Identification and Characterization of a Nonimmunoglobulin Factor in Human Saliva That Inhibits Streptococcus mutans Glucosyltransferase

Abstract

Saliva contains an array of nonimmunoglobulin defense factors which are thought to contribute to the protection of the hard and soft tissue surfaces of the oral cavity by modulating microbial colonization and metabolism. Here we report the discovery of a putative innate defense factor in human saliva that inhibits the glucosyltransferase (GTF) of Streptococcus mutans, a virulence enzyme involved in oral colonization by this pathogen. The GTF-inhibiting factor (GIF) was initially identified as a nonimmunoglobulin salivary component that interfered with detection of antibodies to the glucan-binding region (GLU) of GTF by an enzyme-linked immunosorbent assay. This inhibitory activity was present in whole saliva and submandibular-sublingual saliva, but it was essentially absent from parotid saliva. GIF inhibited the recognition of S. mutans cell surface-associated GTF by specific antibodies but had no effect on antibodies to other cell surface antigens, suggesting that GIF specifically binds to GTF on S. mutans. GIF purified by size exclusion or affinity chromatography was used for biochemical and functional characterization. Analysis of GIF by sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed a high-molecular-weight glycoprotein after staining with Coomassie blue or Schiff's reagent. Heating and reduction with 2-mercaptoethanol of GIF resulted in the release of a ∼58-kDa protein that was identified as α-amylase by Western blotting using anti-α-amylase antibodies. GLU bound blotted α-amylase, suggesting that the latter molecule is the GLU-binding component of the GIF complex. The ability of GTF to synthesize extracellular glucans was inhibited by GIF but not by uncomplexed α-amylase or an unrelated high-molecular-weight glycoprotein. In conclusion, our findings demonstrate that in human saliva, there is a high-molecular-weight glycoprotein-α-amylase complex which is capable of inhibiting GTF and may contribute to control of S. mutans colonization in the oral cavity.

In addition to the adaptive or specific immunity that is mediated predominantly by secretory immunoglobulin A (IgA) antibodies, human saliva also contains an array of antimicrobial molecules whose presence does not depend on previous exposure to microbial antigens. These nonimmunoglobulin defense factors contribute to the protection of the dental and mucosal surfaces of the oral cavity by modulating microbial colonization and metabolism (16, 28, 35). Submandibular-sublingual mucins and other salivary glycoproteins, such as the parotid salivary agglutinin, are capable of aggregating oral microorganisms in the fluid phase, which results in clearance of the microorganisms from the mouth by swallowing (27, 32, 34). Microbial metabolic processes can be inhibited by various factors, including lactoferrin, which deprives bacteria of iron, and the salivary peroxidase system, which can reduce bacterial acid production and the subsequent damaging effect on dental enamel (23)_.

Innate humoral defense factors present in saliva may act alone or with each other in a synergistic or antagonistic manner (23, 35). One type of interaction is via the formation of heterotypic complexes (e.g., mucins form complexes with various molecules, including lysozyme, cystatins, and α-amylase), which in certain cases may have properties distinct from those of the individual components (4, 7, 11). The complexity of the role of saliva in host defense is further illustrated by the fact that individual salivary molecules may have more than one function. Also, different molecules may have similar activities, and salivary molecules not only may act in defense of the host but also may be used by the microorganisms for their own benefit (7, 35). These properties of the salivary defense components, in addition to the variability of the salivary secretion, may provide a plausible explanation for why clinical studies designed to associate levels of individual salivary molecules with oral disease activity in general have been inconclusive (25).

The enzyme glucosyltransferase (GTF) is an important virulence factor of Streptococcus mutans (15). GTF synthesizes adhesive glucans from sucrose which are essential for the establishment of cohesive streptococcal masses on the tooth surface and subsequent caries development (15, 17). This enzyme contains an N-terminal catalytic site (CAT) and a C-terminal repetitive glucan-binding region (GLU) which is presumably involved in chain extension of the growing glucan polymers (14, 21, 36). Antibodies to either CAT or especially GLU inhibit glucan synthesis by GTF (12), and intranasal immunization of mice with GLU inhibits S. mutans colonization (13).

