INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial insertion sequences (IS) can mediate profound genetic effects in their host cells. First, they may create mutations via transposition that can lead to the inactivation of single genes by simple insertion or of multiple genes by virtue of polarity effects (13). Second, their transposition to new sites may result in the transcriptional activation of dormant genes by outward-firing IS promoters or by the creation of new promoters resulting from the insertion event (38). Third, IS elements may promote genomic rearrangements (deletions, inversions, etc.) as the result of their transpositional mobility or by acting as sites for homologous recombination. Examples of IS activity with implications for bacterial pathogenicity have been reported elsewhere, including the transposition (3) and activation (22, 38, 39) of antibiotic resistance genes, the movement and dissemination of virulence genes (14, 50), and the inactivation of virulence genes (34, 46).

Porphyromonas gingivalis is important to the etiology of adult-onset periodontitis (9, 28, 45, 48, 49). Although there are several properties of this organism that are likely to promote periodontal pathology, cysteine protease activity has been associated with virulence. An allelic-exchange mutant deficient in arginine-specific protease activity displayed reduced virulence in a soft tissue infection rodent model (12). Other P. gingivalis mutants constructed in similar fashion have displayed defects in vitro, suggesting specific roles for proteases in virulence (21, 30, 51). A number of related genes encoding cysteine proteases with either arginine or lysine specificity have been cloned, and their nucleotide sequences have been determined (Table 1). There is growing evidence that the repertoire of proteases with arginine specificity produced by P. gingivalis originates from two different genes, rgp-1 and rgpB (30, 43). Recently, the cloning and sequencing of genes encoding lysine-specific cysteine proteases, prtP (W12), kgp (H66), kgp (381), and prtK (W50), have been reported (2, 33, 35). prtP has been reported to have both Arg-X and Lys-X specificity, while kgp apparently has only Lys-X specificity. Structure-function studies of cysteine proteases have revealed that some protease gene sequences are comprised of domains specifying differing functions (e.g., protease activity and hemagglutinin activity) (2, 33, 36, 37).

Protease-encoding genes have been shown to contain multiple copies of repeated nucleotide sequences, and related sequences also have been found in hemagglutinin genes. Multiple DNA fragments bearing these repeats have been detected by hybridizational or direct DNA sequence analysis (2, 8, 12, 18, 36). While studying these repeated sequences, we discovered a DNA fragment cloned from P. gingivalis W83 that contained sequences with near identity to the previously reported prtP gene encoding a protease with reportedly both Arg-X and Lys-X specificity (2). However, this prtP homolog was interrupted by an IS-like element that we have designated IS195. We report here the molecular characterization of this IS-like element. Furthermore, we predicted that the IS195 insertion into prtP inactivated this gene. To test this hypothesis, we have isolated two naturally occurring variants of P. gingivalis W83: one carrying IS195 within prtP and the other devoid of the IS element within the prtP gene. In addition, we constructed an allelic-exchange mutant of P. gingivalis W83 defective in the prtP gene. Here we report on the biochemical characterization of these three strains and an assessment of their virulence in a rodent model.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains and plasmids. P. gingivalis strains were cultivated in brain heart infusion broth (Difco Laboratories, Detroit, Mich.) supplemented with hemin (5 µg/ml), vitamin K₃ (0.5 µg/ml), and cysteine (1%). Cultures were incubated at 37°C in an anaerobic chamber (Coy Manufacturing, Ann Arbor, Mich.) in 10% H₂-10% CO₂-80% N₂. Escherichia coli strains were grown in Luria-Bertani broth (44). Antibiotics were used at the following concentrations: clindamycin, 0.5 µg/ml; erythromycin, 300 µg/ml for E. coli or 0.5 µg/ml for P. gingivalis; and carbenicillin, 50 µg/ml. Plasmid pUC19 was used as a cloning vector (44).

