detection buffer [0.1 M sodium acetate (pH 5.5)], 0.1% Triton X-100, and 10 mM p-nitrophenol phosphate]. The results were quantified by measuring the absorbance at 405 nm using a MicroELISA reader.

Bacterial Fg-adherence assay
The Fg adherence assay was performed as previously described [46]. Briefly, the microtiter plates were coated with 50 μL of purified Fg (0.1 mg/mL in 0.05 M sodium carbonate buffer, pH 9.6). After washing three times with PBS-Tris buffer, plates were blocked with 1% BSA. Overnight-cultured S. mutans wild type and mutant strains were washed, resuspended in 0.5 mL of PBS, and adjusted to an absorbance of 0.9 at 550 nm. The bacterial suspension (50 μL/ well) was then applied to the Fg-coated assay plates and incubated 37˚C at room temperature for 2 h. Nonadherent bacteria were removed with three gentle washes of PBS. Adherent bacteria were fixed with a 15-min incubation at 60˚C. After cooling to room temperature, a 1:5000 dilution of antiserum against S. mutans [51] was added, followed by incubation for 1 h at 37˚C. After washing with PBS three times, a 1:10,000 dilution of horseradish peroxidaselabeled goat anti-rabbit immunoglobulin G (Sigma) was applied, followed by addition of 3,3',5,5'-tetramethylbenzidine substrate (Clinical Science Product Inc. MA, USA). The reactions were stopped by addition of 2 N H 2 SO 4 , and the absorbance at 450 nm was measured using a MicroELISA reader.

Statistical analysis
An unpaired, 2-tailed Student's t-test was used to analyze the statistical significance of the difference between two sets of data. Differences between more than two sets of data were assessed using 1-way analysis of variance followed by the Bonferroni multiple-comparisons test. The Kruskal-Wallis test with subsequent Dunn's test were used to analyze nonparametrically distributed data. A P < 0.05 was considered statistically significant. (B) Biofilms of S. mutans GS5 wild type, Δpcp, and comΔpcp cultured in BHI medium containing 1% sucrose were stained with 0.1% crystal violet and the absorbance quantified at 550 nm. Data are expressed as the means ± standard deviations of triplicate experiments; ns, not significant by 1-way analysis of variance. (TIF)