In this paper, we report the isolation and characterization of a GTF inhibitory factor (GIF). This factor was initially identified as a nonimmunoglobulin salivary component that interfered with antibody recognition of recombinant GLU by an enzyme-linked immunosorbent assay (ELISA). It was subsequently chromatographically purified and characterized as a glycoprotein-α-amylase complex. The binding of this salivary factor to GLU interfered with the enzymatic activity of GTF in a manner analogous to that of anti-GLU antibodies (12). Our results provide in vitro evidence that there may be an innate defense mechanism in saliva against the cariogenic bacterium S. mutans.

(This study was performed by Christina Jespersgaard in partial fulfillment of the requirements for a Ph.D. from The University of Aarhus, Aarhus, Denmark.)

MATERIALS AND METHODS

Saliva sample collection.

Unstimulated whole, parotid, and submandibular-sublingual saliva samples were collected from five healthy, caries-free human subjects. Whole saliva was collected on ice with a sterile 50-ml plastic tube. Parotid saliva was collected in a Schaefer cup (30) placed over Stensen's duct, while at the same time submandibular-sublingual saliva was collected from under the tongue with a pipette. The saliva samples were clarified by centrifugation prior to use. The individual saliva samples had insignificant levels of anti-GTF and anti-GLU specific antibodies (specific IgA/total IgA ratio, <0.1%). All saliva samples were collected by using the guidelines of the University of Alabama at Birmingham Institutional Review Board and with the approval of this board.

ELISA.

The levels of specific GTF-binding activity in saliva samples were determined by using Maxisorp microtiter plates (Nunc, Roskilde, Denmark) coated with GLU (1 μg/ml) or whole cells of S. mutans PC3379 (5 × 10⁸ CFU/ml). The tetracycline- and erythromycin-resistant strain S. mutans PC3379 was grown overnight anaerobically in Todd-Hewitt broth containing 10 μg of tetracycline and erythromycin per ml (13). The bacteria were harvested and washed prior to use in ELISA. Whole, parotid, and submandibular-sublingual saliva samples from four healthy subjects were serially diluted in 0.01 M phosphate buffer (pH 7.2) containing 0.5 M NaCl and 0.15% Tween 20 (PBS-Tween) and incubated at 4°C overnight on ELISA plates precoated with either GLU or whole cells of S. mutans. To determine the effect of saliva on inhibition of anti-GLU or anti-CAT specific antibody binding to the coating antigens, the plates were subsequently incubated with biotinylated rabbit anti-GLU or anti-CAT antibodies (12). Peroxidase-labeled streptavidin (Sigma, St. Louis, Mo.) was used as the developing reagent, followed by o-phenylenediamine substrate with H₂O₂. Inhibition of anti-S. mutans specific antibody binding by saliva was determined in a similar way by using mouse antiserum to S. mutans (13) and peroxidase-labeled anti-mouse IgG antibodies (Southern Biotechnology Associates, Birmingham, Ala.) as the developing reagent, followed by o-phenylenediamine substrate with H₂O₂.

Purification of salivary factor.

Pooled whole saliva from five healthy individuals was suspended in buffer A (4 M guanidine-HCl, 50 mM Tris-HCl [pH 7.5], 5 mM EDTA, 0.02% azide) and stirred overnight at 4°C. The saliva was then fractionated by size exclusion chromatography on a Superose 12 HR 16/50 column (Amersham Pharmacia Biotech, Piscataway, N.J.) equilibrated in buffer A. Fractions were tested to determine their abilities to inhibit GLU specific antibody activity by performing ELISA as described above. Fractions that resulted in more than 80% inhibition were pooled and used for further experiments immediately after dialysis against sodium phosphate-buffered saline-0.2% sodium azide (PBSA) (pH 6.5).

