Functional characterization of the propeptide of Plasmodium falciparum subtilisin-like protease-1.

Erythrocyte invasion by the malaria merozoite is prevented by serine protease inhibitors. Various aspects of the biology of Plasmodium falciparum subtilisin-like protease-1 (PfSUB-1), including the timing of its expression and its apical location in the merozoite, suggest that this enzyme is involved in invasion. Recombinant PfSUB-1 expressed in a baculovirus system is secreted in the p54 form, noncovalently bound to its cognate propeptide, p31. To understand the role of p31 in PfSUB-1 maturation, we examined interactions between p31 and both recombinant and native enzymes. CD analyses revealed that recombinant p31 (rp31) possesses significant secondary structure on its own, comparable with that of folded propeptides of some bacterial subtilisins. Kinetic studies demonstrated that rp31 is a fast binding, high affinity inhibitor of PfSUB-1. Inhibition of two bacterial subtilisins by rp31 was much less effective, with inhibition constants 49-60-fold higher than that for PfSUB-1. Single (at the P4 or P1 position) or double (at P4 and P1 positions) point mutations of residues within the C-terminal region of rp31 had little effect on its inhibitory activity, and truncation of 11 residues from the rp31 C terminus substantially reduced, but did not abolish, inhibition. None of these modifications prevented binding to the PfSUB-1 catalytic domain or rendered the propeptide susceptible to proteolytic digestion by PfSUB-1. These studies provide new insights into the function of the propeptide in PfSUB-1 activation and shed light on the structural requirements for interaction with the catalytic domain.

Parasites of the genus Plasmodium are responsible for the debilitating disease malaria, which affects 300 -500 million people each year, mostly in subtropical regions (1). The continued emergence of drug-resistant parasites imposes an urgent need for a new generation of control measures. The life cycle of the malaria parasite in the human is complex, involving intracellular replication in both hepatocytes and erythrocytes, with the pathophysiological manifestations closely allied to intraerythrocytic replication. Plasmodium merozoites actively penetrate erythrocytes via a tightly regulated series of events involving proteolytic processing and release of proteins from the apical organelles and surface of the parasite (2)(3)(4). Functional maturation of many invasion-related proteins requires their proteolytic modification and, in some cases, their eventual removal from the parasite surface. These processing events can occur both in the apical organelles and at the parasite surface (4 -6), suggesting a role for proteases in different compartments of the parasite. Specific inhibitors of cysteine and serine proteases can block invasion, providing compelling evidence that these classes of parasite proteases play a major role in the invasion process (6,7). In the case of Plasmodium falciparum, the species responsible for most fatal cases of malaria, two subtilisin-like serine proteases (subtilases) have been identified, called PfSUB-1 1 and PfSUB-2 (8 -10). PfSUB-1 is expressed in the latter stages of schizogony, is located in a subset of dense granules at the apical end of the merozoite, and is secreted during erythrocyte invasion (8), consistent with an involvement in invasion. Sequence comparisons and homology modeling have shown that PfSUB-1 is structurally most closely related to the bacterial subtilisins; unlike most bacterial subtilisins, however, PfSUB-1 has a highly restricted substrate specificity, suggesting a highly specialized, non-degradative physiological role in the parasite (11,12). It may be a suitable target for drug development. A better understanding of the function, enzymology, and specificity of PfSUB-1, along with the characterization of endogenous inhibitors, may facilitate the development of specific inhibitors as a novel antimalarial therapy.
Subtilases are synthesized as zymogens, which consist minimally of a signal peptide, a propeptide domain, and a catalytic domain. The propeptide functions as an intramolecular chaperone, being essential for correct folding of the catalytic domain and for enzyme maturation (13). Folding is generally rapidly followed by autoproteolytic cleavage at the junction between the propeptide and mature domain. The cleaved propeptide often remains transiently associated with the catalytic domain in a noncovalent complex. In many cases, subtilase propeptides are potent inhibitors of their cognate catalytic domains, with inhibition constants in the nanomolar range (14 -18), and so eventual enzyme activation generally requires further proteolytic digestion and release of the propeptide (14,15). PfSUB-1 maturation during secretory transport in the parasite involves two sequential autocatalytic processing events that convert the precursor first to a p54 form by cleavage of the propeptide and subsequently to a p47 form by further cleavage within the N-terminal segment of p54 (11). Interestingly, when produced in a baculovirus expression system, recombinant PfSUB-1 (rPfSUB-1) is mostly secreted in the p54 form, a substantial fraction of which is noncovalently linked to its cognate propeptide, p31 (11,12).
To better understand the activation of PfSUB-1, we have examined the role of the propeptide and explored the possibility that it can act as a specific inhibitor. After recombinant expression and characterization of the propeptide, we have investigated its structure, its interactions with both baculovirus-derived and native enzymes, and the specificity and stability of these interactions. Analyses of various mutants of the propeptide have indicated the regions and residues that determine its interactions with the catalytic domain.

