F. nucleatum subsp. polymorphum ATCC 10953 was grown in brain heart infusion (BHI) broth (Difco, USA) supplemented with 5 μg/ml hemin (Sigma-Aldrich, USA), 1 μg/ml vitamin K (Sigma-Aldrich, USA), and 0.5% yeast extract (Difco, USA). S. gordonii DL1 was grown in BHI broth. Both bacterial strains were grown under anaerobic conditions (N₂ 90%, CO₂ 5%, H₂ 5%) at 37°C.

Coaggregation assays were performed in modified coaggregation buffer (CAB) containing 150 mM NaCl, 1 mM Tris HCl pH 8, 0.1 mM CaCl₂, and 0.1 mM MgCl₂ as previously described (Kaplan et al., 2009; Kaplan et al., 2014; Lima et al., 2017). The bacterial cells were collected at the late exponential phase of growth. The optical density at 600nm (OD_600nm) of F. nucleatum subsp. polymorphum was measured to be 0.80 (~10⁹ CFU/mL), and the OD_600nm of S. gordonii was around 0.65 (~10⁹ CFU/mL). The colony-forming units (CFUs) of bacteria was quantified by incubating F. nucleatum subsp. polymorphum on 5% sheep blood agar plates and incubating S. gordonii on BHI agar plates in serial dilutions under anaerobic conditions. Bacterial cells were cleaned and resuspended in CAB to a final concentration of ~2×10⁹ CFU/mL. Equal numbers of bacterial cells from each species were added together and vortexed for 10 seconds in a new reaction tube. The suspensions were settled at room temperature for 10 min to allow the bacteria to coaggregate with each other. The reaction tube was centrifuged at low speed (100×g) for 1 min to pellet coaggregated bacterial cells while leaving the nonaggregated cells in the supernatant. The supernatant was collected carefully for OD_600nm measurement. The coaggregation index (C.I.) was calculated as follows (Kaplan et al., 2009; Kaplan et al., 2014): C.I. = (OD_600nm(Fnp) + OD_600nm (Sg)-OD_600nm (Fnp-Sg))/[OD_600nm (Fnp)+OD_600nm (Sg)]. In this formula, OD_600nm(Fnp) and OD_600nm (Sg) were the optical density of F. nucleatum subsp. polymorphum and S. gordonii respectively, while OD_600nm (Fnp-Sg) was the optical density of the supernatant after coaggregation. Because saliva is the common coaggregation buffer in the oral cavity, the coaggregation index of F. nucleatum subsp. polymorphum and S. gordonii in different concentrations of artificial saliva (Phygene, China) was also calculated. The coaggregation and autoaggregation of the two bacterial species in CAB at different time points were also evaluated and observed with phase contrast microscopy. The autoaggregation index was calculated as follows (Nagaoka et al., 2008; Karched et al., 2015; Toh et al., 2019): (OD_600nm (time zero value)- OD_600nm (sample value))/(OD_600nm (time zero value).

F. nucleatum subsp. polymorphum and S. gordonii were cultured to the late-exponential phase. Bacterial cells were washed three times and resuspended in sterile PBS. For visualization, F. nucleatum subsp. polymorphum was stained green with 5-(and-6)-carboxyfluorescein succinimidyl ester (CFSE) (Thermo Fisher, USA), while S. gordonii was stained red with hexidium iodide (Thermo Fisher, USA) according to the manufacturer’s instructions. Samples were incubated for 15 min in darkness at room temperature. Fluorescently stained bacteria were washed three times with sterile PBS and resuspended in CAB. The coaggregated F. nucleatum subsp. polymorphum and S. gordonii (Fnp-Sg) were obtained as described above. Coculture of the two species (Fnp+Sg) in PBS, where they did not coaggregate with each other but only mixed physically, were used as controls. After coaggregation reactions, 10 μL of coaggregated Fnp-Sg was transferred to a glass slide and covered with a cover glass. The coaggregation and coculture samples were visualized by an Olympus confocal microscope (FV3000, Olympus, Japan) using excitation (Ex) at 492 nm and emission (Em) at 517 nm for CFSE and Ex/Em = 518 nm/600 nm for hexidium iodide.

