Molecular Mechanism for Connective Tissue Destruction by Dipeptidyl Aminopeptidase IV Produced by the Periodontal Pathogen Porphyromonas gingivalis

The P. gingivalis strains used in this study are described in Table 1 and were cultured as described previously (21, 22). Primary cultures of human gingival fibroblasts (HGFs) excised from healthy gingival tissue were obtained from M. Inoue (Department of Biochemistry, Nippon Dental University). NIH 3T3 cells were obtained from T. Sasakawa (Department of Bacteriology, Institute of Medical Science, University of Tokyo). HGF cells and NIH 3T3 cells were grown as monolayer cultures in Dulbecco's modified Eagle's medium-nutrient mixture F-12 (Life Technologies, Rockville, Md.) and Dulbecco's modified Eagle's medium, respectively, supplemented with 10% fetal bovine serum (Life Technologies).

Human MMP-1 (Sigma-Aldrich, St. Louis, Mo.) and human MMP-2 (Sigma-Aldrich or Biogenesis, Poole, England) were activated by incubation with 2 mM aminophenylmercuric acetate (Sigma-Aldrich) for 1 h at 37°C. Recombinant P. gingivalis DPPIV was purified from Escherichia coli DPPRWT and DPPRSA, expressing the wild-type and mutant enzymes, respectively, as described previously (21) and was dialyzed against 10 mM potassium phosphate buffer (pH 7.5). For the gelatinase assay, 2 μg of acid-soluble human type I collagen (catalog no. C-7774; Sigma-Aldrich) was denatured by incubation for 5 min at 95°C and incubated for 10 h with DPPIV, MMP-2 (Sigma-Aldrich), or DPPIV plus MMP-2 in 20 mM Tris-HCl buffer (pH 7.5) at 37°C. To assess collagen degradation, human type I collagen was incubated with MMP-1 in the absence or presence of DPPIV in 50 mM Tris-HCl (pH 7.5) buffer for 6 h at 25 or 37°C. A serine protease inhibitor, diisopropyl fluorophosphate (DFP) (Wako, Osaka, Japan), or a metalloproteinase inhibitor, 1,10-phenanthroline (Nacalai Tesque, Kyoto, Japan), was added to the reaction mixture as required in the gelatinase or collagenase assay. Degradation of human plasma fibronectin (catalog no. F-0895; Sigma-Aldrich) and human laminin (catalog no. L-6274; Sigma-Aldrich) by DPPIV was examined by incubation in 20 mM Tris-HCl buffer (pH 7.5) at 37°C. Reaction mixtures were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (24), and protein bands were stained with Coomassie brilliant blue.

The samples containing gelatin or collagen were precipitated with 80% ethanol. The amino acids and low-molecular-weight peptides are not precipitated under these conditions. The peptides obtained were hydrolyzed in vacuo in 5.7 M HCl containing 1% phenol and 1% 2-mercaptoethanol for 24 h at 108°C. The amino acid composition was determined by the orthophthalic acid method (16) with high-pressure liquid column chromatography. The amounts of precipitated gelatin and collagen were determined from the glycine contents of the amino acids.

The binding activity of DPPIV with acid-soluble human type I collagen, human fibronectin, or human laminin was examined by solid-phase enzyme-linked immunosorbent assay. Microtiter plates (96 wells, flat bottom, polystyrene) (Corning Costar, Cambridge, Mass.) were coated with 1 μg of ECM protein solution diluted with phosphate-buffered saline (PBS) by incubation for 12 h at 4°C. As a negative control, wells were coated with 1 μg of casein (Wako) in PBS. The supernatant was removed, wells were washed three times with PBS, and nonspecific binding sites were blocked by incubation with 10 μg of heat-denatured casein for 2 h at room temperature. After three washes with PBS containing 0.05% Tween 20 (PBST), 100 μl of purified DPPIV diluted with PBS was added to each well and incubated for 2 h at 37°C. Wells were washed three times with PBST and incubated with 100 μl of rabbit anti-DPPIV serum (11) (diluted 1:500 in PBS) for 2 h at room temperature. After three washes with PBST, wells were incubated with 100 μl of alkaline phosphatase-conjugated goat anti-rabbit immunoglobulin G (IgG) (Zymed, South San Francisco, Calif.) (diluted 1:1,000 in PBS) for 2 h at room temperature. The wells were washed twice with PBST and once with 100 mM Gly-NaOH buffer (pH 10.0) containing 1 mM ZnCl₂ and 1 mM MgCl₂ (AP buffer) and then were incubated at room temperature with 100 μl of AP buffer containing 5 mM p-nitrophenyl phosphate (Wako). The color reaction was stopped with 50 μl of 1 M NaCl, and the optical density at 405 nm was measured with an NJ-2001 ImmunoReader (InterMed). The binding of DPPIV to insoluble type I collagen (catalog no. C-9879; Sigma-Aldrich) was examined as described previously (26).

