Folding of the Plasmodium falciparum Cysteine Protease Falcipain-2 Is Mediated by a Chaperone-like Peptide and Not the Prodomain*

Papain-family cysteine proteases of the malaria parasite Plasmodium falciparum, known as falcipains, are hemoglobinases and potential drug targets. Available data suggest that papain-family proteases require prodomains for correct folding into functional conformations. However, in prior studies of falcipain-2, anEscherichia coli-expressed construct containing only a small portion of the prodomain refolded efficiently, suggesting that this enzyme differs in this regard from other papain-family enzymes. To better characterize the determinants of folding for falcipain-2, we expressed multiple pro- and mature constructs of the enzyme in E. coli and assessed their abilities to refold. Mature falcipain-2 refolded into active protease with very similar properties to those of proteins resulting from the refolding of proenzyme constructs. Deletion of a 17-amino acid amino-terminal segment of the mature protease yielded a construct incapable of correct folding, but inclusion of this segment in trans allowed folding to active falcipain-2. The prodomain was a potent, competitive, and reversible inhibitor of mature falcipain-2 (K i 10−10 m). Our results identify a chaperone-like function of an amino-terminal segment of mature falcipain-2 and suggest that protease inhibition, but not the mediation of folding, is a principal function of the falcipain-2 prodomain.

Malaria remains one of the most important infectious diseases in the world. Plasmodium falciparum, the most virulent human malaria parasite, is responsible for hundreds of millions of illnesses and more than one million deaths each year (1). Because available antimalarial agents are limited by drug resistance, toxicity, and cost, new drugs, ideally directed against new targets, are urgently needed (2). Among promising drug targets are proteases that hydrolyze hemoglobin during the erythrocytic stage of infection and thereby provide amino acids for parasite protein synthesis (3). Cysteine protease inhibitors blocked parasite hemoglobin hydrolysis and development in vitro (4) and cured Plasmodium-infected mice (5), suggesting that plasmodial cysteine proteases are appropriate drug targets. The best characterized cysteine proteases of P. falciparum are the papain-family enzymes falcipain-2 (6) and falcipain-3 (7). Each of these proteases cleaves hemoglobin under the acidic conditions of the plasmodial food vacuole, the site of hemoglobin hydrolysis, and they are likely the principal targets of antimalarial cysteine protease inhibitors.
In initial studies of the in vitro refolding of Escherichia coli-expressed falcipain-2 and falcipain-3, constructs including the mature proteases and only short (35 and 33 amino acid, respectively) carboxyl-terminal portions of the prodomains folded into active enzymes (6 -8). These results were surprising, because the prodomains of papain-family proteases have previously been deemed essential for correct folding (9 -11). Mature falcipain-2 and falcipain-3 contain an unusual aminoterminal segment that is not seen in any other described papain-family enzymes. We hypothesized that the amino-terminal extension of falcipain-2 plays a role in protein folding. To determine the roles of the falcipain-2 prodomain and aminoterminal extension of the mature domain in folding, we evaluated the abilities of multiple falcipain-2 constructs to refold into active protease. We found that the falcipain-2 prodomain is not required for folding, that folding is mediated by the aminoterminal extension, and that, as is the case with other proteases, the prodomain is a potent inhibitor of mature falcipain-2.

EXPERIMENTAL PROCEDURES
Materials-Benzyloxycarbonyl-Phe-Arg-7-amino-4-methyl coumarin (Z-Phe-Arg-AMC) 1 was from Bachem, Z-Leu-Arg-AMC was from Peptides International, and Z-Phe-Arg-fluoromethyl ketone (Z-Phe-Arg-FMK) was a gift from Dr. Robert Smith (Prototek). Restriction endonucleases and Vent DNA polymerase were from New England Biolabs. Oligonucleotides were synthesized at the Biomolecular Resource Center, University of California, San Francisco. The synthetic peptide was from Genemed Synthesis. All other reagents were from Sigma or as mentioned in the text.
