Identification and characterization of a novel Fusobacterium nucleatum adhesin involved in physical interaction and biofilm formation with Streptococcus gordonii

Abstract To successfully colonize the oral cavity, bacteria must directly or indirectly adhere to available oral surfaces. Fusobacterium nucleatum plays an important role in oral biofilm community development due to its broad adherence abilities, serving as a bridge between members of the oral biofilm that cannot directly bind to each other. In our efforts to characterize the molecular mechanisms utilized by F. nucleatum to physically bind to key members of the oral community, we investigated the involvement of F. nucleatum outer membrane proteins in its ability to bind to the pioneer biofilm colonizer, Streptococcus gordonii. Here, we present evidence that in addition to the previously characterized fusobacterial adhesin RadD, the interaction between F. nucleatum ATCC 23726 and S. gordonii V288 involves a second outer membrane protein, which we named coaggregation mediating protein A (CmpA). We also characterized the role of CmpA in dual‐species biofilm formation with S. gordonii V288, evaluated growth‐phase‐dependent as well as biofilm expression profiles of radD and cmpA, and confirmed an important role for CmpA, especially under biofilm growth conditions. Our findings underscore the complex set of specific interactions involved in physical binding and thus community integration of interacting bacterial species. This complex set of interactions could have critical implications for the formation and maturation of the oral biofilms in vivo, and could provide clues to the mechanism behind the distribution of organisms inside the human oral cavity.

species present in the oral cavity, these communities can cause a multitude of oral and systemic diseases including dental caries, periodontal disease, and many others (Zarco, Vess, & Ginsburg, 2012).
To form these multispecies communities, oral bacteria must first, directly or indirectly, attach to a surface in the oral cavity. Oral streptococci are among the first species to attach to the surface of the teeth and they comprise the majority of the early colonizers (Avila, Ojcius, & Yilmaz, 2009;Diaz et al., 2006;Dige, Nilsson, Kilian, & Nyvad, 2007;Dige, Nyengaard, Kilian, & Nyvad, 2009;Nyvad & Kilian, 1987, 1990. Once established, the early colonizers alter their microenvironment and serve as anchors for subsequent colonizers of the dental plaque. The Gram-negative bacterium Fusobacterium nucleatum is considered an important species in the development and maturation process of dental plaque (Jorth et al., 2014;Kolenbrander & London, 1993) as it contributes to important structural and metabolic changes. Structurally, F. nucleatum binds to numerous species in the oral cavity, serving as a bridge between early-and late-colonizing species (Guo, He, & Shi, 2014;Kolenbrander, Andersen, & Moore, 1989;Kolenbrander, Parrish, Andersen, & Greenberg, 1995). Metabolically, F. nucleatum is a key contributor to butyrate production (Jorth et al., 2014), which has been linked to the development of periodontal disease (Niederman, Buyle-Bodin, Lu, Robinson, & Naleway, 1997).
Earlier studies have focused on the identification and characterization of molecular components required for the direct cell-to-cell interaction among members of the oral community. Of special interest to our laboratory is the characterization of the interactions between F. nucleatum and other members of the dental plaque community. To date, only two F. nucleatum adhesins have been characterized for their role in interspecies interaction: Fap2 and RadD. Fap2 is a galactose-inhibitable adhesin, which has been implicated in the interaction between F. nucleatum and the periodontal pathogen Porphyromonas gingivalis (Coppenhagen-Glazer et al., 2015). RadD is an arginine-inhibitable adhesin required for the interaction between F. nucleatum and multiple Gram-positive members of the dental plaque, including the early colonizers Actinomyces naeslundii, Streptococcus sanguinis, Streptococcus oralis, and Streptococcus gordonii (Kaplan, Lux, Haake, & Shi, 2009).
In the work described here, we report the identification of a previously uncharacterized adhesin, which we named coaggregation mediating protein A (CmpA), involved in the interaction between F. nucleatum ATCC strain 23726 and S. gordonii V288. Along with RadD, CmpA plays an important role in the ability of F. nucleatum to coaggregate and form dual-species biofilms with S. gordonii V288.

