Vol. 42, No. 2INFECTION AND IMMUNITY, Nov. 1983, p. 459-4700019-9567/83/110459-12$02.00/0Copyright © 1983, American Society for Microbiology

Antigenic Heterogeneity of Bacteroides intermedius as

Recognized by Monoclonal AntibodiesRUDOLF GMUR* AND BERNHARD GUGGENHEIM

Department of Oral Microbiology and General Immunology, Dental Institute, University of Zurich, CH-8028Zurich, Switzerland

Received 24 May 1983/Accepted 12 August 1983

Four hybrid cell lines secreting monoclonal antibodies against antigens ofBacteroides intermedius were generated by fusing murine NSI cells with spleno-cytes from a rat immunized with B. intermedius strain OMZ248. An enzyme-linked immunosorbent assay was used to analyze the distribution of the recog-nized antigens on 39 strains from various Bacteroides species and on 5 strainsfrom other genera. Only Bacteroides species B. intermedius, B. loescheii, B.melaninogenicus, and B. corporis were found to express at least one of therecognized antigens. Strains of the two asaccharolytic black-pigmenting Bacteroi-des species were negative. Among the strains capable of binding to one or more ofthe monoclonal antibodies, five groups with different reactivity patterns could bedistinguished. Two of the monoclonal antibodies were specific for B. intermedius.The B. intermedius strains were metabolically almost identical, expressed at leastthree of the recognized antigens, and fell into three distinct antibody reactivitygroups, suggesting a tentative separation of this species into three new sero-

groups. Oral and nonoral isolates of B. intermedius were, however, not distin-guished by the monoclonal antibodies. One monoclonal antibody was directedagainst an antigen strongly expressed on all saccharolytic black-pigmentingBacteroides strains tested so far, thus confirming the previously noted antigenicrelationship between the species which had emerged from the former B. melanino-genicus subsp. intermedius and B. melaninogenicus subsp. melaninogenicusgroups.

Evidence indicating a major role of bacteria inthe etiology of periodontal disease is convincing.However, the microbial flora associated withgingivitis and periodontitis is complex and ofconsiderable variation from patient to patient aswell as from site to site. Furthermore, varioussites of patients with different forms of periodon-tal disease, such as localized juvenile periodonti-tis or adult periodontitis, appear to harbor char-acteristically different sets of microorganisms(18, 21, 23, 25). It is only from recent work thatinformation about the composition of the micro-flora of active sites of patients with a given typeof periodontal disease has begun to emerge (12;A. C. R. Tanner and S. S. Socransky, J. Dent.Res. vol. 62, abstr. no. 350, 1983; S. S. So-cransky et al., J. Dent. Res. vol. 62, abstr. no.250, 1983). Among the bacteria encountered themost frequently in periodontally diseased sitesare black-pigmenting Bacteroides species. Par-ticularly, the two species B. gingivalis and B.intermedius proliferate very significantly duringprogressive gingival pocket formation and alveo-lar bone destruction (14, 21-23, 28, 29). Thespecific role, if any, of these two species in the

development of the disease is, however, stillunknown.A widely used approach to assess the etiologi-

cal importance of distinct oral microorganismsin periodontal disease is the analysis of thespecific humoral immune response in both pa-tients and healthy control persons. So far, thistype of analysis has indicated a strong correla-tion of high immunoglobulin G (IgG) and IgAtiters against Actinobacillus actinomycetemco-mitans with the occurrence of localized juvenileperiodontitis (3, 6, 24) and against B. gingivaliswith rapidly progressive and adult periodontitis(6, 16, 24). On the other hand, specific antibodytiters against B. intermedius have been reportedto be high in most individuals regardless of theirperiodontal state of health (6, 19). For severalreasons, the latter data cannot presently beproperly interpreted. (i) The habitat of B. inter-medius is not restricted to the oral cavity. If oraland nonoral forms of B. intermedius are antigen-ically indistinguishable as suggested by severalstudies (11, 17, 20), then nonoral infections orabscess formations in control individuals couldindirectly prevent detection of specific immune

459

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

460 GMUR AND GUGGENHEIM

responses to oral B. intermedius infections inpatients with periodontal disease. (ii) Differentsaccharolytic black-pigmenting Bacteroides spe-cies seem to exhibit considerable immunologicalcross-reactivity, making even the experimentalproduction of high-titer species-specific rabbitantisera a difficult task (11, 17, 20). (iii) Differentisolates of B. intermedius appear to be antigene-tically quite heterogeneous. The heterogeneitymay be such that differences in the reactivity ofa given human serum with several B. interme-dius strains may exceed the titer variationsobserved when several sera are tested against asingle isolate (Gmur, unpublished data).

Clearly, for a better understanding of therelevance of serological reactions against B.intermedius, more information about the anti-genic properties of B. intermedius as well as ofother black-pigmenting species is necessary.

In this paper, we report the production offourmonoclonal antibodies against B. intermediusstrain OMZ248 and describe the distribution ofthe recognized antigens on 28 B. intermediusisolates and on 16 strains from other species.The data emphasize the antigenic heterogeneityof B. intermedius, but also confirm the previous-ly noted antigenic relationship among saccharo-lytic black-pigmenting Bacteroides species.

MATERIALS AND METHODS

Bacterial strains. Of the 44 strains used in this study,38 were black- or brown-pigmenting Bacteroidesstrains of human origin. Seventeen strains, designatedherein as reference strains, have been described previ-ously and are listed in Table 1. Twenty-seven strainswere isolated in this laboratory from periodontal pock-ets (>4-mm pocket depth) of 27 patients.

Preliminary biochemical tests done before this studyidentified 24 of these strains as B. intermedius, 2 asbelonging to the former B. melaninogenicus subsp.melaninogenicus group, and 1 (strain OMZ274) asFusobacterium nucleatum. All strains were stored inliquid nitrogen and also were lyophylized.