Polymerase Chain Reaction and Sequencing-All DNA fragments were amplified from the pTOP-FP2 plasmid (6) using Vent DNA Polymerase and the primers mentioned below. The sequence of each construct was confirmed by DNA sequencing at the Biomolecular Resource Center, University of California, San Francisco.
Both pRSET-B and pQE-30 expression vectors add His-tag fusions at the amino terminus of the foreign proteins that allow one-step purification of recombinant proteins by nickel-nitrilotriacetic acid (Ni-NTA) chromatography. Expression was induced with isopropyl ␤-D-thiogalactopyranoside, and expressed proteins were purified by Ni-NTA chromatography under denaturing conditions as previously described (8).
Ϫ188FP2 and Ϫ89FP2 were further purified by binding to a Q-Sepharose column (Amersham Biosciences) and elution with a step gradient of NaCl in 8 M urea, 20 mM Tris-Cl, pH 8.0.
For large scale refolding, equal amounts (100 nmol) of each denatured-reduced protein were diluted 100-fold in 200 ml of the optimal refolding buffer for each protein, containing 1 mM each GSH and GSSG (buffers B for Ϫ188FP2, C for Ϫ89FP2, and E for Ϫ35FP2 and FP2) and incubated at 4°C for 20 h. Insoluble protein was removed using a 0.22-m membrane. Each refolding reaction was concentrated to 20 ml, brought to pH 5.8 and 5 mM DTT at room temperature for 1.5 h (1 h for Ϫ35FP2 and FP2), the pH was raised to 6.25, the protein was applied to a Q-Sepharose column, and active protease was eluted using a gradient of 1 M NaCl in 20 mM bis-Tris, pH 6.25. Elution fractions containing protease activity (indicated by Z-Leu-Arg-AMC hydrolysis) were combined, concentrated to 1-2 ml, and diluted with an equal volume of glycerol for storage. Quantities of unfolded expressed proteases were estimated by the Bradford assay (12). To estimate recovery of active protease, enzyme concentrations were determined by active site titration with Z-Phe-Arg-FMK (13) before (Ϫ35FP2 and FP2) or after (Ϫ188FP2 and Ϫ89FP2) purification.
Amino-terminal Sequencing and Substrate SDS-PAGE-For aminoterminal sequencing, purified active products of Ϫ188FP2, Ϫ89FP2, Ϫ35FP2, and FP2 were electrophoresed in a 12% SDS-polyacrylamide gel, transferred to an Immobilon-P SQ membrane (Millipore), stained with Coomassie Blue, excised, and sequenced at the Protein and Nucleic Acid Facility, Stanford University Medical Center. For substrate SDS-PAGE, protease samples were mixed with non-reducing SDS-PAGE sample buffer and electrophoresed in a 12% SDS-polyacrylamide gel copolymerized with 0.1% gelatin. The gel was washed twice with 2.5% Triton X-100 for 30 min each at room temperature, incubated overnight at 37°C in 100 mM sodium acetate, 10 mM DTT, pH 5.5, and then stained with Coomassie Blue.
Progress of Refolding-Samples (150 pmol) of denatured-reduced recombinant proteins (in 8 M urea, 20 mM Tris-Cl, 200 mM imidazole, 10 mM DTT, pH 8.0) were diluted 100-fold in 297 l of ice-cold refolding buffer containing 1 mM each of reduced and oxidized glutathione (buffers B for Ϫ188FP2 and Ϫ188dFP2; C for Ϫ89FP2 and Ϫ89dFP2; E for Ϫ35FP2, FP2 and dFP2) and incubated at 4°C. At multiple time points, 15 l of each refolding reaction was incubated in 325 l of 100 mM sodium acetate, 10 mM DTT, pH 5.5, at room temperature for 25 min to allow processing, 10 l of Z-Leu-Arg-AMC was added (final concentration 50 M), and fluorescence, representing protease activity, was monitored continuously for 10 min at room temperature (8).