| Bacteria and culture conditions
All bacterial strains and plasmids used in this study are listed in Table 1. Unless otherwise stated, F. nucleatum strains were grown in Columbia broth or on Columbia agar plates (BD Difco, Detroit, MI) supplemented with 5% defibrinated sheep blood (Colorado Serum Company, Denver, CO) under anaerobic conditions (5% H 2 , 5% CO 2 , 90% N 2 ) at 37°C. When necessary, Thiamphenicol and Clindamycin (MP Biomedicals, Irvine, CA) at 5 μg/ml and 0.2 μg/ml, respectively, were added to the media. For P. gingivalis growth, Columbia broth was supplemented with hemin and menadione at 5 μg/ml and 1 μg/ml, respectively. Columbia agar plates were supplemented with 5% defibrinated sheep blood. S. sanguinis and S. gordonii were grown in Todd Hewitt (TH) broth or agar plates (BD Difco, Detroit MI) at 37°C under anaerobic conditions. Streptococcus gordonii selection was carried out with 5 μg/ml erythromycin added to the media. Escherichia coli was grown aerobically at 37°C in Luria-Bertani (LB) broth or agar plates (BD Difco, Detroit, MI). Escherichia coli selection was carried out with 100 μg/ml erythromycin or ampicillin added to the media.

| Coaggregation assay
Coaggregations were performed in coaggregation buffer (CAB) (150 mmol/L NaCl, 1 mmol/L Tris, 0.1 mmol/L CaCl 2 , 0.1 mmol/L MgCl 2 ) as previously described with minor modifications (Kaplan et al., 2009). Overnight cultures of S. gordonii and F. nucleatum were diluted 10-fold into fresh medium in the morning and grown until they reached OD 600 1.5 for S. gordonii and OD 600 2.0 for F. nucleatum. Bacterial cells were washed with CAB and resuspended, also with CAB. Equal numbers of bacterial cells were combined to a final concentration of 2 × 10 9 cells ml −1 in a 1.5 ml microcentrifuge tube. To account for autoaggregation, the cells were also incubated in the absence of the binding partner.
The tubes were vortexed for 10 s and incubated for 10 min at room temperature. After incubation, the bacterial mixtures were centrifuged at low speed (100g) for 1 min to pellet cellular aggregates, while leaving the nonaggregated bacteria in suspension. The supernatant was then removed without disturbing the pellet, and the optical density of the recovered supernatant was measured at 600 nm. For coaggregation inhibition assays, 50 mmol/L ofarginine was added to the reaction tube containing the different F. nucleatum strains and vortexed before addition of the partner strain. Relative coaggregation was determined by subtracting the turbidity of the recovered supernatant after coaggregation from the turbidity of the cell mixture before coaggregation and dividing the results by the turbidity before coaggregation.
Mutants were also confirmed by PCR analysis.

| mCherry + S. gordonii
To construct the mCherry-expressing S. gordonii strain BL98, we transformed plasmid pVA8912 (Vickerman, Mansfield, Zhu, Walters, & Banas, 2015) into the wild-type S. gordonii V288 according to previously published protocol (Warren, Lund, Jones, & Hruby, 2007), utilizing competence-stimulating peptide N-DVRSNKIRLWWENIFFNKK (Pepmic, Suzhou, China). The mCherry encoding gene was inserted into S. gordonii attB site and is expressed under the control of the ldh promoter (for a full description of the construct, see Vickerman et al., 2015). The mCherry expression had no effect on coaggregation or dual-species biofilm formation with F. nucleatum 23726.

| Biofilm growth
One milliliter of SHI-FSMS (50% SHI medium, Tian et al., 2010, 25% filtered saliva, 0.5% mannose, 0.5% sucrose; de Avila et al., 2015) containing 1 × 10 8 F. nucleatum cells and 5 × 10 4 S. gordonii cells diluted from overnight cultures were added to the wells of a 24-well polystyrene culture plates (Thermo Fisher Scientific, Waltham, MA, USA) and incubated overnight under anaerobic conditions (5% H 2 , 5% CO 2 , 90% N 2 ) at 37°C. After overnight growth, the planktonic cells were removed and the biofilm was washed three times with 500 μl of prereduced, sterile phosphate-buffered saline (PBS). The biofilm that remained attached to the wells was either processed for confocal laser scanning microscopy (CLSM) analysis or collected for DNA isolation.

