INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Dental caries is a transmissible infectious disease in which mutans streptococci (MS) play the major role. Streptococcus mutans, the primary etiological agent, possesses several virulence factors that allow it to accumulate within the dental biofilm and to produce and tolerate the acids that cause carious lesions. Biofilm development occurs in two distinct phases; during the first, bacterial surface proteins interact with host or bacterial products adsorbed on the tooth surface. In the second phase, a biofilm forms as bacteria accumulate by aggregation with the same or other species and produce an extracellular polysaccharide matrix (13). Bacterial components associated with the accumulation phase of MS include glucosyltransferases, their glucan products, and proteins that bind glucan. At least three S. mutans glucan binding proteins (Gbp) have been identified: GbpA (23), GbpB (26), and GbpC (24). After cloning and sequencing, the gbpA gene product was found to share homology with the putative glucan binding domain of glucosyltransferase (2) and the gbpA gene was found to encode a constitutively expressed secreted protein (1, 2). Cell surface-associated GbpC was related to the Spa family of streptococcal proteins and was expressed only during conditions of stress (24). GbpB was immunologically distinct from other Gbps expressed by S. mutans and Streptococcus sobrinus and also differed in size and purification properties (26).

For successful colonization of the oral cavity, MS need nonshedding tooth surfaces; thus, the period of highest colonization occurs from approximately 18 to 30 months of age, coinciding with the eruption of primary molars (3, 12). At this age, the mucosal immune system is mature enough to respond to immune interventions against infecting MS (27) and MS colonization of young children results in a measurable immunoglobulin A response to GbpB (27). Furthermore, in experimental infection of rats, systemic or mucosal immunization with GbpB induced protective immunity to dental caries, indicating that GbpB may be an important target for the development of caries vaccines (29). Preliminary studies have shown that GbpB is expressed in all laboratory and clinical S. mutans strains tested so far (D. J. Smith, W. F. King, and M. A. Taubman, J. Dent. Res. 74:123, 1995). However, the biological function of GbpB and its role in the virulence of S. mutans are still unclear. The cloning and sequencing of gbpB were briefly reported previously (S. Jin, M. J. Duncan, M. A. Taubman, and D. J. Smith, J. Dent. Res. 79:224, 2000). The protein shows homology to a putative peptidoglycan hydrolase from group B streptococcus, suggesting that GbpB plays a role in peptidoglycan biosynthesis. In this study, we examined the genotypic and protein diversity of gbpB in clinical isolates of S. mutans. We also determined the production and localization of GbpB protein in clinical isolates and laboratory strains and amounts of protein produced correlated positively with biofilm growth in an in vitro assay.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. S. mutans strain SJ32 was described previously (26, 27). Additional S. mutans strains were obtained from a subset of children derived from a previously described larger population (16, 17). Clinical strains were genotyped by arbitrarily primed PCR (15) and included 44 distinct S. mutans amplitypes and an S. sobrinus strain. Laboratory strains of S. mutans strains included in this study were UA130, UA159, T8 (kindly provided by P. W. Caufield, University of Alabama), and GS5 (kindly provided by H. K. Kuramitsu, State University of New York at Buffalo). MS strains were grown in either Todd Hewitt broth (THB) or brain heart infusion broth or plates and in chemically defined medium (COM) (31) under anaerobic conditions, 10% H₂-10% CO₂-80% N₂. Escherichia coli strains DH5 and BL21 (Novagen, Madison, Wis.) were grown in Luria Bertani (LB) broth or plates, and ampicillin (100 µg/ml) was used to select and maintain recombinant plasmids. Unless stated otherwise, all chemicals were obtained from Sigma Chemical Co., St. Louis, Mo.

DNA isolation. S. mutans chromosomal DNA was isolated using a MasterPure DNA purification kit from Epicentre Technologies (Madison, Wis.). Recombinant plasmid DNA was isolated using a Perfectprep kit from Eppendorf Scientific Inc. (Westbury, N.Y.).

S. mutans library construction. Chromosomal DNA was partially digested with Sau3AI, and 2.5- to 5.0-kb fragments were gel purified using a QIAquick gel extraction kit (QIAGEN Inc., Valencia, Calif.). Fragments were ligated to BamHI-digested, calf intestinal phosphatase-treated pUC19 and used to transform competent E. coli DH5 cells. Ampicillin-resistant recombinant clones were selected on plates of LB and ampicillin and containing X-Gal (5-bromo-4-chloro-3-indolyl--D-galactopyranoside) and IPTG (isopropyl--D-thiogalactopyranoside) to screen for insert-containing clones.