P. gingivalis strains cultured to the logarithmic phase of growth were harvested, washed twice with PBS, and resuspended in PBS to an optical density at 600 nm of 5. Microtiter plates were coated with 100 μl of the bacterial suspension, dried at 37°C under anaerobic conditions, washed with PBS, and blocked. Various amounts of human fibronectin were added, and bound fibronectin was measured by enzyme-linked immunosorbent assay with a rabbit anti-human fibronectin antibody (Transformation Research, Framingham, Mass., or Sigma-Aldrich).

HGF cells and NIH 3T3 cells (see “Bacterial strains and cells” above) were used for the cell adhesion assay. Cover glasses (diameter, 12 mm) (Matsunami, Tokyo, Japan) placed in wells of a 24-well microtiter plate (Corning Costar) were coated with 5 μg of fibronectin in PBS by incubation for 12 h at 4°C and washed three times with PBS. The nonspecific binding sites were blocked with 1% bovine serum albumin (BSA) in PBS by incubation for 30 min at room temperature. Nearly confluent cells (70 to 80% confluent) were detached from culture dishes by treatment with 0.25% trypsin (Life Technologies). Trypsin inhibitor (Sigma-Aldrich) (0.25 mg/ml) was added to the cell suspension, and cells were washed twice with PBS. After the cover glasses were washed with PBS three times, 5 × 10⁴ cells suspended in serum-free medium were plated, and purified DPPIV or heat-denatured BSA as a negative control was added to each well following incubation for 30 min in a CO₂ (5%) incubator at 37°C. To examine the inhibition of cell attachment by anti-integrin antibodies, cells were pretreated with 1 μg of an anti-human integrin β1 mouse monoclonal antibody (clone DE9) (Upstate Biotechnology, Lake Placid, N.Y.) or an anti-human integrin α5 mouse monoclonal antibody (clone PID6) (Chemicon, Temecula, Calif.) in PBS for 30 min at room temperature, centrifuged, resuspended in serum-free medium, and plated. Cells were fixed with 4% paraformaldehyde (Chiyoda Junyaku, Tokyo, Japan) in PBS and stained with 1% crystal violet (Wako). The cover glasses were mounted with glycerol, and cell morphology was observed at a magnification of 100-fold under a microscope (MICROPHOT-FX; Nikon, Tokyo, Japan) equipped with an HC-300Zi digital camera (Nikon Fujix, Tokyo, Japan). Micrographs of more than 50% of the fields in each cover glass were transferred to computer hardware with Photograb version 2.0. The attached area of cells was measured with Image J version 1.17.

Wells of microtiter plates were coated with fibronectin and blocked as described above. Human α5β1 integrin (Chemicon) diluted in 100 μl of PBS was added and incubated in the absence or of DPPIV. The amount of bound α5β1 integrin was measured by using mouse anti-β1 integrin monoclonal antibody (Chemicon) and alkaline phosphatase-conjugated goat anti-mouse IgG (Bio-Rad). The inhibition by DPPIV of the binding of α5β1 integrin to fibronectin was examined by incubating DPPIV in the fibronectin-coated wells in the presence of α5β1 integrin and measuring the amount of bound DPPIV with anti-DPPIV antiserum.

Anti-DPPIV antibody (IgG) was purified from anti-DPPIV antiserum (21) by ammonium sulfate precipitation (ca. 25% saturation) following chromatography on a column of DEAE equilibrated with 15 mM potassium phosphate buffer (pH 8.0) and eluted with a linear gradient of 15 to 300 mM potassium phosphate buffer (pH 8.0). DPPIV and the actin filaments of cells attached to cover glasses were stained as follows. After fixation with 4% paraformaldehyde, cells were washed three times with PBS, quenched with 50 mM NH₄Cl in Tris-buffered saline (TBS) for 10 min, permeabilized with 0.2% Triton X-100 in TBS for 10 min, and blocked with 2% BSA in TBS for 30 min at room temperature. Cells were incubated with 100 μl of a rabbit anti-DPPIV antibody (diluted 1:500 in TBS) for 1 h at 37°C, washed three times with TBS containing 0.05% Tween 20 (TBST), and incubated with 100 μl of fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit IgG (Sigma-Aldrich) (diluted 1:100 in TBS) for 1 h at 37°C. After three washes with TBST, the cells were incubated with 100 μl of rhodamine-conjugated phalloidin (Molecular Probes, Eugene, Oreg.) (diluted 1:100 in TBS) for 30 min at 37°C. After three washes with TBST and two washes with distilled water, the cover glasses were mounted in Vectashield (Vector Laboratories, Burlingame, Calif.) and observed under a fluorescence microscope (LABOPHOT-2; Nikon). For staining vinculin, a mouse antivinculin antibody (Molecular Probes) and FITC-conjugated goat anti-mouse IgG (Sigma-Aldrich) were used.

