Porphyromonas gingivalis diffusible signaling molecules enhance Fusobacterium nucleatum biofilm formation via gene expression modulation

ABSTRACT Background Periodontitis is caused by a dysbiotic shift in the dental plaque microbiome. Fusobacterium nucleatum is involved in the colonization of Porphyromonas gingivalis, which plays a key role in dysbiosis, via coaggregation and synergy with this microorganism. Aim We investigated the effect of diffusible signaling molecules from P. gingivalis ATCC 33277 on F. nucleatum TDC 100 to elucidate the synergistic mechanisms involved in dysbiosis. Methods The two species were cocultured separated with an 0.4-µm membrane in tryptic soy broth, and F. nucleatum gene expression profiles in coculture with P. gingivalis were compared with those in monoculture. Results RNA sequencing revealed 139 genes differentially expressed between the coculture and monoculture. The expression of 52 genes was upregulated, including the coaggregation ligand-coding gene. Eighty-seven genes were downregulated. Gene Ontology analysis indicated enrichment for the glycogen synthesis pathway and a decrease in de novo synthesis of purine and pyrimidine. Conclusion These results indicate that diffusible signaling molecules from P. gingivalis induce metabolic changes in F. nucleatum, including an increase in polysaccharide synthesis and reduction in de novo synthesis of purine and pyrimidine. The metabolic changes may accelerate biofilm formation by F. nucleatum with P. gingivalis. Further, the alterations may represent potential therapeutic targets for preventing dysbiosis.


Introduction
Periodontitis is a highly prevalent disease, with a global age-standardized prevalence of approximately 10% [1]. Inflammation caused by this disease disrupts the periodontal ligament, induces resorption of the alveolar bone, and results in tooth loss [2,3]. A major feature of this disease is dysbiosis, which is a shift in the composition and abundance of the subgingival microbiota toward more pathogenic species [4]; however, the details of the dysbiotic shift are yet to be elucidated. Understanding the intricacies of dysbiosis is essential for developing treatment and prevention strategies for this condition. Periodontopathic bacteria implicated in dysbiosis include Porphyromonas gingivalis, Tannerella forsythia, Treponema denticola, and Filifactor alocis, which are abundant in periodontitis lesions [5][6][7]. Among them, P. gingivalis is a key microorganism for the induction of this dysbiotic shift of the microbiome [8].
Coaggregation, which is adherence between bacteria, is essential for dental plaque formation [9][10][11]. Fusobacterium nucleatum is one of the most frequently detected bacterial species in both the healthy oral microbiome and periodontal disease lesions. This microorganism is associated with systemic conditions such as preterm birth and colorectal cancer [5,[12][13][14].
In an in vivo study, F. nucleatum was detected between the basal layer and top layer of the dental plaque, and P. gingivalis was detected in the top layer [15]. In another in vivo study, Fusobacterium was detected in the annulus area between the tooth side and peripheral area of the plaque. Porphyromonas was detected in the peripheral area and was found adjacent to Fusobacterium [16]. In vitro studies revealed that F. nucleatum may coaggregate with periodontopathic bacteria such as P. gingivalis, T. denticola, and Prevotella intermedia [17][18][19] and exert a synergistic effect with periodontopathic bacteria, including P. gingivalis, on pathogenicity or biofilm formation [18,[20][21][22]. F. nucleatum has a coaggregation ligand Fap2 that mediates the adhesion of F. nucleatum to P. gingivalis and red blood cells [17] and promotes the growth of P. gingivalis in aerobic and CO 2 depleted conditions [23]. The localization of F. nucleatum and P. gingivalis in the plaque structure and their coaggregation and synergy indicate that F. nucleatum assists the colonization of P. gingivalis. In addition, Feuille et al. found an increase in the virulence of P. gingivalis cocultured with F. nucleatum [24]. Metzger et al. also reported the synergistic pathogenicity of P. gingivalis and F. nucleatum using a murine subcutaneous chamber model [21]. Recent proteomic analyses have provided insights into synergic interactions via cross-feeding among species such as Streptococcus gordonii, F. nucleatum, and P. gingivalis using a chemically defined medium or brain heart infusion broth [25,26]. In these studies, S. gordonii showed synergism with F. nucleatum by providing ArcD-regulated ornithine as an energy source, whereas F. nucleatum facilitated P. gingivalis biofilm formation by providing putrescine. These studies provide insights into how dysbiosis is involved in biofilm formation. However, the effect of P. gingivalis on F. nucleatum was not fully investigated.
