Requirement for Intercellular Adhesion Molecule 1 and Caveolae in Invasion of Human Oral Epithelial Cells by Porphyromonas gingivalis

Abstract

Porphyromonas gingivalis, a periodontopathic bacterium, is known to invade oral epithelial cells in periodontal lesions, although the mechanism is unclear. In the present study, goat polyclonal anti-intercellular adhesion molecule 1 (anti-ICAM-1) antibody inhibited the invasion of P. gingivalis into KB cells (human oral epithelial cells). Further, the P. gingivalis fimbria, a pathogenic adhesion molecule, bound to recombinant human ICAM-1, as shown by enzyme-linked immunosorbent assay. P. gingivalis was also found to colocalize with ICAM-1 on KB cells, as seen with an immunofluorescence microscope, and the knockdown of ICAM-1 in KB cells resulted in the inhibition of P. gingivalis invasion by RNA interference. In addition, methyl-β-cyclodextrin, a cholesterol-binding agent, inhibited the colocalization of P. gingivalis with ICAM-1 and invasion by the microorganism. The colocalization of caveolin-1, a caveolar marker protein, on KB cells with P. gingivalis was also shown, and the knockdown of caveolin-1 in KB cells caused a reduced level of P. gingivalis invasion. These results suggest that ICAM-1 and caveolae are required for the invasion of P. gingivalis into human oral epithelial cells, and these molecules appear to be associated with the primary stages of the development and progression of chronic periodontitis.

Porphyromonas gingivalis is known as a major etiological agent in the development and progression of periodontal diseases (32), and it has been shown to invade epithelial and endothelial cells (5, 30). Such invasion is a common strategy used by various pathogens to establish host diseases, and, especially, the invasion of nonphagocytic cells is a method used to escape detection by the host immune system (11).

A molecule known as intercellular adhesion molecule 1 (ICAM-1), a member of the immunoglobulin supergene family, is expressed on both epithelial and endothelial cells. Increased ICAM-1 expression induced by various pathogens was shown to mediate cell-to-cell adhesion in inflamed tissues (13), while P. gingivalis infection is known to upregulate ICAM-1 expression (14). Furthermore, Bartonella henselae accumulates ICAM-1 for invasion into endothelial cells (4), and the clustering of ICAM-1 induces an endocytic pathway (19).

It was recently reported that caveolae are the point of entry for the invasion of various pathogens, including Escherichia coli, Chlamydia, and simian virus 40 (15). Caveolae are flask-shaped invaginations found in the plasma membranes of many cell types and consist of caveolin-1, a cholesterol-binding membrane protein (18, 29). Caveolae form caveolar vesicles, which are caveolin-1-containing endocytic compartments (1, 27) that are targeted to early endosomes in a Rab5-dependent pathway (26). It has been demonstrated that P. gingivalis colocalizes with Rab5 after internalization (8); however, the entry of P. gingivalis into host cells at the molecular level has not been elucidated. In the present study, we demonstrated that ICAM-1 and caveolae participate in the invasion of human oral epithelial cells by P. gingivalis.

MATERIALS AND METHODS

Cells, bacteria, antibodies, and reagents.

KB cells, derived from an oral epidermoid carcinoma, were obtained from the Institute of Development, Aging and Cancer, Tohoku University. P. gingivalis strain 381 was anaerobically grown in GAM broth (Nissui, Tokyo, Japan) supplemented with hemin (5 μg/ml) and menadione (5 μg/ml) at 37°C. Fimbriae were isolated from P. gingivalis strain 381 and purified as described previously (23). Recombinant human ICAM-1, mouse monoclonal antibody specific for ICAM-1, and goat polyclonal antibody specific for E-cadherin were purchased from R&D Systems Inc. (Minneapolis, Minn.). Goat polyclonal antibody specific for ICAM-1 and goat immunoglobulin G (IgG) were obtained from Genzyme Techne (Minneapolis, Minn.). Mouse monoclonal antibody specific for caveolin-1 was purchased from BD Biosciences (San Jose, Calif.). Rabbit polyclonal antibody specific for caveolin-1 was obtained from Santa Cruz Biotechnology, Inc. (Santa Cruz, Calif.). Alexa 488-conjugated donkey anti-goat IgG F(ab′)₂ antibody, Alexa 488-conjugated goat anti-rabbit IgG F(ab′)₂ antibody, and Alexa 568-conjugated goat anti-mouse IgG F(ab′)₂ antibody were purchased from Molecular Probes (Carlsbad, Calif.). Peroxidase-conjugated anti-rabbit IgG was obtained from Bio-Rad (Hercules, Calif.). Mouse monoclonal and rabbit polyclonal anti-P. gingivalis fimbria antibodies were produced as described previously (22). Human serum albumin (HSA) and methyl-β-cyclodextrin (MβCD) was purchased from Sigma (St. Louis, Mo.).