Culture conditions and biochemical tests. Wilkins-Chalgren anaerobe agar (Oxoid Ltd.) supplementedwith 5% (vol/vol) hemolyzed human blood was used assolid medium. All strains were able to grow in amodification of a fluid medium (FUM) originally de-scribed by Loesche et al. (13). FUM contained (perliter of distilled water): 10 g of tryptone, 5 g of yeastextract, 3 g of glucose, 2 mg of hemin, 1 mg ofmenadione, 0.5 g of cysteine hydrochloride, 0.1 g ofdithiothreitol, 2.9 g of NaCl, 0.5 g of Na2CO3, 1 g ofKNO3, 0.45 g of K2HPO4, 0.45 g of KH2PO4, 0.9 g of(NH4)2SO4, and 0.188 g of MgSO4 * 7H20 and had apH of 7.1. Heat-inactivated, filter-sterilized horse se-rum was finally added to this medium, which wasprepared and sterilized as originally described (13). Allcultures were incubated anaerobically in jars (BBLMicrobiology Systems) at 37°C for 3 to 5 days.Gram stains, catalase activity tests, and growth

experiments in air gave negative results for all black-

pigmenting strains. In addition, the strains were char-acterized as follows.

Fermentation of glucose, lactose, mannitol, andtrehalose was tested in glucose-free FUM containing1% of the respective filter-sterilized compounds. Cul-tures in FUM were additionally analyzed for indoleproduction. In later experiments performed to reevalu-ate the indole production of several strains, a carbohy-drate- and KNO3-free FUM medium was used. Escu-lin hydrolysis was tested in FUM containing 1%esculin. The production of lipase and lecithinase wastested by standard procedures (27), using Wilkins-Chalgren agar as a base for the egg yolk supplement.Milk proteolysis was examined on Wilkins-Chalgrenagar plates containing 5% skim milk powder; 5 g wassuspended in 100 ml of water, sterilized by repeatedheating in flowing steam, and added to 900 ml ofprecooled (57°C) medium. Finally, the capacity ofthese strains to hydrolyze soluble starch and purifiedgelatin (E. Merck AG) was tested by adding 1% of therespective compounds to the same medium.

Identification of fermentation products. The bacteriawere incubated in FUM with 1% glucose for 40 to 120h. The cultures were then centrifuged, and the super-natants were acidified with 4 N HCI to pH 2.0. A 1-mlvolume of diethyl ether was thoroughly mixed with a2-ml sample of the acidified culture fluid. The organicphase was separated from the water phase by freezing(-20°C) and was then analyzed by gas-liquid chroma-tography (2).

Cell lines. The murine myeloma cell line NSI-AG 4/1(10) was provided by B. K. Grove, Institute for CellBiology, Swiss Federal Institute of Technology, Zu-rich, Switzerland. The cells were grown in Dulbeccominimal essential medium supplemented with 10%heat-inactivated newborn bovine serum (GIBCO Eu-rope), 15 p.g of 8-azaguanine per ml, 2 mM glutamine,100 IU of penicillin per ml, and 100 ,ug of streptomycinper ml. The isolation and cultivation of diploid fetal ratfibroblasts have been described previously (5).

Production of hybrid ceils. An inbred RIC (rat inbredat Carworth Farm, New York, N.Y.)-Sprague-Dawleyrat was subcutaneously immunized with approximate-ly 108 bacteria of B. intermedius OMZ248, emulsifiedin 0.1 ml of saline and an equal volume of incompleteFreund adjuvant. Intraperitoneal booster injectionswith 100 ,ug of a lyophilized broken-cell supernatant(1) of strain OMZ248 bacteria were administered after4 and 9 months. Three days after the last injection, therat was sacrificed, and splenocytes were isolated bygently pressing the spleen through a stainless steelmesh in Ca2+- and Mg2+-free Dulbecco phosphate-buffered saline. Erythrocytes were lysed by hypotonicshock with 0.83% NH4Cl, and 9 x 10' spleen cellswere fused with 9 x 10' NSI cells as described byGmur et al. (7). After fusion, the cell suspension wasplated in 10 24-well tissue culture plates (Nunc) inselection medium that consisted of Iscove-modifiedminimal essential medium (KC Biologicals Inc.) sup-plemented with 10%o heat-inactivated newborn bovineserum, 13.6 ,ug of hypoxanthine per ml, 1.76 p.g ofaminopterin per ml, 3.9 ,ug of thymidine per ml, 0.225,ug of glycine per ml, 100 IU of penicillin per ml, 100 ,ugof streptomycin per ml, and 1% (vol/vol) anti-PPLOagent (GIBCO Europe). Wells with growth of cellhybrids were tested for the presence of antibodies

INFECT. IMMUN.

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

B. INTERMEDIUS STRAIN HETEROGENEITY 461

TABLE 1. Reference strains used in this studySpecies and strain Site of origin Received froma:

B. asaccharolyticusATCC25260 (VP14198)NCTC9337 (VP18945)

B. gingivalis3811021Wl (Loesche W)

B. intermediusNonoralATCC25261 (VPI4203)ATCC25611 (VP14197)

OralH187M107-74 (VP19146)

B. corporis532-70A (ATCC33547, VP19342)

EmpyemaInfected hemorrhoids

Subgingival plaqueSubgingival plaqueSubgingival plaque

Laryngotomy woundEmpyema

Subgingival plaqueSubgingival plaque

Cervical swab

T. J. M. van SteenbergenG. H. Bowden

S. S. SocranskyJ. SlotsN. P. Lang

T. J. M. van SteenbergenT. J. M. van Steenbergen

T. J. M. van SteenbergenG. H. Bowden

T. J. M. van Steenbergen

B. melaninogenicusVP19343 Great toe T. J. M. van Steenbergen

B. loescheiiATCC15930 (VP10037) Gingival crevice T. J. M. van Steenbergen

B. fragilisATCC25285 Appendix abscess

F. nucleatumOMZ274 Subgingival plaque

A. actinomycetemcomitansY-4 Subgingival plaque

Supragingival plaqueRat plaque

Supragingival plaque

B. F. HammondJ. S. van der Hoeven

Our isolate

a T. J. M. van Steenbergen, Free University, Amsterdam, The Netherlands; G. H. Bowden, University ofManitoba, Winnipeg, Canada; S. S. Socransky, Forsyth Dental Center, Boston, Mass.; J. Slots, State Universityat Buffalo, Buffalo, N. Y.; N. P. Lang, University of Beme, Switzerland; J. S. van der Hoeven, University ofNijmegen, The Netherlands; J. Wust, University of Zurich, Switzerland; B. F. Hammond, University ofPennsylvania, Philadelphia.

