Identification of functional domains of the minor fimbrial antigen involved in the interaction of Porphyromonas gingivalis with oral streptococci

Abstract Porphyromonas gingivalis is associated with chronic periodontitis and may initially colonize the oral cavity by adhering to streptococci. Adhesion to streptococci is driven by interaction of the minor fimbrial antigen (Mfa1) with streptococcal antigen I/II. We identified the region of antigen I/II required for this interaction and developed small molecule mimetics that inhibited P. gingivalis adherence. However, the functional motifs of Mfa1 involved in the interaction with antigen I/II remain uncharacterized. A series of N‐ and C‐terminal peptide fragments of Mfa1 were expressed and tested for inhibition of P. gingivalis adherence to S. gordonii. This approach identified residues 225–400 of Mfa1 as essential for P. gingivalis adherence. Using the three‐dimensional structure of Mfa1, a putative binding cleft was identified using SiteMap and five small molecule mimetics could dock in this site. Site‐specific mutation of residues in the predicted cleft, including R240A, W275A, D321A and A357P inhibited the interaction of Mfa1 with streptococci, whereas mutation of residues not in the predicted cleft (V238A, I252F and ΔK253) had no effect. Complementation of an Mfa1‐deficient P. gingivalis strain with wild‐type mfa1 restored adherence to streptococci, whereas complementation with full‐length mfa1 containing the R240A or A357P mutations did not restore adherence. The mutations did not affect polymerization of Mfa1, suggesting that the complemented strains produced intact minor fimbriae. These results identified specific residues and structural motifs required for the Mfa1‐antigen I/II interaction and will facilitate the design of small molecule therapeutics to prevent P. gingivalis colonization of the oral cavity.

gingivalis is strongly associated with chronic adult periodontitis and is an important pathogen that is capable of modulating the host immune response and disrupting normal host/microbe homeostasis (Hajishengallis, 2015;Olsen, Lambris, & Hajishengallis, 2017).
This can lead to the development of a dysbiotic microbial community which can induce uncontrolled inflammation leading to the destruction of tooth supporting tissues, and ultimately tooth loss (Hajishengallis & Lamont, 2014Lamont & Hajishengallis, 2015). Periodontitis is also associated with increased risk of other systemic diseases such as rheumatoid arthritis, cardiovascular disease, some cancers and chronic respiratory disease (Bingham & Moni, 2013;Kim & Amar, 2006;Winning & Linden, 2017).
The primary niche for P. gingivalis is the subgingival pocket but the organism also adheres efficiently to supragingival bacteria such as various commensal streptococci (Brooks, Demuth, Gil, & Lamont, 1997;Demuth, Irvine, Costerton, Cook, & Lamont, 2001;Lamont, Hersey, & Rosan, 1992). Indeed, adherence to streptococci can modulate the pathogenic potential of P. gingivalis (Daep, Novak, Lamont, & Demuth, 2011;Kuboniwa et al., 2017;Kuboniwa & Lamont, 2010) and may also be important for the initial colonization of the oral cavity by the organism. Initial colonization of the oral cavity by P. gingivalis is thought to occur at more available sites such as the supragingival tooth surface (Quirynen et al., 2005;Socransky, Haffajee, Ximenez-Fyvie, Feres, & Mager, 1999;Takazoe, Nakamura, & Okuda, 1984) and oral introduction of P. gingivalis in human volunteers results in the organism locating almost exclusively on streptococcal-rich supragingival plaque (Slots & Gibbons, 1978). In addition, in patients with periodontal disease, the levels of supragingival P.
gingivalis have been shown to correlate with subgingival levels of the organism (Mayanagi, Sato, Shimauchi, & Takahashi, 2004). Thus, adherence of P. gingivalis to streptococci represents a viable target for therapeutic intervention.
Deap et al. identified several discrete structural motifs in SspB that are essential for adherence and suggested that this functional region resembles the eukaryotic nuclear receptor (NR) box protein-protein interaction domain (Daep, Lamont, & Demuth, 2008). In addition, a synthetic peptide (BAR) that encompasses this region potently inhibited P. gingivalis/streptococcal adherence in vitro and significantly reduces P. gingivalis virulence in vivo (Daep et al., 2011). Subsequently, small molecule BAR peptidomimetics that potently inhibit P. gingivalis adherence were developed (Patil, Luzzio, & Demuth, 2015;Patil, Tan, Demuth, & Luzzio, 2016). Although the binding region in antigen I/II has been well characterized, little is known about the binding domains or motifs of Mfa1 that contribute to this protein-protein interaction.
In this study, N-and C-terminal truncated Mfa1 polypeptides were shown to inhibit P. gingivalis/streptococcal adherence and suggested that Mfa1 functional motifs are present between residues 225-400 of the protein. Using the three-dimensional structure of Mfa1 (Hall, Hasegawa, Yoshimura, & Persson, 2018), a putative binding cleft was identified using the prediction tool SiteMap. Sitespecific mutation of amino acids in the predicted cleft, for example, R240A, W275A, D321A and A357P inhibited the interaction of Mfa1 with streptococci. Finally, complementation of an Mfa1-deficient P.
gingivalis strain with wild-type mfa1 restored adherence to streptococci, whereas complementation with the site-specific mfa1 mutants resulted in significantly reduced levels of adherence. Together, these results identify specific residues and motifs that are important for the Mfa1/SspB protein-protein interaction.

