Insertional inactivation of the prtP gene of Treponema denticola confirms dentilisin's disruption of epithelial junctions.

The purified chymotrypsin-like protease of Treponema denticola, designated dentilisin or PrtP (DDBJ accession no. D83264), can disrupt cell-cell junctions and impair the barrier function of epithelial monolayers in vitro. Serine protease inhibitors block these effects. Yet, the protease is apparently less significant in perturbing intracellular signaling pathways and cytoskeletal rearrangement in fibroblasts. The purpose of this study was to use a PrtP-deficient mutant of T. denticola to confirm that the cytopathic effects of whole bacteria and its outer membrane on epithelial cell junctions were primarily accounted for by the activity of this protease. The prtP gene of ATCC 35405 was inactivated by insertion of an erythromycin-resistance cassette, yielding mutant K1. In contrast to wildtype ATCC 35405, mutant K1 grew in tight cell aggregates; the cells had a disrupted outer sheath, as determined by electron microscopy. When compared by silver stained SDS-PAGE of sonicated extracts of whole cells, the extract of mutant K1 was missing a band at approximately 90 kDa that was present in the wildtype ATCC 35405 strain. Whole cells and Triton X-100 outer membrane (OM) extracts of K1 and the wildtype strains were compared 1) for SAAPNA degrading activity by a colorimetric assay, 2) for stress fiber disruption in human gingival fibroblasts (HGF) by fluorescence microscopy of TRITC-phalloidin stained cells, and 3) the OM extracts only for perturbation of HEp-2 epithelial monolayers by electrical cell-substrate impedance sensing (ECIS). Mutant K-1 cells and OM had no SAPPNA degrading activity that is characteristic of dentilisin. K1 cells had HGF stress fiber disrupting activity (86 +/- 4.5% of HGFs affected) equivalent to both 35405 wildtype strains (84 +/- 3.9% and 71 +/- 14.1% of HGF, respectively). Yet, mutant K1 OM had diminished stress fiber disrupting activity (12.9 +/- 4.6% of HGF) compared with its parent 35405's OM (94.6 +/- 2.9%). The major cytopathogenic difference between the K1 mutant and wildtype strains was in their OM's effect on epithelial cell junctions. ATCC 35405 OM completely disrupted epithelial resistance in a concentration - dependent manner; mutant K1 OM had negligible effects. These data confirm that inactivation of the prtP gene completely reverses T. denticola's disruption of epithelial junctions, but there are pleiotropic effects of the mutation that may account for its apparently diminished effects on the cytoskeleton of HGF when the cells were challenged with OM extracts.

Filamentous actin disruption and diminished inositol phosphate response in gingival fibroblasts caused by Treponema denticola.

Previous reports have shown that Treponema denticola causes rearrangement of filamentous actin (F-actin) in human gingival fibroblasts (HGF). The purpose of this investigation was to determine the effect of T. denticola on the generation of inositol phosphates (IPs) in relation to a time course for F-actin disruption in HGF. Cultured HGF were exposed to washed cells of T. denticola ATCC 35405 for 140 min. Changes in the fluorescence intensity of rhodamine-phalloidin-labeled F-actin in serial optical sections of single HGF were quantified by confocal microscopy image analysis. The percentage of cells with stress fiber disruption was also determined by fluorescence microscopy. Challenge with T. denticola caused a significant reduction in F-actin within the first hour, especially at the expense of F-actin in the ventral third of the cells, and a significant increase in the percentage of HGF with altered stress fiber patterns. Significant concentration-dependent disruption of stress fibers was also caused by HGF exposure to a Triton X-100 extract of T. denticola outer membrane (OM). IPs were measured by a radiotracer assay based on the incorporation of myo-[3H]inositol into IPs in HGF incubated with LiCl to inhibit endogenous phosphatases. HGF challenge with several strains of T. denticola and the OM extract of T. denticola ATCC 35405 resulted in a diminished accumulation of radiolabeled IPs relative to both 15 and 1% fetal bovine serum, which served as strongly positive and background control agonists, respectively. The significantly diminished IP response to T. denticola ATCC 35405 occurred within 60 min, concomitant with significant reduction of total F-actin and disruption of stress fibers. Pretreatment with the proteinase inhibitor phenylmethylsulfonyl fluoride, which had previously been found to block T. denticola's degradation of endogenous fibronectin and detachment of HGF from the extracellular matrix, had little effect on F-actin stress fiber disruption and the IP response. Therefore, in addition to its major surface chymotrypsin-like properties, T. denticola expresses cytopathogenic activities that diminish the generation of IPs during the time course associated with significant cytoskeletal disruption in fibroblasts.