Human gingival epithelial cells (hGECs) were obtained from the American Type Culture Collection (ATCC CRL-3397) and incubated in DMEM containing 10% fetal bovine serum (FBS) (Gibco, USA) at 37°C in the presence of 5% CO₂ (Huang et al., 2020). Cells were seeded at 3.5 × 10⁵ cells per well in 6-well cell culture plates (Corning, USA). hGECs were infected with F. nucleatum subsp. polymorphum monoculture (Fnp), S. gordonii monoculture (Sg), coinfection of F. nucleatum subsp. polymorphum and S. gordonii (Fnp+Sg), and coaggregation of F. nucleatum subsp. polymorphum and S. gordonii (Fnp-Sg) at an MOI of 100, respectively. The coinfection of F. nucleatum subsp. polymorphum and S. gordonii (Fnp+Sg) was only a physical mixture of bacteria in PBS where they did not coaggregate with each other. To ensure the number of bacterial cells in coaggregates was similar with monocultures, the coaggregates were resuspended in PBS, vigorously vortexed and disrupted until no visible pellet existed with validation under a microscope (Liu et al., 2021). The CFU of the coaggregates were determined by incubating the resuspension solution on 5% sheep blood agar plates and BHI agar plates in serial dilutions. The volumes of bacterial cells used in the coaggregation group were adjusted to ensure the number of bacterial cells in coaggregates were similar with monoculture groups. After incubation at 37°C in 5% CO₂ for 4 hours, the culture medium containing bacteria was removed, and the bacteria were washed with PBS three times to remove planktonic bacteria. Cells in each well were added to 2 mL of DMEM containing 10% FBS, 200 μg/mL metronidazole (Solarbio, China) and 300 μg/mL of gentamicin (Solarbio, China) and incubated at 37°C in 5% CO₂ for 60 min, 90 min, and 120 min to test the antibiotic treatment time for completely killing of extracellular bacteria in all groups. In detail, hGECs infected with Fnp, Sg, Fnp+Sg, and Fnp-Sg of the same antibiotic treatment time were digested and mixed together. The effect of killing extracellular bacteria was confirmed by incubating the digested cells mixture on a plate containing 10% sterile sheep’s blood at 37°C with 90% N₂ + 5% CO₂ + 5% H₂ for 2-3 days. If bacterial colonies grew on the plate, it meant not all groups achieved a complete killing of extracellular bacteria. After killing the extracellular bacteria, cells were washed with PBS three times and incubated at 37°C in 5% CO₂ for different time points. The experiment was performed three times.

To examine bacterial infection, CFSE-labeled F. nucleatum subsp. polymorphum and S. gordonii were cocultivated with hGECs for 4 hours on cell slides. After infection, hGECs were washed 3 times with PBS to remove planktonic bacteria. The cells were fixed with 4% paraformaldehyde for 15 min and treated with 0.1% Triton X-100 (Beyotime, China) for 10 min. The cytoskeleton was stained with phalloidin (Thermo Fisher, USA) for 30 min, and the nucleus was stained with DAPI (ZSGB-BIO, China). All CLSM images were obtained by an Olympus confocal microscope using Ex/Em = 492 nm/517 nm for CFSE, Ex/Em = 540 nm/565 nm for phalloidin and Ex/Em = 340 nm/488 nm for DAPI.

hGECs were infected with Fnp, Sg, Fnp+Sg, and Fnp-Sg at a MOI of 100. After 4 hours of infection, the cells were washed 3 times with PBS to remove the planktonic bacteria and lysed in sterile water for 90 min to release intracellular bacteria. The total number of F. nucleatum subsp. polymorphum and S. gordonii adhering to and invading hGECs was counted by serial dilution and plating on BHI agar supplemented with yeast extract, hemin, and vitamin K. The agar plates were incubated anaerobically at 37°C with 90% N₂ + 5% CO₂ + 5% H₂ for 2-3 days. For the invasion assay, after 4 hours of infection, the cells were washed 3 times with PBS to remove planktonic bacteria and treated with fresh DMEM supplemented with 10% FBS, 200 μg/mL metronidazole, and 300 μg/mL gentamicin for 120 min to kill extracellular bacteria. Cells were lysed in sterile water for 90 min and the number of intracellular bacteria was determined by serial dilution and plating as described above.

