Two Gene Determinants Are Differentially Involved in the Biogenesis of Fap1 Precursors in Streptococcus parasanguis

ABSTRACT Mature Fap1, a 200-kDa fimbria-associated adhesin, is required for fimbrial biogenesis and biofilm formation in Streptococcus parasanguis. Fap1-like proteins are found in the genomes of many streptococcal and staphylococcal species. Fap1 is a serine-rich glycoprotein modified by O-linked glycan moieties. In this study, we identified a seven-gene cluster including secY2, orf1, orf2, orf3, secA2, gtf1, and gtf2 that is localized immediately downstream of fap1. The lower G+C contents and the presence of a putative transposase element suggest that this gene cluster was horizontally transferred from other bacteria and represents a genomic island. At least two genes in this island mediated Fap1 biogenesis. Mutation of a glucosyltransferase (Gtf1) gene led to accumulation of a Fap1 precursor, which had no detectable glycan moieties. Inactivation of a gene coding for an accessory Sec protein (SecY2) resulted in expression of a distinct Fap1 precursor, which reacted with one glycan-specific Fap1 antibody but not with another glycan-specific antibody. Furthermore, partially glycosylated Fap1 was detected on the cell surface and in the culture supernatant. These data suggest that SecY2 has a role in complete glycosylation of Fap1 and imply that SecY2 is not the only translocation channel for the Fap1 precursor and that alternative secretion machinery exists. Together, Gtf1 and SecY2 are involved in biogenesis of two distinct Fap1 precursors in S. parasanguis. Discovery of the effect of an accessory Sec protein on Fap1 glycosylation suggests that Fap1 secretion and glycosylation are coupled during Fap1 biogenesis.

As a primary colonizer of the tooth surface, Streptococcus parasanguis is critical to the formation of the most complicated human biofilm, dental plaque. This complex biofilm has been implicated in development of oral diseases, such as dental caries and periodontal disease (5,15,21). The long fimbriae of S. parasanguis FW213 are major cell surface structures that mediate streptococcal adhesion to salivacoated hydroxylapatite, an in vitro model of bacterial adhesion to teeth (10,11,14).
A Fap1 glycoprotein has been identified as a major fimbrial subunit of S. parasanguis FW213 and plays an important role in bacterial adhesion (25,35,36). Mature Fap1 migrates at an apparent molecular mass of 200 kDa in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE). Sequence analyses have shown that Fap1 is comprised of two serine-rich repeat regions, an unusually long N-terminal signal sequence, and a classic C-terminal cell wall sorting motif (34). Fap1-like molecules have recently been implicated in oral colonization by Streptococcus cristatus (8), Streptococcus salivarius (19), and Streptococcus gordonii (16,26), as well as in binding to human platelets by S. gordonii and Streptococcus sanguis (4,23,29). Genes similar to fap1 have also been found in the genomes of other human pathogens, including Streptococcus pneumoniae, Staphylococcus aureus, Staphylococcus epidermidis, and Streptococcus agalactiae (http://www.ncbi.nlm.nih.gov/sutils/genom table. cgi?), indicating that Fap1 homologues may have a role in colonization and pathogenesis.
A majority of the Fap1-like molecules are annotated as hypothetical proteins encoded by genes in bacterial genomes, and their functional characteristics are not well known. Fap1 is the first protein identified in this family, and studies have demonstrated that it is a glycoprotein. Monosaccharide composition analyses have shown that mature Fap1 contains a variety of monosaccharide residues, including glucose, galactose, rhamnose, N-acetylglucosamine, and N-acetylgalactosamine (25). Glycopeptides purified from pronase-digested Fap1 have the same monosaccharide composition as the intact mature protein, suggesting that a homogenous glycan chain is associated with Fap1. The four predominant amino acids (serine, valine, isoleucine, and glutamic acid) in the Fap1 repeat regions also are the predominant amino acids in the isolated glycopeptides, suggesting that there is a link between the repeat region and glycan modification. These observations, together with the finding that N-glycanase does not cleave glycans from Fap1, suggest that the glycan moiety of Fap1 is O linked within the serine-rich repeat regions (25). Recently, three Fap1 homologues, GspB and Hsa of S. gordonii and SrpA of S. sanguis, have been shown to be glycosylated as well (3,29). Several glycosyltransferase genes downstream of gspB have been found at the same locus as genes coding for an accessory Sec system (28,29). The role of these accessory Sec genes in protein secretion has been documented. However, it is not known whether they are implicated in protein glycosylation. We identified an accessory Sec protein, SecA2, during our study of Fap1 glycosylation and demonstrated that SecA2 is required for Fap1 secretion and affects the abundance of Fap1 glycosylation (6), indicating that there is a functional association between SecA2 and Fap1 glycosylation. However, the detailed mechanism of Fap1 biogenesis and what other genes are required for Fap1 glycosylation are not known. Characterization of genes involved in Fap1 biosynthesis should increase our understanding of the Fap1 glycosylation pathway. In this paper, we report identification of a genomic island for Fap1 glycosylation and secretion. Glucosyltransferase Gtf1 and an accessory Sec protein, SecY2, were involved in biogenesis of two new Fap1 precursors, preFap1A and preFap1B. Analyses of these precursors suggested that Gtf1 is required for Fap1 glycosylation, whereas SecY2 plays a role in complete glycosylation of mature Fap1.

