Materials and Methods

Saliva Samples of unstimulated whole saliva were obtained from 7 infants and from 5 healthy adult volunteers, by use of a method described elsewhere [15]. All infants were inpatients in the Neonatal Intensive Care Unit of Nagano Children’s Hospital, and all were receiving mothers’ milk. Before the study, formal permission was obtained from the hospital’s ethics committee, and informed consent was obtained from the parents of each infant and from each adult volunteer

Strains and growth conditions of microorganisms The microorganisms used were Streptococcus sanguinis ATCC10556^T, Streptococcus gordonii ATCC10558^T, Streptococcus oralis NCTC11427^T, Streptococcus mitis NCTC12261^T, catalase-negative Staphylococcus aureus TW 4632, and clinical isolates of Streptococcus sanguinis, C. albicans and MRSA. These strains were cultured on brain-heart infusion (BHI) broth (Nippon Becton Dickinson). Identification and susceptibility testing were accomplished by use of either MicroScan WalkAway (Dade) or a method described elsewhere [16]

Depletion of IgA or IgG from saliva SIgA or secretory IgG was depleted from saliva by repeatedly passing the latter through a monoclonal anti-IgA–coupled (Hytest) or anti-IgG–coupled (Japan Biotest) affinity column (1 mL, HiTrap N-hydroxy-succinimide–activated high performance; Amersham Biosciences), until no residual antibodies were detected by ELISA. The concentration of salivary IgA was determined by use of a commercially available ELISA protein-detector kit (Kirkegaard & Perry Laboratories), according to the manufacturer’s instructions

Aggregation and coaggregation Viridans streptococci, MRSA (unstained or stained with ethidium bromide), or both were mixed, on a glass slide for several minutes, with either intact saliva, IgA- or IgG-depleted saliva, preadsorbed saliva (saliva adsorbed with MRSA), or colostral SIgA (Cappel), and the aggregates were observed macro- and microscopically. For study of the formation of bacterial aggregates in vivo, oral cavities were swabbed and were stained with May-Grünwald-Giemsa solution

Flow cytometry MRSA (1×10⁷ cfu) was incubated, for 30 min at room temperature, in either 0.5 mL of 50% saliva in PBS, 2 mg of colostral SIgA, or PBS. After being washed in PBS, MRSA was stained with rabbit F(ab)′₂ anti–human IgA conjugated to fluorescein isothiocyanate (Dako) and were fixed with 2% paraformaldehyde. A total of ∼10⁴ cells were counted in a gated region comprising unclamped cells, by use of a FACScan (Becton Dickinson)

2-Dimensional gel electrophoresis, and analysis by matrix-assisted laser-desorption ionization time-of-flight mass spectrometry (MALDI TOF MS) analysis Proteins were analyzed according to the method described by Yao et al. [17]. A 2-mL aliquot of Streptococcus sanguinis cell sediments (a result of centrifugation at 1300 g for 30 min at 4°C) was incubated, for 60 min at room temperature, with 10 mL of saliva and was washed 3 times with PBS. Salivary components bound to the cell surface were extracted with 4 mol/L NaCl. Streptococcus sanguinis proteins were extracted with a mixture of 8 mol/L urea, 2 mol/L thiourea, and 4% 3-[(cholamidopropyl)dimethylammonio]-1-propane sulfonate [18, 19]. The solvents of these extracted proteins were replaced with either PBS or 8 mol/L urea, by use of a centrifugal filter device (Millipore). Streptococcus sanguinis proteins bound to SIgA were prepared, by use of a colostral SIgA–coupled affinity column. These proteins were analyzed by 2-dimensional gel electrophoresis and MALDI TOF MS (Applied Biosystems). Peptide-mass fingerprints were used to screen tryptic-fragment libraries, by use of Mascot (Matrix Sciences) and Protein Prospector (University of California, San Francisco). All experiments were performed 3 times

Assay of catalase activity MRSA or C. albicans (100 μL) was incubated in 2.9 mL of substrate solution (0.1 mL of 30% H₂O₂ in 50 mL of 0.05 mol/L PBS [pH 7.0]), and the decrease in absorbance at 240 nm at 25°C was measured 2 times. One catalase unit was defined as the amount that results in decomposition at a rate of 1.0 μmol of H₂O₂/min by 1×10⁸ cfu, at pH 7.0 and 25°C [2]

