Porphyromonas gingivalis-derived RgpA-Kgp Complex Activates the Macrophage Urokinase Plasminogen Activator System

Background: We recently found that uPA−/− mice are resistant to experimental periodontitis following oral infection with P. gingivalis. Results: P. gingivalis-derived RgpA-Kgp complex activates the macrophage urokinase plasminogen activator. Conclusion: P. gingivalis activates a critical host proteolytic pathway to promote tissue destruction. Significance: A new host-pathogen interaction may promote tissues destruction and pathogen virulence in periodontitis. Urokinase plasminogen activator (uPA) converts plasminogen to plasmin, resulting in a proteolytic cascade that has been implicated in tissue destruction during inflammation. Periodontitis is a highly prevalent chronic inflammatory disease characterized by destruction of the tissue and bone that support the teeth. We demonstrate that stimulation of macrophages with the arginine- and lysine-specific cysteine protease complex (RgpA-Kgp complex), produced by the keystone pathogen Porphyromonas gingivalis, dramatically increased their ability to degrade matrix in a uPA-dependent manner. We show that the RgpA-Kgp complex cleaves the inactive zymogens, pro-uPA (at consensus sites Lys158-Ile159 and Lys135-Lys136) and plasminogen, yielding active uPA and plasmin, respectively. These findings are consistent with activation of the uPA proteolytic cascade by P. gingivalis being required for the pathogen to induce alveolar bone loss in a model of periodontitis and reveal a new host-pathogen interaction in which P. gingivalis activates a critical host proteolytic pathway to promote tissue destruction and pathogen virulence.

Major virulence factors produced by P. gingivalis are the trypsin-like cysteine proteinases called gingipains (22). The gingipains comprise the arginine-specific (RgpA and RgpB) and the lysine-specific (Kgp) proteinases based on their ability to cleave Arg-Xaa or Lys-Xaa peptide bonds, respectively (22). RgpA and Kgp form an extracellular cysteine protease complex (RgpA-Kgp complex) common to all characterized P. gingivalis strains and is a major factor contributing to its pathogenicity (23,24). The proteolytic activity of the gingipains enables P. gingivalis to degrade a range of host proteins including cytokines (e.g. IL-4, IL-12, and IFN␥), cell surface receptors (e.g. CD2, CD4, and CD14) and matrix proteins (e.g. fibrin, fibrinogen) (25). The peptides produced from such breakdown provide P. gingivalis with the nutrients it requires for growth and survival (18). Not only do the gingipains degrade host proteins, they also activate host systems, most notably the complement cascade, which enables P. gingivalis to subvert the host immune response and drive PD (20,21).
uPA has been detected in inflamed periodontal tissue (26) and gingival crevicular fluid (27). Additionally, polymorphisms in the genes for uPA and one of its inhibitors have been associated with periodontal bone loss (28). However, the role of uPA in PD has not been defined. We recently found that uPA Ϫ/Ϫ mice are resistant to experimental PD following oral infection with P. gingivalis, possibly via a macrophage-dependent mechanism(s) (29). In agreement with that study, we also showed that depletion of macrophages, which can be a major source of uPA in inflammation (30), protected mice from P. gingivalisinduced PD (31) and that uPA regulates macrophage-mediated matrix degradation (8), a hallmark of chronic PD (16,17). Therefore, in this study, we investigated the interaction of P. gingivalis-derived RgpA-Kgp complex with the uPA/plasminogen pathway in macrophages.

Experimental Procedures
Mice-Female C57BL/6 mice (6 -8 weeks) were from Monash University (Malvern East, Australia) (32). The uPA gene-deficient mice (uPA Ϫ/Ϫ ) from Dr. P. Carmeliet (University of Leuven, Leuven, Belgium) were backcrossed onto the C57BL/6 background for 11 generations. The protocols used in this document were approved by the University of Melbourne Ethics Committee for Animal Experimentation (approval number 081049) and were conducted in accordance with the Declaration of Helsinki. All human blood donors gave written informed consent.
Bacterial Strains, Growth Conditions, and Purification of the RgpA-Kgp Complexes-P. gingivalis strain W50 (ATCC 53978) was grown and harvested as before (34). The purification of the RgpA-Kgp complexes was performed as before (35) and were activated prior to use with 10 mM L-cysteine in 0.5 M Tris/HCl, pH 7.4 buffer.