MATERIALS AND METHODS
Bacterial strains and growth conditions. The bacterial strains and plasmids used in this study are listed in Table 1. Escherichia coli strains were grown with aeration at 37°C in Luria-Bertani broth or on Luria-Bertani agar plates supplemented with ampicillin (100 g/m), tetracycline (10 g/ml), erythromycin (300 g/ml), or kanamycin (25 g/ml) as needed. S. parasanguis strains were grown in Todd-Hewitt broth or on Todd-Hewitt agar plates supplemented with kanamycin (125 g/ml), erythromycin (10 g/ml), or tetracycline (12.5 g/ml) as necessary and incubated statically in the presence of 5% CO 2 at 37°C.
Antibodies. Monoclonal antibodies (MAbs) F51, E42, and D10 were produced using hybridoma technology (10). MAb E42 is specific for the unique peptides in the N terminus of the Fap1 protein, whereas MAbs F51 and D10 are specific for glycan moieties of Fap1 (25). A polyclonal antiserum directed against the 20-kDa thiol peroxidase (Tpx) of S. parasanguis was prepared as described previously (13).
A polyclonal antibody reacting with Fap1 repetitive regions was generated as follows. A synthetic peptide representing the epitopes available in the repeat sequences of the Fap1 protein (NH 2 -[GC]VSESVSESISESVSESVSES-OH) was conjugated to the carrier protein keyhole limpet hemocyanin, and the conjugated peptide was used to immunize two rabbits (BioSynthesis Incorporated, Lewisville, TX). A polyclonal antiserum designated rpAB (repeat antibody) was obtained from three successive bleeds at 6, 8, and 10 weeks after immunization, and the identity of rpAB was confirmed on the basis of its specific reactivity with the repeat region of Fap1 in Western blot analyses.
Preparation of whole-cell protein extracts and culture supernatants. Wholecell protein extracts were prepared from 10-ml bacterial cultures grown to the early logarithmic phase. Harvested bacterial cells were resuspended in 0.2 ml of 20 mM sodium phosphate buffer (pH 6.0) containing lysozyme (2.5 mg/ml) and mutanolysin (300 U/ml) and then incubated at 37°C for 30 min. SDS at a final concentration of 0.5% was added to the treated bacterial cells, which were lysed for an additional 10 min at 37°C. The cell lysates were mixed with the glass beads (150 to 212 m), vortexed at the highest speed for 1 min, and then centrifuged at 3,000 ϫ g to remove cell debris. The supernatants were triturated to shear the DNA using a 23-gauge needle attached to a 3-ml syringe and then centrifuged at 10,000 ϫ g for 10 min. The resulting supernatants were used as whole-cell protein extracts.
Culture supernatant fractions were harvested from media in which bacteria were grown to the early logarithmic phase. The collected culture supernatants were prepared as described previously (17). Briefly, 0.5 ml of culture supernatant was collected from 0.6 ml of a bacterial culture by centrifugation. One milliliter of ice-cold ethanol was added to the 0.5 ml of clarified culture supernatant and incubated at Ϫ80°C for 5 min. The mixture was then centrifuged at 10,000 ϫ g for 10 min to precipitate the proteins. The protein pellets were resuspended in SDS sample buffer and subjected to SDS-PAGE analyses as culture supernatant fractions.
Whole-cell enzyme-linked immunosorbent assay. A whole-cell enzyme-linked immunosorbent assay (BactELISA) was performed as described previously (9). Streptococcal cells were dried onto 96-well plates and incubated with 1% bovine serum albumin in phosphate-buffered saline (PBS) for 1 h at room temperature. The treated plates were washed three times with PBS prior to incubation with primary antibodies for 1 h at room temperature. Horseradish peroxidase-conjugated secondary antibody was used to detect the binding of primary antibodies after the wells were rinsed using PBS with 0.1% Tween 20. The activities of conjugated horseradish peroxidase were quantified with the o-phenylenediamine substrate at 490 nm using a 96-well EL311 microplate autoreader (Biotek Instruments, Winooski, VT). The antibody reactivities with each strain were determined in triplicate in three independent experiments. SDS-PAGE, protein staining, and Western blot analyses. SDS-PAGE analyses were carried out using commercially available precast 4 to 12% gradient gels (BMA, Rockland, ME). Protein samples were boiled in sample buffer (0.0625 M Tris [pH 6.8], 2% SDS, 10% glycerol, 0.01% bromophenol blue) for 10 min prior to loading. All gels were electrophoresed at 125 V until the bromophenol blue reached the bottom of the gel. Total-protein staining was carried out using Pro-Q Emerald 300 SYPRO Ruby protein gel stain kits according to the manufacturer's instructions (Molecular Probes, Eugene, OR). For the Western blotting analyses, proteins were transferred from gels to nitrocellulose membranes. Nonspecific binding sites on nitrocellulose blots were blocked by incubating the blots with 5% nonfat dry milk in PBS for 1 h prior to probing with various primary antibodies diluted in PBS with 0.1% Tween 20. Horseradish peroxidaseconjugated secondary antibody and luminol-based reagents (ECL Western blot detection reagents; Amersham Pharmacia Biotech, Little Chalfont, Buckinghamshire, England) were used to detect protein bands that reacted specifically with primary antibodies.