Construction of an Escherichia coli PfSUB-1 Propeptide Expression
System-DNA encoding the propeptide of PfSUB-1 (Lys 26 -Asp 219 of the 690-residue-long primary sequence) (8,11) was amplified by PCR from a synthetic gene (12) using the oligonucleotide primers 5Ј-GGTATTG-AGGGTCGCAAGGAGGTACGTAGCGAG-3Ј (called F1) and 5Ј-AGAG-GAGAGTTAGAGCCTTAATCAGCGGACACCAACTTGTC-3Ј and cloned into pET-30Xa/LIC (Novagen) by ligation-independent cloning according to the manufacturer's instructions. The recombinant propeptide was called rp31. A truncated version of rp31 (called rp31⌬11) lacking 11 residues at the C terminus was also generated by PCR using the F1 forward oligonucleotide and the reverse oligonucleotide 5Ј-AGAGG-AGAGTTAGAGCCTTAAGCACCCTTCTCTTCCAAGATC-3Ј. The nucleotide sequences of the cloned products were confirmed on both strands by sequencing on an automated sequencer (Applied Biosystems Model 377).
Site-directed Mutagenesis-The desired mutations within the PfSUB-1 propeptide were introduced by PCR-mediated site-directed mutagenesis using the F1 forward primer in combination with the reverse primers indicated below. Three different rp31 mutants (called rp31-VSAL, rp31-ASAD and rp31-ASAL) were generated. In the rp31-VSAL mutant, the C-terminal residue Asp 219 was mutated to Leu by replacement of the terminal CTA codon with CTT using the reverse oligonucleotide 5Ј-AGAGGAGAGTTAGAGCCTTAAAGAGCGGACAC-CAACTTGTC-3Ј (the mismatched codon is underlined). In the rp31-A-SAD mutant, Val 216 was similarly mutated to Ala using the reverse oligonucleotide 5Ј-AGAGGAGAGTTAGAGCCTTAATCAGCGGAAGCC-AACTTGTCAGAC-3Ј. For the rp31-ASAL double mutant, Asp 219 was mutated to Leu and Val 216 was replaced with Ala using the reverse oligonucleotide5Ј-AGAGGAGAGTTAGAGCCTTAAAGAGCGGAAGCC-AACTTGTCAGAC-3Ј. DNA fragments encoding the mutants were cloned into pET-30Xa/LIC, and the mutations were verified on both strands by sequencing as described above.
Protein Expression and Purification and Antiserum Production-rp31 and mutant rp31 sequences were expressed in E. coli BL21-Gold (DE3) pLysS (Stratagene). One liter of LB medium containing 30 g/ml kanamycin was inoculated with 4 ml of fresh overnight culture and grown at 37°C to A 600 ϭ 0.5 prior to induction with 1 mM isopropyl-␤-D-thiogalactopyranoside. After 4 h of induction, the cells were pelleted, resuspended in 50 ml of 50 mM phosphate buffer (pH 8.0) containing 300 mM NaCl, and lysed by sonication, and the insoluble material was removed by centrifugation. The resulting supernatant was supplemented with 1 mM phenylmethylsulfonyl fluoride, 1 M pepstatin, 100 M leupeptin, and 0.3 M aprotinin and clarified by passage through a 0.22-m filter.
Purification of all of the recombinant propeptide constructs to homogeneity was by the same three-step procedure. To the ϳ50-ml supernatant was added 40 ml of nickel-nitrilotriacetic acid-agarose (QIAGEN Inc.). The gel suspension was mixed on ice for 30 min, transferred to a sintered glass funnel, washed with 5 gel volumes of 2 mM imidazole in 50 mM phosphate buffer (pH 8.0) containing 300 mM NaCl, and eluted with 2 gel volumes of 150 mM imidazole in the same buffer. Eluted proteins were concentrated by ultrafiltration at 4°C with a YM-10 membrane (Amicon, Inc.) and then chromatographed on a Superdex 200 26/60 prep-grade column (Amersham Biosciences) equilibrated in 20 mM Tris-HCl (pH 8.2) containing 150 mM NaCl. The propeptide peak was detected by SDS-PAGE (19) and staining with Coomassie Blue R-250. Pooled peak fractions were loaded onto a 1-ml RESOURCE Q column (Amersham Biosciences) pre-equilibrated in 20 mM Tris-HCl (pH 8.2) and 150 mM NaCl. The column was washed with the same buffer and then eluted with a linear gradient of NaCl (150 -600 mM) in the same buffer. The propeptide peak was pooled and stored in aliquots at Ϫ70°C. Electrospray mass spectrometric analysis of the purified protein was performed as described previously (11). A mouse polyclonal antiserum was raised against the purified protein using standard procedures. Expression of baculovirus-derived rPfSUB-1 in insect cells and protease purification were as described previously (12).
Cleavage of the His 6 and S Tags from rp31-The N-terminal fusion peptide containing the His 6 and S tags was removed from rp31 using Factor Xa (Roche Applied Science). A solution of purified rp31 was adjusted to 50 mM Tris-HCl (pH 8.2), 100 mM NaCl, and 1 mM CaCl 2 ; Factor Xa was added to obtain a protease/substrate ratio of 1:56 (w/w); and digestion was allowed to proceed at room temperature for 4 h. Protease activity was stopped by the addition of 1 mM phenylmethylsulfonyl fluoride. An equal volume of nickel-nitrilotriacetic acid-agarose was added to the digest, and unbound material containing cleaved rp31 (rp31⌬SH) was recovered, supplemented with 10 mM EGTA, and purified by gel filtration as described above. The rp31⌬SH peak was concentrated by ultrafiltration at 4°C using a YM-3 membrane (Amicon, Inc.) and stored at Ϫ70°C.