Southern hybridization. Hybridization conditions and signal development were as recommended in the Enhanced Chemiluminescence gene detection system (Amersham-Pharmacia, Piscataway, N.J.).

Cloning of gbpB. GbpB was purified by anion exchange chromatography in the presence of urea as previously described (28). Peptide sequences were obtained at the Molecular Biology Core Facilities, Dana-Farber Cancer Institute, Boston, Mass., and the Harvard Microchemistry Facility, Harvard University, Cambridge, Mass., by trypsin digestion, high-performance liquid chromatography, matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) (mass spectrometry), and Edman degradation. PCR primers derived from these peptides and used to isolate gbpB are described in Table 1. Reaction mixture volumes were 50 to 100 µl and contained 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 2.5 mM MgCl₂, 50 µM deoxynucleoside triphosphates, approximately 0.5 µM primers, 0.1 to 0.5 µg of template, and 2.5 U of AmpliTaq Gold DNA polymerase (Perkin-Elmer, Branchburg, N.J.). Thermal conditions were as follows: 10 min at 95°C; 6 cycles of 30 s at 95°C, 30 s at 40°C, and 1.5 min at 72°C with 5-s increments per cycle; 30 cycles of 30 s at 95°C, 30 s at 46°C, and 2 min at 72°C with 5-s increments per cycle; and 5 min at 72°C. PCR and specific sequencing primers were obtained from The Midland Certified Reagent Company (Midland, Tex). A Peltier Thermal Cycler model PTC-200 (MJ Research, Cambridge, Mass.) was used for PCR. Sequencing was carried out with either dRhodamine or Big Dye Terminator cycle sequencing kits (Perkin-Elmer, Foster City, Calif.) using a PE 9700 Thermocycler. Reactions were run on Perkin-Elmer ABI 377 Sequencer.

Expression of gbpB in E. coli The gbpB gene from S. mutans SJ32 genomic DNA was amplified using a 5' PCR primer with an NcoI restriction enzyme recognition site for insertion of the amplicon into an expression vector and for sequence coding for the first seven amino acids of the mature GbpB protein. The 3' primer contained an XhoI restriction site and sequence coding for the last eight amino acids of GbpB, but excluding the stop codon. The 1,212-kb PCR product was ligated to the NcoI-XhoI-digested vector pET22B and then transformed into E. coli BL21(DE3). GbpB expression in E. coli was induced with IPTG, protein extraction, and purification of the His-tag protein on nickel columns as described by the supplier (Novagen).

RFLP analysis of gbpB. gbpB genotypes were detected by restriction fragment length polymorphism (RFLP) analysis of PCR-amplified gbpB genes from 44 amplitypes of S. mutans clinical isolates. Laboratory strains SJ32, T8, UA130, UA159, and GS5 were also included in the analysis. The gbpB DNA sequence from SJ32 was aligned with that of strain UA159 (S. mutans genome database [http://www.genome.ou.edu/smutans.html]), and from the latter, primers were designed to amplify gbpB plus 162 bp of the upstream and 195 bp of the downstream sequences flanking the open reading frame (ORF) of GbpB (Table 1).

Measurement of GbpB in culture supernatants and cell extracts. After overnight growth in CDM, bacterial suspensions were adjusted to the same optical density (A₅₅₀) so that the same size inoculum was added to fresh CDM (4 ml) containing 10 µM hydrochloride 4-(2-aminoethyl)-benzolsulfonylfluoride (AEBSF) from Roche Diagnostics (Indianapolis, Ind.) to inhibit protease activity. After growth for 18 h, cells were pelleted from 2 ml of culture, and the supernatant was filtered through 0.22-µm-pore-size Spin-X filter membranes (Costar, New York, N.Y.) and immediately frozen at 70°C. Cell-associated GbpB was extracted with urea as previously described (8). For direct comparisons, culture supernatants and urea extracts were dialyzed overnight at 4°C against 0.02 M sodium phosphate buffer (PB) (pH 6.5) with 5 mM -mercaptoethanol. All samples were stored at 70°C.

