INFECTION AND IMMUNITY, Feb. 1987, p. 438-445 Vol. 55, No. 20019-9567/87/020438-08$02.00/0Copyright © 1987, American Society for Microbiology

Relationship of Cell Surface Morphology and Composition ofStreptococcus salivarius K+ to Adherence and Hydrophobicity

ANTON H. WEERKAMP,1* HENNY C. VAN DER MEI,1 AND JAN W. SLOT2Department of Oral Biology, Dental School, University of Groningen, 9713 AV Groningen,l and Department of Electron

Microscopy, Medical School, University of Utrecht, Utrecht,2 The Netherlands

Received 16 April 1986/Accepted 9 October 1986

The cell surfaces of a range of variants of Streptococcus salivarius HB, altered in cell wail antigencomposition, were compared with those of the parent with respect to adherence, ability to adsorb tohexadecane, morphology, and exposure of lipoteichoic acid (LTA). Adherence to host surfaces was measuredby using both saliva-coated hydroxyapatite beads and tissue-cultured HeLa cells, and interbacterial adherencewas measured by using Veillonella alcalescens Vl cells. Progressive loss of the protease-sensitive fibril classeswas generally associated with decreasing ability to adsorb to hexadecane. However, increased exposure ofprotein antigen C (AgC) increased the apparent hydrophobicity of the cell. This correlated with the finding thatAgC was the most hydrophobic of the solubilized fibrillar cell wall antigens. Collectively, this demonstrates thatadsorption to hydrophobic ligands is directly related to the density of the fibrillar layer on the cells and theproperties and surface exposure of specific fibril classes. The involvement of hydrophobic interactions inAgC-associated attachment was suggested by its sensitivity to low levels of the hydrophobic bond-breakingagent tetramethyl urea, although the reduction was not to the level of adherence observed with strains lackingAgC. However, hydrophobicity was less essential to other adherence reactions. Circumstantial evidence,including (i) immunoelectron microscopy, showing that LTA was virtually absent from the fibrillar layer, (ii)whole-cell enzyme-linked immunosorbent assay, suggesting that surface exposure of LTA related inversely tothe density of the fibrillar layer, and (iii) agarose gel electrophoresis, showing that LTA was not specificallyassociated with protein fibrillar antigens, strongly suggested that LTA does not confer hydrophobic propertiesto these cells and is not involved in adherence reactions associated with the cell wall protein antigens.

Microbial adherence to tissues and to other microorgan-isms is thought to be an important factor in the formation ofthe specific microfloras found in association with a host. Inthe human oral cavity, distinct bacterial ecosystems thusform at specific sites, partly owing to interactions betweenbacterial surface components and the oral surfaces. A widevariety of interactions, both long and short range, have beenimplicated in the complex adhesion process, including vander Waals dispersion, hydrogen bonding, and electrostaticand hydrophobic interactions (25).

Several studies have suggested an important role forhydrophobic interactions in microbial adherence (24). Con-sidering their ability to adhere to hydrophobic ligands (3, 9,20-22, 36), many freshly isolated oral bacteria indeed appearto have hydrophobic domains on their surface. However,contact angle measurements on layers of oral streptococcalstrains suggested that most strains possess a relatively lowoverall surface hydrophobicity, particularly after beingcoated with saliva (34). Nonhydrophobic variants of oralstreptococci (8, 10, 18, 34, 37) and cells grown under variousgrowth conditions (12) have been used to relate hydrophobiccell surface properties with the ability to adhere to oralsurfaces. However, these experiments could not conclu-sively demonstrate a direct relationship, since a variety ofsurface components is usually lost from these cells simulta-neously. Recently (3), evidence was supplied by use of oralactinomycetes which showed that although hydrophobicityand adherence to saliva-coated hydroxyapatite (SHA) werestatistically correlated traits, hydrophobic interactions ap-peared not to be of major importance for the adherencereaction itself.

* Corresponding author.

According to one recent model (4), hydrophobic bondsadjacent to stereospecific surface molecules stabilize inter-actions mediated by these compounds. According to macro-scopic models of bacterial adhesion (2, 34) the bacterialsurface hydrophobicity (surface free energy) is an importantfactor in the interaction with other surfaces by virtue of itseffect on the interfacial free energy of adhesion.

Despite the apparent involvement of hydrophobic do-mains in adherence reactions, little is known about thenature of the molecules which confer hydrophobicity to thecell surface of oral and other streptococci. Several authorshave demonstrated a relationship between the presence ofsurface appendages on these bacteria and hydrophobicity (8,10, 18, 26), similar to observations made with gram-negativebacteria (7). Recently, McBride and co-workers (14, 18)showed that the loss of hydrophobicity of Streptococcussanguis and Streptococcus mutans strains coincided withthe loss of several high-molecular-weight proteins from thecell wall. However, no conclusive evidence was given thatone or more of these proteins were directly responsible forhydrophobicity. Similarly, a relationship between the Mprotein of group A streptococci and surface hydrophobicityhas been suggested (26), but later studies have presentedevidence that lipoteichoic acid (LTA) is the major hydropho-bicity-conferring component in these bacteria (15, 16). BothM protein and LTA have been implicated in the adherence ofgroup A streptococci to epithelial surfaces (1, 5).We previously identified two high-molecular-weight

proteinaceous cell wall-associated antigens in Streptococcussalivarius (29), respectively suggested to be involved inadherence to host tissues, including pellicle-coated toothsurfaces, and to mediate coaggregation with Veillonella cellsand other gram-negative bacteria. The latter antigen is