Previously, we showed that F. nucleatum TDC 100 isolated from apical periodontitis lesions strongly adheres to type-I collagen and shows enhanced biofilm formation through a synergistic relationship with P. gingivalis [27]. Coinfection of these species also enhanced the invasion of epithelial cells by P. gingivalis [28,29]. Enhanced F. nucleatum biofilm formation by P. gingivalis was observed in a coculture separated with an 0.4-µm membrane, which was significantly higher than that observed for other bacterial species [27], indicating that a factor produced by P. gingivalis plays an important role in the synergy between F. nucleatum and P. gingivalis. Therefore, in this study, we investigated the effect of diffusible signaling molecules of P. gingivalis on the mRNA and protein expression in F. nucleatum to clarify the mechanism of enhanced F. nucleatum biofilm formation by P. gingivalis.

Induction of biofilm formation using a two-compartment system
The interaction of the diffusible signaling molecules between F. nucleatum and P. gingivalis was evaluated via coculture separated with an 0.4-µm membrane using a two-compartment system as described previously [27]. F. nucleatum and P. gingivalis were inoculated into tryptic soy broth (TSB) containing 5 µg/mL hemin and 0.5 µg/mL menadione (TSBhm) and precultured anaerobically for 1 day at 37°C. Then, each culture was diluted with fresh TSBhm to an OD 660 = 0.1 and cultured anaerobically. After 24 h, each culture was diluted 1:2 with fresh TSBhm. Next, 750 µL of the resulting F. nucleatum culture was added into the wells (lower well) of a 12-well plate, coated with type-I collagen and polystyrene (Iwaki, Tokyo, Japan). Upper wells (Transwell, Corning, NY) were then inserted into the lower well, followed by the addition of 750 µL of the resulting P. gingivalis culture to the upper wells. The organisms were then cocultured while being physically separated by an 0.4-µm pore size membrane. For monoculture, only 750 µL of TSBhm was added to the upper well. After anaerobic incubation at 37°C for 2 days, the upper wells were removed. Biofilm formation was evaluated using crystal violet staining with Spectra MAX M5 (Molecular Device, Sunnyvale, CA) [31].

F. nucleatum biofilm protein profiling
F. nucleatum biofilms were lysed with a sample buffer containing a reducing agent, and the lysate proteins were resolved using 10-20% sodium dodecyl sulfatepolyacrylamide electrophoresis (SDS-PAGE). The protein profiles were determined by staining the gel with Coomassie Blue. In the cocultured F. nucleatum protein profile, the 39-kDa protein showed considerably enhanced staining intensity. To identify the protein band, the electrophoresed bands were transferred from the gel to a polyvinylidene difluoride (PVDF) membrane (Immobilon-P; Millipore, Billerica, MA) using Trans-Blot SD Semi-Dry Transfer Cell (Bio-Rad, Hercules, CA). After staining the membrane with Coomassie Blue, the 39-kDa protein band was excised from the membrane, and the N-terminal amino acids were sequenced using the automatic peptide sequencer Procise 494 cLC (Applied Biosystems, Foster City, CA). The peptide sequences were matched with the genomic sequences of F. nucleatum in the NCBI database (http://blast. ncbi.nlm.nih.gov).
Based on the obtained gene sequence, an antibody against the protein was prepared. The peptide ESEVKNWRWQPTAW showed the highest similarity to residues 352-365 of Fusobacterium periodonticum outer membrane protein A (FomA), which has an additional N-terminal Cys residue. This sequence was identical between F. nucleatum and F. periodonticum except for the sixth residue. The peptide was synthesized using F-moc chemistry and was conjugated to keyhole limpet hemocyanin using maleimidobenzoic acid N-hydroxy succinimide ester, and the antibody was prepared as described previously [32]. To quantify FomA, immunoblotting was performed as described earlier. Ten micrograms of biofilm organized in the lower well, which was monocultured and cocultured with P. gingivalis, was separated using SDS-PAGE and transferred onto a PVDF membrane. FomA was stained with anti-FomA antibody (1:1,000 dilution) and goat antirabbit IgG (1:3,000 dilution). The membrane was scanned, and the staining intensity of the developed 39-kDa bands was evaluated using Image-Quant TL software v.8.1 (GE Healthcare, Little Chalfont, UK).