Biofilm formation in microtiter plates. Biofilm formation in microtiter plates was assayed by the method described by O'Toole and Kolter (21). An aliquot from a 5-ml overnight THB culture was diluted 1:100 in fresh THB, and 200 ml was transferred to sterile polystyrene U-bottom microtiter plates (Dynatech Lab, Chantilly, Va.). Plates were incubated anaerobically for 18 h, and biofilm growth was revealed and quantified by staining with crystal violet (21). Crystal violet absorbance was determined with a plate reader at 575 nm (Dynatech, Winooski, Vt.). The absorbance (A₅₅₀) of planktonic cultures grown under the same conditions was measured to monitor growth. Biofilm formation for all strains was measured in triplicate plates. The laboratory strains S. mutans UA130 and SJ32 were also included in the same experiments.

Statistical analysis. Pearson correlation analyses were used to evaluate associations between production of GbpB, biofilm growth, and planktonic growth of cultures. To avoid bias because of GbpB production by different S. mutans amplitypes, only one of each amplitype identified per child was used after random selection.

Nucleotide sequence accession numbers. The nucleotide sequence of the gbpB gene from S. mutans strain SJ32 was assigned GenBank accession number AY046410; from strain 3VF4, AY046411; from 15JP2, AY046412; from 3SN1, AY046413; and from 5SM3, AY046414.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Cloning of gbpB. Cloning was carried out during the beginning stages of the S. mutans genome project, and before gbpB-related sequences were available. Therefore, mature GbpB was purified from culture supernatants of S. mutans SJ32 and the N-terminal and two internal peptide sequences were determined after tryptic digestion, MALDI-TOF (mass spectrometry), and cycle sequencing. The N-terminal sequences of peptide 1 (DDF?AQIASCD[A]KI[V][N][T]) and internal peptides 2 (GWFNPGSVSYIYP[L]) and 3 (LEAQSATLGQQIQTLSSSK) were obtained. However, the positions of 2 and 3 relative to each other were unknown, since at that time there were no homologues in GenBank, nor did they have identity with sequences released from the genome sequencing project. Degenerate oligonucleotide primers 1 and 3 (Table 1) were designed from peptides 1 and 3, respectively (Fig. 1A), since peptide 2 was unsuitable for primer design. Southern blot analysis showed that primer 1, designed from N-terminal peptide 1, hybridized to single fragments of restriction enzyme-digested S. mutans SJ32 chromosomal DNA (data not shown), indicating that at least the 5' terminus of the gene existed in single copy in the genome. The steps involved in cloning gbpB are depicted in Fig. 1. The gbpB gene was subcloned into the pET22B expression vector and transformed into E. coli BL23. Following induction with IPTG, cell-associated protein extracts and culture supernatants contained proteins that reacted with polyclonal antibody to purified GbpB (Fig. 1G), providing evidence that we had cloned the correct protein.

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Strategy for cloning gbpB from S. mutans SJ32. (A) From S. mutans SJ32 template DNA, PCR primers 1 and 3 yielded a 180-bp fragment with 100% DNA and amino acid identity to GenBank accession (Acc.) no. 4098503, a partially cloned but unknown S. mutans gene. Peptides 1 and 3 were found in the partial putative ORF; thus, peptide order within GbpB was 1-3-2. N-term, N terminus. (B) Specific 5' and 3' primers designed from accession no. 4098503 generated a 691-bp amplicon. (C) The purified amplicon was used to probe an S. mutans SJ32 genomic library for clones containing gbpB-related sequences. A single hybridizing clone contained a 3.5-kb insert with the 5' promoter region, a 300-amino-acid (aa) ORF comprising a signal sequence, the N terminus of mature GbpB, peptide 3, and 89 amino acids more than the gene fragment reported in GenBank accession no. 4098503. The ORF did not contain peptide 2 or a stop codon. (D) To obtain the 3' region, the PCR primer specific to the 5' sequence of GenBank accession no. 4098503 was used with a 3' specific primer designed after BLAST searching the S. mutans genome database with peptide 2. A 765-bp amplicon was aligned with already obtained sequences, but still without the stop codon. (E) A randomly primed PCR strategy (4) was used to generate the C terminus of the protein and the 3' noncoding region. First-round PCR products were generated with specific primer 1 derived from the 765-bp amplicon and random primer 2 that contained a 3' EcoRV sequence. The second round 5' specific primer 3 was again derived from sequences within the 765-bp amplicon, and the 3' random primer 4 retained the EcoRV sequence tag. A 1,054-bp fragment was generated that contained the C terminus, the stop codon, and 3' downstream untranslated sequence. (F) All sequences were aligned, and PCR primers were designed for the sequence 100 bp up- and downstream of the promoter and stop codons to generate the complete gene from SJ32. (G) GbpB expressed in E. coli reacted with antibody to the native protein. Shown is a Western blot of GbpB expressed in E. coli. Lane 1, induced GbpB; lane 2, induced vector alone; lane 3, purified GbpB.

