Two Autonomous Structural Modules in the Fimbrial Shaft Adhesin FimA Mediate Actinomyces Interactions with Streptococci and Host Cells during Oral Biofilm Development

Abstract

By combining X-ray crystallography and modeling, we describe here the atomic structure of distinct adhesive moieties of FimA, the shaft fimbrillin of Actinomyces type 2 fimbriae, which uniquely mediates the receptor-dependent intercellular interactions between Actinomyces and oral streptococci as well as host cells during the development of oral biofilms. The FimA adhesin is built with three IgG-like domains, each of which harbors an intramolecular isopeptide bond, previously described in several Gram-positive pilins. Genetic and biochemical studies demonstrate that although these isopeptide bonds are dispensable for fimbrial assembly, cell-cell interactions and biofilm formation, they contribute significantly to the proteolytic stability of FimA. Remarkably, FimA harbors two autonomous adhesive modules, which structurally resemble the Staphylococcus aureus Cna B domain. Each isolated module can bind the plasma glycoprotein asialofetuin as well as the polysaccharide receptors present on the surface of oral streptococci and epithelial cells. Thus, FimA should serve as an excellent paradigm for the development of therapeutic strategies and elucidating the precise molecular mechanisms underlying the interactions between cellular receptors and Gram-positive fimbriae.

Introduction

Specific interactions between microbes in the oral cavity are the key steps in the development of the oral biofilm community known as dental plaque. Development of this community begins with the attachment of early colonizers, such as Actinomyces and oral streptococci, to tooth enamel (Nyvad & Kilian, 1990), forming a dynamic layer that attracts bridging and late colonizers, which includes Gram-negative species that have been implicated in the etiology of periodontal diseases (Kolenbrander et al., 2006). Consistent with the important role of interbacterial adhesion in biofilm development, recent studies on human teeth provide physical evidence for the spatial distribution of various microbes in dental plaque as well as vibrant interactions between these species (Zijnge et al., 2010). For Actinomyces spp., fimbriae or pili are major adhesins that mediate bacterial adherence to host cells, tooth surface and etiological partners including oral streptococci (Yeung, 1999). Actinomyces oris, formerly Actinomyces naeslundii genospecies 2 (Henssge et al., 2009), which is known to predominantly colonize the human oral cavity (Kamma et al., 2000, Papaioannou et al., 2009, Preza et al., 2008), expresses two types of heterodimeric fimbriae. The type 1 fimbriae are required for bacterial adherence to salivary proline-rich proteins (PRPs) that coat the tooth surface (Gibbons et al., 1988), while the type 2 fimbriae promote Actinomyces adherence to oral streptococci and various host cells, including erythrocytes, epithelial cells, and polymorphonuclear leukocytes (McIntire et al., 1978, Yeung, 1999, Cisar et al., 1997). For the type 2 fimbriae, GalNAcβ1-3Gal and Galβ1-3GalNAc, present in cell-surface receptor polysaccharides (RPS) of oral streptococci or host glycoconjugates, serve as fimbrial receptors (Cisar et al., 1995, Ruhl et al., 2000, Ruhl et al., 1996, Stromberg & Karlsson, 1990). Consistently, Actinomyces adherence to epithelial cells was shown to be inhibited by methyl-β-D-galactoside, and N-acetyl-D-galactosamine, but not by methyl-α-D-galactoside, cellobiose, N-acetyl-D-glucosamine, L-fucose, or D-mannose (Brennan et al., 1984).

The type 2 fimbria is composed of FimA, which constitutes the fimbrial shaft, and FimB, a subunit located at the tip (Mishra et al., 2007). FimA belongs to a growing group of Gram-positive major shaft pilins that possess several conserved features required for pilus assembly. FimA harbors a typical carboxy-terminal cell wall sorting signal (CWS), which consists of a LPXTG motif, followed by a hydrophobic region and a positively charged tail (Navarre & Schneewind, 1999). Recognized by a transpeptidase enzyme called sortase (Mazmanian et al., 1999), the CWS is necessary and sufficient for the cell wall anchoring of a surface protein by the sortase-catalyzed mechanism first described for Staphylococcus aureus protein A (Mazmanian et al., 2001). Common to the Gram-positive shaft pilins, FimA also contains a putative pilin motif YPK (Ton-That et al., 2004), which is involved in the lysine-mediated transpeptidation reaction that crosslinks pilins into pilus polymers by a pilin-specific (i.e. class C) sortase (Dramsi et al., 2005) that was first described in Corynebacterium diphtheriae (Ton-That & Schneewind, 2003). C. diphtheriae harbors five pilin-specific sortases (srtA-srtE) organized in three pilus gene clusters, i.e. the spaA, spaD and spaH gene clusters encoding the three distinct types of pili (Ton-That & Schneewind, 2003). Formation of well studied SpaA pili depends on pilin-specific sortase SrtA (Ton-That & Schneewind, 2003), which cleaves the LPXTG motif of a SpaA precursor between threonine and glycine, forming an acyl enzyme intermediate with SpaA (Ton-That & Schneewind, 2004). This intermediate is resolved by a nucleophilic attack by the lysine residue within the pilin motif of an adjacent SpaA-SrtA intermediate, leading to the covalent linkage of two subunits. Subsequent cyclic addition of a free intermediate to the growing polymer extends the pilus structure. Predictably, mutations of the SpaA LPXTG motif or the lysine residue of the pilin motif abrogate pilus assembly (Ton-That et al., 2004). Interestingly, when expressed in C. diphtheriae, Actinomyces FimA precursors are polymerized by SrtD, the pilin-specific sortase required for the polymerization of the SpaH pili (Ton-That et al., 2004). Furthermore, phylogenetic analysis revealed that FimA is closely related to SpaH (Mishra et al., 2007), indicating similarity between the two pilus systems.

In addition to the intermolecular isopeptide linkage formed between the threonine residue within the LPXTG motif and the lysine residue within the pilin motif of an adjacent pilin, recent structural studies of several FimA homologs have revealed additional features. All possess multiple tandem immunoglobulin (IgG)-like domains initially described for the CnaB domains of the collagen binding protein Cna from S. aureus (Deivanayagam et al., 2000). Secondly, they contain internal isopeptide bonds formed between lysine and asparagines that are independent of sortase catalysis, and these intramolecular bonds are typically involved in protein stability (Kang & Baker, 2011). Sequence alignment and Blast analysis have revealed several conserved residues between FimA and its homologs, presumably involving intra- and intermolecular linkages (Mishra et al., 2007). However, these analyses did not illuminate the molecular basis for the multivalent functions of FimA that have been reported recently (Mishra et al., 2010). In that study, we showed that FimA is essential for the receptor-mediated co-aggregation of A. oris and S. oralis as well as for the haemagglutination of red blood cells. Furthermore, we showed that FimA is required for the monospecies biofilm that forms when Actinomyces is grown in media containing sucrose or human saliva.