Carbohydrate staining with the ECL glycoprotein detection system. Carbohydrate staining of Fap1 species was performed as described in the manufacturer's instructions (ECL glycoprotein detection system; Amersham Bioscience, Rockford, IL). In brief, proteins were prepared from bacterial culture supernatants as described above. The protein samples were electrophoresed on 4 to 12% SDSpolyacrylamide gels and transferred to nitrocellulose membranes. The membranes were treated with 10 mM sodium metaperiodate in 100 mM acetate buffer (pH 5.5) for 20 min, washed with the acetate buffer twice, and then incubated with 40 l of 0.125 mM biotin hydrazide in 20 ml of 100 mM acetate buffer (pH 5.5). The rinsed membranes were incubated with 5% membrane blocking agent (Amersham Bioscience, Rockford, IL) to block nonspecific binding. Horseradish peroxidase-conjugated streptavidin at a 1:1,000 dilution was used to stain the membranes, and the stained membranes were processed for ECL detection.
DNA manipulations. DNA-modifying enzymes, including restriction endonucleases, T4 DNA ligase, and T4 DNA polymerase, were purchased from Invitrogen (San Diego, CA). Plasmid DNA was isolated using a Qiaprep Spin Miniprep kit (QIAGEN Inc., Santa Clarita, CA). Recombinant plasmids were introduced into E. coli by electroporation using a Gene Pulser II apparatus (Bio-Rad, Hercules, CA). Genomic DNA from S. parasanguis strains was isolated using a PureGene gram-positive DNA kit (Gentra Systems, Minneapolis, MN). Southern blot analyses were performed as described previously using Hybond-N nitrocellulose membranes (36), probes were labeled with the ECL

Isolation and identification of a glucosyltransferase gene.
A search of the S. gordonii, S. pneumoniae, and S. aureus genomes identified a gene encoding a highly conserved glucosyltransferase (Gtf); this gene is located adjacent to a gene coding for a serine-rich high-molecular-mass (HMM) Fap1-like protein. A pair of degenerate oligonucleotide primers based on conserved Gtf domains were designed. The forward primer, GGNGTNGARTAYGCNCARGC, corresponded to a conserved amino acid domain from amino acid position 16 to amino acid position 22, GVEYAQA. The reverse primer, CCARTCDATRTGYTT YTC, matched the complementary strand coding for another conserved amino acid region, EKHIDW (positions 332 to 337). This primer pair was used to amplify a gtf homologue from S. parasanguis genomic DNA by PCR. DNA amplification was performed with a Techne thermocycler (Techne Inc., Princeton, NJ). One microgram of S. parasanguis genomic DNA was used as the PCR template. The PCR conditions were as follows: one cycle at 94°C for 3 min, followed by 30 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 45 s. The amplification reaction was catalyzed with Taq DNA polymerase (Invitrogen, San Diego, CA) and was completed with an additional cycle consisting of 72°C for 10 min. An expected 966-bp PCR product was generated in the PCR and isolated from a 1.2% of agarose gel with a QIAquick gel extraction kit (QIAGEN Inc., Santa Clarita, CA). The gel-purified PCR product was ligated into the pGEM-T Easy vector (Promega, Madison, WI) by using T4 DNA polymerase; the ligation mixture was transformed into E. coli. The cloned PCR product was verified by EcoRI restriction endonuclease digestion and sequencing of both strands with universal primers T7 and SP6 (Promega, Madison, WI). Plasmid pVT1579 with a correct PCR product was isolated and used in this study.
Isolation of the secY2-gtf2 locus. The gtf1 PCR product was used to probe an S. parasanguis genomic library constructed in the EMBL3 lambda vector and expressed in bacteriophages as described previously (36), with the following modifications. The probe was labeled with the ECL direct nucleic acid labeling system, and detection was performed with the ECL detection system (Amersham Life Science, Little Chalfont, Buckinghamshire, England). Recombinant bacteriophage plaques reacting with the gtf1 probe were isolated and purified by three additional rounds of screening. Isolated phage plaques were amplified and examined to determine their reactivities with the gtf1 probe. A gtf1-positive bacteriophage (VT1584) was selected. The phage DNA was isolated by following the manufacturer's instructions (Promega, Madison, WI) and was directly used as a template for nucleotide sequencing.
Nucleotide sequencing and sequence analyses. Nucleotide sequencing analyses were performed at the DNA analysis facility at the Vermont Cancer Center at the University of Vermont using the dideoxy chain termination method and an ABI 1372A DNA sequencer. The nucleotide sequence was determined for both strands. Nucleotide and amino acid sequence searches and analyses were carried out using the BLASTX and BLASTp programs at the National Center for Biotechnology Information (1). Total 22,761-bp DNA sequences of S. parasanguis were compiled from fimA loci and other 20 sequenced genes of S. parasanguis and used for GϩC content analysis with DNAStar software.
