The Glycan Moieties and the N-Terminal Polypeptide Backbone of a Fimbria-Associated Adhesin, Fap1, Play Distinct Roles in the Biofilm Development of Streptococcus parasanguinis

Abstract

Fap1, a fimbria-associated glycoprotein, is essential for biofilm formation of Streptococcus parasanguinis and mediates bacterial attachment to saliva-coated hydroxylapatite, an in vitro tooth model (E. H. Froeliger and P. M. Fives-Taylor, Infect. Immun. 69:2512-2519, 2001; H. Wu and P. M. Fives-Taylor, Mol. Microbiol. 34:1070-1081, 1999; H. Wu et al., Mol. Microbiol. 28:487-500, 1998). Fap1 belongs to a growing family of high-molecular-weight serine-rich proteins found in streptococcal and staphylococcal species and possesses two serine-rich repeat regions. The glycan moiety of Fap1 appears to be O linked within the repeat regions (A. E. Stephenson et al., Mol. Microbiol. 43:147-157, 2002). In the present study, we identified a gene cluster immediately upstream of fap1 that encodes three putative glycosyltransferases and one nucleotide-sugar synthetase-like protein. Inactivation of one glycosyltransferase gene galT2 abolished the expression of two glycan epitopes; however, it did not alter bacterial ability to adhere to both SHA and saliva-conditioned biofilm surfaces. In contrast, the biofilms formed by the galT2 mutant were shallow and had a 70% decrease in biomass accumulation, suggesting that these glycan moieties mediated by GalT2 are not required for the initial adhesion but are important for biofilm formation. A recombinant N-terminal Fap1 polypeptide was shown to interact with a 53-kDa salivary protein and block and displace bacterial attachment, further demonstrating the role of the Fap1 polypeptide in bacterial adhesion. Taken together, these results suggest that Fap1 glycosylation plays an important role in bacterial biofilm formation, whereas the nonglycosylated Fap1 peptide mediates bacterial initial attachment during the process of biofilm formation.

Biofilms are a complex bacterial community. Biofilm formation is a multistep process that is initiated by bacteria attachment to biological or nonbiological surfaces and promoted by the formation and networking of bacterial extrapolysaccharide matrix (6). The steps involved in biofilm development and maturation have been well documented for gram-negative bacteria (22-24). Less is known about the biofilm formation process of gram-positive bacteria, especially oral streptococci, despite their adaptation to a biofilm lifestyle in oral environments (4) and their engagement in the development of the most common oral infections, such as dental caries and periodontal disease (16). In addition, the biofilm formation process is also critical to the pathogenesis of oral streptococcus- mediated endocarditis (7).

Adhesion of oral streptococci to the tooth surface is the first step in the formation of dental plaque, the most complex human biofilms (17). Many surface components and signaling molecules of oral streptococci are critical for biofilm formation (17). Fimbriae, the major surface structure of Streptococcus parasanguinis, a primary colonizer of the tooth surface, are required for bacterial attachment. Fap1, the structural subunit of fimbriae, is a serine-rich glycoprotein and required for bacterial adhesion and biofilm formation (12). Fap1 is composed of glycan moieties and a polypeptide backbone. The relative contribution of glycan and peptide to bacterial adhesion and biofilm formation is not known. Glycan moieties of serine-rich proteins appear to be O linked to the serine residues found in serine-rich repeat regions (37-39). Fap1 glycosylation is mediated by a seven-gene cluster that includes accessory secretion components and glycosyltransferases (36). One glycosyltransferase (Gtf) gene product appears to glycosylate the Fap1 polypeptide; however, inactivation of the gtf gene renders the nonglycosylated Fap1 more susceptible to protein degradation and less stable. Therefore, it is difficult to assess the function of unstable Fap1 using the gtf mutant.

Fap1-like proteins have also been described in many streptococcal and staphylococcal species and play important role in bacterial interaction with host components (26, 34, 35, 40). For instance, GspB and Has of Streptococcus gordonii (29), SrpA of Streptococcus sanguinis (25), and SraP of Staphylococcus aureus (27) interact with human platelets, and SrpA of Streptococcus cristatus mediates its coaggregation with fusobacterium (13). GspB and SrpA are also glycosylated. Glycosylation and biosynthesis of GspB are mediated by an accessory secretion system (31). Many glycosyltransferase-related genes, such as gly and nss, linked to the accessory secretion system of S. gordonii are not found in the identified fap1 downstream locus. We sought to identify additional glycosyltransferases that are involved in Fap1 glycosylation and to determine the function of these gene products in Fap1 glycosylation and Fap1-mediated bacterial adhesion and biofilm formation.

