Host Cell Invasion by TRYPANOSOMA cRUZI Is Potentiated by Activation of Bradykinin B₂ Receptors

Chagas' disease, the chronic infection by the parasitic protozoan Trypanosoma cruzi, is a major cause of cardiomyopathy in rural Latin America. Transmitted by blood-sucking triatomine insects, the infective forms of T. cruzi (trypomastigotes) rapidly enter the bloodstream, from where they disseminate the infection to multiple tissues. After invading macrophages, muscle, and other nucleated cells, the trypomastigotes escape from endocytic vacuoles and migrate into the cytoplasm where they transform into round-shaped amastigotes, the replicating forms. Within 5–6 d, the host cells rupture, releasing large numbers of trypomastigotes and amastigotes into interstitial spaces. Acute pathology and parasite tissue load subside with the onset of immunity, but the pathogen is not eradicated. After years of asymptomatic infection, 10–24% of the patients develop a severe chronic cardiomyopathy characterized by myocarditis, fibrosis, microcirculatory lesions, cardiomegaly, and conduction system abnormalities 1,2,3.

At the cellular level, T. cruzi trypomastigotes invade nonphagocytic cells by a unique mechanism distinct from phagocytosis 4,5. Penetration by tissue culture trypomastigotes (TCTs) is preceded by energy-dependent adhesive interactions 6 involving the parasites' surface glycoproteins 7,8 and negatively charged host surface molecules 9. Depending on the host cell–parasite combination studied, invasion requires activation of the TGF-β signaling pathway 10 or stimulation of host cell receptors coupled to heterotrimeric G proteins 11,12. Efforts to characterize the hitherto unknown Ca²⁺-signaling agonist pointed to a crucial role of a cytosolic parasitic serine protease of 80 kD, oligopeptidase B 13. Although null mutants generated by targeted deletion of the oligopeptidase B gene were poorly infective 14, purified or recombinant oligopeptidase B alone failed to induce intracellular free calcium ([Ca²⁺]_i) transients in the mammalian cells 13. Because addition of recombinant oligopeptidase B to null parasite extracts reconstituted [Ca²⁺]_i signaling, it was suggested that the agonistic activity was generated by oligopeptidase B–mediated processing of a cytoplasmic T. cruzi precursor molecule 14.

Other clues to understand the role of T. cruzi proteases in host cell invasion emerged from in vitro assays performed with synthetic inhibitors of cruzipain 15, the parasite's major cysteine proteinase 16,17,18. Encoded by multiple polymorphic genes 19,20, this cathepsin L–like proteinase is the most extensively characterized isoform expressed by replicating forms of the parasite 16,17,18,21. Given the broad pH range of the activity profile and the high stability of cruzipain 17, the finding of antigen deposits of this molecule in foci of myocardial inflammation 22 suggested that this proteinase may contribute to pathology. Our findings that the substrate specificity of cruzipain resembles that of tissue kallikrein and that cruzipain releases the bradykinin (BK)-like vasoactive peptide lysyl-bradykinin (“kallidin”) from its large precursor forms, high (H-) and low (L-) molecular weight kininogens 23, suggested that T. cruzi may directly trigger the kinin system through the activity of this cysteine proteinase.

Here we demonstrate that the short-lived kinin peptides and their cognate G protein–coupled cellular receptors 24 are engaged in the signaling mechanisms leading to T. cruzi invasion. We also show that invasion of cells that overexpress the constitutive B₂ subtype of BK receptor is critically modulated by the kinin-degrading activity of host kininase II, also known as the angiotensin I–converting enzyme (ACE). The finding that activation of the proinflammatory kinin cascade by trypomastigotes potentiates invasion may shed light on the molecular basis of Chagas' disease pathophysiology.

Chinese hamster ovary (CHO) cells transfected with the cDNA encoding the rat B₂ type of BK receptor (B₂R; CHO-B₂R) or mock-transfected CHO cells (CHO-mock) were used 25. Subclone rB2CHO12/4 showed a maximum ³H-BK binding activity of 1.3 pmol/mg of protein at passage 2. CHO cells were cultured in HAM's F12, each supplemented with 10% (vol/vol) of FCS at 37°C in a humidified atmosphere containing 5% CO₂. Vero cells were cultivated in DMEM with 10% FCS. Human primary umbilical vein endothelial cells (HUVECs) were obtained by treatment of umbilical veins with a 0.1% (wt/vol) collagenase IV solution (Sigma-Aldrich). Primary HUVECs were seeded in 25-cm² flasks (Corning) coated with 2% porcine skin gelatin, and grown in M199 medium supplemented with 2 mM glutamine, 2.5 μg/ml amphotericin B, 100 μg/ml penicillin, 100 μg/ml gentamycin, 0.13% sodium bicarbonate, and 20% FCS. Cells were maintained at 37°C in a humidified 5% CO₂ atmosphere until they reached confluency. After treatment with 0.02% trypsin/0.02% EDTA, HUVECs were seeded into 24-well plates with gelatin-coated glass coverslips and cultivated at 37°C for several days before being used in invasion assays.

