INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Porphyromonas gingivalis has been implicated as a major etiological agent in the onset and progression of chronic periodontitis, a destructive inflammatory disease of the supporting tissues of the teeth which affects between 10 and 15% of dentate adults (10, 20, 49). In a recent study, Griffen et al. (11) analyzed plaque samples from 311 subjects for the presence of heteroduplex types of P. gingivalis. Of the 11 different heteroduplex types detected, P. gingivalis W83/W50-like strains were found to be associated with periodontitis, whereas other strains, including 381-like strains, were not found to be associated with disease. This finding extends earlier animal studies in which strains W83 and W50 were classified as invasive based on their ability to cause ulcerative spreading lesions distant from the injection site, whereas strains 381 and ATCC 33277 were classified as noninvasive as they produced a localized abscess at the site of injection (29, 54). These results therefore suggest that W50 and related strains are more virulent in both animals and humans.

The pathogenicity of P. gingivalis has been attributed to a number of virulence factors, such as fimbriae (4), hemagglutinins (12, 13), lipopolysaccharide (LPS) (14), and the extracellular and cell-associated Arg-x- and Lys-x-specific cysteine proteinases and their associated adhesins (31, 33, 36, 45). Among these factors, the extracellular Arg-x- and Lys-x-specific cysteine proteinases are believed to play a major role in the pathogenesis of periodontal disease, as they are able to degrade a variety of host proteins and have the potential to dysregulate host defense (53).

Three genes encode the major extracellular Arg-x- and Lys-x-specific cysteine proteinases of P. gingivalis, and these are designated rgpA, rgpB, and kgp (6). We have previously characterized the proteins encoded by rgpA and kgp of strain W50 as a cell-associated complex of noncovalently associated proteinases and adhesins, designated the RgpA-Kgp proteinase-adhesin complexes, formerly the PrtR-PrtK proteinase-adhesin complexes (3). The RgpA-Kgp complexes of P. gingivalis strain W50 are composed of a 45-kDa Arg-x-specific proteinase (RgpA45, formerly PrtR45) associated with four sequence-related adhesins, RgpA44, RgpA15, RgpA17, and RgpA27, all encoded by rgpA (Fig. 1). The RgpA-Kgp complexes are also characterized by a 48-kDa Lys-x-specific proteinase (Kgp48, formerly PrtK48) associated with three sequence-related adhesins, Kgp39, Kgp15, and Kgp44, all encoded by kgp (3, 46, 47) (Fig. 1).

View larger version (42K): [in this window] [in a new window]   FIG. 1.   Schematic representation of the processing of the RgpA and Kgp polyproteins and RgpB. The white areas indicate the catalytic domains of the proteinases, the shaded areas indicate the adhesins, and the filled C-terminal areas show the conserved C-terminal sequence proposed to be involved in secretion and cell attachment.  marks the proposed outer membrane attachment to LPS. All processing sites are preceded by Arg or Lys residues (3, 46, 47).  shows the location of an adhesin-binding motif proposed to be involved in binding of the RgpA45 and Kgp48 catalytic domains into large noncovalent complexes with the adhesins and in autoaggregation of the adhesins (46).

We have previously characterized the extracellular Arg-x-specific cysteine proteinase encoded by rgpB of strain W50 (46). This proteinase, designated RgpB, is not associated with adhesins, as the rgpB gene does not encode adhesins or an adhesin binding motif (Fig. 1) that is present in the RgpA and Kgp catalytic domains (46). This adhesin binding motif is also present in some of the adhesin domains of RgpA and Kgp (Fig. 1) and is proposed to be responsible for the incorporation of the RgpA and Kgp catalytic domains into noncovalently associated complexes with adhesins and for the autoaggregation of the adhesins into large complexes (46). The RgpB proteinase has been isolated as a 70- to 80-kDa membrane-associated protein and as a discrete 50-kDa protein from the culture supernatant (40, 46).

