INFECTION AND IMMUNITY, OCt. 1982, p. 233-242 Vol. 38, No. 1 0019-9567/82/100233-10$02.00/0 Copyright C 1982, American Society for

Cell Wall-Associated Protein Antigens of salivarius: Purification, Properties, and Function in Adherence ANTON H. WEERKAMPt* AND TON JACOBS Department ofMicrobiology, Faculty of Science, University of Nijmegen, Toernooiveld, NL-6525 ED Nijmegen, The Netherlands Received 17 February 1982/Accepted 7 June 1982 Three -associated protein antigens (antigens b, c, and d) were isolated from mutanolysin-solubilized cell walls of Streptococcus salivarius HB and purified to apparent homogeneity by a combination of ion-exchange chromatogra- phy, gel filtration, and immunoadsorption chromatography. Antigens b and c were also isolated from culture supernatants. Antigen b consisted of more than 80% protein and had an apparent molecular weight as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis of 320,000. Antigen c consisted of 57% protein, about 30% neutral sugar, and about 13% amino sugar, and its glycopro- tein nature was confirmed by specific staining techniques. During sodium dodecyl sulfate-polyacrylamide gel electrophoresis antigen c resolved into two or more bands, depending on the source or the isolation procedure, in the molecular weight range from 220,000 to 280,000. Antigen d consisted of 95% protein and was observed in sodium dodecyl sulfate-polyacrylamide gel electrophoresis as two bands with molecular weights of 129,000 and 121,000. Under nondenaturing conditions all three antigens had molecular weights in the range from 1 x 106 to 3 x 106 as determined by gel filtration. The amino acid compositions of antigens b, c, and d were characterized by low amounts of basic amino acids and relatively high levels of nonpolar amino acids. Among oral streptococcal species antigens b and c were virtually restricted to strains of S. salivarius and most often to serotype I strains. Antigen b was recognized as the factor that mediates coaggregation of S. salivarius with Veillonella strains. The purified protein retained its biological activity. Antigen c could be linked to functions relating to adhesion of the streptococci to host tissues on the basis of its absence in mutant strains and blocking by specific antisera. The purified molecule had no detectable biological activity. Antigen d could not be linked to an established adhesion function.

The primary event in the initial colonization of a lectin-like mechanism has been proposed on a host by most indigenous and pathogenic bacte- the basis of the inhibition of attachment by ria is the specific attachment of bacterial cells to certain sugars. In previous studies we have used certain surfaces in the host (8). However, in Streptococcus salivarius as a model organism to most cases the precise nature of the bacterial study the relationship between surface proper- surface components which mediate these inter- ties of and the ability of bacteria to actions is not known. Consequently, the mode adhere to and colonize suitable surfaces in a host of interaction between these components and (30-33). ligands on the host tissues is still poorly defined. S. salivarius, which is a successful inhabitant Surface appendages of the bacteria, which are of the human oral cavity, colonizes preferably often referred to as fibrils, fimbriae, or pili, have the tongue dorsum and buccal epithelium (8, 30). been implicated in the initial attachment of vari- This organism is also able to attach to tooth ous gram-negative and gram-positive bacteria to surfaces by means of saliva-derived components host tissues and other bacteria (7, 9, 20). In (30), but it is found only in low numbers in dental many cases proteins have been suggested as the plaque in vivo (8). S. salivarius readily forms adhesion receptors on the bacterial surface, and aggregates with several oral anaerobic, gram- negative bacteria (31). Such interaction may promote a with t Present address: Department of Oral Biology, State Uni- symbiotic relationship (e.g., versity of Groningen, Ant. Deusinglaan 1, 9713 AV Groning- veillonellae [32, 33]) or may serve as a mecha- en, The Netherlands. nism to foster the primary colonization and

233 234 WEERKAMP AND JACOBS INFECT. IMMUN. protection of oxygen-sensitive anaerobes (29, dissolved in a small amount of 50 mM Tris buffer (pH was removed nature of the cell-bound re- 7.5). Insoluble material by centrifuga- 31). Recently, the tion. ceptors involved in these adhesion reactions was The clear preparation obtained from 1.8 g (dry partially elucidated. The adhesion of streptococci weight) of cell walls was chromatographed on a to human oral surfaces requires protease-sensitive DEAE-Sephadex A-25 column (37 by 3.5 cm) in the components (31), and different proteinaceous cell same buffer and eluted at a flow rate of 6 cm/h. After wall constituents mediate coaggregation with Veil- passage of the nonadsorbed material, the column was lonella alcalescens (31, 33) and Fusobacterium further eluted with 800 ml of a linear 0 to 0.5 M NaCl and Bacteroides species (31). The specific nature gradient in Tris buffer. Fractions were monitored for of the adhesion characteristics was corroborated absorbance at 220 nm and for the presence of antigens are absent in by rocket immunoelectrophoresis. Antigen-containing by the finding that specific proteins peak fractions were pooled and concentrated by ultra- the cell walls of mutants which are defective in filtration, using a PM30 filter (Amicon B.V., Ooster- certain adhesion properties (33). hout, The Netherlands). The fractions were purified Cell wall proteins similar to those found in S. further by gel filtration on Sepharose CL-2B columns salivarius have been detected in Streptococcus (50 