Preparation and analysis of DNA. DNA for contour-clamped homogeneous electric field (CHEF) gel electrophoresis (6) was prepared in agarose plugs. P. gingivalis W83 was grown to mid-exponential phase in 30 ml of brain heart infusion broth. Washed cells (10⁹ cells/ml) were mixed with an equal volume of 2% molten SeaPlaque agarose (FMC), and the mixture was allowed to solidify in 80-µl rectangular molds. The agarose plugs were incubated for 18 h at 50°C in a solution containing 50 mM Tris-HCl (pH 8.0), 1% sodium dodecyl sulfate, 50 mM EDTA (pH 8.0), and 2 mg of proteinase K per ml. Plugs were stored in TE buffer (10 mM Tris-HCl, 0.1 mM EDTA, pH 8.0). Chromosomal DNA for gel electrophoresis (0.75% agarose, TAE buffer [0.4 M Tris-acetate, 0.001 M EDTA, pH 8.0]) was extracted by the method of Marmur (27). Plasmid DNA isolation was performed with a Qiagen kit according to the manufacturer's instructions (Qiagen Inc., Chatsworth, Calif.). DNA was digested with restriction enzymes as specified by the manufacturer (GIBCO/BRL, Gaithersburg, Md., or New England Biolabs, Beverly, Mass.). Electrophoretically separated DNA was transferred to positively charged nylon membranes (Boehringer Mannheim Corp., Indianapolis, Ind.), and hybridization was performed as described by Fletcher et al. (11). Labeling was with [-³²P]dCTP by nick translation with a Promega kit (Promega Corp., Madison, Wis.) or with positively charged peroxidase (ECL kit; Amersham Corp., Arlington Heights, Ill.). Autoradiography was performed as described previously (25).

PCR. Two sets of primers were utilized to amplify P. gingivalis DNA (see Fig. 1, top). Primer set 1 (F1, 5'-GAGACGGTCTTTTCGTCACG-3', and R1, 5'-CACCGTCTTCTTCGAATGTCG-3') amplified DNA encompassing IS195 and sequences of the prtP gene surrounding the IS-like element. Primer set 2, F2 and R2, amplified sequence internal to the IS-like element (F2, 5'-CTGATTAGTGGTAAACGCCCA-3', and R2, 5'-CTGTCCTGTAACGATAACGTC-3'). Primers were synthesized at the Nucleic Acid Core Facility, Virginia Commonwealth University, Richmond. PCR amplification was performed with a Perkin-Elmer DNA Thermal Cycler 480 (Perkin-Elmer Corporation, Norwalk, Conn.) in reaction mixtures (100 µl) containing 700 ng of each primer, 70 ng of template DNA, and 0.5 µl (2.5 U) of AmpliTaq polymerase (Perkin-Elmer Corporation). The PCR consisted of 32 cycles with a temperature profile of 30 s at 95°C, 20 s at 50°C, and 2 min at 72°C, followed by 7 min at 72°C. The amplified DNAs were analyzed on an 0.7% TAE agarose gel.

Construction of allelic-exchange mutant. An allelic-exchange mutation of the prtP gene was made by ligating the ermF-ermAM (12) cassette into the KpnI-BamHI-cleaved pUC19 vector carrying an 0.45-kb HindIII fragment encompassing the prepropeptide region of the prtP gene (see Fig. 7A). This construct was used to electroporate P. gingivalis W83 (12), and clindamycin-resistant colonies were analyzed by Southern blot analysis to confirm the disruption of the prtP gene (see Fig. 7B).

Enzyme assays. Extracellular protease activity was assayed by the method of Grenier and Mayrand (16) with some modifications. Proteins present in culture supernatants were precipitated by addition of ammonium sulfate to 55% saturation. Vesicles were stored in 1 ml of Tris buffer (50 mM; pH 7.5) and kept at 20°C. The presence of trypsin-like proteolytic activity with Lys-X and Arg-X specificity was determined by using chromogenic substrates N--benzyloxycarbonyl-L-lysine-p-nitroanilide (Z-Lys-pNA) (Novabiochem, La Jolla, Calif.) and N--benzoyl-DL-arginine-p-nitroanilide (BApNA) (Sigma Chemical Co., St. Louis, Mo.), respectively.