Preparation of Mouse Bone Marrow-derived Macrophages (BMM) and Human Monocyte-derived Macrophages (MDM)-Murine BMM and human MDM were prepared as before (36). Bone marrow cells from the femurs of mice were cultured in RPMI 1640 medium with 10% FCS, 2 mM GlutaMAX-1, 100 units/ml penicillin, 100 g/ml streptomycin, and M-CSF (2000 units/ml). On day 7, adherent cells were harvested. Human monocytes were purified from buffy coats (Red Cross Blood Bank, Melbourne, Australia) and MDM were generated from M-CSF-treated cultures (36).
Matrix Degradation Assay-Macrophage matrix degradation was determined as before (8). Briefly, macrophages (3 ϫ 10 4 ) were cultured on FITC-coupled gelatin for 24 -48 h in the presence of M-CSF (2000 units/ml), plasminogen (500 nM), RgpA-Kgp (5-50 nM), anti-uPA mAb (267 nM), and isotype control mAb (267 nM) as indicated. Cells were processed for F-actin and DAPI staining and visualized by fluorescent microscopy (Zeiss Axioskop 2, at ϫ40 magnification). Images were captured by a Zeiss AxioCam MRm and quantification of matrix degradation by calculating the area of gelatin degradation per total cell area (ImageJ, version 1.46) (8). Six random fields were analyzed for each group, and the assays were performed in four independent experiments.
Quantitative PCR-Total RNA was extracted using RNeasy kits (Qiagen) and reverse transcribed using SuperScript III (Invitrogen). Quantitative PCR was performed using an ABI PRISM 7900HT sequence detection system (Applied Biosystems, Carlsbad, CA) and TaqMan probe/primer combinations for human and murine uPA, uPAR, and Ubiquitin C (Applied Biosystems) (32). All samples were performed in triplicate, and results are expressed relative to UBC from five independent experiments.
Analysis of Pro-uPA Activation by Western Blot and SDS-PAGE-For Western blotting, human pro-uPA (250 nM) was incubated with plasmin (10 nM) or RgpA-Kgp complex (10 nM) in reaction buffer (0.5 M Tris/HCl, pH 7.4) at 37°C for up to 60 min. Reactions were terminated by 5ϫ SDS sample buffer and heating to 95°C for 10 min. Samples were analyzed under reducing conditions by SDS-PAGE (Invitrogen), and uPA was detected with an anti-human uPA mAb (clone U-16) by Western blotting. Alternatively, human pro-uPA (4.5 M) was incubated with plasmin (50 nM) or RgpA-Kgp complex (50 nM) at 37°C for 10 min. Reactions were terminated by 5ϫ SDS and heated to 95°C for 10 min. Samples were analyzed by SDS-PAGE with proteins bands visualized with EzBlue gel staining reagent (Sigma).
Cleavage Site Determination by Sulfo-NHS-acetate Labeling and Tandem Mass Spectrometry-RgpA-Kgp complex (50 nM) was incubated for 10 min with human pro-uPA (4.5 M) at 37°C in PBS. Samples were labeled as before (37). Briefly, sam-ples were precipitated at 4°C for 30 min and air-dried at room temperature. The pellets were resuspended in 0.2% SDS and 100 mM NaHCO 3 , pH 8.5, and sonicated for 20 min, labeled with 10 mM sulfo-NHS-acetate for 60 min, and quenched with 10 mM of Tris/HCl, pH 8.0. Denatured samples were run on SDS-PAGE with trypsin in-gel digestion performed after reduction with DTT and alkylation with iodoacetamide (37). Tryptic digests were analyzed by LC-MS/MS using a Dionex Ultimate 3000 system (Thermo Scientific) coupled to an HCTultra ion trap mass spectrometer (37). Proteins were identified by MS/MS Ion Search using Mascot v 2.2 (Matrix Science) against the SwissProt database (June 2014) with taxonomy restricted to Homo sapiens. Search parameters were: enzyme ϭ trypsin (with results of semi-trypsin search also checked); missed cleavages ϭ 3; fixed modifications ϭ carbamidomethyl (Cys); optional modifications ϭ acetyl (Lys), acetyl (N terminus), and oxidation (Met); MS tolerance ϭ 1.5 Da; and MS/MS tolerance ϭ 0.5 Da.
Statistical Analysis-Statistical comparisons between groups were performed using unpaired Student's t test (GraphPad Prism 4 software). p values Յ0.05 indicate significance. The data were plotted using GraphPad Prism 4.03 software. The data for alveolar bone resorption (mm2) were analyzed using one-way analysis of variance, with a Dunnett's T post hoc test.