Determination of Protein Purity and Concentration-The amino acid composition and concentration of purified rp31 were determined in duplicate amino acid analyses by Dr. Peter Sharratt (Protein and Nucleic Acid Chemistry Facility, Cambridge University, Cambridge, UK). The concentrations of the various purified rp31 mutants were then determined by a comparative reversed-phase high pressure liquid chromatography (RP-HPLC) method. Samples of purified rp31 and its mutant forms were chromatographed on a Vydac C 18 RP-HPLC column (4.6 mm ϫ 25 cm) and eluted with a 9 -27% (v/v) gradient of acetonitrile in 0.1% (v/v) trifluoroacetic acid, with absorbance monitored at 215 nm. Each recombinant protein eluted as a single peak under these conditions. The concentration of each mutant was extrapolated by comparing its peak height upon RP-HPLC with the peak height of a known amount of rp31.
CD Measurements-CD spectra of rp31 and rp31⌬SH were recorded from 260 to 195 nm (or from 260 to 210 nm in the presence of urea) at 20°C in a fused silica cuvette of 1-mm path length using a Jasco J-715 spectropolarimeter. Protein concentrations were ϳ0.1 mg/ml, and the buffer was 5 mM Tris-HCl (pH 8.2) and 67.5 mM NaCl with or without 5 M urea (for rp31) or 16.6 mM Tris-HCl (pH 8.2) and 125 mM NaCl (for rp31⌬SH). The spectra are the averages of multiple scans recorded using a scan speed of 200 nm/min and a response time of 0.25 s. Appropriate buffer blanks were subtracted from all spectra. For deconvolution, spectra were subjected to multicomponent secondary structure analysis using the SELCON3 algorithm (20).
PfSUB-1 Enzyme Activity and Inhibition Assays-Kinetic experiments assessing the effect of recombinant propeptide constructs on the activity of baculovirus-derived rPfSUB-1 were as described previously, based on the cleavage of the tetramethylrhodamine-labeled fluorogenic peptide substrate pepF1-6R (12,21). Briefly, 90 l of pepF1-6R (2 M final concentration) in 50 mM Tris-HCl (pH 8.2), 12 mM CaCl 2 , and 0.05% (w/v) Nonidet P-40 was added to a 3 ϫ 3-mm path length cell. The solution was allowed to stabilize to 37°C in the spectrofluorometer for 2 min before the addition of 10 l of purified rPfSUB-1 (74 nM final concentration). Increase in fluorescence, reflecting cleavage of pepF1-6R, was measured over the ensuing 4 min. To determine the inhibitory activity of the recombinant propeptides, 1 l of different concentrations of the recombinant propeptides or 2 l of an inhibitor control (3,4dichloroisocoumarin at 80 M final concentration) was added directly to the reaction with rapid mixing, and the hydrolysis rate was measured for an additional 5 min. All experiments were performed in triplicate and used concentrations of pepF1-6R well below its K m , under which conditions, K i(app) ϳ K i (22). In the absence of inhibitors, the rate of hydrolysis over this period was linear, and ϳ10% of the total substrate was hydrolyzed. Following each assay, the cell was thoroughly cleaned by treatment with chromic acid.
P. falciparum schizont extracts containing endogenous PfSUB-1 were prepared by successive solubilizations with Triton X-100 and CHAPS as described previously (11,12). To determine inhibition of native PfSUB-1 by rp31, the schizont extract was incubated with an unlabeled synthetic peptide substrate called PEP1 (2 mM final concentration) in the absence or presence of purified rp31 (84 nM final concentration) for 4 h at 37°C, and then cleavage of PEP1 was assessed by RP-HPLC as described previously (12).
Selectivity of Inhibition by rp31-The selectivity of rp31 as a subtilase inhibitor was assessed by measuring its capacity to inhibit the activity of subtilisin BPNЈ (Sigma P-4789) or subtilisin Carlsberg (Sigma P-5380). The assay was based on the hydrolysis of N-succinyl-L-Ala-L-Ala-L-Pro-L-Phe-p-nitroanilide (Suc-AAPF-pNA) by these proteases, essentially as described by DelMar et al. (23), in 50 mM Tris-HCl (pH 8.2), 12 mM CaCl 2 , and 0.05% (w/v) Nonidet P-40 at room temperature. Kinetic measurements were performed using a Unicam UV1 spectroflu-orometer, measuring absorbance at 412 nm. For K m determination, concentrations of Suc-AAPF-pNA between 60 M and 1 mM were used to determine initial hydrolysis rates, which were plotted against substrate concentration, fitting the data to the hyperbolic Michaelis-Menten rate equation by nonlinear regression using the program GraFit 5 (Version 5.0.4, Erithacus Software Ltd., Horley, UK) (24). For inhibition assays, 500 l of the substrate at 60 M in a 10-mm path length cell was supplemented with 2.5 l of subtilisin BPNЈ or subtilisin Carlsberg (25 nM final concentration). Increase in absorbance was measured over the ensuing 15 min to obtain an initial hydrolysis rate, before direct addition to the mixture of various concentrations of purified rp31 (16.8, 42, 84, and 168 nM final concentrations) or phenylmethylsulfonyl fluoride (0.5 mM final concentration) as an inhibitor control. The subsequent hydrolysis rate was measured for an additional 10 min. Note that, in the absence of inhibitors, the rate of hydrolysis was linear over the entire period of measurement, and no more than 10% of the total available substrate was hydrolyzed.