Hemagglutination assays. Hemagglutination activity was assayed according to the method of Chandad et al. (5) with total extracellular proteins. Round-bottomed 96-well microtiter plates were utilized. One hundred microliters of phosphate-buffered saline (PBS) was added to each well. Then 100 µl of extracellular protein suspension (containing 35 µg of protein) was added to a single well and then diluted serially across several wells. Finally, 100 µl of 1% sheep erythrocytes was added to each well. Defibrinated sheep blood (BBL, Becton Dickinson Microbiology Systems, Cockeysville, Md.) was washed two times in PBS, and washed erythrocytes were prepared as a 1% (vol/vol) suspension. Plates were incubated overnight at 4°C. Hemagglutination was assessed visually, and the reciprocal of the highest dilution displaying a positive agglutination of erythrocytes was recorded.

Virulence studies. P. gingivalis W83 variants, with or without IS195 in the prtP gene and the allelic-exchange mutant, were tested for invasiveness in a mouse model as previously described by Neiders et al. (31). Strains were grown for 18 h in tryptic soy broth supplemented with hemin (1 µg/ml), vitamin K₃ (1 µg/ml), and dithiothreitol (0.5 µg/ml). The cells were centrifuged, washed twice in sterile PBS (0.147 M NaCl, 0.01 M sodium phosphate) under anaerobic conditions, and counted in a Petroff-Hausser chamber. With these counts, various cell concentrations were prepared and their optical densities were measured at 660 nm. These measurements were used to generate a standard curve which was employed thereafter to prepare desired cell densities in PBS. Mice were challenged with subcutaneous injections of 0.1 ml of bacterial suspension at two sites on the dorsal surface. Mice were then examined daily to assess their general health status. The presence and location of lesions were evaluated. Weights were determined for all surviving mice. These experiments were performed under the authorization of an institutionally approved animal use protocol.

DNA sequencing and computer analysis. The dideoxy chain-termination sequencing method was employed with an ABI Prism DNA sequencing kit (Perkin-Elmer Corp.). Recombinant pUC19 plasmids carrying various P. gingivalis W83 sequences were used as templates in sequencing reactions. Nucleic acid and deduced amino acid sequences were analyzed by FASTA and BLAST (Genetics Computer Group Inc., Madison, Wis.) software.

Nucleotide sequence accession number. The nucleotide sequences of IS195 and prtP were deposited in the GenBank nucleotide sequence database under the accession numbers U83995 and AF017059, respectively.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Nucleotide and amino acid sequence analysis. In order to better understand the mechanism of expression of protease genes in P. gingivalis W83, we subcloned and determined the nucleotide sequence of DNA fragments containing the conserved repeated sequences present in protease genes (2, 12, 18, 36). During this work, we discovered that one of the DNA fragments cloned from P. gingivalis W83 was nearly identical to the prtP gene encoding porphypain, a cysteine protease initially characterized in P. gingivalis W12 (2). We compared five genes encoding P. gingivalis cysteine proteases with Lys-X specificity: prtP from P. gingivalis W12, the prtP homolog from P. gingivalis W83, the kgp gene from P. gingivalis 381, the kgp gene from P. gingivalis H66, and the prtK gene from W50 (Table 1). The prtP homologs from W12 and W83 were 99.9% identical except for the insertion of a 1,068-bp sequence at nucleotide position 3129 in the W83 gene. In the 5' region of the gene, the 4,200-bp nucleotide sequence of kgp and prtP was 96% identical while the 3' end of the gene encoding hemagglutinin contained a duplication of different segments of the gene (Fig. 1, bottom). Both kgp and prtP (from W83 and W12) were flanked at their 3' ends by IS1126 sequences (26). The prtP gene was followed by an incomplete copy of IS1126 (IS1126), which had a deletion of 451 bp within its putative transposase gene. The kgp gene was followed by complete copies of IS1126 (1,338 bp) and a 1,068-bp sequence identical to that seen in the prtP gene of P. gingivalis W83 (Fig. 1, bottom).