Refolding in the Presence of Propeptide Constructs-Refolding of extension-deleted mature falcipain-2 (dFP2) alone or with different concentrations of prodomain constructs (Ϫ188P 17 , Ϫ89P 17 , Ϫ188P, or Ϫ89P; molar ratio of dFP2:prodomain constructs 1:1-1:3) or a synthetic peptide (molar ratio of dFP2:peptide 1:1-1:50) identical to the falcipain-2 amino-terminal extension was studied in 11 refolding buffers (described above; each containing 1 mM each GSH and GSSG). An equal amount of mature falcipain-2 (FP2) was also refolded in buffer E. For each refolding reaction, 3-l aliquots containing 150 pmol of denatured reduced recombinant protein(s) were added to 297 l of ice-cold refolding buffer, samples were incubated at 4°C for 20 h to allow refolding, 25 l of each refolding reaction was added to 315 l of 100 mM sodium acetate, 10 mM DTT, pH 5.5, at room temperature for 25 min to allow processing, and refolding to active enzyme was assessed as relative hydrolysis of Z-Leu-Arg-AMC over 30 min. Inhibition of Falcipain-2 by the Prodomain-To refold prodomain Ϫ188P for inhibition studies, the denatured protein was 100-fold diluted (final protein concentration 15-20 g/ml) in 100 mM Tris-Cl, 1 mM EDTA, 500 mM L-arginine-HCl, pH 7.5, incubated at 4°C for 20 h, and concentrated (using a 10-kDa cut-off membrane, Millipore) to 10 ml. Insoluble protein was removed using a 0.22-m syringe filter. Different concentrations of Ϫ188P were preincubated with 0.96 nM falcipain-2 derived from 4 different constructs in 100 mM sodium acetate, 10 mM DTT, pH 5.5, for 10 min at room temperature. The substrate Z-Leu-Arg-AMC (4 M; ϳ2-fold below K m ) was added, and fluorescence was continuously measured for 10 min. Pseudo-first order conditions were assumed (substrate hydrolysis was shown to be Ͻ Ͻ10%) (15). The results were plotted as (V o /V i ) Ϫ 1 versus I, where V o is the uninhibited rate, V i the rate in the presence of the inhibitor, and I is the concentration of inhibitor. From this plot, the K i,app was determined (slope ϭ 1/K i,app ) and used to calculate the true K i using the relationship K i ϭ K i,app /(1ϩ [S]/K m ), where [S] is the substrate concentration.

Unusual Features of Plasmodial Cysteine Proteases-Falci-
pain-2 and falcipain-3 are fairly typical papain-family proteases, but they share some unusual features, including large prodomains, a putative type II membrane-spanning region near the amino terminus, and some unique inserts in the mature protease sequence (6,7). Most relevant for this study, the two plasmodial proteases lack the typical papain-family mature protease cleavage site, where cleavage occurs immediately upstream of the sequence XP, where X represents a nonpolar amino acid about 25 amino acids upstream of the catalytic cysteine (16) (Fig. 1). The conserved proline is seen in nearly all known papain-family proteases other than falcipain-2 and falcipain-3. Determination of the amino-terminal sequences of native mature falcipain-2 and processed recombinant falcipain-2 and falcipain-3 showed that these two enzymes contain an amino-terminal extension, with an additional 17 amino acids in falcipain-2 and 26 amino acids in falcipain-3 relative to other papain-family proteases (Fig. 1). The sequences of the two plasmodial amino-terminal extensions are 47% identical and do not show significant identity with any other protein sequences in available data bases.