Indigo carmine–oxidation reaction, in PBS and in saliva The method described by Wentworth et al. [14] was used to determine the indigo carmine–oxidation reaction. In each well of a 96-well microtiter plate, 3 μL of indigo carmine (1 mmol/L), which reacts with O₃, ¹O₂, and H₂O₃, was mixed with a solution of PBS (pH 7.4), 5 μL of MRSA in PBS at various concentrations (2.5×10⁵, 1.0×10⁶, 2.0×10⁶, and 4.0×10⁶), 1 μL of 1% H₂O₂, and 91 μL of either PBS or saliva and was incubated at 37°C. At various time points, the decrease in absorbance at 600 nm was measured, 3 times, by use of a microtiter plate reader (Toso). In some experiments, various doses (0.03–300 mU) of either Aspergillus niger catalase (Sigma-Aldrich) or Staphylococcus aureus TW4632 (catalase-negative strain) were used instead of MRSA. To measure SIgA interference with the reaction, 0, 0.5, or 2.0 mg of human colostral SIgA was added to the reaction mixture comprising indigo carmine, MRSA (2.5×10⁵ cfu), H₂O₂, and PBS. Each experiment was performed 5 times

Assay of 4-vinyl benzoic acid oxidation The method described by Wentworth et al. [14] was used to measure 4-vinyl benzoic acid oxidation. Saliva (150 μL) was mixed with a solution of 1 mmol/L of 4-vinyl benzoic acid, which reacts with O₃, in 150 μL of PBS (pH 7.4), 30 μL of 1% H₂O₂, and 1.1 mg of human colostral SIgA in 20 μL of PBS and was incubated for 3 h at room temperature. Aliquots (20 μL) were removed and diluted 1:3 in an acetonitrile:water (1:1) mixture. The product composition was determined by reversed-phase high-performance liquid chromatography performed by use of a Jasco LC-800 apparatus (Nihon Bunko) with a Shim-Pack CLC-ODS column. Products were detected by UV light at 254 nm (retention time [RT] for 4-vinyl benzoic acid, 11.47 min; RT for 4-carboxybenzaldehyde, 3.83 min; RT for 4-oxiranyl-benzoic acid, 4.31 min). In the control assay, 4-vinyl benzoic acid was oxidized by irradiation on a transilluminator (312 nm, 0.8 mW/cm²), at room temperature in the presence of SIgA, and the composition of the products was determined by a similar method

Killing of MRSA by Streptococcus sanguinis. Killing of MRSA by Streptococcus sanguinis was measured by use of 2 small (0.4 cm³) chambers separated by a dialyzing-cellulose membrane (Wako Pure Chemical) made with silicon-rubber sheets, between 2 glass slides. To 1 of these chambers, 1×10⁴ cfu of MRSA in BHI broth was added; to the other chamber, 5×10⁶ cfu of Streptococcus sanguinis was added; and both chambers were cultured for 16 h at 37°C. Cells in each chamber were recovered and were stained with 0.01% acridine orange in PBS for 45 s and then were viewed under a fluorescence microscope [20, 21]. In the control study, the same doses of MRSA were cultured in a similar device but in only 1 chamber. SIgA’s augmentation of the killing of MRSA by Streptococcus sanguinis was measured. MRSA (25 μL, 230 cfu) was mixed with various doses (200–3200 cfu) of Streptococcus sanguinis in intact, IgA- or IgG-depleted saliva (125 μL), and the solution was incubated for 4 h at 37°C. MRSA viability was measured on the basis of the rate of recovery in the number of colony-forming units in ortho-phthaldialdehyde/Staphylococcus medium (Nippon Becton Dickinson). These experiments were performed 3 times

Killing of MRSA by Streptococcus sanguinis in 50% saliva–agarose plates MRSA (3 mL, 1×10⁸ cfu) in PBS was mixed with 9 mL of BHI broth (preincubated for 10 min at 50°C) containing 50 mg of low-melting-point agarose (Takara) and 6 mL of saliva, and the solution was poured onto a plastic plate. After the solution solidified, 3.3 mg of colostral SIgA (20 μL) was spotted onto 3 points on the plate and was allowed to soak into the gel, after which 1×10⁸, 1×10⁷, or 1×10⁶ cfu of Streptococcus sanguinis (ATCC10556^T) (10 μL) were each spotted onto 1 of these SIgA-soaked points; the same dose of Streptococcus sanguinis solution, but without SIgA, was similarly spotted onto 3 points on the plate. The plates were cultured overnight at 35°C in 5% CO₂. This experiment was performed 2 times

