Inhibition of Trypanosomal Cysteine Proteinases by Their Propeptides*

The ability of the prodomains of trypanosomal cysteine proteinases to inhibit their active form was studied using a set of 23 overlapping 15-mer peptides covering the whole prosequence of congopain, the major cysteine proteinase of Trypanosoma congolense. Three consecutive peptides with a common 5-mer sequence YHNGA were competitive inhibitors of congopain. A shorter synthetic peptide consisting of this 5-mer sequence flanked by two Ala residues (AYHNGAA) also inhibited purified congopain. No residue critical for inhibition was identified in this sequence, but a significant improvement in K i value was obtained upon N-terminal elongation. Procongopain-derived peptides did not inhibit lysosomal cathepsins B and L but did inhibit native cruzipain (from Dm28c clone epimastigotes), the major cysteine proteinase ofTrypanosoma cruzi, the proregion of which also contains the sequence YHNGA. The positioning of the YHNGA inhibitory sequence within the prosegment of trypanosomal proteinases is similar to that covering the active site in the prosegment of cysteine proteinases, the three-dimensional structure of which has been resolved. This strongly suggests that trypanosomal proteinases, despite their long C-terminal extension, have a prosegment that folds similarly to that in related mammal and plant cysteine proteinases, resulting in reverse binding within the active site. Such reverse binding could also occur for short procongopain-derived inhibitory peptides, based on their resistance to proteolysis and their ability to retain inhibitory activity after prolonged incubation. In contrast, homologous peptides in related cysteine proteinases did not inhibit trypanosomal proteinases and were rapidly cleaved by these enzymes.

Cysteine proteinases of the papain superfamily have very similar sequences and a common catalytic mechanism; they are found in bacteria, protozoa, plants, and mammals (1,2). These enzymes play a critical role in the life cycle of protozoan parasites, in host invasion and alteration of the host immune response (3,4). Parasite cysteine proteinases from the genus Trypanosoma and Plasmodium, have been reported as putative targets for chemotherapeutic inhibitors (5,6). The main draw-back of this strategy is that cysteine proteinases have a broad substrate specificity, which makes it difficult to develop inhibitors that target individual proteinases.
Congopain is a cathepsin L-like cysteine proteinase of Trypanosoma congolense, and cruzipain is the equivalent from Trypanosoma cruzi. These protozoan parasites cause bovine trypanosomiase in Africa and Chagas disease in South America respectively (3,4). Cruzipain (also called cruzain) is a highly mannosylated glycoprotein and an immunodominant antigen (GP57/51) that is among the best characterized of parasitic cysteine proteinases (7)(8)(9)(10)(11). The three-dimensional structure of its recombinant central catalytic domain (215 residues) is very similar to those of papain and cathepsin L (11,12). Congopain differs only slightly from cruzipain in its enzymatic specificity due to the presence of a Leu instead of Glu at position 205 (papain numbering) in the S2 subsite (13,14). Unlike mammalian cathepsins B, H, L, S, and K, and falcipain, the related hemoglobin-degrading cysteine proteinase of Plasmodium falciparum (15), which is also a putative target for antimalarial therapy, cruzipain and congopain have a long C-terminal extension that is linked to the catalytic domain by a repeated sequence (poly(Thr) in cruzipain, poly(Pro) in congopain) whose function is unclear (16).