| Nucleic acid isolation
Genomic DNA was extracted from biofilms using MasterPure™ DNA Purification Kit (Epicenter ® , Madison, WI, USA) according to

| qPCR
To quantify the relative proportions of each species in the respective dual-species biofilms, previously designed species-specific primer pairs were used (Park, Shokeen, Haake, & Lux, 2016). For Scientific, Waltham, MA). Each PCR run was carried out with an initial incubation of 10 min at 95°C followed by 40 cycles of denaturing at 95°C for 15 s and annealing and elongation at 60°C for 1 min.
After the 40 cycles of amplification, an additional denaturing step was performed at 95°C for 1 min followed by annealing and elongation at 60°C for 1 min. A melting curve analysis was completed after each run. In addition, gel electrophoresis was utilized during optimization step to determine size and number of amplicons. The DNA concentrations (ng ml −1 ) were calculated with standard curves obtained by 10-fold serial dilutions of previously purified and quantified bacterial genomic DNA. Three independent qPCR runs were performed with three technical replicates for each sample to assess reproducibility and inter-run variability.

| Transcriptional analysis
To determine radD and cmpA expression pattern, 1 μg of total RNA was used for cDNA synthesis using SuperScript ® III First-Strand

| Statistical analysis
Student's t-test was performed to determine statistical significance using Excel 2010 (Microsoft, Seattle, WA, USA).

| Multiple F. nucleatum adhesins are involved in its coaggregation with S. gordonii
For detailed characterization of the physical interaction between F. nucleatum and S. gordonii, we collected F. nucleatum cells, ATCC strain 23726, that were in stationary phase, after overnight growth, and cells that were in late exponential phase to perform coaggregation experiments with midexponential phase S. gordonii V288 cells.
We observed less coaggregation with stationary phase F. nucleatum cultures (63.03% ± 2.39) when compared to those in late exponential phase (80.85% ± 0.5) (Figure 1a). Surprisingly, the coaggregation defect usually observed for the radD mutant derivative of F. nucleatum was less evident at the late exponential phase, compared to F. nucleatum collected the stationary growth phase (Figure 1a). This phenotype was not evident among other commonly used S. gordonii strains (DL1, ATCC 10558, and ATCC 51656) ( Figure 1b). Thus, F. nucleatum coaggregation with S. gordonii V288 might require an additional, and yet uncharacterized, surface adhesin that is likely expressed as cells enter the exponential phase of growth.

| Identification of an additional adhesin involved in the interaction between F. nucleatum and S. gordonii
The For further characterization of these OMPs in binding to S. gordonii V288, we constructed a double radD cmpA mutant, by using previously established approaches (Haake et al., 2000;Kaplan et al., 2005Kaplan et al., , 2009) to inactivate the ORF encoding cmpA in the radD mutant background (Figure 3a and b). Consistent with the idea that cmpA and radD function independently from each and are the two major adhesins involved in the F. nucleatum 23726 interaction with S. gordonii V288, the double mutant had a stronger coaggregation defect than either single mutant, and the defect in coaggregation was similar to that observed when the coaggregation inhibitor arginine was added to the buffer (Figure 4a).

| CmpA coaggregation involvement is S. gordonii strain dependent
In contrast to above findings with S. gordonii V288, we had previously failed to observe a CmpA involvement in the interaction between We also investigated if CmpA was involved in F. nucleatum interaction with the periodontal pathogen P. gingivalis (strain 4612) and with another streptococcal species closely related to S. gordonii and S. sanguinis (ATCC 10556). However, we did not observe any difference in coaggregation compared to the wild-type strains (data not shown).
Thus, to the extent that we have tested, CmpA seems to be largely involved in the specific interaction between F. nucleatum strain 23726 and S. gordonii V288.

| CmpA is required for dual-species biofilm formation with S. gordonii
The interaction between streptococcal species and F. nucleatum is hypothesized to be an important step in oral biofilm development (Kolenbrander, 1989;Kolenbrander & London, 1993). Previous studies demonstrate that F. nucleatum requires radD to form a dual-species biofilm with at least one of the oral streptococci species, S. sanguinis (Kaplan et al., 2009;Lancy, Dirienzo, Appelbaum, Rosan, & Holt, 1983). Since our data demonstrate that both RadD and CmpA are involved in the physical binding between F. nucleatum and S. gordonii V288, we investigated whether these adhesins were also involved in dual-species biofilm formation. To differentiate between S. gordonii

| radD and cmpA expression pattern
To further characterize radD and cmpA, we measured their mRNA level in wild-type F. nucleatum throughout planktonic growth in Columbia broth from the early exponential to the stationary phase as well as from a single time point from the overnight biofilm. Both radD and cmpA mRNA level increased as the cultures entered stationary phase, but quickly decreased in the following time points (Figure 6a and b) implicating a growth-dependent regulation of adhesins expression.
Under overnight biofilm conditions, radD expression was about threefold lower while cmpA expression was about threefold higher compared to the planktonic cells collected from the same wells ( Figure 6c).