Spontaneous P. gingivalis mutants with reduced Arg-x and Lys-x proteinase activity and wild-type cells treated with a protease inhibitor (N-p-tosyl-L-lysine chloromethyl ketone [TLCK]) have been reported to be avirulent in animal models (16). Furthermore, a nonpigmented mutant of P. gingivalis W50/BE1, which has reduced Lys-x and Arg-x proteinase activity, is reported to be avirulent in animal models (26). However, this mutant also lacks gelatinase, collagenase, and dipeptidylaminopeptidase and shows reduced hemagglutinin activity, fimbrination, and extracellular vesicle production (5, 26, 43). Tokuda et al. (50, 51) have reported that two constructed isogenic mutants of P. gingivalis strain 381 lacking either RgpA or RgpB exhibited reduced aggregation, hemagglutination, and binding to matrix proteins relative to the wild-type 381 strain. Further, a triple mutant based on strain ATCC 33277 lacking RgpA, RgpB, and Kgp was reported not to agglutinate erythrocytes, bind to hemoglobin, or grow in defined medium containing bovine serum albumin (BSA) as the sole carbon and energy source (44). These studies suggest that the rgpA, rgpB, and kgp genes are important for the virulence of P. gingivalis.

The aim of this study therefore was to determine the virulence of isogenic mutants of the invasive W50 strain that lack the RgpA, RgpB, and Kgp proteinases in a murine lesion model.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Bacterial strains and growth conditions. The bacteria and plasmids used in this study are listed in Table 1. Lyophilized cultures of P. gingivalis W50 and mutant strains (501, D7, and K1A) were grown anaerobically at 37°C on lysed horse blood agar plates supplemented with 5 µg of hemin, 0.5 µg of cysteine (HB agar) and appropriate antibiotics (1.0 µg of tetracycline and/or 10 µg of erythromycin) (<10 passages) per ml. After 3 to 4 days, colonies were used to inoculate brain heart infusion medium containing 5 µg of hemin, 0.5 µg of cysteine (25), and appropriate antibiotics (0.5 µg of tetracycline and/or 5 µg of erythromycin) per ml. Batch cultures were grown anaerobically in an MK3 Anaerobic Workstation (Don Whitley Scientific Ltd., Adelaide, Australia). Cells were harvested during exponential growth phase by centrifugation (5,000 × g, 30 min, 4°C) and washed twice with PG buffer (50 mM Tris-HCl, 150 mM NaCl, 5 mM CaCl₂, 5 mM cysteine-HCl, pH 8.0) in the anaerobic workstation for the whole-cell proteinase assays and the murine lesion model experiments. Growth of batch cultures was monitored at 650 nm using a spectrophotometer (model 295E; Perkin-Elmer). Culture purity was checked routinely by Gram stain, microscopic examination, and a variety of biochemical tests according to Slots (48).

Construction of pNJR12::rgpB-complemented RgpB mutant. A PCR-derived fragment containing the rgpB coding region and 5' untranslated region was generated from the GEM-12 P. gingivalis genomic clone (46) using Elongase (Life Technologies) according to the manufacturer's instructions on a PC-960 thermal sequencer (Corbett Research). The PCR was performed using the forward primer (5'-GCGCGCTCTAGAGGACAGTATCTGCAACCGTCG-3') that consists of a six base-buffer, an XbaI site, and bases 495 to 515 of rgpB (previously prtRII; GenBank accession no. AF007124) and the reverse primer (5'-CCGAATGGATTCCTCGGC-3') that consists of bases 3150 to 3167 of rgpB in an antisense orientation. The 2.7-kb PCR product was purified using the QIAquick PCR purification kit (Qiagen Pty Ltd.) and ligated into pGEM-T Easy (Promega Corporation), transformed into Escherichia coli JM109 (Promega Corporation), and selected on Luria-Bertani medium (LB) containing ampicillin (100 µg/ml) following standard procedures (42).