RgpA-Kgp Complex Potentiates Mouse Macrophage Matrix
Degradation in a uPA-dependent Manner-Oral inoculation of uPA Ϫ/Ϫ mice with P. gingivalis failed to induce significant alveolar bone loss (p Ͻ 0.001; Fig. 1A) in contrast to wild-type mice, which developed significant bone loss relative to uninoculated mice. Also uPA Ϫ/Ϫ mice had lower levels of P. gingivalis in the subgingival plaque compared with wild-type mice (data not shown), suggesting that uPA is necessary for the emergence of the pathogen (29). We have previously demonstrated that macrophage uPA conversion of plasminogen to active plasmin plays a major role in its ability to degrade matrix (8) and that depletion of macrophages protects mice from P. gingivalis-induced PD (31). Given that PD is characterized by the irreversible destruction of periodontal tissue (17), we speculated that the major virulence factor from P. gingivalis, the RgpA-Kgp complex, might enhance the ability of macrophages to degrade matrix. To address this hypothesis and to begin to explore potential mechanistic explanations for the protection of uPA Ϫ/Ϫ mice in experimental PD, murine BMM were seeded on FITC-gelatin-coated coverslips in the presence of plasminogen and/or RgpA-Kgp complex and matrix degradation quantified by fluorescent microscopy.
As shown in Fig. 1B, BMM had a low capacity to degrade matrix, which was significantly increased by the addition of plasminogen (p Ͻ 0.05; Fig. 1C for quantification). Strikingly, this plasminogen-induced matrix breakdown could be further enhanced by the addition of the RgpA-Kgp complex with a 4 -5-fold increase in FITC-gelatin degradation by macrophages co-treated with RgpA-Kgp complex and plasminogen versus those treated with plasminogen alone (p Ͻ 0.05; Fig. 1, B and C). A dose response was observed where increasing concentrations of RgpA-Kgp complex enhanced BMM FITC-gelatin degradation (p Ͻ 0.05; Fig. 2, A and B for quantification). Treatment of BMM with the RgpA-Kgp complex in the absence of plasminogen did not enhance matrix breakdown (Fig. 1, B and C). It was also observed that in the absence of macrophages, the RgpA-Kgp complex either in the absence or presence of plasminogen did not degrade the FITC-gelatin matrix (data not shown). These findings show that the gingipain itself did not degrade the matrix and that macrophages are required for matrix degradation. Importantly, we found that treatment of BMM with the anti-uPA mAb, mU1, which blocks the function of murine uPA in vivo (33), abrogated the ability of the RgpA-Kgp complex and plasminogen combination to potentiate macrophagemediated matrix degradation (p Ͻ 0.05; Fig. 1, B and C). One possible mechanism for the enhancement of uPA-dependent matrix breakdown is increased uPA and uPAR gene expression in macrophages treated with RgpA-Kgp complex. However, when BMM were treated with the RgpA-Kgp complex, we did not observe any significant changes in uPA or uPAR gene expression relative to untreated cells (Fig. 1D). These data demonstrate that P. gingivalis-derived RgpA-Kgp complex potentiates macrophage matrix breakdown in a uPA-dependent manner, which could not be explained by the ability of the complex to modulate uPA and uPAR expression.
RgpA-Kgp Complex Potentiates Human Macrophage Matrix Degradation in a uPA-dependent Manner-We next addressed whether RgpA-Kgp complex could potentiate human MDM matrix degradation. Similar to BMM (Fig. 1B), MDM matrix degradation was significantly enhanced by the addition of plasminogen (p Ͻ 0.05; Fig. 3, A and B for quantification) (8). Again, we found a striking increase in matrix breakdown (an approximately 3-4-fold increase in FITC-gelatin degradation) by MDM co-treated with the RgpA-Kgp complex and plasminogen versus those treated with plasminogen alone. Neutralizing anti-human uPA mAb (U-16) potently blocked MDM matrix degradation induced by plasminogen and RgpA-Kgp complex (p Ͻ 0.05; Fig. 3, A and B). In MDM, as in BMM (Fig. 1D), the RgpA-Kgp complex did not regulate uPA or uPAR gene expression (Fig. 3C), further substantiating that the RgpA-Kgp com-plex potentiation of macrophage degradation is not caused by increased gene expression of uPA or its receptor.