438

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

SURFACE FIBRILS AND HYDROPHOBICITY OF S. SALIVARIUS

TABLE 1. Antigenic composition of S. salivarius strains

Strain Cell wall antigensa Reference

HB B+ C+ D+ E+ F+ 29HB-7 B+ C-b D+ E- F+ 28, 29HB-V5 B- C+ D+ E+ F+ 29, 30HB-V51 B- C-b D+ E- F+ 26HB-C12 B- C- D-c E- F+ This study

a All strains are Lancefield group K+.b Accumulated intracellularly.c Excreted in increased amounts compared to parent strain.

thought to involve a galactose-specific lectin (32, 33). S.salivarius (serogroup K) cells possess a typical fibrillarsurface layer (28), which consists of fibrils of at least threedifferent lengths (28, 35). Recently, we demonstrated thattwo specific fibril classes could be identified with the twoantigens involved in adhesion (35), and we described a seriesof mutant strains which lack specific fibrils and particularadhesive properties (28). This knowledge of the complexsurface topology of these bacteria could be used to locateand identify the molecules or structures which confer hydro-phobicity to the cell surface.

In the present paper, we demonstrate that hydrophobicproperties of the cell surface are associated with specificfibrillar protein antigens on the surface of S. salivarius cells.It is proposed that the hydrophobicity of these compounds isan intrinsic property rather than the result of associationwith LTA.

MATERIALS AND METHODS

Bacterial strains and growth conditions. S. salivarius HBand mutant strains HB-7, HB-V5, and HB-VS1 have beendescribed previously (28, 29, 31). Strain HB-C12 was aspontaneous stable variant obtained after prolonged cultiva-tion of strain HB in a chemostat at high specific growth rate.It lacked the ability to attach to buccal epithelial cells andsaliva-coated SHA and was impaired in coaggregation withveillonellae. The antigenic structures of the strains areshown in Table 1.For each experiment, the bacteria were grown from a

frozen stock in batch culture in Todd-Hewitt broth for 16 hat 37°C. Cells were harvested by centrifugation for 5 min at8,000 x g, washed twice with phosphate-buffered saline(PBS; pH 7.0), and resuspended in PBS.

Preparation of cell wall digests. Procedures for the prepa-ration of cell walls by shaking them with glass beads in aBraun cell disintegrator and subsequently washing them withTriton X-100 were described previously (29). For somepurposes, the Triton X-100 treatment was omitted and thewalls were washed only with PBS. Mutanolysin digests weremade of freeze-dried walls (20 mg [dry weight]) per ml,containing 100 U of mutanolysin; Sigma Chemical Co.) bypreviously described methods (29).

Immunological procedures. Polyvalent monospecific anti-sera against wall antigens B (AgB), C (AgC), and D (AgD)were prepared by immunoaffinity chromatography of wholeantiserum against S. salivarius HB walls on columns con-taining the immobilized purified antigens, as previouslydescribed (28, 29).

Specific antiserum against the polyglycerolphosphatebackbone of Lactobacillus teichoic acid (lot 644; 2.07 mg ofantibody per ml), was kindly donated by K. W. Knox,

School of Dental Medicine, University of Sydney, Sydney,Australia.

Standard crossed-immunoelectrophoretic techniques wereperformed with 1% agarose gels made in Tris-Veronal(Winthrop Laboratories) buffer (pH 8.6), as described pre-viously (29).Immunoblotting on nitrocellulose filter paper (Bio-Rad

Laboratories), washing, blocking, and immunodetectionprocedures with colloidal gold-labeled protein A followed bysilver amplification were carried out by the method ofMoeremans et al. (17). Protein blotting from agarose gelswas done simply by overlaying the gel with nitrocellulosepaper, several layers of filter paper, and a light weight, thusallowing fluid to be removed from the gel by suction.Hexadecane adsorption test. A modification of the method

described by Rosenberg et al. (23) was used for the hexa-decane assays carried out on modified cells. An initial cellconcentration of approximately 2 x 109 cells in a totalvolume of 1.2 ml of PBS and 75 ,u1 of hexadecane was used.The agitation time was 60 s. For comparison of the mutantstrains, the more accurate assay recently described byLichtenberg et al. (13), which is based on a kinetic analysisof the adsorption to hexadecane, was used. The data wereexpressed as the removal coefficient (k), which is the slope ofa linear function between the hexadecane/water ratio and therate constant of removal of the bacteria from the bulkaqueous phase to the hexadecane interface. Although thelines did not pass through the origin as stated in the originalpublication (13), the linear correlation coefficients werealways better than 0.96.Hydrophobic interaction chromatography. To determine

the relative hydrophobicity of cell wall-associated antigens,mutanolysin digests of crude cell wall preparations (nottreated with Triton X-100) were applied to a column ofOctyl-Sepharose 4B (7 by 1 cm; Pharmacia, Inc.)preequilibrated in 50 mM potassium phosphate buffer (pH7.3) containing 25% saturated ammonium sulfate. The sam-ple, previously brought to the same ammonium sulfateconcentration, was eluted with a linear gradient of decreas-ing concentration of ammonium sulfate (25 to 0% saturation)and simultaneously increasing concentration of ethyleneglycol (0 to 50%, vol/vol). Subsequently, the column waseluted with 1 M tetramethyl urea in phosphatee buffer (onecolumn volume) and finally with buffer containing 1%(vol/vol) Triton X-100. Fractions were collected, dialyzedextensively against distilled water, and freeze-dried.