Arg-x-specific and Lys-x-specific whole-cell proteinase assays. P. gingivalis cells (W50 and mutants 501, D7, and K1A) were harvested under anaerobic conditions at early exponential, mid-late exponential, and stationary growth phases (0.5, 0.9, and 1.25 O.D.₆₅₀, respectively) by centrifugation (5,000 × g, 30 min, 4°C), washed, and resuspended in PG buffer (1 ml). Resuspended cells were analyzed immediately for Arg-x and Lys-x proteolytic activity using N--benzoyl-L-Arg-p-nitroanilide (Bz-Arg-pNA; Sigma) and benzyloxycarbonyl-L-Lys-p-nitroanilide (Bz-Lys-pNA; Sigma). These enzyme substrates were prepared as follows: 2 mM Bz-Arg-pNA or 2 mM Bz-Lys-pNA in 3 ml of isopropyl alcohol was subjected to sonication for 10 min, after which 7 ml of enzyme buffer (400 mM Tris-HCl, 100 mM NaCl, 20 mM cysteine, pH 8.0) was added. Amidolytic activity is expressed as units (micromoles of substrate converted per minute) at 37°C.

SDS-PAGE and Western blot analysis of P. gingivalis W50 and mutant cell sonicates. P. gingivalis strain W50 wild-type, mutants, and pNJR12::rgpB-complemented RgpB mutant were grown in batch culture and harvested at late exponential phase by centrifugation (5,000 × g, 20 min, 4°C). Cells were washed once with 50 ml of TC buffer (20 mM Tris-HCl, 5 mM CaC1₂) containing 150 mM NaCl, pH 7.4, and sonicated at 4°C as previously described (3). The harvesting of cells and sonication were also performed with 10 mM TLCK in the buffer, with fresh 10 mM TLCK being added at every step. The sonicates were recentrifuged at 4°C for 10 min, and the collected supernatants were stored at 70°C. The protein concentration of each cell sonicate was determined by using the Bradford protein assay (Bio-Rad, North Ryde, New South Wales, Australia) with BSA as the standard. Each cell sonicate (10 µg of protein) was subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) in gels of 12.5% (wt/vol) acrylamide (1 mm) by the method of Laemmli (18) with a minigel system (Bio-Rad, North Ryde, New South Wales, Australia). Proteins were electrophoretically transferred onto a polyvinylidene difluoride (PVDF) membrane as described previously (8). After sectioning of the membrane, one section was stained with 0.1% (wt/vol) CBB R250, and the remaining section was blocked for 1 h at 20°C with 5% (wt/vol) nonfat skim milk powder in TN buffer (50 mM Tris-HCl [pH 7.4], 100 mM NaCl) and incubated with proteinase-specific antibody (32) diluted 1:25 with TN buffer.

Murine lesion model. The murine lesion model experiments were approved by the University of Melbourne Ethics Committee for Animal Experimentation and were conducted essentially as described previously (32). BALB/c mice 6 to 8 weeks old (10 animals per group) were challenged with either 3.0 × 10⁹, 6.0 × 10⁹, or 1.2 × 10¹⁰ viable cells of P. gingivalis strain W50 and mutants 501, D7, and K1A by subcutaneous injection (100 µl) in the abdomen, and lesion size and mortality were monitored over 14 days as described previously (32). Mice were also challenged with 3.0 × 10⁹ and 6.0 × 10⁹ viable cells of the RgpB mutant D7 carrying the pNJR12 vector and the pNJR12::rgpB-complemented RgpB mutant. The P. gingivalis inocula were prepared using PG buffer in the anaerobic workstation as described above. The number of viable cells in each inoculum was verified by enumeration on HB agar. The maximum sizes of the lesions developed were statistically analyzed using the Kruskal-Wallis test and Mann-Whitney U-Wilcoxon rank sum test with a Bonferroni correction for type 1 error (30).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Arg-x and Lys-x whole-cell proteinase activity of RgpA, RgpB, and Kgp mutants of P. gingivalis W50. Whole-cell Arg-x and Lys-x proteinase activity was measured at the early exponential, mid-late exponential, and early stationary phases of growth for each of the isogenic mutants and wild-type strain (Fig. 2 and 3). The whole-cell Arg-x activity (in units per 10¹¹ cells) (Fig. 2) of the Kgp strain was found not to be significantly different at each growth phase measured compared with the wild-type strain. For both the Kgp mutant and wild-type strain, the Arg-x activity at the early exponential phase of growth was significantly (P < 0.05) less (28%) than that at later growth phases. There was a significant (P < 0.001) difference in the Arg-x activity of both the RgpA and RgpB mutants compared with the wild-type strain. The Arg-x-specific whole-cell activity of both these mutants was consistently lower (3- to 4-fold) than that of the wild-type strain at each growth phase measured. However, the Arg-x activities of the RgpA and RgpB mutants were not significantly different from each other and did not increase with growth.