| DISCUSSION
The organization of the oral microbial community is thought to involve a complex network of interactions, often mediated by sur-  (Kaplan et al., 2009). Most significantly, these data add an additional layer of complexity to the interaction between F. nucleatum and S. gordonii. Similar complexity seems to be present in the interaction between F. nucleatum and P. gingivalis; while Fap2 appears to be a major adhesin for the interaction of F. nucleatum with P. gingivalis (Coppenhagen-Glazer et al., 2015;Park et al., 2016), RadD plays an additional role in binding to strain 4612 (Park et al., 2016).  (Park et al., 2016).
While it is widely accepted that the initial interaction between oral bacteria can be assessed in vitro by measuring the ability of their planktonic cells to coaggregate, in the oral cavity, the ability of individual cells or group of cells to integrate into a biofilm is crucial for their survival and maintenance in the oral cavity (Kolenbrander, Palmer, Periasamy, & Jakubovics, 2010). Cells that integrate into a biofilm undergo significant physiological changes compared to their planktonic counterparts. These changes include, but are not limited to (1) gene and protein expression patterns, (2) metabolic preferences, and (3) replication rates (Cook, Costerton, & Lamont, 1998;Resch, Rosenstein, Nerz, & Gotz, 2005).
Our results demonstrate that during formation of dual-species biofilm with S. gordonii V288 in vitro, both RadD and CmpA are key players (Figure 5a and b). Interestingly, while cmpA expression was increased under biofilm conditions, radD expression was decreased (Figure 6c), suggesting that these proteins are likely to be involved in different physiological processes under biofilm conditions. Perhaps, RadD could play a more important role in the initial binding, while CmpA could be more involved in the subsequent stages of biofilm development.
Fusobacterium nucleatum encodes at least eight autotransporterlike OMPs with molecular weights greater than 200 kDa. RadD, Fap2, and Aim1 have been previously characterized as interspecies adhesins (Coppenhagen-Glazer et al., 2015;Kaplan et al., 2009) and apoptosisinducing proteins (Kaplan et al., 2005). The involvement of CmpA in interspecies interaction leaves five OMPs to be characterized. These autotransporter-like OMPs possess very similar characteristics and are predicted to contain a core β-barrel structure (Kaplan et al., 2009). In other bacteria, this structure possesses multiple activities that function in adherence and biofilm formation (Charbonneau & Mourez, 2007;Korotkova et al., 2006;Laarmann, Cutter, Juehne, Barenkamp, & ST Geme, 2002). The wide array of adherence properties found in F. nucleatum strains could be mediated by a combination of OMPs with varying degrees of affinity for different species and/or strains of bacteria present in the oral cavity.
In summary, the data presented here support the existence of a potentially complex interaction network between F. nucleatum and S. gordonii, which seems to be mediated by varying degrees of preferences for different F. nucleatum adhesins. This variation in adhesin preference could have a profound impact on community composition and species distribution in the oral microbiome, especially if such phenotype is also observed in the interaction among other members of the oral microbial community.
F I G U R E 6 radD and cmpA expression: WT Fusobacterium nucleatum was grown in Columbia broth for 25 hr. Cell samples were collected every 3 hr and (a) OD600 was measured, as well as (b) radD and cmpA expression by qRT-PCR. Gene expression was normalized to rpoB and compare to the first time point. The dashed line was added to aid in the comparison. (c) radD and cmpA expression were also measured in cells from Fn biofilm grown overnight in SHI-FSMS. Gene expression was normalized to rpoB and compare to planktonic cells grown under the same conditions. Each value represents means and standard deviation of at least three independent experiments. To aid with visualization, the dashed line represents no change