by 1.5 cm) and subsequently by chromatography mutans and have received attention as possible on immunoaffinity columns (6 by 1.5 cm) containing immunogens for caries prevention (15). Three specific antibodies directed against cell wall antigens such proteins were isolated by Russell and co- to remove contaminant antigens. The latter technique workers (23-26, 37) and were shown not to was used instead of binding to the complementary of the low of these columns or to cross-react weakly immu- antibody because capacity cross-react only and because elution under extreme or denaturing nologically with S. salivarius (23, 24). Another conditions (e.g., elution by chaotropic agents or at a study suggested that some of these proteins may low pH) could thus be avoided. be covalently linked to the struc- Immunoaffinity columns were prepared by covalent- ture of the cell wall (19). ly linking the immunoglobulin fractions of the specific In this paper we describe the purification and antisera to cyanogen bromide-activated Sepharose properties of three cell wall-bound protein anti- beads (Pharmacia Fine Chemicals) according to the gens of S. salivarius HB, two of which appear to specifications of the manufacturer. The immunoglob- be related to specific adhesion properties of this ulins were isolated by using DEAE-Affigel Blue (Bio- Rad Laboratories), concentrated and washed by using bacterium. a membrane filter (type PM30; Amicon Corp.), and finally dissolved in coupling buffer. We used 25 mg of immunoglobulin per g (dry weight) of gel. MATERIALS AND METHODS Antigens that were excreted into the culture medium Bacteria. S. salivarius HB and mutant strains HB-7 during growth were partially purified by ammonium and HB-V5 have been described previously (31, 33). sulfate precipitation of neutralized culture superna- Unless otherwise stated, these strains were grown in tants (70% saturation). The precipitate was dissolved batch cultures in a Trypticase soy broth yeast extract in 50 mM Tris buffer (pH 7.5), and the solution was medium for 16 h at 37°C (33). subsequently brought to 65% (vol/vol) with ethanol. The oral streptococcal strains that were used for The precipitate that formed was collected by centrifu- reference were from our laboratory culture collection. gation and dissolved in about 1% of the original culture Purification of antigens. Triton X-100-treated cell volume of Tris buffer. walls were prepared from S. salivarius HB by the Preparation of antisera. The production of antisera procedure described previously (33). Cell walls were against cell walls of S. salivarius HB in rabbits has solubilized by incubation with mutanolysin (Ml en- been described previously (33). Specific antisera were zyme), an N-acetylmuramidase produced by Strepto- prepared by adsorbing anti-HB serum onto whole cells myces globisporus (preparation generously donated by of mutants HB-V5 and HB-7 (10 mg [dry weight] of K. Yokogawa [36]). Stock solutions of the enzyme (1 cells per ml of antiserum). The specific sera showed mg/ml) were prepared by dissolving the lyophilized single precipitation lines on Grabar-Williams immuno- powder in 40 mM sodium phosphate buffer (pH 6.2) electrophoresis gels with complete cell wall digests of containing 0.05% sodium azide and blending with a strain HB. Vortex mixer. The slightly turbid solution was clari- Streptococcal group K typing serum was obtained fied by centrifugation for 15 min at 40,000 x g. The from Wellcome Reagents Ltd., Beckenham, England. final preparation exhibited a slight proteolytic activity Immunological procedures. Standard immunoelec- (28). The cell walls were treated with the enzyme in a trophoretic and immunodiffusion techniques were per- wall-to-enzyme ratio of 200:1 (wt/wt) for 24 h at 37°C. formed with agarose gels (-Mr = 0.13; Bio-Rad) made After centrifugation for 20 min at 40,000 x g, the in Tris-Veronal buffer (ionic strength, 0.1 M; pH 8.6) insoluble residue was incubated for 24 h in one-half of containing 0.02% sodium azide and 1.2 mM calcium the original volume of enzyme solution. This proce- lactate. Electrophoresis was carried out at 10°C and 10 dure solubilized 70 to 80% of the cell wall dry weight. V/cm (33). The supernatants of both incubation mixtures were For localization of antigens in polyacrylamide gels, combined, and ethanol was added to a final concentra- an immunooverlay technique was used. Samples were tion of 65% (vol/vol). After standing at room tempera- run in duplicate on each half of a 5 to 9%o gradient slab ture for 20 min, the preparation was centrifuged for 10 gel in the presence of sodium dodecyl sulfate (SDS). min at 8,000 x g. The dried pellet was subsequently After separation the gel was sliced in half lengthwise, VOL. 38, 1982 S. SALIVARIUS CELL WALL PROTEIN ANTIGENS 235 and one-half was overlaid with 0.6% agarose in Tris- Veronal buffer containing an appropriate antiserum at 0 a concentration of 10o (vol/vol). After overnight incu- bation at room temperature in a moist chamber, the agarose was floated off in 0.1 M sodium chloride, ...... dried, washed extensively, and finally stained with Coomassie brilliant blue. The other half of the gel was stained directly with Coomassie brilliant blue to devel- op the corresponding protein pattern. In some experi- 0 ments it was necessary to transfer the proteins electro- phoretically to the agarose overlay. To do this, the gel sandwich was covered with a 1.5-cm thickness offilter paper prewetted in Tris-Veronal buffer and then placed in an electroblotting cell with the agarose facing the anode. Electrophoresis was done overnight at 5 V/ 0 cm. Analytical procedures. Polyacrylamide gel electro- phoresis (PAGE) in slab gels in the presence of SDS was performed by the method