View larger version (46K): [in this window] [in a new window]   FIG. 2.   Whole-cell Arg-x proteolytic activity of P. gingivalis W50 and RgpA, RgpB, and Kgp isogenic mutants and pNJR12::rgpB-complemented RgpB mutant at different phases of growth. P. gingivalis cells were harvested anaerobically at early (A) and late (B) exponential and stationary (C) growth phases. Harvested cells were centrifuged, resuspended in PG buffer, and assayed for Arg-x proteolytic activity. Data presented represent the mean ± standard deviation (SD) of three separate experiments. *, groups significantly different (P < 0.05) from activity at later phases of growth. **, groups significantly different (P < 0.001) from activity of P. gingivalis wild-type W50 strain.

View larger version (35K): [in this window] [in a new window]   FIG. 3.   Whole-cell Lys-x proteolytic activity of P. gingivalis W50 and RgpA, RgpB and Kgp isogenic mutants at different phases of growth. P. gingivalis cells were harvested anaerobically at early (A) and late (B) exponential and stationary (C) growth phases. Harvested cells were centrifuged, resuspended in PG buffer, and assayed for Lys-x proteolytic activity. Data presented represent the mean ± SD of three separate experiments. *, no whole-cell Lys-x proteolytic activity was detected for the Kgp mutant.

SDS-PAGE and Western blot analysis of RgpA, RgpB, and Kgp mutants of P. gingivalis W50. Cell sonicates of P. gingivalis W50 and RgpA, RgpB, and Kgp mutants were subjected to SDS-PAGE and Western blot analysis using the proteinase-specific antibody (Fig. 4). The Western blot of the wild-type W50 sonicate prepared in the absence of TLCK (Fig. 4A) revealed diffuse immunoreactive bands at 45 kDa, 48 kDa, and 70 to 90 kDa. The same analysis of the proteinase isogenic mutants showed that the 48-kDa band in Kgp, the 45-kDa band in RgpA and the 70- to 90-kDa diffuse band in RgpB were not detected, confirming the genotype of the three mutants and the previous assignment of these bands (3, 40). However, when the cell sonicates were prepared in the presence of 10 mM TLCK, the bands obtained on the Western blot were less diffuse, suggesting that TLCK was inhibiting proteolytic processing during preparation of the sample (Fig. 4B).

View larger version (61K): [in this window] [in a new window]   FIG. 4.   Western blot analysis of cell sonicates of isogenic mutants RgpA, RgpB, and Kgp, pNJR12::rgpB-complemented RgpB mutant, and wild-type strain using proteinase-specific sera. (A) Cell sonicate. (B) Cell sonicate prepared in the presence of 10 mM TLCK. The cell sonicates were subjected to SDS-PAGE and transferred onto a PVDF membrane. The membranes were probed with sera (1:25 in TN buffer) from mice immunized with a synthetic peptide which has the sequence HGSETAW that is common to each proteinase (32). Positions of molecular mass markers are indicated (in kilodaltons).

Characterization of pNJR12::rgpB-complemented RgpB mutant. A cell sonicate of the pNJR12::rgpB-complemented RgpB mutant was subjected to SDS-PAGE and Western blot analysis using the proteinase-specific antibodies (Fig. 4). This analysis showed that the pNJR12::rgpB-complemented mutant had immunoreactive bands corresponding to the RgpA (45 kDa) and Kgp (48 kDa) catalytic domains as well as the 72- and 80-kDa immunoreactive protein bands corresponding to the RgpB proteinase. The intensity of the 72- and 80-kDa immunoreactive bands was similar to the same immunoreactive bands for the wild-type W50 strain. The bands, however, were absent in the Western blot of the RgpB mutant D7 (Fig. 4).