In the present study we identified a gene cluster containing glycosyltransferase and nucleotide-sugar synthetase genes. Characterization of one glycosyltransferase GalT2 indicates that GalT2-mediated Fap1 glycosylation plays a critical role in biofilm formation. Furthermore, we show that the N-terminal region of the Fap1 polypeptide mediates S. parasanguinis adhesion. Therefore, we conclude peptide and glycan moieties of Fap1 differentially mediate S. parasanguinis biofilm formation.

MATERIALS AND METHODS

Bacterial strains, plasmids, and reagents.

The bacterial and phage strains and plasmids used in the present study are listed in Table 1. Escherichia coli strains DH5α and JM109 and their derivatives were grown in LB with appropriate antibiotics as specified. S. parasanguinis FW213 and its mutant derivatives were grown in Todd-Hewitt broth with the specified antibiotics. The Fap1 mutant VT1393 (fap1::Kan^r) was constructed as described previously (37). The antibiotics used in the present study were as follows: ampicillin (50 μg/ml), chloramphenicol (34 μg/ml), erythromycin (300 μg/ml), kanamycin (25 μg/ml), and tetracycline (12 μg/ml) for E. coli and erythromycin (10 μg/ml), kanamycin (100 μg/ml), and tetracycline (12 μg/ml) for S. parasanguinis. A low-copy-number plasmid pHSG576 was used as a cloning vector, whereas pSF151 was used as the integration vector for insertional inactivation of the target gene. The monoclonal antibodies (MAbs) used in the present study are described elsewhere (10, 28). E42 (a peptide-specific MAb), F51, and D10 are glycan-specific MAbs.

TABLE 1.

Bacterial and phage strains and plasmids used in this study

Strain or plasmid	Relevant characteristicsa	Source or reference
Strains
FW213	S. parasanguinis parent strain	5
VT1393	FW213, fap1::aphA3; Kanr	39
VT1583	FW213, gtf1::aphA3; Kanr	36
VT1584	Phage, fap1 positive	39
AL105	FW213, galT2::Kanr	This study
AL108	FW213, galT2::aphA3/pVA838::Pgly-galT2; Kanr Ermr	This study
AL109	FW213, galT2::aphA3/pVA838; Kanr Ermr	This study
Plasmids
pAL100	3.9-kb fap1 upstream fragment in pGEM-T Easy; Ampr	This study
pAL101	2.5-kb HaeII fragment in pGEM-T Easy; Ampr	This study
pAL102	700-bp HindIII fragment in pHSG576; Cmr	This study
pAL103	3.2-kb HindIII-PstI fragment in pHSG576; Cmr	This study
pAL104	350-bp galT2 internal fragment in pSF151, Kanr	This study
pAL106	6.0-kb fap1 upstream region in pVA838; Ermr	This study
pAL107	Pgly-galT2 in pVA838; Ermr	This study
pVA838	E. coli-streptococcal shuttle vector	19
pSHG576	Low-copy-number cloning vector	32
pSF151	Streptococcal integration vector	33

^aKan^r, kanamycin resistance; Erm^r, erythromycin resistance; Amp^r, ampicillin resistance; Cm^r, chloramphenicol resistance.

PCR amplification and primers.

Recombinant phage DNA positive for fap1 was isolated from FW213 λEMBL3 genomic library as described previously (39). A 3.9-kb fragment that is located upstream of fap1 was amplified from the fap1-positive phage DNA using the λEMBL3 left arm vector-specific sequence 5′-CGCTTATCTGCTTCTCATAG-3′ (EMBL F1) and the 5′ end of the fap1 sequence, 5′-CCATTTTAGCACCGTATTCC-3′ (800bp2), as primers. This upstream fragment was also amplified from FW213 chromosomal DNA by using corresponding primers to confirm that this region is present in the FW213 genome. To isolate a region further upstream of this fragment, we performed an inverse PCR. HaeII-digested FW213 genomic DNA was self-ligated at a concentration of 10 ng of DNA/ml, the ligation mixture was used as a template to amplify a 2.5-kb PCR fragment with the primers nss5′-1 (5′-CAGTTGTTCAAGGAATGTGG-3′) and nss5′-2 (5′-GCTTGAGCACAACAACATCTG-3′). PCR was carried out in a Techne Genius PCR apparatus using Vent DNA polymerase (New England Biolabs). The reaction was initiated at 94°C for 2 min and continued for 30 cycles of 94°C for 30 s, 60°C for 30 s, and 72°C for 2 to 4 min. The PCR fragment was purified using QIAGEN PCR purification column.

DNA sequencing and analysis.