Killing of C. albicans by H ₂ O ₂ in 50% saliva–agarose plates C. albicans (1×10⁸ cfu) was embedded in saliva/BHI-broth–agarose plates by use of the method described above, and colostral SIgA was soaked into the gel. A. niger catalase (0.3 or 5.0 U) was spotted onto the points soaked with SIgA and onto 2 points not soaked with SIgA. Then, 1% H₂O₂ (10 μL) was spotted onto the points soaked with SIgA, catalase, or both and onto a point not soaked with either SIgA or catalase. The plate was cultured overnight at 35°C in 5% CO₂. This experiment was performed 2 times

Statistical analysis Student’s paired t test was used to compare, at the 3 time points (30, 60, and 90 min), differences in either the growth of MRSA cocultured with Streptococcus sanguinis or the absorbance in the indigo carmine–oxidation reaction. P<.05 was considered to be significant

Results

Aggregation and coaggregation Figure 1 A shows a representative sample of bacterial aggregate in the oral cavity of an infant. Although the bacterial composition of this aggregate was not determined, similar aggregates were easily formed in vitro by simple agitation of viridans streptococci, either in saliva (figure 1 B) or in colostral SIgA (table 1). However, depletion of IgA (figure 1 C), but not of IgG, from saliva almost completely abolished aggregation. These data indicate that both SIgA and viridans streptococci are important contributors to bacterial aggregation in vivo. The concentration of IgA in the 7 infants’ saliva (33.0±30.4 μg/mL) was not significantly different from that in the 5 adults’ saliva (78.2±90.2 μg/mL) (P=.406) (see table 1); in contrast, none of the saliva and none of the colostral SIgA formed MRSA aggregates, although both saliva from the 5 adults and from 1 infant (infant 1) and colostral SIgA contained anti-MRSA SIgA (figure 2). Coaggregation between MRSA and Streptococcus sanguinis occurred in the saliva of all 5 adults (figure 3 A and 3 C) and in that of 1 infant (infant 4, without anti-MRSA SIgA). The saliva of infant 1 contained anti-MRSA SIgA but did not form coaggregates (see table 1). Preadsorption of the 5 adults’ saliva with MRSA successfully inhibited coaggregation but not Streptococcus sanguinis aggregation (figure 3 B and 3 D), which suggests that a salivary component other than anti-MRSA SIgA is necessary to bind viridans streptococci aggregates to MRSA

Binding of salivary proteins to the Streptococcus sanguinis cell surface and binding of colostral SIgA to S. sanguinis.  Salivary IgA and β-catenin bound to the cell surface of Streptococcus sanguinis; of the cell-surface proteins of Streptococcus sanguinis fructose-bisphosphate aldolase and some manganese-dependent dismutase bound to the SIgA-coupled affinity column (table 2)

Decomposition of H ₂ O ₂ by bacterial cells, in PBS and in saliva The average catalase activity of 5 strains of MRSA was 0.43±0.24 U/1×10⁸ cfu; that of C. albicans was 0.03±0.01 U/1×10⁸ cfu. In PBS, MRSA catalase decomposed H₂O₂ (figure 4 A line m), whereas reagent H₂O₂ remained stable (figure 4 A line s). Very small but detectable amounts of oxidants were produced during the spontaneous degradation of H₂O₂ in the presence of indigo carmine (figure 4 B line s), and the velocity of the reaction was proportional to the H₂O₂ concentration (data not shown); MRSA catalase inhibited this oxidant formation, in a dose-dependent manner (as shown by figure 4 B lines m1 [2.5×10⁵ cfu of MRSA, equivalent to the activity with 1.4 mU of catalase] and m2 [1.0×10⁶ cfu of MRSA, equivalent to the activity with 5.5 mU of catalase]; overall, the comparative oxidant formation in these 3 situations, expressed in terms of the lines in figure 4 B was s > m1 [P<.02] and m1 > m2 [P<.04]). A. niger catalase (30 mU) also inhibited this oxidant formation (figure 4 B line c). These data show that, in PBS, both A. niger catalase and MRSA catalase reduce the bactericidal potency of H₂O₂