Virulence of RgpA, RgpB, Kgp, and pNJR12::rgpB-complemented RgpB mutants in the murine lesion model. To evaluate the virulence of the isogenic mutants, BALB/c mice were challenged subcutaneously with the RgpA, RgpB and Kgp mutants as well as the wild-type W50 strain at three doses, 3.0 × 10⁹, 6.0 × 10⁹ and 1.2 × 10¹⁰ viable cells. Lesions that developed were measured, and the maximum lesion size for each strain is shown in Fig. 5 A, B, and C. At the 3.0 × 10⁹ viable-cell dose, only mice challenged with the wild-type W50 strain developed lesions. All the mice challenged with the wild-type W50 strain at the 6.0 × 10⁹ viable-cell dose died 3 days after the inoculation. Both the RgpA and RgpB mutants induced lesions in all of the mice at the 6.0 × 10⁹ viable-cell dose, and the lesion sizes were not significantly different from those induced by the wild-type W50 strain at the 3.0 × 10⁹ viable-cell dose. However, no lesions were recorded for mice challenged with the Kgp mutant at the 6.0 × 10⁹ viable-cell dose. The 1.2 × 10¹⁰ viable-cell dose induced 100% mortality in mice inoculated with the RgpA mutant and wild-type W50 strain. Challenge with the RgpB mutant at the 1.2 × 10¹⁰ viable-cell dose resulted in only 20% mortality and an average increase of 35% in the lesion size compared with the lesions induced by the same mutant at the 6.0 × 10⁹ viable-cell dose. In contrast, none of the mice challenged with the Kgp mutant at the 1.2 × 10¹⁰ viable-cell dose died, and only 40% of the animals developed lesions. The mean lesion size produced by the Kgp mutant at this dose was significantly (P < 0.001) smaller than that induced by the RgpB mutant at the same dose and by the wild-type W50 strain at the 3.0 × 10⁹ viable-cell dose.

View larger version (32K): [in this window] [in a new window]   FIG. 5.   Maximum lesion size of mice challenged with P. gingivalis W50, RgpA, RgpB, and Kgp isogenic mutants, and pNJR12::rgpB-complemented RgpB mutant. BALB/c mice were challenged subcutaneously with 3.0 × 109 (A), 6.0 × 109 (B), or 1.2 × 1010 (C) viable P. gingivalis W50 and isogenic mutant cells and with 3.0 × 109 (D) and 6.0 × 109 (E) viable P. gingivalis W50, RgpB isogenic mutant, and pNJR12::rgpB-complemented RgpB mutant cells. Animals were monitored over a 14-day period for mortality and lesion size. Data presented in panels A to E represent mean ± SD (n = 10) from separate experiments and were analyzed using the nonparametric Mann-Whitney U-Wilcoxon rank sum test. *, group significantly different (P < 0.001) from RgpB at same dose and from wild-type W50 strain at the 3.0 × 109 viable-cell dose. , 100% mortality. Nil, no lesions observed.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

In this study, the role of the Arg-x (RgpA and RgpB) and Lys-x (Kgp) proteinases in virulence of P. gingivalis was investigated in the murine lesion model using three isogenic mutants lacking the proteinases. Subjecting each of the isogenic mutants to SDS-PAGE and Western blot analysis with a proteinase-specific antibody showed that each of these proteinases was absent in the Western blot. The analysis of the form of the RgpA, RgpB, and Kgp proteinases of the wild type and the three isogenic mutants by SDS-PAGE and Western blot revealed that RgpB appeared as 72- and 80-kDa bands and the majority of RgpA and Kgp appeared as fully processed 45-kDa and 48-kDa catalytic domains, respectively. These results therefore are consistent with the previous characterization of these cell surface proteins (3, 46).