A 3.9-kb PCR product and a 2.5-kb inverse PCR product were cloned into pGEM-T Easy to construct pAL100 and pAL101, respectively. The purified 3.9-kb PCR fragment was digested with HindIII and PstI. Two fragments resulting from the digest, a 700-bp HindIII and a 3.2-kb HindIII-PstI fragment were subcloned into pHSG576 to construct pAL102 and pAL103, respectively. The resulting plasmids were used as sequencing templates. The nucleotide sequences of both strands of each insert were determined by primer walking at the Vermont Cancer Center DNA Analysis Facility at the University of Vermont. Nucleotide and protein similarity searches were conducted with the BLAST programs via the National Center for Biotechnology Information BLAST server.

Construction of a galT2 mutant and its complemented strains.

A 350-bp internal fragment of galT2 was amplified from S. parasanguinis chromosomal DNA by PCR with BamHI-embedded (italicized) galT2-1 (5′-GATCGGATCCGTGCCTATATATAATGCAG-3′) and galT2-2 primers (5′-GATCGGATCCGCAAAGCTCTGTCTTTCCTC-3′). The amplified internal fragment of galT2 was then cloned into the BamHI site of pSF151 (33) to construct pAL104. The resulting plasmid was propagated in E. coli JM109 prior to introduction to S. parasanguinis FW213. Transformation of pAL104 into FW213 was accomplished by electroporation as described previously (11). Transformants were selected on 100-μg/ml kanamycin TH agar plates. Chromosomal DNA was purified from putative transformants. Recombination that occurred at the expected site of the chromosomal DNA was confirmed by PCR and Southern blot analysis as described previously (39). A galT2 mutant, AL105, was also generated and used in the present study.

The full-length galT2 gene on the E. coli-streptococcus shuttle vector pVA838 was constructed as follows. The fap1 upstream region spanning from the gly promoter to galT2 was amplified from FW213 genomic DNA with BamHI-embedded (italicized) primers gly5′-1 (5′-GATCGGATCCTCGATTCTTATGAGGTGTTG-3′) and galT2-3′ (5′-GATCGGATCCAATGAAGCGAGAATCAG-3′). The PCR product was digested with BamHI and then cloned into BamHI-digested pVA838 vector, resulting in the plasmid pAL106. This plasmid contained Pgly-gly-nss-galT1-galT2. The gly-nss-galT1 fragment was deleted by inverse PCR using the KpnI-embedded (italicized) primers KpnI-gly5′-2 (5′-GATCGGTACCATAACTCTTCTAATCATTTAAACG-3′) and KpnI-galT2-5′ (5′-GATAGGTACCGATTGAAAATTATCCGAAGGAG-3′). The resulting inverse PCR product was digested with KpnI and self-ligated to generate a plasmid, pAL107, that contained Pgly-galT2. This new plasmid and its parental plasmid pVA838 were transformed into the galT2 mutant. Transformants were selected based on erythromycin resistance, confirmed by plasmid preparation, and designated AL108 and AL109, respectively.

Immunoprecipitation of Fap1 species of S. parasanguinis.

A total of 200 μg of cell lysates prepared from FW213 and AL105 was incubated with 5 μg of the Fap1-specific MAb E42 at 4°C for 2 h. After incubation, 50 μl of treated rProtein AG beads was added to the antibody-antigen mixture, followed by incubation at 4°C overnight. The incubated mixtures were collected by centrifugation at 1,000 rpm for 2 min and washed three times with NETN washing buffer (20 mM Tris-HCl [pH 7.0], 100 mM NaCl, 1mM EDTA, 0.2% NP-40) at 4°C. The proteins bound to rProtein AG beads were eluted by boiling in sodium dodecyl sulfate (SDS) sample loading buffer and subjected to SDS-polyacrylamide gel electrophoresis (PAGE) and immunoblot analyses with succinyl wheat germ agglutinin and MAb E42.

BacterELISA analysis.

Surface expression of Fap1 by S. parasanguinis and its derivatives was monitored by using BacterELISA as described previously (9). In brief, wild-type or mutant strains were dried into each well of 96-well plates, and wells were blocked with Tris-buffered saline plus 0.1% Tween 20 (TBST). Primary antibody binding was detected with horseradish peroxidase-conjugated secondary antibody. After incubation with the secondary antibody, antibody binding was detected with an ECL system as described previously (39). The plate was washed three times with TBST following primary and secondary antibody incubations. The antibody reactivity with each strain was tested in triplicate by three independent experiments.

RNA isolation.