T. cruzi epimastigotes (Dm28c clone) were cultivated at 28°C in LIT medium containing 10% FCS. TCTs were harvested from the supernatants of infected Vero cultures maintained in DMEM supplemented with 2% FCS (TCT-FCS). TCT transfectants overexpressing Dm28c genes encoding the major cruzipain 18 isoform (for simplicity, hereafter designated cruzipain-1) or cruzipain-2 20,23 were obtained by cloning full-length copies of each of these into the SmaI-HindIII sites of pTEX plasmid 26. Log phase Dm28c epimastigotes were transfected by electroporation with a single pulse of 450 kV, 500 μF in an electroporator (Bio-Rad Laboratories). The parasites were selected for growth in LIT medium containing 10% FCS and 200 μg/ml of geneticin (Sigma-Aldrich) for six consecutive weeks and reselected at 800 μg/ml of geneticin for four additional weeks. Metacyclogenesis was done by incubating stationary phase–transfected epimastigotes in Grace's medium, pH 5.5, including 800 μg/ml of geneticin for 7 d at 27°C. TCT transfectants were collected from Vero cell supernatants 3–4 d after infection with the metacyclics. Plasmid contents were stable for at least 7 wk of culture in the absence of the selecting drug; TCT transfectants were tested in invasion assays after a 3-wk passage. The cysteine proteinase activity contained in cell lysates from transfected or wild-type parasites was measured as the rate of hydrolysis of ε-l-NH₂-Cap-l-(SBz)C-MCA (20 μM) in Na₂HPO₄, 50 mM Na₂HPO₄, 200 mM NaCl, and 5 mM EDTA, pH 7.0, supplemented with 2.5 mM dithiothreitol (DTT), at 37°C. To prepare the parasite cell lysates, freshly released TCTs were washed twice in HBSS and resuspended in 300 μl of PBS, pH 7.2, containing 2 mM EDTA. Then, parasites were subjected to freeze and thaw cycles (two times), followed by the addition of Triton X-100 to 1%. Samples were kept on ice for 10 min and soluble material was recovered by centrifugation at 13,000 g. Protein concentration was determined by the Dc-protein kit (Bio-Rad Laboratories). Peptidase activity was measured in lysates normalized to 2 μg/ml protein (final concentration). Enzyme stability tests were performed by mixing 2 μl of lysates (1 mg/ml) to 100 μl of 0.1 M glycine, pH 12, for 5 s. Assay buffer was added to 1 ml and the peptidase activity was measured as described above.

CHO-B₂R, CHO-mock, or native HUVECs were plated on 13-mm round coverslips at a density of 2.5 × 10⁴ cells/cm² in appropriate medium supplemented with 10% FCS and cultivated in 24-well plates for 48 h at 37°C in a 5% CO₂ atmosphere. Before addition of TCTs, coverslips with attached cells were washed three times with HBSS and kept in serum-free medium containing 1 mg/ml BSA. The parasites added to the wells were freshly released from cultures of infected Vero cells cultivated in DMEM-FCS (2%). After removing cellular debris by low speed centrifugation (169 g), the parasite suspension was diluted three times in HBSS, spun down at 2,000 g, and gently resuspended in DMEM-BSA or M199-BSA (1 mg/ml each). Invasion assays with CHO cells or HUVECs were done in a total volume of 0.6 ml of DMEM-BSA or M199-BSA, respectively, at a parasite/host cell ratio of 2:1 for 3 h at 37°C, unless otherwise specified. When required, the medium was supplemented with 25 μM of captopril (CAP medium). The effects of receptor antagonists or antibodies were assayed by adding to CAP medium 0.1 μM HOE 140 (Aventis), 5–100 nM BK (Calbiochem), or 200 nM of mAbs. In some experiments, CAP medium was supplemented with purified H-kininogen (9 nM) 5 min before adding TCTs to HUVECs. Likewise, the effects of protease inhibitors were tested by adding 10 or 75 μM 1-trans-epoxysuccinyl-l-leucylamido-(4-guanidino) butane (E-64), 10 μM leupeptin (Sigma-Aldrich), 10 μM Z-(SBz)Cys-Phe-CHN₂ 15, or 1 μM recombinant human cystatin C (from Dr. Magnus Abrahamson, Lund University, Lund, Sweden) to the CAP medium. The effect of cruzipain-1 on invasion was evaluated by adding 10-fold dilutions of the activated protease to wells (0–10 nM final concentration) immediately after the addition of the parasites. The interaction was stopped by removing TCTs and washing cells three times with HBSS. Monolayers were fixed with Bouin and stained with Giemsa. Invasion was quantified by counting the number of intracellular parasites in a total of 100 cells per coverslip. Values represent means ± SD of at least three independent experiments, each done in triplicate under “blinded” conditions. Statistical analysis was done by one-way analysis of variance at a P = 0.05 significance level.