P. gingivalis strains were grown and injected subcutaneously into mice as described previously (21). General health, weight, and the presence and location of lesions were examined daily for 14 days. For histopathological analysis, animals challenged with P. gingivalis strains were euthanatized 3 days after injection. Skin specimens including lesions were dissected, fixed in 4% paraformaldehyde-0.1 M sodium cacodylate buffer (pH 7.3), and then embedded in paraffin. Serially cut sections (4 μm thick) were stained with hematoxylin-eosin for routine diagnosis and with Elastica van Gieson stain for detection of collagen and fibrous structures. Inflammatory cells were identified by their morphology. Two sections from the central areas of the most severe host reactions and two sections from the peripheral area of lesions from each animal were examined. Two fields were randomly selected from within the most severely inflamed areas of lesions from each animal. The numbers of inflammatory cells and bacteria per unit area of the fields were counted at magnifications of ×400 and ×1,000, respectively, with ImageGauge version 3.12 (Fuji Film).

We studied the degradation of an ECM protein composing the connective tissue, type I collagen, by P. gingivalis DPPIV. Since type I collagen contains many Gly-Pro sequences, we hypothesized that P. gingivalis DPPIV is involved in the degradation of collagen via an endopeptidase activity that digests gelatin (namely gelatinase activity) and/or via an exopeptidase activity that digests degradation products of gelatin produced by the activity of MMP2. Initially we examined the gelatinase activity of DPPIV toward heat-denatured collagen (gelatin) type I. No detectable decrease in the amount of α1 and α2 chains was observed when gelatin prepared from type I collagen (by heat denaturation as described in Materials and Methods) was incubated without any enzyme, indicating that no heat-resistant gelatinolytic enzyme contaminated the gelatin substrate (data not shown). When the gelatin solution was incubated with DPPIV at 37°C, the amount of the two chains did not decrease (Fig. 1 lane 4), indicating that DPPIV does not possess endopeptidase activity toward gelatin. We next explored the effect of DPPIV activity on gelatin in combination with a host-derived gelatinase, MMP-2. Gelatin prepared from type I collagen by heating at 95°C was degraded slightly by MMP-2 (Fig. 1, lane 2). This activity of MMP-2 was inhibited in the presence of a metalloproteinase inhibitor, EDTA (Fig. 1. lane 3). When DPPIV was added to the reaction mixture containing MMP-2, further digestion of gelatin was observed in proportion to the amount of DPPIV (Fig. 1, lanes 5 to 7). This activity of DPPIV was inhibited by a serine protease inhibitor, DFP (Fig. 1, lane 11), which blocks the exopeptidase activity of DPPIV. No digestion of gelatin by DPPIV and MMP-2 was observed in the presence of the MMP-2 inhibitor EDTA, which does not inhibit DPPIV activity (Fig. 1, lane 8). DPPIV did not catalyze the conversion of pro-MMP-2 to native MMP-2 (data not shown), suggesting that DPPIV degrades gelatin by exhibiting exopeptidase activity toward the degradation products generated by MMP-2 and not via endopeptidase activity to activate MMP-2. DPPIV could not degrade ECM components (fibronectin or laminin) as substrates (data not shown).