The Falcipain-2 Protease Domain Refolds without Its Prodomain-To investigate the role of the falcipain-2 prodomain, we expressed four constructs including mature falcipain-2 with different portions of the prodomain (Ϫ188FP2, Ϫ89FP2, and Ϫ35FP2) and mature falcipain-2 alone (FP2) in E. coli (Fig. 2,  A and B). Each of the proteins was refolded at alkaline pH and underwent autohydrolysis at acid pH to a product of identical size (Fig. 2C). Amino-terminal sequencing of the active products revealed the native mature falcipain-2 terminus for the FP2 construct (QMNYEE) and microheterogeneity (DQM-NYEE and MNYEE) for the three prodomain constructs. All four refolded proteins were enzymatically active, as demonstrated by hydrolysis of the substrates Z-Leu-Arg-AMC and gelatin (Fig. 2D). The refolding of each construct to active protein varied with the composition of refolding buffer and redox couple, as seen previously (8), but optimized recovery for each product was similar ( Fig. 2A). The progress of refolding to active enzyme for each protein was very similar, with maximum activity achieved after ϳ18 h (Fig. 3). Thus, in contrast to all other described papain-family proteases, the prodomain of falcipain-2 is not required for folding of the active enzyme. Furthermore, including the prodomain in expression constructs does not improve the speed or efficiency of folding over that of the mature protease.
Prodomain Does Not Alter the Catalytic Behavior of Falcipain-2-It has recently been shown that alterations in prodomains can result in differential properties of mature proteases that are identical in sequence but have been processed from different proenzyme constructs (17). To determine whether this "memory" phenomenon was evident in falcipain-2, we compared catalytic properties of the four active proteases generated from different falcipain-2 constructs. The four proteases cleaved peptide substrates with very similar kinetics (Table I). Thus, the falcipain-2 prodomain does not appear to impart folding characteristics on the mature protease that impact on catalytic activity.

FIG. 2. Expression and folding of falcipain-2 constructs.
A, schematic of expressed constructs. Percentage recovery was calculated as the quantity of active protease (based on active site titration with Z-Phe-Arg-FMK) divided by the starting quantity of unfolded protein (quantified by Bradford assay). Purified recombinant proteins were evaluated by SDS-PAGE before (B) and after (C) refolding at alkaline pH, and to assess activity, refolded proteins were also evaluated by gelatin-substrate SDS-PAGE (D). Gels were stained with Coomassie Blue. The positions of molecular mass standards are shown in kilodaltons. FIG. 1. Domain organization of preprofalcipain-2. The schematic shows locations of a predicted type II transmembrane domain (TM), the mature protease domain processing site (arrow), an unusual amino-terminal extension of the mature protease (filled box), and highly conserved catalytic amino acids (C, H, N). The sequence flanking the protease domain-processing site of falcipain-2 (FP2) was aligned using the Clustal method (38) with corresponding regions of falcipain-3 (FP3), papain (Pap), human cathepsin K (hCK), human cathepsin L (hCL), and cruzain (Cru). Fully conserved amino acids are shaded red, identities with FP2 are shaded green, protease domain cleavage sites are shown by arrowheads, the amino-terminal extensions of FP2 and FP3 are underlined, and "ERFNIN" prodomain residues (18) are labeled with asterisks.
The Amino-terminal Extension of Mature Falcipain-2 Is Essential for Folding-Considering the lack of a requirement for the prodomain for folding of mature falcipain-2, we considered the role of the unusual amino-terminal extension of this mature protease. Three constructs from which the 17-amino acid extension was deleted were expressed in E. coli (Fig. 4, A and  C), purified, and refolded. Proenzyme constructs from which the 17-amino acid extension was deleted (Ϫ188dFP2, Ϫ89dFP2) and mature falcipain-2 lacking the 17-amino acid extension (dFP2) were all incapable of folding to active enzyme in any of the 176 tested buffers (Fig. 3). Thus, the short aminoterminal extension of falcipain-2 is required for correct folding of the mature protease.