| Bacterial strains and growth conditions
The strains and plasmids used in this study are shown in Table 1.
P. gingivalis ATCC 33277 was grown in TSBY medium comprised of trypticase soy broth (Difco) supplemented with 2% yeast extract, 1 µg/ml hemin and 5 µg/ml menadione. For growth on plates, this medium was further supplemented with 1.5% agar and 5% sheep blood. The Mfa1-deficient and complemented strains were cultured in medium containing the appropriate antibiotics, that is, 1 and 5 µg/ ml of tetracycline and erythromycin, respectively. All P. gingivalis cultures were incubated under anaerobic conditions (10% CO 2 , 10% H 2 and 80% N 2 ). Brain heart infusion agar (Difco) supplemented with 5% yeast extract was used to grow S. gordonii DL-1. E. coli strains were maintained in LB medium supplemented with the appropriate antibiotic. Where necessary, the final concentration of ampicillin was 100 µg/ml. All bacterial stocks were stored at −80°C in the appropriate medium supplemented with 30% glycerol.

| Recombinant protein/peptide constructs
Nucleotide primers used in this study are shown in Table 2. To generate the full-length Mfa1 construct lacking the signal sequence, the mfa1 sequence from 61 bp to 1689 bp was amplified using the forward and reverse primers listed in Table 2 21-194, 21-225, 21-279, 21-400 and a C-terminal fragment encoding residues 280-563 of Mfa1 were amplified using the primers shown in Table 2 from P. gingivalis ATCC33277 genomic DNA. A similar approach to that described above was followed to clone these fragments in pGEX6p1 expression vector and introduce the constructs into E. coli BL21.

Mfa1 peptides
To express the full-length and truncated Mfa1 proteins, an overnight culture containing the desired construct was diluted in prewarmed LB medium supplemented with 100 µg/ml ampicillin to an OD 600 nm of 0.1 and incubated at 37°C in a rotating shaker at speed of 220 rpm. When the OD 600 nm reached 0.5, protein expression was induced by adding IPTG to a final concentration of 1mM. After further incubation for 4 hr at 37°C, cells were harvested by centrifugation at 3,000 g for 10 min and the cell pellets were frozen. One gram of frozen cell pellet was suspended in 5 ml of CellLytic B (Sigma-Aldrich) containing lysozyme (0.2 mg/ml) and Benzonase (50 U/ml).
A protease inhibitor cocktail (Thermo Fisher Scientific) was added as per manufacturer recommendations and incubated at 25°C for 30 min with gentle shaking. To complete the disruption of the cells, brief sonication was carried out using a Vibra-Cell ultrasonic Liquid Processor VCX 130 (Sonics). Cells were pulsed at 20 kHz for 2 min using a 10 s short burst followed by a 30 s cooling interval. During sonication, all steps were carried out in ice. Cell debris was removed by centrifugation at 13,000 g for 20 min and the supernatant was transferred to a fresh tube.
Purification of GST-tagged fusion proteins was carried out using the Pierce GST Spin Purification Kit (ThermoFisher). Briefly, spin columns were equilibrated with the equilibration/wash buffer supplied with the kit at 4°C, and then 15 ml of cell lysate supernatant was added to each column and incubated overnight at 4°C. Columns were washed three times with 15 ml of wash buffer by centrifugation at 700 g for 2 min. In column, cleavage with Precision Protease (ThermoFisher) was carried out by overnight  (Gardner, Russell, Wilson, Wang, & Shoemaker, 1996) pTCOW-Mfa1 pT-COW containing a 2.5-kb fragment containing the upstream and coding region of the mfa1 gene (Park et al., 2005) pTCOW-Mfa1 Arg240/ Ala