Quantitation of LTA. LTA in culture media and cell walldigests was determined in a semiquantitative manner byusing an immunodot-blotting test. Twofold serial dilutions ofthe test solution (2 IlI) were spotted on nitrocellulose filterand allowed to dry. The nitrocellulose was quenched inTris-buffered saline (pH 7.4) containing 1% bovine serumalbumin for 30 min. Subsequently, the filters were incubatedwith LTA antiserum (diluted 1:200) in TBS containing 0.1%bovine serum albumin or control preimmune serum for 2 h atroom temperature. The filters were than treated as describedabove.The relative concentration of cell surface-located LTA

was determined by a modification of the procedure describedby Miorner et al. (16). A 50-,u aliquot of a washed-cellsuspension containing approximately 5 x 109 cells in Tris-buffered saline-0.4% bovine serum albumin, was incubatedfor 60 min at 30°C with 100 RId of LTA antiserum (diluted1:100) or control preimmune serum. The cells were centri-fuged in an Eppendorf type 5414S centrifuge for 3 min,washed twice with Tris-buffered saline-bovine serum albu-

VOL. 55, 1987 439

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

440 WEERKAMP ET AL.

TABLE 2. Adherence properties and surface hydrophobicity of S. salivarius strains

Adsorption to hexadecane Adherence to SHA Adherence to HeLa cellsb (no. Coaggregation withStrain (k [min-']) (% of HB) (B/U)a (% of HB) of bacteria/cell) (% of HB) Veillonella cells (m)

HB 52 (100) 1.01 (100) 12.3 (100) 0.132 (100)HB-7 44 (84.6) 0.08 (7.9) 2.0 (16.3) NTHB-V5 76 (146.1) 0.92 (91.1) 12.6 (102.4) 0.010 (7.5)HB-V51 27 (51.9) 0.11 (10.9) 2.4 (19.5) 0.009 (6.7)HB-C12 2 (3.8) 0.06 (5.9) 0.4 (3.3) 0.065 (49.2)aExpressed as the ratio of bound over unbound cells.b Average of three different cultures; standard deviation ± 10%.

min, and subsequently incubated with alkaline phosphatase-labeled goat anti-rabbit immunoglobulin G antiserum (di-luted 1:1,000). After another 60 min, the cells were washedtwice and incubated with 200 p.1 of enzyme substrate. Thereaction was terminated after 20 min by the addition of 50 ,u1of 3 N NaOH. The A405 of the supernatant minus that of thecontrol serum was taken as a measure of the surface LTAconcentration. Since with many strains the nonspecific bind-ing of rabbit immunoglobulin G could not be completelyblocked, care was taken to apply the same amount of totalimmunoglobulin G in both the experimental and controlincubations to obtain more realistic base-line values.Immunoelectron microscopy. The subcellular localization

of LTA and cell wall-associated protein antigens was deter-mined by using immunoelectron microscopy on ultrathincryosections. The sandwich labeling technique involving theuse of colloidal gold-complexed protein A was describedpreviously (28).Adherence and aggregation experiments. Adherence of

bacteria to SHA was determined essentially as describedpreviously (30). All experiments were performed in tripli-cate. Since the adherence of S. salivarius HB to SHAinvolves a single high-affinity type of interaction, the indirectmethod of calculating the number of bound cells was used(19).Adherence to confluent monolayers of tissue-cultured

HeLa cells was measured as follows. HeLa cells were grownin RPMI 1640 medium (GIBCO Laboratories) containing10% fetal calf serum in tissue culture polystyrene petridishes (25 cm2). After aspiration of the medium, the celllayers were incubated for 90 min at 37°C with 2 ml of thebacterial suspensions containing 108 cells per ml. Duringincubation, the dishes were slightly agitated on a reciprocalshaking platform (20 strokes per min). The cells werewashed four times with PBS, fixed with 2% glutaraldehyde,and stained with crystal violet. The number of bacteriaattached per cell was estimated under a light microscope bycounting at least 50 cells randomly distributed over thesurface of the petri dish. The average standard deviation inthe counts was 10%.Coaggregation of the streptococci with Veillonella

alcalescens Vl was estimated in a spectrophotometric assayby the method of Ericson and Rundegren (6). To 5 ml of acell suspension in PBS (optical density at 660 nm, 0.60) wasadded 400 p.1 of a V. alcalescens suspension (optical density,2.5), and the extinction was monitored continuously. Thedata were linearized according to the formula (6) ln [(Ao -A)IA] = mt + b, where Ao and A are the optical densities attimes zero and t, respectively. The value of m was used as ameasure of the aggregating activity. Replicate experimentswith the same culture were reproducible within 5% standarddeviation.