Bacteroides gingivalis vesicles bind to and aggregate Actinomyces viscosus.

Isolated Bacteroides gingivalis 2561 vesicles aggregated suspensions of Actinomyces viscosus and Actinomyces naeslundii of all taxonomy clusters. Vesicles bound near A. viscosus cell walls and among its surface fibrils. Tritiated vesicles bound slightly better to saliva-coated hydroxyapatite (SHA) than to SHA coated with A. viscosus; saturation was approached at the concentrations that were tested. Pretreatment of A. viscosus-coated SHA with vesicles impaired the subsequent adherence of B. gingivalis whole cells.

Coaggregation among periodontal pathogens, emphasizing Bacteroides gingivalis--Actinomyces viscosus cohesion on a saliva-coated mineral surface.

Teeth offer nonshedding surfaces on which a wide range of bacterial species accumulate as thick, cohesive plaques. Intergeneric coaggregation mediated by specific recognition between surface "cohesins" is thought to contribute to both the cohesiveness of plaque and the sequence in which bacteria colonize the tooth surface. There is some evidence that Gram-positive species, like the efficient tooth colonizer Actinomyces viscosus, enhance subsequent tooth colonization by the more virulent periodontal pathogen Bacteroides gingivalis. To study their mechanism of cohesion, we have developed an in vitro assay that measures the sequential binding of tritium-labeled B. gingivalis to A. viscosus adsorbed to saliva-coated hydroxyapatite beads, mimicking teeth (actinobeads). The assay yields equilibrium and kinetics data amenable to statistical analysis. The presence of A. viscosus significantly increased the number of B. gingivalis cells bound. Inhibition studies were conducted to test the sensitivity of binding to heat; to various saccharides and sugar amines; to proteolytic treatment of Bacteroides; and to incorporation of various chaotropic agents, increased KCl, and saliva in the suspension buffer. Heating the Bacteroides cells but not the actinobeads diminished Bacteroides adherence. Proteolysis and various saccharides had little, if any, effect. Among chaotropic agents, NaSCN and LiCl reduced numbers of cells bound by 40%, but tetramethylurea had no effect. Increasing the ionic concentration of KCl reduced binding by 50 to 60%. Diluted saliva showed a concentration-dependent inhibition of B. gingivalis adherence to actinobeads. To begin examining B. gingivalis surface molecules significant to these reactions, lipopolysaccharide was extracted by the phenol-water method and analyzed by biochemical assays and polyacrylamide gel electrophoresis.(ABSTRACT TRUNCATED AT 250 WORDS)

Bacteroides gingivalis-Actinomyces viscosus cohesive interactions as measured by a quantitative binding assay.

There is limited evidence, mostly indirect, to suggest that the adherence of Bacteroides gingivalis to teeth may be enhanced by the presence of gram-positive dental plaque bacteria like Actinomyces viscosus. The purpose of this study was to carry out direct quantitative assessments of the cohesion of B gingivalis and A. viscosus by using an in vitro assay modeled on the natural sequence in which these two species colonize the teeth. The assay allowed comparisons to be made of the adherence of 3H-labeled B. gingivalis 2561 and 381 to saliva-coated hydroxyapatite beads (S-HA) and A. viscosus WVU627- or T14V-coated S-HA (actinobeads) in equilibrium and kinetics binding studies. A series of preliminary binding studies with 3H-labeled A. viscosus and parallel studies by scanning electron microscopy with unlabeled A. viscosus were conducted to establish a protocol by which actinobeads suitable for subsequent Bacteroides adherence experiments could be prepared. By scanning electron microscopy, the actinobeads had only small gaps of exposed S-HA between essentially irreversibly bound A. viscosus cells. Furthermore, B. gingivalis cells appeared to bind preferentially to the Actinomyces cells instead of the exposed S-HA. B. gingivalis binding to both S-HA and actinobeads was saturable with at least 2 X 10(9) to 3 X 10(9) cells per ml, and equilibrium with saturating concentrations was reached within 10 to 20 min. B. gingivalis always bound in greater numbers to the actinobeads than to S-HA. These findings provide direct measurements supporting the concept that cohesion with dental plaque bacteria like A. viscosus may foster the establishment of B. gingivalis on teeth by enhancing its adherence.