Construction of a gtf1 mutant. An inverse PCR strategy (22) was used to construct an StuI site in pVT1579 to facilitate insertion of a nonpolar kanamycin resistance cassette, aphA3 (a gift from Allen Honeyman) (18) into the gtf1 fragment. In brief, primers with embedded StuI sites (GCAGAGGCCTCTCA TTTTGAGTGATTTATG and CGACAGGCCTACCCTCATAACGAATAC TTC; StuI sites underlined) were used to inverse PCR amplify a 3.90-kb fragment from pVT1579. The amplified fragment was digested with restriction endonuclease StuI and self-ligated to generate pVT1580, in which a 48-bp internal fragment of gtf1 was deleted. pVT1580 was then digested with StuI and gel purified. StuI-linearized and gel-purified plasmid pVT1580 was fused with a nonpolar kanamycin resistance cassette (aphA3) to form pVT1581, in which gtf1 was interrupted by insertion of aphA3. The mutated gtf1 allele was excised from pVT1581 by EcoRI digestion and then ligated into S. parasanguis suicide vector pSF143 (31) to create pVT1582. This plasmid with the mutated gtf1 allele was used to transform S. parasanguis FW213 via electroporation by the method described previously (12). Inactivation of the genomic copy of gtf1 by a doublecrossover recombination event was confirmed by PCR and Southern blot hybridization. A confirmed mutant, VT1583, was used in subsequent studies.
Construction of an isogenic secY2 mutant. The secY2 gene and its flanking regions were amplified by PCR using forward primer GAACACTTCCGAATA CTGGTG and reverse primer GTGATATTCTGGATATCATC. The amplified PCR fragment was subcloned into the pGEM-T Easy vector (Promega, Madison, WI) to generate pVT1585. An inverse PCR approach was used to delete a 700-bp secY2 internal fragment from pVT1585. In brief, forward and reverse primers with embedded HindIII sites, GCGGATGGAAGCTTCAGTATGTTGTAAAC TCATTTGCGC and TCGCGCGCAAGCTTCCAGCTGCCAATCAATT TAAA (HindIII sites underlined), were used to amplify a 4.0-kb fragment from pVT1585. The PCR product was digested with HindIII and self-ligated to construct pVT1586, in which a 700-bp secY2 internal fragment was deleted. HindIIIdigested pVT1586 was ligated in frame with an 830-bp aphA3 nonpolar kanamycin resistance cassette to create a secY2-aphA3 in-frame fusion construct, pVT1587. The correct in-frame fusion was confirmed by sequencing. pVT1587 was digested with restriction endonucleases SphI and SalI to release the secY2-aphA3 fusion fragment. The released fragment was isolated by gel electrophoresis and ligated with SphI-and SalI-treated pSF143 to construct pVT1588. The plasmid with the mutated secY2 gene was then delivered into S. parasanguis via electroporation. Putative mutants were selected based on their abilities to resist kanamycin and their susceptibilities to tetracycline. Replacement of the wild-type secY2 gene with the mutated allele by a double-crossover recombination event was confirmed by PCR and Southern blot hybridization. S. parasanguis strain VT1589 with the confirmed mutant secY2 allele was used in this study.
Complementation of the gtf1 and secY2 mutants. The full-length gtf1 gene was amplified with primers Gtf-SalI-FF (TGACGTCGACATGACAATCTATAAT ATTAATTTAGGG; the SalI site is underlined) and Gtf-KpnI-RR (TGACGG TACCCATCTCCTCAATAAAATCTTTCC; the KpnI site is underlined). The full-length secY2 gene, including its own promoter, was amplified using two primers with embedded BamHI sites, GATTGGATCCTGGATTAGTTGGAT TAAGTGG and GGATCCTTCTTGTCCCTGCAAGAAC (BamHI sites underlined), and the PCR products were digested with SalI and KpnI or BamHI and then ligated into the corresponding restriction enzyme-treated E. coli and streptococcal shuttle vectors (pVPT and VA838, respectively) (20). The ligation mixtures were then transformed into E. coli DH10B cells. The desired plasmids, pAL50 and pVT1590, were isolated from putative transformants and confirmed by restriction digestion and DNA sequencing. pVPT and pAL50 were transformed into the gtf1 mutant to generate AL51 and AL50, respectively, whereas pVA838 and pVT1590 were transformed into the secY2 mutant to generate VT1591 and VT1590, respectively. The ability of the cloned gtf1 and secY2 genes to restore mature Fap1 expression in AL50 and VT1590 was examined using BactELISA or Western blot analyses with Fap1-specific antibodies.
Subcellular localization of Fap1. Exponentially grown S. parasanguis wild-type strain FW213 cells and secY2 mutant cells were harvested by centrifugation at 5,000 ϫ g at 4°C for 10 min. The cell pellets collected from 1-ml cultures were washed with PBS buffer once and then resuspended in 45 l spheroplasting buffer (10 mM Tris HCl, 2 mM MgCl 2 , 26% raffinose, 1 mM phenylmethylsulfonyl fluoride) along with 5 l of a freshly prepared amidase solution. The suspensions were incubated on ice with occasional mixing for 20 min. The spheroplasts were harvested by centrifugation. The resultant supernatants were transferred to clean tubes and used as cell wall-associated protein fractions. Each pellet was dissolved in 50 l spheroplasting buffer and used as a cytoplasmic protein fraction.