Salivary peroxidase decomposed H₂O₂ and produced more indigo carmine–reactive oxidant (figure 5 A line m0) than was produced in the spontaneous decomposition of H₂O₂ (figure 5 A line s) (P<.001). However, in contrast to the oxidation-reaction outcome in PBS, MRSA in saliva did not always inhibit the oxidant formation by salivary peroxidase: at smaller doses (2.5×10⁵ cfu [figure 5 B line m1] and 1.0×10⁶ cfu [figure 5 B line m2) of MRSA, MRSA catalase increased the oxidant-producing activity of salivary peroxidase (figure 5 B line m0) (for m0 vs. m1, P<.02); at larger doses (2×10⁶ cfu [equivalent to the activity with 11.0 mU of catalase] [figure 5 B line m3] and 4×10⁶ cfu [equivalent to the activity with 22.0 mU of catalase] [figure 5 B line m4]) of MRSA, this activity was inhibited (for m0 vs. m4, P<.05). For A. niger in saliva, the results were similar: 0.03 mU (figure 5 C line c1) of A. niger catalase increased the oxidant-producing activity of salivary peroxidase (figure 5 C line m0) (P<.003), 0.3 mU of this catalase (figure 5 C line c2) had no effect on this activity, and 300 mU of this catalase (figure 5 C line c3) inhibited this activity (P<.001). Staphylococcus aureus (catalase-negative strain TW4632) in saliva had no effect on the oxidant-producing activity of salivary peroxidase, at any of the 3 doses used (2.5×10⁵ cfu [figure 5 D line a1], 2.5×10⁶ cfu [figure 5 D line a2], and 5×10⁶ cfu [figure 5 D line a3)

SIgA-mediated interference with the indigo carmine–oxidation reaction After 180 min, exogenous SIgA (0.5 or 2.0 mg) decreased both the rate of the indigo carmine–oxidation reaction by salivary peroxidase and the rate of the MRSA catalase–catalyzed H₂O₂-decomposition reaction, by 11.9% and 28.6% (P<.001), respectively (data not shown). This result suggests that SIgA mediates the production of an indigo carmine–nonreactive product, presumably the reformation of H₂O₂. If so, then the indigo carmine–reactive components—ozone and ¹O₂—are both products of the reaction. Next, we tried to ascertain whether the reaction products included ozone

Assay of 4-vinyl benzoic acid oxidation Ozonolysis of 4-vinyl benzoic acid was measured by reversed-phase high-performance liquid chromatography. In the presence of SIgA and saliva, H₂O₂ oxidized 4-vinyl benzoic acid (RT, 11.47 min) and produced 4-carboxybenzaldehyde (RT, 3.83 min) and 4-oxiranyl-benzoic acid (RT, 4.31 min), whereas neither H₂O₂ in PBS nor SIgA and saliva without H₂O₂ oxidized it; and 4-vinyl benzoic acid was similarly oxidized by ozone produced by UV irradiation (data not shown). These data show that H₂O₂ is decomposed by saliva and is converted to ozone by the catalytic activity of SIgA

Killing of MRSA by Streptococcus sanguinis. Staining with acridine orange showed that the effector molecule(s) produced by Streptococcus sanguinis passed through the cellulose membrane and killed MRSA in the neighboring chamber (figure 6 A no fluorescence), whereas Streptococcus sanguinis (figure 6 B orange fluorescence) and MRSA in a single-chamber device (figure 6 C) survived. MRSA killed by reagent H₂O₂ did not emit fluorescence (figure 6 D). SIgA augmented this bactericidal activity of Streptococcus sanguinis. In the coaggregate (intact saliva), Streptococcus sanguinis decreased the survival rate of MRSA in a dose-dependent manner, compared with the rate when MRSA was incubated alone (P<.001). Initially, complete killing of 230 cfu of MRSA required >7 times more (⩾1600 cfu of) Streptococcus sanguinis (figure 7). However, Streptococcus sanguinis decreased the survival rate of MRSA similarly in IgA-depleted saliva; the survival rate of MRSA recovered completely in SIgA-depleted saliva, whereas it did not recover in either intact or IgG-depleted saliva (P<.001)

Killing of MRSA by Streptococcus sanguinis and killing of C. albicans by H ₂ O ₂ , in a 50% saliva–agarose plate MRSA cultured on a 50% saliva–agarose plate produced confluent opaque colonies. Around the Streptococcus sanguinis colonies without exogenous SIgA, MRSA was killed by 1×10⁸ cfu of Streptococcus sanguinis and by 1×10⁷ cfu of Streptococcus sanguinis and the plate remained clear; however, MRSA was not killed by 1×10⁶ cfu of Streptococcus sanguinis. To assess the level of killing, we compared the squares of the radii of the clear zones. The addition of 3.3 mg of colostral SIgA, which contained anti-MRSA SIgA (see figure 2), augmented the killing of MRSA, and larger areas remained clear around Streptococcus sanguinis at 1×10⁸ cfu, 1×10⁷ cfu (P<.001), and 1×10⁶ cfu (P<.004) (table 3 A)