| Dual species biofilm model
Interspecies adherence and biofilm formation between P. gingivalis and S. gordonii were carried out essentially as previously described (Patil et al., 2016). S. gordonii cells were harvested from a 10 ml of overnight culture by centrifugation at 3,000 g for 5 min,

| SiteMap prediction of a putative binding cleft in Mfa1
The Mfa1

| Site-directed mutagenesis
Site-specific mutation reactions were carried out using the Quick-

| Complementation of P. gingivalis SMF1 with mutated Mfa1
Complementation of the Mfa1-deficient P. gingivalis strain SMF1 was carried out using a modification of the protocol previously described (Park et al., 2005). P. gingivalis SMF1 was grown on a blood agar plate under anaerobic conditions for 2-3 days and donor E. coli S17-1 was grown aerobically on a LB agar plate. pTCOW-mfa1 containing the desired mutation was electroporated into E.
coli strain S17-1 and subsequently conjugated with P. gingivalis SMF1 using the agar plate method. Briefly, cells from both donor and recipient were scraped from the agar plates and spread on a 4 cm 2 area on a blood agar plate containing no antibiotics. After incubation for 24 hr, mixed cells from blood agar plates were collected and incubated in TSBY supplemented with hemin (1 µl/ml) and menadione (5 µl/ml) for 1 hr at 37°C under anaerobic conditions. Subsequently, 0.1 ml of the cell suspension was plated on blood agar containing 20 µg of erythromycin per ml and 200 µg of gentamicin per ml and was incubated anaerobically at 37°C for 10 days. Transconjugants were subsequently grown in the presence of antibiotics and the purified plasmid was confirmed to possess the desired mutation by DNA sequencing.

| Cell surface expression of mutated Mfa1 polypeptides
Cell surface expression of Mfa1 by the transconjugants was determined using an enzyme-linked immunosorbent assay (ELISA) after adsorption of P. gingivalis strains onto Maxisorp plates (Nunc). Briefly, P. gingivalis cells were centrifuged at 3,000 g for 5 min and cell pellets were washed there times with PBS. Subsequently, 10 7 cells were incubated in each well for 2 hr at 4°C followed by washing with PBS to remove unbound bacteria. Bound cells were incubated with rabbit rMfa monoclonal antibodies (1:5,000 dilution) (Covance, Denver,

| Mfa1 polymerization
Whole cell lysates were prepared using a modification of the procedure previously described by Hasegawa (Hasegawa et al., 2016).
Briefly, P. gingivalis strains were grown until early stationary phase in TSBY media supplemented with hemin (5 µg/ml) and menadione (1 µg/ ml). Following centrifugation at 6,000 g for 5 min, cell pellets were collected and suspended in 1× NuPAGE LDS sample buffer (Thermo Fisher) at the final OD of 2. The cell suspensions were then denatured by incubation either at 60°C or 100°C for 10 min. Following the heat treatment, the whole cell lysate was centrifuged at 20,000 g for 10 min to remove cellular debris, electrophoresed in 12% Bis-Tris Plus gel (Thermo Fisher) and Mfa1 was visualized by ELISA as described above.

| Localization of Mfa1 functional domains
To identify regions of Mfa1 that contribute to the interaction with antigen I/II, a series of N-and C-terminal truncated Mfa1/ GST fusion proteins were constructed and expressed. The truncated Mfa1 polypeptides were purified by removing the GST tag by in column cleavage and were designated as N194, N225, N279, N400 and C280 as shown in Figure 1. The functional activity of these peptides was determined by assessing their ability to inhibit P. gingivalis/S. gordonii adherence and biofilm formation using the dual species biofilm model described previously by Patil (Patil et al., 2016). Representative images of biofilms formed in the presence of each peptide are shown in Figure 2a and inhibition results are summarized in Figure 2b. Peptides N194 and N225 were relatively poor inhibitors of P. gingivalis adherence (~20% inhibition) compared to the full-length Mfa1 protein (80% inhibition). In contrast, peptide N279 exhibited 70% inhibition and adherence inhibition by peptide N400 was similar to that of the full-length Mfa1 protein. Peptide C280 exhibited reduced activity (~40% inhibition) compared to peptides N279 and N400 but was significantly more active than peptides N194 and N225. Together, these results suggest that essential functional residues that contribute to P. gingivalis adherence to streptococci reside in the region of Mfa1 comprising amino acids 225-400.