against B. intermedius OMZ248 on days 11 and 17 byan enzyme-linked immunosorbent assay (ELISA) as

described below. To prevent overgrowth by nonpro-ducer cells, wells with apparent specific antibodyproduction were split at low density, whereby one partof the cells was used for cloning by limiting dilution in96-well tissue culture plates (Nunc) and the other partwas transferred into new plates and expanded forfreezing. Most of the initial splits failed to yield clonesbecause of rather low plating efficiencies of the hybridcells. This problem was overcome by propagating allcultures on feeder layers of irradiated (2,000 rad)diploid fetal rat fibroblasts.The fusion finally yielded four independent hybrid

cell lines producing antibodies against B. intermedius.The four lines were cloned by limiting dilution, ex-

panded to mass cultures, and then immediately re-cloned again. After being cloned once, all cultureswere switched to selection medium lacking aminop-terin (HT medium). All experiments described hereinwere done with pools of culture supernatant harvestedfrom fibroblast feeder layer-supported mass culturesof clonal hybridoma lines.For isotyping of monoclonal antibodies, undiluted

culture supernatants were tested with rat immunoglob-ulin class-specific reagents (Nordic ImmunologicalLaboratories) in Ouchterlony gel diffusion.ELISA. Both the screening of hybridoma culture

A. viscosusT14VNyl

S. mutansOMZ176

J. Wust

Our isolate

J. Slots

VOL. 42, 1983

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

462 GMUR AND GUGGENHEIM

fluids for the presence of antibodies against B. inter-medius and the analysis of the strain distribution of theantigens recognized by the isolated monoclonal re-agents were done with a modified ELISA (4). Round-bottom 96-well microtest plates (Greiner) were coatedwith antigen by adding 100 p.l of bacterial broken-cellsupernatant per well (10 jig of protein per ml ofphosphate-buffered saline) and incubating the platesovernight at 4°C. The wells were then washed twicewith borate-buffered saline (BS; 0.14 M boric acid,0.025 M sodium tetraborate, 0.075 M sodium chloride,pH 8.1), refilled with 100 ,ul of 0.5% bovine serumalbumin (BSA) and 0.05% Tween 20 per well in BS andallowed to stand for 2 h at 37°C. After being washedtwice with BS and once with distilled water, the wellswere dried over water-free calcium chloride. At thatpoint, the plates were sealed tightly and transferred to4°C so that they could be stored for several monthswithout loss of antigenic activity. Alternatively, bacte-ria were grown in fluid cultures for 24 to 72 h, washedtwice with 0.9% sodium chloride, suspended in PBS,and diluted to a standard optical density of 0.5 at 550nm. Flat-bottom ELISA microtest plates (Inotech)which had been pretreated with 100 p.l of 0.1% glutar-aldehyde per well in BS for 30 min at room tempera-ture were filled with 200 p.l of bacterial suspension perwell and centrifuged (10 min at 1,300 x g). Withoutdecantation, an additional 100 p.1 of 0.1% glutaralde-hyde in BS was then added carefully to each well, andthe plates were allowed to stand for 30 min at roomtemperature. After two washing steps with BS, thewells were filled with 200 p.l of 0.1% BSA per well inBS, and the plates were sealed tightly and stored at4°C until used. No loss of antigenic activity wasobserved for as long as 12 months.The antigen-antibody binding assay procedure in-

volved three steps. (i) Antigen-coated microtest plateswere washed twice with BS and incubated overnight at4°C with monoclonal reagent or antisera that had beendiluted with BS containing 0.5% BSA and 0.05%Tween 20. Unbound antibodies were then removed byflicking the plates, and wells were washed four timeswith ELISA wash solution (0.9% sodium chloride,0.05% Tween 20, 0.01% sodium azide). (ii) a 100-,ulvolume of horseradish peroxidase-conjugated rabbitanti-rat IgG per well, prepared by the method of Milleret al. (15) and diluted 1/2,000 in BS containing 0.5%BSA and 0.05% Tween 20, was added to each micro-test well. The plates were incubated for 90 min at 37°C,and the wells were washed eight times with ELISAwash solution before being refilled (step iii) with 200 p.1of enzyme substrate solution (3.7 mM 1,2-phenylene-diamine, 0.012% hydrogen peroxide in 24.3 mM citricacid, 25.7 mM disodium hydrogen phosphate, pH 5.1).After a 15-min incubation at room temperature in thedark, the enzyme reaction was stopped by addition of50 p.l of 2.5 M sulfuric acid per well, and the extent ofthe reaction was determined spectrophotometrically at492 mm by using a Titertek Multiscan (Flow Labora-tories Ltd.).

Screening of cell hybrid culture fluids. Fluids fromcultures with proliferating hybrid cells were testedwith ELISA for the production of antibodies against B.intermedius by adding 50 p.l (cloning cultures) or 100p.l (mass cultures) to individual wells of both microtestplates coated with strain OMZ248 broken-cell super-natant and to plates coated with native strain OMZ248

INFECT. IMMUN.bacteria from fluid cultures. Positive control wellscontained 1/500-diluted serum obtained from the ratwhose spleen was used for hybridization.

Analysis of strain distribution of the antigens recog-nized by monoclonal antibodies. Analysis was done by(i) directly measuring with ELISA the binding of themonoclonal reagents to bacteria of various strains andby (ii) quantitative absorption of the monoclonal re-agents with various strains.

In the first analysis procedure, the wells of eighthorizontal rows of ELISA microtest plates were coat-ed with suspensions of eight different bacterial strains.All suspensions were adjusted to identical opticaldensities (0.5 at 550 nm). For reference purposes, thewells of one row always contained strain OMZ248bacteria. The wells of each row were then filled withserial dilutions of one of the monoclonal reagents andprocessed by the ELISA procedure outlined above.