Immune electron microscopy studies. S. parasanguis FW213 strains grown to an A 470 of 0.5 were harvested and washed in 0.1 M phosphate buffer. The washed cell pellets were fixed in 3% paraformaldehyde, embedded in agarose, dehydrated, and polymerized in a TTP010 low-temperature polymerization apparatus (Balzer Union). Ultrathin sections were cut with a Reichert Jung Ultracut E microtome (Leica) and collected on Formvar-coated nickel grids. Sections on the grids were incubated with 0.5% ovalbumin in PBS for 30 min, with MAb D10 overnight at 4°C, and with 10-nm gold particle-conjugated protein A for 30 min. After incubation, the grids were stained with uranyl acetate and lead citrate. The samples on the grids were visualized and photographed with a JEM-1210 transmission electron microscope.

Identification and isolation of a glucosyltransferase gene.
Protein glycosylation is initiated by transferring sugar residues to the targeted polypeptide backbone. This modification step is controlled by glycosyltransferases. As Fap1 is a glycoprotein, we hypothesized that certain glycosyltransferases are required for Fap1 glycosylation. By searching bacterial genomic databases, a conserved gene coding for glucosyltransferase (Gtf) was identified in the genomes of S. gordonii (http://www.tigr .org/tdb/mdb/mdbinprogress.html), S. pneumoniae (32), and S. aureus (2). The gtf gene is localized downstream of fap1 ho-mologues in these species. We reasoned that such a gtf homologous gene is also localized adjacent to the fap1 open reading frame (ORF). Based on two well-conserved domains of Gtf from S. pneumoniae, S. gordonii, and S. aureus (310 amino acids apart), we designed degenerate primers and amplified a 966-bp fragment from the S. parasanguis FW213 chromosomal DNA. Determination of the DNA nucleotide sequence of this PCR fragment revealed a partial ORF encoding a glucosyltransferase-like protein. We designated this ORF Gtf1. Gtf1 exhibits 50%, 52%, and 46% identity, at the amino acid level, with the corresponding regions of Gtf from S. gordonii (4), S. pneumoniae (32), and S. aureus (2), respectively.
Identification of a genomic island localized downstream of fap1. No genome sequence is available for S. parasanguis. In an effort to determine whether additional genes are present in the gtf1 locus, we screened an S. parasanguis genomic library constructed in lambda bacteriophages and identified a 14-kb gtf1positive DNA fragment from the library. Analyses of the DNA sequence revealed that the 14-kb genomic fragment contained the entire gtf1 coding sequence, as well as regions upstream and downstream of gtf1 (Fig. 1).
The region flanking the gtf1 gene contained six additional ORFs (Fig. 1). Two ORFs, designated secY2 and secA2, encode proteins that are homologous to components of the general secretion pathway. Three genes, designated orf1, orf2, and orf3, exhibit homology to the asp1, asp2, and asp3 genes implicated in GspB export in S. gordonii, whereas gtf2 is similar to a newly recognized glycosyltransferase gene, gtfB (29).
In addition, a gene coding for a putative transposase element was found downstream of gtf2 and was transcribed in the direction opposite the direction of secY2-gtf2 locus transcription (Fig. 1). Furthermore, the GϩC content of the entire region, including fap1 and secY2-gtf2, was calculated to be 35.6%, which is significantly lower than the GϩC content of other known genes of S. parasanguis (44.8%) (22,761 bp of DNA compiled from fimA loci and 20 other genes available for S. parasanguis in the GenBank database). These data suggest that the gene cluster at the fap1 locus was acquired and disseminated via horizontal transfer. Therefore, we designated the fap1-secY2-gtf2 region a genomic island.
Gtf1 is required for Fap1 glycosylation. Genes coding for glucosyltransferase are conserved among a variety of streptococcal and staphylococcal species possessing Fap1-like proteins (29). Construction of a defined gtf1 mutant should help determine the function of gtf1 in Fap1 glycosylation. A gtf1 mutant was constructed by insertion of a nonpolar kanamycin resistance cassette (13) into the wild-type gtf1 allele. A reverse transcription-PCR experiment was performed to determine the effect of gtf1 mutation on expression of gtf2, a downstream gene. Similar levels of gtf2 expression were observed in both wild-type and gtf1 mutant strains (data not shown), indicating that insertional inactivation of gtf1 by the kanamycin resistance cassette did not have a polar effect on expression of the downstream gene.
To examine the effect of the gtf1 mutation on glycosylation of Fap1, two glycan-specific Fap1 antibodies (F51 and D10) and one peptide-specific antibody (E42) were used to monitor Fap1 glycosylation and production in the gtf1 mutant by BactE-LISA. The gtf1 mutant did not react with the glycan-specific   genes, secY2, orf1, orf2, orf3, secA2, gtf1, gtf2, and tn.