RESULTS

Cell surface properties of the bacterial strains. Table 1shows the antigenic composition of the cell walls of S.salivarius HB and the mutant strains derived. At least fiveantigens, labeled B to F, could be observed by crossedimmunoelectrophoresis in the mutanolysin digest of strainHB. The major antigenic components AgB, AgC, and AgDwere identified previously (29). The identity and possiblefunctions ofAgE and AgF have not yet been established. Allstrains possessed similar amounts ofAgF and also possessedthe carbohydrate group K antigen. Mutant HB-7 lacked thehost attachment factor AgC and, in addition, appeared tolack AgE. In strain HB-V5, the veillonella-binding proteinAgB was absent, but the walls were otherwise similar tothose of the parent strain HB. Strain HB-V51 lacked AgB,AgC, and AgE, whereas strain HB-C12 was missing virtuallyall cell wall antigens that react with anti-HB antiserum,except AgF. None of the mutant strains excreted the anti-gens in higher amounts than did the parent strain duringgrowth, with the exception of HB-C12, which excretedincreased amounts of AgD.The cell surface hydrophobicity of the strains as probed by

the adherence to hexadecane is shown in Table 2. Thehydrophobicity of the cells ranges on a relative scale fromvery strong to weak, correlating with the progressive loss ofcell wall antigenic compounds in these strains. Mutant HB-7adsorbed significantly less well to hexadecane than did theparent strain HB but still retained a relatively high level ofhydrophobicity compared with those of many other strepto-cocci. A further decrease was observed with mutantHB-V51, and HB-C12 bound only very weakly to hexadec-ane. In contrast, mutant HB-V5, which lacks only AgB,appeared significantly more hydrophobic than parent strainHB and in fact was the most hydrophobic in this test of about20 different oral streptococci tested (unpublished observa-tion).Comparison of the biological adherence properties of the

strains with the adsorption to hexadecane showed an incom-plete correlation with respect to both adherence to SHA andHeLa cells and showed no correlation in the case of Veil-lonella coaggregation (Table 2). Microscopic observation ofthe adherence of the various strains to human buccal epithe-lial cells yielded results very similar to those obtained forHeLa cells, ranging from buccal epithelial cells carryingmany bacteria (HB and HB-V5) to those nearly free of anyattached cells (HB-7, HB-V51, and HB-C12) (results notshown).

Surface properties of modified cells. Whole cells of thestrains were subjected to various treatments and subse-quently tested for the ability to bind to hexadecane (Table 3).Only treatment with proteases and extraction with sodium

INFECT. IMMUN.

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

SURFACE FIBRILS AND HYDROPHOBICITY OF S. SALIVARIUS

TABLE 3. Effect of cell surface modifications on thehydrophobicity of S. salivarius HB

% Adsorption toTreatment hexadecane (%

of untreated value)

CellsNone ................................ ..... 89 (100)100°C, 20 min ................................... 94 (105.6)3% SDS, 2 h, room temp ....................... 28 (31.5)45% aqueous phenol, 1 h, room temp ...... .... 71(79.8)10% trichloroacetic acid, 2 h, 60°C ...... ....... 78 (87.6)Sonication, 6x, 30 s each .......... ............ 75 (84.3)Pronase (1 mg/ml), 3 h, 37°C ........ ........... 7 (7.9)Pronase (heat-inactivated) .......... ............ 85 (95.5)Trypsin (1 mg/ml), 3 h, 37°C .................... 45 (50.6)Trypsin + trypsin inhibitor (each 1 mg/ml) ..... 83 (93.3)

Cell wallsNone ..................................... 582% Triton X-100 extracted...................... 67

ANTIGEN CONC.(% of total)

8OF

60w

4O*B

L1. D20O

0-15 16-25 26-35% ETHYLENEGLYCOL

dodecyl sulfate (SDS) followed by extensive washing signif-icantly reduced surface hydrophobicity. In contrast, allother treatments had only a limited effect. Methods designedto extract LTA from the cells, such as aqueous phenol andtrichloracetic acid extraction, appeared to impair adsorptionto hexadecane only slightly, whereas boiling resulted in aslightly increased adsorption.

Isolated cell walls, both before and after treatment withTriton X-100, retained a surface hydrophobicity which wasrelatively high but which was significantly reduced com-pared with that of whole cells.

Relative hydrophobicity of solubilized cell wall antigens.Mutanolysin digests of strain HB cell walls which had notbeen treated with Triton X-100 were subjected to hydropho-bic interaction chromatography on a column of Octyl-Sepharose 4B to determine the relative hydrophobicity ofthe wall antigens. When the digest was applied directly to thecolumn in 50 mM phosphate buffer, all of the AgC but onlypart of AgB was bound. Therefore, samples were subse-quently applied in a 25% saturated ammonium sulfate solu-tion, under which conditions all antigens completely boundto the column.

Elution of the column with decreasing ammonium sulfateand simultaneously increasing ethylene glycol concentra-tions to increase the polarity of the solvent selectivelydesorbed some of the antigens (Fig. 1). AgB was the mostreadily released and hence the least hydrophobic compound,followed by AgD. AgC could be only partly released in thisway but was largely recovered upon elution of the columnwith tetramethyl urea, an agent known to break hydrophobicbonds. Quantitative recovery of this antigen required elutionwith Triton X-100, indicating that the compound has stronglyhydrophobic properties.

Effect of tetramethyl urea on adherence. To assess the roleof hydrophobic bonds in adherence, we studied the influenceof the presence of the hydrophobic bond-breaking agenttetramethyl urea (20). Adherence to SHA, in which AgC isinvolved, and coaggregation with Veillonella spp., mediatedby AgB, were chosen because of the ability to quantitatethese reactions accurately. Concentrations of TMU as lowas 50 mM reduce the adherence of strain HB to SHA (Fig. 2).Complete inhibition was not achieved even at concentrations50-fold higher. The half-maximal effect was reached atapproximately 0.2 M. In contrast to this, the effect of urea