In the second analysis procedure, bacteria of a givenstrain were harvested from at least two 14-ml fluidcultures, pooled, and quantitated by measuring theoptical density at 550 nm of a threefold-diluted sample.The bacteria were then washed twice with phosphate-buffered saline and centrifuged again, and the pelletswere suspended in an amount of HT medium calculat-ed by the formula y = az, where y is the amount(milliliters) of HT medium used for the suspension ofthe pellet, z is the previously determined opticaldensity, and a is a dilution factor. The concentration ofbacteria in a suspension obtained with a = 1.67 wasarbitrarily defined as 1 x. It was determined in apreliminary experiment (data not shown) that the 1 xconcentration equals that obtained when 100 pl ofcentrifugation-sedimented bacteria was suspended in 2ml of HT medium. Absorptions of monoclonal re-agents were always performed with at least threedifferent concentrations of bacteria. To this end, 100p.1 of bacterial suspensions (either 1 x, 1/5 x and 1/25 xor 4x, 2x, lx, 1/2x and 1/4x depending on theexperiment) were mixed in 96-well round-bottom mi-crotest plates with 50 p.1 of monoclonal reagent. Thedilution of the monoclonal reagent was chosen suchthat the unabsorbed reagent yielded 70 to 90% ofmaximal binding in ELISA with strain OMZ248 bacte-ria (optical density, 0.3 at 550 nm) as antigen. Theplates were then tightly sealed and incubated at 37°Con a shaker. After 2 h, the plates were centrifuged(1,300 x g for 10 min), 100 p.1 of the supernatant fromeach well was transferred into wells of ELISA micro-test plates coated with strain OMZ248 bacteria (opticaldensity, 0.3 at 550 nm), and ELISA was performed asdescribed above.

RESULTSIsolation of hybrid cell lines producing mono-

clonal antibodies against B. intermedius. Culturesupernatants from 130 wells with growing hybridcolonies were screened in ELISA for activityagainst intact bacteria and broken-cell superna-tants of strain OMZ248. Nine supernatants werefound to be strongly positive. From four cellpopulations, hybrid cell lines producing antibod-ies against B. intermedius could be isolated andcloned. The four clonal cell lines and the anti-bodies produced by them were named 37BI6.1,38B11, 39BI1.1, and 40BI3.2, where the first

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

B. INTERMEDIUS STRAIN HETEROGENEITY 463

numbers identify the original cell line, BI is theimmunogen used, and the last numbers indicatethe clone obtained from the original isolate.The supernatants from all four hybridoma

lines reacted with both strain OMZ248 broken-cell supernatant and OMZ248 bacteria, but tovery different extents. With the supernatant asthe antigen source, 38B11 yielded a strong reac-tion and a 50% endpoint titer of 1:320, whereasthe supernatants of 39BII. and 40BI3.2 were oflower titers and produced much less intenseELISA reactions. Monoclonal antibody 37BI6.1bound only in barely detectable amounts tobroken-cell supernatant-coated ELISA wells. Inmarked contrast, all four monoclonal antibodieswere found to react strongly with intact strainOMZ248 bacteria (Fig. 1).

In Ouchterlony gel diffusion tests, the mono-clonal antibodies 38B11, 39BI1.1, and 40B13.2were all identified as IgG2a, whereas 37BI6.1was IgG2b.

Analysis of strain distribution of the antigensrecognized by monoclonal antibodies. Figure 1shows the binding pattern of the monoclonalantibodies with the four B. intermedius strainsATCC25611, ATCC25261, H187, OMZ248, andwith B. corporis 532-70A. Only two of thesestrains, the nonoral isolate ATCC25611 and theoral strain OMZ248, were able to bind all fourantibodies. At high antibody concentrations,binding reached plateau levels, and 50% end-point titers were between 1/100 (37BI6.1) and1/4,400 (40BI3.2). Two other strains,ATCC25261 and H187, bound three of themonoclonal antibodies but did not react with40B13.2. Interestingly, these two strains couldbe distinguished by their reactivities with anti-body 39BI.1. Whereas strain ATCC25611showed the same type of dose-response curvewith 39BI1.1 as did strain ATCC25261 orOMZ248, the strain H187 bacteria could only belabeled with a 1,000-fold-higher antibody con-centration. Furthermore, the dose-responsecurve never reached plateau levels (Fig. lc).Strain 532-70A, the type strain ofB. corporis (9),reacted only with 38B11. Taken together, thefour B. intermedius strains revealed as many asthree different reactivity patterns with the mono-clonal antibodies but also shared one of therecognized antigens with the B. corporis typestrain.These results prompted us to study a larger

panel of B. intermedius strains and, to furtheranalyze the species specificity of the monoclonalantibodies, to include a variety of strains fromother Bacteroides species and from other generain the analysis. Results from these experimentsare summarized in Fig. 2. Of the 28 B. interme-dius strains tested, 16 (57%) reacted with all fourmonoclonal antibodies, whereas 12 (43%) failed

to bind antibody 40B13.2. The latter group ofstrains further separated into two subgroups ofseven (25%) and five (18%) strains, respectively.The 16 strains of the first group and the 7 strainsfalling into the larger subgroup of the secondgroup exhibited the same type of dose-responsecurve with 39B11.1 as did strain ATCC25611 orOMZ248 (Fig. lc). All five strains of the smallersubgroup, however, revealed dose-responsecurves nearly identical to the one described inFig. lc for strain H187.Unable to react with the four antibodies were

the two B. asaccharolyticus strains tested, aswell as all three strains of B. gingivalis. Alsonegative were all strains analyzed of the fourgenera other than Bacteroides.

Finally, four strains showed the same reactiv-ity patterns as that ofB. corporis 532-70A in thatthey bound only monoclonal antibody 38BI1.Note that all four were black-pigmenting Bacte-roides strains positive for lactose fermentationand esculin hydrolysis and negative for indoleproduction (see Table 5). By these criteria, theybelong neither to B. intermedius, B. asaccharo-lyticus, B. gingivalis, nor B. corporis.

Characterization of strain heterogeneity by ab-sorption analysis. In the course of the experi-ments summarized in Fig. 2, we observed that afew strains were coating the wells of ELISAplates less efficiently than others. Particularly,strain 532-70A appeared to barely adsorb to thewells. We therefore reinvestigated the resultsshown in Fig. 2 by absorption analysis. Increas-ing numbers of bacteria from representativestrains were incubated with adequately dilutedmonoclonal reagents, and the absorbed reagentswere subsequently tested on strain OMZ248bacteria for remaining activity. The results areshown in Table 2; for ease of comparison, onlyabsorption data from one concentration of bac-teria are shown. For strains belonging to thethree largest groups observed in Fig. 2, i.e., thereactivity groups including strains ATCC25611,OMZ245, and ATCC25260, respectively, the ab-sorption experiments confirmed the resultsshown above. The strains of B. intermediusforming the subgroup containing strain H187(Fig. 2), however, were unable to absorb39BI1.1, in spite of the fact that these strainscould bind this antibody in direct tests. Thisresult may suggest that these strains express aslightly cross-reactive antigen. Finally, the fivestrains that in direct ELISA-binding tests wererecognized by 38B11 only surprisingly split intotwo subgroups. The three strains OMZ254,OMZ255, and ATCC15930 quite clearly ab-sorbed antibody 37BI6.1 at the highest bacterialconcentration used for absorption (1 x), but didnot at 1/5x and 1/25x concentrations. That thisabsorption is not due to unspecific antibody

VOL. 42, 1983

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

464 GMUR AND GUGGENHEIM INFECT. IMMUN.

a

E

ol

0.5-

1.5-

O.A-''''u A

E

(.4C 1.000

0.