DNA and protein analysis of GbpB. The DNA sequence of gbpB together with the predicted coding sequence is shown in Fig. 2. A putative Pribnow box, T₉₇AATATA₁₀₃, was found next to the predicted transcription start site at A₁₀₃ (Promoter Prediction by Neural Network [http://www.hgc.lbl.gov.gov/projects/promoter.html]) (10; M. G. Reese, N. L. Harris, and F. H. Eeckman, Proc. 1996 Pacific Symp. Biocomputing, 1996), and a ribosome binding site (₁₄₇AGGA₁₅₀) was identified 12 bp 5' to the ATG start codon. The coding sequence was 1,295 bp in length, and following the TAA termination codon, a hairpin loop was detected (₁₄₆₈A-₁₄₉₉T) with 40.3 kcal of free energy per ml. Therefore, it appears that gbpB is an independently regulated gene.

View larger version (54K): [in this window] [in a new window]   FIG. 2.   DNA and encoded protein sequence of gbpB from S. mutans SJ32. The putative Pribnow box in the 5' untranslated sequence is underlined, and the predicted transcription start is shown in bold. The ribosome binding site is shown in bold and underlined. Within the protein sequence the leucine zipper is underlined.

View larger version (46K): [in this window] [in a new window]   FIG. 3.   Alignment of deduced amino acid sequences of GbpB and PcsB precursor proteins. The ClustalW multiple alignment program (http://npsa-pbil.ibcp.fr/cgi-bin/align_clustalw.pl) was used to align amino acids. Identical and similar amino acids are marked at the bottom as asterisks (strong similarity) and dots (weak similarity), respectively. A conserved cysteine residue at the C-terminal region is observed at position 350.

Comparative genomic analysis of gbpB region. Evidence from comparative genomics also suggests that GbpB is involved in cell wall biology. Analysis of the S. mutans genome database (http://www.genome.ou.edu) revealed that two genes upstream from and in the same orientation as gbpB, encoded homologues of cell shape-determining proteins from other gram-positive bacteria. These ORFs contained approximately 50% amino acid similarity to cell shape-determining proteins MreC and MreD from Streptococcus pneumoniae, Lactococcus lactis, Enterococcus faecalis, and Enterococcus faecium (Fig. 4). Downstream from gbpB, other genes were identified that encoded homologues of proteins from gram-positive organisms (L. lactis, Bacillus subtilis, L. monocytogenes, Streptococus gordonii, and Streptococcus sanguis) that were involved in amino acid and lipid synthesis, ATP-dependent transport system, DNA repair, and competence. The gene immediately downstream from pcsB in group B streptococci encoded a polypeptide with high similarity to phosphoribosyl pyrophosphate synthetase (23), and the same locus was found immediately downstream from gbpB. Thus, in several gram-positive bacteria, including S. mutans, there appears to be conservation of genomic structure suggesting a functional relationship between genes involved in cell shape and cell wall maintenance.

View larger version (17K): [in this window] [in a new window]   FIG. 4.   Comparative genomics of the S. mutans gbpB locus. Genes flanking gbpB were examined for homologies in gram-positive bacteria. ORFs 1 and 2 showed homology to MreC and MreD, described as cell shape-determining proteins. GbpB shares homology with Usp45 from S. pneumoniae and L. lactis and to SagA from E. faecalis and E. faecium, which are secreted proteins related to peptidoglycan hydrolases from gram-positive bacteria. ORFs 4 and 5 were variable with regard to protein function. Accession numbers were obtained from the GenBank and The Institute for Genomic Research (*) databases.