P. gingivalis invasion assay.

Semiconfluent KB cells (1 × 10⁵ cells/well) in 24-well plates (BD Biosciences) were incubated with 1 × 10⁷ P. gingivalis cells in culture medium at 37°C for 90 min in a humidified 5% CO₂ incubator. The monolayers were washed three times with minimum essential medium (Sigma), and further incubated in experimental medium containing gentamicin (300 μg/ml) and metronidazole (200 μg/ml) for 1 h to kill the extracellular bacteria. The monolayers were washed again three times and then lysed with distilled water for 20 min. The intracellular bacteria were enumerated by plating on tryptic soy agar plates supplemented with 5% horse blood, hemin, and menadione. In some experiments, KB cells were pretreated with various inhibitors for 30 min prior to the addition of the bacteria. The effects of these inhibitors on KB cells were assessed by an lactate dehydrogenase cytotoxic assay, which showed that they did not affect cell viability. The lactate dehydrogenase cytotoxic assay was performed according to the manufacturer's instructions (Cytotoxicity Detection Kit; Roche Diagnostics, Rotkreuz, Switzerland).

Enzyme-linked immunosorbent assay.

Recombinant human ICAM-1 or HSA (1 μg/well) samples were immobilized in the wells of a 96-well microplate in 50 mM of carbonate buffer, pH 9.6, at 4°C for 16 h. P. gingivalis fimbriae were diluted with 20 mM of Tris-HCl (pH 7.4) buffer containing 2% bovine serum albumin. After the unbound proteins were washed with phosphate-buffered saline (PBS) containing 0.05% Tween 20, P. gingivalis fimbriae (a polymeric form of the fimbrillin) at various concentrations were reacted with ICAM-1 at 37°C for 1 h. The wells were washed three times with PBS containing 0.05% Tween 20, and then rabbit polyclonal anti-P. gingivalis fimbria antibody (1:500) was added and incubated at 37°C for 1 h. The microplate was washed three times, and peroxidase-conjugated anti-rabbit IgG was added. After being incubated at 25°C for 1 h, 3,3′,5,5′-tetramethylbenzidine microwell peroxidase substrate (Kirkegaard & Perry Laboratories, Gaithersburg, Md.) was added after the wells were washed, and absorbance at 450 nm was measured.

Immunofluorescence microscopy.

KB cells were seeded onto cover glasses in a 35-mm dish (BD Biosciences) at 5 × 10⁵ cells and cultured at 37°C for 24 h in a humidified 5% CO₂ incubator. The cells were cocultured with P. gingivalis 381 at a multiplicity of infection of 100 at 37°C for 30 min. The glasses were washed four times with PBS containing 0.1% Tween 20, fixed with 3.7% formaldehyde for 15 min, washed three times, and blocked with PBS containing 5% donkey serum at 25°C for 1 h. After being washed three times, the cover glasses were incubated with mouse monoclonal anti-P. gingivalis fimbria antibody (1:1,000). All antibodies were diluted in PBS containing 5% donkey serum. The cells were labeled with Alexa 568-labeled goat anti-mouse IgG (heavy plus light chains [H+L]) (1:500) and then incubated with goat polyclonal anti-ICAM-1 or E-cadherin antibody (1:200) followed by Alexa 488-labeled donkey anti-goat IgG (H+L) (1:500). Each sample was mounted in FluoroGuard antifade reagent (Bio-Rad) and observed using an MRC1024 microscope (Bio-Rad). For caveolin-1 labeling, the cells were permeabilized with 0.01% Triton X-100 at 25°C for 5 min, washed with PBS containing 0.1% Tween 20, and incubated with rabbit polyclonal anti-caveolin-1 antibody (1:10) followed by Alexa 488-labeled goat anti-rabbit IgG (H+L) (1:500). The observation data were saved using LaserSharp2000 software (Bio-Rad). Randomly chosen images are shown and are representative of three separate experiments. Each shows at least five fields containing an average of 40 cells per specimen.

Transfection of siRNA.

