Involvement of NADH Oxidase in Biofilm Formation in Streptococcus sanguinis

Biofilms play important roles in microbial communities and are related to infectious diseases. Here, we report direct evidence that a bacterial nox gene encoding NADH oxidase is involved in biofilm formation. A dramatic reduction in biofilm formation was observed in a Streptococcus sanguinis nox mutant under anaerobic conditions without any decrease in growth. The membrane fluidity of the mutant bacterial cells was found to be decreased and the fatty acid composition altered, with increased palmitic acid and decreased stearic acid and vaccenic acid. Extracellular DNA of the mutant was reduced in abundance and bacterial competence was suppressed. Gene expression analysis in the mutant identified two genes with altered expression, gtfP and Idh, which were found to be related to biofilm formation through examination of their deletion mutants. NADH oxidase-related metabolic pathways were analyzed, further clarifying the function of this enzyme in biofilm formation.


Introduction
Biofilms are ubiquitous in nature, in that nearly every microbial species has mechanisms for adhering to surfaces and other cells in mixed-species communities. According to the National Institutes of Health (NIH), biofilms could be responsible for over 80% of microbial infections [1]. In the United States alone, biofilm-related infections account for an estimated 1.7 million infections and 99,000 associated deaths each year (Centers for Disease Control and Prevention Report, 2007). Biofilms are known to help protect bacteria from abiotic stresses, antimicrobials and assaults of the host immune system. Within biofilms, bacterial cells are encased in an extracellular matrix that is typically composed of exopolysaccharides (EPS), proteins, lipids and nucleic acids [2]. Biofilm matrix is closely associated with adhesion of bacteria to biotic and abiotic surfaces and with cohesion of cells in the biofilm [2,3]. Numerous studies have focused on various signals and mechanisms that control expression of the genes required for matrix production and biofilm formation. Quorum-sensing signals such as acyl-homoserine lactone, Pseudomonas quinolone signal and autoinducer-2 have been reported to play important roles in biofilm formation in bacteria [4][5][6].
Streptococcus sanguinis is a normal inhabitant of the oral cavity and one of the pioneer colonizers of tooth surfaces. Biofilm formation is important for S. sanguinis to colonize and interact with other bacterial inhabitants or pathogens in mixed-species communities in the oral cavity. In our previous study, we identified several genes involved in biofilm formation in this bacterium [7].
During glycolysis, bacterial cells produce NADH from NAD + . NADH oxidase has been recognized as playing an important role in maintaining glycolysis by producing NAD + from NADH and, thus, maintaining NAD + /NADH balance. In Streptococcus agalactiae, inactivation of the nox gene encoding NADH oxidase was shown to reduce growth under aerobic conditions [8], while growth was not affected by nox inactivation in Streptococcus pneumoniae under aerobic or anaerobic conditions [9]. In Streptococcus mutans, Nox is responsible for the majority of NADH-dependent oxygen consumption and is involved in adaption to acidic and oxidative stresses [10]. In S. pneumoniae and S. agalactiae, nox inactivation attenuates virulence in animal models [8,9]. In addition, the efficiency of competence for genetic transformation was significantly altered in an S. pneumoniae nox mutant [9]. Therefore, the nox gene plays important roles in many biological functions in streptococci.
In Streptococcus gordonii, the expression of the nox gene was increased in biofilms relative to planktonic cells [11]. Mutation of the rex gene, encoding a NADH-and NAD + -sensing transcriptional regulator, caused biofilms to decrease and to exhibit a more porous and rugged architecture in S. mutans [12]. Rex was shown to interact with the nox gene in a regulatory loop [13]. These results suggest that the nox gene may be involved in biofilm formation in streptococci; however, direct evidence is lacking.
Over the course of screening a library of S. sanguinis mutants for biofilm deficiency, we identified the S. sanguinis nox gene (SSA_1127) as being involved in biofilm formation. The nox gene was annotated as encoding an H 2 O-forming NADH dehydrogenase in S. sanguinis. In another study, we report that the a recombinant Nox protein (rNox) of S. sanguinis has the activity of an H 2 O-forming NADH oxidase under aerobic conditions (Ge et al., submitted for publication), which is consistent with NADH oxidases in S. pneumoniae [9], S. mutans [14] and S. agalactiae [8]. In this study, we confirmed a role for the nox gene in biofilm formation under anaerobic conditions and examined possible mechanisms by which the nox gene could influence biofilm formation. We found that rNox possesses NADH dehydrogenase activity under anaerobic conditions and that the nox mutant exhibited a dramatic reduction in biofilm formation. The nox mutant also exhibited alterations in fatty acid composition and decreases in membrane fluidity, extracellular DNA (eDNA) and bacterial competence. The gtfP and Idh genes, which were significantly down-regulated in the nox mutant, were shown to be involved in biofilm formation.