RFLP analysis of the gbpB gene. The gbpB gene was obtained by PCR from 44 amplitypes of S. mutans strains (15), and a fragment of the predicted size, 1,653 kb, was obtained from all strains (data not shown). Figure 5A shows the predicted restriction map of gbpB, and endonucleases PvuII and Sau3AI were selected as the most informative for RFLP analysis because of the higher number of cutting sites and yield of fragments that could be well separated by electrophoresis. By PvuII digestion, the gbpB genes could be divided into three classes, and the most prevalent class, represented by strain 5SM3 in Fig. 5B, was found in 63.7% of the strains, including SJ32, from which gbpB was originally sequenced, as well as laboratory strains T8 and UA130. Also, 34% of the amplitypes showed a second pattern, represented by strain 3VF4 (Fig. 5B) and strains UA159 and GS5. Sau3AI-generated polymorphisms were less informative since only four amplitypes (9.1%) showed a different pattern, as shown with strain 3SN1 in Fig. 5C.

View larger version (44K): [in this window] [in a new window]   FIG. 5.   RFLP analysis of gbpB from clinical isolates of S. mutans. (A) Restriction map of 1,653-bp amplicon that included 162 bp upstream and 195 bp downstream flanking ORF of GbpB from UA159. (B) RFLP patterns with restriction enzyme PvuII. (C) Restriction patterns obtained with Sau3AI. Lanes M, molecular size markers.

View larger version (67K): [in this window] [in a new window]   FIG. 6.   Polymorphisms of gbpB in different S. mutans strains lead to changes in protein sequence. Out of 18 mutations identified, 9 caused changes in the protein level (highlighted bars) and 5 represented changes to residues of different classes.

Production of GbpB by clinical isolates of S. mutans The amounts of GbpB measured in culture supernatants from 76 clinical isolates were found to be highly variable. Values were normalized to microgram of protein/A₅₅₀ of culture, and the mean (± standard deviation [SD]) production level was 5.58 ± 0.167 µg/ml of culture supernatant. The A₅₅₀ of cultures did not vary significantly (mean, 0.147; SD, 0.116); thus, there was no correlation between growth and GbpB production (Pearson, r = 0.05, P = 0.667). Strain 20A3 produced the largest amounts of GbpB in the culture supernatant (8.52 µg/ml), and strain 8VS3 produced the smallest amounts (1.16 µg/ml). Twenty isolates with the widest spectrum of GbpB protein levels, including S. mutans laboratory strains UA130 and SJ32 and S. sobrinus as a negative control, were selected for further analysis of the distribution of GbpB between cells and supernatants (Fig. 7). It then became apparent that levels of GbpB in culture supernatants alone did not reflect total levels since in some strains a large amount of protein remained cell associated, e.g., strain 8VS3. On the other hand, 20A3 contained little cell-associated protein and other strains produced small amounts of GbpB in both culture supernatants and cell extracts, e.g., 4JP2. Out of 19 S. mutans clinical strains analyzed, eight (42.1%) retained most of the GbpB they produced in a cell-associated form, e.g., 3A1 and 5SM3, while in 11 strains (57.9%), most of the protein was secreted into the culture supernatants. Among the laboratory strains, most of the GbpB produced by SJ32 was found in culture supernatants, while for UA130, most of the protein was in a cell-associated form. As expected, S. sobrinus strain 3SSA1 did not produce GbpB. The patterns of GbpB distribution between cell surface and culture supernatants were reproducible, and in similar control experiments, GbpA was extracted from 20 strains and the majority of this protein was in supernatants and was not cell associated (data not shown). Thus, the protein localization patterns appear to be intrinsic to GbpB. However, different amplitypes of S. mutans with the same gbpB RFLP pattern differed in GbpB distribution (e.g., 5SM3 and SJ32); thus, other factors influence GbpB localization.

View larger version (28K): [in this window] [in a new window]   FIG. 7.   Distribution of GbpB in culture supernatants and cells of 22 MS strains. Strain 3SSA1 is S. sobrinus, a negative control. Shaded bars, amounts of protein measured in the culture supernatants; solid bars, GbpB obtained after urea extraction of cells.