0.-

2 18 162 1458 13122 2 18 162 1458 13122

RECIPROCAL DILUTIONS OF HYBRIDOMA SUPERNATANTS

FIG. 1. Binding of the four monoclonal antibodies (a) 37BI6.1, (b) 38BI1, (c) 39BI1.1, and (d) 40B13.2 to thefour B. intermedius strains OMZ248 (M), ATCC25611 (A), ATCC25261 (@), and H187 (0) and to B. corporis 532-70A (V). Binding was measured by ELISA with intact bacteria coated to 96-well microtest plates. OD 492 nm,Optical density at 492 nm.

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

B. INTERMEDIUS STRAIN HETEROGENEITY

zw0w

z00z0

00

2zwiid

1501

125-

100

I %ii75 5-w P % i a > > aw a

150-

125

100

75-

5J

25

0~~~ ~ ~ ~ ~ ~ ~ ~ ~ ~

150-

1251-

1001

7- -LL.

50- u @S @S °I% ~ ~ ~

* 37516.1* 38B1v 39S1 1.1* 40B1 3.2

,14.% %4AO6/ 1

_ __,.4 _.

_ _

.._

._ _ _

-_ _. ._ _, *O _ _-, <', aAshFIG. 2. Binding of the monoclonal antibodies 37BI6.1 (A), 38B11 (0), 39BI1.1 (V), and 40B13.2 (U) to 38

strains of various Bacteroides species and six strains from other genera (see the text). Antibody-binding wasdetermined by ELISA with intact bacteria coated to microtest plates. The reactivity of each monoclonalantibody with the immunizing strain OMZ248 was arbitrarily defined as 10lo.

trapping is shown in Fig. 3, which presentsresults from experiments in which still largeramounts of bacteria were used. Obviously, ab-sorption of the monoclonal reagent followed adose-dependent pattern unique for these threestrains.Table 3 provides a summary of the results

presented thus far. Apparently the four mono-clonal antibodies could react only with black-pigmenting strains that are either B. intermediusor belonged to the group of strains formerlyclassified as B. melaninogenicus subsp. melan-inogenicus. They separated these strains intofive distinct reactivity groups, three of whichcontained only B. intermedius strains of bothoral and nonoral origins.

Metabolic properties of black-pigmentingstrains used in this study. This obvious antigenicheterogeneity of the analyzed B. intermediusstrains appears to contrast with the previouslynoted phenotypic homogeneity of the species.Since most of the B. intermedius strains used inthis study were new and phenotypically not yetcarefully characterized isolates, we decided to

analyze them as well as the other black-pigment-ing strains used for a number of characteristicmetabolic properties. Results are summarized inTables 4 and 5. All strains classified in theserological reactivity groups I, II, and IIIshowed remarkably similar metabolic profiles(Table 4). By fermenting glucose but not lactose(two exceptions), hydrolizing gelatin (one ex-ception) but not esculin, producing indole, ace-tic acid, isobutyric acid, and isovaleric acid and,at the most, only trace amounts of butyric acid,they exhibited the characteristic features of B.intermedius. The fermentation of lactose by twoof the strains is exceptional. Nevertheless, weconsider them as B. intermedius because of theirother metabolical features. Furthermore, one ofthe two strains (H187) has been shown to belongto the B. intermedius DNA homology group B(26). Among the reactivity groups IV and V, fourstrains fermented glucose and lactose, hydro-lyzed esculin, and were unable to produce in-dole. Two of these strains, ATCC15930 andVPI9343, have been very recently assigned tothe two new Bacteroides species B. loescheii

VOL. 42, 1983 465

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

466 GMUR AND GUGGENHEIM

TABLE 2. Absorption analysis of the expression of the antigens recognized by four monoclonal antibodieson black-pigmented Bacteroides strains

Monoclonal reagentsaStrain

37BI6.1 38B11 39BI1.1 40B13.2

OMZ248 72.7b 65.9 89.6 92.5OMZ257 80.4 85.2 92.5 89.3OMZ258 79.9 88.3 91.7 85.8OMZ267 73.9 85.2 64.3 92.8OMZ272 76.3 83.3 85.8 91.2ATCC25261 100.0 100.0 82.5 0OMZ251 76.7 67.0 95.8 4.4OMZ260 75.6 64.2 70.6 0OMZ276 68.9 44.0 88.5 0M107-74 64.1 79.0 0 1.9H187 58.3 66.4 3.0 0OMZ261 67.4 46.0 2.4 1.3OMZ265 78.1 75.4 8.1 , 0OMZ275 79.2 74.6 0 0VP19343 8.3 100.0 0 0ATCC15930 62.8 99.7 7.5 0532-70A 0 88.5 0 1.7OMZ254 35.0 96.3 0 3.5OMZ255 66.2 92.3 1.7 0NCTC9337 0 2.1 0 0ATCC25260 0 6.7 7.2 0

a Hybridoma supernatants were absorbed at concentrations sufficient to yield 70 to 90% maximum binding inELISA with strain OMZ248 bacteria as the antigen source.

b Absorptions were performed in triplicate with at least three concentrations of bacteria. However, for ease ofcomparison, data from only one concentration (1 x) are shown. The ELISA data are expressed as percentreduction of binding of the monoclonal reagent to the target bacteria (strain OMZ248) after absorption with agiven bacterial test strain.

and B. melaninogenicus, respectively (8),whereas the other two strains, OMZ254 andOMZ255, presently cannot definitely be classi-fied because of the lack ofDNA homology data.The five asaccharolytic strains included in thisstudy were all well described reference strains.They did not express the antigens recognized bythe monoclonal antibodies. Three of them arephenotypically characterized as B. gingivalis bytheir production of phenylacetic acid. The othertwo strains had the metabolic profile of B.asaccharolyticus, although with the unexpectedfeature of weak but significant glucose fermenta-tion by strain ATCC25260.