Like mammalian cysteine proteinases, parasite proteinases are synthetized as zymogens that are converted to the mature form by proteolytic cleavage and the release of the proregion. The propeptide of mammalian and vegetal cysteine proteinases takes part in the proper folding, intracellular trafficking, or secretion of the mature proteinase, and in the control of proteolytic activity by blocking the active site (17)(18)(19). Isolated propeptides may still bind and inhibit their mature enzyme even when released from the proenzyme (20,21). The recent x-ray elucidation of the three-dimensional structures of procaricain and procathepsins B and L has provided evidence that the prosegments inhibit papain-like proteinases in a similar fashion, although they differ slightly in length, fold, and sequence (22)(23)(24)(25). The prosegment contacts the enzyme in two main areas in all three crystal structures. One is a prosegment binding loop involving mainly aromatic side chains in hydrophobic interactions, and the other is along the substrate binding cleft but in the opposite direction to that of the substrate. The relative contribution of each region of cathepsin L propeptide to binding to mature enzyme has been measured by successive truncations of the recombinant propeptide (25,26). We have used a different approach, based on the construction of synthetic peptides overlapping the sequence of the prosegment of cathepsin B, to identify the critical residues interacting with the active site of the mature enzyme (27). But the primary sequences of the proregion of parasite cysteine proteinases differ significantly from those of their mammalian and plant homologues, raising the question of whether the proteolytic activity of parasite proteinases is similarly regulated by their proregion. In addition, we do not know whether the presence of a C-terminal extension in trypanosomal cathepsin L-like proteinases modifies the overall folding of the proenzyme, leading to a different interaction of the prosegment with the catalytic site. A series of overlapping peptides spanning the proregion of congopain was synthetized and used to investigate their ability to inhibit the mature enzyme and thus identify residues specifically interacting with the active site. Any such sequence within the prosegment of parasite cysteine proteinases could be used to develop new compounds for the diagnosis or chemotherapy of trypanosomiasis. EXPERIMENTAL PROCEDURES L-3-Carboxy-trans-2,3-epoxypropionyl-leucylamido- (4-guanidino) butane (E-64) and DL-dithiothreitol (DTT) 1 were purchased from Sigma. Benzyloxycarbonyl-Phe-Arg-AMC was from Bachem Biochimie (Voisins-Le-Bretonneux, France).
Circular Dichroism Spectroscopy-The CD spectra of the 17-mer peptide Pcp25 (40 M) and 34-mer peptide Pcp27 (20 M) were measured at room temperature (25°C) over 280 -200 nm using a Jobin-Yvon Mark IV discometer. The 29-mer peptide HIV1-V3-13 (26 M) was used as a control (29). The path length of the cell was 1 mm, and data from repeated scans were averaged. Experiments were made in 0.1 M phosphate buffer, pH 6.0, or in 0.1 M phosphate buffer, pH 6.0, 50% TFE (v/v). Prediction of the secondary structure of the three peptides was performed by the Chou and Fasman method, running the MacPromass software (Beckman Research Institute).
Enzymes-Congopain from T. congolense was purified as reported by Authié et al. (30), and cruzipain was purified from Dm28c clone epimastigotes of T. cruzi according to Lima et al. (31). The recombinant catalytic domain of cruzipain was a gift from Dr. James McKerrow (University of California, San Francisco, CA). Cathepsins B and L were purified from rat liver (32). The activation buffers for enzyme assays were 0.1 M phosphate buffer, pH 6.0, containing 6 mM DTT and 2 mM EDTA for congopain, and 0.1 M phosphate buffer, pH 6.0, containing 10 mM DTT for cruzipain, 0.1 M phosphate buffer, pH 6.0, containing 1 mM EDTA, 2 mM DTT for cathepsins B and L. Titration of the active site was made using E64 (33).
Kinetic Measurement-The activities of trypanosomal enzymes were measured with acetyl-Phe-Arg-AMC. K m values were determined from Hanes linear plot (triplicate experiments). The activities of mammalian cathepsins were recorded with benzyloxycarbonyl-Phe-Arg-AMC (32).

TABLE I
Synthetic peptides derived from the proregion of congopain All peptides have been synthetized as peptidyl amides. For the sake of clarity, peptides Pcp1 to Pcp14 are not shown. 23 overlapping 15-mer peptides (i.e. peptides Pcp1-Pcp23) spanning the whole pre-proregion of congopain with a five-residue stagger were screened for their inhibitory capacity toward congopain. Only YHNGA-containing peptides inhibited congopain. Peptides Pcp24 to Pcp27 were derived from peptides Pcp19 -Pcp21, and all contain the YHNGA sequence. Peptides Pcp28 to Pcp31 result from Ala scanning of Pcp25, and peptide Pcp32 results from the substitution Gly 3 Ile in Pcp25. Peptides Pcp33, Pcp34, and PB8 are the homologous peptides in the proregions of cruzipain, falcipain, and rat cathepsin B, respectively.