The GLU-binding salivary factor was also purified by affinity chromatography by using a CNBr-activated Sepharose 4B column (Amersham Pharmacia Biotech) coupled with purified GLU (13). Briefly, 1.9 g of resin was suspended in 1 mM HCl and washed for 15 min on a sintered glass filter. The resin was saturated with coupling buffer containing 0.1 M NaHCO₃ (pH 8.3) and 0.5 M NaCl before it was mixed with 25 mg of affinity-purified GLU. The GLU polypeptide was coupled to the resin for 2 h at room temperature with gentle end-over-end rotation. The efficiency of the coupling process was determined by measuring the concentration of free GLU with a bicinchoninic acid (BCA) kit (Pierce, Rockford, Ill.) before and after coupling. The efficiency of coupling GLU to the CNBr-activated Sepharose was 48%; approximately 2 mg of GLU per ml of resin was coupled. Uncoupled GLU was removed by washing the resin twice with 20 ml of PBS-Tween. The remaining coupling sites on the resin were blocked by incubation with 0.1 M Tris-HCl (pH 8.0) for 2 h. The column was finally washed by using three alternating cycles of 0.1 M acetate buffer (pH 4.0) containing 0.5 M NaCl and 0.1 M Tris-HCl (pH 8.0) containing 0.5 M NaCl. The specificity of column binding was established by determining the fraction of biotinylated anti-GLU antibody captured by GLU-coupled or blocked resin. The GLU-coupled resin captured 84% of the GLU-specific antibodies, whereas the level of nonspecific binding of the GLU-specific antibodies to the blocked resin was <2%.

Whole saliva was incubated with GLU-coupled resin in PBS-Tween by gentle end-over-end rotation overnight at 4°C. The column was then washed twice with 20 ml of PBS-Tween before elution of the salivary factor with 5 ml of 0.1 M Tris-HCl (pH 10.0) containing 1 M NaCl for 2 h at room temperature. The eluted salivary factor was used for experiments immediately after dialysis against PBS-Tween.

SDS-PAGE and Western blot analysis.

Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and Western blot analysis were performed by using the Mini-PROTEAN 3 and Mini Trans-Blot cell systems, respectively, as suggested by the manufacturer (Bio-Rad, Hercules, Calif.). Gels were stained with either Bio-Safe Coomassie blue (Bio-Rad), silver stain (Bio-Rad), or Schiff's reagent (Sigma) as recommended by the manufacturers. Human salivary α-amylase type IX-A, control mucin from bovine submaxillary glands (type I-S), and human salivary α-amylase-specific rabbit antibody were purchased from Sigma. Biotinylated goat anti-rabbit IgG (Southern Biotechnology Associates) was used to develop Western blots probed with anti-α-amylase antibodies, whereas biotinylated rabbit anti-GLU was used to develop the Western blot probed with purified GLU; this was followed by incubation with alkaline phosphatase-conjugated streptavidin and by treatment with 5-bromo-4-chloro-3-indolyl phosphate (BCIP)-nitroblue tetrazolium tablets (Sigma) for visualization.

GTF preparation.

A DNA fragment containing gtfB from S. mutans was removed from plasmid pYNB13 (36) (provided by H. K. Kuramitsu, State University of New York at Buffalo, Buffalo, N.Y.) by digestion with XhoI and NotI and was subcloned into expression vector pET32b(+) (Novagen, Madison, Wis.). The plasmid containing the sequence encoding gtfB, pET32b(+)-gtfb, was then electroporated into Escherichia coli BL21(DE3), and transformed colonies were selected on Luria-Bertani agar plates (1% tryptone, 0.5% yeast extract, 1% NaCl, 0.1% dextrose, 1.8% agar) containing 50 μg of carbenicillin per ml for selection of the plasmid. The transformants were examined for the presence of a 10.3-kb plasmid. An overnight culture of E. coli BL21(DE3)/pET32b(+)-gtfb (2.5 ml) was used for inoculation of 500 ml of Luria-Bertani broth containing 50 μg of carbenicillin per ml. The cells were grown to the mid-log phase at 30°C (approximately 6 h) before they were induced with 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) overnight. Soluble proteins were recovered by resuspending pelleted cells in TTE buffer (50 mM Tris-HCl [pH 8.0], 0.1% Triton X-100, 2 mM EDTA), and the protein extract was stored for 1 h at −70°C. Following thawing, the cell lysate was sonicated twice for 10 s and then subjected to 60% saturated ammonium sulfate precipitation at 4°C overnight. The precipitated proteins were resuspended in PBS-Tween and dialyzed at 4°C overnight against PBS-Tween in order to remove any remaining ammonium sulfate.

