Kininogenase Activity by the Major Cysteinyl Proteinase (Cruzipain) from Trypanosoma cruzi *

The major isoform of Trypanosoma cruzi cysteinyl proteinase (cruzipain) has generated Lys-bradykinin (Lys-BK or kallidin), a proinflammatory peptide, by proteolysis of kininogen. The releasing of this peptide was demonstrated by mass spectrometry, radioimmunoassay, and ileum contractile responses. The kinin-releasing activity was immunoabsorbed selectively by monoclonal antibodies to the characteristic COOH-terminal domain of cruzipain. To determine the hydrolysis steps that account for the kininogenase activity of cruzipain, we synthesized a fluorogenic peptide (o-aminobenzoyl-Leu-Gly-Met-Ile-Ser-Leu-Met-Lys-Arg-Pro-Pro-Gly-Phe-Ser-Pro-Phe-Arg389-Ser390-Ser-Arg-Ile-NH2) based on the sequence Leu373 to Ile393 of the human high molecular weight kininogen. The hydrolysis products from this peptide were isolated by high performance liquid chromatography, and Lys-BK was characterized as the major released kinin by mass spectrometry. Intramolecularly quenched fluorogenic peptides spanning the Met379-Lys380 and Arg389-Ser390 bradykinin-flanking sequences were then used to assess the substrate specificity requirements of the parasite-derived protease compared with two COOH-terminal truncated recombinant isoforms (cruzain and cruzipain 2). In contrast to the high catalytic efficiency of parasite-derived cruzipain, the recombinant proteinases cleaved the bradykinin-flanking sites at markedly different rates. In addition, we also demonstrated that cruzipain activates plasmatic prekallikrein, which would be a second and indirect way of the parasite protease to release bradykinin.

Trypanosoma cruzi, the parasitic protozoan that causes Chagas' disease, undergoes an obligatory stage of intracellular replication in the mammalian host (1)(2)(3). The infection is propagated to deep tissues by bloodstream trypomastigotes, but the mechanisms that enable their traverse across capillary vessels are not known. Following the parasite adherence, different signaling pathways are triggered in the mammalian cell (4,5), the process being accompanied by cytosolic-free Ca 2ϩ transients in T. cruzi (6). Soon after invasion, T. cruzi lyses the membrane of the parasitophorous vacuole (7) to multiply as amastigotes. The parasitized host cells collapse, releasing newly transformed trypomastigotes to tissue fluids, and then they return to the bloodstream.
In the past few years, there has been significant progress in the characterization of cysteinyl proteinases from T. cruzi (8 -12). Encoded by approximately 130 closely related genes (13), cruzipain(s) are synthesized as preproproteins that undergo processing by autocatalytic mechanisms (10). The mature form of these enzymes contains a papain-like catalytic domain in addition to a long and structurally unique carboxyl-terminal extension whose function remains unknown (10,14). Despite the structural similarity to mammalian cathepsin L (8 -10), the substrate-specificity properties of cruzipain are somewhat reminiscent of cathepsin B (15). The finding that cruzipain expression is increased markedly in replicating forms of this intracellular parasite (12,16) has stimulated efforts to develop synthetic inhibitors as anti-parasite drugs (17)(18)(19)(20).
Despite the wealth of structural and biochemical information on cruzipain(s), their biological role remains unclear. Recent studies revealed that some polymorphic genes are transcribed by the parasite, suggesting that the parasite may express several isoforms at different stages of development and/or stress conditions (21). Sequence analysis of some of these variant genes revealed that non-conservative amino acid substitutions tend to cluster in the catalytic domain, some of the changes being localized to positions that could conceivably influence the subsite specificity.
In the course of studies aimed at characterizing the substrate specificity of parasite-derived cruzipain (22,23) we noticed that this enzyme shared some interesting properties with the human tissue kallikrein (24,25), namely, the ability to hydrolyze efficiently substrates containing Arg or a hydrophobic amino acid at the P 1 position. In the present work we demonstrate that cruzipain releases bioactive kinins from human kininogen as well as from human plasma, even though kininogen, a member of the cystatin superfamily of inhibitors of cysteinyl proteases (26), has the ability to inactivate cruzipain (27,28). Lys-bradykinin (Lys-BK) 1 was demonstrated by mass spectrometry to be the released kinin. The cruzipain kininogenase activity was depleted by affinity chromatography using a monoclonal antibody to the characteristic COOH-terminal domain of the protease. In addition, the sites of cleavage were systematically confirmed, using a synthetic fragment of human kininogen labeled at the NH 2 terminus with o-aminobenzoic acid (Abz-(Leu 373 -Ile 393 )-hKng-NH 2 ) and related internally quenched fluorogenic peptides. We then used synthetic substrates based on the NH 2 -terminal and COOH-terminal flanking regions of bradykinin in human kininogen to compare the substrate specificity requirements of two recombinant isoforms, namely cruzain and cruzipain 2, with the parasitederived cruzipain. Finally, we also demonstrated that cruzipain was able to act via contact phase activation cascade by converting plasma prekallikrein into active kallikrein.