hGECs were inoculated into a 96-well plate (200 μL/well) at a density of 1.0×10⁴ cells per well. Fnp, Sg, Fnp+Sg and Fnp-Sg were added to cells at an MOI of 100. hGECs without any bacterial stimuli were used as the blank control group (control). After 4 hours, samples were treated with DMEM supplemented with 10% FBS, 200 μg/mL metronidazole, and 300 μg/mL gentamicin for 120 min and fresh DMEM supplemented with 10% FBS was added. The proliferation activity of hGECs was determined by the Cell Counting Kit-8 (CCK8 kit, Dojindo, Japan). After the addition of 10 μL of CCK8 solution to each well, the plate was incubated at 37°C in 5% CO₂ for 1-4 h. The absorbance at 450 nm (OD_450nm) was detected by a microplate reader. The effect of antibiotics alone on the proliferation activity of hGECs was also determined.

hGECs were inoculated into a 6-well plate at 3.5×10⁵ cells per well. Bacterial stimuli were added as described above and hGECs without any bacterial stimuli were used as blank control. After antibiotic treatment, cells were cultured at 37°C in 5% CO₂ and digested with trypsin without EDTA at different time points. Cells were washed with PBS twice and collected in flow cytometry tubes with 1~5×10⁵ cells by centrifugation at 1000 rpm for 5 min. An Annexin V-FITC Apoptosis Kit (BD, USA) was used to detect the apoptosis of hGECs according to the manufacturer’s instructions. After 500 μL of binding buffer was used to resuspend the cells, 5 μL of Annexin V-FITC was added and mixed gently. Samples were placed on ice for 15 min, mixed with 5 μL of propodium iodide (PI) and then detected by Beckman Coulter CytoFLEX immediately using Ex/Em = 488 nm/530 nm. Cells without Annexin V-FITC and PI were used as negative controls.

The expression levels of TLR2 and TLR4 in hGECs were determined by quantitative reverse transcription PCR (RT–qPCR). Total RNA was isolated from hGECs using RNAzol according to the manufacturer’s protocol (Sigma-Aldrich, USA). The concentrations of RNA samples were determined by a NanoDrop 2000C Spectrophotometer (Thermo Fisher, USA). cDNA was synthesized using PrimeScript RT Master Mix (Takara, Japan). RT–qPCR analysis was performed in a 20-μL reaction mixture containing 10-μL of master mix (Hieff qPCR SYBR Green Master Mix, Yeasen) using a Light Cycler 480 (Roche Applied Science, Germany). The reaction product was quantified by the standard curve method. Levels of GAPDH mRNA served as internal controls. The primer sequences were as follows (F/R): TLR2 (ATCAGGCTTCTCTGTCTTGTG/TCTGTAGGTCACTGTTGCTAATG); TLR4 (GGAAGGAGCAGAATCAGGATATG/CTCCATTCACTCCACTAACCAC); and GAPDH (AATCCCATCACCATCTTCCAG/AAATGAGCCCCAGCCTTC).

The hGECs were stimulated with Fnp, Sg, Fnp+Sg and Fnp-Sg as described above. Cell-free supernatants were harvested and stored at -80°C for cytokine assays. Cytokine levels (TNF-α, IL-6, -8, -10 and TGF-β1) in the culture supernatants were measured by ELISA kits (Neobioscience, Shenzhen, China) according to the manufacturer’s instructions.

After removing the supernatant, the cells were washed twice with PBS at 4°C. RIPA lysis buffer (Beyotime, China) containing 1% protease inhibitor (Sigma–Aldrich, USA) and 1% serine protease inhibitor (Sigma–Aldrich, USA) was added for 30 min to lyse cells and extract proteins from each sample. The concentrations of total proteins were detected by a BCA protein assay kit (Beyotime, China) according to the manufacturer’s instructions. Protein samples were mixed with 5× loading buffer (ThermoFisher, USA) at a ratio of 4:1 and boiled at 99°C for 10 min. Samples were loaded and run on SDS–PAGE gels (CWBIO, China) and transferred onto PVDF membranes (Millipore, USA). Membranes were blocked with 2% skim milk (BD, USA) for 1 h at room temperature and then incubated with anti-IKKα, anti-IKKβ, anti-pIKKα/β, anti-p65, anti-pp65, anti-p38, anti-pp38, anti-SAPK/JNK, anti-pSAPK/JNK, and anti-GAPDH primary antibodies (Abcam, UK) overnight at 4°C. After primary incubation, blots were washed and incubated with secondary goat anti-rabbit or goat anti-mouse HRP (Abcam, UK) for 1 hour. Membranes were washed and exposed to chemiluminescent HRP substrate (Millipore, USA). Images were obtained using the GeneGnome XRQ system (Syngene, USA) and analyzed using ImageJ software.

