Porphyromonas gingivalis, a gram-negative anaerobe, is thought to be a major etiological agent associated with adult periodontitis (
1,
4,
6,
8). Recently, much interest has been focused on a possible implication of oral bacteria in systemic diseases.
P. gingivalis has been suggested to be involved in aspiration pneumonia in the elderly and also to be responsible for the development of other systemic diseases, such as endocarditis and pulmonary infections, as a putative blood-borne pathogen (
1,
2,
6,
10). However, little is known about the mechanisms by which
P. gingivalis penetrates the systemic circulation to cause infections at distant sites. The mouse abscess model has been used to investigate the virulence of
P. gingivalis (
6). Several components, such as trypsin-like proteases, are considered to be potential virulence factors on the basis of results obtained with this model, involving infection by strains harboring either wild-type or mutant structural genes for the putative virulence factors (
5,
6). However, there is still a paucity of information about specific roles or the relative importance of the potential virulence factors of the bacterium.
Dipeptidyl aminopeptidase IV (DPPIV; EC 3.4.14.5 ) is a serine protease that cleaves X-Pro or X-Ala dipeptide from the N-terminal ends of polypeptide chains. Eukaryotic DPPIV has been postulated to be involved in various biological processes, such as T-cell activation, interaction with collagen and/or fibronectin, and degradation of biologically active peptides, including chemokines and others (
3). However, the physiological and pathological functions of bacterial DPPIV have not yet been clarified. The gene (
dpp) coding for DPPIV has been cloned from
P. gingivalis W83, a
dpp null mutant (4351) has been constructed, and it has been shown that
P. gingivalis DPPIV plays a role as a virulence factor (
9).
In the present study, to assess the role of DPPIV in the virulence of
P. gingivalis, histopathological differences in the lesions caused by W83 and 4351 in a mouse abscess model were investigated. Ten BALB/c mice (male, 11 weeks old; Tokyo Laboratory Animal Science Co. Ltd., Tokyo, Japan) each were challenged with
P. gingivalis W83 and 4351 by dorsal subcutaneous injection as described previously (
9). Protocols conformed to the guidelines for the care and use of laboratory animals at Nippon Dental University. W83 and 4351 were harvested immediately after reaching the stationary phase, as determined by measurement of the optical densities, because the optical densities of both W83 and 4351 are proportional to viable cell numbers by this phase, as previously shown (
9). No significant differences were found between the two strains in growth rates at the logarithmic phase, viable counts after reaching the stationary phase, and viability during exposure to the atmosphere. The dose of bacteria used for injection was determined as reported previously (
5,
6). Phosphate-buffered saline was injected as a negative control.
Half of the animals injected with either W83 or 4351 were euthanatized 3 days after injection. Four of the remaining five mice inoculated with W83 and two of the five inoculated with 4351 were surviving at 3 days after inoculation but died by 6 days. Surviving animals (one mouse injected with W83 and three mice injected with 4351) were euthanatized 14 days after inoculation. Autopsy of the animals was conducted to examine whether infectious lesions had spread to adjacent internal organs or to distant places. Blood specimens were obtained from their tails 2 days after injection. The numbers of viable bacteria were counted by spreading blood specimens on brain heart infusion (Difco, Detroit, Mich.) agar plates supplemented with 5 μg of hemin (Sigma-Aldrich, St. Louis, Mo.)/ml, 0.5 μg of menadione (Sigma-Aldrich)/ml, and 5% defibrinated horse blood (Nippon Bio-supp. Center, Tokyo, Japan), followed by anaerobic incubation of the plates.
Skin specimens, including those from dorsal, lateral, and ventral sites, as well as internal organs (i.e., lungs, spleen, liver, kidneys, heart, and intestines) were dissected for pathological analysis. All tissue specimens were fixed in 4% paraformaldehyde in 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 Elastica van Gieson (EVG) and Azan stains for detection of collagen and fibrous structures. Inflammatory cells were identified by their morphology. For skin specimens, two sections from peripheral areas and two sections from the central, most-severe areas of lesions from every animal were examined. Two fields were randomly selected from within the most-severe areas of lesions from each of four animals inoculated with W83 and 4351 after euthanasia at 3 days. The numbers of inflammatory cells and bacteria in the fields were counted at ×400 and ×1,000 magnifications, respectively.
Macroscopic observations revealed no local abscesses or no signs of illness after injection of phosphate-buffered saline. At autopsy of seriously ill animals challenged with W83 or 4351 and euthanatized at day 3, hemorrhage of intra-abdominal cavities and/or mesentery was detected. Lesions caused by W83 or 4351 on the dorsal surface and abdominal area were similar to those described previously (
9). In short, all mice exhibited no severe skin lesions, except for redness, at the site of injection on the dorsal surface. Animals challenged with W83 exhibited a large abscess which covered the abdominal area and which progressed to ulceration with associated gangrenous skin necrosis and subcutaneous inflammation. In contrast, mice inoculated with 4351 developed merely a small focus of abscess on the abdomen which was resolved or substantially healed by the end of the observation period, with an increase in body weight beginning at 6 days after injection.