EXPERIMENTAL PROCEDURES
Enzymes-Cruzipain (GP57/51) was isolated from crude aqueous extracts of Dm28c strain epimastigotes as described previously (15). A single band of 51 kDa was observed on sodium dodecyl sulfate-polyacrylamide gel electrophoresis performed under denaturing conditions. As already reported by Murta et al. (9), the presence of the 57-kDa band can only be distinguished when the electrophoresis is carried out under native conditions. MALDI-TOF mass spectrometry yielded a mass of 43 kDa at the center of a wide peak typical of glycosylation. Two wide peaks of low intensity (60 kDa, corresponding to GP57 and 15 kDa corresponding to a degradation product) were also noticed. We subjected the enzyme to immunoabsorption with a monoclonal antibody to cruzipain (JO1, IgG isotype), which binds to the epitope located in the COOH-terminal extension of cruzipain (29). The immunoabsorption was carried out by treating an agarose-protein G resin (Pharmacia Biotech Inc.) with 100 l of either JO1 ascites or an unrelated ascites as control. The antibody-coated beads were washed with phosphate-buffered saline, 0.05% Tween, and 0.5 mg/ml bovine serum albumin and thereafter incubated with 2 volumes of cruzipain solution at 120 g/ml. The supernatants were collected and assayed for hydrolytic activity using the fluorogenic substrate Z-Phe-Arg-MCA as already described (17).
Cruzain, the recombinant protein expressed in Escherichia coli without the COOH-terminal domain (10), was kindly supplied by Drs. J. H. McKerrow and J. C. Engel, from the University of California, San Francisco. Recombinant cruzipain 2 was expressed in Saccharomyces cerevisae essentially as described by Vernet et al. (30). Briefly, the final construct consisted of a chimera containing the preproregion of the S. cerevisae ␣-factor gene and the procruzipain 2 gene where the carboxylterminal sequence region was deleted. Transformation and culture conditions were carried out as described in the aforementioned study. The transformed yeast cells were lysed in phosphate-buffered saline with glass beads in a Brown homogenizer. After removing the cellular debris by centrifugation at 15,000 ϫ g for 20 min, the supernatant was treated with 1% Triton X-100 to improve the solubilization of active recombinant protease. After precipitation with 1 volume of 100 mM acetate buffer, pH 5, the supernatant was extracted with 1 volume of saturated butanol. Recovered in the aqueous phase, the active protease was fractionated further by affinity chromatography on a thiopropyl-Sepharose 6B (Pharmacia). Partially purified cruzipain 2 migrated on sodium dodecyl sulfate-polyacrylamide gels as a 29-kDa band.
Substrates-Intramolecularly quenched fluorogenic peptides have EDDnp attached to COOH-terminal glutamine, a necessary result of the solid phase peptide synthesis strategy employed, the details of which were published elsewhere (32). An automated benchtop simultaneous multiple solid phase peptide synthesizer (PSSM-8 system from Shimadzu) was used for the solid phase synthesis of all of the peptides by the Fmoc (N-(9-fluorenyl)methoxycarbonyl) procedure. High and low molecular weight human kininogens (HMWK and LMWK, respectively) were purchased from Calbiochem. We also used heated human plasma as a source of kininogen, as described previously (33).