Type I collagen is composed of two α1(I) chains and one α2(I) chain (Fig. 2, lane 2). No decrease in the amount of collagen was detected after incubation without any enzyme at 25 or at 37°C (not shown), indicating that there is no collagenolytic enzyme in this solution. Degradation of type I collagen was observed in the presence of MMP-1 at 25°C (Fig. 2, lane 3); two 3/4 fragments (containing two α1A chains and two α1B chains) and one 1/4 fragment (containing one α2A chain and one α2B chain) were produced from type I collagen but were not digested further by additional incubation at 25°C. The 3/4 and 1/4 fragments are thought to be denatured at 37°C (6). The denaturation temperatures of the 3/4 and 1/4 fragments are reported to be 32 and 28°C, respectively, both of which are lower than that of the intact form (40 to 42°C). Therefore, the 3/4 and 1/4 fragments are thought to be denatured at 37°C, and digested by the gelatinase activity of MMP-1. When type I collagen was incubated with MMP-1 at 37°C, the α1A and α2A chains were further digested because of the gelatinolytic activity of MMP-1 (Fig. 3, lane 2). While the intensity of the band on the sodium dodecyl sulfate-polyacrylamide gel was obviously reduced early at 37°C, the results of the amino acid analysis indicated 88% recovery after ethanol precipitation (Fig. 3, lane 2). The underestimation can be explained by precipitation of the products, α1A, α1B, α2A, and α2B, together with the substrates, α1 and α2. After a long incubation at 37°C, the recovery drastically decreased, indicating further degradation of the products by the gelatinase activity of MMP-1 (data not shown). This collagenolytic reaction by MMP-1 was inhibited completely by EDTA, a metalloproteinase inhibitor (Fig. 3, lane 3). Type I collagen was not degraded when it was incubated with P. gingivalis DPPIV at 25°C (not shown) or at 37°C (Fig. 3, lane 4), indicating that native collagen and the gelatin produced by the native collagen on further incubation at 37°C are not substrates for DPPIV. When type I collagen was incubated with MMP-1 plus DPPIV (Fig. 3, lanes 5 to 7) at 37°C, enhancement of the degradation of collagen in a dose-dependent manner was observed. The enhancing effect of DPPIV was abolished by a serine protease inhibitor, DFP (Fig. 3, lane 11), which inhibits the exopeptidase activity of DPPIV. No collagen was digested by DPPIV and MMP-1 in the presence of the MMP-1 inhibitor EDTA, which does not affect DPPIV activity (Fig. 3, lane 8). DPPIV did not show catalytic activity to convert pro-MMP-1 to mature MMP (not shown).

To confirm that the exopeptidase activity of DPPIV is associated with the function to degrade collagen and gelatin, we examined the degradation of collagen and gelatin by a mutant DPPIV, in which the catalytic Ser was changed to Ala (DPPSA), in combination with MMP-1 or MMP-2. The degradation of collagen and gelatin by MMP-1 and MMP-2, respectively, was not enhanced by the mutant (Fig. 4, lanes 4).

We examined the binding of DPPIV to ECM proteins. DPPIV binds to immobilized fibronectin but not to casein as a negative control (data not shown). This binding was dependent on the amount of DPPIV used and was nearly saturable (Fig. 5A). The binding was not inhibited by a 2 mM concentration of a serine protease inhibitor, DFP (data not shown). The efficiency with which the mutant DPPIV (DPPSA) bound to fibronectin was almost the same as that of wild-type DPPIV, suggesting that the active site for peptidase activity is not associated with the fibronectin-binding activity. No significant binding of wild-type or mutant DPPIV to nondenatured acid-soluble human type I collagen, to nondenatured pepsin-extracted type IV collagen, or to laminin was found in our assay system. Binding of insoluble type I collagen also was not observed (not shown).

To investigate whether the adhesion of P. gingivalis to fibronectin is mediated by DPPIV, the binding of immobilized dpp⁺ W83 and dpp null mutant 4351 bacteria to fibronectin was measured. As shown in Fig. 6, the binding capacity of 4351 was significantly less than that of W83 (P < 0.001 by Student's t test). We have previously constructed a shuttle vector that is mobilized from Escherichia coli to P. gingivalis and is maintained stably in both bacteria (Table 1) (22). A P. gingivalis strain overproducing wild-type DPPIV (4351ADPP) was prepared by introducing the shuttle vector containing dpp⁺ into host P. gingivalis having recA and dpp null mutations (4351A). The binding capacity of P. gingivalis 4351ADPP was significantly greater than that of the control strain (4351AVEC, expressing no DPPIV), and was similar to that of 4351ADPPSA (expressing mutant DPPSA) (Fig. 5B), indicating that both the wild type and mutant DPPIV mediate the bacterial binding to fibronectin, and this property is well correlated with the result obtained with the purified DPPIV protein (Fig. 5A).