Before bacterial stimuli, hGECs were pretreated the TLR2/4 signaling pathway antagonist OxPAPC (In vivoGen, USA) at 30 μg/mL, 400 nM p38 inhibitor SB 239063 (MCE, USA), 2 μM JNK inhibitor SP600125 (MCE, USA), and 4 μM NF-κB signaling pathway inhibitor BAY 11-7082 (MCE, USA) for 1 h. hGECs treated without any inhibitors and bacterial stimuli were used as blank controls, hGECs treated with inhibitors but without any bacterial stimuli were used as negative controls, and hGECs treated with Fnp-Sg infection but without any inhibitors were used as positive controls. Total proteins were extracted at 24 h and 48 h after incubation. The phosphorylation of NF-κB pathway protein p65, MAPK pathway protein p38 and JNK was detected by western blot as described above. The supernatant was collected and used to detect the changes in the secretion of inflammatory cytokines by ELISA.

The bacterial counts data were log (10) transformed for subsequent analysis. All data were presented as the mean ± standard deviation and assessed for normality by Kolmogorov-Smirnov test. The results showed that the data fitted a normal distribution. Differences between two groups were analyzed by Student’s t tests. Differences in the quantitative data between multiple groups were evaluated by one-way ANOVA combined with Bonferroni’s post hoc test. P values less than 0.05 were designated as significant differences. Statistical analyses were conducted by SPSS Statistics v.20 software (IBM, Inc., Chicago, IL, USA) and GraphPad Prism 9 software (GraphPad Software, Inc., San Diego, CA, USA).

Studies have confirmed that different bacterial strains coaggregate adequately with each other in CAB (Kaplan et al., 2009; Kaplan et al., 2014). Saliva was also used as a coaggregation buffer, of which the composition was complex and included various enzymes, immunoglobulins, and mucins (Heller et al., 2017; Carpenter, 2020). In the present study, F. nucleatum subsp. polymorphum coaggregated strongly with S. gordonii in 10 min with large numbers of coaggregation clumps formed at the bottom of the centrifuge tube and a clear upper suspension. The coaggregation index (C.I.) was 89.370% ± 3.269% (mean ± standard deviation). There was no significant difference between coaggregation indices in CAB and different concentrations of artificial saliva ( Figure 1A ). The coaggregation was stable in 10-90 min with a range of coaggregation indices from 89.370% ± 3.269% to 94.450% ± 1.161% ( Figures 1B, C ). Thus, CAB was used in the present study to exclude the influence of saliva components on the results. Autoaggregation is the adhesion of bacteria of the same strain, which is common with oral bacteria (Khemaleelakul et al., 2006; Merritt et al., 2009). It is mediated by autoagglutinins which is related to surface proteins in general or related to carbohydrates, particularly exopolysaccharides in some cases (Trunk et al., 2018; Yakovlieva and Walvoort, 2020). Although F. nucleatum had been extensively studied on coaggregation, little was known about its autoaggregation. Previous studies showed the autoaggregation of F. nucleatum was strain-dependent and occurs via both saliva-dependent and -independent mechanisms (Merritt et al., 2009; Karched et al., 2015). There were few studies focused on the autoaggregation of S. gordonii. A previous study showed no autoaggregation of S. gordonii (Levin-Sparenberg et al., 2016), while another study showed an increased autoaggregation of S. gordonii depending on the concentration of composite resin containing surface reaction-type pre-reacted glass ionomer eluate used (Shimazu et al., 2016). However, the specific mechanisms of autoaggregation in F. nucleatum and S. gordonii still needed further investigations. In the present study, the autoaggregation of F. nucleatum subsp. polymorphum significantly increased at 20 min ( Figure 1D ) which was consistent with a previous study (Karched et al., 2015). We chose 10 min as the coaggregation time in the present study when there was little autoaggregation of F. nucleatum subsp. polymorphum and S. gordonii ( Figures 1E, F ).