To understand the structural basis of the multivalent role of FimA in fimbrial assembly and cellular adhesion, we attempted to crystallize recombinant FimA. We report here the 1.9Å resolution crystal structure of a carboxy-terminal fragment of FimA that contains two IgG-like N2 and N3 domains, each stabilized by an intramolecular isopeptide bond commonly found in Gram-positive pilins. We show that these linkages contribute significantly to the proteolytic stability of FimA. Based on the available structural homologs, we built a model of the full-length FimA that harbors an additional IgG-like domain at its amino-terminus, named the N1 domain. Intriguingly, either the N1 or N3 domain is sufficient to mediate adherence to a fimbrial receptor molecule that can be competitively inhibited. This study thus provides an experimental model to elucidate the molecular mechanism of interactions between cellular receptors and Gram-positive fimbriae.

Results

Determination of FimA structure by X-ray crystallography

A FimA construct of A. oris T14V consisting of residues 33–488 (FimA_33–488), which lacks both the N-terminal leader peptide sequence and the C-terminal cell wall sorting signal, was expressed in Escherichia coli and purified by affinity chromatography (see Experimental Procedure). Attempts to crystallize this recombinant FimA_33–488 protein were not successful. FimA_33–488 was then treated with trypsin to generate a stable fragment that could be purified and crystallized. The N-terminal sequencing of the purified tryptic fragment revealed the loss of the N-terminal 33–198 residues (Figure 1A), wherein Lys198 of the conserved pilin motif YPK is involved in the intermolecular isopeptide linkages forming the pilus shaft (Ton-That & Schneewind, 2004, Kang et al., 2007) (see below). The crystal structure of FimA_199–488 was then solved by the multiple isomorphous replacement (MIR) method using iridium and samarium derivatives and refined to a resolution of 1.9Å with an R/R_free value of 20.9/24.9%. The asymmetric unit contains one molecule of FimA_199–488, having two IgG-like domains (N2 and N3) (Figure 1), and the final model contains 215 water molecules and one Zn²⁺ ion located at the interface of two crystal symmetry related molecules. Two loops consisting of residues 364 to 375 (AB loop) and 473 to 480 (FG loop) and three residues at the N-terminus were disordered in the crystal structure.

Figure 1.

Modules of FimA. (A) Schematic representation of the FimA modules. Backbone pilin FimA contains a signal peptide (S) at the N-terminus and a C-terminal cell wall sorting signal (CWS), which consists of a LPXTG motif, membrane-spanning hydrophobic domain and positively charged tail. The middle portion contains three IgG-like domains referred to as N1, N2 and N3. The pilin motif (VPKN) is present near the N1-N2 domain linker. The solid lines at the bottom and top indicate the intra-molecular isopeptide bonds and disulfide bridges. The isopeptide and disulfide bonds in the N1 domain are predicted based on the homology modeling. (B) Ribbon representation of the crystal structure of FimA_199–488 containing two IgG-like domains (N2 and N3), with strands from each domain colored in rainbow style from red (N-terminus) to violet (C-terminus). The N2 domain exhibits a CnaA/DEv-IgG fold, while the N3 domain is an IgG-rev fold/cnaB. Both the N2 and N3 domains contain an isopeptide bond and a conserved acidic residue, shown as sticks. The disulfide bridge in the N3 domain is shown in yellow stick. A Zn²⁺ binding site in the N2 domain is shown in grey. (C) Topology diagram of the crystal structure FimA_199–488. The core β strands of N2 (DEv-IgG fold) and N3 (IgG-rev fold) domain are labeled A to G corresponding to 1B. The horizontal black and yellow lines indicate positions of the two isopeptide bonds and a disulfide bridge.

The crystal structure of the stable fragment FimA_199–488 is of elongated shape with roughly 92 Å in length and 35 Å in width (Figure 1A & 1B). It is made of two IgG-like domains arranged in tandem, and a `small middle domain,' consisting of a three (anti-parallel) stranded β-sheets, is tucked between them. Both IgG-like domains, sporting a β-sandwich shape and named as N2 (middle domain) and N3 (C-terminal domain) (Figure 1), respectively, are made of two anti-parallel β-sheets similar to the other major pilin structures reported to date (Budzik et al., 2009, Kang et al., 2007, Kang et al., 2009, Vengadesan et al., 2011). The linker joining the N2 and N3 domains from head to toe appears as a β-strand in the `small middle domain' (Figure 1 B & C) and the N2/N3 domain interface is large, covering an area of 14,348 Ų.

The N2 domain (residues 201–356) is composed of two β-sheet folds (I and II), each made of 4 β-strands of various lengths and resembles the DEv-IgG/cnaA type fold first observed in S. aureus MSCRAMM CNA (Symersky et al., 1997). Four anti-parallel strands (A, B, E, D) form the β-sheet I and the β-sheet II, which is typically made of five strands (C, F, G, DI, DII) in the conventional DEv-IgG fold, is composed of four anti-parallel strands (C, F, DI, DII), as a prolin-rich segment distorts the G strand. A 3-turn α-helix is seen linking the two inserted antiparallel strands (DI and DII) to the E-strand. The `small middle domain' at the interface is made of three anti-parallel β strands (I1, I2, I3), where the N2–N3 domain linker joining the G strand of N2 domain to A strand of N3 domain forms the I2 strand, and the EF and DE loops of N3 domain form I1 and I3 strands, respectively. The N2-domain harbors a Zn²⁺ ion that is in coordination with the acidic residues from an α helix (Glu288), a long loop connecting DI and D strands (Asp268, Asp271) and a proline-rich segment (Glu337) of a neighboring molecule (Figure 1B & C).

The N3-domain (residues 357 – 488) is of the IgG-rev/cnaB type fold, first identified in the B-region sub-domains of collagen binding MSCRAMM CNA of S. aureus (Deivanayagam et al., 2000). The N3-domain is formed by two β sheet folds (I and II), which are made of three (D, A, G) and four strands (C, B, E, F), respectively (Figure 1C). A 22-residue long loop in the N3-domain, connecting D and E strands, crosses over the domain interface towards N2 domain and contributes a β-strand (I3) to the `small middle domain'. The corresponding loop of the N2-domain is 41 residues long and forms two extra β-strands (DI and DII) and the 3-turn helix, as discussed above. Another 27-residue long loop joining B and C strands of the N3-domain covers the outer surface of β-sheet II, and the corresponding loop is only 7 residues long in the N2-domain. This flexible BC loop is stabilized by a disulfide bridge between its Cys455 and Cys394 present on the E strand.

Intra-molecular isopeptide bonds

The intra-molecular isopeptide bonds in Gram-positive adhesins were first identified in the structure of Spy0128 from S. pyogenes also containing two tandem IgG-like domains (Kang et al., 2007). These isopeptide bonds were shown to be formed by intra-molecular reactions between the NZ atom of Lys and the CG of Asn, facilitated by an acidic residue present in close proximity (Kang et al., 2007).