of Lugtenberg et al. C M - --- -.WNW----. (16), except that the acrylamide concentration in the running gel was either 7% (wtlvol) or a 5 to 9%o linear COMo :' gradient. Proteins were stained with Coomassie bril- liant blue, and were detected with the a b c d periodate-Schiff reagent. Isoelectric points were estimated by isoelectric fo- FIG. 1 Immunoelectrophoretic analysis of mutano- cusing in agarose-IEF (Pharmacia), using a pH 2.5 to lysin-solubilized cell walls and purified cell wall anti- 6.0 gradient, as previously described (33). gens. The troughs contained complete antiserum Protein concentrations were measured by using a against strain HB cell walls. COM, Crude cell wall modified Lowry procedure, as described by Peterson digest. (21), and bovine serum albumin as the standard. Total hexose was estimated with the anthrone reagent, using glucose as the standard and the micromethod de- tested for the presence of antigens. The effectiveness scribed by Jermyn (13). Phosphorus determinations of the procedure was tested by wet weight determina- were carried out by the method of Chen et al. (3). tions of the cell pellets before and after mutanolysin Amino acid and amino sugar analyses were performed treatment and in no case was significantly lower than with a Jeol amino acid analyzer, after the samples the effectiveness with strain HB. The general suscepti- were hydrolyzed with 5.7 N HCI in vacuo for 24 h at bility of oral streptococci to this enzyme has been 105°C and for 6 h at 100°C, respectively. Norleucin demonstrated previously (35). A 10-fold-diluted prepa- was added to the samples before hydrolysis as an ration of strain HB clearly allowed the detection of all internal standard. Muramic acid was also measured antigens tested. colorimetrically by the method of Hadzija (10). Neu- The presence of group K antigen was detected by tral sugars were analyzed by thin-layer chromatogra- counterimmunoelectrophoresis of extracts prepared phy after hydrolysis of the samples in 2 N sulfuric acid by the method of Ederer et al. (6). for 2 h at 100°C. Hydrolysates were neutralized with Aggregation and adhesion assays. All assays were barium hydroxyde and centrifuged, and the clear su- done as described in detail previously (31, 33). Incuba- pernatants were eluted on a column of Dowex 50-X4 tion mixtures contained (in a final volume of 150 ,u) 50 (H+ form) in water. Fractions were lyophilized, dis- p.l of a Veillonella cell suspension (absorbance at 660 solved in a small volume of water, and spotted onto nm, 4.0) or 50 ,ul of a suspension of formalinized silica gel-coated sheets (E. Merck AG, Darmstandt, erythrocytes (5%, vol/vol) and, if appropriate, 50 pJ of Germany). The plates were developed by the method an S. salivarius HB cell suspension (absorbance at 660 of Hay et al. (12) and sprayed with N-(1-napthyl)ethyl- nm, 2.0). To test the aggregating potency of the enediamine reagent (1). Cochromatography of known purified antigens, these antigens were added to final quantities of reference sugars allowed semiquantita- concentrations of up to 1 mg (dry weight) per ml. The tive estimations of the sugar residues in the antigen effects of antibodies on the aggregation reactions were preparations. tested by preincubating the streptococci or purified Presence of antigens in oral streptococcal strains. antigens with the appropriate antisera (final concentra- Screening for the presence of antigens b and c in a tion, 5%, vol/vol) for 15 min before admixing the variety of laboratory strains and freshly isolated bacte- aggregating substrate. Aggregation was scored on a ria was performed by immuno double diffusion and scale from 0 (no aggregation) to 4+ (rapid aggregation, counterimmunoelectrophoresis of mutanolysin digests with complete settling of the aggregates). of whole bacterial cells against complete anti-HB serum and antigen b- and c-specific antisera. Cells RESULTS from 10-ml overnight cultures were washed twice with cell wall-asso- phosphate-buffered saline and finally suspended in 50 Identification and purification of >1 of lysis buffer containing 50 1Lg of mutanolysin per ciated antigens. Solubilization of Triton X-100- ml. After 48 h of incubation at 37°C, each suspension treated cell walls of S. salivarius HB by using was centrifuged, and the resulting supernatant was either lysozyme or Ml muramidase enabled the 236 WEERKAMP AND JACOBS INFECT. IMMUN. A B C 0 in mutant HB-7 (33). Antigens b and c were also detected in supematants of strain HB cultures. I ii I 1 i H i ii When crude solubilized cell wall preparations

"' ",- and concentrated culture supernatants were sub- ..o" jected to SDS-PAGE, a relatively simple protein pattern was observed, which mainly consisted of 3 relatively high-molecular-weight proteins (Fig. -220- 2). To identify the positions of the antigens in the polyacrylamide gel, gel strips were overlaid with agarose containing either complete anti-HB se- 135 rum or specific antisera against antigens b and c. The gel sandwiches were then subjected to elec- trophoresis to transfer the proteins to the antise- rum-laden agarose gel. The electrophoretic transfer was necessary because of the very low diffusion rate of the high-molecular-weight pro- tein, which prevented its simple diffusion into the immunooverlay. Figure 2 shows that the highest-molecular-weight band reacted with antigen b-specific antibodies to form an insolu- ble precipitate in the agarose gel, whereas a group of three or four proteins of somewhat FIG. 2. Identification of antigens by SDS-PAGE. lower molecular weight reacted specifically with Lanes i, Mutanolysin digest of strain HB cell walls; the antigen c-specific antiserum. By using the lanes ii, concentrated culture supernatant. (A) Coo- serum