CLSM images at low magnification showed coaggregation between F. nucleatum subsp. polymorphum and S. gordonii in the form of a large number of clumps (white arrows), while the coculture F. nucleatum subsp. polymorphum and S. gordonii were noncoaggregated and distributed separately ( Figure 2A ). Under high magnification, the coaggregation group showed that F. nucleatum subsp. polymorphum and S. gordonii cells adhered to each other tightly and were distributed quite evenly throughout coaggregates ( Figure 2B ).

The results showed that the number of extracellular bacteria decreased with prolonged antibiotic treatment time (** p <0.01, *** p <0.001). After 120 min, no visible bacterial colonies grew, indicating the complete killing of extracellular bacteria among all the groups with bacterial stimuli ( Figure S1A ). The CCK-8 results showed that antibiotic treatment had no significant effect on the proliferation of hGECs ( Figure S1B ). Therefore, the extracellular bacteria were killed by antibiotic treatment for 120 min in the present study.

As the CLSM images showed, cell cytoskeleton was stained red with phalloidin and the nucleus was stained blue with DAPI ( Figure 3A ). There was a large number of F. nucleatum subsp. polymorphum infected- hGECs ( Figure 3B ), while S. gordonii hardly infected hGECs ( Figure 3C ). Compared with the coinfection group ( Figure 3D ), the number of F. nucleatum subsp. polymorphum and S. gordonii that infected hGECs seemed reduced in the coaggregation group ( Figure 3E ).

To further explore the infection of F. nucleatum subsp. polymorphum and S. gordonii to hGECs, we quantified the bacteria that adhered to and invaded hGECs by serial dilution and plating, respectively. The results showed that coaggregation significantly inhibited F. nucleatum subsp. polymorphum adhesion and invasion of hGECs (** p <0.01, *** p <0.001) ( Figures 4A, C ). For S. gordonii, coaggregation and coinfection both enhanced the adhesion of hGECs, with the coinfection group showing a stronger effect than the coaggregation group (* p <0.05, ** p <0.01, *** p <0.001) ( Figure 4B ). Nevertheless, the invasion ability of S. gordonii in hGECs was weak among all groups and neither coinfection nor coaggregation influenced the invasion ability of S. gordonii ( Figure 4D ).

The results showed that the monocultures, coinfection, and coaggregation of F. nucleatum subsp. polymorphum and S. gordonii had no significant effect on the proliferation activity of hGECs after 24 h of infection ( Figure 5A ). After 48 h of infection, the proliferation activity of hGECs was significantly reduced in both the coinfection and coaggregation groups with no significant difference between the two groups (* p <0.05). The monoculture of S. gordonii inhibited the proliferation activity of hGECs after 72 h of infection (* p <0.05), while F. nucleatum subsp. polymorphum monoculture had no significant influence on the proliferation activity of hGECs at various time points.

As shown in Figure 5B , F. nucleatum subsp. polymorphum showed a relatively weak ability to promote cell apoptosis after 2 h of infection and sustained through 24 h. After 12 h of infection, the coaggregation and coinfection of F. nucleatum subsp. polymorphum and S. gordonii significantly promoted hGECs apoptosis compared with the other groups (^### p <0.001, ** p <0.01, *** p <0.001, ^#compared with the control group). There was no significant difference between the coinfection and coaggregation groups within 12 h of infection, however, the coinfection group significantly promoted cell apoptosis at 24 h compared with the coaggregation group (*** p <0.001). After 24 h of infection, the ability of S. gordonii to promote cell apoptosis was enhanced and was the strongest after 48 h. The images of flow cytometry were shown in Figure S2 .

