The antigenic complexity of three strains of Bacteroides gingivalis and four strains resembling B. gingivalis isolated from animals was analyzed and compared by crossed immunoelectrophoresis. Thirteen antigens of the human biotype were present in all human strains and six antigens of the animal biotype were present in all animal strains, indicating a marked serological homogeneity within each biotype. Four antigens cross-reacting between the human B. gingivalis and the animal strains were identified. This antigenic relatedness defined the serological homogeneity of the two biotypes within the species and allowed recognition of four species-specific antigens. Two antigens specific to the human strains and two antigens specific to the animal strains were identified, indicating that serotype-specific antigens can distinguish each biotype. It is thus proposed that the oral, black-pigmented asaccharolytic Bacteroides strains of animal origin be classified as catalase-positive variants of B. gingivalis. It is also proposed that two serotypes be recognized within the species B. gingivalis. Serotype 1 includes the catalase-negative human biotype, and serotype 2 includes the catalase-positive animal biotype.

We have compared outer membranes (OM) of Bacteroides gingivalis ATCC 33277 isolated by the following 3 techniques: 1) high speed centrifugation after mechanical cell shearing; 2) sonication of the bacteria, followed by solubilization of the cytoplasmic membrane with N-Laurylsarconsinate (Sarkosyl), after which the Sarkosyl-insoluble membranes were recovered by centrifugation; 3) ammonium sulfate precipitation of extracellular vesicules from culture supernatant, followed by centrifugation and dialysis. Electron microscopy showed that the 3 preparations consisted of closed vesicules. Analysis by SDS-PAGE revealed that all 3 contained up to 28 polypeptides, most of which were common to each extract. The extracellular vesicules and Sarkosyl-insoluble preparation yielded similar protein patterns, although quantitative differences were observed. The sheared-cell preparation contained 8 additional proteins. The level of contamination of OM material by peptidoglycan and cytosol components was 1.8% in the sheared-cell preparation, and was null or lower than 0.8% in the other preparations. All 3 preparations showed the presence of LPS with a multiple banding pattern typical of smooth LPS. The sheared-cell preparation had a slightly lower LPS content than the other 2 preparations. Since extracellular vesicules are naturally released during bacterial growth, and are relatively simple to obtain, such native entities seem an appropriate source of OM components for use in studying the immunobiology of B. gingivalis surface antigens.

A cell-bound hemagglutinating adhesin (HA-Ag2) of Bacteroides gingivalis was identified by crossed immunoaffinity electrophoresis as one of the common antigens of the species. A polyclonal antiserum with a restricted specificity for HA-Ag2 was produced by immunizing with the relevant immunoprecipitate excised from crossed-immunoelectrophoresis gels. The immunoglobulin G fraction of this monospecific antiserum inhibited hemagglutination. The antiserum was used against a cell surface extract of B. gingivalis in immunoblotting experiments, and we detected two antigens with apparent molecular masses of 33 and 38 kilodaltons in B. gingivalis ATCC 33277 and W83. Monoclonal antibody, C1.17, produced in another laboratory against B. gingivalis 381 and characterized as showing reactivity with a hemagglutinin of this strain (Y. Naito, K. Okuda, T. Kato, and I. Takazoe, Infect. Immun. 50:231-235, 1985), was also used to produce immunoblots of extracts of strains ATCC 33277 and W83. The apparent molecular masses of the major polypeptides recognized by monoclonal C1.17 in the immunoblots were the same as those detected by the polyclonal monospecific antiserum, i.e., 33 and 38 kilodaltons. Significantly, none of the polypeptides identified in this study corresponded to the polypeptide appearing in the 41- to 43-kilodalton region and identified by Yoshimura and co-workers (F. Yoshimura, K. Takahashi, N. Yoshinobu, and T. Suzuki, J. Bacteriol. 160:949-957, 1984) as the fimbrial protein characteristic of the species. Enzyme-linked immunosorbent assay inhibition experiments with the monospecific antiserum indicated that the cell surface extracts from strains ATCC 33277 and W83 were strong inhibitors, whereas the fimbria-enriched preparations from both strains failed to inhibit binding of antibodies to the cell surface antigens. As a whole, our study indicates that a nonfimbrial surface protein complex demonstrating erythrocyte-binding capacity, HA-Ag2, is common to three strains of B. gingivalis and is composed of at least two associated polypeptides with apparent molecular masses of 33 and 38 kilodaltons which share at least one antigenic determinant.

Colloidal gold particles 3-6 nm in diameter were prepared and stabilized with the IgG fraction of polyspecific rabbit antisera produced against four different oral bacteria. The immunogold markers were used in homologous reactions to label the bacteria in a preembedding procedure for electron microscopy. An indirect immunofluorescence procedure was concurrently used to optimize the labelling conditions before observation with the electron microscope. The immunogold markers labelled fibrillar structures extending outward 50-275 nm from the Gram-positive cell envelopes and a fuzzy 5-10 nm thick capsulelike layer on the outer aspect of Bacteroides gingivalis. The immunogold method appears to be a simple, rapid, and inexpensive procedure suitable for the study of bacterial surface antigens and can be upgraded with the use of monospecific antibodies.