Human ICAM-1, caveolin-1, and control small interfering RNAs (siRNAs) were obtained from Santa Cruz Biotechnology, Inc. The day before transfection, KB cells were seeded into a 12-well plate and incubated at 37°C for 16 h to 50% confluence without antibiotics in a humidified 5% CO₂ incubator. Prior to transfection, 0.125, 0.25, or 0.5 μg of each siRNA was added to 12.5, 25, or 50 μl of siRNA transfection medium (Santa Cruz Biotechnology, Inc.), mixed gently, and kept at 25°C for 5 min. In addition, 0.5, 1, or 2 μl of Lipofectamine 2000 (Invitrogen Corp., Carlsbad, Calif.) was added to 12.5, 25, or 50 μl of siRNA transfection medium, mixed gently and kept at 25°C for 5 min. After 5 min, the prepared solutions were mixed to form siRNA-Lipofectamine 2000 complex and incubated at 25°C for 20 min. The medium was removed, and fresh minimal essential medium with 10% serum was added. The KB cells were transfected with the siRNA-Lipofectamine 2000 complex in siRNA transfection medium and incubated at 37°C for the indicated times. The cells were analyzed by reverse transcription-PCR (RT-PCR) and flow cytometric analysis as described below and used for the invasion assay.

RT-PCR.

Total cellular RNA from KB cells was extracted with TRIzol reagent (Invitrogen Corp.) according to the manufacturer's instructions. RNase-free DNase (Takara Biochemicals, Shiga, Japan) was used to remove genomic DNA based on a method described previously (6). Extracted RNA (0.1 μg) was reverse transcribed into first-strand cDNA at 42°C for 40 min as described previously (2). PCR amplification for human β-actin was performed using oligonucleotide-specific primers (5′-GTGGGGCGCCCCAGGCACCA-3′ and 5′-CTCCTTAATGTCACGCACGATTTC-3′), and 5 μg of cDNA from the samples was amplified with 0.2 μM of the sense and antisense primers for the target gene in a 50-μl reaction mixture containing 75 U/ml of Ex Taq polymerase (Takara Biochemicals). After initial denaturation at 94°C for 2 min, 28 cycles of denaturation (94°C for 30 s), annealing (58°C for 1 min), and extension (72°C for 1 min) for the respective target genes were performed using a PCR EXPRESS thermal cycler (HYBAID, Middlesex, United Kingdom). PCR assays for ICAM-1 and caveolin-1 were performed according to the following RT-PCR primer protocol (Santa Cruz Biotechnology, Inc.): for ICAM-1, 16 cycles at 94°C for 30 s, 63°C for 1 min, and 72°C for 1 min with primer A and 1.0 mM of MgCl₂, repeated one more time with primer B; for caveolin-1, 16 cycles at 94°C for 30 s, 65°C for 1 min, and 72°C for 1 min, repeated once with 1.5 mM of MgCl₂. The primers for β-actin, ICAM-1, and caveolin-1 were constructed to generate fragments of 506, 435, and 660 bp, respectively. For a negative control, a sample without reverse transcriptase was amplified by PCR.

Flow cytometry.

The KB cells transfected with or without siRNA (0.25 μg/well) at 37°C for 24 h were incubated at 25°C for 15 min with mouse monoclonal antibody to human ICAM-1 or caveolin-1. For the controls, the cells were incubated with mouse IgG1κ isotype (Dako, Glostrup, Denmark). After being washed with PBS-0.1% azide, the cells were incubated at 25°C for 15 min with fluorescein isothiocyanate-conjugated goat anti-mouse IgG (eBioscience, San Diego, Calif.). The cells were washed with PBS-0.1% azide and then fixed with 1% paraformaldehyde. The stained cells were analyzed with a FACSCalibur using Cell Quest software (BD Bioscience). For caveolin-1 detection, the cells were incubated with BD FACS lysing solution followed by BD FACS permeabilizing solution 2 (BD Bioscience) according to the manufacturer's instructions before addition of anti-caveolin-1 antibody.

Statistical analysis.

Data were analyzed using a one-way analysis of variance and the Bonferroni or Dunn method, with the results presented as the mean ± standard error of the mean (SEM).

RESULTS

ICAM-1 is related to P. gingivalis invasion.