Substrate Specificity of the T. cruzi Kininogenase-The hydrolytic activity of the parasite protease was initially measured using fluorogenic substrates containing the specific cleavage sites to the Lys-BK or bradykinin releasing. Hydrolysis of the fluorogenic peptide substrates at 37°C in 100 mM sodium phosphate buffer, 400 mM NaCl, pH 6.3, containing 10 mM EDTA was followed by measuring the fluorescence at em ϭ 420 nm and ex ϭ 320 nm in a Hitachi F-2000 spectrofluorometer. Cruzipain and recombinant enzymes (0.4 -4 nM) were preactivated with 5 mM dithiothreitol for 10 min at 37°C in 200 l of the above described buffer and were kept in ice before use. The 1-cm path length cuvette containing 1.8 ml of the substrate solution was placed in the thermostatted cell compartment for 10 min before the enzyme solution was added, and the increase in fluorescence with time was recorded continuously for 10 min. The slope was converted into mol of substrate hydrolyzed/min based on the fluorescence curves for standard peptide solutions before and after total enzymatic hydrolysis. A solution of Abz-Phe-Arg-OH was used as standard for the fluorescence measurements, which was prepared from tryptic hydrolysis of Abz-Phe-Arg-pnitroanilide (Abz-Phe-Arg-pNA), and its concentration was determined from the absorbance at 405 nm, assuming ⑀ 405 ϭ 8,900 M Ϫ1 ⅐cm Ϫ1 for p-nitroanilide. The enzyme concentrations for initial rate determinations were chosen so as to hydrolyze less than 5% of the substrate. The nonlinear regression data analysis program GraFit (34) fit all data. Some of the second-order rate constant k cat /K m was measured working under pseudo first-order conditions.
Assay of Abz-[Leu 373 -Ile 393 ]-hKng-NH 2 -The cleavage sites of the hydrolysis of the kininogen fragment Abz-(Leu 373 -Ile 393 )-hKng-NH 2 by cruzipain or recombinant cruzain were identified by HPLC on a Novapak C-18 column (3.2 ϫ 150 mm) equilibrated with 0.1% H 3 PO 4 (solvent A). The column was eluted at a flow rate of 1.7 ml/min with 0 -10% gradient of solvent B (90% ACN, 0.1% H 3 PO 4 ,v/v) over 5 min followed by a 10 -80% gradient of the same solvent over 15 min. The elution profile was determined at 220 nm and by fluorescence at 420 nm after excitation at 320 nm. Abz-[Leu 373 -Ile 393 ]-hKng-NH 2 (40 M) was incubated with activated cruzipain or recombinant proteases (final concentration 4 nM) in the buffer described above for the fluorometric enzyme assays. After the addition of the enzymes, 200-l aliquots were removed from these assay mixtures at 15-min intervals and mixed with 600 l of 0.1% H 3 PO 4 to stop the reaction. This reaction mixture was analyzed by liquid chromatography/mass spectrometry using Fisons-VG-Platform mass spectrometer as a courtesy of Dr. Jörg von Hendel from Fisons Instruments, Mainz-Kastel, Germany, as follows. A 5-l sample of the enzymatic reaction mixture was applied in a C-4 column (1 ϫ 250 mm) and two solvent systems: (A) trifluoroacetic acid/H 2 O (1:2,000) and (B) trifluoroacetic acid/ACN (1:2,000). The column was eluted to a mass spectrometer and UV detector (214 nm) at a flow rate of 50 l/min with a 5-60% gradient of system B over 90 min and then to 90% of B in 10 min.
Kininogenase Activity on Human HMWK and LMWK-The ability of cruzipain to cleave the human kininogens was evaluated incubating 1 l of HMWK and LMWK (100 g/ml) with active cruzipain (1.8 nM) in reaction mixtures containing 100 mM sodium phosphate buffer, 400 mM NaCl, 10 mM EDTA, pH 6.3, at 37°C for 2 h. Ethanol (3:1, v/v) was added, and the mixture was centrifuged at 1,000 ϫ g for 15 min. The kinin content in the supernatant was measured by radioimmunoassay, as described previously (35). After 4 h of incubation the kinin released by cruzipain from HMWK (molar ratio 1:40) was identified by MALDI-TOF mass spectrometry, using a TofSpecE from Micromass, Manchester, U. K., after previous purification on the HPLC system.