DISCUSSIONIn this paper we describe the production of

four hybrid cell lines that produce monoclonalantibodies directed against antigens of B. inter-medius OMZ248 and report the distribution ofthe recognized antigens on various strains of thegenus Bacteroides. The few strains from othergenera, namely F. nucleatum, A. actinomyce-temcomitans, A. viscosus, and S. mutans, whichwere analyzed for the expression of these anti-gens, proved to be negative.

Analysis of the distribution of these antigenswas performed with two different ELISA tech-niques. With the first technique, intact bacteria

were coated onto the bottom of ELISA micro-test plates and antigen expression was thenassessed by the determination of the amount ofmonoclonal antibodies capable of binding totheir target antigens. This method yields onlysemiquantitative results since the ability of thebacteria to coat the ELISA microtest platesvaried from strain to strain. A characteristicfeature of this approach is that antigen-antibodybinding may occur under conditions of vastantibody excess, probably enabling the detec-tion of antigens with only low affinity to themonoclonal antibodies. In a second set of ex-periments many strains were retested by quanti-tative absorption for expression of the antigensrecognized by the four monoclonal antibodies.In contrast to the first technique, antigen-anti-body binding occurs here under excess of anti-gen.The two assays yielded different strain distri-

butions for each of the four recognized antigens.It may be concluded, therefore, that the fourmonoclonal antibodies bind to four differentmolecules. The antigen detected by antibody38B11 was expressed by all saccharolytic black-pigmenting Bacteroides strains tested so far,regardless of whether they were isolates fromoral or nonoral sites. Among the positive strainswere the type strain (ATCC25611) of B. interme-

INFECT. IMMUN.

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

B. INTERMEDIUS STRAIN HETEROGENEITY 467

E -c

0

0.4-

1/4x 1/2x lx 2x 4x

CONCENTRATION OF BACTERIA DURING ABSORPTIONOF MONOCLONAL REAGENT

FIG. 3. Absorption analysis of reactivity of mono-clonal antibody 37BI6.1 with bacteria from variousstrains. After absorption with M107-74 (0), 532-70A(O), OMZ265 (A), OMZ254 (0), OMZ255 (A),ATCC15930 (A), or VPI9343 (V), the residual bindingactivity of the monoclonal reagent was tested byELISA on bacteria of strain OMZ248. The concentra-tions of bacteria used for absorption are defined in thetext. Unabsorbed monoclonal reagent yielded inELISA an optical density at 492 nm (OD 492 nm) of1.175 ± 0.057 (standard deviation).

dius (9), the type strain (532-70A) of B. corporis(9), the type strain (ATCC15930) of B. loescheii(8), and a strain (VP19343) belonging to thesecond DNA homology group of B. melanino-genicus (8). In contrast, the antigen could not bedetected on any of five tested strains of theasaccharolytic black-pigmenting species B.asaccharolyticus and B. gingivalis, nor on theintestinal organism B. fragilis. It will certainlybe of interest to analyze still more Bacteroidesspecies (e.g., B. denticola, B. oralis, B. levii,and B. macacae) to determine how widespreadthis common antigen is. The detection of such an

antigen with monoclonal antibodies is not sur-

prising and confirms earlier observations ofcross-reactivities among antisera raised againststrains of B. intermedius, B. corporis, and B.melaninogenicus (11, 17, 20).As for antibody 38BI1, monoclonal antibody

37BI6.1 recognized an antigen present on atleast two different Bacteroides species. Howev-

er, only B. intermedius strains expressed readilydetectable amounts of the antigen. The frequen-cy of expression among the strains of this spe-cies was 100%. Only very low amounts of theantigen were found on three strains, one ofwhich was B. loescheii, whereas the two othershave not been classified beyond belonging to theformer B. melaninogenicus subsp. melaninogen-icus group. The antigen levels on these threestrains were too low to be detected in directantibody binding tests, but absorption analysisof the monoclonal reagent with large amounts ofintact bacteria yielded unambiguous, positiveresults.The antigens defined by 39BI1.1 and 40BI3.2

are apparently restricted to B. intermedius. Toour surprise, monoclonal antibody 39BI1.1 re-vealed two distinct types of dose-responsecurves with B. intermedius strains. Whereas forthe majority of the strains, normal antibodybinding was observed, a group of five strainscould only be labeled in direct ELISA antibody-binding test at very high concentrations of anti-body. Absorption tests with antibody 39B11.1and these strains, however, gave consistentlynegative results. At present, we do not have asatisfactory explanation for the seemingly con-tradictory results obtained with the two assaysystems. The altered dose-response curves ob-served with five strains could be interpreted asreflecting expression of an antigen with a slightlyaltered antigenic binding site. We therefore op-eratively have designated this molecule as across-reactive antigen. Certainly, this hypothe-sis will have to be tested by additional experi-ments, which will require extensive purificationof the target antigen of 39BI.1.Monoclonal antibody 40BI3.2 bound only to

57% of the B. intermedius strains tested, therebyseparating the strains into two distinct groups.Taken together, the results on the relativeexpression of these antigens on various black-pigmenting Bacteroides strains suggest a separa-tion of B. intermedius into two main reactivitygroups (either 40B13.2 positive or negative), ofwhich one may be further subdivided on thebasis of the different capacity of the strains tobind 39BI1.1 antibodies. In agreement with re-cently published genetic and immunological data(9, 20), nonoral isolates of B. intermedius did notreveal a specific antigenic profile, but ratherproved to be phenotypically indistinguishablefrom oral strains in all parameters tested.