Two intra-domain isopeptide bonds were found in the FimA_199–488 structure, identified through difference electron density maps. In the N2-domain of FimA_199–488, an isopeptide bond is found between Lys206 and Asn319 that links the A strand and the N-terminus of the F strand (I1) (Figure 1C & 2A) in cis configuration. An acidic residue Asp246, present in close proximity on the C strand, facilitates the interaction by hydrogen bonding with N-H and C=O moieties in the isopeptide bond. The isopeptide bond is surrounded by hydrophobic residues, Phe224, Leu248, Val226, Val352, Ala321, and Phe305.

Figure 2.

The intra-molecular isopeptide bonds in the N2 and N3 domains from FimA. The isopeptide bond, its conserved acidic residue, and surrounding hydrophobic residues are shown using stick models. The dotted lines represent hydrogen bonds. The electron density is from 2Fo-Fc map contoured at 1.5σ. (A) The N2-domain isopeptide bond links two anti-parallel strands from opposite side β-sheets (A, N-terminus of F) by Lys206 and Asn319, with catalytic Asp246 present on C strand. (B) The N3-domain isopeptide bond connects adjacent parallel A and G strands through Lys363 and Asn485, with catalytic Glu450 from E strand.

The covalent isopeptide bond in the N3 domain is formed between Lys363 of strand A and Asn485 present at the C-terminal end of G strand in a trans configuration (Figure 1C & 2B). The acidic residue Glu450 is located at a distance of 4.5Å on E strand, likely catalyzing this bond formation (Kang et al., 2007). Interestingly, a water molecule resides between Glu450 and the isopeptide bond, hydrogen bonding with the OD1 moiety of Asn485 and the OE2 moiety of Glu450, indicating indirect involvement of Glu450 through a water molecule. The isopeptide bond Lys363–Asn485 is surrounded by the hydrophobic residues Ile483, Pro454, Pro460, Leu376, Phe381, Ile361 and Ala379.

Comparison with other gram-positive pilin structures

The Dali server (Holm & Rosenstrom, 2010) identified major pilins, SpaA of C. diphtheriae (PDB code 3HTL; Z-score of 20.1), BcpA of B. cereus (PDB code 3KPT; Z-score of 14.1), RrgB of S. pneumoniae (PDB code 2X9W; Z-score 13.4), minor pilin GBS52 of S. agalactiae (PDB code 3PHS, Z-score 12.4) and GBS80 of S. agalactiae (PDP code 3PF2; Z-score of 15.4) as the best structural homologues for FimA_199–488. When the N3 domain of FimA alone was used for the search, the N2 domain of GBS52 was the top hit. The conserved intra-domain isopeptide bonds seen in the DEv-IgG fold and IgG-rev fold of the FimA are structurally similar in the respective domains of other major pilins such SpaA, BcpA, GBS80, and RrgA with variations in the loop lengths and conformations. However, a significant structural deviation is seen in the proline-rich segment (₃₃₂PPTPETPPTDPENPP₃₄₇) located at the FG loop of FimA. The primary and tertiary structural comparison with the other major pilins (SpaA, BcpA, GBS80 and RrgB) reveals the presence of a proline-rich segment only in the FimA, though it shares the other common features such as the pilin motif, conserved residues for intra-domain isopeptide bonds, and the sorting motif, as pointed out in the introduction. Note that proline-rich segments are also identified in other Gram-positive pilins including GBS52 and FctB (Kang & Baker, 2009, Krishnan et al., 2007), but these are located near the LPXTG motif of their CWSS.

The superposition of FimA_199–488 onto its best structural homolog SpaA resulted in an rms deviation of 1.9 Å for 150 common Cα atoms. An incomplete disulfide bond between neighboring B and E strands was observed in the N3-domain of the C. diphtheriae major pilin SpaA crystal structure (Kang et al., 2009), speculated to be due to DTT used in crystallization. The disulfide bridge in N3 domain of FimA is similarly located; however, it appears as linking E-strand and BC loop due to an insertion of 10 residues at the C-terminal end of the B-strand and variations of BC and DE loops. There is an additional β-strand in the DE loop of SpaA, which aligns with three-stranded interface β-sheet. The FimA N2-domain G-strand is linked to the A-strand of the N3-domain through the middle strand (I2) of the `small middle domain' β-sheet (Figure 1B & C), whereas in SpaA, the linker between the N2 and N3 domains appears as an extended continuous β-strand.

Modeling of FimA amino-terminal domain N1

The sequence comparison between FimA and SpaA helped us model the FimA N1 domain as IgG-rev or cnaB type fold (Figure 3). The resulting full length model of FimA revealed a possible intra-domain isopeptide bond in the N1 domain formed between residues Lys53 and Asn199, where the later residue resides next to the pilin motif Lys198, and the required catalytic acidic residue as Glu158. The N1 domain model also suggests a possible disulfide bridge between Cys116 and Cys157. Interestingly, this disulfide bond also connects the E strand and BC loop as seen in the N3 domain. Thus, in addition to three isopeptide bonds, which are shown to enhance the thermodynamic stability and resistance to proteolysis in Gram-positive pilins, the two disulfide bridges are expected to further increase the stability in the FimA. Many Gram-positive bacterial major pilins (GBS80, BcpA, RrgB) lack Cys residues for disulfide formation with exception to SpaA, which has only one disulfide bond in the N3 domain. Whether or not the disulfide bridges are involved in the aforementioned functions remains to be investigated.

Figure 3.

The FimA model containing N1, N2 and N3 domains. The N1 domain is modeled based on the SpaA structure of C. diphtheriae. The strands are colored in rainbow style as in Figure 1B. The residues of isopeptide bonds, a conserved acidic residue are shown as sticks. The pilin motif Lys198 and free proximal Cys116 and Cys154 for possible disulfide bond formation are marked by arrows.

Requirements for fimbrial assembly and proteolytic stability

As mentioned above, FimA contains a putative pilin motif, with the conserved lysine residue K198 proposed to be involved in the formation of intermolecular isopeptide bonds, and two intramolecular isopeptide bonds involving K206–N319 and K363–N485. To determine whether these residues are critical for fimbrial assembly, we generated several alanine-substitution mutants from the parental pFimA vector (Mishra et al., 2010) and transformed the resulting plasmids into the fimA deficient mutant (ΔfimA) of Actinomyces. We next examined by Western blotting fimbrial polymers isolated from liquid cultures of these strains grown to mid-log phase, representing the bacteria-free culture medium (marked as M) and the cell wall fraction (W) obtained by muramidase treatment. Both culture medium and cell wall fractions were obtained quantitatively by TCA precipitation, acetone wash and subjected to SDS-PAGE and immunobloting analyses using antibodies against FimA (α-FimA). In wild type MG-1 bacteria, most FimA polymers (FimA_HMW) were detected abundantly in cell wall fractions (Figure 4, MG-1 lanes). A similar phenotype was observed in a ΔfimA strain expressing FimA from plasmid, whereas no specific FimA signal was detected in the ΔfimA mutant alone (Figure 4, next four lanes). Alanine substitution of the lysine residue K198 in the FimA pilin motif completely abrogated fimbrial polymers with apparent accumulation of FimA monomers (FimA_M) (Figure 4, K198A lanes), demonstrating the essential role of K198 in intermolecular bond formation in fimbrial polymers. In sharp contrast, fimbrial assembly was not compromised by alanine substitution of residues involved in the N2 domain isopeptide bond, i.e. K206, N319 and their catalytic residue D246, and the N3 domain isopeptide bond, i.e. K363, N485 and their catalytic residue E450 (Figure 4, remaining lanes). Also, no defect was apparent for the assembly of the tip fimbrillin FimB with these alanine substituted FimA mutants (data not shown).