Assignment of Actinomyces viscosus and Actinomyces naeslundii strains to numerical taxonomy clusters by immunofluorescence based on antifibril antisera.

A previous observation, using a few representative laboratory strains, that rabbit antisera raised against isolated surface fibrils might be useful in identifying Actinomyces viscosus and Actinomyces naeslundii isolates to their correct numerical taxonomy clusters was reexamined with a large culture collection, including clinical isolates from three different laboratories. Strains were first identified by a slower standard procedure, including agglutination with anti-whole-cell antisera, catalase test, and gas and paper chromatography as needed. Indirect immunofluorescence tests indicated that antisera raised against fibrils of strains representing the three principal taxonomic clusters could separate most strains into their correct clusters. The antisera were particularly successful in separating two taxonomically distinct groups of A. naeslundii, clusters 3 and 5, but could not separate A. viscosus belonging to the principal cluster (no. 1) and minor clusters (no. 2, 4, 6).

Chemical and immunological comparison of surface fibrils of strains representing six taxonomic groups of Actinomyces viscosus and Actinomyces naeslundii.

Human isolates of Actinomyces viscosus and Actinomyces naeslundii have been divided into six clusters in a numerical taxonomy study. Surface fibrils of strains representing these clusters were isolated and purified. Chemical analyses revealed that the major component of all fibrils was protein and that although differences in percentages of specific amino acid residues were found, the relative proportions of basic, acidic, polar uncharged, and nonpolar amino acids were rather similar among clusters. All of the fibrils except those from strain B236 (cluster 2) either failed to migrate or penetrated only slightly into gels during sodium dodecyl sulfate-polyacrylamide gel electrophoresis, even after boiling, reduction, or alkylation. Immunological studies by electron microscopic examination of fibril-antibody immunocomplexes, whole bacterial cell agglutination, inhibition of hemagglutination, and immunofluorescence by using antifibril antisera and antibodies demonstrated that strains of typical A. naeslundii (cluster 5) have a specific fibril-associated antigen(s) distinct from those of strains of other clusters. Cross-reactions for atypical A. naeslundii (cluster 3) were few. The fibrils from A. viscosus clusters 1, 2, 4, and 6 demonstrated several cross-reactions. By absorbing antifibril antibodies with cross-reactive strains it was possible to obtain cluster-specific antibodies, as determined by whole cell agglutination, only for cluster 5. Absorbed antifibril antisera for both A. naeslundii clusters 3 and 5 were specific by indirect immunofluorescence, whereas anti-cluster 1 fibril antisera cross-reacted only with other A. viscosus cluster representatives. Purification of Actinomyces fibrils by methods used for appendages of other species yields preparations containing common antigens among taxonomic groups. However, absorbing antifibril antisera, gamma globulin, or both has promise for producing cluster-specific reagents useful in identification.

Purification and characterization of surface fibrils from taxonomically typical Actinomyces viscosus WVU627.

Fibrils of Actinomyces viscosus WVU627 (numerical taxonomy cluster 1) were obtained by homogenization and purified by ultrafiltration, ammonium sulfate precipitations, gel filtration, and ion-exchange chromatography. Electron microscopy and resolution of a single band by sodium dodecyl sulfate-polyacrylamide gel electrophoresis attested to the purity of the preparation. Purified fibrils were composed mainly of protein; small quantities of carbohydrate and phosphorus were detected. Immunoelectrophoresis revealed only a single precipitable antigen, which migrated slightly toward the anode, in reactions between purified fibrils and antiserum raised against either whole bacterial cells or the purified fibrils themselves. Immunoelectron microscopy with ferritin-conjugated antifibril antibody hemagglutination inhibition, and bacterial agglutination tests demonstrated that fibrils of Actinomyces viscosus cluster 1 strains shared some common antigens with clusters 2, 3, 4 and 6, but did not cross-react with typical Actinomyces naeslundii of cluster 5. Stability tests revealed that after heat or alkali treatment, the fibrils lost their antigenicity and disappeared from electron micrographs. They were affected less by sodium dodecyl sulfate, sonic, or acid treatments.