Correlation between biofilm formation and GbpB production. S. mutans strains were compared for their ability to form biofilms in microtiter plates as measured by crystal violet staining (21). Figure 8 illustrates the differences observed in biofilm formation of a subset of clinical isolates. The mean A₅₇₅ of crystal violet-stained biofilms varied from 0.09 (SD, 0.03) to 0.58 (SD, 0.01), indicating a high level of variability in biofilm growth between strains, although values for planktonic growth were very similar (Fig. 8). The amounts of GbpB in culture supernatants and biofilm growth were very similar when isolates of the same S. mutans amplitype were compared (Fig. 8). In addition, strains in which most of the GbpB was cell associated also grew well as biofilms, as observed for strains 8ID3, 5SM3, and 8VS3. Within the 44 clinical amplitypes, there was a significantly positive relationship between the amounts of GbpB present in culture supernatants and the level of biofilm growth (Pearson correlation, r = 0.315, P < 0.05; Fig. 9). However, there was no significant relationship between biofilm formation and the amount of either GbpA (Pearson correlation, r = 0.101; P = 0.512) or glucosyltransferases (Pearson correlation, r = 0.005; P = 0.976) that were quantified in the same samples by immuno dot blot analysis (Fig. 9). There was no significant association between amounts of GbpB in culture supernatant with planktonic growth (Pearson correlation, r = 0.24; P = 0.114), nor was biofilm growth related to that of planktonic cultures (Pearson correlation, r = 0.146, P = 0.344), although variability in planktonic growth was much lower than that in biofilms. Even though only 21 S. mutans strains were analyzed for GbpB distribution, a positive relationship was identified between amounts of cell-associated GbpB and biofilm formation (Pearson correlation, r = 0.45, P < 0.05, n = 21).

View larger version (59K): [in this window] [in a new window]   FIG. 8.   Biofilm formation, planktonic growth, and GbpB production in clinical strains of S. mutans. Strains within vertical lines were isolated from the same child. One child (10ST) was colonized by two different amplitypes: one represented by isolates 10ST1 and 10ST4 and the other by isolates 10ST2 and 10ST3. In all other samples, one amplitype per child was identified. The lower panels show the amounts of GbpB present in culture supernatants that were quantified by immuno dot blot analysis.

View larger version (13K): [in this window] [in a new window]   FIG. 9.   (A) A positive relationship was detected between amounts of GbpB present in culture supernatants and biofilm formation among the 44 amplitypes of S. mutans. No significant associations were detected between amounts of GpbA (B) and glucosyl transferases (GTF) (C) and biofilm growth in the same strains.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Dental caries is a biofilm-dependent infectious disease in which MS are the major pathogenic bacteria. However, the molecular mechanisms of MS participation in biofilm development and maintenance are not completely understood. Accumulation of MS depends on the production of an extracellular polysaccharide matrix of water-insoluble glucans (30) that also modifies the physicochemical, and hence cariogenic, properties of dental plaque (34). Glucans are synthesized exclusively from sucrose by glucosyltransferases, a group of extracellular enzymes. Glucan binding proteins also promote bacterial aggregation, and although three different Gbps, A to C, have been characterized so far (23, 24, 26), the mechanisms of attachment between glucans and the bacterial cell surface are unknown. In this study, we cloned gbpB from SJ32, the S. mutans strain from which the protein was first identified (26). Sequencing of gbpB confirmed our previous immunological results that the protein was novel and had no homology to other Gbps (26). A 27-amino-acid signal peptide was predicted immediately followed by the sequence encoding the N terminus of the purified, secreted mature protein. Apart from its unusual alanine-and glutamine-rich amino acid composition, an interesting feature was that the protein contained heptad motifs characteristic of leucine zippers, and because GbpB is found in culture medium or on the cell surface, we presume this sequence is involved in interactions with other proteins. RFLP analysis of gbpB from 44 S. mutans amplitypes revealed that most sequence changes were clustered in the central region of the protein, suggesting functional conservation of sequences within the N- and C-terminal domains. Recent results indicate that the N terminus is immunologically active (unpublished data) and may possess the protective epitopes associated with the native molecule (29).