Figure 4.

Requirements of inter- and intra-molecular isopeptide bonds in FimA fimbrial polymerization. Culture medium (M) and cell wall (W) fractions were collected from MG-1 and its isogenic derivatives grown in early log phase. Equivalent protein samples were separated on 4–12% tris glycine gradient gel and detected by immunoblotting with antibodies against FimA (α-FimA). The positions of monomers (FimA_M), high molecular mass products (FimA_HMW) and molecular mass markers are indicated.

To confirm that these mutations do not affect surface assembly of FimA structures, we examined Actinomyces fimbriae by immuno-electron microscopy (IEM), whereby cells of various strains were immobilized on nickel grids, reacted with α-FimA orα-FimB followed by gold particles conjugated to IgG, stained with 1% uranyl acetate and then viewed by an electron microscope (Figure 5). Consistent with the results above, mutation of K198 abolished assembly of FimA (i.e. the type 2 fimbriae), without affecting the assembly of type 1 fimbriae as expected (Figure 5D). Note that a few FimA- and FimB-stained gold particles were observed in the FimA-K198A mutant (compare Figure 5B & 5D and 5F & 5H), indicating that surface display of monomeric FimA and FimB did not require the FimA lysine K198 residue, reminiscent of what we previously reported in C. diphtheriae (Ton-That et al., 2004). Furthermore, consistent with the Western blotting results (Figure 4), mutations in the residues involved in the two isopeptide bonds did not affect surface assembly of type 2 fimbriae (data not shown).

Figure 5.

The conserved lysine residue K198 within the FimA pilin motif is required for surface assembly of type 2 fimbriae. Bacterial cells were immobilized on nickel grids, stained with α-FimA (A, B, C and D) or α-FimB (E, F, G and H) and goat anti-rabbit IgG conjugated to 12 nm gold particles. Samples were then stained with 1% uranyl acetate and viewed by transmission electron microscopy. Note that MG-1 and its isogenic mutants still express type 1 fimbriae. Scale bars indicate the length of 0.2 μm.

Intramolecular isopeptide bonds in FimA confer proteolytic stability

We next examined whether intra- and inter-molecular isopeptide bonds confer fimbrial stability by subjecting isolated cell wall-linked fimbriae to tryptic digestion. The cell wall-linked fimbriae were isolated by treating Actinomyces cells with muramidase (See Experimental Procedure). Prior to TCA precipitation and acetone wash for SDS-PAGE and immunoblotting with α-FimA as described in Figure 4, the cell wall-linked fimbriae were either treated with increased concentrations of trypsin or mock-treated. Wild type fimbriae (expressed from the chromosome, lane MG-1, or plasmid, lane pFimA) were highly resistant to trypsin (Figure 6, first six lanes). Monomeric FimA, accumulated in the K198A mutant, also displayed high resistance to tryptic digestion (Figure 6, lanes K198A). Strikingly, mutations that disrupted the formation of the N2 domain isopeptide bond sensitized fimbrial degradation by trypsin as evident by the appearance of multiple fimbrial fragments (Figure 6, lanes K206A, N319A and K206A/N319A). By comparison, the alanine substitution of D246, proposed to facilitate the formation of the K206-N319 linkage, produced a lesser effect (Figure 6, lanes D246A), suggesting that D246 may not be the only facilitator of intramolecular isopeptide bond formation. Similarly, cell wall-linked fimbriae with mutations that targeted the formation of the C-domain isopeptide bond were also sensitive to trypsin cleavage (Figure 6, the remaining lanes). Clearly, the intramolecular isopeptide bonds of FimA contribute to its proteolytic stability, although these bonds are not required for fimbrial assembly.

Figure 6.

Proteolytic stability of type 2 fimbriae is mediated by intramolecular isopeptide bonds. Cell wall fractions were collected from MG-1 and its isogenic derivatives grown in early log phase and treated with increasing quantities of trypsin (0, 40 and 80 μg/ml) for two hours at 37°C. Equivalent protein samples were separated on 4–12% tris glycine gradient gel and detected by immunoblotting with α-FimA. The positions of monomers (FimA_M), high molecular mass products (FimA_HMW) and molecular mass markers are indicated.

Intramolecular linkages are dispensable for biofilm formation and Actinomyces coaggregation with streptococci

The data presented above predicts that intramolecular linkages may not be critical in receptor-mediated interactions between Actinomyces and other etiological partners of dental plaque such as Streptococcus oralis. One of these interactions is the self-association interaction between Actinomyces cells leading to the formation of a monospecies biofilm that is dependent on sucrose (Mishra et al., 2010). To investigate this possibility, various strains of Actinomyces expressing individual FimA mutants or wild type FimA were subjected to the biofilm assay, simply by growing Actinomyces cells for 48 h at 37°C with 5% CO₂ in the presence of sugar molecules. Biofilm quantification was then measured by optical density at 450 nm after staining the biofilms with crystal violet (see Experimental Procedure) (Figure 7A, C.V.). In parallel, surface expression of FimA in these strains was determined by dot-blotting with α-FimA (Figure 7). Consistent with our previous report (Mishra et al., 2010), wild type FimA-expressing cells formed monospecies biofilm in the presence of sucrose (Figure 7A, lane MG-1), fructose and glucose (data not shown). That this biofilm formation is dependent on FimA is evident from the fact that deletion of fimA abrogated its biofilm formation (Figure 7A, lane ΔfimA), a defect that could be rescued by expressing FimA in this mutant by plasmid (Figure 6A, lane pFimA). Surprisingly, although the K198A mutation did not abolish surface display of FimA monomers (FimA_M), as shown in Figure 5D, the FimA-K198A mutant failed to form biofilm (Figure 6A, lane K198A). Presumably, this defect is due to the drastically decreased level of FimA on the bacterial surface as determined by dot-blotting (Figure 7A, lane K198A). By contrast, none of the mutations involving intramolecular linkages affected biofilm formation, except for the D246A mutant that exhibited some defect, likely related to a decreased level of FimA observed with this mutant (Figure 7A, the remaining lanes).

Figure 7.

Requirements for FimA-mediated biofilm formation and coaggregation of Actinomyces with oral streptococci. (A) Wild type MG-1 and its isogenic derivatives were grown in 96-well plates at 37°C with 5% CO₂ for 48 h and the generated biofilms were stained with crystal violet (C.V.) and quantified optical density (OD) at 580 nm. The values shown are the means of three independent experiments and the error bars represent standard deviations. The same strains were also examined for FimA surface expression by dot immunoblotting. (B) Stationary-phase cultures of wild type MG-1 and its isogenic derivatives were examined for co-aggregation with RPS-bearing S.oralis 34.