An anti-GLU specific affinity column was prepared in order to purify sufficient enzyme for the GTF inhibition assay. Six milligrams of affinity-purified rabbit anti-GLU specific antibodies (12) was coupled to 0.35 g of CNBr-activated Sepharose 4B resin (Amersham Pharmacia Biotech) as described above. The efficiency of the coupling process was determined by measuring the concentration of anti-GLU antibody with a BCA kit before and after coupling. Rabbit anti-GLU antibodies were coupled to CNBr-activated resin with 98.5% efficiency (approximately 9.4 mg of antibody per ml of resin). The anti-GLU antibody-coupled resin was washed as described above and was then washed twice with PBS-Tween and once with 100 mM glycine buffer (pH 2.5). The anti-GLU resin was equilibrated with PBS-Tween and mixed with partially purified GTF from whole-cell lysates of E. coli BL21(DE3)/pET32b(+)-gtfb for 4 h at 4°C by gentle end-over-end rotation. The resin was washed with 20 column volumes of PBS-Tween and then with 20 column volumes of 10 mM phosphate buffer (pH 7.0). GTF was eluted with 10 column volumes of 100 mM glycine buffer (pH 2.5), and the flow was stopped for 15 min. The eluted fractions were collected in tubes containing pH-adjusting 1 M potassium phosphate buffer (pH 8.2), and the purity of the purified GTF was confirmed by SDS-PAGE. GTF was dialyzed against PBSA (pH 6.5) and stored at −70°C.

GTF inhibition.

Glass test tubes were precoated with 80 μl of fractionated saliva, control mucin, or α-amylase in PBSA (pH 6.5) containing 0.05% bovine serum albumin. The concentrations of the fractionated saliva and the control mucin samples were standardized to 0.2 mg/ml based on Schiff's reagent staining of the carbohydrate (18), whereas the concentration of the α-amylase sample was 0.2 mg/ml based on the total weight of protein and carbohydrate. Next, 20 μl of GTF (6.25 μg/ml) in PBSA was added to obtain a final GTF concentration of 0.125 μg/tube. This mixture (100 μl) was preincubated at 37°C for 2 h. Then 0.85 mg of sucrose containing 19 nCi of [¹⁴C]glucose-sucrose (approximately 40,000 cpm; NEN, Boston, Mass.) in 100 μl of PBSA (pH 6.5) was added. The tubes were incubated at 37°C for 3 h, and then the reactions were stopped by placing the tubes on ice. The samples were vortexed, and 100 μl of each reaction mixture was pipetted onto a 13-mm-diameter Whatman GF/C glass fiber filter (Whatman, Clifton, N.J.) that had been prewashed with 0.5 ml of PBSA (pH 6.5) in a 13-mm-diameter stainless steel filter holder (Millipore, Bedford, Mass.). The filters were washed twice with 0.5 ml of PBSA (pH 6.5) and air dried, and the amount of radioactivity was measured with a scintillation counter.

RESULTS

Identification of GLU-binding factor in saliva.

Initial analysis of saliva samples from mice immunized intranasally with GLU failed to reveal an IgA anti-GLU response, despite the presence of IgA antibodies to GLU in serum and vaginal secretions (13). This result prompted us to investigate whether a salivary component interfered in the ELISA by binding to the GLU antigen used to coat the plates. Evidence that there is a nonantibody salivary factor was obtained when preincubation of immobilized GLU with saliva samples from unimmunized mice was found to inhibit the subsequent binding of affinity-purified rabbit IgG anti-GLU antibodies (data not shown). The inability to detect anti-GLU activity was resolved by centrifugation of the mouse saliva after freezing, a procedure which precipitated mucins and molecules complexed with them without eliminating salivary IgA antibodies (13). Therefore, we decided to determine if human saliva contains a similar GLU-binding factor.