Stability of rp31 and rp31 Mutants-The stability of rp31 and its various mutants to digestion by rPfSUB-1 in the presence of the substrate pepF1-6R was assessed by SDS-PAGE and Western blotting. To 90 l of 2 M pepF1-6R in 50 mM Tris-HCl (pH 8.2), 12 mM CaCl 2 , and 0.05% (w/v) Nonidet P-40 were added 10 l of purified rPfSUB-1 (61.5 nM final concentration) and 1 l of recombinant propeptide at various concentrations. The mixture was incubated at 37°C. At intervals, aliquots were removed, and protease activity was terminated by boiling in Laemmli dissociation buffer containing 100 mM dithiothreitol. All samples were then subjected to SDS-PAGE and electroblotted onto Hybond-C membrane (Amersham Biosciences). The membranes were blocked by incubation in 5% (w/v) nonfat milk in phosphate-buffered saline (PBS) and incubated with mouse anti-rp31 antiserum, followed by incubation with horseradish peroxidase-conjugated anti-mouse IgG antibody. Antigen-antibody interactions were visualized by enhanced chemiluminescence (SuperSignal West Pico chemiluminescent substrate, Pierce).
Binding of rPfSUB-1 to rp31 and rp31 Mutants-Recombinant propeptide constructs, along with molecular mass marker proteins (Amersham Biosciences) as controls for nonspecific binding of PfSUB-1, were subjected to SDS-PAGE, and the proteins were either stained with Coomassie Blue R-250 or transferred to Hybond-C membrane. The membranes were blocked by incubation in 5% (w/v) nonfat milk in PBS and then incubated for 1 h at room temperature in PBS containing purified rPfSUB-1 (ϳ3 nM final concentration) before washing and probing with a rabbit antiserum specific for the catalytic domain of PfSUB-1 (8), followed by horseradish peroxidase-conjugated anti-rabbit IgG antibody. Antigen-antibody interactions were visualized by enhanced chemiluminescence.

Recombinant Expression and Characterization of the
PfSUB-1 Propeptide (rp31)-Plasmid pET-30Xa/LIC carrying the p31 sequence ( Fig. 1A) was transformed into E. coli strain BL21-Gold (DE3) pLysS, and protein expression was induced with isopropyl-␤-D-thiogalactopyranoside as described under "Experimental Procedures." The recombinant product (rp31) is predicted to comprise Lys 26 -Asp 219 of the PfSUB-1 sequence precisely fused to a 46-residue-long N-terminal extension derived from the expression plasmid and containing a His 6 tag and an S tag, followed by a Factor Xa cleavage site. The expressed protein was mostly soluble (data not shown) and was purified to homogeneity (Fig. 1B) using a three-step procedure consisting of chelating nickel, gel filtration, and ion exchange chromatography. The purified product migrated on SDS-polyacrylamide gel as a single species (Fig. 1B). Electrospray mass spectrometry revealed a molecular mass for the pure product of 26,929 Ϯ 2.90 Da, which is in good agreement with its predicted mass of 26,928.63 Da. Amino acid analysis confirmed its identity and purity and also allowed determination of the precise concentration of a large batch of purified rp31, which was 8.4 M (data not shown). Note that rp31 contains no tryptophan residues and only a single tyrosine residue. Based on the algorithm of Pace et al. (26), its predicted molar absorption coefficient at 280 nm in water (⑀ 280 nm ) is therefore only 1490 M Ϫ1 cm Ϫ1 . Using the purified material and assuming 100% purity, the experimentally determined ⑀ 280 nm was found to be 6785 M Ϫ1 cm Ϫ1 . A sample of purified rp31 was treated with Factor Xa to remove the N-terminal His 6 and S tags, and the resulting protein (called rp31⌬SH) was analyzed by CD under nondenaturing conditions in parallel with intact rp31 (Fig. 1C). The secondary structure of rp31⌬SH measured from the CD spectrum was calculated to be 33.4% unordered. The spectrum also displayed some negative ellipticity at 208 and 222 nm, indicating the presence of secondary structure that, upon deconvolution, could be attributed to 19.5% ␣-helix and 26.3% ␤-sheet (Fig. 1C). It is worth noting that, as mentioned above, the PfSUB-1 propeptide sequence is not rich in aromatic residues, side chains of which can contribute to the CD spectrum and provide potentially misleading data (25). Our results indicate that the recombinant propeptide adopts significant secondary structure in isolation, without the presence of the cognate PfSUB-1 catalytic domain. Under similar conditions, the CD spectrum of intact rp31 revealed an only slightly lower degree of secondary structure, consisting of 13.8% ␣-helix and 21.9% ␤-sheet. Importantly, this could be removed by denaturation with 5 M urea (Fig. 1C). Our data indicate that the N-terminal sequence comprising the His 6 and S tags does not contribute to the secondary structure of the protein and, more importantly, does not prevent folding of its propeptide component. This result validated the use of rp31 for the additional studies described below.