We next determined whether a disruption of the intramolecular isopeptide bonds interferes with FimA-mediated interbacterial adhesion between Actinomyces and S. oralis. For these experiments, Actinomyces and streptococcal cells at equal optical density were mixed together in microtiter plates and coaggregation was assessed by photography. In agreement with the results in Figure 7A, only ΔfimA and FimA-K198A mutants failed to coaggregate with S. oralis 34, while other mutants displayed coaggregation with a level comparable to that of the wild type and its isogenic fimA mutant expressing FimA (Figure 7B), with an exception of the D246A mutant, which exhibited somewhat reduced coaggregation compared to the wild type (Figure 7B, lane FimA-D246A).

Independent adherence of two FimA modules to glycoprotein asialofetuin

A key unresolved question about the function of the shaft fimbrillin FimA is whether it is sufficient and necessary to mediate interactions with cellular receptors, such as polysaccharides on the streptococcal surface and host cells. Secondly, as Gram-positive pilins are so diverse in their primary sequences, which structural domains of Gram-positive pilins confer cellular interactions has so far remained obscured. As shown in Figure 1, the N2-domain of FimA represents the CnaA-type/Dev-IgG fold first seen in the A-region sub-domains of the collagen binding MSCRAMM CNA of S. aureus. In addition, our bioinformatics and BLAST analyses revealed that the N3-domain also contains a putative CnaB-type/IgG-rev fold domain (Figure 8), suggesting that the N2/N3-domains might be involved in binding to cellular molecules such as asialofetuin. Asialofetuin is a monomeric plasma glycoprotein containing three O-linked disaccharide chains with the Galβ1-3GalNAc motif (Nilsson & Svensson, 1979), and as shown previously asialofetuin-coated latex beads are capable of aggregating Actinomyces oris T14V (Heeb et al., 1982) and asialofetuin blocks binding of Actinomyces to S. orils (Mishra and Ton-That, unpublished data).

Figure 8.

Identification of two adhesive modules of FimA for plasma glycoprotein asialofetuin. (A) Shown is a schematic diagram of wild type FimA along with its N-terminal signal peptide, pilin motif, CnaB domain, C-terminal cell wall sorting signal. Positions of truncated derivatives of FimA are indicated. (B) Binding of FimA proteins to asialofetuin was measured by ELISA at OD of 450 nm. The results are presented as means of three independent experiments, with error bars showing standard deviations. Streptococcal RrgB was used as a control.

To identify the receptor binding site(s) of FimA, we generated several recombinant FimA proteins that were expressed in E. coli, purified by affinity chromatography and examined for their binding to asialofetuin by ELISA (see Experimental Procedure). As shown in Figure 8, the FimA_33–497 recombinant protein, which lacks the leader peptide sequence and the CWS (Figure 1A), adhered strongly to asialofetuin-coated plates. As a control, recombinant protein RrgB did not bind (Figure 8B); RrgB is the major pilin subunit of Streptococcus pneumoniae (Barocchi et al., 2006, LeMieux et al., 2006). Surprisingly, the recombinant FimA_33–212, which is devoid of the N2 and N3 domains, bound to asialofetuin with only 40% reduced affinity compared to the FimA_33–497 recombinant, indicating the N1 domain represents an autonomous binding domain. Remarkably, FimA mutants lacking this N1 domain of CnaB-type fold, missing the N3 domain of CnaB-type fold or containing only the N1 domain displayed similar binding affinities, whereas the FimA mutant containing only the N2 domain did not bind asialofetuin. Thus, the N1 and N3 domains of FimA having the CnaB-type fold are the two binding domains for asialofetuin. Incidentally, S. aureus B-region domains with similar folding did not have a role in collagen binding (Hartford et al., 1999). Conceivably, the CnaB-type fold may be specific for receptor polysaccharides.

Interactions of two FimA adhesive modules with Streptococcus oralis and epithelial cells

Similar to asialofetuin, the receptor polysaccharides (RPS) expressed by early colonizing oral streptococci and host cells such as KB oral epithelial cells contain the Galβ1-3GalNAc motif, which is known to be a core fimbrial receptor (Cisar et al., 1995). We asked if the two domains as shown above displayed comparable binding affinity to S. oralis 34 and oral epithelial KB cells by performing experiments similar to the one described for asialofetuin in Figure 8. In this case, asialofetuin was replaced by S. oralis 34 or oral epithelial KB cells grown in the microtiter plates. Remarkably, comparable binding of the different FimA recombinant proteins to the KB epithelial cells was also observed consistently (Figure 9A), supporting the independent roles of the N1 and N3 domains of FimA in receptor binding.

Figure 9.

Interactions of two adhesive modules of FimA with cellular receptors present on epithelial cells and Streptococcus oralis. Recombinant FimA proteins were added to ELISA plates coated with oral epithelial KB cells (A) or RPS-bearing S. oralis 34 (B), with RPS-negative S. oralis OC1 used as a control (asterisk). In (C), FimA33-497 protein, preincubated with an increasing dose of asialofetuin (0, 0.1 and 0.2 mg), was added to ELISA plates coated with asialofetuin, KB cells or S. oralis 34. Binding of FimA was quantified by OD at 450 nm. The results are presented as means of three independent experiments, with error bars showing standard deviations. Streptococcal RrgB was used as a control.

FimA binding to oral streptococci was next examined using S. oralis 34 and its isogenic mutant OC1 lacking RPS (Yoshida et al., 2005). The FimA_33–497 recombinant protein bound strongly to S. oralis 34 and this binding was specific to RPS since the same protein did not bind to the OC1 mutant (Figure 9B, lane with asterisk). In general agreement with the above results, binding of recombinant FimA proteins containing the N1 only was significantly reduced as compared to the FimA_33–497 protein, and the isolated N2 domain was unable to bind. Most remarkably, the CnaB domain bound to S. oralis 34 with similar affinity as FimA_33–497, pinpointing the CnaB domain of FimA as the major adhesive module for the RPS of oral streptococci.

All the data above support the notion that FimA binds to the common core molecule of these cellular receptors that contain the Galβ1-3GalNAc motif. To probe this, we went on to determine if asialofetuin can block the FimA binding to KB cell as well as S. oralis 34, by pre-incubating the recombinant proteins with asialofetuin at different concentrations and using the preincubated mixtures in the binding assays as described above. Indeed, asialofetuin preincubation not only prevented its own binding to the recombinant FimA in control experiments (Figure 9C), it also blocked binding of FimA to KB cells and S. oralis 34 (Figure 9C).