Pre-pro region of congopain: Proteinases were activated for 5 min at 37°C in their respective assay buffer before kinetic measurements (using a Kontron SFM 25 spectrofluorimeter, with excitation and emission wavelenghts of 350 and 460 nm). Peptides Pcp1-Pcp23 were first screened for their capacity to inhibit congopain by microfluorimetry (Dynatech microreader). Congopain (3.3 nM) was incubated with peptide (10 and 100 M) in the activating buffer at 37°C for 5 min (final volume in the microfluor plate ϭ 100 l), and the enzymatic reaction was triggered by adding acetyl-Phe-Arg-AMC (10 M); the residual enzymatic activity was recorded. The inhibitory constants (K i ) were determined (triplicate experiments) for all the peptides that showed significant inhibition under the above conditions. Congopain (3.3 nM) was incubated with inhibitory peptides for 5 min at 37°C before adding substrate (1-10 M). The Peptide Stability-A mixture of peptide Pcp25 (60 M) and parasite proteinase (congopain/cruzipain (3.3 nM)) was incubated in the activation buffer at 37°C for 0 to 5 h. At various times, two aliquots were removed, one for measuring the residual enzymatic activity by spectrofluorimetry after adding acetyl-Phe-Arg-AMC (10 M), and the other was mixed with E64 (10 M) to inactivate the enzyme. This sample was chromatographed by reverse phase-HPLC on a C18 OD 300 Brownlee column using a linear 0 -60% gradient of acetonitrile in 0.1% trifluoroacetic acid. The elution profiles were analyzed by the software Spectacle (ThermoQuest, les Ulis), and cleavage sites were located by N-terminal sequencing (ABI 477A sequencer, Perkin-Elmer). The same experiments were done with peptides Pcp24, Pcp26, Pcp27, and Pcp33, the procathepsin B-derived peptide PB8 (60 M), and the profalcipainderived peptide Pcp34 (60 M). Cathepsin L (4 nM) was incubated for 5 min to 1 h with peptide Pcp25 (60 M) at 37°C in 0.1 M phosphate buffer, pH 6.0, containing 1 mM EDTA, 2 mM DTT. The reaction was stopped by adding E64 (10 M), and proteolysis products were fractionated by reverse-phase chromatography, as described above.

Inhibition of Congopain by Synthetic Proregion-derived Pep-
tides-A critical segment that inhibits the catalytic domain of mature rat cathepsin B was previously identified using overlapping peptides spanning the whole proregion (27). The almost simultaneous elucidation of the three-dimensional structures of human and rat procathepsin B has shown that this segment lies close to essential residues within the active site cleft (23,24). We looked to see if this could occur for trypanosomal cysteine proteinases using 15-mer peptidyl amides  spanning the whole pre-proregion of congopain. These overlapping peptides (peptides Pcp1-Pcp23), prepared by Fmoc solidphase multisynthesis, are reported in Table I. They were assayed (final concentration: 0.1 mM) for their ability to inhibit purified congopain using acetyl-Phe-Arg-AMC as substrate (K m ϭ 10 Ϯ 1 M). Three consecutive peptides Pcp19, Pcp20, and Pcp21 inhibited congopain competitively with K i values of 40, 8, and 60 M. Peptides Pcp19, Pcp20, and Pcp21 have a common pentapeptide sequence: Tyr74p-His75p-Asn76p-Gly77p-Ala78p (procathepsin L numbering). This short sequence flanked by Ala residues (Pcp24) is also inhibitory, although the K i is higher (K i ϭ 225 M). Replacements by Ala of single amino acids from position 74p to 77p (peptides Pcp28 -31), or substitution of Gly77p by a bulky isoleucyl residue (Pcp32) did not significantly change the K i values (4, 5, 5, 7, and 7 