| In silico prediction of a putative Ag I/II binding cleft in Mfa1
To further highlight the functional region(s) of Mfa1, we took advantage of the recently published three-dimensional structure of Mfa1 (Hall et al., 2018) to predict putative binding clefts using SiteMap. In addition, a series of in silico docking experiments were conducted using five peptidomimetic compounds that mimic the BAR peptide and were previously shown by Patil et al. (Patil et al., 2015 to be potent competitive inhibitors of P. gingivalis/S. gordonii adherence. As shown in Figure 3a, all five of the mimetic compounds could be docked in the putative binding cleft that exhibited the highest sitescore by SiteMap. Amino acids of the Mfa1 protein that comprise the putative binding cleft are highlighted in red and underlined in Figure 3b.
To validate the predicted binding cleft, a series of Mfa1 site-specific mutant peptides that targeted residues and putative motifs predicted in Figure 3b to comprise the binding cleft were constructed and expressed. Since the results of the truncated Mfa1 peptides in Figure 2b indicated that the region comprising residues 226-279 was important for P. gingivalis adherence, site-specific mutations R240A and W275A were introduced into peptide N279 since both of these residues are predicted by SiteMap to be part of the binding cleft. Additional mutations, D321A and A357P, were also constructed in peptide N400 to disrupt two predicted amphipathic helices in the putative binding cleft (residues 321-329 and 351-364). Finally, several additional residues in peptide N279 that were not predicted to comprise the binding cleft were tested (e.g., V238A, I252F and ΔK253). As shown in Figure 4, peptides N279 and N400 inhibited P. gingivalis adherence to streptococci by 66% and 79%, respectively, consistent with the results shown in Figure 2b. Polypeptide N279 containing the R240A or F I G U R E 1 Schematic representation of the series of Mfa1 peptide fragments. The full-length Mfa1 lacking the signal peptide (21-563aa residues), and N-terminal peptide fragments N194, N225, N279 and N400 encoding residues 21-194, 21-225, 21-279 and 21-400, respectively, are shown. The C-terminal peptide fragment, C280, is comprised of Mfa1 residues 280-563 N275A mutations were significantly less potent inhibitors relative to the parent N279 peptide, exhibiting only 32% and 38% inhibition of P.
gingivalis adherence, respectively. In addition, peptide N279 containing both mutations, R240A and W275A exhibited significantly lower inhibitory activity than either of the peptide fragments containing a single mutation. In contrast, peptide N279 containing the mutations V238A, I252F or ΔK253 showed no significant reductions in inhibitory activity. Furthermore, mutations D321A and A357P, intended to disrupt the two putative helices, also reduced inhibitory activity relative to the parent N400 peptide (79% to 59% and 79% to 38%, respectively).
Together, these results provide preliminary validation of the binding cleft predicted by SiteMap and identify specific Mfa1 residues that contribute to adherence of P. gingivalis to streptococci.

| Complementation of Mfa1-deficient P. gingivalis with site-specific Mfa1 mutants
To further confirm the functional roles for R240 and A357, full-length Mfa1 polypeptides containing the R240A and A357P mutations were constructed and introduced into the Mfa1-deficient strain P. gingivalis SMF1. As shown in Figure 5a, cell surface expression of Mfa1 was significantly reduced in P. gingivalis SMF1 compared to the wild-type strain, P. gingivalis 33277. Complementation of P. gingivalis SMF1 with wild-type mfa1 or with the site-specific mutants restored cell surface expression of Mfa1 to wild-type levels (Figure 5a). Consistent with its level of cell surface expression, adherence of P. gingivalis SMF1 to streptococci was significantly reduced relative to the parent strain 33277 but was restored to wild-type levels by complementation with full-length mfa1, as shown in Figure 5b. In contrast, although complementation with Mfa1 containing either the R240A or A357P mutations restored cell surface expression, both of these complemented strains showed significantly reduced levels of adherence to streptococci.