Ethics Statement
The study was conducted in accordance with the Declaration of Helsinki and saliva was collected using a protocol approved by the Virginia Commonwealth University Institutional Review Board (protocol HM10244).

Saliva collection
Subjects at least 21 years of age, not on medication or ill, and who had not eaten or drunk anything other than water, or brushed their teeth within 60 min prior to collection were recruited. Saliva collection and processing were performed as described previously [15,16]. Briefly, subjects chewed on paraffin to stimulate saliva production and then expectorated into a 50-ml centrifuge tube for 5 minutes. The tube was capped and placed on ice. After collection, saliva was mixed at 4°C for 20 minutes with 2.5 mM dithiothreitol to prevent protein aggregation and then centrifuged at 5000 x g for 20 min. The supernatant was transferred to a new tube, mixed with 3 volume of sterile dH 2 O, and filter sterilized. Samples from 6-10 subjects were pooled and stored at -20°C until use.

Deletion and complementation of the nox gene
The nox open reading frame (ORF) in S. sanguinis SK36 was replaced by a promoterless kanamycin cassette (aphA-3) as described previously [17]. Briefly, three pairs of primers nox_F1 and nox_R1, nox_F3 and nox_R3, and kan_F2 and kan_R2 (Table 1) were used for PCRamplification of 1-kb upstream and downstream flanking regions of the nox gene, and promoterless aphA-3, respectively. The three PCR-amplified fragments were combined by secondround PCR amplification using primers nox_F1 and nox_R3. The final linear recombinant PCR amplicon was transformed into S. sanguinis SK36 to obtain the nox-deleted mutant using kanamycin for selection. The mutant was confirmed by colony-PCR amplification using nox_F1 and nox_R3, followed by amplicon sequencing.
The nox mutant was complemented by a similar strategy. Upstream (1-kb) plus ORF of the nox gene, promoterless erythromycin cassette (erm) and 1-kb downstream of the nox gene were PCR amplified and then combined to obtain a recombinant PCR amplicon in which the nox ORF was followed by the erm cassette. The recombined amplicon was transformed into the nox mutant to obtain a complemented strain of the nox mutant using erythromycin for selection. The strain was confirmed as for the nox mutant. The primers used are listed in Table 1.

Biofilm formation analysis
Biofilm formation of S. sanguinis wild-type and mutants was examined in 96-well or 12-well microtiter plates (Greiner Bio-One, Monroe, North Carolina). Overnight BHI-cultured strains were diluted 100-fold in BM on plates and incubated 16 h at 37°C under anaerobic conditions for biofilm formation [7]. After measuring absorbance at 450 nm for bacterial growth, the plate wells were gently washed with deionized water (dH 2 O), and stained with 50 μl of 0.4% (w/v) crystal violet (Fisher scientific, Pittsburgh, PA) for 15 min at room temperature. After 3 times washing with dH 2 O, the biofilm stain was dissolved in 200 μl of 33% (v/v) acetic acid and then 100 μl transferred for measuring absorbance at 600 nm. For confocal laser scanning microscopy, S. sanguinis biofilm formed on 12-well plates was stained with 10 μM SYTO 9 (Life Technologies, Grand Island, NY) for 15 min at room temperature. After washing, the biofilm was observed with a Leica TCS-SP2 AOBS confocal laser scanning microscope (VCU core facilities) using a laser wavelength of 488 nm and emission wavelengths of 495-525 nm [18]. A series of green fluorescent x-y sections in the z plane of the biofilm was scanned and obtained. Images were analyzed with Image J (National Institutes of Health) [18]. For biofilm formation on human protein-coated plates, 96-well plates were coated with 100 μg/ml fibrinogen, 10 μg/ml fibronectin, 10 μg/ml collagen type I or 10 μg/ml laminin (Sigma-Aldrich, St. Louis, MO) overnight at 4°C, respectively [19]. After washing, the coated plates were used for biofilm formation assays as above. Biofilm formation was performed in human saliva via pre-coating plate wells with human saliva or mixing saliva with BM [20]. For the pre-coating experiment, 100 μl of 25% filter-sterilized saliva was added to the plate wells and incubated at 37°C with gentle shaking. After 2 h incubation, the plate wells were washed with sterile PBS and used for biofilm formation analysis as above. For the experiment with saliva-mixed medium, saliva was added to BM medium to a final concentration of 10% (v/v) prior to analysis as above. PBS was used as a control for saliva.