VOL. 189, 2007
Fap1 GLYCOSYLATION BY Gtf1 AND SecY2 1393 antibodies F51 and D10 but reacted with the peptide-specific antibody E42 (Fig. 2), suggesting that the gtf1 mutant has a defect in glycosylation of Fap1. Moreover, the gtf1 mutant displayed elevated reactivity with rpAB, an antibody specific for repetitive sequences of Fap1, indicating that the gtf1 mutant has more rpAB epitopes exposed than wild-type bacteria have exposed.
To further characterize the gtf1 mutant, expression of Fap1 was determined by Western blot analyses. The wild-type strain expressed a mature 200-kDa Fap1 when it was probed with the peptide-specific Fap1 antibody E42 (Fig. 3A, lane 1). The gtf1 mutant did not express 200-kDa mature Fap1; instead, it expressed a new HMM protein that migrated slower than the mature Fap1 (Fig. 3A, lane 2). We designated this band Fap1 precursor A (preFap1A). Mature Fap1 (200 kDa) reacted with monoclonal antibodies E42, D10, and F51 (Fig. 3A, lanes 1, 3,  and 5). However, preFap1A reacted only with peptide-specific Fap1 antibody E42 (Fig. 3A, lanes 2, 4, and 6), indicating that preFap1A lacks detectable glycosylation. To confirm that the observed phenotypes can be attributed to gtf1, we performed a complementation experiment. Introduction of the full-length gtf1 gene into the gtf1 mutant restored mature Fap1 expression (Fig. 3B, lane 5), whereas transformation of the gtf1 mutant with an empty vector did not change the Fap1 precursor (Fig.  3B, lane 4), demonstrating that gtf1 is required for Fap1 glycosylation. To further determine the glycosylation nature of preFap1A, we performed carbohydrate staining using culture supernatant samples prepared from gtf1 mutant and wild-type cells that produced preFap1A and mature 200-kDa Fap1, respectively. The preFap1A reacted only with peptide-specific monoclonal antibody E42 (Fig. 3C, lane 1), and it did not react with the carbohydrate-staining reagent (Fig. 3C, lane 3). In contrast, mature Fap1 reacted with both E42 and the carbohydrate-staining reagent (Fig. 3C, lanes 2 and 4), suggesting that preFap1A is not glycosylated and may represent the polypeptide backbone of Fap1 prior to any detectable glycosylation.
secY2 gene product is required for complete glycosylation of mature Fap1. A defined secY2 mutant was constructed and utilized to determine the function of secY2 in Fap1 glycosylation and secretion. As a SecY2 homologue has been implicated in protein secretion (4), we first examined protein secretion profiles of the secY2 mutant. Culture supernatant fractions representative of secreted proteins were prepared and stained to detect the total proteins; no apparent difference in the number or relative intensity of protein bands was observed between the wild-type strain and the secY2 mutant strain (Fig.  4A), suggesting that SecY2 did not affect protein secretion in general. Expression of Fap1 was then examined by immunological assays using Fap1-specific antibodies. The surface expression of Fap1 by the secY2 mutant was first analyzed by performing a BactELISA (Fig. 4B). The wild-type strain re- acted with all peptide-and glycan-specific Fap1 antibodies. The secY2 mutant reacted with peptide-specific antibody E42 and one glycan-specific antibody, D10, but did not react with the other glycan-specific antibody, F51, suggesting that the secY2 mutant expressed a partially glycosylated Fap1. Interestingly, the binding of the E42 antibody by the secY2 mutant was reduced compared with the binding by the wild-type strain, but the binding of the D10 antibody was greater than the binding by the wild-type strain. To confirm that the observed phenotypes were directly linked to mutation of secY2, we performed a complementation experiment using full-length secY2. The F51 reactivity was restored when the secY2 mutant was supplemented in trans with pVA838::secY2, whereas complementation of the secY2 mutant with the pVA838 vector alone did not restore the F51 reactivity (Fig. 4B).
To define the molecular identity of this D10-positive and F51-negative Fap1 species, we examined proteins produced by the secY2 mutant using immunoblot analyses. A peptide-specific antibody, E42, and one glycan-specific antibody, D10, both recognized an HMM protein in the cell extracts of the secY2 mutant (Fig. 4C, lanes 2 and 5). The other glycan-specific antibody, F51, did not react with this HMM Fap1 species (Fig.  4C, lane 8). In contrast, all the monoclonal antibodies reacted with mature Fap1 at the position generated by the wild-type bacteria (200 kDa) (Fig. 4C, lanes 1, 4, and 7). Together, these results indicate that the HMM Fap1 species is only partially glycosylated by D10-reactive epitopes. A complementation experiment showed that the full-length secY2 gene on plasmid pVA838 was able to restore expression of mature and fully glycosylated 200-kDa Fap1 (Fig. 4C, lanes 3, 6, and 9). To further characterize the glycosylation nature of this HMM Fap1 species, we used the ECL glycoprotein detection system to stain the protein. As shown in Fig. 4D, both mature Fap1 and the HMM Fap1 species reacted strongly with the carbohydrate-staining reagent, indicating that the HMM Fap1 species was still glycosylated even though it had lost F51-reactive epitopes. To determine if this HMM Fap1 migrated differently than the previously identified preFap1A, we performed Western blot analyses with both gtf1 and secY2 mutants. Apparently, the HMM Fap1 protein generated by the secY2 mutant migrated slower than preFap1A generated by the gtf1 mutant migrated (Fig. 4E, lanes 3 and 4). Therefore, this new form of Fap1 is distinct from preFap1A produced by the gtf1 mutant, so we designated it preFap1B. Production of partially glycosylated preFap1B by the secY2 mutant indicates that SecY2 plays a role in Fap1 glycosylation.