S. parasanguinis bacteria were grown in 10 ml of Todd-Hewitt broth to an optical density of 0.6 at 470 nm. The 10⁹ cells were resuspended in diethyl pyrocarbonate-treated water and added to 2-ml FastPrep Blue vials containing extraction solutions described by the manufacturer (Bio 101, La Jolla, CA). The bacteria were lysed with a FastPrep instrument (Bio 101) at setting 6 for 30 s and immediately placed on ice for 1 min. Samples were centrifuged at 12,000 × g for 15 min at 4°C. The upper aqueous phase was extracted with chloroform-isoamyl alcohol and precipitated with isopropanol. Samples were centrifuged at 12,000 × g for 5 min at 4°C, and pellets were washed with RNase-free salt-ethanol solution. RNA was suspended in 50 μl of diethyl pyrocarbonate-treated water and stored at −80°C. The yield and integrity of the RNA was assessed by using a spectrophotometer and nondenaturing agarose gel electrophoresis.

RT-PCR.

A trace amount of genomic DNA was removed from the total RNA preparation using amplification-grade DNase I (Gibco-BRL). The first-strand cDNA synthesis from total RNA was carried out with the SuperScript first-strand synthesis system for reverse transcriptase PCR (RT-PCR). Briefly, gene-specific oligonucleotides, fap1-RT1 (5′-GAAGTGAAATTGTCAAG-3′) and galT1-RT1 (5′-TACTGGTTTGTCTCCCTG-3′), were primed with 1 μg of DNase I-treated RNA, respectively; the first-strand was synthesized by reverse transcription using SuperScript II RT. The synthesized DNA was amplified directly using PCR with corresponding gene-specific primer pairs. A 396-bp fap1 fragment was amplified by fap1 RT1 and 2 (5′-GATCAATTTACGAGCAGCC-3′), a 438-bp galT1 fragment was amplified by galT1-RT1 and galT1-RT2 (5′-TACTGGTTTGTCTCCCTG-3′ and 5′-AGTTGTCAGATCCGCAAC-3′). To confirm the purity of the RNA samples and the complete elimination of genomic DNA contamination by DNase I treatment, a no-RT control was used in every RT-PCR experiment.

Adhesion of S. parasanguinis to SHA: an in vitro tooth model.

Adhesion of S. parasanguinis to saliva-coated hydroxylapatite (SHA) was performed as described previously (39). Cells of S. parasanguinis FW213, VT1393, and the galT2 mutant were labeled with 1 mCi of [methyl-³H]thymidine/ml to specific activities of 5.26 × 10³ ± 395, 5.01 × 10³ ± 327, and 5.97 × 10³ ± 287 CFU/cpm, respectively (means of three experiments, three determinations of each, ± the standard deviation). Radiolabeled bacteria were incubated with SHA beads at 37°C for 1 h to allow bacterial binding to occur. Supernatants and washed beads were transferred to scintillation vials, and their respective radioactivities were determined. Radioactive counts associated with the SHA beads represent bound bacteria, whereas those present in the supernatant fractions represent unbound, free-living bacteria. Adhesion was determined by dividing the radioactive counts recovered from the bound bacteria by those from both bound and free-living bacteria. Wild-type strain FW213 and the fap1-null mutant were used as positive and negative controls, respectively.

Adhesion blocking and displacement assays.

S. parasanguinis cells were labeled with 1 mCi of [³H]thymidine/ml and grown to a cell density at 6 × 10⁸ CFU/ml. The radiolabeled cells were washed three times and resuspended in 50 mM phosphate buffer (pH 6.0; adhesion buffer) to reach the same cell density. The bacterial suspension was sonicated with an ultrasonic cuphorn system (Heat Systems-Ultrasonics., Inc.) to break bacterial chains apart. Recombinant Fap1 was expressed and purified as described previously (28), and the purified rFap1 was dialyzed overnight against the adhesion buffer. The recombinant Fap1 or bovine serum albumin (BSA) reconstituted in adhesion buffer was added into adhesion buffer equilibrated SHA beads and incubated at 37°C for 1 h on a rotating shaker. For the rFap1 displacement experiment, rFap1 was added only after incubation of SHA with wild-type bacteria. S. parasanguinis FW213 was incubated with SHA for 1 h and washed as usual. One milliliter of a 100-μg/ml concentration of rFap1 was added, followed by incubation for 1 h. The unbound protein(s) were then removed, and the beads were washed extensively. The treated SHAs were subjected to a regular adhesion assay procedure as described previously.

Overlay assay of saliva blots.

Multiple 20-μl portions of clarified saliva were resolved by SDS-12% PAGE and transferred to a nitrocellulose membrane blot. Saliva blots were incubated either with 100 μg of fimbria preparations/ml (28) or with 50 μg of purified recombinant Fap1/ml dissolved in TBST or with control buffer (TBST) for 1 h at room temperature. After three washes, the blots were processed for immunoblotting. The polyclonal fimbria-specific antibody or MAb Fap1-specific E42 was used as the primary antibody to detect the binding of the fimbria preparation or rFap1 to salivary protein(s), respectively.