RNA-sequencing
F. nucleatum TDC 100 was grown as a monoculture and cocultured with P. gingivalis ATCC 33277 for 2 days using the two-compartment system. After removing the upper wells, three wells containing F. nucleatum were combined into one sample for RNA extraction, and the total RNA was isolated using TRIzol (ThermoFisher Scientific, Tokyo, Japan). The total RNA was obtained from three independent replicates of monocultured and cocultured F. nucleatum. DNA was removed using the TURBO DNA-free kit (Thermo Fisher Scientific, Waltham, MA), and ribosomal RNA was extracted from the sample using the Ribo-Zero rRNA Removal kit (Bacteria) (Illumina). A library was constructed from the extracted RNA using the Truseq Stranded mRNA Sample Prep kit (Illumina) according to the manufacturer's instructions. The obtained library was sequenced on HiSeq 2500 (Illumina); 100-bp paired-end reads were trimmed using cutadapt (https://cutadapt.readthedocs. org/en/stable/) and Trimmomatic (http://www.usadel lab.org/cms/index.php?page=trimmomatic). As the genomic DNA of F. nucleatum TDC 100 has not been sequenced, the genomes of F. nucleatum subsp. nucleatum ATCC 25586, F. nucleatum subsp. vincentii ATCC 49256, F. nucleatum subsp. vincentii 3_1_36A2, and F. nucleatum KCOM 2931 were used as references to map the obtained sequences. Among these subspecies, F. nucleatum KCOM 2931 showed the highest mapping rate (89.7%), and its 16S rRNA sequence showed 99% identity with that of F. nucleatum TDC 100. Thereafter, the obtained sequences were mapped using TopHat (http://ccb.jhu.edu/software/tophat/index.shtml) and Bowtie1 (http://bowtie-bio.sourceforge.net/index.shtml) with F. nucleatum KCOM 2931 as the reference strain. From the mapped sequences, the gene expression in monocultured and cocultured F. nucleatum was evaluated using cufflinks and cuffdiff. The difference in expression between monoculture and coculture was compared using TCC-GUI [33]. Normalization was carried out using the TMM method, and differentially expressed genes were identified using edgeR [34], where findings with a p-value of 0.05 were considered significant and the false discovery rate (FDR) was set to 0.05, with differences in gene expression evaluated using log2 fold ≥1.5. Differentially expressed genes were subjected to Gene Ontology (GO) analysis using goatools v. 1.1.6 [35].

Statistical analysis
To investigate the effect of coculture on biofilm formation, the level of biofilm formation and FomA expression were analyzed using Student's t-test at a 5% level of significance.

Effect of P. gingivalis coculture on F. nucleatum protein profile
Biofilm formation under P. gingivalis coculture was greater than that under F. nucleatum TDC 100 monoculture ( Figure 1b, p < 0.01). A comparison of protein profiles revealed a band at approximately 39 kDa with increased staining intensity for F. nucleatum TDC 100 cocultured with P. gingivalis ATCC 33277 (Figure 1b). The N-terminal amino acid sequence of the band at 39 kDa was EVTPAWRPNG. The BLAST analysis revealed that this sequence was identical to that of residues 49-58 of FomA. The predicted molecular mass of amino acids 1-48 was 5179.17, indicating that the 39-kDa band was FomA without these 48 amino acid residues. The antibody reactivity to the 39-kDa band ( Figure 2a) also supported this result.
The level of the major outer membrane protein FomA was investigated using immunoblotting with rabbit anti-FomA antibody. Image analysis of the immunoblots using Imagequant TL showed that staining of the 39-kDa and faint 35-kDa bands detected with the anti-FomA antibody in cocultured F. nucleatum was 1.7 times more intense than that for F. nucleatum monoculture (Figure 2b).