Whole saliva from four caries-free individuals strongly inhibited the binding of affinity-purified rabbit IgG anti-GLU antibodies to immobilized GLU (Fig. 1A). The inhibitory effect was reduced by dilution but was still present in saliva diluted 1/500. To identify the salivary origin of this binding factor, we compared the ability of parotid saliva to inhibit GLU recognition by rabbit IgG anti-GLU antibodies with the ability of submandibular-sublingual saliva from a healthy individual with undetectable anti-GTF antibody activity to inhibit GLU recognition by rabbit IgG anti-GLU antibodies. The submandibular-sublingual saliva was as potent as whole saliva for inhibiting the binding of IgG anti-GLU antibodies to GLU, whereas the parotid saliva was much less effective (Fig. 1B). Specifically, the submandibular-sublingual and whole saliva samples were potent inhibitors even when they were diluted 1/500, whereas a 1/4 dilution was required to obtain similar inhibitory activity with parotid saliva (Fig. 1B).

FIG. 1.

Inhibition of rabbit IgG anti-GLU binding to GLU by a factor in human whole saliva (A) and identification of the salivary gland secreting this inhibitory activity (B). ELISA plates coated with recombinant GLU polypeptide were preincubated with dilutions of saliva or with buffer alone (as a control), washed, and then allowed to react with affinity-purified IgG anti-GLU antibodies. Saliva was collected from caries-free individuals with undetectable levels of anti-GTF antibodies. The data (means based on duplicate determinations) are percentages of inhibition of antibody binding to GLU by saliva calculated by comparison with the inhibition of antibody binding to GLU by buffer alone. (A) Symbols: □, subject 1; ♦, subject 2; ∗, subject 3; ⊕, subject 4. (B) Symbols: □, whole saliva; ⊕, parotid saliva; •, submandibular-sublingual saliva.

Binding of anti-GLU and anti-CAT rabbit IgG antibodies to whole cells of S. mutans was also strongly inhibited by human whole saliva (Fig. 2). These findings suggest that the GLU-binding salivary factor recognized native GTF on the cell surface of S. mutans. The inability of the whole saliva to effectively inhibit the binding of anti-S. mutans antibodies to immobilized S. mutans (Fig. 2) underscores the specificity of the GLU-GTF whole-saliva interactions.

FIG. 2.

Binding of anti-GLU, anti-CAT, or anti-whole-cell S. mutans antibodies to S. mutans cell surface is inhibited by a factor in human whole saliva. ELISA plates coated with whole S. mutans cells were preincubated with buffer alone or with various dilutions of clarified whole saliva with undetectable anti-GTF antibody activity. After washing, the plates were incubated with anti-GLU (▪), anti-CAT (○), or anti-whole-cell S. mutans (∗) antibodies. The data (means based on duplicate determinations) are percentages of inhibition of antibody binding to S. mutans by saliva calculated by comparison with the inhibition of antibody binding to S. mutans by buffer alone.

Purification of the salivary factor.

To isolate the GLU-binding salivary factor, whole saliva was fractionated on a Superose 12 size exclusion column (Fig. 3). Saliva fractions containing the GLU-binding factor were identified on the basis of the ability to inhibit the interaction between purified GLU and affinity-purified antibodies. The main GLU-binding activity eluted as a high-molecular-weight component in the protein absorbance peak corresponding to fractions 9 to 14. A protein absorbance peak corresponding to fractions 15 to 21 also exhibited GLU-binding properties; however, only fractions 9 to 14, which resulted in more than 80% inhibition of GLU-specific antibody binding, were pooled and used for further analysis. The GLU-binding factor was also purified by affinity chromatography by using a column containing agarose-immobilized GLU. Different elution buffers were tested by using a quantitative ELISA performed with GLU-coated plates to measure the ability of the GLU-binding factor to maintain inhibitory activity after exposure to the buffers (data not shown). The elution strategies assessed included buffers containing urea or guanidium-HCl and low and high pHs in the presence of high-salt-concentration buffers. None of the buffers tested was able to entirely release the GLU-binding factor (data not shown). In this regard, a component of the GLU-binding factor, identified as α-amylase (see below), was released from GLU-Sepharose by heating and reduction, as determined by SDS-PAGE. Uncoupled control resin produced no bands following the same treatment. The salivary factor was eluted from the affinity column with 0.1 M Tris-HCl (pH 10.0) buffer containing 1 M NaCl, which appeared to be the most efficient elution buffer tested.

FIG. 3.