We examined the relationship between ICAM-1 and P. gingivalis invasion into KB cells by using an invasion assay. The cells were preincubated with goat polyclonal anti-ICAM-1 antibody at 37°C for 30 min before the addition of P. gingivalis, and the antibody inhibited the bacterial invasion in a dose-dependent manner (Fig. 1). The anti-ICAM-1 antibody at a dilution of 1:1,000 exhibited significant inhibitory activity towards the invasion of P. gingivalis into KB cells. In contrast, P. gingivalis invasion was not prevented by goat IgG, the control antibody. In addition, goat polyclonal anti-E-cadherin antibody, even at a dilution of 1:100, did not inhibit the invasion.

FIG. 1.

Inhibition of P. gingivalis invasion into KB cells by anti-ICAM-1 antibody. The cells were preincubated with or without either goat polyclonal anti-ICAM-1 antibody, goat polyclonal anti-E-cadherin antibody, or goat IgG at 37°C for 30 min as indicated, and then P. gingivalis organisms were incubated with the cells at 37°C for 90 min. Invasion of the cells by P. gingivalis at a multiplicity of infection of 100 was determined by an invasion assay. The assays were carried out in triplicate as described in Materials and Methods. The CFU of P. gingivalis invasion without antibodies was set at 100%. **, significantly different (P < 0.01) from the mean value for preincubation without antibody. Error bars indicate standard errors of the means.

P. gingivalis fimbriae bind to ICAM-1.

P. gingivalis fimbriae are known to be a pathogenic factor in the adhesion to host cells. Our results showed that P. gingivalis fimbriae bound to ICAM-1 in a dose-dependent manner (Fig. 2). On the other hand, the fimbriae scarcely bound to HSA.

FIG. 2.

P. gingivalis fimbriae bind to ICAM-1. Binding was detected using rabbit polyclonal anti-P. gingivalis fimbria antibody, peroxidase-conjugated secondary antibody, and 3,3′,5,5′-tetramethylbenzidine substrate with enzyme-linked immunosorbent assay. The assays were carried out in triplicate as described in Materials and Methods. * and **, significantly different (P < 0.05 and P < 0.01, respectively) from the mean value for each dose of P. gingivalis fimbriae with HSA coating. Error bars indicate standard errors of the means.

P. gingivalis colocalizes with ICAM-1.

Immunofluorescence microscopy showed the colocalization of ICAM-1 with P. gingivalis when KB cells were coincubated with P. gingivalis for 30 min and then treated with mouse monoclonal anti-P. gingivalis fimbria antibody, with ICAM-1 clustered around the P. gingivalis cells (Fig. 3G). These findings suggest that P. gingivalis fimbriae are able to bind to the ICAM-1 molecule. On the other hand, colocalization of E-cadherin with P. gingivalis was not observed (Fig. 3H).

FIG.3.

Colocalization of P. gingivalis with ICAM-1 on KB cells. The cells were incubated without (A and B) or with (C through H) P. gingivalis at 37°C for 30 min and then fixed and stained with antibodies specific for ICAM-1 (green) and P. gingivalis fimbriae (red). Colocalization of ICAM-1 and P. gingivalis is shown in yellow in the merged image (G). However, the organisms were not colocalized with E-cadherin (H). The cells were examined by laser scanning confocal microscopy using a 60× high-numerical-aperture oil immersion objective on an MRC 1024 microscope.

Repression of P. gingivalis invasion in ICAM-1 knockdown KB cells.

To confirm the contribution of ICAM-1 in P. gingivalis invasion, we examined whether P. gingivalis is able to invade ICAM-1 knockdown KB cells. The invasion assays were carried out following the transfection of ICAM-1-specific siRNA at a concentration of 0.25 μg/well for 24 h (Fig. 4A, B, and C). The siRNA-mediated knockdown of ICAM-1 expression on KB cells reduced the internalization of P. gingivalis into the cells (Fig. 4D). On the other hand, the control siRNA-transfected and nontransfected KB cells allowed P. gingivalis invasion. These results suggest that ICAM-1 is required for the invasion of P. gingivalis into KB cells.

FIG. 4.

Inhibition of P. gingivalis invasion into KB cells following siRNA-mediated knockdown of the ICAM-1 gene. (A and B) RT-PCR analysis of ICAM-1 mRNA expression in KB cells transfected with ICAM-1-specific siRNA was performed. Semiconfluent KB cells were incubated with siRNA for the indicated times at a concentration of 0.25 μg/well (A) or at the indicated concentrations for 24 h (B). Total RNA was then extracted from the cells and cDNA prepared. (C) ICAM-1 expression in KB cells transfected with siRNA was determined by flow cytometry. Semiconfluent KB cells were incubated without siRNA, with control siRNA, or with ICAM-1-specific siRNA at a concentration of 0.25 μg/well for 24 h. The cells were stained with ICAM-1-specific antibody (solid lines) or its isotype control (dotted lines). (D) The invasion assay was performed using ICAM-1 knockdown KB cells. Semiconfluent KB cells were incubated with or without siRNA as described for panel C. The invasion assays were carried out in triplicate as described in Materials and Methods. The CFU of P. gingivalis invasion without siRNA was set at 100%. **, significantly different (P < 0.01) from the mean value for P. gingivalis invasion in nontransfected cells. Error bars indicate standard errors of the means.