RgpA-Kgp Complex Cleaves Latent Pro-uPA-Given that pro-uPA requires proteolytic cleavage at a lysine consensus site to become active (11), we hypothesized that the intrinsic lysine activity of the Kgp gingipain may allow it to cleave and activate pro-uPA, potentially explaining the ability of the RgpA-Kgp  complex to enhance uPA-dependent macrophage-mediated matrix degradation (Figs. 1B and 3A). To test this, cleavage of human pro-uPA by the RgpA-Kgp complex was compared with cleavage by the major physiologic regulator of uPA, plasmin. Plasmin cleavage of single-chain pro-uPA can occur at two major sites (12): proteolysis at the primary site (Lys 158 -Ile 159 ) generates the A-chain (20 kDa) and B-chain (30 kDa), whereas processing at the secondary site (Lys 135 -Lys 136 ) yields LMW-uPA (33 kDa) and the N-terminal fragment (18 kDa).
Accordingly, when human pro-uPA was incubated with plasmin for 10 min and the resulting fragments were separated by SDS-PAGE, four major fragments were generated corresponding to LMW-uPA (band I, 33 kDa), B-chain (band II, 30 kDa), A-chain (band III, 20 kDa), and N-terminal fragment (band IV, 18 kDa) (Fig. 4A, left lane). Incubation of RgpA-Kgp complex with pro-uPA for 10 min yielded identically sized fragments at 33 kDa (band I) and 30 kDa (band II), consistent with the generation of LMW-uPA and B-chain, respectively (Fig. 4A, lane  2). Identical fragments at 20 kDa (Fig. 4A, band III) and 18 kDa (Fig. 4A, band IV) were also generated by the RgpA-Kgp complex but were much weaker than those generated by plasmin.
RgpA-Kgp complex cleavage of pro-uPA also generated a smaller band of ϳ15 kDa (Fig. 4A, band V) that was not generated by plasmin.
We next analyzed the kinetics of pro-uPA cleavage by RgpA-Kgp complex versus plasmin by Western blot using an anti-uPA mAb targeting the protease domain. Plasmin rapidly and completely cleaved single-chain pro-uPA (labeled P, 55 kDa) to generate a protease domain-containing band even after 1 min (Fig.  4B). This is consistent with the complete cleavage of pro-uPA by plasmin seen by SDS-PAGE in Fig. 4A. In comparison, RgpA-Kgp cleavage of pro-uPA was slower with a protease domain-containing band not detected until 30 min (Fig. 4B); also the cleavage of single-chain pro-uPA was incomplete, with intact pro-uPA still present at 60 min. The bands generated by RgpA-Kgp complex versus plasmin cleavage of pro-uPA were identical in size (ϳ30 kDa) and were similar in size to the protease domain-containing LMW-uPA (band I, ϳ33 kDa) and B-chain (band II, ϳ30 kDa) observed in Fig. 4A; however, these two distinct bands could not be clearly identified by Western blot. Together, these data show that the RgpA-Kgp complex cleavage of pro-uPA generates fragments similar to those generated by its physiologic activator, plasmin, but with slower kinetics.

RgpA-Kgp Complex Cleaves Pro-uPA at the Major Consensus
Sites-To confirm that the RgpA-Kgp complex cleaves human pro-uPA at the same sites as plasmin (Lys 158 -Ile 159 and Lys 135 -Lys 136 ) (11,13), the RgpA-Kgp-cleaved pro-uPA fragments (Fig. 4A, lane 2) were labeled and identified by mass spectrometry. Specifically, the N terminus of each pro-uPA fragment was acetylated with sulfo-NHS-acetate, which acetylates free primary amines on the side chain of Lys residues and the N terminus of polypeptides. The labeled fragments were then separated by SDS-PAGE (Fig. 4A, lane 2), digested with trypsin, and identified by LC-MS/MS.
Peptides identified from the 33-kDa band (Fig. 4A, band I) indicated that this fragment included amino acids from Lys 136 to the C terminus of the protein at Leu 411 , conforming to the sequence of LMW-uPA (Fig. 4C, peptides identified are shown in bold and underlined). Importantly, the most N-terminal peptide identified ( 136 KPSSPPEELK 145 ) (Fig. 4C, arrowhead) was doubly acetylated at the N-terminal K residue, indicating that both the Lys side chain and the N terminus were acetylated. The N-terminal acetylation shows that this amino group was free at the time of labeling thereby, confirming that the RgpA-Kgp complex cleaved pro-uPA at the consensus site Lys 135 -Lys 136 (13). The MS/MS data for this peptide ( 136 KPSSPPEELK 145 ) did not give a significant Mascot score because of the highly favored fragmentation of the three Xaa-Pro bonds in the sequence (38); however, the fragmentation pattern was consistent with the sequence (Fig. 4D).