Fractionation on a Superose 12 column of clarified whole saliva collected from healthy individuals without detectable salivary anti-GTF antibody activity. Symbols: ♦, protein absorbance; ○, GIF inhibition. OD 280 nm, optical density at 280 nm.

Samples obtained by size exclusion or affinity chromatography which contained the GLU-binding factor were examined by SDS-PAGE (Fig. 4). Samples obtained after both purification methods were used produced a high-molecular-mass band (>200 kDa), as well as a band at ∼58 kDa. Almost identical band patterns were obtained with the different salivary factor samples. The main difference between samples was the higher level of the 58-kDa component in the Superose 12 size exclusion chromatography-purified sample.

FIG. 4.

SDS-PAGE analysis of different batches of affinity-purified GLU-binding salivary factor (lanes 1, 5, and 6), Superose 12-fractionated GLU-binding salivary factor (lane 2), purified human salivary α-amylase (lane 3), and an unrelated mucin from the bovine submaxillary gland (lane 4). Samples were boiled for 5 min in the presence of 2-mercaptoethanol, analyzed by SDS-PAGE, and visualized by silver and Coomassie blue staining.

Identification of the salivary factor components.

High levels of α-amylase (100 to 2,600 μg/ml) are found in human parotid saliva (2), and the molecular weight of this enzyme is comparable to the molecular weight of the low-molecular-weight component of the GLU-binding factor. Therefore, to establish whether α-amylase is the low-molecular-weight component of the GLU-binding salivary factor, we performed a Western blot analysis with anti-α-amylase antibodies (Fig. 5A). The anti-α-amylase antibodies reacted with the high-molecular-weight band in addition to the lower-molecular-weight-band in the unreduced and unheated sample. All anti-α-amylase antibody-binding activity was released from the high-molecular-weight complex and was found in the low-molecular-weight band in the reduced and boiled sample. This finding suggested that α-amylase is a component of the high-molecular-weight GLU-binding factor but is released upon reduction and heating. No bands that were not reactive with anti-α-amylase antibodies were present, as revealed by comparison of the Western blot (Fig. 5A) with a Coomassie blue-stained SDS-PAGE gel (Fig. 5B). A Western blot analysis was then performed with anti-α-amylase antibodies and whole saliva, and the pattern observed was similar to that obtained with the purified GLU-binding factor (Fig. 5C). The blot showed that there was a shift of the α-amylase bands from 55 and 58 kDa to approximately 60 kDa after heating and reduction of the disulfide bonds in the molecule (24). The ability of α-amylase to bind GLU was demonstrated by a Western blot analysis (Fig. 5D) in which purified GLU specifically bound the α-amylase bands on membranes following transfer of separated parotid saliva. The high-molecular-weight component of the salivary factor was identified as a glycoprotein by SDS-PAGE and subsequent staining with Schiff's reagent (Fig. 6). The SDS-PAGE migration pattern of the glycoprotein moiety of the salivary factor was the same after boiling and reduction.

FIG. 5.

Identification of α-amylase as a component of the GLU-binding salivary factor. (A) Purified GLU-binding salivary factor in loading buffer was either not treated (lane 1) or boiled for 5 min in the presence of 2-mercaptoethanol (lane 2) and analyzed by Western blotting using rabbit IgG anti-human α-amylase antibodies. (B) Same preparation of untreated GLU-binding salivary factor as in panel A, visualized on a Coomassie blue-stained SDS-PAGE gel. (C) Whole clarified saliva in loading buffer was either boiled in the presence of 2-mercaptoethanol (lane 1) or not treated (lane 2) and analyzed by Western blotting in a manner similar to that used for purified GLU-binding salivary factor. (D) Parotid saliva analyzed by Western blotting using purified GLU polypeptide (lane 1) or PBS-Tween (lane 2) for incubation, followed by detection of rabbit anti-GLU antibodies.

FIG. 6.

Identification of the GLU-binding salivary factor as a high-molecular-weight glycoprotein. Samples of GLU-binding salivary factor in loading buffer with 2-mercaptoethanol (2-ME) (lanes 1 and 2) or without 2-mercaptoethanol (lanes 3 and 4) were boiled for 5 min (lanes 2 and 3) or not treated (lanes 1 and 4), analyzed by SDS-PAGE, and stained with either Coomassie blue (A) or Schiff's reagent (B).