The changes in [Ca²⁺]_i were determined using the fluorescent dye Fura 2-AM (Molecular Probes). Customized chambers used in these experiments were designed as follows: 30-mm plastic Petri dishes were drilled leaving 1-cm diameter holes in the center covered at the bottom by thin glass coverslips (0.7 mm) mounted with silicone glue. The plates were extensively washed and sterilized under a UV lamp for 20 min before use; coverslips for HUVECs contained 2% porcine gelatin. CHO cells and HUVECs were plated at 2 × 10⁵ cells per dish in the appropriate medium supplemented with 10 and 20% of FCS, respectively, and cultivated for 24 h at 37°C in 5% CO₂. The monolayers were washed three times with HBSS and incubated for 20 min at 37°C in a 5% CO₂ atmosphere using serum-free HAM's F12 medium, pH 7.4, supplemented with 12.5 mM Hepes, 1 mg/ml BSA, 2.5 mM probenecid (Sigma-Aldrich), 25 μM captopril, and 6 μM Fura 2-AM. After rinsing three times to remove extracellular dye, the cells were maintained at 37°C for 15 min in the same medium without Fura 2-AM. Fura-loaded cells were analyzed in an Axiovert 100 microscope under an oil immersion 40× objective (ZEISS). Fluorescence images were collected by a digital CCD camera using a 510-nm filter. [Ca²⁺]_i was monitored at 36–37°C by alternating the excitation wavelengths between 334 and 380 nm using the Attofluor Ratio System (ZEISS). Raw fluorescence images were digitalized to a pixel assay, point density readings were taken for each image, and a visual display of the 340/380-nm ratio was produced. Before adding parasites or purified proteases, the variation of [Ca²⁺]_i of all cells in the field was monitored for 2 min; 20–30 cells which did not present spontaneous [Ca²⁺]_i transients were chosen as regions of interest. After initial monitoring, the cellular responsiveness to kinin receptor agonists was assessed by adding BK (5–50 nM) to CHO-B₂R or HUVECs preloaded with Fura 2-AM, using CAP medium. Vigorous [Ca²⁺]_i transients were observed for nearly 100% of the monitored cells. Specificity of BK-induced transients was checked by adding 100 nM of HOE 140 to the monolayers before 50-nM agonist exposure. CHO-mock failed to induce significant [Ca²⁺]_i elevations up to 50 nM BK. When indicated, 100 nM of HOE 140 was applied to the cells before the application of enzyme or parasites. Assays with TCTs were carried out at a parasite/host cell ratio of 10:1. Responding cells were observed under light phase to confirm interaction with the parasites. The Attograph software was used to generate tracings representing [Ca²⁺]_i transients of individual cell responses, as well as average responses of 20–30 cells (n = 30). Purified cruzipain-1 was tested at 5 nM, after a 15-min activation of a stock solution (10-fold) with 2.5 mM DTT in PBS, pH 7.2. Controls included the addition of DTT-containing buffer alone or activated cruzipain-1 pretreated with 75 μM E-64 for 30 min. The specificity of HOE 140 was tested by pretreating HUVECs with 100 nM of the B₂R antagonist before stimulating the cells with 14 nM α-thrombin (Dr. Russolina Zingali, Federal University of Rio de Janeiro, Rio de Janeiro, Brazil) or with 50 nM BK. In the last series of experiments, the [Ca²⁺]_i concentrations were determined by the two-point calibration in vivo method 27 after sequentially adding 20 μM ionomycin (Sigma-Aldrich) and 10 mM EGTA to the cultures.