Peptides identified from the 30-kDa band (Fig. 4A, band II) included amino acids from Ile 159 to the C terminus of the protein at Leu 411 , conforming to the sequence of the B-chain of uPA (Fig. 4E, peptides identified are shown in bold and underlined). The N-terminal peptide ( 159 IIG . . . IYR 178 ) was acetylated at the N terminus and gave a rich fragmentation pattern yielding a high Mascot score of 94 (Fig. 4F), confirming that the RgpA-Kgp complex, like plasmin, cleaves the Lys 158 -Ile 159 bond of pro-uPA (Fig. 4F, arrowhead) (11). The analysis of the 20-and 18-kDa bands (Fig. 4A, bands III and IV) proved inconclusive and did not allow us to identify these bands as the A-chain (band III) or the N-terminal fragment (band IV) of uPA, respectively. However, these data confirm for the first time that the P. gingivalis-derived RgpA-Kgp complex cleaves pro-uPA at the common activation sites (Lys 158 -Ile 159 and Lys 135 -Lys 136 ) (11,13) to generate the protease domain-containing LMW-uPA and B-chain (Fig. 4G).

RgpA-Kgp Complex Activates
Pro-uPA-Having found that the RgpA-Kgp complex cleaves human pro-uPA at the major consensus sites, we next assessed the enzymatic function of the pro-uPA cleavage products generated by the RgpA-Kgp complex using an enzymatic assay. Human pro-uPA was incubated with RgpA-Kgp complex or plasmin in the presence of the uPAspecific chromogenic substrate (S2444) (33), and the resultant uPA activity measured over 90 min (Fig. 5A). Plasmin rapidly induced uPA activity with maximum activation reached at 25-30 min. In comparison, the RgpA-Kgp complex initially led to a slower rate of uPA activity, which is consistent with the kinetics for the generation, by the gingipain complex, of the protease domain-containing band by Western blot (Fig. 4B). Ultimately, uPA activity induced by the RgpA-Kgp complex, reached a comparable peak to that induced by plasmin at 90 min (Fig. 5A). Importantly, the RgpA-Kgp complex did not  directly cleave the uPA-specific chromogenic substrate (Fig.  5A). The specific plasmin inhibitor, ␣2-antiplasmin, blocked plasmin-induced activation of pro-uPA, whereas the uPA inhibitor, PAI-1, completely blocked uPA activity (generated by plasmin activation of pro-uPA) (Fig. 5A). In comparison, ␣2-antiplasmin had no effect on the ability of the RgpA-Kgp complex to activate pro-uPA, whereas PAI-1 blocked the uPA activity generated by RgpA-Kgp complex activation of pro-uPA (Fig. 5A).
When pro-uPA was incubated with different concentrations of the RgpA-Kgp complex or plasmin and uPA activity measured at 60 min (Fig. 5B), we found that low concentrations (i.e. a molar ratio of 1:500 to 1:50 of RgpA-Kgp to pro-uPA) of the RgpA-Kgp complex was inefficient at activating pro-uPA compared with plasmin. For example, at a molar ratio of 1:50, the RgpA-Kgp complex induced ϳ15% of the plasmin-induced uPA activity. At higher concentrations (i.e. a molar ratio of 1:25 to 1:2 of RgpA-Kgp to pro-uPA), the RgpA-Kgp complex was much more efficient at activating pro-uPA with ϳ70% of the plasmin-induced uPA activity at a molar ratio of 1:2 (Fig. 5B). These results demonstrate that cleavage of pro-uPA by the RgpA-Kgp complex produces enzymatically active uPA, which is consistent with its cleavage at the major consensus site within pro-uPA (Fig. 4G).
RgpA-Kgp Complex Cleaves and Activates Plasminogen-The major substrate of uPA is plasminogen, which is a zymogen that requires proteolytic cleavage to form the active serine protease, plasmin. This cleavage occurs at the Arg 561 -Val 562 peptide bond to yield the active two-chain protein, which is comprised of a heavy chain (ϳ63 kDa) and a protease domaincontaining light chain (ϳ25 kDa) that are linked by a disulfide bond (11). Given the intrinsic arginine activity of the RgpA gingipain, we next sought to determine whether the RgpA-Kgp complex could also cleave and activate plasminogen.