Expression and purification of rNox protein
The cloning, expression and purification of recombinant S. sanguinis Nox protein in E. coli was performed as described previously [21]. Briefly, the S. sanguinis nox gene was PCR-amplified using primers nox_rp_F and nox_rp_R (Table 1), cloned into pET-46 Ek/LIC vector (Novagen, Madison, WI) and expressed in E. coli BL21(DE3)pLysS (Novagen, Madison, WI) according to the manufacturer's protocols. The expressed rNox protein with N-terminal His tag was isolated using BugBuster buffer (Novagen, Madison, WI) and purified using a His•Bind1 Column Chromatography kit (Novagen, Madison, WI) as described in the manufacturer's protocols.

NADH dehydrogenase activity assay
To identify alternative electron receptors for the rNox in the absence of oxygen, the oxidation of NADH to NAD + by the rNox was analyzed under anaerobic conditions. Reagent preparation and the assays by NanoDrop1 ND 1000 Spectrophotometer (Thermo scientific, Wilmington, DE) were performed in an anaerobic chamber. Each reaction mixture contained 50 mM potassium phosphate buffer (pH 7.0), 1 mM β-NADH, 1 mM each of the potential electron receptors and the purified rNox [22]. The reduction of the acceptors by rNox was monitored at the following wavelengths: FAD, 450 nm; dichlorophenolindophenol (DCIP), 600 nm; menadione, 340 nm; K 3 (FeCN 3 ) 6 , 420 nm; cytochrome c, 550 nm; CoQ 10 , 340 nm; and XTT, 480 nm [22]. Protein concentrations were determined by the Bradford method [23] using bovine serum albumin as a standard. All chemicals were purchased from Sigma-Aldrich (St. Louis, MO).

Membrane fluidity
Membrane fluidity was quantified by fluorescence anisotropy of 1,6-diphenyl 1,3,5-hexatriene (DPH; Sigma-Aldrich, St. Louis, MO) as described previously [24]. Briefly, exponentialgrowth-phase S. sanguinis planktonic cells cultured anaerobically in BM medium were harvested, washed twice with PBS (pH 7.4) and then incubated with 5 μM DPH at 37°C for 1 h. Unlabeled cells were used as a scattering reference. The fluorescence polarization was measured at 37°C using a Cary Eclipse Fluorescence spectrophotometer (VCU Department of Chemistry Instrumentation Facility) with excitation at 360 nm and emission at 430 nm. Fluorescence anisotropy was calculated by the formula A = (I II -I ? )/(I II + 2I ? ), where I II and I ? were the fluorescence intensities parallel and perpendicular to the direction of excitation light, respectively.

Determination of bacterial fatty acid composition
The bacterial cell fatty acid composition was determined by GC/MS as described [25]. S. sanguinis cells were harvested from overnight cultures in BM medium under anaerobic conditions by centrifugation and washed twice with dH 2 O. The cell pellets were mixed with 0.5 ml 1 N sodium methoxide for 1 min and the fatty acid methyl esters were extracted by addition of 0.3 ml hexane containing methyl-10-undecenoate (Sigma-Aldrich, St. Louis, MO) as an internal standard. GC/MS analysis was carried out on Varian Saturn GC/MS (VCU core facilities) equipped with a Restek Stabilwax-DA column (30 m × 0.25 mm × 0.5 μm). Carried gas (H 2 ) velocity was 1 ml/min. Injection and detection temperature were 230°C and 260°C, respectively. The oven temperature was increased from 100°C to 240°C at 5°C/min and maintained for 20 min. The peaks with their retention times were identified by MS. Each fatty acid composition was expressed as percentage of the total content of fatty acids. eDNA assay S. sanguinis cultures incubated overnight in BM medium anaerobically were centrifuged at 6,000 rpm for 10 min at 4°C and the supernatants were collected and passed through 0.45 μm filters (EMD Millipore, Billerica, MA) to eliminate residual cells. Half milliliter of the supernatant was mixed with 0.5 ml TE buffer saturated with phenol-chloroform-isoamyl alcohol (25:24:1; Sigma-Aldrich, St. Louis, MO) [26]. After vortexing for 30 s, the mixture was centrifuged at 16,000 ×g for 5 min at 4°C. The aqueous phase (0.4 ml) was transferred into a new tube and mixed with 40 μl 3M sodium acetate (pH5.2) and 1 ml 100% ethanol. After placement at -20°C for 10 min, the mixture was centrifuged at 16,000 ×g for 10 min at 4°C and the pellets were collected. After air drying, the pellet was suspended in 100 μl dH 2 O and then DNA concentration was determined by measuring absorbance at 260 nm using a NanoDrop1 ND 1000 Spectrophotometer.