RNA-sequencing of F. nucleatum TDC 100 cocultured with P. gingivalis ATCC 33277
To evaluate the effect of P. gingivalis on F. nucleatum in coculture, RNA-sequencing analysis was performed. At FDR <0.05, 176 differentially expressed genes were obtained ( Figure 3a). Among these, the genes from the coculture showing a log2-fold difference of >1.5 compared with their expression in the monoculture of F. nucleatum, p < 0.05, and FDR <0.05 were selected (Figure 3b). In F. nucleatum cocultured with P. gingivalis, 139 genes (4.9%) were differentially expressed compared with those in monocultured F. nucleatum. Among them, 52 genes were upregulated (Table 1), including the gene encoding the galactose inhibitable autotransporter adhesion Fap2, a protein involved in porphyrin metabolism (cobK and Cobartprecorrin 5A hydrolase), chaperone proteins (ClpB, DnaK, and DnaJ), proteins involved in polysaccharide synthesis (glycogen synthase, GlgG, glucose-1-phosphate adenylyltransferase, 1,4-alpha-glucan branching protein GlgB, and GalU), and proteins involved in membrane transport (basic amino acid ABC transporter substrate-binding protein, amino acid ABC transporter ATP-binding protein, DMT family transporter, AzlC family ABC transporter permease, branchedchain amino acid transporter permease, basic amino acid ABC transporter substrate-binding protein, efflux RND transporter permease subunit, ABC transporter ATP-binding protein, energy-coupling factor transporter transmembrane protein EcfT, autotransporter serine protease fusolisin, ABC transporter permease, and ABC transporter ATP-binding protein). Interestingly, the expression of fomA did not show an increase in the RNA-sequencing.
The GO enrichment analysis of the 139 differentially expressed genes showed significant changes in de novo synthesis of inosinic acid and uridylic acid and glycogen biosynthesis in biological process (p < 0.05, Figure 4).

Discussion
In the present study, the effect of P. gingivalis diffusible signaling molecules on F. nucleatum was examined to clarify the synergy between these microorganisms involved in a dysbiotic shift of the dental plaque microbiome. F. nucleatum cocultured with P. gingivalis enhanced biofilm formation, and 139 differentially expressed genes were detected compared with those in monocultured F. nucleatum, including the ligand for adherence (fap2) to P. gingivalis. The GO analysis suggested sugar metabolism shift to glycogen biosynthesis, whereas the de novo purine and pyrimidine synthesis were decreased. The expression of fap2 increased significantly in cocultured F. nucleatum. Fap2 is a galactose inhibitable adherence factor involved in coaggregation between F. nucleatum and P. gingivalis, hemagglutination by F. nucleatum [17], and localization of F. nucleatum to colon tumors [36]. The results indicate that the sensing of diffusible signaling molecules from P. gingivalis by F. nucleatum induces an increase in fap2. The increase in Fap 2 may contribute to binding between the species and enhance these intimate interactions, although further analysis is required to confirm the magnitude of contribution. Fap2 is also involved in hemagglutination by F. nucleatum [17]. P. gingivalis has hemagglutination activity, which is involved in heme acquisition [37]. In the cocultured F. nucleatum, cobK and cbiJ expression increased.
These genes are involved in porphyrin metabolism. It is possible that hemagglutination is involved in iron acquisition in F. nucleatum. Therefore, the increase in Fap2 expression may be one of the adaptations for establishing a synergistic community of F. nucleatum and P. gingivalis.