The data in table 3 B show that very few C. albicans cells, even those just below the spot of reagent H₂O₂ (10 μL, 1%), were killed in the absence of SIgA, even though the plate contained saliva and A. niger catalase—and that the addition of SIgA significantly augmented the killing. Large doses (5 U) of A. niger catalase inhibited the killing of C. albicans even in the presence of SIgA, compared with the killing in cultures without this catalase; however, large doses of this catalase did not completely eliminate production of ¹O₂, because the augmentation in the killing of C. albicans in the presence of SIgA, compared with the killing in the absence of SIgA, indicates that ¹O₂ is present; and small doses (0.3 U) of the catalase augmented the killing of C. albicans in the presence of SIgA, compared with the level of killing in the absence of the catalase (P<.002)

Discussion

The data in the present study demonstrate a previously unrecognized process that may help to maintain homeostasis in the oral cavity and thereby defend the oral environment against pathogens. Coaggregation between viridans streptococci and pathogens, in conjunction with the catabolism of H₂O₂ in the aggregates to produce ozone by SIgA, seems to be the key event in this process. SIgA is responsible for mucosal defense, not only by immune exclusion [22], but also by direct killing of pathogens

SIgA contains various bacteria-binding sites in its glycans, in addition to its 4 Fab binding sites [7, 8], and it binds to bacterial surface proteins, such as fructose bisphosphate aldolase of Streptococcus pneumoniae [23]. In fact, the present study found that SIgA binds to viridans streptococci and forms aggregates. However, such aggregation was not found when MRSA was used, presumably because of the latter’s significant ζ potential (i.e., the electrical potential of the interface between the bacterial surface and the aqueous environment) [24], a characteristic that is an advantage in the development of an MRSA-killing process: even in the presence of anti-MRSA SIgA, MRSA cells invading the oral cavity may remain unaggregated until they associate with numerous viridans streptococci cells [25] already aggregated to various extents. Additional salivary component(s) that can reduce ζ potential may be necessary for coaggregation; however, if such coaggregation occurs, ζ potential may augment bactericidal activity, even in infants’ saliva without anti-MRSA SIgA

Salivary peroxidase decomposes H₂O₂ into an indigo carmine–reactive oxidant, presumably either ¹O₂, O₃, or H₂O₃ [12, 14]. However, this oxidant is not strong enough to kill some resistant microorganisms, such as C. albicans [2]. In the present study, the addition of SIgA and small doses of A. niger catalase significantly augmented this oxidant’s bactericidal and fungicidal effects, and more C. albicans cells were killed by the newly synthesized oxidant. Ozone may be produced by the SIgA-catalyzed water-oxidation reaction [14]. If so, then the oxidant produced before the addition of SIgA is ¹O₂, one of the starting materials of the reaction [12], and the products of 4-vinyl benzoic acid oxidation strongly suggest that ozone is present [14]. In this situation, MRSA catalase may potentiate not the survival but, instead, the suicide of MRSA

None of the viridans streptococci tested in our earlier study were low-H₂O₂–producing strains [2]; and the present study used physiological amounts of SIgA, A. niger catalase, and reagent H₂O₂. In the present study, the concentration of SIgA that was soaked into the plates was 5 times greater than that in intact saliva, and this concentration may be achieved by the coaggregation formation even in infants. Similarly, the concentrations of H₂O₂ and A. niger catalase were equivalent to the amounts produced by 1×10⁸ cfu of either viridans streptococci [25] or MRSA and may be achievable in the oral cavity [2]. Ozone may be toxic to oral epithelia [26]; however, because it is short lived [14], tissue damage is limited to a narrow region around the coaggregates. In BHI broth, >500 viridans streptococci cells are necessary to kill 1 MRSA cell [2], whereas the results of the present study show that only 7 viridans streptococci cells are necessary to kill each MRSA cell in the coaggregates. Therefore, the killing of MRSA and of C. albicans may actually occur in the oral cavity. Mucosal surfaces other than the oral cavity, such as the vaginal mucosa, which harbors H₂O₂-producing lactobacilli, may have a similar system to destroy pathogens [27]

In conclusion, the results of the present study show that, in cooperation with H₂O₂-producing viridans streptococci and salivary components, SIgA can kill MRSA and C. albicans—and that this killing potential is augmented by bacterial catalase