Although sequences with partial homology to GbpB were found in the GenBank database, there were no suggestions as to function, apart from the ability to bind glucan. Recently, a differential display reverse transcription-PCR screen for S. mutans stress-responsive genes identified a gene with complete identity to gbpB (5). Expression increased under conditions of high osmolarity and temperature, and it was speculated that the encoded protein might be a structural component of the cell wall or membrane. More significantly with regard to GbpB function, a protein designated PcsB was recently identified in group B streptococcus (22) and was shown to have good homology to GbpB (Fig. 3). Generating a pcsB mutant proved difficult, and by using osmotically protective conditions for both transformation and the subsequent growth of candidates, one mutant was obtained. In addition to osmotic sensitivity, the mutant grew slowly, was hypersensitive to several antibiotics, and showed abnormal septum formation during growth. From these characteristics, together with some homology to a peptidoglycan hydrolase from L. monocytogenes, it was concluded that PcsB was involved in wall separation during cell division. Database analyses indicated that PcsB-like proteins were present in several gram-positive bacteria; these proteins included P45 of L. monocytogenes, SagA from E. faecium and E. faecalis, and Usp45 from Streptococcus pneumoniae and L. lactis (7, 32, 33). GbpB homologues are preceded by two ORFs that code for cell shape-determining proteins (Fig. 4). As with GbpB and PcsB, these proteins were both secreted and cell associated and contained a conserved cysteine residue in the C-terminal domain that was necessary for peptidoglycan hydrolytic activity (35). We observed polymorphic forms of GbpB in clinical isolates of S. mutans; however, most of the changes were clustered in the central region of the protein (Fig. 6) while the N- and C-terminal regions were conserved, suggesting functional roles. Because of the close homology of GbpB to PcsB, and in turn to the other proteins with sequence similarities that include the conserved cysteine active site residue, we hypothesize that GbpB has a similar function in peptidoglycan biosynthesis. To test this hypothesis, we attempted to isolate gbpB mutants in several strains using different integrating constructs. Inactivation of gbpB by simple Campbell-like integration of a suicide vector yielded viable mutants containing multiple and deleted versions of gbpB, as indicated by PCR and sequencing analysis (R. O. Mattos-Graner et al., unpublished data). Scanning electron micrographs of one class of slow-growing mutants showed clumps of enlarged cells instead of the chains of cells observed with the parent strain, a growth phenotype similar to that of pcsB mutants (22). Thus, although glucan-binding properties were originally ascribed to GbpB, our results to date indicate that the protein plays other roles in the biology of S. mutans.

GbpB was previously described as a secreted protein in several laboratory strains of S. mutans (D. J. Smith et al., J. Dent. Res), however, we observed large variations in the amounts of GbpB secreted in the culture supernatants of 74 clinical isolates of S. mutans. In the majority of clinical isolates, most of the GbpB was secreted into culture medium, but in some amplitypes the majority of the protein remained cell associated (Fig. 7). Unlike GbpC, GbpB does not possess the C-terminal cell wall anchor motif, LPXTG, found in several surface proteins of gram-positive bacteria (18). However, this cell-sorting signal is not contained in all surface proteins of gram-positive organisms (19) and was not present in P45 from L. monocytogenes (25).

The amounts of GbpB secreted by strains of the same amplitype were very similar, suggesting that distribution of the protein between cells and culture medium is intrinsic to specific clones of S. mutans. Different genotypes, even when isolated from the same host, showed different amounts of secreted GbpB, as exemplified by strains 10ST1 and 10ST4 (Fig. 8), which have the same amplitype and secrete large amounts of GbpB, and strains 10ST2 and 10ST3, which were isolated from the same child but have a different DNA fingerprint and secrete less GbpB. Since GbpB may be involved in cell wall formation, it would be interesting to determine the response of each amplitype to high osmolarity, temperature, pH, and antibiotic exposure.

In this study, a large number of S. mutans clinical isolates were examined for the ability to form biofilms in low-sucrose medium (THB) (9) using an in vitro assay that was previously used to screen for biofilm-defective mutants in both gram-negative and gram-positive bacteria (14, 21). Despite high variability between isolates, a significant finding was a positive correlation between the ability of strains to grow as biofilms and GbpB production. Even strains in which a large proportion of the protein remained cell associated showed good growth in biofilms. That no correlations were observed between production of GbpA or glucosyltransferases and biofilm formation also underscores the importance of GbpB in this mode of growth (Fig. 9). However, for several strains, the level of secreted GbpB did not correlate with biofilm growth, a predictable finding since many unknown factors, beside GbpB, are involved in the complex process of biofilm formation. Since the virulence of S. mutans is directly related to its ability to colonize tooth surfaces and accumulate in the dental biofilm, it is important to define a function for GbpB and its role in biofilm formation. Given the extent of homology between GbpB and PscB from group B streptococcus, one hypothesis is that GbpB is involved in cell wall formation. Several biofilm-defective mutants of S. gordonii Challis had disruptions in genes involved in peptidoglycan biosynthesis, and a putative osmoregulatory role in biofilm formation was suggested (14). This provides circumstantial supporting evidence for our hypothesis that GbpB is also involved in peptidoglycan biosynthesis.