massie brilliant blue stained. (B) Immunooverlay with complete anti-HB we identified an addi- antiserum against strain HB cell walls. (C) Immuno- tional antigen that was associated with two overlay with anti-b serum. (D) Immunooverlay with minor protein bands in the muramidase digest, anti-c serum. The markers used were ferritin (halfunit; but not in the medium supernatant. molecular weight, 220,000) and p-galactosidase (mo- For the purification of the antigens, a starting lecular weight, 135,000). material consisting of the mutanolysin digest of 1.8 g (dry weight) of strain HB cell walls was used. The purification procedure which we used detection of at least four antigens (Fig. 1) (33), is described above. Fractionation ofthe prepara- designated antigens a through d. Previously, it tion obtained after ethanol precipitation by has been shown that antigen b is absent in DEAE-Sephadex ion-exchange chromatography mutant strain HB-V5 and that antigen c is absent resulted in the resolution of separate fractions

a -4

44 -0

FRACTION No FIG. 3. Elution profile of DEAE-Sephadex A-25 chromatography of mutanolysin-solubilized S. salivarius HB cell walls. The arrows indicate the start and end of a 0 to 0.5 M sodium chloride gradient, which was followed by a 1 M sodium chloride wash. Solid line, Absorbance at 220 nm; dashed lines, peak height in rocket immunoelectrophoresis. Symbols: V, antigen b; 0, antigen c; *, antigen d. VOL. 38, 1982 S. SALIVARIUS CELL WALL PROTEIN ANTIGENS 237 ii iii iv TABLE 1. Chemical compositions of S. salivarius HB cell wall-associated antigens Composition (%) t".~.U. Antigen Protein Hexose Hexosamine Phosphorus b 83.0 6.4 1.0 0.17 c 57.1 29.2 12.6 0.37 220- 'F d 95.0 10.2 0.9 0.13

135- Nature and compositions of the antigens. The results of SDS-PAGE of the purified antigens are shown in Fig. 4, and these results appeared to be consistent with the location of the antigens in the polyacrylamide gel when an immunoover- 93- lay was used (Fig. 2). Antigen b was identified as a single, homogenous protein band with an ap- parent molecular weight of 320,000. This band was not detectable after staining with periodate- Schiff reagent. Identical results were obtained for this protein when lysozyme-solubilized walls were used and when medium supernatant was used. Antigen c was resolved into two major bands FIG. 4. SDS-PAGE of purified cell wall proteins on (molecular weights, 245,000 and 229,000) and a a 5 to 9o gradient gel. Lane i, Antigen b; lane ii, few additional minor bands in the same molecu- antigen c; lane iii, antigen d; lane iv, crude cell wall lar weight range. The gel patterns obtained with digest. The markers used were ferritin (half unit; antigen c varied slightly depending on whether molecular weight, 220,000), I-galactosidase (molecu- the antigen was extracellular or wall derived and lar weight, 135,000), and phosphorylase B (molecular on the solubilization method used. In all prepa- weight, 93,000). rations the molecular weights were in the range from 220,000 to 280,000. Periodate-Schiff rea- gent clearly stained all protein bands associated containing antigen c, b, and d activities (Fig. 3), with antigen c, which demonstrated the glyco- corresponding to the elution of UV-absorbing protein nature of this compound. No additional material. The weak antigen (antigen a) could not protein or was detected on the gel. be detected in any of the column fractions. Preparations of antigen d produced two pro- Additional UV-absorbing material which did not tein bands during SDS-PAGE; these bands had react with the complete anti-HB serum was molecular weights of 129,000 and 121,000. The eluted from the column. To remove low- and positions of the bands corresponded to the posi- high-molecular-weight materials, the pooled and tions of the unknown precipitates in immuno- concentrated peak fractions were separately overlays with complete anti-HB serum (Fig. 2). eluted on a Sepharose CL-2B column. Such a No staining was obtained with periodate-Schiff procedure could not remove the small amounts reagent. of cross-contaminating antigens, since all three The chemical compositions of the three anti- antigens eluted from the column in about the gen preparations are shown in Table 1. Relative- same fractions (estimated molecular weights, 1 ly small amounts of carbohydrate, phosphorus, x 106 to 3 x 106). Therefore, cross-contaminat- and hexosamine were detected in antigens b and ing antigens were removed by immunoaffinity d. Antigen c contained substantial amounts of chromatography on columns containing either neutral carbohydrate, which is consistent with antigen b- or antigen c-specific immunoglobulin the results obtained by SDS-PAGE. A thin-layer G coupled to CNBr-activated Sepharose beads. chromatographic analysis of the neutral carbo- The final yields of the purified antigens were 3.1 hydrates revealed the presence of glucose, ga- mg (antigen b), 8.5 mg (antigen c), and 1.2 mg lactose, and rhamnose in an approximate molar (antigen d). The purified antigens displayed sin- ratio of 6:1.5:1. No other spots could be detect- gle precipitation lines on Grabar-Williams immu- ed on the chromatograms. In addition to neutral noelectrophoresis gels with complete anti-HB sugar, glucosamine was detected in significant serum (Fig. 1) and were also homogenous after amounts, which suggests the presence of pepti- Laurell rocket immunoelectrophoresis with this doglycan. Although muramic acid was not de- antiserum. tected in hydrolysates of antigen c when the 238 WEERKAMP AND JACOBS INFECT. IMMUN. TABLE 2. Amino acid compositions of S. salivarius HB cell wall protein antigens % of total amino acids in: No. of residues per 1,000 in: Amino acid type Antigen Antigen Antigen