To begin to address this question, we used an enzymatic assay in which plasminogen is incubated with uPA or the RgpA-Kgp complex in the presence of the plasmin-specific chromogenic substrate (S2251) (33), and the resulting plasmin activity is then measured over 180 min. Fig. 6A shows that the activation of plasminogen by uPA reached maximum levels at ϳ60 min. The RgpA-Kgp complex induced plasminogen activation but with delayed kinetics relative to uPA (Fig. 6A). Early on (0 -30 min), the RgpA-Kgp complex only weakly induced plasmin activity, but at 180 min the plasmin activity induced by the RgpA-Kgp complex was ϳ75% of the uPA-induced plasmin activity. Neither the RgpA-Kgp complex nor uPA directly cleaved the plasmin-specific chromogenic substrate (Fig. 6A).
The abilities of uPA and RgpA-Kgp complex to activate plasminogen were further compared by incubating plasminogen with different concentrations of uPA or the RgpA-Kgp complex and measuring plasmin activity at 120 min (Fig. 6B). At low concentrations (i.e. a molar ratio of 1:500 to 1:250 of RgpA-Kgp to plasminogen), the RgpA-Kgp complex was inefficient at activating plasminogen compared with uPA. For example, at a molar ratio of 1:250, the RgpA-Kgp complex induced ϳ17% of the uPA-induced plasmin activity (Fig. 6B). At higher concentrations (i.e. a molar ratio of 1:10 to 1:2 of RgpA-Kgp to plasminogen), the RgpA-Kgp complex was much more efficient at activating plasminogen with ϳ67% of the uPA-induced plasmin activity at a molar ratio of 1:2 (Fig. 6B). These findings demonstrate that not only can the RgpA-Kgp complex activate pro-uPA to uPA, but it can also activate plasminogen to plasmin.
We have previously shown that unlike wild-type cells, uPA Ϫ/Ϫ macrophages do not respond to exogenous plasminogen for increased matrix degradation (8). Given we show that the RgpA-Kgp complex has intrinsic uPA-like activity in being able to activate plasminogen to plasmin, we reasoned that treatment of uPA Ϫ/Ϫ macrophages with the RgpA-Kgp complex might "restore" their ability to degrade matrix in the presence of plasminogen. To test this, FITC-gelatin matrix degradation by uPA Ϫ/Ϫ BMM was determined. As before (8), uPA Ϫ/Ϫ BMM treated with plasminogen did not degrade FITC-gelatin (Fig.  6C). However, when they were co-treated with the RgpA-Kgp complex and plasminogen, a significant increase in FITC-gelatin degradation was observed (p Ͻ 0.05; Fig. 6, C and D, for quantification). Indeed, the levels of matrix degradation by wild-type versus uPA Ϫ/Ϫ BMM co-treated with RgpA-Kgp complex and plasminogen were similar (compare Figs. 1B and 6C). These data suggest that in the absence of uPA, the intrinsic arginine activity of the RgpA gingipain can activate plasminogen to plasmin and restore macrophage matrix degradation.

Discussion
PD is characterized by a high rate of matrix turnover that depends on the regulated expression of proteolytic enzymes (39). P. gingivalis gingipains are powerful proteolytic enzymes that can degrade host proteins but can also activate host pathways. This unique ability allows P. gingivalis to generate nutrients in the form of tissue breakdown products and to subvert the host immune response, leading to periodontal inflammation and dysbiosis (21,22). uPA has been identified in inflamed periodontal tissue (26) and in the gingival crevicular fluid (27). Our findings here provide evidence of a new host pathway that can be activated by P. gingivalis gingipains, namely the uPA/ plasminogen pathway. Specifically, we find that the RgpA-Kgp complex has intrinsic plasmin-like and uPA-like enzymatic activities in being able to activate host zymogens, pro-uPA, and plasminogen, respectively. In the case of pro-uPA, the RgpA-Kgp complex cleaved at the major consensus sites (Lys 158 -Ile 159 and Lys 135 -Lys 136 ) (11,13) to yield enzymatically active uPA. The RgpA-Kgp complex also activated plasminogen, which is present in micromolar concentrations in extracellular fluids (40) to yield active plasmin.