36-50. TMU TRITON

FIG. 1. Hydrophobic interaction chromatography on Octyl-Sepharose 4B of mutanolysin-solubilized cell walls of S. salivariusHB. A digest of crude cell walls, not treated with Triton X-100, wasapplied to the column and sequentially eluted with a linear 0 to 50%(vol/vol) gradient of ethylene glycol, 0.5 M tetramethyl urea, and 1%Triton X-100. Concentrations of antigens in pooled fractions wereestimated by crossed immunoelectrophoresis and expressed as thepercentage of the totally eluted amount of the antigens. Symbols:EZJ, AgB; cm, AgC; _, AgD.

was negligible at these concentrations. Approximately 4 Murea was required to obtain a 50% reduction in adherence toSHA. As a control, a similar experiment was done withmutant strain HB-7, which shows only a very weak adher-

B/uA

1.0 A

0.5 \

0.01 OQ05 0.1 0.5 1 5CONCENTRATION (M)

FIG. 2. Effect of urea (A, L) and tetramethyl urea (0, 0) on theadherence of S. salivarius to SHA. The results are expressed as theratio of bound (B) over unbound (U) cells. Solid symbols, strain HB;open symbols, strain HB-7.

VOL. 55, 1987 441

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

442 WEERKAMP ET AL.

ence to SHA. At all inhibitor concentrations tested, thisstrain yielded similar low adherence values.The effects of both tetramethyl urea and urea were negli-

gible up to concentrations of 2 M, when the interbacterialcoaggregation was studied. Hence, this reaction seems notto involve essential hydrophobic bonds, in contrast to ad-herence to SHA.

Association of protein antigens with LTA. To determinewhether, in analogy with group A streptococci (14, 15), LTAmay be responsible for the surface hydrophobicity of S.salivarius, several tests were applied. The presence of LTAat the cell surface of various mutant strains and modifiedcells were measured by using a modified whole-cell enzyme-linked immunosorbent assay technique involving antiserumagainst the polyglycerolphosphate backbone ofLTA (16). Asa control, the technique was applied to specific antiseraagainst cell wall antigens of strain HB and was shown toaccurately detect the presence of these antigens in appropri-ate strains (Table 4). The results with anti-LTA show aninverse relationship between the antigenic complexity of thecell wall and the relative amount of LTA detectable at thecell surface. Comparison of the range of mutants indicatedthat the progressive loss of protein antigens correlates withincreased binding of anti-LTA antibodies. Concomitantly, adecrease in the binding of nonspecific rabbit immunoglobulinG was observed in the mutants (not shown). These resultssuggest that the protein surface antigens shield LTA fromthe external environment. This is further supported by theobservation that pronase treatment of cells resulted in expo-sure of LTA (Table 4).

In a second experiment, the relative amount of LTA in cellwall digests of the strains was measured. Similar titers(1:512) were obtained with all strains, except that digests ofstrain HB-V5 seemed to contain a slightly higher amount. Tosee whether LTA in the digest of HB cell walls was associ-ated with any of the cell wall antigens, we subjected thepreparation to electrophoresis in an agarose gel, followed byblotting to nitrocellulose and immunodetection of LTA.LTA was not specifically associated with the wall proteinantigens but migrated close to the electrophoresis front (Fig.3). Aqueous phenol-extracted LTA from the same strain thatwas partially purified by Sepharose CL-2B gel filtration andextraction with chloroform-methanol (11) also migratedclose to the electrophoresis front. The somewhat differentelectrophoretic behavior of the two preparations may reflectdifferences in association with other compounds.

TABLE 4. Surface exposure of LTA and protein antigens inS. salivarius strains

LTA AgB AgC

Strain A405' (% A405 (% A405 (%Control Control Controlvalue) value) value)

HB 0.42 (100) 1.48 (100) 1.68 (100)HB-7 0.69 (164) 1.60 (116) 0.78 (46.4)HB-V5 0.48 (115) 0.02 (1) NTbHB-V51 0.78 (206) 0.08 (5) NTHB-C12 1.14 (273) 0.24 (16) 0.20 (12.5)HB (pronase) 0.88 (210)HB-7 (pronase)Y 1.28 (186)

a Whole-cell enzyme-linked immunosorbent assay; expressed as the netincrease in A405 compared with preimmune serum. Average standard devia-tion in all tests was 7%.bNT, Not tested.c Treated with pronase (1 mg/ml) for 3 h at 370C.

1 2 3 4

D~I~

c99

B

FIG. 3. Agarose gel electrophoresis of a mutanolysin digest of S.salivarius HB cell walls (lanes 1 to 3). Antigens were detected afterimmunoblotting to nitrocellulose with protein A-colloidal gold andtotal antiserum against HB walls (lane 1), serum specific for AgC(lane 2), and serum specific for LTA (lane 3). B, C, and D refer to thepositions of the antigens in the gel. Lane 4 shows the electrophoresisof partially purified LTA after immunodetection with anti-LTA. Theorigin of the electrophoresis is at the bottom.

In a further test, immunoelectron microscopy was done tolocalize LTA on cryosections of HB cells. The results showthat LTA is clearly associated with the cytoplasmic mem-brane, but very little or none is found in the periferal fibrillarlayer of the cell envelope (Fig. 4). For comparison, it isshown that AgC is almost exclusively associated with thefibrillar layer.