Antigen Antigen Antigen b c d b c d Acidic 25.8 21.8 20.6 Aspartic acid 147 99 106 Glutamic acid 111 119 100 Basic 12.3 9.1 7.8 Lysine 100 70 64 Histidine 7 8 2 Arginine 16 13 12 Polar 27.1 37.3 34.8 Threonine 103 84 179 uncharged Serine 72 193 60 Glycine 68 59 91 Tyrosine 28 37 18 Cystine 0 0 0 Nonpolar 34.2 31.3 36.8 Proline 43 49 94 90 117 93 Valine 85 71 9 Methionine 6 2 5 Isoleucine 32 29 29 Leucine 47 31 17 Phenylalanine 39 14 34 amino acid analyzer was used, the presence of in, but they were resistant to pepsin at pH 5.5. this compound could have been obscured by Antigen d appeared to be resistant to the action amino acids under the operating conditions em- of trypsin, as determined by both immunoelec- ployed. Therefore, a specific colorimetric test trophoresis and SDS-PAGE. Furthermore, this was performed with samples of the purified antigen was only partially affected by treatment antigens. Only trace amounts of muramic acid with subtilisin, resulting in a similar pair of could be detected, indicating that no significant protein bands but at a somewhat lower molecu- amounts of peptidoglycan were present. lar weight (about 100,000). Antigen d was com- We found that the gross amino acid composi- pletely destroyed by pronase treatment. tions of the antigens are essentially similar (Ta- Estimation of the amounts of antigens in cell ble 2) and are characterized mainly by low walls. To estimate the absolute amounts of the amounts of basic amino acids and relatively high antigens in Triton X-100-treated cell walls of levels of nonpolar amino acids. However, each strain HB, known quantities of triplicate cell of the antigens is characterized by specific pro- wall preparations were solubilized by treatment portions ofthe individual amino acids. Antigen b with Ml muramidase. The concentrations of the contains high levels of aspartic acid, glutamic antigens in the extracts were then calculated by acid, and threonine; antigen c is rich in serine, using rocket immunoelectrophoresis calibrated followed by glutamic acid and alanine; and in with serial dilutions of known quantities of the antigen d threonine is by far the most abundant purified antigen preparations. The amounts of amino acid, followed by aspartic and glutamic antigens per milligram (dry weight) of wall were acid. estimated to be as follows: antigen b, 30.7 ± 2.2 The isoelectric points of the purified proteins ,ug (mean ± standard error of the mean); antigen were commensurate with their amino acid com- c, 34.2 ± 1.6 ,ug; antigen d, 11.4 ± 0.5 ,ug. positions and thus were found to be in the acidic Assuming that the antigens were uniformly dis- range. The isoelectric points were as follows: tributed over the solubilized and insolubilized antigen b, pH 4.58; antigen c, pH 4.28; antigen wall fractions, and since only relatively small d, pH 3.65. amounts of the antigens were detected in the Effect of proteolysis. To confirm the protein cytoplasmic and membrane fractions (data not nature of the antigens, the influence of proteo- shown), these values also relate to the amounts lytic attack was tested. Measurement of the of antigens per cell. antigen activity by rocket immunoelectrophore- Presence of antigens b and c in oral streptococ- sis showed that treatment with either pronase or ci. A variety of reference strains of oral strepto- subtilisin strongly affected antigens b and c. cocci and freshly isolated S. salivarius strains Trypsin also reduced the activities of these were tested for the presence of antigens b and c. antigens. The loss of antigen activity was ac- Counterimmunoelectrophoresis enabled the de- companied by the disappearance of the corre- tection of both of these antigens in nearly all S. sponding protein bands in SDS-PAGE gels. In salivarius strains which possess the streptococ- addition, the antigens were susceptible to papa- cal group K antigen (Table 3). In addition, a few VOL. 38, 1982 S. SALIVARIUS CELL WALL PROTEIN ANTIGENS 239 TABLE 3. Presence of antigens b and c in TABLE 4. Role of purified antigens and specific mutanolysine digests of oral streptococci antisera in aggregation reactions No. of strains Nofstrains positive Aggregation Reaction with: Organism No.testedof for: substrate Antigen b Antigen c Antigen d Antigen b Antigen c V. alcalescens 4+ (4+)a 0 (4+) 0 (4+) S. salivarius 6 6 5 VI (K+) V. parvula 0 (0) NTb NT S. salivarius 9 1 2 ATCC 10790 (K-) V. alcalescens 0 (0) NT NT S. mutans 11 0 0 Vl + anti-b (serotypes serum a through g) V. alcalescens 4+ (4+) NT NT Streptococcus 3 0 0 Vl + anti-HB- sanguis VS serum Streptococcus 1 0 0 Erythrocytes 0 (4+) 0 (4+) 0 (4+) mitis Erythrocytes + NT 0 (0) NT anti-c serum Erythrocytes + NT 0 (4+) NT anti-HB-7 strains that lacked detectable group K antigen serum showed the presence of either antigen b or a antigen c. Representative strains of other oral The values in parentheses represent incubation streptococcal species lacked detectable levels of mixtures which contained S. salivarius HB cells. these antigens. Control experiments, which b NT, Not tested. were carried out with digests of S. salivarius HB cells, showed that precipitation lines were readi- ly obtained with 10-fold-diluted preparations. bodies was used as an additional test for the Since with most strains the amount of solubi- involvement of antigens in the reaction (Table lized cell material was at least equal to the 4). Antiserum against antigen c completely amount obtained with strain HB, we concluded blocked the reaction of strain HB cells with that antigens b and c are restricted to S. salivar- erythrocytes. In contrast, treatment of bacteria ius. Essentially similar results were obtained by with an antiserum directed against mutant HB-7, using immunofluorescence microscopy