Competence assay
Competence of S. sanguinis strains was determined by transformation with pJFP96, a suicide plasmid containing the spectinomycin resistance gene (aad9) with~1 kb upstream and downstream of the SSA_0169 [27]. Briefly, overnight cultures of S. sanguinis strains were diluted 200-fold into pre-warmed TH-HS and incubated microaerobically for 2-5 hrs. Aliquots of 330 μl culture at various incubation times were transferred into pre-warmed microfuge tubes containing 70 ng S. sanguinis competence-stimulating peptide (CSP) and 50 ng pJFP96, and incubated at 37°C for 1 h. Cells were serially diluted, plated on BHI agar plates with and without spectinomycin and grown microaerobically at 37°C for 2 d. The efficiency of transformation was defined as the ratio of spectinomycin-resistant colonies to total CFU.

Microarray analysis
S. sanguinis SK36, nox mutant and complemented strain cells cultured anaerobically in BM medium were harvested from the exponential growth phase and used for microarray analysis [17]. RNA from the sample cells was isolated through lysozyme lysis, mechanical disruption with FastPrep1 lysing matrix B and purified using RNeasy mini kit (Qiagen, Valencia, CA) [17]. DNA was removed by treatment in columns with DNase I during purification. The microarray was performed on spotted microarray slides obtained from the Pathogen Functional Genomics Resource Center at JCVI according to the manufacturer's protocol [17]. Cy3 and Cy5 were used for labeling cDNA of SK36 and the nox mutant/complemented strain, respectively. The microarray data were obtained from scanning of hybridization slides using a GenePix 4100A microarray scanner (Molecular Devices, Downingtown, PA), and analyzed using Spotfinder and Midas programs (TM4 software suite) to obtain the expression ratio of nox mutant/complementation to the wild-type. The microarray data have been deposited in the NCBI Gene Expression Omnibus (GEO) under the accession number of GSE68492.
qRT-PCR S. sanguinis RNA from the microarray assay was used to perform real-time reverse transcription PCR (qRT-PCR) for detecting the expression change of genes ldh and gtfP in the nox mutant. For all reactions, first-strand cDNA was synthesized in a 20 μl reaction mixture containing 4 μl of 5× first-strand buffer, 100 ng RNA, 1.5 μg random primers, 1 μl of 10 mM dNTP mix, 1 μl of 0.1 M DTT, 1 μl of RNaseOUT™ Recombinant RNase Inhibitor (40 U/μl) and 1 μl of SuperScript™ III reverse transcriptase (200 U/μl) following the manufacturer's protocol (Life Technologies, Grand Island, NY). The qRT-PCR reaction was composed of 5 μl 2× SYBR Green PCR master mix (Life Technologies, Grand Island, NY), 10 pmol of each paired primer, 1 μl of 50-fold diluted cDNA template and dH 2 O up to 10 μl using Applied Biosystems 7500 Fast Real-Time PCR system (Life Technologies, Grand Island, NY). The housekeeping gene gyrA was used as a normalization control.

Statistical analysis
All data were obtained in at least triplicates. For data on confocal microscopy and qRT-PCR, one sample t-test was applied to analyze the values of mutant or complemented mutant that differ from 1. The electron receptor data were statistically analyzed by ANOVA followed by Tukey's HSD test. The microarray data were analyzed by t-test with correction of false discovery rate (FDR). Other data were statistically analyzed by t-test. The significance was set as Pvalue <0.05.

Results
The nox gene is involved in biofilm formation We first examined anaerobic growth of overnight cultures of a nox mutant, a complemented nox mutant, and the wild-type parent, SK36, in BHI broth by dilution plating. There were no significant differences among the three strains (data not shown). We next examined biofilm formation of the nox mutant in BM medium because it has been shown to support abundant biofilm formation by S. sanguinis SK36 cells [7]. Biofilms were formed on the polystyrene surfaces of microtiter plate wells overnight under anaerobic conditions and then examined using confocal laser scanning microscopy. The results showed that the amount of biofilm formed by the nox mutant was significantly less than the wild-type (P-value < 0.01) and had a very different structure, whereas biofilm amounts and structure were restored by complementation of the nox mutation (Fig 1).
To examine whether the nox mutant formed reduced biofilm on host cell constituents, human plasma protein fibrinogen and extracellular matrix proteins including fibronectin, collagen and laminin [28][29][30][31] were coated on the plate surface prior to assaying biofilm formation. The results confirmed that the nox mutant exhibited a significant reduction in biofilm formation on different extracellular matrix constituent-coated surfaces compared to the wild-type (Fig 2A). Complementation of the nox mutation restored biofilm formation to the same level as the wild-type. Further, the biofilm formation of the wild-type, mutant and complemented mutant in human saliva were examined. Both in wells pre-coated with saliva or BM medium mixed with saliva, biofilm formation was also dramatically decreased in the nox mutant compared to the wild-type (Fig 2B). There were no significant growth differences among the experimental strains in either medium, suggesting that changes in growth were not responsible for the reduced biofilm formation of the mutant (Fig 2). These results confirmed that the nox gene is involved in biofilm formation in S. sanguinis.