Inhibition of P. gingivalis invasion by MβCD.

MβCD, a cholesterol-binding agent, prevented ICAM-1 clustering on KB cells cocultured with P. gingivalis (Fig. 5). MβCD has also been shown to destroy caveolae, which are the entry point for various types of bacterial invasion (15). To examine the inhibitory effect of MβCD on P. gingivalis invasion, KB cells were treated with various concentrations of MβCD prior to coincubation with P. gingivalis cells. MβCD at a dose of 10 mM caused 50% inhibition of the invasion by P. gingivalis (Fig. 6).

FIG. 5.

Inhibition of colocalization of P. gingivalis with ICAM-1 on KB cells by MβCD. After KB cells were pretreated without (A) or with (B) MβCD at 37°C for 30 min, they were cultured with P. gingivalis at 37°C for 30 min and then fixed and stained with antibodies specific for ICAM-1 (green) and P. gingivalis fimbriae (red). The cells were examined as described for Fig. 3.

FIG. 6.

Inhibition of P. gingivalis invasion in KB cells by MβCD. After KB cells were pretreated with MβCD at 37°C for 30 min, invasion assays were carried out in triplicate as described in Materials and Methods. The CFU of P. gingivalis invasion without MβCD was set at 100%. **, significantly different (P < 0.01) from the mean value for P. gingivalis invasion into cells without MβCD pretreatment. Error bars indicate standard errors of the means.

P. gingivalis colocalizes with caveolin-1.

Next, we examined the relationship between P. gingivalis and caveolin-1, a caveolar marker that is an integral component of the structure of caveolae, in KB cells by using immunofluorescence microscopy. Merged images showed the colocalization of caveolin-1 in the cytosol of KB cells with P. gingivalis (Fig. 7).

FIG. 7.

Colocalization of P. gingivalis with caveolin-1 on KB cells. The cells were incubated with P. gingivalis at 37°C for 30 min and then fixed and stained with antibodies specific for caveolin-1 (green) and P. gingivalis fimbriae (red). Colocalization of caveolin-1 and P. gingivalis is shown in yellow in the merged image. The cells were examined as described for Fig. 3.

Repression of P. gingivalis invasion in caveolin-1 knockdown KB cells.

Next, we knocked down caveolin-1 in KB cells by RNA interference. Transfection of caveolin-1 siRNA resulted in suppression of caveolin-1 mRNA and protein expression (Fig. 8A, B, and C). P. gingivalis invasion was inhibited by a reduction of caveolin-1 with specific siRNA at a concentration of 0.25 μg/well for 24 h (Fig. 8D). These findings indicate that caveolae are strongly associated with the invasion of KB cells by P. gingivalis.

FIG. 8.

Inhibition of P. gingivalis invasion into KB cells following siRNA-mediated knockdown of caveolin-1. (A and B) RT-PCR analysis of caveolin-1 mRNA expression in KB cells transfected with caveolin-1-specific siRNA was performed. Semiconfluent KB cells were incubated with siRNA for the indicated times at a concentration of 0.25 μg/well (A) or at the indicated concentrations for 24 h (B). Total RNA was then extracted from the cells and cDNA prepared. (C) Caveolin-1 expression in KB cells transfected with caveolin-1-specific siRNA was determined by flow cytometry. Semiconfluent KB cells were incubated without siRNA, with control siRNA, or with caveolin-1-specific siRNA at a concentration of 0.25 μg/well for 24 h. The cells were stained with caveolin-1-specific antibody (solid lines) or its isotype control (dotted lines). (D) The invasion assay was performed using caveolin-1 knockdown KB cells. Semiconfluent KB cells were incubated with or without siRNA as described for panel C. The invasion assays were carried out in triplicate as described in Materials and Methods. The CFU of P. gingivalis invasion without siRNA was set at 100%. **, significantly different (P < 0.01) from the mean value for P. gingivalis invasion into nontransfected cells. Error bars indicate standard errors of the means.