View larger version (41K): [in this window] [in a new window]   FIG. 1.   Maps of P. gingivalis protease genes. The W83 prtP homolog originally cloned and sequenced in this work is shown at the top of the figure. The numbers below the map refer to size in kilobases. The solid black line indicates flanking upstream DNA. "s" signifies the signal peptide known to be cleaved off during secretion. The prepropeptide (propt.) sequence is believed to be removed from the catalytic protease domain (protease) during posttranslational processing. The hemagglutinin domain is also cleaved during processing of the protein. The shadings and patterns are meant to convey similar nucleotide sequences. The predicted stop codon of the gene is depicted by the rightward arrowhead (at approximately 7 kb). The IS195 sequences interrupting the prtP gene of strain W83 and also seen at the 3' ends of the kgp genes are shown in black. The positions of primers used to amplify the prtP gene are shown by the small arrows below the prtP map of W83. Primer set 1 amplified IS195 and neighboring prtP sequences: F1 (forward) was complementary to bp coordinates 5' 2863 to 2882 3', and R1 (reverse) was complementary to bp coordinates 5' 4498 to 4477 3'. Primer set 2 amplified only IS195 sequences: F2 (forward) was designed to complement 5' 3169 to 3190 3', and R2 (reverse) was designed to complement 5' 4220 to 4200 3' (nucleotide numbers of the prtP gene homolog of strain W83). Similarly constructed maps of other prtP homologs are shown below the W83 homolog. IS elements (IS195 and IS1126) present within or flanking prtP or kgp genes are indicated above or at the ends of the linear maps. ORF, open reading frame.

View larger version (38K): [in this window] [in a new window]   FIG. 2.   Nucleotide sequence of IS195 and flanking regions. The presumed target sequence (9-bp direct repeats) is represented in lowercase letters. The terminal 11-bp inverted repeats of IS195 are underlined twice. The putative promoter regions (35 and 10) and start and stop codons are underlined once. A schematic representation of the nucleotide sequence of IS195 is seen below the nucleotide sequence (IR, inverted repeats; ORF, open reading frame).

View larger version (52K): [in this window] [in a new window]   FIG. 3.   Sequence alignment of putative transposases. The origin of the compared protein sequences is indicated to the left. Amino acids conserved with IS195 are indicated in boldface. The nucleotide sequences of the terminal inverted repeats of these elements are compared and illustrated at the bottom of the figure.

IS195 copy number and distribution in P. gingivalis. We evaluated the presence and approximated the number of IS195-like sequences in different strains of P. gingivalis by Southern blot hybridization. BamHI-digested genomic DNA was electrophoresed, Southern blotted, and hybridized to an internal fragment of IS195. The analysis revealed six hybridizing components in P. gingivalis W83 (Fig. 4A, lane 1). P. gingivalis ATCC 33277 (Fig. 4A, lane 2) and several clinical isolates of P. gingivalis were also examined and found to contain multiple fragments hybridizing to IS195 (Fig. 4A, lanes 4 to 10). Interestingly, an allelic-exchange mutant (Fig. 4A, lane 3) derived from P. gingivalis W83 (12) contained only five fragments that hybridized to the IS195 probe. The size of the hybridizing fragments ranged from 1.4 to 13 kb (Fig. 4A). Since the IS195 element did not contain a BamHI site, a single hybridizing fragment was assumed to represent a single copy of the element. This approach yields a conservative estimate of IS195 copies since more than one copy may reside on the same fragment or there may be two or more comigrating fragments which hybridize with the probe. Alternatively, the hybridizing fragments could contain only a portion of IS195. The minimum number of complete copies of the IS-like element was estimated by PCR amplification of sucrose gradient-fractionated, BamHI-digested genomic DNA of P. gingivalis W83 (Fig. 4B). The chromosomal DNA was digested to completion with BamHI. DNA fractions containing fragments corresponding in length to the six IS195-hybridizing components served as template. We used primers that were internal but close to the ends of the IS-like element so that a PCR product of about 1 kb was assumed to correspond to a complete copy of the IS195 element. All DNA fractions analyzed gave rise to predicted fragments corresponding to an intact IS195 (Fig. 4C). Based on the potential overlap between the fractionated BamHI fragments, we estimated that there are at least three complete copies of the element in the genome of P. gingivalis W83 (Fig. 4C).