During the preparation of this manuscript,Johnson and Holdeman (9) proposed to dividethe former B. melaninogenicus subsp. interme-dius group into two new species, namely, B.intermedius and B. corporis. Although their datain support of such a division arose primarilyfrom studies ofDNA homologies, both metabol-

VOL. 42, 1983

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

468 GMUR AND GUGGENHEIM INFECT. IMMUN.

TABLE 3. Formation of reactivity groups based on the bacterial capacity to bind the monoclonal antibodies

Expression of antigens recognized byReactivity Strains monoclonal antibodies:group 37BI6.1 38B11 39B11.1 40B13.2

I ATCC25611, OMZ55, 248, 257-259, 262-264, and 266-272 + + + +

II ATCC25261, OMZ245, 251-253, 260, and 276 + + +

III H187, M107-74, OMZ261, 265, and 275 + + wa

IV ATCC15930, OMZ254 and 255 + b +

V 532-70A, VP19343 - + - -

VI ATCC25260, NCTC9337, 381, 1021, Wl, ATCC25285, Y-4, - - - -OMZ274, OMZ176, Nyl, T14V

a w, Expression of a weakly crossreacting antigen.b Quantitatively very weak, but significant expression.

TABLE 4. Metabolic properties of B. intermedius strains used in this studyReaction

ATCC25611,OMZ248, ATCC25261 OMZ261,Metabolic properties 2257-259, OMZ 268 OMZ252, OMZ251 265, and H187262-264, OZ28 253, 260. 275266, 267, and 276

and 269-272

FermentationGlucose + + + + + +Lactose - + - - - +

HydrolysisEsculin - - - - - -Gelatine + + + - + +

Starch + + + + + +

Proteolysis of milk + + + + + +

Production of:Indole + + + + + +Lipase + + + + + +

Products from FUMAcetic acid + + + + + +Isobutyric acid + + + + + +Butyric acidbIsovaleric acid + + + + + +Valeric acidcPhenylacetic acid

Reactivity group I II IIIwith monoclonalantibodies

a All strains were gram negative, black pigmented on blood agar, unable to ferment mannitol or trehalose, andunable to produce catalase or lecithinase.

b Traces found with 69% of BI strains tested.' Traces found with 10% of BI strains tested (OMZ260, 262, and 263)

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

B. INTERMEDIUS STRAIN HETEROGENEITY 469

TABLE 5. Metabolic properties of black-pigmenting Bacteroides strains used in this study (excluding B.intermedius)

Reaction

Metabolic propertieSa B Ih Unidentified B. melanino- B B. asaccharolyticus B. gingivalisToescheii genicus corports NTCC 9337 and Wi, 1021,ATCC15930 OMZ254 OMZ 255 VP19343 532-70A ATCC 25260 and 381

FermentationGlucose + + + + + b

Lactose + + + +

HydrolysisEsculin + + + + - -Gelatine - - - - - + +Starch - + + + - -

Proteolysis of milk - + - - - + +

Production of:Indole - - - - - + +Lipase + +

Products from FUMAcetic acid + + + + + + +Isobutyric acid - _b + + + + +Butyric acid b _ b _ _b + +Isovaleric acid _ b _b + + + + +Valeric acid - -

Phenylacetic acid - - - - - - +

Reactivity groups IV V VIwith monoclonalantibodiesa All strains were gram negative, black or brown pigmented on blood agar, unable to ferment mannitol or

trehalose, and unable to produce catalase or lecithinase.b Trace amounts were produced.

ic and antigenic differences between strains ofthe new species have been known for severalyears (11) and were confirmed again in thisstudy. More interestingly, however, was that thestrains united in the new species B. intermediusseparated into two genetically distinct but phe-notypically indistinguishable DNA homologygroups (9). Remarkably, the results presented inthis paper divide B. intermedius strains into twomain groups (either 40BI3.2 positive or negative)as well. It is obvious to ask now whether theantigen recognized by monoclonal antibody40BI3.2 is a phenotypic marker of strains be-longing to one of the two DNA homologygroups. If so, this would allow the rapid classifi-cation of new isolates (e.g., by immunofluores-cence) without the necessity of laborious DNAhomology studies. We cannot yet answer thequestion because for only three of the B. inter-medius strains analyzed in our study is the DNAhomology group known. Nevertheless, the clas-sifications of these three strains obtained eitherby DNA homology or, alternatively, by 40BI3.2monoclonal antibody typing are in remarkable

agreement; strain ATCC25611 has been reportedto belong to the B. intermedius homology groupI (9, 26) and is shown in this paper to fall intoreactivity group I (expressing all the recognizedantigens), whereas strains ATCC25261 andH187 are known to be part of DNA homologygroup II (9, 26), and both failed to express the40BI3.2 defined antigen. In view of these re-sults, we tentatively describe reactivity groupsI, II, and III as three new serogroups of B.intermedius. Certainly, further studies will benecessary to determine whether this promisingcorrelation of genetic and phenotypic parame-ters is true for all B. intermedius strains. In theaffirmative case, the use of monoclonal antibod-ies 40B13.2 and 39BI1.1 as markers of newserogroups will be justified and probably veryhelpful.

In conclusion, we have produced a set ofmonoclonal antibodies that define four indepen-dent B. intermedius antigens partially shared byother black-pigmenting saccharolytic Bacter-oides species. The expression of the antigens bythe strains is not random but specific and may

VOL. 42, 1983

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from

470 GMUR AND GUGGENHEIM

serve as an indicator of their antigenic relation-ship. None of the antigens has been biochemi-cally characterized. Nevertheless, our observa-tions that all four antigens can be readilydetected by direct labeling of intact bacteria withthe monoclonal antibodies or by absorption ofthe antibodies with freshly collected bacteriastrongly suggest that the antigens are located inthe outer membrane or in the layer covering thisstructure. Although not demonstrated here, theease with which the antigens can be detected inimmunofluorescence tests could make thesemonoclonal antibodies interesting tools for theidentification of black-pigmenting saccharolyticBacteroides isolates, in particular the species B.intermedius.

ACKNOWLEDGMENTS

We thank Angela Lang, Hanni Schneider, and MartinGander for excellent technical assistance; Eva Ebneter fortyping; and Christoph Wyss for critically reading the manu-script. The generous gift of bacterial strains by G. H. Bowden,B. F. Hammond, N. P. Lang, J. Slots, S. S. Socransky, J. S.van der Hoeven, T. J. M. van Steenbergen, and J. Wust isgratefully acknowledged.

LITERATURE CITED

1. Burckhardt, J. J. 1978. Rat memory T lymphocytes: invitro proliferation induced by antigens of Actinomycesviscosus. Scand. J. Immunol. 7:167-172.

2. Carlsson, J. 1973. Simplified gas chromatographic proce-dure for identification of bacterial metabolic products.Appl. Microbiol. 25:287-289.