It has been proposed previously (3, 46) that the RgpA and Kgp polyproteins are secreted and then attached to the outer membrane, possibly through the conserved C-terminal segment (Fig. 1), as this segment is not found in the soluble RgpA-Kgp complexes released from the outer membrane (3, 36; unpublished data). As the N-terminal residue of the activated catalytic domains of RgpA, Kgp, and RgpB is a residue C-terminal to an Arg (3, 46), it is likely that the polyproteins are processed (activated) at these Arg residues to remove their profragments to produce mature 160-kDa (RgpA) and 163-kDa (Kgp) forms (Fig. 1). These mature RgpA and Kgp polyproteins must then undergo further processing at Arg and Lys residues (probably autolytic) to release the proteinase catalytic domains and several C-terminal sequence-related adhesin domains (Fig. 1) that have been characterized previously (3, 36, 47).

We have shown (46) that both the RgpA and Kgp catalytic domains, but not RgpB, contain a C-terminal adhesin-binding motif that is also found in the released adhesins (Fig. 1). Through this adhesin-binding motif, the proteinase catalytic domains may bind to their respective adhesins, which in turn may aggregate and bind to the putative anchored C-terminal adhesin, localizing RgpA and Kgp as noncovalently associated processed domains on the cell surface. The 50-kDa mature RgpB is presumably attached directly to the cell surface to form the membrane-associated 72- and 80-kDa forms (Fig. 1). This is supported by the work of Curtis et al. (7), who have shown that the 70- to 80-kDa membrane-associated form of RgpB, but not the 50-kDa isoform found in the culture supernatant, is immunoreactive with anti-LPS antibodies. In the presence of 10 mM TLCK, RgpB appeared as two major bands at 72 kDa and 80 kDa in the current study, which presumably represent two differently LPS-modified isoforms of the enzyme. The proposed C-terminal attachment of RgpB is also consistent with the work of Potempa et al. (39), who have shown that the 50-kDa isoform in the culture supernatant is C-terminally truncated by processing before the putative anchor sequence.

The processing of RgpA and Kgp, once secreted, to release the proteinase catalytic domains and the adhesin domains is thought to be autolytic, as the processing sites always involve Arg or Lys residues (Fig. 1). The slight change in molecular weight of the processed domains obtained in the Western blots of each of the isogenic mutants (Fig. 4) is consistent with a processing role for these enzymes. Furthermore, the 3- to 4-fold reduction in cell surface Arg-x-specific proteolytic activity upon inactivation of either rgpA or rgpB is also consistent with a processing role for both of these enzymes. It is interesting, however, that inactivating any one of the three proteinase genes did not abolish secretion and processing of the other two gene products. As multiple Arg and Lys residues exist between the processed domains of the polyproteins, it is very likely that any one of the proteinases can facilitate processing of the others.

The whole-cell Arg-x-specific proteolytic activity of both the Kgp mutant and W50 wild-type strain was significantly less at the early exponential phase of growth and increased to a plateau at the mid-exponential phase. This increase in activity was not observed for the Lys-x activity of the RgpA and RgpB mutants or the wild-type W50 strain. In fact, the Lys-x-specific proteolytic activity of the wild-type and RgpA and RgpB mutants was similar at all phases of growth. The unchanged Lys-x proteolytic activity for the RgpA and RgpB mutants is in contrast to the findings of Tokuda et al. (52), who have reported that inactivation of the RgpA gene in strain 381 resulted in downregulation of the kgp gene, as indicated by Northern blot analysis. However, whole-cell Arg-x and Lys-x proteolytic activities were not measured in that study. The data reported here suggest that Arg-x activity but not Lys-x activity may be influenced by growth phase or environmental factors. This is consistent with several earlier reports that have shown that environmental factors, such as hemin availability, pH, and temperature, can affect the Arg-x-specific proteolytic activity of P. gingivalis (22, 24, 35).