View larger version (58K): [in this window] [in a new window]   FIG. 4.   Analysis of genomic DNA. (A) Genomic DNA from P. gingivalis W83 and several clinical isolates was probed with an internal fragment of IS195. BamHI-digested genomic DNA from W83 (lane 1), ATCC 33277 (lane 2), V2296 (lane 3), V2299 (lane 4), V2300 (lane 5), V2302 (lane 6), V2305 (lane 7), V2306 (lane 8), V2307 (lane 9), and V2308 (lane 10) was electrophoretically separated on a 0.75% agarose gel. The Southern blot was probed with an [-32P]dCTP-labeled 1.1-kb DNA fragment PCR amplified from plasmid pVA2538. (B) Agarose gel electrophoresis of sucrose gradient-separated fractions of W83 genomic DNA. BamHI-digested genomic DNA of P. gingivalis W83 was fractionated by sucrose gradient centrifugation and analyzed electrophoretically on a 0.75% agarose gel. (C) Agarose gel electrophoresis of PCR-amplified DNA with, as a template, fractions shown in panel B and primer set 2 (F2 and R2, designed to amplify only IS195) sequences. The same lanes for panels B and C correspond to the same sucrose gradient fractions. Lane 1, 1-kb ladder; lane 2, fraction 45; lane 3, fraction 47; lane 4, fraction 50; lane 5, fraction 52; lane 6, fraction 55; lane 7, fraction 57; lane 8, fraction 65.

View larger version (34K): [in this window] [in a new window]   FIG. 5.   Southern blot analysis of AvrII fragment of P. gingivalis W83 DNA. AvrII-cleaved DNA was subjected to CHEF gel analysis as described in Materials and Methods. Lane 1, molecular size markers (sizes given to left of photograph); lane 2, W83 AvrII fragments; lane 3, Southern blot analysis of lane 2, with an IS195 probe.

Disruption of the prtP-like gene of P. gingivalis W83 by IS195. Our discovery of IS195 in the prtP gene of strain W83 was fortuitous. Using PCR analysis, we evaluated the possibility that our stock culture might contain cells with a wild-type prtP allele in addition to the IS195-interrupted version of this gene. Oligonucleotide primers that should amplify the region immediately flanking the insertion site gave rise to two amplicons (Fig. 6, lane 2). The 500-bp amplicon represented the expected result for the wild-type gene while the 1,600-bp amplicon represented the IS195-inactivated prtP gene. We then did an analysis with DNA extracted from single colonies of P. gingivalis W83 as a template and the same primer sets (Fig. 6, lanes 3 and 4). These results allowed us to isolate two separate strains, one carrying the intact prtP gene and the other carrying this gene inactivated by the IS195 insertion.

View larger version (37K): [in this window] [in a new window]   FIG. 6.   PCR analysis of P. gingivalis W83. Genomic DNA was used as a template. The primer set 1 (F1 and R1) (see primer locations in Fig. 1) to the prtP gene surrounding IS195 was utilized. Lane 1, 1-kb DNA ladder; lane 2, W83 mixed bacterial population; lane 3, W83 variant containing IS195 within the prtP gene; lane 4, W83 variant containing an intact prtP gene.

View larger version (30K): [in this window] [in a new window]   FIG. 7.   Construction of allelic-exchange mutant. (A) Maps of the wild-type prtP gene and its mutant allele containing pVA2534 are shown. The W83 prtP homolog encodes a polyprotein composed of signal peptide (s), prepropeptide (propt.), protease domain (protease), and hemagglutinin domain(s) (hemagglutinin). Relevant restriction enzyme sites are shown at the bottom of the maps. B, BamHI; K, KpnI; H, HindIII. An insertion of a suicide vector, pVA2543, containing the sequence of the gene encoding prepropeptide resulted in disruption of the prtP homolog and duplication of the target site sequence, yielding strain V2577. (B) Southern blot analysis of BamHI-digested DNA of W83 (lane 1) and V2577 (lane 2) with sequences depicted by outward-facing arrowheads in panel A as a probe revealed two differently sized bands, 3.5 and 5.9 kb. (C) Results of the same blot as in panel B but probed with the ermF-ermAM cassette. ORF, open reading frame.