Inhibition of GTF activity by the salivary factor.

Purified GTF was used in an activity assay to examine the inhibitory effect of the GLU-binding salivary factor purified by size exclusion chromatography. The salivary factor inhibited insoluble glucan production by GTF by approximately 60%, whereas bovine submaxillary gland mucin, used as a control, had no effect (Fig. 7). Purified α-amylase did not inhibit GTF activity, although it bound GLU (Fig 5D). The inability of isolated α-amylase to inhibit GTF is consistent with the fact that no significant activity inhibiting GLU recognition by specific antibodies was detected in the low-molecular-weight fractions after fractionation of saliva on Superose 12 (Fig. 3).

FIG. 7.

Inhibition of S. mutans GTF insoluble glucan production by the GLU-binding salivary factor purified by size exclusion column chromatography. Bovine submaxillary gland mucin and purified α-amylase were used as controls. The percentage of inhibition of S. mutans GTF was calculated as follows: [(amount of glucan produced in buffer − amount of glucan produced in the presence of sample)/amount of glucan produced in buffer] × 100. The data are means ± standard deviations based on triplicate determinations. Dark cross-hatched bar, fractionated saliva; light cross-hatched bar, control mucin; solid bar, α-amylase.

DISCUSSION

In the present study, we obtained evidence that a high-molecular-weight glycoprotein fraction obtained from human saliva recognizes the GLU domain of S. mutans GTF and functionally inhibits the activity of this enzyme. This glycoprotein fraction was designated GIF and may represent a novel innate defense mechanism against S. mutans colonization of tooth surfaces.

The GLU-binding salivary factor consisted of a high-molecular-weight glycoprotein complexed with α-amylase and was present in submandibular-sublingual saliva and, to a lesser extent, in parotid saliva. The submandibular-sublingual saliva is rich in high-molecular-weight mucous glycoproteins (7) which form complexes with various salivary molecules, including α-amylase (11). On the other hand, parotid saliva is the main source of free α-amylase (2). The binding of GLU by α-amylase is a novel function of this starch-digesting enzyme. It is thought that α-amylase affects bacterial colonization and metabolism through its ability to selectively bind several oral streptococci (29). Bacterial surface-bound α-amylase activity may provide sugar substrates which can be metabolized by the colonizing bacteria to lactic acid, thus contributing to tooth demineralization (29).

Free α-amylase appears to interact with GLU, but it must be complexed with a high-molecular-weight glycoprotein to effectively inhibit GTF activity. Distinct properties of heterotypic complexes of α-amylase and other salivary proteins compared to the properties of uncomplexed α-amylase have been described previously (25). Specifically, Streptococcus sanguis has been shown to bind an amylase-secretory IgA complex, even though S. sanguis does not bind purified amylase. Also, layers of Streptococcus oralis implanted in the oral cavity were able to bind α-amylase despite the absence of any known amylase receptor (26), and it was speculated that amylase may have complexed with a mucin which facilitated the binding. Similarly, the heterotypic complexing of α-amylase and mucin could account for the distinct properties of free and complexed α-amylase.

There are two plausible explanations for why the high-molecular-weight glycoprotein-α-amylase complex (GIF) is a strong inhibitor of GTF activity, in contrast to free α-amylase. GIF may inhibit GTF-substrate interactions via effective steric hindrance due to its relatively large size compared to the size of free α-amylase. Alternatively, the high-molecular-weight glycoprotein may act by concentrating α-amylase molecules, and thus the resulting binding by GIF to GLU would be of greater avidity than that of free α-amylase. In this regard, the presence of repeats in the amino acid sequence of GLU (3, 31) makes it possible that a single GTF molecule could be bound by multiple high-molecular-weight glycoprotein-associated α-amylase molecules. These are not mutually exclusive mechanisms and may act cooperatively.