Inactivation of the rgpA, rgpB, and kgp genes, as well as reducing whole-cell proteolytic activity, also resulted in a measurable reduction in the pathogenicity of each mutant in the murine lesion model. No lesions were recorded for mice challenged with the 3.0 × 10⁹ viable-cell dose for any of the mutants, but increasing the challenge dose resulted in differences in virulence for each of the isogenic mutants. Interestingly, although the Arg-x activity of the RgpA and RgpB mutants was similar, there was a significant difference in virulence at the 1.2 × 10¹⁰ viable-cell dose, with 100% mortality of mice challenged with the RgpA mutant. The RgpB mutant, however, at the same dose induced only 20% mortality.

The rgpA gene in W50 produces a noncovalently associated proteinase-adhesin complex (3). The rgpB gene in W50 produces two isoforms of RgpB that are not associated with adhesins, 72- and 80-kDa membrane-attached forms (40;this study) and a 50-kDa discrete proteinase in the culture supernatant (40, 46). The adhesins of the RgpA proteinase-adhesin complex have been shown to bind to host substrates, facilitating proteolysis (37). This targeting role of the adhesins has been speculated to increase the virulence of the proteinase. However, the results of this study suggest that RgpB may have a greater role in virulence. Recently, Tokuda et al. (50, 51) constructed RgpA and RgpB isogenic mutants of strain 381 and reported marked changes in surface and binding properties of the two mutants relative to the wild type. The authors noted differences between the two mutants, with the RgpB mutant exhibiting a reduced ability to bind to epithelial cells. These and other studies (15, 28), showing pleiotropic effects of inactivation of the rgpA and rgpB genes, suggest that the Arg-x-specific proteolytic activity of P. gingivalis is involved in processing of not only the proteinases but also other surface proteins. Therefore, a greater role in pathogenicity for RgpB may reflect a greater role in processing for this enzyme.

The least virulent of the isogenic mutants in the murine lesion model was the Kgp mutant, as only 40% of mice challenged with the 1.2 × 10¹⁰ viable-cell dose developed lesions and these were significantly smaller than the lesions induced by the wild-type strain at the 3.0 × 10⁹ viable-cell dose. Kgp has been reported to be involved in hemoglobin binding and degradation and heme accumulation (17, 19, 34, 38). The nonpigmented phenotype of the Kgp mutant (K1A) is consistent with the proposed role of this proteinase in heme accumulation. Heme has been reported to be essential for the growth and virulence of P. gingivalis (23), and thus the Kgp mutant's reduced ability to accumulate heme may account for its reduced virulence in the murine lesion model.

In this study we also investigated the complementation of the RgpB mutant with a plasmid containing the rgpB gene. Whole-cell Arg-x-specific proteolytic activity was fully restored by the complementation. The pNJR12::rgpB-complemented RgpB mutant also displayed the same pattern of Arg-x activity/cell as the wild-type strain, with a lower activity at early exponential growth phase which increased in later phases of growth. Although pNJR12 is a low-copy-number plasmid, it may have been expected that the pNJR12::rgpB-complemented RgpB mutant would have exhibited higher Arg-x activity than the wild-type W50 strain due to the presence of multiple copies of the plasmid and therefore of the rgpB gene. As an increase in Arg-x-specific activity was not observed, this suggests that the expression of the rgpB gene is regulated, which is again consistent with earlier reports of Arg-x proteolytic activity regulation (22, 24, 35).

The characteristic 72- and 80-kDa RgpB bands were also observed for the complemented RgpB mutant, indicating that the plasmid-expressed RgpB was secreted and attached to the outer membrane in the same manner as the protein in the wild-type W50 strain. As well as restoring the enzymatic and protein profile, the pNJR12::rgpB complementation fully restored pathogenicity in the murine lesion model. Using techniques similar to those described here for mutant complementation, others have shown the importance of specific genes and their products for the virulence of Listeria monocytogenes, Proteus mirabilis, Vibrio anguillarum, and Yersinia enterocolitica (2, 27, 41, 55).

In conclusion, by characterizing the virulence of P. gingivalis isogenic mutants 501, D7, and K1A, lacking RgpA, RgpB, and Kgp, respectively, and the pNJR12::rgpB-complemented RgpB mutant in the murine lesion model, we have shown that the three proteinases all contributed to virulence in this model and that the order of contribution was Kgp RgpB  RgpA.