Discussion

Actinomyces fimbriae are arguably the first pilus system of Gram-positive bacteria for which fimbrial interactions with cellular receptors have been characterized extensively before the elucidation of the mechanism of their assembly. The major hurdle of the latter has been the lack of efficient genetic tools and methodology for a systematic molecular genetic and biochemical investigation. Recent development of facile gene deletion systems for Actinomyces has advanced this area (Mishra et al., 2010, Wu & Ton-That, 2010), leading to the identification of genetic components of the two fimbrial types in Actinomyces (Mishra et al., 2007, Mishra et al., 2010). Consequently, the genetic dissection of the fimbrial components has uncovered the important fact that FimA is a multivalent adhesin that mediates bacterial coaggregation with oral streptococci and adherence to red blood cells and biofilm formation, in addition to its role as the fimbrial shaft of type 2 fimbriae (Mishra et al., 2010). This was unexpected and contrasts with the fact that the tip fimbrillin FimB of type 2 fimbriae is not involved in any of these processes (Mishra et al., 2010). While these intercellular interactions are known to involve a polysaccharide receptor present on the surface of streptococci (Cisar et al., 1995, Cisar et al., 1997) and host cells (Ruhl et al., 2000, Ruhl et al., 1996, Stromberg & Karlsson, 1990), dissection of the precise structural determinants of FimA that give rise to its multivalent functions became a key issue. We addressed that in this report, by combining x-ray crystallography and structural modeling with genetic and biochemical dissection of the functional modules. Our work revealed a multitude of structural features that contribute to FimA's functionality and a number of structural principles that are conserved in Gram-positive pilins in spite of their gross dissimilarity in primary sequence.

The fact that the major fimbrillin FimA is essential for the adherence of Actinomyces to itself as well as for the adherence of Actinomyces to the oral streptococci and host cells (Mishra et al., 2010) indicated to us that FimA might contain distinct structural modules that might mediate these diverse interactions. The structure of the pilus adhesin GBS52 of S. agalactiae is a case in point (Krishnan et al., 2007), which consists of two IgG-like domains, N1 and N2. It has been shown that only the N2 domain is specifically involved in host cell adherence. Although our attempts to crystallize FimA was successful for only a portion of the protein, and revealed two distinct IgG-like modules in the structure, we were able to build a model of the missing N-terminal part based on a close homolog, SpaA of corynebacteria, whose structure was described recently (Kang et al., 2009). According to this model, FimA appears to have three familiar IgG-like domains – N1 (deduced by modeling) and N2 and N3 solved by a high resolution structure (Figures 1 & 3). Each of these modules possesses an intramolecular isopeptide bond that has been recognized as a determinant of structural stability of pilins that confers proteolytic stability (Kang & Baker, 2011). Consistent with studies of the Streptococcus pyogenes major pilin Spy0128 known as a Lancefield T1 antigen (Kang & Baker, 2009), our genetic and biochemical data show that mutations disrupting the isopeptide linkage of FimA did not interfere with fimbrial assembly but compromised its resistance to proteolysis (Figures 4 & 6). We also showed that while the intramolecular isopeptide bonds are dispensable for fimbrial assembly of Actinomyces, the lysine residue in the FimA pilin motif is essential for formation of intermolecular linkages (Figure 4 & 5), as is true for many other Gram-positive pilins (Quigley et al., 2009, Budzik et al., 2008, Okura et al., 2011, Kang & Baker, 2009, Ton-That et al., 2004). Unexpectedly, when this lysine was mutated to alanine, surface assembly of monomeric FimA was drastically reduced (Figures 5 and 7), suggesting fimbrilin FimA may be more primed for fimbrial assembly than cell wall linkage as a non-fimbrial entity. As a consequence, Actinomyces cells expressing this mutant FimA failed to form mono-species biofilm in the presence of sugar molecules (sucrose, fructose or glucose). Importantly, mutations of the residues required for intramolecular linkages did not abrogate biofilm formation (Figure 7), indicating that intramolecular linkages are not involved in cell-to-cell interactions.

Although FimA possesses three structurally autonomous IgG-like domains, only two of these (N1 and N3 separated by N2) have so far been identified as functional modules by virtue of their independent binding to the plasma glycoprotein asiafetuin, oral epithelial cells and S. oralis. Common to these etiological agents of the oral biofilm is the presence of disaccharide motifs GalNAcβ1-3Gal or Galβ1-3GalNAc. It is highly likely that the N1 and N3 domains bind directly to GalNAcβ1-3Gal or Galβ1-3GalNAc. This is supported by the evidence that recombinant FimA failed to bind to the surface of S. oralis lacking this specific receptor molecule (Figure 9) and that binding of FimA to the three agents was competitively inhibited by preincubation with asialofetuin. The question thus arises as to why the N2 domain did not have any binding affinity to GalNAcβ1-3Gal or Galβ1-3GalNAc, given that it has a similar IgG-like fold. It is tempting to speculate that the N2 module may bind to a different receptor molecule yet that is present on the surface of bacteria in the oral biofilm. Moreover, the spatial arrangement of individual strains within the biofilm may dictate its specificity for a particular receptor. Future studies should explore FimA interaction with other prominent members of the oral biofilm community. The solution of co-structures of FimA adhesive modules with specific receptor molecules, which would reveal specific binding sites of the N1/N3 domains for the disaccharide motifs, should provide deeper insights into the mechanism of receptor-mediated adhesion by Actinomyces in particular and the Gram-positive pilins in general.

Experimental Procedure

Bacterial strains, plasmids, media and cell culture

Bacterial strains and plasmids used in this study are listed in Table S1. Actinomyces were grown in hearth infusion broth (HIB) or on HIB agar plates. Streptococci were grown in CAMG media with or without erythromycin (10 μg/ml) as needed for the maintenance of ermAM. E. coli DH5α, used as host for molecular cloning experiment, was grown in Luria Broth (LB). Kanamycin was added at 50 μg/ml wherever required. Rabbit-raised polyclonal antibodies against recombinant fimbrial proteins were previously described (Mishra et al., 2007). Reagents were purchased from Sigma unless indicated otherwise.

Human oral epithelial cells (KB) were cultured and maintained in RPMI-1640 medium (Thermo-scientific) supplemented with 100 U penicillin, 100 μg/ml streptomycin and 10% fetal bovine serum (FBS; Gibco, USA) at 37°C under 5% CO₂.

Recombinant plasmids

-pFimA_33–488 – Primers FimA-33-Nde-F and FimA-488-BgIII-R (Table S2) were used to PCR-amplify the coding region of FimA spanning residues 33–488 from chromosomal DNA of Actinomyces oris T14V, while appending NdeI and BglII sites. The DNA fragment, digested with BamHI and NdeI, was cloned into pET34b (Novagen) precut with BamHI and NdeI enzymes. The resulting plasmid was transformed into E. coli BL21 (DE3) and confirmed by DNA sequencing.

- pFimA – To construct pFimA, the 5' promoter sequence and UTR of fimB (508 bp) were fused to the coding sequence of fimA. For PCR amplification, primers Pro508B-Bam-F and Pro508-BXba-R (Table S2) were used to amplify the 5' promoter sequence of fimB with flanking BamHI and XbaI sites from chromosomal DNA of MG-1. Two primers FimA-Xba-F and FimA-Eco-R (Table S2) were used for PCR amplification of the coding region of fimA including its ribosomal binding site while appending XbaI and EcoRI sites. The two PCR fragments were digested with BamHI and XbaI or XbaI and EcoRI enzymes and ligated into vector pJRD215 precut with BamHI and EcoRI. The resulting plasmid was transformed into E. coli DH5α and A. oris ΔfimA.