Upon histological examination, congestion of lungs, spleen, liver, and kidneys was found. Hepatic cells were hyperplastic, with intraparenchymal focal accumulation of mononuclear cells but without gross necrotic patches or large areas of inflammation. Neither ulceration nor massive inflammatory reactions were found at the injection site. During the development of abscess, the inflammatory infiltrate, comprising mainly inflammatory cells, spread between the cutaneous trunk muscle and the external oblique muscle toward the abdominal region. Bacteria were detected in the spreading infiltrate.
The pathological features of the abdominal regions were markedly different depending on the strain injected. Inoculation of W83 evoked an accumulation of (1.0 ± 0.5) × 10
3inflammatory cells (
n = 8)/mm
2together with (2.1 ± 0.6) × 10
5bacteria (
n = 8)/mm
2; the fasciae appeared to be partially collapsed, giving rise to diffuse penetration of inflammatory cells into the underlying musculature (Fig.
1). In sharp contrast, challenge with 4351 induced an accumulation of (5.0 ± 1.0) × 10
3 inflammatory cells (
n = 8)/mm
2 above the fasciae together with (5.4 ± 3.6) × 10
4 bacteria (
n = 8)/mm
2; almost no inflammatory cells penetrated into the deep muscular layer, and there was no apparent destruction of the fasciae (Fig.
1). The differences in the numbers of both inflammatory cells and bacteria evoked by W83 and 4351 were statistically significant, with a
P value of <0.001 for inflammatory cells and a
P value of <0.02 for bacteria, as determined by unpaired nonparametric analysis (Mann-Whitney test).
W83 caused a spreading type of infection, often accompanied by massive lysis of epithelial and subcutaneous tissues. These lesions generally degraded into gangrenous necrosis of the skin. Staining with EVG or Azan stain revealed a considerable loss of collagen fibers and associated supportive fibrous structures in the dermis and fasciae. In contrast, lesions caused by 4351 appeared far less destructive in connective tissues than those caused by W83 (Fig.
1). A gradual shift from the acute inflammatory phase to the reparative phase was observed in lesions caused by 4351 within 14 days after injection; this shift featured the formation of a subcutaneous granulomatous focus accompanied by blood capillary formation and a number of inflammatory cells and macrophages, leading to the lesions being resolved.
The histopathological differences in lesions caused by W83 and 4351 are likely to be consistent with the findings obtained by in vitro assays using purified DPPIV. DPPIV was found to participate in the degradation of type I collagen together with tissue- or inflammatory cell-derived matrix metalloproteinases 1 and 8 (unpublished data). Furthermore, it has been demonstrated that eukaryotic DPPIV cleaves C-C and C-X-C chemokines at the C terminus of the proline residue on the penultimate position of the N terminus and that truncated chemokines have activity as chemotactic inhibitors (
3). Previous studies proved that recombinant DPPIV of
P. gingivaliswas also capable of cleaving two synthetic substrates possessing the same N-terminal sequences to human RANTES and human MCP1 (
9). On the basis of these data,
P. gingivalis DPPIV is likely to cleave chemokines similarly, leading to the disturbance in the mobilization of inflammatory cells and thus in the host defense mechanism. It is suggested that in lesions caused by W83, a less effective accumulation of inflammatory cells provokes an increase in the level of bacteria, most likely resulting in more-severe tissue destruction by proteases produced by
P. gingivalis, i.e., trypsin-like proteases and DPPIV.
Between 10
3 and 10
6bacteria/ml of blood were detected in animals injected with W83 but not 4351 when the plates were incubated anaerobically. All colonies on plates were black pigmented, specifically,
P. gingivalis. DPPIV is not essential for bacterial growth in vitro, since the rates of growth of W83 and 4351 are similar (
9). In contrast, DPPIV may be required for
P. gingivalis to survive in vivo. This property of DPPIV meets one of the modern criteria for virulence factors (
7), and the difference in the virulence of W83 and 4351 is likely attributable to the lower viability of 4351 in vivo, as implied by the difficulty in isolating the mutant from blood.
P. gingivalisDPPIV also may be involved in dissemination of the bacteria through the destruction of connective tissue, so that bacteria can enter the bloodstream. This process may confer upon the bacterium a mechanism to provoke systemic diseases as a blood-borne pathogen.
We are currently attemting to develop an inoculation model using the oral cavity of mice to investigate activities of DPPIV in the pathogenesis of periodontitis and other systemic diseases.
Acknowledgments
We thank Ayako Yajima for skillful assistance in animal experiments.
This work was supported by grants-in-aid for scientific research 07457071 and 08307004 from the Ministry of Education, Science, Culture and Sports, Tokyo, Japan.