Determination of proteolytic activity of P. gingivalis W83 variants. We examined P. gingivalis W83, V2543, and V2577 for protease activity. The trypsin-like proteolytic activity with Lys-X specificity of the P. gingivalis W83 variant containing IS195 within the coding region of the prtP gene (strain V2543) was reduced about four- to fivefold compared to that of wild-type P. gingivalis W83 (Table 2). A similar reduction in the Lys-X activity was also seen in comparing the wild-type strain to the allelic-exchange mutant (V2577). Interestingly, the Arg-X activity of the insertion mutant (V2543) was slightly depressed compared to that of the wild-type strain. This was not seen in the case of the allelic-exchange mutant.

Hemagglutination studies. We assessed the hemagglutination potential of P. gingivalis W83, V2443, and V2577. While the activity of the allelic-exchange mutant was comparable to that of the wild type, the hemagglutination capability of V2543 (prtP::IS195) was significantly reduced (fourfold dilution [Fig. 8]).

View larger version (95K): [in this window] [in a new window]   FIG. 8.   Hemagglutinating activity of P. gingivalis strains. Hemagglutination activity of 35 µg of extracellular proteins was analyzed as described in Materials and Methods. The reciprocals of serial twofold dilutions are shown to the left of the figure. The strain being tested is indicated at the top of each column of wells. PBS was used as a negative control.

Virulence studies. P. gingivalis W83 wild type and the insertion and allelic-exchange mutants all were tested for virulence in a rodent model. At a dose of 10¹⁰ bacteria per animal, P. gingivalis W83 wild type (no IS195 within the prtP gene) induced swelling of the abdomen and the ventral site of the neck by 24 h. No swelling was observed at the dorsal site of injection. At 24 h, the mice appeared cachectic and hunched with ruffled hair, and by 30 h, all animals challenged with this strain died (Fig. 9). In contrast, all mice challenged with a variant containing IS195 within the prtP gene (V2543) survived the 2-day observation period. The mice appeared cachectic by 24 h, but significant improvement in general health was observed after 48 h. These mice were euthanized after 72 h in order to recover bacteria for further study. PCR analysis of the recovered bacteria from V2543-infected mice revealed the presence of both forms of the prtP gene (data not shown). Challenge with the allelic-exchange mutant (V2577) resulted in the death of all animals but at times that were two to five times as late as that seen for the wild type. We assume that this delayed time of death reflects some type of reduction in the virulence of V2577.

View larger version (15K): [in this window] [in a new window]   FIG. 9.   Virulence of P. gingivalis strains in mice. Ten-week-old female BALB/c mice were inoculated with 1010 bacteria/animal. Survival of mice is represented as a function of time postchallenge. , animals challenged with W83 (wild type); , animals challenged with V2543 (prtP::IS195 insertion mutant); , animals challenged with V2577 (prtP defective allelic-exchange mutant).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We have identified the prtP homolog from P. gingivalis W83. This gene was initially described for P. gingivalis W12 as a cysteine protease with both Arg-X and Lys-X specificity (2). Homologs (called prtK and kgp) have also been characterized for other strains, including W50, H66, and 381 (Table 1). Biochemical analysis of these gene products favors the notion that the enzymatic activity of these gene products is solely a Lys-X specificity. The role of proteases with Arg-X specificity in the virulence of P. gingivalis infections has been suggested by both in vivo and in vitro experiments. Fletcher et al. (12) constructed an allelic-exchange mutant of P. gingivalis W83 carrying a defective copy of the Arg-X protease gene (called rgp-1 [36], prpR1 [1, 43], prtR [47], or rgpA [30]). Their report of this mutant's construction and characterization erroneously called the sequence "prtH" (11, 12). Subsequent analysis of this region revealed this sequence to encode the C terminus of the rgp-1 gene. When translated, this region of the protease is processed to form a subunit or subunits with hemagglutinating ability, and these subunits associate with the protease domain derived from the rgp-1-encoded protein to form a protease-adhesin complex. This mutant showed a significant decrease in Arg-X protease activity and was dramatically less virulent in a mouse abscess model (12).