M, respectively). This would indicate that no single residue of the inhibitory sequence is essential for peptide binding to mature congopain unless detection is impaired by the weakness of the interactions between short peptides and the protease active site. The pentapeptide YHNGA sequence in procongopain is in the same position relative to the C terminus of the procongopain sequence, as are the residues that contact the substrate binding cleft of cathepsin B, cathepsin L, and caricain ( Fig. 1) (22). This suggests that they interact similarly and that the structural organization of the prosegment of trypanosomal cathepsin L-like proteinases is comparable with that of related plant and mammalian proteinases. However the primary sequences of these segments are completely different, suggesting that the interaction with the cognate enzyme is specific. The three-dimensional structure of procathepsin L indicates that additional residues take part in the interaction of the prosequence with the catalytic domain. This is the case for residues Phe63p and Phe71p through aromatic interactions in the vicinity of the active site and for residue Phe56p, which contribute to the stabilization of the prosegment via interactions with two tyrosyl residues of an aromatic cluster of the protein binding loop (Tyr-146 and Tyr-151) (25). Because these aromatic residues are also present in procongopain, their contribution to congopain binding was investigated using longer peptides derived from peptide Pcp20 (Table I). The inhibitory activity of peptide Pcp27, which includes both the YHNGA sequence and the aromatic amino acids reported above, was assayed toward congopain to investigate the effect of the 17residue N-terminal extension. This extension significantly enhanced the binding to the mature enzyme (Student's t test; p Ͻ 0.01) ( Table II). The aromatic residues Phe63p and Phe56p, which tightly contact cathepsin L via its protein binding loop, may provide a secondary anchoring site for congopain/Pcp27 interaction, thus stabilizing the complex. But peptides Pcp25 and Pcp26 inhibited congopain similarly to peptide Pcp20. Coulombe et al. (25) have shown that residues 68p-75p in procathepsin L form a two-turn ␣-helix (␣3p helix) (Fig. 1). A Chou and Fasman analysis of the secondary structure of the congopain proregion predicted that a similar arrangement could occur (Fig. 1). Accordingly, the CD spectra of the N-terminalextended peptide Pcp27 (residues 51p-85p) showed a partial ␣-helical conformation that was amplified by the presence of TFE (34). This was not the case for the shorter peptide Pcp25 (Fig. 2) and for HIV1-V3-13 (29), used as a control peptide of similar length, that did not show any ␣-helical structure, even in 50% TFE (v/v) (data not shown).
Resistance of Procongopain-derived Peptides to Proteolysis-The three-dimensional structure of papain-related proenzymes shows that this part of the prosegment, which is in close contact with the active-site cleft, is positioned in a reverse direction to that of a substrate, which probably makes it more resistant to proteolysis (24,25). But whether short synthetic peptides derived from the prosequence of cysteine proteinases also obey this reverse binding mode is not known. Measurements of their resistance to proteolysis would suggest an orientation similar to that of the whole prosegment, all the more that peptides contain a Phe-Arg dipeptide, which is one of the preferred pairs of residues accomodated at S2 and S1 subsites in papain-like cysteine proteinases (35).