| D ISCUSS I ON
Heterotypic community formation of P. gingivalis with oral streptococci is driven by a protein-protein interaction between the minor fimbrial antigen (Mfa1) of P. gingivalis and streptococcal surface antigen I/II, for example, SspB of S. gordonii (Brooks et al., 1997;Demuth et al., 2001;Park et al., 2005). This interaction has been shown to modulate the virulence potential of P. gingivalis (Hajishengallis & Lamont, 2016;Kuboniwa & Lamont, 2010) and may also be important for initial colonization of the oral cavity by P. gingivalis.
Therefore, disruption of heterotypic community formation by targeting the Mfa1/antigen I/II interaction may represent a potential therapeutic approach to control P. gingivalis colonization and virulence (Daep et al., 2011;Sztukowska, Roky, & Demuth, 2019;Tan et al., 2018). The region of antigen I/II involved in the interaction with F I G U R E 3 (a) Three-dimensional structure of the Mfa1 with a composite of five peptidomimetic adherence inhibitory compounds docked in a putative binding cleft. The residues that comprise the predicted binding cleft shown in "a" are shown in red underlined text in the Mfa1 sequence (b) or highlighted in red in the Mfa1 structure (c). The positions of residues R240 and W275 (see text) are shown in green and cyan, respectively F I G U R E 4 Inhibition of P. gingivalis adherence to S. gordonii by mutated Mfa1 peptides. Biofilms treated with parent and mutated peptides were compared and analyzed using an unpaired T test. ***p < .05, ns, not statistically significant Mfa1 has been extensively characterized (Daep, James, Lamont, & Demuth, 2006;Daep et al., 2008Daep et al., , 2011Daep, Novak, Lamont, & Demuth, 2010) and these studies led to the development of both peptide and small molecule peptidomimetics that potently inhibit P.
gingivalis/streptococcal adherence in vitro and significantly reduced P. gingivalis virulence in vivo (Daep et al., 2011;Patil et al., 2016;Patil, Tan, Demuth, & Luzzio, 2019;Tan et al., 2018). However, the inter- (b) Adherence of P. gingivalis to streptococci was determined using a two species biofilm model as described in Materials and Methods. Adherence data were normalized to the level of adherence of the wild-type P. gingivalis 33277 and data were analyzed using an unpaired T test. ***p < .001, **p < .05, ns, not statistically significant F I G U R E 6 Denaturation of P. gingivalis minor fimbriae. P. gingivalis cells were suspended in 1× LDS buffer and incubated either at (a) 100°C or (b) 60°C for 10 min. Extracts were electrophoresed in a 12% Bis-Tris gel and after transfer, Mfa1 was visualized using polyclonal anti-Mfa1 antibodies. Lanes 1, P. gingivalis ATCC 33277; 2, P. gingivalis SMF1; 3, P. gingivalis cSMF1; 4, P. gingivalis cMF1-R240A; and 5, P. gingivalis cSMF1-A357P; M, size markers however, this interaction by itself is insufficient to promote stable P. gingivalis-streptococcal biofilms (Lamont et al., 2002 (Alaei, Park, Walker, & Thanassi, 2019). Peptide CT2 also inhibited P. gingivalis-streptococcal biofilm formation and functioned by interfering with minor fimbrial biogenesis. In contrast, the BAR peptide and the BAR peptidomimetics function as competitive inhibitors of streptococcal adherence and have no effect on minor fimbrial biogenesis. The mature minor fimbriae of P. gingivalis also contain three additional tip proteins, Mfa3, Mfa4 and Mfa5. These proteins appear to play a role in the assembly of the tip complex itself and its incorporation into the fimbrial shaft and are required for optimal surface expression of the minor fimbriae (Hasegawa et al., 2016(Hasegawa et al., , 2013Ikai et al., 2015).
While Mfa1 has been shown to interact with Mfa3 (Lee et al., 2018), there is little information to suggest that the tip proteins contribute directly to P. gingivalis adherence to streptococci. Indeed, purified recombinant Mfa1 in the absence of the tip proteins potently inhibits P. gingivalis adherence, suggesting that Mfa3, Mfa4 and Mfa5 do not play a major role in the interaction with streptococcal antigen I/II.
Although several specific amino acids and/or structural motifs of Mfa1 were shown to be important for its interaction with Ag I/ II, the functional properties of other residues predicted to comprise the ligand binding cleft have yet to be determined. For example, K70 and Mfa1 amino acids 180-194 were identified as putative cleft residues; however, the truncated peptide N225 was only a poor inhibitor of P. gingivalis adherence. This suggests that these residues may not interact directly with Ag I/II (or the BAR peptide), or alternatively, that they may also require the contribution of downstream residues.
Since Mfa1 has been recently crystallized (Hall et al., 2018), it may be possible to co-crystallize the protein with the BAR peptide or with recently developed peptidomimetics of BAR (Patil et al., 2019) to generate a more complete picture of the Mfa1-Ag I/II interacting interface. Ultimately, a thorough understanding of the mechanism of the Mfa1/Ag I/II interaction will facilitate structure-based drug design and the development of potential therapeutics that may limit P. gingivalis colonization of the oral cavity.

ACK N OWLED G M ENTS
This study was supported by grants R01DE014605 and R01DE023206 from the National Institute for Dental and Craniofacial Research.
Modeling and docking studies were carried out in the Brown Cancer Center Molecular Modeling Core Facility at the University of Louisville which was supported by grant 1P30GM106396.

CO N FLI C T O F I NTE R E S T
The authors declare no conflicts of interest.