Inhibitory Effect of rp31 on Recombinant and Authentic PfSUB-1-The ability of rp31 to inhibit the proteolytic activity of rPfSUB-1 was determined in kinetic assays using the fluorescent substrate pepF1-6R. Preliminary experiments indicated that the addition of a molar excess of rp31 to an ongoing hydrolysis reaction resulted in immediate conversion of the linear progress curve to a less steep steady-state linear curve, with no evidence of a hyperbolic transition period, suggesting that the propeptide can act as a typical rapid equilibrium competitive inhibitor. Additional experiments showed that this effect was reproducible and concentration-dependent ( Fig. 2A). The apparent tight binding inhibition constant (K i(app) ) of the interaction was calculated using the method of Nicklin and Barrett (22), which uses the slope of the progress curve in the absence of the inhibitor under test (the initial velocity, v o ) and the slope of the progress curve immediately following the addition of inhibitor (the inhibition velocity, v i ). A plot of (v o /v i ) Ϫ 1 versus inhibitor concentration has a slope of 1/K i(app) (Fig.  2B). Using this method, the experimentally determined K i(app) was 5.3 Ϯ 0.4 nM (Table I). At the substrate concentration used, which is substantially below its K m , this is unlikely to be significantly different from the K i (21,22). Another E. coli A solution of the fluorogenic substrate pepF1-6R (2 M) was allowed to stabilize to 37°C in the spectrofluorometer for 2 min before the addition of purified rPfSUB-1 (74 nM). Increase in fluorescence was measured at 550 nm over the ensuing 4 min. One microliter of purified rp31 was then added with rapid mixing to achieve a variety of different final concentrations (shown in the inset), and the hydrolysis rate was measured for an additional 5 min. The arrows represent the points of addition to the reaction mixture of rPfSUB-1 and rp31. B, method for determination of K i(app) for inhibition of rPfSUB-1 by rp31. For each concentration of rp31, the initial hydrolysis rate (v o ) and the inhibition hydrolysis rate (v i ) were obtained from progress curves as shown in A. The slope of the (v o /v i ) Ϫ 1 versus propeptide concentration curve is equal to 1/K i(app) . C, inhibition of authentic PfSUB-1 activity by rp31.
The Triton X-100-insoluble fraction of P. falciparum schizonts was incubated with the substrate PEP1 (2 mM final concentration) for 4 h at 37°C in the presence or absence of 84 nM purified rp31 before analysis of the digestion mixture by RP-HPLC as described previously (12). Cleavage of PEP1 by endogenous PfSUB-1 resulted in its conversion to two peptide products that eluted at 9 and 9.5 min. Note that prolonged incubation with the schizont extract resulted in complete conversion to the cleavage products (data not shown). derived recombinant fusion protein possessing the same Nterminal fusion partner as rp31, but containing an irrelevant malarial protein sequence, displayed no detectable inhibitory activity against rPfSUB-1 (data not shown).
To evaluate any effect of the N-terminal His 6 and S tagcontaining sequence on the inhibition mediated by rp31, we next investigated the inhibitory properties of rp31⌬SH, which comprises exclusively the PfSUB-1 propeptide sequence (Lys 26 -Asp 219 ). The determined K i(app) for inhibition by rp31⌬SH was 12.5 nM (Table I). This value is slightly higher than that found for rp31, but is still in the low nanomolar range and confirms that the inhibition mediated by rp31 is due to the propeptide component of its sequence. The difference between this value and that found for intact rp31 is most likely the result of experimental error in determination of the protein concentration of rp31⌬SH; as shown in Fig. 1C (inset), there was a degree of unavoidable heterogeneity in the rp31⌬SH preparation, presumably caused by limited cleavage at secondary sites during digestion of rp31 with Factor Xa. The presence of minor truncated species in the preparation that do not interact with rPfSUB-1 would have the effect of lowering the specific activity of the preparation. Additionally, any weakly interacting forms might compete with intact rp31⌬SH for binding to the protease, also effectively leading to an underestimate of the inhibition constant. Whatever the case, our data show unambiguously that the PfSUB-1 propeptide is a fast binding, high affinity inhibitor of rPfSUB-1.
We have previously shown that the Triton X-100-insoluble fraction of P. falciparum schizont extracts contains endogenous PfSUB-1 activity (11,12). The addition of the unlabeled peptide substrate PEP1 to these extracts results in its conversion to two dominant cleavage products diagnostic of cleavage by PfSUB-1. Treatment of the schizont extract with purified rp31 was found to completely deplete the activity (Fig. 2C). This result shows that the activity of authentic, parasite-derived PfSUB-1 can be potently inhibited by the propeptide. This result is of particular interest because, whereas baculovirusderived rPfSUB-1 is secreted in the p54 form, endogenous PfSUB-1 in schizont extracts is predominantly in the form of the terminal p47 processing product (11,12). Our results therefore indicate that rp31 is able to bind to and inhibit both p54 and p47 forms of PfSUB-1, showing, in turn, that conversion of p54 to p47 does not per se abolish the ability of the protease to interact with its propeptide.