Procongopain-derived peptides (60 M) were incubated with congopain (3.3 nM) for 0 -5 h at 37°C, and the mixtures were analyzed by reverse-phase HPLC and by measuring their inhibitory potential. Peptide Pcp24 (AYHNGAA) was completely resistant to proteolysis by congopain as shown by reverse phase-HPLC and retained its full inhibitory capacity after incubation for 5 h. Longer peptides containing the Phe-Arg pair (Pcp25-27, Pcp28 -31) also inhibited congopain after incubation for 5 h. But N-terminal sequencing revealed some cleavage after the Phe-Arg pair. This hydrolysis, however, was much slower than that of peptides derived from procathepsin B (peptide PB8) and profalcipain (Pcp34) from P. falciparum, which did not inhibit mature congopain, and were totally degraded by congopain in less than 30 min under the same experimental conditions. Such a rapid cleavage was also observed reacting procongopain-derived peptide Pcp25 with cathepsin L. This supports the hypothesis that peptides Pcp25-27 and Pcp28 -31 bind to congopain in a reverse binding mode, leading to enzyme inhibition and conferring resistance to proteolysis but that there may be some minor, substrate-like binding, leading to slow proteolysis. By contrast, the unrelated peptides PB8 and Pcp34 bind to congopain as substrates, as does procongopainderived peptide Pcp25 when interacting with cathepsin L.
Selective Inhibition of Trypanosomal versus Mammalian Enzymes by Peptides from Their Proregions-The active-site-interacting sequences of the proregions of papain-like proteinases from plants, mammals, and parasites differ significantly. However, the procongopain Y74pHNGA78p sequence (procathepsin L numbering) is a consensus sequence in the propeptides of trypanosomal cysteine proteinases (Fig. 3). This suggests that YHNGA-containing peptides can specifically inhibit cysteine proteinases from other trypanosomal species but not mammalian lysosomal cathepsins. We checked this by measuring the activities of the procongopain-derived peptides Pcp25 to Pcp27 toward cruzipain, the homologous cysteine proteinase of T.cruzi (13,14), and cathepsins B and L. The corresponding YHNGA-containing peptide from the cruzipain prosegment (peptide Pcp33) was also synthesized and assayed with congopain. None of the procongopain-derived peptides altered the enzymatic activity of mammalian cathepsins B or L, but peptides Pcp25, Pcp26, and Pcp 27 all inhibited cruzipain Dm28c, acting as competitive inhibitors with the K i comparable with those reported for congopain (Table II). A recombinant form of cruzipain (also called cruzain) corresponding to the catalytic domain, which therefore lacks the C-terminal extension (12), was inhibited as well as the full-length cruzipain Dm28c by Pcp25, Pcp26, and Pcp27. Therefore, the C-terminal extension does not impair the catalytic activity of cruzipain, suggesting that the C-terminal domain is not in the vicinity of the active site of cruzipain. As expected, the procruzipain-derived peptide Pcp33 inhibited cruzipain Dm28c, recombinant cruzain, and congopain, with K i values in the same range as those for Pcp25, Pcp26, and Pcp27 (Table II), but it did not affect the enzymatic activity of cathepsins B and L. Peptide Pcp33 was also fully inhibitory after incubation with cruzipain for 5 h.
The almost identical positioning of the inhibitory sequences in the prosequences of congopain and cruzipain with those in mammal and plant cysteine proteinases whose three-dimensional structures are known suggests that the prosegments of the trypanosomal enzymes are folded similarly despite the presence of a C-terminal extension. Although proregion-derived peptides can specifically interact with their cognate enzyme, the K i values are too high to make these peptides efficient inhibitors. Nevertheless these values compare with those found for procathepsin-B-derived peptides (27,36). Crystallographic data have shown that this part of the proregion, which is flexible and fits into the active site of the catalytic domain, contributes weakly to the overall binding energy of interaction between the enzyme and the propeptide (25). Experiments carried out with recombinant truncated proregions of cathepsin L have demonstrated that the two N-terminal ␣-helices (␣1p and ␣2p) are essential for propeptide binding and contribute greatly to stability (25,26). The short ␣3p helix (residues 68p-75p), which is close to the active-site-interacting sequence, also helps anchor the residues dipping into the substrate binding site. This could explain why peptide Pcp27 from procongopain, which has a partial ␣-helical structure, unlike the related inhibitory peptide Pcp25, is also a slightly better inhibitor. The sequence YHNGA represents a promising new feature that may contribute to the design of proteinase-directed antiparasitic drugs of therapeutic interest.