It is intriguing that mutation of secY2, a putative secretory protein, did not eliminate preFap1B secretion but rather affected the glycosylation of Fap1. It is reasonable to question whether the presence of preFap1B on the cell surface and in the culture supernatant was due to cell lysis during experimental processes. To investigate this possibility, we performed enzyme-linked immunosorbent assays with whole-cell and cytoplasmic fractions using a cytoplasmic-specific marker, thiol peroxidase (Tpx) of S. parasanguis (6,13). The signal detected on the cell surfaces of all strains, including the wild type and secY2 and fap1 mutants, was not any stronger than the signal observed for the Tpx mutant (Fig. 5A), suggesting that no cytoplasmic proteins were present on the cell surface. Furthermore, no Tpx expression was detected in the culture superna-tants and cell wall-associated fractions (Fig. 5B, bottom panel, lanes 1, 3, 4, and 6), whereas Tpx expression was readily detected in the cytoplasmic fractions of the wild-type and secY2 strains (Fig. 5B, lanes 2 and 5). Therefore, we suggest that the cell surface and culture supernatant fractions of the secY2 mutant were not contaminated with cytoplasmic components and that the cell surface localization of preFap1B is supported by a SecY2-independent mechanism.
To further characterize the distribution of preFap1B, we determined the subcellular localization of Fap1. Mature Fap1 produced by wild-type bacteria was detected in culture supernatants, cytoplasmic fractions, and cell wall-associated fractions (Fig. 5B, lanes 1, 2, and 3); a similar distribution pattern was observed for the secY2 mutant (Fig. 5B, lanes 4, 5, and 6), and detection of Tpx in cytoplasmic fractions but not in other fractions (Fig. 5B, lanes 3 and 5) confirmed the purity of these cell fractions. These data indicated that preFap1B was distributed on the cell surface. To substantiate the surface localization of the D10-reactive Fap1 species, we performed immune electron microscopy studies using the D10 antibody. In wildtype cells, gold particles representative of D10-reactive Fap1 species were distributed around the cell surfaces, as well as in cytoplasmic regions (Fig. 6A). In the secY2 mutant, a considerable amount of D10-reactive epitopes was localized to a surface region where two cells were connected. A significant portion of D10-reactive epitopes was retained in cytoplasmic regions (Fig. 6B). No gold particles were evident in a negative control (a fap1 mutant), indicating the specificity of the labeling.

DISCUSSION
Mature Fap1 is characterized as a glycosylated protein with an apparent molecular mass of 200 kDa (36). The structural organization of Fap1 is remarkably similar to that of a growing family of serine-rich HMM cell surface proteins encoded by streptococcal and staphylococcal genomes, including plateletbinding proteins Hsa and GspB of S. gordonii (3,26) and SrpA of S. sanguis SK36 (22), as well as the virulence factor Srr-2 of serotype III S. agalactiae (24). In an effort to determine the molecular mechanisms involved in Fap1 glycosylation, we screened an S. parasanguis lambda genomic library and identified a gene cluster consisting of seven genes, secY2, orf1 to orf3, secA2, gtf1, and gtf2. We found that a putative transposase element was associated with the secY2-gtf2 locus, and the GϩC content of the entire locus was significantly lower than that of other known genes identified for S. parasanguis (35.6% versus 44.8%). These features are hallmarks of mobile elements in bacteria, suggesting that the gene cluster was acquired and disseminated via horizontal transfer. Therefore, the fap1-secY2-gtf2 locus may represent a genomic island. The organization of the secY2-gtf2 locus is highly conserved in streptococci and staphylococci (27). Mutagenesis of genes in this locus revealed that three functionally distinct gene families are involved in Fap1 glycosylation and secretion. SecA2 is required for export of Fap1 to the cell surface and modulates Fap1 glycosylation (6), Gtf1 is essential for Fap1 glycosylation, and SecY2 contributes to the complete glycosylation of Fap1. The similar gene organization of S. gordonii has been identified and attributed to glycosylation and secretion of GspB (3,26). However, the role of the accessory Sec genes in protein glycosylation is not known. The principal contribution of the current study is the discovery of two new forms of Fap1, precursors A and B, and the association of Gtf1 and SecY2 with the biogenesis of these precursors.