Kinin Bioassays-The biological activity of the released kinin was measured as isotonic contraction on isolated guinea pig terminal ileum. The isolated organ was suspended in a bath of 2-ml capacity in Tyrode's solution at 37°C containing 135 mM NaCl, 2.7 mM KCl, 1.36 mM CaCl 2 , 0.5 mM MgCl 2 , 0.36 mM NaH 2 PO 4 , 11.9 mM NaHCO 3 , 5.04 mM glucose, pH 7.4. The sensitivity of the response was calibrated with standard solutions of bradykinin and Lys-BK (1-10 nM). Heat-treated human plasma (100 l) and human HMWK (100 l, 1 mg/ml) were added to the ileum preparation bath. After the equilibration period of 3 min, the activated cruzipain or recombinant cruzain solutions (0.4 -40 nM) were added and the isotonic contraction recorded. The amounts of enzyme and substrates were adjusted in a manner that the released kinin fit inside the dose-response curve for bradykinin. Similar experiments were performed in the presence of the bradykinin ␤ 2 -receptor antagonist (Hoe-140, D-Arg-Arg-Pro-Hyp-Gly-␤-(2-thienyl)Ala-Ser-D-tetrahydroisoquiniline-3-carboxylic acid-octahydroindole-2-carboxylic acid-Arg) (10 nM) and with E-64-pretreated cruzipain.
Kinetic Analysis of the Activation of Purified Human Prekallikrein-Plasma prekallikrein (2 g) was incubated with active cruzipain (0.6 nM final concentration) in 100 mM sodium phosphate buffer, 400 mM NaCl, 10 mM EDTA, pH 6.3, for 30 min to 1 h at 37°C. After the incubation period cruzipain was inactivated irreversibly with E-64 (10 M), and the kallikrein activity was measured using 20 M D-Pro-Phe-Arg-MCA in 50 mM Tris-HCl buffer containing 0.015 M NaCl, pH 7.5, at 37°C. As experimental controls of the employed preparation of human plasma prekallikrein, it was also activated for 1 h with 0.6 nM human factor XIIa. and Arg-Ser bonds (peptide 3), as in the bradykinin region of human kininogen, were synthesized to study the behavior of the parasite-derived cruzipain as well the two recombinant isoforms, namely cruzain and cruzipain 2 ( Table I). The HPLC analysis using authentic synthesized fragments demonstrated that all of the peptides were hydrolyzed only at Met-Lys or Arg-Ser bonds, except for the peptide 1, which was at low peptide concentrations hydrolyzed at the Gly-Met and subsequently at the Met 379 -Lys 380 bond. A similar pattern was observed when these peptides were hydrolyzed by two recombinant isoforms. As demonstrated by the k cat /K m values, Met-Lys or Gly-Met bonds were not hydrolyzed as efficiently by cruzain as by parasite-derived cruzipain. In contrast to the behavior of this isoform, recombinant cruzipain 2 has hydrolyzed the Met-Lys bond quite efficiently, being similar to the parasite-derived enzyme in this respect. On the other hand, the Arg-Ser bond in peptide 3 was hydrolyzed efficiently by cruzain, whereas cruzipain 2 displayed a very low k cat /K m value. These results revealed that these two recombinant isoforms have distinct requirements for the S 1 -P 1 interaction, suggesting that their substrate specificities might not be identical. Table II shows all the fragments resulting from the hydrolysis of Abz-(Leu 373 -Ile 393 )-hKng-NH 2 by cruzipain, following analysis by mass spectroscopy. The HPLC profiles (Fig. 1) show the time course of enzymatic hydrolysis of the peptide Abz-(Leu 373 -Ile 393 )-hKng-NH 2 . The fragment Abz-LGMISLMKRPPGFSPFR was observed to accumulate, indicating that the initial hydrolysis occurs at the Arg-Ser bond. The fragments Abz-LG and KRPPGFSPFR (Lys-bradykinin) were detected as the reaction proceeded, suggesting that Gly-Met and Met-Lys bonds have similar susceptibility to hydrolysis by cruzipain under these conditions. Only traces of bradykinin were detected after a 45-min incubation, indicating that the Lys-Arg bond is not cleaved readily by cruzipain. It is noteworthy that Lys-BK and bradykinin were resistant to hydrolysis by cruzipain (20 nM) for up to 6 h of incubation. The same fragments were detected upon hydrolysis of Abz-(Leu 373 -Ile 393 )-hKng-NH 2 by recombinant cruzain; but in contrast to the activity of the parasite-derived protease the Met-Lys bond was only slowly hydrolyzed by the recombinant enzyme. This observation is consistent with the data presented in Table I, which also indicate the lower susceptibility of Met-Lys bond in the peptide Abz-(Leu 373 -Ile 393 )-hKng-NH 2 . This peptide was resistant to hydrolysis by cruzipain 2.