Biofilm formation of S. parasanguinis derivatives.

Experimental analyses of biofilm formation were performed by two methods: a microtiter plate assay and a confocal laser scanning microscopic (CLSM) assay (12). For the microtiter plate assay, overnight cultures of S. parasanguinis strains were diluted to 1:100 in Todd-Hewitt broth-1% glucose, and 200 μl of this cell suspension was used to inoculate sterile saliva-coated 96-well polystyrene microtiter plates. After 12 h at 37°C, the wells were gently washed three times with 400 μl of distilled H₂O, dried in an inverted position, and stained with 0.1% crystal violet for 15 min. The wells were rinsed, and the crystal violet was solubilized in 250 μl of acetic acid (30% [vol/vol]). The optical density at 562 nm was determined by using a microplate reader. Each assay was performed in triplicate and repeated three times. Bacterial adherence to saliva-coated polystyrene coverslips was examined by CLSM under the following modifications. Portions (3 ml) of diluted overnight cultures were inoculated into the wells of 24-well microtiter dishes containing sterile saliva-coated polystyrene coverslips. After incubation for 16 h at 37°C, the coverslips were gently rinsed three times with distilled H₂O, stained with SYTO 9 from a Live/Dead BacLight bacterial viability kit (Molecular Probes, Eugene, OR) and mounted on a microscope for imaging.

A Bio-Rad MRC-1000 CLSM system in combination with a Nikon Diaphot 300 inverted microscope was used to characterize a larger field view of real-time biofilms on the surfaces of saliva-coated polystyrene coverslips. The microscope was equipped with a 40×, 1.15-numerical-aperture water immersion lens and a krypton-argon laser. Excitation at 488/10 nm was used to detect the green fluorescence. Images of bacterial biofilms were collected as x-y and x-z sections and were captured as digital files with the packaged software (Bio-Rad); the pixel fluorescent intensity was measured over a 512 × 512 μm field.

RESULTS

Identification of a gene cluster that encodes three glycosyltransferases and one nucleotide-sugar synthetase-like proteins.

Fap1 is a glycoprotein (28). We have identified a gene cluster located at the fap1 downstream region (36); however, mutation of a glycosyltransferase renders the Fap1 unstable relative to mature Fap1. Therefore, the mutant precludes us to further evaluate the function of Fap1 glycosylation. To define the function of glycosylation in bacterial adhesion, we sought to identify additional genes that are important for Fap1 modification. We have isolated a fap1-positive phage DNA from S. parasanguinis lambda genomic library (37). PCR analysis of the positive clone revealed the presence of a 3.9-kb fragment upstream of fap1 (Fig. 1A). An additional 2.5-kb upstream HaeII fragment was amplified by inverse PCR using the HaeII- digested and religated S. parasanguinis genomic DNA as a template (Fig. 1B). The complete nucleotide sequence of the 6.4-kb fragment was determined. Four open reading frames that code for glycosyltransferases (gly, galT1, and galT2) and a nucleotide-sugar synthetase (nss) gene were identified (Fig. 1). Comparison of the deduced amino acid sequences of these open reading frames with the available database showed that this locus was related to glycosyltransferase systems. Gly, GalT1, and GalT2 all contained a family 2 glycosyltransferase domain. The domain overlapped significantly with WcaA domain that mediates cell wall biogenesis of many bacteria.

FIG. 1.

Diagram of gly, nss, galT1, galT2, and fap1 gene loci. (A) Restriction map of the fap1-positive recombinant phage DNA. A 3.9-kb fragment was PCR amplified from the recombinant phage DNA using a λEMBL3 left-arm-specific sequence (F1) and a fap1 5′-end-specific sequence (800bp2) as primers. The left and right arms represent λEMBL3 cloning vectors. (B) Organization of the fap1 upstream region. Genes that share homology with glycosyltransferase (gly), nucleotide-sugar synthetase (nss), and galactosyltransferase genes (galT1 and galT2) are located upstream of the fap1 locus. The fap1 gene is 618 bp from the stop codon of galT2. The HaeII genomic DNA fragment was amplified by inverse PCR using nss5′-1 and nss5′-2 primers.

GalT2 is important for Fap1 glycosylation.

To determine the role of GalT2 in Fap1 glycosylation, we mutated the galT2 gene by insertional mutagenesis. Since galT2 is the last gene in this cluster, it is likely that the insertional inactivation of galT2 would generate a nonpolar mutation. RT-PCR analysis demonstrated expression of the flanking genes, fap1 and galT1was not affected by the galT2 mutation (data not shown), confirming that the galT2mutation is nonpolar.