The Amino-terminal Extension Can Mediate Folding in Trans-To determine whether the amino-terminal extension must be covalently linked to the remainder of the mature protease for folding, prodomain constructs with (Ϫ188P 17 , Ϫ89P 17 ) or without ( Ϫ188P, Ϫ89P) the amino-terminal extension were expressed (Fig. 4, B and D) and incubated with the mature protease lacking the extension (dFP2) in standard folding buffers (Fig. 5). The extension-deleted mature protease (dFP2) was successfully folded to active enzyme in combinations that included extended prodomain constructs ( Ϫ188P 17 or Ϫ89P 17 ) but not in those including prodomains without the extension ( Ϫ188P, Ϫ89P; Fig.  5). Thus, presentation of the amino-terminal extension of mature falcipain-2 in trans was adequate to mediate correct folding of truncated mature falcipain-2. When a synthetic peptide identical to the 17-amino acid extension was incubated with the extensiondeleted mature protease (dFP2), folding to active enzyme was not seen (Fig. 5). Our results indicate that the short amino-terminal falcipain-2 extension does not need to be covalently bound to the mature protease to mediate folding. However, the free peptide could not mediate folding, suggesting that, when it is not bound to the remainder of the mature protease, the extension must be presented by the prodomain, presumably after this domain binds to the mature protease via structural features maintained by well conserved "ERFNIN" and other residues in many papainfamily enzymes ( Fig. 1; Refs. 18 and 19).
The Prodomain Is a Potent Inhibitor of Falcipain-2-In addition to their roles as mediators of folding, the prodomains of many papain-family (20,21) and other (22,23) proteases are potent inhibitors of their cognate proteases, but it is not clear if the folding and inhibitory functions of these domains are linked. It was of interest to determine whether the falcipain-2 prodomain retained an inhibitory function independent of folding activity. The inhibition of falcipain-2 by the prodomain was compared for the active products of four different constructs (Ϫ188FP2, Ϫ89FP2, Ϫ35FP2, and FP2; Fig. 2). The prodomain was a potent inhibitor of the mature protease, with similar inhibition constants for each active product (Table II). Inhibition of the protease was competitive and reversible, as indicated by very similar V MAX values but increasing K m values with increasing prodomain concentrations. For example, the K m for Z-Leu-Arg-AMC was 8.2 M (V MAX , 2.7ϫ10 Ϫ7 mol s Ϫ1 ) in the absence of the prodomain and 73.3 M (V MAX , 2.7ϫ10 Ϫ7 mol s Ϫ1 ) in the presence of 18 nM prodomain. DISCUSSION We have characterized folding determinants of the P. falciparum cysteine protease falcipain-2 and shown that, unlike other studied papain-family proteases, falcipain-2 does not require its prodomain for folding. Rather, falcipain-2 utilizes an unusual amino-terminal peptide to mediate folding. Although the falcipain-2 prodomain differs from other papain-family enzymes in this regard, it shares another function in that it is a potent inhibitor of the mature protease.
Molecular chaperones mediate folding of many proteins (24). A subset of chaperones, including the proforms of many serine (11,(25)(26)(27) and cysteine (9, 10) proteases, act intramolecularly by mediating folding of the cognate mature protein before processing and removal of the chaperone domain. Our results identify a novel intramolecular chaperone-like function for the amino-terminal extension of falcipain-2. It is described as chaperone-like, because unlike true chaperones, it remains bound to the mature protease.
A role in folding for small amino-terminal peptides has previously been hypothesized but not, to our knowledge, identified (28). It is likely that many other proteins utilize small intramolecular peptides to mediate folding, in particular proteins that lack prodomains and those, such as falcipain-2 and some other proteases (e.g. trypsin, chymotrypsin, and cathepsin D), that do not require their prodomains for folding to the active enzymes (29 -31). Of note, the prodomain of falcipain-2 acts as a potent inhibitor of the mature protease, but it does not affect the rate or efficiency of protease folding or impact on the catalytic properties of the mature protease.