of whole which contains antibodies against all antigens cells of streptococci treated with antisera against except antigen c, did not significantly affect antigens b and c and fluorescein-labeled goat hemagglutination. Likewise, specific anti-b se- anti-rabbit immunoglobulin G serum. However, rum inhibited Veillonella coaggregation, but se- cross-reactive antigens were detected in some S. rum against mutant HB-V5 did not. So far, mutans strains by immuno double diffusion tests antigen d has not been implicated in adhesion and immunofluorescence microscopy when the reactions. complete anti-HB serum was used. Cross-reac- tivity between oral streptococcal species has DISCUSSION been reported previously (2, 24, 27). Cell walls of S. salivarius HB contain at least Role of protein antigens in adhesion reactions. four different antigens, designated antigens a To test the possible involvement of the purified through d (33) according to their mobility during antigens in the adhesion of strain HB cells to V. immunoelectrophoresis; three of these antigens alcalescens or erythrocytes, the antigen prepa- have now been identified as the major proteins rations were incubated with these substrates present in the cell wall. Together, these antigens both in the presence and in the absence of strain constitute 5.7% of the total dry weight of puri- HB cells. Table 4 shows that purified antigen b fied cell walls, which is in accordance with the was able to induce agglutination of V. alcales- previously reported content of non-peptidogly- cens Vl, but not of ATCC can amino acids in cell walls of this strain, which 10790, a strain which does not aggregate with was estimated to be about 6% of the dry wall strain HB. Dilution of antigen b preparations mass (33). A similar protein content has been down to 0.1 ,ug (dry weight) per ml induced clear found in the cell walls of S. mutans (5) and other agglutination of a suspension containing 108 oral streptococcal species (Weerkamp, unpub- Veillonella cells per ml. In contrast, prepara- lished data). At least one of our proteins (33), as tions of antigens c and d did not induce a direct well as S. mutans wall-associated protein (19), is agglutination reaction, nor did they inhibit the presumably covalently linked to the peptidogly- aggregation of whole cells of strain HB with can structure. either Veillonella cells or erythrocytes. The concentrations of antigens b and c in the The blocking of aggregation by specific anti- cell walls of batch-grown cultures were very 240 WEERKAMP AND JACOBS INFECT. IMMUN. similar, whereas the concentration of antigen d tigation, proteolytic cleavage may account for was one-third of the other concentrations. How- the somewhat lower molecular weight of this ever, preliminary experiments with chemostat- form of antigen c compared with preparations grown cultures have indicated that the relative obtained with lysozyme or derived from the proportions of the antigens may vary considera- culture supernatant. However, an immunologi- bly depending on the growth conditions (B. C. cally detectable, lower-molecular-weight frag- McBride and A. H. Weerkamp, unpublished ment has not been found. It seems plausible that data). The latter results suggest that the synthe- a common protein component, which bears the sis or incorporation into the cell wall of each antigenic determinant, can be associated with protein is regulated independently. Certain sur- variable amounts of carbohydrate. However, face properties of S. mutans and the release of a the smallest unit detected (220,000 daltons) still cell wall protein antigen into the culture super- stained well for carbohydrate. Peptidoglycan is natant were also reported to depend on the presumably not present in antigen c preparations growth rate and carbon source (11). in significant amounts. A more detailed study of The cell wall antigens described here are partial degradation products of this molecule different from the streptococcal group K carbo- should provide a more definite insight in its hydrate antigen, since solubilized cell wall prep- molecular structure. arations of strain HB did not react with group K In the absence of SDS all three antigens typing serum. Whole cells of this strain pos- formed larger aggregates, with estimated molec- sessed the group K antigen since it was detected ular weights in the range from 1 x 106 to 3 x 106. in extracts prepared by the Lancefield method Since the mobility in SDS-PAGE gels did not or by protease treatment. In addition, mutant change when ,B-mercaptoethanol was omitted, it strains HB-V5 and HB-7, which lack antigens b is not likely that the formation of disulfide bonds and c, respectively, both possess the group K accounts for the increased molecular weight. antigen (33). It is also unlikely that the antigenic Moreover, an amino acid analysis did not detect determinant of any of antigens b, c, and d is the presence of cysteine. A similar phenomenon glycerol teichoic acid, the presence of which has was noted previously for S. mutans wall-associ- been demonstrated in many oral streptococci ated protein antigen I/II (24, 37). Such a behav- (17), because of the susceptibility to proteases, ior could be explained by spontaneous autoag- the relatively low phosphorus content, and the gregation of the solubilized proteins. However, absence of cross-reactivity with other oral strep- we have obtained substantial evidence by using tococci. However, the presence of an antibody cleavable cross-linking agents and differential that cross-reacts with some S. mutans strains immunoprecipitation that in the cell walls the has been detected in complete antisera against proteins are present as homologous, high-molec- strain HB cell walls. So far, this