Cruzipain (GP57/51) was isolated from crude aqueous extracts of Dm28c epimastigotes as described 28. Recombinant cruzain (cruzipain-1) lacking the COOH-terminal extension 18 was expressed in Escherichia coli (Dr. J.H. McKerrow, University of California at San Francisco, San Francisco, CA). Recombinant cruzipain-2 devoid of the COOH terminus 20 was expressed in Saccharomyces cerevisiae 23 and partially purified by affinity chromatography on thiolpropyl–Sepharose 6B (Amersham Pharmacia Biotech); SDS-PAGE revealed a major band of 29 kD. Purified H-kininogen was from Calbiochem. mAbs MBK3 (IgG₁) recognizing the BK epitope in domain D4 of human and bovine H-/L-kininogens, HKL16 directed to domain D6_H of human H-kininogens 29, mAb212 directed to cruzipain-1 17, and IgG₁ myeloma protein (MOPC 31c; Sigma-Aldrich) were used. Antiserum to cruzipain-2 (rab 222) was raised in rabbit chimeric protein, where a fragment (Tyr 147 to Trp 187) from the central domain of cruzipain-2 20 had been fused to glutathione transferase.

TCTs were harvested from supernatants of infected Vero cells, washed with HBSS, and resuspended at 2 × 10⁷ cells/ml in DMEM without FCS. After incubation at 37°C for 2 h, the cell suspension was centrifuged at 3,000 g, and the supernatant was filtered through 0.2-μm Millipore filters. 200 μl of the supernatant was tested for proteolytic activity (see above). For the control, 30 μM E-64 (final concentration) was added to the supernatants. Immunoabsorption of cruzipain isozymes was carried out by treating 600 μl of the TCT supernatant for 2 h at room temperature with 10% of protein A–agarose beads (Sigma-Aldrich) previously coated with 10 μl of mAb212 ascites or with 50 μl of rab 222 antiserum. For controls, irrelevant mAb or preimmune serum was used. After washing with PBS, the hydrolytic activity associated to the resin was measured with 50 μM fluorogenic substrate in the presence or absence of E-64.

The kinin concentration in reaction mixtures of human H-kininogen and cruzipain was measured by competitive ELISA (Markit-M; Dainippon). After activating stock solutions of Dm28c cruzipain-1 (50 nM) with 2.5 mM DTT, dilutions of the protease were mixed (2.5–10 nM) with H-kininogen (10 nM) in 200 μl of a reaction buffer containing 50 mM Na₂HPO₄, pH 6.5, 200 mM NaCl, 5 mM EDTA, and 0.25 mM DTT. After incubation for 30 min at 37°C, the reaction was stopped by adding 75 μM E-64, 1 mg/ml BSA, and 25 μM captopril; samples were deproteinized with ice-cold TCA. After diluting the soluble fractions with the supplier's buffer, the samples were applied to the ELISA plate. A standard curve was constructed with synthetic BK. Assays with recombinant cruzipain isozymes (20 nM) were performed as described above. Controls performed with enzyme buffer alone did not induce kinin release. The kinin-releasing activity of trypomastigotes was measured after incubating 10 nM of H-kininogen with TCTs (3 × 10⁶ cells) in 250 μl of Ham's F12, 12.5 mM Hepes, 25 μM captopril, and 1 mg/ml BSA, pH 6.5, for 30 min at 37°C. For the control, TCTs were preincubated with 75 μM E-64 for 30 min at 37°C before the addition of H-kininogen. The reaction was stopped by adding E-64 (final concentration of 10 μM) to the cell suspension. Parasites were centrifuged at 3,000 rpm for 10 min, the supernatants were filtered through 0.2-μm Millipore filters, and kinin release was quantified as above. Experiments were done in triplicates.

