Extracellular Arg-x- and Lys-x-specific cysteine proteinases are considered important virulence factors and pathogenic markers for Porphyromonas gingivalis, a bacterium implicated as a major etiological agent of chronic periodontitis. Three genes. rgpA, rgpB, and kgp, encode an Arg-x-specific proteinase and adhesins (RgpA), an Arg-x-specific proteinase (RgpB), and a Lys-x-specific proteinase and adhesins (Kgp), respectively. The contribution to pathogenicity of each of the proteinase genes of P. gingivalis W50 was investigated in a murine lesion model using isogenic mutants lacking RgpA, RgpB, and Kgp. Whole-cell Arg-x-specific proteolytic activity of both the RgpA(-) and RgpB(-) isogenic mutants was significantly reduced (3- to 4-fold) relative to that of the wild-type W50. However, for the Kgp(-) isogenic mutant, whole-cell Arg-x activity was similar to that of the wild-type strain. Whole-cell Lys-x proteolytic activity of the RgpA(-) and RgpB(-) mutants was not significantly different from that of wild-type W50, whereas the Kgp(-) mutant was devoid of Lys-x whole-cell proteolytic activity. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Western blot analysis using proteinase-specific antibodies of cell sonicates of the wild-type and mutant strains showed that the proteinase catalytic domain of each of the mutants was not expressed. This analysis further showed that RgpB appeared as 72- and 80-kDa bands, and the catalytic domains of RgpA and Kgp appeared as processed 45-kDa and 48-kDa bands, respectively. In the murine lesion model, mice were challenged with three doses of each mutant and wild-type strain. At the lower dose (3.0 x 10(9) viable-cells), no lesions were recorded for each of the mutants, whereas wild-type W50 induced large ulcerative lesions. At a dose of 6.0 x 10(9) viable-cells, all the mice challenged with the wild-type strain died, whereas mice challenged with the RgpA(-) and RgpB(-) isogenic mutants did not die but developed lesions. Mice challenged with the Kgp(-) isogenic mutant at this dose did not develop lesions. At a 1.2 x 10(10) viable-cell dose, only 40% of mice challenged with the Kgp(-) mutant developed lesions, and these lesions were significantly smaller than lesions induced by the wild-type strain at the 3.0 x 10(9) viable-cell dose. All the mice challenged with the RgpA(-) mutant died at the 1.2 x 10(10) viable-cell dose, whereas only 20% died when challenged with the RgpB(-) mutant at this dose. Wild-type phenotype was restored to the RgpB(-) mutant by complementation with plasmid pNJR12::rgpB containing the rgpB gene. There was no difference between the pNJR12::rgpB-complemented RgpB(-) mutant and the wild-type W50 strain in whole-cell Arg-x activity, protein profile, or virulence in the murine lesion model. These results show that the three proteinases, RgpA, RgpB, and Kgp, all contributed to virulence of P. gingivalis W50 in the murine lesion model and that the order in which they contributed was Kgp >> RgpB > or = RgpA.