common anti- ular-weight complexes (manuscript in prepara- gen has not been characterized. The existence of tion). In conjunction with a location at the such common antigens in oral streptococci has periphery of the cell, the concept of the situation been reported previously (2, 24, 27). of each antigen on specialized fimbrial classes An analysis of the purified antigens showed seems attractive. The similarity both in the size that antigen b is a protein which may be associ- of the multimeric form and in the gross amino ated with a low amount of carbohydrate and has acid composition compared with fibrillar pro- an apparent molecular weight of 320,000 as teins of other gram-positive bacteria, such as determined by SDS-PAGE. The homogeneity of Actinomyces viscosus (18, 34) and Corynebacte- the preparations, irrespective of whether the rium renale (14), suggests that each of the anti- source is extracellular or cell wall derived by gen complexes may even represent a single using either lysozyme or mutanolysin, suggests fimbrial structure. Such a concept is supported that this form represents the smallest entity. by the recently reported detection of antigenical- Antigen d was similar to antigen b in composi- ly divergent fibrils in a strain of A. viscosus (4). tion, except that it showed two adjoining bands The relatively high content of nonpolar amino of lower molecular weight in preparations ob- acid residues in S. salivarius wall antigens, as tained with lysozyme and mutanolysin. Antigen well as in fibrils of Actinomyces and Corynebac- d was not detected in culture supernatants. In terium species (14, 18, 34), suggests the pres- contrast, antigen c differed in that it contained ence of hydrophobic areas within the molecules. significant amounts of neutral carbohydrate and It should also be noted that S. mutans cell wall- glucosamine and showed micro-heterogeneity associated proteins have been reported to have a depending on its source and the preparation similar composition (23, 25). These observations procedure. Since antigen c is susceptible to a could be of some importance since it has been variety of proteolytic enzymes and some prote- suggested that hydrophobic interactions, partic- ase activity was present in the mutanolysin ularly when associated with fibrillar surface ap- preparation that was used throughout our inves- pendages, may facilitate adhesion by counter- VOL. 38, 1982 S. SALIVARIUS CELL WALL PROTEIN ANTIGENS 241 acting electrostatic repulsive forces between the by monoclonal antibodies. J. Immunol. 127:1318-1322. interacting surfaces (20, 22). 5. Cooper, H. R., F. W. Chorpenning, and S. Rosen. 1975. Preparation and chemical composition of the cell wall of In a previous report we demonstrated that the Streptococcus mutans. Infect. Immun. 11:823-828. protein which corresponds to antigen b is indeed 6. Ederer, G. M., M. M. Herrmann, R. Bruce, J. M. Mat- directly involved in the coaggregation of S. sen, and S. S. Chapman. 1972. Rapid extraction method salivarius serotype I strains with strains of the with Pronase B for grouping a-hemolytic streptococci. Appl. Microbiol. 23:283-288. genus Veillonella (32, 33). Therefore, this pro- 7. Gibbons, R. J. 1977. Adherence of bacteria to host tissue, tein was named Veillonella-binding protein. As p. 395-406. In D. Schlessinger (ed.), Microbiology-1977. little as 0.1 ,ug of the8purified protein induces American Society for Microbiology, Washington, D.C. rapid aggregation of 10 V. alcalescens cells 8. Gibbons, R. J., and J. van Houte. 1975. Bacterial adher- Vl ence in oral . Annu. Rev. Microbiol. under physiological conditions. This Veillonella- 29:19-44. binding protein-induced aggregation is inhibited 9. Gibbons, R. J., and J. van Houte. 1980. Bacterial adher- by the presence of galactose or some galactose ence and the formation of dental plaques, p. 61-104. In E. H. Beachey (ed.), Bacterial adherence. Receptors and analogs (32), suggesting a lectin-like interaction. recognition, ser. B, vol. 6. Chapman and Hall, London. The purification of Veillonella-binding protein 10. Hadzija, 0. 1974. A simple method for the quantitative should now enable a more detailed study of the determination of muramic acid. Anal. Biochem. 60:512- binding characteristics of this protein, which is 517. in 11. Hardy, L., N. A. Jaques, H. Forrester, L. K. Campbell, currently progress. Involvement of antigen c K. W. Knox, and A. J. WIcken. 1981. Effect of fructose in adhesion reactions is less readily demonstrat- and other carbohydrates on the surface properties, lipotei- ed, but the following observations suggest that choic acid production, and extracellular proteins of Strep- this glycoprotein mediates host-related adhesion tococcus mutans Ingbritt grown in continuous culture. Infect. Immun. 31:78-87. functions of the bacteria, such as the binding of 12. Hay, G. W., B. A. Lewis, and F. Smith. 1963. Thin-film epithelial cells or erythrocytes and aggregation chromatography in the study of carbohydrates. J. Chro- with salivary components (31): (i) the glycopro- matogr. 11:479-486. tein is absent in mutant HB-7, which lacks the 13. Jermyn, M. A. 1975. Increasing the sensitivity of the anthrone method for carbohydrate. Anal. Biochem. host-related adhesion functions (31); (ii) these 68:332-335. activities are susceptible to the action of proteo- 14. Kumazawa, N., and R. Yanagawa. 1972. Chemical proper- lytic enzymes; (iii) Triton X-100-treated cell ties of the pili of renale. Infect. Immun. walls retain the adhesive properties, including 5:27-30. 15. Lehner, T., M. W. Russell, J. Caldwell, and R. Smith. sensitivity to protease, but contain only a few 1981. Immunization with purified antigens from Strepto- detectable proteins; and (iv) adhesion to buccal mutans against dental caries in rhesus monkeys. epithelial cells or erythrocytes is specifically Infect. Immun. 