Despite using several approaches to elute GIF from Sepharose-immobilized GLU, including chaotropic agents such as urea and guanidium-HCl (data not shown), as well as low and high pH in the presence of high salt concentrations, we could not completely elute GIF from the column. The use of 0.1 M Tris-HCl (pH 10.0) containing 1 M NaCl gave the best results, although substantial amounts of α-amylase remained bound to the GLU-Sepharose. It is possible that the eluted portion of GIF was the portion with the lowest avidity for GLU binding due to lower amounts of complexed α-amylase. Indeed, the affinity-purified GIF had a substantially lower α-amylase content than the GIF purified by size exclusion chromatography, although the SDS-PAGE patterns of heated and reduced samples and the intensities of the bands corresponding to the high-molecular-weight glycoprotein component were similar.

As the GLU domain is required for glucan synthesis by GTF (1, 14), it follows that blocking of this domain by GIF suppresses GTF enzymatic activity. However, an additional mechanism may be the concurrent blocking of the CAT domain of GTF due to the large size of GIF. In this regard, we have shown that GIF blocks the binding of anti-CAT antibodies to cell-associated GTF, although it did not substantially affect the binding of anti-S. mutans antibodies to the same substrate (i.e., the S. mutans cell surface).

The exact nature of the GIF-GTF interaction is not known, although it is possible that the amino acid sequence repeats of GLU are pattern recognition sites. Pattern recognition is mediated by germ line-encoded molecules which evolved by natural selection to detect pathogens (19, 20). Because such recognition involves molecular patterns rather than specific amino acid sequences, the interactions are not highly specific and do not preclude binding of several microbial products by the same host molecule. The parotid salivary agglutinin which interacts with the alanine-rich repeat region of S. mutans AgI/II adhesin (8, 9) is an example. This agglutinin was recently shown to be identical to the lung scavenger receptor gp-340, which appears to be a pattern recognition receptor that recognizes at least 14 other bacteria, including Helicobacter pylori, Streptococcus pyogenes, and Streptococcus agalactiae (22). We do not know whether GIF is a pattern recognition complex capable of recognizing other virulence proteins from oral pathogens and thus whether it has more general biological significance in the oral cavity. This may be true for the α-amylase component of GIF, which in addition to GLU recognizes distinct amylase-binding proteins of several oral streptococci, such as Streptococcus gordonii, Streptococcus mitis, Streptococcus cristatus, Streptococcus anginosus, and Streptococcus parasanguinis (5).

GTF is a major colonization factor of S. mutans and has attracted considerable interest as a target for a mucosal vaccine against dental caries (10). GIF may contribute to the control of S. mutans colonization in the oral cavity by functional inhibition of GTF. The inhibitory effect of GIF may depend on its concentration in saliva, and epidemiological studies could provide clinical data supporting the biologic significance of GIF. However, it has been difficult to obtain in vivo evidence that a postulated defense factor has a significant effect on oral ecology or disease prevalence (25). This is particularly true for a disease with a multifactorial etiology, like dental caries. In this regard, colonization by S. mutans is probably influenced by a variety of factors, including (but not limited to) GIF, salivary IgA antibodies to GTF and other virulence factors, and the amount of sucrose in the diet. The role of GIF may be especially important in young children before their mucosal immune systems have had time to develop specific immune responses. Specific immunity against S. mutans virulence factors is not generally present during the colonization period of S. mutans (33), which extends from 19 to 31 months of age (6). During this time, GIF probably provides a first line of defense before the induction of adaptive immune responses. Although GTF, like other bacterial proteins, could mutate to avoid GIF recognition, its GLU domain contains molecular motifs that are critical for the survival of S. mutans. Therefore, GLU probably remains relatively invariant (and thus potentially recognizable by innate immunity components, such as GIF) despite negative selective pressure by the innate immune system.

In conclusion, we identified a putative innate defense factor in human saliva which inhibits GTF, a major colonization factor of S. mutans. This salivary component, designated GIF, consists of a high-molecular-weight glycoprotein complexed to α-amylase. Both components appear to be required for inhibitory activity against GTF. GIF may contribute to the regulation of oral colonization by S. mutans, especially prior to the induction of specific secretory immunity against this cariogenic organism. If similar complexing occurs between pancreatic α-amylase and mucins or other high-molecular-weight glycoproteins, it is possible that they play a role in regulation of microbial colonization in the gut. Studies designed to examine possible interactions between GIF and other bacterial pathogens could determine if GIF has more general significance for the microbial ecology of the oral cavity or even other places where α-amylase occurs.