Selectivity of Inhibition by rp31-The PfSUB-1 catalytic domain is phylogenetically most closely related to the families of bacterial subtilisins and bacterial pyrolysin subtilases (8,12,27). Members of these families sharing the highest amino acid homology with PfSUB-1 include subtilisin BPNЈ and subtilisin Carlsberg (32 and 31% identities, respectively, within the catalytic domains) (12). To determine whether the inhibitory activity of rp31 is selective for its cognate enzyme, we investigated its effect on the activity of these two bacterial subtilisins in kinetic assays by monitoring the effects of rp31 on hydrolysis of Suc-AAPF-pNA as described under "Experimental Procedures." Inhibition of both enzymes by rp31 was substantially poorer than that seen with rPfSUB-1, with calculated K i(app) values of 0.33 Ϯ 0.02 M for subtilisin BPNЈ and 0.36 Ϯ 0.05 M for subtilisin Carlsberg. The experimentally determined K m values for cleavage of Suc-AAPF-pNA by these proteases under the conditions used were 242 Ϯ 8 M for subtilisin BPNЈ and 434 Ϯ 51 M for subtilisin Carlsberg (data not shown), so the above K i(app) values convert to K i values of ϳ0.26 M for subtilisin BPNЈ and 0.32 M for subtilisin Carlsberg (22). Relative to inhibition of rPfSUB-1 (K i ϳ 5.3 nM), these values correspond to selectivity indices of ϳ49 and 60, respectively. Our data suggest that rp31 is a relatively selective inhibitor of its cognate enzyme.
Effect of Point Mutations and C-terminal Truncation on the Inhibitory Properties of rp31-The S1 and S4 binding pockets are thought to dominate substrate preference in subtilases (27). Examination of the amino acid sequences flanking the cleavage sites used during autocatalytic processing of PfSUB-1 has revealed two conserved residues at the P1 (Asp) and P4 (Val) positions relative to the scissile bonds (11). Subsequent studies have shown that, although the P1 Asp is not crucial for substrate cleavage, it is the preferred residue at this position for efficient substrate cleavage in trans by PfSUB-1, and its replacement with Leu completely prevents cleavage (12). Similarly, the P4 Val plays an important role in substrate recognition (12). On the basis that these residues might also be important for the inhibition mediated by rp31, we generated, by mutagenesis, point mutations in appropriate positions of the recombinant propeptide. Three mutant forms of rp31 were investigated: a mutant with replacement of P4 Val 216 with Ala (rp31-ASAD), a mutant with replacement of P1 Asp 219 with Leu (rp31-VSAL), and a double mutant with replacement of both P1 Asp 219 with Leu and P4 Val 216 with Ala (rp31-ASAL) ( Fig. 1; see Fig. 4A, third through fifth lanes).
Inhibition assays with the mutant proteins showed that all remained potent inhibitors of rPfSUB-1, with calculated K i(app) values of 5.1 Ϯ 0.3, 2.6 Ϯ 1.3, and 11.9 Ϯ 3.6 nM for rp31-ASAD, rp31-VSAL, and rp31-ASAL, respectively (Table I). These results were unexpected, indicating that the modified residues play little if any role in the potent inhibition mediated by the propeptide. To further explore the mechanism of inhibition by rp31, we therefore constructed, expressed, and purified to homogeneity another construct identical to rp31 except for the truncation of 11 residues (Leu 209 -Asp 219 ) from its C terminus ( Fig. 1; see Fig. 4A, second lane). Surprisingly, this construct (called rp31⌬11) also inhibited the activity of rPfSUB-1, although with a considerably reduced efficiency compared with rp31 or the point mutants described above, with a calculated K i(app) of 76.6 Ϯ 15.9 nM (Table I). This indicates a requirement for interactions between the C-terminal "tail" of the propeptide and PfSUB-1 for optimal inhibition, but also shows that other interactions involving residues upstream of this domain play an important role in binding of the propeptide to the catalytic domain.
Stability of rp31 and rp31 Mutants-The propeptides of certain subtilases are readily and rapidly digested by their cognate enzymes. One practical consequence of this it that it can lead to potential inherent errors when attempting to experimentally define inhibition constants due to a time-dependent decrease in the concentration of the propeptide (e.g. Ref. 28). Although rp31 and its various mutants did not exhibit any discernible time-dependent decrease in their inhibitory capacity during the inhibition assays described above, we felt it important to assess the capacity of the propeptide to act as a values for inhibition of rPfSUB-1 by rp31 and rp31 mutants Inhibition of rPfSUB-1 activity by purified rp31, rp31⌬SH, rp31⌬11, and rp31 mutants was measured as described under "Experimental Procedures" and in the legend to Fig. 2. Each K i (app) value shown is the mean Ϯ SD of three completely independent determinations. Functional Characterization of the PfSUB-1 Propeptide substrate for rPfSUB-1 and to verify that the concentration of each polypeptide was stable during the assay. To assess this, purified rp31 or the various mutant propeptides were incubated at 37°C in the presence of the substrate pepF1-6R and rPfSUB-1 under conditions identical to those used in the assays performed to measure their inhibitory activity. Two concentrations of each propeptide were tested, corresponding to the lowest and highest concentrations used in the inhibition assays. Fig. 3 shows that, even at the lowest propeptide concentration used, the concentration of each propeptide remained stable over time, and no digestion products could be detected upon SDS-PAGE analysis. Similar results were obtained using the highest concentration of each propeptide used in the inhibition assays (data not shown). These data clearly demonstrate that rp31 is completely resistant to digestion by PfSUB-1 over the time course of the assay. Moreover, the truncation and the various mutations made to the propeptide did not reduce this resistance to digestion.