34:407-415. blocked by anti-c serum, but not significantly 16. Lugtenberg, B., J. Meijers, R. Peters, P. van der Hoek, and L. van Alphen. 1975. Electrophoretic resolution of the affected by a complete antiserum against cell "major outer membrane protein" of Escherichia coli K12 walls of mutant HB-7, which only lacks antibod- into four bands. FEBS Lett. 58:254-258. ies against antigen c. Although the last effect 17. Markham, J. L., K. W. Knox, A. J. Wicken, and M. J. could have been produced through the masking Hewett. 1975. Formation of extracellular lipotheichoic acid by oral streptococci and lactobacilli. Infect. Immun. of neighboring sites, the body of evidence is 12:378-386. strongly in favor of a direct involvement of this 18. Masuda, N., R. P. Ellen, and D. A. Grove. 1981. Purifica- glycoprotein in adhesion reactions. So far, a tion and characterization of surface fibrils from taxonomi- possible role of antigen d in adhesion reactions cally typical Actinomyces viscosus WVU 627. J. Bacteri- ol. 147:1095-1104. has not been demonstrated, but such a role may 19. Nesbitt, W. E., R. H. Staat, B. Rosan, K. G. Taylor, and be proposed on the basis of its similarity to the R. J. Doyle. 1980. Association of protein with the cell wall other factors. Likewise, such a role could be of Streptococcus mutans. Infect. Immun. 28:118-126. postulated for the comparable S. mutans surface 20. Ofek, I., and E. H. Beachey. 1980. General concepts and principles of bacterial adherence in animals and man, p. 1- proteins. This possibility is supported by the 29. In E. H. Beachey (ed.), Bacterial adherence. Recep- significant caries-reductive effect obtained after tors and recognition, ser. B, vol. 6. Chapman and Hall, immunization with some of these proteins (15). London. 21. Peterson, G. L. 1977. A simplification of the protein assay method of Lowry et al. which is more generally applica- LITERATURE CITED ble. Anal. Biochem. 83:346-356. 1. Bounias, M. 1980. N-(1-naphtyl)ethylenediamine dihy- 22. Pethica, B. A. 1961. The physical chemistry of cell adhe- drochloride as a new reagent for nanomole quantification sion. Exp. Cell Res. Suppl. 8:123-140. of sugars on thin-layer plates by a mathematical calibra- 23. Russell, M. W. 1979. Purification and properties of a tion process. Anal. Biochem. 106:291-295. protein surface antigen of Streptococcus mutans. Micro- 2. Bratthafl, D., and R. M. Pettersson. 1976. Common and bios 25:7-18. unique antigens of Streptococcus mutans. J. Dent. Res. 24. Russell, M. W., L. A. Bergmeier, E. D. Zanders, and T. 55(Special Issue A):A60-A64. Lehner. 1980. Protein antigens of Streptococcus mutans: 3. Chen, P. S., T. Y. Toribara, and H. Warren. 1956. Micro- purification and properties of a double antigen and its determination of phosphorus. Anal. Chem. 28:1756-1758. protease-resistant component. Infect. Immun. 28:486- 4. Cisar, J. O., E. L. Barsumlan, S. H. Curl, A. E. Vatter, 493. A. L. Sandberg, and R. P. Siraganian. 1981. Detection 25. Russell, M. W., E. D. Zanders, L. A. Bergmeler, and T. and localization of a lectin on Actinomyces viscosus T14V Lehner. 1980. Affinity purification and characterization of 242 WEERKAMP AND JACOBS INFECT. IMMUN. protease-susceptible antigen I of Streptococcus mutans. varius. Infect. Immun. 29:459-468. Infect. Immun. 29:999-1006. 32. Weerkamp, A. H., and B. C. McBride. 1980. Studies on a 26. Russell, R. R. B. 1979. Wall-associated protein antigens of veillonella-binding protein of Streptococcus salivarius, p. Streptococcus mutans. J. Gen. Microbiol. 114:109-115. 521-523. In R. C. W. Berkeley et al. (ed.), Microbial 27. Sch6ller, M., J. P. Klein, and R. M. Frank. 1981. Com- adhesion to surfaces. Ellis Horwood Ltd., Chichester, mon antigens of streptococcal and non-streptococcal oral England. bacteria: immunochemical studies of extracellular and cell 33. WVeerkamp, A. H., and B. C. McBride. 1980. Identifica- wall-associated antigens from Streptococcus sanguis, tion of a Streptococcus salivarius cell wall component Streptococcus mutans, Lactobacillus salivarius, and Acti- mediating coaggregation with Veillonella alcalescens VI. nomyces viscosus. Infect. Immun. 31:52-60. Infect. Immun. 32:723-730. 28. Siegel, J. L., S. F. Hurst, E. S. Liberman, S. E. Coleman, 34. Wheeler, T. T., and W. B. Clark. 1980. Fibril-mediated and A. S. Bleiweis. 1981. Mutanolysin-induced sphero- adherence of Actinomyces viscosus to saliva-treated hy- plasts of Streptococcus mutans are true protoplasts. In- droxyapatite. Infect. Immun. 28:577-584. fect. Immun. 31:808-815. 35. Yokogawa, K., S. Kawata, S. Nishimura, Y. Ikeda, and Y. 29. Slots, J., and R. J. Gibbons. 1978. Attachment of Bacte- Yoshimura. 1974. Mutanolysin, bacteriolytic agent for roides melaninogenicus subsp. asaccharolyticus to oral cariogenic streptococci; partial purification and proper- surfaces and its possible role in colonization of the mouth ties. Antimicrob. Agents Chemother. 6:156-165. and of periodontal pockets. Infect. Immun. 19:254-264. 36. Yokogawa, K., S. Kawata, T. Takesmura, and Y. Yoshi- 30. Weerkamp, A. H., and B. C. McBride. 1980. Adherence mura. 1975. Purification and properties of lytic enzymes of Streptococcus salivarius HB and HB-7 to oral surfaces from Streptomyces globisporus 1829. Agric. Biol. Chem. and saliva-coated hydroxyapatite. Infect. Immun. 30:150- 39:1533-1543. 158. 37. Zanders, E. D., and T. Lehner. 1981. Separation and 31. Weerkamp, A. H., and B. C. McBride. 1980. Character- characterization of a protein antigen from cells of Strepto- ization of the adherence properties of Streptococcus sali- coccus mutans. J. Gen. Microbiol. 122:217-225.