- Site-directed mutagenesis of recombinant plasmids – PCR-based site-directed mutagenesis of double-stranded DNA was used to introduce point mutations in FimA as previously described (Swierczynski & Ton-That, 2006). Briefly, recombinant plasmid DNA (pCR2.1-fimA) was used as a template for PCR amplification with Pfu DNA polymerase (Stratagene) using appropriate primer pairs (Table S2). After PCR amplification and overnight treatment with DpnI at 37°C, the resulting plasmids were transformed into E. coli DH5α and confirmed by DNA sequencing. The plasmids carrying the desired mutations of fimA were then used as a template for PCR amplification of mutated fimA fragments, which were then cloned into plasmid pJRD215 containing the fimB promoter (see above). Lastly, the recombinant plasmids were transformed into A. oris ΔfimA by electroporation.

-Ligation-Independent Cloning (LIC) – Various truncated FimA proteins were cloned into vector pMCSG7 using LIC to produce recombinant proteins with an N-terminal hexa-histidine tag as previously described with some modifications (Stols et al., 2002). Briefly, appropriate oligonucleotide primers (Table S2, LIC primers) and A. oris MG-1 genomic DNA were used to PCR-amplify the corresponding sequences of fimA. Generated PCR products were treated with T4 DNA polymerase (Novagen) in the presence of dCTP (2.5 mM) as the sole nucleotide in 10 μl reactions. Similarly, SspI-cut pMCSG7 was treated with T4 DNA polymerase in the presence of dGTP. The treated PCR products and vector above were mixed together at a molarity ratio of 1:4 (vector: insert) and allowed to anneal at 70°C for 30 seconds, followed by 5 min incubation at room temperature after adding 2.5 mM EDTA. The resulting vectors were directly transformed into E. coli DH5α. Bacterial colonies were screened by colony-PCR to identify positive clones with respective genes, which were further verified by DNA sequencing. The recombinant plasmids were introduced into E. coli BL21 (DE3) for protein expression.

Protein purification

For binding studies, various FimA proteins were expressed by pMCSG7 vector (see above) in E. coli BL21 (DE3). One liter cultures of E. coli was grown at 37°C in LB medium containing 100 μg/ml ampicillin until OD₆₀₀ ~ 0.6. FimA expression was induced by addition of 1 mM isopropyl-ß-D-thiogalactopyranosidase (IPTG) at 30°C. After 4 h of induction, cells were harvested by centrifugation and resuspended in 25 ml of lysis buffer (50 mM NaH₂PO₄, 300 mM NaCl, 10 mM imidazole, pH 8.0). The cells were lysed by passage through a French press cell twice at 11,000 psi. The clear lysates obtained by centrifugation were subject to purification by affinity chromatography as previous described. Purified proteins were subsequently dialyzed twice against 2 liters of EQ buffer (150 mM NaCl, 50 mM Tris.HCl, pH 7.5) at 4 °C and stored at −20 °C in 10% glycerol.

For structural analysis, FimA (residues 33–488) of A. oris T14V, which lacks its leader peptide sequence and the C-terminal sorting signal, was expressed in E. coli. E. coli BL21 (DE3) harboring pFimA_33–488 was grown in LB media at 37°C in the presence of 100 μg/ml kanamycin until OD₆₀₀ ~ 0.6. After overnight induction with 0.2 mM IPTG at 25°C, cells were harvested by centrifugation, resuspended in lysis buffer (50 mM phosphate pH 7.8, 300 mM NaCl, 10% (v/v) glycerol, 10mM imidazole) and lysed by sonication. The clear lysates obtained by centrifugation were loaded onto charged 5ml NiCl₂ HiTrap chelating column and eluted with a linear gradient of 0–300 mM imidazole of elution buffer (300 mM imidazole in lysis buffer). Fractions containing protein, confirmed by SDS-PAGE, were dialyzed against TBS (20mM Tris-HCl pH 7.0, 150mM NaCl, 5mM EDTA, 40mM L-Arg and L-Glu). The concentrated protein, using a Millipore Amicon 10KDa centrifugal device, was subject to Sephacryl gel filtration with a S100 (26/60) column for further studies. For trypsin cleavage, purified FimA was treated 50 mM Trypsin at 37°C and the resulting stable fragment FimA_199–488, determined by N-terminal sequencing, was purified and crystallized.

Crystallization, data collection and structure determination

The protease resistant fragment FimA_199–488 was crystallized using the hanging-drop vapor-diffusion method at 22°C. Small plate-like crystals were obtained from the well solution containing 0.1 M imidazole buffer, 0.2M zinc acetate and 17% PEG 2000 MME and FimA at ~ 20 mg/ml. Diffracting-quality crystals were obtained by streak-seeding. Small plate-like crystals were initially crushed and diluted in the stabilization solution (0.1 M imidazole buffer, 0.2M zinc acetate and 20% PEG 2000 MME), and human hair was used to streak the seeds into the hanging drop. These crystals belonged in a monoclinic P2₁ space group, with cell dimensions a = 47.5 Å, b = 39.8 Å, c = 77.1, β = 98.3° and one molecule per asymmetric unit. The crystals were equilibrated in the cryo-protectant 15 % glycerol and stabilizing solution. The low resolution native data and diffraction data on crystals soaked with heavy-atom derivatives was collected on a RIGAKU R-axis IV imaging plate detector using CuKα radiation on an in-house rotating-anode X-ray generator and processed by D*trek (Pflugrath, 1999). Native diffraction data to 1.9 Å were collected at beam line 24ID-C at the Advanced Photon Source (Argonne, IL) and were processed and scaled using HKL2000 (Otwinowski & Minor, 1997)

Multiple isomorphous replacement method was used to determine the FimA crystal structure using two heavy-atom derivatives (Sm(CH₃COO)₃ and K₃IrCl₆). Diffraction data sets were processed and scaled using D*trek. The heavy-atom determination and phase calculations, performed by autoSHARP program (Bricogne et al., 2003, de La Fortelle & Bricogne, 1997), found 4 heavy atoms sites from two heavy atom-derivatives and yielded excellent phasing statistics with an initial high quality map. Automatic model-building was carried out using ARP/wARP (Perrakis et al., 1999). Iterative model building and refinement were carried out using COOT (Emsley & Cowtan, 2004) and REFMAC (Murshudov et al., 1997). 215 water molecules were added in the final cycle's refinement with REFMAC. The final R_free and R_factor are 24.9% and 20.9%, respectively. The data collection and refinement statistics are presented in Table 1.

Table 1.

Data collection, processing, phasing and refinement statistics.