To characterize the mutant, we performed BacterELISA and immunoblot analysis with two glycan-specific MAbs and one peptide-specific MAb. The galT2 mutant exhibited significant reactivity with MAb E42, a peptide-specific antibody, indicating that the galT2 mutant retained the intact peptide backbone of Fap1. However, it did not react with MAbF51 and D10, the glycan-specific antibodies (Fig. 2A); implying the galT2 mutant has a defect in Fap1 glycosylation. The fap1 mutant did not react with any of these MAbs, demonstrating the antibodies are specific for Fap1. Western blot analysis of the galT2 mutant showed that a band binding to the peptide-specific MAb E42 was slightly smaller than 200kDa (Fig. 2B, top panel, lane 2), indicating this band is devoid of some modification. Further, this band did not bind to the glycan-specific antibodies (Fig. 2B, bottom panel, lane 2), suggesting the presence of glycosylation defects. These results support the notion that GalT2 mediates Fap1 glycosylation. In addition, introduction of the full-length galT2 gene back into the galT2 mutant restored the expression of all epitopes recognized by these Fap1-specific antibodies (Fig. 2A and B), and the migration of Fap1 at the mature Fap1 position (Fig. 2B, lane 3) was also evident, while such a mature Fap1 was not detected when an empty vector pVA838 used as a control (Fig. 2B, lane 4). Therefore, we conclude that the mutation of galT2 is responsible for the Fap1 glycosylation defect we observed in the galT2 mutant.

FIG. 2.

Expression of Fap1 by the galT2 mutant. Peptide-specific MAb E42 and glycan-specific MAbs (F51 and D10) were used in BacterELISA (A), Western blot (B), and immunoprecipitation (C) analyses to monitor Fap1 expression and glycosylation. (A) Expression of Fap1 by the galT2 mutant was compared to that of wild type and the fap1-negative mutant. Bars indicate absorbance at 490 nm and represent an indirect measurement of Fap1 expression level. (B) Western blotting analyses of Fap1 expression and glycosylation by the galT2 mutant. Cell lysates were subjected to Western blot analyses with peptide-specific MAb E42 (top panel) and glycan-specific MAb F51 (bottom panel). Lane 1, FW213; lane 2, AL105; lane 3, AL108 and lane 4, AL109. (C) The galT2 mutant produces GlcNAc-positive Fap1 species. Cell lysates of FW213 (lane 1) and AL105 (lane 2) were immunoprecipitated with MAbE42 and subjected to immunoblot analysis with lectin succinyl wheat germ agglutinin (sWGA; left panel) and MAb E42 (right panel).

Our previous studies revealed that the gtf mutant failed to express any detectable glycan moieties which include two glycan-epitopes recognized by MAbs F51 and D10 (36). To determine whether the Fap1 species generated by the galT2 mutant is still glycosylated by other sugar residues, we performed lectin blot analyses on MAb E42 immunoprecipitated fractions from both the wild type and the galT2 mutant. The GalT2 mutant expressed GlcNAc-positive Fap1 species and migrated slightly faster than the wild-type species (Fig. 2C), demonstrating that the Fap1 molecule expressed by the galT2 mutant possesses GlcNAc-positive epitope(s).

Role of galT2 in S. parasanguinis adhesion.

Fap1 is required for bacterial adhesion to SHA, an in vitro tooth model (37). Previous antibody blocking studies indicated that glycan(s) associated with Fap1 is involved in bacterial adhesion in vitro (28). To define the role of galT2 in bacterial adhesion, we performed adhesion assays with different concentrations of input bacterial cells. Much to our surprise, the galT2 mutant exhibited a level of adhesion similar to that of the wild-type bacteria, whereas the fap1 mutant displayed an 80% reduction in bacterial adhesion (Fig. 3). These results suggest GalT2-mediated Fap1 glycosylation does not contribute to bacterial adhesion to SHA.

FIG. 3.

Adhesion of S. parasanguinis and its mutant derivatives to an in vitro tooth model, SHA. Bars represent the binding of [³H]thymidine-labeled bacteria to SHA and are the mean of values from three independent experiments. The adhesion values of AL105 and VT1393 are presented as a relative percentage of the adhesion by the wild-type FW213 strain (3.03 × 10⁴ cpm).

The N-terminal region of Fap1 is involved in the adhesion of S. parasanguinis to SHA.