Hydrolysis of Abz-(Leu 373 -Ile 393 )-hKng-NH 2 : Characterization of the Cleavage Products-
Cruzipain-released Bioactive Kinins-The releasing assay of kinin was carried out with heat-treated human plasma (to inactivate preferentially the aminopeptidases that readily inactivate kinins) and purified form of bovine LMWK. The kinin released was detected by guinea pig ileum contraction assay, as shown in Figs. 2 and 3. At the concentrations used, human heat-treated plasma or cruzipain alone did not induce detectable smooth muscle contractile activity. The reaction was only observed at enzyme concentrations above 5 nM; no contraction was seen when the human plasma was incubated with E-64treated cruzipain, confirming the reaction dependence on a thiol proteinase. Importantly, the addition of Hoe-140 has abrogated the ileum contractile response. Similar results were observed with human and bovine kininogen. The HPLC profile (Fig. 3) of the reaction mixture of purified bovine or human LMWK with cruzipain shows that Lys-BK was the major kinin released. Under the same assay conditions, recombinant cruzain and cruzipain 2 did not show any detectable kininogenase activity when incubated with human plasma or purified kininogens. Radioimmunoassay experiment confirmed that cruzipain releases kinin from human HMWK and LMWK as shown in Table III. MALDI-TOF mass spectrometry analyses of the material collected from the HPLC of the reaction mixture of human HMWK in the same conditions of Fig. 3 has shown that the Lys-BK was the major kinin released. The agarose gel affinity column containing a monoclonal antibody to the COOH-terminal domain of cruzipain depleted completely the proteolytic activity of the enzyme solution compared with the agarose gel column devoid of the antibody (Table III).
Cruzipain Activity on Prekallikrein-After being converted to ␣-kallikrein by factor XIIa, plasma prekallikrein possesses the ability to generate the chemical mediator bradykinin from HMWK. To test the possibility of cruzipain indirectly influencing kininogenase activity via contact phase activators, human plasma prekallikrein was incubated with active cruzipain, and samples were analyzed by fluorometric assay. The time course of cruzipain activity on prekallikrein using the fluorogenic substrate of plasma kallikrein, D-Pro-Phe-Arg-MCA, revealed significant amidolytic activity compared with the activation produced by the active human Hageman factor after 1 h of incubation. Fig. 4 shows the activation produced by cruzipain at the first 30-min incubation period. In contrast, after 1 h of incubation with cruzipain, prekallikrein activation was not detectable any more because of the progressive digestion of kallikrein by the parasite enzyme. DISCUSSION The recognition that proteolytic enzymes, other than kallikreins, were also capable of releasing kinins upon incubation with blood plasma was reported long ago by Rocha e Silva et al. (36), in their enzymatic studies with trypsin and Bothrops jararaca venom proteases. Subsequent work indicated that ficin, papain, and a cysteine-proteinase secreted by Clostridium histolyticum could also generate bradykinin-like peptides from plasma (37). More recently it has been reported that cysteine proteinases from Streptococcus pyogenes (38) and from Pophyromas gingivalis, the major causative agent in the devel-  FIG. 2. Guinea pig ileum contraction induced by kinin generated by cruzipain using heat-treated human plasma. Guinea pig ileum contraction was carried out with heat-treated human plasma (Pl). Solutions of 100 l of human plasma and 20 nM cruzipain (Cz) were held in Tyrode's solution bath for at least 1 min, and the response of guinea pig ileum was recorded to certify if the protein substrates themselves present contractile activity. Cruzipain solutions (5-40 nM), preactivated with 5 mM dithiothreitol for 10 min, were added directly to the bath containing 100 l of heat-treated human plasma, and the isotonic contraction was recorded. Hoe-140 (10 nM) was supplemented to the bath 3 min before adding the enzyme and heat-treated plasma. To assess the participation of cysteine proteinases in the reaction, 5 nM dithiothreitol-treated cruzipain was treated with 1,000-fold excess of E-64 for 10 -30 min at 37°C immediately before use in the ileum contraction assay. opment of periodontitis, could rapidly release bradykinin from human kininogen, the reaction being proposed as a correlate for clinical indices of inflammation (39 -42).