Given the evidence that the major cysteine proteinase from T. cruzi, cruzipain, acts as a kininogenase 23, we set out to investigate if short-lived kinins were also generated when trypomastigotes interact with mammalian cells in vitro. As the B₂Rs are preferentially coupled to G proteins of the G_q subtype, we reasoned that target cells that express elevated levels of this receptor should swiftly respond to kinins released upon trypomastigote contact. We addressed this question by first examining if the TCT could trigger [Ca²⁺]_i transients in Fura-2–loaded CHO cells that were transfected with CHO-B₂R. These assays were run in CAP medium to prevent rapid degradation of the short-lived kinins. Asynchronous and repetitive [Ca²⁺]_i transients were promptly induced ∼50 s after TCTs had been added to CHO-B₂R (Fig. 1A, Fig. C, and Fig. E). Average responses (Fig. 1 A) involved 67% of the CHO-B₂R exposed to the parasites. The data registered with two single CHO-B₂R cells (Fig. 1C and Fig. E) show that TCT induces multiple peaks of [Ca²⁺]_i, although the intensity and frequency of the peaks varied from one cell to another, presumably due to the varying number of productive contacts established by the parasites. By contrast, [Ca²⁺]_i transients evolved much slower in CHO-mock and were of low frequency and intensity (Fig. 1G and Fig. H), involving only a small fraction (7%) of the cell population (Fig. 1 H, average responses). The overall response of CHO-B₂R was markedly reduced when the B₂R-selective antagonist HOE 140 was added to the CAP medium before the addition of TCTs (Fig. 1, individual cell responses in D and F, average responses in B), indicating that the B₂R may play an important role for the TCT-induced [Ca²⁺]_i transients. We next evaluated if this signaling pathway was also engaged when the parasites interact with nontransfected host cells expressing high levels of endogenous B₂R, such as primary HUVECs. The addition of TCTs to CAP medium likewise induced [Ca²⁺]_i transients in HUVECs (Fig. 2, single cell tracing in A and average cell responses in D). The response, which involved 29% of the host cells, was markedly reduced by HOE 140 (Fig. 2B and Fig. E). Control experiments showed that HOE 140 did not reduce the [Ca²⁺]_i transients induced by α-thrombin (Fig. 2G and Fig. H). Estimates of the cytosolic [Ca²⁺]_i by in vivo calibration methods 28 indicated that the responses induced by α-thrombin and BK (in both cases involving nearly 100% of the HUVECs) peaked at ∼500 and 770 nM, respectively, starting from a base level of ∼70 nM. By contrast, HOE 140 attenuated the vigorous [Ca²⁺]_i transients which BK, applied at 250 nM, elicited in HUVECs (Fig. 2I and Fig. K). These results confirmed that HOE 140 acts as a specific antagonist of B₂R in the HUVEC system. The addition of epimastigotes failed to elicit significant transients in these cells (Fig. 2C and Fig. F). Collectively, these data suggest that Dm28c TCT signals in HUVECs through the constitutively expressed B₂R.

As it is well established that the induction of [Ca²⁺]_i transients is crucial for invasion by T. cruzi of many different cell types, we reasoned that the [Ca²⁺]_i responses which TCTs stimulated in CHO-B₂R cells may have increased their susceptibility to infection. Consistent with this prediction, the invasion index for CHO-B₂R in the presence of CAP medium was nearly 4.5-fold higher than that for CHO-mock (Fig. 3 A), whereas captopril had no effect on the invasion of CHO-mock (not shown). Similarly to the inhibitory effects of HOE 140 on TCT-induced [Ca²⁺]_i transients, addition of the B₂R antagonist decreased the invasion of CHO-B₂R in CAP medium, whereas it had no effect on the invasion of CHO-mock (Fig. 3 A). Similar effects were noted when we attempted to downregulate B₂R by applying high concentrations of the agonist, i.e., 0.1 μM BK, to these cultures. Using primary cultures of HUVECs, we noted that parasite invasion was enhanced threefold in the presence of 25 μM captopril, whereas absence of captopril or inclusion of HOE 140 in CAP medium significantly attenuated invasion (Fig. 3 B). Although HUVECs were less sensitive to invasion than CHO-B₂R, TCT invasion of HUVECs in the presence of CAP medium was likewise reduced by the addition of high concentrations (0.1 μM) BK (Fig. 3 B). In conclusion, TCT invasion of target cells that overexpress B₂R is markedly enhanced in the presence of captopril, suggesting that the short-lived kinin agonist is protected from degradation by the kininase II inhibitor.

The fact that TCT invasion is potentiated by kininase II inhibitors and reduced in cultures exposed to specific B₂R antagonist or to high levels of exogenous BK pointed to the possibility that a low amount of plasma kininogen either remains associated with the cell surfaces of target cells and/or is displayed by the parasites, in either way serving as a source for precursors of kinin-signaling peptides. Consistent with this concept, the addition of an exogenous supply of H-kininogen (9 nM) to the serum-free CAP medium resulted in a greater than threefold increase in parasite invasion (Fig. 4 A). To test the hypothesis that HUVEC susceptibility to TCT invasion is influenced by the levels of kinin precursors and to further substantiate our notion that kinins are indeed involved in the signaling mechanism, we tested the effects of an mAB (MBK3) to BK that cross-reacts with the kinin segment of human and bovine kininogens 27. Our data show that preincubation of HUVECs with 300 nM MBK3 in CAP medium significantly reduced the potentiating effects of captopril, whereas the addition of unrelated Ab (IgG₁) failed to interfere with TCT invasion, and HKL16, an mAb directed to an epitope of kininogen domain D6_H, had only marginal inhibitory activity (Fig. 4 B). The results suggest that MBK3 compromises parasite invasion of HUVECs, either by blocking the processing of cell-bound kininogens and/or by “neutralizing” the liberated effector kinin peptide.