NADH dehydrogenase activity of rNox in the absence of oxygen
Under aerobic conditions, oxygen is the electron acceptor that allows the nox-encoded NADH oxidase to oxidize NADH to NAD + in S. sanguinis (Ge et al., submitted for publication). Here, other potential electron acceptors to allow the rNox protein to oxidize NADH to NAD + were analyzed under anaerobic conditions. As shown in

Membrane fluidity decrease in the nox mutant
To further understand how the nox gene affects biofilm formation, we examined NADH oxidase-related metabolic pathways. We considered that the nox mutant might have an altered cell membrane, because NADH oxidase has been associated with fatty acid biosynthesis in S. agalactiae and S. mutans [8,10]. The membrane fluidity in the nox mutant and the wild-type was assessed using a DPH anisotropy approach. In this assay, an increase in DPH anisotropy represents a decrease in membrane fluidity, and vice versa [24]; small changes in DPH anisotropy may reflect marked changes in membrane microviscosity [32]. In the nox mutant, the DPH anisotropy was significantly greater than the wild-type (Fig 4), indicating reduced membrane fluidity.

Fatty acid composition change in the nox mutant
Membrane fluidity has been found to be linked with fatty acid composition [24]. The finding of membrane fluidity reduction implied that the levels of individual membrane lipids might change in the nox mutant. Thus, we ascertained the composition of fatty acids in the nox mutant and compared to the wild type. As shown in Table 2, the results indicated palmitic acid (C16:0) content was significantly increased but the levels of stearic acid (C18:0) and cis-vaccenic acid (C18:1, ω7) were significantly decreased in the nox mutant compared to the wild-type. These results suggested that deletion of the nox gene led to increase in the content of C16 fatty acids and to decrease in the content of C18 fatty acids in the membrane of S. sanguinis cells. The decrease in unsaturated fatty acid (cis-vaccenic acid) would be expected to result in decreased membrane fluidity [33], in agreement with our findings (Fig 4). Diminishment of eDNA in the nox mutant DNA release has been demonstrated to play an important role in biofilm formation and is associated with bacterial competence [34][35][36][37]. Thus, levels of eDNA in the nox mutant were examined and compared to the wild-type. As shown in Fig 5, the eDNA concentration in the nox mutant was significantly less than that in the wild-type.

Competence suppression in the nox mutant
We also compared the competence of the nox mutant and the wild-type by transformation with pJFP96 [27], a plasmid that inserts a spectinomycin resistance gene into the chromosome at a dispensable location by allelic exchange [38]. As shown in Fig 6, transformation with pJFP96 was significantly suppressed in the nox mutant compared to wild-type, even though CSP was added. This result differs from that seen in S. pneumoniae, where mutation of the nox gene lowered transformation efficiency only when exogenous CSP was not added [9]. It's not clear why the timing of competence development was slightly altered in the complemented  mutant, but the important outcome is that the nox mutant was significantly less competent than SK36 at the first three time points, and longer incubation did not increase its competence.