Much of the damage to the periodontal tissue in PD is now recognized to be caused by the destructive host inflammatory response (41). We have shown that macrophages (31) and uPA (29) are important in the P. gingivalis-induced PD model. Interestingly, protection of uPA Ϫ/Ϫ mice in the P. gingivalis-induced PD model was associated with reduced numbers of macrophages in the gingival tissue (29), consistent with our data showing a critical role for uPA in promoting macrophage adhesion and motility (8). Indeed, we have speculated (8) that uPA may play an important role in controlling macrophage retention within certain inflammatory lesions, which may help to explain the benefit of its targeting in inflammatory disease (7,9,10,42); our prior data showing a uPA dependence for three systemic arthritis models (9,10) suggest that elevated uPA activity could be a common feature of PD and rheumatoid arthritis, perhaps helping to explain their association (43). The ability of the RgpA-Kgp complex to amplify the uPA/plasminogen-mediated matrix degradation by macrophages provides a potential mechanism for enhanced matrix turnover in PD. Because macrophages express high levels of uPAR and various receptors for plasminogen (e.g. annexin A2 and Plg-R KT ) (44,45), thereby localizing the activation of pro-uPA and plasminogen to the cell surface (4), it is likely that the gingipain complex is acting in close proximity to the cell surface uPAR-uPA complex. Indeed, it is known that Kgp binds vitronectin, a component of the extracellular matrix and a ligand for uPAR (46,47). A functional consequence of the focal proteolysis mediated by the uPAR-uPA complex is facilitation of cell migration through complex matrices (4,15). Whether RgpA-Kgp complex activation of the macrophage uPA/plasminogen pathway regulates cell motility is currently unknown. Its effects on other cell types that express uPA (e.g. neutrophils and fibroblasts) (4) are also unknown.
Activation of pro-uPA (to uPA) by plasmin and plasminogen (to plasmin) by uPA occur at lysine and arginine residues, respectively (11,12). Therefore, it would not be unreasonable to propose that pro-uPA is a substrate for lysine-specific Kgp, whereas plasminogen is a substrate for arginine-specific RgpA. The dual induction of uPA-like and plasmin-like activity by the RgpA-Kgp complex was observed in the context of macrophage matrix degradation. In this system, the intrinsic plasminlike enzymatic activity (i.e. activation of pro-uPA to uPA) of the gingipain complex was evidenced by the potent blockade of matrix degradation by anti-uPA mAbs. In turn, the intrinsic uPA-like enzymatic activity (i.e. activation of plasminogen to plasmin) of the complex was evidenced by its ability to restore the plasminogen-mediated matrix degradation by uPA Ϫ/Ϫ macrophages.
Given this dual induction of uPA and plasmin activity by the gingipain complex, it was expected that uPA blockade (via anti-uPA mAbs) would only partially block macrophage matrix degradation in the presence of plasminogen. However, we were surprised to find an almost complete abrogation of macrophage matrix degradation, when cells were treated with a combination of RgpA-Kgp complex and plasminogen in the presence of the blocking uPA mAbs. The activation of pro-uPA and plas-minogen are greatly enhanced when both are bound to the cell surface (15,48). On monocytes-macrophages, the plasminogen receptors Plg-R KT (45,49) and annexin A2 (50) are co-localized with uPAR (and presumably uPA), with this proximity being central to the reciprocal cell surface activation of uPA and plasmin (45). For the RgpA-Kgp complex to activate pro-uPA to uPA and/or plasminogen to plasmin, it would presumably have to be in very close proximity to the cell-surface uPAR-uPA and plasminogen "complex." It is possible that the neutralizing anti-uPA mAbs (mU1 and U-16), which block the interaction with plasminogen at the active site by steric hindrance (33,51), may also sterically hinder the large (ϳ250 kDa) gingipain complex from the plasminogen active site. This is currently under investigation because it may explain the nearly complete abrogation of macrophage matrix degradation by anti-uPA mAbs when cells were co-treated with RgpA-Kgp complex and plasminogen.