DISCUSSION

The availability of mutants of S. salivarius HB which wereoriginally selected on the basis of the absence of specificadhesive properties and were subsequently found to lackdistinct fibrillar surface antigens (28, 29, 35) created theopportunity to investigate the role of these structures inhydrophobic properties of the cell surface. The results of thisstudy demonstrate that hydrophobicity, assessed from theability to adsorb to hydrophobic ligands, is a function of boththe fibrillar density on the cell surface and the properties andexposure of specific fibril classes. An association of hydro-phobic properties with cell surface proteins and fibrillarstructures of streptococci and oral actinomycetes has beensuggested in previous studies (8, 10, 14, 15, 18, 26, 38), butno conclusive evidence was supplied, or conflicting findingswere reported (8, 16, 26).At least three classes of fibrils on the surface of S.

salivarius HB cells have been recognized (28, 35). Thelongest (168 nm) fibrils are apparently not proteinaceous innature since they cannot be removed by protease treatmentof the cells, in contrast to the other fibril classes. Therefore,the long fibrils do not seem to contribute significantly tosurface hydrophobicity, since treatment with protease re-moves most or all hydrophobic sites from the cell surface.The function and nature of this fibril class is still unknown. Aclass of 72-nm, protease-sensitive fibrils corresponds to thehost attachment factor AgC (29) and is absent from the cellsurface of mutant HB-7. Mutants of this type had a lowerability to adsorb to hexadecane than did the parent strain;

INFECT. IMMUN.

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

SURFACE FIBRILS AND HYDROPHOBICITY OF S. SALIVARIUS

FIG. 4. Ultrathin cryosections of S. salivarius HB indirectly immunolabeled with a protein A-colloidal gold (6 nm) probe and anti-LTAantiserum. The insert shows immunolabeling with anti-AgC antiserum. Bar, 0.5 ,um.

this corresponds to the finding that solubilized AgC was themost hydrophobic of the fibrillar antigens. It correspondsalso to the finding that mutant HB-V5, which lacks themedium-sized (91 nm) fibrils, is more hydrophobic than theparent strain, since an increased surface exposure of AgCwould be expected in this mutant. The 91-nm fibrils, corre-sponding to AgB, the veillonella-binding protein (33), haveonly weak hydrophobic properties as indicated by hydropho-bic interaction chromatography. However, they apparentlystill contribute significantly to hydrophobic properties of thecell surface, since mutant cells having predominantly AgB atthe cell surface (i.e., HB-7) retain a relatively high surfacehydrophobicity. This is further supported by the observationthat the simultaneous absence of both AgB and AgC inmutant HB-V51 results in a strong decrease in hydrophobic-ity, although not its complete loss. Mutant HB-V51 waspreviously shown to carry very sparsely distributed short (63nm) fibrils (28), which may have been responsible for theresidual hydrophobicity. Indeed, the supplementary reduc-tion in hydrophobicity observed for strain HB-C12 corre-sponds with the virtual absence of all fibrils from this strain(P. Handley, personal communication).

Further evidence for the association of hydrophobic prop-erties with surface fibrils is provided by the cell modificationstudies, which suggested that this property is associated withproteinaceous compounds. Circumstantial evidence for thenoninvolvement of LTA in surface hydrophobicity wasprovided by our experiments, which showed that (i) surfaceexposure of LTA is inversely correlated with cell surfacehydrophobicity, (ii) LTA is not specifically associated withisolated fibrillar proteins, and (iii) LTA localizes predomi-nantly in the cytoplasmic membrane region of the cell

envelope but is virtually absent from the fibrillar layer inimmunoelectronmicroscopy. These results contrast with themodel accounting for surface hydrophobicity of S. pyogenes(15, 16). Surprisingly, treatment of S. salivarius with SDS atroom temperature strongly reduced the ability of the organ-ism to adsorb to hexadecane but did not extract significantamounts of the antigens (not shown). This suggests that SDSis bound to and masks hydrophobic sites on the cell surface.In contrast to our findings, Morris et al. (18) did not observea significant reduction in hydrophobicity of S. sanguis, evenafter it was boiled in the presence of SDS. Boiling did notdestroy hydrophobicity nor extract antigens from the S.salivarius cells; this finding is similar to results for S. sanguis(18) but contrasts with findings for S. mutans and S.pyogenes (14, 21). However, it was noted that boilingremoved cell surface proteins from S. mutans, which sug-gests that in S. salivarius and S. sanguis, the proteins areincorporated into the cell wall in a quite different way.Another contrasting finding with respect to S. mutans wasthat in nonhydrophobic variants of S. salivarius, the missingwall protein antigens were not excreted into the culturemedium during growth, with the exception of AgD, whichwas excreted by strain HB-C12. On the basis of all evidence,it appears that hydrophobic properties of the cell surface areassociated with the presence of a variety of specific (fibrillar)proteins, which exhibit various degrees of intrinsic hydro-phobicity.

In a previous report on the subcellular localization of cellwall proteins in parent and mutant strains (28), we suggestedthat synthesis, translocation, and incorporation of thesecompounds are independent processes. This may explainour failure to select stable, nonhydrophobic mutants of S.