Nox-controlled biofilm-related genes
To reveal whether nox deletion possibly influenced the expression of other genes involved in biofilm formation, global gene expression of the nox mutant and its complement were compared to the wild-type in BM medium using two-color spotted microarrays. All three strains were grown under anaerobic conditions. The microarray data showed that 13 genes were significantly down-regulated and 71 genes up-regulated in the nox mutant, and most were partially or fully restored in the complemented mutant strain (S1 Table). From the metabolic pathway analysis, three of the down-regulated genes, ldh and acoA and acoB (components of acetoin dehydrogenase Acdh), are responsible for converting pyruvate into lactate and acetyl-CoA in pyruvate metabolism. Amongst the up-regulated genes, there were three, ald, ackA and adhE, that participate in pyruvate metabolism. In addition, several operons associated with sucrose and galactose metabolism were up-regulated in the nox mutant. These include gal (SSA_1003 to SSA_1010) containing gtfA (SSA_1006) and lac operons (SSA_1692 to SSA_1699) for galactose metabolism from sucrose. It was interesting that genes SSA_0509 to SSA_0531, containing an operon for ethanolamine metabolism, showed increased expression. This operon was acquired by horizontal gene transfer [39]. NADH oxidase has been demonstrated to function as an important NADH-oxidizing system in glycolysis in Streptococcus agalactiae [8]. Therefore, these changes in gene expression may be a response to the nox deficiency to ensure normal glycolysis in S. sanguinis. (See below.) Using our comprehensive single-gene deletion library [17], we picked mutants deleted for each of the 13 down-regulated genes and examined their biofilm formation. The vast majority of these mutants did not display significant changes in biofilm formation (data not shown) except that the ldh mutant exhibited an uneven biofilm surface and lower biofilm abundance than the wild-type (Fig 7A). The significant decrease in expression of ldh in the nox mutant was also verified by qRT-PCR analysis (Fig 7B).
The gtf genes have been implicated in biofilm formation in Streptococcus mutans [40,41]. These include gtfB, gtfC and gtfD. S. sanguinis possesses only a single gtf gene, gtfP. We examined gtfP expression in the microarray assay of the nox mutant but were unable to draw a conclusion because its expression levels were too low in most of the experiments. We further quantified its expression using qRT-PCR. The qRT-PCR data indicated gtfP expression was detectable and significantly lower in the nox mutant than in the wild-type (Fig 7B). We then examined the biofilm formation of a gtfP deletion mutant and found that it exhibited a 12-fold reduction compared to the wild-type (Fig 7C). These results suggest that reduced expression of ldh and gtfP may be involved in the biofilm reduction of the nox mutant.