Increased expression of genes involved in the glycogen synthase pathway in cocultured F. nucleatum was detected through the GO analysis, including those encoding GlgG, glucose-1-phosphate adenylyltransferase, GlgB, and GauU. The increase in GlgG and glucose-1-phosphate adenylyltransferase expression in F. nucleatum biofilms formed during coculture with P. gingivalis compared with their expression in a single-species biofilm of F. nucleatum has also been reported in a proteomic study [38]. GalU also  plays an important role in glycogenesis and cell wall synthesis. Mutant galU affects biofilm formation in Escherichia coli [39]. In Pseudomonas aeruginosa, galU is required for lipopolysaccharide-core organization [40]. In Xanthomonas citri, galU is involved in exopolysaccharide and capsular polysaccharide synthesis [41]. galU is also essential for biofilm formation in Vibrio cholerae [42]. In F. nucleatum, the major energy source is amino acid catabolism; however, this microorganism can also metabolize fructose [43]. The study showed that fructose metabolism is stopped, and polysaccharides are organized under amino-acidrich conditions. P. gingivalis has Arg-gingipain; Lysgingipain [44,45]; dipeptidyl peptidases such as DPP4, DPP5, DPP7, and DPP11; as well as serine exopeptidases [46]. The strong proteolytic activity of P. gingivalis is considered to enhance F. nucleatum growth [12]. Here, the expression of genes involved in amino acid metabolism was not altered, but Hendrickson et al. [38] reported that the coculture of P. gingivalis and F. nucleatum inhibited more numbers of amino acid metabolic pathways compared with F. nucleatum alone. F. nucleatum and P. gingivalis were incubated in PBS for 18 h in their study, whereas in this study, microorganisms were inoculated in TSBhm for 2 days, and therefore, we observed large amounts of amino acids. In addition, the subspecies of F. nucleatum used were different between the studies. These differences in the experimental conditions may contribute to the difference in the results. The increase in the expression of genes involved in glycogen synthesis strongly suggested that  the shift of energy source from sugar to amino acid was promoted by the increase in the levels of available amino acids upon coculture with P. gingivalis. In contrast, the expression of pelF and pelG in F. nucleatum was reduced in the coculture. pelF and pelG are involved in the synthesis and transport of the exopolysaccharides Psl and Pel, and their protein production is regulated via cyclic-di-GMP in P. aeruginosa [47,48]. It is possible that the downregulation of these genes results in an increase in biofilm formation in cocultured F. nucleatum, although further analysis of the role of these genes in F. nucleatum is required.
Fusolisin, which is an autotransporter protease [49], showed increased expression in cocultured F. nucleatum. The expression of branched-chain amino acid transporter permease, basic amino acid ABC transporter substrate-binding protein, and amino acid ABC transporter ATP-binding protein was also increased in cocultured F. nucleatum. As mentioned earlier, P. gingivalis shows a strong proteolytic activity. These genes may be involved in peptide digestion or amino acid import. However, in previous proteomic analyses, the expression levels of these transporters did not change in F. nucleatum cocultured with P. gingivalis [38,50] or S. gordonii. In the previous studies, F. nucleatum was cocultured without a membrane, whereas in this study, F. nucleatum was cocultured with P. gingivalis separated by an 0.4-µm pore membrane. One study used Brucella broth [50] and another used Todd Hewitt broth [38]. In addition, in the former study, the samples of biofilms were harvested after 4 days of culture, and in the latter, samples were obtained after 18 h of incubation in PBS; in this study, we harvested samples after 2 days. This might explain the differences in the results.
Genes for de novo synthesis of inosinic acid and uridylic acid were downregulated in our study. These processes involve the synthesis of pyrimidine and purine using amino acids. Previous proteomic studies have found only minor changes in the proteome of F. nucleatum cocultured with P. gingivalis, primarily a decrease in the production of specific proteins [50]. In the present study, the numbers of decreased and increased proteins were comparable.
Here, the expression of genes involved in riboflavin synthesis from GTP was reduced, such as ribD and the riboflavin synthase gene. In eukaryotic cells, metabolic convergence of glutamine toward nucleotide biosynthesis observed in the process of malignant progression and inhibition of the shift reduced the proliferation of cancer cells [51]. It is possible that these reductions associated with purine and pyrimidine reflect changes in the external conditions caused by P. gingivalis, including the amino-acid-rich condition induced by the proteolytic activity, and the downregulation of genes involved in nucleotide biosynthesis from glutamine supports the glutamine for energy production. Furthermore, P. gingivalis and F. nucleatum produce indole that induces increased biofilm formation [52]. The biofilm mass in cocultured F. nucleatum was almost double that in the monoculture. In the biofilm, gradients of nutrients and oxygen induce a gradient of growth rate that results in fast-growing cells at the surface and slowgrowing cells in the deeper layer of the biofilm [53]. It is possible that F. nucleatum growth decreases with increased biofilm formation and, therefore, the expression of the enzymes involved in pyrimidine and purine synthesis is decreased.