Since GalT2-mediated Fap1 glycosylation is not required for bacterial adhesion in vitro and Fap1 possesses both glycan and peptide epitopes, we hypothesize that the Fap1 polypeptide plays a predominant role in bacterial adhesion. To test this hypothesis, we expressed the N-terminal Fap1 polypeptide in a recombinant format (rFap1) and used rFap1 in adhesion blocking, displacement, and saliva overlay assays to evaluate its binding property. In adhesion blocking assays, a dose-dependent decrease in the bacterial attachment to SHA was observed when the amounts of rFap1 preincubated with SHA were increased. In contrast, preincubation of SHA with increased concentrations of BSA did not inhibit the binding of S. parasanguinis to SHA significantly (Fig. 4A). Furthermore, the addition of 100 μg of rFap1 to the bacteria that had already bound to SHA was able to displace the attached bacteria and reduced the adhesion by 50%, whereas the similar amount of BSA was not able to displace the bound bacteria (Fig. 4B). These data suggest that the N terminus of Fap1 contains an adhesive motif. To further support this, we examined the potential interaction between rFap1 and salivary proteins using a saliva overlay assay. Both rFap1 and fimbrial preparations interacted with a 53-kDa salivary protein (Fig. 4C, left and right panels, lane 1); other controls that lacked rFap1 (lane 2) and fimbrial preparation (lane 5), or primary antibodies (lanes 3 and 6) did not give rise to the same salivary protein, suggesting that this interaction is specific and that this 53-kDa salivary protein is a receptor for native Fap1. The nature of this receptor remains to be determined. Taken together, we conclude that rFap1 mediates bacterial adhesion. These results explain why the galT2 mutant still retains its ability to bind to SHA.

FIG. 4.

Involvement of the N-terminal region of the Fap1 polypeptide (rFap1) in S. parasanguinis adhesion. (A) rFap1 blocked adhesion of S. parasanguinis to SHA. An adhesion blocking assay was performed as described in Materials and Methods. Radiolabeled S. parasanguinis (1.05 × 10⁵ cpm) was incubated with the different concentrations of rFap1 or the control protein BSA; the pretreated bacteria were then incubated with SHA. Bacteria that bound and did not bind to SHA were separated and examined for radioactivity. The radioactivity associated with cells was presented as a mean percentage of the binding ± the standard deviation calculated from three independent experimental determinations. (B) rFap1 displaced prebound bacteria. A total of 100 μg of rFap1 was added to the bacteria that had already bound to SHA for 1 h, the mixture was incubated for 1 h and washed twice, and the radiolabeled bacteria that remained bound to SHA was measured. The percentage of the bound bacteria displaced by incubation with rFap1 and BSA was calculated and is presented. (C) Interaction of rFap1 and fimbrial preparation of S. parasanguinis with a 53-kDa salivary protein. An overlay of saliva blotting analysis was performed to determine the interaction of rFap1 and fimbrial preparation with salivary components. Clarified saliva were separated onto SDS-12% polyacrylamide gels and then transferred to nitrocellulose membrane blots. One blot was incubated with purified rFap1 and probed with Fap1-specific MAb E42 (left panel), whereas the other one was incubated with fimbrial preparation and probed with a fimbrial specific antibody (right panel).

Role of GalT2 in biofilm formation.

Fap1 is required for the biofilm formation of S. parasanguinis (12). However, the functional contribution of glycosylation to biofilm formation is not known. Using the glycosylation-deficient and peptide-stable mutant galT2, we examined the contribution of Fap1 glycosylation to the biofilm formation. Visual observation of the microtiter wells revealed that the biofilms formed by the galT2 mutant were much thinner than the wild-type biofilms (data not shown). The galT2 mutant displayed a 70% decrease in biofilm mass accumulation relative to that of its parent strain in a microtiter plate assay (Fig. 5). It should be noted that the wild type and the galT2 and fap1 mutants grew equally in suspension under biofilm conditions (data not shown). These results were further confirmed by CLSM studies. The wild-type S. parasanguinis biofilm displayed thicker confluent cells after 16 h. The galT2 mutant exhibited a consistent pattern of thinner biofilms, and the pixel fluorescent intensity was reduced by 65% relative to the wild-type biofilms. In contrast, the fap1 mutant displayed sporadic minimal attachment (Fig. 5). Taken together, these results suggest that Fap1 glycosylation plays an important role in biofilm formation.

FIG. 5.

Analyses of biofilm formation. (A) Microtiter plate quantitative assays. In vitro biofilm assays were performed as described in Materials and Methods. Biofilms stained by crystal violet in the microtiter wells were extracted and quantified by measuring absorbance at 562 nm. Bars represent the mean measurements from three independent experiments. (B) CLSM analyses of biofilms. Bacteria were allowed to grow on the saliva-conditioned polystyrene coverslips. Biofilm formed on the coverslips was stained and examined by CLSM. The top panels were reconstructed from a collection of 10 optical sections in the x-y plane of the wild-type FW213 and galT2-defective AL105 biofilms. The step size between each section was 1 μm. The bottom panels represent the x-z plane. Bar, 50 μm.