Discussion
Nox orthologs have been demonstrated to possess NADH oxidase activity in S. pneumoniae [9], S. mutans [14] and S. agalactiae [8]. We have also verified that S. sanguinis Nox does indeed function as an H 2 O-forming NADH oxidase (Ge et al., submitted for publication). Without oxygen, however, other electron receptors, such as DCIP, menadione and FAD [14] (Fig 3), could be reduced by the Nox protein in the presence of NADH. Although we do not know the exact electron acceptor(s) employed by the Nox in S. sanguinis cells, FAD is a possible candidate because S. sanguinis has the metabolic pathway for FAD synthesis and BM medium contains riboflavin. Regardless, to our knowledge, this is the first time a streptococcal NADH oxidase has been shown to possess NADH dehydrogenase activity under anaerobic conditions. Remarkably, we identified a marked deficiency in biofilm formation, changes in membrane fluidity and fatty acid composition, release of DNA, and changes in gene expression in the nox mutant under anaerobic conditions. Thus, Nox is important for cell physiology even in the absence of oxygen.
Using confocal laser scanning microscopy, biofilm formation was observed to significantly decrease in the S. sanguinis nox mutant (Fig 1). Similar findings were also shown on plate surfaces coated with human extracellular matrix or plasma proteins as well as human saliva (Fig  2). These results indicate that the nox gene is involved in biofilm formation in S. sanguinis not only on an abiotic surface but also on biotic surfaces. Under aerobic conditions, nox inactivation was demonstrated to impair cell growth in S. agalactiae [8]. The same study also showed that deficiency in aerobic growth of the nox mutant was mostly caused by an underlying defect in fatty acid biosynthesis due to reduction in total fatty acid yield, and suggested that NAD + depletion in a nox mutant probably affects acetyl-CoA production, a precursor for fatty acid biosynthesis. In our study, biofilm formation was defective in the nox mutant of S. sanguinis, but bacterial growth was not influenced by deletion of nox under anaerobic conditions (Fig 2A). Furthermore, fatty acid abundance was not significantly changed in the nox mutant compared to the wild type in the same culture conditions (data not shown).
Microarray data indicated that adhE, ackA, and genes required for ethanolamine metabolism were significantly elevated in their expression in the nox mutant (S1 Table). The adhE gene encodes a bifunctional enzyme that acts as an acetaldehyde dehydrogenase, which converts acetyl-CoA into acetaldehyde with concomitant oxidation of NADH to NAD + , as well as an alcohol dehydrogenase that converts acetaldehyde to ethanol. The ackA gene encodes acetate kinase, which produces ATP from acetyl phosphate. Ethanolamine metabolism has been proposed to contribute to pathogenesis in bacteria such as Salmonella enterica subsp. enterica serovar Typhimurium and E. coli [42]. Ethanolamine catabolism occurs in an ethanolaminespecific microcompartment, where acetaldehyde derived from ethanolamine can be catabolized to ethanol by EutG (which encodes an alcohol dehydrogenase), or to acetyl-CoA by EutE (which encodes an acetaldehyde dehydrogenase) and then converted into acetyl phosphate by EutD (which encodes a phosphotransacetylase). Acetyl phosphate can diffuse freely into the cytoplasm and generates ATP through AckA. Within the microcompartment, CoA and NAD + /NADH can be recycled by EutD, EutE and EutG. In the bacterial cytoplasm, Pta plays the same role as EutD, converting acetyl-CoA into acetyl phosphate [42].
In S. sanguinis, the orthologs of eutE (SSA_0523) and eutG (SSA_0514) are present in the ethanolamine operon but eutD is missing. However, another phosphotransacetylase encoded by pduL (SSA_0527), which has been demonstrated in S. enterica subsp. enterica serovar Typhimurium [43], is present and located within the ethanolamine operon in S. sanguinis. These three genes showed increased expression in the nox mutant (S1 Table). In addition, the expression of the pta gene (SSA_1207) did not change in the microarray analysis of the nox mutant (data not shown), suggesting cytoplasmic phosphotransacetylase encoded by pta might not be involved in the metabolism above. These results suggest that deletion of S. sanguinis nox may cause AdhE to oxidize NADH into NAD + , concomitantly converting acetyl-CoA into acetaldehyde which may enter the ethanolamine specific microcompartment to be catabolized into ethanol by EutG and acetyl phosphate by EutE and PduL, with concomitant recycling of CoA and NAD + /NADH in the microcompartment, and acetyl phosphate could then diffuse into the cytoplasm and generate ATP via AckA. In addition, the ald gene, involved in metabolizing pyruvate into alanine with concomitant oxidation of NADH to NAD + , was up-regulated in the nox mutant, which could enhance oxidation of NADH to NAD + . The down-regulation of two acdh genes acoA and acoB, involved in reduction of NAD + into NADH, could conserve NAD + in the nox mutant. (See Fig 8) Therefore, it is possible that the elimination of Nox-mediated conversion of NADH to NAD + in S. sanguinis cells due to nox deletion may be offset by decreases in acoA and acoB and increases in ald, adhE, ackA and ethanolamine catabolism. These compensatory changes may lead to diminished levels of acetyl-CoA, which though sufficient for normal cell growth, may affect downstream metabolism such as fatty acid biosynthesis (see below).
The composition of fatty acids was shown to be altered in the S. sanguinis nox mutant, as reflected by decreases in stearic acid and cis-vaccenic acid and an increase in palmitic acid (Table 2). Obviously, these results indicated that nox deletion affected the composition of fatty acids in S. sanguinis despite the unchanged total amount of fatty acids. These changes may be due to the limited level of acetyl-CoA in the nox mutant (see above). Moreover, the membrane fluidity of the nox mutant was decreased compared to the wild-type (Fig 4). Fourier transform infrared spectrometry has demonstrated that changes in membrane fluidity are directly associated with the level of unsaturated fatty acids in bacterial biological membranes [33]. Thus, the decrease in membrane fluidity in the nox mutant may result from the reduction in the contents of unsaturated fatty acids, such as cis-vaccenic acid. In Pseudomonas aeruginosa, lipid analysis showed a significant decrease in the uneven-numbered chain phospholipids and a slight increase in long