	FimA-N2N3	Sm(CH3COO)3-FimA-K3IrCl6-FimA-N2N3 N2N3
Resolution range (Å)	40–1.9 (1.97–1.9)	40–2.4 (2.49–2.4)	40–2.2 (2.28–2.2)
Space group	P21	P21	P21
Unit-cell parameters (a, b, c in Å; β in °)	47.5, 39.8, 77.1; 98.3	47.4, 39.8, 77.3; 98.0	47.6, 39.5, 76.4; 99.0
Unique reflections	22677	11046	13380
Multiplicity	3.5 (3.5)	3.7 (3.7)	2.7 (2.7)
Mean I/σI (I)	41.7 (11.8)	12.3 (5.2)	19.5 (7.3)
Completeness (%)	99.3 (99.5)	96.5(94.5)	92.2 (92.0)
Rmerge (%)	6.3 (20.3)	7.0 (23.8)	3.3 (11.7)
Phasing
Phasing Power ISO (acentric/centric)		0.51/0.46	0.87/0.85
Phasing Power ANO		0.27	0.51
Rcullis ISO (acentric/centric)		0.88/0.87	0.71/0.75
Rcullis ANO		0.9	0.93
Figure of Merit (acentric/centric)	0.28/0.27
Refinement
Rcryst/Rfree (%)	20.0/24.9
Average B value (Å2)	28.7
Rmsd, bonds Å)	0.01
Rmsd, angles (°)	1.11
No. of protein/metal atoms	2040/1
No. of solvent atoms	199
Ramachandran plot	95.5/4.5/0.0
Preferred/Allowed/outliers (%)

The numbers in parentheses correspond to the values in the highest resolution shell.

Cell fractionation, Western blotting and dot-blotting

Overnight grown actinomyces were used to inoculate mid-log phase cultures (1:50 dilution) at 37°C in HIB medium. Kanamycin was added to a final concentration of 50 μg ml-1 as needed. All strains were grown till OD₆₀₀ ~ 0.3 and normalized aliquots were fractionated into supernatants – representing culture medium fractions (M) – and cell pellets by centrifugation. The cell pellets were treated with mutanolysin (300 U ml-1) in SMM buffer (0.5 M sucrose, 10 mM MgCl2, and 10 mM maleate, pH 6.8) for overnight at 37°C. After mutanolysin treatment, soluble fractions – represent cell wall fractions (W) – were separated from the protoplasts by centrifugation. The culture medium and cell wall fractions were subjected to TCA precipitation and acetone wash prior to boiling in SDS-containing sample buffer, separated on 4–12% Tris-glycine gradient gels (Invitrogen). Fimbriae were detected by immunoblotting with specific rabbit antisera (1:10,000 for α-FimA and 1:2000 for α-FimB & α-FimO) and Horseradish Peroxidase (HRP)-Conjugated goat anti-rabbit IgG for chemiluminescence. For trypsin treatment, cell fractionation was carried out as the above with the exception that the cell wall fractions were treated with 40 or 80 μg/ml trypsin in 0.1M Tris pH 8.0 for 2 h at 37°C prior to the TCA precipitation step.

Cell surface expression of FimA in A. oris was determined by dot-blotting according to a published protocol (Mishra et al., 2010). Briefly, overnight grown cultures of various Actinomyces strains were washed and suspended in PBS. Approximately 5 × 10⁵ CFU in 5 μl drops of each strain were spotted on nitrocellulose membrane and allowed to air-dry completely. Membrane was blocked in 5 % skim milk in phosphate buffered saline (PBS) containing 0.2% Tween-20 (PBST) for 1 h prior to incubation with rabbit-raised α-FimA (1:10,000). Membrane was washed, reacted with anti-rabbit IgG-horseradish peroxidase conjugate and analyzed by chemiluminescence.

Bacterial coaggregation

Co-aggregation was performed with various strains of Actinomyces and S. oralis 34 and OC1 as previously described (Mishra et al., 2010). Briefly, stationary-phase cultures of bacterial strains grown in CAMG compex medium with 0.2% glucose, were harvested by centrifugation, washed in Tris buffered saline (pH 7.5) containing 0.1 mM CaCl₂ and suspended to equal cell density of approximately 2 × 10⁹ ml⁻¹. For co-aggregation, 250 μl of Actinomyces and streptococcal cell suspensions was mixed in 24-well plates for a few minutes on a rotator shaker and photographed by a FluorChem Q Imager (AlphaInnotech).

Immuno-electron microscopy

Actinomyces cells grown on HIB agar plates were suspended in 0.1 M NaCl, washed with and resuspended in PBS. For immunogold labeling, a drop of bacterial suspension in PBS was placed on nickel grids with formvar carbon support (Electron Microscopy Sciences), washed three times with PBS containing 2% BSA and blocked for 1 h in PBS with 0.1% gelatin. Fimbriae were stained with primary antibodies diluted in PBS (1:100 for α-FimA and 1:50 for α-FimB) containing 2% BSA for 1 h, followed by washing with the same buffer. Samples were then stained with 12-nm gold-goat anti-rabbit IgG (Jackson ImmunoResearch) diluted 1:20 in PBS containing 2% BSA for 1 h. The grids were finally washed five times with water, stained with 1% uranyl acetate and viewed in a JEOL JEM-1400 electron microscope.

Biofilm formation assay

Biofilm assays were followed as previously described with some modifications (Mishra et al., 2010). Briefly, overnight cultures of Actinomyces were diluted 1:100 in HIB containing 1% sucrose, glucose or fructose and 200 μl was dispensed into 96-well polystyrene plates (Corning, NY), followed by incubation at 37°C with 5% CO₂ for 48 h. Stationary-phase cells with similar optical density at 600 nm were gently washed three times with sterile PBS and air dried for 30 min. Biofilms were stained with 0.5% crystal violet for 30 min, washed extensively to remove the unbound dye and air dried. Biofilm suspensions in ethanol/acetone (80:20) were quantified using an Infinite M1000 microplate reader (Tecan). Samples were performed in quadruplicate in three independent experiments.

Binding of recombinant proteins

A solid-phase binding assay was employed to determine whether FimA recombinant proteins bind to asialofetuin (Sigma-Aldrich), KB cells and S. oralis. For asialofetuin, 96-well microplates (Greiner bio-one, medium binding) were coated with 100 μl of asialofetuin (16 μg/ml), centrifuged at 2000 rpm for 5 min, and incubated for 2h at 37 °C. The plates were blocked with 5% milk for 1h at room temperature and washed three times with PBST. Recombinant proteins were added at the final concentration of 40 μM. After centrifugation, the plates were incubated at 37 °C for 1h, followed by washing and staining with anti-penta His-HRP labeled antibodies. After 1h incubation at room temperature, the wells were washed three times with PBST, and incubated with tetramethylbenzidine for 30 min. The reactions were stopped by sulfuric acid and read at OD₄₅₀. All assays were performed in triplicate in three independent experiments.

For KB cells and S. oralis, the epithelial and streptococcal cells were used in place of asialofetuin in 96-well plates. FimA binding to these cells was measured as described above. For blocking experiments, recombinant FimA protein (40 μM) was incubated with asialofetuin (100 μg) for 2h at 37 °C prior to use in ELISA as described above.