RT–qPCR was used to detect the effect of coaggregation between F. nucleatum subsp. polymorphum and S. gordonii on the TLR2 and TLR4 mRNA expression levels in hGECs ( Figure 6 ). The results showed that after 6 h of infection, the TLR2 and TLR4 mRNA expression levels were significantly increased in the S. gordonii monoculture, coinfection, and coaggregation groups compared with the control group (^# p < 0.05, ^### p < 0.001). After 24 h of infection, the TLR2 and TLR4 mRNA expression levels were significantly decreased in the coaggregation group compared with the coinfection group (***p <0.001). In the present study, S. gordonii monoculture increased both TLR2 and TLR4 mRNA expression levels in hGECs, while F. nucleatum subsp. polymorphum monoculture had no significant influence on the expression levels.

ELISA results showed that the secretion level of TNF-α by hGECs in the coaggregation group was significantly higher than that in other groups at 24 h of infection (***p <0.001, ^### p <0.001, ^#compared with the control group), while no difference was found between groups at 0.5 h, 2 h, and 6 h. Afterwards, the secretion level of TNF-α in the coaggregation group decreased with no difference compared with the coinfection group ( Figure 7A ). The secretion level of IL-6 steadily reached the highest in the coaggregation group at 24 h of infection, showing a significantly higher level than that in other groups (***p <0.001, ^### p <0.001, ^#compared with the control group). The secretion level of IL-6 in the coinfection group increased within 6 h and decreased at 24 h of infection. However, the secretion level of IL-6 in the coinfection group restored and was dramatically higher than that in the coaggregation group at 48 h of infection ( Figure 7B ). As for IL-8, F. nucleatum subsp. polymorphum monoculture and coinfection groups resulted in high secretion levels at 24 h of infection and then decreased. The coaggregation group did not significantly promote IL-8 secretion at various time points ( Figure 7C ). The secretion levels of TGF-β1 were higher in the control and F. nucleatum subsp. polymorphum monoculture groups than that in other groups at 48 h and 72 h of infection (***p <0.001, ^### p <0.001, ^#compared with the control group). However, the secretion level of TGF-β1 in the coinfection group increased significantly at 24 h with no difference compared with the control and F. nucleatum subsp. polymorphum monoculture groups and decreased afterwards ( Figure 7E ). No significant difference was found in the secretion of IL-1β and IL-10 among the groups ( Figures 7D, F ).

The western blot and semiquantitative analysis results showed that the phosphorylation level of p65 (p-p65) protein in the coaggregation group was higher than that of the other groups after 0.5 h and 2 h of infection (* p <0.05, ** p <0.01, *** p <0.001, ^#p <0.05, ^###p <0.001, ^#compared with the control group) ( Figure 8 ). Moreover, the phosphorylation level of p38 (p-p38) protein in the coaggregation group was higher than that of the other groups after 6 h and 12 h of infection (* p <0.05, ** p <0.01, *** p <0.001, ^#p <0.05, ^##p <0.01, ^###p <0.001, ^#compared with the control group) ( Figure 9 ). After 12 h of infection, the phosphorylation level of JNK (p-JNK) protein in the coaggregation group was higher than that in the other groups (* p <0.05, ** p <0.01, *** p <0.001, ^#p <0.05, ^##p <0.01, ^###p <0.001, ^#compared with the control group) ( Figure 9 ).

By pretreatment of hGECs with the TLR2/4 antagonist OxPAPC for 1 h, the protein expression of p-p65, p-p38, and p-JNK in the coaggregation group was decreased ( Figures 10A, B ). After pretreatment with the NF-κB inhibitor BAY 11-7082, the protein expression of p-p65 decreased significantly ( Figure 10C ). Pretreatment with the p38 MAPK inhibitor SB 239063 decreased the p-p38 protein level; thus, the p-JNK protein level showed a compensatory increase ( Figure 10D ). Pretreatment with the JNK MAPK inhibitor SP 600125 significantly decreased the p-JNK protein level. Because the targets of the inhibitor SP 600125 also included the upstream kinases MKK3, MKK4, and MKK6 in the p38 MAPK signaling pathway, the p-p38 protein level was decreased significantly ( Figure 10E ).

The changes in the secretion of inflammatory cytokines by hGECs were also evaluated by ELISA. By pretreatment of hGECs with antagonist or inhibitors, the levels of TNF-α and IL-6 decreased in the coaggregation group after 24 h and 48 h of infection (** p < 0.01, *** p < 0.001) ( Figures 11A, B, D, E ), but no significant difference was detected in the secretion of TGF-β1 ( Figures 11C, F ).