chain phosphatidylethanolamine in biofilm compared to planktonic cells [44]. The results revealed that a decrease in membrane fluidity and lipid stability in the bilayer may Deletion of nox leads to cessation of the nox-mediated oxidation of NADH to NAD + . This causes decreased expression (green) of ldh and two genes encoding components of acetoin dehydrogenase, acoA and acoB, and increased expression (red) of adhE, ackA, ald and ethanolamine utilization genes including eutG, eutE and pdu. These gene expression changes affect the associated pathways to partially compensate for the imbalance of NAD + /NADH caused by nox deletion. Meanwhile, gtfA (red) expression increases and gtfP (green) expression decreases, which may cause more sucrose influx into energy metabolism rather than exopolysaccharide biosynthesis (e.g. glucan). Increased consumption of acetyl-CoA by AdhE may influence fatty acid biosynthesis to alter the membrane fatty acid components and to subsequently decrease the membrane fluidity and may suppress the release of eDNA. The inhibition of DNA release and glucan biosynthesis impairs biofilm matrix and results in defective biofilm formation. Green arrow, reduced reaction or function in the nox mutant; black arrow, normal reaction or function; red arrow, increased reaction or function; hollow arrow, function lost as a result of the nox deletion. be required for biofilm formation in P. aeruginosa. In S. mutans, anthraquinones could suppress biofilm formation on hydroxyapatite and the anthraquinone-treated S. mutans cells showed a significant decrease in membrane fluidity. Based on these reports, our results suggest the change in composition of fatty acids and subsequent decrease in membrane fluidity may be one of the causes of the biofilm defects observed in the nox mutant. Membrane fluidity is critical for maintaining the properties of the bacterial membrane and the functions of membraneassociated proteins, such as permeability of the lipid bilayer, protein mobility, protein-protein interactions and active transport processes [33,43,[45][46][47]. Alterations in the membrane permeability and the functions of membrane-associated proteins due to decrease in membrane fluidity may affect the transport/translocation or assembly of eDNA (see below) and extracellular biofilm matrix in the S. sanguinis nox mutant.
Bacterial eDNA has been identified to play an important role in biofilm formation as well as dispersal [37]. The crucial role of eDNA in stabilizing bacterial biofilm matrix has been revealed in both Gram-negative and Gram-positive bacteria such as P. aeruginosa, Staphylococcus aureus and S. gordonii [36,48,49]. Additionally, eDNA has been shown to promote dispersal in biofilm by inhibiting settling of motile progeny cells in Caulobacter crescentus [50]. In our study, eDNA was significantly reduced in the nox mutant compared to the wild-type ( Fig  5). This result suggests that eDNA decrease may be one of the reasons for the impaired biofilm formation seen in the nox mutant. In streptococci, biofilm biomass has been revealed to be decreased in competence-defective mutants [34,51] and DNA release has been demonstrated to be associated with cell competence [34,35]. Our results showed that transformation frequency was significantly decreased in the nox mutant (Fig 6), suggesting competence was attenuated due to deletion of the nox gene. These results suggest that eDNA reduction in the S. sanguinis nox mutant may be also affected by a decrease in cell competence. In addition, the decrease in membrane fluidity may also affect the uptake of exogenous DNA to attenuate competence.
The gtfB, gtfC and gtfD genes have been reported to act as glucosyltransferases for exopolysaccharide biosynthesis by utilizing sucrose [52,53]. Moreover, these three genes have been reported to be involved in biofilm formation in S. mutans [54]. In S. sanguinis, like many other streptococci, gtfP is the only ortholog of these three genes. Our results showed that deletion of gtfP caused almost complete loss of S. sanguinis biofilm formation (Fig 7C). Apart from gtfB, gtfC and gtfD, S. mutans has another gtf gene, gtfA [52][53][54]. The ortholog of gtfA gene is extensively present in other streptococci and encodes sucrose phosphorylase responsible for converting sucrose into D-fructose and alpha-D-glucose 1-phosphate. In S. sanguinis, gtfA gene is located within a gal operon for galactose metabolism and its mutant did not show a decrease in biofilm formation (data not shown). It is interesting that the expression of gtfP was decreased (Fig 7B) but the expression of gal and lac operons containing gtfA was increased in the S. sanguinis nox mutant (S1 Table). These results indicate that the biofilm defect in the nox mutant may also be associated with reduction in gtfP expression. Meanwhile, an increase in gtfA expression and a decrease in gtfP expression in the nox mutant might cause more sucrose influx into energy metabolism rather than exopolysaccharide biosynthesis for biofilm formation. Furthermore, we also found deletion of ldh led to a more uneven biofilm surface and less biofilm formation than the wild-type ( Fig 7A) and that expression of ldh was significantly decreased in the nox mutant (Fig 7B), possibly implicating ldh in the biofilm defect caused by the nox deletion. L-lactate dehydrogenase, encoded by ldh, converts pyruvate to L-lactate with concomitant oxidation of NADH to NAD + to assure continued glycolysis in lactic acid bacteria including Streptococcus and Lactobacillus [55,56]. It is possible that these functions of ldh may be replaced by ethanolamine catabolism in combination with adhE and ackA activity in the S. sanguinis nox mutant.
In conclusion, an S. sanguinis nox mutant exhibited a significant decrease in biofilm formation. The data suggest that the deficiency of biofilm formation by deletion of nox may be due to changes in metabolic pathways associated with NAD + /NADH balance and sucrose utilization, which might contribute to alteration in membrane fatty acid and subsequent decrease in membrane fluidity, suppression of DNA release, and reduction in the expression of biofilm-associated gtfP and ldh (Fig 8). Since the nox gene is widespread in other species of streptococci, continuation of this work may lead to a comprehensive elucidation of the underlying mechanisms of biofilm formation for streptococci.
Supporting Information S1 Table. Genes significantly down-regulated and up-regulated in the nox mutant and restored in the nox complemented strain. (XLSX)