DISCUSSION

Fap1 is a serine-rich glycoprotein (18, 35). It is required for fimbrial and biofilm formation and mediates adhesion of S. parasanguinis FW213 to SHA, an in vitro model of teeth (37-39). Fap1-like serine-rich proteins are responsible for the binding of respective bacteria to different host components, such as human platelets and salivary components (2, 15), and represent a new family of bacterial adhesins (3, 28). We have identified a seven-gene cluster downstream of fap1, this gene cluster consisting of accessory secretion proteins and glycosyltransferases has been implicated in Fap1 glycosylation and secretion (36). In the present studies, we isolated and identified an additional gene cluster which comprises three glycosyl transferases (Gly, GalT1, and GalT2) and one nucleotide-sugar synthetase (Nss). This gene cluster is located immediately upstream of the fap1 locus and shares significant homology with other glycosyltransferase genes. Gly, Nss, and GalT1 homologues have been identified in S. gordonii and implicated in the glycosylation of Fap1-like protein, GspB (2, 31). No homology was found for GalT2 in genes that are likely modulating serine-rich glycoproteins, suggesting that GalT2 is important for Fap1 glycosylation. Biochemical analyses of the galT2 mutant demonstrated that galT2 is involved in Fap1 glycosylation. Since GalT2 shares significant homology with known glycosyltransferases and mediates Fap1 glycosylation, it is likely that GalT2 functions as a transferase to catalyze the direct transfer of some of previously identified monosaccharide residues (28) to the Fap1 substrate.

Interestingly, we found that the galT2 mutation did not inhibit bacterial adhesion in an in vitro tooth model, indicating GalT2-directed glycosylation is not critical for Fap1-mediated bacterial adhesion. In contrast, we report that the glycan-specific MAbs F51 and D10 are able to block bacterial adhesion to SHA, suggesting the role of these glycan moieties in bacterial adhesion (28). The apparent discrepancy in two studies is likely due to the nature of antibody blocking experiments. Binding of antibodies to their targeted protein(s) could alter protein conformation and mask their adhesive domains that are otherwise accessible for binding. In support of the concept that GalT2-directed Fap1 glycosylation is not required for bacterial adhesion, we demonstrated that the N-terminal portion of recombinant Fap1 interacted with a 53-kDa salivary protein and was able to block and displace bacterial adhesion. Since the recombinant protein was expressed in and purified from E. coli, it is unlikely that rFap1 is glycosylated; therefore, we conclude that the N-terminal portion of the Fap1 polypeptide is responsible for bacterial adhesion. Intriguingly, an N-terminal nonrepetitive region of the GspB polypeptide has been identified as the principal receptor for human platelets (30) despite the fact that glycosylation of GspB is implicated in the binding of S. gordonii to human platelets (1). This finding is also consistent with our conclusion that a nonrepetitive peptide region of serine-rich proteins possesses adhesive function.

Biofilm formation can be divided into three discrete steps: initial attachment of bacteria to a surface, microcolony formation, and biofilm maturation. Many bacterial surface components engaged in initial adhesion are essential for biofilm formation. Fap1 is a bacterial surface adhesin and required for biofilm formation. Inactivation of the fap1 gene completely abolishes biofilm formation (12). In the present study, we have dissected the function of glycan and peptide epitopes of Fap1 and determined that mutation of galT2 did not inhibit bacterial attachment to SHA surfaces. The galT2 mutant maintained a significant portion of attachment to the biofilm surface. This attachment is likely to be accomplished by the N-terminal recombinant Fap1 (rFap1) since rFap1 interacted with a 53-kDa salivary protein and was able to block and displace the adhesion of S. parasanguinis to SHA. Despite the successful initial attachment of the galT2 mutant, the mutant had a significant decrease in biofilm mass accumulation, suggesting the glycosylation is important for biofilm formation. In gram-negative bacteria, the production of extracellular matrix exopolysaccharides (EPS) is essential for biofilm formation. EPS production does not affect bacterial ability to initiate attachment to biofilm surfaces. However, it does affect bacterial biofilm development and biomass accumulation (8, 14, 21). The precise mechanism of how the galT2 mutant failed to develop mature biofilm is not known. The difference between the wild type and the galT2 mutant is the status of the Fap1 glycosylation. The mutant may have produced a less stable intercellular matrix for microcolony formation than the glycosylated wild-type bacteria. This concept was consistent with the fact that galT2 mutant exhibited less biofilm mass and was defective in mature biofilm formation. Glycosylated Fap1 may substitute the function of EPS to glue bacterial layers together. EPS production and biofilm formation are important virulence determinants in endocarditis mediated by many oral streptococci (20). Glycosylation of Fap1 is important in the development of biofilm and whether it plays a role in S. parasanguinis-mediated endocarditis remains to be determined.