Given the indications that the major cysteine proteinase isoform, cruzipain-1, releases kinins from H-kininogen 23, and that an accessible BK epitope is required for parasite invasion (Fig. 4 B), we sought to examine if living trypomastigotes are able to release kinins from purified H-kininogen. To this end, washed TCTs were resuspended in CAP medium and 10 nM of purified H-kininogen was added to the suspension. ELISA measurements of the kinin concentrations in culture supernatants revealed the presence of this peptide within 30 min (Fig. 5 A). The processing reaction, which evolved in the absence of reducing agents, converted a small fraction (∼10%) of available H-kininogens into immunoreactive kinins over a limited time period. Significantly, the kinin-releasing activity was completely abolished by pretreating the parasites with an irreversible inhibitor of cysteine proteinases, E-64. These data suggest that TCTs engage cysteine proteinases in releasing kinins from soluble H-kininogens. As trypomastigotes are poorly endocytic 30, the membrane-impermeable E-64 inhibitor either binds to secreted cysteine proteases or acts on activated enzymes as they reach the parasites' flagellar pocket 17.

As pointed out in a previous study 23, the Met-Lys and Arg-Ser flanking site bonds of BK are cleaved at different efficiencies by two distinct recombinant cruzipain isoforms, namely cruzain (cruzipain-1) and cruzipain-2. Assays performed with each of these genetically engineered proteinases (both truncated on the COOH-terminal extension) confirmed that they are able to release immunoreactive kinins from purified H-kininogen, albeit with different efficiencies (Fig. 5 B). Consistent with the demonstration of kinin-releasing activity by living trypomastigotes, analysis of the supernatants from parasite suspensions revealed the presence of peptidases that were activated by DTT and inhibited by E-64 (Fig. 6 A). Immunoprecipitation using isoform-specific Abs mAb212 (anti–cruzipain-1) and rab 222 (anti–cruzipain-2) followed by fluorogenic substrate assay suggested that both isoforms were present in the supernatants (Fig. 6 B). Hence, TCTs released cysteine protease isoforms that may be involved in parasite-induced kinin release.

As both recombinant cruzipain-1 and cruzipain-2 act as kininogenases in vitro, we investigated if the engagement of B₂R by the TCT may be linked to differential expression of these isoforms. To this end, we used the episomal pTEX plasmid to generate TCT transfectants containing full-length copies of the gene encoding for the major cruzipain-1 isoform (TCT-cz1) or for the cruzipain-2 gene (TCT-cz2). Consistent with the higher stability of recombinant cruzipain-2 at alkaline pH (Lima, A.P.C.A., unpublished observations), we found that a significant proportion of the overall cysteine proteinase activity present in TCT-cz2 lysates (30%) was attributed to enzymes that resist inactivation at pH 12 (Fig. 6 C), whereas only a minor fraction (8%) of alkaline-resistant cysteine peptidases were present in TCT-cz1 (not shown) or wild-type cells (Fig. 6 C). Hence, the profile of isoforms expressed by TCT-cz2, although still dominated by the alkaline-sensitive major cruzipain-1 isoform, is skewed towards increased production of the cruzipain-2–like isoforms. Remarkably, invasion assays performed with target cells that do not overexpress B₂R, i.e., CHO-mock or Vero cells, revealed that TCT-cz1 was consistently less infective (exemplified for CHO-mock; Fig. 6 D, gray bar at left) compared with TCT-cz2 (Fig. 6 D, gray bar at right) or wild-type parasites (Fig. 3 A). TCT-cz1 efficiently invaded CHO-B₂R cells, and this effect was reduced almost to the reference level of CHO-mock by the addition of HOE 140 (Fig. 6 D, black bars at right), indicating that the B₂R pathway is used by the cz1 transfectants. By contrast, overexpression of cruzipain-2 did not significantly enhance invasion of CHO-B₂R cells compared with CHO-mock (Fig. 6 D, right). Furthermore, HOE 140 did not reduce TCT-cz2 infectivity for CHO-B₂R (Fig. 6 D, right), suggesting that these parasites signal through kinin-independent pathways. Collectively, these data suggest that T. cruzi's ability to engage the kinin signaling pathway of invasion is preferentially linked to a dominant expression of the major cysteine proteinase isoform (cruzipain-1).