Cleavage of either pro-uPA or plasminogen by the RgpA-Kgp complex leads ultimately to the generation of the broad spectrum serine protease, plasmin: pro-uPA is most likely to be cellular-derived, whereas plasminogen would be readily available because it circulates at very high levels (14). Plasmin can directly activate components of the complement cascade, such as C3 and C5 (52), and we have shown that uPA is upstream of C5a activation in peritonitis model and is required for optimal C5a generation (10). Activation of the complement cascade has been invoked as a key aspect of the role of P. gingivalis as a keystone pathogen (20). Similar to the findings in the C5aR Ϫ/Ϫ mice, which are resistant to P. gingivalis colonization and are thus protected from alveolar bone loss (53), we recently found that P. gingivalis poorly colonized the subgingival plaque in uPA Ϫ/Ϫ mice compared with wild-type mice (29). This likely explains the protection of uPA Ϫ/Ϫ mice from P. gingivalis-induced alveolar bone loss but also suggests that the pathogen depends on uPA to colonize the host and drive inflammation. More specifically, it may suggest the involvement of Kgp-mediated activation of pro-uPA to uPA in P. gingivalis colonization. In this context, we have found that of the three proteinases, Kgp is the major contributor to P. gingivalis virulence in the murine PD model used here (54). Apart from activating the extracellular uPA proteolytic cascade, the generation of uPA at the cell surface by the gingipain complex may also promote uPA-uPAR signaling, which is independent of uPA proteolytic activity (55), and can regulate cell adhesion, survival and motility (4) (Fig. 7). Interestingly, in the context of our observation that uPA Ϫ/Ϫ mice were protected from P. gingivalis-induced alveolar bone loss (29), a recent study found that uPA-uPAR signaling promoted osteoclastogenesis (56). Whether RgpA-Kgp complex activation of the host uPA/plasminogen pathway occurs in PD as discussed above (Fig. 7) and whether this pathway is linked to the complement-driven dysbiosis and inflammation involved in PD progression are unknown.
Similar to its ability to activate the complement cascade (21), the activation of pro-uPA and plasminogen would likely benefit the pathogen in several important ways. P. gingivalis survival and growth is dependent on the peptides that are released during tissue degradation (18). The ability to activate the destructive host uPA/plasminogen pathway leading to heightened plasmin activity could confer a significant survival advantage, allowing it to generate nutrients locally and enable it to survive deep in the periodontal pocket (25) (Fig. 7). Indeed, studies have shown that the highest cell numbers of P. gingivalis are found in the deepest periodontal pockets (57). Like many other pathogens, P. gingivalis has a strict requirement for hemin (18), and the well documented fibrinolytic activity of plasmin would be expected to promote bleeding and enhance its ability to acquire hemin. The ability to access hemin is directly linked to P. gingivalis virulence (Fig. 7) (58). Plasmin can degrade both the noncollagenous components (59) via activation of latent metalloproteinases and the collagenous components (1) of gingival tissue and bone matrix, which may also help to explain the protection of uPA Ϫ/Ϫ mice from P. gingivalis-induced bone loss (Fig. 7) (29). It should be noted that there are other oral bacterial species that possess enzymes with trypsin-like activity. For example, Treponema denticola has a chymotrypsin-like proteinase (60,61). Despite its expression of chymotrypsin-like proteinase, we found that inoculation of mice with T. denticola alone failed to induce alveolar bone loss in a murine PD model (62). Relatedly, the discovery that apparently innocuous oral commensals were capable of inducing plasmin activity (via streptokinase) (63) suggests that plasmin activity per se may not contribute to virulence but might act as a general bacterial sur-FIGURE 7. Model of P. gingivalis activation of the destructive uPA/plasminogen proteolytic cascade. P. gingivalis-derived RgpA-Kgp complex activation of the uPA/plasminogen pathway (via cleavage and activation of the zymogens, pro-uPA and plasminogen, respectively) initiates a downstream proteolytic cascade promoting periodontal tissue breakdown, fibrinolysis and bleeding, complement activation, and alveolar bone loss (see "Discussion"). This provides the pathogen with a source of nutrients in the form of tissue breakdown products and hemin, contributing to its colonization and virulence, as well as promoting PD. In cells that express uPAR and plasminogen receptors (e.g. macrophages), RgpA-Kgp complex activation of the uPA/ plasminogen pathway occurs at the cell surface to promote pericellular proteolysis of the extracellular matrix. RgpA-Kgp-induced activation of uPA at the cell surface may also promote uPA-uPAR signaling, which is known to play a role in cell adhesion, survival, and motility (4). vival strategy. As discussed above, for macrophages the gingipain complex is likely to be acting in close proximity to the cell surface uPAR-uPA and plasminogen receptors (e.g. Plg-R KT) . In this context, any cell-bound plasmin generated would be largely protected from its inhibitor ␣2-antiplasmin (64), enabling the enhancement, for example, of macrophage matrix destruction.
We show here that P. gingivalis-derived RgpA-Kgp complex specifically targets and activates the uPA/plasminogen pathway. Given the nature of the diverse downstream effector pathways that can be activated by the uPA/plasminogen pathway, including tissue breakdown and fibrinolysis, its activation by P. gingivalis would provide it with a significant survival advantage and helps explain its role as a keystone pathogen.