The unusual folding properties of falcipain-2 are probably shared by other plasmodial cysteine proteases. Falcipain-3 (but not falcipain-1) shares the presence of an unusual sequence at the amino terminus of the mature protease domain FIG. 3. Progress of refolding. Denatured-reduced recombinant proteins were diluted 100-fold in refolding buffer containing 1 mM each of reduced and oxidized glutathione (buffers B for Ϫ188FP2 and Ϫ188dFP2, C for Ϫ89FP2 and Ϫ89dFP2, and E for Ϫ35FP2, FP2, and dFP2; see "Experimental Procedures" for description of buffers) and incubated at 4°C. Aliquots from each reaction were tested for Z-Leu-Arg-AMC hydrolysis as described under "Experimental Procedures," and results are expressed as the percentage of maximum activity (fluorescence units/min) obtained with any protease.  (7). In addition, a chimeric protein in which the aminoterminal extension of falcipain-2 was replaced by the corresponding region of falcipain-3 refolded with very similar ef-ficiency to that of falcipain-2. 2 Similar sequences are also seen in homologous proteases of murine malaria parasites 3 but not in any other reported cysteine proteases. Thus, the unusual folding properties of a subset of the falcipains appear to be required due to a specific feature of the biology of malaria parasites. Why has a plasmodial cysteine protease developed an unusual intramolecular control of protein folding? We hypothesize that this architecture may have been generated to allow the protease to perform two quite different cellular functions. First, falcipain-2 cleaves hemoglobin in the acidic P. falciparum food vacuole (4,6). Second, the protease appears to be responsible for the hydrolysis of erythrocyte cytoskeletal proteins that is probably required to allow the egress of mature parasites from the erythrocyte. A role for falcipain-2 in the hydrolysis of cytoskeletal proteins is supported by the observation that cysteine protease inhibitors block the rupture of erythrocytes by mature malaria parasites (32,33), the identification of P. falciparum cysteine protease activity that cleaves the erythrocyte cytoskeletal proteins ankyrin and protein 4.1 (34), and the and extension-deleted falcipain-2 (dFP2) were incubated alone or with prodomain constructs (as labeled; equimolar concentrations) or a synthetic 17-mer peptide identical to the falcipain-2 amino-terminal extension (Peptide; dFP2:peptide ratio of 1:50) in refolding buffer (containing 1 mM each GSH and GSSG). Results represent hydrolysis of Z-Leu-Arg-AMC by 25 l of each refolding reaction (arbitrary fluorescence units) in the optimal refolding buffer for each protein (buffer E for FP2 and buffer G for dFP2 and all its combinations). Error bars represent S.D. of the results from four refolding reactions. FU, fluorescence units. The hydrolysis of Z-Leu-Arg-AMC by falcipain-2 derived from four different constructs was carried out in the presence of different concentrations of the prodomain (Ϫ188P), and inhibition constants (K i ) were calculated as discussed under "Experimental Procedures." demonstration that this activity is identical to that of recombinant falcipain-2 (35).
To access hemoglobin, falcipain-2 is likely transported to the parasite food vacuole by vesicular trafficking from the cytostome organelle after insertion into the parasite plasma membrane, as has also been proposed for aspartic plasmodial proteases (36). To access the erythrocyte cytoskeleton, we propose that falcipain-2 is cleaved from the plasma membrane by limited autohydrolysis or the action of exogenous proteases. This process likely occurs at the cytostome, the plasmodial organelle through which erythrocyte cytosol transits into the parasite, and where the parasite plasma membrane appears to be in direct apposition with erythrocyte cytosol (37). If this cleavage into erythrocyte cytosol liberates truncated, incompletely folded falcipain-2, the parasite may have evolved an intramolecular system for protease folding that is independent of the prodomain to allow the elaboration of hydrolytic activity against cytoskeletal targets. This model must take into account the need for control of falcipain-2 activity in the erythrocyte, presumably via inhibition by the free falcipain-2 prodomain or exogenous cysteine protease inhibitors such as cystatins. In any event, it explains the unique need in malaria parasites for a cysteine protease that can mediate its folding after cleavage of the prodomain.
More generally, our data show that inhibitory and folding functions can be independent in papain-family enzymes. This independent control of enzyme inhibition and folding may well be a common feature of proteases, including those that may utilize different portions of the prodomain for the two functions. Furthermore, our results identify the use of a small internal peptide to direct folding of a mature protease and suggest that many other proteins may incorporate similar systems in which internal sequences provide mediation of protein folding.