In the present study, we demonstrate that the major cysteine protease from the pathogenic parasite T. cruzi also displays kinin-releasing activity. The biochemical characterization of this reaction was initially carried out with the fluorescent labeled peptide, Abz-(Leu 373 -Ile 393 )-hKng-NH 2 , the data indicating that Arg-Ser, Gly-Met, and Met-Lys bonds were successively cleaved, thus releasing Lys-BK. The same pattern of cleavage was observed using a series of internally quenched fluorescent peptides. The results suggest that both Arg-Ser and Met-Lys bonds are susceptible to hydrolysis by parasite-derived protease. However, it is noteworthy that the Met-Lys bond is hydrolyzed poorly by cruzain, in contrast to cruzipain 2, which hardly cleaved Arg-Ser bond. In agreement with the results obtained with synthetic substrates, we characterized Lys-BK as the major kinin released from human HMWK and bovine LMWK. The kininogenasic activity of cruzipain was abolished by treating the enzyme with E-64, thus confirming that it was mediated by cysteine-proteinase rather than by traces of other classes of proteolytic enzymes. The enzyme preparation used in the present study has the same purity as described previously (9), and its MALDI-TOF analysis was in agreement with molecular weight estimates based on electrophoretic behavior. Importantly, we showed that the kininogenase activity of the parasite is mediated by the native form of parasite-derived cruzipain, since the biological activity was specifically depleted by immunoaffinity columns prepared with a monoclonal antibody that reacts with their characteristic COOH-terminal domain.
It is surprising that cruzipain is included in the restricted group of proteases that generates Lys-BK upon hydrolysis of native kininogen because kininogens are potent inhibitors of many thiol proteases (43). It is unclear how cruzipain acts as kinin-releasing enzyme because the proteinase binds tightly to kininogens (27), the inhibition constants being in the pM-nM range depending on the species of kininogen used (28). Notwithstanding these effects, we have observed that human HMWK and LMWK are hydrolyzed completely by cruzipain in fragments after 2 h of incubation (data not shown). The substrate specificity requirements of recombinant cruzain, cruzipain 2, and parasite-derived cruzipain were compared using synthetic peptides. The data showed that all enzymes have hydrolyzed the same peptide bonds. The recombinant cruzain has a marked preference for Arg over Met at the P 1 position of the synthetic substrates (Table I), in contrast to cruzipain 2, which hardly hydrolyzes Arg-Ser bond but cleaves Met-Lys bond with almost the same efficiency of the parasite-derived cruzipain. Accordingly, the relative endurance of Met-Lys or Arg-Ser bonds to the action of recombinant cruzain or cruzipain 2, respectively, might explain their failure to develop significant kininogenase activity in assays with human plasma. Notwithstanding the above arguments, it should be pointed out that the genetically engineered cruzain (10) and cruzipain 2 do not contain the highly glycosylated COOH-terminal domain (130 residues) present in the native form of cruzipain. The functional role of this long and unique extension remains unknown, but it is conceivable that its presence in the intact proteinase might sterically hinder the binding. As a consequence, it decreases the susceptibility to inhibition by the cystatin-like inhibitory domains of human/bovine kininogen slowing the rate of association to the inhibitory domain; however, the kinetics of the kinin-releasing reaction might be favored. It   4. Cruzipain activity on prekallikrein. Plasma prekallikrein (2 g) was incubated with factor XIIa for 1 h (q) and activated cruzipain (final concentration of 0.6 nM) for 30 min (Ⅺ) and 1 h (f) in the same buffer described under "Experimental Procedures." After the incubation period cruzipain was inhibited completely with 10 M E-64. Aliquots of the reaction mixtures were removed, and the kallikrein activity was identified using the fluorogenic substrate D-Pro-Phe-Arg-MCA by the production of 7-amino-4-methylcoumarin registered continuously at 37°C for 10 min at ex 360 nm and ex 480 nm. Prekallikrein activity (‚) was also tested as a control experiment.
is worthwhile mentioning that reversible conformational changes are thought to affect the enzymatic properties of cruzipain, this phenomenon being tentatively attributed to the COOH-terminal extension (15). Independently from the role, if any, of the COOH-terminal domain, the ability of cruzipain to cleave the flanking sequences of BK must be the primary requirement for the expression of a kininogenase activity. The identification of polymorphic variants of cruzain in the parasite genome (21) suggests that cruzipain isoforms displaying different substrate specificities and/or organelle compartmentalization may exist. Furthermore, we have demonstrated that cruzipain activates human plasma prekallikrein; therefore cruzipain can act directly on the physiological kininogenase system generating active plasma kallikrein to target the kallikrein-mediated processing cascade. The capability to generate vasoactive kinins in the bloodstream or interstitial fluids may qualify the structurally diverse T. cruzi cysteine proteinases as factors of virulence in Chagas' disease.