Given the indications that cruzipain-1 is found in the supernatants from wild-type TCTs and that living trypomastigotes release kinins through E-64–sensitive protease(s), we tested the ability of purified cruzipain-1 to induce [Ca²⁺]_i transients. Indeed, we found that the parasite proteinase triggered robust [Ca²⁺]_i transients in CHO-B₂R loaded with Fura-2 (Fig. 7 B). This response was almost nullified by adding HOE 140 at 50 nM (Fig. 7 D) to the CAP medium, whereas [des Arg⁹-Leu⁸-BK], an antagonist of the B₁R subtype, failed to interfere (data not shown). Importantly, the catalytic activity of cruzipain-1 was essentially required for the B₂R stimulation because pretreatment with E-64 abrogated the [Ca²⁺]_i response (Fig. 7 C). These data suggest that cruzipain-1 generates a kinin-like signaling factor by processing kininogen displayed by CHO cells. Unlike the strong [Ca²⁺]_i transients in CHO-B₂R, cruzipain-1 induced only a minor [Ca²⁺]_i response in CHO-mock (Fig. 7 A). This low response was not inhibited by HOE 140 or by [des Arg⁹-Leu⁸-BK] (data not shown), suggesting that the minor [Ca²⁺]_i transients induced in CHO-mock resulted from a kinin-independent signaling pathway.

To further investigate the interrelation between cruzipain activity and the parasite's ability to invade target cells by the kinin pathway, dose–response curves were separately established for BK (up to 100 nM) or for purified cruzipain. Parasite invasion peaked at 1 nM BK, whereas higher concentrations significantly reduced invasion to levels even below baseline, most likely due to receptor desensitization (Fig. 8 A). Thus, stimulation of B₂R by kinin concentrations in the low nanomolar range significantly increases HUVECs' susceptibility to invasion by T. cruzi, whereas higher concentrations of the same peptide protected the host cells. Given that cruzipain-1 is able to release kinins from soluble H-kininogen (Fig. 8 B) and that it induces potent [Ca²⁺]_i transients through B₂R (Fig. 7), we checked if variations in the levels of activated protease could likewise influence the effectiveness of the invasion process. HUVEC invasion was modulated upon addition of increasing concentrations of cruzipain-1 to CAP medium, peaking at 5 nM activated protease (Fig. 8 C). The similar concentration range observed for peak activities of exogenous BK and purified cruzipain-1 suggests that the parasite competence to generate the kinin agonist may be influenced by quantitative and/or qualitative differences of cysteine protease expressed by the parasite. As a further attempt to characterize the involvement of the kinin-releasing cysteine proteases in the invasion process, the effects of synthetic or natural cysteine protease inhibitors were investigated. Unexpectedly, addition of E-64, an irreversible inhibitor of papain-like enzymes, failed to protect HUVECs from infection even when tested at 75 μM (<1% inhibition; Fig. 8 D). Likewise, invasion was not impaired in cultures either supplemented with leupeptin (10 μM) or by cystatin C (1 μM), a tight binding protein inhibitor present in human plasma. Considering the facts that (a) E-64 prevents kinin release from H-kininogen by TCTs (Fig. 5 A), and that (b) trypomastigotes are poorly endocytic in cell suspensions 30, we reasoned that the failure of membrane-impermeable cruzipain inhibitors to impair invasion could be an indication that kinin liberation occurs within the secluded sites formed by the juxtaposition of host and parasite plasma membranes. To address this possibility, we tested the effects of a membrane-permeable, irreversible inhibitor of cruzipain, Z-(SBz)Cys-Phe-CHN₂ 15 and found that it effectively inhibited (85%) HUVEC invasion, the extent of protection conferred by this compound being higher than that observed with HOE 140 (Fig. 8 D).

In this study, we demonstrate that trypomastigote invasion of host cells expressing BK B₂Rs is drastically increased due to signaling by kinin peptides. The presence of captopril in the culture medium was a crucial prerequisite for the increased invasiveness of target cells expressing B₂R, most likely because BK was protected from rapid degradation by ACE/kininase II 31. The effects induced by captopril suggest that T. cruzi tissue tropism may be influenced by variable levels and activities of kinin-degrading peptidases expressed in different host tissues. Indeed, highly vascularized tissues such as kidney parenchyma and lungs abundantly expressing ACE 32 are virtually spared from T. cruzi parasitism, whereas other tissues undergo massive infection and destruction during acute infection 33. Thus, variable expression levels of ACE may modulate the severity of lesions and fibrosis induced by this pathogen in the cardiovascular system of chagasic patients.