We next investigated the effect of fibronectin-binding activity of DPPIV in host cells that reside in the connective tissue, i.e., HGFs and NIH 3T3 cells. It has been reported that eukaryotic DPPIV was not detected on membranes of NIH 3T3 cells by flow cytometry and after incubation of the cells with a substrate for the exopeptidase activity of DPPIV, Gly-Pro-pNA (25). We confirmed that there was no apparent expression of DPPIV in NIH 3T3 cells and HGFs excised from two healthy persons by measuring DPPIV exopeptidase activity in sonicated cell extracts. On preincubation of the cells with anti-α5 integrin or anti-β1 integrin monoclonal antibody, cell attachment was inhibited nearly to the basal level on noncoated cover glasses (Table 2). Control IgG did not inhibit the adhesion of the cells to fibronectin. These results are in good agreement with previous reports indicating that cell attachment to fibronectin is mediated mainly by α5β1 integrin in HGFs (17, 37). When cells were plated on fibronectin-coated cover glasses in the presence of DPPIV, the attached area was decreased compared to that in the presence of heat-denatured BSA (Fig. 7 and Table 2). The number of unattached cells, i.e., round cells, was greater in the presence of DPPIV than in the presence of heat-denatured BSA (Table 2). The mutant DPPSA had the same inhibitory activity for the adhesion of cells to fibronectin as the wild-type DPPIV (not shown), indicating that this inhibitory activity is not correlated to the peptidase activity but may be associated with the capacity to bind fibronectin.

This inhibition was judged not to be due to the incorporation of DPPIV into cells from the results of fluorescent staining. Cells were permeabilized with 0.2% Triton X-100 following immunostaining with rabbit anti-DPPIV antibody, FITC-conjugated goat anti-rabbit IgG, and rhodamine-conjugated phalloidin or with mouse antivinculin antibody, FITC-conjugated goat anti-mouse IgG, and rhodamine-conjugated phalloidin. While actin filaments and vinculin were visualized, no fluorescent signal for DPPIV was detected in cells. Fluorographs of DPPIV staining in permeabilized cells were similar to those of DPPIV staining in nonpermeabilized cells (i.e., with no treatment with 0.2% Triton X-100) (data not shown).

Previously, we reported preliminary results on the virulence of P. gingivalis strains in mice (22). In the present study, we made more detailed observations. Fifteen of 24 mice inoculated with 4351AVEC (8 × 10⁹/mouse) developed an abscess on the abdomen in 2 days, and 6 of the mice died in 4 days (Fig. 8). All 22 animals injected with 4351ADPP (8 × 10⁹/mouse) developed an abscess on the abdomen in 2 days, and 20 of the mice died in 6 days (Fig. 8), indicating that the highly virulent phenotype of the dpp⁺ strain was restored by introducing the cloned dpp⁺ gene. When mice were inoculated with 4351ADPPSA (8 × 10⁹), which expresses a peptidase-deficient mutant DPPIV (DPPSA), 21 of 25 animals developed an abscess on the abdomen in 2 days, and 15 of them developed an abscess in 5 days (Fig. 8), indicating that the virulence was partially restored by introducing the cloned mutant dpp gene. An increase in average body weight was observed in 4351AVEC-injected mice at 6 days after inoculation, as observed in 4351-injected animals (43). Two surviving mice inoculated with 4351ADPP did not show an increase in body weight within 14 days. The average body weight of 4351ADPPSA-injected animals was increased 9 days after the inoculation. These results confirm that DPPIV is a virulence factor and suggest that the peptidase activity is essential but not enough for the full virulence of DPPIV.

We compared histopathologically four mice injected with 4351AVEC, five mice injected with 4351ADPP, and eight mice injected with 4351ADPPSA. The average degree of tissue destruction was greater in the mice infected with 4351ADPP than in those infected with 4351AVEC (Fig. 9). Severe destruction of the dermis was observed in the lesions caused by 4351ADPP, mild destruction was observed in the lesions caused by 4351AVEC, and a moderate degree of destruction was observed in the lesion caused by 4351ADPPSA. With regard to the destruction of the fascia, while severe destruction occurred in lesions caused by 4351ADPP, only weak destruction was induced by 4351ADPPSA and 4351AVEC. Although large numbers of inflammatory cells were observed in the lesions caused by 4351AVEC, small numbers were found in the lesions due to 4351ADPPSA and 4351ADPP (Fig. 10A). In contrast, bacterial cell numbers were very high in the lesions caused by 4351ADPP but were low in those caused by 4351AVEC, while the lesions caused by 4351ADPPSA had a moderate number of bacteria (Fig. 10B).