3. Ebersole, J. L., M. A. Taubman, D. J. Smith, R. J. Genco,and D. E. Frey. 1982. Human immune response to oralmicroorganisms. I. Association of localized juvenile per-iodontitis (LJP) with serum antibody responses to Actino-bacillus actinomycetemcomitans. Clin. Exp. Immunol.47:43-52.

4. Engvall, E., and P. Perlmann. 1972. Enzyme-linked im-munosorbent assay, ELISA. III. Quantitation of specificantibodies by enzyme-labeled anti-immunoglobulin inantigen-coated tubes. J. Immunol. 109:129-135.

5. Gaegauf-Zollinger, R., J. J. Burckhardt, R. Gmur, and B.Guggenheim. 1982. Cell-mediated cytotoxicity against ratfibroblasts induced by Actinomyces viscosus. Infect. Im-mun. 37:710-719.

6. Genco, R. J., J. Slots, C. Mouton, and P. Murray. 1980.Systemic immune responses to oral anaerobic organisms,p. 277-293. In D. W. Lambe, R. J. Genco, and K. J.Mayberry-Carson (ed.), Anaerobic bacteria. Plenum Pub-lishing Corp., New York.

7. Gmur, R., D. Solter, and B. B. Knowles. 1980. Indepen-dent regulation of H-2K and H-2D gene expression inmurine teratocarcinoma somatic cell hybrids. J. Exp.Med. 151:1349-1359.

8. Holdeman, L. V., and J. L. Johnson. 1982. Description ofBacteroides loescheii sp. nov. and emendation of thedescriptions of Bacteroides melaninogenicus (Oliver andWherry) Ray and Kelly 1939 and Bacteroides denticolaShah and Collins 1981. Int. J. Syst. Bacteriol. 32:399-409.

9. Johnson, J. L., and L. V. Holdeman. 1983. Bacteroidesintermedius comb. nov. and description of Bacteroidescorporis sp. nov. and Bacteroides levii sp. nov. Int. J.Syst. Bacteriol. 33:15-25.

10. Kohler, G., S. C. Howe, and C. Milstein. 1976. Fusionbetween immunoglobulin-secreting and nonsecreting my-eloma cell lines. Eur. J. Immunol. 6:292-295.

11. Lambe, D. W., and R. C. Jerris. 1976. Description of apolyvalent conjugate and a new serogroup of Bacteroidesmelaninogenicus by fluorescent antibody staining. J. Clin.Microbiol. 3:506-512.

12. Listgarten, M. A., and S. Levin. 1981. Positive correlationbetween the proportions of subgingival spirochetes andmotile bacteria and susceptibility of human subjects toperiodontal deterioration. J. Clin. Periodontol. 8:122-138.

13. Loesche, W. J., R. N. Hockett, and S. A. Syed. 1972. Thepredominant cultivable flora of tooth surface plaque re-moved from institutionalized subjects. Arch. Oral Biol.17:1311-1325.

14. Loesche, W. J., S. A. Syed, B. A. Laughon, and J. Stoll.1982. The bacteriology of acute necrotizing ulcerativegingivitis. J. Periodontol. 53:223-230.

15. Miller, M. H., M. J. Karnovsky, and G. T. Diamandopou-los. 1974. An immunoperoxidase technique for identifyingSV40 V and T antigens by light microcopy. Proc. Soc.Exp. Biol. Med. 146:432-437.

16. Mouton, C., P. G. Hammond, J. Slots, and R. J. Genco.1981. Serum antibodies to oral Bacteroides asaccharolyti-cus (Bacteroides gingivalis): relationship to age and peri-odontal disease. Infect. Immun. 31:182-192.

17. Mouton, C., P. G. Hammond, J. Slots, M. J. Reed, andR. J. Genco. 1981. Identification of Bacteroides gingivalisby fluorescent antibody staining. Ann. Microbiol. (Paris)132B:69-83.

18. Newman, M. G. 1979. The role of Bacteroides melanino-genicus and other anaerobes in periodontal infections.Rev. Infect. Dis. 1:313-323.

19. Patters, M. R., and K. S. Kornman. 1982. Serum antibod-ies to Bacteroides species in human periodontitis. J.Periodontal Res. 17:474-477.

20. Reed, M. J., J. Slots, C. Mouton, and R. J. Genco. 1980.Antigenic studies of oral and nonoral black pigmentedBacteroides strains. Infect. Immun. 29:564-574.

21. Slots, J. 1979. Subgingival microflora and periodontaldisease. J. Clin. Periodontol. 6:351-382.

22. Syed, S. A., M. Svanberg, and G. Svanberg. 1981. Thepredominant cultivable dental plaque flora of beagle dogswith periodontitis. J. Clin. Periodontol. 8:45-46.

23. Tanner, A. C. R., C. Haffer, G. T. Bratthall, R. A.Visconti, and S. S. Socransky. 1979. A study of the bacte-ria associated with advancing periodontitis in man. J.Clin. Periodontol. 6:278-307.

24. Taubman, M. A., J. L. Ebersole, and D. J. Smith. 1982.Association between systemic and local antibody andperiodontal diseases, p. 283-298. In R. J. Genco and S. E.Mergenhagen (ed.), Host-parasite interactions in peri-odontal diseases. American Society for Microbiology,Washington, D.C.

25. Van Palenstein Helderman, W. H. 1981. Microbial etiologyof peridontal disease. J. Clin. Periodontol. 8:261-280.

26. Van Steenbergen, T. J. M., C. A. Vlaanderen, and J. deGraaf. 1982. Deoxyribonucleic acid homologies amongstrains of Bacteroides melaninogenicus and related spe-cies. J. Appl. Bacteriol. 53:269-276.

27. Vera, H. T., and M. Dumoff. 1974. Culture media, p. 901.In E. H. Lennette, E. H. Spaulding, and J. P. Truant(ed.), Manual of clinical microbiology, 2nd ed. AmericanSociety for Microbiology, Washington, D.C.

28. White, D., and D. Mayrand. 1981. Association of oralBacteroides with gingivitis and adult periodontitis. J.Periodontal Res. 16:259-265.

29. Zambon, J. J., H. S. Reynolds, and J. Slots. 1981. Black-pigmented Bacteroides spp. in the human oral cavity.Infect. Immun. 32:198-203.

INFECT. IMMUN.

on March 28, 2021 by guest

http://iai.asm.org/

Dow

nloaded from