-M

m

VOL. 55, 1987 443

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

444 WEERKAMP ET AL.

salivarius by repeated adsorption of mutagenized cells tohexadecane or Octyl-Sepharose. It should be noted in thisrespect that prolonged growth at high specific growth rateapparently induces phenotypes which have lost multiplesurface components (27) and finally leads to the isolation ofmutant HB-C12. In a similar way, this may also explain whysuccessful selection of nonhydrophobic mutants with otheroral streptococci generally resulted in strains which hadsimultaneously lost multiple surface components and a va-riety of adhesive functions (10, 14, 18). A recent experimentin which S. sanguis mutants were selected for specificadhesive functions yielded a range of six phenotypes dif-fering in surface properties; this was more similar to ourfindings (8).The reactions mediated by the fibrillar antigens of S.

salivarius are highly specific and, at least in the case of AgB,involve lectinlike interactions (32, 33). Therefore, the hydro-phobicity in itself cannot be the sole factor determiningspecificity. McBride et al. (18) recently suggested that hy-drophobic sites in S. sanguis are not localized close to theadhesin involved in adsorption to SHA. For oralactinomycetes, it was demonstrated that although hydropho-bicity and adsorption to SHA are statistically correlated,hydrophobic interactions appeared not to be of major impor-tance to the reaction itself (3). However, hydrophobic inter-actions may function to stabilize the specific interactions ofthe cell with the adhesion substratum by forming additionalbonds (4, 9). Indeed, provided that tetramethyl urea disruptsthe bacterium-receptor interaction and does not act byinducing conformational changes in either of the surfaceswhich may prevent their proper interaction, its reductiveeffect on the adherence of S. salivarius HB to SHA maysupport such a model. This reduction was not down to thelevel of adhesion of mutant HB-7, which lacks AgC butretains a fairly high surface hydrophobicity. A reductiveeffect was not noticed on the coaggregation of strain HB withveillonellae, a reaction which is mediated by the less hydro-phobic AgB. Such findings are indicative of a mosaic patternof surface hydrophobicity accommodating the specific re-quirements of various surface functions. In addition tocreating a suitable microenvironment supporting the forma-tion or stabilization of specific bonds, the overall hydropho-bicity of the cell surface may also be of importance toadherence, since it affects the interfacial free energy be-tween the bacterial cell and the substratum (2, 34).

ACKNOWLEDGMENTWe are indebted to Caroline Tijhof for excellent technical assist-

ance.

LITERATURE CITED1. Beachey, E. H., and I. Ofek. 1976. Epithelial cell binding of

group A streptococci by lipoteichoic acid on fimbriae denudedof M protein. J. Exp. Med. 143:759-771.

2. Busscher, H. J., A. H. Weerkamp, H. C. van der Mei, A. W. J.van Pelt, H. P. de Jong, and J. Arends. 1984. Measurement of thesurface free energy of bacterial cell surfaces and its relevancefor adhesion. Appl. Envir. Microbiol. 48:980-983.

3. Clark, W. B., M. D. Lane, J. E. Beem, S. L. Bragg, and T. T.Wheeler. 1985. Relative hydrophobicities of Actinomycesviscosus and Actinomyces naeslundii strains and their adsorp-tion to saliva-treated hydroxyapatite. Infect. Immun. 47:730-736.

4. Doyle, R. J., W. E. Nesbitt, and K. G. Taylor. 1982. On themechanism of adherence of Streptococcus sanguis to hydrox-ylapatite. FEMS Microbiol. Lett. 15:1-5.

5. Ellen, R. P., and R. J. Gibbons. 1972. M-protein-associated

adherence of Streptococcus pyogenes to epithelial surfaces:prerequisite for virulence. Infect. Immun. 5:826-830.

6. Ericson, T., and J. Rundegren. 1983. Characterization of asalivary agglutinin reacting with a serotype c strain of Strepto-coccus mutans. Eur. J. Biochem. 133:255-261.

7. Faris, A., T. Wadstrom, and J. M. Freer. 1981. Hydrophobicadsorptive and haemagglutinating properties of Escherichia colipossessing colonizing factor antigens (CFA/I and CFA/II), type1 pili and other pili. Curr. Microbiol. 5:67-72.

8. Fives-Taylor, P. M., and D. W. Thompson. 1985. Surfaceproperties of Streptococcus sanguis FW213 mutants nonadher-ent to saliva-coated hydroxyapatite. Infect. Immun. 47:752-759.

9. Gibbons, R. J., and I. Etherden. 1983. Comparative hydropho-bicities of oral bacteria and their adherence to salivary pellicles.Infect. Immun. 41:1109-1196.

10. Gibbons, R. J., I. Etherden, and Z. Skobe. 1983. Association offimbriae with the hydrophobicity of Streptococcus sanguis FC-1and adherence to salivary pellicles. Infect. Immun, 41:414-417.

11. Josephson, S. L., M. W. Stinson, S. J. Miliar, and R. E. Cohen.1986. Purification of lipoteichoic acid by chromatography inwater-organic solvent systems. Infect. Immun. 51:378-384.

12. Knox, K. W., L. N. Hardy, L. J. Markevics, J. D. Evans, andA. J. Wicken. 1985. Comparative studies on the effect of growthconditions on adhesion, hydrophobicity, and extracellular pro-tein profile of Streptococcus sanguis G9B. Infect. Immun.50:545-554.

13. Lichtenberg, D., M. Rosenberg, N. Sharfman, and I. Ofek. 1985.A kinetic approach to bacterial adherence to hydrocarbon. J.Microbiol. Methods 4:141-146.

14. McBride, B. C., M. Song, B. Krasse, and J. Olsson. 1984.Biochemical and immunological differences between hydropho-bic and hydrophilic strains of Streptococcus mutans. Infect.Immun. 44:68-75.

15, Miorner, H., J. Havlicek, and G. Kronvall. 1984. Surfacecharacteristics of group A streptococci with and without M-protein. Acta Pathol. Microbiol. Immunol. Scand. Sect. B92:23-30.