The prtP allele that we cloned from strain W83 was interrupted by a sequence of 1,068 bp which occurred at bp 3129 of the protease open reading frame. Given the growing evidence for the role of proteases in virulence, this naturally occurring insertion in the W83 prtP gene suggested that cells carrying this mutation were altered in their ability to cause infection. We investigated the sequence interrupting the prtP gene and concluded that it was a novel IS-like element. We have designated this element IS195. Although the tools to demonstrate serial transposition of this sequence in P. gingivalis are not available, our nucleotide sequence analyses argue strongly that this sequence is a functional IS element. First, the entire sequence contained perfect 11-bp inverted repeats at its termini, characteristic of transposable elements. Second, the intervening DNA contained an open reading frame that encoded a predicted protein with sequence similarity related to other putative transposases (Fig. 3) (20, 52). This is strengthened by the DNA sequence homology of the terminal inverted repeat sequences that define each of these putative mobile elements (Fig. 3). Third, consistent with the typical size of IS, all of the elements represented in Fig. 3 are approximately 1 kb. Fourth, we note the presence of a DDE-like motif in the predicted IS195 protein. This motif is believed to play a critical role in transposition reactions and may be the catalytic site of some transposases (40). Normally, this domain consists of an aspartate-aspartate-glutamate (DDE) triad, with the first two aspartates separated by 55 to 64 amino acids. The second aspartate residue and the glutamate residue are separated by 35 amino acids, about half of which are preferred (40). Although this precise spacing was not observed in our work, reports of variability in the architecture of this motif have been made elsewhere (10). Finally, the IS195 copy that we characterized was bordered by a 9-bp duplication of the prtP gene. This is the signature of IS movement into a new location and suggests that IS195 was inserted into the prtP gene via serial transposition.

Taken together, our genetic and biochemical data (Table 2 and Fig. 7) indicate that prtP is the only locus in P. gingivalis W83 encoding a Lys-X protease activity. The behavior of P. gingivalis V2543 with an insertionally inactivated prtP gene suggested that the Lys-X cysteine protease is a virulence factor in soft tissue infections (Fig. 9). However, these in vivo data must be considered in light of the virulence of the allelic-exchange mutant (V2577), which shows a reduction in virulence distinct from that of V2543. We are unable to explain this difference without further experimentation, but some possibilities exist. First, because we do not know the history of V2543, it is possible that other mutational events have occurred and that it is truly not isogenic with W83. Thus, secondary mutations may be contributing to virulence reduction. V2577 was constructed with the W83 strain, and it was biochemically indistinguishable from W83 except for Lys-X protease activity. The difference in virulence seen in comparing V2577 with V2543 suggests undetected alterations in the strain carrying the IS195-inactivated prtP gene. For example, V2543 might carry secondary mutations affecting virulence. Alternatively, it is possible that a truncated gene product produced by the prtP::IS195 gene is an active protease which is unable to be secreted by the cell. The position of the IS195 insertion predicts this (Fig. 1). This trapped protease might have pleiotropic effects on the cell, including interference with the production or secretion of the Arg-X protease(s). Exploring such possibilities awaits further experimentation.

IS elements have been previously reported to insertionally inactivate bacterial virulence genes (7, 19, 34). It is reasonable to speculate that, because such an insertion could be reversible, transposition might function as a means to control virulence gene expression. This is particularly attractive in the case of protease gene inactivation in P. gingivalis. The transcription and translation of large genes like prtP would not be economical for cells grown in vitro where protein degradation would not be needed to acquire nutrients. Thus, it is logical to predict that IS195 inactivation may be related to some selective advantage, and we intend to test this hypothesis. It is also important to note that IS elements may be found in proximity to known or suspected virulence genes. This has implications in terms of both the translocation (14) and the modification (42) of such genes. Finally, the presence of known IS elements and repeated sequences which may be IS-like is one of the hallmarks of pathogenicity islands: genomic blocks encoding selected virulence genes which are present in pathogenic strains of microorganisms (4, 17).