Binding of rp31 and rp31 Mutants to Baculovirus-derived PfSUB-1-The above inhibition data clearly imply tight binding of rp31 and its various derivatives to rPfSUB-1. To directly visualize this, we used a modification of a binding assay described by Kessler and Safrin (29) and Markaryan et al. (30). The various recombinant propeptides were subjected to SDS-PAGE and electroblotted onto a membrane. The membrane was subsequently incubated with PBS containing purified rPfSUB-1 and then probed with a rabbit antiserum specific for the PfSUB-1 catalytic domain. A single band was observed in each case (Fig. 4B), and comparison with Coomassie Bluestained duplicate gels showed that, in each case, the position of the band corresponded exactly to the position of migration of the corresponding propeptide construct (Fig. 4A). No signal was observed on control blots, which were not probed with rPfSUB-1 (Fig. 4C), demonstrating that the signals observed in Fig. 4B were not due to nonspecific recognition of the propeptides by the rabbit antibodies or the horseradish peroxidaseconjugated anti-rabbit IgG secondary antibodies used. The specificity of PfSUB-1 binding was demonstrated using a set of standard molecular mass marker proteins, all loaded at a higher concentration than the propeptide constructs. Two species in the marker lanes, corresponding to phosphorylase b (94 kDa) and carbonic anhydrase (30 kDa), were visualized on the blot probed with rPfSUB-1 (Fig. 4B); however, both were also detected on the control blot (Fig. 4C), indicating that the recognition of these proteins was solely due to nonspecific binding of either the primary antibody or the horseradish peroxidaseconjugated secondary antibody and not to nonspecific binding of rPfSUB-1. Our results provide direct support for the above inhibition data, demonstrating that rPfSUB-1 is able to interact with all of the recombinant propeptide constructs in a highly specific manner. DISCUSSION In several protease families, propeptides function as intramolecular chaperones that are essential for correct folding of the catalytic domain during secretory transport. Propeptides are also often potent inhibitors of their associated catalytic domains and are thought to act as regulators of protease activity during secretory transport, preventing potentially damag- ing premature activation (13, 29 -32). Initial studies on secretion of rPfSUB-1 from insect cells demonstrated that at least a proportion of the secreted protease remains noncovalently linked with its propeptide, p31 (11). These early observations raised the question of the role of p31 in the activation and function of PfSUB-1 and prompted us to examine its structure and activity in detail.
CD analysis of the recombinant propeptide both with (rp31) and without (rp31⌬SH) an N-terminal fusion peptide derived from the expression plasmid demonstrated that it possesses secondary structure in the absence of its cognate catalytic domain. Previous studies by others have indicated that subtilase propeptides broadly divide into two groups according to their folding capacity. Propeptides of bacterial subtilisin BPNЈ and subtilisin Carlsberg are unfolded when expressed on their own (17,33,34), whereas propeptides of aqualysin I and of some of the subtilisin-like proprotein convertases (SPCs) adopt significant secondary structure (18,35,36). Thus, in the case of aqualysin I, the isolated propeptide was estimated to be composed of 21% ␣-helix and 24% ␤-sheet (35). The propeptides of several SPCs have been shown to possess a similar degree of secondary structure, with, for example, the propeptide of mouse SPC3 being 22% ␣-helical and 30% ␤-sheet (18). The observed ␣-␤ content of rp31 is in good agreement with these values. It has been suggested that propeptides with significant secondary structure are more stable and more efficient in both folding and inhibiting the catalytic domain than those that lack structure. In the case of subtilisin BPNЈ, Kojima et al. (33) demonstrated, through a series of amino acid replacements, that an increase in secondary structure in the propeptide could be correlated with a slower rate of decrease in the inhibitory activity and a greater resistance to proteolytic digestion in the presence of the cognate catalytic domain, both leading to a lower K i than that for the wild-type propeptide. Likewise, the aqualysin I propeptide is more efficient in the folding and inhibition of subtilisin BPNЈ compared with the subtilisin BPNЈ propeptide itself, which lacks structure (35). Unstructured propeptides are thought to adopt structure only upon interaction with their cognate catalytic domains (34,37), so these observations are consistent with a model in which possession of some pre-existing structure accelerates productive binding to the catalytic domain because the structural transition required for binding is minimized (35). Our CD data indicate that the PfSUB-1 propeptide belongs to the subclass of subtilase propeptides that can fold in isolation from their cognate proteases, and our finding that rp31 is a fast binding, high affinity inhibitor of rPfSUB-1 is consistent with this.
Subtilase propeptides are not necessarily selective inhibitors of their cognate enzymes and, in some cases, are able to potently inhibit other subtilisins (14,17,18,35). Indeed, the aqualysin I propeptide is a better inhibitor of subtilisin BPNЈ compared with the subtilisin BPNЈ propeptide (35), and the subtilisin E propeptide inhibits both subtilisin BPNЈ and subtilisin Carlsberg with K i values in the low nanomolar range (18). The degree of sequence homology between bacterial propeptides correlates with inhibitory activity against the heterologous protease domain (e.g. Ref. 17). We observed 49 -60fold higher K i(app) values for inhibition of subtilisin BPNЈ and subtilisin Carlsberg by rp31 compared with its inhibition of PfSUB-1, suggesting that the propeptide has some selectivity for its cognate enzyme. There is no obvious sequence similarity between p31 and the subtilisin BPNЈ and subtilisin Carlsberg propeptides (8), so this result was not unexpected.