The expression of chaperones, including ClpB, DnaK, and DnaJ, was increased in the cocultured F. nucleatum. Although chaperones are generally cytoplasmic, six chaperones, including DnaK and ClpB, were detected in the matrix of F. nucleatum monoculture biofilm [54]. This indicates a relationship between chaperone translocation to the biofilm matrix and increased expression of these genes in F. nucleatum stimulated by P. gingivalis in coculture.
SDS-PAGE and immunoblotting indicated an increase in FomA expression in cocultured F. nucleatum. In the RNA-sequencing analysis, fomA expression did not show a difference between monoculture and coculture. FomA is a major porin protein in the matrix of biofilms formed by F. nucleatum [54]; however, an increase in FomA expression has not been observed in other coculture studies [55]. A recent study reported that the small RNA FoxI acts as a post-transcriptional repressor of FomA induced by oxygen [56]. Therefore, posttranscriptional regulation by FoxI or turnover of the protein in F. nucleatum cells may be responsible for the increase in FomA expression. In this analysis, an internal control was not evaluated, although the quantity of the protein was adjusted and the bands on SDS-PAGE showed almost identical levels. Therefore, further analysis using an internal control is required.
In the present study, F. nucleatum TDC 100 was used because of its strong synergy with P. gingivalis. The result of RNA-sequencing showed that this strain belongs to F. nucleatum subsp. vincentii. Previous studies on such gene or protein expression profiling used F. nucleatum subspecies nucleatum. Subspecies may show differing characteristics, including biofilm formation [57]. Human skeletons from the 18 th and 19 th centuries show high abundance of F. nucleatum subsp. vincentii with P. gingivalis and Prevotella pleuritidis, and this is maintained in modern samples [58]. F. nucleatum subsp. vincentii and P. gingivalis are useful for diagnosing periodontitis in mass patient screening using saliva samples [59]. High abundance of F. nucleatum subsp. vincentii is associated with periodontal lesions, and it further highlights the importance of this subspecies in the periodontopathic microbiome. To fully understand F. nucleatum-associated dysbiotic shift, further analysis of gene expression using multiple subspecies is required.
In this study, the gene expression was evaluated at a single time point before stationary phase, and therefore a single growth phase, of F. nucleatum. Several factors, other than coculture with P. gingivalis, such as differences in the growth phase, affect gene expression. To elucidate the influence of coculture, the evaluation of gene expression at multiple time points and in different ratios between the bacteria is required. TSBhm was used in the current assay; in other proteomics analyses, Brucella broth [50] or Todd-Hewitt broth [38] containing hemin and menadione were used. In the latter studies, the investigation was conducted in PBS. These media can be used to detect synergistic interactions; however, differences in the medium content may result in different results. In addition, a precise investigation on the synergistic effect requires the conditions that simulate those in the oral cavity. To fully comprehend the synergism between F. nucleatum and P. gingivalis, additional research using media that mimic the physiological condition of the oral cavity is required.
In conclusion, the diffusible signaling molecules from P. gingivalis induced an increase in the ligand for coaggregation with P. gingivalis and led to metabolic changes, including the promotion of polysaccharide synthesis and inhibition of the de novo synthesis of purine and pyrimidine in F. nucleatum. The metabolic changes may play a key role in the acceleration of biofilm formation by F. nucleatum cocultured with P. gingivalis. Therefore, the metabolic pathway in which differentially expressed genes involved in this study could be used as potential therapeutic targets for preventing dysbiosis.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
This work was partially supported by JSPS KAKENHI under Grant numbers 245927783 (KI) and 15K1102 (KI).

Data availability statement
The sequence raw data obtained using RNA-Sequencing analysis has been deposited to the Sequence Read Archive in the DNA Data Bank of Japan (DDBJ/DRA) (https://www.ddbj. nig.ac.jp/dra) under submission ID: DRA014805.