Sialidase-enhanced lectin-like mechanism for Actinomyces viscosus and Actinomyces naeslundii hemagglutination.

Laboratory strains representing six numerical taxonomy clusters and fresh isolates of human Actinomyces viscosus and Actinomyces naeslundii were studied by standard flocculation slide tests for the ability to hemagglutinate erythrocytes (RBC) from various animal species. Human AB and horse RBC were agglutinated more frequently and rapidly than others; guinea pig RBC were agglutinated by only a few strains. Human AB RBC were selected for studies of hemagglutination mechanisms. Treatment of RBC with clostridial neuraminidase (NTRBC) greatly enhanced hemagglutination for almost all strains. In hapten inhibition experiments in which various concentrations of sugars were used, beta-galactosides were the most effective inhibitors of hemagglutination for both RBC and NTRBC; inhibition of NTRBC agglutination required higher concentrations. Soybean lectin agglutinated both RBC and NTRBC but not Actinomyces cells. NTRBC agglutinated at a 125-fold-lower concentration. Hemagglutination was sensitive to ethylenediaminetetraacetate for one strain tested. Hemagglutination reactions were reversible by addition of beta-galactosides. The ability of Actinomyces strains to "prime" RBC for hemagglutination by removing sialic acid to expose more penultimate beta-galactoside sites was studied by recycling Actinomyces-agglutinated RBC which were dispersed with a lactose solution and washed free of bacteria (primed RBC). Priming in this manner augmented subsequent hemagglutination by indicator Actinomyces strains and made the RBC more sensitive to agglutination by soybean lectin. The priming ability of Actinomyces strains generally correlated with the amount of sialic acid removed from primed RBC. Strains representing the numerical taxonomy clusters differed in both their hemagglutinating and priming activities. Cluster 5 strains (typical A. naeslundii) were good agglutinators of RBC, NTRBC, and primed RBC but were poor primers. Cluster 3 strains (atypical A. naeslundii) were the weakest hemagglutinators but could prime RBC adequately for subsequent agglutination by other strains. Together, these data indicate that Actinomyces hemagglutination proceeds via a two-step mechanism: (i) neuraminidase removal of terminal sialic acid and (ii) lectin-like binding to exposed beta-galactoside-associated sites on the RBC. Strains differ in the extent to which they can perform the two functions, and this specificity may relate to their taxonomic classification.

Mannose-contaminating agglutinin for Actinomyces viscosus and Actinomyces naeslundii.

Rapid agglutination of Actinomyces viscosus and Actinomyces naeslundii cells by D-mannose solutions was observed during studies of their attachment to mammalian cells in vitro. The specificity of the agglutination reaction was studied by slide agglutination tests and by measuring the rate of decrease in optical density of bacterial phosphate buffer suspensions caused by the setting of bacterial aggregates. Actinomyces cells were agglutinated by protein-containing mannose solutions of several chemical suppliers. Solutions of sugars other than D-mannose and solutions of mannitol and mannan all failed to agglutinate A. viscsus and A. naeslundii. "Mannose-enhanced" agglutination was impaired by boiling or autoclaving the mannose but was not affected by heating the bacteria, the presence of chloramphenicol, running the assay in the cold, or incorporating any of several commercially purchased sugars in the reaction mixture. During these hapten inhibition experiments, only 6-deoxy-L-talcose-containing extracts of an A. viscosus strain retarded the rate of mannose-enhanced agglutination. Protein-containing fractions of D-mannose mother liquors also agglutinated cells of A. viscosus and A. naeslundii. Other species of oral gram-positive rods were not agglutinated by mannose solutions. Together the data indicate that plant seed-derived D-mannose contains a protein-associated agglutinin for A. viscosus and A. naeslundii which may function via a "lectin-like" selective affinity for the unique cell wall sugar 6-deoxy-L-talose.