16. Miorner, H., G. Johansson, and G. Kronvall. 1983. Lipoteichoicacid is the major cell wall component responsible for surfacehydrophobicity of group A streptococci. Infect. Immun. 39:336-343.

17. Moeremans, M., G. Daneels, A. van Dijck, G. Langanger, and J.de Mey. 1984. Sensitive visualization of the antigen-antibodyreaction in dot and blot immuno-overlay assays with im-munogold and immunogold/silver staining. J. Immunol. Meth-ods 74:353-360.

18. Morris, E. J., N. Ganeshkumar, and B. C. McBride. 1985. Cellsurface components of Streptococcus sanguis: relationship toaggregation, adherence, and hydrophobicity. J. Bacteriol. 164:255-262.

19. Morris, E. J., and B. C. McBride. 1984. Adherence of Strepto-coccus sanguis to saliva-coated hydroxyapatite: evidence fortwo binding sites. Infect. Immun. 43:656-663.

20. Nesbitt, W. E., R. J. Doyle, and K. G. Taylor. 1982. Hydropho-bic interactions and the adherence of Streptococcus sanguis tohydroxyapatite. Infect. Immun. 38:637-644.

21. Ofek, I., E. Whitnack, and E. H. Beachey. 1983. Hydrophobicinteractions of group A streptococci with hexadecane droplets.J. Bacteriol. 154:139-145.

22. Olsson, J., and G. Westergren. 1982. Hydrophobic surfaceproperties of oral streptococci. FEMS Microbiol. Lett. 15:319-323.

23. Rosenberg, M., D. Gutnick, and E. Rosenberg. 1980. Adherenceof bacteria to hydrocarbons: a simple method for measuringcell-surface hydrophobicity. FEMS Microbiol. Lett. 9:29-33.

24. Rosenberg, M., and S. Kjelleberg. 1986. Hydrophobic interac-tions: role in bacterial adhesion. Microbiol. Ecol. 8:353-393.

25. Rutter, P. R., and B. Vincent. 1980. The adhesion of microor-ganisms to surfaces: physicochemical aspects, p. 79-92. InR. C. W. Berkeley, J. M. Lynch, J. Melling, P. R. Rutter, andB. Vincent (ed.), Microbial adhesion to surfaces. Ellis HorwoodLtd., Chichester, England.

26. Tylewska, S. K., S. Hjerten, and T. Wadstrom. 1979. Contribu-

INFECT. IMMUN.

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/

SURFACE FIBRILS AND HYDROPHOBICITY OF S. SALIVARIUS

tion of M protein to the hydrophobic properties of Streptococ-cus pyogenes. FEMS Microbiol. Lett. 6:249-253.

27. Weerkamp, A. H., and P. S. Handley. 1986. The growth rateregulates the composition and density of the fibrillar coat on thesurface of Streptococcus salivarius K+ cells. FEMS Microbiol.Lett. 33:179-183.

28. Weerkamp, A. H., P. S. Handley, A. Baars, and J. W. Slot. 1986.Negative staining and immunoelectron microscopy of adhesion-deficient mutants of Streptococcus salivarius reveal that theadhesive protein antigens are separate classes of cell surfacefibril. J. Bacteriol. 165:746-755.

29. Weerkamp, A. H., and T. Jacobs. 1982. Cell wall-associatedprotein antigens of Streptococcus salivarius: purification, prop-erties, and function in adherence. Infect. Immun. 38:233-242.

30. Weerkamp, A. H., and B. C. McBride. 1980. Adherence ofStreptococcus salivarius HB and HB-7 to oral surfaces andsaliva-coated hydroxyapatite. Infect. Immun. 30:150-158.

31. Weerkamp, A. H., and B. C. McBride. 1980. Characterization ofthe adherence properties of Streptococcus salivarius. Infect.Immun. 29:459-468.

32. Weerkamp, A. H., and B. C. McBride. 1980. Studies on aveillonella-binding protein of Streptococcus salivarius, p.521-523. In R. C. W. Berkeley, J. M. Lynch, J. Melling, P. R.

Rutter, and B. Vincent (ed.), Microbial adhesion to surfaces.Ellis Horwood Ltd., Chichester, England.

33. Weerkamp, A. H., and B. C. McBride. 1981. Identification of aStreptococcus salivarius cell wall component mediating coag-gregation with Veillonella alcalescens Vl. Infect. Immun.32:723-730.

34. Weerkamp, A. H., H. C. van der Mei, and H. J. Busscher. 1985.The surface free energy of oral streptococci after being coatedwith saliva and its relation to adhesion in the mouth. J. Dent.Res. 64:1204-1220.

35. Weerkamp, A. H., H. C. van der Mei, and R. S. B. Liem. 1986.Structural properties of fibrillar proteins isolated from the cellsurface and cytoplasm of Streptococcus salivarius (K+) cellsand nonadhesive mutants. J. Bacteriol. 165:756-762.

36. Weiss, E., M. Rosenberg, H. Judes, and E. Rosenberg. 1982.Cell-surface hydrophobicity of adherent oral bacteria. Curr.Microbiol. 7:125-128.

37. Westergren, G., and J. Olsson. 1983. Hydrophobicity and ad-herence of oral streptococci after repeated subculture in vitro.Infect. Immun. 40:432-435.

38. Wheeler, T. T., and W. B. Clark. 1980. Fibril-mediated adher-ence of Actinomyces viscosus to saliva-treated hydroxyapatite.Infect. Immun. 28:577-584.

VOL. 55, 1987 445

on March 29, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

http://iai.asm.org/