Gtf1 exhibited a high level of similarity with glucosyltransferases from S. gordonii and many other streptococci and staphylococci. Inactivation of gtf1 resulted in expression of HMM preFap1A, which was recognized only by a peptidespecific antibody and not by any glycan-specific antibodies. While the antibody specific for nonrepetitive Fap1 peptides was able to detect the HMM protein, the detection sensitivity was greatly enhanced by use of an antibody specific for the serine-rich repeat peptides (rpAB) (Fig. 2). The increased sensitivity was likely due to the complete exposure of multiple antibody binding sites (serine-rich repeat regions) in the Fap1 molecule when Fap1 was not glycosylated. This finding is consistent with the concept that serine-rich repeat regions are not glycosylated by a gtf1 mutant. Moreover, preFap1A did not react with a carbohydrate-staining reagent, further supporting the notion that preFap1A is not glycosylated. Hence, we concluded that Gtf1 is required for glycosylation of Fap1. Gtf is also essential for glycosylation of GspB by S. gordonii. However, inactivation of gtf in S. gordonii completely mitigates the expression of GspB (29), and no unglycosylated GspB was evident. On the other hand, we identified a nonglycosylated Fap1 precursor, preFap1A. Discovery of preFap1A should facilitate design of in vitro Fap1 glycosylation reactions as this Fap1 species can be a useful native substrate for Gtf1-mediated Fap1 glycosylation.
Two genes of S. parasanguis, designated secA2 and secY2, exhibited significant homology with genes coding for putative bacterial accessory secretion components of gram-positive bacteria, including S. gordonii, S. sanguis, S. agalactiae, S. pneumoniae, S. aureus, and S. epidermidis. SecA2 and SecY2 have been implicated in secretion of the Fap1-like protein GspB of S. gordonii (4). We determined previously that SecA2 not only is important in Fap1 secretion but also is critical for maintaining the Fap1 glycosylation level (6). Here we demonstrated that SecY2, the other secretion protein, is required for complete glycosylation and maturation of Fap1. Mutation of secY2 led to loss of expression of an F51-reactive glycan-specific epitope(s), as well as accumulation of a new HMM Fap1 species, preFap1B. preFap1B was only partially glycosylated by one glycan-specific antibody, D10. Expression of this new Fap1 species by the secY2 mutant highlights the role of SecY2 in Fap1 glycosylation. The mechanism of SecY2-mediated glycosylation is not known. It is unlikely that SecY2 is directly involved in glycosylation as a glycosyltransferase. It is possible that SecY2 mediates the activity and specificity of glycosyltransferases that are important for complete Fap1 glycosylation. In fact, such an export-mediated glycosylation process has FIG. 6. Localization of D10-reactive epitopes by the secY2 mutant. S. parasanguis cells prepared from the wild type (A) and secY2 (B) and fap1 (C) mutants were sectioned and probed with MAb D10. Protein A conjugated with 10-nm gold particles was used to determine the Fap1 localization. Arrows indicate gold particles. been proposed for Mycobacterium and eukaryotic cells. Manipulation of the Sec-dependent leader sequence alters protein mannosylation in Mycobacterium (33). While preFap1B was accumulated by the SecY2 mutant, it was barely detected in wild-type cells. The difficulty in detecting preFap1B in wildtype cells also indicates that completion of Fap1 glycosylation is relatively efficient in normal physiological conditions. preFap1B represents another intermediate in the biogenesis of mature Fap1.
Surprisingly, preFap1B was evident on the cell surface and was also found in the culture supernatant of the secY2 mutant. These findings differ significantly from the findings described previously for the function of SecY2 in S. gordonii. In S. gordonii, inactivation of secY2 does not alter glycosylation of GspB but does affect secretion of mature GspB (3). Given the perception and previous experimental finding that SecY2 is a primary translocation channel for export of Fap1-like proteins to the cell surface, it is intriguing that preFap1B exposed on the cell surface was identified, while SecY2 was absent. One possible explanation for this result is that the cell integrity is compromised when SecY2 is disrupted. However, neither the cell surface nor the culture supernatant of the secY2 mutant was contaminated by a cytoplasm-specific protein, Tpx, arguing strongly that the cell integrity was maintained in the secY2 mutant. Furthermore, an HMM GspB precursor was present in the cell wall-associated fractions when putative homologues of SecE and SecG of S. gordonii were inhibited, indicating that the accessory secretion components are not essential for export of the GspB precursor as well (30). Secretion of partially glycosylated Fap1 and Fap1-like species can be completed by an unknown SecY2-independent mechanism. Interestingly, the binding of the E42 antibody by the secY2 mutant appeared to be reduced in comparison with the binding by the wild-type strain; this indicates that translocation of the Fap1 precursor by an alternative pathway is not optimal when the SecY2 system is disrupted and that SecY2 may be a primary efficient translocation system for mature Fap1. On the other hand, the inability to detect mature Fap1 in the secY2 mutant suggests that the presence of SecY2 is required for expression of mature and fully glycosylated Fap1. We therefore concluded that SecY2 is required for complete glycosylation of mature Fap1 and that secretion of preFap1B is carried out by an inefficient SecY2-independent mechanism.
In summary, we identified gtf1 and secY2, genes that are responsible for the production of two new Fap1 species. Demonstration of the effect of the accessory secretory component SecY2 on Fap1 glycosylation indicated that glycosylation and secretion of Fap1 are coupled.