DISCUSSION

P. gingivalis has been detected in dental pockets and is known to have various virulence factors that lead to chronic periodontal diseases (32). This pathogenic bacterium initially attaches to oral epithelial cells to exert its virulence and invades host cells. In this study, we demonstrated that P. gingivalis invasion was inhibited by pretreatment of oral epithelial cells with the anti-ICAM-1 antibody and by ICAM-1 knockdown with RNA interference (Fig. 1 and 4). Furthermore, immunofluorescence microscopy showed that P. gingivalis was colocalized with ICAM-1 on KB cells after 30 min of coculture (Fig. 3). These findings indicate that ICAM-1 is associated with P. gingivalis invasion. B. henselae also accumulates ICAM-1 for invasion into endothelial cells. In this case, ICAM-1 was enriched in the membrane protrusions entrapping the bacterial aggregation (4). Furthermore, ICAM-1 plays an important role in the adherence of Candida albicans to pulmonary vascular endothelial cells (37).

In the case of invasion into epithelial cells by Listeria monocytogenes, the organisms require E-cadherin, which is a adhesion molecule and the receptor for internalin B, a cell wall-associated protein of L. monocytogenes (16, 24). Recently, ClpP, the protein which has homology with internalin, was detected in P. gingivalis, and the protein was associated with invasion of the organisms, because insertional inactivation of clpP resulted in approximately a 50% reduction in invasion by P. gingivalis (38). Therefore, we examined the colocalization of P. gingivalis with E-cadherin, but it was not detected by immunofluorescence microscopy(Fig. 3).

P. gingivalis fimbriae are thought to be required for the adhesion to and invasion of host cells, because the number of fimbria-deficient mutant cells which invade oral epithelial cells is much less than that of the wild type (7, 20, 36). In our experiments, P. gingivalis fimbriae were shown to bind to ICAM-1, indicating that the fimbriae participate in the invasion by P. gingivalis (Fig. 2). Fimbriae also play an essential role in the activation of human cells (23, 25), and P. gingivalis fimbriae utilize Toll-like receptor 2 (TLR2) for NF-κB activation, which enhances ICAM-1 expression (10, 21). TLR2 is a main signaling protein that displays sensitivity to a variety of microbial pathogens, including P. gingivalis fimbriae on KB cells and human gingival epithelial cells (2, 34). TLR2 also mediates tyrosine phosphorylation of Src family kinases (12). Tilghman et al. (35) showed that MβCD inhibits the association of Src family kinases with ICAM-1, and Src was found to be required for the clustering of adhesion molecules. We found that P. gingivalis did not colocalize with ICAM-1 on MβCD-pretreated KB cells, because the clustering of ICAM-1 was inhibited (Fig. 5). Therefore, TLR2 on oral epithelial cells might be associated with the process of P. gingivalis invasion related to ICAM-1 clustering.

We also demonstrated that caveolin-1 is associated with P. gingivalis invasion. Caveolar vesicles, which are caveolin-1-containing endocytic compartments, are targeted to early endosomes in a Rab5-dependent pathway (1, 26, 27). These events correlate with the finding that vesicles containing P. gingivalis colocalize with Rab5 after internalization (8). Further, the pathogen is trafficked to the autophagosome, which is fused to the lysosome (8, 9). However, Dorn et al. (8) indicated that a functional autophagic system is necessary for P. gingivalis survival in endothelial cells because they could escape from the endocytic pathway by the transfer to the autophagosome. P. gingivalis might gain time until the lysis by utilizing caveolar vesicles and autophagosomes.

In the case of E. coli, FimH, a mannose-binding lectin contained in the surface type 1 fimbriae, is an important factor for invasion (3). FimH also binds to CD48, which was shown to colocalize with caveolae in mast cells (31). Further, another study found that outer membrane protein A-deficient E. coli K1 could not activate protein kinase Cα and enhance the association of this molecule with caveolin-1, and thus the number of outer membrane protein A deletion mutant cells which invaded endothelial cells was much less than that of the wild type (33). On the other hand, Streptococcus pyogenes was shown to invade respiratory epithelial cells via caveolae by binding to human fibronectin on the cells (17, 28). Streptococcal fibronectin binding protein 1, which is present in S. pyogenes but not in Streptococcus gordonii, may trigger the accumulation of caveolae, and P. gingivalis fimbriae might have acted like those proteins in the present experiments. Taken together, these results demonstrate that ICAM-1 and caveolae are important for P. gingivalis invasion.