X-ray crystal structures of two bacterial subtilisin-propeptide complexes, stabilized by appropriate mutation of the active-site serine, have shown that the bound propeptide forms a structure consisting of two ␣-helices and a four-stranded ␤-sheet. This ␤-sheet packs against two parallel surface ␣-helices of the catalytic domain, whereas the C-terminal tail of the propeptide lies in a substrate-like manner within the activesite cleft of the enzyme, blocking access of exogenous substrate. The complex is stabilized by numerous hydrogen bonds between the two polypeptides, several of which involve the four C-terminal residues of the propeptide tail in the active-site cleft (38,39). A recent NMR structure of the 83-residue-long PC1 propeptide showed that the isolated molecule adopts a fold similar to that of the enzyme-bound bacterial propeptide, despite the absence of any sequence homology to the bacterial protein (36). Homology modeling of the PfSUB-1 catalytic domain has revealed an overall structure similar to that of subtilisin BPNЈ, including the presence of the two surface ␣-helices (12), so the manner in which PfSUB-1 interacts with its propeptide is probably broadly similar to that of the bacterial subtilisins. Our finding that point mutations within the four residues at the C terminus of rp31 had little effect on its inhibitory capacity and that even removal of 11 residues from the C terminus did not abolish inhibition (although there was an ϳ14-fold increase in K i(app) ) was therefore unexpected. Our data clearly demonstrate that, in the case of the PfSUB-1 propeptide, interactions within the active-site cleft may be important for optimal inhibition, but are not essential for complex formation. Since inhibition presumably involves steric blockade of the PfSUB-1 active-site groove, it is not clear how this is mediated by the truncated propeptide; however, it is important to note that p31 is substantially larger than any of the subtilase propeptides whose three-dimensional structures have been solved, and it is possible that, in the complex formed between rPfSUB-1 and rp31⌬11, structures formed by propeptide residues distant in the primary sequence from its C terminus can occlude access to the active site.
Some subtilisin propeptides are efficiently degraded by their cognate enzymes (e.g. Refs. 17 and 27). This can cause practical difficulties in kinetic measurements, during which the propeptide concentration needs to remain stable. It may also be important for eventual autocatalytic removal of the propeptide from the protease catalytic domain during activation. Our finding that rp31 was not degraded over the course of the inhibition assays validates the accuracy of the K i(app) values calculated for all of the propeptide constructs. It is also consistent with previous indications that PfSUB-1 has a fairly stringent substrate specificity. Studies using synthetic peptide substrates have indicated that the favored minimal recognition sequence for PfSUB-1 is Val-X-X-Asp2, where X is any amino acid residue (12). No such motif occurs within p31 (besides that at the extreme C terminus). This apparent specificity of PfSUB-1 raises the question of how activation of the protease ultimately takes place during trafficking through the parasite secretory transport system since it presumably requires removal of the propeptide. In many cases, including that of the SPC furin, subtilase propeptide removal involves its proteolytic cleavage, mediated by the cognate protease in cis or in trans (40,41). Our data suggest that this is unlikely to be so in the case of Pf-SUB-1. We have previously shown that, in the parasite, p54 is bound to its propeptide (11) and that conversion of p54 to p47 occurs only in a post-endoplasmic reticulum compartment (8). Our demonstration here that the propeptide can bind to and inhibit parasite-derived p47 implies that, whatever the function of the conversion of p54 to p47, this processing step is insufficient per se to remove the bound propeptide. Propeptide release may be mediated by a simple environmental trigger (such as a change in pH or calcium concentration) associated with transport beyond the parasite endoplasmic reticulum.
The identity of the physiological substrate(s) of PfSUB-1 is unknown. Given the apparent high substrate specificity of the enzyme, it is unlikely to be involved in degradative processes; a role in specific post-translational protein modifications is more likely, which could take place either within the secretory organelles in which p47 accumulates or following protease release during erythrocyte invasion. We currently favor the former possibility since we have found that the addition of high concentrations of rp31 to P. falciparum cultures has no effect on the efficiency of erythrocyte invasion, and extensive experiments have indicated that rPfSUB-1 has no detectable capacity to modify surface proteins of the intact merozoite or the host red blood cell under nondenaturing conditions. 2 In light of the findings described here, rp31 could be used to regulate the activity of PfSUB-1 in a specific and informative manner; it may be possible, for example, to subtly and specifically downregulate levels of functional PfSUB-1 in the parasite by constitutive overexpression of the propeptide from a suitable transgene (42). The demonstration that rp31 inhibits both authentic and baculovirus-derived PfSUB-1 validates the use of the recombinant protease as a tool for high throughput screening for low molecular mass inhibitors (21). Defining the important inhibitory sequences in rp31 could aid in developing potent inhibitors with potential as antimalarial compounds. Indeed, endogenous inhibitors are generally a good starting point for further development of smaller but highly selective inhibitors (43)(44)(45)(46)(47).