Because the presence of captopril or HOE 140 did not interfere with parasite invasion of CHO-mock (this study), Vero cells (not shown), or L₆E₉ myoblasts (Docampo, R., personal communication), parasites likely invade these cell lines by alternative mechanisms such as the TGF-β–dependent transducing pathway 7 or oligopeptidase B–mediated production of [Ca²⁺]_i signaling agonists 13,14. Whereas the latter pathway is thought to involve proteolytic processing of a cytoplasmic precursor protein of the parasite 14, the kinin-mediated signal transduction route described herein likely depends on the processing of a host-derived precursor, namely kininogen(s). Preliminary analysis of a limited panel of well-established T. cruzi strains and clones indicates that invasion through the kinin-transducing pathway is not ubiquitous (Scharfstein, J., unpublished observations), indicating that clones from this highly diverse parasite species 34,35,36,37 may vary with respect to the ability to release proinflammatory kinins.

One of the conundrums of this study is the source of kininogens involved in the invasion process. Kininogens are secretory proteins displaying high affinity binding sites for heparan sulfate proteoglycans exposed on most cell surfaces 38, and thus we speculate that serum-derived kininogens attach to HUVECs or CHO cells during cultivation in vitro. At first sight, the finding that purified cruzipain is able to release kinins from cell-bound kininogens seems paradoxical because papain-like enzymes are rather sensitive to inhibition by the cystatin-like domains of kininogens 39. However, it is well known that the cystatin inhibitory sites in domain D3 of kininogen's heavy chain overlaps with the cell binding site that docks it to endothelial cells 40, and therefore may not be available for cruzipain inhibition. Our finding that FITC-labeled cruzipain failed to bind to HUVECs (data not shown) is consistent with this interpretation. Further, our demonstration that activated cruzipain stimulates vigorous Ca²⁺ transients in CHO-B₂R, and that HOE 140 and E-64 block this effect, points to proteolytic release of kinin agonists, either directly by secreted forms of the proteinase or indirectly by the activation of a kallikrein-like zymogen 23. We have previously shown that cruzipain-2 efficiently cleaves the NH₂-terminal flanking site of BK in kininogen domain D4, whereas the major isoform, cruzain, herein referred as cruzipain-1, preferentially cleaves the COOH-terminal flanking site 23. In this study, we show the presence of both isozymes in TCT supernatants; however, invasion assays performed with TCT transfectants showed that overexpression of the major isoform cruzipain-1, but not of cruzipain-2, favors the engagement of the kinin-mediated invasion pathway. Hence, our findings point to a causal link between parasitic protease cruzipain-1 and host kinin signaling pathway.

Previous studies have revealed that [Ca²⁺]_i transients induced by G protein–coupled receptor agonists stimulate the recruitment and fusion of lysosomes to the plasma membrane at sites where trypomastigotes are attached to host cells 4,5. Considering that adhesive interactions between TCTs and host cells favor their reciprocal activation 41, it is conceivable that kinin signaling is most effective when the processing reaction is restricted to sites formed by juxtaposition of host and parasite cell membranes. Hence, signals emanating from host contacts with a few trypomastigotes, e.g., in CHO-mock or in HOE 140–treated CHO-B₂R, may promote the delivery of host lysosomal cargos to secluded intercellular spaces, and this incipient activation process may generate a reduced environment in these segregated sites thereby sustaining activation of cruzipain-type kininogenases. The finding that membrane-permeable cysteine proteinase inhibitor Z-(SBz)Cys-Phe-CHN₂, but not hydrophilic inhibitor E-64, conveyed partial albeit significant protection of HUVECs, is consistent with such a “dual signaling” model.

As B₂R is constitutively expressed by endothelial cells, smooth muscle, fibroblasts, epithelial cells, and neuronal cells 24, and the B₁ subtype of kinin receptors is upregulated during inflammation 42, increased cellular invasion may be just one example of a physiopathological response that kinins and their metabolites may induce in tissues exposed to T. cruzi. For example, triggering of kinin receptors from vascular endothelial cells may facilitate the transmigration of the trypomastigotes across capillaries, modulate the expression of vascular adhesion molecules, and/or promote plasma leakage into interstitial spaces, and thus contribute to the microvascular changes observed in patients with chronic cardiomyopathy 43. Unraveling the delicate interplay between the parasite cysteine proteinases and the multiple components of the kinin cascade system may provide fresh insights into the molecular pathogenesis of Chagas' disease.

We thank Leila F.C. Duarte and Alda M. Alves for technical assistance, and Dr. Anibal Gil Lopes for free access to digital imaging fluorescence microscopy.

This work was supported by grants from Pronex II, Conselho Nacional de Pesquisas, Fundacão de Amparo à Pesquisa do Rio de Janeiro, Fundacão Universitária José Bonif ácio, and the Deutsche Forschungsgemeinschaft. The authors dedicate this work to the memory of the late Prof. Carlos Chagas Filho.